annotate src/spellsuggest.c @ 35140:9ba95f736abb default tip

Added tag v9.1.0399 for changeset 617f9087f97c05289903b9e88f487ef74d8869fc
author Christian Brabandt <cb@256bit.org>
date Wed, 08 May 2024 22:30:04 +0200
parents fd15dc02c223
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
32670
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1 /* vi:set ts=8 sts=4 sw=4 noet:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2 *
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3 * VIM - Vi IMproved by Bram Moolenaar
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4 *
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
5 * Do ":help uganda" in Vim to read copying and usage conditions.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
6 * Do ":help credits" in Vim to see a list of people who contributed.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
7 * See README.txt for an overview of the Vim source code.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
8 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
9
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
10 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
11 * spellsuggest.c: functions for spelling suggestions
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
12 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
13
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
14 #include "vim.h"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
15
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
16 #if defined(FEAT_SPELL) || defined(PROTO)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
17
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
18 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
19 * Use this to adjust the score after finding suggestions, based on the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
20 * suggested word sounding like the bad word. This is much faster than doing
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
21 * it for every possible suggestion.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
22 * Disadvantage: When "the" is typed as "hte" it sounds quite different ("@"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
23 * vs "ht") and goes down in the list.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
24 * Used when 'spellsuggest' is set to "best".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
25 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
26 #define RESCORE(word_score, sound_score) ((3 * (word_score) + (sound_score)) / 4)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
27
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
28 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
29 * Do the opposite: based on a maximum end score and a known sound score,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
30 * compute the maximum word score that can be used.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
31 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
32 #define MAXSCORE(word_score, sound_score) ((4 * (word_score) - (sound_score)) / 3)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
33
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
34 // only used for su_badflags
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
35 #define WF_MIXCAP 0x20 // mix of upper and lower case: macaRONI
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
36
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
37 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
38 * Information used when looking for suggestions.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
39 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
40 typedef struct suginfo_S
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
41 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
42 garray_T su_ga; // suggestions, contains "suggest_T"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
43 int su_maxcount; // max. number of suggestions displayed
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
44 int su_maxscore; // maximum score for adding to su_ga
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
45 int su_sfmaxscore; // idem, for when doing soundfold words
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
46 garray_T su_sga; // like su_ga, sound-folded scoring
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
47 char_u *su_badptr; // start of bad word in line
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
48 int su_badlen; // length of detected bad word in line
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
49 int su_badflags; // caps flags for bad word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
50 char_u su_badword[MAXWLEN]; // bad word truncated at su_badlen
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
51 char_u su_fbadword[MAXWLEN]; // su_badword case-folded
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
52 char_u su_sal_badword[MAXWLEN]; // su_badword soundfolded
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
53 hashtab_T su_banned; // table with banned words
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
54 slang_T *su_sallang; // default language for sound folding
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
55 } suginfo_T;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
56
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
57 // One word suggestion. Used in "si_ga".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
58 typedef struct suggest_S
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
59 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
60 char_u *st_word; // suggested word, allocated string
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
61 int st_wordlen; // STRLEN(st_word)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
62 int st_orglen; // length of replaced text
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
63 int st_score; // lower is better
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
64 int st_altscore; // used when st_score compares equal
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
65 int st_salscore; // st_score is for soundalike
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
66 int st_had_bonus; // bonus already included in score
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
67 slang_T *st_slang; // language used for sound folding
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
68 } suggest_T;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
69
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
70 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
71
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
72 // TRUE if a word appears in the list of banned words.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
73 #define WAS_BANNED(su, word) (!HASHITEM_EMPTY(hash_find(&(su)->su_banned, word)))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
74
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
75 // Number of suggestions kept when cleaning up. We need to keep more than
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
76 // what is displayed, because when rescore_suggestions() is called the score
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
77 // may change and wrong suggestions may be removed later.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
78 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 130 ? 150 : (su)->su_maxcount + 20)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
79
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
80 // Threshold for sorting and cleaning up suggestions. Don't want to keep lots
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
81 // of suggestions that are not going to be displayed.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
82 #define SUG_MAX_COUNT(su) (SUG_CLEAN_COUNT(su) + 50)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
83
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
84 // score for various changes
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
85 #define SCORE_SPLIT 149 // split bad word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
86 #define SCORE_SPLIT_NO 249 // split bad word with NOSPLITSUGS
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
87 #define SCORE_ICASE 52 // slightly different case
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
88 #define SCORE_REGION 200 // word is for different region
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
89 #define SCORE_RARE 180 // rare word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
90 #define SCORE_SWAP 75 // swap two characters
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
91 #define SCORE_SWAP3 110 // swap two characters in three
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
92 #define SCORE_REP 65 // REP replacement
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
93 #define SCORE_SUBST 93 // substitute a character
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
94 #define SCORE_SIMILAR 33 // substitute a similar character
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
95 #define SCORE_SUBCOMP 33 // substitute a composing character
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
96 #define SCORE_DEL 94 // delete a character
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
97 #define SCORE_DELDUP 66 // delete a duplicated character
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
98 #define SCORE_DELCOMP 28 // delete a composing character
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
99 #define SCORE_INS 96 // insert a character
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
100 #define SCORE_INSDUP 67 // insert a duplicate character
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
101 #define SCORE_INSCOMP 30 // insert a composing character
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
102 #define SCORE_NONWORD 103 // change non-word to word char
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
103
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
104 #define SCORE_FILE 30 // suggestion from a file
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
105 #define SCORE_MAXINIT 350 // Initial maximum score: higher == slower.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
106 // 350 allows for about three changes.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
107
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
108 #define SCORE_COMMON1 30 // subtracted for words seen before
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
109 #define SCORE_COMMON2 40 // subtracted for words often seen
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
110 #define SCORE_COMMON3 50 // subtracted for words very often seen
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
111 #define SCORE_THRES2 10 // word count threshold for COMMON2
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
112 #define SCORE_THRES3 100 // word count threshold for COMMON3
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
113
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
114 // When trying changed soundfold words it becomes slow when trying more than
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
115 // two changes. With less than two changes it's slightly faster but we miss a
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
116 // few good suggestions. In rare cases we need to try three of four changes.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
117 #define SCORE_SFMAX1 200 // maximum score for first try
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
118 #define SCORE_SFMAX2 300 // maximum score for second try
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
119 #define SCORE_SFMAX3 400 // maximum score for third try
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
120
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
121 #define SCORE_BIG (SCORE_INS * 3) // big difference
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
122 #define SCORE_MAXMAX 999999 // accept any score
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
123 #define SCORE_LIMITMAX 350 // for spell_edit_score_limit()
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
124
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
125 // for spell_edit_score_limit() we need to know the minimum value of
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
126 // SCORE_ICASE, SCORE_SWAP, SCORE_DEL, SCORE_SIMILAR and SCORE_INS
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
127 #define SCORE_EDIT_MIN SCORE_SIMILAR
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
128
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
129 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
130 * For finding suggestions: At each node in the tree these states are tried:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
131 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
132 typedef enum
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
133 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
134 STATE_START = 0, // At start of node check for NUL bytes (goodword
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
135 // ends); if badword ends there is a match, otherwise
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
136 // try splitting word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
137 STATE_NOPREFIX, // try without prefix
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
138 STATE_SPLITUNDO, // Undo splitting.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
139 STATE_ENDNUL, // Past NUL bytes at start of the node.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
140 STATE_PLAIN, // Use each byte of the node.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
141 STATE_DEL, // Delete a byte from the bad word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
142 STATE_INS_PREP, // Prepare for inserting bytes.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
143 STATE_INS, // Insert a byte in the bad word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
144 STATE_SWAP, // Swap two bytes.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
145 STATE_UNSWAP, // Undo swap two characters.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
146 STATE_SWAP3, // Swap two characters over three.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
147 STATE_UNSWAP3, // Undo Swap two characters over three.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
148 STATE_UNROT3L, // Undo rotate three characters left
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
149 STATE_UNROT3R, // Undo rotate three characters right
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
150 STATE_REP_INI, // Prepare for using REP items.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
151 STATE_REP, // Use matching REP items from the .aff file.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
152 STATE_REP_UNDO, // Undo a REP item replacement.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
153 STATE_FINAL // End of this node.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
154 } state_T;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
155
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
156 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
157 * Struct to keep the state at each level in suggest_try_change().
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
158 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
159 typedef struct trystate_S
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
160 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
161 state_T ts_state; // state at this level, STATE_
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
162 int ts_score; // score
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
163 idx_T ts_arridx; // index in tree array, start of node
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
164 short ts_curi; // index in list of child nodes
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
165 char_u ts_fidx; // index in fword[], case-folded bad word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
166 char_u ts_fidxtry; // ts_fidx at which bytes may be changed
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
167 char_u ts_twordlen; // valid length of tword[]
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
168 char_u ts_prefixdepth; // stack depth for end of prefix or
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
169 // PFD_PREFIXTREE or PFD_NOPREFIX
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
170 char_u ts_flags; // TSF_ flags
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
171 char_u ts_tcharlen; // number of bytes in tword character
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
172 char_u ts_tcharidx; // current byte index in tword character
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
173 char_u ts_isdiff; // DIFF_ values
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
174 char_u ts_fcharstart; // index in fword where badword char started
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
175 char_u ts_prewordlen; // length of word in "preword[]"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
176 char_u ts_splitoff; // index in "tword" after last split
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
177 char_u ts_splitfidx; // "ts_fidx" at word split
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
178 char_u ts_complen; // nr of compound words used
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
179 char_u ts_compsplit; // index for "compflags" where word was spit
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
180 char_u ts_save_badflags; // su_badflags saved here
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
181 char_u ts_delidx; // index in fword for char that was deleted,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
182 // valid when "ts_flags" has TSF_DIDDEL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
183 } trystate_T;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
184
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
185 // values for ts_isdiff
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
186 #define DIFF_NONE 0 // no different byte (yet)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
187 #define DIFF_YES 1 // different byte found
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
188 #define DIFF_INSERT 2 // inserting character
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
189
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
190 // values for ts_flags
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
191 #define TSF_PREFIXOK 1 // already checked that prefix is OK
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
192 #define TSF_DIDSPLIT 2 // tried split at this point
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
193 #define TSF_DIDDEL 4 // did a delete, "ts_delidx" has index
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
194
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
195 // special values ts_prefixdepth
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
196 #define PFD_NOPREFIX 0xff // not using prefixes
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
197 #define PFD_PREFIXTREE 0xfe // walking through the prefix tree
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
198 #define PFD_NOTSPECIAL 0xfd // highest value that's not special
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
199
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
200 static long spell_suggest_timeout = 5000;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
201
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
202 static void spell_find_suggest(char_u *badptr, int badlen, suginfo_T *su, int maxcount, int banbadword, int need_cap, int interactive);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
203 #ifdef FEAT_EVAL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
204 static void spell_suggest_expr(suginfo_T *su, char_u *expr);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
205 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
206 static void spell_suggest_file(suginfo_T *su, char_u *fname);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
207 static void spell_suggest_intern(suginfo_T *su, int interactive);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
208 static void spell_find_cleanup(suginfo_T *su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
209 static void suggest_try_special(suginfo_T *su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
210 static void suggest_try_change(suginfo_T *su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
211 static void suggest_trie_walk(suginfo_T *su, langp_T *lp, char_u *fword, int soundfold);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
212 static void go_deeper(trystate_T *stack, int depth, int score_add);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
213 static void find_keepcap_word(slang_T *slang, char_u *fword, char_u *kword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
214 static void score_comp_sal(suginfo_T *su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
215 static void score_combine(suginfo_T *su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
216 static int stp_sal_score(suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
217 static void suggest_try_soundalike_prep(void);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
218 static void suggest_try_soundalike(suginfo_T *su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
219 static void suggest_try_soundalike_finish(void);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
220 static void add_sound_suggest(suginfo_T *su, char_u *goodword, int score, langp_T *lp);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
221 static int soundfold_find(slang_T *slang, char_u *word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
222 static int similar_chars(slang_T *slang, int c1, int c2);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
223 static void add_suggestion(suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus, slang_T *slang, int maxsf);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
224 static void check_suggestions(suginfo_T *su, garray_T *gap);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
225 static void add_banned(suginfo_T *su, char_u *word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
226 static void rescore_suggestions(suginfo_T *su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
227 static void rescore_one(suginfo_T *su, suggest_T *stp);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
228 static int cleanup_suggestions(garray_T *gap, int maxscore, int keep);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
229 static int soundalike_score(char_u *goodsound, char_u *badsound);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
230 static int spell_edit_score(slang_T *slang, char_u *badword, char_u *goodword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
231 static int spell_edit_score_limit(slang_T *slang, char_u *badword, char_u *goodword, int limit);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
232 static int spell_edit_score_limit_w(slang_T *slang, char_u *badword, char_u *goodword, int limit);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
233
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
234 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
235 * Return TRUE when the sequence of flags in "compflags" plus "flag" can
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
236 * possibly form a valid compounded word. This also checks the COMPOUNDRULE
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
237 * lines if they don't contain wildcards.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
238 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
239 static int
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
240 can_be_compound(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
241 trystate_T *sp,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
242 slang_T *slang,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
243 char_u *compflags,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
244 int flag)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
245 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
246 // If the flag doesn't appear in sl_compstartflags or sl_compallflags
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
247 // then it can't possibly compound.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
248 if (!byte_in_str(sp->ts_complen == sp->ts_compsplit
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
249 ? slang->sl_compstartflags : slang->sl_compallflags, flag))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
250 return FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
251
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
252 // If there are no wildcards, we can check if the flags collected so far
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
253 // possibly can form a match with COMPOUNDRULE patterns. This only
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
254 // makes sense when we have two or more words.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
255 if (slang->sl_comprules != NULL && sp->ts_complen > sp->ts_compsplit)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
256 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
257 int v;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
258
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
259 compflags[sp->ts_complen] = flag;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
260 compflags[sp->ts_complen + 1] = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
261 v = match_compoundrule(slang, compflags + sp->ts_compsplit);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
262 compflags[sp->ts_complen] = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
263 return v;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
264 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
265
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
266 return TRUE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
267 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
268
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
269 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
270 * Adjust the score of common words.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
271 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
272 static int
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
273 score_wordcount_adj(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
274 slang_T *slang,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
275 int score,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
276 char_u *word,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
277 int split) // word was split, less bonus
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
278 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
279 hashitem_T *hi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
280 wordcount_T *wc;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
281 int bonus;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
282 int newscore;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
283
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
284 hi = hash_find(&slang->sl_wordcount, word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
285 if (HASHITEM_EMPTY(hi))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
286 return score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
287
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
288 wc = HI2WC(hi);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
289 if (wc->wc_count < SCORE_THRES2)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
290 bonus = SCORE_COMMON1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
291 else if (wc->wc_count < SCORE_THRES3)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
292 bonus = SCORE_COMMON2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
293 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
294 bonus = SCORE_COMMON3;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
295 if (split)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
296 newscore = score - bonus / 2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
297 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
298 newscore = score - bonus;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
299 if (newscore < 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
300 return 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
301 return newscore;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
302 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
303
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
304 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
305 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
306 * capital. So that make_case_word() can turn WOrd into Word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
307 * Add ALLCAP for "WOrD".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
308 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
309 static int
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
310 badword_captype(char_u *word, char_u *end)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
311 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
312 int flags = captype(word, end);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
313 int c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
314 int l, u;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
315 int first;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
316 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
317
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
318 if (!(flags & WF_KEEPCAP))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
319 return flags;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
320
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
321 // Count the number of UPPER and lower case letters.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
322 l = u = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
323 first = FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
324 for (p = word; p < end; MB_PTR_ADV(p))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
325 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
326 c = PTR2CHAR(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
327 if (SPELL_ISUPPER(c))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
328 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
329 ++u;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
330 if (p == word)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
331 first = TRUE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
332 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
333 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
334 ++l;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
335 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
336
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
337 // If there are more UPPER than lower case letters suggest an
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
338 // ALLCAP word. Otherwise, if the first letter is UPPER then
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
339 // suggest ONECAP. Exception: "ALl" most likely should be "All",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
340 // require three upper case letters.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
341 if (u > l && u > 2)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
342 flags |= WF_ALLCAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
343 else if (first)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
344 flags |= WF_ONECAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
345
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
346 if (u >= 2 && l >= 2) // maCARONI maCAroni
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
347 flags |= WF_MIXCAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
348
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
349 return flags;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
350 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
351
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
352 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
353 * Opposite of offset2bytes().
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
354 * "pp" points to the bytes and is advanced over it.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
355 * Returns the offset.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
356 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
357 static int
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
358 bytes2offset(char_u **pp)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
359 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
360 char_u *p = *pp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
361 int nr;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
362 int c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
363
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
364 c = *p++;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
365 if ((c & 0x80) == 0x00) // 1 byte
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
366 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
367 nr = c - 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
368 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
369 else if ((c & 0xc0) == 0x80) // 2 bytes
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
370 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
371 nr = (c & 0x3f) - 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
372 nr = nr * 255 + (*p++ - 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
373 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
374 else if ((c & 0xe0) == 0xc0) // 3 bytes
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
375 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
376 nr = (c & 0x1f) - 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
377 nr = nr * 255 + (*p++ - 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
378 nr = nr * 255 + (*p++ - 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
379 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
380 else // 4 bytes
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
381 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
382 nr = (c & 0x0f) - 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
383 nr = nr * 255 + (*p++ - 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
384 nr = nr * 255 + (*p++ - 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
385 nr = nr * 255 + (*p++ - 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
386 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
387
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
388 *pp = p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
389 return nr;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
390 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
391
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
392 // values for sps_flags
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
393 #define SPS_BEST 1
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
394 #define SPS_FAST 2
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
395 #define SPS_DOUBLE 4
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
396
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
397 static int sps_flags = SPS_BEST; // flags from 'spellsuggest'
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
398 static int sps_limit = 9999; // max nr of suggestions given
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
399
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
400 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
401 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
402 * Sets "sps_flags" and "sps_limit".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
403 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
404 int
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
405 spell_check_sps(void)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
406 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
407 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
408 char_u *s;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
409 char_u buf[MAXPATHL];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
410 int f;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
411
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
412 sps_flags = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
413 sps_limit = 9999;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
414
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
415 for (p = p_sps; *p != NUL; )
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
416 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
417 copy_option_part(&p, buf, MAXPATHL, ",");
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
418
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
419 f = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
420 if (VIM_ISDIGIT(*buf))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
421 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
422 s = buf;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
423 sps_limit = getdigits(&s);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
424 if (*s != NUL && !VIM_ISDIGIT(*s))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
425 f = -1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
426 }
33399
95db67c7b754 patch 9.0.1958: cannot complete option values
Christian Brabandt <cb@256bit.org>
parents: 32670
diff changeset
427 // Note: Keep this in sync with p_sps_values.
