Mercurial > vim
annotate src/regexp_nfa.c @ 4671:b3c59716e700 v7.3.1083
updated for version 7.3.1083
Problem: New regexp engine: Does not support \%^ and \%$.
Solution: Support matching start and end of file.
author | Bram Moolenaar <bram@vim.org> |
---|---|
date | Fri, 31 May 2013 22:14:52 +0200 |
parents | 2d15582e132f |
children | 05d57d7c2d55 |
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 { \ | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
236 if (post_ptr >= post_end && realloc_post_list() == FAIL) \ |
4444 | 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. |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
259 * When it is still not enough realloc_post_list() will be used. */ |
4458 | 260 nstate_max += 1000; |
4454 | 261 |
262 /* Size for postfix representation of expr. */ | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
263 postfix_size = sizeof(int) * 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 /* | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
280 * Allocate more space for post_start. Called when |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
281 * running above the estimated number of states. |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
282 */ |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
283 static int |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
284 realloc_post_list() |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
285 { |
4667
9e7ef781d494
updated for version 7.3.1081
Bram Moolenaar <bram@vim.org>
parents:
4661
diff
changeset
|
286 int nstate_max = (int)(post_end - post_start); |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
287 int new_max = nstate_max + 1000; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
288 int *new_start; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
289 int *old_start; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
290 |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
291 new_start = (int *)lalloc(new_max * sizeof(int), TRUE); |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
292 if (new_start == NULL) |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
293 return FAIL; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
294 mch_memmove(new_start, post_start, nstate_max * sizeof(int)); |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
295 vim_memset(new_start + nstate_max, 0, 1000 * sizeof(int)); |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
296 old_start = post_start; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
297 post_start = new_start; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
298 post_ptr = new_start + (post_ptr - old_start); |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
299 post_end = post_start + new_max; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
300 vim_free(old_start); |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
301 return OK; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
302 } |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
303 |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
304 /* |
4444 | 305 * Search between "start" and "end" and try to recognize a |
306 * character class in expanded form. For example [0-9]. | |
307 * On success, return the id the character class to be emitted. | |
308 * On failure, return 0 (=FAIL) | |
309 * Start points to the first char of the range, while end should point | |
310 * to the closing brace. | |
311 */ | |
312 static int | |
313 nfa_recognize_char_class(start, end, extra_newl) | |
314 char_u *start; | |
315 char_u *end; | |
316 int extra_newl; | |
317 { | |
318 int i; | |
319 /* Each of these variables takes up a char in "config[]", | |
320 * in the order they are here. */ | |
321 int not = FALSE, af = FALSE, AF = FALSE, az = FALSE, AZ = FALSE, | |
322 o7 = FALSE, o9 = FALSE, underscore = FALSE, newl = FALSE; | |
323 char_u *p; | |
324 #define NCONFIGS 16 | |
325 int classid[NCONFIGS] = { | |
326 NFA_DIGIT, NFA_NDIGIT, NFA_HEX, NFA_NHEX, | |
327 NFA_OCTAL, NFA_NOCTAL, NFA_WORD, NFA_NWORD, | |
328 NFA_HEAD, NFA_NHEAD, NFA_ALPHA, NFA_NALPHA, | |
329 NFA_LOWER, NFA_NLOWER, NFA_UPPER, NFA_NUPPER | |
330 }; | |
4450 | 331 char_u myconfig[10]; |
