Mercurial > vim
annotate src/regexp_nfa.c @ 4682:2f51ee8825db v7.3.1088
updated for version 7.3.1088
Problem: New regexp engine: \@<= and \@<! are not implemented.
Solution: Implement look-behind matching. Fix off-by-one error in old
regexp engine.
author | Bram Moolenaar <bram@vim.org> |
---|---|
date | Sat, 01 Jun 2013 19:54:43 +0200 |
parents | 4d92b873acef |
children | 8db697ae406a |
rev | line source |
---|---|
4444 | 1 /* vi:set ts=8 sts=4 sw=4: |
2 * | |
3 * NFA regular expression implementation. | |
4 * | |
5 * This file is included in "regexp.c". | |
6 */ | |
7 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
8 /* |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
9 * Logging of NFA engine. |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
10 * |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
11 * The NFA engine can write four log files: |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
12 * - Error log: Contains NFA engine's fatal errors. |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
13 * - Dump log: Contains compiled NFA state machine's information. |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
14 * - Run log: Contains information of matching procedure. |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
15 * - Debug log: Contains detailed information of matching procedure. Can be |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
16 * disabled by undefining NFA_REGEXP_DEBUG_LOG. |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
17 * The first one can also be used without debug mode. |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
18 * The last three are enabled when compiled as debug mode and individually |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
19 * disabled by commenting them out. |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
20 * The log files can get quite big! |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
21 * Do disable all of this when compiling Vim for debugging, undefine DEBUG in |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
22 * regexp.c |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
23 */ |
4444 | 24 #ifdef DEBUG |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
25 # define NFA_REGEXP_ERROR_LOG "nfa_regexp_error.log" |
4444 | 26 # define ENABLE_LOG |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
27 # define NFA_REGEXP_DUMP_LOG "nfa_regexp_dump.log" |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
28 # define NFA_REGEXP_RUN_LOG "nfa_regexp_run.log" |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
29 # define NFA_REGEXP_DEBUG_LOG "nfa_regexp_debug.log" |
4444 | 30 #endif |
31 | |
32 enum | |
33 { | |
34 NFA_SPLIT = -1024, | |
35 NFA_MATCH, | |
36 NFA_SKIP_CHAR, /* matches a 0-length char */ | |
37 NFA_END_NEG_RANGE, /* Used when expanding [^ab] */ | |
38 | |
39 NFA_CONCAT, | |
40 NFA_OR, | |
4675
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
41 NFA_STAR, /* greedy * */ |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
42 NFA_STAR_NONGREEDY, /* non-greedy * */ |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
43 NFA_QUEST, /* greedy \? */ |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
44 NFA_QUEST_NONGREEDY, /* non-greedy \? */ |
4444 | 45 NFA_NOT, /* used for [^ab] negated char ranges */ |
46 | |
47 NFA_BOL, /* ^ Begin line */ | |
48 NFA_EOL, /* $ End line */ | |
49 NFA_BOW, /* \< Begin word */ | |
50 NFA_EOW, /* \> End word */ | |
51 NFA_BOF, /* \%^ Begin file */ | |
52 NFA_EOF, /* \%$ End file */ | |
53 NFA_NEWL, | |
54 NFA_ZSTART, /* Used for \zs */ | |
55 NFA_ZEND, /* Used for \ze */ | |
56 NFA_NOPEN, /* Start of subexpression marked with \%( */ | |
57 NFA_NCLOSE, /* End of subexpr. marked with \%( ... \) */ | |
58 NFA_START_INVISIBLE, | |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
59 NFA_START_INVISIBLE_BEFORE, |
4444 | 60 NFA_END_INVISIBLE, |
61 NFA_COMPOSING, /* Next nodes in NFA are part of the | |
62 composing multibyte char */ | |
63 NFA_END_COMPOSING, /* End of a composing char in the NFA */ | |
64 | |
65 /* The following are used only in the postfix form, not in the NFA */ | |
66 NFA_PREV_ATOM_NO_WIDTH, /* Used for \@= */ | |
67 NFA_PREV_ATOM_NO_WIDTH_NEG, /* Used for \@! */ | |
68 NFA_PREV_ATOM_JUST_BEFORE, /* Used for \@<= */ | |
69 NFA_PREV_ATOM_JUST_BEFORE_NEG, /* Used for \@<! */ | |
70 NFA_PREV_ATOM_LIKE_PATTERN, /* Used for \@> */ | |
71 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
72 NFA_BACKREF1, /* \1 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
73 NFA_BACKREF2, /* \2 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
74 NFA_BACKREF3, /* \3 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
75 NFA_BACKREF4, /* \4 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
76 NFA_BACKREF5, /* \5 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
77 NFA_BACKREF6, /* \6 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
78 NFA_BACKREF7, /* \7 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
79 NFA_BACKREF8, /* \8 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
80 NFA_BACKREF9, /* \9 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
81 NFA_SKIP, /* Skip characters */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
82 |
4444 | 83 NFA_MOPEN, |
84 NFA_MCLOSE = NFA_MOPEN + NSUBEXP, | |
85 | |
86 /* NFA_FIRST_NL */ | |
87 NFA_ANY = NFA_MCLOSE + NSUBEXP, /* Match any one character. */ | |
88 NFA_ANYOF, /* Match any character in this string. */ | |
89 NFA_ANYBUT, /* Match any character not in this string. */ | |
90 NFA_IDENT, /* Match identifier char */ | |
91 NFA_SIDENT, /* Match identifier char but no digit */ | |
92 NFA_KWORD, /* Match keyword char */ | |
93 NFA_SKWORD, /* Match word char but no digit */ | |
94 NFA_FNAME, /* Match file name char */ | |
95 NFA_SFNAME, /* Match file name char but no digit */ | |
96 NFA_PRINT, /* Match printable char */ | |
97 NFA_SPRINT, /* Match printable char but no digit */ | |
98 NFA_WHITE, /* Match whitespace char */ | |
99 NFA_NWHITE, /* Match non-whitespace char */ | |
100 NFA_DIGIT, /* Match digit char */ | |
101 NFA_NDIGIT, /* Match non-digit char */ | |
102 NFA_HEX, /* Match hex char */ | |
103 NFA_NHEX, /* Match non-hex char */ | |
104 NFA_OCTAL, /* Match octal char */ | |
105 NFA_NOCTAL, /* Match non-octal char */ | |
106 NFA_WORD, /* Match word char */ | |
107 NFA_NWORD, /* Match non-word char */ | |
108 NFA_HEAD, /* Match head char */ | |
109 NFA_NHEAD, /* Match non-head char */ | |
110 NFA_ALPHA, /* Match alpha char */ | |
111 NFA_NALPHA, /* Match non-alpha char */ | |
112 NFA_LOWER, /* Match lowercase char */ | |
113 NFA_NLOWER, /* Match non-lowercase char */ | |
114 NFA_UPPER, /* Match uppercase char */ | |
115 NFA_NUPPER, /* Match non-uppercase char */ | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
116 |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
117 NFA_CURSOR, /* Match cursor pos */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
118 NFA_LNUM, /* Match line number */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
119 NFA_LNUM_GT, /* Match > line number */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
120 NFA_LNUM_LT, /* Match < line number */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
121 NFA_COL, /* Match cursor column */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
122 NFA_COL_GT, /* Match > cursor column */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
123 NFA_COL_LT, /* Match < cursor column */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
124 NFA_VCOL, /* Match cursor virtual column */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
125 NFA_VCOL_GT, /* Match > cursor virtual column */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
126 NFA_VCOL_LT, /* Match < cursor virtual column */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
127 |
4444 | 128 NFA_FIRST_NL = NFA_ANY + ADD_NL, |
129 NFA_LAST_NL = NFA_NUPPER + ADD_NL, | |
130 | |
131 /* Character classes [:alnum:] etc */ | |
132 NFA_CLASS_ALNUM, | |
133 NFA_CLASS_ALPHA, | |
134 NFA_CLASS_BLANK, | |
135 NFA_CLASS_CNTRL, | |
136 NFA_CLASS_DIGIT, | |
137 NFA_CLASS_GRAPH, | |
138 NFA_CLASS_LOWER, | |
139 NFA_CLASS_PRINT, | |
140 NFA_CLASS_PUNCT, | |
141 NFA_CLASS_SPACE, | |
142 NFA_CLASS_UPPER, | |
143 NFA_CLASS_XDIGIT, | |
144 NFA_CLASS_TAB, | |
145 NFA_CLASS_RETURN, | |
146 NFA_CLASS_BACKSPACE, | |
147 NFA_CLASS_ESCAPE | |
148 }; | |
149 | |
150 /* Keep in sync with classchars. */ | |
151 static int nfa_classcodes[] = { | |
152 NFA_ANY, NFA_IDENT, NFA_SIDENT, NFA_KWORD,NFA_SKWORD, | |
153 NFA_FNAME, NFA_SFNAME, NFA_PRINT, NFA_SPRINT, | |
154 NFA_WHITE, NFA_NWHITE, NFA_DIGIT, NFA_NDIGIT, | |
155 NFA_HEX, NFA_NHEX, NFA_OCTAL, NFA_NOCTAL, | |
156 NFA_WORD, NFA_NWORD, NFA_HEAD, NFA_NHEAD, | |
157 NFA_ALPHA, NFA_NALPHA, NFA_LOWER, NFA_NLOWER, | |
158 NFA_UPPER, NFA_NUPPER | |
159 }; | |
160 | |
161 static char_u e_misplaced[] = N_("E866: (NFA regexp) Misplaced %c"); | |
162 | |
163 /* | |
164 * NFA errors can be of 3 types: | |
165 * *** NFA runtime errors, when something unknown goes wrong. The NFA fails | |
166 * silently and revert the to backtracking engine. | |
167 * syntax_error = FALSE; | |
168 * *** Regexp syntax errors, when the input regexp is not syntactically correct. | |
169 * The NFA engine displays an error message, and nothing else happens. | |
170 * syntax_error = TRUE | |
171 * *** Unsupported features, when the input regexp uses an operator that is not | |
172 * implemented in the NFA. The NFA engine fails silently, and reverts to the | |
173 * old backtracking engine. | |
174 * syntax_error = FALSE | |
175 * "The NFA fails" means that "compiling the regexp with the NFA fails": | |
176 * nfa_regcomp() returns FAIL. | |
177 */ | |
178 static int syntax_error = FALSE; | |
179 | |
180 /* NFA regexp \ze operator encountered. */ | |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
181 static int nfa_has_zend; |
4444 | 182 |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
183 /* NFA regexp \1 .. \9 encountered. */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
184 static int nfa_has_backref; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
185 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
186 /* Number of sub expressions actually being used during execution. 1 if only |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
187 * the whole match (subexpr 0) is used. */ |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
188 static int nfa_nsubexpr; |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
189 |
4444 | 190 static int *post_start; /* holds the postfix form of r.e. */ |
191 static int *post_end; | |
192 static int *post_ptr; | |
193 | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
194 static int nstate; /* Number of states in the NFA. Also used when |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
195 * executing. */ |
4444 | 196 static int istate; /* Index in the state vector, used in new_state() */ |
197 | |
198 | |
199 static int nfa_regcomp_start __ARGS((char_u*expr, int re_flags)); | |
200 static int nfa_recognize_char_class __ARGS((char_u *start, char_u *end, int extra_newl)); | |
201 static int nfa_emit_equi_class __ARGS((int c, int neg)); | |
202 static int nfa_regatom __ARGS((void)); | |
203 static int nfa_regpiece __ARGS((void)); | |
204 static int nfa_regconcat __ARGS((void)); | |
205 static int nfa_regbranch __ARGS((void)); | |
206 static int nfa_reg __ARGS((int paren)); | |
207 #ifdef DEBUG | |
208 static void nfa_set_code __ARGS((int c)); | |
209 static void nfa_postfix_dump __ARGS((char_u *expr, int retval)); | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
210 static void nfa_print_state __ARGS((FILE *debugf, nfa_state_T *state)); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
211 static void nfa_print_state2 __ARGS((FILE *debugf, nfa_state_T *state, garray_T *indent)); |
4444 | 212 static void nfa_dump __ARGS((nfa_regprog_T *prog)); |
213 #endif | |
214 static int *re2post __ARGS((void)); | |
215 static nfa_state_T *new_state __ARGS((int c, nfa_state_T *out, nfa_state_T *out1)); | |
216 static nfa_state_T *post2nfa __ARGS((int *postfix, int *end, int nfa_calc_size)); | |
217 static int check_char_class __ARGS((int class, int c)); | |
218 static void st_error __ARGS((int *postfix, int *end, int *p)); | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
219 static void nfa_set_neg_listids __ARGS((nfa_state_T *start)); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
220 static void nfa_set_null_listids __ARGS((nfa_state_T *start)); |
4444 | 221 static void nfa_save_listids __ARGS((nfa_state_T *start, int *list)); |
222 static void nfa_restore_listids __ARGS((nfa_state_T *start, int *list)); | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
223 static int nfa_re_num_cmp __ARGS((long_u val, int op, long_u pos)); |
4444 | 224 static long nfa_regtry __ARGS((nfa_state_T *start, colnr_T col)); |
225 static long nfa_regexec_both __ARGS((char_u *line, colnr_T col)); | |
226 static regprog_T *nfa_regcomp __ARGS((char_u *expr, int re_flags)); | |
227 static int nfa_regexec __ARGS((regmatch_T *rmp, char_u *line, colnr_T col)); | |
228 static long nfa_regexec_multi __ARGS((regmmatch_T *rmp, win_T *win, buf_T *buf, linenr_T lnum, colnr_T col, proftime_T *tm)); | |
229 | |
230 /* helper functions used when doing re2post() ... regatom() parsing */ | |
231 #define EMIT(c) do { \ | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
232 if (post_ptr >= post_end && realloc_post_list() == FAIL) \ |
4444 | 233 return FAIL; \ |
234 *post_ptr++ = c; \ | |
235 } while (0) | |
236 | |
237 /* | |
238 * Initialize internal variables before NFA compilation. | |
239 * Return OK on success, FAIL otherwise. | |
240 */ | |
241 static int | |
242 nfa_regcomp_start(expr, re_flags) | |
243 char_u *expr; | |
244 int re_flags; /* see vim_regcomp() */ | |
245 { | |
4458 | 246 size_t postfix_size; |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
247 int nstate_max; |
4444 | 248 |
249 nstate = 0; | |
250 istate = 0; | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
251 /* A reasonable estimation for maximum size */ |
4673
05d57d7c2d55
updated for version 7.3.1084
Bram Moolenaar <bram@vim.org>
parents:
4671
diff
changeset
|
252 nstate_max = (int)(STRLEN(expr) + 1) * 25; |
4444 | 253 |
4454 | 254 /* Some items blow up in size, such as [A-z]. Add more space for that. |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
255 * When it is still not enough realloc_post_list() will be used. */ |
4458 | 256 nstate_max += 1000; |
4454 | 257 |
258 /* Size for postfix representation of expr. */ | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
259 postfix_size = sizeof(int) * nstate_max; |
4454 | 260 |
4444 | 261 post_start = (int *)lalloc(postfix_size, TRUE); |
262 if (post_start == NULL) | |
263 return FAIL; | |
264 vim_memset(post_start, 0, postfix_size); | |
265 post_ptr = post_start; | |
4454 | 266 post_end = post_start + nstate_max; |
4444 | 267 nfa_has_zend = FALSE; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
268 nfa_has_backref = FALSE; |
4444 | 269 |
270 regcomp_start(expr, re_flags); | |
271 | |
272 return OK; | |
273 } | |
274 | |
275 /* | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
276 * Allocate more space for post_start. Called when |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
277 * running above the estimated number of states. |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
278 */ |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
279 static int |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
280 realloc_post_list() |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
281 { |
4667
9e7ef781d494
updated for version 7.3.1081
Bram Moolenaar <bram@vim.org>
parents:
4661
diff
changeset
|
282 int nstate_max = (int)(post_end - post_start); |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
283 int new_max = nstate_max + 1000; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
284 int *new_start; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
285 int *old_start; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
286 |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
287 new_start = (int *)lalloc(new_max * sizeof(int), TRUE); |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
288 if (new_start == NULL) |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
289 return FAIL; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
290 mch_memmove(new_start, post_start, nstate_max * sizeof(int)); |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
291 vim_memset(new_start + nstate_max, 0, 1000 * sizeof(int)); |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
292 old_start = post_start; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
293 post_start = new_start; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
294 post_ptr = new_start + (post_ptr - old_start); |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
295 post_end = post_start + new_max; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
296 vim_free(old_start); |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
297 return OK; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
298 } |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
299 |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
300 /* |
4444 | 301 * Search between "start" and "end" and try to recognize a |
302 * character class in expanded form. For example [0-9]. | |
303 * On success, return the id the character class to be emitted. | |
304 * On failure, return 0 (=FAIL) | |
305 * Start points to the first char of the range, while end should point | |
306 * to the closing brace. | |
307 */ | |
308 static int | |
309 nfa_recognize_char_class(start, end, extra_newl) | |
310 char_u *start; | |
311 char_u *end; | |
312 int extra_newl; | |
313 { | |
314 int i; | |
315 /* Each of these variables takes up a char in "config[]", | |
316 * in the order they are here. */ | |
317 int not = FALSE, af = FALSE, AF = FALSE, az = FALSE, AZ = FALSE, | |
318 o7 = FALSE, o9 = FALSE, underscore = FALSE, newl = FALSE; | |
319 char_u *p; | |
320 #define NCONFIGS 16 | |
321 int classid[NCONFIGS] = { | |
322 NFA_DIGIT, NFA_NDIGIT, NFA_HEX, NFA_NHEX, | |
323 NFA_OCTAL, NFA_NOCTAL, NFA_WORD, NFA_NWORD, | |
324 NFA_HEAD, NFA_NHEAD, NFA_ALPHA, NFA_NALPHA, | |
325 NFA_LOWER, NFA_NLOWER, NFA_UPPER, NFA_NUPPER | |
326 }; | |
4450 | 327 char_u myconfig[10]; |
4444 | 328 char_u config[NCONFIGS][9] = { |
329 "000000100", /* digit */ | |
330 "100000100", /* non digit */ | |
331 "011000100", /* hex-digit */ | |
332 "111000100", /* non hex-digit */ | |
333 "000001000", /* octal-digit */ | |
334 "100001000", /* [^0-7] */ | |
335 "000110110", /* [0-9A-Za-z_] */ | |
336 "100110110", /* [^0-9A-Za-z_] */ | |
337 "000110010", /* head of word */ | |
338 "100110010", /* not head of word */ | |
339 "000110000", /* alphabetic char a-z */ | |
340 "100110000", /* non alphabetic char */ | |
341 "000100000", /* lowercase letter */ | |
342 "100100000", /* non lowercase */ | |
343 "000010000", /* uppercase */ | |
344 "100010000" /* non uppercase */ | |
345 }; | |
346 | |
347 if (extra_newl == TRUE) | |
348 newl = TRUE; | |
349 | |
350 if (*end != ']') | |
351 return FAIL; | |
352 p = start; | |
353 if (*p == '^') | |
354 { | |
355 not = TRUE; | |
356 p ++; | |
357 } | |
358 | |
359 while (p < end) | |
360 { | |
361 if (p + 2 < end && *(p + 1) == '-') | |
362 { | |
363 switch (*p) | |
364 { | |
365 case '0': | |
366 if (*(p + 2) == '9') | |
367 { | |
368 o9 = TRUE; | |
369 break; | |
370 } | |
371 else | |
372 if (*(p + 2) == '7') | |
373 { | |
374 o7 = TRUE; | |
375 break; | |
376 } | |
377 case 'a': | |
378 if (*(p + 2) == 'z') | |
379 { | |
380 az = TRUE; | |
381 break; | |
382 } | |
383 else | |
384 if (*(p + 2) == 'f') | |
385 { | |
386 af = TRUE; | |
387 break; | |
388 } | |
389 case 'A': | |
390 if (*(p + 2) == 'Z') | |
391 { | |
392 AZ = TRUE; | |
393 break; | |
394 } | |
395 else | |
396 if (*(p + 2) == 'F') | |
397 { | |
398 AF = TRUE; | |
399 break; | |
400 } | |
401 /* FALLTHROUGH */ | |
402 default: | |
403 return FAIL; | |
404 } | |
405 p += 3; | |
406 } | |
407 else if (p + 1 < end && *p == '\\' && *(p + 1) == 'n') | |
408 { | |
409 newl = TRUE; | |
410 p += 2; | |
411 } | |
412 else if (*p == '_') | |
413 { | |
414 underscore = TRUE; | |
415 p ++; | |
416 } | |
417 else if (*p == '\n') | |
418 { | |
419 newl = TRUE; | |
420 p ++; | |
421 } | |
422 else | |
423 return FAIL; | |
424 } /* while (p < end) */ | |
425 | |
426 if (p != end) | |
427 return FAIL; | |
428 | |
429 /* build the config that represents the ranges we gathered */ | |
430 STRCPY(myconfig, "000000000"); | |
431 if (not == TRUE) | |
432 myconfig[0] = '1'; | |
433 if (af == TRUE) | |
434 myconfig[1] = '1'; | |
435 if (AF == TRUE) | |
436 myconfig[2] = '1'; | |
437 if (az == TRUE) | |
438 myconfig[3] = '1'; | |
439 if (AZ == TRUE) | |
440 myconfig[4] = '1'; | |
441 if (o7 == TRUE) | |
442 myconfig[5] = '1'; | |
443 if (o9 == TRUE) | |
444 myconfig[6] = '1'; | |
445 if (underscore == TRUE) | |
446 myconfig[7] = '1'; | |
447 if (newl == TRUE) | |
448 { | |
449 myconfig[8] = '1'; | |
450 extra_newl = ADD_NL; | |
451 } | |
452 /* try to recognize character classes */ | |
453 for (i = 0; i < NCONFIGS; i++) | |
4450 | 454 if (STRNCMP(myconfig, config[i], 8) == 0) |
4444 | 455 return classid[i] + extra_newl; |
456 | |
457 /* fallthrough => no success so far */ | |
458 return FAIL; | |
459 | |
460 #undef NCONFIGS | |
461 } | |
462 | |
463 /* | |
464 * Produce the bytes for equivalence class "c". | |
465 * Currently only handles latin1, latin9 and utf-8. | |
466 * Emits bytes in postfix notation: 'a,b,NFA_OR,c,NFA_OR' is | |
467 * equivalent to 'a OR b OR c' | |
468 * | |
469 * NOTE! When changing this function, also update reg_equi_class() | |
470 */ | |
471 static int | |
472 nfa_emit_equi_class(c, neg) | |
473 int c; | |
474 int neg; | |
475 { | |
476 int first = TRUE; | |
477 int glue = neg == TRUE ? NFA_CONCAT : NFA_OR; | |
478 #define EMIT2(c) \ | |
479 EMIT(c); \ | |
480 if (neg == TRUE) { \ | |
481 EMIT(NFA_NOT); \ | |
482 } \ | |
483 if (first == FALSE) \ | |
484 EMIT(glue); \ | |
485 else \ | |
486 first = FALSE; \ | |
487 | |
488 #ifdef FEAT_MBYTE | |
489 if (enc_utf8 || STRCMP(p_enc, "latin1") == 0 | |
490 || STRCMP(p_enc, "iso-8859-15") == 0) | |
491 #endif | |
492 { | |
493 switch (c) | |
494 { | |
495 case 'A': case '\300': case '\301': case '\302': | |
496 case '\303': case '\304': case '\305': | |
497 EMIT2('A'); EMIT2('\300'); EMIT2('\301'); | |
498 EMIT2('\302'); EMIT2('\303'); EMIT2('\304'); | |
499 EMIT2('\305'); | |
500 return OK; | |
501 | |
502 case 'C': case '\307': | |
503 EMIT2('C'); EMIT2('\307'); | |
504 return OK; | |
505 | |
506 case 'E': case '\310': case '\311': case '\312': case '\313': | |
507 EMIT2('E'); EMIT2('\310'); EMIT2('\311'); | |
508 EMIT2('\312'); EMIT2('\313'); | |
509 return OK; | |
510 | |
511 case 'I': case '\314': case '\315': case '\316': case '\317': | |
512 EMIT2('I'); EMIT2('\314'); EMIT2('\315'); | |
513 EMIT2('\316'); EMIT2('\317'); | |
514 return OK; | |
515 | |
516 case 'N': case '\321': | |
517 EMIT2('N'); EMIT2('\321'); | |
518 return OK; | |
519 | |
520 case 'O': case '\322': case '\323': case '\324': case '\325': | |
521 case '\326': | |
522 EMIT2('O'); EMIT2('\322'); EMIT2('\323'); | |
523 EMIT2('\324'); EMIT2('\325'); EMIT2('\326'); | |
524 return OK; | |
525 | |
526 case 'U': case '\331': case '\332': case '\333': case '\334': | |
527 EMIT2('U'); EMIT2('\331'); EMIT2('\332'); | |
528 EMIT2('\333'); EMIT2('\334'); | |
529 return OK; | |
530 | |
531 case 'Y': case '\335': | |
532 EMIT2('Y'); EMIT2('\335'); | |
533 return OK; | |
534 | |
535 case 'a': case '\340': case '\341': case '\342': | |
536 case '\343': case '\344': case '\345': | |
537 EMIT2('a'); EMIT2('\340'); EMIT2('\341'); | |
538 EMIT2('\342'); EMIT2('\343'); EMIT2('\344'); | |
539 EMIT2('\345'); | |
540 return OK; | |
541 | |
542 case 'c': case '\347': | |
543 EMIT2('c'); EMIT2('\347'); | |
544 return OK; | |
545 | |
546 case 'e': case '\350': case '\351': case '\352': case '\353': | |
547 EMIT2('e'); EMIT2('\350'); EMIT2('\351'); | |
548 EMIT2('\352'); EMIT2('\353'); | |
549 return OK; | |
550 | |
551 case 'i': case '\354': case '\355': case '\356': case '\357': | |
552 EMIT2('i'); EMIT2('\354'); EMIT2('\355'); | |
553 EMIT2('\356'); EMIT2('\357'); | |
554 return OK; | |
555 | |
556 case 'n': case '\361': | |
557 EMIT2('n'); EMIT2('\361'); | |
558 return OK; | |
559 | |
560 case 'o': case '\362': case '\363': case '\364': case '\365': | |
561 case '\366': | |
562 EMIT2('o'); EMIT2('\362'); EMIT2('\363'); | |
563 EMIT2('\364'); EMIT2('\365'); EMIT2('\366'); | |
564 return OK; | |
565 | |
566 case 'u': case '\371': case '\372': case '\373': case '\374': | |
567 EMIT2('u'); EMIT2('\371'); EMIT2('\372'); | |
568 EMIT2('\373'); EMIT2('\374'); | |
569 return OK; | |
570 | |
571 case 'y': case '\375': case '\377': | |
572 EMIT2('y'); EMIT2('\375'); EMIT2('\377'); | |
573 return OK; | |
574 | |
575 default: | |
576 return FAIL; | |
577 } | |
578 } | |
579 | |
580 EMIT(c); | |
581 return OK; | |
582 #undef EMIT2 | |
583 } | |
584 | |
585 /* | |
586 * Code to parse regular expression. | |
587 * | |
588 * We try to reuse parsing functions in regexp.c to | |
589 * minimize surprise and keep the syntax consistent. | |
590 */ | |
591 | |
592 /* | |
593 * Parse the lowest level. | |
594 * | |
595 * An atom can be one of a long list of items. Many atoms match one character | |
596 * in the text. It is often an ordinary character or a character class. | |
597 * Braces can be used to make a pattern into an atom. The "\z(\)" construct | |
598 * is only for syntax highlighting. | |
599 * | |
600 * atom ::= ordinary-atom | |
601 * or \( pattern \) | |
602 * or \%( pattern \) | |
603 * or \z( pattern \) | |
604 */ | |
605 static int | |
606 nfa_regatom() | |
607 { | |
608 int c; | |
609 int charclass; | |
610 int equiclass; | |
611 int collclass; | |
612 int got_coll_char; | |
613 char_u *p; | |
614 char_u *endp; | |
615 #ifdef FEAT_MBYTE | |
616 char_u *old_regparse = regparse; | |
617 #endif | |
618 int extra = 0; | |
619 int first; | |
620 int emit_range; | |
621 int negated; | |
622 int result; | |
623 int startc = -1; | |
624 int endc = -1; | |
625 int oldstartc = -1; | |
626 int cpo_lit; /* 'cpoptions' contains 'l' flag */ | |
627 int cpo_bsl; /* 'cpoptions' contains '\' flag */ | |
628 int glue; /* ID that will "glue" nodes together */ | |
629 | |
630 cpo_lit = vim_strchr(p_cpo, CPO_LITERAL) != NULL; | |
631 cpo_bsl = vim_strchr(p_cpo, CPO_BACKSL) != NULL; | |
632 | |
633 c = getchr(); | |
634 switch (c) | |
635 { | |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
636 case NUL: |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
637 syntax_error = TRUE; |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
638 EMSG_RET_FAIL(_("E865: (NFA) Regexp end encountered prematurely")); |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
639 |
4444 | 640 case Magic('^'): |
641 EMIT(NFA_BOL); | |
642 break; | |
643 | |
644 case Magic('$'): | |
645 EMIT(NFA_EOL); | |
646 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
647 had_eol = TRUE; | |
648 #endif | |
