Mercurial > vim
annotate src/regexp_nfa.c @ 4647:857f6c53f117 v7.3.1071
updated for version 7.3.1071
Problem: New regexp engine: backreferences don't work correctly.
Solution: Add every possible start/end position on the state stack.
author | Bram Moolenaar <bram@vim.org> |
---|---|
date | Thu, 30 May 2013 17:05:39 +0200 |
parents | 5679b8ddd8cc |
children | 9f502d8d5165 |
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 /* Upper limit allowed for {m,n} repetitions handled by NFA */ | |
33 #define NFA_BRACES_MAXLIMIT 10 | |
34 /* For allocating space for the postfix representation */ | |
35 #define NFA_POSTFIX_MULTIPLIER (NFA_BRACES_MAXLIMIT + 2)*2 | |
36 | |
37 enum | |
38 { | |
39 NFA_SPLIT = -1024, | |
40 NFA_MATCH, | |
41 NFA_SKIP_CHAR, /* matches a 0-length char */ | |
42 NFA_END_NEG_RANGE, /* Used when expanding [^ab] */ | |
43 | |
44 NFA_CONCAT, | |
45 NFA_OR, | |
46 NFA_STAR, | |
47 NFA_PLUS, | |
48 NFA_QUEST, | |
49 NFA_QUEST_NONGREEDY, /* Non-greedy version of \? */ | |
50 NFA_NOT, /* used for [^ab] negated char ranges */ | |
51 | |
52 NFA_BOL, /* ^ Begin line */ | |
53 NFA_EOL, /* $ End line */ | |
54 NFA_BOW, /* \< Begin word */ | |
55 NFA_EOW, /* \> End word */ | |
56 NFA_BOF, /* \%^ Begin file */ | |
57 NFA_EOF, /* \%$ End file */ | |
58 NFA_NEWL, | |
59 NFA_ZSTART, /* Used for \zs */ | |
60 NFA_ZEND, /* Used for \ze */ | |
61 NFA_NOPEN, /* Start of subexpression marked with \%( */ | |
62 NFA_NCLOSE, /* End of subexpr. marked with \%( ... \) */ | |
63 NFA_START_INVISIBLE, | |
64 NFA_END_INVISIBLE, | |
65 NFA_COMPOSING, /* Next nodes in NFA are part of the | |
66 composing multibyte char */ | |
67 NFA_END_COMPOSING, /* End of a composing char in the NFA */ | |
68 | |
69 /* The following are used only in the postfix form, not in the NFA */ | |
70 NFA_PREV_ATOM_NO_WIDTH, /* Used for \@= */ | |
71 NFA_PREV_ATOM_NO_WIDTH_NEG, /* Used for \@! */ | |
72 NFA_PREV_ATOM_JUST_BEFORE, /* Used for \@<= */ | |
73 NFA_PREV_ATOM_JUST_BEFORE_NEG, /* Used for \@<! */ | |
74 NFA_PREV_ATOM_LIKE_PATTERN, /* Used for \@> */ | |
75 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
76 NFA_BACKREF1, /* \1 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
77 NFA_BACKREF2, /* \2 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
78 NFA_BACKREF3, /* \3 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
79 NFA_BACKREF4, /* \4 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
80 NFA_BACKREF5, /* \5 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
81 NFA_BACKREF6, /* \6 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
82 NFA_BACKREF7, /* \7 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
83 NFA_BACKREF8, /* \8 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
84 NFA_BACKREF9, /* \9 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
85 NFA_SKIP, /* Skip characters */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
86 |
4444 | 87 NFA_MOPEN, |
88 NFA_MCLOSE = NFA_MOPEN + NSUBEXP, | |
89 | |
90 /* NFA_FIRST_NL */ | |
91 NFA_ANY = NFA_MCLOSE + NSUBEXP, /* Match any one character. */ | |
92 NFA_ANYOF, /* Match any character in this string. */ | |
93 NFA_ANYBUT, /* Match any character not in this string. */ | |
94 NFA_IDENT, /* Match identifier char */ | |
95 NFA_SIDENT, /* Match identifier char but no digit */ | |
96 NFA_KWORD, /* Match keyword char */ | |
97 NFA_SKWORD, /* Match word char but no digit */ | |
98 NFA_FNAME, /* Match file name char */ | |
99 NFA_SFNAME, /* Match file name char but no digit */ | |
100 NFA_PRINT, /* Match printable char */ | |
101 NFA_SPRINT, /* Match printable char but no digit */ | |
102 NFA_WHITE, /* Match whitespace char */ | |
103 NFA_NWHITE, /* Match non-whitespace char */ | |
104 NFA_DIGIT, /* Match digit char */ | |
105 NFA_NDIGIT, /* Match non-digit char */ | |
106 NFA_HEX, /* Match hex char */ | |
107 NFA_NHEX, /* Match non-hex char */ | |
108 NFA_OCTAL, /* Match octal char */ | |
109 NFA_NOCTAL, /* Match non-octal char */ | |
110 NFA_WORD, /* Match word char */ | |
111 NFA_NWORD, /* Match non-word char */ | |
112 NFA_HEAD, /* Match head char */ | |
113 NFA_NHEAD, /* Match non-head char */ | |
114 NFA_ALPHA, /* Match alpha char */ | |
115 NFA_NALPHA, /* Match non-alpha char */ | |
116 NFA_LOWER, /* Match lowercase char */ | |
117 NFA_NLOWER, /* Match non-lowercase char */ | |
118 NFA_UPPER, /* Match uppercase char */ | |
119 NFA_NUPPER, /* Match non-uppercase char */ | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
120 |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
121 NFA_CURSOR, /* Match cursor pos */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
122 NFA_LNUM, /* Match line number */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
123 NFA_LNUM_GT, /* Match > line number */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
124 NFA_LNUM_LT, /* Match < line number */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
125 NFA_COL, /* Match cursor column */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
126 NFA_COL_GT, /* Match > cursor column */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
127 NFA_COL_LT, /* Match < cursor column */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
128 NFA_VCOL, /* Match cursor virtual column */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
129 NFA_VCOL_GT, /* Match > cursor virtual column */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
130 NFA_VCOL_LT, /* Match < cursor virtual column */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
131 |
4444 | 132 NFA_FIRST_NL = NFA_ANY + ADD_NL, |
133 NFA_LAST_NL = NFA_NUPPER + ADD_NL, | |
134 | |
135 /* Character classes [:alnum:] etc */ | |
136 NFA_CLASS_ALNUM, | |
137 NFA_CLASS_ALPHA, | |
138 NFA_CLASS_BLANK, | |
139 NFA_CLASS_CNTRL, | |
140 NFA_CLASS_DIGIT, | |
141 NFA_CLASS_GRAPH, | |
142 NFA_CLASS_LOWER, | |
143 NFA_CLASS_PRINT, | |
144 NFA_CLASS_PUNCT, | |
145 NFA_CLASS_SPACE, | |
146 NFA_CLASS_UPPER, | |
147 NFA_CLASS_XDIGIT, | |
148 NFA_CLASS_TAB, | |
149 NFA_CLASS_RETURN, | |
150 NFA_CLASS_BACKSPACE, | |
151 NFA_CLASS_ESCAPE | |
152 }; | |
153 | |
154 /* Keep in sync with classchars. */ | |
155 static int nfa_classcodes[] = { | |
156 NFA_ANY, NFA_IDENT, NFA_SIDENT, NFA_KWORD,NFA_SKWORD, | |
157 NFA_FNAME, NFA_SFNAME, NFA_PRINT, NFA_SPRINT, | |
158 NFA_WHITE, NFA_NWHITE, NFA_DIGIT, NFA_NDIGIT, | |
159 NFA_HEX, NFA_NHEX, NFA_OCTAL, NFA_NOCTAL, | |
160 NFA_WORD, NFA_NWORD, NFA_HEAD, NFA_NHEAD, | |
161 NFA_ALPHA, NFA_NALPHA, NFA_LOWER, NFA_NLOWER, | |
162 NFA_UPPER, NFA_NUPPER | |
163 }; | |
164 | |
165 static char_u e_misplaced[] = N_("E866: (NFA regexp) Misplaced %c"); | |
166 | |
167 /* | |
168 * NFA errors can be of 3 types: | |
169 * *** NFA runtime errors, when something unknown goes wrong. The NFA fails | |
170 * silently and revert the to backtracking engine. | |
171 * syntax_error = FALSE; | |
172 * *** Regexp syntax errors, when the input regexp is not syntactically correct. | |
173 * The NFA engine displays an error message, and nothing else happens. | |
174 * syntax_error = TRUE | |
175 * *** Unsupported features, when the input regexp uses an operator that is not | |
176 * implemented in the NFA. The NFA engine fails silently, and reverts to the | |
177 * old backtracking engine. | |
178 * syntax_error = FALSE | |
179 * "The NFA fails" means that "compiling the regexp with the NFA fails": | |
180 * nfa_regcomp() returns FAIL. | |
181 */ | |
182 static int syntax_error = FALSE; | |
183 | |
184 /* NFA regexp \ze operator encountered. */ | |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
185 static int nfa_has_zend; |
4444 | 186 |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
187 /* NFA regexp \1 .. \9 encountered. */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
188 static int nfa_has_backref; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
189 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
190 /* 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
|
191 * the whole match (subexpr 0) is used. */ |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
192 static int nfa_nsubexpr; |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
193 |
4444 | 194 static int *post_start; /* holds the postfix form of r.e. */ |
195 static int *post_end; | |
196 static int *post_ptr; | |
197 | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
198 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
|
199 * executing. */ |
4444 | 200 static int istate; /* Index in the state vector, used in new_state() */ |
201 | |
202 | |
203 static int nfa_regcomp_start __ARGS((char_u*expr, int re_flags)); | |
204 static int nfa_recognize_char_class __ARGS((char_u *start, char_u *end, int extra_newl)); | |
205 static int nfa_emit_equi_class __ARGS((int c, int neg)); | |
206 static int nfa_regatom __ARGS((void)); | |
207 static int nfa_regpiece __ARGS((void)); | |
208 static int nfa_regconcat __ARGS((void)); | |
209 static int nfa_regbranch __ARGS((void)); | |
210 static int nfa_reg __ARGS((int paren)); | |
211 #ifdef DEBUG | |
212 static void nfa_set_code __ARGS((int c)); | |
213 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
|
214 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
|
215 static void nfa_print_state2 __ARGS((FILE *debugf, nfa_state_T *state, garray_T *indent)); |
4444 | 216 static void nfa_dump __ARGS((nfa_regprog_T *prog)); |
217 #endif | |
218 static int *re2post __ARGS((void)); | |
219 static nfa_state_T *new_state __ARGS((int c, nfa_state_T *out, nfa_state_T *out1)); | |
220 static nfa_state_T *post2nfa __ARGS((int *postfix, int *end, int nfa_calc_size)); | |
221 static int check_char_class __ARGS((int class, int c)); | |
222 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
|
223 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
|
224 static void nfa_set_null_listids __ARGS((nfa_state_T *start)); |
4444 | 225 static void nfa_save_listids __ARGS((nfa_state_T *start, int *list)); |
226 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
|
227 static int nfa_re_num_cmp __ARGS((long_u val, int op, long_u pos)); |
4444 | 228 static long nfa_regtry __ARGS((nfa_state_T *start, colnr_T col)); |
229 static long nfa_regexec_both __ARGS((char_u *line, colnr_T col)); | |
230 static regprog_T *nfa_regcomp __ARGS((char_u *expr, int re_flags)); | |
231 static int nfa_regexec __ARGS((regmatch_T *rmp, char_u *line, colnr_T col)); | |
232 static long nfa_regexec_multi __ARGS((regmmatch_T *rmp, win_T *win, buf_T *buf, linenr_T lnum, colnr_T col, proftime_T *tm)); | |
233 | |
234 /* helper functions used when doing re2post() ... regatom() parsing */ | |
235 #define EMIT(c) do { \ | |
236 if (post_ptr >= post_end) \ | |
237 return FAIL; \ | |
238 *post_ptr++ = c; \ | |
239 } while (0) | |
240 | |
241 /* | |
242 * Initialize internal variables before NFA compilation. | |
243 * Return OK on success, FAIL otherwise. | |
244 */ | |
245 static int | |
246 nfa_regcomp_start(expr, re_flags) | |
247 char_u *expr; | |
248 int re_flags; /* see vim_regcomp() */ | |
249 { | |
4458 | 250 size_t postfix_size; |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
251 int nstate_max; |
4444 | 252 |
253 nstate = 0; | |
254 istate = 0; | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
255 /* A reasonable estimation for maximum size */ |
4458 | 256 nstate_max = (int)(STRLEN(expr) + 1) * NFA_POSTFIX_MULTIPLIER; |
4444 | 257 |
4454 | 258 /* Some items blow up in size, such as [A-z]. Add more space for that. |
259 * TODO: some patterns may still fail. */ | |
4458 | 260 nstate_max += 1000; |
4454 | 261 |
262 /* Size for postfix representation of expr. */ | |
4444 | 263 postfix_size = sizeof(*post_start) * nstate_max; |
4454 | 264 |
4444 | 265 post_start = (int *)lalloc(postfix_size, TRUE); |
266 if (post_start == NULL) | |
267 return FAIL; | |
268 vim_memset(post_start, 0, postfix_size); | |
269 post_ptr = post_start; | |
4454 | 270 post_end = post_start + nstate_max; |
4444 | 271 nfa_has_zend = FALSE; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
272 nfa_has_backref = FALSE; |
4444 | 273 |
274 regcomp_start(expr, re_flags); | |
275 | |
276 return OK; | |
277 } | |
278 | |
279 /* | |
280 * Search between "start" and "end" and try to recognize a | |
281 * character class in expanded form. For example [0-9]. | |
282 * On success, return the id the character class to be emitted. | |
283 * On failure, return 0 (=FAIL) | |
284 * Start points to the first char of the range, while end should point | |
285 * to the closing brace. | |
286 */ | |
287 static int | |
288 nfa_recognize_char_class(start, end, extra_newl) | |
289 char_u *start; | |
290 char_u *end; | |
291 int extra_newl; | |
292 { | |
293 int i; | |
294 /* Each of these variables takes up a char in "config[]", | |
295 * in the order they are here. */ | |
296 int not = FALSE, af = FALSE, AF = FALSE, az = FALSE, AZ = FALSE, | |
297 o7 = FALSE, o9 = FALSE, underscore = FALSE, newl = FALSE; | |
298 char_u *p; | |
299 #define NCONFIGS 16 | |
300 int classid[NCONFIGS] = { | |
301 NFA_DIGIT, NFA_NDIGIT, NFA_HEX, NFA_NHEX, | |
302 NFA_OCTAL, NFA_NOCTAL, NFA_WORD, NFA_NWORD, | |
303 NFA_HEAD, NFA_NHEAD, NFA_ALPHA, NFA_NALPHA, | |
304 NFA_LOWER, NFA_NLOWER, NFA_UPPER, NFA_NUPPER | |
305 }; | |
4450 | 306 char_u myconfig[10]; |
4444 | 307 char_u config[NCONFIGS][9] = { |
308 "000000100", /* digit */ | |
309 "100000100", /* non digit */ | |
310 "011000100", /* hex-digit */ | |
311 "111000100", /* non hex-digit */ | |
312 "000001000", /* octal-digit */ | |
313 "100001000", /* [^0-7] */ | |
314 "000110110", /* [0-9A-Za-z_] */ | |
315 "100110110", /* [^0-9A-Za-z_] */ | |
316 "000110010", /* head of word */ | |
317 "100110010", /* not head of word */ | |
318 "000110000", /* alphabetic char a-z */ | |
319 "100110000", /* non alphabetic char */ | |
320 "000100000", /* lowercase letter */ | |
321 "100100000", /* non lowercase */ | |
322 "000010000", /* uppercase */ | |
323 "100010000" /* non uppercase */ | |
324 }; | |
325 | |
326 if (extra_newl == TRUE) | |
327 newl = TRUE; | |
328 | |
329 if (*end != ']') | |
330 return FAIL; | |
331 p = start; | |
332 if (*p == '^') | |
333 { | |
334 not = TRUE; | |
335 p ++; | |
336 } | |
337 | |
338 while (p < end) | |
339 { | |
340 if (p + 2 < end && *(p + 1) == '-') | |
341 { | |
342 switch (*p) | |
343 { | |
344 case '0': | |
345 if (*(p + 2) == '9') | |
346 { | |
347 o9 = TRUE; | |
348 break; | |
349 } | |
350 else | |
351 if (*(p + 2) == '7') | |
352 { | |
353 o7 = TRUE; | |
354 break; | |
355 } | |
356 case 'a': | |
357 if (*(p + 2) == 'z') | |
358 { | |
359 az = TRUE; | |
360 break; | |
361 } | |
362 else | |
363 if (*(p + 2) == 'f') | |
364 { | |
365 af = TRUE; | |
366 break; | |
367 } | |
368 case 'A': | |
369 if (*(p + 2) == 'Z') | |
370 { | |
371 AZ = TRUE; | |
372 break; | |
373 } | |
374 else | |
375 if (*(p + 2) == 'F') | |
376 { | |
377 AF = TRUE; | |
378 break; | |
379 } | |
380 /* FALLTHROUGH */ | |
381 default: | |
382 return FAIL; | |
383 } | |
384 p += 3; | |
385 } | |
386 else if (p + 1 < end && *p == '\\' && *(p + 1) == 'n') | |
387 { | |
388 newl = TRUE; | |
389 p += 2; | |
390 } | |
391 else if (*p == '_') | |
392 { | |
393 underscore = TRUE; | |
394 p ++; | |
395 } | |
396 else if (*p == '\n') | |
397 { | |
398 newl = TRUE; | |
399 p ++; | |
400 } | |
401 else | |
402 return FAIL; | |
403 } /* while (p < end) */ | |
404 | |
405 if (p != end) | |
406 return FAIL; | |
407 | |
408 /* build the config that represents the ranges we gathered */ | |
409 STRCPY(myconfig, "000000000"); | |
410 if (not == TRUE) | |
411 myconfig[0] = '1'; | |
412 if (af == TRUE) | |
413 myconfig[1] = '1'; | |
414 if (AF == TRUE) | |
415 myconfig[2] = '1'; | |
416 if (az == TRUE) | |
417 myconfig[3] = '1'; | |
418 if (AZ == TRUE) | |
419 myconfig[4] = '1'; | |
420 if (o7 == TRUE) | |
421 myconfig[5] = '1'; | |
422 if (o9 == TRUE) | |
423 myconfig[6] = '1'; | |
424 if (underscore == TRUE) | |
425 myconfig[7] = '1'; | |
426 if (newl == TRUE) | |
427 { | |
428 myconfig[8] = '1'; | |
429 extra_newl = ADD_NL; | |
430 } | |
431 /* try to recognize character classes */ | |
432 for (i = 0; i < NCONFIGS; i++) | |
4450 | 433 if (STRNCMP(myconfig, config[i], 8) == 0) |
4444 | 434 return classid[i] + extra_newl; |
435 | |
436 /* fallthrough => no success so far */ | |
437 return FAIL; | |
438 | |
439 #undef NCONFIGS | |
440 } | |
441 | |
442 /* | |
443 * Produce the bytes for equivalence class "c". | |
444 * Currently only handles latin1, latin9 and utf-8. | |
445 * Emits bytes in postfix notation: 'a,b,NFA_OR,c,NFA_OR' is | |
446 * equivalent to 'a OR b OR c' | |
447 * | |
448 * NOTE! When changing this function, also update reg_equi_class() | |
449 */ | |
450 static int | |
451 nfa_emit_equi_class(c, neg) | |
452 int c; | |
453 int neg; | |
454 { | |
455 int first = TRUE; | |
456 int glue = neg == TRUE ? NFA_CONCAT : NFA_OR; | |
457 #define EMIT2(c) \ | |
458 EMIT(c); \ | |
459 if (neg == TRUE) { \ | |
460 EMIT(NFA_NOT); \ | |
461 } \ | |
462 if (first == FALSE) \ | |
463 EMIT(glue); \ | |
464 else \ | |
465 first = FALSE; \ | |
466 | |
467 #ifdef FEAT_MBYTE | |
468 if (enc_utf8 || STRCMP(p_enc, "latin1") == 0 | |
469 || STRCMP(p_enc, "iso-8859-15") == 0) | |
470 #endif | |
471 { | |
472 switch (c) | |
473 { | |
474 case 'A': case '\300': case '\301': case '\302': | |
475 case '\303': case '\304': case '\305': | |
476 EMIT2('A'); EMIT2('\300'); EMIT2('\301'); | |
477 EMIT2('\302'); EMIT2('\303'); EMIT2('\304'); | |
478 EMIT2('\305'); | |
479 return OK; | |
480 | |
481 case 'C': case '\307': | |
482 EMIT2('C'); EMIT2('\307'); | |
483 return OK; | |
484 | |
485 case 'E': case '\310': case '\311': case '\312': case '\313': | |
486 EMIT2('E'); EMIT2('\310'); EMIT2('\311'); | |
487 EMIT2('\312'); EMIT2('\313'); | |
488 return OK; | |
489 | |
490 case 'I': case '\314': case '\315': case '\316': case '\317': | |
491 EMIT2('I'); EMIT2('\314'); EMIT2('\315'); | |
492 EMIT2('\316'); EMIT2('\317'); | |
493 return OK; | |
494 | |
495 case 'N': case '\321': | |
496 EMIT2('N'); EMIT2('\321'); | |
497 return OK; | |
498 | |
499 case 'O': case '\322': case '\323': case '\324': case '\325': | |
500 case '\326': | |
501 EMIT2('O'); EMIT2('\322'); EMIT2('\323'); | |
502 EMIT2('\324'); EMIT2('\325'); EMIT2('\326'); | |
503 return OK; | |
504 | |
505 case 'U': case '\331': case '\332': case '\333': case '\334': | |
506 EMIT2('U'); EMIT2('\331'); EMIT2('\332'); | |
507 EMIT2('\333'); EMIT2('\334'); | |
508 return OK; | |
509 | |
510 case 'Y': case '\335': | |
511 EMIT2('Y'); EMIT2('\335'); | |
512 return OK; | |
513 | |
514 case 'a': case '\340': case '\341': case '\342': | |
515 case '\343': case '\344': case '\345': | |
516 EMIT2('a'); EMIT2('\340'); EMIT2('\341'); | |
517 EMIT2('\342'); EMIT2('\343'); EMIT2('\344'); | |
518 EMIT2('\345'); | |
519 return OK; | |
520 | |
521 case 'c': case '\347': | |
522 EMIT2('c'); EMIT2('\347'); | |
523 return OK; | |
524 | |
525 case 'e': case '\350': case '\351': case '\352': case '\353': | |
526 EMIT2('e'); EMIT2('\350'); EMIT2('\351'); | |
527 EMIT2('\352'); EMIT2('\353'); | |
528 return OK; | |
529 | |
530 case 'i': case '\354': case '\355': case '\356': case '\357': | |
531 EMIT2('i'); EMIT2('\354'); EMIT2('\355'); | |
532 EMIT2('\356'); EMIT2('\357'); | |
533 return OK; | |
534 | |
535 case 'n': case '\361': | |
536 EMIT2('n'); EMIT2('\361'); | |
537 return OK; | |
538 | |
539 case 'o': case '\362': case '\363': case '\364': case '\365': | |
540 case '\366': | |
541 EMIT2('o'); EMIT2('\362'); EMIT2('\363'); | |
542 EMIT2('\364'); EMIT2('\365'); EMIT2('\366'); | |
543 return OK; | |
544 | |
545 case 'u': case '\371': case '\372': case '\373': case '\374': | |
546 EMIT2('u'); EMIT2('\371'); EMIT2('\372'); | |
547 EMIT2('\373'); EMIT2('\374'); | |
548 return OK; | |
549 | |
550 case 'y': case '\375': case '\377': | |
551 EMIT2('y'); EMIT2('\375'); EMIT2('\377'); | |
552 return OK; | |
553 | |
554 default: | |
555 return FAIL; | |
556 } | |
557 } | |
558 | |
559 EMIT(c); | |
560 return OK; | |
561 #undef EMIT2 | |
562 } | |
563 | |
564 /* | |
565 * Code to parse regular expression. | |
566 * | |
567 * We try to reuse parsing functions in regexp.c to | |
568 * minimize surprise and keep the syntax consistent. | |
569 */ | |
570 | |
571 /* | |
572 * Parse the lowest level. | |
573 * | |
574 * An atom can be one of a long list of items. Many atoms match one character | |
575 * in the text. It is often an ordinary character or a character class. | |
576 * Braces can be used to make a pattern into an atom. The "\z(\)" construct | |
577 * is only for syntax highlighting. | |
578 * | |
579 * atom ::= ordinary-atom | |
580 * or \( pattern \) | |
581 * or \%( pattern \) | |
582 * or \z( pattern \) | |
583 */ | |
584 static int | |
585 nfa_regatom() | |
586 { | |
587 int c; | |
588 int charclass; | |
589 int equiclass; | |
590 int collclass; | |
591 int got_coll_char; | |
592 char_u *p; | |
593 char_u *endp; | |
594 #ifdef FEAT_MBYTE | |
595 char_u *old_regparse = regparse; | |
596 #endif | |
597 int extra = 0; | |
598 int first; | |
599 int emit_range; | |
600 int negated; | |
601 int result; | |
602 int startc = -1; | |
603 int endc = -1; | |
604 int oldstartc = -1; | |
605 int cpo_lit; /* 'cpoptions' contains 'l' flag */ | |
606 int cpo_bsl; /* 'cpoptions' contains '\' flag */ | |
607 int glue; /* ID that will "glue" nodes together */ | |
608 | |
609 cpo_lit = vim_strchr(p_cpo, CPO_LITERAL) != NULL; | |
610 cpo_bsl = vim_strchr(p_cpo, CPO_BACKSL) != NULL; | |
611 | |
612 c = getchr(); | |
613 switch (c) | |
614 { | |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
615 case NUL: |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
616 syntax_error = TRUE; |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
617 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
|
618 |
4444 | 619 case Magic('^'): |
620 EMIT(NFA_BOL); | |
621 break; | |
622 | |
623 case Magic('$'): | |
624 EMIT(NFA_EOL); | |
625 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
626 had_eol = TRUE; | |
627 #endif | |
628 break; | |
629 | |
630 case Magic('<'): | |
631 EMIT(NFA_BOW); | |
632 break; | |
633 | |
634 case Magic('>'): | |
635 EMIT(NFA_EOW); | |
636 break; | |
637 | |
638 case Magic('_'): | |
639 c = no_Magic(getchr()); | |
640 if (c == '^') /* "\_^" is start-of-line */ | |
641 { | |
642 EMIT(NFA_BOL); | |
643 break; | |
644 } | |
645 if (c == '$') /* "\_$" is end-of-line */ | |
646 { | |
647 EMIT(NFA_EOL); | |
648 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
649 had_eol = TRUE; | |
650 #endif | |
651 break; | |
652 } | |
653 | |
654 extra = ADD_NL; | |
655 | |
656 /* "\_[" is collection plus newline */ | |
657 if (c == '[') | |
4517
9a2183bd8295
updated for version 7.3.1006
Bram Moolenaar <bram@vim.org>
parents:
4515
diff
changeset
|
658 goto collection; |
4444 | 659 |
660 /* "\_x" is character class plus newline */ | |
661 /*FALLTHROUGH*/ | |
662 | |
663 /* | |
664 * Character classes. | |
665 */ | |
666 case Magic('.'): | |
667 case Magic('i'): | |
668 case Magic('I'): | |
669 case Magic('k'): | |
670 case Magic('K'): | |