32670
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
428 else if (STRCMP(buf, "best") == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
429 f = SPS_BEST;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
430 else if (STRCMP(buf, "fast") == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
431 f = SPS_FAST;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
432 else if (STRCMP(buf, "double") == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
433 f = SPS_DOUBLE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
434 else if (STRNCMP(buf, "expr:", 5) != 0
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
435 && STRNCMP(buf, "file:", 5) != 0
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
436 && (STRNCMP(buf, "timeout:", 8) != 0
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
437 || (!VIM_ISDIGIT(buf[8])
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
438 && !(buf[8] == '-' && VIM_ISDIGIT(buf[9])))))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
439 f = -1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
440
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
441 if (f == -1 || (sps_flags != 0 && f != 0))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
442 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
443 sps_flags = SPS_BEST;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
444 sps_limit = 9999;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
445 return FAIL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
446 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
447 if (f != 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
448 sps_flags = f;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
449 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
450
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
451 if (sps_flags == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
452 sps_flags = SPS_BEST;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
453
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
454 return OK;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
455 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
456
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
457 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
458 * "z=": Find badly spelled word under or after the cursor.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
459 * Give suggestions for the properly spelled word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
460 * In Visual mode use the highlighted word as the bad word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
461 * When "count" is non-zero use that suggestion.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
462 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
463 void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
464 spell_suggest(int count)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
465 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
466 char_u *line;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
467 pos_T prev_cursor = curwin->w_cursor;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
468 char_u wcopy[MAXWLEN + 2];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
469 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
470 int i;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
471 int c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
472 suginfo_T sug;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
473 suggest_T *stp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
474 int mouse_used;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
475 int need_cap;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
476 int limit;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
477 int selected = count;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
478 int badlen = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
479 int msg_scroll_save = msg_scroll;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
480 int wo_spell_save = curwin->w_p_spell;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
481
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
482 if (!curwin->w_p_spell)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
483 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
484 parse_spelllang(curwin);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
485 curwin->w_p_spell = TRUE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
486 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
487
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
488 if (*curwin->w_s->b_p_spl == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
489 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
490 emsg(_(e_spell_checking_is_not_possible));
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
491 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
492 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
493
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
494 if (VIsual_active)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
495 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
496 // Use the Visually selected text as the bad word. But reject
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
497 // a multi-line selection.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
498 if (curwin->w_cursor.lnum != VIsual.lnum)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
499 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
500 vim_beep(BO_SPELL);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
501 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
502 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
503 badlen = (int)curwin->w_cursor.col - (int)VIsual.col;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
504 if (badlen < 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
505 badlen = -badlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
506 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
507 curwin->w_cursor.col = VIsual.col;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
508 ++badlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
509 end_visual_mode();
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
510 // make sure we don't include the NUL at the end of the line
34540
9e093c96dff6 patch 9.1.0172: More code can use ml_get_buf_len() instead of STRLEN()
Christian Brabandt <cb@256bit.org>
parents: 34309
diff changeset
511 if (badlen > ml_get_curline_len() - (int)curwin->w_cursor.col)
9e093c96dff6 patch 9.1.0172: More code can use ml_get_buf_len() instead of STRLEN()
Christian Brabandt <cb@256bit.org>
parents: 34309
diff changeset
512 badlen = ml_get_curline_len() - (int)curwin->w_cursor.col;
32670
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
513 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
514 // Find the start of the badly spelled word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
515 else if (spell_move_to(curwin, FORWARD, TRUE, TRUE, NULL) == 0
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
516 || curwin->w_cursor.col > prev_cursor.col)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
517 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
518 // No bad word or it starts after the cursor: use the word under the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
519 // cursor.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
520 curwin->w_cursor = prev_cursor;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
521 line = ml_get_curline();
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
522 p = line + curwin->w_cursor.col;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
523 // Backup to before start of word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
524 while (p > line && spell_iswordp_nmw(p, curwin))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
525 MB_PTR_BACK(line, p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
526 // Forward to start of word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
527 while (*p != NUL && !spell_iswordp_nmw(p, curwin))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
528 MB_PTR_ADV(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
529
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
530 if (!spell_iswordp_nmw(p, curwin)) // No word found.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
531 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
532 beep_flush();
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
533 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
534 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
535 curwin->w_cursor.col = (colnr_T)(p - line);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
536 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
537
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
538 // Get the word and its length.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
539
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
540 // Figure out if the word should be capitalised.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
541 need_cap = check_need_cap(curwin, curwin->w_cursor.lnum,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
542 curwin->w_cursor.col);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
543
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
544 // Make a copy of current line since autocommands may free the line.
34540
9e093c96dff6 patch 9.1.0172: More code can use ml_get_buf_len() instead of STRLEN()
Christian Brabandt <cb@256bit.org>
parents: 34309
diff changeset
545 line = vim_strnsave(ml_get_curline(), ml_get_curline_len());
32670
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
546 if (line == NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
547 goto skip;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
548
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
549 // Get the list of suggestions. Limit to 'lines' - 2 or the number in
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
550 // 'spellsuggest', whatever is smaller.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
551 if (sps_limit > (int)Rows - 2)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
552 limit = (int)Rows - 2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
553 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
554 limit = sps_limit;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
555 spell_find_suggest(line + curwin->w_cursor.col, badlen, &sug, limit,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
556 TRUE, need_cap, TRUE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
557
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
558 if (sug.su_ga.ga_len == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
559 msg(_("Sorry, no suggestions"));
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
560 else if (count > 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
561 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
562 if (count > sug.su_ga.ga_len)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
563 smsg(_("Sorry, only %ld suggestions"), (long)sug.su_ga.ga_len);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
564 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
565 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
566 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
567 #ifdef FEAT_RIGHTLEFT
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
568 // When 'rightleft' is set the list is drawn right-left.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
569 cmdmsg_rl = curwin->w_p_rl;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
570 if (cmdmsg_rl)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
571 msg_col = Columns - 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
572 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
573
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
574 // List the suggestions.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
575 msg_start();
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
576 msg_row = Rows - 1; // for when 'cmdheight' > 1
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
577 lines_left = Rows; // avoid more prompt
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
578 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
579 sug.su_badlen, sug.su_badptr);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
580 #ifdef FEAT_RIGHTLEFT
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
581 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
582 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
583 // And now the rabbit from the high hat: Avoid showing the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
584 // untranslated message rightleft.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
585 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
586 sug.su_badlen, sug.su_badptr);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
587 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
588 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
589 msg_puts((char *)IObuff);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
590 msg_clr_eos();
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
591 msg_putchar('\n');
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
592
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
593 msg_scroll = TRUE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
594 for (i = 0; i < sug.su_ga.ga_len; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
595 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
596 int el;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
597
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
598 stp = &SUG(sug.su_ga, i);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
599
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
600 // The suggested word may replace only part of the bad word, add
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
601 // the not replaced part. But only when it's not getting too long.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
602 vim_strncpy(wcopy, stp->st_word, MAXWLEN);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
603 el = sug.su_badlen - stp->st_orglen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
604 if (el > 0 && stp->st_wordlen + el <= MAXWLEN)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
605 vim_strncpy(wcopy + stp->st_wordlen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
606 sug.su_badptr + stp->st_orglen, el);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
607 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
608 #ifdef FEAT_RIGHTLEFT
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
609 if (cmdmsg_rl)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
610 rl_mirror(IObuff);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
611 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
612 msg_puts((char *)IObuff);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
613
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
614 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
615 msg_puts((char *)IObuff);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
616
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
617 // The word may replace more than "su_badlen".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
618 if (sug.su_badlen < stp->st_orglen)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
619 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
620 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
621 stp->st_orglen, sug.su_badptr);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
622 msg_puts((char *)IObuff);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
623 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
624
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
625 if (p_verbose > 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
626 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
627 // Add the score.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
628 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
629 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
630 stp->st_salscore ? "s " : "",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
631 stp->st_score, stp->st_altscore);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
632 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
633 vim_snprintf((char *)IObuff, IOSIZE, " (%d)",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
634 stp->st_score);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
635 #ifdef FEAT_RIGHTLEFT
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
636 if (cmdmsg_rl)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
637 // Mirror the numbers, but keep the leading space.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
638 rl_mirror(IObuff + 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
639 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
640 msg_advance(30);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
641 msg_puts((char *)IObuff);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
642 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
643 msg_putchar('\n');
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
644 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
645
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
646 #ifdef FEAT_RIGHTLEFT
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
647 cmdmsg_rl = FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
648 msg_col = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
649 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
650 // Ask for choice.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
651 selected = prompt_for_number(&mouse_used);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
652 if (mouse_used)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
653 selected -= lines_left;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
654 lines_left = Rows; // avoid more prompt
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
655 // don't delay for 'smd' in normal_cmd()
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
656 msg_scroll = msg_scroll_save;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
657 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
658
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
659 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
660 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
661 // Save the from and to text for :spellrepall.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
662 VIM_CLEAR(repl_from);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
663 VIM_CLEAR(repl_to);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
664
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
665 stp = &SUG(sug.su_ga, selected - 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
666 if (sug.su_badlen > stp->st_orglen)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
667 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
668 // Replacing less than "su_badlen", append the remainder to
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
669 // repl_to.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
670 repl_from = vim_strnsave(sug.su_badptr, sug.su_badlen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
671 vim_snprintf((char *)IObuff, IOSIZE, "%s%.*s", stp->st_word,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
672 sug.su_badlen - stp->st_orglen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
673 sug.su_badptr + stp->st_orglen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
674 repl_to = vim_strsave(IObuff);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
675 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
676 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
677 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
678 // Replacing su_badlen or more, use the whole word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
679 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
680 repl_to = vim_strsave(stp->st_word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
681 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
682
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
683 // Replace the word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
684 p = alloc(STRLEN(line) - stp->st_orglen + stp->st_wordlen + 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
685 if (p != NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
686 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
687 int len_diff = stp->st_wordlen - stp->st_orglen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
688
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
689 c = (int)(sug.su_badptr - line);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
690 mch_memmove(p, line, c);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
691 STRCPY(p + c, stp->st_word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
692 STRCAT(p, sug.su_badptr + stp->st_orglen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
693
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
694 // For redo we use a change-word command.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
695 ResetRedobuff();
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
696 AppendToRedobuff((char_u *)"ciw");
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
697 AppendToRedobuffLit(p + c,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
698 stp->st_wordlen + sug.su_badlen - stp->st_orglen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
699 AppendCharToRedobuff(ESC);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
700
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
701 // "p" may be freed here
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
702 ml_replace(curwin->w_cursor.lnum, p, FALSE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
703 curwin->w_cursor.col = c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
704
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
705 changed_bytes(curwin->w_cursor.lnum, c);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
706 #if defined(FEAT_PROP_POPUP)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
707 if (curbuf->b_has_textprop && len_diff != 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
708 adjust_prop_columns(curwin->w_cursor.lnum, c, len_diff,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
709 APC_SUBSTITUTE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
710 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
711 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
712 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
713 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
714 curwin->w_cursor = prev_cursor;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
715
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
716 spell_find_cleanup(&sug);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
717 skip:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
718 vim_free(line);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
719 curwin->w_p_spell = wo_spell_save;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
720 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
721
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
722 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
723 * Find spell suggestions for "word". Return them in the growarray "*gap" as
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
724 * a list of allocated strings.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
725 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
726 void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
727 spell_suggest_list(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
728 garray_T *gap,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
729 char_u *word,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
730 int maxcount, // maximum nr of suggestions
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
731 int need_cap, // 'spellcapcheck' matched
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
732 int interactive)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
733 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
734 suginfo_T sug;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
735 int i;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
736 suggest_T *stp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
737 char_u *wcopy;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
738
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
739 spell_find_suggest(word, 0, &sug, maxcount, FALSE, need_cap, interactive);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
740
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
741 // Make room in "gap".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
742 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
743 if (ga_grow(gap, sug.su_ga.ga_len) == OK)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
744 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
745 for (i = 0; i < sug.su_ga.ga_len; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
746 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
747 stp = &SUG(sug.su_ga, i);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
748
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
749 // The suggested word may replace only part of "word", add the not
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
750 // replaced part.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
751 wcopy = alloc(stp->st_wordlen
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
752 + (unsigned)STRLEN(sug.su_badptr + stp->st_orglen) + 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
753 if (wcopy == NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
754 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
755 STRCPY(wcopy, stp->st_word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
756 STRCPY(wcopy + stp->st_wordlen, sug.su_badptr + stp->st_orglen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
757 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
758 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
759 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
760
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
761 spell_find_cleanup(&sug);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
762 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
763
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
764 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
765 * Find spell suggestions for the word at the start of "badptr".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
766 * Return the suggestions in "su->su_ga".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
767 * The maximum number of suggestions is "maxcount".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
768 * Note: does use info for the current window.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
769 * This is based on the mechanisms of Aspell, but completely reimplemented.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
770 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
771 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
772 spell_find_suggest(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
773 char_u *badptr,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
774 int badlen, // length of bad word or 0 if unknown
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
775 suginfo_T *su,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
776 int maxcount,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
777 int banbadword, // don't include badword in suggestions
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
778 int need_cap, // word should start with capital
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
779 int interactive)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
780 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
781 hlf_T attr = HLF_COUNT;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
782 char_u buf[MAXPATHL];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
783 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
784 int do_combine = FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
785 char_u *sps_copy;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
786 #ifdef FEAT_EVAL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
787 static int expr_busy = FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
788 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
789 int c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
790 int i;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
791 langp_T *lp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
792 int did_intern = FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
793
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
794 // Set the info in "*su".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
795 CLEAR_POINTER(su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
796 ga_init2(&su->su_ga, sizeof(suggest_T), 10);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
797 ga_init2(&su->su_sga, sizeof(suggest_T), 10);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
798 if (*badptr == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
799 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
800 hash_init(&su->su_banned);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
801
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
802 su->su_badptr = badptr;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
803 if (badlen != 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
804 su->su_badlen = badlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
805 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
806 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL, FALSE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
807 su->su_maxcount = maxcount;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
808 su->su_maxscore = SCORE_MAXINIT;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
809
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
810 if (su->su_badlen >= MAXWLEN)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
811 su->su_badlen = MAXWLEN - 1; // just in case
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
812 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
813 (void)spell_casefold(curwin, su->su_badptr, su->su_badlen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
814 su->su_fbadword, MAXWLEN);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
815 // TODO: make this work if the case-folded text is longer than the original
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
816 // text. Currently an illegal byte causes wrong pointer computations.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
817 su->su_fbadword[su->su_badlen] = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
818
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
819 // get caps flags for bad word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
820 su->su_badflags = badword_captype(su->su_badptr,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
821 su->su_badptr + su->su_badlen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
822 if (need_cap)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
823 su->su_badflags |= WF_ONECAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
824
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
825 // Find the default language for sound folding. We simply use the first
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
826 // one in 'spelllang' that supports sound folding. That's good for when
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
827 // using multiple files for one language, it's not that bad when mixing
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
828 // languages (e.g., "pl,en").
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
829 for (i = 0; i < curbuf->b_s.b_langp.ga_len; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
830 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
831 lp = LANGP_ENTRY(curbuf->b_s.b_langp, i);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
832 if (lp->lp_sallang != NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
833 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
834 su->su_sallang = lp->lp_sallang;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
835 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
836 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
837 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
838
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
839 // Soundfold the bad word with the default sound folding, so that we don't
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
840 // have to do this many times.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
841 if (su->su_sallang != NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
842 spell_soundfold(su->su_sallang, su->su_fbadword, TRUE,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
843 su->su_sal_badword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
844
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
845 // If the word is not capitalised and spell_check() doesn't consider the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
846 // word to be bad then it might need to be capitalised. Add a suggestion
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
847 // for that.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
848 c = PTR2CHAR(su->su_badptr);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
849 if (!SPELL_ISUPPER(c) && attr == HLF_COUNT)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
850 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
851 make_case_word(su->su_badword, buf, WF_ONECAP);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
852 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
853 0, TRUE, su->su_sallang, FALSE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
854 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
855
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
856 // Ban the bad word itself. It may appear in another region.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
857 if (banbadword)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
858 add_banned(su, su->su_badword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
859
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
860 // Make a copy of 'spellsuggest', because the expression may change it.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
861 sps_copy = vim_strsave(p_sps);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
862 if (sps_copy == NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
863 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
864 spell_suggest_timeout = 5000;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
865
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
866 // Loop over the items in 'spellsuggest'.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
867 for (p = sps_copy; *p != NUL; )
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
868 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
869 copy_option_part(&p, buf, MAXPATHL, ",");
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
870
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
871 if (STRNCMP(buf, "expr:", 5) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
872 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
873 #ifdef FEAT_EVAL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
874 // Evaluate an expression. Skip this when called recursively,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
875 // when using spellsuggest() in the expression.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
876 if (!expr_busy)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
877 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
878 expr_busy = TRUE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
879 spell_suggest_expr(su, buf + 5);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
880 expr_busy = FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
881 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
882 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
883 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
884 else if (STRNCMP(buf, "file:", 5) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
885 // Use list of suggestions in a file.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
886 spell_suggest_file(su, buf + 5);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
887 else if (STRNCMP(buf, "timeout:", 8) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
888 // Limit the time searching for suggestions.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
889 spell_suggest_timeout = atol((char *)buf + 8);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
890 else if (!did_intern)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
891 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
892 // Use internal method once.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
893 spell_suggest_intern(su, interactive);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
894 if (sps_flags & SPS_DOUBLE)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
895 do_combine = TRUE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
896 did_intern = TRUE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
897 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
898 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
899
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
900 vim_free(sps_copy);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
901
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
902 if (do_combine)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
903 // Combine the two list of suggestions. This must be done last,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
904 // because sorting changes the order again.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
905 score_combine(su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
906 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
907
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
908 #ifdef FEAT_EVAL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
909 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
910 * Find suggestions by evaluating expression "expr".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
911 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
912 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
913 spell_suggest_expr(suginfo_T *su, char_u *expr)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
914 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
915 list_T *list;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
916 listitem_T *li;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
917 int score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
918 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
919
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
920 // The work is split up in a few parts to avoid having to export
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
921 // suginfo_T.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
922 // First evaluate the expression and get the resulting list.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
923 list = eval_spell_expr(su->su_badword, expr);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
924 if (list != NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
925 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
926 // Loop over the items in the list.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
927 FOR_ALL_LIST_ITEMS(list, li)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
928 if (li->li_tv.v_type == VAR_LIST)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
929 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
930 // Get the word and the score from the items.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
931 score = get_spellword(li->li_tv.vval.v_list, &p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
932 if (score >= 0 && score <= su->su_maxscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
933 add_suggestion(su, &su->su_ga, p, su->su_badlen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
934 score, 0, TRUE, su->su_sallang, FALSE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
935 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
936 list_unref(list);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
937 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
938
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
939 // Remove bogus suggestions, sort and truncate at "maxcount".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
940 check_suggestions(su, &su->su_ga);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
941 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
942 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
943 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
944
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
945 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
946 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
947 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
948 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
949 spell_suggest_file(suginfo_T *su, char_u *fname)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
950 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
951 FILE *fd;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
952 char_u line[MAXWLEN * 2];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
953 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
954 int len;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
955 char_u cword[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
956
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
957 // Open the file.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
958 fd = mch_fopen((char *)fname, "r");
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
959 if (fd == NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
960 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
961 semsg(_(e_cant_open_file_str), fname);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
962 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
963 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
964
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
965 // Read it line by line.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
966 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
967 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
968 line_breakcheck();
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
969
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
970 p = vim_strchr(line, '/');
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
971 if (p == NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
972 continue; // No Tab found, just skip the line.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
973 *p++ = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
974 if (STRICMP(su->su_badword, line) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
975 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
976 // Match! Isolate the good word, until CR or NL.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
977 for (len = 0; p[len] >= ' '; ++len)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
978 ;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
979 p[len] = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
980
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
981 // If the suggestion doesn't have specific case duplicate the case
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
982 // of the bad word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
983 if (captype(p, NULL) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
984 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
985 make_case_word(p, cword, su->su_badflags);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
986 p = cword;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
987 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
988
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
989 add_suggestion(su, &su->su_ga, p, su->su_badlen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
990 SCORE_FILE, 0, TRUE, su->su_sallang, FALSE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
991 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
992 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
993
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
994 fclose(fd);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
995
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
996 // Remove bogus suggestions, sort and truncate at "maxcount".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
997 check_suggestions(su, &su->su_ga);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
998 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
999 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1000
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1001 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1002 * Find suggestions for the internal method indicated by "sps_flags".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1003 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1004 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1005 spell_suggest_intern(suginfo_T *su, int interactive)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1006 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1007 // Load the .sug file(s) that are available and not done yet.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1008 suggest_load_files();
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1009
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1010 // 1. Try special cases, such as repeating a word: "the the" -> "the".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1011 //
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1012 // Set a maximum score to limit the combination of operations that is
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1013 // tried.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1014 suggest_try_special(su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1015
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1016 // 2. Try inserting/deleting/swapping/changing a letter, use REP entries
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1017 // from the .aff file and inserting a space (split the word).