649 break; | |
650 | |
651 case Magic('<'): | |
652 EMIT(NFA_BOW); | |
653 break; | |
654 | |
655 case Magic('>'): | |
656 EMIT(NFA_EOW); | |
657 break; | |
658 | |
659 case Magic('_'): | |
660 c = no_Magic(getchr()); | |
661 if (c == '^') /* "\_^" is start-of-line */ | |
662 { | |
663 EMIT(NFA_BOL); | |
664 break; | |
665 } | |
666 if (c == '$') /* "\_$" is end-of-line */ | |
667 { | |
668 EMIT(NFA_EOL); | |
669 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
670 had_eol = TRUE; | |
671 #endif | |
672 break; | |
673 } | |
674 | |
675 extra = ADD_NL; | |
676 | |
677 /* "\_[" is collection plus newline */ | |
678 if (c == '[') | |
4517
9a2183bd8295
updated for version 7.3.1006
Bram Moolenaar <bram@vim.org>
parents:
4515
diff
changeset
|
679 goto collection; |
4444 | 680 |
681 /* "\_x" is character class plus newline */ | |
682 /*FALLTHROUGH*/ | |
683 | |
684 /* | |
685 * Character classes. | |
686 */ | |
687 case Magic('.'): | |
688 case Magic('i'): | |
689 case Magic('I'): | |
690 case Magic('k'): | |
691 case Magic('K'): | |
692 case Magic('f'): | |
693 case Magic('F'): | |
694 case Magic('p'): | |
695 case Magic('P'): | |
696 case Magic('s'): | |
697 case Magic('S'): | |
698 case Magic('d'): | |
699 case Magic('D'): | |
700 case Magic('x'): | |
701 case Magic('X'): | |
702 case Magic('o'): | |
703 case Magic('O'): | |
704 case Magic('w'): | |
705 case Magic('W'): | |
706 case Magic('h'): | |
707 case Magic('H'): | |
708 case Magic('a'): | |
709 case Magic('A'): | |
710 case Magic('l'): | |
711 case Magic('L'): | |
712 case Magic('u'): | |
713 case Magic('U'): | |
714 p = vim_strchr(classchars, no_Magic(c)); | |
715 if (p == NULL) | |
716 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
717 EMSGN("INTERNAL: Unknown character class char: %ld", c); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
718 return FAIL; |
4444 | 719 } |
720 #ifdef FEAT_MBYTE | |
721 /* When '.' is followed by a composing char ignore the dot, so that | |
722 * the composing char is matched here. */ | |
723 if (enc_utf8 && c == Magic('.') && utf_iscomposing(peekchr())) | |
724 { | |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
725 old_regparse = regparse; |
4444 | 726 c = getchr(); |
727 goto nfa_do_multibyte; | |
728 } | |
729 #endif | |
730 EMIT(nfa_classcodes[p - classchars]); | |
731 if (extra == ADD_NL) | |
732 { | |
733 EMIT(NFA_NEWL); | |
734 EMIT(NFA_OR); | |
735 regflags |= RF_HASNL; | |
736 } | |
737 break; | |
738 | |
739 case Magic('n'): | |
740 if (reg_string) | |
741 /* In a string "\n" matches a newline character. */ | |
742 EMIT(NL); | |
743 else | |
744 { | |
745 /* In buffer text "\n" matches the end of a line. */ | |
746 EMIT(NFA_NEWL); | |
747 regflags |= RF_HASNL; | |
748 } | |
749 break; | |
750 | |
751 case Magic('('): | |
752 if (nfa_reg(REG_PAREN) == FAIL) | |
753 return FAIL; /* cascaded error */ | |
754 break; | |
755 | |
756 case Magic('|'): | |
757 case Magic('&'): | |
758 case Magic(')'): | |
759 syntax_error = TRUE; | |
4450 | 760 EMSGN(_(e_misplaced), no_Magic(c)); |
4444 | 761 return FAIL; |
762 | |
763 case Magic('='): | |
764 case Magic('?'): | |
765 case Magic('+'): | |
766 case Magic('@'): | |
767 case Magic('*'): | |
768 case Magic('{'): | |
769 /* these should follow an atom, not form an atom */ | |
770 syntax_error = TRUE; | |
4450 | 771 EMSGN(_(e_misplaced), no_Magic(c)); |
4444 | 772 return FAIL; |
773 | |
774 case Magic('~'): /* previous substitute pattern */ | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
775 /* TODO: Not supported yet */ |
4444 | 776 return FAIL; |
777 | |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
778 case Magic('1'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
779 case Magic('2'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
780 case Magic('3'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
781 case Magic('4'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
782 case Magic('5'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
783 case Magic('6'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
784 case Magic('7'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
785 case Magic('8'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
786 case Magic('9'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
787 EMIT(NFA_BACKREF1 + (no_Magic(c) - '1')); |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
788 nfa_has_backref = TRUE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
789 break; |
4444 | 790 |
791 case Magic('z'): | |
792 c = no_Magic(getchr()); | |
793 switch (c) | |
794 { | |
795 case 's': | |
796 EMIT(NFA_ZSTART); | |
797 break; | |
798 case 'e': | |
799 EMIT(NFA_ZEND); | |
800 nfa_has_zend = TRUE; | |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
801 break; |
4444 | 802 case '1': |
803 case '2': | |
804 case '3': | |
805 case '4': | |
806 case '5': | |
807 case '6': | |
808 case '7': | |
809 case '8': | |
810 case '9': | |
811 case '(': | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
812 /* TODO: \z1...\z9 and \z( not yet supported */ |
4444 | 813 return FAIL; |
814 default: | |
815 syntax_error = TRUE; | |
4450 | 816 EMSGN(_("E867: (NFA) Unknown operator '\\z%c'"), |
4444 | 817 no_Magic(c)); |
818 return FAIL; | |
819 } | |
820 break; | |
821 | |
822 case Magic('%'): | |
823 c = no_Magic(getchr()); | |
824 switch (c) | |
825 { | |
826 /* () without a back reference */ | |
827 case '(': | |
828 if (nfa_reg(REG_NPAREN) == FAIL) | |
829 return FAIL; | |
830 EMIT(NFA_NOPEN); | |
831 break; | |
832 | |
833 case 'd': /* %d123 decimal */ | |
834 case 'o': /* %o123 octal */ | |
835 case 'x': /* %xab hex 2 */ | |
836 case 'u': /* %uabcd hex 4 */ | |
837 case 'U': /* %U1234abcd hex 8 */ | |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
838 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
839 int nr; |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
840 |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
841 switch (c) |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
842 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
843 case 'd': nr = getdecchrs(); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
844 case 'o': nr = getoctchrs(); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
845 case 'x': nr = gethexchrs(2); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
846 case 'u': nr = gethexchrs(4); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
847 case 'U': nr = gethexchrs(8); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
848 default: nr = -1; break; |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
849 } |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
850 |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
851 if (nr < 0) |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
852 EMSG2_RET_FAIL( |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
853 _("E678: Invalid character after %s%%[dxouU]"), |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
854 reg_magic == MAGIC_ALL); |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
855 /* TODO: what if a composing character follows? */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
856 EMIT(nr); |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
857 } |
4444 | 858 break; |
859 | |
860 /* Catch \%^ and \%$ regardless of where they appear in the | |
861 * pattern -- regardless of whether or not it makes sense. */ | |
862 case '^': | |
863 EMIT(NFA_BOF); | |
864 break; | |
865 | |
866 case '$': | |
867 EMIT(NFA_EOF); | |
868 break; | |
869 | |
870 case '#': | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
871 EMIT(NFA_CURSOR); |
4444 | 872 break; |
873 | |
874 case 'V': | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
875 /* TODO: not supported yet */ |
4444 | 876 return FAIL; |
877 break; | |
878 | |
879 case '[': | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
880 /* TODO: \%[abc] not supported yet */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
881 return FAIL; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
882 |
4444 | 883 default: |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
884 { |
4655
779ca415f8e1
updated for version 7.3.1075
Bram Moolenaar <bram@vim.org>
parents:
4651
diff
changeset
|
885 int n = 0; |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
886 int cmp = c; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
887 |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
888 if (c == '<' || c == '>') |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
889 c = getchr(); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
890 while (VIM_ISDIGIT(c)) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
891 { |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
892 n = n * 10 + (c - '0'); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
893 c = getchr(); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
894 } |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
895 if (c == 'l' || c == 'c' || c == 'v') |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
896 { |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
897 EMIT(n); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
898 if (c == 'l') |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
899 EMIT(cmp == '<' ? NFA_LNUM_LT : |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
900 cmp == '>' ? NFA_LNUM_GT : NFA_LNUM); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
901 else if (c == 'c') |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
902 EMIT(cmp == '<' ? NFA_COL_LT : |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
903 cmp == '>' ? NFA_COL_GT : NFA_COL); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
904 else |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
905 EMIT(cmp == '<' ? NFA_VCOL_LT : |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
906 cmp == '>' ? NFA_VCOL_GT : NFA_VCOL); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
907 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
908 } |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
909 else if (c == '\'') |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
910 /* TODO: \%'m not supported yet */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
911 return FAIL; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
912 } |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
913 syntax_error = TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
914 EMSGN(_("E867: (NFA) Unknown operator '\\%%%c'"), |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
915 no_Magic(c)); |
4444 | 916 return FAIL; |
917 } | |
918 break; | |
919 | |
920 case Magic('['): | |
4517
9a2183bd8295
updated for version 7.3.1006
Bram Moolenaar <bram@vim.org>
parents:
4515
diff
changeset
|
921 collection: |
4444 | 922 /* |
923 * Glue is emitted between several atoms from the []. | |
924 * It is either NFA_OR, or NFA_CONCAT. | |
925 * | |
926 * [abc] expands to 'a b NFA_OR c NFA_OR' (in postfix notation) | |
927 * [^abc] expands to 'a NFA_NOT b NFA_NOT NFA_CONCAT c NFA_NOT | |
928 * NFA_CONCAT NFA_END_NEG_RANGE NFA_CONCAT' (in postfix | |
929 * notation) | |
930 * | |
931 */ | |
932 | |
933 | |
934 /* Emit negation atoms, if needed. | |
935 * The CONCAT below merges the NOT with the previous node. */ | |
936 #define TRY_NEG() \ | |
937 if (negated == TRUE) \ | |
938 { \ | |
939 EMIT(NFA_NOT); \ | |
940 } | |
941 | |
942 /* Emit glue between important nodes : CONCAT or OR. */ | |
943 #define EMIT_GLUE() \ | |
944 if (first == FALSE) \ | |
945 EMIT(glue); \ | |
946 else \ | |
947 first = FALSE; | |
948 | |
949 p = regparse; | |
950 endp = skip_anyof(p); | |
951 if (*endp == ']') | |
952 { | |
953 /* | |
954 * Try to reverse engineer character classes. For example, | |
955 * recognize that [0-9] stands for \d and [A-Za-z_] with \h, | |
956 * and perform the necessary substitutions in the NFA. | |
957 */ | |
958 result = nfa_recognize_char_class(regparse, endp, | |
959 extra == ADD_NL); | |
960 if (result != FAIL) | |
961 { | |
962 if (result >= NFA_DIGIT && result <= NFA_NUPPER) | |
963 EMIT(result); | |
964 else /* must be char class + newline */ | |
965 { | |
966 EMIT(result - ADD_NL); | |
967 EMIT(NFA_NEWL); | |
968 EMIT(NFA_OR); | |
969 } | |
970 regparse = endp; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
971 mb_ptr_adv(regparse); |
4444 | 972 return OK; |
973 } | |
974 /* | |
975 * Failed to recognize a character class. Use the simple | |
976 * version that turns [abc] into 'a' OR 'b' OR 'c' | |
977 */ | |
978 startc = endc = oldstartc = -1; | |
979 first = TRUE; /* Emitting first atom in this sequence? */ | |
980 negated = FALSE; | |
981 glue = NFA_OR; | |
982 if (*regparse == '^') /* negated range */ | |
983 { | |
984 negated = TRUE; | |
985 glue = NFA_CONCAT; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
986 mb_ptr_adv(regparse); |
4444 | 987 } |
988 if (*regparse == '-') | |
989 { | |
990 startc = '-'; | |
991 EMIT(startc); | |
992 TRY_NEG(); | |
993 EMIT_GLUE(); | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
994 mb_ptr_adv(regparse); |
4444 | 995 } |
996 /* Emit the OR branches for each character in the [] */ | |
997 emit_range = FALSE; | |
998 while (regparse < endp) | |
999 { | |
1000 oldstartc = startc; | |
1001 startc = -1; | |
1002 got_coll_char = FALSE; | |
1003 if (*regparse == '[') | |
1004 { | |
1005 /* Check for [: :], [= =], [. .] */ | |
1006 equiclass = collclass = 0; | |
1007 charclass = get_char_class(®parse); | |
1008 if (charclass == CLASS_NONE) | |
1009 { | |
1010 equiclass = get_equi_class(®parse); | |
1011 if (equiclass == 0) | |
1012 collclass = get_coll_element(®parse); | |
1013 } | |
1014 | |
1015 /* Character class like [:alpha:] */ | |
1016 if (charclass != CLASS_NONE) | |
1017 { | |
1018 switch (charclass) | |
1019 { | |
1020 case CLASS_ALNUM: | |
1021 EMIT(NFA_CLASS_ALNUM); | |
1022 break; | |
1023 case CLASS_ALPHA: | |
1024 EMIT(NFA_CLASS_ALPHA); | |
1025 break; | |
1026 case CLASS_BLANK: | |
1027 EMIT(NFA_CLASS_BLANK); | |
1028 break; | |
1029 case CLASS_CNTRL: | |
1030 EMIT(NFA_CLASS_CNTRL); | |
1031 break; | |
1032 case CLASS_DIGIT: | |
1033 EMIT(NFA_CLASS_DIGIT); | |
1034 break; | |
1035 case CLASS_GRAPH: | |
1036 EMIT(NFA_CLASS_GRAPH); | |
1037 break; | |
1038 case CLASS_LOWER: | |
1039 EMIT(NFA_CLASS_LOWER); | |
1040 break; | |
1041 case CLASS_PRINT: | |
1042 EMIT(NFA_CLASS_PRINT); | |
1043 break; | |
1044 case CLASS_PUNCT: | |
1045 EMIT(NFA_CLASS_PUNCT); | |
1046 break; | |
1047 case CLASS_SPACE: | |
1048 EMIT(NFA_CLASS_SPACE); | |
1049 break; | |
1050 case CLASS_UPPER: | |
1051 EMIT(NFA_CLASS_UPPER); | |
1052 break; | |
1053 case CLASS_XDIGIT: | |
1054 EMIT(NFA_CLASS_XDIGIT); | |
1055 break; | |
1056 case CLASS_TAB: | |
1057 EMIT(NFA_CLASS_TAB); | |
1058 break; | |
1059 case CLASS_RETURN: | |
1060 EMIT(NFA_CLASS_RETURN); | |
1061 break; | |
1062 case CLASS_BACKSPACE: | |
1063 EMIT(NFA_CLASS_BACKSPACE); | |
1064 break; | |
1065 case CLASS_ESCAPE: | |
1066 EMIT(NFA_CLASS_ESCAPE); | |
1067 break; | |
1068 } | |
1069 TRY_NEG(); | |
1070 EMIT_GLUE(); | |
1071 continue; | |
1072 } | |
1073 /* Try equivalence class [=a=] and the like */ | |
1074 if (equiclass != 0) | |
1075 { | |
1076 result = nfa_emit_equi_class(equiclass, negated); | |
1077 if (result == FAIL) | |
1078 { | |
1079 /* should never happen */ | |
1080 EMSG_RET_FAIL(_("E868: Error building NFA with equivalence class!")); | |
1081 } | |
1082 EMIT_GLUE(); | |
1083 continue; | |
1084 } | |
1085 /* Try collating class like [. .] */ | |
1086 if (collclass != 0) | |
1087 { | |
1088 startc = collclass; /* allow [.a.]-x as a range */ | |
1089 /* Will emit the proper atom at the end of the | |
1090 * while loop. */ | |
1091 } | |
1092 } | |
4677
c1622ff9ed8d
updated for version 7.3.1086
Bram Moolenaar <bram@vim.org>
parents:
4675
diff
changeset
|
1093 /* Try a range like 'a-x' or '\t-z'. Also allows '-' as a |
c1622ff9ed8d
updated for version 7.3.1086
Bram Moolenaar <bram@vim.org>
parents:
4675
diff
changeset
|
1094 * start character. */ |
c1622ff9ed8d
updated for version 7.3.1086
Bram Moolenaar <bram@vim.org>
parents:
4675
diff
changeset
|
1095 if (*regparse == '-' && oldstartc != -1) |
4444 | 1096 { |
1097 emit_range = TRUE; | |
1098 startc = oldstartc; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1099 mb_ptr_adv(regparse); |
4444 | 1100 continue; /* reading the end of the range */ |
1101 } | |
1102 | |
1103 /* Now handle simple and escaped characters. | |
1104 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim | |
1105 * accepts "\t", "\e", etc., but only when the 'l' flag in | |
1106 * 'cpoptions' is not included. | |
1107 * Posix doesn't recognize backslash at all. | |
1108 */ | |
1109 if (*regparse == '\\' | |
1110 && !cpo_bsl | |
1111 && regparse + 1 <= endp | |
1112 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL | |
1113 || (!cpo_lit | |
1114 && vim_strchr(REGEXP_ABBR, regparse[1]) | |
1115 != NULL) | |
1116 ) | |
1117 ) | |
1118 { | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1119 mb_ptr_adv(regparse); |
4444 | 1120 |
4507
9dbbddb2ed10
updated for version 7.3.1001
Bram Moolenaar <bram@vim.org>
parents:
4503
diff
changeset
|
1121 if (*regparse == 'n') |
4444 | 1122 startc = reg_string ? NL : NFA_NEWL; |
1123 else | |
1124 if (*regparse == 'd' | |
1125 || *regparse == 'o' | |
1126 || *regparse == 'x' | |
1127 || *regparse == 'u' | |
1128 || *regparse == 'U' | |
1129 ) | |
1130 { | |
1131 /* TODO(RE) This needs more testing */ | |
1132 startc = coll_get_char(); | |
1133 got_coll_char = TRUE; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1134 mb_ptr_back(old_regparse, regparse); |
4444 | 1135 } |
1136 else | |
1137 { | |
1138 /* \r,\t,\e,\b */ | |
1139 startc = backslash_trans(*regparse); | |
1140 } | |
1141 } | |
1142 | |
1143 /* Normal printable char */ | |
1144 if (startc == -1) | |
4677
c1622ff9ed8d
updated for version 7.3.1086
Bram Moolenaar <bram@vim.org>
parents:
4675
diff
changeset
|
1145 startc = PTR2CHAR(regparse); |
4444 | 1146 |
1147 /* Previous char was '-', so this char is end of range. */ | |
1148 if (emit_range) | |
1149 { | |
4677
c1622ff9ed8d
updated for version 7.3.1086
Bram Moolenaar <bram@vim.org>
parents:
4675
diff
changeset
|
1150 endc = startc; |
c1622ff9ed8d
updated for version 7.3.1086
Bram Moolenaar <bram@vim.org>
parents:
4675
diff
changeset
|
1151 startc = oldstartc; |
4444 | 1152 if (startc > endc) |
1153 EMSG_RET_FAIL(_(e_invrange)); | |
1154 #ifdef FEAT_MBYTE | |
1155 if (has_mbyte && ((*mb_char2len)(startc) > 1 | |
1156 || (*mb_char2len)(endc) > 1)) | |
1157 { | |
1158 if (endc > startc + 256) | |
1159 EMSG_RET_FAIL(_(e_invrange)); | |
1160 /* Emit the range. "startc" was already emitted, so | |
1161 * skip it. */ | |
1162 for (c = startc + 1; c <= endc; c++) | |
1163 { | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1164 EMIT(c); |
4444 | 1165 TRY_NEG(); |
1166 EMIT_GLUE(); | |
1167 } | |
1168 } | |
1169 else | |
1170 #endif | |
1171 { | |
1172 #ifdef EBCDIC | |
1173 int alpha_only = FALSE; | |
1174 | |
1175 /* for alphabetical range skip the gaps | |
1176 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */ | |
1177 if (isalpha(startc) && isalpha(endc)) | |
1178 alpha_only = TRUE; | |
1179 #endif | |
1180 /* Emit the range. "startc" was already emitted, so | |
1181 * skip it. */ | |
1182 for (c = startc + 1; c <= endc; c++) | |
1183 #ifdef EBCDIC | |
1184 if (!alpha_only || isalpha(startc)) | |
1185 #endif | |
1186 { | |
1187 EMIT(c); | |
1188 TRY_NEG(); | |
1189 EMIT_GLUE(); | |
1190 } | |
1191 } | |
4677
c1622ff9ed8d
updated for version 7.3.1086
Bram Moolenaar <bram@vim.org>
parents:
4675
diff
changeset
|
1192 emit_range = FALSE; |
c1622ff9ed8d
updated for version 7.3.1086
Bram Moolenaar <bram@vim.org>
parents:
4675
diff
changeset
|
1193 startc = -1; |
4444 | 1194 } |
1195 else | |
1196 { | |
1197 /* | |
1198 * This char (startc) is not part of a range. Just | |
1199 * emit it. | |
1200 * | |
1201 * Normally, simply emit startc. But if we get char | |
1202 * code=0 from a collating char, then replace it with | |
1203 * 0x0a. | |
1204 * | |
1205 * This is needed to completely mimic the behaviour of | |
1206 * the backtracking engine. | |
1207 */ | |
1208 if (got_coll_char == TRUE && startc == 0) | |
1209 EMIT(0x0a); | |
1210 else | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1211 EMIT(startc); |
4444 | 1212 TRY_NEG(); |
1213 EMIT_GLUE(); | |
1214 } | |
1215 | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1216 mb_ptr_adv(regparse); |
4444 | 1217 } /* while (p < endp) */ |
1218 | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1219 mb_ptr_back(old_regparse, regparse); |
4444 | 1220 if (*regparse == '-') /* if last, '-' is just a char */ |
1221 { | |
1222 EMIT('-'); | |
1223 TRY_NEG(); | |
1224 EMIT_GLUE(); | |
1225 } | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1226 mb_ptr_adv(regparse); |
4444 | 1227 |
1228 /* skip the trailing ] */ | |
1229 regparse = endp; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1230 mb_ptr_adv(regparse); |
4444 | 1231 if (negated == TRUE) |
1232 { | |
1233 /* Mark end of negated char range */ | |
1234 EMIT(NFA_END_NEG_RANGE); | |
1235 EMIT(NFA_CONCAT); | |
1236 } | |
4615
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1237 |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1238 /* \_[] also matches \n but it's not negated */ |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1239 if (extra == ADD_NL) |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1240 { |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1241 EMIT(reg_string ? NL : NFA_NEWL); |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1242 EMIT(NFA_OR); |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1243 } |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1244 |
4444 | 1245 return OK; |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1246 } /* if exists closing ] */ |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1247 |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1248 if (reg_strict) |
4444 | 1249 { |
1250 syntax_error = TRUE; | |
1251 EMSG_RET_FAIL(_(e_missingbracket)); | |
1252 } | |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1253 /* FALLTHROUGH */ |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1254 |
4444 | 1255 default: |
1256 { | |
1257 #ifdef FEAT_MBYTE | |
1258 int plen; | |
1259 | |
1260 nfa_do_multibyte: | |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1261 /* plen is length of current char with composing chars */ |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1262 if (enc_utf8 && ((*mb_char2len)(c) |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1263 != (plen = (*mb_ptr2len)(old_regparse)) |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1264 || utf_iscomposing(c))) |
4444 | 1265 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
1266 int i = 0; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
1267 |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
1268 /* A base character plus composing characters, or just one |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
1269 * or more composing characters. |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1270 * This requires creating a separate atom as if enclosing |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1271 * the characters in (), where NFA_COMPOSING is the ( and |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1272 * NFA_END_COMPOSING is the ). Note that right now we are |
4444 | 1273 * building the postfix form, not the NFA itself; |
1274 * a composing char could be: a, b, c, NFA_COMPOSING | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1275 * where 'b' and 'c' are chars with codes > 256. */ |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1276 for (;;) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1277 { |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1278 EMIT(c); |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1279 if (i > 0) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1280 EMIT(NFA_CONCAT); |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1281 if ((i += utf_char2len(c)) >= plen) |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1282 break; |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1283 c = utf_ptr2char(old_regparse + i); |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1284 } |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1285 EMIT(NFA_COMPOSING); |
4444 | 1286 regparse = old_regparse + plen; |
1287 } | |
1288 else | |
1289 #endif | |
1290 { | |
1291 c = no_Magic(c); | |
1292 EMIT(c); | |
1293 } | |
1294 return OK; | |
1295 } | |
1296 } | |
1297 | |
1298 #undef TRY_NEG | |
1299 #undef EMIT_GLUE | |
1300 | |
1301 return OK; | |
1302 } | |
1303 | |
1304 /* | |
1305 * Parse something followed by possible [*+=]. | |
1306 * | |
1307 * A piece is an atom, possibly followed by a multi, an indication of how many | |
1308 * times the atom can be matched. Example: "a*" matches any sequence of "a" | |
1309 * characters: "", "a", "aa", etc. | |
1310 * | |
1311 * piece ::= atom | |
1312 * or atom multi | |
1313 */ | |
1314 static int | |
1315 nfa_regpiece() | |
1316 { | |
1317 int i; | |
1318 int op; | |
1319 int ret; | |
1320 long minval, maxval; | |
1321 int greedy = TRUE; /* Braces are prefixed with '-' ? */ | |
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4677
diff
changeset
|
1322 parse_state_T old_state; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4677
diff
changeset
|
1323 parse_state_T new_state; |
4444 | 1324 int c2; |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1325 int old_post_pos; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1326 int my_post_start; |
4444 | 1327 int quest; |
1328 | |
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4677
diff
changeset
|
1329 /* Save the current parse state, so that we can use it if <atom>{m,n} is |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4677
diff
changeset
|
1330 * next. */ |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4677
diff
changeset
|
1331 save_parse_state(&old_state); |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4677
diff
changeset
|
1332 |
4444 | 1333 /* store current pos in the postfix form, for \{m,n} involving 0s */ |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1334 my_post_start = (int)(post_ptr - post_start); |
4444 | 1335 |
1336 ret = nfa_regatom(); | |
1337 if (ret == FAIL) | |
1338 return FAIL; /* cascaded error */ | |
1339 | |
1340 op = peekchr(); | |
1341 if (re_multi_type(op) == NOT_MULTI) | |
1342 return OK; | |
1343 | |
1344 skipchr(); | |
1345 switch (op) | |
1346 { | |
1347 case Magic('*'): | |
1348 EMIT(NFA_STAR); | |
1349 break; | |
1350 | |
1351 case Magic('+'): | |
1352 /* | |
1353 * Trick: Normally, (a*)\+ would match the whole input "aaa". The | |
1354 * first and only submatch would be "aaa". But the backtracking | |
1355 * engine interprets the plus as "try matching one more time", and | |
1356 * a* matches a second time at the end of the input, the empty | |
1357 * string. | |
1358 * The submatch will the empty string. | |
1359 * | |
4673
05d57d7c2d55
updated for version 7.3.1084
Bram Moolenaar <bram@vim.org>
parents:
4671
diff
changeset
|
1360 * In order to be consistent with the old engine, we replace |
05d57d7c2d55
updated for version 7.3.1084