671 case Magic('f'): | |
672 case Magic('F'): | |
673 case Magic('p'): | |
674 case Magic('P'): | |
675 case Magic('s'): | |
676 case Magic('S'): | |
677 case Magic('d'): | |
678 case Magic('D'): | |
679 case Magic('x'): | |
680 case Magic('X'): | |
681 case Magic('o'): | |
682 case Magic('O'): | |
683 case Magic('w'): | |
684 case Magic('W'): | |
685 case Magic('h'): | |
686 case Magic('H'): | |
687 case Magic('a'): | |
688 case Magic('A'): | |
689 case Magic('l'): | |
690 case Magic('L'): | |
691 case Magic('u'): | |
692 case Magic('U'): | |
693 p = vim_strchr(classchars, no_Magic(c)); | |
694 if (p == NULL) | |
695 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
696 EMSGN("INTERNAL: Unknown character class char: %ld", c); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
697 return FAIL; |
4444 | 698 } |
699 #ifdef FEAT_MBYTE | |
700 /* When '.' is followed by a composing char ignore the dot, so that | |
701 * the composing char is matched here. */ | |
702 if (enc_utf8 && c == Magic('.') && utf_iscomposing(peekchr())) | |
703 { | |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
704 old_regparse = regparse; |
4444 | 705 c = getchr(); |
706 goto nfa_do_multibyte; | |
707 } | |
708 #endif | |
709 EMIT(nfa_classcodes[p - classchars]); | |
710 if (extra == ADD_NL) | |
711 { | |
712 EMIT(NFA_NEWL); | |
713 EMIT(NFA_OR); | |
714 regflags |= RF_HASNL; | |
715 } | |
716 break; | |
717 | |
718 case Magic('n'): | |
719 if (reg_string) | |
720 /* In a string "\n" matches a newline character. */ | |
721 EMIT(NL); | |
722 else | |
723 { | |
724 /* In buffer text "\n" matches the end of a line. */ | |
725 EMIT(NFA_NEWL); | |
726 regflags |= RF_HASNL; | |
727 } | |
728 break; | |
729 | |
730 case Magic('('): | |
731 if (nfa_reg(REG_PAREN) == FAIL) | |
732 return FAIL; /* cascaded error */ | |
733 break; | |
734 | |
735 case Magic('|'): | |
736 case Magic('&'): | |
737 case Magic(')'): | |
738 syntax_error = TRUE; | |
4450 | 739 EMSGN(_(e_misplaced), no_Magic(c)); |
4444 | 740 return FAIL; |
741 | |
742 case Magic('='): | |
743 case Magic('?'): | |
744 case Magic('+'): | |
745 case Magic('@'): | |
746 case Magic('*'): | |
747 case Magic('{'): | |
748 /* these should follow an atom, not form an atom */ | |
749 syntax_error = TRUE; | |
4450 | 750 EMSGN(_(e_misplaced), no_Magic(c)); |
4444 | 751 return FAIL; |
752 | |
753 case Magic('~'): /* previous substitute pattern */ | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
754 /* TODO: Not supported yet */ |
4444 | 755 return FAIL; |
756 | |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
757 case Magic('1'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
758 case Magic('2'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
759 case Magic('3'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
760 case Magic('4'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
761 case Magic('5'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
762 case Magic('6'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
763 case Magic('7'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
764 case Magic('8'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
765 case Magic('9'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
766 EMIT(NFA_BACKREF1 + (no_Magic(c) - '1')); |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
767 nfa_has_backref = TRUE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
768 break; |
4444 | 769 |
770 case Magic('z'): | |
771 c = no_Magic(getchr()); | |
772 switch (c) | |
773 { | |
774 case 's': | |
775 EMIT(NFA_ZSTART); | |
776 break; | |
777 case 'e': | |
778 EMIT(NFA_ZEND); | |
779 nfa_has_zend = TRUE; | |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
780 break; |
4444 | 781 case '1': |
782 case '2': | |
783 case '3': | |
784 case '4': | |
785 case '5': | |
786 case '6': | |
787 case '7': | |
788 case '8': | |
789 case '9': | |
790 case '(': | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
791 /* TODO: \z1...\z9 and \z( not yet supported */ |
4444 | 792 return FAIL; |
793 default: | |
794 syntax_error = TRUE; | |
4450 | 795 EMSGN(_("E867: (NFA) Unknown operator '\\z%c'"), |
4444 | 796 no_Magic(c)); |
797 return FAIL; | |
798 } | |
799 break; | |
800 | |
801 case Magic('%'): | |
802 c = no_Magic(getchr()); | |
803 switch (c) | |
804 { | |
805 /* () without a back reference */ | |
806 case '(': | |
807 if (nfa_reg(REG_NPAREN) == FAIL) | |
808 return FAIL; | |
809 EMIT(NFA_NOPEN); | |
810 break; | |
811 | |
812 case 'd': /* %d123 decimal */ | |
813 case 'o': /* %o123 octal */ | |
814 case 'x': /* %xab hex 2 */ | |
815 case 'u': /* %uabcd hex 4 */ | |
816 case 'U': /* %U1234abcd hex 8 */ | |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
817 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
818 int nr; |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
819 |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
820 switch (c) |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
821 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
822 case 'd': nr = getdecchrs(); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
823 case 'o': nr = getoctchrs(); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
824 case 'x': nr = gethexchrs(2); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
825 case 'u': nr = gethexchrs(4); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
826 case 'U': nr = gethexchrs(8); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
827 default: nr = -1; break; |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
828 } |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
829 |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
830 if (nr < 0) |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
831 EMSG2_RET_FAIL( |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
832 _("E678: Invalid character after %s%%[dxouU]"), |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
833 reg_magic == MAGIC_ALL); |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
834 /* TODO: what if a composing character follows? */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
835 EMIT(nr); |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
836 } |
4444 | 837 break; |
838 | |
839 /* Catch \%^ and \%$ regardless of where they appear in the | |
840 * pattern -- regardless of whether or not it makes sense. */ | |
841 case '^': | |
842 EMIT(NFA_BOF); | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
843 /* TODO: Not yet supported */ |
4444 | 844 return FAIL; |
845 break; | |
846 | |
847 case '$': | |
848 EMIT(NFA_EOF); | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
849 /* TODO: Not yet supported */ |
4444 | 850 return FAIL; |
851 break; | |
852 | |
853 case '#': | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
854 EMIT(NFA_CURSOR); |
4444 | 855 break; |
856 | |
857 case 'V': | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
858 /* TODO: not supported yet */ |
4444 | 859 return FAIL; |
860 break; | |
861 | |
862 case '[': | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
863 /* TODO: \%[abc] not supported yet */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
864 return FAIL; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
865 |
4444 | 866 default: |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
867 { |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
868 long_u n = 0; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
869 int cmp = c; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
870 |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
871 if (c == '<' || c == '>') |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
872 c = getchr(); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
873 while (VIM_ISDIGIT(c)) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
874 { |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
875 n = n * 10 + (c - '0'); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
876 c = getchr(); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
877 } |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
878 if (c == 'l' || c == 'c' || c == 'v') |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
879 { |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
880 EMIT(n); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
881 if (c == 'l') |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
882 EMIT(cmp == '<' ? NFA_LNUM_LT : |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
883 cmp == '>' ? NFA_LNUM_GT : NFA_LNUM); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
884 else if (c == 'c') |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
885 EMIT(cmp == '<' ? NFA_COL_LT : |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
886 cmp == '>' ? NFA_COL_GT : NFA_COL); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
887 else |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
888 EMIT(cmp == '<' ? NFA_VCOL_LT : |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
889 cmp == '>' ? NFA_VCOL_GT : NFA_VCOL); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
890 break; |
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 else if (c == '\'') |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
893 /* TODO: \%'m not supported yet */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
894 return FAIL; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
895 } |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
896 syntax_error = TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
897 EMSGN(_("E867: (NFA) Unknown operator '\\%%%c'"), |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
898 no_Magic(c)); |
4444 | 899 return FAIL; |
900 } | |
901 break; | |
902 | |
903 case Magic('['): | |
4517
9a2183bd8295
updated for version 7.3.1006
Bram Moolenaar <bram@vim.org>
parents:
4515
diff
changeset
|
904 collection: |
4444 | 905 /* |
906 * Glue is emitted between several atoms from the []. | |
907 * It is either NFA_OR, or NFA_CONCAT. | |
908 * | |
909 * [abc] expands to 'a b NFA_OR c NFA_OR' (in postfix notation) | |
910 * [^abc] expands to 'a NFA_NOT b NFA_NOT NFA_CONCAT c NFA_NOT | |
911 * NFA_CONCAT NFA_END_NEG_RANGE NFA_CONCAT' (in postfix | |
912 * notation) | |
913 * | |
914 */ | |
915 | |
916 | |
917 /* Emit negation atoms, if needed. | |
918 * The CONCAT below merges the NOT with the previous node. */ | |
919 #define TRY_NEG() \ | |
920 if (negated == TRUE) \ | |
921 { \ | |
922 EMIT(NFA_NOT); \ | |
923 } | |
924 | |
925 /* Emit glue between important nodes : CONCAT or OR. */ | |
926 #define EMIT_GLUE() \ | |
927 if (first == FALSE) \ | |
928 EMIT(glue); \ | |
929 else \ | |
930 first = FALSE; | |
931 | |
932 p = regparse; | |
933 endp = skip_anyof(p); | |
934 if (*endp == ']') | |
935 { | |
936 /* | |
937 * Try to reverse engineer character classes. For example, | |
938 * recognize that [0-9] stands for \d and [A-Za-z_] with \h, | |
939 * and perform the necessary substitutions in the NFA. | |
940 */ | |
941 result = nfa_recognize_char_class(regparse, endp, | |
942 extra == ADD_NL); | |
943 if (result != FAIL) | |
944 { | |
945 if (result >= NFA_DIGIT && result <= NFA_NUPPER) | |
946 EMIT(result); | |
947 else /* must be char class + newline */ | |
948 { | |
949 EMIT(result - ADD_NL); | |
950 EMIT(NFA_NEWL); | |
951 EMIT(NFA_OR); | |
952 } | |
953 regparse = endp; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
954 mb_ptr_adv(regparse); |
4444 | 955 return OK; |
956 } | |
957 /* | |
958 * Failed to recognize a character class. Use the simple | |
959 * version that turns [abc] into 'a' OR 'b' OR 'c' | |
960 */ | |
961 startc = endc = oldstartc = -1; | |
962 first = TRUE; /* Emitting first atom in this sequence? */ | |
963 negated = FALSE; | |
964 glue = NFA_OR; | |
965 if (*regparse == '^') /* negated range */ | |
966 { | |
967 negated = TRUE; | |
968 glue = NFA_CONCAT; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
969 mb_ptr_adv(regparse); |
4444 | 970 } |
971 if (*regparse == '-') | |
972 { | |
973 startc = '-'; | |
974 EMIT(startc); | |
975 TRY_NEG(); | |
976 EMIT_GLUE(); | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
977 mb_ptr_adv(regparse); |
4444 | 978 } |
979 /* Emit the OR branches for each character in the [] */ | |
980 emit_range = FALSE; | |
981 while (regparse < endp) | |
982 { | |
983 oldstartc = startc; | |
984 startc = -1; | |
985 got_coll_char = FALSE; | |
986 if (*regparse == '[') | |
987 { | |
988 /* Check for [: :], [= =], [. .] */ | |
989 equiclass = collclass = 0; | |
990 charclass = get_char_class(®parse); | |
991 if (charclass == CLASS_NONE) | |
992 { | |
993 equiclass = get_equi_class(®parse); | |
994 if (equiclass == 0) | |
995 collclass = get_coll_element(®parse); | |
996 } | |
997 | |
998 /* Character class like [:alpha:] */ | |
999 if (charclass != CLASS_NONE) | |
1000 { | |
1001 switch (charclass) | |
1002 { | |
1003 case CLASS_ALNUM: | |
1004 EMIT(NFA_CLASS_ALNUM); | |
1005 break; | |
1006 case CLASS_ALPHA: | |
1007 EMIT(NFA_CLASS_ALPHA); | |
1008 break; | |
1009 case CLASS_BLANK: | |
1010 EMIT(NFA_CLASS_BLANK); | |
1011 break; | |
1012 case CLASS_CNTRL: | |
1013 EMIT(NFA_CLASS_CNTRL); | |
1014 break; | |
1015 case CLASS_DIGIT: | |
1016 EMIT(NFA_CLASS_DIGIT); | |
1017 break; | |
1018 case CLASS_GRAPH: | |
1019 EMIT(NFA_CLASS_GRAPH); | |
1020 break; | |
1021 case CLASS_LOWER: | |
1022 EMIT(NFA_CLASS_LOWER); | |
1023 break; | |
1024 case CLASS_PRINT: | |
1025 EMIT(NFA_CLASS_PRINT); | |
1026 break; | |
1027 case CLASS_PUNCT: | |
1028 EMIT(NFA_CLASS_PUNCT); | |
1029 break; | |
1030 case CLASS_SPACE: | |
1031 EMIT(NFA_CLASS_SPACE); | |
1032 break; | |
1033 case CLASS_UPPER: | |
1034 EMIT(NFA_CLASS_UPPER); | |
1035 break; | |
1036 case CLASS_XDIGIT: | |
1037 EMIT(NFA_CLASS_XDIGIT); | |
1038 break; | |
1039 case CLASS_TAB: | |
1040 EMIT(NFA_CLASS_TAB); | |
1041 break; | |
1042 case CLASS_RETURN: | |
1043 EMIT(NFA_CLASS_RETURN); | |
1044 break; | |
1045 case CLASS_BACKSPACE: | |
1046 EMIT(NFA_CLASS_BACKSPACE); | |
1047 break; | |
1048 case CLASS_ESCAPE: | |
1049 EMIT(NFA_CLASS_ESCAPE); | |
1050 break; | |
1051 } | |
1052 TRY_NEG(); | |
1053 EMIT_GLUE(); | |
1054 continue; | |
1055 } | |
1056 /* Try equivalence class [=a=] and the like */ | |
1057 if (equiclass != 0) | |
1058 { | |
1059 result = nfa_emit_equi_class(equiclass, negated); | |
1060 if (result == FAIL) | |
1061 { | |
1062 /* should never happen */ | |
1063 EMSG_RET_FAIL(_("E868: Error building NFA with equivalence class!")); | |
1064 } | |
1065 EMIT_GLUE(); | |
1066 continue; | |
1067 } | |
1068 /* Try collating class like [. .] */ | |
1069 if (collclass != 0) | |
1070 { | |
1071 startc = collclass; /* allow [.a.]-x as a range */ | |
1072 /* Will emit the proper atom at the end of the | |
1073 * while loop. */ | |
1074 } | |
1075 } | |
1076 /* Try a range like 'a-x' or '\t-z' */ | |
1077 if (*regparse == '-') | |
1078 { | |
1079 emit_range = TRUE; | |
1080 startc = oldstartc; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1081 mb_ptr_adv(regparse); |
4444 | 1082 continue; /* reading the end of the range */ |
1083 } | |
1084 | |
1085 /* Now handle simple and escaped characters. | |
1086 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim | |
1087 * accepts "\t", "\e", etc., but only when the 'l' flag in | |
1088 * 'cpoptions' is not included. | |
1089 * Posix doesn't recognize backslash at all. | |
1090 */ | |
1091 if (*regparse == '\\' | |
1092 && !cpo_bsl | |
1093 && regparse + 1 <= endp | |
1094 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL | |
1095 || (!cpo_lit | |
1096 && vim_strchr(REGEXP_ABBR, regparse[1]) | |
1097 != NULL) | |
1098 ) | |
1099 ) | |
1100 { | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1101 mb_ptr_adv(regparse); |
4444 | 1102 |
4507
9dbbddb2ed10
updated for version 7.3.1001
Bram Moolenaar <bram@vim.org>
parents:
4503
diff
changeset
|
1103 if (*regparse == 'n') |
4444 | 1104 startc = reg_string ? NL : NFA_NEWL; |
1105 else | |
1106 if (*regparse == 'd' | |
1107 || *regparse == 'o' | |
1108 || *regparse == 'x' | |
1109 || *regparse == 'u' | |
1110 || *regparse == 'U' | |
1111 ) | |
1112 { | |
1113 /* TODO(RE) This needs more testing */ | |
1114 startc = coll_get_char(); | |
1115 got_coll_char = TRUE; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1116 mb_ptr_back(old_regparse, regparse); |
4444 | 1117 } |
1118 else | |
1119 { | |
1120 /* \r,\t,\e,\b */ | |
1121 startc = backslash_trans(*regparse); | |
1122 } | |
1123 } | |
1124 | |
1125 /* Normal printable char */ | |
1126 if (startc == -1) | |
1127 #ifdef FEAT_MBYTE | |
1128 startc = (*mb_ptr2char)(regparse); | |
1129 #else | |
1130 startc = *regparse; | |
1131 #endif | |
1132 | |
1133 /* Previous char was '-', so this char is end of range. */ | |
1134 if (emit_range) | |
1135 { | |
1136 endc = startc; startc = oldstartc; | |
1137 if (startc > endc) | |
1138 EMSG_RET_FAIL(_(e_invrange)); | |
1139 #ifdef FEAT_MBYTE | |
1140 if (has_mbyte && ((*mb_char2len)(startc) > 1 | |
1141 || (*mb_char2len)(endc) > 1)) | |
1142 { | |
1143 if (endc > startc + 256) | |
1144 EMSG_RET_FAIL(_(e_invrange)); | |
1145 /* Emit the range. "startc" was already emitted, so | |
1146 * skip it. */ | |
1147 for (c = startc + 1; c <= endc; c++) | |
1148 { | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1149 EMIT(c); |
4444 | 1150 TRY_NEG(); |
1151 EMIT_GLUE(); | |
1152 } | |
1153 emit_range = FALSE; | |
1154 } | |
1155 else | |
1156 #endif | |
1157 { | |
1158 #ifdef EBCDIC | |
1159 int alpha_only = FALSE; | |
1160 | |
1161 /* for alphabetical range skip the gaps | |
1162 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */ | |
1163 if (isalpha(startc) && isalpha(endc)) | |
1164 alpha_only = TRUE; | |
1165 #endif | |
1166 /* Emit the range. "startc" was already emitted, so | |
1167 * skip it. */ | |
1168 for (c = startc + 1; c <= endc; c++) | |
1169 #ifdef EBCDIC | |
1170 if (!alpha_only || isalpha(startc)) | |
1171 #endif | |
1172 { | |
1173 EMIT(c); | |
1174 TRY_NEG(); | |
1175 EMIT_GLUE(); | |
1176 } | |
1177 emit_range = FALSE; | |
1178 } | |
1179 } | |
1180 else | |
1181 { | |
1182 /* | |
1183 * This char (startc) is not part of a range. Just | |
1184 * emit it. | |
1185 * | |
1186 * Normally, simply emit startc. But if we get char | |
1187 * code=0 from a collating char, then replace it with | |
1188 * 0x0a. | |
1189 * | |
1190 * This is needed to completely mimic the behaviour of | |
1191 * the backtracking engine. | |
1192 */ | |
1193 if (got_coll_char == TRUE && startc == 0) | |
1194 EMIT(0x0a); | |
1195 else | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1196 EMIT(startc); |
4444 | 1197 TRY_NEG(); |
1198 EMIT_GLUE(); | |
1199 } | |
1200 | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1201 mb_ptr_adv(regparse); |
4444 | 1202 } /* while (p < endp) */ |
1203 | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1204 mb_ptr_back(old_regparse, regparse); |
4444 | 1205 if (*regparse == '-') /* if last, '-' is just a char */ |
1206 { | |
1207 EMIT('-'); | |
1208 TRY_NEG(); | |
1209 EMIT_GLUE(); | |
1210 } | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1211 mb_ptr_adv(regparse); |
4444 | 1212 |
1213 /* skip the trailing ] */ | |
1214 regparse = endp; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1215 mb_ptr_adv(regparse); |
4444 | 1216 if (negated == TRUE) |
1217 { | |
1218 /* Mark end of negated char range */ | |
1219 EMIT(NFA_END_NEG_RANGE); | |
1220 EMIT(NFA_CONCAT); | |
1221 } | |
4615
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1222 |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1223 /* \_[] also matches \n but it's not negated */ |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1224 if (extra == ADD_NL) |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1225 { |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1226 EMIT(reg_string ? NL : NFA_NEWL); |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1227 EMIT(NFA_OR); |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1228 } |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1229 |
4444 | 1230 return OK; |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1231 } /* if exists closing ] */ |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1232 |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1233 if (reg_strict) |
4444 | 1234 { |
1235 syntax_error = TRUE; | |
1236 EMSG_RET_FAIL(_(e_missingbracket)); | |
1237 } | |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1238 /* FALLTHROUGH */ |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1239 |
4444 | 1240 default: |
1241 { | |
1242 #ifdef FEAT_MBYTE | |
1243 int plen; | |
1244 | |
1245 nfa_do_multibyte: | |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1246 /* 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
|
1247 if (enc_utf8 && ((*mb_char2len)(c) |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1248 != (plen = (*mb_ptr2len)(old_regparse)) |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1249 || utf_iscomposing(c))) |
4444 | 1250 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
1251 int i = 0; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
1252 |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
1253 /* 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
|
1254 * or more composing characters. |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1255 * 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
|
1256 * 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
|
1257 * NFA_END_COMPOSING is the ). Note that right now we are |
4444 | 1258 * building the postfix form, not the NFA itself; |
1259 * 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
|
1260 * 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
|
1261 for (;;) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1262 { |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1263 EMIT(c); |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1264 if (i > 0) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1265 EMIT(NFA_CONCAT); |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1266 if ((i += utf_char2len(c)) >= plen) |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1267 break; |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1268 c = utf_ptr2char(old_regparse + i); |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1269 } |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1270 EMIT(NFA_COMPOSING); |
4444 | 1271 regparse = old_regparse + plen; |
1272 } | |
1273 else | |
1274 #endif | |
1275 { | |
1276 c = no_Magic(c); | |
1277 EMIT(c); | |
1278 } | |
1279 return OK; | |
1280 } | |
1281 } | |
1282 | |
1283 #undef TRY_NEG | |
1284 #undef EMIT_GLUE | |
1285 | |
1286 return OK; | |
1287 } | |
1288 | |
1289 /* | |
1290 * Parse something followed by possible [*+=]. | |
1291 * | |
1292 * A piece is an atom, possibly followed by a multi, an indication of how many | |
1293 * times the atom can be matched. Example: "a*" matches any sequence of "a" | |
1294 * characters: "", "a", "aa", etc. | |
1295 * | |
1296 * piece ::= atom | |
1297 * or atom multi | |
1298 */ | |
1299 static int | |