4444 | 332 char_u config[NCONFIGS][9] = { |
333 "000000100", /* digit */ | |
334 "100000100", /* non digit */ | |
335 "011000100", /* hex-digit */ | |
336 "111000100", /* non hex-digit */ | |
337 "000001000", /* octal-digit */ | |
338 "100001000", /* [^0-7] */ | |
339 "000110110", /* [0-9A-Za-z_] */ | |
340 "100110110", /* [^0-9A-Za-z_] */ | |
341 "000110010", /* head of word */ | |
342 "100110010", /* not head of word */ | |
343 "000110000", /* alphabetic char a-z */ | |
344 "100110000", /* non alphabetic char */ | |
345 "000100000", /* lowercase letter */ | |
346 "100100000", /* non lowercase */ | |
347 "000010000", /* uppercase */ | |
348 "100010000" /* non uppercase */ | |
349 }; | |
350 | |
351 if (extra_newl == TRUE) | |
352 newl = TRUE; | |
353 | |
354 if (*end != ']') | |
355 return FAIL; | |
356 p = start; | |
357 if (*p == '^') | |
358 { | |
359 not = TRUE; | |
360 p ++; | |
361 } | |
362 | |
363 while (p < end) | |
364 { | |
365 if (p + 2 < end && *(p + 1) == '-') | |
366 { | |
367 switch (*p) | |
368 { | |
369 case '0': | |
370 if (*(p + 2) == '9') | |
371 { | |
372 o9 = TRUE; | |
373 break; | |
374 } | |
375 else | |
376 if (*(p + 2) == '7') | |
377 { | |
378 o7 = TRUE; | |
379 break; | |
380 } | |
381 case 'a': | |
382 if (*(p + 2) == 'z') | |
383 { | |
384 az = TRUE; | |
385 break; | |
386 } | |
387 else | |
388 if (*(p + 2) == 'f') | |
389 { | |
390 af = TRUE; | |
391 break; | |
392 } | |
393 case 'A': | |
394 if (*(p + 2) == 'Z') | |
395 { | |
396 AZ = TRUE; | |
397 break; | |
398 } | |
399 else | |
400 if (*(p + 2) == 'F') | |
401 { | |
402 AF = TRUE; | |
403 break; | |
404 } | |
405 /* FALLTHROUGH */ | |
406 default: | |
407 return FAIL; | |
408 } | |
409 p += 3; | |
410 } | |
411 else if (p + 1 < end && *p == '\\' && *(p + 1) == 'n') | |
412 { | |
413 newl = TRUE; | |
414 p += 2; | |
415 } | |
416 else if (*p == '_') | |
417 { | |
418 underscore = TRUE; | |
419 p ++; | |
420 } | |
421 else if (*p == '\n') | |
422 { | |
423 newl = TRUE; | |
424 p ++; | |
425 } | |
426 else | |
427 return FAIL; | |
428 } /* while (p < end) */ | |
429 | |
430 if (p != end) | |
431 return FAIL; | |
432 | |
433 /* build the config that represents the ranges we gathered */ | |
434 STRCPY(myconfig, "000000000"); | |
435 if (not == TRUE) | |
436 myconfig[0] = '1'; | |
437 if (af == TRUE) | |
438 myconfig[1] = '1'; | |
439 if (AF == TRUE) | |
440 myconfig[2] = '1'; | |
441 if (az == TRUE) | |
442 myconfig[3] = '1'; | |
443 if (AZ == TRUE) | |
444 myconfig[4] = '1'; | |
445 if (o7 == TRUE) | |
446 myconfig[5] = '1'; | |
447 if (o9 == TRUE) | |
448 myconfig[6] = '1'; | |
449 if (underscore == TRUE) | |
450 myconfig[7] = '1'; | |
451 if (newl == TRUE) | |
452 { | |
453 myconfig[8] = '1'; | |
454 extra_newl = ADD_NL; | |
455 } | |
456 /* try to recognize character classes */ | |
457 for (i = 0; i < NCONFIGS; i++) | |
4450 | 458 if (STRNCMP(myconfig, config[i], 8) == 0) |
4444 | 459 return classid[i] + extra_newl; |
460 | |
461 /* fallthrough => no success so far */ | |
462 return FAIL; | |
463 | |
464 #undef NCONFIGS | |
465 } | |
466 | |
467 /* | |
468 * Produce the bytes for equivalence class "c". | |
469 * Currently only handles latin1, latin9 and utf-8. | |
470 * Emits bytes in postfix notation: 'a,b,NFA_OR,c,NFA_OR' is | |
471 * equivalent to 'a OR b OR c' | |
472 * | |
473 * NOTE! When changing this function, also update reg_equi_class() | |
474 */ | |
475 static int | |
476 nfa_emit_equi_class(c, neg) | |
477 int c; | |
478 int neg; | |
479 { | |
480 int first = TRUE; | |
481 int glue = neg == TRUE ? NFA_CONCAT : NFA_OR; | |
482 #define EMIT2(c) \ | |
483 EMIT(c); \ | |
484 if (neg == TRUE) { \ | |
485 EMIT(NFA_NOT); \ | |
486 } \ | |
487 if (first == FALSE) \ | |
488 EMIT(glue); \ | |
489 else \ | |
490 first = FALSE; \ | |
491 | |
492 #ifdef FEAT_MBYTE | |
493 if (enc_utf8 || STRCMP(p_enc, "latin1") == 0 | |
494 || STRCMP(p_enc, "iso-8859-15") == 0) | |
495 #endif | |
496 { | |
497 switch (c) | |
498 { | |
499 case 'A': case '\300': case '\301': case '\302': | |
500 case '\303': case '\304': case '\305': | |
501 EMIT2('A'); EMIT2('\300'); EMIT2('\301'); | |
502 EMIT2('\302'); EMIT2('\303'); EMIT2('\304'); | |
