vim: src/spellsuggest.c annotate

annotate src/spellsuggest.c @ 34051:f96f16107d61

runtime(sh): Update sh syntax and add local keyword for bash (#13806) Commit: https://github.com/vim/vim/commit/b16fc9805535dc6eb084142da0c87210fc102494 Author: Lucien Grondin <grondilu@yahoo.fr> Date: Mon Jan 1 19:00:41 2024 +0100 runtime(sh): Update sh syntax and add local keyword for bash (https://github.com/vim/vim/issues/13806) add `local` in shStatement Signed-off-by: Lucien Grondin <grondilu@yahoo.fr> Signed-off-by: Christian Brabandt <cb@256bit.org>

author	Christian Brabandt <cb@256bit.org>
date	Mon, 01 Jan 2024 19:15:03 +0100
parents	3b8089d550eb
children	d7cfd8fb1d75

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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	511 line = ml_get_curline();
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	512 if (badlen > (int)STRLEN(line) - (int)curwin->w_cursor.col)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	513 badlen = (int)STRLEN(line) - (int)curwin->w_cursor.col;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	514 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	515 // Find the start of the badly spelled word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	516 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	517 \|\| curwin->w_cursor.col > prev_cursor.col)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	518 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	519 // 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	520 // cursor.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	521 curwin->w_cursor = prev_cursor;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	522 line = ml_get_curline();
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	523 p = line + curwin->w_cursor.col;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	524 // Backup to before start of word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	525 while (p > line && spell_iswordp_nmw(p, curwin))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	526 MB_PTR_BACK(line, p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	527 // Forward to start of word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	528 while (*p != NUL && !spell_iswordp_nmw(p, curwin))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	529 MB_PTR_ADV(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	530
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	531 if (!spell_iswordp_nmw(p, curwin)) // No word found.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	532 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	533 beep_flush();
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	534 return;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	535 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	536 curwin->w_cursor.col = (colnr_T)(p - line);
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	539 // Get the word and its length.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	540
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	541 // Figure out if the word should be capitalised.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	542 need_cap = check_need_cap(curwin, curwin->w_cursor.lnum,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	543 curwin->w_cursor.col);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	544
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	545 // Make a copy of current line since autocommands may free the line.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	546 line = vim_strsave(ml_get_curline());
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	547 if (line == NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	548 goto skip;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	549
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	550 // 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	551 // 'spellsuggest', whatever is smaller.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	552 if (sps_limit > (int)Rows - 2)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	553 limit = (int)Rows - 2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	554 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	555 limit = sps_limit;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	556 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	557 TRUE, need_cap, TRUE);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	558
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	559 if (sug.su_ga.ga_len == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	560 msg(_("Sorry, no suggestions"));
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	561 else if (count > 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	562 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	563 if (count > sug.su_ga.ga_len)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	564 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	565 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	566 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	567 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	568 #ifdef FEAT_RIGHTLEFT
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	569 // When 'rightleft' is set the list is drawn right-left.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	570 cmdmsg_rl = curwin->w_p_rl;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	571 if (cmdmsg_rl)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	572 msg_col = Columns - 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	573 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	574
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	575 // List the suggestions.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	576 msg_start();
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	577 msg_row = Rows - 1; // for when 'cmdheight' > 1
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	578 lines_left = Rows; // avoid more prompt
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	579 vim_snprintf((char )IObuff, IOSIZE, _("Change \"%.s\" to:"),
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	580 sug.su_badlen, sug.su_badptr);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	581 #ifdef FEAT_RIGHTLEFT
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	582 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	583 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	584 // 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	585 // untranslated message rightleft.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	586 vim_snprintf((char )IObuff, IOSIZE, ":ot \"%.s\" egnahC",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	587 sug.su_badlen, sug.su_badptr);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	588 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	589 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	590 msg_puts((char *)IObuff);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	591 msg_clr_eos();
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	592 msg_putchar('\n');
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	593
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	594 msg_scroll = TRUE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	595 for (i = 0; i < sug.su_ga.ga_len; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	596 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	597 int el;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	598
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	599 stp = &SUG(sug.su_ga, i);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	600
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	601 // 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	602 // 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	603 vim_strncpy(wcopy, stp->st_word, MAXWLEN);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	604 el = sug.su_badlen - stp->st_orglen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	605 if (el > 0 && stp->st_wordlen + el <= MAXWLEN)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	606 vim_strncpy(wcopy + stp->st_wordlen,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	607 sug.su_badptr + stp->st_orglen, el);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	608 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	609 #ifdef FEAT_RIGHTLEFT
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	610 if (cmdmsg_rl)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	611 rl_mirror(IObuff);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	612 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	613 msg_puts((char *)IObuff);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	614
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	615 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	616 msg_puts((char *)IObuff);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	617
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	618 // The word may replace more than "su_badlen".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	619 if (sug.su_badlen < stp->st_orglen)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	620 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	621 vim_snprintf((char )IObuff, IOSIZE, _(" < \"%.s\""),
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	622 stp->st_orglen, sug.su_badptr);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	623 msg_puts((char *)IObuff);
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	626 if (p_verbose > 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	627 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	628 // Add the score.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	629 if (sps_flags & (SPS_DOUBLE \| SPS_BEST))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	630 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	631 stp->st_salscore ? "s " : "",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	632 stp->st_score, stp->st_altscore);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	633 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	634 vim_snprintf((char *)IObuff, IOSIZE, " (%d)",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	635 stp->st_score);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	636 #ifdef FEAT_RIGHTLEFT
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	637 if (cmdmsg_rl)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	638 // Mirror the numbers, but keep the leading space.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	639 rl_mirror(IObuff + 1);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	640 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	641 msg_advance(30);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	642 msg_puts((char *)IObuff);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	643 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	644 msg_putchar('\n');
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	647 #ifdef FEAT_RIGHTLEFT
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	648 cmdmsg_rl = FALSE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	649 msg_col = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	650 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	651 // Ask for choice.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	652 selected = prompt_for_number(&mouse_used);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	653 if (mouse_used)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	654 selected -= lines_left;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	655 lines_left = Rows; // avoid more prompt
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	656 // don't delay for 'smd' in normal_cmd()
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	657 msg_scroll = msg_scroll_save;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	660 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	661 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	662 // Save the from and to text for :spellrepall.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	663 VIM_CLEAR(repl_from);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	664 VIM_CLEAR(repl_to);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	665
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	666 stp = &SUG(sug.su_ga, selected - 1);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	667 if (sug.su_badlen > stp->st_orglen)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	668 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	669 // Replacing less than "su_badlen", append the remainder to
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	670 // repl_to.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	671 repl_from = vim_strnsave(sug.su_badptr, sug.su_badlen);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	672 vim_snprintf((char )IObuff, IOSIZE, "%s%.s", stp->st_word,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	673 sug.su_badlen - stp->st_orglen,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	674 sug.su_badptr + stp->st_orglen);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	675 repl_to = vim_strsave(IObuff);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	676 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	677 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	678 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	679 // Replacing su_badlen or more, use the whole word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	680 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	681 repl_to = vim_strsave(stp->st_word);
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	684 // Replace the word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	685 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	686 if (p != NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	687 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	688 int len_diff = stp->st_wordlen - stp->st_orglen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	689
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	690 c = (int)(sug.su_badptr - line);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	691 mch_memmove(p, line, c);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	692 STRCPY(p + c, stp->st_word);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	693 STRCAT(p, sug.su_badptr + stp->st_orglen);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	694
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	695 // For redo we use a change-word command.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	696 ResetRedobuff();
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	697 AppendToRedobuff((char_u *)"ciw");
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	698 AppendToRedobuffLit(p + c,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	699 stp->st_wordlen + sug.su_badlen - stp->st_orglen);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	700 AppendCharToRedobuff(ESC);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	701
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	702 // "p" may be freed here
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	703 ml_replace(curwin->w_cursor.lnum, p, FALSE);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	704 curwin->w_cursor.col = c;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	705
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	706 changed_bytes(curwin->w_cursor.lnum, c);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	707 #if defined(FEAT_PROP_POPUP)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	708 if (curbuf->b_has_textprop && len_diff != 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	709 adjust_prop_columns(curwin->w_cursor.lnum, c, len_diff,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	710 APC_SUBSTITUTE);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	711 #endif
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	714 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	715 curwin->w_cursor = prev_cursor;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	716
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	717 spell_find_cleanup(&sug);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	718 skip:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	719 vim_free(line);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	720 curwin->w_p_spell = wo_spell_save;
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	724 * 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	725 * a list of allocated strings.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	726 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	727 void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	728 spell_suggest_list(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	729 garray_T *gap,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	730 char_u *word,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	731 int maxcount, // maximum nr of suggestions
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	732 int need_cap, // 'spellcapcheck' matched
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	733 int interactive)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	734 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	735 suginfo_T sug;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	736 int i;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	737 suggest_T *stp;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	738 char_u *wcopy;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	739
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	740 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	741
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	742 // Make room in "gap".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	743 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	744 if (ga_grow(gap, sug.su_ga.ga_len) == OK)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	745 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	746 for (i = 0; i < sug.su_ga.ga_len; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	747 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	748 stp = &SUG(sug.su_ga, i);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	749
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	750 // 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	751 // replaced part.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	752 wcopy = alloc(stp->st_wordlen
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	753 + (unsigned)STRLEN(sug.su_badptr + stp->st_orglen) + 1);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	754 if (wcopy == NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	755 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	756 STRCPY(wcopy, stp->st_word);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	757 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	758 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	762 spell_find_cleanup(&sug);
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	766 * 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	767 * Return the suggestions in "su->su_ga".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	768 * The maximum number of suggestions is "maxcount".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	769 * Note: does use info for the current window.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	770 * 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	771 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	772 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	773 spell_find_suggest(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	774 char_u *badptr,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	775 int badlen, // length of bad word or 0 if unknown
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	776 suginfo_T *su,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	777 int maxcount,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	778 int banbadword, // don't include badword in suggestions
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	779 int need_cap, // word should start with capital
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	780 int interactive)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	781 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	782 hlf_T attr = HLF_COUNT;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	783 char_u buf[MAXPATHL];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	784 char_u *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	785 int do_combine = FALSE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	786 char_u *sps_copy;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	787 #ifdef FEAT_EVAL
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	788 static int expr_busy = FALSE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	789 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	790 int c;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	791 int i;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	792 langp_T *lp;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	793 int did_intern = FALSE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	794
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	795 // Set the info in "*su".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	796 CLEAR_POINTER(su);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	797 ga_init2(&su->su_ga, sizeof(suggest_T), 10);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	798 ga_init2(&su->su_sga, sizeof(suggest_T), 10);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	799 if (*badptr == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	800 return;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	801 hash_init(&su->su_banned);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	802
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	803 su->su_badptr = badptr;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	804 if (badlen != 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	805 su->su_badlen = badlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	806 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	807 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	808 su->su_maxcount = maxcount;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	809 su->su_maxscore = SCORE_MAXINIT;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	810
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	811 if (su->su_badlen >= MAXWLEN)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	812 su->su_badlen = MAXWLEN - 1; // just in case
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	813 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	814 (void)spell_casefold(curwin, su->su_badptr, su->su_badlen,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	815 su->su_fbadword, MAXWLEN);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	816 // 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	817 // text. Currently an illegal byte causes wrong pointer computations.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	818 su->su_fbadword[su->su_badlen] = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	819
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	820 // get caps flags for bad word
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	821 su->su_badflags = badword_captype(su->su_badptr,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	822 su->su_badptr + su->su_badlen);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	823 if (need_cap)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	824 su->su_badflags \|= WF_ONECAP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	825
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	826 // 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	827 // 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	828 // 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	829 // languages (e.g., "pl,en").
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	830 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	831 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	832 lp = LANGP_ENTRY(curbuf->b_s.b_langp, i);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	833 if (lp->lp_sallang != NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	834 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	835 su->su_sallang = lp->lp_sallang;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	836 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	840 // 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	841 // have to do this many times.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	842 if (su->su_sallang != NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	843 spell_soundfold(su->su_sallang, su->su_fbadword, TRUE,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	844 su->su_sal_badword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	845
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	846 // 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	847 // 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	848 // for that.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	849 c = PTR2CHAR(su->su_badptr);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	850 if (!SPELL_ISUPPER(c) && attr == HLF_COUNT)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	851 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	852 make_case_word(su->su_badword, buf, WF_ONECAP);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	853 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	854 0, TRUE, su->su_sallang, FALSE);
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	857 // 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	858 if (banbadword)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	859 add_banned(su, su->su_badword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	860
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	861 // 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	862 sps_copy = vim_strsave(p_sps);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	863 if (sps_copy == NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	864 return;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	865 spell_suggest_timeout = 5000;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	866
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	867 // Loop over the items in 'spellsuggest'.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	868 for (p = sps_copy; *p != NUL; )
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	869 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	870 copy_option_part(&p, buf, MAXPATHL, ",");
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	871
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	872 if (STRNCMP(buf, "expr:", 5) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	873 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	874 #ifdef FEAT_EVAL
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	875 // Evaluate an expression. Skip this when called recursively,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	876 // when using spellsuggest() in the expression.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	877 if (!expr_busy)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	878 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	879 expr_busy = TRUE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	880 spell_suggest_expr(su, buf + 5);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	881 expr_busy = FALSE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	882 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	883 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	884 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	885 else if (STRNCMP(buf, "file:", 5) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	886 // Use list of suggestions in a file.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	887 spell_suggest_file(su, buf + 5);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	888 else if (STRNCMP(buf, "timeout:", 8) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	889 // Limit the time searching for suggestions.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	890 spell_suggest_timeout = atol((char *)buf + 8);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	891 else if (!did_intern)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	892 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	893 // Use internal method once.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	894 spell_suggest_intern(su, interactive);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	895 if (sps_flags & SPS_DOUBLE)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	896 do_combine = TRUE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	897 did_intern = TRUE;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	901 vim_free(sps_copy);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	902
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	903 if (do_combine)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	904 // 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	905 // because sorting changes the order again.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	906 score_combine(su);
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	909 #ifdef FEAT_EVAL
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	910 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	911 * Find suggestions by evaluating expression "expr".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	912 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	913 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	914 spell_suggest_expr(suginfo_T su, char_u expr)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	915 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	916 list_T *list;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	917 listitem_T *li;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	918 int score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	919 char_u *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	920
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	921 // 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	922 // suginfo_T.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	923 // First evaluate the expression and get the resulting list.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	924 list = eval_spell_expr(su->su_badword, expr);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	925 if (list != NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	926 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	927 // Loop over the items in the list.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	928 FOR_ALL_LIST_ITEMS(list, li)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	929 if (li->li_tv.v_type == VAR_LIST)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	930 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	931 // Get the word and the score from the items.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	932 score = get_spellword(li->li_tv.vval.v_list, &p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	933 if (score >= 0 && score <= su->su_maxscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	934 add_suggestion(su, &su->su_ga, p, su->su_badlen,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	935 score, 0, TRUE, su->su_sallang, FALSE);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	936 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	937 list_unref(list);
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	940 // Remove bogus suggestions, sort and truncate at "maxcount".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	941 check_suggestions(su, &su->su_ga);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	942 (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	943 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	944 #endif
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	947 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	948 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	949 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	950 spell_suggest_file(suginfo_T su, char_u fname)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	951 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	952 FILE *fd;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	953 char_u line[MAXWLEN * 2];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	954 char_u *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	955 int len;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	956 char_u cword[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	957
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	958 // Open the file.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	959 fd = mch_fopen((char *)fname, "r");
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	960 if (fd == NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	961 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	962 semsg(_(e_cant_open_file_str), fname);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	963 return;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	966 // Read it line by line.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	967 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	968 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	969 line_breakcheck();
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	970
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	971 p = vim_strchr(line, '/');
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	972 if (p == NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	973 continue; // No Tab found, just skip the line.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	974 *p++ = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	975 if (STRICMP(su->su_badword, line) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	976 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	977 // Match! Isolate the good word, until CR or NL.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	978 for (len = 0; p[len] >= ' '; ++len)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	979 ;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	980 p[len] = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	981
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	982 // 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	983 // of the bad word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	984 if (captype(p, NULL) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	985 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	986 make_case_word(p, cword, su->su_badflags);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	987 p = cword;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	990 add_suggestion(su, &su->su_ga, p, su->su_badlen,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	991 SCORE_FILE, 0, TRUE, su->su_sallang, FALSE);
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	995 fclose(fd);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	996
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	997 // Remove bogus suggestions, sort and truncate at "maxcount".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	998 check_suggestions(su, &su->su_ga);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	999 (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	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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1003 * Find suggestions for the internal method indicated by "sps_flags".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1004 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1005 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1006 spell_suggest_intern(suginfo_T *su, int interactive)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1007 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1008 // 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	1009 suggest_load_files();
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1010
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1011 // 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	1012 //
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1013 // 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	1014 // tried.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1015 suggest_try_special(su);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1016
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1017 // 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	1018 // 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	1019 suggest_try_change(su);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1020
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1021 // 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	1022 if (sps_flags & SPS_DOUBLE)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1023 score_comp_sal(su);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1024
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1025 // 3. Try finding sound-a-like words.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1026 if ((sps_flags & SPS_FAST) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1027 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1028 if (sps_flags & SPS_BEST)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1029 // 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	1030 // they sound like.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1031 rescore_suggestions(su);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1032
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1033 // 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	1034 // 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	1035 // and "su_sfmaxscore" the rescored score, which is set by
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1036 // cleanup_suggestions().