Bram Moolenaar <bram@vim.org>
parents:
4671
diff
changeset
|
1361 * <atom>+ with <atom><atom>* |
4444 | 1362 */ |
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4677
diff
changeset
|
1363 restore_parse_state(&old_state); |
4444 | 1364 curchr = -1; |
1365 if (nfa_regatom() == FAIL) | |
1366 return FAIL; | |
1367 EMIT(NFA_STAR); | |
1368 EMIT(NFA_CONCAT); | |
1369 skipchr(); /* skip the \+ */ | |
1370 break; | |
1371 | |
1372 case Magic('@'): | |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1373 c2 = getdecchrs(); |
4444 | 1374 op = no_Magic(getchr()); |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1375 i = 0; |
4444 | 1376 switch(op) |
1377 { | |
1378 case '=': | |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1379 /* \@= */ |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1380 i = NFA_PREV_ATOM_NO_WIDTH; |
4444 | 1381 break; |
4661
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
1382 case '!': |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1383 /* \@! */ |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1384 i = NFA_PREV_ATOM_NO_WIDTH_NEG; |
4661
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
1385 break; |
4444 | 1386 case '<': |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1387 op = no_Magic(getchr()); |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1388 if (op == '=') |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1389 /* \@<= */ |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1390 i = NFA_PREV_ATOM_JUST_BEFORE; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1391 else if (op == '!') |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1392 /* \@<! */ |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1393 i = NFA_PREV_ATOM_JUST_BEFORE_NEG; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1394 break; |
4444 | 1395 case '>': |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1396 /* \@> Not supported yet */ |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1397 /* i = NFA_PREV_ATOM_LIKE_PATTERN; */ |
4444 | 1398 return FAIL; |
1399 } | |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1400 if (i == 0) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1401 { |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1402 syntax_error = TRUE; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1403 EMSGN(_("E869: (NFA) Unknown operator '\\@%c'"), op); |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1404 return FAIL; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1405 } |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1406 EMIT(i); |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1407 if (i == NFA_PREV_ATOM_JUST_BEFORE |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1408 || i == NFA_PREV_ATOM_JUST_BEFORE_NEG) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1409 EMIT(c2); |
4444 | 1410 break; |
1411 | |
1412 case Magic('?'): | |
1413 case Magic('='): | |
1414 EMIT(NFA_QUEST); | |
1415 break; | |
1416 | |
1417 case Magic('{'): | |
1418 /* a{2,5} will expand to 'aaa?a?a?' | |
1419 * a{-1,3} will expand to 'aa??a??', where ?? is the nongreedy | |
1420 * version of '?' | |
1421 * \v(ab){2,3} will expand to '(ab)(ab)(ab)?', where all the | |
1422 * parenthesis have the same id | |
1423 */ | |
1424 | |
1425 greedy = TRUE; | |
1426 c2 = peekchr(); | |
1427 if (c2 == '-' || c2 == Magic('-')) | |
1428 { | |
1429 skipchr(); | |
1430 greedy = FALSE; | |
1431 } | |
1432 if (!read_limits(&minval, &maxval)) | |
1433 { | |
1434 syntax_error = TRUE; | |
1435 EMSG_RET_FAIL(_("E870: (NFA regexp) Error reading repetition limits")); | |
1436 } | |
1437 /* <atom>{0,inf}, <atom>{0,} and <atom>{} are equivalent to | |
1438 * <atom>* */ | |
4675
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1439 if (minval == 0 && maxval == MAX_LIMIT) |
4444 | 1440 { |
4675
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1441 if (greedy) |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1442 /* \{}, \{0,} */ |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1443 EMIT(NFA_STAR); |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1444 else |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1445 /* \{-}, \{-0,} */ |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1446 EMIT(NFA_STAR_NONGREEDY); |
4444 | 1447 break; |
1448 } | |
1449 | |
1450 /* Special case: x{0} or x{-0} */ | |
1451 if (maxval == 0) | |
1452 { | |
1453 /* Ignore result of previous call to nfa_regatom() */ | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1454 post_ptr = post_start + my_post_start; |
4444 | 1455 /* NFA_SKIP_CHAR has 0-length and works everywhere */ |
1456 EMIT(NFA_SKIP_CHAR); | |
1457 return OK; | |
1458 } | |
1459 | |
1460 /* Ignore previous call to nfa_regatom() */ | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1461 post_ptr = post_start + my_post_start; |
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4677
diff
changeset
|
1462 /* Save parse state after the repeated atom and the \{} */ |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4677
diff
changeset
|
1463 save_parse_state(&new_state); |
4444 | 1464 |
1465 quest = (greedy == TRUE? NFA_QUEST : NFA_QUEST_NONGREEDY); | |
1466 for (i = 0; i < maxval; i++) | |
1467 { | |
1468 /* Goto beginning of the repeated atom */ | |
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4677
diff
changeset
|
1469 restore_parse_state(&old_state); |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1470 old_post_pos = (int)(post_ptr - post_start); |
4444 | 1471 if (nfa_regatom() == FAIL) |
1472 return FAIL; | |
1473 /* after "minval" times, atoms are optional */ | |
1474 if (i + 1 > minval) | |
4673
05d57d7c2d55
updated for version 7.3.1084
Bram Moolenaar <bram@vim.org>
parents:
4671
diff
changeset
|
1475 { |
05d57d7c2d55
updated for version 7.3.1084
Bram Moolenaar <bram@vim.org>
parents:
4671
diff
changeset
|
1476 if (maxval == MAX_LIMIT) |
4675
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1477 { |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1478 if (greedy) |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1479 EMIT(NFA_STAR); |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1480 else |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1481 EMIT(NFA_STAR_NONGREEDY); |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1482 } |
4673
05d57d7c2d55
updated for version 7.3.1084
Bram Moolenaar <bram@vim.org>
parents:
4671
diff
changeset
|
1483 else |
05d57d7c2d55
updated for version 7.3.1084
Bram Moolenaar <bram@vim.org>
parents:
4671
diff
changeset
|
1484 EMIT(quest); |
05d57d7c2d55
updated for version 7.3.1084
Bram Moolenaar <bram@vim.org>
parents:
4671
diff
changeset
|
1485 } |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1486 if (old_post_pos != my_post_start) |
4444 | 1487 EMIT(NFA_CONCAT); |
4673
05d57d7c2d55
updated for version 7.3.1084
Bram Moolenaar <bram@vim.org>
parents:
4671
diff
changeset
|
1488 if (i + 1 > minval && maxval == MAX_LIMIT) |
05d57d7c2d55
updated for version 7.3.1084
Bram Moolenaar <bram@vim.org>
parents:
4671
diff
changeset
|
1489 break; |
4444 | 1490 } |
1491 | |
1492 /* Go to just after the repeated atom and the \{} */ | |
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4677
diff
changeset
|
1493 restore_parse_state(&new_state); |
4444 | 1494 curchr = -1; |
1495 | |
1496 break; | |
1497 | |
1498 | |
1499 default: | |
1500 break; | |
1501 } /* end switch */ | |
1502 | |
1503 if (re_multi_type(peekchr()) != NOT_MULTI) | |
1504 { | |
1505 /* Can't have a multi follow a multi. */ | |
1506 syntax_error = TRUE; | |
1507 EMSG_RET_FAIL(_("E871: (NFA regexp) Can't have a multi follow a multi !")); | |
1508 } | |
1509 | |
1510 return OK; | |
1511 } | |
1512 | |
1513 /* | |
1514 * Parse one or more pieces, concatenated. It matches a match for the | |
1515 * first piece, followed by a match for the second piece, etc. Example: | |
1516 * "f[0-9]b", first matches "f", then a digit and then "b". | |
1517 * | |
1518 * concat ::= piece | |
1519 * or piece piece | |
1520 * or piece piece piece | |
1521 * etc. | |
1522 */ | |
1523 static int | |
1524 nfa_regconcat() | |
1525 { | |
1526 int cont = TRUE; | |
1527 int first = TRUE; | |
1528 | |
1529 while (cont) | |
1530 { | |
1531 switch (peekchr()) | |
1532 { | |
1533 case NUL: | |
1534 case Magic('|'): | |
1535 case Magic('&'): | |
1536 case Magic(')'): | |
1537 cont = FALSE; | |
1538 break; | |
1539 | |
1540 case Magic('Z'): | |
1541 #ifdef FEAT_MBYTE | |
1542 regflags |= RF_ICOMBINE; | |
1543 #endif | |
1544 skipchr_keepstart(); | |
1545 break; | |
1546 case Magic('c'): | |
1547 regflags |= RF_ICASE; | |
1548 skipchr_keepstart(); | |
1549 break; | |
1550 case Magic('C'): | |
1551 regflags |= RF_NOICASE; | |
1552 skipchr_keepstart(); | |
1553 break; | |
1554 case Magic('v'): | |
1555 reg_magic = MAGIC_ALL; | |
1556 skipchr_keepstart(); | |
1557 curchr = -1; | |
1558 break; | |
1559 case Magic('m'): | |
1560 reg_magic = MAGIC_ON; | |
1561 skipchr_keepstart(); | |
1562 curchr = -1; | |
1563 break; | |
1564 case Magic('M'): | |
1565 reg_magic = MAGIC_OFF; | |
1566 skipchr_keepstart(); | |
1567 curchr = -1; | |
1568 break; | |
1569 case Magic('V'): | |
1570 reg_magic = MAGIC_NONE; | |
1571 skipchr_keepstart(); | |
1572 curchr = -1; | |
1573 break; | |
1574 | |
1575 default: | |
1576 if (nfa_regpiece() == FAIL) | |
1577 return FAIL; | |
1578 if (first == FALSE) | |
1579 EMIT(NFA_CONCAT); | |
1580 else | |
1581 first = FALSE; | |
1582 break; | |
1583 } | |
1584 } | |
1585 | |
1586 return OK; | |
1587 } | |
1588 | |
1589 /* | |
1590 * Parse a branch, one or more concats, separated by "\&". It matches the | |
1591 * last concat, but only if all the preceding concats also match at the same | |
1592 * position. Examples: | |
1593 * "foobeep\&..." matches "foo" in "foobeep". | |
1594 * ".*Peter\&.*Bob" matches in a line containing both "Peter" and "Bob" | |
1595 * | |
1596 * branch ::= concat | |
1597 * or concat \& concat | |
1598 * or concat \& concat \& concat | |
1599 * etc. | |
1600 */ | |
1601 static int | |
1602 nfa_regbranch() | |
1603 { | |
1604 int ch; | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1605 int old_post_pos; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1606 |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1607 old_post_pos = (int)(post_ptr - post_start); |
4444 | 1608 |
1609 /* First branch, possibly the only one */ | |
1610 if (nfa_regconcat() == FAIL) | |
1611 return FAIL; | |
1612 | |
1613 ch = peekchr(); | |
1614 /* Try next concats */ | |
1615 while (ch == Magic('&')) | |
1616 { | |
1617 skipchr(); | |
1618 EMIT(NFA_NOPEN); | |
1619 EMIT(NFA_PREV_ATOM_NO_WIDTH); | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1620 old_post_pos = (int)(post_ptr - post_start); |
4444 | 1621 if (nfa_regconcat() == FAIL) |
1622 return FAIL; | |
1623 /* if concat is empty, skip a input char. But do emit a node */ | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1624 if (old_post_pos == (int)(post_ptr - post_start)) |
4444 | 1625 EMIT(NFA_SKIP_CHAR); |
1626 EMIT(NFA_CONCAT); | |
1627 ch = peekchr(); | |
1628 } | |
1629 | |
1630 /* Even if a branch is empty, emit one node for it */ | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1631 if (old_post_pos == (int)(post_ptr - post_start)) |
4444 | 1632 EMIT(NFA_SKIP_CHAR); |
1633 | |
1634 return OK; | |
1635 } | |
1636 | |
1637 /* | |
1638 * Parse a pattern, one or more branches, separated by "\|". It matches | |
1639 * anything that matches one of the branches. Example: "foo\|beep" matches | |
1640 * "foo" and matches "beep". If more than one branch matches, the first one | |
1641 * is used. | |
1642 * | |
1643 * pattern ::= branch | |
1644 * or branch \| branch | |
1645 * or branch \| branch \| branch | |
1646 * etc. | |
1647 */ | |
1648 static int | |
1649 nfa_reg(paren) | |
1650 int paren; /* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */ | |
1651 { | |
1652 int parno = 0; | |
1653 | |
1654 #ifdef FEAT_SYN_HL | |
1655 #endif | |
1656 if (paren == REG_PAREN) | |
1657 { | |
1658 if (regnpar >= NSUBEXP) /* Too many `(' */ | |
1659 { | |
1660 syntax_error = TRUE; | |
1661 EMSG_RET_FAIL(_("E872: (NFA regexp) Too many '('")); | |
1662 } | |
1663 parno = regnpar++; | |
1664 } | |
1665 | |
1666 if (nfa_regbranch() == FAIL) | |
1667 return FAIL; /* cascaded error */ | |
1668 | |
1669 while (peekchr() == Magic('|')) | |
1670 { | |
1671 skipchr(); | |
1672 if (nfa_regbranch() == FAIL) | |
1673 return FAIL; /* cascaded error */ | |
1674 EMIT(NFA_OR); | |
1675 } | |
1676 | |
1677 /* Check for proper termination. */ | |
1678 if (paren != REG_NOPAREN && getchr() != Magic(')')) | |
1679 { | |
1680 syntax_error = TRUE; | |
1681 if (paren == REG_NPAREN) | |
1682 EMSG2_RET_FAIL(_(e_unmatchedpp), reg_magic == MAGIC_ALL); | |
1683 else | |
1684 EMSG2_RET_FAIL(_(e_unmatchedp), reg_magic == MAGIC_ALL); | |
1685 } | |
1686 else if (paren == REG_NOPAREN && peekchr() != NUL) | |
1687 { | |
1688 syntax_error = TRUE; | |
1689 if (peekchr() == Magic(')')) | |
1690 EMSG2_RET_FAIL(_(e_unmatchedpar), reg_magic == MAGIC_ALL); | |
1691 else | |
1692 EMSG_RET_FAIL(_("E873: (NFA regexp) proper termination error")); | |
1693 } | |
1694 /* | |
1695 * Here we set the flag allowing back references to this set of | |
1696 * parentheses. | |
1697 */ | |
1698 if (paren == REG_PAREN) | |
1699 { | |
1700 had_endbrace[parno] = TRUE; /* have seen the close paren */ | |
1701 EMIT(NFA_MOPEN + parno); | |
1702 } | |
1703 | |
1704 return OK; | |
1705 } | |
1706 | |
1707 #ifdef DEBUG | |
1708 static char_u code[50]; | |
1709 | |
1710 static void | |
1711 nfa_set_code(c) | |
1712 int c; | |
1713 { | |
1714 int addnl = FALSE; | |
1715 | |
1716 if (c >= NFA_FIRST_NL && c <= NFA_LAST_NL) | |
1717 { | |
1718 addnl = TRUE; | |
1719 c -= ADD_NL; | |
1720 } | |
1721 | |
1722 STRCPY(code, ""); | |
1723 switch (c) | |
1724 { | |
1725 case NFA_MATCH: STRCPY(code, "NFA_MATCH "); break; | |
1726 case NFA_SPLIT: STRCPY(code, "NFA_SPLIT "); break; | |
1727 case NFA_CONCAT: STRCPY(code, "NFA_CONCAT "); break; | |
1728 case NFA_NEWL: STRCPY(code, "NFA_NEWL "); break; | |
1729 case NFA_ZSTART: STRCPY(code, "NFA_ZSTART"); break; | |
1730 case NFA_ZEND: STRCPY(code, "NFA_ZEND"); break; | |
1731 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1732 case NFA_BACKREF1: STRCPY(code, "NFA_BACKREF1"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1733 case NFA_BACKREF2: STRCPY(code, "NFA_BACKREF2"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1734 case NFA_BACKREF3: STRCPY(code, "NFA_BACKREF3"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1735 case NFA_BACKREF4: STRCPY(code, "NFA_BACKREF4"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1736 case NFA_BACKREF5: STRCPY(code, "NFA_BACKREF5"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1737 case NFA_BACKREF6: STRCPY(code, "NFA_BACKREF6"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1738 case NFA_BACKREF7: STRCPY(code, "NFA_BACKREF7"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1739 case NFA_BACKREF8: STRCPY(code, "NFA_BACKREF8"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1740 case NFA_BACKREF9: STRCPY(code, "NFA_BACKREF9"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1741 case NFA_SKIP: STRCPY(code, "NFA_SKIP"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1742 |
4444 | 1743 case NFA_PREV_ATOM_NO_WIDTH: |
1744 STRCPY(code, "NFA_PREV_ATOM_NO_WIDTH"); break; | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
1745 case NFA_PREV_ATOM_NO_WIDTH_NEG: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
1746 STRCPY(code, "NFA_PREV_ATOM_NO_WIDTH_NEG"); break; |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1747 case NFA_PREV_ATOM_JUST_BEFORE: |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1748 STRCPY(code, "NFA_PREV_ATOM_JUST_BEFORE"); break; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1749 case NFA_PREV_ATOM_JUST_BEFORE_NEG: |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1750 STRCPY(code, "NFA_PREV_ATOM_JUST_BEFORE_NEG"); break; |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
1751 case NFA_NOPEN: STRCPY(code, "NFA_NOPEN"); break; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
1752 case NFA_NCLOSE: STRCPY(code, "NFA_NCLOSE"); break; |
4444 | 1753 case NFA_START_INVISIBLE: STRCPY(code, "NFA_START_INVISIBLE"); break; |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1754 case NFA_START_INVISIBLE_BEFORE: |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
1755 STRCPY(code, "NFA_START_INVISIBLE_BEFORE"); break; |
4444 | 1756 case NFA_END_INVISIBLE: STRCPY(code, "NFA_END_INVISIBLE"); break; |
1757 | |
1758 case NFA_COMPOSING: STRCPY(code, "NFA_COMPOSING"); break; | |
1759 case NFA_END_COMPOSING: STRCPY(code, "NFA_END_COMPOSING"); break; | |
1760 | |
1761 case NFA_MOPEN + 0: | |
1762 case NFA_MOPEN + 1: | |
1763 case NFA_MOPEN + 2: | |
1764 case NFA_MOPEN + 3: | |
1765 case NFA_MOPEN + 4: | |
1766 case NFA_MOPEN + 5: | |
1767 case NFA_MOPEN + 6: | |
1768 case NFA_MOPEN + 7: | |
1769 case NFA_MOPEN + 8: | |
1770 case NFA_MOPEN + 9: | |
1771 STRCPY(code, "NFA_MOPEN(x)"); | |
1772 code[10] = c - NFA_MOPEN + '0'; | |
1773 break; | |
1774 case NFA_MCLOSE + 0: | |
1775 case NFA_MCLOSE + 1: | |
1776 case NFA_MCLOSE + 2: | |
1777 case NFA_MCLOSE + 3: | |
1778 case NFA_MCLOSE + 4: | |
1779 case NFA_MCLOSE + 5: | |
1780 case NFA_MCLOSE + 6: | |
1781 case NFA_MCLOSE + 7: | |
1782 case NFA_MCLOSE + 8: | |
1783 case NFA_MCLOSE + 9: | |
1784 STRCPY(code, "NFA_MCLOSE(x)"); | |
1785 code[11] = c - NFA_MCLOSE + '0'; | |
1786 break; | |
1787 case NFA_EOL: STRCPY(code, "NFA_EOL "); break; | |
1788 case NFA_BOL: STRCPY(code, "NFA_BOL "); break; | |
1789 case NFA_EOW: STRCPY(code, "NFA_EOW "); break; | |
1790 case NFA_BOW: STRCPY(code, "NFA_BOW "); break; | |
4671
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
1791 case NFA_EOF: STRCPY(code, "NFA_EOF "); break; |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
1792 case NFA_BOF: STRCPY(code, "NFA_BOF "); break; |
4444 | 1793 case NFA_STAR: STRCPY(code, "NFA_STAR "); break; |
4675
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1794 case NFA_STAR_NONGREEDY: STRCPY(code, "NFA_STAR_NONGREEDY "); break; |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1795 case NFA_QUEST: STRCPY(code, "NFA_QUEST"); break; |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
1796 case NFA_QUEST_NONGREEDY: STRCPY(code, "NFA_QUEST_NON_GREEDY"); break; |
4444 | 1797 case NFA_NOT: STRCPY(code, "NFA_NOT "); break; |
1798 case NFA_SKIP_CHAR: STRCPY(code, "NFA_SKIP_CHAR"); break; | |
1799 case NFA_OR: STRCPY(code, "NFA_OR"); break; | |
1800 case NFA_END_NEG_RANGE: STRCPY(code, "NFA_END_NEG_RANGE"); break; | |
1801 case NFA_CLASS_ALNUM: STRCPY(code, "NFA_CLASS_ALNUM"); break; | |
1802 case NFA_CLASS_ALPHA: STRCPY(code, "NFA_CLASS_ALPHA"); break; | |
1803 case NFA_CLASS_BLANK: STRCPY(code, "NFA_CLASS_BLANK"); break; | |
1804 case NFA_CLASS_CNTRL: STRCPY(code, "NFA_CLASS_CNTRL"); break; | |
1805 case NFA_CLASS_DIGIT: STRCPY(code, "NFA_CLASS_DIGIT"); break; | |
1806 case NFA_CLASS_GRAPH: STRCPY(code, "NFA_CLASS_GRAPH"); break; | |
1807 case NFA_CLASS_LOWER: STRCPY(code, "NFA_CLASS_LOWER"); break; | |
1808 case NFA_CLASS_PRINT: STRCPY(code, "NFA_CLASS_PRINT"); break; | |
1809 case NFA_CLASS_PUNCT: STRCPY(code, "NFA_CLASS_PUNCT"); break; | |
1810 case NFA_CLASS_SPACE: STRCPY(code, "NFA_CLASS_SPACE"); break; | |
1811 case NFA_CLASS_UPPER: STRCPY(code, "NFA_CLASS_UPPER"); break; | |
1812 case NFA_CLASS_XDIGIT: STRCPY(code, "NFA_CLASS_XDIGIT"); break; | |
1813 case NFA_CLASS_TAB: STRCPY(code, "NFA_CLASS_TAB"); break; | |
1814 case NFA_CLASS_RETURN: STRCPY(code, "NFA_CLASS_RETURN"); break; | |
1815 case NFA_CLASS_BACKSPACE: STRCPY(code, "NFA_CLASS_BACKSPACE"); break; | |
1816 case NFA_CLASS_ESCAPE: STRCPY(code, "NFA_CLASS_ESCAPE"); break; | |
1817 | |
1818 case NFA_ANY: STRCPY(code, "NFA_ANY"); break; | |
1819 case NFA_IDENT: STRCPY(code, "NFA_IDENT"); break; | |
1820 case NFA_SIDENT:STRCPY(code, "NFA_SIDENT"); break; | |
1821 case NFA_KWORD: STRCPY(code, "NFA_KWORD"); break; | |
1822 case NFA_SKWORD:STRCPY(code, "NFA_SKWORD"); break; | |
1823 case NFA_FNAME: STRCPY(code, "NFA_FNAME"); break; | |
1824 case NFA_SFNAME:STRCPY(code, "NFA_SFNAME"); break; | |
1825 case NFA_PRINT: STRCPY(code, "NFA_PRINT"); break; | |
1826 case NFA_SPRINT:STRCPY(code, "NFA_SPRINT"); break; | |
1827 case NFA_WHITE: STRCPY(code, "NFA_WHITE"); break; | |
1828 case NFA_NWHITE:STRCPY(code, "NFA_NWHITE"); break; | |
1829 case NFA_DIGIT: STRCPY(code, "NFA_DIGIT"); break; | |
1830 case NFA_NDIGIT:STRCPY(code, "NFA_NDIGIT"); break; | |
1831 case NFA_HEX: STRCPY(code, "NFA_HEX"); break; | |
1832 case NFA_NHEX: STRCPY(code, "NFA_NHEX"); break; | |
1833 case NFA_OCTAL: STRCPY(code, "NFA_OCTAL"); break; | |
1834 case NFA_NOCTAL:STRCPY(code, "NFA_NOCTAL"); break; | |
1835 case NFA_WORD: STRCPY(code, "NFA_WORD"); break; | |
1836 case NFA_NWORD: STRCPY(code, "NFA_NWORD"); break; | |
1837 case NFA_HEAD: STRCPY(code, "NFA_HEAD"); break; | |
1838 case NFA_NHEAD: STRCPY(code, "NFA_NHEAD"); break; | |
1839 case NFA_ALPHA: STRCPY(code, "NFA_ALPHA"); break; | |
1840 case NFA_NALPHA:STRCPY(code, "NFA_NALPHA"); break; | |
1841 case NFA_LOWER: STRCPY(code, "NFA_LOWER"); break; | |
1842 case NFA_NLOWER:STRCPY(code, "NFA_NLOWER"); break; | |
1843 case NFA_UPPER: STRCPY(code, "NFA_UPPER"); break; | |
1844 case NFA_NUPPER:STRCPY(code, "NFA_NUPPER"); break; | |
1845 | |
1846 default: | |
1847 STRCPY(code, "CHAR(x)"); | |
1848 code[5] = c; | |
1849 } | |
1850 | |
1851 if (addnl == TRUE) | |
1852 STRCAT(code, " + NEWLINE "); | |
1853 | |
1854 } | |
1855 | |
1856 #ifdef ENABLE_LOG | |
1857 static FILE *log_fd; | |
1858 | |
1859 /* | |
1860 * Print the postfix notation of the current regexp. | |
1861 */ | |
1862 static void | |
1863 nfa_postfix_dump(expr, retval) | |
1864 char_u *expr; | |
1865 int retval; | |
1866 { | |
1867 int *p; | |
1868 FILE *f; | |
1869 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
1870 f = fopen(NFA_REGEXP_DUMP_LOG, "a"); |
4444 | 1871 if (f != NULL) |
1872 { | |
1873 fprintf(f, "\n-------------------------\n"); | |
1874 if (retval == FAIL) | |
1875 fprintf(f, ">>> NFA engine failed ... \n"); | |
1876 else if (retval == OK) | |
1877 fprintf(f, ">>> NFA engine succeeded !\n"); | |
1878 fprintf(f, "Regexp: \"%s\"\nPostfix notation (char): \"", expr); | |
4464 | 1879 for (p = post_start; *p && p < post_end; p++) |
4444 | 1880 { |
1881 nfa_set_code(*p); | |
1882 fprintf(f, "%s, ", code); | |
1883 } | |
1884 fprintf(f, "\"\nPostfix notation (int): "); | |
4464 | 1885 for (p = post_start; *p && p < post_end; p++) |
4444 | 1886 fprintf(f, "%d ", *p); |
1887 fprintf(f, "\n\n"); | |
1888 fclose(f); | |
1889 } | |
1890 } | |
1891 | |
1892 /* | |
1893 * Print the NFA starting with a root node "state". | |
1894 */ | |
1895 static void | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1896 nfa_print_state(debugf, state) |
4444 | 1897 FILE *debugf; |
1898 nfa_state_T *state; | |
1899 { | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1900 garray_T indent; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1901 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1902 ga_init2(&indent, 1, 64); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1903 ga_append(&indent, '\0'); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1904 nfa_print_state2(debugf, state, &indent); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1905 ga_clear(&indent); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1906 } |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1907 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1908 static void |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1909 nfa_print_state2(debugf, state, indent) |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1910 FILE *debugf; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1911 nfa_state_T *state; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1912 garray_T *indent; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1913 { |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1914 char_u *p; |
4444 | 1915 |
1916 if (state == NULL) | |
1917 return; | |
1918 | |
1919 fprintf(debugf, "(%2d)", abs(state->id)); | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1920 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1921 /* Output indent */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1922 p = (char_u *)indent->ga_data; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1923 if (indent->ga_len >= 3) |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1924 { |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1925 int last = indent->ga_len - 3; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1926 char_u save[2]; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1927 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1928 STRNCPY(save, &p[last], 2); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1929 STRNCPY(&p[last], "+-", 2); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1930 fprintf(debugf, " %s", p); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1931 STRNCPY(&p[last], save, 2); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1932 } |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1933 else |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1934 fprintf(debugf, " %s", p); |
4444 | 1935 |
1936 nfa_set_code(state->c); | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1937 fprintf(debugf, "%s%s (%d) (id=%d)\n", |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1938 state->negated ? "NOT " : "", code, state->c, abs(state->id)); |
4444 | 1939 if (state->id < 0) |
1940 return; | |
1941 | |
1942 state->id = abs(state->id) * -1; | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1943 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1944 /* grow indent for state->out */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1945 indent->ga_len -= 1; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1946 if (state->out1) |
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1947 ga_concat(indent, (char_u *)"| "); |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1948 else |
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1949 ga_concat(indent, (char_u *)" "); |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1950 ga_append(indent, '\0'); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1951 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1952 nfa_print_state2(debugf, state->out, indent); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1953 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1954 /* replace last part of indent for state->out1 */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1955 indent->ga_len -= 3; |
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1956 ga_concat(indent, (char_u *)" "); |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1957 ga_append(indent, '\0'); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1958 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1959 nfa_print_state2(debugf, state->out1, indent); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1960 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1961 /* shrink indent */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1962 indent->ga_len -= 3; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1963 ga_append(indent, '\0'); |