503 EMIT2('\305'); | |
504 return OK; | |
505 | |
506 case 'C': case '\307': | |
507 EMIT2('C'); EMIT2('\307'); | |
508 return OK; | |
509 | |
510 case 'E': case '\310': case '\311': case '\312': case '\313': | |
511 EMIT2('E'); EMIT2('\310'); EMIT2('\311'); | |
512 EMIT2('\312'); EMIT2('\313'); | |
513 return OK; | |
514 | |
515 case 'I': case '\314': case '\315': case '\316': case '\317': | |
516 EMIT2('I'); EMIT2('\314'); EMIT2('\315'); | |
517 EMIT2('\316'); EMIT2('\317'); | |
518 return OK; | |
519 | |
520 case 'N': case '\321': | |
521 EMIT2('N'); EMIT2('\321'); | |
522 return OK; | |
523 | |
524 case 'O': case '\322': case '\323': case '\324': case '\325': | |
525 case '\326': | |
526 EMIT2('O'); EMIT2('\322'); EMIT2('\323'); | |
527 EMIT2('\324'); EMIT2('\325'); EMIT2('\326'); | |
528 return OK; | |
529 | |
530 case 'U': case '\331': case '\332': case '\333': case '\334': | |
531 EMIT2('U'); EMIT2('\331'); EMIT2('\332'); | |
532 EMIT2('\333'); EMIT2('\334'); | |
533 return OK; | |
534 | |
535 case 'Y': case '\335': | |
536 EMIT2('Y'); EMIT2('\335'); | |
537 return OK; | |
538 | |
539 case 'a': case '\340': case '\341': case '\342': | |
540 case '\343': case '\344': case '\345': | |
541 EMIT2('a'); EMIT2('\340'); EMIT2('\341'); | |
542 EMIT2('\342'); EMIT2('\343'); EMIT2('\344'); | |
543 EMIT2('\345'); | |
544 return OK; | |
545 | |
546 case 'c': case '\347': | |
547 EMIT2('c'); EMIT2('\347'); | |
548 return OK; | |
549 | |
550 case 'e': case '\350': case '\351': case '\352': case '\353': | |
551 EMIT2('e'); EMIT2('\350'); EMIT2('\351'); | |
552 EMIT2('\352'); EMIT2('\353'); | |
553 return OK; | |
554 | |
555 case 'i': case '\354': case '\355': case '\356': case '\357': | |
556 EMIT2('i'); EMIT2('\354'); EMIT2('\355'); | |
557 EMIT2('\356'); EMIT2('\357'); | |
558 return OK; | |
559 | |
560 case 'n': case '\361': | |
561 EMIT2('n'); EMIT2('\361'); | |
562 return OK; | |
563 | |
564 case 'o': case '\362': case '\363': case '\364': case '\365': | |
565 case '\366': | |
566 EMIT2('o'); EMIT2('\362'); EMIT2('\363'); | |
567 EMIT2('\364'); EMIT2('\365'); EMIT2('\366'); | |
568 return OK; | |
569 | |
570 case 'u': case '\371': case '\372': case '\373': case '\374': | |
571 EMIT2('u'); EMIT2('\371'); EMIT2('\372'); | |
572 EMIT2('\373'); EMIT2('\374'); | |
573 return OK; | |
574 | |
575 case 'y': case '\375': case '\377': | |
576 EMIT2('y'); EMIT2('\375'); EMIT2('\377'); | |
577 return OK; | |
578 | |
579 default: | |
580 return FAIL; | |
581 } | |
582 } | |
583 | |
584 EMIT(c); | |
585 return OK; | |
586 #undef EMIT2 | |
587 } | |
588 | |
589 /* | |
590 * Code to parse regular expression. | |
591 * | |
592 * We try to reuse parsing functions in regexp.c to | |
593 * minimize surprise and keep the syntax consistent. | |
594 */ | |
595 | |
596 /* | |
597 * Parse the lowest level. | |
598 * | |
599 * An atom can be one of a long list of items. Many atoms match one character | |
600 * in the text. It is often an ordinary character or a character class. | |
601 * Braces can be used to make a pattern into an atom. The "\z(\)" construct | |
602 * is only for syntax highlighting. | |
603 * | |
604 * atom ::= ordinary-atom | |
605 * or \( pattern \) | |
606 * or \%( pattern \) | |
607 * or \z( pattern \) | |
608 */ | |
609 static int | |
610 nfa_regatom() | |
611 { | |
612 int c; | |
613 int charclass; | |
614 int equiclass; | |
615 int collclass; | |
616 int got_coll_char; | |
617 char_u *p; | |
618 char_u *endp; | |
619 #ifdef FEAT_MBYTE | |
620 char_u *old_regparse = regparse; | |
621 #endif | |
622 int extra = 0; | |
623 int first; | |
624 int emit_range; | |
625 int negated; | |
626 int result; | |
627 int startc = -1; | |
628 int endc = -1; | |
629 int oldstartc = -1; | |
630 int cpo_lit; /* 'cpoptions' contains 'l' flag */ | |
631 int cpo_bsl; /* 'cpoptions' contains '\' flag */ | |
632 int glue; /* ID that will "glue" nodes together */ | |
633 | |
634 cpo_lit = vim_strchr(p_cpo, CPO_LITERAL) != NULL; | |
635 cpo_bsl = vim_strchr(p_cpo, CPO_BACKSL) != NULL; | |
636 | |
637 c = getchr(); | |
638 switch (c) | |
639 { | |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
640 case NUL: |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