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1018 suggest_try_change(su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1019
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1020 // For the resulting top-scorers compute the sound-a-like score.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1021 if (sps_flags & SPS_DOUBLE)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1022 score_comp_sal(su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1023
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1024 // 3. Try finding sound-a-like words.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1025 if ((sps_flags & SPS_FAST) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1026 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1027 if (sps_flags & SPS_BEST)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1028 // Adjust the word score for the suggestions found so far for how
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1029 // they sound like.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1030 rescore_suggestions(su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1031
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1032 // While going through the soundfold tree "su_maxscore" is the score
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1033 // for the soundfold word, limits the changes that are being tried,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1034 // and "su_sfmaxscore" the rescored score, which is set by
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1035 // cleanup_suggestions().
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1036 // First find words with a small edit distance, because this is much
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1037 // faster and often already finds the top-N suggestions. If we didn't
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1038 // find many suggestions try again with a higher edit distance.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1039 // "sl_sounddone" is used to avoid doing the same word twice.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1040 suggest_try_soundalike_prep();
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1041 su->su_maxscore = SCORE_SFMAX1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1042 su->su_sfmaxscore = SCORE_MAXINIT * 3;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1043 suggest_try_soundalike(su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1044 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1045 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1046 // We didn't find enough matches, try again, allowing more
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1047 // changes to the soundfold word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1048 su->su_maxscore = SCORE_SFMAX2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1049 suggest_try_soundalike(su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1050 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1051 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1052 // Still didn't find enough matches, try again, allowing even
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1053 // more changes to the soundfold word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1054 su->su_maxscore = SCORE_SFMAX3;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1055 suggest_try_soundalike(su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1056 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1057 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1058 su->su_maxscore = su->su_sfmaxscore;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1059 suggest_try_soundalike_finish();
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1060 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1061
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1062 // When CTRL-C was hit while searching do show the results. Only clear
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1063 // got_int when using a command, not for spellsuggest().
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1064 ui_breakcheck();
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1065 if (interactive && got_int)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1066 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1067 (void)vgetc();
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1068 got_int = FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1069 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1070
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1071 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1072 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1073 if (sps_flags & SPS_BEST)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1074 // Adjust the word score for how it sounds like.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1075 rescore_suggestions(su);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1076
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1077 // Remove bogus suggestions, sort and truncate at "maxcount".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1078 check_suggestions(su, &su->su_ga);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1079 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1080 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1081 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1082
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1083 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1084 * Free the info put in "*su" by spell_find_suggest().
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1085 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1086 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1087 spell_find_cleanup(suginfo_T *su)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1088 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1089 int i;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1090
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1091 // Free the suggestions.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1092 for (i = 0; i < su->su_ga.ga_len; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1093 vim_free(SUG(su->su_ga, i).st_word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1094 ga_clear(&su->su_ga);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1095 for (i = 0; i < su->su_sga.ga_len; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1096 vim_free(SUG(su->su_sga, i).st_word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1097 ga_clear(&su->su_sga);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1098
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1099 // Free the banned words.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1100 hash_clear_all(&su->su_banned, 0);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1101 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1102
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1103 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1104 * Try finding suggestions by recognizing specific situations.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1105 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1106 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1107 suggest_try_special(suginfo_T *su)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1108 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1109 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1110 size_t len;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1111 int c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1112 char_u word[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1113
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1114 // Recognize a word that is repeated: "the the".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1115 p = skiptowhite(su->su_fbadword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1116 len = p - su->su_fbadword;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1117 p = skipwhite(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1118 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1119 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1120 // Include badflags: if the badword is onecap or allcap
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1121 // use that for the goodword too: "The the" -> "The".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1122 c = su->su_fbadword[len];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1123 su->su_fbadword[len] = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1124 make_case_word(su->su_fbadword, word, su->su_badflags);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1125 su->su_fbadword[len] = c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1126
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1127 // Give a soundalike score of 0, compute the score as if deleting one
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1128 // character.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1129 add_suggestion(su, &su->su_ga, word, su->su_badlen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1130 RESCORE(SCORE_REP, 0), 0, TRUE, su->su_sallang, FALSE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1131 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1132 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1133
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1134 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1135 * Change the 0 to 1 to measure how much time is spent in each state.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1136 * Output is dumped in "suggestprof".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1137 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1138 #if 0
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1139 # define SUGGEST_PROFILE
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1140 proftime_T current;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1141 proftime_T total;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1142 proftime_T times[STATE_FINAL + 1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1143 long counts[STATE_FINAL + 1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1144
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1145 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1146 prof_init(void)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1147 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1148 for (int i = 0; i <= STATE_FINAL; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1149 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1150 profile_zero(&times[i]);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1151 counts[i] = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1152 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1153 profile_start(&current);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1154 profile_start(&total);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1155 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1156
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1157 // call before changing state
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1158 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1159 prof_store(state_T state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1160 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1161 profile_end(&current);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1162 profile_add(&times[state], &current);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1163 ++counts[state];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1164 profile_start(&current);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1165 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1166 # define PROF_STORE(state) prof_store(state);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1167
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1168 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1169 prof_report(char *name)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1170 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1171 FILE *fd = fopen("suggestprof", "a");
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1172
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1173 profile_end(&total);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1174 fprintf(fd, "-----------------------\n");
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1175 fprintf(fd, "%s: %s\n", name, profile_msg(&total));
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1176 for (int i = 0; i <= STATE_FINAL; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1177 fprintf(fd, "%d: %s (%ld)\n", i, profile_msg(&times[i]), counts[i]);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1178 fclose(fd);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1179 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1180 #else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1181 # define PROF_STORE(state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1182 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1183
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1184 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1185 * Try finding suggestions by adding/removing/swapping letters.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1186 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1187 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1188 suggest_try_change(suginfo_T *su)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1189 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1190 char_u fword[MAXWLEN]; // copy of the bad word, case-folded
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1191 int n;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1192 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1193 int lpi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1194 langp_T *lp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1195
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1196 // We make a copy of the case-folded bad word, so that we can modify it
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1197 // to find matches (esp. REP items). Append some more text, changing
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1198 // chars after the bad word may help.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1199 STRCPY(fword, su->su_fbadword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1200 n = (int)STRLEN(fword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1201 p = su->su_badptr + su->su_badlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1202 (void)spell_casefold(curwin, p, (int)STRLEN(p), fword + n, MAXWLEN - n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1203
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1204 // Make sure the resulting text is not longer than the original text.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1205 n = (int)STRLEN(su->su_badptr);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1206 if (n < MAXWLEN)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1207 fword[n] = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1208
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1209 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1210 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1211 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1212
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1213 // If reloading a spell file fails it's still in the list but
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1214 // everything has been cleared.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1215 if (lp->lp_slang->sl_fbyts == NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1216 continue;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1217
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1218 // Try it for this language. Will add possible suggestions.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1219 #ifdef SUGGEST_PROFILE
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1220 prof_init();
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1221 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1222 suggest_trie_walk(su, lp, fword, FALSE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1223 #ifdef SUGGEST_PROFILE
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1224 prof_report("try_change");
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1225 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1226 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1227 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1228
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1229 // Check the maximum score, if we go over it we won't try this change.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1230 #define TRY_DEEPER(su, stack, depth, add) \
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1231 ((depth) < MAXWLEN - 1 && (stack)[depth].ts_score + (add) < (su)->su_maxscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1232
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1233 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1234 * Try finding suggestions by adding/removing/swapping letters.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1235 *
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1236 * This uses a state machine. At each node in the tree we try various
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1237 * operations. When trying if an operation works "depth" is increased and the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1238 * stack[] is used to store info. This allows combinations, thus insert one
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1239 * character, replace one and delete another. The number of changes is
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1240 * limited by su->su_maxscore.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1241 *
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1242 * After implementing this I noticed an article by Kemal Oflazer that
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1243 * describes something similar: "Error-tolerant Finite State Recognition with
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1244 * Applications to Morphological Analysis and Spelling Correction" (1996).
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1245 * The implementation in the article is simplified and requires a stack of
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1246 * unknown depth. The implementation here only needs a stack depth equal to
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1247 * the length of the word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1248 *
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1249 * This is also used for the sound-folded word, "soundfold" is TRUE then.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1250 * The mechanism is the same, but we find a match with a sound-folded word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1251 * that comes from one or more original words. Each of these words may be
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1252 * added, this is done by add_sound_suggest().
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1253 * Don't use:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1254 * the prefix tree or the keep-case tree
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1255 * "su->su_badlen"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1256 * anything to do with upper and lower case
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1257 * anything to do with word or non-word characters ("spell_iswordp()")
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1258 * banned words
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1259 * word flags (rare, region, compounding)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1260 * word splitting for now
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1261 * "similar_chars()"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1262 * use "slang->sl_repsal" instead of "lp->lp_replang->sl_rep"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1263 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1264 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1265 suggest_trie_walk(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1266 suginfo_T *su,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1267 langp_T *lp,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1268 char_u *fword,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1269 int soundfold)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1270 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1271 char_u tword[MAXWLEN]; // good word collected so far
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1272 trystate_T stack[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1273 char_u preword[MAXWLEN * 3]; // word found with proper case;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1274 // concatenation of prefix compound
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1275 // words and split word. NUL terminated
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1276 // when going deeper but not when coming
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1277 // back.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1278 char_u compflags[MAXWLEN]; // compound flags, one for each word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1279 trystate_T *sp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1280 int newscore;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1281 int score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1282 char_u *byts, *fbyts, *pbyts;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1283 idx_T *idxs, *fidxs, *pidxs;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1284 int depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1285 int c, c2, c3;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1286 int n = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1287 int flags;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1288 garray_T *gap;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1289 idx_T arridx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1290 int len;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1291 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1292 fromto_T *ftp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1293 int fl = 0, tl;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1294 int repextra = 0; // extra bytes in fword[] from REP item
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1295 slang_T *slang = lp->lp_slang;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1296 int fword_ends;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1297 int goodword_ends;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1298 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1299 // Stores the name of the change made at each level.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1300 char_u changename[MAXWLEN][80];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1301 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1302 int breakcheckcount = 1000;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1303 #ifdef FEAT_RELTIME
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1304 proftime_T time_limit;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1305 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1306 int compound_ok;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1307
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1308 // Go through the whole case-fold tree, try changes at each node.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1309 // "tword[]" contains the word collected from nodes in the tree.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1310 // "fword[]" the word we are trying to match with (initially the bad
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1311 // word).
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1312 depth = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1313 sp = &stack[0];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1314 CLEAR_POINTER(sp);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1315 sp->ts_curi = 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1316
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1317 if (soundfold)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1318 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1319 // Going through the soundfold tree.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1320 byts = fbyts = slang->sl_sbyts;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1321 idxs = fidxs = slang->sl_sidxs;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1322 pbyts = NULL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1323 pidxs = NULL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1324 sp->ts_prefixdepth = PFD_NOPREFIX;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1325 sp->ts_state = STATE_START;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1326 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1327 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1328 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1329 // When there are postponed prefixes we need to use these first. At
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1330 // the end of the prefix we continue in the case-fold tree.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1331 fbyts = slang->sl_fbyts;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1332 fidxs = slang->sl_fidxs;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1333 pbyts = slang->sl_pbyts;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1334 pidxs = slang->sl_pidxs;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1335 if (pbyts != NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1336 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1337 byts = pbyts;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1338 idxs = pidxs;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1339 sp->ts_prefixdepth = PFD_PREFIXTREE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1340 sp->ts_state = STATE_NOPREFIX; // try without prefix first
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1341 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1342 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1343 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1344 byts = fbyts;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1345 idxs = fidxs;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1346 sp->ts_prefixdepth = PFD_NOPREFIX;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1347 sp->ts_state = STATE_START;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1348 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1349 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1350 #ifdef FEAT_RELTIME
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1351 // The loop may take an indefinite amount of time. Break out after some
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1352 // time.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1353 if (spell_suggest_timeout > 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1354 profile_setlimit(spell_suggest_timeout, &time_limit);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1355 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1356
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1357 // Loop to find all suggestions. At each round we either:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1358 // - For the current state try one operation, advance "ts_curi",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1359 // increase "depth".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1360 // - When a state is done go to the next, set "ts_state".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1361 // - When all states are tried decrease "depth".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1362 while (depth >= 0 && !got_int)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1363 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1364 sp = &stack[depth];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1365 switch (sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1366 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1367 case STATE_START:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1368 case STATE_NOPREFIX:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1369 // Start of node: Deal with NUL bytes, which means
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1370 // tword[] may end here.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1371 arridx = sp->ts_arridx; // current node in the tree
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1372 len = byts[arridx]; // bytes in this node
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1373 arridx += sp->ts_curi; // index of current byte
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1374
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1375 if (sp->ts_prefixdepth == PFD_PREFIXTREE)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1376 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1377 // Skip over the NUL bytes, we use them later.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1378 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1379 ;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1380 sp->ts_curi += n;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1381
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1382 // Always past NUL bytes now.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1383 n = (int)sp->ts_state;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1384 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1385 sp->ts_state = STATE_ENDNUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1386 sp->ts_save_badflags = su->su_badflags;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1387
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1388 // At end of a prefix or at start of prefixtree: check for
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1389 // following word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1390 if (depth < MAXWLEN - 1
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1391 && (byts[arridx] == 0 || n == (int)STATE_NOPREFIX))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1392 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1393 // Set su->su_badflags to the caps type at this position.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1394 // Use the caps type until here for the prefix itself.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1395 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1396 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1397 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1398 n = sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1399 flags = badword_captype(su->su_badptr, su->su_badptr + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1400 su->su_badflags = badword_captype(su->su_badptr + n,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1401 su->su_badptr + su->su_badlen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1402 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1403 sprintf(changename[depth], "prefix");
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1404 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1405 go_deeper(stack, depth, 0);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1406 ++depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1407 sp = &stack[depth];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1408 sp->ts_prefixdepth = depth - 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1409 byts = fbyts;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1410 idxs = fidxs;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1411 sp->ts_arridx = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1412
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1413 // Move the prefix to preword[] with the right case
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1414 // and make find_keepcap_word() works.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1415 tword[sp->ts_twordlen] = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1416 make_case_word(tword + sp->ts_splitoff,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1417 preword + sp->ts_prewordlen, flags);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1418 sp->ts_prewordlen = (char_u)STRLEN(preword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1419 sp->ts_splitoff = sp->ts_twordlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1420 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1421 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1422 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1423
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1424 if (sp->ts_curi > len || byts[arridx] != 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1425 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1426 // Past bytes in node and/or past NUL bytes.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1427 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1428 sp->ts_state = STATE_ENDNUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1429 sp->ts_save_badflags = su->su_badflags;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1430 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1431 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1432
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1433 // End of word in tree.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1434 ++sp->ts_curi; // eat one NUL byte
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1435
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1436 flags = (int)idxs[arridx];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1437
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1438 // Skip words with the NOSUGGEST flag.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1439 if (flags & WF_NOSUGGEST)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1440 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1441
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1442 fword_ends = (fword[sp->ts_fidx] == NUL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1443 || (soundfold
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1444 ? VIM_ISWHITE(fword[sp->ts_fidx])
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1445 : !spell_iswordp(fword + sp->ts_fidx, curwin)));
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1446 tword[sp->ts_twordlen] = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1447
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1448 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1449 && (sp->ts_flags & TSF_PREFIXOK) == 0
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1450 && pbyts != NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1451 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1452 // There was a prefix before the word. Check that the prefix
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1453 // can be used with this word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1454 // Count the length of the NULs in the prefix. If there are
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1455 // none this must be the first try without a prefix.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1456 n = stack[sp->ts_prefixdepth].ts_arridx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1457 len = pbyts[n++];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1458 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1459 ;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1460 if (c > 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1461 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1462 c = valid_word_prefix(c, n, flags,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1463 tword + sp->ts_splitoff, slang, FALSE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1464 if (c == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1465 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1466
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1467 // Use the WF_RARE flag for a rare prefix.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1468 if (c & WF_RAREPFX)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1469 flags |= WF_RARE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1470
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1471 // Tricky: when checking for both prefix and compounding
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1472 // we run into the prefix flag first.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1473 // Remember that it's OK, so that we accept the prefix
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1474 // when arriving at a compound flag.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1475 sp->ts_flags |= TSF_PREFIXOK;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1476 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1477 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1478
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1479 // Check NEEDCOMPOUND: can't use word without compounding. Do try
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1480 // appending another compound word below.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1481 if (sp->ts_complen == sp->ts_compsplit && fword_ends
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1482 && (flags & WF_NEEDCOMP))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1483 goodword_ends = FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1484 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1485 goodword_ends = TRUE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1486
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1487 p = NULL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1488 compound_ok = TRUE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1489 if (sp->ts_complen > sp->ts_compsplit)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1490 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1491 if (slang->sl_nobreak)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1492 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1493 // There was a word before this word. When there was no
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1494 // change in this word (it was correct) add the first word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1495 // as a suggestion. If this word was corrected too, we
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1496 // need to check if a correct word follows.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1497 if (sp->ts_fidx - sp->ts_splitfidx
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1498 == sp->ts_twordlen - sp->ts_splitoff
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1499 && STRNCMP(fword + sp->ts_splitfidx,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1500 tword + sp->ts_splitoff,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1501 sp->ts_fidx - sp->ts_splitfidx) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1502 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1503 preword[sp->ts_prewordlen] = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1504 newscore = score_wordcount_adj(slang, sp->ts_score,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1505 preword + sp->ts_prewordlen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1506 sp->ts_prewordlen > 0);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1507 // Add the suggestion if the score isn't too bad.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1508 if (newscore <= su->su_maxscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1509 add_suggestion(su, &su->su_ga, preword,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1510 sp->ts_splitfidx - repextra,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1511 newscore, 0, FALSE,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1512 lp->lp_sallang, FALSE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1513 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1514 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1515 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1516 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1517 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1518 // There was a compound word before this word. If this
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1519 // word does not support compounding then give up
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1520 // (splitting is tried for the word without compound
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1521 // flag).