4444 | 1964 } |
1965 | |
1966 /* | |
1967 * Print the NFA state machine. | |
1968 */ | |
1969 static void | |
1970 nfa_dump(prog) | |
1971 nfa_regprog_T *prog; | |
1972 { | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
1973 FILE *debugf = fopen(NFA_REGEXP_DUMP_LOG, "a"); |
4444 | 1974 |
1975 if (debugf != NULL) | |
1976 { | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1977 nfa_print_state(debugf, prog->start); |
4444 | 1978 fclose(debugf); |
1979 } | |
1980 } | |
1981 #endif /* ENABLE_LOG */ | |
1982 #endif /* DEBUG */ | |
1983 | |
1984 /* | |
1985 * Parse r.e. @expr and convert it into postfix form. | |
1986 * Return the postfix string on success, NULL otherwise. | |
1987 */ | |
1988 static int * | |
1989 re2post() | |
1990 { | |
1991 if (nfa_reg(REG_NOPAREN) == FAIL) | |
1992 return NULL; | |
1993 EMIT(NFA_MOPEN); | |
1994 return post_start; | |
1995 } | |
1996 | |
1997 /* NB. Some of the code below is inspired by Russ's. */ | |
1998 | |
1999 /* | |
2000 * Represents an NFA state plus zero or one or two arrows exiting. | |
2001 * if c == MATCH, no arrows out; matching state. | |
2002 * If c == SPLIT, unlabeled arrows to out and out1 (if != NULL). | |
2003 * If c < 256, labeled arrow with character c to out. | |
2004 */ | |
2005 | |
2006 static nfa_state_T *state_ptr; /* points to nfa_prog->state */ | |
2007 | |
2008 /* | |
2009 * Allocate and initialize nfa_state_T. | |
2010 */ | |
2011 static nfa_state_T * | |
2012 new_state(c, out, out1) | |
2013 int c; | |
2014 nfa_state_T *out; | |
2015 nfa_state_T *out1; | |
2016 { | |
2017 nfa_state_T *s; | |
2018 | |
2019 if (istate >= nstate) | |
2020 return NULL; | |
2021 | |
2022 s = &state_ptr[istate++]; | |
2023 | |
2024 s->c = c; | |
2025 s->out = out; | |
2026 s->out1 = out1; | |
2027 | |
2028 s->id = istate; | |
2029 s->lastlist = 0; | |
2030 s->negated = FALSE; | |
2031 | |
2032 return s; | |
2033 } | |
2034 | |
2035 /* | |
2036 * A partially built NFA without the matching state filled in. | |
2037 * Frag_T.start points at the start state. | |
2038 * Frag_T.out is a list of places that need to be set to the | |
2039 * next state for this fragment. | |
2040 */ | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2041 |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2042 /* Since the out pointers in the list are always |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2043 * uninitialized, we use the pointers themselves |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2044 * as storage for the Ptrlists. */ |
4444 | 2045 typedef union Ptrlist Ptrlist; |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2046 union Ptrlist |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2047 { |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2048 Ptrlist *next; |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2049 nfa_state_T *s; |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2050 }; |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2051 |
4444 | 2052 struct Frag |
2053 { | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2054 nfa_state_T *start; |
4444 | 2055 Ptrlist *out; |
2056 }; | |
2057 typedef struct Frag Frag_T; | |
2058 | |
2059 static Frag_T frag __ARGS((nfa_state_T *start, Ptrlist *out)); | |
2060 static Ptrlist *list1 __ARGS((nfa_state_T **outp)); | |
2061 static void patch __ARGS((Ptrlist *l, nfa_state_T *s)); | |
2062 static Ptrlist *append __ARGS((Ptrlist *l1, Ptrlist *l2)); | |
2063 static void st_push __ARGS((Frag_T s, Frag_T **p, Frag_T *stack_end)); | |
2064 static Frag_T st_pop __ARGS((Frag_T **p, Frag_T *stack)); | |
2065 | |
2066 /* | |
4456 | 2067 * Initialize a Frag_T struct and return it. |
4444 | 2068 */ |
2069 static Frag_T | |
2070 frag(start, out) | |
2071 nfa_state_T *start; | |
2072 Ptrlist *out; | |
2073 { | |
4456 | 2074 Frag_T n; |
2075 | |
2076 n.start = start; | |
2077 n.out = out; | |
4444 | 2078 return n; |
2079 } | |
2080 | |
2081 /* | |
2082 * Create singleton list containing just outp. | |
2083 */ | |
2084 static Ptrlist * | |
2085 list1(outp) | |
2086 nfa_state_T **outp; | |
2087 { | |
2088 Ptrlist *l; | |
2089 | |
2090 l = (Ptrlist *)outp; | |
2091 l->next = NULL; | |
2092 return l; | |
2093 } | |
2094 | |
2095 /* | |
2096 * Patch the list of states at out to point to start. | |
2097 */ | |
2098 static void | |
2099 patch(l, s) | |
2100 Ptrlist *l; | |
2101 nfa_state_T *s; | |
2102 { | |
2103 Ptrlist *next; | |
2104 | |
2105 for (; l; l = next) | |
2106 { | |
2107 next = l->next; | |
2108 l->s = s; | |
2109 } | |
2110 } | |
2111 | |
2112 | |
2113 /* | |
2114 * Join the two lists l1 and l2, returning the combination. | |
2115 */ | |
2116 static Ptrlist * | |
2117 append(l1, l2) | |
2118 Ptrlist *l1; | |
2119 Ptrlist *l2; | |
2120 { | |
2121 Ptrlist *oldl1; | |
2122 | |
2123 oldl1 = l1; | |
2124 while (l1->next) | |
2125 l1 = l1->next; | |
2126 l1->next = l2; | |
2127 return oldl1; | |
2128 } | |
2129 | |
2130 /* | |
2131 * Stack used for transforming postfix form into NFA. | |
2132 */ | |
2133 static Frag_T empty; | |
2134 | |
2135 static void | |
2136 st_error(postfix, end, p) | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2137 int *postfix UNUSED; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2138 int *end UNUSED; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2139 int *p UNUSED; |
4444 | 2140 { |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2141 #ifdef NFA_REGEXP_ERROR_LOG |
4444 | 2142 FILE *df; |
2143 int *p2; | |
2144 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2145 df = fopen(NFA_REGEXP_ERROR_LOG, "a"); |
4444 | 2146 if (df) |
2147 { | |
2148 fprintf(df, "Error popping the stack!\n"); | |
2149 #ifdef DEBUG | |
2150 fprintf(df, "Current regexp is \"%s\"\n", nfa_regengine.expr); | |
2151 #endif | |
2152 fprintf(df, "Postfix form is: "); | |
2153 #ifdef DEBUG | |
2154 for (p2 = postfix; p2 < end; p2++) | |
2155 { | |
2156 nfa_set_code(*p2); | |
2157 fprintf(df, "%s, ", code); | |
2158 } | |
2159 nfa_set_code(*p); | |
2160 fprintf(df, "\nCurrent position is: "); | |
2161 for (p2 = postfix; p2 <= p; p2 ++) | |
2162 { | |
2163 nfa_set_code(*p2); | |
2164 fprintf(df, "%s, ", code); | |
2165 } | |
2166 #else | |
2167 for (p2 = postfix; p2 < end; p2++) | |
2168 { | |
2169 fprintf(df, "%d, ", *p2); | |
2170 } | |
2171 fprintf(df, "\nCurrent position is: "); | |
2172 for (p2 = postfix; p2 <= p; p2 ++) | |
2173 { | |
2174 fprintf(df, "%d, ", *p2); | |
2175 } | |
2176 #endif | |
2177 fprintf(df, "\n--------------------------\n"); | |
2178 fclose(df); | |
2179 } | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2180 #endif |
4444 | 2181 EMSG(_("E874: (NFA) Could not pop the stack !")); |
2182 } | |
2183 | |
2184 /* | |
2185 * Push an item onto the stack. | |
2186 */ | |
2187 static void | |
2188 st_push(s, p, stack_end) | |
2189 Frag_T s; | |
2190 Frag_T **p; | |
2191 Frag_T *stack_end; | |
2192 { | |
2193 Frag_T *stackp = *p; | |
2194 | |
2195 if (stackp >= stack_end) | |
2196 return; | |
2197 *stackp = s; | |
2198 *p = *p + 1; | |
2199 } | |
2200 | |
2201 /* | |
2202 * Pop an item from the stack. | |
2203 */ | |
2204 static Frag_T | |
2205 st_pop(p, stack) | |
2206 Frag_T **p; | |
2207 Frag_T *stack; | |
2208 { | |
2209 Frag_T *stackp; | |
2210 | |
2211 *p = *p - 1; | |
2212 stackp = *p; | |
2213 if (stackp < stack) | |
2214 return empty; | |
2215 return **p; | |
2216 } | |
2217 | |
2218 /* | |
2219 * Convert a postfix form into its equivalent NFA. | |
2220 * Return the NFA start state on success, NULL otherwise. | |
2221 */ | |
2222 static nfa_state_T * | |
2223 post2nfa(postfix, end, nfa_calc_size) | |
2224 int *postfix; | |
2225 int *end; | |
2226 int nfa_calc_size; | |
2227 { | |
2228 int *p; | |
2229 int mopen; | |
2230 int mclose; | |
2231 Frag_T *stack = NULL; | |
2232 Frag_T *stackp = NULL; | |
2233 Frag_T *stack_end = NULL; | |
2234 Frag_T e1; | |
2235 Frag_T e2; | |
2236 Frag_T e; | |
2237 nfa_state_T *s; | |
2238 nfa_state_T *s1; | |
2239 nfa_state_T *matchstate; | |
4484 | 2240 nfa_state_T *ret = NULL; |
4444 | 2241 |
2242 if (postfix == NULL) | |
2243 return NULL; | |
2244 | |
4456 | 2245 #define PUSH(s) st_push((s), &stackp, stack_end) |
4444 | 2246 #define POP() st_pop(&stackp, stack); \ |
2247 if (stackp < stack) \ | |
2248 { \ | |
2249 st_error(postfix, end, p); \ | |
2250 return NULL; \ | |
2251 } | |
2252 | |
2253 if (nfa_calc_size == FALSE) | |
2254 { | |
2255 /* Allocate space for the stack. Max states on the stack : nstate */ | |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
2256 stack = (Frag_T *)lalloc((nstate + 1) * sizeof(Frag_T), TRUE); |
4444 | 2257 stackp = stack; |
4462 | 2258 stack_end = stack + (nstate + 1); |
4444 | 2259 } |
2260 | |
2261 for (p = postfix; p < end; ++p) | |
2262 { | |
2263 switch (*p) | |
2264 { | |
2265 case NFA_CONCAT: | |
2266 /* Catenation. | |
2267 * Pay attention: this operator does not exist | |
2268 * in the r.e. itself (it is implicit, really). | |
2269 * It is added when r.e. is translated to postfix | |
2270 * form in re2post(). | |
2271 * | |
2272 * No new state added here. */ | |
2273 if (nfa_calc_size == TRUE) | |
2274 { | |
4458 | 2275 /* nstate += 0; */ |
4444 | 2276 break; |
2277 } | |
2278 e2 = POP(); | |
2279 e1 = POP(); | |
2280 patch(e1.out, e2.start); | |
2281 PUSH(frag(e1.start, e2.out)); | |
2282 break; | |
2283 | |
2284 case NFA_NOT: | |
2285 /* Negation of a character */ | |
2286 if (nfa_calc_size == TRUE) | |
2287 { | |
4458 | 2288 /* nstate += 0; */ |
4444 | 2289 break; |
2290 } | |
2291 e1 = POP(); | |
2292 e1.start->negated = TRUE; | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2293 #ifdef FEAT_MBYTE |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2294 if (e1.start->c == NFA_COMPOSING) |
4444 | 2295 e1.start->out1->negated = TRUE; |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2296 #endif |
4444 | 2297 PUSH(e1); |
2298 break; | |
2299 | |
2300 case NFA_OR: | |
2301 /* Alternation */ | |
2302 if (nfa_calc_size == TRUE) | |
2303 { | |
4458 | 2304 nstate++; |
4444 | 2305 break; |
2306 } | |
2307 e2 = POP(); | |
2308 e1 = POP(); | |
2309 s = new_state(NFA_SPLIT, e1.start, e2.start); | |
2310 if (s == NULL) | |
4484 | 2311 goto theend; |
4444 | 2312 PUSH(frag(s, append(e1.out, e2.out))); |
2313 break; | |
2314 | |
2315 case NFA_STAR: | |
4675
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2316 /* Zero or more, prefer more */ |
4444 | 2317 if (nfa_calc_size == TRUE) |
2318 { | |
4458 | 2319 nstate++; |
4444 | 2320 break; |
2321 } | |
2322 e = POP(); | |
2323 s = new_state(NFA_SPLIT, e.start, NULL); | |
2324 if (s == NULL) | |
4484 | 2325 goto theend; |
4444 | 2326 patch(e.out, s); |
2327 PUSH(frag(s, list1(&s->out1))); | |
2328 break; | |
2329 | |
4675
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2330 case NFA_STAR_NONGREEDY: |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2331 /* Zero or more, prefer zero */ |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2332 if (nfa_calc_size == TRUE) |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2333 { |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2334 nstate++; |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2335 break; |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2336 } |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2337 e = POP(); |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2338 s = new_state(NFA_SPLIT, NULL, e.start); |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2339 if (s == NULL) |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2340 goto theend; |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2341 patch(e.out, s); |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2342 PUSH(frag(s, list1(&s->out))); |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2343 break; |
811a4c9b51d8
updated for version 7.3.1085
Bram Moolenaar <bram@vim.org>
parents:
4673
diff
changeset
|
2344 |
4444 | 2345 case NFA_QUEST: |
2346 /* one or zero atoms=> greedy match */ | |
2347 if (nfa_calc_size == TRUE) | |
2348 { | |
4458 | 2349 nstate++; |
4444 | 2350 break; |
2351 } | |
2352 e = POP(); | |
2353 s = new_state(NFA_SPLIT, e.start, NULL); | |
2354 if (s == NULL) | |
4484 | 2355 goto theend; |
4444 | 2356 PUSH(frag(s, append(e.out, list1(&s->out1)))); |
2357 break; | |
2358 | |
2359 case NFA_QUEST_NONGREEDY: | |
2360 /* zero or one atoms => non-greedy match */ | |
2361 if (nfa_calc_size == TRUE) | |
2362 { | |
4458 | 2363 nstate++; |
4444 | 2364 break; |
2365 } | |
2366 e = POP(); | |
2367 s = new_state(NFA_SPLIT, NULL, e.start); | |
2368 if (s == NULL) | |
4484 | 2369 goto theend; |
4444 | 2370 PUSH(frag(s, append(e.out, list1(&s->out)))); |
2371 break; | |
2372 | |
2373 case NFA_SKIP_CHAR: | |
2374 /* Symbol of 0-length, Used in a repetition | |
2375 * with max/min count of 0 */ | |
2376 if (nfa_calc_size == TRUE) | |
2377 { | |
4458 | 2378 nstate++; |
4444 | 2379 break; |
2380 } | |
2381 s = new_state(NFA_SKIP_CHAR, NULL, NULL); | |
2382 if (s == NULL) | |
4484 | 2383 goto theend; |
4444 | 2384 PUSH(frag(s, list1(&s->out))); |
2385 break; | |
2386 | |
2387 case NFA_PREV_ATOM_NO_WIDTH: | |
4661
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
2388 case NFA_PREV_ATOM_NO_WIDTH_NEG: |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
2389 case NFA_PREV_ATOM_JUST_BEFORE: |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
2390 case NFA_PREV_ATOM_JUST_BEFORE_NEG: |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2391 /* The \@= operator: match the preceding atom with zero width. |
4661
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
2392 * The \@! operator: no match for the preceding atom. |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
2393 * The \@<= operator: match for the preceding atom. |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
2394 * The \@<! operator: no match for the preceding atom. |
4444 | 2395 * Surrounds the preceding atom with START_INVISIBLE and |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2396 * END_INVISIBLE, similarly to MOPEN. */ |
4444 | 2397 |
2398 if (nfa_calc_size == TRUE) | |
2399 { | |
2400 nstate += 2; | |
2401 break; | |
2402 } | |
2403 e = POP(); | |
2404 s1 = new_state(NFA_END_INVISIBLE, NULL, NULL); | |
2405 if (s1 == NULL) | |
4484 | 2406 goto theend; |
4444 | 2407 patch(e.out, s1); |
2408 | |
2409 s = new_state(NFA_START_INVISIBLE, e.start, s1); | |
2410 if (s == NULL) | |
4484 | 2411 goto theend; |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
2412 if (*p == NFA_PREV_ATOM_NO_WIDTH_NEG |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
2413 || *p == NFA_PREV_ATOM_JUST_BEFORE_NEG) |
4661
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
2414 { |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
2415 s->negated = TRUE; |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
2416 s1->negated = TRUE; |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
2417 } |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
2418 if (*p == NFA_PREV_ATOM_JUST_BEFORE |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
2419 || *p == NFA_PREV_ATOM_JUST_BEFORE_NEG) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
2420 { |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
2421 s->val = *++p; /* get the count */ |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
2422 ++s->c; /* NFA_START_INVISIBLE -> NFA_START_INVISIBLE_BEFORE */ |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
2423 } |
4661
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
2424 |
4444 | 2425 PUSH(frag(s, list1(&s1->out))); |
2426 break; | |
2427 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2428 #ifdef FEAT_MBYTE |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2429 case NFA_COMPOSING: /* char with composing char */ |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2430 #if 0 |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2431 /* TODO */ |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2432 if (regflags & RF_ICOMBINE) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2433 { |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
2434 /* use the base character only */ |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2435 } |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2436 #endif |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2437 /* FALLTHROUGH */ |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2438 #endif |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2439 |
4444 | 2440 case NFA_MOPEN + 0: /* Submatch */ |
2441 case NFA_MOPEN + 1: | |
2442 case NFA_MOPEN + 2: | |
2443 case NFA_MOPEN + 3: | |
2444 case NFA_MOPEN + 4: | |
2445 case NFA_MOPEN + 5: | |
2446 case NFA_MOPEN + 6: | |
2447 case NFA_MOPEN + 7: | |
2448 case NFA_MOPEN + 8: | |
2449 case NFA_MOPEN + 9: | |
2450 case NFA_NOPEN: /* \%( "Invisible Submatch" */ | |
2451 if (nfa_calc_size == TRUE) | |
2452 { | |
2453 nstate += 2; | |
2454 break; | |
2455 } | |
2456 | |
2457 mopen = *p; | |
2458 switch (*p) | |
2459 { | |
2460 case NFA_NOPEN: | |
2461 mclose = NFA_NCLOSE; | |
2462 break; | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2463 #ifdef FEAT_MBYTE |
4444 | 2464 case NFA_COMPOSING: |
2465 mclose = NFA_END_COMPOSING; | |
2466 break; | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2467 #endif |
4444 | 2468 default: |
2469 /* NFA_MOPEN(0) ... NFA_MOPEN(9) */ | |
2470 mclose = *p + NSUBEXP; | |
2471 break; | |
2472 } | |
2473 | |
2474 /* Allow "NFA_MOPEN" as a valid postfix representation for | |
2475 * the empty regexp "". In this case, the NFA will be | |
2476 * NFA_MOPEN -> NFA_MCLOSE. Note that this also allows | |
2477 * empty groups of parenthesis, and empty mbyte chars */ | |
2478 if (stackp == stack) | |
2479 { | |
2480 s = new_state(mopen, NULL, NULL); | |
2481 if (s == NULL) | |
4484 | 2482 goto theend; |
4444 | 2483 s1 = new_state(mclose, NULL, NULL); |
2484 if (s1 == NULL) | |
4484 | 2485 goto theend; |
4444 | 2486 patch(list1(&s->out), s1); |
2487 PUSH(frag(s, list1(&s1->out))); | |
2488 break; | |
2489 } | |
2490 | |
2491 /* At least one node was emitted before NFA_MOPEN, so | |
2492 * at least one node will be between NFA_MOPEN and NFA_MCLOSE */ | |
2493 e = POP(); | |
2494 s = new_state(mopen, e.start, NULL); /* `(' */ | |
2495 if (s == NULL) | |
4484 | 2496 goto theend; |
4444 | 2497 |
2498 s1 = new_state(mclose, NULL, NULL); /* `)' */ | |
2499 if (s1 == NULL) | |
4484 | 2500 goto theend; |
4444 | 2501 patch(e.out, s1); |
2502 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2503 #ifdef FEAT_MBYTE |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2504 if (mopen == NFA_COMPOSING) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2505 /* COMPOSING->out1 = END_COMPOSING */ |
4444 | 2506 patch(list1(&s->out1), s1); |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2507 #endif |
4444 | 2508 |
2509 PUSH(frag(s, list1(&s1->out))); | |
2510 break; | |
2511 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2512 case NFA_BACKREF1: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2513 case NFA_BACKREF2: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2514 case NFA_BACKREF3: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2515 case NFA_BACKREF4: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2516 case NFA_BACKREF5: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2517 case NFA_BACKREF6: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2518 case NFA_BACKREF7: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2519 case NFA_BACKREF8: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2520 case NFA_BACKREF9: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2521 if (nfa_calc_size == TRUE) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2522 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2523 nstate += 2; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2524 break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2525 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2526 s = new_state(*p, NULL, NULL); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2527 if (s == NULL) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2528 goto theend; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2529 s1 = new_state(NFA_SKIP, NULL, NULL); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2530 if (s1 == NULL) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2531 goto theend; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2532 patch(list1(&s->out), s1); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2533 PUSH(frag(s, list1(&s1->out))); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2534 break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2535 |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2536 case NFA_LNUM: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2537 case NFA_LNUM_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2538 case NFA_LNUM_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2539 case NFA_VCOL: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2540 case NFA_VCOL_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2541 case NFA_VCOL_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2542 case NFA_COL: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2543 case NFA_COL_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2544 case NFA_COL_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2545 if (nfa_calc_size == TRUE) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2546 { |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2547 nstate += 1; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2548 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2549 } |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2550 e1 = POP(); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2551 s = new_state(*p, NULL, NULL); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2552 if (s == NULL) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2553 goto theend; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2554 s->val = e1.start->c; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2555 PUSH(frag(s, list1(&s->out))); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2556 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2557 |
4444 | 2558 case NFA_ZSTART: |
2559 case NFA_ZEND: | |
2560 default: | |
2561 /* Operands */ | |
2562 if (nfa_calc_size == TRUE) | |
2563 { | |
4458 | 2564 nstate++; |
4444 | 2565 break; |
2566 } | |
2567 s = new_state(*p, NULL, NULL); | |
2568 if (s == NULL) | |
4484 | 2569 goto theend; |
4444 | 2570 PUSH(frag(s, list1(&s->out))); |
2571 break; | |
2572 | |
2573 } /* switch(*p) */ | |
2574 | |
2575 } /* for(p = postfix; *p; ++p) */ | |
2576 | |
2577 if (nfa_calc_size == TRUE) | |
2578 { | |
4458 | 2579 nstate++; |
4484 | 2580 goto theend; /* Return value when counting size is ignored anyway */ |
4444 | 2581 } |
2582 | |
2583 e = POP(); | |
2584 if (stackp != stack) | |
2585 EMSG_RET_NULL(_("E875: (NFA regexp) (While converting from postfix to NFA), too many states left on stack")); | |
2586 | |
2587 if (istate >= nstate) | |
2588 EMSG_RET_NULL(_("E876: (NFA regexp) Not enough space to store the whole NFA ")); | |
2589 | |
2590 matchstate = &state_ptr[istate++]; /* the match state */ | |
2591 matchstate->c = NFA_MATCH; | |
2592 matchstate->out = matchstate->out1 = NULL; | |
2593 | |
2594 patch(e.out, matchstate); | |
4484 | 2595 ret = e.start; |
2596 | |
2597 theend: | |
2598 vim_free(stack); | |
2599 return ret; | |
4444 | 2600 |
2601 #undef POP1 | |
2602 #undef PUSH1 | |
2603 #undef POP2 | |
2604 #undef PUSH2 | |
2605 #undef POP | |
2606 #undef PUSH | |
2607 } | |
2608 | |
2609 /**************************************************************** | |
2610 * NFA execution code. | |
2611 ****************************************************************/ | |
2612 | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2613 typedef struct |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2614 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2615 int in_use; /* number of subexpr with useful info */ |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2616 |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2617 /* When REG_MULTI is TRUE list.multi is used, otherwise list.line. */ |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2618 union |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2619 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2620 struct multipos |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2621 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2622 lpos_T start; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2623 lpos_T end; |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2624 } multi[NSUBEXP]; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2625 struct linepos |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2626 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2627 char_u *start; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2628 char_u *end; |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2629 } line[NSUBEXP]; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2630 } list; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2631 } regsub_T; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2632 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2633 /* nfa_thread_T contains execution information of a NFA state */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2634 typedef struct |
4444 | 2635 { |
2636 nfa_state_T *state; | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2637 int count; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2638 regsub_T sub; /* submatch info, only party used */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2639 } nfa_thread_T; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2640 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2641 /* nfa_list_T contains the alternative NFA execution states. */ |
4444 | 2642 typedef struct |