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1037 // 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	1038 // 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	1039 // find many suggestions try again with a higher edit distance.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1040 // "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	1041 suggest_try_soundalike_prep();
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1042 su->su_maxscore = SCORE_SFMAX1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1043 su->su_sfmaxscore = SCORE_MAXINIT * 3;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1044 suggest_try_soundalike(su);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1045 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1046 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1047 // We didn't find enough matches, try again, allowing more
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1048 // changes to the soundfold word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1049 su->su_maxscore = SCORE_SFMAX2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1050 suggest_try_soundalike(su);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1051 if (su->su_ga.ga_len < SUG_CLEAN_COUNT(su))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1052 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1053 // Still didn't find enough matches, try again, allowing even
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1054 // more changes to the soundfold word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1055 su->su_maxscore = SCORE_SFMAX3;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1056 suggest_try_soundalike(su);
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1059 su->su_maxscore = su->su_sfmaxscore;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1060 suggest_try_soundalike_finish();
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1063 // 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	1064 // got_int when using a command, not for spellsuggest().
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1065 ui_breakcheck();
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1066 if (interactive && got_int)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1067 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1068 (void)vgetc();
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1069 got_int = FALSE;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1072 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	1073 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1074 if (sps_flags & SPS_BEST)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1075 // Adjust the word score for how it sounds like.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1076 rescore_suggestions(su);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1077
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1078 // Remove bogus suggestions, sort and truncate at "maxcount".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1079 check_suggestions(su, &su->su_ga);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1080 (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	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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1085 * Free the info put in "*su" by spell_find_suggest().
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1086 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1087 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1088 spell_find_cleanup(suginfo_T *su)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1089 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1090 int i;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1091
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1092 // Free the suggestions.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1093 for (i = 0; i < su->su_ga.ga_len; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1094 vim_free(SUG(su->su_ga, i).st_word);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1095 ga_clear(&su->su_ga);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1096 for (i = 0; i < su->su_sga.ga_len; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1097 vim_free(SUG(su->su_sga, i).st_word);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1098 ga_clear(&su->su_sga);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1099
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1100 // Free the banned words.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1101 hash_clear_all(&su->su_banned, 0);
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1105 * Try finding suggestions by recognizing specific situations.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1106 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1107 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1108 suggest_try_special(suginfo_T *su)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1109 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1110 char_u *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1111 size_t len;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1112 int c;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1113 char_u word[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1114
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1115 // Recognize a word that is repeated: "the the".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1116 p = skiptowhite(su->su_fbadword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1117 len = p - su->su_fbadword;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1118 p = skipwhite(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1119 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	1120 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1121 // Include badflags: if the badword is onecap or allcap
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1122 // use that for the goodword too: "The the" -> "The".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1123 c = su->su_fbadword[len];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1124 su->su_fbadword[len] = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1125 make_case_word(su->su_fbadword, word, su->su_badflags);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1126 su->su_fbadword[len] = c;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1127
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1128 // 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	1129 // character.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1130 add_suggestion(su, &su->su_ga, word, su->su_badlen,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1131 RESCORE(SCORE_REP, 0), 0, TRUE, su->su_sallang, FALSE);
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1136 * 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	1137 * Output is dumped in "suggestprof".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1138 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1139 #if 0
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1140 # define SUGGEST_PROFILE
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1141 proftime_T current;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1142 proftime_T total;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1143 proftime_T times[STATE_FINAL + 1];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1144 long counts[STATE_FINAL + 1];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1145
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1146 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1147 prof_init(void)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1148 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1149 for (int i = 0; i <= STATE_FINAL; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1150 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1151 profile_zero(&times[i]);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1152 counts[i] = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1153 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1154 profile_start(&current);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1155 profile_start(&total);
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1158 // call before changing state
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1159 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1160 prof_store(state_T state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1161 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1162 profile_end(&current);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1163 profile_add(&times[state], &current);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1164 ++counts[state];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1165 profile_start(&current);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1166 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1167 # define PROF_STORE(state) prof_store(state);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1168
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1169 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1170 prof_report(char *name)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1171 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1172 FILE *fd = fopen("suggestprof", "a");
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1173
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1174 profile_end(&total);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1175 fprintf(fd, "-----------------------\n");
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1176 fprintf(fd, "%s: %s\n", name, profile_msg(&total));
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1177 for (int i = 0; i <= STATE_FINAL; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1178 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	1179 fclose(fd);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1180 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1181 #else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1182 # define PROF_STORE(state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1183 #endif
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1186 * Try finding suggestions by adding/removing/swapping letters.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1187 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1188 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1189 suggest_try_change(suginfo_T *su)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1190 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1191 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	1192 int n;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1193 char_u *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1194 int lpi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1195 langp_T *lp;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1196
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1197 // 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	1198 // 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	1199 // chars after the bad word may help.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1200 STRCPY(fword, su->su_fbadword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1201 n = (int)STRLEN(fword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1202 p = su->su_badptr + su->su_badlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1203 (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	1204
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1205 // 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	1206 n = (int)STRLEN(su->su_badptr);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1207 if (n < MAXWLEN)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1208 fword[n] = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1209
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1210 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	1211 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1212 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1213
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1214 // 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	1215 // everything has been cleared.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1216 if (lp->lp_slang->sl_fbyts == NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1217 continue;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1218
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1219 // Try it for this language. Will add possible suggestions.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1220 #ifdef SUGGEST_PROFILE
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1221 prof_init();
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1222 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1223 suggest_trie_walk(su, lp, fword, FALSE);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1224 #ifdef SUGGEST_PROFILE
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1225 prof_report("try_change");
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1226 #endif
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1230 // 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	1231 #define TRY_DEEPER(su, stack, depth, add) \
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1232 ((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	1233
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1234 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1235 * Try finding suggestions by adding/removing/swapping letters.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1236 *
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1237 * 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	1238 * 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	1239 * 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	1240 * 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	1241 * limited by su->su_maxscore.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1242 *
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1243 * 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	1244 * describes something similar: "Error-tolerant Finite State Recognition with
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1245 * Applications to Morphological Analysis and Spelling Correction" (1996).
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1246 * 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	1247 * 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	1248 * the length of the word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1249 *
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1250 * 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	1251 * 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	1252 * 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	1253 * added, this is done by add_sound_suggest().
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1254 * Don't use:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1255 * the prefix tree or the keep-case tree
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1256 * "su->su_badlen"
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1257 * anything to do with upper and lower case
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1258 * 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	1259 * banned words
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1260 * word flags (rare, region, compounding)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1261 * word splitting for now
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1262 * "similar_chars()"
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1263 * 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	1264 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1265 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1266 suggest_trie_walk(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1267 suginfo_T *su,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1268 langp_T *lp,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1269 char_u *fword,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1270 int soundfold)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1271 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1272 char_u tword[MAXWLEN]; // good word collected so far
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1273 trystate_T stack[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1274 char_u preword[MAXWLEN * 3]; // word found with proper case;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1275 // concatenation of prefix compound
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1276 // words and split word. NUL terminated
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1277 // when going deeper but not when coming
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1278 // back.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1279 char_u compflags[MAXWLEN]; // compound flags, one for each word
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1280 trystate_T *sp;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1281 int newscore;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1282 int score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1283 char_u byts, fbyts, *pbyts;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1284 idx_T idxs, fidxs, *pidxs;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1285 int depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1286 int c, c2, c3;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1287 int n = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1288 int flags;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1289 garray_T *gap;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1290 idx_T arridx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1291 int len;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1292 char_u *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1293 fromto_T *ftp;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1294 int fl = 0, tl;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1295 int repextra = 0; // extra bytes in fword[] from REP item
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1296 slang_T *slang = lp->lp_slang;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1297 int fword_ends;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1298 int goodword_ends;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1299 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1300 // Stores the name of the change made at each level.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1301 char_u changename[MAXWLEN][80];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1302 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1303 int breakcheckcount = 1000;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1304 #ifdef FEAT_RELTIME
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1305 proftime_T time_limit;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1306 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1307 int compound_ok;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1308
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1309 // 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	1310 // "tword[]" contains the word collected from nodes in the tree.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1311 // "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	1312 // word).
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1313 depth = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1314 sp = &stack[0];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1315 CLEAR_POINTER(sp);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1316 sp->ts_curi = 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1317
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1318 if (soundfold)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1319 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1320 // Going through the soundfold tree.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1321 byts = fbyts = slang->sl_sbyts;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1322 idxs = fidxs = slang->sl_sidxs;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1323 pbyts = NULL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1324 pidxs = NULL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1325 sp->ts_prefixdepth = PFD_NOPREFIX;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1326 sp->ts_state = STATE_START;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1327 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1328 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1329 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1330 // 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	1331 // 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	1332 fbyts = slang->sl_fbyts;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1333 fidxs = slang->sl_fidxs;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1334 pbyts = slang->sl_pbyts;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1335 pidxs = slang->sl_pidxs;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1336 if (pbyts != NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1337 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1338 byts = pbyts;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1339 idxs = pidxs;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1340 sp->ts_prefixdepth = PFD_PREFIXTREE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1341 sp->ts_state = STATE_NOPREFIX; // try without prefix first
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1342 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1343 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1344 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1345 byts = fbyts;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1346 idxs = fidxs;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1347 sp->ts_prefixdepth = PFD_NOPREFIX;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1348 sp->ts_state = STATE_START;
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1351 #ifdef FEAT_RELTIME
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1352 // 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	1353 // time.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1354 if (spell_suggest_timeout > 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1355 profile_setlimit(spell_suggest_timeout, &time_limit);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1356 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1357
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1358 // Loop to find all suggestions. At each round we either:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1359 // - For the current state try one operation, advance "ts_curi",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1360 // increase "depth".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1361 // - 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	1362 // - When all states are tried decrease "depth".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1363 while (depth >= 0 && !got_int)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1364 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1365 sp = &stack[depth];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1366 switch (sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1367 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1368 case STATE_START:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1369 case STATE_NOPREFIX:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1370 // Start of node: Deal with NUL bytes, which means
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1371 // tword[] may end here.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1372 arridx = sp->ts_arridx; // current node in the tree
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1373 len = byts[arridx]; // bytes in this node
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1374 arridx += sp->ts_curi; // index of current byte
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1375
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1376 if (sp->ts_prefixdepth == PFD_PREFIXTREE)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1377 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1378 // Skip over the NUL bytes, we use them later.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1379 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1380 ;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1381 sp->ts_curi += n;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1382
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1383 // Always past NUL bytes now.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1384 n = (int)sp->ts_state;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1385 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1386 sp->ts_state = STATE_ENDNUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1387 sp->ts_save_badflags = su->su_badflags;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1388
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1389 // 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	1390 // following word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1391 if (depth < MAXWLEN - 1
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1392 && (byts[arridx] == 0 \|\| n == (int)STATE_NOPREFIX))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1393 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1394 // 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	1395 // Use the caps type until here for the prefix itself.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1396 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1397 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1398 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1399 n = sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1400 flags = badword_captype(su->su_badptr, su->su_badptr + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1401 su->su_badflags = badword_captype(su->su_badptr + n,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1402 su->su_badptr + su->su_badlen);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1403 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1404 sprintf(changename[depth], "prefix");
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1405 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1406 go_deeper(stack, depth, 0);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1407 ++depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1408 sp = &stack[depth];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1409 sp->ts_prefixdepth = depth - 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1410 byts = fbyts;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1411 idxs = fidxs;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1412 sp->ts_arridx = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1413
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1414 // Move the prefix to preword[] with the right case
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1415 // and make find_keepcap_word() works.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1416 tword[sp->ts_twordlen] = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1417 make_case_word(tword + sp->ts_splitoff,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1418 preword + sp->ts_prewordlen, flags);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1419 sp->ts_prewordlen = (char_u)STRLEN(preword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1420 sp->ts_splitoff = sp->ts_twordlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1421 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1422 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1425 if (sp->ts_curi > len \|\| byts[arridx] != 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1426 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1427 // Past bytes in node and/or past NUL bytes.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1428 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1429 sp->ts_state = STATE_ENDNUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1430 sp->ts_save_badflags = su->su_badflags;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1431 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1434 // End of word in tree.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1435 ++sp->ts_curi; // eat one NUL byte
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1436
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1437 flags = (int)idxs[arridx];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1438
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1439 // Skip words with the NOSUGGEST flag.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1440 if (flags & WF_NOSUGGEST)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1441 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1442
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1443 fword_ends = (fword[sp->ts_fidx] == NUL
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1444 \|\| (soundfold
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1445 ? VIM_ISWHITE(fword[sp->ts_fidx])
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1446 : !spell_iswordp(fword + sp->ts_fidx, curwin)));
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1447 tword[sp->ts_twordlen] = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1448
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1449 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1450 && (sp->ts_flags & TSF_PREFIXOK) == 0
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1451 && pbyts != NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1452 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1453 // 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	1454 // can be used with this word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1455 // 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	1456 // none this must be the first try without a prefix.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1457 n = stack[sp->ts_prefixdepth].ts_arridx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1458 len = pbyts[n++];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1459 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1460 ;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1461 if (c > 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1462 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1463 c = valid_word_prefix(c, n, flags,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1464 tword + sp->ts_splitoff, slang, FALSE);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1465 if (c == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1466 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1467
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1468 // Use the WF_RARE flag for a rare prefix.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1469 if (c & WF_RAREPFX)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1470 flags \|= WF_RARE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1471
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1472 // Tricky: when checking for both prefix and compounding
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1473 // we run into the prefix flag first.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1474 // 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	1475 // when arriving at a compound flag.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1476 sp->ts_flags \|= TSF_PREFIXOK;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1480 // Check NEEDCOMPOUND: can't use word without compounding. Do try
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1481 // appending another compound word below.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1482 if (sp->ts_complen == sp->ts_compsplit && fword_ends
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1483 && (flags & WF_NEEDCOMP))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1484 goodword_ends = FALSE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1485 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1486 goodword_ends = TRUE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1487
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1488 p = NULL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1489 compound_ok = TRUE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1490 if (sp->ts_complen > sp->ts_compsplit)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1491 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1492 if (slang->sl_nobreak)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1493 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1494 // 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	1495 // 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	1496 // as a suggestion. If this word was corrected too, we
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1497 // need to check if a correct word follows.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1498 if (sp->ts_fidx - sp->ts_splitfidx
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1499 == sp->ts_twordlen - sp->ts_splitoff
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1500 && STRNCMP(fword + sp->ts_splitfidx,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1501 tword + sp->ts_splitoff,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1502 sp->ts_fidx - sp->ts_splitfidx) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1503 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1504 preword[sp->ts_prewordlen] = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1505 newscore = score_wordcount_adj(slang, sp->ts_score,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1506 preword + sp->ts_prewordlen,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1507 sp->ts_prewordlen > 0);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1508 // Add the suggestion if the score isn't too bad.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1509 if (newscore <= su->su_maxscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1510 add_suggestion(su, &su->su_ga, preword,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1511 sp->ts_splitfidx - repextra,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1512 newscore, 0, FALSE,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1513 lp->lp_sallang, FALSE);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1514 break;
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1517 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1518 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1519 // There was a compound word before this word. If this
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1520 // word does not support compounding then give up
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1521 // (splitting is tried for the word without compound
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1522 // flag).