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1522 if (((unsigned)flags >> 24) == 0
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1523 || sp->ts_twordlen - sp->ts_splitoff
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1524 < slang->sl_compminlen)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1525 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1526 // For multi-byte chars check character length against
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1527 // COMPOUNDMIN.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1528 if (has_mbyte
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1529 && slang->sl_compminlen > 0
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1530 && mb_charlen(tword + sp->ts_splitoff)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1531 < slang->sl_compminlen)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1532 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1533
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1534 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1535 compflags[sp->ts_complen + 1] = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1536 vim_strncpy(preword + sp->ts_prewordlen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1537 tword + sp->ts_splitoff,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1538 sp->ts_twordlen - sp->ts_splitoff);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1539
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1540 // Verify CHECKCOMPOUNDPATTERN rules.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1541 if (match_checkcompoundpattern(preword, sp->ts_prewordlen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1542 &slang->sl_comppat))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1543 compound_ok = FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1544
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1545 if (compound_ok)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1546 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1547 p = preword;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1548 while (*skiptowhite(p) != NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1549 p = skipwhite(skiptowhite(p));
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1550 if (fword_ends && !can_compound(slang, p,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1551 compflags + sp->ts_compsplit))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1552 // Compound is not allowed. But it may still be
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1553 // possible if we add another (short) word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1554 compound_ok = FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1555 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1556
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1557 // Get pointer to last char of previous word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1558 p = preword + sp->ts_prewordlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1559 MB_PTR_BACK(preword, p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1560 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1561 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1562
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1563 // Form the word with proper case in preword.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1564 // If there is a word from a previous split, append.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1565 // For the soundfold tree don't change the case, simply append.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1566 if (soundfold)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1567 STRCPY(preword + sp->ts_prewordlen, tword + sp->ts_splitoff);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1568 else if (flags & WF_KEEPCAP)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1569 // Must find the word in the keep-case tree.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1570 find_keepcap_word(slang, tword + sp->ts_splitoff,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1571 preword + sp->ts_prewordlen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1572 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1573 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1574 // Include badflags: If the badword is onecap or allcap
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1575 // use that for the goodword too. But if the badword is
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1576 // allcap and it's only one char long use onecap.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1577 c = su->su_badflags;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1578 if ((c & WF_ALLCAP)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1579 && su->su_badlen == (*mb_ptr2len)(su->su_badptr))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1580 c = WF_ONECAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1581 c |= flags;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1582
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1583 // When appending a compound word after a word character don't
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1584 // use Onecap.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1585 if (p != NULL && spell_iswordp_nmw(p, curwin))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1586 c &= ~WF_ONECAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1587 make_case_word(tword + sp->ts_splitoff,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1588 preword + sp->ts_prewordlen, c);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1589 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1590
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1591 if (!soundfold)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1592 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1593 // Don't use a banned word. It may appear again as a good
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1594 // word, thus remember it.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1595 if (flags & WF_BANNED)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1596 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1597 add_banned(su, preword + sp->ts_prewordlen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1598 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1599 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1600 if ((sp->ts_complen == sp->ts_compsplit
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1601 && WAS_BANNED(su, preword + sp->ts_prewordlen))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1602 || WAS_BANNED(su, preword))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1603 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1604 if (slang->sl_compprog == NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1605 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1606 // the word so far was banned but we may try compounding
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1607 goodword_ends = FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1608 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1609 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1610
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1611 newscore = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1612 if (!soundfold) // soundfold words don't have flags
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1613 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1614 if ((flags & WF_REGION)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1615 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1616 newscore += SCORE_REGION;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1617 if (flags & WF_RARE)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1618 newscore += SCORE_RARE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1619
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1620 if (!spell_valid_case(su->su_badflags,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1621 captype(preword + sp->ts_prewordlen, NULL)))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1622 newscore += SCORE_ICASE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1623 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1624
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1625 // TODO: how about splitting in the soundfold tree?
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1626 if (fword_ends
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1627 && goodword_ends
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1628 && sp->ts_fidx >= sp->ts_fidxtry
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1629 && compound_ok)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1630 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1631 // The badword also ends: add suggestions.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1632 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1633 if (soundfold && STRCMP(preword, "smwrd") == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1634 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1635 int j;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1636
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1637 // print the stack of changes that brought us here
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1638 smsg("------ %s -------", fword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1639 for (j = 0; j < depth; ++j)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1640 smsg("%s", changename[j]);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1641 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1642 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1643 if (soundfold)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1644 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1645 // For soundfolded words we need to find the original
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1646 // words, the edit distance and then add them.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1647 add_sound_suggest(su, preword, sp->ts_score, lp);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1648 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1649 else if (sp->ts_fidx > 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1650 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1651 // Give a penalty when changing non-word char to word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1652 // char, e.g., "thes," -> "these".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1653 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1654 MB_PTR_BACK(fword, p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1655 if (!spell_iswordp(p, curwin) && *preword != NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1656 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1657 p = preword + STRLEN(preword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1658 MB_PTR_BACK(preword, p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1659 if (spell_iswordp(p, curwin))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1660 newscore += SCORE_NONWORD;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1661 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1662
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1663 // Give a bonus to words seen before.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1664 score = score_wordcount_adj(slang,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1665 sp->ts_score + newscore,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1666 preword + sp->ts_prewordlen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1667 sp->ts_prewordlen > 0);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1668
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1669 // Add the suggestion if the score isn't too bad.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1670 if (score <= su->su_maxscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1671 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1672 add_suggestion(su, &su->su_ga, preword,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1673 sp->ts_fidx - repextra,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1674 score, 0, FALSE, lp->lp_sallang, FALSE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1675
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1676 if (su->su_badflags & WF_MIXCAP)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1677 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1678 // We really don't know if the word should be
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1679 // upper or lower case, add both.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1680 c = captype(preword, NULL);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1681 if (c == 0 || c == WF_ALLCAP)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1682 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1683 make_case_word(tword + sp->ts_splitoff,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1684 preword + sp->ts_prewordlen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1685 c == 0 ? WF_ALLCAP : 0);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1686
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1687 add_suggestion(su, &su->su_ga, preword,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1688 sp->ts_fidx - repextra,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1689 score + SCORE_ICASE, 0, FALSE,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1690 lp->lp_sallang, FALSE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1691 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1692 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1693 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1694 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1695 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1696
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1697 // Try word split and/or compounding.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1698 if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1699 // Don't split halfway a character.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1700 && (!has_mbyte || sp->ts_tcharlen == 0))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1701 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1702 int try_compound;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1703 int try_split;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1704
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1705 // If past the end of the bad word don't try a split.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1706 // Otherwise try changing the next word. E.g., find
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1707 // suggestions for "the the" where the second "the" is
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1708 // different. It's done like a split.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1709 // TODO: word split for soundfold words
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1710 try_split = (sp->ts_fidx - repextra < su->su_badlen)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1711 && !soundfold;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1712
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1713 // Get here in several situations:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1714 // 1. The word in the tree ends:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1715 // If the word allows compounding try that. Otherwise try
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1716 // a split by inserting a space. For both check that a
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1717 // valid words starts at fword[sp->ts_fidx].
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1718 // For NOBREAK do like compounding to be able to check if
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1719 // the next word is valid.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1720 // 2. The badword does end, but it was due to a change (e.g.,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1721 // a swap). No need to split, but do check that the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1722 // following word is valid.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1723 // 3. The badword and the word in the tree end. It may still
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1724 // be possible to compound another (short) word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1725 try_compound = FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1726 if (!soundfold
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1727 && !slang->sl_nocompoundsugs
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1728 && slang->sl_compprog != NULL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1729 && ((unsigned)flags >> 24) != 0
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1730 && sp->ts_twordlen - sp->ts_splitoff
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1731 >= slang->sl_compminlen
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1732 && (!has_mbyte
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1733 || slang->sl_compminlen == 0
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1734 || mb_charlen(tword + sp->ts_splitoff)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1735 >= slang->sl_compminlen)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1736 && (slang->sl_compsylmax < MAXWLEN
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1737 || sp->ts_complen + 1 - sp->ts_compsplit
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1738 < slang->sl_compmax)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1739 && (can_be_compound(sp, slang,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1740 compflags, ((unsigned)flags >> 24))))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1741
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1742 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1743 try_compound = TRUE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1744 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1745 compflags[sp->ts_complen + 1] = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1746 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1747
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1748 // For NOBREAK we never try splitting, it won't make any word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1749 // valid.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1750 if (slang->sl_nobreak && !slang->sl_nocompoundsugs)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1751 try_compound = TRUE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1752
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1753 // If we could add a compound word, and it's also possible to
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1754 // split at this point, do the split first and set
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1755 // TSF_DIDSPLIT to avoid doing it again.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1756 else if (!fword_ends
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1757 && try_compound
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1758 && (sp->ts_flags & TSF_DIDSPLIT) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1759 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1760 try_compound = FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1761 sp->ts_flags |= TSF_DIDSPLIT;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1762 --sp->ts_curi; // do the same NUL again
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1763 compflags[sp->ts_complen] = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1764 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1765 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1766 sp->ts_flags &= ~TSF_DIDSPLIT;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1767
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1768 if (try_split || try_compound)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1769 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1770 if (!try_compound && (!fword_ends || !goodword_ends))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1771 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1772 // If we're going to split need to check that the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1773 // words so far are valid for compounding. If there
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1774 // is only one word it must not have the NEEDCOMPOUND
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1775 // flag.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1776 if (sp->ts_complen == sp->ts_compsplit
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1777 && (flags & WF_NEEDCOMP))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1778 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1779 p = preword;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1780 while (*skiptowhite(p) != NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1781 p = skipwhite(skiptowhite(p));
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1782 if (sp->ts_complen > sp->ts_compsplit
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1783 && !can_compound(slang, p,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1784 compflags + sp->ts_compsplit))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1785 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1786
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1787 if (slang->sl_nosplitsugs)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1788 newscore += SCORE_SPLIT_NO;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1789 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1790 newscore += SCORE_SPLIT;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1791
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1792 // Give a bonus to words seen before.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1793 newscore = score_wordcount_adj(slang, newscore,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1794 preword + sp->ts_prewordlen, TRUE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1795 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1796
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1797 if (TRY_DEEPER(su, stack, depth, newscore))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1798 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1799 go_deeper(stack, depth, newscore);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1800 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1801 if (!try_compound && !fword_ends)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1802 sprintf(changename[depth], "%.*s-%s: split",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1803 sp->ts_twordlen, tword, fword + sp->ts_fidx);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1804 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1805 sprintf(changename[depth], "%.*s-%s: compound",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1806 sp->ts_twordlen, tword, fword + sp->ts_fidx);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1807 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1808 // Save things to be restored at STATE_SPLITUNDO.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1809 sp->ts_save_badflags = su->su_badflags;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1810 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1811 sp->ts_state = STATE_SPLITUNDO;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1812
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1813 ++depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1814 sp = &stack[depth];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1815
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1816 // Append a space to preword when splitting.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1817 if (!try_compound && !fword_ends)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1818 STRCAT(preword, " ");
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1819 sp->ts_prewordlen = (char_u)STRLEN(preword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1820 sp->ts_splitoff = sp->ts_twordlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1821 sp->ts_splitfidx = sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1822
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1823 // If the badword has a non-word character at this
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1824 // position skip it. That means replacing the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1825 // non-word character with a space. Always skip a
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1826 // character when the word ends. But only when the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1827 // good word can end.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1828 if (((!try_compound && !spell_iswordp_nmw(fword
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1829 + sp->ts_fidx,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1830 curwin))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1831 || fword_ends)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1832 && fword[sp->ts_fidx] != NUL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1833 && goodword_ends)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1834 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1835 int l;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1836
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1837 l = mb_ptr2len(fword + sp->ts_fidx);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1838 if (fword_ends)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1839 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1840 // Copy the skipped character to preword.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1841 mch_memmove(preword + sp->ts_prewordlen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1842 fword + sp->ts_fidx, l);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1843 sp->ts_prewordlen += l;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1844 preword[sp->ts_prewordlen] = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1845 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1846 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1847 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1848 sp->ts_fidx += l;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1849 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1850
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1851 // When compounding include compound flag in
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1852 // compflags[] (already set above). When splitting we
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1853 // may start compounding over again.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1854 if (try_compound)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1855 ++sp->ts_complen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1856 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1857 sp->ts_compsplit = sp->ts_complen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1858 sp->ts_prefixdepth = PFD_NOPREFIX;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1859
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1860 // set su->su_badflags to the caps type at this
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1861 // position
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1862 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1863 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1864 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1865 n = sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1866 su->su_badflags = badword_captype(su->su_badptr + n,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1867 su->su_badptr + su->su_badlen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1868
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1869 // Restart at top of the tree.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1870 sp->ts_arridx = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1871
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1872 // If there are postponed prefixes, try these too.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1873 if (pbyts != NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1874 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1875 byts = pbyts;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1876 idxs = pidxs;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1877 sp->ts_prefixdepth = PFD_PREFIXTREE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1878 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1879 sp->ts_state = STATE_NOPREFIX;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1880 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1881 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1882 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1883 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1884 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1885
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1886 case STATE_SPLITUNDO:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1887 // Undo the changes done for word split or compound word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1888 su->su_badflags = sp->ts_save_badflags;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1889
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1890 // Continue looking for NUL bytes.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1891 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1892 sp->ts_state = STATE_START;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1893
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1894 // In case we went into the prefix tree.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1895 byts = fbyts;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1896 idxs = fidxs;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1897 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1898
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1899 case STATE_ENDNUL:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1900 // Past the NUL bytes in the node.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1901 su->su_badflags = sp->ts_save_badflags;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1902 if (fword[sp->ts_fidx] == NUL && sp->ts_tcharlen == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1903 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1904 // The badword ends, can't use STATE_PLAIN.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1905 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1906 sp->ts_state = STATE_DEL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1907 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1908 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1909 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1910 sp->ts_state = STATE_PLAIN;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1911 // FALLTHROUGH
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1912
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1913 case STATE_PLAIN:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1914 // Go over all possible bytes at this node, add each to tword[]
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1915 // and use child node. "ts_curi" is the index.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1916 arridx = sp->ts_arridx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1917 if (sp->ts_curi > byts[arridx])
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1918 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1919 // Done all bytes at this node, do next state. When still at
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1920 // already changed bytes skip the other tricks.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1921 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1922 if (sp->ts_fidx >= sp->ts_fidxtry)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1923 sp->ts_state = STATE_DEL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1924 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1925 sp->ts_state = STATE_FINAL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1926 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1927 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1928 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1929 arridx += sp->ts_curi++;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1930 c = byts[arridx];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1931
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1932 // Normal byte, go one level deeper. If it's not equal to the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1933 // byte in the bad word adjust the score. But don't even try
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1934 // when the byte was already changed. And don't try when we
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1935 // just deleted this byte, accepting it is always cheaper than
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1936 // delete + substitute.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1937 if (c == fword[sp->ts_fidx]
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1938 || (sp->ts_tcharlen > 0 && sp->ts_isdiff != DIFF_NONE))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1939 newscore = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1940 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1941 newscore = SCORE_SUBST;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1942 if ((newscore == 0
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1943 || (sp->ts_fidx >= sp->ts_fidxtry
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1944 && ((sp->ts_flags & TSF_DIDDEL) == 0
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1945 || c != fword[sp->ts_delidx])))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1946 && TRY_DEEPER(su, stack, depth, newscore))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1947 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1948 go_deeper(stack, depth, newscore);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1949 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1950 if (newscore > 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1951 sprintf(changename[depth], "%.*s-%s: subst %c to %c",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1952 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1953 fword[sp->ts_fidx], c);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1954 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1955 sprintf(changename[depth], "%.*s-%s: accept %c",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1956 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1957 fword[sp->ts_fidx]);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1958 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1959 ++depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1960 sp = &stack[depth];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1961 if (fword[sp->ts_fidx] != NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1962 ++sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1963 tword[sp->ts_twordlen++] = c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1964 sp->ts_arridx = idxs[arridx];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1965 if (newscore == SCORE_SUBST)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1966 sp->ts_isdiff = DIFF_YES;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1967 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1968 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1969 // Multi-byte characters are a bit complicated to
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1970 // handle: They differ when any of the bytes differ
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1971 // and then their length may also differ.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1972 if (sp->ts_tcharlen == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1973 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1974 // First byte.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1975 sp->ts_tcharidx = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1976 sp->ts_tcharlen = MB_BYTE2LEN(c);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1977 sp->ts_fcharstart = sp->ts_fidx - 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1978 sp->ts_isdiff = (newscore != 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1979 ? DIFF_YES : DIFF_NONE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1980 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1981 else if (sp->ts_isdiff == DIFF_INSERT
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1982 && sp->ts_fidx > 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1983 // When inserting trail bytes don't advance in the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1984 // bad word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1985 --sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1986 if (++sp->ts_tcharidx == sp->ts_tcharlen)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1987 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1988 // Last byte of character.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1989 if (sp->ts_isdiff == DIFF_YES)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1990 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1991 // Correct ts_fidx for the byte length of the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1992 // character (we didn't check that before).
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1993 sp->ts_fidx = sp->ts_fcharstart
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1994 + mb_ptr2len(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1995 fword + sp->ts_fcharstart);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1996 // For changing a composing character adjust
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1997 // the score from SCORE_SUBST to
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1998 // SCORE_SUBCOMP.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
1999 if (enc_utf8
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2000 && utf_iscomposing(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2001 utf_ptr2char(tword
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2002 + sp->ts_twordlen
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2003 - sp->ts_tcharlen))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2004 && utf_iscomposing(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2005 utf_ptr2char(fword
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2006 + sp->ts_fcharstart)))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2007 sp->ts_score -=
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2008 SCORE_SUBST - SCORE_SUBCOMP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2009
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2010 // For a similar character adjust score from
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2011 // SCORE_SUBST to SCORE_SIMILAR.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2012 else if (!soundfold
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2013 && slang->sl_has_map
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2014 && similar_chars(slang,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2015 mb_ptr2char(tword
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2016 + sp->ts_twordlen
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2017 - sp->ts_tcharlen),
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2018 mb_ptr2char(fword
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2019 + sp->ts_fcharstart)))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2020 sp->ts_score -=
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2021 SCORE_SUBST - SCORE_SIMILAR;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2022 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2023 else if (sp->ts_isdiff == DIFF_INSERT
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2024 && sp->ts_twordlen > sp->ts_tcharlen)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2025 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2026 p = tword + sp->ts_twordlen - sp->ts_tcharlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2027 c = mb_ptr2char(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2028 if (enc_utf8 && utf_iscomposing(c))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2029 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2030 // Inserting a composing char doesn't
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2031 // count that much.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2032 sp->ts_score -= SCORE_INS - SCORE_INSCOMP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2033 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2034 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2035 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2036 // If the previous character was the same,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2037 // thus doubling a character, give a bonus
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2038 // to the score. Also for the soundfold
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2039 // tree (might seem illogical but does
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2040 // give better scores).