1300 nfa_regpiece() | |
1301 { | |
1302 int i; | |
1303 int op; | |
1304 int ret; | |
1305 long minval, maxval; | |
1306 int greedy = TRUE; /* Braces are prefixed with '-' ? */ | |
1307 char_u *old_regparse, *new_regparse; | |
1308 int c2; | |
1309 int *old_post_ptr, *my_post_start; | |
1310 int old_regnpar; | |
1311 int quest; | |
1312 | |
1313 /* Save the current position in the regexp, so that we can use it if | |
1314 * <atom>{m,n} is next. */ | |
1315 old_regparse = regparse; | |
1316 /* Save current number of open parenthesis, so we can use it if | |
1317 * <atom>{m,n} is next */ | |
1318 old_regnpar = regnpar; | |
1319 /* store current pos in the postfix form, for \{m,n} involving 0s */ | |
1320 my_post_start = post_ptr; | |
1321 | |
1322 ret = nfa_regatom(); | |
1323 if (ret == FAIL) | |
1324 return FAIL; /* cascaded error */ | |
1325 | |
1326 op = peekchr(); | |
1327 if (re_multi_type(op) == NOT_MULTI) | |
1328 return OK; | |
1329 | |
1330 skipchr(); | |
1331 switch (op) | |
1332 { | |
1333 case Magic('*'): | |
1334 EMIT(NFA_STAR); | |
1335 break; | |
1336 | |
1337 case Magic('+'): | |
1338 /* | |
1339 * Trick: Normally, (a*)\+ would match the whole input "aaa". The | |
1340 * first and only submatch would be "aaa". But the backtracking | |
1341 * engine interprets the plus as "try matching one more time", and | |
1342 * a* matches a second time at the end of the input, the empty | |
1343 * string. | |
1344 * The submatch will the empty string. | |
1345 * | |
1346 * In order to be consistent with the old engine, we disable | |
1347 * NFA_PLUS, and replace <atom>+ with <atom><atom>* | |
1348 */ | |
1349 /* EMIT(NFA_PLUS); */ | |
1350 regnpar = old_regnpar; | |
1351 regparse = old_regparse; | |
1352 curchr = -1; | |
1353 if (nfa_regatom() == FAIL) | |
1354 return FAIL; | |
1355 EMIT(NFA_STAR); | |
1356 EMIT(NFA_CONCAT); | |
1357 skipchr(); /* skip the \+ */ | |
1358 break; | |
1359 | |
1360 case Magic('@'): | |
1361 op = no_Magic(getchr()); | |
1362 switch(op) | |
1363 { | |
1364 case '=': | |
1365 EMIT(NFA_PREV_ATOM_NO_WIDTH); | |
1366 break; | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1367 case '0': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1368 case '1': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1369 case '2': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1370 case '3': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1371 case '4': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1372 case '5': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1373 case '6': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1374 case '7': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1375 case '8': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1376 case '9': |
4444 | 1377 case '!': |
1378 case '<': | |
1379 case '>': | |
1380 /* Not supported yet */ | |
1381 return FAIL; | |
1382 default: | |
1383 syntax_error = TRUE; | |
4450 | 1384 EMSGN(_("E869: (NFA) Unknown operator '\\@%c'"), op); |
4444 | 1385 return FAIL; |
1386 } | |
1387 break; | |
1388 | |
1389 case Magic('?'): | |
1390 case Magic('='): | |
1391 EMIT(NFA_QUEST); | |
1392 break; | |
1393 | |
1394 case Magic('{'): | |
1395 /* a{2,5} will expand to 'aaa?a?a?' | |
1396 * a{-1,3} will expand to 'aa??a??', where ?? is the nongreedy | |
1397 * version of '?' | |
1398 * \v(ab){2,3} will expand to '(ab)(ab)(ab)?', where all the | |
1399 * parenthesis have the same id | |
1400 */ | |
1401 | |
1402 greedy = TRUE; | |
1403 c2 = peekchr(); | |
1404 if (c2 == '-' || c2 == Magic('-')) | |
1405 { | |
1406 skipchr(); | |
1407 greedy = FALSE; | |
1408 } | |
1409 if (!read_limits(&minval, &maxval)) | |
1410 { | |
1411 syntax_error = TRUE; | |
1412 EMSG_RET_FAIL(_("E870: (NFA regexp) Error reading repetition limits")); | |
1413 } | |
1414 /* <atom>{0,inf}, <atom>{0,} and <atom>{} are equivalent to | |
1415 * <atom>* */ | |
1416 if (minval == 0 && maxval == MAX_LIMIT && greedy) | |
1417 { | |
1418 EMIT(NFA_STAR); | |
1419 break; | |
1420 } | |
1421 | |
1422 if (maxval > NFA_BRACES_MAXLIMIT) | |
1423 { | |
1424 /* This would yield a huge automaton and use too much memory. | |
1425 * Revert to old engine */ | |
1426 return FAIL; | |
1427 } | |
1428 | |
1429 /* Special case: x{0} or x{-0} */ | |
1430 if (maxval == 0) | |
1431 { | |
1432 /* Ignore result of previous call to nfa_regatom() */ | |
1433 post_ptr = my_post_start; | |
1434 /* NFA_SKIP_CHAR has 0-length and works everywhere */ | |
1435 EMIT(NFA_SKIP_CHAR); | |
1436 return OK; | |
1437 } | |
1438 | |
1439 /* Ignore previous call to nfa_regatom() */ | |
1440 post_ptr = my_post_start; | |
1441 /* Save pos after the repeated atom and the \{} */ | |
1442 new_regparse = regparse; | |
1443 | |
1444 quest = (greedy == TRUE? NFA_QUEST : NFA_QUEST_NONGREEDY); | |
1445 for (i = 0; i < maxval; i++) | |
1446 { | |
1447 /* Goto beginning of the repeated atom */ | |
1448 regparse = old_regparse; | |
1449 curchr = -1; | |
1450 /* Restore count of parenthesis */ | |
1451 regnpar = old_regnpar; | |
1452 old_post_ptr = post_ptr; | |
1453 if (nfa_regatom() == FAIL) | |
1454 return FAIL; | |
1455 /* after "minval" times, atoms are optional */ | |
1456 if (i + 1 > minval) | |
1457 EMIT(quest); | |
1458 if (old_post_ptr != my_post_start) | |
1459 EMIT(NFA_CONCAT); | |
1460 } | |
1461 | |
1462 /* Go to just after the repeated atom and the \{} */ | |
1463 regparse = new_regparse; | |
1464 curchr = -1; | |
1465 | |
1466 break; | |
1467 | |
1468 | |
1469 default: | |
1470 break; | |
1471 } /* end switch */ | |
1472 | |
1473 if (re_multi_type(peekchr()) != NOT_MULTI) | |
1474 { | |
1475 /* Can't have a multi follow a multi. */ | |
1476 syntax_error = TRUE; | |
1477 EMSG_RET_FAIL(_("E871: (NFA regexp) Can't have a multi follow a multi !")); | |
1478 } | |
1479 | |
1480 return OK; | |
1481 } | |
1482 | |
1483 /* | |
1484 * Parse one or more pieces, concatenated. It matches a match for the | |
1485 * first piece, followed by a match for the second piece, etc. Example: | |
1486 * "f[0-9]b", first matches "f", then a digit and then "b". | |
1487 * | |
1488 * concat ::= piece | |
1489 * or piece piece | |
1490 * or piece piece piece | |
1491 * etc. | |
1492 */ | |
1493 static int | |
1494 nfa_regconcat() | |
1495 { | |
1496 int cont = TRUE; | |
1497 int first = TRUE; | |
1498 | |
1499 while (cont) | |
1500 { | |
1501 switch (peekchr()) | |
1502 { | |
1503 case NUL: | |
1504 case Magic('|'): | |
1505 case Magic('&'): | |
1506 case Magic(')'): | |
1507 cont = FALSE; | |
1508 break; | |
1509 | |
1510 case Magic('Z'): | |
1511 #ifdef FEAT_MBYTE | |
1512 regflags |= RF_ICOMBINE; | |
1513 #endif | |
1514 skipchr_keepstart(); | |
1515 break; | |
1516 case Magic('c'): | |
1517 regflags |= RF_ICASE; | |
1518 skipchr_keepstart(); | |
1519 break; | |
1520 case Magic('C'): | |
1521 regflags |= RF_NOICASE; | |
1522 skipchr_keepstart(); | |
1523 break; | |
1524 case Magic('v'): | |
1525 reg_magic = MAGIC_ALL; | |
1526 skipchr_keepstart(); | |
1527 curchr = -1; | |
1528 break; | |
1529 case Magic('m'): | |
1530 reg_magic = MAGIC_ON; | |
1531 skipchr_keepstart(); | |
1532 curchr = -1; | |
1533 break; | |
1534 case Magic('M'): | |
1535 reg_magic = MAGIC_OFF; | |
1536 skipchr_keepstart(); | |
1537 curchr = -1; | |
1538 break; | |
1539 case Magic('V'): | |
1540 reg_magic = MAGIC_NONE; | |
1541 skipchr_keepstart(); | |
1542 curchr = -1; | |
1543 break; | |
1544 | |
1545 default: | |
1546 if (nfa_regpiece() == FAIL) | |
1547 return FAIL; | |
1548 if (first == FALSE) | |
1549 EMIT(NFA_CONCAT); | |
1550 else | |
1551 first = FALSE; | |
1552 break; | |
1553 } | |
1554 } | |
1555 | |
1556 return OK; | |
1557 } | |
1558 | |
1559 /* | |
1560 * Parse a branch, one or more concats, separated by "\&". It matches the | |
1561 * last concat, but only if all the preceding concats also match at the same | |
1562 * position. Examples: | |
1563 * "foobeep\&..." matches "foo" in "foobeep". | |
1564 * ".*Peter\&.*Bob" matches in a line containing both "Peter" and "Bob" | |
1565 * | |
1566 * branch ::= concat | |
1567 * or concat \& concat | |
1568 * or concat \& concat \& concat | |
1569 * etc. | |
1570 */ | |
1571 static int | |
1572 nfa_regbranch() | |
1573 { | |
1574 int ch; | |
1575 int *old_post_ptr; | |
1576 | |
1577 old_post_ptr = post_ptr; | |
1578 | |
1579 /* First branch, possibly the only one */ | |
1580 if (nfa_regconcat() == FAIL) | |
1581 return FAIL; | |
1582 | |
1583 ch = peekchr(); | |
1584 /* Try next concats */ | |
1585 while (ch == Magic('&')) | |
1586 { | |
1587 skipchr(); | |
1588 EMIT(NFA_NOPEN); | |
1589 EMIT(NFA_PREV_ATOM_NO_WIDTH); | |
1590 old_post_ptr = post_ptr; | |
1591 if (nfa_regconcat() == FAIL) | |
1592 return FAIL; | |
1593 /* if concat is empty, skip a input char. But do emit a node */ | |
1594 if (old_post_ptr == post_ptr) | |
1595 EMIT(NFA_SKIP_CHAR); | |
1596 EMIT(NFA_CONCAT); | |
1597 ch = peekchr(); | |
1598 } | |
1599 | |
1600 /* Even if a branch is empty, emit one node for it */ | |
1601 if (old_post_ptr == post_ptr) | |
1602 EMIT(NFA_SKIP_CHAR); | |
1603 | |
1604 return OK; | |
1605 } | |
1606 | |
1607 /* | |
1608 * Parse a pattern, one or more branches, separated by "\|". It matches | |
1609 * anything that matches one of the branches. Example: "foo\|beep" matches | |
1610 * "foo" and matches "beep". If more than one branch matches, the first one | |
1611 * is used. | |
1612 * | |
1613 * pattern ::= branch | |
1614 * or branch \| branch | |
1615 * or branch \| branch \| branch | |
1616 * etc. | |
1617 */ | |
1618 static int | |
1619 nfa_reg(paren) | |
1620 int paren; /* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */ | |
1621 { | |
1622 int parno = 0; | |
1623 | |
1624 #ifdef FEAT_SYN_HL | |
1625 #endif | |
1626 if (paren == REG_PAREN) | |
1627 { | |
1628 if (regnpar >= NSUBEXP) /* Too many `(' */ | |
1629 { | |
1630 syntax_error = TRUE; | |
1631 EMSG_RET_FAIL(_("E872: (NFA regexp) Too many '('")); | |
1632 } | |
1633 parno = regnpar++; | |
1634 } | |
1635 | |
1636 if (nfa_regbranch() == FAIL) | |
1637 return FAIL; /* cascaded error */ | |
1638 | |
1639 while (peekchr() == Magic('|')) | |
1640 { | |
1641 skipchr(); | |
1642 if (nfa_regbranch() == FAIL) | |
1643 return FAIL; /* cascaded error */ | |
1644 EMIT(NFA_OR); | |
1645 } | |
1646 | |
1647 /* Check for proper termination. */ | |
1648 if (paren != REG_NOPAREN && getchr() != Magic(')')) | |
1649 { | |
1650 syntax_error = TRUE; | |
1651 if (paren == REG_NPAREN) | |
1652 EMSG2_RET_FAIL(_(e_unmatchedpp), reg_magic == MAGIC_ALL); | |
1653 else | |
1654 EMSG2_RET_FAIL(_(e_unmatchedp), reg_magic == MAGIC_ALL); | |
1655 } | |
1656 else if (paren == REG_NOPAREN && peekchr() != NUL) | |
1657 { | |
1658 syntax_error = TRUE; | |
1659 if (peekchr() == Magic(')')) | |
1660 EMSG2_RET_FAIL(_(e_unmatchedpar), reg_magic == MAGIC_ALL); | |
1661 else | |
1662 EMSG_RET_FAIL(_("E873: (NFA regexp) proper termination error")); | |
1663 } | |
1664 /* | |
1665 * Here we set the flag allowing back references to this set of | |
1666 * parentheses. | |
1667 */ | |
1668 if (paren == REG_PAREN) | |
1669 { | |
1670 had_endbrace[parno] = TRUE; /* have seen the close paren */ | |
1671 EMIT(NFA_MOPEN + parno); | |
1672 } | |
1673 | |
1674 return OK; | |
1675 } | |
1676 | |
1677 #ifdef DEBUG | |
1678 static char_u code[50]; | |
1679 | |
1680 static void | |
1681 nfa_set_code(c) | |
1682 int c; | |
1683 { | |
1684 int addnl = FALSE; | |
1685 | |
1686 if (c >= NFA_FIRST_NL && c <= NFA_LAST_NL) | |
1687 { | |
1688 addnl = TRUE; | |
1689 c -= ADD_NL; | |
1690 } | |
1691 | |
1692 STRCPY(code, ""); | |
1693 switch (c) | |
1694 { | |
1695 case NFA_MATCH: STRCPY(code, "NFA_MATCH "); break; | |
1696 case NFA_SPLIT: STRCPY(code, "NFA_SPLIT "); break; | |
1697 case NFA_CONCAT: STRCPY(code, "NFA_CONCAT "); break; | |
1698 case NFA_NEWL: STRCPY(code, "NFA_NEWL "); break; | |
1699 case NFA_ZSTART: STRCPY(code, "NFA_ZSTART"); break; | |
1700 case NFA_ZEND: STRCPY(code, "NFA_ZEND"); break; | |
1701 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1702 case NFA_BACKREF1: STRCPY(code, "NFA_BACKREF1"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1703 case NFA_BACKREF2: STRCPY(code, "NFA_BACKREF2"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1704 case NFA_BACKREF3: STRCPY(code, "NFA_BACKREF3"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1705 case NFA_BACKREF4: STRCPY(code, "NFA_BACKREF4"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1706 case NFA_BACKREF5: STRCPY(code, "NFA_BACKREF5"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1707 case NFA_BACKREF6: STRCPY(code, "NFA_BACKREF6"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1708 case NFA_BACKREF7: STRCPY(code, "NFA_BACKREF7"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1709 case NFA_BACKREF8: STRCPY(code, "NFA_BACKREF8"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1710 case NFA_BACKREF9: STRCPY(code, "NFA_BACKREF9"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1711 case NFA_SKIP: STRCPY(code, "NFA_SKIP"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1712 |
4444 | 1713 case NFA_PREV_ATOM_NO_WIDTH: |
1714 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
|
1715 case NFA_PREV_ATOM_NO_WIDTH_NEG: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
1716 STRCPY(code, "NFA_PREV_ATOM_NO_WIDTH_NEG"); break; |
4444 | 1717 case NFA_NOPEN: STRCPY(code, "NFA_MOPEN_INVISIBLE"); break; |
1718 case NFA_NCLOSE: STRCPY(code, "NFA_MCLOSE_INVISIBLE"); break; | |
1719 case NFA_START_INVISIBLE: STRCPY(code, "NFA_START_INVISIBLE"); break; | |
1720 case NFA_END_INVISIBLE: STRCPY(code, "NFA_END_INVISIBLE"); break; | |
1721 | |
1722 case NFA_COMPOSING: STRCPY(code, "NFA_COMPOSING"); break; | |
1723 case NFA_END_COMPOSING: STRCPY(code, "NFA_END_COMPOSING"); break; | |
1724 | |
1725 case NFA_MOPEN + 0: | |
1726 case NFA_MOPEN + 1: | |
1727 case NFA_MOPEN + 2: | |
1728 case NFA_MOPEN + 3: | |
1729 case NFA_MOPEN + 4: | |
1730 case NFA_MOPEN + 5: | |
1731 case NFA_MOPEN + 6: | |
1732 case NFA_MOPEN + 7: | |
1733 case NFA_MOPEN + 8: | |
1734 case NFA_MOPEN + 9: | |
1735 STRCPY(code, "NFA_MOPEN(x)"); | |
1736 code[10] = c - NFA_MOPEN + '0'; | |
1737 break; | |
1738 case NFA_MCLOSE + 0: | |
1739 case NFA_MCLOSE + 1: | |
1740 case NFA_MCLOSE + 2: | |
1741 case NFA_MCLOSE + 3: | |
1742 case NFA_MCLOSE + 4: | |
1743 case NFA_MCLOSE + 5: | |
1744 case NFA_MCLOSE + 6: | |
1745 case NFA_MCLOSE + 7: | |
1746 case NFA_MCLOSE + 8: | |
1747 case NFA_MCLOSE + 9: | |
1748 STRCPY(code, "NFA_MCLOSE(x)"); | |
1749 code[11] = c - NFA_MCLOSE + '0'; | |
1750 break; | |
1751 case NFA_EOL: STRCPY(code, "NFA_EOL "); break; | |
1752 case NFA_BOL: STRCPY(code, "NFA_BOL "); break; | |
1753 case NFA_EOW: STRCPY(code, "NFA_EOW "); break; | |
1754 case NFA_BOW: STRCPY(code, "NFA_BOW "); break; | |
1755 case NFA_STAR: STRCPY(code, "NFA_STAR "); break; | |
1756 case NFA_PLUS: STRCPY(code, "NFA_PLUS "); break; | |
1757 case NFA_NOT: STRCPY(code, "NFA_NOT "); break; | |
1758 case NFA_SKIP_CHAR: STRCPY(code, "NFA_SKIP_CHAR"); break; | |
1759 case NFA_OR: STRCPY(code, "NFA_OR"); break; | |
1760 case NFA_QUEST: STRCPY(code, "NFA_QUEST"); break; | |
1761 case NFA_QUEST_NONGREEDY: STRCPY(code, "NFA_QUEST_NON_GREEDY"); break; | |
1762 case NFA_END_NEG_RANGE: STRCPY(code, "NFA_END_NEG_RANGE"); break; | |
1763 case NFA_CLASS_ALNUM: STRCPY(code, "NFA_CLASS_ALNUM"); break; | |
1764 case NFA_CLASS_ALPHA: STRCPY(code, "NFA_CLASS_ALPHA"); break; | |
1765 case NFA_CLASS_BLANK: STRCPY(code, "NFA_CLASS_BLANK"); break; | |
1766 case NFA_CLASS_CNTRL: STRCPY(code, "NFA_CLASS_CNTRL"); break; | |
1767 case NFA_CLASS_DIGIT: STRCPY(code, "NFA_CLASS_DIGIT"); break; | |
1768 case NFA_CLASS_GRAPH: STRCPY(code, "NFA_CLASS_GRAPH"); break; | |
1769 case NFA_CLASS_LOWER: STRCPY(code, "NFA_CLASS_LOWER"); break; | |
1770 case NFA_CLASS_PRINT: STRCPY(code, "NFA_CLASS_PRINT"); break; | |
1771 case NFA_CLASS_PUNCT: STRCPY(code, "NFA_CLASS_PUNCT"); break; | |
1772 case NFA_CLASS_SPACE: STRCPY(code, "NFA_CLASS_SPACE"); break; | |
1773 case NFA_CLASS_UPPER: STRCPY(code, "NFA_CLASS_UPPER"); break; | |
1774 case NFA_CLASS_XDIGIT: STRCPY(code, "NFA_CLASS_XDIGIT"); break; | |
1775 case NFA_CLASS_TAB: STRCPY(code, "NFA_CLASS_TAB"); break; | |
1776 case NFA_CLASS_RETURN: STRCPY(code, "NFA_CLASS_RETURN"); break; | |
1777 case NFA_CLASS_BACKSPACE: STRCPY(code, "NFA_CLASS_BACKSPACE"); break; | |
1778 case NFA_CLASS_ESCAPE: STRCPY(code, "NFA_CLASS_ESCAPE"); break; | |
1779 | |
1780 case NFA_ANY: STRCPY(code, "NFA_ANY"); break; | |
1781 case NFA_IDENT: STRCPY(code, "NFA_IDENT"); break; | |
1782 case NFA_SIDENT:STRCPY(code, "NFA_SIDENT"); break; | |
1783 case NFA_KWORD: STRCPY(code, "NFA_KWORD"); break; | |
1784 case NFA_SKWORD:STRCPY(code, "NFA_SKWORD"); break; | |
1785 case NFA_FNAME: STRCPY(code, "NFA_FNAME"); break; | |
1786 case NFA_SFNAME:STRCPY(code, "NFA_SFNAME"); break; | |
1787 case NFA_PRINT: STRCPY(code, "NFA_PRINT"); break; | |
1788 case NFA_SPRINT:STRCPY(code, "NFA_SPRINT"); break; | |
1789 case NFA_WHITE: STRCPY(code, "NFA_WHITE"); break; | |
1790 case NFA_NWHITE:STRCPY(code, "NFA_NWHITE"); break; | |
1791 case NFA_DIGIT: STRCPY(code, "NFA_DIGIT"); break; | |
1792 case NFA_NDIGIT:STRCPY(code, "NFA_NDIGIT"); break; | |
1793 case NFA_HEX: STRCPY(code, "NFA_HEX"); break; | |
1794 case NFA_NHEX: STRCPY(code, "NFA_NHEX"); break; | |
1795 case NFA_OCTAL: STRCPY(code, "NFA_OCTAL"); break; | |
1796 case NFA_NOCTAL:STRCPY(code, "NFA_NOCTAL"); break; | |
1797 case NFA_WORD: STRCPY(code, "NFA_WORD"); break; | |
1798 case NFA_NWORD: STRCPY(code, "NFA_NWORD"); break; | |
1799 case NFA_HEAD: STRCPY(code, "NFA_HEAD"); break; | |
1800 case NFA_NHEAD: STRCPY(code, "NFA_NHEAD"); break; | |
1801 case NFA_ALPHA: STRCPY(code, "NFA_ALPHA"); break; | |
1802 case NFA_NALPHA:STRCPY(code, "NFA_NALPHA"); break; | |
1803 case NFA_LOWER: STRCPY(code, "NFA_LOWER"); break; | |
1804 case NFA_NLOWER:STRCPY(code, "NFA_NLOWER"); break; | |
1805 case NFA_UPPER: STRCPY(code, "NFA_UPPER"); break; | |
1806 case NFA_NUPPER:STRCPY(code, "NFA_NUPPER"); break; | |
1807 | |
1808 default: | |
1809 STRCPY(code, "CHAR(x)"); | |
1810 code[5] = c; | |
1811 } | |
1812 | |
1813 if (addnl == TRUE) | |
1814 STRCAT(code, " + NEWLINE "); | |
1815 | |
1816 } | |
1817 | |
1818 #ifdef ENABLE_LOG | |
1819 static FILE *log_fd; | |
1820 | |
1821 /* | |
1822 * Print the postfix notation of the current regexp. | |
1823 */ | |
1824 static void | |
1825 nfa_postfix_dump(expr, retval) | |
1826 char_u *expr; | |
1827 int retval; | |
1828 { | |
1829 int *p; | |
1830 FILE *f; | |
1831 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
1832 f = fopen(NFA_REGEXP_DUMP_LOG, "a"); |
4444 | 1833 if (f != NULL) |
1834 { | |
1835 fprintf(f, "\n-------------------------\n"); | |
1836 if (retval == FAIL) | |
1837 fprintf(f, ">>> NFA engine failed ... \n"); | |
1838 else if (retval == OK) | |
1839 fprintf(f, ">>> NFA engine succeeded !\n"); | |
1840 fprintf(f, "Regexp: \"%s\"\nPostfix notation (char): \"", expr); | |
4464 | 1841 for (p = post_start; *p && p < post_end; p++) |
4444 | 1842 { |
1843 nfa_set_code(*p); | |
1844 fprintf(f, "%s, ", code); | |
1845 } | |
1846 fprintf(f, "\"\nPostfix notation (int): "); | |
4464 | 1847 for (p = post_start; *p && p < post_end; p++) |
4444 | 1848 fprintf(f, "%d ", *p); |
1849 fprintf(f, "\n\n"); | |
1850 fclose(f); | |
1851 } | |
1852 } | |
1853 | |
1854 /* | |
1855 * Print the NFA starting with a root node "state". | |
1856 */ | |
1857 static void | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1858 nfa_print_state(debugf, state) |
4444 | 1859 FILE *debugf; |
1860 nfa_state_T *state; | |
1861 { | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1862 garray_T indent; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1863 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1864 ga_init2(&indent, 1, 64); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1865 ga_append(&indent, '\0'); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1866 nfa_print_state2(debugf, state, &indent); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1867 ga_clear(&indent); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1868 } |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1869 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1870 static void |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1871 nfa_print_state2(debugf, state, indent) |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1872 FILE *debugf; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1873 nfa_state_T *state; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1874 garray_T *indent; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1875 { |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1876 char_u *p; |
4444 | 1877 |
1878 if (state == NULL) | |
1879 return; | |
1880 | |
1881 fprintf(debugf, "(%2d)", abs(state->id)); | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1882 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1883 /* Output indent */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1884 p = (char_u *)indent->ga_data; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1885 if (indent->ga_len >= 3) |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1886 { |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1887 int last = indent->ga_len - 3; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1888 char_u save[2]; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1889 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1890 STRNCPY(save, &p[last], 2); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1891 STRNCPY(&p[last], "+-", 2); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1892 fprintf(debugf, " %s", p); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1893 STRNCPY(&p[last], save, 2); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1894 } |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1895 else |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1896 fprintf(debugf, " %s", p); |
4444 | 1897 |
1898 nfa_set_code(state->c); | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1899 fprintf(debugf, "%s%s (%d) (id=%d)\n", |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1900 state->negated ? "NOT " : "", code, state->c, abs(state->id)); |
4444 | 1901 if (state->id < 0) |
1902 return; | |
1903 | |
1904 state->id = abs(state->id) * -1; | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1905 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1906 /* grow indent for state->out */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1907 indent->ga_len -= 1; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1908 if (state->out1) |
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1909 ga_concat(indent, (char_u *)"| "); |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1910 else |
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1911 ga_concat(indent, (char_u *)" "); |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1912 ga_append(indent, '\0'); |