2643 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2644 nfa_thread_T *t; /* allocated array of states */ |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2645 int n; /* nr of states currently in "t" */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2646 int len; /* max nr of states in "t" */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2647 int id; /* ID of the list */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2648 } nfa_list_T; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2649 |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2650 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2651 static void |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2652 log_subexpr(sub) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2653 regsub_T *sub; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2654 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2655 int j; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2656 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2657 for (j = 0; j < sub->in_use; j++) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2658 if (REG_MULTI) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2659 fprintf(log_fd, "\n *** group %d, start: c=%d, l=%d, end: c=%d, l=%d", |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2660 j, |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2661 sub->list.multi[j].start.col, |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2662 (int)sub->list.multi[j].start.lnum, |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2663 sub->list.multi[j].end.col, |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2664 (int)sub->list.multi[j].end.lnum); |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2665 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2666 fprintf(log_fd, "\n *** group %d, start: \"%s\", end: \"%s\"", |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2667 j, |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2668 (char *)sub->list.line[j].start, |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2669 (char *)sub->list.line[j].end); |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2670 fprintf(log_fd, "\n"); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2671 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2672 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2673 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2674 /* Used during execution: whether a match has been found. */ |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2675 static int nfa_match; |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2676 |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2677 static int sub_equal __ARGS((regsub_T *sub1, regsub_T *sub2)); |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2678 static void addstate __ARGS((nfa_list_T *l, nfa_state_T *state, regsub_T *sub, int off)); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2679 static void addstate_here __ARGS((nfa_list_T *l, nfa_state_T *state, regsub_T *sub, int *ip)); |
4444 | 2680 |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2681 /* |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2682 * Return TRUE if "sub1" and "sub2" have the same positions. |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2683 */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2684 static int |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2685 sub_equal(sub1, sub2) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2686 regsub_T *sub1; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2687 regsub_T *sub2; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2688 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2689 int i; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2690 int todo; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2691 linenr_T s1, e1; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2692 linenr_T s2, e2; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2693 char_u *sp1, *ep1; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2694 char_u *sp2, *ep2; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2695 |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2696 todo = sub1->in_use > sub2->in_use ? sub1->in_use : sub2->in_use; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2697 if (REG_MULTI) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2698 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2699 for (i = 0; i < todo; ++i) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2700 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2701 if (i < sub1->in_use) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2702 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2703 s1 = sub1->list.multi[i].start.lnum; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2704 e1 = sub1->list.multi[i].end.lnum; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2705 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2706 else |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2707 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2708 s1 = 0; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2709 e1 = 0; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2710 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2711 if (i < sub2->in_use) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2712 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2713 s2 = sub2->list.multi[i].start.lnum; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2714 e2 = sub2->list.multi[i].end.lnum; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2715 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2716 else |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2717 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2718 s2 = 0; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2719 e2 = 0; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2720 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2721 if (s1 != s2 || e1 != e2) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2722 return FALSE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2723 if (s1 != 0 && sub1->list.multi[i].start.col |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2724 != sub2->list.multi[i].start.col) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2725 return FALSE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2726 if (e1 != 0 && sub1->list.multi[i].end.col |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2727 != sub2->list.multi[i].end.col) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2728 return FALSE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2729 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2730 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2731 else |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2732 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2733 for (i = 0; i < todo; ++i) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2734 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2735 if (i < sub1->in_use) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2736 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2737 sp1 = sub1->list.line[i].start; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2738 ep1 = sub1->list.line[i].end; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2739 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2740 else |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2741 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2742 sp1 = NULL; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2743 ep1 = NULL; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2744 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2745 if (i < sub2->in_use) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2746 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2747 sp2 = sub2->list.line[i].start; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2748 ep2 = sub2->list.line[i].end; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2749 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2750 else |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2751 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2752 sp2 = NULL; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2753 ep2 = NULL; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2754 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2755 if (sp1 != sp2 || ep1 != ep2) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2756 return FALSE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2757 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2758 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2759 |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2760 return TRUE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2761 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2762 |
4444 | 2763 static void |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2764 addstate(l, state, sub, off) |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2765 nfa_list_T *l; /* runtime state list */ |
4444 | 2766 nfa_state_T *state; /* state to update */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2767 regsub_T *sub; /* pointers to subexpressions */ |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2768 int off; /* byte offset, when -1 go to next line */ |
4444 | 2769 { |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2770 int subidx; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2771 nfa_thread_T *thread; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2772 lpos_T save_lpos; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2773 int save_in_use; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2774 char_u *save_ptr; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2775 int i; |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2776 #ifdef ENABLE_LOG |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2777 int did_print = FALSE; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2778 #endif |
4444 | 2779 |
2780 if (l == NULL || state == NULL) | |
2781 return; | |
2782 | |
2783 switch (state->c) | |
2784 { | |
2785 case NFA_SPLIT: | |
2786 case NFA_NOT: | |
2787 case NFA_NOPEN: | |
2788 case NFA_NCLOSE: | |
2789 case NFA_MCLOSE: | |
2790 case NFA_MCLOSE + 1: | |
2791 case NFA_MCLOSE + 2: | |
2792 case NFA_MCLOSE + 3: | |
2793 case NFA_MCLOSE + 4: | |
2794 case NFA_MCLOSE + 5: | |
2795 case NFA_MCLOSE + 6: | |
2796 case NFA_MCLOSE + 7: | |
2797 case NFA_MCLOSE + 8: | |
2798 case NFA_MCLOSE + 9: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2799 /* These nodes are not added themselves but their "out" and/or |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2800 * "out1" may be added below. */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2801 break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2802 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2803 case NFA_MOPEN: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2804 case NFA_MOPEN + 1: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2805 case NFA_MOPEN + 2: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2806 case NFA_MOPEN + 3: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2807 case NFA_MOPEN + 4: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2808 case NFA_MOPEN + 5: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2809 case NFA_MOPEN + 6: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2810 case NFA_MOPEN + 7: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2811 case NFA_MOPEN + 8: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2812 case NFA_MOPEN + 9: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2813 /* These nodes do not need to be added, but we need to bail out |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2814 * when it was tried to be added to this list before. */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2815 if (state->lastlist == l->id) |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2816 goto skip_add; |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2817 state->lastlist = l->id; |
4444 | 2818 break; |
2819 | |
2820 default: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2821 if (state->lastlist == l->id) |
4444 | 2822 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2823 /* This state is already in the list, don't add it again, |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2824 * unless it is an MOPEN that is used for a backreference. */ |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2825 if (!nfa_has_backref) |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2826 { |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2827 skip_add: |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2828 #ifdef ENABLE_LOG |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2829 nfa_set_code(state->c); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2830 fprintf(log_fd, "> Not adding state %d to list %d. char %d: %s\n", |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2831 abs(state->id), l->id, state->c, code); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2832 #endif |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2833 return; |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2834 } |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2835 |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2836 /* See if the same state is already in the list with the same |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2837 * positions. */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2838 for (i = 0; i < l->n; ++i) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2839 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2840 thread = &l->t[i]; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2841 if (thread->state->id == state->id |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2842 && sub_equal(&thread->sub, sub)) |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2843 goto skip_add; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2844 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2845 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2846 |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2847 /* when there are backreferences the number of states may be (a |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2848 * lot) bigger */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2849 if (nfa_has_backref && l->n == l->len) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2850 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2851 int newlen = l->len * 3 / 2 + 50; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2852 |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2853 l->t = vim_realloc(l->t, newlen * sizeof(nfa_thread_T)); |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2854 l->len = newlen; |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2855 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2856 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2857 /* add the state to the list */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2858 state->lastlist = l->id; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2859 thread = &l->t[l->n++]; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2860 thread->state = state; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2861 thread->sub.in_use = sub->in_use; |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2862 if (sub->in_use > 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2863 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2864 /* Copy the match start and end positions. */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2865 if (REG_MULTI) |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2866 mch_memmove(&thread->sub.list.multi[0], |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2867 &sub->list.multi[0], |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2868 sizeof(struct multipos) * sub->in_use); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2869 else |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2870 mch_memmove(&thread->sub.list.line[0], |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2871 &sub->list.line[0], |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2872 sizeof(struct linepos) * sub->in_use); |
4444 | 2873 } |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2874 #ifdef ENABLE_LOG |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2875 { |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2876 int col; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2877 |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2878 if (thread->sub.in_use <= 0) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2879 col = -1; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2880 else if (REG_MULTI) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2881 col = thread->sub.list.multi[0].start.col; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2882 else |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2883 col = (int)(thread->sub.list.line[0].start - regline); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2884 nfa_set_code(state->c); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2885 fprintf(log_fd, "> Adding state %d to list %d. char %d: %s (start col %d)\n", |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2886 abs(state->id), l->id, state->c, code, col); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2887 did_print = TRUE; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2888 } |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2889 #endif |
4444 | 2890 } |
2891 | |
2892 #ifdef ENABLE_LOG | |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2893 if (!did_print) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2894 { |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2895 int col; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2896 |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2897 if (sub->in_use <= 0) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2898 col = -1; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2899 else if (REG_MULTI) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2900 col = sub->list.multi[0].start.col; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2901 else |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2902 col = (int)(sub->list.line[0].start - regline); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2903 nfa_set_code(state->c); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2904 fprintf(log_fd, "> Processing state %d for list %d. char %d: %s (start col %d)\n", |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2905 abs(state->id), l->id, state->c, code, col); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2906 } |
4444 | 2907 #endif |
2908 switch (state->c) | |
2909 { | |
2910 case NFA_MATCH: | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2911 nfa_match = TRUE; |
4444 | 2912 break; |
2913 | |
2914 case NFA_SPLIT: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2915 addstate(l, state->out, sub, off); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2916 addstate(l, state->out1, sub, off); |
4444 | 2917 break; |
2918 | |
2919 #if 0 | |
2920 case NFA_END_NEG_RANGE: | |
2921 /* Nothing to handle here. nfa_regmatch() will take care of it */ | |
2922 break; | |
2923 | |
2924 case NFA_NOT: | |
2925 EMSG(_("E999: (NFA regexp internal error) Should not process NOT node !")); | |
2926 #ifdef ENABLE_LOG | |
2927 fprintf(f, "\n\n>>> E999: Added state NFA_NOT to a list ... Something went wrong ! Why wasn't it processed already? \n\n"); | |
2928 #endif | |
2929 break; | |
2930 | |
2931 case NFA_COMPOSING: | |
2932 /* nfa_regmatch() will match all the bytes of this composing char. */ | |
2933 break; | |
2934 #endif | |
2935 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2936 case NFA_SKIP_CHAR: |
4444 | 2937 case NFA_NOPEN: |
2938 case NFA_NCLOSE: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2939 addstate(l, state->out, sub, off); |
4444 | 2940 break; |
2941 | |
2942 case NFA_MOPEN + 0: | |
2943 case NFA_MOPEN + 1: | |
2944 case NFA_MOPEN + 2: | |
2945 case NFA_MOPEN + 3: | |
2946 case NFA_MOPEN + 4: | |
2947 case NFA_MOPEN + 5: | |
2948 case NFA_MOPEN + 6: | |
2949 case NFA_MOPEN + 7: | |
2950 case NFA_MOPEN + 8: | |
2951 case NFA_MOPEN + 9: | |
2952 case NFA_ZSTART: | |
2953 if (state->c == NFA_ZSTART) | |
2954 subidx = 0; | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2955 else |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2956 subidx = state->c - NFA_MOPEN; |
4444 | 2957 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2958 /* Set the position (with "off") in the subexpression. Save and |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2959 * restore it when it was in use. Otherwise fill any gap. */ |
4649
9f502d8d5165
updated for version 7.3.1072
Bram Moolenaar <bram@vim.org>
parents:
4647
diff
changeset
|
2960 save_ptr = NULL; |
4444 | 2961 if (REG_MULTI) |
2962 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2963 if (subidx < sub->in_use) |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2964 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2965 save_lpos = sub->list.multi[subidx].start; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2966 save_in_use = -1; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2967 } |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2968 else |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2969 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2970 save_in_use = sub->in_use; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2971 for (i = sub->in_use; i < subidx; ++i) |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2972 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2973 sub->list.multi[i].start.lnum = -1; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2974 sub->list.multi[i].end.lnum = -1; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2975 } |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2976 sub->in_use = subidx + 1; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2977 } |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2978 if (off == -1) |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2979 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2980 sub->list.multi[subidx].start.lnum = reglnum + 1; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2981 sub->list.multi[subidx].start.col = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2982 } |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2983 else |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2984 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2985 sub->list.multi[subidx].start.lnum = reglnum; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2986 sub->list.multi[subidx].start.col = |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2987 (colnr_T)(reginput - regline + off); |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2988 } |
4444 | 2989 } |
2990 else | |
2991 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2992 if (subidx < sub->in_use) |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2993 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2994 save_ptr = sub->list.line[subidx].start; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2995 save_in_use = -1; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2996 } |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2997 else |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2998 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2999 save_in_use = sub->in_use; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3000 for (i = sub->in_use; i < subidx; ++i) |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3001 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3002 sub->list.line[i].start = NULL; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3003 sub->list.line[i].end = NULL; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3004 } |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3005 sub->in_use = subidx + 1; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3006 } |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3007 sub->list.line[subidx].start = reginput + off; |
4444 | 3008 } |
3009 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3010 addstate(l, state->out, sub, off); |
4444 | 3011 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3012 if (save_in_use == -1) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3013 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3014 if (REG_MULTI) |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3015 sub->list.multi[subidx].start = save_lpos; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3016 else |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3017 sub->list.line[subidx].start = save_ptr; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3018 } |
4444 | 3019 else |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3020 sub->in_use = save_in_use; |
4444 | 3021 break; |
3022 | |
3023 case NFA_MCLOSE + 0: | |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3024 if (nfa_has_zend) |
4444 | 3025 { |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
3026 /* Do not overwrite the position set by \ze. If no \ze |
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
3027 * encountered end will be set in nfa_regtry(). */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3028 addstate(l, state->out, sub, off); |
4444 | 3029 break; |
3030 } | |
3031 case NFA_MCLOSE + 1: | |
3032 case NFA_MCLOSE + 2: | |
3033 case NFA_MCLOSE + 3: | |
3034 case NFA_MCLOSE + 4: | |
3035 case NFA_MCLOSE + 5: | |
3036 case NFA_MCLOSE + 6: | |
3037 case NFA_MCLOSE + 7: | |
3038 case NFA_MCLOSE + 8: | |
3039 case NFA_MCLOSE + 9: | |
3040 case NFA_ZEND: | |
3041 if (state->c == NFA_ZEND) | |
3042 subidx = 0; | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3043 else |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3044 subidx = state->c - NFA_MCLOSE; |
4444 | 3045 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3046 /* We don't fill in gaps here, there must have been an MOPEN that |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3047 * has done that. */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3048 save_in_use = sub->in_use; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3049 if (sub->in_use <= subidx) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3050 sub->in_use = subidx + 1; |
4444 | 3051 if (REG_MULTI) |
3052 { | |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3053 save_lpos = sub->list.multi[subidx].end; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3054 if (off == -1) |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3055 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3056 sub->list.multi[subidx].end.lnum = reglnum + 1; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3057 sub->list.multi[subidx].end.col = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3058 } |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3059 else |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3060 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3061 sub->list.multi[subidx].end.lnum = reglnum; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3062 sub->list.multi[subidx].end.col = |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3063 (colnr_T)(reginput - regline + off); |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3064 } |
4444 | 3065 } |
3066 else | |
3067 { | |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3068 save_ptr = sub->list.line[subidx].end; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3069 sub->list.line[subidx].end = reginput + off; |
4444 | 3070 } |
3071 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3072 addstate(l, state->out, sub, off); |
4444 | 3073 |
3074 if (REG_MULTI) | |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3075 sub->list.multi[subidx].end = save_lpos; |
4444 | 3076 else |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3077 sub->list.line[subidx].end = save_ptr; |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3078 sub->in_use = save_in_use; |
4444 | 3079 break; |
3080 } | |
3081 } | |
3082 | |