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2041 MB_PTR_BACK(tword, p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2042 if (c == mb_ptr2char(p))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2043 sp->ts_score -= SCORE_INS
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2044 - SCORE_INSDUP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2045 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2046 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2047
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2048 // Starting a new char, reset the length.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2049 sp->ts_tcharlen = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2050 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2051 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2052 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2053 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2054 // If we found a similar char adjust the score.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2055 // We do this after calling go_deeper() because
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2056 // it's slow.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2057 if (newscore != 0
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2058 && !soundfold
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2059 && slang->sl_has_map
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2060 && similar_chars(slang,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2061 c, fword[sp->ts_fidx - 1]))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2062 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2063 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2064 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2065 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2066 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2067
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2068 case STATE_DEL:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2069 // When past the first byte of a multi-byte char don't try
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2070 // delete/insert/swap a character.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2071 if (has_mbyte && sp->ts_tcharlen > 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2072 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2073 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2074 sp->ts_state = STATE_FINAL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2075 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2076 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2077 // Try skipping one character in the bad word (delete it).
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2078 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2079 sp->ts_state = STATE_INS_PREP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2080 sp->ts_curi = 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2081 if (soundfold && sp->ts_fidx == 0 && fword[sp->ts_fidx] == '*')
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2082 // Deleting a vowel at the start of a word counts less, see
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2083 // soundalike_score().
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2084 newscore = 2 * SCORE_DEL / 3;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2085 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2086 newscore = SCORE_DEL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2087 if (fword[sp->ts_fidx] != NUL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2088 && TRY_DEEPER(su, stack, depth, newscore))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2089 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2090 go_deeper(stack, depth, newscore);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2091 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2092 sprintf(changename[depth], "%.*s-%s: delete %c",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2093 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2094 fword[sp->ts_fidx]);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2095 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2096 ++depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2097
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2098 // Remember what character we deleted, so that we can avoid
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2099 // inserting it again.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2100 stack[depth].ts_flags |= TSF_DIDDEL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2101 stack[depth].ts_delidx = sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2102
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2103 // Advance over the character in fword[]. Give a bonus to the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2104 // score if the same character is following "nn" -> "n". It's
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2105 // a bit illogical for soundfold tree but it does give better
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2106 // results.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2107 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2108 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2109 c = mb_ptr2char(fword + sp->ts_fidx);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2110 stack[depth].ts_fidx += mb_ptr2len(fword + sp->ts_fidx);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2111 if (enc_utf8 && utf_iscomposing(c))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2112 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2113 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2114 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2115 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2116 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2117 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2118 ++stack[depth].ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2119 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2120 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2121 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2122 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2123 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2124 // FALLTHROUGH
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2125
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2126 case STATE_INS_PREP:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2127 if (sp->ts_flags & TSF_DIDDEL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2128 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2129 // If we just deleted a byte then inserting won't make sense,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2130 // a substitute is always cheaper.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2131 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2132 sp->ts_state = STATE_SWAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2133 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2134 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2135
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2136 // skip over NUL bytes
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2137 n = sp->ts_arridx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2138 for (;;)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2139 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2140 if (sp->ts_curi > byts[n])
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2141 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2142 // Only NUL bytes at this node, go to next state.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2143 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2144 sp->ts_state = STATE_SWAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2145 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2146 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2147 if (byts[n + sp->ts_curi] != NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2148 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2149 // Found a byte to insert.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2150 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2151 sp->ts_state = STATE_INS;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2152 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2153 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2154 ++sp->ts_curi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2155 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2156 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2157
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2158 // FALLTHROUGH
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2159
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2160 case STATE_INS:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2161 // Insert one byte. Repeat this for each possible byte at this
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2162 // node.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2163 n = sp->ts_arridx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2164 if (sp->ts_curi > byts[n])
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2165 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2166 // Done all bytes at this node, go to next state.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2167 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2168 sp->ts_state = STATE_SWAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2169 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2170 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2171
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2172 // Do one more byte at this node, but:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2173 // - Skip NUL bytes.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2174 // - Skip the byte if it's equal to the byte in the word,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2175 // accepting that byte is always better.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2176 n += sp->ts_curi++;
33863
3b8089d550eb patch 9.0.2141: [security]: buffer-overflow in suggest_trie_walk
Christian Brabandt <cb@256bit.org>
parents: 33399
diff changeset
2177
3b8089d550eb patch 9.0.2141: [security]: buffer-overflow in suggest_trie_walk
Christian Brabandt <cb@256bit.org>
parents: 33399
diff changeset
2178 // break out, if we would be accessing byts buffer out of bounds
3b8089d550eb patch 9.0.2141: [security]: buffer-overflow in suggest_trie_walk
Christian Brabandt <cb@256bit.org>
parents: 33399
diff changeset
2179 if (byts == slang->sl_fbyts && n >= slang->sl_fbyts_len)
3b8089d550eb patch 9.0.2141: [security]: buffer-overflow in suggest_trie_walk
Christian Brabandt <cb@256bit.org>
parents: 33399
diff changeset
2180 {
3b8089d550eb patch 9.0.2141: [security]: buffer-overflow in suggest_trie_walk
Christian Brabandt <cb@256bit.org>
parents: 33399
diff changeset
2181 got_int = TRUE;
3b8089d550eb patch 9.0.2141: [security]: buffer-overflow in suggest_trie_walk
Christian Brabandt <cb@256bit.org>
parents: 33399
diff changeset
2182 break;
3b8089d550eb patch 9.0.2141: [security]: buffer-overflow in suggest_trie_walk
Christian Brabandt <cb@256bit.org>
parents: 33399
diff changeset
2183 }
32670
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2184 c = byts[n];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2185 if (soundfold && sp->ts_twordlen == 0 && c == '*')
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2186 // Inserting a vowel at the start of a word counts less,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2187 // see soundalike_score().
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2188 newscore = 2 * SCORE_INS / 3;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2189 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2190 newscore = SCORE_INS;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2191 if (c != fword[sp->ts_fidx]
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2192 && TRY_DEEPER(su, stack, depth, newscore))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2193 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2194 go_deeper(stack, depth, newscore);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2195 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2196 sprintf(changename[depth], "%.*s-%s: insert %c",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2197 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2198 c);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2199 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2200 ++depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2201 sp = &stack[depth];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2202 tword[sp->ts_twordlen++] = c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2203 sp->ts_arridx = idxs[n];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2204 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2205 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2206 fl = MB_BYTE2LEN(c);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2207 if (fl > 1)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2208 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2209 // There are following bytes for the same character.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2210 // We must find all bytes before trying
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2211 // delete/insert/swap/etc.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2212 sp->ts_tcharlen = fl;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2213 sp->ts_tcharidx = 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2214 sp->ts_isdiff = DIFF_INSERT;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2215 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2216 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2217 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2218 fl = 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2219 if (fl == 1)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2220 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2221 // If the previous character was the same, thus doubling a
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2222 // character, give a bonus to the score. Also for
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2223 // soundfold words (illogical but does give a better
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2224 // score).
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2225 if (sp->ts_twordlen >= 2
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2226 && tword[sp->ts_twordlen - 2] == c)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2227 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2228 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2229 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2230 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2231
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2232 case STATE_SWAP:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2233 // Swap two bytes in the bad word: "12" -> "21".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2234 // We change "fword" here, it's changed back afterwards at
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2235 // STATE_UNSWAP.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2236 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2237 c = *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2238 if (c == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2239 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2240 // End of word, can't swap or replace.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2241 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2242 sp->ts_state = STATE_FINAL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2243 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2244 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2245
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2246 // Don't swap if the first character is not a word character.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2247 // SWAP3 etc. also don't make sense then.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2248 if (!soundfold && !spell_iswordp(p, curwin))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2249 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2250 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2251 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2252 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2253 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2254
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2255 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2256 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2257 n = MB_CPTR2LEN(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2258 c = mb_ptr2char(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2259 if (p[n] == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2260 c2 = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2261 else if (!soundfold && !spell_iswordp(p + n, curwin))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2262 c2 = c; // don't swap non-word char
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2263 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2264 c2 = mb_ptr2char(p + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2265 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2266 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2267 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2268 if (p[1] == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2269 c2 = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2270 else if (!soundfold && !spell_iswordp(p + 1, curwin))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2271 c2 = c; // don't swap non-word char
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2272 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2273 c2 = p[1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2274 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2275
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2276 // When the second character is NUL we can't swap.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2277 if (c2 == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2278 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2279 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2280 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2281 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2282 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2283
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2284 // When characters are identical, swap won't do anything.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2285 // Also get here if the second char is not a word character.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2286 if (c == c2)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2287 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2288 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2289 sp->ts_state = STATE_SWAP3;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2290 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2291 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2292 if (c2 != NUL && TRY_DEEPER(su, stack, depth, SCORE_SWAP))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2293 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2294 go_deeper(stack, depth, SCORE_SWAP);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2295 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2296 sprintf(changename[depth], "%.*s-%s: swap %c and %c",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2297 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2298 c, c2);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2299 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2300 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2301 sp->ts_state = STATE_UNSWAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2302 ++depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2303 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2304 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2305 fl = mb_char2len(c2);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2306 mch_memmove(p, p + n, fl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2307 mb_char2bytes(c, p + fl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2308 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2309 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2310 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2311 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2312 p[0] = c2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2313 p[1] = c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2314 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2315 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2316 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2317 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2318 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2319 // If this swap doesn't work then SWAP3 won't either.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2320 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2321 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2322 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2323 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2324
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2325 case STATE_UNSWAP:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2326 // Undo the STATE_SWAP swap: "21" -> "12".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2327 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2328 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2329 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2330 n = mb_ptr2len(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2331 c = mb_ptr2char(p + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2332 mch_memmove(p + mb_ptr2len(p + n), p, n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2333 mb_char2bytes(c, p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2334 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2335 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2336 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2337 c = *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2338 *p = p[1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2339 p[1] = c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2340 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2341 // FALLTHROUGH
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2342
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2343 case STATE_SWAP3:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2344 // Swap two bytes, skipping one: "123" -> "321". We change
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2345 // "fword" here, it's changed back afterwards at STATE_UNSWAP3.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2346 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2347 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2348 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2349 n = MB_CPTR2LEN(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2350 c = mb_ptr2char(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2351 fl = MB_CPTR2LEN(p + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2352 c2 = mb_ptr2char(p + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2353 if (!soundfold && !spell_iswordp(p + n + fl, curwin))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2354 c3 = c; // don't swap non-word char
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2355 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2356 c3 = mb_ptr2char(p + n + fl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2357 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2358 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2359 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2360 c = *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2361 c2 = p[1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2362 if (!soundfold && !spell_iswordp(p + 2, curwin))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2363 c3 = c; // don't swap non-word char
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2364 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2365 c3 = p[2];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2366 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2367
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2368 // When characters are identical: "121" then SWAP3 result is
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2369 // identical, ROT3L result is same as SWAP: "211", ROT3L result is
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2370 // same as SWAP on next char: "112". Thus skip all swapping.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2371 // Also skip when c3 is NUL.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2372 // Also get here when the third character is not a word character.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2373 // Second character may any char: "a.b" -> "b.a"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2374 if (c == c3 || c3 == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2375 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2376 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2377 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2378 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2379 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2380 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2381 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2382 go_deeper(stack, depth, SCORE_SWAP3);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2383 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2384 sprintf(changename[depth], "%.*s-%s: swap3 %c and %c",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2385 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2386 c, c3);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2387 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2388 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2389 sp->ts_state = STATE_UNSWAP3;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2390 ++depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2391 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2392 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2393 tl = mb_char2len(c3);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2394 mch_memmove(p, p + n + fl, tl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2395 mb_char2bytes(c2, p + tl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2396 mb_char2bytes(c, p + fl + tl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2397 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2398 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2399 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2400 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2401 p[0] = p[2];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2402 p[2] = c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2403 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2404 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2405 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2406 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2407 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2408 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2409 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2410 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2411 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2412
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2413 case STATE_UNSWAP3:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2414 // Undo STATE_SWAP3: "321" -> "123"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2415 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2416 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2417 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2418 n = mb_ptr2len(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2419 c2 = mb_ptr2char(p + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2420 fl = mb_ptr2len(p + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2421 c = mb_ptr2char(p + n + fl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2422 tl = mb_ptr2len(p + n + fl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2423 mch_memmove(p + fl + tl, p, n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2424 mb_char2bytes(c, p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2425 mb_char2bytes(c2, p + tl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2426 p = p + tl;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2427 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2428 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2429 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2430 c = *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2431 *p = p[2];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2432 p[2] = c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2433 ++p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2434 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2435
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2436 if (!soundfold && !spell_iswordp(p, curwin))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2437 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2438 // Middle char is not a word char, skip the rotate. First and
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2439 // third char were already checked at swap and swap3.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2440 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2441 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2442 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2443 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2444
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2445 // Rotate three characters left: "123" -> "231". We change
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2446 // "fword" here, it's changed back afterwards at STATE_UNROT3L.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2447 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2448 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2449 go_deeper(stack, depth, SCORE_SWAP3);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2450 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2451 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2452 sprintf(changename[depth], "%.*s-%s: rotate left %c%c%c",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2453 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2454 p[0], p[1], p[2]);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2455 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2456 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2457 sp->ts_state = STATE_UNROT3L;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2458 ++depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2459 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2460 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2461 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2462 n = MB_CPTR2LEN(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2463 c = mb_ptr2char(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2464 fl = MB_CPTR2LEN(p + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2465 fl += MB_CPTR2LEN(p + n + fl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2466 mch_memmove(p, p + n, fl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2467 mb_char2bytes(c, p + fl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2468 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2469 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2470 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2471 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2472 c = *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2473 *p = p[1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2474 p[1] = p[2];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2475 p[2] = c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2476 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2477 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2478 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2479 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2480 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2481 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2482 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2483 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2484 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2485
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2486 case STATE_UNROT3L:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2487 // Undo ROT3L: "231" -> "123"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2488 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2489 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2490 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2491 n = mb_ptr2len(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2492 n += mb_ptr2len(p + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2493 c = mb_ptr2char(p + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2494 tl = mb_ptr2len(p + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2495 mch_memmove(p + tl, p, n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2496 mb_char2bytes(c, p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2497 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2498 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2499 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2500 c = p[2];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2501 p[2] = p[1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2502 p[1] = *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2503 *p = c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2504 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2505
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2506 // Rotate three bytes right: "123" -> "312". We change "fword"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2507 // here, it's changed back afterwards at STATE_UNROT3R.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2508 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2509 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2510 go_deeper(stack, depth, SCORE_SWAP3);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2511 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2512 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2513 sprintf(changename[depth], "%.*s-%s: rotate right %c%c%c",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2514 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2515 p[0], p[1], p[2]);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2516 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2517 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2518 sp->ts_state = STATE_UNROT3R;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2519 ++depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2520 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2521 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2522 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2523 n = MB_CPTR2LEN(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2524 n += MB_CPTR2LEN(p + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2525 c = mb_ptr2char(p + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2526 tl = MB_CPTR2LEN(p + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2527 mch_memmove(p + tl, p, n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2528 mb_char2bytes(c, p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2529 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2530 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2531 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2532 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2533 c = p[2];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2534 p[2] = p[1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2535 p[1] = *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2536 *p = c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2537 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2538 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2539 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2540 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2541 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2542 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2543 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2544 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2545 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2546
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2547 case STATE_UNROT3R:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2548 // Undo ROT3R: "312" -> "123"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2549 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2550 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2551 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2552 c = mb_ptr2char(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2553 tl = mb_ptr2len(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2554 n = mb_ptr2len(p + tl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2555 n += mb_ptr2len(p + tl + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2556 mch_memmove(p, p + tl, n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2557 mb_char2bytes(c, p + n);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2558 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2559 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2560 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2561 c = *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2562 *p = p[1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2563 p[1] = p[2];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2564 p[2] = c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2565 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2566 // FALLTHROUGH
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2567
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2568 case STATE_REP_INI:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2569 // Check if matching with REP items from the .aff file would work.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2570 // Quickly skip if:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2571 // - there are no REP items and we are not in the soundfold trie
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2572 // - the score is going to be too high anyway
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2573 // - already applied a REP item or swapped here
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2574 if ((lp->lp_replang == NULL && !soundfold)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2575 || sp->ts_score + SCORE_REP >= su->su_maxscore
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2576 || sp->ts_fidx < sp->ts_fidxtry)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2577 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2578 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2579 sp->ts_state = STATE_FINAL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2580 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2581 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2582
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2583 // Use the first byte to quickly find the first entry that may
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2584 // match. If the index is -1 there is none.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2585 if (soundfold)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2586 sp->ts_curi = slang->sl_repsal_first[fword[sp->ts_fidx]];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2587 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2588 sp->ts_curi = lp->lp_replang->sl_rep_first[fword[sp->ts_fidx]];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2589
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2590 if (sp->ts_curi < 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2591 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2592 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2593 sp->ts_state = STATE_FINAL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2594 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2595 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2596
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2597 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2598 sp->ts_state = STATE_REP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2599 // FALLTHROUGH
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2600
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2601 case STATE_REP:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2602 // Try matching with REP items from the .aff file. For each match
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2603 // replace the characters and check if the resulting word is
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2604 // valid.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2605 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2606
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2607 if (soundfold)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2608 gap = &slang->sl_repsal;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2609 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2610 gap = &lp->lp_replang->sl_rep;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2611 while (sp->ts_curi < gap->ga_len)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2612 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2613 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2614 if (*ftp->ft_from != *p)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2615 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2616 // past possible matching entries
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2617 sp->ts_curi = gap->ga_len;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2618 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2619 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2620 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2621 && TRY_DEEPER(su, stack, depth, SCORE_REP))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2622 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2623 go_deeper(stack, depth, SCORE_REP);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2624 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2625 sprintf(changename[depth], "%.*s-%s: replace %s with %s",
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2626 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2627 ftp->ft_from, ftp->ft_to);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2628 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2629 // Need to undo this afterwards.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2630 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2631 sp->ts_state = STATE_REP_UNDO;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2632
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2633 // Change the "from" to the "to" string.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2634 ++depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2635 fl = (int)STRLEN(ftp->ft_from);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2636 tl = (int)STRLEN(ftp->ft_to);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2637 if (fl != tl)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2638 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2639 STRMOVE(p + tl, p + fl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2640 repextra += tl - fl;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2641 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2642 mch_memmove(p, ftp->ft_to, tl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2643 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2644 stack[depth].ts_tcharlen = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2645 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2646 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2647 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2648
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2649 if (sp->ts_curi >= gap->ga_len && sp->ts_state == STATE_REP)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2650 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2651 // No (more) matches.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2652 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2653 sp->ts_state = STATE_FINAL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2654 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2655
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2656 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2657
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2658 case STATE_REP_UNDO:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2659 // Undo a REP replacement and continue with the next one.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2660 if (soundfold)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2661 gap = &slang->sl_repsal;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2662 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2663 gap = &lp->lp_replang->sl_rep;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2664 ftp = (fromto_T *)gap->ga_data + sp->ts_curi - 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2665 fl = (int)STRLEN(ftp->ft_from);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2666 tl = (int)STRLEN(ftp->ft_to);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2667 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2668 if (fl != tl)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2669 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2670 STRMOVE(p + fl, p + tl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2671 repextra -= tl - fl;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2672 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2673 mch_memmove(p, ftp->ft_from, fl);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2674 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2675 sp->ts_state = STATE_REP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2676 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2677
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2678 default:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2679 // Did all possible states at this level, go up one level.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2680 --depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2681
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2682 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2683 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2684 // Continue in or go back to the prefix tree.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2685 byts = pbyts;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2686 idxs = pidxs;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2687 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2688
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2689 // Don't check for CTRL-C too often, it takes time.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2690 if (--breakcheckcount == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2691 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2692 ui_breakcheck();
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2693 breakcheckcount = 1000;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2694 #ifdef FEAT_RELTIME
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2695 if (spell_suggest_timeout > 0
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2696 && profile_passed_limit(&time_limit))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2697 got_int = TRUE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2698 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2699 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2700 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2701 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2702 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2703
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2704
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2705 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2706 * Go one level deeper in the tree.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2707 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2708 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2709 go_deeper(trystate_T *stack, int depth, int score_add)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2710 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2711 stack[depth + 1] = stack[depth];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2712 stack[depth + 1].ts_state = STATE_START;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2713 stack[depth + 1].ts_score = stack[depth].ts_score + score_add;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2714 stack[depth + 1].ts_curi = 1; // start just after length byte
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2715 stack[depth + 1].ts_flags = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2716 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2717
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2718 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2719 * "fword" is a good word with case folded. Find the matching keep-case
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2720 * words and put it in "kword".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2721 * Theoretically there could be several keep-case words that result in the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2722 * same case-folded word, but we only find one...