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1523 if (((unsigned)flags >> 24) == 0
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1524 \|\| sp->ts_twordlen - sp->ts_splitoff
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1525 < slang->sl_compminlen)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1526 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1527 // For multi-byte chars check character length against
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1528 // COMPOUNDMIN.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1529 if (has_mbyte
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1530 && slang->sl_compminlen > 0
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1531 && mb_charlen(tword + sp->ts_splitoff)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1532 < slang->sl_compminlen)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1533 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1534
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1535 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1536 compflags[sp->ts_complen + 1] = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1537 vim_strncpy(preword + sp->ts_prewordlen,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1538 tword + sp->ts_splitoff,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1539 sp->ts_twordlen - sp->ts_splitoff);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1540
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1541 // Verify CHECKCOMPOUNDPATTERN rules.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1542 if (match_checkcompoundpattern(preword, sp->ts_prewordlen,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1543 &slang->sl_comppat))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1544 compound_ok = FALSE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1545
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1546 if (compound_ok)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1547 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1548 p = preword;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1549 while (*skiptowhite(p) != NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1550 p = skipwhite(skiptowhite(p));
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1551 if (fword_ends && !can_compound(slang, p,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1552 compflags + sp->ts_compsplit))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1553 // Compound is not allowed. But it may still be
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1554 // possible if we add another (short) word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1555 compound_ok = FALSE;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1558 // Get pointer to last char of previous word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1559 p = preword + sp->ts_prewordlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1560 MB_PTR_BACK(preword, p);
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1564 // Form the word with proper case in preword.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1565 // If there is a word from a previous split, append.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1566 // 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	1567 if (soundfold)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1568 STRCPY(preword + sp->ts_prewordlen, tword + sp->ts_splitoff);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1569 else if (flags & WF_KEEPCAP)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1570 // Must find the word in the keep-case tree.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1571 find_keepcap_word(slang, tword + sp->ts_splitoff,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1572 preword + sp->ts_prewordlen);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1573 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1574 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1575 // Include badflags: If the badword is onecap or allcap
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1576 // 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	1577 // allcap and it's only one char long use onecap.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1578 c = su->su_badflags;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1579 if ((c & WF_ALLCAP)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1580 && su->su_badlen == (*mb_ptr2len)(su->su_badptr))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1581 c = WF_ONECAP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1582 c \|= flags;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1583
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1584 // 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	1585 // use Onecap.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1586 if (p != NULL && spell_iswordp_nmw(p, curwin))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1587 c &= ~WF_ONECAP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1588 make_case_word(tword + sp->ts_splitoff,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1589 preword + sp->ts_prewordlen, c);
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1592 if (!soundfold)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1593 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1594 // 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	1595 // word, thus remember it.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1596 if (flags & WF_BANNED)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1597 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1598 add_banned(su, preword + sp->ts_prewordlen);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1599 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1600 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1601 if ((sp->ts_complen == sp->ts_compsplit
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1602 && WAS_BANNED(su, preword + sp->ts_prewordlen))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1603 \|\| WAS_BANNED(su, preword))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1604 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1605 if (slang->sl_compprog == NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1606 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1607 // 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	1608 goodword_ends = FALSE;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1612 newscore = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1613 if (!soundfold) // soundfold words don't have flags
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1614 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1615 if ((flags & WF_REGION)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1616 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1617 newscore += SCORE_REGION;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1618 if (flags & WF_RARE)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1619 newscore += SCORE_RARE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1620
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1621 if (!spell_valid_case(su->su_badflags,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1622 captype(preword + sp->ts_prewordlen, NULL)))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1623 newscore += SCORE_ICASE;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1626 // TODO: how about splitting in the soundfold tree?
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1627 if (fword_ends
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1628 && goodword_ends
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1629 && sp->ts_fidx >= sp->ts_fidxtry
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1630 && compound_ok)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1631 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1632 // The badword also ends: add suggestions.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1633 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1634 if (soundfold && STRCMP(preword, "smwrd") == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1635 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1636 int j;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1637
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1638 // print the stack of changes that brought us here
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1639 smsg("------ %s -------", fword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1640 for (j = 0; j < depth; ++j)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1641 smsg("%s", changename[j]);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1642 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1643 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1644 if (soundfold)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1645 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1646 // For soundfolded words we need to find the original
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1647 // words, the edit distance and then add them.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1648 add_sound_suggest(su, preword, sp->ts_score, lp);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1649 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1650 else if (sp->ts_fidx > 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1651 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1652 // Give a penalty when changing non-word char to word
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1653 // char, e.g., "thes," -> "these".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1654 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1655 MB_PTR_BACK(fword, p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1656 if (!spell_iswordp(p, curwin) && *preword != NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1657 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1658 p = preword + STRLEN(preword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1659 MB_PTR_BACK(preword, p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1660 if (spell_iswordp(p, curwin))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1661 newscore += SCORE_NONWORD;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1664 // Give a bonus to words seen before.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1665 score = score_wordcount_adj(slang,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1666 sp->ts_score + newscore,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1667 preword + sp->ts_prewordlen,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1668 sp->ts_prewordlen > 0);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1669
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1670 // Add the suggestion if the score isn't too bad.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1671 if (score <= su->su_maxscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1672 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1673 add_suggestion(su, &su->su_ga, preword,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1674 sp->ts_fidx - repextra,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1675 score, 0, FALSE, lp->lp_sallang, FALSE);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1676
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1677 if (su->su_badflags & WF_MIXCAP)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1678 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1679 // We really don't know if the word should be
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1680 // upper or lower case, add both.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1681 c = captype(preword, NULL);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1682 if (c == 0 \|\| c == WF_ALLCAP)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1683 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1684 make_case_word(tword + sp->ts_splitoff,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1685 preword + sp->ts_prewordlen,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1686 c == 0 ? WF_ALLCAP : 0);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1687
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1688 add_suggestion(su, &su->su_ga, preword,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1689 sp->ts_fidx - repextra,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1690 score + SCORE_ICASE, 0, FALSE,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1691 lp->lp_sallang, FALSE);
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1698 // Try word split and/or compounding.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1699 if ((sp->ts_fidx >= sp->ts_fidxtry \|\| fword_ends)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1700 // Don't split halfway a character.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1701 && (!has_mbyte \|\| sp->ts_tcharlen == 0))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1702 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1703 int try_compound;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1704 int try_split;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1705
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1706 // 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	1707 // Otherwise try changing the next word. E.g., find
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1708 // suggestions for "the the" where the second "the" is
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1709 // different. It's done like a split.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1710 // TODO: word split for soundfold words
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1711 try_split = (sp->ts_fidx - repextra < su->su_badlen)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1712 && !soundfold;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1713
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1714 // Get here in several situations:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1715 // 1. The word in the tree ends:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1716 // If the word allows compounding try that. Otherwise try
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1717 // 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	1718 // valid words starts at fword[sp->ts_fidx].
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1719 // 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	1720 // the next word is valid.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1721 // 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	1722 // 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	1723 // following word is valid.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1724 // 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	1725 // be possible to compound another (short) word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1726 try_compound = FALSE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1727 if (!soundfold
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1728 && !slang->sl_nocompoundsugs
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1729 && slang->sl_compprog != NULL
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1730 && ((unsigned)flags >> 24) != 0
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1731 && sp->ts_twordlen - sp->ts_splitoff
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1732 >= slang->sl_compminlen
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1733 && (!has_mbyte
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1734 \|\| slang->sl_compminlen == 0
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1735 \|\| mb_charlen(tword + sp->ts_splitoff)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1736 >= slang->sl_compminlen)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1737 && (slang->sl_compsylmax < MAXWLEN
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1738 \|\| sp->ts_complen + 1 - sp->ts_compsplit
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1739 < slang->sl_compmax)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1740 && (can_be_compound(sp, slang,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1741 compflags, ((unsigned)flags >> 24))))
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 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1744 try_compound = TRUE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1745 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1746 compflags[sp->ts_complen + 1] = NUL;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1749 // 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	1750 // valid.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1751 if (slang->sl_nobreak && !slang->sl_nocompoundsugs)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1752 try_compound = TRUE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1753
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1754 // 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	1755 // split at this point, do the split first and set
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1756 // TSF_DIDSPLIT to avoid doing it again.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1757 else if (!fword_ends
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1758 && try_compound
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1759 && (sp->ts_flags & TSF_DIDSPLIT) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1760 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1761 try_compound = FALSE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1762 sp->ts_flags \|= TSF_DIDSPLIT;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1763 --sp->ts_curi; // do the same NUL again
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1764 compflags[sp->ts_complen] = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1765 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1766 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1767 sp->ts_flags &= ~TSF_DIDSPLIT;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1768
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1769 if (try_split \|\| try_compound)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1770 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1771 if (!try_compound && (!fword_ends \|\| !goodword_ends))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1772 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1773 // 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	1774 // words so far are valid for compounding. If there
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1775 // is only one word it must not have the NEEDCOMPOUND
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1776 // flag.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1777 if (sp->ts_complen == sp->ts_compsplit
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1778 && (flags & WF_NEEDCOMP))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1779 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1780 p = preword;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1781 while (*skiptowhite(p) != NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1782 p = skipwhite(skiptowhite(p));
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1783 if (sp->ts_complen > sp->ts_compsplit
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1784 && !can_compound(slang, p,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1785 compflags + sp->ts_compsplit))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1786 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1787
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1788 if (slang->sl_nosplitsugs)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1789 newscore += SCORE_SPLIT_NO;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1790 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1791 newscore += SCORE_SPLIT;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1792
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1793 // Give a bonus to words seen before.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1794 newscore = score_wordcount_adj(slang, newscore,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1795 preword + sp->ts_prewordlen, TRUE);
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1798 if (TRY_DEEPER(su, stack, depth, newscore))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1799 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1800 go_deeper(stack, depth, newscore);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1801 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1802 if (!try_compound && !fword_ends)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1803 sprintf(changename[depth], "%.*s-%s: split",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1804 sp->ts_twordlen, tword, fword + sp->ts_fidx);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1805 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1806 sprintf(changename[depth], "%.*s-%s: compound",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1807 sp->ts_twordlen, tword, fword + sp->ts_fidx);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1808 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1809 // Save things to be restored at STATE_SPLITUNDO.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1810 sp->ts_save_badflags = su->su_badflags;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1811 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1812 sp->ts_state = STATE_SPLITUNDO;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1813
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1814 ++depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1815 sp = &stack[depth];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1816
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1817 // Append a space to preword when splitting.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1818 if (!try_compound && !fword_ends)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1819 STRCAT(preword, " ");
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1820 sp->ts_prewordlen = (char_u)STRLEN(preword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1821 sp->ts_splitoff = sp->ts_twordlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1822 sp->ts_splitfidx = sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1823
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1824 // If the badword has a non-word character at this
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1825 // position skip it. That means replacing the
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1826 // non-word character with a space. Always skip a
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1827 // character when the word ends. But only when the
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1828 // good word can end.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1829 if (((!try_compound && !spell_iswordp_nmw(fword
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1830 + sp->ts_fidx,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1831 curwin))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1832 \|\| fword_ends)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1833 && fword[sp->ts_fidx] != NUL
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1834 && goodword_ends)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1835 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1836 int l;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1837
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1838 l = mb_ptr2len(fword + sp->ts_fidx);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1839 if (fword_ends)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1840 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1841 // Copy the skipped character to preword.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1842 mch_memmove(preword + sp->ts_prewordlen,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1843 fword + sp->ts_fidx, l);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1844 sp->ts_prewordlen += l;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1845 preword[sp->ts_prewordlen] = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1846 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1847 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1848 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1849 sp->ts_fidx += l;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1852 // When compounding include compound flag in
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1853 // compflags[] (already set above). When splitting we
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1854 // may start compounding over again.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1855 if (try_compound)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1856 ++sp->ts_complen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1857 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1858 sp->ts_compsplit = sp->ts_complen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1859 sp->ts_prefixdepth = PFD_NOPREFIX;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1860
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1861 // set su->su_badflags to the caps type at this
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1862 // position
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1863 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1864 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1865 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1866 n = sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1867 su->su_badflags = badword_captype(su->su_badptr + n,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1868 su->su_badptr + su->su_badlen);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1869
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1870 // Restart at top of the tree.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1871 sp->ts_arridx = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1872
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1873 // If there are postponed prefixes, try these too.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1874 if (pbyts != NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1875 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1876 byts = pbyts;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1877 idxs = pidxs;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1878 sp->ts_prefixdepth = PFD_PREFIXTREE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1879 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1880 sp->ts_state = STATE_NOPREFIX;
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1885 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1886
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1887 case STATE_SPLITUNDO:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1888 // Undo the changes done for word split or compound word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1889 su->su_badflags = sp->ts_save_badflags;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1890
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1891 // Continue looking for NUL bytes.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1892 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1893 sp->ts_state = STATE_START;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1894
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1895 // In case we went into the prefix tree.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1896 byts = fbyts;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1897 idxs = fidxs;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1898 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1899
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1900 case STATE_ENDNUL:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1901 // Past the NUL bytes in the node.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1902 su->su_badflags = sp->ts_save_badflags;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1903 if (fword[sp->ts_fidx] == NUL && sp->ts_tcharlen == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1904 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1905 // The badword ends, can't use STATE_PLAIN.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1906 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1907 sp->ts_state = STATE_DEL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1908 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1909 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1910 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1911 sp->ts_state = STATE_PLAIN;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1912 // FALLTHROUGH
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1913
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1914 case STATE_PLAIN:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1915 // 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	1916 // and use child node. "ts_curi" is the index.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1917 arridx = sp->ts_arridx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1918 if (sp->ts_curi > byts[arridx])
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1919 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1920 // 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	1921 // already changed bytes skip the other tricks.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1922 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1923 if (sp->ts_fidx >= sp->ts_fidxtry)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1924 sp->ts_state = STATE_DEL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1925 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1926 sp->ts_state = STATE_FINAL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1927 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1928 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1929 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1930 arridx += sp->ts_curi++;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1931 c = byts[arridx];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1932
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1933 // 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	1934 // 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	1935 // 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	1936 // just deleted this byte, accepting it is always cheaper than
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1937 // delete + substitute.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1938 if (c == fword[sp->ts_fidx]
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1939 \|\| (sp->ts_tcharlen > 0 && sp->ts_isdiff != DIFF_NONE))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1940 newscore = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1941 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1942 newscore = SCORE_SUBST;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1943 if ((newscore == 0
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1944 \|\| (sp->ts_fidx >= sp->ts_fidxtry
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1945 && ((sp->ts_flags & TSF_DIDDEL) == 0
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1946 \|\| c != fword[sp->ts_delidx])))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1947 && TRY_DEEPER(su, stack, depth, newscore))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1948 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1949 go_deeper(stack, depth, newscore);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1950 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1951 if (newscore > 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1952 sprintf(changename[depth], "%.*s-%s: subst %c to %c",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1953 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1954 fword[sp->ts_fidx], c);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1955 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1956 sprintf(changename[depth], "%.*s-%s: accept %c",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1957 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1958 fword[sp->ts_fidx]);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1959 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1960 ++depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1961 sp = &stack[depth];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1962 if (fword[sp->ts_fidx] != NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1963 ++sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1964 tword[sp->ts_twordlen++] = c;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1965 sp->ts_arridx = idxs[arridx];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1966 if (newscore == SCORE_SUBST)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1967 sp->ts_isdiff = DIFF_YES;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1968 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1969 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1970 // Multi-byte characters are a bit complicated to
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1971 // handle: They differ when any of the bytes differ
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1972 // and then their length may also differ.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1973 if (sp->ts_tcharlen == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1974 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1975 // First byte.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1976 sp->ts_tcharidx = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1977 sp->ts_tcharlen = MB_BYTE2LEN(c);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1978 sp->ts_fcharstart = sp->ts_fidx - 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1979 sp->ts_isdiff = (newscore != 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1980 ? DIFF_YES : DIFF_NONE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1981 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1982 else if (sp->ts_isdiff == DIFF_INSERT
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1983 && sp->ts_fidx > 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1984 // When inserting trail bytes don't advance in the
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1985 // bad word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1986 --sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1987 if (++sp->ts_tcharidx == sp->ts_tcharlen)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1988 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1989 // Last byte of character.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1990 if (sp->ts_isdiff == DIFF_YES)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1991 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1992 // Correct ts_fidx for the byte length of the
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1993 // character (we didn't check that before).