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 nfa_print_state2(debugf, state->out, indent); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1915 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1916 /* replace last part of indent for state->out1 */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1917 indent->ga_len -= 3; |
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1918 ga_concat(indent, (char_u *)" "); |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1919 ga_append(indent, '\0'); |
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 nfa_print_state2(debugf, state->out1, indent); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1922 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1923 /* shrink indent */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1924 indent->ga_len -= 3; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1925 ga_append(indent, '\0'); |
4444 | 1926 } |
1927 | |
1928 /* | |
1929 * Print the NFA state machine. | |
1930 */ | |
1931 static void | |
1932 nfa_dump(prog) | |
1933 nfa_regprog_T *prog; | |
1934 { | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
1935 FILE *debugf = fopen(NFA_REGEXP_DUMP_LOG, "a"); |
4444 | 1936 |
1937 if (debugf != NULL) | |
1938 { | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1939 nfa_print_state(debugf, prog->start); |
4444 | 1940 fclose(debugf); |
1941 } | |
1942 } | |
1943 #endif /* ENABLE_LOG */ | |
1944 #endif /* DEBUG */ | |
1945 | |
1946 /* | |
1947 * Parse r.e. @expr and convert it into postfix form. | |
1948 * Return the postfix string on success, NULL otherwise. | |
1949 */ | |
1950 static int * | |
1951 re2post() | |
1952 { | |
1953 if (nfa_reg(REG_NOPAREN) == FAIL) | |
1954 return NULL; | |
1955 EMIT(NFA_MOPEN); | |
1956 return post_start; | |
1957 } | |
1958 | |
1959 /* NB. Some of the code below is inspired by Russ's. */ | |
1960 | |
1961 /* | |
1962 * Represents an NFA state plus zero or one or two arrows exiting. | |
1963 * if c == MATCH, no arrows out; matching state. | |
1964 * If c == SPLIT, unlabeled arrows to out and out1 (if != NULL). | |
1965 * If c < 256, labeled arrow with character c to out. | |
1966 */ | |
1967 | |
1968 static nfa_state_T *state_ptr; /* points to nfa_prog->state */ | |
1969 | |
1970 /* | |
1971 * Allocate and initialize nfa_state_T. | |
1972 */ | |
1973 static nfa_state_T * | |
1974 new_state(c, out, out1) | |
1975 int c; | |
1976 nfa_state_T *out; | |
1977 nfa_state_T *out1; | |
1978 { | |
1979 nfa_state_T *s; | |
1980 | |
1981 if (istate >= nstate) | |
1982 return NULL; | |
1983 | |
1984 s = &state_ptr[istate++]; | |
1985 | |
1986 s->c = c; | |
1987 s->out = out; | |
1988 s->out1 = out1; | |
1989 | |
1990 s->id = istate; | |
1991 s->lastlist = 0; | |
1992 s->negated = FALSE; | |
1993 | |
1994 return s; | |
1995 } | |
1996 | |
1997 /* | |
1998 * A partially built NFA without the matching state filled in. | |
1999 * Frag_T.start points at the start state. | |
2000 * Frag_T.out is a list of places that need to be set to the | |
2001 * next state for this fragment. | |
2002 */ | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2003 |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2004 /* 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
|
2005 * uninitialized, we use the pointers themselves |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2006 * as storage for the Ptrlists. */ |
4444 | 2007 typedef union Ptrlist Ptrlist; |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2008 union Ptrlist |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2009 { |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2010 Ptrlist *next; |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2011 nfa_state_T *s; |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2012 }; |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2013 |
4444 | 2014 struct Frag |
2015 { | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2016 nfa_state_T *start; |
4444 | 2017 Ptrlist *out; |
2018 }; | |
2019 typedef struct Frag Frag_T; | |
2020 | |
2021 static Frag_T frag __ARGS((nfa_state_T *start, Ptrlist *out)); | |
2022 static Ptrlist *list1 __ARGS((nfa_state_T **outp)); | |
2023 static void patch __ARGS((Ptrlist *l, nfa_state_T *s)); | |
2024 static Ptrlist *append __ARGS((Ptrlist *l1, Ptrlist *l2)); | |
2025 static void st_push __ARGS((Frag_T s, Frag_T **p, Frag_T *stack_end)); | |
2026 static Frag_T st_pop __ARGS((Frag_T **p, Frag_T *stack)); | |
2027 | |
2028 /* | |
4456 | 2029 * Initialize a Frag_T struct and return it. |
4444 | 2030 */ |
2031 static Frag_T | |
2032 frag(start, out) | |
2033 nfa_state_T *start; | |
2034 Ptrlist *out; | |
2035 { | |
4456 | 2036 Frag_T n; |
2037 | |
2038 n.start = start; | |
2039 n.out = out; | |
4444 | 2040 return n; |
2041 } | |
2042 | |
2043 /* | |
2044 * Create singleton list containing just outp. | |
2045 */ | |
2046 static Ptrlist * | |
2047 list1(outp) | |
2048 nfa_state_T **outp; | |
2049 { | |
2050 Ptrlist *l; | |
2051 | |
2052 l = (Ptrlist *)outp; | |
2053 l->next = NULL; | |
2054 return l; | |
2055 } | |
2056 | |
2057 /* | |
2058 * Patch the list of states at out to point to start. | |
2059 */ | |
2060 static void | |
2061 patch(l, s) | |
2062 Ptrlist *l; | |
2063 nfa_state_T *s; | |
2064 { | |
2065 Ptrlist *next; | |
2066 | |
2067 for (; l; l = next) | |
2068 { | |
2069 next = l->next; | |
2070 l->s = s; | |
2071 } | |
2072 } | |
2073 | |
2074 | |
2075 /* | |
2076 * Join the two lists l1 and l2, returning the combination. | |
2077 */ | |
2078 static Ptrlist * | |
2079 append(l1, l2) | |
2080 Ptrlist *l1; | |
2081 Ptrlist *l2; | |
2082 { | |
2083 Ptrlist *oldl1; | |
2084 | |
2085 oldl1 = l1; | |
2086 while (l1->next) | |
2087 l1 = l1->next; | |
2088 l1->next = l2; | |
2089 return oldl1; | |
2090 } | |
2091 | |
2092 /* | |
2093 * Stack used for transforming postfix form into NFA. | |
2094 */ | |
2095 static Frag_T empty; | |
2096 | |
2097 static void | |
2098 st_error(postfix, end, p) | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2099 int *postfix UNUSED; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2100 int *end UNUSED; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2101 int *p UNUSED; |
4444 | 2102 { |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2103 #ifdef NFA_REGEXP_ERROR_LOG |
4444 | 2104 FILE *df; |
2105 int *p2; | |
2106 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2107 df = fopen(NFA_REGEXP_ERROR_LOG, "a"); |
4444 | 2108 if (df) |
2109 { | |
2110 fprintf(df, "Error popping the stack!\n"); | |
2111 #ifdef DEBUG | |
2112 fprintf(df, "Current regexp is \"%s\"\n", nfa_regengine.expr); | |
2113 #endif | |
2114 fprintf(df, "Postfix form is: "); | |
2115 #ifdef DEBUG | |
2116 for (p2 = postfix; p2 < end; p2++) | |
2117 { | |
2118 nfa_set_code(*p2); | |
2119 fprintf(df, "%s, ", code); | |
2120 } | |
2121 nfa_set_code(*p); | |
2122 fprintf(df, "\nCurrent position is: "); | |
2123 for (p2 = postfix; p2 <= p; p2 ++) | |
2124 { | |
2125 nfa_set_code(*p2); | |
2126 fprintf(df, "%s, ", code); | |
2127 } | |
2128 #else | |
2129 for (p2 = postfix; p2 < end; p2++) | |
2130 { | |
2131 fprintf(df, "%d, ", *p2); | |
2132 } | |
2133 fprintf(df, "\nCurrent position is: "); | |
2134 for (p2 = postfix; p2 <= p; p2 ++) | |
2135 { | |
2136 fprintf(df, "%d, ", *p2); | |
2137 } | |
2138 #endif | |
2139 fprintf(df, "\n--------------------------\n"); | |
2140 fclose(df); | |
2141 } | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2142 #endif |
4444 | 2143 EMSG(_("E874: (NFA) Could not pop the stack !")); |
2144 } | |
2145 | |
2146 /* | |
2147 * Push an item onto the stack. | |
2148 */ | |
2149 static void | |
2150 st_push(s, p, stack_end) | |
2151 Frag_T s; | |
2152 Frag_T **p; | |
2153 Frag_T *stack_end; | |
2154 { | |
2155 Frag_T *stackp = *p; | |
2156 | |
2157 if (stackp >= stack_end) | |
2158 return; | |
2159 *stackp = s; | |
2160 *p = *p + 1; | |
2161 } | |
2162 | |
2163 /* | |
2164 * Pop an item from the stack. | |
2165 */ | |
2166 static Frag_T | |
2167 st_pop(p, stack) | |
2168 Frag_T **p; | |
2169 Frag_T *stack; | |
2170 { | |
2171 Frag_T *stackp; | |
2172 | |
2173 *p = *p - 1; | |
2174 stackp = *p; | |
2175 if (stackp < stack) | |
2176 return empty; | |
2177 return **p; | |
2178 } | |
2179 | |
2180 /* | |
2181 * Convert a postfix form into its equivalent NFA. | |
2182 * Return the NFA start state on success, NULL otherwise. | |
2183 */ | |
2184 static nfa_state_T * | |
2185 post2nfa(postfix, end, nfa_calc_size) | |
2186 int *postfix; | |
2187 int *end; | |
2188 int nfa_calc_size; | |
2189 { | |
2190 int *p; | |
2191 int mopen; | |
2192 int mclose; | |
2193 Frag_T *stack = NULL; | |
2194 Frag_T *stackp = NULL; | |
2195 Frag_T *stack_end = NULL; | |
2196 Frag_T e1; | |
2197 Frag_T e2; | |
2198 Frag_T e; | |
2199 nfa_state_T *s; | |
2200 nfa_state_T *s1; | |
2201 nfa_state_T *matchstate; | |
4484 | 2202 nfa_state_T *ret = NULL; |
4444 | 2203 |
2204 if (postfix == NULL) | |
2205 return NULL; | |
2206 | |
4456 | 2207 #define PUSH(s) st_push((s), &stackp, stack_end) |
4444 | 2208 #define POP() st_pop(&stackp, stack); \ |
2209 if (stackp < stack) \ | |
2210 { \ | |
2211 st_error(postfix, end, p); \ | |
2212 return NULL; \ | |
2213 } | |
2214 | |
2215 if (nfa_calc_size == FALSE) | |
2216 { | |
2217 /* Allocate space for the stack. Max states on the stack : nstate */ | |
4462 | 2218 stack = (Frag_T *) lalloc((nstate + 1) * sizeof(Frag_T), TRUE); |
4444 | 2219 stackp = stack; |
4462 | 2220 stack_end = stack + (nstate + 1); |
4444 | 2221 } |
2222 | |
2223 for (p = postfix; p < end; ++p) | |
2224 { | |
2225 switch (*p) | |
2226 { | |
2227 case NFA_CONCAT: | |
2228 /* Catenation. | |
2229 * Pay attention: this operator does not exist | |
2230 * in the r.e. itself (it is implicit, really). | |
2231 * It is added when r.e. is translated to postfix | |
2232 * form in re2post(). | |
2233 * | |
2234 * No new state added here. */ | |
2235 if (nfa_calc_size == TRUE) | |
2236 { | |
4458 | 2237 /* nstate += 0; */ |
4444 | 2238 break; |
2239 } | |
2240 e2 = POP(); | |
2241 e1 = POP(); | |
2242 patch(e1.out, e2.start); | |
2243 PUSH(frag(e1.start, e2.out)); | |
2244 break; | |
2245 | |
2246 case NFA_NOT: | |
2247 /* Negation of a character */ | |
2248 if (nfa_calc_size == TRUE) | |
2249 { | |
4458 | 2250 /* nstate += 0; */ |
4444 | 2251 break; |
2252 } | |
2253 e1 = POP(); | |
2254 e1.start->negated = TRUE; | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2255 #ifdef FEAT_MBYTE |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2256 if (e1.start->c == NFA_COMPOSING) |
4444 | 2257 e1.start->out1->negated = TRUE; |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2258 #endif |
4444 | 2259 PUSH(e1); |
2260 break; | |
2261 | |
2262 case NFA_OR: | |
2263 /* Alternation */ | |
2264 if (nfa_calc_size == TRUE) | |
2265 { | |
4458 | 2266 nstate++; |
4444 | 2267 break; |
2268 } | |
2269 e2 = POP(); | |
2270 e1 = POP(); | |
2271 s = new_state(NFA_SPLIT, e1.start, e2.start); | |
2272 if (s == NULL) | |
4484 | 2273 goto theend; |
4444 | 2274 PUSH(frag(s, append(e1.out, e2.out))); |
2275 break; | |
2276 | |
2277 case NFA_STAR: | |
2278 /* Zero or more */ | |
2279 if (nfa_calc_size == TRUE) | |
2280 { | |
4458 | 2281 nstate++; |
4444 | 2282 break; |
2283 } | |
2284 e = POP(); | |
2285 s = new_state(NFA_SPLIT, e.start, NULL); | |
2286 if (s == NULL) | |
4484 | 2287 goto theend; |
4444 | 2288 patch(e.out, s); |
2289 PUSH(frag(s, list1(&s->out1))); | |
2290 break; | |
2291 | |
2292 case NFA_QUEST: | |
2293 /* one or zero atoms=> greedy match */ | |
2294 if (nfa_calc_size == TRUE) | |
2295 { | |
4458 | 2296 nstate++; |
4444 | 2297 break; |
2298 } | |
2299 e = POP(); | |
2300 s = new_state(NFA_SPLIT, e.start, NULL); | |
2301 if (s == NULL) | |
4484 | 2302 goto theend; |
4444 | 2303 PUSH(frag(s, append(e.out, list1(&s->out1)))); |
2304 break; | |
2305 | |
2306 case NFA_QUEST_NONGREEDY: | |
2307 /* zero or one atoms => non-greedy match */ | |
2308 if (nfa_calc_size == TRUE) | |
2309 { | |
4458 | 2310 nstate++; |
4444 | 2311 break; |
2312 } | |
2313 e = POP(); | |
2314 s = new_state(NFA_SPLIT, NULL, e.start); | |
2315 if (s == NULL) | |
4484 | 2316 goto theend; |
4444 | 2317 PUSH(frag(s, append(e.out, list1(&s->out)))); |
2318 break; | |
2319 | |
2320 case NFA_PLUS: | |
2321 /* One or more */ | |
2322 if (nfa_calc_size == TRUE) | |
2323 { | |
4458 | 2324 nstate++; |
4444 | 2325 break; |
2326 } | |
2327 e = POP(); | |
2328 s = new_state(NFA_SPLIT, e.start, NULL); | |
2329 if (s == NULL) | |
4484 | 2330 goto theend; |
4444 | 2331 patch(e.out, s); |
2332 PUSH(frag(e.start, list1(&s->out1))); | |
2333 break; | |
2334 | |
2335 case NFA_SKIP_CHAR: | |
2336 /* Symbol of 0-length, Used in a repetition | |
2337 * with max/min count of 0 */ | |
2338 if (nfa_calc_size == TRUE) | |
2339 { | |
4458 | 2340 nstate++; |
4444 | 2341 break; |
2342 } | |
2343 s = new_state(NFA_SKIP_CHAR, NULL, NULL); | |
2344 if (s == NULL) | |
4484 | 2345 goto theend; |
4444 | 2346 PUSH(frag(s, list1(&s->out))); |
2347 break; | |
2348 | |
2349 case NFA_PREV_ATOM_NO_WIDTH: | |
2350 /* The \@= operator: match the preceding atom with 0 width. | |
2351 * Surrounds the preceding atom with START_INVISIBLE and | |
2352 * END_INVISIBLE, similarly to MOPEN. | |
2353 */ | |
2354 /* TODO: Maybe this drops the speed? */ | |
4484 | 2355 goto theend; |
4444 | 2356 |
2357 if (nfa_calc_size == TRUE) | |
2358 { | |
2359 nstate += 2; | |
2360 break; | |
2361 } | |
2362 e = POP(); | |
2363 s1 = new_state(NFA_END_INVISIBLE, NULL, NULL); | |
2364 if (s1 == NULL) | |
4484 | 2365 goto theend; |
4444 | 2366 patch(e.out, s1); |
2367 | |
2368 s = new_state(NFA_START_INVISIBLE, e.start, s1); | |
2369 if (s == NULL) | |
4484 | 2370 goto theend; |
4444 | 2371 PUSH(frag(s, list1(&s1->out))); |
2372 break; | |
2373 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2374 #ifdef FEAT_MBYTE |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2375 case NFA_COMPOSING: /* char with composing char */ |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2376 #if 0 |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2377 /* TODO */ |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2378 if (regflags & RF_ICOMBINE) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2379 { |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
2380 /* use the base character only */ |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2381 } |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2382 #endif |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2383 /* FALLTHROUGH */ |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2384 #endif |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2385 |
4444 | 2386 case NFA_MOPEN + 0: /* Submatch */ |
2387 case NFA_MOPEN + 1: | |
2388 case NFA_MOPEN + 2: | |
2389 case NFA_MOPEN + 3: | |
2390 case NFA_MOPEN + 4: | |
2391 case NFA_MOPEN + 5: | |
2392 case NFA_MOPEN + 6: | |
2393 case NFA_MOPEN + 7: | |
2394 case NFA_MOPEN + 8: | |
2395 case NFA_MOPEN + 9: | |
2396 case NFA_NOPEN: /* \%( "Invisible Submatch" */ | |
2397 if (nfa_calc_size == TRUE) | |
2398 { | |
2399 nstate += 2; | |
2400 break; | |
2401 } | |
2402 | |
2403 mopen = *p; | |
2404 switch (*p) | |
2405 { | |
2406 case NFA_NOPEN: | |
2407 mclose = NFA_NCLOSE; | |
2408 break; | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2409 #ifdef FEAT_MBYTE |
4444 | 2410 case NFA_COMPOSING: |
2411 mclose = NFA_END_COMPOSING; | |
2412 break; | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2413 #endif |
4444 | 2414 default: |
2415 /* NFA_MOPEN(0) ... NFA_MOPEN(9) */ | |
2416 mclose = *p + NSUBEXP; | |
2417 break; | |
2418 } | |
2419 | |
2420 /* Allow "NFA_MOPEN" as a valid postfix representation for | |
2421 * the empty regexp "". In this case, the NFA will be | |
2422 * NFA_MOPEN -> NFA_MCLOSE. Note that this also allows | |
2423 * empty groups of parenthesis, and empty mbyte chars */ | |
2424 if (stackp == stack) | |
2425 { | |
2426 s = new_state(mopen, NULL, NULL); | |
2427 if (s == NULL) | |
4484 | 2428 goto theend; |
4444 | 2429 s1 = new_state(mclose, NULL, NULL); |
2430 if (s1 == NULL) | |
4484 | 2431 goto theend; |
4444 | 2432 patch(list1(&s->out), s1); |
2433 PUSH(frag(s, list1(&s1->out))); | |
2434 break; | |
2435 } | |
2436 | |
2437 /* At least one node was emitted before NFA_MOPEN, so | |
2438 * at least one node will be between NFA_MOPEN and NFA_MCLOSE */ | |
2439 e = POP(); | |
2440 s = new_state(mopen, e.start, NULL); /* `(' */ | |
2441 if (s == NULL) | |
4484 | 2442 goto theend; |
4444 | 2443 |
2444 s1 = new_state(mclose, NULL, NULL); /* `)' */ | |
2445 if (s1 == NULL) | |
4484 | 2446 goto theend; |
4444 | 2447 patch(e.out, s1); |
2448 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2449 #ifdef FEAT_MBYTE |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2450 if (mopen == NFA_COMPOSING) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2451 /* COMPOSING->out1 = END_COMPOSING */ |
4444 | 2452 patch(list1(&s->out1), s1); |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2453 #endif |
4444 | 2454 |
2455 PUSH(frag(s, list1(&s1->out))); | |
2456 break; | |
2457 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2458 case NFA_BACKREF1: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2459 case NFA_BACKREF2: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2460 case NFA_BACKREF3: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2461 case NFA_BACKREF4: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2462 case NFA_BACKREF5: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2463 case NFA_BACKREF6: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2464 case NFA_BACKREF7: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2465 case NFA_BACKREF8: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2466 case NFA_BACKREF9: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2467 if (nfa_calc_size == TRUE) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2468 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2469 nstate += 2; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2470 break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2471 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2472 s = new_state(*p, NULL, NULL); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2473 if (s == NULL) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2474 goto theend; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2475 s1 = new_state(NFA_SKIP, NULL, NULL); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2476 if (s1 == NULL) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2477 goto theend; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2478 patch(list1(&s->out), s1); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2479 PUSH(frag(s, list1(&s1->out))); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2480 break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2481 |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2482 case NFA_LNUM: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2483 case NFA_LNUM_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2484 case NFA_LNUM_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2485 case NFA_VCOL: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2486 case NFA_VCOL_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2487 case NFA_VCOL_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2488 case NFA_COL: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2489 case NFA_COL_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2490 case NFA_COL_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2491 if (nfa_calc_size == TRUE) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2492 { |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2493 nstate += 1; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2494 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2495 } |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2496 e1 = POP(); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2497 s = new_state(*p, NULL, NULL); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2498 if (s == NULL) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2499 goto theend; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2500 s->val = e1.start->c; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2501 PUSH(frag(s, list1(&s->out))); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2502 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2503 |
4444 | 2504 case NFA_ZSTART: |
2505 case NFA_ZEND: | |
2506 default: | |
2507 /* Operands */ | |
2508 if (nfa_calc_size == TRUE) | |
2509 { | |
4458 | 2510 nstate++; |
4444 | 2511 break; |
2512 } | |
2513 s = new_state(*p, NULL, NULL); | |
2514 if (s == NULL) | |
4484 | 2515 goto theend; |
4444 | 2516 PUSH(frag(s, list1(&s->out))); |
2517 break; | |
2518 | |
2519 } /* switch(*p) */ | |
2520 | |
2521 } /* for(p = postfix; *p; ++p) */ | |
2522 | |
2523 if (nfa_calc_size == TRUE) | |
2524 { | |
4458 | 2525 nstate++; |
4484 | 2526 goto theend; /* Return value when counting size is ignored anyway */ |
4444 | 2527 } |
2528 | |
2529 e = POP(); | |
2530 if (stackp != stack) | |
2531 EMSG_RET_NULL(_("E875: (NFA regexp) (While converting from postfix to NFA), too many states left on stack")); | |
2532 | |
2533 if (istate >= nstate) | |
2534 EMSG_RET_NULL(_("E876: (NFA regexp) Not enough space to store the whole NFA ")); | |
2535 | |
2536 matchstate = &state_ptr[istate++]; /* the match state */ | |
2537 matchstate->c = NFA_MATCH; | |
2538 matchstate->out = matchstate->out1 = NULL; | |
2539 | |
2540 patch(e.out, matchstate); | |
4484 | 2541 ret = e.start; |
2542 | |
2543 theend: | |
2544 vim_free(stack); | |
2545 return ret; | |
4444 | 2546 |
2547 #undef POP1 | |
2548 #undef PUSH1 | |
2549 #undef POP2 | |
2550 #undef PUSH2 | |
2551 #undef POP | |
2552 #undef PUSH | |
2553 } | |
2554 | |
2555 /**************************************************************** | |
2556 * NFA execution code. | |
2557 ****************************************************************/ | |
2558 | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2559 typedef struct |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2560 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2561 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
|
2562 |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2563 /* 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
|
2564 union |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2565 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2566 struct multipos |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2567 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2568 lpos_T start; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2569 lpos_T end; |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2570 } multi[NSUBEXP]; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2571 struct linepos |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2572 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2573 char_u *start; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2574 char_u *end; |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2575 } line[NSUBEXP]; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2576 } list; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2577 } regsub_T; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2578 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2579 /* 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