3083 /* | |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3084 * Like addstate(), but the new state(s) are put at position "*ip". |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3085 * Used for zero-width matches, next state to use is the added one. |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3086 * This makes sure the order of states to be tried does not change, which |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3087 * matters for alternatives. |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3088 */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3089 static void |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3090 addstate_here(l, state, sub, ip) |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3091 nfa_list_T *l; /* runtime state list */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3092 nfa_state_T *state; /* state to update */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3093 regsub_T *sub; /* pointers to subexpressions */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3094 int *ip; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3095 { |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3096 int tlen = l->n; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3097 int count; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3098 int i = *ip; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3099 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3100 /* first add the state(s) at the end, so that we know how many there are */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3101 addstate(l, state, sub, 0); |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3102 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3103 /* when "*ip" was at the end of the list, nothing to do */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3104 if (i + 1 == tlen) |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3105 return; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3106 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3107 /* re-order to put the new state at the current position */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3108 count = l->n - tlen; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3109 if (count == 1) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3110 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3111 /* overwrite the current state */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3112 l->t[i] = l->t[l->n - 1]; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3113 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3114 else if (count > 1) |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3115 { |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3116 /* make space for new states, then move them from the |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3117 * end to the current position */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3118 mch_memmove(&(l->t[i + count]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3119 &(l->t[i + 1]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3120 sizeof(nfa_thread_T) * (l->n - i - 1)); |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3121 mch_memmove(&(l->t[i]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3122 &(l->t[l->n - 1]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3123 sizeof(nfa_thread_T) * count); |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3124 } |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3125 --l->n; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3126 *ip = i - 1; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3127 } |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3128 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3129 /* |
4444 | 3130 * Check character class "class" against current character c. |
3131 */ | |
3132 static int | |
3133 check_char_class(class, c) | |
3134 int class; | |
3135 int c; | |
3136 { | |
3137 switch (class) | |
3138 { | |
3139 case NFA_CLASS_ALNUM: | |
4464 | 3140 if (c >= 1 && c <= 255 && isalnum(c)) |
4444 | 3141 return OK; |
3142 break; | |
3143 case NFA_CLASS_ALPHA: | |
4464 | 3144 if (c >= 1 && c <= 255 && isalpha(c)) |
4444 | 3145 return OK; |
3146 break; | |
3147 case NFA_CLASS_BLANK: | |
3148 if (c == ' ' || c == '\t') | |
3149 return OK; | |
3150 break; | |
3151 case NFA_CLASS_CNTRL: | |
4464 | 3152 if (c >= 1 && c <= 255 && iscntrl(c)) |
4444 | 3153 return OK; |
3154 break; | |
3155 case NFA_CLASS_DIGIT: | |
3156 if (VIM_ISDIGIT(c)) | |
3157 return OK; | |
3158 break; | |
3159 case NFA_CLASS_GRAPH: | |
4464 | 3160 if (c >= 1 && c <= 255 && isgraph(c)) |
4444 | 3161 return OK; |
3162 break; | |
3163 case NFA_CLASS_LOWER: | |
3164 if (MB_ISLOWER(c)) | |
3165 return OK; | |
3166 break; | |
3167 case NFA_CLASS_PRINT: | |
3168 if (vim_isprintc(c)) | |
3169 return OK; | |
3170 break; | |
3171 case NFA_CLASS_PUNCT: | |
4464 | 3172 if (c >= 1 && c <= 255 && ispunct(c)) |
4444 | 3173 return OK; |
3174 break; | |
3175 case NFA_CLASS_SPACE: | |
3176 if ((c >=9 && c <= 13) || (c == ' ')) | |
3177 return OK; | |
3178 break; | |
3179 case NFA_CLASS_UPPER: | |
3180 if (MB_ISUPPER(c)) | |
3181 return OK; | |
3182 break; | |
3183 case NFA_CLASS_XDIGIT: | |
3184 if (vim_isxdigit(c)) | |
3185 return OK; | |
3186 break; | |
3187 case NFA_CLASS_TAB: | |
3188 if (c == '\t') | |
3189 return OK; | |
3190 break; | |
3191 case NFA_CLASS_RETURN: | |
3192 if (c == '\r') | |
3193 return OK; | |
3194 break; | |
3195 case NFA_CLASS_BACKSPACE: | |
3196 if (c == '\b') | |
3197 return OK; | |
3198 break; | |
3199 case NFA_CLASS_ESCAPE: | |
3200 if (c == '\033') | |
3201 return OK; | |
3202 break; | |
3203 | |
3204 default: | |
3205 /* should not be here :P */ | |
3206 EMSG_RET_FAIL(_("E877: (NFA regexp) Invalid character class ")); | |
3207 } | |
3208 return FAIL; | |
3209 } | |
3210 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3211 static int match_backref __ARGS((regsub_T *sub, int subidx, int *bytelen)); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3212 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3213 /* |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3214 * Check for a match with subexpression "subidx". |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3215 * return TRUE if it matches. |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3216 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3217 static int |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3218 match_backref(sub, subidx, bytelen) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3219 regsub_T *sub; /* pointers to subexpressions */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3220 int subidx; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3221 int *bytelen; /* out: length of match in bytes */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3222 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3223 int len; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3224 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3225 if (sub->in_use <= subidx) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3226 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3227 retempty: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3228 /* backref was not set, match an empty string */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3229 *bytelen = 0; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3230 return TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3231 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3232 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3233 if (REG_MULTI) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3234 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3235 if (sub->list.multi[subidx].start.lnum < 0 |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3236 || sub->list.multi[subidx].end.lnum < 0) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3237 goto retempty; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3238 /* TODO: line breaks */ |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3239 len = sub->list.multi[subidx].end.col |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3240 - sub->list.multi[subidx].start.col; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3241 if (cstrncmp(regline + sub->list.multi[subidx].start.col, |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3242 reginput, &len) == 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3243 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3244 *bytelen = len; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3245 return TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3246 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3247 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3248 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3249 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3250 if (sub->list.line[subidx].start == NULL |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3251 || sub->list.line[subidx].end == NULL) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3252 goto retempty; |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3253 len = (int)(sub->list.line[subidx].end - sub->list.line[subidx].start); |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3254 if (cstrncmp(sub->list.line[subidx].start, reginput, &len) == 0) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3255 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3256 *bytelen = len; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3257 return TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3258 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3259 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3260 return FALSE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3261 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3262 |
4444 | 3263 /* |
3264 * Set all NFA nodes' list ID equal to -1. | |
3265 */ | |
3266 static void | |
3267 nfa_set_neg_listids(start) | |
3268 nfa_state_T *start; | |
3269 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3270 if (start != NULL && start->lastlist >= 0) |
4444 | 3271 { |
3272 start->lastlist = -1; | |
3273 nfa_set_neg_listids(start->out); | |
3274 nfa_set_neg_listids(start->out1); | |
3275 } | |
3276 } | |
3277 | |
3278 /* | |
3279 * Set all NFA nodes' list ID equal to 0. | |
3280 */ | |
3281 static void | |
3282 nfa_set_null_listids(start) | |
3283 nfa_state_T *start; | |
3284 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3285 if (start != NULL && start->lastlist == -1) |
4444 | 3286 { |
3287 start->lastlist = 0; | |
3288 nfa_set_null_listids(start->out); | |
3289 nfa_set_null_listids(start->out1); | |
3290 } | |
3291 } | |
3292 | |
3293 /* | |
3294 * Save list IDs for all NFA states in "list". | |
3295 */ | |
3296 static void | |
3297 nfa_save_listids(start, list) | |
3298 nfa_state_T *start; | |
3299 int *list; | |
3300 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3301 if (start != NULL && start->lastlist != -1) |
4444 | 3302 { |
3303 list[abs(start->id)] = start->lastlist; | |
3304 start->lastlist = -1; | |
3305 nfa_save_listids(start->out, list); | |
3306 nfa_save_listids(start->out1, list); | |
3307 } | |
3308 } | |
3309 | |
3310 /* | |
3311 * Restore list IDs from "list" to all NFA states. | |
3312 */ | |
3313 static void | |
3314 nfa_restore_listids(start, list) | |
3315 nfa_state_T *start; | |
3316 int *list; | |
3317 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3318 if (start != NULL && start->lastlist == -1) |
4444 | 3319 { |
3320 start->lastlist = list[abs(start->id)]; | |
3321 nfa_restore_listids(start->out, list); | |
3322 nfa_restore_listids(start->out1, list); | |
3323 } | |
3324 } | |
3325 | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3326 static int |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3327 nfa_re_num_cmp(val, op, pos) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3328 long_u val; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3329 int op; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3330 long_u pos; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3331 { |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3332 if (op == 1) return pos > val; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3333 if (op == 2) return pos < val; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3334 return val == pos; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3335 } |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3336 |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3337 static int nfa_regmatch __ARGS((nfa_state_T *start, regsub_T *submatch, regsub_T *m, save_se_T *endp)); |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3338 |
4444 | 3339 /* |
3340 * Main matching routine. | |
3341 * | |
3342 * Run NFA to determine whether it matches reginput. | |
3343 * | |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3344 * When "endp" is not NULL it is a required end-of-match position. |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3345 * |
4444 | 3346 * Return TRUE if there is a match, FALSE otherwise. |
3347 * Note: Caller must ensure that: start != NULL. | |
3348 */ | |
3349 static int | |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3350 nfa_regmatch(start, submatch, m, endp) |
4444 | 3351 nfa_state_T *start; |
3352 regsub_T *submatch; | |
3353 regsub_T *m; | |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3354 save_se_T *endp; |
4444 | 3355 { |
3356 int result; | |
3357 int size = 0; | |
3358 int flag = 0; | |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3359 int go_to_nextline = FALSE; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3360 nfa_thread_T *t; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3361 nfa_list_T list[3]; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3362 nfa_list_T *listtbl[2][2]; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3363 nfa_list_T *ll; |
4444 | 3364 int listid = 1; |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3365 int listidx; |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3366 nfa_list_T *thislist; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3367 nfa_list_T *nextlist; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3368 nfa_list_T *neglist; |
4444 | 3369 int *listids = NULL; |
4460 | 3370 #ifdef NFA_REGEXP_DEBUG_LOG |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3371 FILE *debug = fopen(NFA_REGEXP_DEBUG_LOG, "a"); |
4444 | 3372 |
3373 if (debug == NULL) | |
3374 { | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3375 EMSG2(_("(NFA) COULD NOT OPEN %s !"), NFA_REGEXP_DEBUG_LOG); |
4444 | 3376 return FALSE; |
3377 } | |
3378 #endif | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3379 nfa_match = FALSE; |
4444 | 3380 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3381 /* Allocate memory for the lists of nodes. */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3382 size = (nstate + 1) * sizeof(nfa_thread_T); |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3383 list[0].t = (nfa_thread_T *)lalloc_clear(size, TRUE); |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3384 list[0].len = nstate + 1; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3385 list[1].t = (nfa_thread_T *)lalloc_clear(size, TRUE); |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3386 list[1].len = nstate + 1; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3387 list[2].t = (nfa_thread_T *)lalloc_clear(size, TRUE); |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3388 list[2].len = nstate + 1; |
4444 | 3389 if (list[0].t == NULL || list[1].t == NULL || list[2].t == NULL) |
3390 goto theend; | |
3391 | |
3392 #ifdef ENABLE_LOG | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3393 log_fd = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 3394 if (log_fd != NULL) |
3395 { | |
3396 fprintf(log_fd, "**********************************\n"); | |
3397 nfa_set_code(start->c); | |
3398 fprintf(log_fd, " RUNNING nfa_regmatch() starting with state %d, code %s\n", | |
3399 abs(start->id), code); | |
3400 fprintf(log_fd, "**********************************\n"); | |
3401 } | |
3402 else | |
3403 { | |
3404 EMSG(_("Could not open temporary log file for writing, displaying on stderr ... ")); | |
3405 log_fd = stderr; | |
3406 } | |
3407 #endif | |
3408 | |
3409 thislist = &list[0]; | |
3410 thislist->n = 0; | |
3411 nextlist = &list[1]; | |
3412 nextlist->n = 0; | |
3413 neglist = &list[2]; | |
3414 neglist->n = 0; | |
3415 #ifdef ENABLE_LOG | |
3416 fprintf(log_fd, "(---) STARTSTATE\n"); | |
3417 #endif | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3418 thislist->id = listid; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3419 addstate(thislist, start, m, 0); |
4444 | 3420 |
3421 /* There are two cases when the NFA advances: 1. input char matches the | |
3422 * NFA node and 2. input char does not match the NFA node, but the next | |
3423 * node is NFA_NOT. The following macro calls addstate() according to | |
3424 * these rules. It is used A LOT, so use the "listtbl" table for speed */ | |
3425 listtbl[0][0] = NULL; | |
3426 listtbl[0][1] = neglist; | |
3427 listtbl[1][0] = nextlist; | |
3428 listtbl[1][1] = NULL; | |
3429 #define ADD_POS_NEG_STATE(node) \ | |
3430 ll = listtbl[result ? 1 : 0][node->negated]; \ | |
3431 if (ll != NULL) \ | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3432 addstate(ll, node->out , &t->sub, clen); |
4444 | 3433 |
3434 | |
3435 /* | |
3436 * Run for each character. | |
3437 */ | |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3438 for (;;) |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3439 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3440 int curc; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3441 int clen; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3442 |
4444 | 3443 #ifdef FEAT_MBYTE |
3444 if (has_mbyte) | |
3445 { | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3446 curc = (*mb_ptr2char)(reginput); |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3447 clen = (*mb_ptr2len)(reginput); |
4444 | 3448 } |
3449 else | |
3450 #endif | |
3451 { | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3452 curc = *reginput; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3453 clen = 1; |
4444 | 3454 } |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3455 if (curc == NUL) |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3456 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3457 clen = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3458 go_to_nextline = FALSE; |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3459 } |
4444 | 3460 |
3461 /* swap lists */ | |
3462 thislist = &list[flag]; | |
3463 nextlist = &list[flag ^= 1]; | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3464 nextlist->n = 0; /* clear nextlist */ |
4444 | 3465 listtbl[1][0] = nextlist; |
3466 ++listid; | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3467 thislist->id = listid; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3468 nextlist->id = listid + 1; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3469 neglist->id = listid + 1; |
4444 | 3470 |
3471 #ifdef ENABLE_LOG | |
3472 fprintf(log_fd, "------------------------------------------\n"); | |
3473 fprintf(log_fd, ">>> Reginput is \"%s\"\n", reginput); | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3474 fprintf(log_fd, ">>> Advanced one character ... Current char is %c (code %d) \n", curc, (int)curc); |
4444 | 3475 fprintf(log_fd, ">>> Thislist has %d states available: ", thislist->n); |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3476 { |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3477 int i; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3478 |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3479 for (i = 0; i < thislist->n; i++) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3480 fprintf(log_fd, "%d ", abs(thislist->t[i].state->id)); |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3481 } |
4444 | 3482 fprintf(log_fd, "\n"); |
3483 #endif | |
3484 | |
4460 | 3485 #ifdef NFA_REGEXP_DEBUG_LOG |
4444 | 3486 fprintf(debug, "\n-------------------\n"); |
3487 #endif | |
4480 | 3488 /* |
3489 * If the state lists are empty we can stop. | |
3490 */ | |
3491 if (thislist->n == 0 && neglist->n == 0) | |
3492 break; | |
4444 | 3493 |
3494 /* compute nextlist */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3495 for (listidx = 0; listidx < thislist->n || neglist->n > 0; ++listidx) |
4444 | 3496 { |
3497 if (neglist->n > 0) | |
3498 { | |
3499 t = &neglist->t[0]; | |
4468 | 3500 neglist->n--; |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3501 listidx--; |
4444 | 3502 } |
3503 else | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3504 t = &thislist->t[listidx]; |
4444 | 3505 |
4460 | 3506 #ifdef NFA_REGEXP_DEBUG_LOG |
4444 | 3507 nfa_set_code(t->state->c); |
3508 fprintf(debug, "%s, ", code); | |
3509 #endif | |
3510 #ifdef ENABLE_LOG | |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3511 { |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3512 int col; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3513 |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3514 if (t->sub.in_use <= 0) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3515 col = -1; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3516 else if (REG_MULTI) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3517 col = t->sub.list.multi[0].start.col; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3518 else |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3519 col = (int)(t->sub.list.line[0].start - regline); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3520 nfa_set_code(t->state->c); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3521 fprintf(log_fd, "(%d) char %d %s (start col %d) ... \n", |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3522 abs(t->state->id), (int)t->state->c, code, col); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3523 } |
4444 | 3524 #endif |
3525 | |
3526 /* | |
3527 * Handle the possible codes of the current state. | |
3528 * The most important is NFA_MATCH. | |
3529 */ | |
3530 switch (t->state->c) | |
3531 { | |
3532 case NFA_MATCH: | |
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3533 { |
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3534 int j; |
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3535 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3536 nfa_match = TRUE; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3537 submatch->in_use = t->sub.in_use; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3538 if (REG_MULTI) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3539 for (j = 0; j < submatch->in_use; j++) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3540 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3541 submatch->list.multi[j].start = |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3542 t->sub.list.multi[j].start; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3543 submatch->list.multi[j].end = t->sub.list.multi[j].end; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3544 } |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3545 else |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3546 for (j = 0; j < submatch->in_use; j++) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3547 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3548 submatch->list.line[j].start = |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3549 t->sub.list.line[j].start; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3550 submatch->list.line[j].end = t->sub.list.line[j].end; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3551 } |
4444 | 3552 #ifdef ENABLE_LOG |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3553 log_subexpr(&t->sub); |
4444 | 3554 #endif |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3555 /* Found the left-most longest match, do not look at any other |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3556 * states at this position. When the list of states is going |
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3557 * to be empty quit without advancing, so that "reginput" is |
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3558 * correct. */ |
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3559 if (nextlist->n == 0 && neglist->n == 0) |
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3560 clen = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3561 goto nextchar; |
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3562 } |
4444 | 3563 |
3564 case NFA_END_INVISIBLE: | |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3565 /* This is only encountered after a NFA_START_INVISIBLE or |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3566 * NFA_START_INVISIBLE_BEFORE node. |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3567 * They surround a zero-width group, used with "\@=", "\&", |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3568 * "\@!", "\@<=" and "\@<!". |
4444 | 3569 * If we got here, it means that the current "invisible" group |
3570 * finished successfully, so return control to the parent | |
3571 * nfa_regmatch(). Submatches are stored in *m, and used in | |
3572 * the parent call. */ | |
3573 if (start->c == NFA_MOPEN + 0) | |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3574 /* TODO: do we ever get here? */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3575 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3576 else |
3577 { | |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3578 #ifdef ENABLE_LOG |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3579 if (endp != NULL) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3580 { |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3581 if (REG_MULTI) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3582 fprintf(log_fd, "Current lnum: %d, endp lnum: %d; current col: %d, endp col: %d\n", |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3583 (int)reglnum, |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3584 (int)endp->se_u.pos.lnum, |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3585 (int)(reginput - regline), |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3586 endp->se_u.pos.col); |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3587 else |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3588 fprintf(log_fd, "Current col: %d, endp col: %d\n", |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3589 (int)(reginput - regline), |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3590 (int)(endp->se_u.ptr - reginput)); |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3591 } |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3592 #endif |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3593 /* It's only a match if it ends at "endp" */ |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3594 if (endp != NULL && (REG_MULTI |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3595 ? (reglnum != endp->se_u.pos.lnum |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3596 || (int)(reginput - regline) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3597 != endp->se_u.pos.col) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3598 : reginput != endp->se_u.ptr)) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3599 break; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3600 |