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2723 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2724 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2725 find_keepcap_word(slang_T *slang, char_u *fword, char_u *kword)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2726 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2727 char_u uword[MAXWLEN]; // "fword" in upper-case
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2728 int depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2729 idx_T tryidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2730
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2731 // The following arrays are used at each depth in the tree.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2732 idx_T arridx[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2733 int round[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2734 int fwordidx[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2735 int uwordidx[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2736 int kwordlen[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2737
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2738 int flen, ulen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2739 int l;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2740 int len;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2741 int c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2742 idx_T lo, hi, m;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2743 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2744 char_u *byts = slang->sl_kbyts; // array with bytes of the words
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2745 idx_T *idxs = slang->sl_kidxs; // array with indexes
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2746
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2747 if (byts == NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2748 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2749 // array is empty: "cannot happen"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2750 *kword = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2751 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2752 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2753
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2754 // Make an all-cap version of "fword".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2755 allcap_copy(fword, uword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2756
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2757 // Each character needs to be tried both case-folded and upper-case.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2758 // All this gets very complicated if we keep in mind that changing case
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2759 // may change the byte length of a multi-byte character...
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2760 depth = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2761 arridx[0] = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2762 round[0] = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2763 fwordidx[0] = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2764 uwordidx[0] = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2765 kwordlen[0] = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2766 while (depth >= 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2767 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2768 if (fword[fwordidx[depth]] == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2769 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2770 // We are at the end of "fword". If the tree allows a word to end
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2771 // here we have found a match.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2772 if (byts[arridx[depth] + 1] == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2773 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2774 kword[kwordlen[depth]] = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2775 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2776 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2777
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2778 // kword is getting too long, continue one level up
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2779 --depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2780 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2781 else if (++round[depth] > 2)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2782 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2783 // tried both fold-case and upper-case character, continue one
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2784 // level up
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2785 --depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2786 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2787 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2788 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2789 // round[depth] == 1: Try using the folded-case character.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2790 // round[depth] == 2: Try using the upper-case character.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2791 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2792 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2793 flen = MB_CPTR2LEN(fword + fwordidx[depth]);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2794 ulen = MB_CPTR2LEN(uword + uwordidx[depth]);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2795 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2796 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2797 ulen = flen = 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2798 if (round[depth] == 1)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2799 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2800 p = fword + fwordidx[depth];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2801 l = flen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2802 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2803 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2804 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2805 p = uword + uwordidx[depth];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2806 l = ulen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2807 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2808
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2809 for (tryidx = arridx[depth]; l > 0; --l)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2810 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2811 // Perform a binary search in the list of accepted bytes.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2812 len = byts[tryidx++];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2813 c = *p++;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2814 lo = tryidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2815 hi = tryidx + len - 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2816 while (lo < hi)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2817 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2818 m = (lo + hi) / 2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2819 if (byts[m] > c)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2820 hi = m - 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2821 else if (byts[m] < c)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2822 lo = m + 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2823 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2824 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2825 lo = hi = m;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2826 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2827 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2828 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2829
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2830 // Stop if there is no matching byte.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2831 if (hi < lo || byts[lo] != c)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2832 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2833
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2834 // Continue at the child (if there is one).
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2835 tryidx = idxs[lo];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2836 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2837
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2838 if (l == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2839 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2840 // Found the matching char. Copy it to "kword" and go a
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2841 // level deeper.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2842 if (round[depth] == 1)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2843 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2844 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2845 flen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2846 kwordlen[depth + 1] = kwordlen[depth] + flen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2847 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2848 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2849 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2850 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2851 ulen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2852 kwordlen[depth + 1] = kwordlen[depth] + ulen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2853 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2854 fwordidx[depth + 1] = fwordidx[depth] + flen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2855 uwordidx[depth + 1] = uwordidx[depth] + ulen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2856
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2857 ++depth;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2858 arridx[depth] = tryidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2859 round[depth] = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2860 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2861 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2862 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2863
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2864 // Didn't find it: "cannot happen".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2865 *kword = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2866 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2867
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2868 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2869 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2870 * su->su_sga.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2871 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2872 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2873 score_comp_sal(suginfo_T *su)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2874 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2875 langp_T *lp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2876 char_u badsound[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2877 int i;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2878 suggest_T *stp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2879 suggest_T *sstp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2880 int score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2881 int lpi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2882
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2883 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2884 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2885
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2886 // Use the sound-folding of the first language that supports it.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2887 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2888 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2889 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2890 if (lp->lp_slang->sl_sal.ga_len > 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2891 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2892 // soundfold the bad word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2893 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2894
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2895 for (i = 0; i < su->su_ga.ga_len; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2896 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2897 stp = &SUG(su->su_ga, i);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2898
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2899 // Case-fold the suggested word, sound-fold it and compute the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2900 // sound-a-like score.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2901 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2902 if (score < SCORE_MAXMAX)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2903 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2904 // Add the suggestion.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2905 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2906 sstp->st_word = vim_strsave(stp->st_word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2907 if (sstp->st_word != NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2908 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2909 sstp->st_wordlen = stp->st_wordlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2910 sstp->st_score = score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2911 sstp->st_altscore = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2912 sstp->st_orglen = stp->st_orglen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2913 ++su->su_sga.ga_len;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2914 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2915 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2916 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2917 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2918 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2919 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2920 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2921
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2922 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2923 * Combine the list of suggestions in su->su_ga and su->su_sga.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2924 * They are entwined.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2925 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2926 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2927 score_combine(suginfo_T *su)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2928 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2929 int i;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2930 int j;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2931 garray_T ga;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2932 garray_T *gap;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2933 langp_T *lp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2934 suggest_T *stp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2935 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2936 char_u badsound[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2937 int round;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2938 int lpi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2939 slang_T *slang = NULL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2940
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2941 // Add the alternate score to su_ga.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2942 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2943 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2944 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2945 if (lp->lp_slang->sl_sal.ga_len > 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2946 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2947 // soundfold the bad word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2948 slang = lp->lp_slang;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2949 spell_soundfold(slang, su->su_fbadword, TRUE, badsound);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2950
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2951 for (i = 0; i < su->su_ga.ga_len; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2952 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2953 stp = &SUG(su->su_ga, i);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2954 stp->st_altscore = stp_sal_score(stp, su, slang, badsound);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2955 if (stp->st_altscore == SCORE_MAXMAX)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2956 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2957 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2958 stp->st_score = (stp->st_score * 3
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2959 + stp->st_altscore) / 4;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2960 stp->st_salscore = FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2961 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2962 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2963 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2964 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2965
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2966 if (slang == NULL) // Using "double" without sound folding.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2967 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2968 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2969 su->su_maxcount);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2970 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2971 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2972
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2973 // Add the alternate score to su_sga.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2974 for (i = 0; i < su->su_sga.ga_len; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2975 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2976 stp = &SUG(su->su_sga, i);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2977 stp->st_altscore = spell_edit_score(slang,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2978 su->su_badword, stp->st_word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2979 if (stp->st_score == SCORE_MAXMAX)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2980 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2981 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2982 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2983 stp->st_salscore = TRUE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2984 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2985
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2986 // Remove bad suggestions, sort the suggestions and truncate at "maxcount"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2987 // for both lists.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2988 check_suggestions(su, &su->su_ga);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2989 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2990 check_suggestions(su, &su->su_sga);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2991 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2992
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2993 ga_init2(&ga, sizeof(suginfo_T), 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2994 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2995 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2996
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2997 stp = &SUG(ga, 0);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2998 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
2999 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3000 // round 1: get a suggestion from su_ga
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3001 // round 2: get a suggestion from su_sga
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3002 for (round = 1; round <= 2; ++round)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3003 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3004 gap = round == 1 ? &su->su_ga : &su->su_sga;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3005 if (i < gap->ga_len)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3006 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3007 // Don't add a word if it's already there.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3008 p = SUG(*gap, i).st_word;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3009 for (j = 0; j < ga.ga_len; ++j)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3010 if (STRCMP(stp[j].st_word, p) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3011 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3012 if (j == ga.ga_len)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3013 stp[ga.ga_len++] = SUG(*gap, i);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3014 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3015 vim_free(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3016 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3017 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3018 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3019
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3020 ga_clear(&su->su_ga);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3021 ga_clear(&su->su_sga);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3022
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3023 // Truncate the list to the number of suggestions that will be displayed.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3024 if (ga.ga_len > su->su_maxcount)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3025 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3026 for (i = su->su_maxcount; i < ga.ga_len; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3027 vim_free(stp[i].st_word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3028 ga.ga_len = su->su_maxcount;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3029 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3030
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3031 su->su_ga = ga;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3032 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3033
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3034 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3035 * For the goodword in "stp" compute the soundalike score compared to the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3036 * badword.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3037 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3038 static int
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3039 stp_sal_score(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3040 suggest_T *stp,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3041 suginfo_T *su,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3042 slang_T *slang,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3043 char_u *badsound) // sound-folded badword
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3044 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3045 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3046 char_u *pbad;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3047 char_u *pgood;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3048 char_u badsound2[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3049 char_u fword[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3050 char_u goodsound[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3051 char_u goodword[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3052 int lendiff;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3053
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3054 lendiff = (int)(su->su_badlen - stp->st_orglen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3055 if (lendiff >= 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3056 pbad = badsound;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3057 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3058 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3059 // soundfold the bad word with more characters following
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3060 (void)spell_casefold(curwin,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3061 su->su_badptr, stp->st_orglen, fword, MAXWLEN);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3062
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3063 // When joining two words the sound often changes a lot. E.g., "t he"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3064 // sounds like "t h" while "the" sounds like "@". Avoid that by
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3065 // removing the space. Don't do it when the good word also contains a
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3066 // space.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3067 if (VIM_ISWHITE(su->su_badptr[su->su_badlen])
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3068 && *skiptowhite(stp->st_word) == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3069 for (p = fword; *(p = skiptowhite(p)) != NUL; )
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3070 STRMOVE(p, p + 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3071
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3072 spell_soundfold(slang, fword, TRUE, badsound2);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3073 pbad = badsound2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3074 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3075
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3076 if (lendiff > 0 && stp->st_wordlen + lendiff < MAXWLEN)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3077 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3078 // Add part of the bad word to the good word, so that we soundfold
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3079 // what replaces the bad word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3080 STRCPY(goodword, stp->st_word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3081 vim_strncpy(goodword + stp->st_wordlen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3082 su->su_badptr + su->su_badlen - lendiff, lendiff);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3083 pgood = goodword;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3084 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3085 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3086 pgood = stp->st_word;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3087
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3088 // Sound-fold the word and compute the score for the difference.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3089 spell_soundfold(slang, pgood, FALSE, goodsound);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3090
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3091 return soundalike_score(goodsound, pbad);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3092 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3093
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3094 // structure used to store soundfolded words that add_sound_suggest() has
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3095 // handled already.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3096 typedef struct
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3097 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3098 short sft_score; // lowest score used
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3099 char_u sft_word[1]; // soundfolded word, actually longer
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3100 } sftword_T;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3101
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3102 static sftword_T dumsft;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3103 #define HIKEY2SFT(p) ((sftword_T *)((p) - (dumsft.sft_word - (char_u *)&dumsft)))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3104 #define HI2SFT(hi) HIKEY2SFT((hi)->hi_key)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3105
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3106 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3107 * Prepare for calling suggest_try_soundalike().
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3108 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3109 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3110 suggest_try_soundalike_prep(void)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3111 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3112 langp_T *lp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3113 int lpi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3114 slang_T *slang;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3115
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3116 // Do this for all languages that support sound folding and for which a
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3117 // .sug file has been loaded.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3118 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3119 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3120 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3121 slang = lp->lp_slang;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3122 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3123 // prepare the hashtable used by add_sound_suggest()
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3124 hash_init(&slang->sl_sounddone);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3125 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3126 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3127
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3128 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3129 * Find suggestions by comparing the word in a sound-a-like form.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3130 * Note: This doesn't support postponed prefixes.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3131 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3132 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3133 suggest_try_soundalike(suginfo_T *su)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3134 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3135 char_u salword[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3136 langp_T *lp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3137 int lpi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3138 slang_T *slang;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3139
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3140 // Do this for all languages that support sound folding and for which a
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3141 // .sug file has been loaded.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3142 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3143 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3144 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3145 slang = lp->lp_slang;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3146 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3147 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3148 // soundfold the bad word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3149 spell_soundfold(slang, su->su_fbadword, TRUE, salword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3150
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3151 // try all kinds of inserts/deletes/swaps/etc.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3152 // TODO: also soundfold the next words, so that we can try joining
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3153 // and splitting
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3154 #ifdef SUGGEST_PROFILE
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3155 prof_init();
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3156 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3157 suggest_trie_walk(su, lp, salword, TRUE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3158 #ifdef SUGGEST_PROFILE
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3159 prof_report("soundalike");
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3160 #endif
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3161 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3162 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3163 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3164
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3165 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3166 * Finish up after calling suggest_try_soundalike().
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3167 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3168 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3169 suggest_try_soundalike_finish(void)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3170 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3171 langp_T *lp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3172 int lpi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3173 slang_T *slang;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3174 int todo;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3175 hashitem_T *hi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3176
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3177 // Do this for all languages that support sound folding and for which a
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3178 // .sug file has been loaded.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3179 for (lpi = 0; lpi < curwin->w_s->b_langp.ga_len; ++lpi)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3180 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3181 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3182 slang = lp->lp_slang;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3183 if (slang->sl_sal.ga_len > 0 && slang->sl_sbyts != NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3184 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3185 // Free the info about handled words.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3186 todo = (int)slang->sl_sounddone.ht_used;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3187 FOR_ALL_HASHTAB_ITEMS(&slang->sl_sounddone, hi, todo)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3188 if (!HASHITEM_EMPTY(hi))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3189 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3190 vim_free(HI2SFT(hi));
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3191 --todo;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3192 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3193
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3194 // Clear the hashtable, it may also be used by another region.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3195 hash_clear(&slang->sl_sounddone);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3196 hash_init(&slang->sl_sounddone);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3197 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3198 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3199 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3200
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3201 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3202 * A match with a soundfolded word is found. Add the good word(s) that
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3203 * produce this soundfolded word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3204 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3205 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3206 add_sound_suggest(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3207 suginfo_T *su,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3208 char_u *goodword,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3209 int score, // soundfold score
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3210 langp_T *lp)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3211 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3212 slang_T *slang = lp->lp_slang; // language for sound folding
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3213 int sfwordnr;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3214 char_u *nrline;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3215 int orgnr;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3216 char_u theword[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3217 int i;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3218 int wlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3219 char_u *byts;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3220 idx_T *idxs;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3221 int n;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3222 int wordcount;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3223 int wc;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3224 int goodscore;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3225 hash_T hash;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3226 hashitem_T *hi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3227 sftword_T *sft;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3228 int bc, gc;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3229 int limit;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3230
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3231 // It's very well possible that the same soundfold word is found several
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3232 // times with different scores. Since the following is quite slow only do
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3233 // the words that have a better score than before. Use a hashtable to
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3234 // remember the words that have been done.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3235 hash = hash_hash(goodword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3236 hi = hash_lookup(&slang->sl_sounddone, goodword, hash);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3237 if (HASHITEM_EMPTY(hi))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3238 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3239 sft = alloc(offsetof(sftword_T, sft_word) + STRLEN(goodword) + 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3240 if (sft != NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3241 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3242 sft->sft_score = score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3243 STRCPY(sft->sft_word, goodword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3244 hash_add_item(&slang->sl_sounddone, hi, sft->sft_word, hash);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3245 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3246 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3247 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3248 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3249 sft = HI2SFT(hi);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3250 if (score >= sft->sft_score)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3251 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3252 sft->sft_score = score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3253 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3254
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3255 // Find the word nr in the soundfold tree.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3256 sfwordnr = soundfold_find(slang, goodword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3257 if (sfwordnr < 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3258 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3259 internal_error("add_sound_suggest()");
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3260 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3261 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3262
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3263 // go over the list of good words that produce this soundfold word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3264 nrline = ml_get_buf(slang->sl_sugbuf, (linenr_T)(sfwordnr + 1), FALSE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3265 orgnr = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3266 while (*nrline != NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3267 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3268 // The wordnr was stored in a minimal nr of bytes as an offset to the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3269 // previous wordnr.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3270 orgnr += bytes2offset(&nrline);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3271
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3272 byts = slang->sl_fbyts;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3273 idxs = slang->sl_fidxs;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3274
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3275 // Lookup the word "orgnr" one of the two tries.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3276 n = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3277 wordcount = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3278 for (wlen = 0; wlen < MAXWLEN - 3; ++wlen)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3279 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3280 i = 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3281 if (wordcount == orgnr && byts[n + 1] == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3282 break; // found end of word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3283
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3284 if (byts[n + 1] == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3285 ++wordcount;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3286
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3287 // skip over the NUL bytes
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3288 for ( ; byts[n + i] == NUL; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3289 if (i > byts[n]) // safety check
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3290 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3291 STRCPY(theword + wlen, "BAD");
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3292 wlen += 3;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3293 goto badword;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3294 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3295
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3296 // One of the siblings must have the word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3297 for ( ; i < byts[n]; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3298 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3299 wc = idxs[idxs[n + i]]; // nr of words under this byte
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3300 if (wordcount + wc > orgnr)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3301 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3302 wordcount += wc;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3303 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3304
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3305 theword[wlen] = byts[n + i];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3306 n = idxs[n + i];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3307 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3308 badword:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3309 theword[wlen] = NUL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3310
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3311 // Go over the possible flags and regions.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3312 for (; i <= byts[n] && byts[n + i] == NUL; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3313 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3314 char_u cword[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3315 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3316 int flags = (int)idxs[n + i];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3317
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3318 // Skip words with the NOSUGGEST flag
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3319 if (flags & WF_NOSUGGEST)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3320 continue;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3321
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3322 if (flags & WF_KEEPCAP)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3323 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3324 // Must find the word in the keep-case tree.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3325 find_keepcap_word(slang, theword, cword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3326 p = cword;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3327 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3328 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3329 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3330 flags |= su->su_badflags;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3331 if ((flags & WF_CAPMASK) != 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3332 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3333 // Need to fix case according to "flags".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3334 make_case_word(theword, cword, flags);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3335 p = cword;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3336 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3337 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3338 p = theword;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3339 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3340
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3341 // Add the suggestion.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3342 if (sps_flags & SPS_DOUBLE)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3343 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3344 // Add the suggestion if the score isn't too bad.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3345 if (score <= su->su_maxscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3346 add_suggestion(su, &su->su_sga, p, su->su_badlen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3347 score, 0, FALSE, slang, FALSE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3348 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3349 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3350 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3351 // Add a penalty for words in another region.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3352 if ((flags & WF_REGION)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3353 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3354 goodscore = SCORE_REGION;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3355 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3356 goodscore = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3357
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3358 // Add a small penalty for changing the first letter from
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3359 // lower to upper case. Helps for "tath" -> "Kath", which is
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3360 // less common than "tath" -> "path". Don't do it when the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3361 // letter is the same, that has already been counted.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3362 gc = PTR2CHAR(p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3363 if (SPELL_ISUPPER(gc))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3364 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3365 bc = PTR2CHAR(su->su_badword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3366 if (!SPELL_ISUPPER(bc)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3367 && SPELL_TOFOLD(bc) != SPELL_TOFOLD(gc))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3368 goodscore += SCORE_ICASE / 2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3369 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3370
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3371 // Compute the score for the good word. This only does letter
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3372 // insert/delete/swap/replace. REP items are not considered,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3373 // which may make the score a bit higher.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3374 // Use a limit for the score to make it work faster. Use
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3375 // MAXSCORE(), because RESCORE() will change the score.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3376 // If the limit is very high then the iterative method is
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3377 // inefficient, using an array is quicker.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3378 limit = MAXSCORE(su->su_sfmaxscore - goodscore, score);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3379 if (limit > SCORE_LIMITMAX)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3380 goodscore += spell_edit_score(slang, su->su_badword, p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3381 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3382 goodscore += spell_edit_score_limit(slang, su->su_badword,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3383 p, limit);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3384
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3385 // When going over the limit don't bother to do the rest.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3386 if (goodscore < SCORE_MAXMAX)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3387 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3388 // Give a bonus to words seen before.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3389 goodscore = score_wordcount_adj(slang, goodscore, p, FALSE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3390
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3391 // Add the suggestion if the score isn't too bad.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3392 goodscore = RESCORE(goodscore, score);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3393 if (goodscore <= su->su_sfmaxscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3394 add_suggestion(su, &su->su_ga, p, su->su_badlen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3395 goodscore, score, TRUE, slang, TRUE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3396 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3397 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3398 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3399 // smsg("word %s (%d): %s (%d)", sftword, sftnr, theword, orgnr);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3400 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3401 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3402
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3403 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3404 * Find word "word" in fold-case tree for "slang" and return the word number.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3405 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3406 static int
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3407 soundfold_find(slang_T *slang, char_u *word)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3408 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3409 idx_T arridx = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3410 int len;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3411 int wlen = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3412 int c;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3413 char_u *ptr = word;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3414 char_u *byts;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3415 idx_T *idxs;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3416 int wordnr = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3417
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3418 byts = slang->sl_sbyts;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3419 idxs = slang->sl_sidxs;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3420
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3421 for (;;)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3422 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3423 // First byte is the number of possible bytes.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3424 len = byts[arridx++];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3425
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3426 // If the first possible byte is a zero the word could end here.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3427 // If the word ends we found the word. If not skip the NUL bytes.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3428 c = ptr[wlen];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3429 if (byts[arridx] == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3430 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3431 if (c == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3432 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3433
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3434 // Skip over the zeros, there can be several.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3435 while (len > 0 && byts[arridx] == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3436 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3437 ++arridx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3438 --len;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3439 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3440 if (len == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3441 return -1; // no children, word should have ended here
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3442 ++wordnr;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3443 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3444
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3445 // If the word ends we didn't find it.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3446 if (c == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3447 return -1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3448
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3449 // Perform a binary search in the list of accepted bytes.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3450 if (c == TAB) // <Tab> is handled like <Space>
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3451 c = ' ';
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3452 while (byts[arridx] < c)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3453 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3454 // The word count is in the first idxs[] entry of the child.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3455 wordnr += idxs[idxs[arridx]];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3456 ++arridx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3457 if (--len == 0) // end of the bytes, didn't find it
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3458 return -1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3459 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3460 if (byts[arridx] != c) // didn't find the byte
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3461 return -1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3462
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3463 // Continue at the child (if there is one).