|
2580 typedef struct |
4444 | 2581 { |
2582 nfa_state_T *state; | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2583 int count; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2584 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
|
2585 } nfa_thread_T; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2586 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2587 /* nfa_list_T contains the alternative NFA execution states. */ |
4444 | 2588 typedef struct |
2589 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2590 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
|
2591 int n; /* nr of states currently in "t" */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2592 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
|
2593 int id; /* ID of the list */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2594 } nfa_list_T; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2595 |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2596 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2597 static void |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2598 log_subexpr(sub) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2599 regsub_T *sub; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2600 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2601 int j; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2602 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2603 for (j = 0; j < sub->in_use; j++) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2604 if (REG_MULTI) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2605 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
|
2606 j, |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2607 sub->list.multi[j].start.col, |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2608 (int)sub->list.multi[j].start.lnum, |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2609 sub->list.multi[j].end.col, |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2610 (int)sub->list.multi[j].end.lnum); |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2611 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2612 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
|
2613 j, |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2614 (char *)sub->list.line[j].start, |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2615 (char *)sub->list.line[j].end); |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2616 fprintf(log_fd, "\n"); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2617 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2618 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2619 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2620 /* 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
|
2621 static int nfa_match; |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2622 |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2623 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
|
2624 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
|
2625 static void addstate_here __ARGS((nfa_list_T *l, nfa_state_T *state, regsub_T *sub, int *ip)); |
4444 | 2626 |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2627 /* |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2628 * 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
|
2629 */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2630 static int |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2631 sub_equal(sub1, sub2) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2632 regsub_T *sub1; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2633 regsub_T *sub2; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2634 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2635 int i; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2636 int todo; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2637 linenr_T s1, e1; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2638 linenr_T s2, e2; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2639 char_u *sp1, *ep1; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2640 char_u *sp2, *ep2; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2641 |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2642 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
|
2643 if (REG_MULTI) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2644 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2645 for (i = 0; i < todo; ++i) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2646 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2647 if (i < sub1->in_use) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2648 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2649 s1 = sub1->list.multi[i].start.lnum; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2650 e1 = sub1->list.multi[i].end.lnum; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2651 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2652 else |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2653 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2654 s1 = 0; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2655 e1 = 0; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2656 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2657 if (i < sub2->in_use) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2658 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2659 s2 = sub2->list.multi[i].start.lnum; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2660 e2 = sub2->list.multi[i].end.lnum; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2661 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2662 else |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2663 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2664 s2 = 0; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2665 e2 = 0; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2666 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2667 if (s1 != s2 || e1 != e2) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2668 return FALSE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2669 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
|
2670 != sub2->list.multi[i].start.col) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2671 return FALSE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2672 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
|
2673 != sub2->list.multi[i].end.col) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2674 return FALSE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2675 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2676 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2677 else |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2678 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2679 for (i = 0; i < todo; ++i) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2680 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2681 if (i < sub1->in_use) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2682 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2683 sp1 = sub1->list.line[i].start; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2684 ep1 = sub1->list.line[i].end; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2685 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2686 else |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2687 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2688 sp1 = NULL; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2689 ep1 = NULL; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2690 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2691 if (i < sub2->in_use) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2692 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2693 sp2 = sub2->list.line[i].start; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2694 ep2 = sub2->list.line[i].end; |
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 else |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2697 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2698 sp2 = NULL; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2699 ep2 = NULL; |
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 (sp1 != sp2 || ep1 != ep2) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2702 return FALSE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2703 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2704 } |
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 return TRUE; |
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 |
4444 | 2709 static void |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2710 addstate(l, state, sub, off) |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2711 nfa_list_T *l; /* runtime state list */ |
4444 | 2712 nfa_state_T *state; /* state to update */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2713 regsub_T *sub; /* pointers to subexpressions */ |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2714 int off; /* byte offset, when -1 go to next line */ |
4444 | 2715 { |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2716 int subidx; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2717 nfa_thread_T *thread; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2718 lpos_T save_lpos; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2719 int save_in_use; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2720 char_u *save_ptr; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2721 int i; |
4444 | 2722 |
2723 if (l == NULL || state == NULL) | |
2724 return; | |
2725 | |
2726 switch (state->c) | |
2727 { | |
2728 case NFA_SPLIT: | |
2729 case NFA_NOT: | |
2730 case NFA_NOPEN: | |
2731 case NFA_NCLOSE: | |
2732 case NFA_MCLOSE: | |
2733 case NFA_MCLOSE + 1: | |
2734 case NFA_MCLOSE + 2: | |
2735 case NFA_MCLOSE + 3: | |
2736 case NFA_MCLOSE + 4: | |
2737 case NFA_MCLOSE + 5: | |
2738 case NFA_MCLOSE + 6: | |
2739 case NFA_MCLOSE + 7: | |
2740 case NFA_MCLOSE + 8: | |
2741 case NFA_MCLOSE + 9: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2742 /* 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
|
2743 * "out1" may be added below. */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2744 break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2745 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2746 case NFA_MOPEN: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2747 case NFA_MOPEN + 1: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2748 case NFA_MOPEN + 2: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2749 case NFA_MOPEN + 3: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2750 case NFA_MOPEN + 4: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2751 case NFA_MOPEN + 5: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2752 case NFA_MOPEN + 6: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2753 case NFA_MOPEN + 7: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2754 case NFA_MOPEN + 8: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2755 case NFA_MOPEN + 9: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2756 /* 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
|
2757 * 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
|
2758 if (state->lastlist == l->id) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2759 return; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2760 state->lastlist = l->id; |
4444 | 2761 break; |
2762 | |
2763 default: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2764 if (state->lastlist == l->id) |
4444 | 2765 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2766 /* 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
|
2767 * 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
|
2768 if (!nfa_has_backref) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2769 return; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2770 |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2771 /* 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
|
2772 * positions. */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2773 for (i = 0; i < l->n; ++i) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2774 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2775 thread = &l->t[i]; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2776 if (thread->state->id == state->id |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2777 && sub_equal(&thread->sub, sub)) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2778 return; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2779 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2780 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2781 |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2782 /* 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
|
2783 * lot) bigger */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2784 if (nfa_has_backref && l->n == l->len) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2785 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2786 int newlen = l->len * 3 / 2 + 50; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2787 |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2788 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
|
2789 l->len = newlen; |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2790 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2791 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2792 /* add the state to the list */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2793 state->lastlist = l->id; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2794 thread = &l->t[l->n++]; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2795 thread->state = state; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2796 thread->sub.in_use = sub->in_use; |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2797 if (sub->in_use > 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2798 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2799 /* Copy the match start and end positions. */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2800 if (REG_MULTI) |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2801 mch_memmove(&thread->sub.list.multi[0], |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2802 &sub->list.multi[0], |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2803 sizeof(struct multipos) * sub->in_use); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2804 else |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2805 mch_memmove(&thread->sub.list.line[0], |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2806 &sub->list.line[0], |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2807 sizeof(struct linepos) * sub->in_use); |
4444 | 2808 } |
2809 } | |
2810 | |
2811 #ifdef ENABLE_LOG | |
2812 nfa_set_code(state->c); | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2813 fprintf(log_fd, "> Adding state %d to list. Character %d: %s\n", |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2814 abs(state->id), state->c, code); |
4444 | 2815 #endif |
2816 switch (state->c) | |
2817 { | |
2818 case NFA_MATCH: | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2819 nfa_match = TRUE; |
4444 | 2820 break; |
2821 | |
2822 case NFA_SPLIT: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2823 addstate(l, state->out, sub, off); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2824 addstate(l, state->out1, sub, off); |
4444 | 2825 break; |
2826 | |
2827 #if 0 | |
2828 case NFA_END_NEG_RANGE: | |
2829 /* Nothing to handle here. nfa_regmatch() will take care of it */ | |
2830 break; | |
2831 | |
2832 case NFA_NOT: | |
2833 EMSG(_("E999: (NFA regexp internal error) Should not process NOT node !")); | |
2834 #ifdef ENABLE_LOG | |
2835 fprintf(f, "\n\n>>> E999: Added state NFA_NOT to a list ... Something went wrong ! Why wasn't it processed already? \n\n"); | |
2836 #endif | |
2837 break; | |
2838 | |
2839 case NFA_COMPOSING: | |
2840 /* nfa_regmatch() will match all the bytes of this composing char. */ | |
2841 break; | |
2842 #endif | |
2843 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2844 case NFA_SKIP_CHAR: |
4444 | 2845 case NFA_NOPEN: |
2846 case NFA_NCLOSE: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2847 addstate(l, state->out, sub, off); |
4444 | 2848 break; |
2849 | |
2850 /* If this state is reached, then a recursive call of nfa_regmatch() | |
2851 * succeeded. the next call saves the found submatches in the | |
2852 * first state after the "invisible" branch. */ | |
2853 #if 0 | |
2854 case NFA_END_INVISIBLE: | |
2855 break; | |
2856 #endif | |
2857 | |
2858 case NFA_MOPEN + 0: | |
2859 case NFA_MOPEN + 1: | |
2860 case NFA_MOPEN + 2: | |
2861 case NFA_MOPEN + 3: | |
2862 case NFA_MOPEN + 4: | |
2863 case NFA_MOPEN + 5: | |
2864 case NFA_MOPEN + 6: | |
2865 case NFA_MOPEN + 7: | |
2866 case NFA_MOPEN + 8: | |
2867 case NFA_MOPEN + 9: | |
2868 case NFA_ZSTART: | |
2869 if (state->c == NFA_ZSTART) | |
2870 subidx = 0; | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2871 else |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2872 subidx = state->c - NFA_MOPEN; |
4444 | 2873 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2874 /* 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
|
2875 * restore it when it was in use. Otherwise fill any gap. */ |
4444 | 2876 if (REG_MULTI) |
2877 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2878 if (subidx < sub->in_use) |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2879 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2880 save_lpos = sub->list.multi[subidx].start; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2881 save_in_use = -1; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2882 } |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2883 else |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2884 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2885 save_in_use = sub->in_use; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2886 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
|
2887 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2888 sub->list.multi[i].start.lnum = -1; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2889 sub->list.multi[i].end.lnum = -1; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2890 } |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2891 sub->in_use = subidx + 1; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2892 } |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2893 if (off == -1) |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2894 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2895 sub->list.multi[subidx].start.lnum = reglnum + 1; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2896 sub->list.multi[subidx].start.col = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2897 } |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2898 else |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2899 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2900 sub->list.multi[subidx].start.lnum = reglnum; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2901 sub->list.multi[subidx].start.col = |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2902 (colnr_T)(reginput - regline + off); |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2903 } |
4444 | 2904 } |
2905 else | |
2906 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2907 if (subidx < sub->in_use) |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2908 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2909 save_ptr = sub->list.line[subidx].start; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2910 save_in_use = -1; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2911 } |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2912 else |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2913 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2914 save_in_use = sub->in_use; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2915 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
|
2916 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2917 sub->list.line[i].start = NULL; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2918 sub->list.line[i].end = NULL; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2919 } |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2920 sub->in_use = subidx + 1; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2921 } |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2922 sub->list.line[subidx].start = reginput + off; |
4444 | 2923 } |
2924 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2925 addstate(l, state->out, sub, off); |
4444 | 2926 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2927 if (save_in_use == -1) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2928 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2929 if (REG_MULTI) |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2930 sub->list.multi[subidx].start = save_lpos; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2931 else |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2932 sub->list.line[subidx].start = save_ptr; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2933 } |
4444 | 2934 else |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2935 sub->in_use = save_in_use; |
4444 | 2936 break; |
2937 | |
2938 case NFA_MCLOSE + 0: | |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
2939 if (nfa_has_zend) |
4444 | 2940 { |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
2941 /* 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
|
2942 * 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
|
2943 addstate(l, state->out, sub, off); |
4444 | 2944 break; |
2945 } | |
2946 case NFA_MCLOSE + 1: | |
2947 case NFA_MCLOSE + 2: | |
2948 case NFA_MCLOSE + 3: | |
2949 case NFA_MCLOSE + 4: | |
2950 case NFA_MCLOSE + 5: | |
2951 case NFA_MCLOSE + 6: | |
2952 case NFA_MCLOSE + 7: | |
2953 case NFA_MCLOSE + 8: | |
2954 case NFA_MCLOSE + 9: | |
2955 case NFA_ZEND: | |
2956 if (state->c == NFA_ZEND) | |
2957 subidx = 0; | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2958 else |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2959 subidx = state->c - NFA_MCLOSE; |
4444 | 2960 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2961 /* 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
|
2962 * has done that. */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2963 save_in_use = sub->in_use; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2964 if (sub->in_use <= subidx) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2965 sub->in_use = subidx + 1; |
4444 | 2966 if (REG_MULTI) |