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1994 sp->ts_fidx = sp->ts_fcharstart
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1995 + mb_ptr2len(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1996 fword + sp->ts_fcharstart);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1997 // For changing a composing character adjust
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1998 // the score from SCORE_SUBST to
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	1999 // SCORE_SUBCOMP.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2000 if (enc_utf8
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2001 && utf_iscomposing(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2002 utf_ptr2char(tword
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2003 + sp->ts_twordlen
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2004 - sp->ts_tcharlen))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2005 && utf_iscomposing(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2006 utf_ptr2char(fword
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2007 + sp->ts_fcharstart)))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2008 sp->ts_score -=
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2009 SCORE_SUBST - SCORE_SUBCOMP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2010
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2011 // For a similar character adjust score from
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2012 // SCORE_SUBST to SCORE_SIMILAR.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2013 else if (!soundfold
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2014 && slang->sl_has_map
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2015 && similar_chars(slang,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2016 mb_ptr2char(tword
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2017 + sp->ts_twordlen
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2018 - sp->ts_tcharlen),
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2019 mb_ptr2char(fword
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2020 + sp->ts_fcharstart)))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2021 sp->ts_score -=
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2022 SCORE_SUBST - SCORE_SIMILAR;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2023 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2024 else if (sp->ts_isdiff == DIFF_INSERT
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2025 && sp->ts_twordlen > sp->ts_tcharlen)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2026 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2027 p = tword + sp->ts_twordlen - sp->ts_tcharlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2028 c = mb_ptr2char(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2029 if (enc_utf8 && utf_iscomposing(c))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2030 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2031 // Inserting a composing char doesn't
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2032 // count that much.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2033 sp->ts_score -= SCORE_INS - SCORE_INSCOMP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2034 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2035 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2036 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2037 // If the previous character was the same,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2038 // thus doubling a character, give a bonus
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2039 // to the score. Also for the soundfold
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2040 // tree (might seem illogical but does
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2041 // give better scores).
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2042 MB_PTR_BACK(tword, p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2043 if (c == mb_ptr2char(p))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2044 sp->ts_score -= SCORE_INS
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2045 - SCORE_INSDUP;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2049 // Starting a new char, reset the length.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2050 sp->ts_tcharlen = 0;
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2053 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2054 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2055 // If we found a similar char adjust the score.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2056 // We do this after calling go_deeper() because
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2057 // it's slow.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2058 if (newscore != 0
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2059 && !soundfold
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2060 && slang->sl_has_map
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2061 && similar_chars(slang,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2062 c, fword[sp->ts_fidx - 1]))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2063 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2067 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2068
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2069 case STATE_DEL:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2070 // 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	2071 // delete/insert/swap a character.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2072 if (has_mbyte && sp->ts_tcharlen > 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2073 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2074 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2075 sp->ts_state = STATE_FINAL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2076 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2077 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2078 // Try skipping one character in the bad word (delete it).
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2079 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2080 sp->ts_state = STATE_INS_PREP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2081 sp->ts_curi = 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2082 if (soundfold && sp->ts_fidx == 0 && fword[sp->ts_fidx] == '*')
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2083 // 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	2084 // soundalike_score().
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2085 newscore = 2 * SCORE_DEL / 3;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2086 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2087 newscore = SCORE_DEL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2088 if (fword[sp->ts_fidx] != NUL
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2089 && TRY_DEEPER(su, stack, depth, newscore))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2090 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2091 go_deeper(stack, depth, newscore);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2092 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2093 sprintf(changename[depth], "%.*s-%s: delete %c",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2094 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2095 fword[sp->ts_fidx]);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2096 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2097 ++depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2098
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2099 // Remember what character we deleted, so that we can avoid
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2100 // inserting it again.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2101 stack[depth].ts_flags \|= TSF_DIDDEL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2102 stack[depth].ts_delidx = sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2103
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2104 // 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	2105 // 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	2106 // 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	2107 // results.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2108 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2109 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2110 c = mb_ptr2char(fword + sp->ts_fidx);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2111 stack[depth].ts_fidx += mb_ptr2len(fword + sp->ts_fidx);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2112 if (enc_utf8 && utf_iscomposing(c))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2113 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2114 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2115 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2116 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2117 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2118 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2119 ++stack[depth].ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2120 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2121 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2122 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2123 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2124 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2125 // FALLTHROUGH
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2126
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2127 case STATE_INS_PREP:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2128 if (sp->ts_flags & TSF_DIDDEL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2129 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2130 // 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	2131 // a substitute is always cheaper.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2132 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2133 sp->ts_state = STATE_SWAP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2134 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2137 // skip over NUL bytes
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2138 n = sp->ts_arridx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2139 for (;;)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2140 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2141 if (sp->ts_curi > byts[n])
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2142 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2143 // Only NUL bytes at this node, go to next state.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2144 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2145 sp->ts_state = STATE_SWAP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2146 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2147 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2148 if (byts[n + sp->ts_curi] != NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2149 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2150 // Found a byte to insert.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2151 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2152 sp->ts_state = STATE_INS;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2153 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2154 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2155 ++sp->ts_curi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2156 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2157 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2158
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2159 // FALLTHROUGH
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2160
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2161 case STATE_INS:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2162 // 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	2163 // node.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2164 n = sp->ts_arridx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2165 if (sp->ts_curi > byts[n])
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2166 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2167 // Done all bytes at this node, go to next state.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2168 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2169 sp->ts_state = STATE_SWAP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2170 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2173 // Do one more byte at this node, but:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2174 // - Skip NUL bytes.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2175 // - 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	2176 // accepting that byte is always better.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2177 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	2178
3b8089d550eb patch 9.0.2141: [security]: buffer-overflow in suggest_trie_walk Christian Brabandt <cb@256bit.org> parents: 33399 diff changeset	2179 // 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	2180 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	2181 {
3b8089d550eb patch 9.0.2141: [security]: buffer-overflow in suggest_trie_walk Christian Brabandt <cb@256bit.org> parents: 33399 diff changeset	2182 got_int = TRUE;
3b8089d550eb patch 9.0.2141: [security]: buffer-overflow in suggest_trie_walk Christian Brabandt <cb@256bit.org> parents: 33399 diff changeset	2183 break;
3b8089d550eb patch 9.0.2141: [security]: buffer-overflow in suggest_trie_walk Christian Brabandt <cb@256bit.org> parents: 33399 diff changeset	2184 }
32670 695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2185 c = byts[n];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2186 if (soundfold && sp->ts_twordlen == 0 && c == '*')
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2187 // 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	2188 // see soundalike_score().
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2189 newscore = 2 * SCORE_INS / 3;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2190 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2191 newscore = SCORE_INS;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2192 if (c != fword[sp->ts_fidx]
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2193 && TRY_DEEPER(su, stack, depth, newscore))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2194 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2195 go_deeper(stack, depth, newscore);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2196 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2197 sprintf(changename[depth], "%.*s-%s: insert %c",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2198 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2199 c);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2200 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2201 ++depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2202 sp = &stack[depth];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2203 tword[sp->ts_twordlen++] = c;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2204 sp->ts_arridx = idxs[n];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2205 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2206 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2207 fl = MB_BYTE2LEN(c);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2208 if (fl > 1)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2209 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2210 // There are following bytes for the same character.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2211 // We must find all bytes before trying
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2212 // delete/insert/swap/etc.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2213 sp->ts_tcharlen = fl;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2214 sp->ts_tcharidx = 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2215 sp->ts_isdiff = DIFF_INSERT;
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2218 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2219 fl = 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2220 if (fl == 1)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2221 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2222 // If the previous character was the same, thus doubling a
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2223 // character, give a bonus to the score. Also for
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2224 // soundfold words (illogical but does give a better
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2225 // score).