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3464 arridx = idxs[arridx];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3465 ++wlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3466
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3467 // One space in the good word may stand for several spaces in the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3468 // checked word.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3469 if (c == ' ')
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3470 while (ptr[wlen] == ' ' || ptr[wlen] == TAB)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3471 ++wlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3472 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3473
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3474 return wordnr;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3475 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3476
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3477 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3478 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3479 * lines in the .aff file.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3480 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3481 static int
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3482 similar_chars(slang_T *slang, int c1, int c2)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3483 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3484 int m1, m2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3485 char_u buf[MB_MAXBYTES + 1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3486 hashitem_T *hi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3487
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3488 if (c1 >= 256)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3489 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3490 buf[mb_char2bytes(c1, buf)] = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3491 hi = hash_find(&slang->sl_map_hash, buf);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3492 if (HASHITEM_EMPTY(hi))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3493 m1 = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3494 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3495 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3496 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3497 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3498 m1 = slang->sl_map_array[c1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3499 if (m1 == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3500 return FALSE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3501
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3502
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3503 if (c2 >= 256)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3504 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3505 buf[mb_char2bytes(c2, buf)] = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3506 hi = hash_find(&slang->sl_map_hash, buf);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3507 if (HASHITEM_EMPTY(hi))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3508 m2 = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3509 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3510 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3511 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3512 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3513 m2 = slang->sl_map_array[c2];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3514
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3515 return m1 == m2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3516 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3517
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3518 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3519 * Add a suggestion to the list of suggestions.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3520 * For a suggestion that is already in the list the lowest score is remembered.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3521 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3522 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3523 add_suggestion(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3524 suginfo_T *su,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3525 garray_T *gap, // either su_ga or su_sga
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3526 char_u *goodword,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3527 int badlenarg, // len of bad word replaced with "goodword"
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3528 int score,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3529 int altscore,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3530 int had_bonus, // value for st_had_bonus
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3531 slang_T *slang, // language for sound folding
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3532 int maxsf) // su_maxscore applies to soundfold score,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3533 // su_sfmaxscore to the total score.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3534 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3535 int goodlen; // len of goodword changed
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3536 int badlen; // len of bad word changed
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3537 suggest_T *stp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3538 suggest_T new_sug;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3539 int i;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3540 char_u *pgood, *pbad;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3541
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3542 // Minimize "badlen" for consistency. Avoids that changing "the the" to
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3543 // "thee the" is added next to changing the first "the" the "thee".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3544 pgood = goodword + STRLEN(goodword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3545 pbad = su->su_badptr + badlenarg;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3546 for (;;)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3547 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3548 goodlen = (int)(pgood - goodword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3549 badlen = (int)(pbad - su->su_badptr);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3550 if (goodlen <= 0 || badlen <= 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3551 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3552 MB_PTR_BACK(goodword, pgood);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3553 MB_PTR_BACK(su->su_badptr, pbad);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3554 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3555 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3556 if (mb_ptr2char(pgood) != mb_ptr2char(pbad))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3557 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3558 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3559 else if (*pgood != *pbad)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3560 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3561 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3562
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3563 if (badlen == 0 && goodlen == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3564 // goodword doesn't change anything; may happen for "the the" changing
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3565 // the first "the" to itself.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3566 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3567
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3568 if (gap->ga_len == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3569 i = -1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3570 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3571 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3572 // Check if the word is already there. Also check the length that is
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3573 // being replaced "thes," -> "these" is a different suggestion from
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3574 // "thes" -> "these".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3575 stp = &SUG(*gap, 0);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3576 for (i = gap->ga_len; --i >= 0; ++stp)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3577 if (stp->st_wordlen == goodlen
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3578 && stp->st_orglen == badlen
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3579 && STRNCMP(stp->st_word, goodword, goodlen) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3580 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3581 // Found it. Remember the word with the lowest score.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3582 if (stp->st_slang == NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3583 stp->st_slang = slang;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3584
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3585 new_sug.st_score = score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3586 new_sug.st_altscore = altscore;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3587 new_sug.st_had_bonus = had_bonus;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3588
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3589 if (stp->st_had_bonus != had_bonus)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3590 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3591 // Only one of the two had the soundalike score computed.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3592 // Need to do that for the other one now, otherwise the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3593 // scores can't be compared. This happens because
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3594 // suggest_try_change() doesn't compute the soundalike
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3595 // word to keep it fast, while some special methods set
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3596 // the soundalike score to zero.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3597 if (had_bonus)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3598 rescore_one(su, stp);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3599 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3600 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3601 new_sug.st_word = stp->st_word;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3602 new_sug.st_wordlen = stp->st_wordlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3603 new_sug.st_slang = stp->st_slang;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3604 new_sug.st_orglen = badlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3605 rescore_one(su, &new_sug);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3606 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3607 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3608
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3609 if (stp->st_score > new_sug.st_score)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3610 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3611 stp->st_score = new_sug.st_score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3612 stp->st_altscore = new_sug.st_altscore;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3613 stp->st_had_bonus = new_sug.st_had_bonus;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3614 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3615 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3616 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3617 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3618
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3619 if (i < 0 && ga_grow(gap, 1) == OK)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3620 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3621 // Add a suggestion.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3622 stp = &SUG(*gap, gap->ga_len);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3623 stp->st_word = vim_strnsave(goodword, goodlen);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3624 if (stp->st_word != NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3625 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3626 stp->st_wordlen = goodlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3627 stp->st_score = score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3628 stp->st_altscore = altscore;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3629 stp->st_had_bonus = had_bonus;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3630 stp->st_orglen = badlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3631 stp->st_slang = slang;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3632 ++gap->ga_len;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3633
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3634 // If we have too many suggestions now, sort the list and keep
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3635 // the best suggestions.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3636 if (gap->ga_len > SUG_MAX_COUNT(su))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3637 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3638 if (maxsf)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3639 su->su_sfmaxscore = cleanup_suggestions(gap,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3640 su->su_sfmaxscore, SUG_CLEAN_COUNT(su));
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3641 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3642 su->su_maxscore = cleanup_suggestions(gap,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3643 su->su_maxscore, SUG_CLEAN_COUNT(su));
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3644 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3645 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3646 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3647 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3648
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3649 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3650 * Suggestions may in fact be flagged as errors. Esp. for banned words and
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3651 * for split words, such as "the the". Remove these from the list here.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3652 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3653 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3654 check_suggestions(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3655 suginfo_T *su,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3656 garray_T *gap) // either su_ga or su_sga
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3657 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3658 suggest_T *stp;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3659 int i;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3660 char_u longword[MAXWLEN + 1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3661 int len;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3662 hlf_T attr;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3663
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3664 if (gap->ga_len == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3665 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3666 stp = &SUG(*gap, 0);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3667 for (i = gap->ga_len - 1; i >= 0; --i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3668 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3669 // Need to append what follows to check for "the the".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3670 vim_strncpy(longword, stp[i].st_word, MAXWLEN);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3671 len = stp[i].st_wordlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3672 vim_strncpy(longword + len, su->su_badptr + stp[i].st_orglen,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3673 MAXWLEN - len);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3674 attr = HLF_COUNT;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3675 (void)spell_check(curwin, longword, &attr, NULL, FALSE);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3676 if (attr != HLF_COUNT)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3677 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3678 // Remove this entry.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3679 vim_free(stp[i].st_word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3680 --gap->ga_len;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3681 if (i < gap->ga_len)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3682 mch_memmove(stp + i, stp + i + 1,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3683 sizeof(suggest_T) * (gap->ga_len - i));
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3684 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3685 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3686 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3687
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3688
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3689 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3690 * Add a word to be banned.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3691 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3692 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3693 add_banned(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3694 suginfo_T *su,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3695 char_u *word)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3696 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3697 char_u *s;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3698 hash_T hash;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3699 hashitem_T *hi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3700
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3701 hash = hash_hash(word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3702 hi = hash_lookup(&su->su_banned, word, hash);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3703 if (!HASHITEM_EMPTY(hi)) // already present
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3704 return;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3705 s = vim_strsave(word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3706 if (s != NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3707 hash_add_item(&su->su_banned, hi, s, hash);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3708 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3709
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3710 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3711 * Recompute the score for all suggestions if sound-folding is possible. This
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3712 * is slow, thus only done for the final results.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3713 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3714 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3715 rescore_suggestions(suginfo_T *su)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3716 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3717 int i;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3718
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3719 if (su->su_sallang != NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3720 for (i = 0; i < su->su_ga.ga_len; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3721 rescore_one(su, &SUG(su->su_ga, i));
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3722 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3723
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3724 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3725 * Recompute the score for one suggestion if sound-folding is possible.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3726 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3727 static void
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3728 rescore_one(suginfo_T *su, suggest_T *stp)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3729 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3730 slang_T *slang = stp->st_slang;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3731 char_u sal_badword[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3732 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3733
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3734 // Only rescore suggestions that have no sal score yet and do have a
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3735 // language.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3736 if (slang != NULL && slang->sl_sal.ga_len > 0 && !stp->st_had_bonus)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3737 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3738 if (slang == su->su_sallang)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3739 p = su->su_sal_badword;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3740 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3741 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3742 spell_soundfold(slang, su->su_fbadword, TRUE, sal_badword);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3743 p = sal_badword;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3744 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3745
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3746 stp->st_altscore = stp_sal_score(stp, su, slang, p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3747 if (stp->st_altscore == SCORE_MAXMAX)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3748 stp->st_altscore = SCORE_BIG;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3749 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3750 stp->st_had_bonus = TRUE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3751 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3752 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3753
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3754 static int sug_compare(const void *s1, const void *s2);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3755
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3756 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3757 * Function given to qsort() to sort the suggestions on st_score.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3758 * First on "st_score", then "st_altscore" then alphabetically.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3759 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3760 static int
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3761 sug_compare(const void *s1, const void *s2)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3762 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3763 suggest_T *p1 = (suggest_T *)s1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3764 suggest_T *p2 = (suggest_T *)s2;
34309
d7cfd8fb1d75 patch 9.1.0089: qsort() comparison functions should be transitive
Christian Brabandt <cb@256bit.org>
parents: 33863
diff changeset
3765 int n;
d7cfd8fb1d75 patch 9.1.0089: qsort() comparison functions should be transitive
Christian Brabandt <cb@256bit.org>
parents: 33863
diff changeset
3766
d7cfd8fb1d75 patch 9.1.0089: qsort() comparison functions should be transitive
Christian Brabandt <cb@256bit.org>
parents: 33863
diff changeset
3767 n = p1->st_score == p2->st_score ? 0 :
d7cfd8fb1d75 patch 9.1.0089: qsort() comparison functions should be transitive
Christian Brabandt <cb@256bit.org>
parents: 33863
diff changeset
3768 p1->st_score > p2->st_score ? 1 : -1;
32670
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3769
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3770 if (n == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3771 {
34309
d7cfd8fb1d75 patch 9.1.0089: qsort() comparison functions should be transitive
Christian Brabandt <cb@256bit.org>
parents: 33863
diff changeset
3772 n = p1->st_altscore == p2->st_altscore ? 0 :
d7cfd8fb1d75 patch 9.1.0089: qsort() comparison functions should be transitive
Christian Brabandt <cb@256bit.org>
parents: 33863
diff changeset
3773 p1->st_altscore > p2->st_altscore ? 1 : -1;
d7cfd8fb1d75 patch 9.1.0089: qsort() comparison functions should be transitive
Christian Brabandt <cb@256bit.org>
parents: 33863
diff changeset
3774
32670
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3775 if (n == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3776 n = STRICMP(p1->st_word, p2->st_word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3777 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3778 return n;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3779 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3780
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3781 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3782 * Cleanup the suggestions:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3783 * - Sort on score.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3784 * - Remove words that won't be displayed.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3785 * Returns the maximum score in the list or "maxscore" unmodified.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3786 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3787 static int
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3788 cleanup_suggestions(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3789 garray_T *gap,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3790 int maxscore,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3791 int keep) // nr of suggestions to keep
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3792 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3793 if (gap->ga_len <= 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3794 return maxscore;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3795
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3796 // Sort the list.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3797 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T),
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3798 sug_compare);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3799
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3800 // Truncate the list to the number of suggestions that will be
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3801 // displayed.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3802 if (gap->ga_len > keep)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3803 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3804 int i;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3805 suggest_T *stp = &SUG(*gap, 0);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3806
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3807 for (i = keep; i < gap->ga_len; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3808 vim_free(stp[i].st_word);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3809 gap->ga_len = keep;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3810 if (keep >= 1)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3811 return stp[keep - 1].st_score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3812 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3813 return maxscore;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3814 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3815
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3816 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3817 * Compute a score for two sound-a-like words.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3818 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3819 * Instead of a generic loop we write out the code. That keeps it fast by
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3820 * avoiding checks that will not be possible.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3821 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3822 static int
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3823 soundalike_score(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3824 char_u *goodstart, // sound-folded good word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3825 char_u *badstart) // sound-folded bad word
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3826 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3827 char_u *goodsound = goodstart;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3828 char_u *badsound = badstart;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3829 int goodlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3830 int badlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3831 int n;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3832 char_u *pl, *ps;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3833 char_u *pl2, *ps2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3834 int score = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3835
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3836 // Adding/inserting "*" at the start (word starts with vowel) shouldn't be
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3837 // counted so much, vowels halfway the word aren't counted at all.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3838 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3839 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3840 if ((badsound[0] == NUL && goodsound[1] == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3841 || (goodsound[0] == NUL && badsound[1] == NUL))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3842 // changing word with vowel to word without a sound
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3843 return SCORE_DEL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3844 if (badsound[0] == NUL || goodsound[0] == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3845 // more than two changes
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3846 return SCORE_MAXMAX;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3847
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3848 if (badsound[1] == goodsound[1]
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3849 || (badsound[1] != NUL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3850 && goodsound[1] != NUL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3851 && badsound[2] == goodsound[2]))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3852 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3853 // handle like a substitute
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3854 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3855 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3856 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3857 score = 2 * SCORE_DEL / 3;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3858 if (*badsound == '*')
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3859 ++badsound;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3860 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3861 ++goodsound;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3862 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3863 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3864
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3865 goodlen = (int)STRLEN(goodsound);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3866 badlen = (int)STRLEN(badsound);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3867
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3868 // Return quickly if the lengths are too different to be fixed by two
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3869 // changes.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3870 n = goodlen - badlen;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3871 if (n < -2 || n > 2)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3872 return SCORE_MAXMAX;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3873
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3874 if (n > 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3875 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3876 pl = goodsound; // goodsound is longest
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3877 ps = badsound;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3878 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3879 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3880 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3881 pl = badsound; // badsound is longest
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3882 ps = goodsound;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3883 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3884
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3885 // Skip over the identical part.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3886 while (*pl == *ps && *pl != NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3887 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3888 ++pl;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3889 ++ps;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3890 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3891
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3892 switch (n)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3893 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3894 case -2:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3895 case 2:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3896 // Must delete two characters from "pl".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3897 ++pl; // first delete
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3898 while (*pl == *ps)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3899 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3900 ++pl;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3901 ++ps;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3902 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3903 // strings must be equal after second delete
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3904 if (STRCMP(pl + 1, ps) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3905 return score + SCORE_DEL * 2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3906
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3907 // Failed to compare.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3908 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3909
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3910 case -1:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3911 case 1:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3912 // Minimal one delete from "pl" required.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3913
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3914 // 1: delete
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3915 pl2 = pl + 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3916 ps2 = ps;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3917 while (*pl2 == *ps2)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3918 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3919 if (*pl2 == NUL) // reached the end
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3920 return score + SCORE_DEL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3921 ++pl2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3922 ++ps2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3923 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3924
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3925 // 2: delete then swap, then rest must be equal
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3926 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3927 && STRCMP(pl2 + 2, ps2 + 2) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3928 return score + SCORE_DEL + SCORE_SWAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3929
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3930 // 3: delete then substitute, then the rest must be equal
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3931 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3932 return score + SCORE_DEL + SCORE_SUBST;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3933
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3934 // 4: first swap then delete
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3935 if (pl[0] == ps[1] && pl[1] == ps[0])
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3936 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3937 pl2 = pl + 2; // swap, skip two chars
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3938 ps2 = ps + 2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3939 while (*pl2 == *ps2)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3940 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3941 ++pl2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3942 ++ps2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3943 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3944 // delete a char and then strings must be equal
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3945 if (STRCMP(pl2 + 1, ps2) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3946 return score + SCORE_SWAP + SCORE_DEL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3947 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3948
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3949 // 5: first substitute then delete
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3950 pl2 = pl + 1; // substitute, skip one char
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3951 ps2 = ps + 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3952 while (*pl2 == *ps2)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3953 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3954 ++pl2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3955 ++ps2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3956 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3957 // delete a char and then strings must be equal
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3958 if (STRCMP(pl2 + 1, ps2) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3959 return score + SCORE_SUBST + SCORE_DEL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3960
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3961 // Failed to compare.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3962 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3963
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3964 case 0:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3965 // Lengths are equal, thus changes must result in same length: An
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3966 // insert is only possible in combination with a delete.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3967 // 1: check if for identical strings
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3968 if (*pl == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3969 return score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3970
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3971 // 2: swap
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3972 if (pl[0] == ps[1] && pl[1] == ps[0])
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3973 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3974 pl2 = pl + 2; // swap, skip two chars
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3975 ps2 = ps + 2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3976 while (*pl2 == *ps2)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3977 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3978 if (*pl2 == NUL) // reached the end
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3979 return score + SCORE_SWAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3980 ++pl2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3981 ++ps2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3982 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3983 // 3: swap and swap again
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3984 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3985 && STRCMP(pl2 + 2, ps2 + 2) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3986 return score + SCORE_SWAP + SCORE_SWAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3987
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3988 // 4: swap and substitute
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3989 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3990 return score + SCORE_SWAP + SCORE_SUBST;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3991 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3992
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3993 // 5: substitute