2967 { | |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2968 save_lpos = sub->list.multi[subidx].end; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2969 if (off == -1) |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2970 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2971 sub->list.multi[subidx].end.lnum = reglnum + 1; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2972 sub->list.multi[subidx].end.col = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2973 } |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2974 else |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2975 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2976 sub->list.multi[subidx].end.lnum = reglnum; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2977 sub->list.multi[subidx].end.col = |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2978 (colnr_T)(reginput - regline + off); |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2979 } |
4444 | 2980 } |
2981 else | |
2982 { | |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2983 save_ptr = sub->list.line[subidx].end; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2984 sub->list.line[subidx].end = reginput + off; |
4444 | 2985 } |
2986 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2987 addstate(l, state->out, sub, off); |
4444 | 2988 |
2989 if (REG_MULTI) | |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2990 sub->list.multi[subidx].end = save_lpos; |
4444 | 2991 else |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2992 sub->list.line[subidx].end = save_ptr; |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2993 sub->in_use = save_in_use; |
4444 | 2994 break; |
2995 } | |
2996 } | |
2997 | |
2998 /* | |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2999 * 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
|
3000 * 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
|
3001 * 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
|
3002 * matters for alternatives. |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3003 */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3004 static void |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3005 addstate_here(l, state, sub, ip) |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3006 nfa_list_T *l; /* runtime state list */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3007 nfa_state_T *state; /* state to update */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3008 regsub_T *sub; /* pointers to subexpressions */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3009 int *ip; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3010 { |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3011 int tlen = l->n; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3012 int count; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3013 int i = *ip; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3014 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3015 /* 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
|
3016 addstate(l, state, sub, 0); |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3017 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3018 /* 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
|
3019 if (i + 1 == tlen) |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3020 return; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3021 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3022 /* 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
|
3023 count = l->n - tlen; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3024 if (count == 1) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3025 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3026 /* overwrite the current state */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3027 l->t[i] = l->t[l->n - 1]; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3028 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3029 else if (count > 1) |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3030 { |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3031 /* 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
|
3032 * end to the current position */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3033 mch_memmove(&(l->t[i + count]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3034 &(l->t[i + 1]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3035 sizeof(nfa_thread_T) * (l->n - i - 1)); |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3036 mch_memmove(&(l->t[i]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3037 &(l->t[l->n - 1]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3038 sizeof(nfa_thread_T) * count); |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3039 } |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3040 --l->n; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3041 *ip = i - 1; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3042 } |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3043 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3044 /* |
4444 | 3045 * Check character class "class" against current character c. |
3046 */ | |
3047 static int | |
3048 check_char_class(class, c) | |
3049 int class; | |
3050 int c; | |
3051 { | |
3052 switch (class) | |
3053 { | |
3054 case NFA_CLASS_ALNUM: | |
4464 | 3055 if (c >= 1 && c <= 255 && isalnum(c)) |
4444 | 3056 return OK; |
3057 break; | |
3058 case NFA_CLASS_ALPHA: | |
4464 | 3059 if (c >= 1 && c <= 255 && isalpha(c)) |
4444 | 3060 return OK; |
3061 break; | |
3062 case NFA_CLASS_BLANK: | |
3063 if (c == ' ' || c == '\t') | |
3064 return OK; | |
3065 break; | |
3066 case NFA_CLASS_CNTRL: | |
4464 | 3067 if (c >= 1 && c <= 255 && iscntrl(c)) |
4444 | 3068 return OK; |
3069 break; | |
3070 case NFA_CLASS_DIGIT: | |
3071 if (VIM_ISDIGIT(c)) | |
3072 return OK; | |
3073 break; | |
3074 case NFA_CLASS_GRAPH: | |
4464 | 3075 if (c >= 1 && c <= 255 && isgraph(c)) |
4444 | 3076 return OK; |
3077 break; | |
3078 case NFA_CLASS_LOWER: | |
3079 if (MB_ISLOWER(c)) | |
3080 return OK; | |
3081 break; | |
3082 case NFA_CLASS_PRINT: | |
3083 if (vim_isprintc(c)) | |
3084 return OK; | |
3085 break; | |
3086 case NFA_CLASS_PUNCT: | |
4464 | 3087 if (c >= 1 && c <= 255 && ispunct(c)) |
4444 | 3088 return OK; |
3089 break; | |
3090 case NFA_CLASS_SPACE: | |
3091 if ((c >=9 && c <= 13) || (c == ' ')) | |
3092 return OK; | |
3093 break; | |
3094 case NFA_CLASS_UPPER: | |
3095 if (MB_ISUPPER(c)) | |
3096 return OK; | |
3097 break; | |
3098 case NFA_CLASS_XDIGIT: | |
3099 if (vim_isxdigit(c)) | |
3100 return OK; | |
3101 break; | |
3102 case NFA_CLASS_TAB: | |
3103 if (c == '\t') | |
3104 return OK; | |
3105 break; | |
3106 case NFA_CLASS_RETURN: | |
3107 if (c == '\r') | |
3108 return OK; | |
3109 break; | |
3110 case NFA_CLASS_BACKSPACE: | |
3111 if (c == '\b') | |
3112 return OK; | |
3113 break; | |
3114 case NFA_CLASS_ESCAPE: | |
3115 if (c == '\033') | |
3116 return OK; | |
3117 break; | |
3118 | |
3119 default: | |
3120 /* should not be here :P */ | |
3121 EMSG_RET_FAIL(_("E877: (NFA regexp) Invalid character class ")); | |
3122 } | |
3123 return FAIL; | |
3124 } | |
3125 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3126 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
|
3127 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3128 /* |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3129 * Check for a match with subexpression "subidx". |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3130 * return TRUE if it matches. |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3131 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3132 static int |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3133 match_backref(sub, subidx, bytelen) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3134 regsub_T *sub; /* pointers to subexpressions */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3135 int subidx; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3136 int *bytelen; /* out: length of match in bytes */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3137 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3138 int len; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3139 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3140 if (sub->in_use <= subidx) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3141 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3142 retempty: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3143 /* backref was not set, match an empty string */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3144 *bytelen = 0; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3145 return TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3146 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3147 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3148 if (REG_MULTI) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3149 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3150 if (sub->list.multi[subidx].start.lnum < 0 |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3151 || sub->list.multi[subidx].end.lnum < 0) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3152 goto retempty; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3153 /* TODO: line breaks */ |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3154 len = sub->list.multi[subidx].end.col |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3155 - sub->list.multi[subidx].start.col; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3156 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
|
3157 reginput, &len) == 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3158 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3159 *bytelen = len; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3160 return TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3161 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3162 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3163 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3164 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3165 if (sub->list.line[subidx].start == NULL |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3166 || sub->list.line[subidx].end == NULL) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3167 goto retempty; |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3168 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
|
3169 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
|
3170 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3171 *bytelen = len; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3172 return TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3173 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3174 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3175 return FALSE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3176 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3177 |
4444 | 3178 /* |
3179 * Set all NFA nodes' list ID equal to -1. | |
3180 */ | |
3181 static void | |
3182 nfa_set_neg_listids(start) | |
3183 nfa_state_T *start; | |
3184 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3185 if (start != NULL && start->lastlist >= 0) |
4444 | 3186 { |
3187 start->lastlist = -1; | |
3188 nfa_set_neg_listids(start->out); | |
3189 nfa_set_neg_listids(start->out1); | |
3190 } | |
3191 } | |
3192 | |
3193 /* | |
3194 * Set all NFA nodes' list ID equal to 0. | |
3195 */ | |
3196 static void | |
3197 nfa_set_null_listids(start) | |
3198 nfa_state_T *start; | |
3199 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3200 if (start != NULL && start->lastlist == -1) |
4444 | 3201 { |
3202 start->lastlist = 0; | |
3203 nfa_set_null_listids(start->out); | |
3204 nfa_set_null_listids(start->out1); | |
3205 } | |
3206 } | |
3207 | |
3208 /* | |
3209 * Save list IDs for all NFA states in "list". | |
3210 */ | |
3211 static void | |
3212 nfa_save_listids(start, list) | |
3213 nfa_state_T *start; | |
3214 int *list; | |
3215 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3216 if (start != NULL && start->lastlist != -1) |
4444 | 3217 { |
3218 list[abs(start->id)] = start->lastlist; | |
3219 start->lastlist = -1; | |
3220 nfa_save_listids(start->out, list); | |
3221 nfa_save_listids(start->out1, list); | |
3222 } | |
3223 } | |
3224 | |
3225 /* | |
3226 * Restore list IDs from "list" to all NFA states. | |
3227 */ | |
3228 static void | |
3229 nfa_restore_listids(start, list) | |
3230 nfa_state_T *start; | |
3231 int *list; | |
3232 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3233 if (start != NULL && start->lastlist == -1) |
4444 | 3234 { |
3235 start->lastlist = list[abs(start->id)]; | |
3236 nfa_restore_listids(start->out, list); | |
3237 nfa_restore_listids(start->out1, list); | |
3238 } | |
3239 } | |
3240 | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3241 static int |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3242 nfa_re_num_cmp(val, op, pos) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3243 long_u val; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3244 int op; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3245 long_u pos; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3246 { |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3247 if (op == 1) return pos > val; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3248 if (op == 2) return pos < val; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3249 return val == pos; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3250 } |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3251 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3252 static int nfa_regmatch __ARGS((nfa_state_T *start, regsub_T *submatch, regsub_T *m)); |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3253 |
4444 | 3254 /* |
3255 * Main matching routine. | |
3256 * | |
3257 * Run NFA to determine whether it matches reginput. | |
3258 * | |
3259 * Return TRUE if there is a match, FALSE otherwise. | |
3260 * Note: Caller must ensure that: start != NULL. | |
3261 */ | |
3262 static int | |
3263 nfa_regmatch(start, submatch, m) | |
3264 nfa_state_T *start; | |
3265 regsub_T *submatch; | |
3266 regsub_T *m; | |
3267 { | |
3268 int result; | |
3269 int size = 0; | |
3270 int flag = 0; | |
3271 int old_reglnum = -1; | |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3272 int go_to_nextline = FALSE; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3273 nfa_thread_T *t; |
4444 | 3274 char_u *old_reginput = NULL; |
3275 char_u *old_regline = NULL; | |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3276 nfa_list_T list[3]; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3277 nfa_list_T *listtbl[2][2]; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3278 nfa_list_T *ll; |
4444 | 3279 int listid = 1; |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3280 int listidx; |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3281 nfa_list_T *thislist; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3282 nfa_list_T *nextlist; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3283 nfa_list_T *neglist; |
4444 | 3284 int *listids = NULL; |
4460 | 3285 #ifdef NFA_REGEXP_DEBUG_LOG |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3286 FILE *debug = fopen(NFA_REGEXP_DEBUG_LOG, "a"); |
4444 | 3287 |
3288 if (debug == NULL) | |
3289 { | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3290 EMSG2(_("(NFA) COULD NOT OPEN %s !"), NFA_REGEXP_DEBUG_LOG); |
4444 | 3291 return FALSE; |
3292 } | |
3293 #endif | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3294 nfa_match = FALSE; |
4444 | 3295 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3296 /* Allocate memory for the lists of nodes. */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3297 size = (nstate + 1) * sizeof(nfa_thread_T); |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3298 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
|
3299 list[0].len = nstate + 1; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3300 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
|
3301 list[1].len = nstate + 1; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3302 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
|
3303 list[2].len = nstate + 1; |
4444 | 3304 if (list[0].t == NULL || list[1].t == NULL || list[2].t == NULL) |
3305 goto theend; | |
3306 | |
3307 #ifdef ENABLE_LOG | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3308 log_fd = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 3309 if (log_fd != NULL) |
3310 { | |
3311 fprintf(log_fd, "**********************************\n"); | |
3312 nfa_set_code(start->c); | |
3313 fprintf(log_fd, " RUNNING nfa_regmatch() starting with state %d, code %s\n", | |
3314 abs(start->id), code); | |
3315 fprintf(log_fd, "**********************************\n"); | |
3316 } | |
3317 else | |
3318 { | |
3319 EMSG(_("Could not open temporary log file for writing, displaying on stderr ... ")); | |
3320 log_fd = stderr; | |
3321 } | |
3322 #endif | |
3323 | |
3324 thislist = &list[0]; | |
3325 thislist->n = 0; | |
3326 nextlist = &list[1]; | |
3327 nextlist->n = 0; | |
3328 neglist = &list[2]; | |
3329 neglist->n = 0; | |
3330 #ifdef ENABLE_LOG | |
3331 fprintf(log_fd, "(---) STARTSTATE\n"); | |
3332 #endif | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3333 thislist->id = listid; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3334 addstate(thislist, start, m, 0); |
4444 | 3335 |
3336 /* There are two cases when the NFA advances: 1. input char matches the | |
3337 * NFA node and 2. input char does not match the NFA node, but the next | |
3338 * node is NFA_NOT. The following macro calls addstate() according to | |
3339 * these rules. It is used A LOT, so use the "listtbl" table for speed */ | |
3340 listtbl[0][0] = NULL; | |
3341 listtbl[0][1] = neglist; | |
3342 listtbl[1][0] = nextlist; | |
3343 listtbl[1][1] = NULL; | |
3344 #define ADD_POS_NEG_STATE(node) \ | |
3345 ll = listtbl[result ? 1 : 0][node->negated]; \ | |
3346 if (ll != NULL) \ | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3347 addstate(ll, node->out , &t->sub, clen); |
4444 | 3348 |
3349 | |
3350 /* | |
3351 * Run for each character. | |
3352 */ | |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3353 for (;;) |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3354 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3355 int curc; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3356 int clen; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3357 |
4444 | 3358 #ifdef FEAT_MBYTE |
3359 if (has_mbyte) | |
3360 { | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3361 curc = (*mb_ptr2char)(reginput); |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3362 clen = (*mb_ptr2len)(reginput); |
4444 | 3363 } |
3364 else | |
3365 #endif | |
3366 { | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3367 curc = *reginput; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3368 clen = 1; |
4444 | 3369 } |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3370 if (curc == NUL) |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3371 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3372 clen = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3373 go_to_nextline = FALSE; |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3374 } |
4444 | 3375 |
3376 /* swap lists */ | |
3377 thislist = &list[flag]; | |
3378 nextlist = &list[flag ^= 1]; | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3379 nextlist->n = 0; /* clear nextlist */ |
4444 | 3380 listtbl[1][0] = nextlist; |
3381 ++listid; | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3382 thislist->id = listid; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3383 nextlist->id = listid + 1; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3384 neglist->id = listid + 1; |
4444 | 3385 |
3386 #ifdef ENABLE_LOG | |
3387 fprintf(log_fd, "------------------------------------------\n"); | |
3388 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
|
3389 fprintf(log_fd, ">>> Advanced one character ... Current char is %c (code %d) \n", curc, (int)curc); |
4444 | 3390 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
|
3391 { |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3392 int i; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3393 |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3394 for (i = 0; i < thislist->n; i++) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3395 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
|
3396 } |
4444 | 3397 fprintf(log_fd, "\n"); |
3398 #endif | |
3399 | |
4460 | 3400 #ifdef NFA_REGEXP_DEBUG_LOG |
4444 | 3401 fprintf(debug, "\n-------------------\n"); |
3402 #endif | |
4480 | 3403 /* |
3404 * If the state lists are empty we can stop. | |
3405 */ | |
3406 if (thislist->n == 0 && neglist->n == 0) | |
3407 break; | |
4444 | 3408 |
3409 /* compute nextlist */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3410 for (listidx = 0; listidx < thislist->n || neglist->n > 0; ++listidx) |
4444 | 3411 { |
3412 if (neglist->n > 0) | |
3413 { | |
3414 t = &neglist->t[0]; | |
4468 | 3415 neglist->n--; |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3416 listidx--; |
4444 | 3417 } |
3418 else | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3419 t = &thislist->t[listidx]; |
4444 | 3420 |
4460 | 3421 #ifdef NFA_REGEXP_DEBUG_LOG |
4444 | 3422 nfa_set_code(t->state->c); |
3423 fprintf(debug, "%s, ", code); | |
3424 #endif | |
3425 #ifdef ENABLE_LOG | |
3426 nfa_set_code(t->state->c); | |
3427 fprintf(log_fd, "(%d) %s, code %d ... \n", abs(t->state->id), | |
3428 code, (int)t->state->c); | |
3429 #endif | |
3430 | |
3431 /* | |
3432 * Handle the possible codes of the current state. | |
3433 * The most important is NFA_MATCH. | |
3434 */ | |
3435 switch (t->state->c) | |
3436 { | |
3437 case NFA_MATCH: | |
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3438 { |
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3439 int j; |
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3440 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3441 nfa_match = TRUE; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3442 submatch->in_use = t->sub.in_use; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3443 if (REG_MULTI) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3444 for (j = 0; j < submatch->in_use; j++) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3445 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3446 submatch->list.multi[j].start = |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3447 t->sub.list.multi[j].start; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3448 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
|
3449 } |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3450 else |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3451 for (j = 0; j < submatch->in_use; j++) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3452 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3453 submatch->list.line[j].start = |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3454 t->sub.list.line[j].start; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3455 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