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2226 if (sp->ts_twordlen >= 2
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2227 && tword[sp->ts_twordlen - 2] == c)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2228 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2231 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2232
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2233 case STATE_SWAP:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2234 // Swap two bytes in the bad word: "12" -> "21".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2235 // We change "fword" here, it's changed back afterwards at
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2236 // STATE_UNSWAP.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2237 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2238 c = *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2239 if (c == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2240 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2241 // End of word, can't swap or replace.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2242 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2243 sp->ts_state = STATE_FINAL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2244 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2247 // 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	2248 // SWAP3 etc. also don't make sense then.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2249 if (!soundfold && !spell_iswordp(p, curwin))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2250 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2251 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2252 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2253 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2256 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2257 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2258 n = MB_CPTR2LEN(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2259 c = mb_ptr2char(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2260 if (p[n] == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2261 c2 = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2262 else if (!soundfold && !spell_iswordp(p + n, curwin))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2263 c2 = c; // don't swap non-word char
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2264 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2265 c2 = mb_ptr2char(p + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2266 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2267 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2268 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2269 if (p[1] == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2270 c2 = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2271 else if (!soundfold && !spell_iswordp(p + 1, curwin))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2272 c2 = c; // don't swap non-word char
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2273 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2274 c2 = p[1];
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2277 // When the second character is NUL we can't swap.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2278 if (c2 == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2279 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2280 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2281 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2282 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2285 // When characters are identical, swap won't do anything.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2286 // 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	2287 if (c == c2)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2288 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2289 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2290 sp->ts_state = STATE_SWAP3;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2291 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2292 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2293 if (c2 != NUL && TRY_DEEPER(su, stack, depth, SCORE_SWAP))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2294 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2295 go_deeper(stack, depth, SCORE_SWAP);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2296 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2297 sprintf(changename[depth], "%.*s-%s: swap %c and %c",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2298 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2299 c, c2);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2300 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2301 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2302 sp->ts_state = STATE_UNSWAP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2303 ++depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2304 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2305 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2306 fl = mb_char2len(c2);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2307 mch_memmove(p, p + n, fl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2308 mb_char2bytes(c, p + fl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2309 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2310 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2311 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2312 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2313 p[0] = c2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2314 p[1] = c;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2315 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2318 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2319 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2320 // 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	2321 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2322 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2323 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2324 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2325
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2326 case STATE_UNSWAP:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2327 // Undo the STATE_SWAP swap: "21" -> "12".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2328 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2329 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2330 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2331 n = mb_ptr2len(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2332 c = mb_ptr2char(p + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2333 mch_memmove(p + mb_ptr2len(p + n), p, n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2334 mb_char2bytes(c, p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2335 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2336 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2337 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2338 c = *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2339 *p = p[1];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2340 p[1] = c;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2341 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2342 // FALLTHROUGH
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2343
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2344 case STATE_SWAP3:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2345 // Swap two bytes, skipping one: "123" -> "321". We change
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2346 // "fword" here, it's changed back afterwards at STATE_UNSWAP3.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2347 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2348 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2349 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2350 n = MB_CPTR2LEN(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2351 c = mb_ptr2char(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2352 fl = MB_CPTR2LEN(p + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2353 c2 = mb_ptr2char(p + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2354 if (!soundfold && !spell_iswordp(p + n + fl, curwin))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2355 c3 = c; // don't swap non-word char
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2356 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2357 c3 = mb_ptr2char(p + n + fl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2358 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2359 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2360 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2361 c = *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2362 c2 = p[1];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2363 if (!soundfold && !spell_iswordp(p + 2, curwin))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2364 c3 = c; // don't swap non-word char
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2365 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2366 c3 = p[2];
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2369 // When characters are identical: "121" then SWAP3 result is
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2370 // 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	2371 // 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	2372 // Also skip when c3 is NUL.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2373 // 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	2374 // Second character may any char: "a.b" -> "b.a"
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2375 if (c == c3 \|\| c3 == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2376 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2377 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2378 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2379 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2380 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2381 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2382 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2383 go_deeper(stack, depth, SCORE_SWAP3);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2384 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2385 sprintf(changename[depth], "%.*s-%s: swap3 %c and %c",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2386 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2387 c, c3);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2388 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2389 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2390 sp->ts_state = STATE_UNSWAP3;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2391 ++depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2392 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2393 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2394 tl = mb_char2len(c3);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2395 mch_memmove(p, p + n + fl, tl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2396 mb_char2bytes(c2, p + tl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2397 mb_char2bytes(c, p + fl + tl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2398 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2399 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2400 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2401 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2402 p[0] = p[2];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2403 p[2] = c;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2404 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2407 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2408 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2409 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2410 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2411 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2412 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2413
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2414 case STATE_UNSWAP3:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2415 // Undo STATE_SWAP3: "321" -> "123"
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2416 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2417 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2418 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2419 n = mb_ptr2len(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2420 c2 = mb_ptr2char(p + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2421 fl = mb_ptr2len(p + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2422 c = mb_ptr2char(p + n + fl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2423 tl = mb_ptr2len(p + n + fl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2424 mch_memmove(p + fl + tl, p, n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2425 mb_char2bytes(c, p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2426 mb_char2bytes(c2, p + tl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2427 p = p + tl;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2428 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2429 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2430 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2431 c = *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2432 *p = p[2];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2433 p[2] = c;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2434 ++p;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2437 if (!soundfold && !spell_iswordp(p, curwin))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2438 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2439 // 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	2440 // third char were already checked at swap and swap3.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2441 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2442 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2443 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2446 // Rotate three characters left: "123" -> "231". We change
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2447 // "fword" here, it's changed back afterwards at STATE_UNROT3L.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2448 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2449 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2450 go_deeper(stack, depth, SCORE_SWAP3);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2451 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2452 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2453 sprintf(changename[depth], "%.*s-%s: rotate left %c%c%c",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2454 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2455 p[0], p[1], p[2]);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2456 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2457 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2458 sp->ts_state = STATE_UNROT3L;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2459 ++depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2460 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2461 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2462 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2463 n = MB_CPTR2LEN(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2464 c = mb_ptr2char(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2465 fl = MB_CPTR2LEN(p + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2466 fl += MB_CPTR2LEN(p + n + fl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2467 mch_memmove(p, p + n, fl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2468 mb_char2bytes(c, p + fl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2469 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2470 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2471 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2472 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2473 c = *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2474 *p = p[1];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2475 p[1] = p[2];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2476 p[2] = c;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2477 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2480 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2481 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2482 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2483 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2484 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2485 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2486
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2487 case STATE_UNROT3L:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2488 // Undo ROT3L: "231" -> "123"
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2489 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2490 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2491 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2492 n = mb_ptr2len(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2493 n += mb_ptr2len(p + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2494 c = mb_ptr2char(p + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2495 tl = mb_ptr2len(p + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2496 mch_memmove(p + tl, p, n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2497 mb_char2bytes(c, p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2498 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2499 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2500 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2501 c = p[2];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2502 p[2] = p[1];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2503 p[1] = *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2504 *p = c;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2507 // Rotate three bytes right: "123" -> "312". We change "fword"
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2508 // here, it's changed back afterwards at STATE_UNROT3R.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2509 if (TRY_DEEPER(su, stack, depth, SCORE_SWAP3))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2510 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2511 go_deeper(stack, depth, SCORE_SWAP3);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2512 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2513 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2514 sprintf(changename[depth], "%.*s-%s: rotate right %c%c%c",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2515 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2516 p[0], p[1], p[2]);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2517 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2518 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2519 sp->ts_state = STATE_UNROT3R;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2520 ++depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2521 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2522 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2523 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2524 n = MB_CPTR2LEN(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2525 n += MB_CPTR2LEN(p + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2526 c = mb_ptr2char(p + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2527 tl = MB_CPTR2LEN(p + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2528 mch_memmove(p + tl, p, n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2529 mb_char2bytes(c, p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2530 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2531 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2532 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2533 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2534 c = p[2];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2535 p[2] = p[1];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2536 p[1] = *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2537 *p = c;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2538 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2541 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2542 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2543 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2544 sp->ts_state = STATE_REP_INI;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2545 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2546 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2547
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2548 case STATE_UNROT3R:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2549 // Undo ROT3R: "312" -> "123"
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2550 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2551 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2552 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2553 c = mb_ptr2char(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2554 tl = mb_ptr2len(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2555 n = mb_ptr2len(p + tl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2556 n += mb_ptr2len(p + tl + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2557 mch_memmove(p, p + tl, n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2558 mb_char2bytes(c, p + n);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2559 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2560 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2561 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2562 c = *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2563 *p = p[1];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2564 p[1] = p[2];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2565 p[2] = c;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2566 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2567 // FALLTHROUGH
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2568
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2569 case STATE_REP_INI:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2570 // 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	2571 // Quickly skip if:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2572 // - 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	2573 // - the score is going to be too high anyway
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2574 // - already applied a REP item or swapped here
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2575 if ((lp->lp_replang == NULL && !soundfold)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2576 \|\| sp->ts_score + SCORE_REP >= su->su_maxscore
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2577 \|\| sp->ts_fidx < sp->ts_fidxtry)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2578 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2579 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2580 sp->ts_state = STATE_FINAL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2581 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2584 // 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	2585 // match. If the index is -1 there is none.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2586 if (soundfold)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2587 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	2588 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2589 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	2590
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2591 if (sp->ts_curi < 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2592 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2593 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2594 sp->ts_state = STATE_FINAL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2595 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2598 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2599 sp->ts_state = STATE_REP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2600 // FALLTHROUGH
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2601
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2602 case STATE_REP:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2603 // 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	2604 // replace the characters and check if the resulting word is
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2605 // valid.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2606 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2607
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2608 if (soundfold)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2609 gap = &slang->sl_repsal;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2610 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2611 gap = &lp->lp_replang->sl_rep;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2612 while (sp->ts_curi < gap->ga_len)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2613 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2614 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2615 if (ftp->ft_from != p)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2616 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2617 // past possible matching entries
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2618 sp->ts_curi = gap->ga_len;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2619 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2620 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2621 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	2622 && TRY_DEEPER(su, stack, depth, SCORE_REP))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2623 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2624 go_deeper(stack, depth, SCORE_REP);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2625 #ifdef DEBUG_TRIEWALK
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2626 sprintf(changename[depth], "%.*s-%s: replace %s with %s",
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2627 sp->ts_twordlen, tword, fword + sp->ts_fidx,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2628 ftp->ft_from, ftp->ft_to);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2629 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2630 // Need to undo this afterwards.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2631 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2632 sp->ts_state = STATE_REP_UNDO;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2633
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2634 // Change the "from" to the "to" string.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2635 ++depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2636 fl = (int)STRLEN(ftp->ft_from);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2637 tl = (int)STRLEN(ftp->ft_to);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2638 if (fl != tl)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2639 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2640 STRMOVE(p + tl, p + fl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2641 repextra += tl - fl;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2642 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2643 mch_memmove(p, ftp->ft_to, tl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2644 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2645 stack[depth].ts_tcharlen = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2646 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2650 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	2651 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2652 // No (more) matches.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2653 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2654 sp->ts_state = STATE_FINAL;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2657 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2658
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2659 case STATE_REP_UNDO:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2660 // Undo a REP replacement and continue with the next one.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2661 if (soundfold)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2662 gap = &slang->sl_repsal;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2663 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2664 gap = &lp->lp_replang->sl_rep;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2665 ftp = (fromto_T *)gap->ga_data + sp->ts_curi - 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2666 fl = (int)STRLEN(ftp->ft_from);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2667 tl = (int)STRLEN(ftp->ft_to);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2668 p = fword + sp->ts_fidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2669 if (fl != tl)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2670 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2671 STRMOVE(p + fl, p + tl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2672 repextra -= tl - fl;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2673 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2674 mch_memmove(p, ftp->ft_from, fl);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2675 PROF_STORE(sp->ts_state)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2676 sp->ts_state = STATE_REP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2677 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2678
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2679 default:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2680 // 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	2681 --depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2682
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2683 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2684 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2685 // Continue in or go back to the prefix tree.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2686 byts = pbyts;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2687 idxs = pidxs;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2690 // 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	2691 if (--breakcheckcount == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2692 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2693 ui_breakcheck();
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2694 breakcheckcount = 1000;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2695 #ifdef FEAT_RELTIME
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2696 if (spell_suggest_timeout > 0
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2697 && profile_passed_limit(&time_limit))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2698 got_int = TRUE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2699 #endif
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2707 * Go one level deeper in the tree.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2708 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2709 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2710 go_deeper(trystate_T *stack, int depth, int score_add)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2711 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2712 stack[depth + 1] = stack[depth];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2713 stack[depth + 1].ts_state = STATE_START;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2714 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	2715 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	2716 stack[depth + 1].ts_flags = 0;
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2720 * "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	2721 * words and put it in "kword".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2722 * 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	2723 * same case-folded word, but we only find one...
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2724 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2725 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2726 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	2727 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2728 char_u uword[MAXWLEN]; // "fword" in upper-case
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2729 int depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2730 idx_T tryidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2731
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2732 // 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	2733 idx_T arridx[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2734 int round[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2735 int fwordidx[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2736 int uwordidx[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2737 int kwordlen[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2738
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2739 int flen, ulen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2740 int l;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2741 int len;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2742 int c;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2743 idx_T lo, hi, m;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2744 char_u *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2745 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	2746 idx_T *idxs = slang->sl_kidxs; // array with indexes
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2747
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2748 if (byts == NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2749 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2750 // array is empty: "cannot happen"
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2751 *kword = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2752 return;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2755 // Make an all-cap version of "fword".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2756 allcap_copy(fword, uword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2757
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2758 // 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	2759 // 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	2760 // may change the byte length of a multi-byte character...
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2761 depth = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2762 arridx[0] = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2763 round[0] = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2764 fwordidx[0] = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2765 uwordidx[0] = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2766 kwordlen[0] = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2767 while (depth >= 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2768 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2769 if (fword[fwordidx[depth]] == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2770 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2771 // 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	2772 // here we have found a match.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2773 if (byts[arridx[depth] + 1] == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2774 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2775 kword[kwordlen[depth]] = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2776 return;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2779 // kword is getting too long, continue one level up
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2780 --depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2781 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2782 else if (++round[depth] > 2)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2783 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2784 // tried both fold-case and upper-case character, continue one
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2785 // level up
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2786 --depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2787 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2788 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2789 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2790 // round[depth] == 1: Try using the folded-case character.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2791 // round[depth] == 2: Try using the upper-case character.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2792 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2793 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2794 flen = MB_CPTR2LEN(fword + fwordidx[depth]);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2795 ulen = MB_CPTR2LEN(uword + uwordidx[depth]);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2796 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2797 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2798 ulen = flen = 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2799 if (round[depth] == 1)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2800 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2801 p = fword + fwordidx[depth];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2802 l = flen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2803 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2804 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2805 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2806 p = uword + uwordidx[depth];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2807 l = ulen;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2810 for (tryidx = arridx[depth]; l > 0; --l)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2811 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2812 // Perform a binary search in the list of accepted bytes.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2813 len = byts[tryidx++];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2814 c = *p++;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2815 lo = tryidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2816 hi = tryidx + len - 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2817 while (lo < hi)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2818 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2819 m = (lo + hi) / 2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2820 if (byts[m] > c)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2821 hi = m - 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2822 else if (byts[m] < c)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2823 lo = m + 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2824 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2825 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2826 lo = hi = m;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2827 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2831 // Stop if there is no matching byte.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2832 if (hi < lo \|\| byts[lo] != c)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2833 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2834
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2835 // Continue at the child (if there is one).