4661
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
3601 /* do not set submatches for \@! */ |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
3602 if (!t->state->negated) |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
3603 /* TODO: only copy positions in use. */ |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
3604 *m = t->sub; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3605 nfa_match = TRUE; |
4444 | 3606 } |
3607 break; | |
3608 | |
3609 case NFA_START_INVISIBLE: | |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3610 case NFA_START_INVISIBLE_BEFORE: |
4669
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3611 { |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3612 char_u *save_reginput = reginput; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3613 char_u *save_regline = regline; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3614 int save_reglnum = reglnum; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3615 int save_nfa_match = nfa_match; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3616 save_se_T endpos; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3617 save_se_T *endposp = NULL; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3618 |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3619 if (t->state->c == NFA_START_INVISIBLE_BEFORE) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3620 { |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3621 /* The recursive match must end at the current position. */ |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3622 endposp = &endpos; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3623 if (REG_MULTI) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3624 { |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3625 endpos.se_u.pos.col = (int)(reginput - regline); |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3626 endpos.se_u.pos.lnum = reglnum; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3627 } |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3628 else |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3629 endpos.se_u.ptr = reginput; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3630 |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3631 /* Go back the specified number of bytes, or as far as the |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3632 * start of the previous line, to try matching "\@<=" or |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3633 * not matching "\@<!". */ |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3634 if (t->state->val <= 0) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3635 { |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3636 if (REG_MULTI) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3637 { |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3638 regline = reg_getline(--reglnum); |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3639 if (regline == NULL) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3640 /* can't go before the first line */ |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3641 regline = reg_getline(++reglnum); |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3642 } |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3643 reginput = regline; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3644 } |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3645 else |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3646 { |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3647 if (REG_MULTI |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3648 && (int)(reginput - regline) < t->state->val) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3649 { |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3650 /* Not enough bytes in this line, go to end of |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3651 * previous line. */ |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3652 regline = reg_getline(--reglnum); |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3653 if (regline == NULL) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3654 { |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3655 /* can't go before the first line */ |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3656 regline = reg_getline(++reglnum); |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3657 reginput = regline; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3658 } |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3659 else |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3660 reginput = regline + STRLEN(regline); |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3661 } |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3662 if ((int)(reginput - regline) >= t->state->val) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3663 { |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3664 reginput -= t->state->val; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3665 #ifdef FEAT_MBYTE |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3666 if (has_mbyte) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3667 reginput -= mb_head_off(regline, reginput); |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3668 #endif |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3669 } |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3670 else |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3671 reginput = regline; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3672 } |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3673 } |
4669
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3674 |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3675 /* Call nfa_regmatch() to check if the current concat matches |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3676 * at this position. The concat ends with the node |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3677 * NFA_END_INVISIBLE */ |
4444 | 3678 if (listids == NULL) |
3679 { | |
4669
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3680 listids = (int *)lalloc(sizeof(int) * nstate, TRUE); |
4444 | 3681 if (listids == NULL) |
3682 { | |
3683 EMSG(_("E878: (NFA) Could not allocate memory for branch traversal!")); | |
3684 return 0; | |
3685 } | |
3686 } | |
3687 #ifdef ENABLE_LOG | |
3688 if (log_fd != stderr) | |
3689 fclose(log_fd); | |
3690 log_fd = NULL; | |
3691 #endif | |
3692 /* Have to clear the listid field of the NFA nodes, so that | |
3693 * nfa_regmatch() and addstate() can run properly after | |
3694 * recursion. */ | |
3695 nfa_save_listids(start, listids); | |
3696 nfa_set_null_listids(start); | |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
3697 result = nfa_regmatch(t->state->out, submatch, m, endposp); |
4444 | 3698 nfa_set_neg_listids(start); |
3699 nfa_restore_listids(start, listids); | |
4669
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3700 |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3701 /* restore position in input text */ |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3702 reginput = save_reginput; |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3703 regline = save_regline; |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3704 reglnum = save_reglnum; |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3705 nfa_match = save_nfa_match; |
4444 | 3706 |
3707 #ifdef ENABLE_LOG | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3708 log_fd = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 3709 if (log_fd != NULL) |
3710 { | |
3711 fprintf(log_fd, "****************************\n"); | |
3712 fprintf(log_fd, "FINISHED RUNNING nfa_regmatch() recursively\n"); | |
3713 fprintf(log_fd, "MATCH = %s\n", result == TRUE ? "OK" : "FALSE"); | |
3714 fprintf(log_fd, "****************************\n"); | |
3715 } | |
3716 else | |
3717 { | |
3718 EMSG(_("Could not open temporary log file for writing, displaying on stderr ... ")); | |
3719 log_fd = stderr; | |
3720 } | |
3721 #endif | |
4661
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
3722 /* for \@! it is a match when result is FALSE */ |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
3723 if (result != t->state->negated) |
4444 | 3724 { |
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3725 int j; |
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3726 |
4444 | 3727 /* Copy submatch info from the recursive call */ |
3728 if (REG_MULTI) | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3729 for (j = 1; j < m->in_use; j++) |
4444 | 3730 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3731 t->sub.list.multi[j].start = m->list.multi[j].start; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3732 t->sub.list.multi[j].end = m->list.multi[j].end; |
4444 | 3733 } |
3734 else | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3735 for (j = 1; j < m->in_use; j++) |
4444 | 3736 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3737 t->sub.list.line[j].start = m->list.line[j].start; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3738 t->sub.list.line[j].end = m->list.line[j].end; |
4444 | 3739 } |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3740 if (m->in_use > t->sub.in_use) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3741 t->sub.in_use = m->in_use; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3742 |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3743 /* t->state->out1 is the corresponding END_INVISIBLE node; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3744 * Add it to the current list (zero-width match). */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3745 addstate_here(thislist, t->state->out1->out, &t->sub, |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3746 &listidx); |
4444 | 3747 } |
3748 break; | |
4669
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3749 } |
4444 | 3750 |
3751 case NFA_BOL: | |
3752 if (reginput == regline) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3753 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3754 break; |
3755 | |
3756 case NFA_EOL: | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3757 if (curc == NUL) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3758 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3759 break; |
3760 | |
3761 case NFA_BOW: | |
3762 { | |
3763 int bow = TRUE; | |
3764 | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3765 if (curc == NUL) |
4444 | 3766 bow = FALSE; |
3767 #ifdef FEAT_MBYTE | |
3768 else if (has_mbyte) | |
3769 { | |
3770 int this_class; | |
3771 | |
3772 /* Get class of current and previous char (if it exists). */ | |
4503 | 3773 this_class = mb_get_class_buf(reginput, reg_buf); |
4444 | 3774 if (this_class <= 1) |
3775 bow = FALSE; | |
3776 else if (reg_prev_class() == this_class) | |
3777 bow = FALSE; | |
3778 } | |
3779 #endif | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3780 else if (!vim_iswordc_buf(curc, reg_buf) |
4503 | 3781 || (reginput > regline |
3782 && vim_iswordc_buf(reginput[-1], reg_buf))) | |
4444 | 3783 bow = FALSE; |
3784 if (bow) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3785 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3786 break; |
3787 } | |
3788 | |
3789 case NFA_EOW: | |
3790 { | |
3791 int eow = TRUE; | |
3792 | |
3793 if (reginput == regline) | |
3794 eow = FALSE; | |
3795 #ifdef FEAT_MBYTE | |
3796 else if (has_mbyte) | |
3797 { | |
3798 int this_class, prev_class; | |
3799 | |
3800 /* Get class of current and previous char (if it exists). */ | |
4503 | 3801 this_class = mb_get_class_buf(reginput, reg_buf); |
4444 | 3802 prev_class = reg_prev_class(); |
3803 if (this_class == prev_class | |
3804 || prev_class == 0 || prev_class == 1) | |
3805 eow = FALSE; | |
3806 } | |
3807 #endif | |
4503 | 3808 else if (!vim_iswordc_buf(reginput[-1], reg_buf) |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3809 || (reginput[0] != NUL |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3810 && vim_iswordc_buf(curc, reg_buf))) |
4444 | 3811 eow = FALSE; |
3812 if (eow) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3813 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3814 break; |
3815 } | |
3816 | |
4671
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3817 case NFA_BOF: |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3818 if (reglnum == 0 && reginput == regline |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3819 && (!REG_MULTI || reg_firstlnum == 1)) |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3820 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3821 break; |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3822 |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3823 case NFA_EOF: |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3824 if (reglnum == reg_maxline && curc == NUL) |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3825 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3826 break; |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3827 |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3828 #ifdef FEAT_MBYTE |
4444 | 3829 case NFA_COMPOSING: |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3830 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3831 int mc = curc; |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3832 int len = 0; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3833 nfa_state_T *end; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3834 nfa_state_T *sta; |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3835 int cchars[MAX_MCO]; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3836 int ccount = 0; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3837 int j; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3838 |
4444 | 3839 sta = t->state->out; |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3840 len = 0; |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3841 if (utf_iscomposing(sta->c)) |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3842 { |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3843 /* Only match composing character(s), ignore base |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3844 * character. Used for ".{composing}" and "{composing}" |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3845 * (no preceding character). */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3846 len += mb_char2len(mc); |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3847 } |
4551
f0c37f5cf4ea
updated for version 7.3.1023
Bram Moolenaar <bram@vim.org>
parents:
4549
diff
changeset
|
3848 if (ireg_icombine && len == 0) |
4444 | 3849 { |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3850 /* If \Z was present, then ignore composing characters. |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3851 * When ignoring the base character this always matches. */ |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3852 /* TODO: How about negated? */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3853 if (len == 0 && sta->c != curc) |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3854 result = FAIL; |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3855 else |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3856 result = OK; |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3857 while (sta->c != NFA_END_COMPOSING) |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3858 sta = sta->out; |
4444 | 3859 } |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3860 |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3861 /* Check base character matches first, unless ignored. */ |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3862 else if (len > 0 || mc == sta->c) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3863 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3864 if (len == 0) |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3865 { |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3866 len += mb_char2len(mc); |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3867 sta = sta->out; |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3868 } |
4444 | 3869 |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3870 /* We don't care about the order of composing characters. |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3871 * Get them into cchars[] first. */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3872 while (len < clen) |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3873 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3874 mc = mb_ptr2char(reginput + len); |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3875 cchars[ccount++] = mc; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3876 len += mb_char2len(mc); |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3877 if (ccount == MAX_MCO) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3878 break; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3879 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3880 |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3881 /* Check that each composing char in the pattern matches a |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3882 * composing char in the text. We do not check if all |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3883 * composing chars are matched. */ |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3884 result = OK; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3885 while (sta->c != NFA_END_COMPOSING) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3886 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3887 for (j = 0; j < ccount; ++j) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3888 if (cchars[j] == sta->c) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3889 break; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3890 if (j == ccount) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3891 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3892 result = FAIL; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3893 break; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3894 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3895 sta = sta->out; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3896 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3897 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3898 else |
4525
36ddcf4cecbc
updated for version 7.3.1010
Bram Moolenaar <bram@vim.org>
parents:
4517
diff
changeset
|
3899 result = FAIL; |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3900 |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3901 end = t->state->out1; /* NFA_END_COMPOSING */ |
4444 | 3902 ADD_POS_NEG_STATE(end); |
3903 break; | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3904 } |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3905 #endif |
4444 | 3906 |
3907 case NFA_NEWL: | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3908 if (curc == NUL && !reg_line_lbr && REG_MULTI |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3909 && reglnum <= reg_maxline) |
4444 | 3910 { |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3911 go_to_nextline = TRUE; |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3912 /* Pass -1 for the offset, which means taking the position |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3913 * at the start of the next line. */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3914 addstate(nextlist, t->state->out, &t->sub, -1); |
4444 | 3915 } |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3916 else if (curc == '\n' && reg_line_lbr) |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3917 { |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3918 /* match \n as if it is an ordinary character */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3919 addstate(nextlist, t->state->out, &t->sub, 1); |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3920 } |
4444 | 3921 break; |
3922 | |
3923 case NFA_CLASS_ALNUM: | |
3924 case NFA_CLASS_ALPHA: | |
3925 case NFA_CLASS_BLANK: | |
3926 case NFA_CLASS_CNTRL: | |
3927 case NFA_CLASS_DIGIT: | |
3928 case NFA_CLASS_GRAPH: | |
3929 case NFA_CLASS_LOWER: | |
3930 case NFA_CLASS_PRINT: | |
3931 case NFA_CLASS_PUNCT: | |
3932 case NFA_CLASS_SPACE: | |
3933 case NFA_CLASS_UPPER: | |
3934 case NFA_CLASS_XDIGIT: | |
3935 case NFA_CLASS_TAB: | |
3936 case NFA_CLASS_RETURN: | |
3937 case NFA_CLASS_BACKSPACE: | |
3938 case NFA_CLASS_ESCAPE: | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3939 result = check_char_class(t->state->c, curc); |
4444 | 3940 ADD_POS_NEG_STATE(t->state); |
3941 break; | |
3942 | |
3943 case NFA_END_NEG_RANGE: | |
3944 /* This follows a series of negated nodes, like: | |
3945 * CHAR(x), NFA_NOT, CHAR(y), NFA_NOT etc. */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3946 if (curc > 0) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3947 addstate(nextlist, t->state->out, &t->sub, clen); |
4444 | 3948 break; |
3949 | |
3950 case NFA_ANY: | |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3951 /* Any char except '\0', (end of input) does not match. */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3952 if (curc > 0) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3953 addstate(nextlist, t->state->out, &t->sub, clen); |
4444 | 3954 break; |
3955 | |
3956 /* | |
3957 * Character classes like \a for alpha, \d for digit etc. | |
3958 */ | |
3959 case NFA_IDENT: /* \i */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3960 result = vim_isIDc(curc); |
4444 | 3961 ADD_POS_NEG_STATE(t->state); |
3962 break; | |
3963 | |
3964 case NFA_SIDENT: /* \I */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3965 result = !VIM_ISDIGIT(curc) && vim_isIDc(curc); |
4444 | 3966 ADD_POS_NEG_STATE(t->state); |
3967 break; | |
3968 | |
3969 case NFA_KWORD: /* \k */ | |
4503 | 3970 result = vim_iswordp_buf(reginput, reg_buf); |
4444 | 3971 ADD_POS_NEG_STATE(t->state); |
3972 break; | |
3973 | |
3974 case NFA_SKWORD: /* \K */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3975 result = !VIM_ISDIGIT(curc) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3976 && vim_iswordp_buf(reginput, reg_buf); |
4444 | 3977 ADD_POS_NEG_STATE(t->state); |
3978 break; | |
3979 | |
3980 case NFA_FNAME: /* \f */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3981 result = vim_isfilec(curc); |
4444 | 3982 ADD_POS_NEG_STATE(t->state); |
3983 break; | |
3984 | |
3985 case NFA_SFNAME: /* \F */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3986 result = !VIM_ISDIGIT(curc) && vim_isfilec(curc); |
4444 | 3987 ADD_POS_NEG_STATE(t->state); |
3988 break; | |
3989 | |
3990 case NFA_PRINT: /* \p */ | |
4470 | 3991 result = ptr2cells(reginput) == 1; |
4444 | 3992 ADD_POS_NEG_STATE(t->state); |
3993 break; | |
3994 | |
3995 case NFA_SPRINT: /* \P */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3996 result = !VIM_ISDIGIT(curc) && ptr2cells(reginput) == 1; |
4444 | 3997 ADD_POS_NEG_STATE(t->state); |
3998 break; | |
3999 | |
4000 case NFA_WHITE: /* \s */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4001 result = vim_iswhite(curc); |
4444 | 4002 ADD_POS_NEG_STATE(t->state); |
4003 break; | |
4004 | |
4005 case NFA_NWHITE: /* \S */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4006 result = curc != NUL && !vim_iswhite(curc); |
4444 | 4007 ADD_POS_NEG_STATE(t->state); |
4008 break; | |
4009 | |
4010 case NFA_DIGIT: /* \d */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4011 result = ri_digit(curc); |
4444 | 4012 ADD_POS_NEG_STATE(t->state); |
4013 break; | |
4014 | |
4015 case NFA_NDIGIT: /* \D */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4016 result = curc != NUL && !ri_digit(curc); |
4444 | 4017 ADD_POS_NEG_STATE(t->state); |
4018 break; | |
4019 | |
4020 case NFA_HEX: /* \x */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4021 result = ri_hex(curc); |
4444 | 4022 ADD_POS_NEG_STATE(t->state); |
4023 break; | |
4024 | |
4025 case NFA_NHEX: /* \X */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4026 result = curc != NUL && !ri_hex(curc); |
4444 | 4027 ADD_POS_NEG_STATE(t->state); |
4028 break; | |
4029 | |
4030 case NFA_OCTAL: /* \o */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4031 result = ri_octal(curc); |
4444 | 4032 ADD_POS_NEG_STATE(t->state); |
4033 break; | |
4034 | |
4035 case NFA_NOCTAL: /* \O */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4036 result = curc != NUL && !ri_octal(curc); |
4444 | 4037 ADD_POS_NEG_STATE(t->state); |
4038 break; | |
4039 | |
4040 case NFA_WORD: /* \w */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4041 result = ri_word(curc); |
4444 | 4042 ADD_POS_NEG_STATE(t->state); |
4043 break; | |
4044 | |
4045 case NFA_NWORD: /* \W */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4046 result = curc != NUL && !ri_word(curc); |
4444 | 4047 ADD_POS_NEG_STATE(t->state); |
4048 break; | |
4049 | |
4050 case NFA_HEAD: /* \h */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4051 result = ri_head(curc); |
4444 | 4052 ADD_POS_NEG_STATE(t->state); |
4053 break; | |
4054 | |
4055 case NFA_NHEAD: /* \H */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4056 result = curc != NUL && !ri_head(curc); |
4444 | 4057 ADD_POS_NEG_STATE(t->state); |
4058 break; | |
4059 | |
4060 case NFA_ALPHA: /* \a */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4061 result = ri_alpha(curc); |
4444 | 4062 ADD_POS_NEG_STATE(t->state); |
4063 break; | |
4064 | |
4065 case NFA_NALPHA: /* \A */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4066 result = curc != NUL && !ri_alpha(curc); |
4444 | 4067 ADD_POS_NEG_STATE(t->state); |
4068 break; | |
4069 | |
4070 case NFA_LOWER: /* \l */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4071 result = ri_lower(curc); |
4444 | 4072 ADD_POS_NEG_STATE(t->state); |
4073 break; | |
4074 | |
4075 case NFA_NLOWER: /* \L */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4076 result = curc != NUL && !ri_lower(curc); |
4444 | 4077 ADD_POS_NEG_STATE(t->state); |
4078 break; | |
4079 | |
4080 case NFA_UPPER: /* \u */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4081 result = ri_upper(curc); |
4444 | 4082 ADD_POS_NEG_STATE(t->state); |
4083 break; | |
4084 | |
4085 case NFA_NUPPER: /* \U */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4086 result = curc != NUL && !ri_upper(curc); |
4444 | 4087 ADD_POS_NEG_STATE(t->state); |
4088 break; | |