|
3456 } |
4444 | 3457 #ifdef ENABLE_LOG |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3458 log_subexpr(&t->sub); |
4444 | 3459 #endif |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3460 /* 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
|
3461 * 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
|
3462 * 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
|
3463 * correct. */ |
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3464 if (nextlist->n == 0 && neglist->n == 0) |
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3465 clen = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3466 goto nextchar; |
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3467 } |
4444 | 3468 |
3469 case NFA_END_INVISIBLE: | |
3470 /* This is only encountered after a NFA_START_INVISIBLE node. | |
3471 * They surround a zero-width group, used with "\@=" and "\&". | |
3472 * If we got here, it means that the current "invisible" group | |
3473 * finished successfully, so return control to the parent | |
3474 * nfa_regmatch(). Submatches are stored in *m, and used in | |
3475 * the parent call. */ | |
3476 if (start->c == NFA_MOPEN + 0) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3477 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3478 else |
3479 { | |
3480 *m = t->sub; | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3481 nfa_match = TRUE; |
4444 | 3482 } |
3483 break; | |
3484 | |
3485 case NFA_START_INVISIBLE: | |
3486 /* Save global variables, and call nfa_regmatch() to check if | |
3487 * the current concat matches at this position. The concat | |
3488 * ends with the node NFA_END_INVISIBLE */ | |
3489 old_reginput = reginput; | |
3490 old_regline = regline; | |
3491 old_reglnum = reglnum; | |
3492 if (listids == NULL) | |
3493 { | |
3494 listids = (int *) lalloc(sizeof(int) * nstate, TRUE); | |
3495 if (listids == NULL) | |
3496 { | |
3497 EMSG(_("E878: (NFA) Could not allocate memory for branch traversal!")); | |
3498 return 0; | |
3499 } | |
3500 } | |
3501 #ifdef ENABLE_LOG | |
3502 if (log_fd != stderr) | |
3503 fclose(log_fd); | |
3504 log_fd = NULL; | |
3505 #endif | |
3506 /* Have to clear the listid field of the NFA nodes, so that | |
3507 * nfa_regmatch() and addstate() can run properly after | |
3508 * recursion. */ | |
3509 nfa_save_listids(start, listids); | |
3510 nfa_set_null_listids(start); | |
3511 result = nfa_regmatch(t->state->out, submatch, m); | |
3512 nfa_set_neg_listids(start); | |
3513 nfa_restore_listids(start, listids); | |
3514 | |
3515 #ifdef ENABLE_LOG | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3516 log_fd = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 3517 if (log_fd != NULL) |
3518 { | |
3519 fprintf(log_fd, "****************************\n"); | |
3520 fprintf(log_fd, "FINISHED RUNNING nfa_regmatch() recursively\n"); | |
3521 fprintf(log_fd, "MATCH = %s\n", result == TRUE ? "OK" : "FALSE"); | |
3522 fprintf(log_fd, "****************************\n"); | |
3523 } | |
3524 else | |
3525 { | |
3526 EMSG(_("Could not open temporary log file for writing, displaying on stderr ... ")); | |
3527 log_fd = stderr; | |
3528 } | |
3529 #endif | |
3530 if (result == TRUE) | |
3531 { | |
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3532 int j; |
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3533 |
4444 | 3534 /* Restore position in input text */ |
3535 reginput = old_reginput; | |
3536 regline = old_regline; | |
3537 reglnum = old_reglnum; | |
3538 /* Copy submatch info from the recursive call */ | |
3539 if (REG_MULTI) | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3540 for (j = 1; j < m->in_use; j++) |
4444 | 3541 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3542 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
|
3543 t->sub.list.multi[j].end = m->list.multi[j].end; |
4444 | 3544 } |
3545 else | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3546 for (j = 1; j < m->in_use; j++) |
4444 | 3547 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3548 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
|
3549 t->sub.list.line[j].end = m->list.line[j].end; |
4444 | 3550 } |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3551 t->sub.in_use = m->in_use; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3552 |
4444 | 3553 /* t->state->out1 is the corresponding END_INVISIBLE node */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3554 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
|
3555 &listidx); |
4444 | 3556 } |
3557 else | |
3558 { | |
3559 /* continue with next input char */ | |
3560 reginput = old_reginput; | |
3561 } | |
3562 break; | |
3563 | |
3564 case NFA_BOL: | |
3565 if (reginput == regline) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3566 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3567 break; |
3568 | |
3569 case NFA_EOL: | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3570 if (curc == NUL) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3571 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3572 break; |
3573 | |
3574 case NFA_BOW: | |
3575 { | |
3576 int bow = TRUE; | |
3577 | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3578 if (curc == NUL) |
4444 | 3579 bow = FALSE; |
3580 #ifdef FEAT_MBYTE | |
3581 else if (has_mbyte) | |
3582 { | |
3583 int this_class; | |
3584 | |
3585 /* Get class of current and previous char (if it exists). */ | |
4503 | 3586 this_class = mb_get_class_buf(reginput, reg_buf); |
4444 | 3587 if (this_class <= 1) |
3588 bow = FALSE; | |
3589 else if (reg_prev_class() == this_class) | |
3590 bow = FALSE; | |
3591 } | |
3592 #endif | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3593 else if (!vim_iswordc_buf(curc, reg_buf) |
4503 | 3594 || (reginput > regline |
3595 && vim_iswordc_buf(reginput[-1], reg_buf))) | |
4444 | 3596 bow = FALSE; |
3597 if (bow) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3598 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3599 break; |
3600 } | |
3601 | |
3602 case NFA_EOW: | |
3603 { | |
3604 int eow = TRUE; | |
3605 | |
3606 if (reginput == regline) | |
3607 eow = FALSE; | |
3608 #ifdef FEAT_MBYTE | |
3609 else if (has_mbyte) | |
3610 { | |
3611 int this_class, prev_class; | |
3612 | |
3613 /* Get class of current and previous char (if it exists). */ | |
4503 | 3614 this_class = mb_get_class_buf(reginput, reg_buf); |
4444 | 3615 prev_class = reg_prev_class(); |
3616 if (this_class == prev_class | |
3617 || prev_class == 0 || prev_class == 1) | |
3618 eow = FALSE; | |
3619 } | |
3620 #endif | |
4503 | 3621 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
|
3622 || (reginput[0] != NUL |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3623 && vim_iswordc_buf(curc, reg_buf))) |
4444 | 3624 eow = FALSE; |
3625 if (eow) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3626 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3627 break; |
3628 } | |
3629 | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3630 #ifdef FEAT_MBYTE |
4444 | 3631 case NFA_COMPOSING: |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3632 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3633 int mc = curc; |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3634 int len = 0; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3635 nfa_state_T *end; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3636 nfa_state_T *sta; |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3637 int cchars[MAX_MCO]; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3638 int ccount = 0; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3639 int j; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3640 |
4444 | 3641 sta = t->state->out; |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3642 len = 0; |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3643 if (utf_iscomposing(sta->c)) |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3644 { |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3645 /* Only match composing character(s), ignore base |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3646 * character. Used for ".{composing}" and "{composing}" |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3647 * (no preceding character). */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3648 len += mb_char2len(mc); |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3649 } |
4551
f0c37f5cf4ea
updated for version 7.3.1023
Bram Moolenaar <bram@vim.org>
parents:
4549
diff
changeset
|
3650 if (ireg_icombine && len == 0) |
4444 | 3651 { |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3652 /* If \Z was present, then ignore composing characters. |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3653 * 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
|
3654 /* TODO: How about negated? */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3655 if (len == 0 && sta->c != curc) |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3656 result = FAIL; |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3657 else |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3658 result = OK; |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3659 while (sta->c != NFA_END_COMPOSING) |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3660 sta = sta->out; |
4444 | 3661 } |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3662 |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3663 /* Check base character matches first, unless ignored. */ |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3664 else if (len > 0 || mc == sta->c) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3665 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3666 if (len == 0) |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3667 { |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3668 len += mb_char2len(mc); |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3669 sta = sta->out; |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3670 } |
4444 | 3671 |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3672 /* 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
|
3673 * Get them into cchars[] first. */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3674 while (len < clen) |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3675 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3676 mc = mb_ptr2char(reginput + len); |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3677 cchars[ccount++] = mc; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3678 len += mb_char2len(mc); |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3679 if (ccount == MAX_MCO) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3680 break; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3681 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3682 |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3683 /* 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
|
3684 * 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
|
3685 * composing chars are matched. */ |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3686 result = OK; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3687 while (sta->c != NFA_END_COMPOSING) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3688 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3689 for (j = 0; j < ccount; ++j) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3690 if (cchars[j] == sta->c) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3691 break; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3692 if (j == ccount) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3693 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3694 result = FAIL; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3695 break; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3696 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3697 sta = sta->out; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3698 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3699 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3700 else |
4525
36ddcf4cecbc
updated for version 7.3.1010
Bram Moolenaar <bram@vim.org>
parents:
4517
diff
changeset
|
3701 result = FAIL; |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3702 |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3703 end = t->state->out1; /* NFA_END_COMPOSING */ |
4444 | 3704 ADD_POS_NEG_STATE(end); |
3705 break; | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3706 } |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3707 #endif |
4444 | 3708 |
3709 case NFA_NEWL: | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3710 if (curc == NUL && !reg_line_lbr && REG_MULTI |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3711 && reglnum <= reg_maxline) |
4444 | 3712 { |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3713 go_to_nextline = TRUE; |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3714 /* 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
|
3715 * at the start of the next line. */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3716 addstate(nextlist, t->state->out, &t->sub, -1); |
4444 | 3717 } |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3718 else if (curc == '\n' && reg_line_lbr) |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3719 { |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3720 /* 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
|
3721 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
|
3722 } |
4444 | 3723 break; |
3724 | |
3725 case NFA_CLASS_ALNUM: | |
3726 case NFA_CLASS_ALPHA: | |
3727 case NFA_CLASS_BLANK: | |
3728 case NFA_CLASS_CNTRL: | |
3729 case NFA_CLASS_DIGIT: | |
3730 case NFA_CLASS_GRAPH: | |
3731 case NFA_CLASS_LOWER: | |
3732 case NFA_CLASS_PRINT: | |
3733 case NFA_CLASS_PUNCT: | |
3734 case NFA_CLASS_SPACE: | |
3735 case NFA_CLASS_UPPER: | |
3736 case NFA_CLASS_XDIGIT: | |
3737 case NFA_CLASS_TAB: | |
3738 case NFA_CLASS_RETURN: | |
3739 case NFA_CLASS_BACKSPACE: | |
3740 case NFA_CLASS_ESCAPE: | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3741 result = check_char_class(t->state->c, curc); |
4444 | 3742 ADD_POS_NEG_STATE(t->state); |
3743 break; | |
3744 | |
3745 case NFA_END_NEG_RANGE: | |
3746 /* This follows a series of negated nodes, like: | |
3747 * 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
|
3748 if (curc > 0) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3749 addstate(nextlist, t->state->out, &t->sub, clen); |
4444 | 3750 break; |
3751 | |
3752 case NFA_ANY: | |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3753 /* 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
|
3754 if (curc > 0) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3755 addstate(nextlist, t->state->out, &t->sub, clen); |
4444 | 3756 break; |
3757 | |
3758 /* | |
3759 * Character classes like \a for alpha, \d for digit etc. | |
3760 */ | |
3761 case NFA_IDENT: /* \i */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3762 result = vim_isIDc(curc); |
4444 | 3763 ADD_POS_NEG_STATE(t->state); |
3764 break; | |
3765 | |
3766 case NFA_SIDENT: /* \I */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3767 result = !VIM_ISDIGIT(curc) && vim_isIDc(curc); |
4444 | 3768 ADD_POS_NEG_STATE(t->state); |
3769 break; | |
3770 | |
3771 case NFA_KWORD: /* \k */ | |
4503 | 3772 result = vim_iswordp_buf(reginput, reg_buf); |
4444 | 3773 ADD_POS_NEG_STATE(t->state); |
3774 break; | |
3775 | |
3776 case NFA_SKWORD: /* \K */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3777 result = !VIM_ISDIGIT(curc) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3778 && vim_iswordp_buf(reginput, reg_buf); |
4444 | 3779 ADD_POS_NEG_STATE(t->state); |
3780 break; | |
3781 | |
3782 case NFA_FNAME: /* \f */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3783 result = vim_isfilec(curc); |
4444 | 3784 ADD_POS_NEG_STATE(t->state); |
3785 break; | |
3786 | |
3787 case NFA_SFNAME: /* \F */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3788 result = !VIM_ISDIGIT(curc) && vim_isfilec(curc); |
4444 | 3789 ADD_POS_NEG_STATE(t->state); |
3790 break; | |
3791 | |
3792 case NFA_PRINT: /* \p */ | |
4470 | 3793 result = ptr2cells(reginput) == 1; |
4444 | 3794 ADD_POS_NEG_STATE(t->state); |
3795 break; | |
3796 | |
3797 case NFA_SPRINT: /* \P */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3798 result = !VIM_ISDIGIT(curc) && ptr2cells(reginput) == 1; |
4444 | 3799 ADD_POS_NEG_STATE(t->state); |
3800 break; | |
3801 | |
3802 case NFA_WHITE: /* \s */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3803 result = vim_iswhite(curc); |
4444 | 3804 ADD_POS_NEG_STATE(t->state); |
3805 break; | |
3806 | |
3807 case NFA_NWHITE: /* \S */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3808 result = curc != NUL && !vim_iswhite(curc); |
4444 | 3809 ADD_POS_NEG_STATE(t->state); |
3810 break; | |
3811 | |
3812 case NFA_DIGIT: /* \d */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3813 result = ri_digit(curc); |
4444 | 3814 ADD_POS_NEG_STATE(t->state); |
3815 break; | |
3816 | |
3817 case NFA_NDIGIT: /* \D */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3818 result = curc != NUL && !ri_digit(curc); |
4444 | 3819 ADD_POS_NEG_STATE(t->state); |
3820 break; | |
3821 | |
3822 case NFA_HEX: /* \x */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3823 result = ri_hex(curc); |
4444 | 3824 ADD_POS_NEG_STATE(t->state); |
3825 break; | |
3826 | |
3827 case NFA_NHEX: /* \X */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3828 result = curc != NUL && !ri_hex(curc); |
4444 | 3829 ADD_POS_NEG_STATE(t->state); |
3830 break; | |
3831 | |
3832 case NFA_OCTAL: /* \o */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3833 result = ri_octal(curc); |
4444 | 3834 ADD_POS_NEG_STATE(t->state); |
3835 break; | |
3836 | |
3837 case NFA_NOCTAL: /* \O */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3838 result = curc != NUL && !ri_octal(curc); |
4444 | 3839 ADD_POS_NEG_STATE(t->state); |
3840 break; | |
3841 | |
3842 case NFA_WORD: /* \w */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3843 result = ri_word(curc); |
4444 | 3844 ADD_POS_NEG_STATE(t->state); |
3845 break; | |
3846 | |
3847 case NFA_NWORD: /* \W */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3848 result = curc != NUL && !ri_word(curc); |
4444 | 3849 ADD_POS_NEG_STATE(t->state); |
3850 break; | |
3851 | |
3852 case NFA_HEAD: /* \h */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3853 result = ri_head(curc); |
4444 | 3854 ADD_POS_NEG_STATE(t->state); |
3855 break; | |
3856 | |
3857 case NFA_NHEAD: /* \H */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3858 result = curc != NUL && !ri_head(curc); |
4444 | 3859 ADD_POS_NEG_STATE(t->state); |
3860 break; | |
3861 | |
3862 case NFA_ALPHA: /* \a */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3863 result = ri_alpha(curc); |
4444 | 3864 ADD_POS_NEG_STATE(t->state); |
3865 break; | |
3866 | |
3867 case NFA_NALPHA: /* \A */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3868 result = curc != NUL && !ri_alpha(curc); |
4444 | 3869 ADD_POS_NEG_STATE(t->state); |
3870 break; | |
3871 | |
3872 case NFA_LOWER: /* \l */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3873 result = ri_lower(curc); |
4444 | 3874 ADD_POS_NEG_STATE(t->state); |
3875 break; | |
3876 | |
3877 case NFA_NLOWER: /* \L */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3878 result = curc != NUL && !ri_lower(curc); |
4444 | 3879 ADD_POS_NEG_STATE(t->state); |
3880 break; | |
3881 | |
3882 case NFA_UPPER: /* \u */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3883 result = ri_upper(curc); |
4444 | 3884 ADD_POS_NEG_STATE(t->state); |
3885 break; | |
3886 | |
3887 case NFA_NUPPER: /* \U */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3888 result = curc != NUL && !ri_upper(curc); |
4444 | 3889 ADD_POS_NEG_STATE(t->state); |
3890 break; | |
3891 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3892 case NFA_BACKREF1: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3893 case NFA_BACKREF2: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3894 case NFA_BACKREF3: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3895 case NFA_BACKREF4: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3896 case NFA_BACKREF5: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3897 case NFA_BACKREF6: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3898 case NFA_BACKREF7: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3899 case NFA_BACKREF8: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3900 case NFA_BACKREF9: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3901 /* \1 .. \9 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3902 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3903 int subidx = t->state->c - NFA_BACKREF1 + 1; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3904 int bytelen; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3905 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3906 result = match_backref(&t->sub, subidx, &bytelen); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3907 if (result) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3908 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3909 if (bytelen == 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3910 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3911 /* 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
|
3912 * be used next */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3913 addstate_here(thislist, t->state->out, &t->sub, |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3914 &listidx); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3915 thislist->t[listidx + 1].count = 0; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3916 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3917 else if (bytelen <= clen) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3918 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3919 /* match current character, jump ahead to out of |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3920 * NFA_SKIP */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3921 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
|
3922 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3923 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3924 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3925 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3926 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3927 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3928 /* skip ofer the matched characters, set character |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3929 * count in NFA_SKIP */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3930 addstate(nextlist, t->state->out, &t->sub, bytelen); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3931 nextlist->t[nextlist->n - 1].count = bytelen - clen; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3932 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3933 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3934 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3935 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3936 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3937 } |