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2836 tryidx = idxs[lo];
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2839 if (l == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2840 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2841 // 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	2842 // level deeper.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2843 if (round[depth] == 1)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2844 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2845 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2846 flen);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2847 kwordlen[depth + 1] = kwordlen[depth] + flen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2848 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2849 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2850 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2851 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2852 ulen);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2853 kwordlen[depth + 1] = kwordlen[depth] + ulen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2854 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2855 fwordidx[depth + 1] = fwordidx[depth] + flen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2856 uwordidx[depth + 1] = uwordidx[depth] + ulen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2857
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2858 ++depth;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2859 arridx[depth] = tryidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2860 round[depth] = 0;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2865 // Didn't find it: "cannot happen".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2866 *kword = NUL;
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2870 * 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	2871 * su->su_sga.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2872 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2873 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2874 score_comp_sal(suginfo_T *su)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2875 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2876 langp_T *lp;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2877 char_u badsound[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2878 int i;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2879 suggest_T *stp;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2880 suggest_T *sstp;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2881 int score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2882 int lpi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2883
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2884 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	2885 return;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2886
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2887 // 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	2888 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	2889 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2890 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2891 if (lp->lp_slang->sl_sal.ga_len > 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2892 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2893 // soundfold the bad word
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2894 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2895
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2896 for (i = 0; i < su->su_ga.ga_len; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2897 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2898 stp = &SUG(su->su_ga, i);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2899
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2900 // 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	2901 // sound-a-like score.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2902 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2903 if (score < SCORE_MAXMAX)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2904 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2905 // Add the suggestion.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2906 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2907 sstp->st_word = vim_strsave(stp->st_word);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2908 if (sstp->st_word != NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2909 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2910 sstp->st_wordlen = stp->st_wordlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2911 sstp->st_score = score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2912 sstp->st_altscore = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2913 sstp->st_orglen = stp->st_orglen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2914 ++su->su_sga.ga_len;
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2918 break;
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2924 * 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	2925 * They are entwined.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2926 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2927 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2928 score_combine(suginfo_T *su)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2929 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2930 int i;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2931 int j;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2932 garray_T ga;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2933 garray_T *gap;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2934 langp_T *lp;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2935 suggest_T *stp;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2936 char_u *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2937 char_u badsound[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2938 int round;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2939 int lpi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2940 slang_T *slang = NULL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2941
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2942 // Add the alternate score to su_ga.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2943 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	2944 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2945 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2946 if (lp->lp_slang->sl_sal.ga_len > 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2947 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2948 // soundfold the bad word
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2949 slang = lp->lp_slang;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2950 spell_soundfold(slang, su->su_fbadword, TRUE, badsound);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2951
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2952 for (i = 0; i < su->su_ga.ga_len; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2953 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2954 stp = &SUG(su->su_ga, i);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2955 stp->st_altscore = stp_sal_score(stp, su, slang, badsound);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2956 if (stp->st_altscore == SCORE_MAXMAX)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2957 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2958 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2959 stp->st_score = (stp->st_score * 3
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2960 + stp->st_altscore) / 4;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2961 stp->st_salscore = FALSE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2962 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2963 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2967 if (slang == NULL) // Using "double" without sound folding.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2968 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2969 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2970 su->su_maxcount);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2971 return;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2974 // Add the alternate score to su_sga.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2975 for (i = 0; i < su->su_sga.ga_len; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2976 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2977 stp = &SUG(su->su_sga, i);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2978 stp->st_altscore = spell_edit_score(slang,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2979 su->su_badword, stp->st_word);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2980 if (stp->st_score == SCORE_MAXMAX)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2981 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2982 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2983 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	2984 stp->st_salscore = TRUE;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2987 // Remove bad suggestions, sort the suggestions and truncate at "maxcount"
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2988 // for both lists.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2989 check_suggestions(su, &su->su_ga);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2990 (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	2991 check_suggestions(su, &su->su_sga);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2992 (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	2993
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2994 ga_init2(&ga, sizeof(suginfo_T), 1);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2995 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	2996 return;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2997
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2998 stp = &SUG(ga, 0);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	2999 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	3000 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3001 // round 1: get a suggestion from su_ga
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3002 // round 2: get a suggestion from su_sga
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3003 for (round = 1; round <= 2; ++round)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3004 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3005 gap = round == 1 ? &su->su_ga : &su->su_sga;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3006 if (i < gap->ga_len)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3007 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3008 // Don't add a word if it's already there.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3009 p = SUG(*gap, i).st_word;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3010 for (j = 0; j < ga.ga_len; ++j)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3011 if (STRCMP(stp[j].st_word, p) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3012 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3013 if (j == ga.ga_len)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3014 stp[ga.ga_len++] = SUG(*gap, i);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3015 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3016 vim_free(p);
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3021 ga_clear(&su->su_ga);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3022 ga_clear(&su->su_sga);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3023
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3024 // 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	3025 if (ga.ga_len > su->su_maxcount)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3026 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3027 for (i = su->su_maxcount; i < ga.ga_len; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3028 vim_free(stp[i].st_word);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3029 ga.ga_len = su->su_maxcount;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3032 su->su_ga = ga;
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3036 * 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	3037 * badword.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3038 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3039 static int
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3040 stp_sal_score(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3041 suggest_T *stp,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3042 suginfo_T *su,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3043 slang_T *slang,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3044 char_u *badsound) // sound-folded badword
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3045 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3046 char_u *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3047 char_u *pbad;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3048 char_u *pgood;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3049 char_u badsound2[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3050 char_u fword[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3051 char_u goodsound[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3052 char_u goodword[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3053 int lendiff;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3054
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3055 lendiff = (int)(su->su_badlen - stp->st_orglen);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3056 if (lendiff >= 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3057 pbad = badsound;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3058 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3059 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3060 // soundfold the bad word with more characters following
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3061 (void)spell_casefold(curwin,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3062 su->su_badptr, stp->st_orglen, fword, MAXWLEN);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3063
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3064 // 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	3065 // 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	3066 // 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	3067 // space.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3068 if (VIM_ISWHITE(su->su_badptr[su->su_badlen])
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3069 && *skiptowhite(stp->st_word) == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3070 for (p = fword; *(p = skiptowhite(p)) != NUL; )
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3071 STRMOVE(p, p + 1);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3072
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3073 spell_soundfold(slang, fword, TRUE, badsound2);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3074 pbad = badsound2;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3077 if (lendiff > 0 && stp->st_wordlen + lendiff < MAXWLEN)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3078 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3079 // 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	3080 // what replaces the bad word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3081 STRCPY(goodword, stp->st_word);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3082 vim_strncpy(goodword + stp->st_wordlen,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3083 su->su_badptr + su->su_badlen - lendiff, lendiff);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3084 pgood = goodword;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3085 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3086 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3087 pgood = stp->st_word;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3088
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3089 // 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	3090 spell_soundfold(slang, pgood, FALSE, goodsound);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3091
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3092 return soundalike_score(goodsound, pbad);
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3095 // 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	3096 // handled already.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3097 typedef struct
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3098 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3099 short sft_score; // lowest score used
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3100 char_u sft_word[1]; // soundfolded word, actually longer
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3101 } sftword_T;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3102
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3103 static sftword_T dumsft;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3104 #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	3105 #define HI2SFT(hi) HIKEY2SFT((hi)->hi_key)
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3108 * Prepare for calling suggest_try_soundalike().
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3109 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3110 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3111 suggest_try_soundalike_prep(void)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3112 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3113 langp_T *lp;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3114 int lpi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3115 slang_T *slang;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3116
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3117 // 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	3118 // .sug file has been loaded.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3119 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	3120 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3121 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3122 slang = lp->lp_slang;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3123 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	3124 // prepare the hashtable used by add_sound_suggest()
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3125 hash_init(&slang->sl_sounddone);
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3130 * 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	3131 * Note: This doesn't support postponed prefixes.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3132 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3133 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3134 suggest_try_soundalike(suginfo_T *su)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3135 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3136 char_u salword[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3137 langp_T *lp;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3138 int lpi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3139 slang_T *slang;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3140
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3141 // 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	3142 // .sug file has been loaded.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3143 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	3144 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3145 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3146 slang = lp->lp_slang;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3147 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	3148 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3149 // soundfold the bad word
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3150 spell_soundfold(slang, su->su_fbadword, TRUE, salword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3151
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3152 // try all kinds of inserts/deletes/swaps/etc.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3153 // 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	3154 // and splitting
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3155 #ifdef SUGGEST_PROFILE
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3156 prof_init();
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3157 #endif
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3158 suggest_trie_walk(su, lp, salword, TRUE);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3159 #ifdef SUGGEST_PROFILE
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3160 prof_report("soundalike");
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3161 #endif
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3167 * Finish up after calling suggest_try_soundalike().
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3168 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3169 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3170 suggest_try_soundalike_finish(void)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3171 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3172 langp_T *lp;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3173 int lpi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3174 slang_T *slang;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3175 int todo;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3176 hashitem_T *hi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3177
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3178 // 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	3179 // .sug file has been loaded.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3180 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	3181 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3182 lp = LANGP_ENTRY(curwin->w_s->b_langp, lpi);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3183 slang = lp->lp_slang;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3184 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	3185 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3186 // Free the info about handled words.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3187 todo = (int)slang->sl_sounddone.ht_used;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3188 FOR_ALL_HASHTAB_ITEMS(&slang->sl_sounddone, hi, todo)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3189 if (!HASHITEM_EMPTY(hi))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3190 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3191 vim_free(HI2SFT(hi));
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3192 --todo;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3195 // 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	3196 hash_clear(&slang->sl_sounddone);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3197 hash_init(&slang->sl_sounddone);
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3203 * 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	3204 * produce this soundfolded word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3205 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3206 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3207 add_sound_suggest(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3208 suginfo_T *su,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3209 char_u *goodword,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3210 int score, // soundfold score
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3211 langp_T *lp)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3212 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3213 slang_T *slang = lp->lp_slang; // language for sound folding
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3214 int sfwordnr;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3215 char_u *nrline;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3216 int orgnr;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3217 char_u theword[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3218 int i;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3219 int wlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3220 char_u *byts;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3221 idx_T *idxs;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3222 int n;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3223 int wordcount;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3224 int wc;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3225 int goodscore;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3226 hash_T hash;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3227 hashitem_T *hi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3228 sftword_T *sft;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3229 int bc, gc;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3230 int limit;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3231
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3232 // 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	3233 // 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	3234 // 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	3235 // remember the words that have been done.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3236 hash = hash_hash(goodword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3237 hi = hash_lookup(&slang->sl_sounddone, goodword, hash);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3238 if (HASHITEM_EMPTY(hi))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3239 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3240 sft = alloc(offsetof(sftword_T, sft_word) + STRLEN(goodword) + 1);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3241 if (sft != NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3242 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3243 sft->sft_score = score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3244 STRCPY(sft->sft_word, goodword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3245 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	3246 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3247 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3248 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3249 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3250 sft = HI2SFT(hi);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3251 if (score >= sft->sft_score)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3252 return;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3253 sft->sft_score = score;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3256 // Find the word nr in the soundfold tree.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3257 sfwordnr = soundfold_find(slang, goodword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3258 if (sfwordnr < 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3259 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3260 internal_error("add_sound_suggest()");
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3261 return;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3264 // 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	3265 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	3266 orgnr = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3267 while (*nrline != NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3268 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3269 // 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	3270 // previous wordnr.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3271 orgnr += bytes2offset(&nrline);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3272
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3273 byts = slang->sl_fbyts;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3274 idxs = slang->sl_fidxs;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3275
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3276 // Lookup the word "orgnr" one of the two tries.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3277 n = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3278 wordcount = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3279 for (wlen = 0; wlen < MAXWLEN - 3; ++wlen)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3280 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3281 i = 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3282 if (wordcount == orgnr && byts[n + 1] == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3283 break; // found end of word
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3284
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3285 if (byts[n + 1] == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3286 ++wordcount;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3287
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3288 // skip over the NUL bytes
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3289 for ( ; byts[n + i] == NUL; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3290 if (i > byts[n]) // safety check
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3291 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3292 STRCPY(theword + wlen, "BAD");
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3293 wlen += 3;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3294 goto badword;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3297 // One of the siblings must have the word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3298 for ( ; i < byts[n]; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3299 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3300 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	3301 if (wordcount + wc > orgnr)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3302 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3303 wordcount += wc;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3306 theword[wlen] = byts[n + i];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3307 n = idxs[n + i];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3308 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3309 badword:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3310 theword[wlen] = NUL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3311
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3312 // Go over the possible flags and regions.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3313 for (; i <= byts[n] && byts[n + i] == NUL; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3314 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3315 char_u cword[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3316 char_u *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3317 int flags = (int)idxs[n + i];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3318
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3319 // Skip words with the NOSUGGEST flag
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3320 if (flags & WF_NOSUGGEST)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3321 continue;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3322
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3323 if (flags & WF_KEEPCAP)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3324 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3325 // Must find the word in the keep-case tree.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3326 find_keepcap_word(slang, theword, cword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3327 p = cword;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3328 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3329 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3330 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3331 flags \|= su->su_badflags;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3332 if ((flags & WF_CAPMASK) != 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3333 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3334 // Need to fix case according to "flags".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3335 make_case_word(theword, cword, flags);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3336 p = cword;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3337 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3338 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3339 p = theword;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3342 // Add the suggestion.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3343 if (sps_flags & SPS_DOUBLE)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3344 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3345 // Add the suggestion if the score isn't too bad.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3346 if (score <= su->su_maxscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3347 add_suggestion(su, &su->su_sga, p, su->su_badlen,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3348 score, 0, FALSE, slang, FALSE);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3349 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3350 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3351 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3352 // Add a penalty for words in another region.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3353 if ((flags & WF_REGION)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3354 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3355 goodscore = SCORE_REGION;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3356 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3357 goodscore = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3358
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3359 // Add a small penalty for changing the first letter from
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3360 // lower to upper case. Helps for "tath" -> "Kath", which is
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3361 // 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	3362 // letter is the same, that has already been counted.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3363 gc = PTR2CHAR(p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3364 if (SPELL_ISUPPER(gc))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3365 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3366 bc = PTR2CHAR(su->su_badword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3367 if (!SPELL_ISUPPER(bc)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3368 && SPELL_TOFOLD(bc) != SPELL_TOFOLD(gc))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3369 goodscore += SCORE_ICASE / 2;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3372 // 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	3373 // insert/delete/swap/replace. REP items are not considered,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3374 // which may make the score a bit higher.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3375 // 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	3376 // MAXSCORE(), because RESCORE() will change the score.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3377 // 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	3378 // inefficient, using an array is quicker.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3379 limit = MAXSCORE(su->su_sfmaxscore - goodscore, score);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3380 if (limit > SCORE_LIMITMAX)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3381 goodscore += spell_edit_score(slang, su->su_badword, p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3382 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3383 goodscore += spell_edit_score_limit(slang, su->su_badword,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3384 p, limit);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3385
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3386 // 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	3387 if (goodscore < SCORE_MAXMAX)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3388 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3389 // Give a bonus to words seen before.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3390 goodscore = score_wordcount_adj(slang, goodscore, p, FALSE);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3391
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3392 // Add the suggestion if the score isn't too bad.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3393 goodscore = RESCORE(goodscore, score);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3394 if (goodscore <= su->su_sfmaxscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3395 add_suggestion(su, &su->su_ga, p, su->su_badlen,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3396 goodscore, score, TRUE, slang, TRUE);
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3400 // smsg("word %s (%d): %s (%d)", sftword, sftnr, theword, orgnr);
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3405 * 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	3406 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3407 static int
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3408 soundfold_find(slang_T slang, char_u word)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3409 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3410 idx_T arridx = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3411 int len;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3412 int wlen = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3413 int c;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3414 char_u *ptr = word;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3415 char_u *byts;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3416 idx_T *idxs;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3417 int wordnr = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3418
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3419 byts = slang->sl_sbyts;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3420 idxs = slang->sl_sidxs;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3421
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3422 for (;;)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3423 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3424 // First byte is the number of possible bytes.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3425 len = byts[arridx++];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3426
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3427 // 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	3428 // 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	3429 c = ptr[wlen];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3430 if (byts[arridx] == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3431 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3432 if (c == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3433 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3434
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3435 // Skip over the zeros, there can be several.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3436 while (len > 0 && byts[arridx] == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3437 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3438 ++arridx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3439 --len;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3440 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3441 if (len == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3442 return -1; // no children, word should have ended here
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3443 ++wordnr;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3446 // If the word ends we didn't find it.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3447 if (c == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3448 return -1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3449
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3450 // Perform a binary search in the list of accepted bytes.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3451 if (c == TAB) // <Tab> is handled like <Space>
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3452 c = ' ';
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3453 while (byts[arridx] < c)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3454 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3455 // 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	3456 wordnr += idxs[idxs[arridx]];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3457 ++arridx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3458 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	3459 return -1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3460 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3461 if (byts[arridx] != c) // didn't find the byte
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3462 return -1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3463
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3464 // Continue at the child (if there is one).