641 syntax_error = TRUE; |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
642 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
|
643 |
4444 | 644 case Magic('^'): |
645 EMIT(NFA_BOL); | |
646 break; | |
647 | |
648 case Magic('$'): | |
649 EMIT(NFA_EOL); | |
650 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
651 had_eol = TRUE; | |
652 #endif | |
653 break; | |
654 | |
655 case Magic('<'): | |
656 EMIT(NFA_BOW); | |
657 break; | |
658 | |
659 case Magic('>'): | |
660 EMIT(NFA_EOW); | |
661 break; | |
662 | |
663 case Magic('_'): | |
664 c = no_Magic(getchr()); | |
665 if (c == '^') /* "\_^" is start-of-line */ | |
666 { | |
667 EMIT(NFA_BOL); | |
668 break; | |
669 } | |
670 if (c == '$') /* "\_$" is end-of-line */ | |
671 { | |
672 EMIT(NFA_EOL); | |
673 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
674 had_eol = TRUE; | |
675 #endif | |
676 break; | |
677 } | |
678 | |
679 extra = ADD_NL; | |
680 | |
681 /* "\_[" is collection plus newline */ | |
682 if (c == '[') | |
4517
9a2183bd8295
updated for version 7.3.1006
Bram Moolenaar <bram@vim.org>
parents:
4515
diff
changeset
|
683 goto collection; |
4444 | 684 |
685 /* "\_x" is character class plus newline */ | |
686 /*FALLTHROUGH*/ | |
687 | |
688 /* | |
689 * Character classes. | |
690 */ | |
691 case Magic('.'): | |
692 case Magic('i'): | |
693 case Magic('I'): | |
694 case Magic('k'): | |
695 case Magic('K'): | |
696 case Magic('f'): | |
697 case Magic('F'): | |
698 case Magic('p'): | |
699 case Magic('P'): | |
700 case Magic('s'): | |
701 case Magic('S'): | |
702 case Magic('d'): | |
703 case Magic('D'): | |
704 case Magic('x'): | |
705 case Magic('X'): | |
706 case Magic('o'): | |
707 case Magic('O'): | |
708 case Magic('w'): | |
709 case Magic('W'): | |
710 case Magic('h'): | |
711 case Magic('H'): | |
712 case Magic('a'): | |
713 case Magic('A'): | |
714 case Magic('l'): | |
715 case Magic('L'): | |
716 case Magic('u'): | |
717 case Magic('U'): | |
718 p = vim_strchr(classchars, no_Magic(c)); | |
719 if (p == NULL) | |
720 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
721 EMSGN("INTERNAL: Unknown character class char: %ld", c); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
722 return FAIL; |
4444 | 723 } |
724 #ifdef FEAT_MBYTE | |
725 /* When '.' is followed by a composing char ignore the dot, so that | |
726 * the composing char is matched here. */ | |
727 if (enc_utf8 && c == Magic('.') && utf_iscomposing(peekchr())) | |
728 { | |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
729 old_regparse = regparse; |
4444 | 730 c = getchr(); |
731 goto nfa_do_multibyte; | |
732 } | |
733 #endif | |
734 EMIT(nfa_classcodes[p - classchars]); | |
735 if (extra == ADD_NL) | |
736 { | |
737 EMIT(NFA_NEWL); | |
738 EMIT(NFA_OR); | |
739 regflags |= RF_HASNL; | |
740 } | |
741 break; | |
742 | |
743 case Magic('n'): | |
744 if (reg_string) | |
745 /* In a string "\n" matches a newline character. */ | |
746 EMIT(NL); | |
747 else | |
748 { | |
749 /* In buffer text "\n" matches the end of a line. */ | |
750 EMIT(NFA_NEWL); | |
751 regflags |= RF_HASNL; | |
752 } | |
753 break; | |
754 | |
755 case Magic('('): | |
756 if (nfa_reg(REG_PAREN) == FAIL) | |
757 return FAIL; /* cascaded error */ | |
758 break; | |
759 | |
760 case Magic('|'): | |
761 case Magic('&'): | |
762 case Magic(')'): | |
763 syntax_error = TRUE; | |
4450 | 764 EMSGN(_(e_misplaced), no_Magic(c)); |
4444 | 765 return FAIL; |
766 | |
767 case Magic('='): | |
768 case Magic('?'): | |
769 case Magic('+'): | |
770 case Magic('@'): | |
771 case Magic('*'): | |
772 case Magic('{'): | |
773 /* these should follow an atom, not form an atom */ | |
774 syntax_error = TRUE; | |
4450 | 775 EMSGN(_(e_misplaced), no_Magic(c)); |
4444 | 776 return FAIL; |
777 | |
778 case Magic('~'): /* previous substitute pattern */ | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
779 /* TODO: Not supported yet */ |
4444 | 780 return FAIL; |
781 | |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
782 case Magic('1'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
783 case Magic('2'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