4482 | 3938 break; |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3939 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3940 case NFA_SKIP: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3941 /* charater of previous matching \1 .. \9 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3942 if (t->count - clen <= 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3943 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3944 /* end of match, go to what follows */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3945 addstate(nextlist, t->state->out, &t->sub, clen); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3946 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3947 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3948 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3949 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3950 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3951 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3952 /* add state again with decremented count */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3953 addstate(nextlist, t->state, &t->sub, 0); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3954 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
|
3955 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3956 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3957 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3958 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3959 break; |
4482 | 3960 |
3961 case NFA_SKIP_CHAR: | |
3962 case NFA_ZSTART: | |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
3963 case NFA_ZEND: |
4482 | 3964 /* TODO: should not happen? */ |
3965 break; | |
3966 | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3967 case NFA_LNUM: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3968 case NFA_LNUM_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3969 case NFA_LNUM_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3970 result = (REG_MULTI && |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3971 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
|
3972 (long_u)(reglnum + reg_firstlnum))); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3973 if (result) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3974 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
|
3975 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3976 |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3977 case NFA_COL: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3978 case NFA_COL_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3979 case NFA_COL_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3980 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
|
3981 (long_u)(reginput - regline) + 1); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3982 if (result) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3983 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
|
3984 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3985 |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3986 case NFA_VCOL: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3987 case NFA_VCOL_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3988 case NFA_VCOL_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3989 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
|
3990 (long_u)win_linetabsize( |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3991 reg_win == NULL ? curwin : reg_win, |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3992 regline, (colnr_T)(reginput - regline)) + 1); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3993 if (result) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3994 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
|
3995 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3996 |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3997 case NFA_CURSOR: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3998 result = (reg_win != NULL |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3999 && (reglnum + reg_firstlnum == reg_win->w_cursor.lnum) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4000 && ((colnr_T)(reginput - regline) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4001 == reg_win->w_cursor.col)); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4002 if (result) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4003 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
|
4004 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4005 |
4444 | 4006 default: /* regular character */ |
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4007 { |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4008 int c = t->state->c; |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4009 |
4482 | 4010 /* TODO: put this in #ifdef later */ |
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4011 if (c < -256) |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4012 EMSGN("INTERNAL: Negative state char: %ld", c); |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4013 if (is_Magic(c)) |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4014 c = un_Magic(c); |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4015 result = (c == curc); |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4016 |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4017 if (!result && ireg_ic) |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4018 result = MB_TOLOWER(c) == MB_TOLOWER(curc); |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4019 #ifdef FEAT_MBYTE |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4020 /* 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
|
4021 * 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
|
4022 * character uses NFA_COMPOSING above. */ |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4023 if (result && enc_utf8 && !ireg_icombine |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4024 && clen != utf_char2len(curc)) |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4025 result = FALSE; |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4026 #endif |
4444 | 4027 ADD_POS_NEG_STATE(t->state); |
4028 break; | |
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4029 } |
4444 | 4030 } |
4031 | |
4032 } /* for (thislist = thislist; thislist->state; thislist++) */ | |
4033 | |
4557
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
4034 /* 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
|
4035 * start state to the list of states. |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
4036 * 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
|
4037 * matters! |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
4038 * 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
|
4039 * because recursive calls should only start in the first position. |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
4040 * Also don't start a match past the first line. */ |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4041 if (nfa_match == FALSE && start->c == NFA_MOPEN + 0 |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
4042 && reglnum == 0 && clen != 0 |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
4043 && (ireg_maxcol == 0 |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
4044 || (colnr_T)(reginput - regline) < ireg_maxcol)) |
4444 | 4045 { |
4046 #ifdef ENABLE_LOG | |
4047 fprintf(log_fd, "(---) STARTSTATE\n"); | |
4048 #endif | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4049 addstate(nextlist, start, m, clen); |
4444 | 4050 } |
4051 | |
4052 #ifdef ENABLE_LOG | |
4053 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
|
4054 { |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4055 int i; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4056 |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4057 for (i = 0; i < thislist->n; i++) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4058 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
|
4059 } |
4444 | 4060 fprintf(log_fd, "\n"); |
4061 #endif | |
4062 | |
4063 nextchar: | |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4064 /* 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
|
4065 * finish. */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4066 if (clen != 0) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4067 reginput += clen; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4068 else if (go_to_nextline) |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4069 reg_nextline(); |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4070 else |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4071 break; |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4072 } |
4444 | 4073 |
4074 #ifdef ENABLE_LOG | |
4075 if (log_fd != stderr) | |
4076 fclose(log_fd); | |
4077 log_fd = NULL; | |
4078 #endif | |
4079 | |
4080 theend: | |
4081 /* Free memory */ | |
4082 vim_free(list[0].t); | |
4083 vim_free(list[1].t); | |
4084 vim_free(list[2].t); | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4085 vim_free(listids); |
4444 | 4086 #undef ADD_POS_NEG_STATE |
4460 | 4087 #ifdef NFA_REGEXP_DEBUG_LOG |
4444 | 4088 fclose(debug); |
4089 #endif | |
4090 | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4091 return nfa_match; |
4444 | 4092 } |
4093 | |
4094 /* | |
4095 * Try match of "prog" with at regline["col"]. | |
4096 * Returns 0 for failure, number of lines contained in the match otherwise. | |
4097 */ | |
4098 static long | |
4099 nfa_regtry(start, col) | |
4100 nfa_state_T *start; | |
4101 colnr_T col; | |
4102 { | |
4103 int i; | |
4104 regsub_T sub, m; | |
4105 #ifdef ENABLE_LOG | |
4106 FILE *f; | |
4107 #endif | |
4108 | |
4109 reginput = regline + col; | |
4110 need_clear_subexpr = TRUE; | |
4111 | |
4112 #ifdef ENABLE_LOG | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
4113 f = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 4114 if (f != NULL) |
4115 { | |
4116 fprintf(f, "\n\n\n\n\n\n\t\t=======================================================\n"); | |
4117 fprintf(f, " =======================================================\n"); | |
4118 #ifdef DEBUG | |
4119 fprintf(f, "\tRegexp is \"%s\"\n", nfa_regengine.expr); | |
4120 #endif | |
4121 fprintf(f, "\tInput text is \"%s\" \n", reginput); | |
4122 fprintf(f, " =======================================================\n\n\n\n\n\n\n"); | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
4123 nfa_print_state(f, start); |
4444 | 4124 fprintf(f, "\n\n"); |
4125 fclose(f); | |
4126 } | |
4127 else | |
4128 EMSG(_("Could not open temporary log file for writing ")); | |
4129 #endif | |
4130 | |
4131 if (REG_MULTI) | |
4132 { | |
4133 /* Use 0xff to set lnum to -1 */ | |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4134 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
|
4135 vim_memset(m.list.multi, 0xff, sizeof(struct multipos) * nfa_nsubexpr); |
4444 | 4136 } |
4137 else | |
4138 { | |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4139 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
|
4140 vim_memset(m.list.line, 0, sizeof(struct linepos) * nfa_nsubexpr); |
4444 | 4141 } |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4142 sub.in_use = 0; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4143 m.in_use = 0; |
4444 | 4144 |
4145 if (nfa_regmatch(start, &sub, &m) == FALSE) | |
4146 return 0; | |
4147 | |
4148 cleanup_subexpr(); | |
4149 if (REG_MULTI) | |
4150 { | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4151 for (i = 0; i < sub.in_use; i++) |
4444 | 4152 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4153 reg_startpos[i] = sub.list.multi[i].start; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4154 reg_endpos[i] = sub.list.multi[i].end; |
4444 | 4155 } |
4156 | |
4157 if (reg_startpos[0].lnum < 0) | |
4158 { | |
4159 reg_startpos[0].lnum = 0; | |
4160 reg_startpos[0].col = col; | |
4161 } | |
4162 if (reg_endpos[0].lnum < 0) | |
4163 { | |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
4164 /* pattern has a \ze but it didn't match, use current end */ |
4444 | 4165 reg_endpos[0].lnum = reglnum; |
4166 reg_endpos[0].col = (int)(reginput - regline); | |
4167 } | |
4168 else | |
4169 /* Use line number of "\ze". */ | |
4170 reglnum = reg_endpos[0].lnum; | |
4171 } | |
4172 else | |
4173 { | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4174 for (i = 0; i < sub.in_use; i++) |
4444 | 4175 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4176 reg_startp[i] = sub.list.line[i].start; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4177 reg_endp[i] = sub.list.line[i].end; |
4444 | 4178 } |
4179 | |
4180 if (reg_startp[0] == NULL) | |
4181 reg_startp[0] = regline + col; | |
4182 if (reg_endp[0] == NULL) | |
4183 reg_endp[0] = reginput; | |
4184 } | |
4185 | |
4186 return 1 + reglnum; | |
4187 } | |
4188 | |
4189 /* | |
4190 * Match a regexp against a string ("line" points to the string) or multiple | |
4191 * lines ("line" is NULL, use reg_getline()). | |
4192 * | |
4193 * Returns 0 for failure, number of lines contained in the match otherwise. | |
4194 */ | |
4195 static long | |
4196 nfa_regexec_both(line, col) | |
4197 char_u *line; | |
4198 colnr_T col; /* column to start looking for match */ | |
4199 { | |
4200 nfa_regprog_T *prog; | |
4201 long retval = 0L; | |
4202 int i; | |
4203 | |
4204 if (REG_MULTI) | |
4205 { | |
4206 prog = (nfa_regprog_T *)reg_mmatch->regprog; | |
4207 line = reg_getline((linenr_T)0); /* relative to the cursor */ | |
4208 reg_startpos = reg_mmatch->startpos; | |
4209 reg_endpos = reg_mmatch->endpos; | |
4210 } | |
4211 else | |
4212 { | |
4213 prog = (nfa_regprog_T *)reg_match->regprog; | |
4214 reg_startp = reg_match->startp; | |
4215 reg_endp = reg_match->endp; | |
4216 } | |
4217 | |
4218 /* Be paranoid... */ | |
4219 if (prog == NULL || line == NULL) | |
4220 { | |
4221 EMSG(_(e_null)); | |
4222 goto theend; | |
4223 } | |
4224 | |
4225 /* If the start column is past the maximum column: no need to try. */ | |
4226 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
4227 goto theend; | |
4228 | |
4229 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */ | |
4230 if (prog->regflags & RF_ICASE) | |
4231 ireg_ic = TRUE; | |
4232 else if (prog->regflags & RF_NOICASE) | |
4233 ireg_ic = FALSE; | |
4234 | |
4235 #ifdef FEAT_MBYTE | |
4236 /* If pattern contains "\Z" overrule value of ireg_icombine */ | |
4237 if (prog->regflags & RF_ICOMBINE) | |
4238 ireg_icombine = TRUE; | |
4239 #endif | |
4240 | |
4241 regline = line; | |
4242 reglnum = 0; /* relative to line */ | |
4243 | |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4244 nfa_has_zend = prog->has_zend; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
4245 nfa_has_backref = prog->has_backref; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4246 nfa_nsubexpr = prog->nsubexp; |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4247 |
4444 | 4248 nstate = prog->nstate; |
4249 for (i = 0; i < nstate; ++i) | |
4250 { | |
4251 prog->state[i].id = i; | |
4252 prog->state[i].lastlist = 0; | |
4253 } | |
4254 | |
4255 retval = nfa_regtry(prog->start, col); | |
4256 | |
4257 theend: | |
4258 return retval; | |
4259 } | |
4260 | |
4261 /* | |
4262 * Compile a regular expression into internal code for the NFA matcher. | |
4263 * Returns the program in allocated space. Returns NULL for an error. | |
4264 */ | |
4265 static regprog_T * | |
4266 nfa_regcomp(expr, re_flags) | |
4267 char_u *expr; | |
4268 int re_flags; | |
4269 { | |
4541
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4270 nfa_regprog_T *prog = NULL; |
4458 | 4271 size_t prog_size; |
4444 | 4272 int *postfix; |
4273 | |
4274 if (expr == NULL) | |
4275 return NULL; | |
4276 | |
4277 #ifdef DEBUG | |
4278 nfa_regengine.expr = expr; | |
4279 #endif | |
4280 | |
4281 init_class_tab(); | |
4282 | |
4283 if (nfa_regcomp_start(expr, re_flags) == FAIL) | |
4284 return NULL; | |
4285 | |
4286 /* 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
|
4287 * (and count its size). */ |
4444 | 4288 postfix = re2post(); |
4289 if (postfix == NULL) | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4290 { |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4291 /* TODO: only give this error for debugging? */ |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4292 if (post_ptr >= post_end) |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4293 EMSGN("Internal error: estimated max number of states insufficient: %ld", post_end - post_start); |
4444 | 4294 goto fail; /* Cascaded (syntax?) error */ |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4295 } |
4444 | 4296 |
4297 /* | |
4298 * In order to build the NFA, we parse the input regexp twice: | |
4299 * 1. first pass to count size (so we can allocate space) | |
4300 * 2. second to emit code | |
4301 */ | |
4302 #ifdef ENABLE_LOG | |
4303 { | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
4304 FILE *f = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 4305 |
4306 if (f != NULL) | |
4307 { | |
4308 fprintf(f, "\n*****************************\n\n\n\n\tCompiling regexp \"%s\" ... hold on !\n", expr); | |
4309 fclose(f); | |
4310 } | |
4311 } | |
4312 #endif | |
4313 | |
4314 /* | |
4315 * PASS 1 | |
4316 * Count number of NFA states in "nstate". Do not build the NFA. | |
4317 */ | |
4318 post2nfa(postfix, post_ptr, TRUE); | |
4541
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4319 |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4320 /* Space for compiled regexp */ |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4321 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
|
4322 prog = (nfa_regprog_T *)lalloc(prog_size, TRUE); |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4323 if (prog == NULL) |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4324 goto fail; |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4325 vim_memset(prog, 0, prog_size); |
4444 | 4326 state_ptr = prog->state; |
4327 | |
4328 /* | |
4329 * PASS 2 | |
4330 * Build the NFA | |
4331 */ | |
4332 prog->start = post2nfa(postfix, post_ptr, FALSE); | |
4333 if (prog->start == NULL) | |
4334 goto fail; | |
4335 | |
4336 prog->regflags = regflags; | |
4337 prog->engine = &nfa_regengine; | |
4338 prog->nstate = nstate; | |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4339 prog->has_zend = nfa_has_zend; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
4340 prog->has_backref = nfa_has_backref; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4341 prog->nsubexp = regnpar; |
4444 | 4342 #ifdef ENABLE_LOG |
4343 nfa_postfix_dump(expr, OK); | |
4344 nfa_dump(prog); | |
4345 #endif | |
4346 | |
4347 out: | |
4348 vim_free(post_start); | |
4349 post_start = post_ptr = post_end = NULL; | |
4350 state_ptr = NULL; | |
4351 return (regprog_T *)prog; | |
4352 | |
4353 fail: | |
4354 vim_free(prog); | |
4355 prog = NULL; | |
4356 #ifdef ENABLE_LOG | |
4357 nfa_postfix_dump(expr, FAIL); | |
4358 #endif | |
4359 #ifdef DEBUG | |
4360 nfa_regengine.expr = NULL; | |
4361 #endif | |
4362 goto out; | |
4363 } | |
4364 | |
4365 | |
4366 /* | |
4367 * Match a regexp against a string. | |
4368 * "rmp->regprog" is a compiled regexp as returned by nfa_regcomp(). | |
4369 * Uses curbuf for line count and 'iskeyword'. | |
4370 * | |
4371 * Return TRUE if there is a match, FALSE if not. | |
4372 */ | |
4373 static int | |
4374 nfa_regexec(rmp, line, col) | |
4375 regmatch_T *rmp; | |
4376 char_u *line; /* string to match against */ | |
4377 colnr_T col; /* column to start looking for match */ | |
4378 { | |
4379 reg_match = rmp; | |
4380 reg_mmatch = NULL; | |
4381 reg_maxline = 0; | |
4382 reg_line_lbr = FALSE; | |
4383 reg_buf = curbuf; | |
4384 reg_win = NULL; | |
4385 ireg_ic = rmp->rm_ic; | |
4386 #ifdef FEAT_MBYTE | |
4387 ireg_icombine = FALSE; | |
4388 #endif | |
4389 ireg_maxcol = 0; | |
4390 return (nfa_regexec_both(line, col) != 0); | |
4391 } | |
4392 | |
4393 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
4394 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
4395 | |
4396 static int nfa_regexec_nl __ARGS((regmatch_T *rmp, char_u *line, colnr_T col)); | |
4397 | |
4398 /* | |
4399 * Like nfa_regexec(), but consider a "\n" in "line" to be a line break. | |
4400 */ | |
4401 static int | |
4402 nfa_regexec_nl(rmp, line, col) | |
4403 regmatch_T *rmp; | |
4404 char_u *line; /* string to match against */ | |
4405 colnr_T col; /* column to start looking for match */ | |
4406 { | |
4407 reg_match = rmp; | |
4408 reg_mmatch = NULL; | |
4409 reg_maxline = 0; | |
4410 reg_line_lbr = TRUE; | |
4411 reg_buf = curbuf; | |
4412 reg_win = NULL; | |
4413 ireg_ic = rmp->rm_ic; | |
4414 #ifdef FEAT_MBYTE | |
4415 ireg_icombine = FALSE; | |
4416 #endif | |
4417 ireg_maxcol = 0; | |
4418 return (nfa_regexec_both(line, col) != 0); | |
4419 } | |
4420 #endif | |
4421 | |
4422 | |
4423 /* | |
4424 * Match a regexp against multiple lines. | |
4425 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
4426 * Uses curbuf for line count and 'iskeyword'. | |
4427 * | |
4428 * Return zero if there is no match. Return number of lines contained in the | |
4429 * match otherwise. | |
4430 * | |
4431 * Note: the body is the same as bt_regexec() except for nfa_regexec_both() | |
4432 * | |
4433 * ! Also NOTE : match may actually be in another line. e.g.: | |
4434 * when r.e. is \nc, cursor is at 'a' and the text buffer looks like | |
4435 * | |
4436 * +-------------------------+ | |
4437 * |a | | |
4438 * |b | | |
4439 * |c | | |
4440 * | | | |
4441 * +-------------------------+ | |
4442 * | |
4443 * then nfa_regexec_multi() returns 3. while the original | |
4444 * vim_regexec_multi() returns 0 and a second call at line 2 will return 2. | |
4445 * | |
4446 * FIXME if this behavior is not compatible. | |
4447 */ | |
4448 static long | |
4449 nfa_regexec_multi(rmp, win, buf, lnum, col, tm) | |
4450 regmmatch_T *rmp; | |
4451 win_T *win; /* window in which to search or NULL */ | |
4452 buf_T *buf; /* buffer in which to search */ | |
4453 linenr_T lnum; /* nr of line to start looking for match */ | |
4454 colnr_T col; /* column to start looking for match */ | |
4455 proftime_T *tm UNUSED; /* timeout limit or NULL */ | |
4456 { | |
4457 reg_match = NULL; | |
4458 reg_mmatch = rmp; | |
4459 reg_buf = buf; | |
4460 reg_win = win; | |
4461 reg_firstlnum = lnum; | |
4462 reg_maxline = reg_buf->b_ml.ml_line_count - lnum; | |
4463 reg_line_lbr = FALSE; | |
4464 ireg_ic = rmp->rmm_ic; | |
4465 #ifdef FEAT_MBYTE | |
4466 ireg_icombine = FALSE; | |
4467 #endif | |
4468 ireg_maxcol = rmp->rmm_maxcol; | |
4469 | |
4503 | 4470 return nfa_regexec_both(NULL, col); |
4444 | 4471 } |
4472 | |
4473 #ifdef DEBUG | |
4474 # undef ENABLE_LOG | |
4475 #endif |