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3465 arridx = idxs[arridx];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3466 ++wlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3467
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3468 // 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	3469 // checked word.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3470 if (c == ' ')
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3471 while (ptr[wlen] == ' ' \|\| ptr[wlen] == TAB)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3472 ++wlen;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3475 return wordnr;
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3479 * 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	3480 * lines in the .aff file.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3481 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3482 static int
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3483 similar_chars(slang_T *slang, int c1, int c2)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3484 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3485 int m1, m2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3486 char_u buf[MB_MAXBYTES + 1];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3487 hashitem_T *hi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3488
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3489 if (c1 >= 256)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3490 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3491 buf[mb_char2bytes(c1, buf)] = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3492 hi = hash_find(&slang->sl_map_hash, buf);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3493 if (HASHITEM_EMPTY(hi))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3494 m1 = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3495 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3496 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	3497 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3498 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3499 m1 = slang->sl_map_array[c1];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3500 if (m1 == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3501 return FALSE;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3504 if (c2 >= 256)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3505 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3506 buf[mb_char2bytes(c2, buf)] = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3507 hi = hash_find(&slang->sl_map_hash, buf);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3508 if (HASHITEM_EMPTY(hi))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3509 m2 = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3510 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3511 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	3512 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3513 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3514 m2 = slang->sl_map_array[c2];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3515
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3516 return m1 == m2;
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3520 * Add a suggestion to the list of suggestions.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3521 * 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	3522 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3523 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3524 add_suggestion(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3525 suginfo_T *su,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3526 garray_T *gap, // either su_ga or su_sga
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3527 char_u *goodword,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3528 int badlenarg, // len of bad word replaced with "goodword"
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3529 int score,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3530 int altscore,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3531 int had_bonus, // value for st_had_bonus
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3532 slang_T *slang, // language for sound folding
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3533 int maxsf) // su_maxscore applies to soundfold score,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3534 // su_sfmaxscore to the total score.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3535 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3536 int goodlen; // len of goodword changed
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3537 int badlen; // len of bad word changed
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3538 suggest_T *stp;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3539 suggest_T new_sug;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3540 int i;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3541 char_u pgood, pbad;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3542
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3543 // Minimize "badlen" for consistency. Avoids that changing "the the" to
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3544 // "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	3545 pgood = goodword + STRLEN(goodword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3546 pbad = su->su_badptr + badlenarg;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3547 for (;;)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3548 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3549 goodlen = (int)(pgood - goodword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3550 badlen = (int)(pbad - su->su_badptr);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3551 if (goodlen <= 0 \|\| badlen <= 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3552 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3553 MB_PTR_BACK(goodword, pgood);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3554 MB_PTR_BACK(su->su_badptr, pbad);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3555 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3556 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3557 if (mb_ptr2char(pgood) != mb_ptr2char(pbad))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3558 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3559 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3560 else if (pgood != pbad)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3561 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3564 if (badlen == 0 && goodlen == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3565 // 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	3566 // the first "the" to itself.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3567 return;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3568
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3569 if (gap->ga_len == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3570 i = -1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3571 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3572 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3573 // 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	3574 // being replaced "thes," -> "these" is a different suggestion from
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3575 // "thes" -> "these".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3576 stp = &SUG(*gap, 0);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3577 for (i = gap->ga_len; --i >= 0; ++stp)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3578 if (stp->st_wordlen == goodlen
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3579 && stp->st_orglen == badlen
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3580 && STRNCMP(stp->st_word, goodword, goodlen) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3581 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3582 // Found it. Remember the word with the lowest score.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3583 if (stp->st_slang == NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3584 stp->st_slang = slang;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3585
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3586 new_sug.st_score = score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3587 new_sug.st_altscore = altscore;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3588 new_sug.st_had_bonus = had_bonus;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3589
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3590 if (stp->st_had_bonus != had_bonus)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3591 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3592 // Only one of the two had the soundalike score computed.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3593 // 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	3594 // scores can't be compared. This happens because
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3595 // suggest_try_change() doesn't compute the soundalike
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3596 // word to keep it fast, while some special methods set
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3597 // the soundalike score to zero.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3598 if (had_bonus)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3599 rescore_one(su, stp);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3600 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3601 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3602 new_sug.st_word = stp->st_word;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3603 new_sug.st_wordlen = stp->st_wordlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3604 new_sug.st_slang = stp->st_slang;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3605 new_sug.st_orglen = badlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3606 rescore_one(su, &new_sug);
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3610 if (stp->st_score > new_sug.st_score)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3611 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3612 stp->st_score = new_sug.st_score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3613 stp->st_altscore = new_sug.st_altscore;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3614 stp->st_had_bonus = new_sug.st_had_bonus;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3615 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3616 break;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3620 if (i < 0 && ga_grow(gap, 1) == OK)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3621 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3622 // Add a suggestion.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3623 stp = &SUG(*gap, gap->ga_len);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3624 stp->st_word = vim_strnsave(goodword, goodlen);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3625 if (stp->st_word != NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3626 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3627 stp->st_wordlen = goodlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3628 stp->st_score = score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3629 stp->st_altscore = altscore;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3630 stp->st_had_bonus = had_bonus;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3631 stp->st_orglen = badlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3632 stp->st_slang = slang;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3633 ++gap->ga_len;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3634
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3635 // 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	3636 // the best suggestions.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3637 if (gap->ga_len > SUG_MAX_COUNT(su))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3638 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3639 if (maxsf)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3640 su->su_sfmaxscore = cleanup_suggestions(gap,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3641 su->su_sfmaxscore, SUG_CLEAN_COUNT(su));
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3642 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3643 su->su_maxscore = cleanup_suggestions(gap,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3644 su->su_maxscore, SUG_CLEAN_COUNT(su));
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3651 * 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	3652 * 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	3653 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3654 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3655 check_suggestions(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3656 suginfo_T *su,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3657 garray_T *gap) // either su_ga or su_sga
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3658 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3659 suggest_T *stp;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3660 int i;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3661 char_u longword[MAXWLEN + 1];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3662 int len;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3663 hlf_T attr;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3664
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3665 if (gap->ga_len == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3666 return;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3667 stp = &SUG(*gap, 0);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3668 for (i = gap->ga_len - 1; i >= 0; --i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3669 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3670 // Need to append what follows to check for "the the".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3671 vim_strncpy(longword, stp[i].st_word, MAXWLEN);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3672 len = stp[i].st_wordlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3673 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	3674 MAXWLEN - len);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3675 attr = HLF_COUNT;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3676 (void)spell_check(curwin, longword, &attr, NULL, FALSE);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3677 if (attr != HLF_COUNT)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3678 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3679 // Remove this entry.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3680 vim_free(stp[i].st_word);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3681 --gap->ga_len;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3682 if (i < gap->ga_len)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3683 mch_memmove(stp + i, stp + i + 1,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3684 sizeof(suggest_T) * (gap->ga_len - i));
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3691 * Add a word to be banned.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3692 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3693 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3694 add_banned(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3695 suginfo_T *su,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3696 char_u *word)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3697 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3698 char_u *s;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3699 hash_T hash;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3700 hashitem_T *hi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3701
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3702 hash = hash_hash(word);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3703 hi = hash_lookup(&su->su_banned, word, hash);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3704 if (!HASHITEM_EMPTY(hi)) // already present
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3705 return;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3706 s = vim_strsave(word);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3707 if (s != NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3708 hash_add_item(&su->su_banned, hi, s, hash);
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3712 * 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	3713 * is slow, thus only done for the final results.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3714 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3715 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3716 rescore_suggestions(suginfo_T *su)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3717 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3718 int i;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3719
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3720 if (su->su_sallang != NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3721 for (i = 0; i < su->su_ga.ga_len; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3722 rescore_one(su, &SUG(su->su_ga, i));
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3726 * 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	3727 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3728 static void
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3729 rescore_one(suginfo_T su, suggest_T stp)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3730 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3731 slang_T *slang = stp->st_slang;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3732 char_u sal_badword[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3733 char_u *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3734
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3735 // 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	3736 // language.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3737 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	3738 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3739 if (slang == su->su_sallang)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3740 p = su->su_sal_badword;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3741 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3742 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3743 spell_soundfold(slang, su->su_fbadword, TRUE, sal_badword);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3744 p = sal_badword;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3747 stp->st_altscore = stp_sal_score(stp, su, slang, p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3748 if (stp->st_altscore == SCORE_MAXMAX)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3749 stp->st_altscore = SCORE_BIG;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3750 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3751 stp->st_had_bonus = TRUE;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3755 static int sug_compare(const void s1, const void s2);
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 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3758 * 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	3759 * First on "st_score", then "st_altscore" then alphabetically.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3760 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3761 static int
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3762 sug_compare(const void s1, const void s2)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3763 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3764 suggest_T p1 = (suggest_T )s1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3765 suggest_T p2 = (suggest_T )s2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3766 int n = p1->st_score - p2->st_score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3767
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3768 if (n == 0)
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 n = p1->st_altscore - p2->st_altscore;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3771 if (n == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3772 n = STRICMP(p1->st_word, p2->st_word);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3773 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3774 return n;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3775 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3776
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 * Cleanup the suggestions:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3779 * - Sort on score.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3780 * - Remove words that won't be displayed.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3781 * Returns the maximum score in the list or "maxscore" unmodified.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3782 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3783 static int
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3784 cleanup_suggestions(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3785 garray_T *gap,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3786 int maxscore,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3787 int keep) // nr of suggestions to keep
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3788 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3789 if (gap->ga_len <= 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3790 return maxscore;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3791
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3792 // Sort the list.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3793 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	3794 sug_compare);
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 // 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	3797 // displayed.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3798 if (gap->ga_len > keep)
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 int i;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3801 suggest_T stp = &SUG(gap, 0);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3802
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3803 for (i = keep; i < gap->ga_len; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3804 vim_free(stp[i].st_word);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3805 gap->ga_len = keep;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3806 if (keep >= 1)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3807 return stp[keep - 1].st_score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3808 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3809 return maxscore;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3810 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3811
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 * Compute a score for two sound-a-like words.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3814 * 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	3815 * 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	3816 * avoiding checks that will not be possible.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3817 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3818 static int
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3819 soundalike_score(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3820 char_u *goodstart, // sound-folded good word
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3821 char_u *badstart) // sound-folded bad word
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3822 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3823 char_u *goodsound = goodstart;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3824 char_u *badsound = badstart;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3825 int goodlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3826 int badlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3827 int n;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3828 char_u pl, ps;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3829 char_u pl2, ps2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3830 int score = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3831
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3832 // 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	3833 // 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	3834 if ((badsound == '' \|\| goodsound == '') && badsound != goodsound)
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 if ((badsound[0] == NUL && goodsound[1] == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3837 \|\| (goodsound[0] == NUL && badsound[1] == NUL))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3838 // changing word with vowel to word without a sound
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3839 return SCORE_DEL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3840 if (badsound[0] == NUL \|\| goodsound[0] == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3841 // more than two changes
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3842 return SCORE_MAXMAX;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3843
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3844 if (badsound[1] == goodsound[1]
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3845 \|\| (badsound[1] != NUL
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3846 && goodsound[1] != NUL
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3847 && badsound[2] == goodsound[2]))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3848 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3849 // handle like a substitute
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3850 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3851 else
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 score = 2 * SCORE_DEL / 3;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3854 if (badsound == '')
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3855 ++badsound;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3856 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3857 ++goodsound;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3858 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3859 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3860
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3861 goodlen = (int)STRLEN(goodsound);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3862 badlen = (int)STRLEN(badsound);
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 // 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	3865 // changes.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3866 n = goodlen - badlen;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3867 if (n < -2 \|\| n > 2)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3868 return SCORE_MAXMAX;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3869
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3870 if (n > 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3871 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3872 pl = goodsound; // goodsound is longest
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3873 ps = badsound;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3874 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3875 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3876 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3877 pl = badsound; // badsound is longest
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3878 ps = goodsound;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3879 }
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 // Skip over the identical part.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3882 while (pl == ps && *pl != NUL)
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 ++pl;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3885 ++ps;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3886 }
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 switch (n)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3889 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3890 case -2:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3891 case 2:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3892 // Must delete two characters from "pl".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3893 ++pl; // first delete
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3894 while (pl == ps)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3895 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3896 ++pl;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3897 ++ps;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3898 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3899 // strings must be equal after second delete
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3900 if (STRCMP(pl + 1, ps) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3901 return score + SCORE_DEL * 2;
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 // Failed to compare.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3904 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3905
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3906 case -1:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3907 case 1:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3908 // Minimal one delete from "pl" required.
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 // 1: delete
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3911 pl2 = pl + 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3912 ps2 = ps;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3913 while (pl2 == ps2)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3914 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3915 if (*pl2 == NUL) // reached the end
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3916 return score + SCORE_DEL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3917 ++pl2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3918 ++ps2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3919 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3920
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3921 // 2: delete then swap, then rest must be equal
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3922 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3923 && STRCMP(pl2 + 2, ps2 + 2) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3924 return score + SCORE_DEL + SCORE_SWAP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3925
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3926 // 3: delete then substitute, then the rest must be equal
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3927 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3928 return score + SCORE_DEL + SCORE_SUBST;
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 // 4: first swap then delete
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3931 if (pl[0] == ps[1] && pl[1] == ps[0])
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3932 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3933 pl2 = pl + 2; // swap, skip two chars
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3934 ps2 = ps + 2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3935 while (pl2 == ps2)
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;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3938 ++ps2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3939 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3940 // delete a char and then strings must be equal
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3941 if (STRCMP(pl2 + 1, ps2) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3942 return score + SCORE_SWAP + SCORE_DEL;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3945 // 5: first substitute then delete
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3946 pl2 = pl + 1; // substitute, skip one char
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3947 ps2 = ps + 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3948 while (pl2 == ps2)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3949 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3950 ++pl2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3951 ++ps2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3952 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3953 // delete a char and then strings must be equal
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3954 if (STRCMP(pl2 + 1, ps2) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3955 return score + SCORE_SUBST + SCORE_DEL;
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 // Failed to compare.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3958 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3959
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3960 case 0:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3961 // 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	3962 // insert is only possible in combination with a delete.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3963 // 1: check if for identical strings
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3964 if (*pl == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3965 return score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3966
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3967 // 2: swap
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3968 if (pl[0] == ps[1] && pl[1] == ps[0])
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3969 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3970 pl2 = pl + 2; // swap, skip two chars
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3971 ps2 = ps + 2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3972 while (pl2 == ps2)
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 if (*pl2 == NUL) // reached the end
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3975 return score + SCORE_SWAP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3976 ++pl2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3977 ++ps2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3978 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3979 // 3: swap and swap again
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3980 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3981 && STRCMP(pl2 + 2, ps2 + 2) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3982 return score + SCORE_SWAP + SCORE_SWAP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3983
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3984 // 4: swap and substitute
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3985 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3986 return score + SCORE_SWAP + SCORE_SUBST;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3989 // 5: substitute
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3990 pl2 = pl + 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3991 ps2 = ps + 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3992 while (pl2 == ps2)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3993 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3994 if (*pl2 == NUL) // reached the end
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3995 return score + SCORE_SUBST;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3996 ++pl2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3997 ++ps2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3998 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	3999
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4000 // 6: substitute and swap
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4001 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4002 && STRCMP(pl2 + 2, ps2 + 2) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4003 return score + SCORE_SUBST + SCORE_SWAP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4004
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4005 // 7: substitute and substitute
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4006 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4007 return score + SCORE_SUBST + SCORE_SUBST;
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 // 8: insert then delete
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4010 pl2 = pl;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4011 ps2 = ps + 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4012 while (pl2 == ps2)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4013 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4014 ++pl2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4015 ++ps2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4016 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4017 if (STRCMP(pl2 + 1, ps2) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4018 return score + SCORE_INS + SCORE_DEL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4019
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4020 // 9: delete then insert
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4021 pl2 = pl + 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4022 ps2 = ps;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4023 while (pl2 == ps2)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4024 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4025 ++pl2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4026 ++ps2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4027 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4028 if (STRCMP(pl2, ps2 + 1) == 0)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4029 return score + SCORE_INS + SCORE_DEL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4030
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4031 // Failed to compare.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4032 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4033 }
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 return SCORE_MAXMAX;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4036 }
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 * 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	4040 * deletes/inserts/substitutes/swaps are required the lower the score.