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3994 pl2 = pl + 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3995 ps2 = ps + 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3996 while (*pl2 == *ps2)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3997 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3998 if (*pl2 == NUL) // reached the end
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
3999 return score + SCORE_SUBST;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4000 ++pl2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4001 ++ps2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4002 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4003
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4004 // 6: substitute and swap
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4005 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4006 && STRCMP(pl2 + 2, ps2 + 2) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4007 return score + SCORE_SUBST + SCORE_SWAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4008
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4009 // 7: substitute and substitute
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4010 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4011 return score + SCORE_SUBST + SCORE_SUBST;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4012
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4013 // 8: insert then delete
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4014 pl2 = pl;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4015 ps2 = ps + 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4016 while (*pl2 == *ps2)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4017 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4018 ++pl2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4019 ++ps2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4020 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4021 if (STRCMP(pl2 + 1, ps2) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4022 return score + SCORE_INS + SCORE_DEL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4023
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4024 // 9: delete then insert
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4025 pl2 = pl + 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4026 ps2 = ps;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4027 while (*pl2 == *ps2)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4028 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4029 ++pl2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4030 ++ps2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4031 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4032 if (STRCMP(pl2, ps2 + 1) == 0)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4033 return score + SCORE_INS + SCORE_DEL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4034
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4035 // Failed to compare.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4036 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4037 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4038
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4039 return SCORE_MAXMAX;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4040 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4041
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4042 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4043 * Compute the "edit distance" to turn "badword" into "goodword". The less
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4044 * deletes/inserts/substitutes/swaps are required the lower the score.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4045 *
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4046 * The algorithm is described by Du and Chang, 1992.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4047 * The implementation of the algorithm comes from Aspell editdist.cpp,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4048 * edit_distance(). It has been converted from C++ to C and modified to
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4049 * support multi-byte characters.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4050 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4051 static int
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4052 spell_edit_score(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4053 slang_T *slang,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4054 char_u *badword,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4055 char_u *goodword)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4056 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4057 int *cnt;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4058 int badlen, goodlen; // lengths including NUL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4059 int j, i;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4060 int t;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4061 int bc, gc;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4062 int pbc, pgc;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4063 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4064 int wbadword[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4065 int wgoodword[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4066
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4067 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4068 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4069 // Get the characters from the multi-byte strings and put them in an
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4070 // int array for easy access.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4071 for (p = badword, badlen = 0; *p != NUL; )
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4072 wbadword[badlen++] = mb_cptr2char_adv(&p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4073 wbadword[badlen++] = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4074 for (p = goodword, goodlen = 0; *p != NUL; )
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4075 wgoodword[goodlen++] = mb_cptr2char_adv(&p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4076 wgoodword[goodlen++] = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4077 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4078 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4079 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4080 badlen = (int)STRLEN(badword) + 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4081 goodlen = (int)STRLEN(goodword) + 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4082 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4083
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4084 // We use "cnt" as an array: CNT(badword_idx, goodword_idx).
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4085 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4086 cnt = ALLOC_MULT(int, (badlen + 1) * (goodlen + 1));
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4087 if (cnt == NULL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4088 return 0; // out of memory
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4089
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4090 CNT(0, 0) = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4091 for (j = 1; j <= goodlen; ++j)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4092 CNT(0, j) = CNT(0, j - 1) + SCORE_INS;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4093
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4094 for (i = 1; i <= badlen; ++i)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4095 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4096 CNT(i, 0) = CNT(i - 1, 0) + SCORE_DEL;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4097 for (j = 1; j <= goodlen; ++j)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4098 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4099 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4100 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4101 bc = wbadword[i - 1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4102 gc = wgoodword[j - 1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4103 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4104 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4105 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4106 bc = badword[i - 1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4107 gc = goodword[j - 1];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4108 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4109 if (bc == gc)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4110 CNT(i, j) = CNT(i - 1, j - 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4111 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4112 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4113 // Use a better score when there is only a case difference.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4114 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4115 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4116 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4117 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4118 // For a similar character use SCORE_SIMILAR.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4119 if (slang != NULL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4120 && slang->sl_has_map
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4121 && similar_chars(slang, gc, bc))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4122 CNT(i, j) = SCORE_SIMILAR + CNT(i - 1, j - 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4123 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4124 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4125 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4126
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4127 if (i > 1 && j > 1)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4128 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4129 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4130 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4131 pbc = wbadword[i - 2];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4132 pgc = wgoodword[j - 2];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4133 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4134 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4135 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4136 pbc = badword[i - 2];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4137 pgc = goodword[j - 2];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4138 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4139 if (bc == pgc && pbc == gc)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4140 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4141 t = SCORE_SWAP + CNT(i - 2, j - 2);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4142 if (t < CNT(i, j))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4143 CNT(i, j) = t;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4144 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4145 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4146 t = SCORE_DEL + CNT(i - 1, j);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4147 if (t < CNT(i, j))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4148 CNT(i, j) = t;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4149 t = SCORE_INS + CNT(i, j - 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4150 if (t < CNT(i, j))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4151 CNT(i, j) = t;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4152 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4153 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4154 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4155
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4156 i = CNT(badlen - 1, goodlen - 1);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4157 vim_free(cnt);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4158 return i;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4159 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4160
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4161 typedef struct
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4162 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4163 int badi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4164 int goodi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4165 int score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4166 } limitscore_T;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4167
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4168 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4169 * Like spell_edit_score(), but with a limit on the score to make it faster.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4170 * May return SCORE_MAXMAX when the score is higher than "limit".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4171 *
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4172 * This uses a stack for the edits still to be tried.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4173 * The idea comes from Aspell leditdist.cpp. Rewritten in C and added support
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4174 * for multi-byte characters.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4175 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4176 static int
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4177 spell_edit_score_limit(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4178 slang_T *slang,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4179 char_u *badword,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4180 char_u *goodword,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4181 int limit)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4182 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4183 limitscore_T stack[10]; // allow for over 3 * 2 edits
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4184 int stackidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4185 int bi, gi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4186 int bi2, gi2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4187 int bc, gc;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4188 int score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4189 int score_off;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4190 int minscore;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4191 int round;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4192
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4193 // Multi-byte characters require a bit more work, use a different function
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4194 // to avoid testing "has_mbyte" quite often.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4195 if (has_mbyte)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4196 return spell_edit_score_limit_w(slang, badword, goodword, limit);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4197
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4198 // The idea is to go from start to end over the words. So long as
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4199 // characters are equal just continue, this always gives the lowest score.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4200 // When there is a difference try several alternatives. Each alternative
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4201 // increases "score" for the edit distance. Some of the alternatives are
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4202 // pushed unto a stack and tried later, some are tried right away. At the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4203 // end of the word the score for one alternative is known. The lowest
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4204 // possible score is stored in "minscore".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4205 stackidx = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4206 bi = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4207 gi = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4208 score = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4209 minscore = limit + 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4210
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4211 for (;;)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4212 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4213 // Skip over an equal part, score remains the same.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4214 for (;;)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4215 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4216 bc = badword[bi];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4217 gc = goodword[gi];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4218 if (bc != gc) // stop at a char that's different
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4219 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4220 if (bc == NUL) // both words end
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4221 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4222 if (score < minscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4223 minscore = score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4224 goto pop; // do next alternative
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4225 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4226 ++bi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4227 ++gi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4228 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4229
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4230 if (gc == NUL) // goodword ends, delete badword chars
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4231 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4232 do
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4233 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4234 if ((score += SCORE_DEL) >= minscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4235 goto pop; // do next alternative
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4236 } while (badword[++bi] != NUL);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4237 minscore = score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4238 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4239 else if (bc == NUL) // badword ends, insert badword chars
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4240 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4241 do
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4242 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4243 if ((score += SCORE_INS) >= minscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4244 goto pop; // do next alternative
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4245 } while (goodword[++gi] != NUL);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4246 minscore = score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4247 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4248 else // both words continue
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4249 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4250 // If not close to the limit, perform a change. Only try changes
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4251 // that may lead to a lower score than "minscore".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4252 // round 0: try deleting a char from badword
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4253 // round 1: try inserting a char in badword
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4254 for (round = 0; round <= 1; ++round)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4255 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4256 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4257 if (score_off < minscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4258 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4259 if (score_off + SCORE_EDIT_MIN >= minscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4260 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4261 // Near the limit, rest of the words must match. We
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4262 // can check that right now, no need to push an item
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4263 // onto the stack.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4264 bi2 = bi + 1 - round;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4265 gi2 = gi + round;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4266 while (goodword[gi2] == badword[bi2])
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4267 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4268 if (goodword[gi2] == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4269 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4270 minscore = score_off;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4271 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4272 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4273 ++bi2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4274 ++gi2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4275 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4276 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4277 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4278 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4279 // try deleting/inserting a character later
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4280 stack[stackidx].badi = bi + 1 - round;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4281 stack[stackidx].goodi = gi + round;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4282 stack[stackidx].score = score_off;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4283 ++stackidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4284 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4285 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4286 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4287
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4288 if (score + SCORE_SWAP < minscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4289 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4290 // If swapping two characters makes a match then the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4291 // substitution is more expensive, thus there is no need to
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4292 // try both.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4293 if (gc == badword[bi + 1] && bc == goodword[gi + 1])
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4294 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4295 // Swap two characters, that is: skip them.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4296 gi += 2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4297 bi += 2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4298 score += SCORE_SWAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4299 continue;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4300 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4301 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4302
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4303 // Substitute one character for another which is the same
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4304 // thing as deleting a character from both goodword and badword.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4305 // Use a better score when there is only a case difference.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4306 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4307 score += SCORE_ICASE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4308 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4309 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4310 // For a similar character use SCORE_SIMILAR.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4311 if (slang != NULL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4312 && slang->sl_has_map
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4313 && similar_chars(slang, gc, bc))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4314 score += SCORE_SIMILAR;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4315 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4316 score += SCORE_SUBST;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4317 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4318
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4319 if (score < minscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4320 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4321 // Do the substitution.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4322 ++gi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4323 ++bi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4324 continue;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4325 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4326 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4327 pop:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4328 // Get here to try the next alternative, pop it from the stack.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4329 if (stackidx == 0) // stack is empty, finished
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4330 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4331
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4332 // pop an item from the stack
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4333 --stackidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4334 gi = stack[stackidx].goodi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4335 bi = stack[stackidx].badi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4336 score = stack[stackidx].score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4337 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4338
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4339 // When the score goes over "limit" it may actually be much higher.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4340 // Return a very large number to avoid going below the limit when giving a
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4341 // bonus.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4342 if (minscore > limit)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4343 return SCORE_MAXMAX;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4344 return minscore;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4345 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4346
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4347 /*
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4348 * Multi-byte version of spell_edit_score_limit().
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4349 * Keep it in sync with the above!
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4350 */
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4351 static int
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4352 spell_edit_score_limit_w(
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4353 slang_T *slang,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4354 char_u *badword,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4355 char_u *goodword,
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4356 int limit)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4357 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4358 limitscore_T stack[10]; // allow for over 3 * 2 edits
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4359 int stackidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4360 int bi, gi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4361 int bi2, gi2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4362 int bc, gc;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4363 int score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4364 int score_off;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4365 int minscore;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4366 int round;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4367 char_u *p;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4368 int wbadword[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4369 int wgoodword[MAXWLEN];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4370
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4371 // Get the characters from the multi-byte strings and put them in an
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4372 // int array for easy access.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4373 bi = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4374 for (p = badword; *p != NUL; )
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4375 wbadword[bi++] = mb_cptr2char_adv(&p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4376 wbadword[bi++] = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4377 gi = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4378 for (p = goodword; *p != NUL; )
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4379 wgoodword[gi++] = mb_cptr2char_adv(&p);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4380 wgoodword[gi++] = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4381
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4382 // The idea is to go from start to end over the words. So long as
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4383 // characters are equal just continue, this always gives the lowest score.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4384 // When there is a difference try several alternatives. Each alternative
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4385 // increases "score" for the edit distance. Some of the alternatives are
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4386 // pushed unto a stack and tried later, some are tried right away. At the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4387 // end of the word the score for one alternative is known. The lowest
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4388 // possible score is stored in "minscore".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4389 stackidx = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4390 bi = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4391 gi = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4392 score = 0;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4393 minscore = limit + 1;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4394
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4395 for (;;)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4396 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4397 // Skip over an equal part, score remains the same.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4398 for (;;)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4399 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4400 bc = wbadword[bi];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4401 gc = wgoodword[gi];
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4402
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4403 if (bc != gc) // stop at a char that's different
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4404 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4405 if (bc == NUL) // both words end
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4406 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4407 if (score < minscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4408 minscore = score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4409 goto pop; // do next alternative
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4410 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4411 ++bi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4412 ++gi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4413 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4414
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4415 if (gc == NUL) // goodword ends, delete badword chars
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4416 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4417 do
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4418 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4419 if ((score += SCORE_DEL) >= minscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4420 goto pop; // do next alternative
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4421 } while (wbadword[++bi] != NUL);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4422 minscore = score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4423 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4424 else if (bc == NUL) // badword ends, insert badword chars
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4425 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4426 do
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4427 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4428 if ((score += SCORE_INS) >= minscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4429 goto pop; // do next alternative
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4430 } while (wgoodword[++gi] != NUL);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4431 minscore = score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4432 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4433 else // both words continue
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4434 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4435 // If not close to the limit, perform a change. Only try changes
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4436 // that may lead to a lower score than "minscore".
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4437 // round 0: try deleting a char from badword
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4438 // round 1: try inserting a char in badword
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4439 for (round = 0; round <= 1; ++round)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4440 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4441 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS);
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4442 if (score_off < minscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4443 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4444 if (score_off + SCORE_EDIT_MIN >= minscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4445 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4446 // Near the limit, rest of the words must match. We
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4447 // can check that right now, no need to push an item
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4448 // onto the stack.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4449 bi2 = bi + 1 - round;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4450 gi2 = gi + round;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4451 while (wgoodword[gi2] == wbadword[bi2])
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4452 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4453 if (wgoodword[gi2] == NUL)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4454 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4455 minscore = score_off;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4456 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4457 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4458 ++bi2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4459 ++gi2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4460 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4461 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4462 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4463 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4464 // try deleting a character from badword later
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4465 stack[stackidx].badi = bi + 1 - round;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4466 stack[stackidx].goodi = gi + round;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4467 stack[stackidx].score = score_off;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4468 ++stackidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4469 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4470 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4471 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4472
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4473 if (score + SCORE_SWAP < minscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4474 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4475 // If swapping two characters makes a match then the
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4476 // substitution is more expensive, thus there is no need to
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4477 // try both.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4478 if (gc == wbadword[bi + 1] && bc == wgoodword[gi + 1])
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4479 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4480 // Swap two characters, that is: skip them.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4481 gi += 2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4482 bi += 2;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4483 score += SCORE_SWAP;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4484 continue;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4485 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4486 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4487
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4488 // Substitute one character for another which is the same
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4489 // thing as deleting a character from both goodword and badword.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4490 // Use a better score when there is only a case difference.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4491 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4492 score += SCORE_ICASE;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4493 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4494 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4495 // For a similar character use SCORE_SIMILAR.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4496 if (slang != NULL
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4497 && slang->sl_has_map
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4498 && similar_chars(slang, gc, bc))
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4499 score += SCORE_SIMILAR;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4500 else
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4501 score += SCORE_SUBST;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4502 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4503
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4504 if (score < minscore)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4505 {
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4506 // Do the substitution.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4507 ++gi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4508 ++bi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4509 continue;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4510 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4511 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4512 pop:
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4513 // Get here to try the next alternative, pop it from the stack.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4514 if (stackidx == 0) // stack is empty, finished
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4515 break;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4516
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4517 // pop an item from the stack
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4518 --stackidx;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4519 gi = stack[stackidx].goodi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4520 bi = stack[stackidx].badi;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4521 score = stack[stackidx].score;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4522 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4523
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4524 // When the score goes over "limit" it may actually be much higher.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4525 // Return a very large number to avoid going below the limit when giving a
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4526 // bonus.
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4527 if (minscore > limit)
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4528 return SCORE_MAXMAX;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4529 return minscore;
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4530 }
695b50472e85 Fix line endings issue
Christian Brabandt <cb@256bit.org>
parents: 32669
diff changeset
4531 #endif // FEAT_SPELL