784 case Magic('3'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
785 case Magic('4'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
786 case Magic('5'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
787 case Magic('6'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
788 case Magic('7'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
789 case Magic('8'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
790 case Magic('9'): |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
791 EMIT(NFA_BACKREF1 + (no_Magic(c) - '1')); |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
792 nfa_has_backref = TRUE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
793 break; |
4444 | 794 |
795 case Magic('z'): | |
796 c = no_Magic(getchr()); | |
797 switch (c) | |
798 { | |
799 case 's': | |
800 EMIT(NFA_ZSTART); | |
801 break; | |
802 case 'e': | |
803 EMIT(NFA_ZEND); | |
804 nfa_has_zend = TRUE; | |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
805 break; |
4444 | 806 case '1': |
807 case '2': | |
808 case '3': | |
809 case '4': | |
810 case '5': | |
811 case '6': | |
812 case '7': | |
813 case '8': | |
814 case '9': | |
815 case '(': | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
816 /* TODO: \z1...\z9 and \z( not yet supported */ |
4444 | 817 return FAIL; |
818 default: | |
819 syntax_error = TRUE; | |
4450 | 820 EMSGN(_("E867: (NFA) Unknown operator '\\z%c'"), |
4444 | 821 no_Magic(c)); |
822 return FAIL; | |
823 } | |
824 break; | |
825 | |
826 case Magic('%'): | |
827 c = no_Magic(getchr()); | |
828 switch (c) | |
829 { | |
830 /* () without a back reference */ | |
831 case '(': | |
832 if (nfa_reg(REG_NPAREN) == FAIL) | |
833 return FAIL; | |
834 EMIT(NFA_NOPEN); | |
835 break; | |
836 | |
837 case 'd': /* %d123 decimal */ | |
838 case 'o': /* %o123 octal */ | |
839 case 'x': /* %xab hex 2 */ | |
840 case 'u': /* %uabcd hex 4 */ | |
841 case 'U': /* %U1234abcd hex 8 */ | |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
842 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
843 int nr; |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
844 |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
845 switch (c) |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
846 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
847 case 'd': nr = getdecchrs(); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
848 case 'o': nr = getoctchrs(); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
849 case 'x': nr = gethexchrs(2); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
850 case 'u': nr = gethexchrs(4); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
851 case 'U': nr = gethexchrs(8); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
852 default: nr = -1; break; |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
853 } |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
854 |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
855 if (nr < 0) |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
856 EMSG2_RET_FAIL( |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
857 _("E678: Invalid character after %s%%[dxouU]"), |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
858 reg_magic == MAGIC_ALL); |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
859 /* TODO: what if a composing character follows? */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
860 EMIT(nr); |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
861 } |
4444 | 862 break; |
863 | |
864 /* Catch \%^ and \%$ regardless of where they appear in the | |
865 * pattern -- regardless of whether or not it makes sense. */ | |
866 case '^': | |
867 EMIT(NFA_BOF); | |
868 break; | |
869 | |
870 case '$': | |
871 EMIT(NFA_EOF); | |
872 break; | |
873 | |
874 case '#': | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
875 EMIT(NFA_CURSOR); |
4444 | 876 break; |
877 | |
878 case 'V': | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
879 /* TODO: not supported yet */ |
4444 | 880 return FAIL; |
881 break; | |
882 | |
883 case '[': | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
884 /* TODO: \%[abc] not supported yet */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