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 * The algorithm is described by Du and Chang, 1992.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4043 * The implementation of the algorithm comes from Aspell editdist.cpp,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4044 * 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	4045 * support multi-byte characters.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4046 */
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4047 static int
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4048 spell_edit_score(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4049 slang_T *slang,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4050 char_u *badword,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4051 char_u *goodword)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4052 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4053 int *cnt;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4054 int badlen, goodlen; // lengths including NUL
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4055 int j, i;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4056 int t;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4057 int bc, gc;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4058 int pbc, pgc;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4059 char_u *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4060 int wbadword[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4061 int wgoodword[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4062
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4063 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4064 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4065 // 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	4066 // int array for easy access.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4067 for (p = badword, badlen = 0; *p != NUL; )
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4068 wbadword[badlen++] = mb_cptr2char_adv(&p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4069 wbadword[badlen++] = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4070 for (p = goodword, goodlen = 0; *p != NUL; )
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4071 wgoodword[goodlen++] = mb_cptr2char_adv(&p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4072 wgoodword[goodlen++] = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4073 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4074 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4075 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4076 badlen = (int)STRLEN(badword) + 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4077 goodlen = (int)STRLEN(goodword) + 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4078 }
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 // 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	4081 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4082 cnt = ALLOC_MULT(int, (badlen + 1) * (goodlen + 1));
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4083 if (cnt == NULL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4084 return 0; // out of memory
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4085
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4086 CNT(0, 0) = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4087 for (j = 1; j <= goodlen; ++j)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4088 CNT(0, j) = CNT(0, j - 1) + SCORE_INS;
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 for (i = 1; i <= badlen; ++i)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4091 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4092 CNT(i, 0) = CNT(i - 1, 0) + SCORE_DEL;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4093 for (j = 1; j <= goodlen; ++j)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4094 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4095 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4096 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4097 bc = wbadword[i - 1];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4098 gc = wgoodword[j - 1];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4099 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4100 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4101 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4102 bc = badword[i - 1];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4103 gc = goodword[j - 1];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4104 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4105 if (bc == gc)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4106 CNT(i, j) = CNT(i - 1, j - 1);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4107 else
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 // 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	4110 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4111 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4112 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4113 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4114 // For a similar character use SCORE_SIMILAR.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4115 if (slang != NULL
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4116 && slang->sl_has_map
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4117 && similar_chars(slang, gc, bc))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4118 CNT(i, j) = SCORE_SIMILAR + CNT(i - 1, j - 1);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4119 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4120 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4121 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4122
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4123 if (i > 1 && j > 1)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4124 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4125 if (has_mbyte)
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 pbc = wbadword[i - 2];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4128 pgc = wgoodword[j - 2];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4129 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4130 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4131 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4132 pbc = badword[i - 2];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4133 pgc = goodword[j - 2];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4134 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4135 if (bc == pgc && pbc == gc)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4136 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4137 t = SCORE_SWAP + CNT(i - 2, j - 2);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4138 if (t < CNT(i, j))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4139 CNT(i, j) = t;
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 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4142 t = SCORE_DEL + CNT(i - 1, j);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4143 if (t < CNT(i, j))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4144 CNT(i, j) = t;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4145 t = SCORE_INS + CNT(i, j - 1);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4146 if (t < CNT(i, j))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4147 CNT(i, j) = t;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4148 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4149 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4150 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4151
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4152 i = CNT(badlen - 1, goodlen - 1);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4153 vim_free(cnt);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4154 return i;
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
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4157 typedef struct
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4158 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4159 int badi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4160 int goodi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4161 int score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4162 } limitscore_T;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4163
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4164 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4165 * 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	4166 * 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	4167 *
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4168 * 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	4169 * 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	4170 * for multi-byte characters.
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 static int
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4173 spell_edit_score_limit(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4174 slang_T *slang,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4175 char_u *badword,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4176 char_u *goodword,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4177 int limit)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4178 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4179 limitscore_T stack[10]; // allow for over 3 * 2 edits
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4180 int stackidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4181 int bi, gi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4182 int bi2, gi2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4183 int bc, gc;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4184 int score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4185 int score_off;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4186 int minscore;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4187 int round;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4188
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4189 // 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	4190 // to avoid testing "has_mbyte" quite often.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4191 if (has_mbyte)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4192 return spell_edit_score_limit_w(slang, badword, goodword, limit);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4193
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4194 // 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	4195 // 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	4196 // When there is a difference try several alternatives. Each alternative
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4197 // 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	4198 // 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	4199 // 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	4200 // possible score is stored in "minscore".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4201 stackidx = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4202 bi = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4203 gi = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4204 score = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4205 minscore = limit + 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4206
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4207 for (;;)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4208 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4209 // Skip over an equal part, score remains the same.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4210 for (;;)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4211 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4212 bc = badword[bi];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4213 gc = goodword[gi];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4214 if (bc != gc) // stop at a char that's different
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4215 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4216 if (bc == NUL) // both words end
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4217 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4218 if (score < minscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4219 minscore = score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4220 goto pop; // do next alternative
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 ++bi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4223 ++gi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4224 }
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 if (gc == NUL) // goodword ends, delete badword chars
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4227 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4228 do
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 ((score += SCORE_DEL) >= minscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4231 goto pop; // do next alternative
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4232 } while (badword[++bi] != NUL);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4233 minscore = score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4234 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4235 else if (bc == NUL) // badword ends, insert badword chars
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4236 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4237 do
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 if ((score += SCORE_INS) >= minscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4240 goto pop; // do next alternative
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4241 } while (goodword[++gi] != NUL);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4242 minscore = score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4243 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4244 else // both words continue
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4245 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4246 // 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	4247 // that may lead to a lower score than "minscore".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4248 // round 0: try deleting a char from badword
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4249 // round 1: try inserting a char in badword
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4250 for (round = 0; round <= 1; ++round)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4251 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4252 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4253 if (score_off < minscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4254 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4255 if (score_off + SCORE_EDIT_MIN >= minscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4256 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4257 // Near the limit, rest of the words must match. We
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4258 // 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	4259 // onto the stack.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4260 bi2 = bi + 1 - round;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4261 gi2 = gi + round;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4262 while (goodword[gi2] == badword[bi2])
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4263 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4264 if (goodword[gi2] == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4265 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4266 minscore = score_off;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4267 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4268 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4269 ++bi2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4270 ++gi2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4271 }
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 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4274 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4275 // try deleting/inserting a character later
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4276 stack[stackidx].badi = bi + 1 - round;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4277 stack[stackidx].goodi = gi + round;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4278 stack[stackidx].score = score_off;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4279 ++stackidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4280 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4281 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4282 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4283
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4284 if (score + SCORE_SWAP < minscore)
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 // If swapping two characters makes a match then the
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4287 // substitution is more expensive, thus there is no need to
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4288 // try both.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4289 if (gc == badword[bi + 1] && bc == goodword[gi + 1])
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4290 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4291 // Swap two characters, that is: skip them.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4292 gi += 2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4293 bi += 2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4294 score += SCORE_SWAP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4295 continue;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4296 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4297 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4298
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4299 // Substitute one character for another which is the same
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4300 // thing as deleting a character from both goodword and badword.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4301 // 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	4302 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4303 score += SCORE_ICASE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4304 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4305 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4306 // For a similar character use SCORE_SIMILAR.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4307 if (slang != NULL
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4308 && slang->sl_has_map
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4309 && similar_chars(slang, gc, bc))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4310 score += SCORE_SIMILAR;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4311 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4312 score += SCORE_SUBST;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4313 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4314
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4315 if (score < minscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4316 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4317 // Do the substitution.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4318 ++gi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4319 ++bi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4320 continue;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4321 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4322 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4323 pop:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4324 // 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	4325 if (stackidx == 0) // stack is empty, finished
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4326 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4327
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4328 // pop an item from the stack
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4329 --stackidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4330 gi = stack[stackidx].goodi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4331 bi = stack[stackidx].badi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4332 score = stack[stackidx].score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4333 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4334
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4335 // 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	4336 // 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	4337 // bonus.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4338 if (minscore > limit)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4339 return SCORE_MAXMAX;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4340 return minscore;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4341 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4342
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4343 /*
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4344 * Multi-byte version of spell_edit_score_limit().
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4345 * Keep it in sync with the above!
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 static int
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4348 spell_edit_score_limit_w(
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4349 slang_T *slang,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4350 char_u *badword,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4351 char_u *goodword,
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4352 int limit)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4353 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4354 limitscore_T stack[10]; // allow for over 3 * 2 edits
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4355 int stackidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4356 int bi, gi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4357 int bi2, gi2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4358 int bc, gc;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4359 int score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4360 int score_off;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4361 int minscore;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4362 int round;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4363 char_u *p;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4364 int wbadword[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4365 int wgoodword[MAXWLEN];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4366
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4367 // 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	4368 // int array for easy access.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4369 bi = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4370 for (p = badword; *p != NUL; )
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4371 wbadword[bi++] = mb_cptr2char_adv(&p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4372 wbadword[bi++] = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4373 gi = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4374 for (p = goodword; *p != NUL; )
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4375 wgoodword[gi++] = mb_cptr2char_adv(&p);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4376 wgoodword[gi++] = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4377
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4378 // 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	4379 // 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	4380 // When there is a difference try several alternatives. Each alternative
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4381 // 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	4382 // 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	4383 // 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	4384 // possible score is stored in "minscore".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4385 stackidx = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4386 bi = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4387 gi = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4388 score = 0;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4389 minscore = limit + 1;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4390
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4391 for (;;)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4392 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4393 // Skip over an equal part, score remains the same.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4394 for (;;)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4395 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4396 bc = wbadword[bi];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4397 gc = wgoodword[gi];
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4398
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4399 if (bc != gc) // stop at a char that's different
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4400 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4401 if (bc == NUL) // both words end
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 (score < minscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4404 minscore = score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4405 goto pop; // do next alternative
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 ++bi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4408 ++gi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4409 }
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 if (gc == NUL) // goodword ends, delete badword chars
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4412 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4413 do
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 ((score += SCORE_DEL) >= minscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4416 goto pop; // do next alternative
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4417 } while (wbadword[++bi] != NUL);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4418 minscore = score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4419 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4420 else if (bc == NUL) // badword ends, insert badword chars
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4421 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4422 do
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 if ((score += SCORE_INS) >= minscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4425 goto pop; // do next alternative
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4426 } while (wgoodword[++gi] != NUL);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4427 minscore = score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4428 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4429 else // both words continue
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4430 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4431 // 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	4432 // that may lead to a lower score than "minscore".
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4433 // round 0: try deleting a char from badword
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4434 // round 1: try inserting a char in badword
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4435 for (round = 0; round <= 1; ++round)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4436 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4437 score_off = score + (round == 0 ? SCORE_DEL : SCORE_INS);
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4438 if (score_off < minscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4439 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4440 if (score_off + SCORE_EDIT_MIN >= minscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4441 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4442 // Near the limit, rest of the words must match. We
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4443 // 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	4444 // onto the stack.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4445 bi2 = bi + 1 - round;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4446 gi2 = gi + round;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4447 while (wgoodword[gi2] == wbadword[bi2])
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4448 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4449 if (wgoodword[gi2] == NUL)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4450 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4451 minscore = score_off;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4452 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4453 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4454 ++bi2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4455 ++gi2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4456 }
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 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4459 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4460 // try deleting a character from badword later
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4461 stack[stackidx].badi = bi + 1 - round;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4462 stack[stackidx].goodi = gi + round;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4463 stack[stackidx].score = score_off;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4464 ++stackidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4465 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4466 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4467 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4468
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4469 if (score + SCORE_SWAP < minscore)
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 // If swapping two characters makes a match then the
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4472 // substitution is more expensive, thus there is no need to
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4473 // try both.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4474 if (gc == wbadword[bi + 1] && bc == wgoodword[gi + 1])
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4475 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4476 // Swap two characters, that is: skip them.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4477 gi += 2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4478 bi += 2;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4479 score += SCORE_SWAP;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4480 continue;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4481 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4482 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4483
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4484 // Substitute one character for another which is the same
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4485 // thing as deleting a character from both goodword and badword.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4486 // 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	4487 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4488 score += SCORE_ICASE;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4489 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4490 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4491 // For a similar character use SCORE_SIMILAR.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4492 if (slang != NULL
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4493 && slang->sl_has_map
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4494 && similar_chars(slang, gc, bc))
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4495 score += SCORE_SIMILAR;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4496 else
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4497 score += SCORE_SUBST;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4498 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4499
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4500 if (score < minscore)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4501 {
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4502 // Do the substitution.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4503 ++gi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4504 ++bi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4505 continue;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4506 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4507 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4508 pop:
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4509 // 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	4510 if (stackidx == 0) // stack is empty, finished
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4511 break;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4512
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4513 // pop an item from the stack
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4514 --stackidx;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4515 gi = stack[stackidx].goodi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4516 bi = stack[stackidx].badi;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4517 score = stack[stackidx].score;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4518 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4519
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4520 // 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	4521 // 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	4522 // bonus.
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4523 if (minscore > limit)
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4524 return SCORE_MAXMAX;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4525 return minscore;
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4526 }
695b50472e85 Fix line endings issue Christian Brabandt <cb@256bit.org> parents: 32669 diff changeset	4527 #endif // FEAT_SPELL

Mercurial > vim

annotate src/spellsuggest.c @ 34051:f96f16107d61