4089 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4090 case NFA_BACKREF1: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4091 case NFA_BACKREF2: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4092 case NFA_BACKREF3: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4093 case NFA_BACKREF4: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4094 case NFA_BACKREF5: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4095 case NFA_BACKREF6: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4096 case NFA_BACKREF7: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4097 case NFA_BACKREF8: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4098 case NFA_BACKREF9: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4099 /* \1 .. \9 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4100 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4101 int subidx = t->state->c - NFA_BACKREF1 + 1; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4102 int bytelen; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4103 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4104 result = match_backref(&t->sub, subidx, &bytelen); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4105 if (result) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4106 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4107 if (bytelen == 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4108 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4109 /* empty match always works, add NFA_SKIP with zero to |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4110 * be used next */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4111 addstate_here(thislist, t->state->out, &t->sub, |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4112 &listidx); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4113 thislist->t[listidx + 1].count = 0; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4114 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4115 else if (bytelen <= clen) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4116 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4117 /* match current character, jump ahead to out of |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4118 * NFA_SKIP */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4119 addstate(nextlist, t->state->out->out, &t->sub, clen); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4120 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4121 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4122 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4123 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4124 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4125 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4126 /* skip ofer the matched characters, set character |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4127 * count in NFA_SKIP */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4128 addstate(nextlist, t->state->out, &t->sub, bytelen); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4129 nextlist->t[nextlist->n - 1].count = bytelen - clen; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4130 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4131 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4132 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4133 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4134 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4135 } |
4482 | 4136 break; |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4137 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4138 case NFA_SKIP: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4139 /* charater of previous matching \1 .. \9 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4140 if (t->count - clen <= 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4141 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4142 /* end of match, go to what follows */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4143 addstate(nextlist, t->state->out, &t->sub, clen); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4144 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4145 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4146 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4147 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4148 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4149 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4150 /* add state again with decremented count */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4151 addstate(nextlist, t->state, &t->sub, 0); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4152 nextlist->t[nextlist->n - 1].count = t->count - clen; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4153 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4154 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4155 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4156 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4157 break; |
4482 | 4158 |
4159 case NFA_SKIP_CHAR: | |
4160 case NFA_ZSTART: | |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
4161 case NFA_ZEND: |
4482 | 4162 /* TODO: should not happen? */ |
4163 break; | |
4164 | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4165 case NFA_LNUM: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4166 case NFA_LNUM_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4167 case NFA_LNUM_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4168 result = (REG_MULTI && |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4169 nfa_re_num_cmp(t->state->val, t->state->c - NFA_LNUM, |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4170 (long_u)(reglnum + reg_firstlnum))); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4171 if (result) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4172 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4173 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4174 |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4175 case NFA_COL: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4176 case NFA_COL_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4177 case NFA_COL_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4178 result = nfa_re_num_cmp(t->state->val, t->state->c - NFA_COL, |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4179 (long_u)(reginput - regline) + 1); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4180 if (result) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4181 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4182 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4183 |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4184 case NFA_VCOL: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4185 case NFA_VCOL_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4186 case NFA_VCOL_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4187 result = nfa_re_num_cmp(t->state->val, t->state->c - NFA_VCOL, |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4188 (long_u)win_linetabsize( |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4189 reg_win == NULL ? curwin : reg_win, |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4190 regline, (colnr_T)(reginput - regline)) + 1); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4191 if (result) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4192 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4193 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4194 |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4195 case NFA_CURSOR: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4196 result = (reg_win != NULL |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4197 && (reglnum + reg_firstlnum == reg_win->w_cursor.lnum) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4198 && ((colnr_T)(reginput - regline) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4199 == reg_win->w_cursor.col)); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4200 if (result) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4201 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4202 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4203 |
4444 | 4204 default: /* regular character */ |
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4205 { |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4206 int c = t->state->c; |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4207 |
4482 | 4208 /* TODO: put this in #ifdef later */ |
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4209 if (c < -256) |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4210 EMSGN("INTERNAL: Negative state char: %ld", c); |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4211 if (is_Magic(c)) |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4212 c = un_Magic(c); |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4213 result = (c == curc); |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4214 |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4215 if (!result && ireg_ic) |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4216 result = MB_TOLOWER(c) == MB_TOLOWER(curc); |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4217 #ifdef FEAT_MBYTE |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4218 /* If there is a composing character which is not being |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4219 * ignored there can be no match. Match with composing |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4220 * character uses NFA_COMPOSING above. */ |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4221 if (result && enc_utf8 && !ireg_icombine |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4222 && clen != utf_char2len(curc)) |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4223 result = FALSE; |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4224 #endif |
4444 | 4225 ADD_POS_NEG_STATE(t->state); |
4226 break; | |
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4227 } |
4444 | 4228 } |
4229 | |
4230 } /* for (thislist = thislist; thislist->state; thislist++) */ | |
4231 | |
4557
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
4232 /* Look for the start of a match in the current position by adding the |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
4233 * start state to the list of states. |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
4234 * The first found match is the leftmost one, thus the order of states |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
4235 * matters! |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
4236 * Do not add the start state in recursive calls of nfa_regmatch(), |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
4237 * because recursive calls should only start in the first position. |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4238 * Unless "endp" is not NULL, then we match the end position. |
4557
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
4239 * Also don't start a match past the first line. */ |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4240 if (nfa_match == FALSE |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4241 && ((start->c == NFA_MOPEN + 0 |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4242 && reglnum == 0 |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4243 && clen != 0 |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4244 && (ireg_maxcol == 0 |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4245 || (colnr_T)(reginput - regline) < ireg_maxcol)) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4246 || (endp != NULL |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4247 && (REG_MULTI |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4248 ? (reglnum < endp->se_u.pos.lnum |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4249 || (reglnum == endp->se_u.pos.lnum |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4250 && (int)(reginput - regline) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4251 < endp->se_u.pos.col)) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4252 : reginput < endp->se_u.ptr)))) |
4444 | 4253 { |
4254 #ifdef ENABLE_LOG | |
4255 fprintf(log_fd, "(---) STARTSTATE\n"); | |
4256 #endif | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4257 addstate(nextlist, start, m, clen); |
4444 | 4258 } |
4259 | |
4260 #ifdef ENABLE_LOG | |
4261 fprintf(log_fd, ">>> Thislist had %d states available: ", thislist->n); | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4262 { |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4263 int i; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4264 |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4265 for (i = 0; i < thislist->n; i++) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4266 fprintf(log_fd, "%d ", abs(thislist->t[i].state->id)); |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4267 } |
4444 | 4268 fprintf(log_fd, "\n"); |
4269 #endif | |
4270 | |
4271 nextchar: | |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4272 /* Advance to the next character, or advance to the next line, or |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4273 * finish. */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4274 if (clen != 0) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4275 reginput += clen; |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4276 else if (go_to_nextline || (endp != NULL && REG_MULTI |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4277 && reglnum < endp->se_u.pos.lnum)) |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4278 reg_nextline(); |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4279 else |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4280 break; |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4281 } |
4444 | 4282 |
4283 #ifdef ENABLE_LOG | |
4284 if (log_fd != stderr) | |
4285 fclose(log_fd); | |
4286 log_fd = NULL; | |
4287 #endif | |
4288 | |
4289 theend: | |
4290 /* Free memory */ | |
4291 vim_free(list[0].t); | |
4292 vim_free(list[1].t); | |
4293 vim_free(list[2].t); | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4294 vim_free(listids); |
4444 | 4295 #undef ADD_POS_NEG_STATE |
4460 | 4296 #ifdef NFA_REGEXP_DEBUG_LOG |
4444 | 4297 fclose(debug); |
4298 #endif | |
4299 | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4300 return nfa_match; |
4444 | 4301 } |
4302 | |
4303 /* | |
4304 * Try match of "prog" with at regline["col"]. | |
4305 * Returns 0 for failure, number of lines contained in the match otherwise. | |
4306 */ | |
4307 static long | |
4308 nfa_regtry(start, col) | |
4309 nfa_state_T *start; | |
4310 colnr_T col; | |
4311 { | |
4312 int i; | |
4313 regsub_T sub, m; | |
4314 #ifdef ENABLE_LOG | |
4315 FILE *f; | |
4316 #endif | |
4317 | |
4318 reginput = regline + col; | |
4319 need_clear_subexpr = TRUE; | |
4320 | |
4321 #ifdef ENABLE_LOG | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
4322 f = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 4323 if (f != NULL) |
4324 { | |
4325 fprintf(f, "\n\n\n\n\n\n\t\t=======================================================\n"); | |
4326 fprintf(f, " =======================================================\n"); | |
4327 #ifdef DEBUG | |
4328 fprintf(f, "\tRegexp is \"%s\"\n", nfa_regengine.expr); | |
4329 #endif | |
4330 fprintf(f, "\tInput text is \"%s\" \n", reginput); | |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
4331 fprintf(f, " =======================================================\n\n"); |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
4332 nfa_print_state(f, start); |
4444 | 4333 fprintf(f, "\n\n"); |
4334 fclose(f); | |
4335 } | |
4336 else | |
4337 EMSG(_("Could not open temporary log file for writing ")); | |
4338 #endif | |
4339 | |
4340 if (REG_MULTI) | |
4341 { | |
4342 /* Use 0xff to set lnum to -1 */ | |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4343 vim_memset(sub.list.multi, 0xff, sizeof(struct multipos) * nfa_nsubexpr); |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4344 vim_memset(m.list.multi, 0xff, sizeof(struct multipos) * nfa_nsubexpr); |
4444 | 4345 } |
4346 else | |
4347 { | |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4348 vim_memset(sub.list.line, 0, sizeof(struct linepos) * nfa_nsubexpr); |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4349 vim_memset(m.list.line, 0, sizeof(struct linepos) * nfa_nsubexpr); |
4444 | 4350 } |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4351 sub.in_use = 0; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4352 m.in_use = 0; |
4444 | 4353 |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
4354 if (nfa_regmatch(start, &sub, &m, NULL) == FALSE) |
4444 | 4355 return 0; |
4356 | |
4357 cleanup_subexpr(); | |
4358 if (REG_MULTI) | |
4359 { | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4360 for (i = 0; i < sub.in_use; i++) |
4444 | 4361 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4362 reg_startpos[i] = sub.list.multi[i].start; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4363 reg_endpos[i] = sub.list.multi[i].end; |
4444 | 4364 } |
4365 | |
4366 if (reg_startpos[0].lnum < 0) | |
4367 { | |
4368 reg_startpos[0].lnum = 0; | |
4369 reg_startpos[0].col = col; | |
4370 } | |
4371 if (reg_endpos[0].lnum < 0) | |
4372 { | |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
4373 /* pattern has a \ze but it didn't match, use current end */ |
4444 | 4374 reg_endpos[0].lnum = reglnum; |
4375 reg_endpos[0].col = (int)(reginput - regline); | |
4376 } | |
4377 else | |
4378 /* Use line number of "\ze". */ | |
4379 reglnum = reg_endpos[0].lnum; | |
4380 } | |
4381 else | |
4382 { | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4383 for (i = 0; i < sub.in_use; i++) |
4444 | 4384 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4385 reg_startp[i] = sub.list.line[i].start; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4386 reg_endp[i] = sub.list.line[i].end; |
4444 | 4387 } |
4388 | |
4389 if (reg_startp[0] == NULL) | |
4390 reg_startp[0] = regline + col; | |
4391 if (reg_endp[0] == NULL) | |
4392 reg_endp[0] = reginput; | |
4393 } | |
4394 | |
4395 return 1 + reglnum; | |
4396 } | |
4397 | |
4398 /* | |
4399 * Match a regexp against a string ("line" points to the string) or multiple | |
4400 * lines ("line" is NULL, use reg_getline()). | |
4401 * | |
4402 * Returns 0 for failure, number of lines contained in the match otherwise. | |
4403 */ | |
4404 static long | |
4405 nfa_regexec_both(line, col) | |
4406 char_u *line; | |
4407 colnr_T col; /* column to start looking for match */ | |
4408 { | |
4409 nfa_regprog_T *prog; | |
4410 long retval = 0L; | |
4411 int i; | |
4412 | |
4413 if (REG_MULTI) | |
4414 { | |
4415 prog = (nfa_regprog_T *)reg_mmatch->regprog; | |
4416 line = reg_getline((linenr_T)0); /* relative to the cursor */ | |
4417 reg_startpos = reg_mmatch->startpos; | |
4418 reg_endpos = reg_mmatch->endpos; | |
4419 } | |
4420 else | |
4421 { | |
4422 prog = (nfa_regprog_T *)reg_match->regprog; | |
4423 reg_startp = reg_match->startp; | |
4424 reg_endp = reg_match->endp; | |
4425 } | |
4426 | |
4427 /* Be paranoid... */ | |
4428 if (prog == NULL || line == NULL) | |
4429 { | |
4430 EMSG(_(e_null)); | |
4431 goto theend; | |
4432 } | |
4433 | |
4434 /* If the start column is past the maximum column: no need to try. */ | |
4435 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
4436 goto theend; | |
4437 | |
4438 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */ | |
4439 if (prog->regflags & RF_ICASE) | |
4440 ireg_ic = TRUE; | |
4441 else if (prog->regflags & RF_NOICASE) | |
4442 ireg_ic = FALSE; | |
4443 | |
4444 #ifdef FEAT_MBYTE | |
4445 /* If pattern contains "\Z" overrule value of ireg_icombine */ | |
4446 if (prog->regflags & RF_ICOMBINE) | |
4447 ireg_icombine = TRUE; | |
4448 #endif | |
4449 | |
4450 regline = line; | |
4451 reglnum = 0; /* relative to line */ | |
4452 | |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4453 nfa_has_zend = prog->has_zend; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
4454 nfa_has_backref = prog->has_backref; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4455 nfa_nsubexpr = prog->nsubexp; |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4456 |
4444 | 4457 nstate = prog->nstate; |
4458 for (i = 0; i < nstate; ++i) | |
4459 { | |
4460 prog->state[i].id = i; | |
4461 prog->state[i].lastlist = 0; | |
4462 } | |
4463 | |
4464 retval = nfa_regtry(prog->start, col); | |
4465 | |
4466 theend: | |
4467 return retval; | |
4468 } | |
4469 | |
4470 /* | |
4471 * Compile a regular expression into internal code for the NFA matcher. | |
4472 * Returns the program in allocated space. Returns NULL for an error. | |
4473 */ | |
4474 static regprog_T * | |
4475 nfa_regcomp(expr, re_flags) | |
4476 char_u *expr; | |
4477 int re_flags; | |
4478 { | |
4541
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4479 nfa_regprog_T *prog = NULL; |
4458 | 4480 size_t prog_size; |
4444 | 4481 int *postfix; |
4482 | |
4483 if (expr == NULL) | |
4484 return NULL; | |
4485 | |
4486 #ifdef DEBUG | |
4487 nfa_regengine.expr = expr; | |
4488 #endif | |
4489 | |
4490 init_class_tab(); | |
4491 | |
4492 if (nfa_regcomp_start(expr, re_flags) == FAIL) | |
4493 return NULL; | |
4494 | |
4495 /* Build postfix form of the regexp. Needed to build the NFA | |
4541
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4496 * (and count its size). */ |
4444 | 4497 postfix = re2post(); |
4498 if (postfix == NULL) | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4499 { |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4500 /* TODO: only give this error for debugging? */ |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4501 if (post_ptr >= post_end) |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4502 EMSGN("Internal error: estimated max number of states insufficient: %ld", post_end - post_start); |
4444 | 4503 goto fail; /* Cascaded (syntax?) error */ |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4504 } |
4444 | 4505 |
4506 /* | |
4507 * In order to build the NFA, we parse the input regexp twice: | |
4508 * 1. first pass to count size (so we can allocate space) | |
4509 * 2. second to emit code | |
4510 */ | |
4511 #ifdef ENABLE_LOG | |
4512 { | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
4513 FILE *f = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 4514 |
4515 if (f != NULL) | |
4516 { | |
4517 fprintf(f, "\n*****************************\n\n\n\n\tCompiling regexp \"%s\" ... hold on !\n", expr); | |
4518 fclose(f); | |
4519 } | |
4520 } | |
4521 #endif | |
4522 | |
4523 /* | |
4524 * PASS 1 | |
4525 * Count number of NFA states in "nstate". Do not build the NFA. | |
4526 */ | |
4527 post2nfa(postfix, post_ptr, TRUE); | |
4541
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4528 |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4529 /* Space for compiled regexp */ |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4530 prog_size = sizeof(nfa_regprog_T) + sizeof(nfa_state_T) * nstate; |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4531 prog = (nfa_regprog_T *)lalloc(prog_size, TRUE); |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4532 if (prog == NULL) |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4533 goto fail; |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4534 vim_memset(prog, 0, prog_size); |
4444 | 4535 state_ptr = prog->state; |
4536 | |
4537 /* | |
4538 * PASS 2 | |
4539 * Build the NFA | |
4540 */ | |
4541 prog->start = post2nfa(postfix, post_ptr, FALSE); | |
4542 if (prog->start == NULL) | |
4543 goto fail; | |
4544 | |
4545 prog->regflags = regflags; | |
4546 prog->engine = &nfa_regengine; | |
4547 prog->nstate = nstate; | |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4548 prog->has_zend = nfa_has_zend; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
4549 prog->has_backref = nfa_has_backref; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4550 prog->nsubexp = regnpar; |
4444 | 4551 #ifdef ENABLE_LOG |
4552 nfa_postfix_dump(expr, OK); | |
4553 nfa_dump(prog); | |
4554 #endif | |
4555 | |
4556 out: | |
4557 vim_free(post_start); | |
4558 post_start = post_ptr = post_end = NULL; | |
4559 state_ptr = NULL; | |
4560 return (regprog_T *)prog; | |
4561 | |
4562 fail: | |
4563 vim_free(prog); | |
4564 prog = NULL; | |
4565 #ifdef ENABLE_LOG | |
4566 nfa_postfix_dump(expr, FAIL); | |
4567 #endif | |
4568 #ifdef DEBUG | |
4569 nfa_regengine.expr = NULL; | |
4570 #endif | |
4571 goto out; | |
4572 } | |
4573 | |
4574 | |
4575 /* | |
4576 * Match a regexp against a string. | |
4577 * "rmp->regprog" is a compiled regexp as returned by nfa_regcomp(). | |
4578 * Uses curbuf for line count and 'iskeyword'. | |
4579 * | |
4580 * Return TRUE if there is a match, FALSE if not. | |
4581 */ | |
4582 static int | |
4583 nfa_regexec(rmp, line, col) | |
4584 regmatch_T *rmp; | |
4585 char_u *line; /* string to match against */ | |
4586 colnr_T col; /* column to start looking for match */ | |
4587 { | |
4588 reg_match = rmp; | |
4589 reg_mmatch = NULL; | |
4590 reg_maxline = 0; | |
4591 reg_line_lbr = FALSE; | |
4592 reg_buf = curbuf; | |
4593 reg_win = NULL; | |
4594 ireg_ic = rmp->rm_ic; | |
4595 #ifdef FEAT_MBYTE | |
4596 ireg_icombine = FALSE; | |
4597 #endif | |
4598 ireg_maxcol = 0; | |
4599 return (nfa_regexec_both(line, col) != 0); | |
4600 } | |
4601 | |
4602 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
4603 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
4604 | |
4605 static int nfa_regexec_nl __ARGS((regmatch_T *rmp, char_u *line, colnr_T col)); | |
4606 | |
4607 /* | |
4608 * Like nfa_regexec(), but consider a "\n" in "line" to be a line break. | |
4609 */ | |
4610 static int | |
4611 nfa_regexec_nl(rmp, line, col) | |
4612 regmatch_T *rmp; | |
4613 char_u *line; /* string to match against */ | |
4614 colnr_T col; /* column to start looking for match */ | |
4615 { | |
4616 reg_match = rmp; | |
4617 reg_mmatch = NULL; | |
4618 reg_maxline = 0; | |
4619 reg_line_lbr = TRUE; | |
4620 reg_buf = curbuf; | |
4621 reg_win = NULL; | |
4622 ireg_ic = rmp->rm_ic; | |
4623 #ifdef FEAT_MBYTE | |
4624 ireg_icombine = FALSE; | |
4625 #endif | |
4626 ireg_maxcol = 0; | |
4627 return (nfa_regexec_both(line, col) != 0); | |
4628 } | |
4629 #endif | |
4630 | |
4631 | |
4632 /* | |
4633 * Match a regexp against multiple lines. | |
4634 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
4635 * Uses curbuf for line count and 'iskeyword'. | |
4636 * | |
4637 * Return zero if there is no match. Return number of lines contained in the | |
4638 * match otherwise. | |
4639 * | |
4640 * Note: the body is the same as bt_regexec() except for nfa_regexec_both() | |
4641 * | |
4642 * ! Also NOTE : match may actually be in another line. e.g.: | |
4643 * when r.e. is \nc, cursor is at 'a' and the text buffer looks like | |
4644 * | |
4645 * +-------------------------+ | |
4646 * |a | | |
4647 * |b | | |
4648 * |c | | |
4649 * | | | |
4650 * +-------------------------+ | |
4651 * | |
4652 * then nfa_regexec_multi() returns 3. while the original | |
4653 * vim_regexec_multi() returns 0 and a second call at line 2 will return 2. | |
4654 * | |
4655 * FIXME if this behavior is not compatible. | |
4656 */ | |
4657 static long | |
4658 nfa_regexec_multi(rmp, win, buf, lnum, col, tm) | |
4659 regmmatch_T *rmp; | |
4660 win_T *win; /* window in which to search or NULL */ | |
4661 buf_T *buf; /* buffer in which to search */ | |
4662 linenr_T lnum; /* nr of line to start looking for match */ | |
4663 colnr_T col; /* column to start looking for match */ | |
4664 proftime_T *tm UNUSED; /* timeout limit or NULL */ | |
4665 { | |
4666 reg_match = NULL; | |
4667 reg_mmatch = rmp; | |
4668 reg_buf = buf; | |
4669 reg_win = win; | |
4670 reg_firstlnum = lnum; | |
4671 reg_maxline = reg_buf->b_ml.ml_line_count - lnum; | |
4672 reg_line_lbr = FALSE; | |
4673 ireg_ic = rmp->rmm_ic; | |
4674 #ifdef FEAT_MBYTE | |
4675 ireg_icombine = FALSE; | |
4676 #endif | |
4677 ireg_maxcol = rmp->rmm_maxcol; | |
4678 | |
4503 | 4679 return nfa_regexec_both(NULL, col); |
4444 | 4680 } |
4681 | |
4682 #ifdef DEBUG | |
4683 # undef ENABLE_LOG | |
4684 #endif |