885 return FAIL; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
886 |
4444 | 887 default: |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
888 { |
4655
779ca415f8e1
updated for version 7.3.1075
Bram Moolenaar <bram@vim.org>
parents:
4651
diff
changeset
|
889 int n = 0; |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
890 int cmp = c; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
891 |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
892 if (c == '<' || c == '>') |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
893 c = getchr(); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
894 while (VIM_ISDIGIT(c)) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
895 { |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
896 n = n * 10 + (c - '0'); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
897 c = getchr(); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
898 } |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
899 if (c == 'l' || c == 'c' || c == 'v') |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
900 { |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
901 EMIT(n); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
902 if (c == 'l') |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
903 EMIT(cmp == '<' ? NFA_LNUM_LT : |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
904 cmp == '>' ? NFA_LNUM_GT : NFA_LNUM); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
905 else if (c == 'c') |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
906 EMIT(cmp == '<' ? NFA_COL_LT : |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
907 cmp == '>' ? NFA_COL_GT : NFA_COL); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
908 else |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
909 EMIT(cmp == '<' ? NFA_VCOL_LT : |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
910 cmp == '>' ? NFA_VCOL_GT : NFA_VCOL); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
911 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
912 } |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
913 else if (c == '\'') |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
914 /* TODO: \%'m not supported yet */ |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
915 return FAIL; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
916 } |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
917 syntax_error = TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
918 EMSGN(_("E867: (NFA) Unknown operator '\\%%%c'"), |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
919 no_Magic(c)); |
4444 | 920 return FAIL; |
921 } | |
922 break; | |
923 | |
924 case Magic('['): | |
4517
9a2183bd8295
updated for version 7.3.1006
Bram Moolenaar <bram@vim.org>
parents:
4515
diff
changeset
|
925 collection: |
4444 | 926 /* |
927 * Glue is emitted between several atoms from the []. | |
928 * It is either NFA_OR, or NFA_CONCAT. | |
929 * | |
930 * [abc] expands to 'a b NFA_OR c NFA_OR' (in postfix notation) | |
931 * [^abc] expands to 'a NFA_NOT b NFA_NOT NFA_CONCAT c NFA_NOT | |
932 * NFA_CONCAT NFA_END_NEG_RANGE NFA_CONCAT' (in postfix | |
933 * notation) | |
934 * | |
935 */ | |
936 | |
937 | |
938 /* Emit negation atoms, if needed. | |
939 * The CONCAT below merges the NOT with the previous node. */ | |
940 #define TRY_NEG() \ | |
941 if (negated == TRUE) \ | |
942 { \ | |
943 EMIT(NFA_NOT); \ | |
944 } | |
945 | |
946 /* Emit glue between important nodes : CONCAT or OR. */ | |
947 #define EMIT_GLUE() \ | |
948 if (first == FALSE) \ | |
949 EMIT(glue); \ | |
950 else \ | |
951 first = FALSE; | |
952 | |
953 p = regparse; | |
954 endp = skip_anyof(p); | |
955 if (*endp == ']') | |
956 { | |
957 /* | |
958 * Try to reverse engineer character classes. For example, | |
959 * recognize that [0-9] stands for \d and [A-Za-z_] with \h, | |
960 * and perform the necessary substitutions in the NFA. | |
961 */ | |
962 result = nfa_recognize_char_class(regparse, endp, | |
963 extra == ADD_NL); | |
964 if (result != FAIL) | |
965 { | |
966 if (result >= NFA_DIGIT && result <= NFA_NUPPER) | |
967 EMIT(result); | |
968 else /* must be char class + newline */ | |
969 { | |
970 EMIT(result - ADD_NL); | |
971 EMIT(NFA_NEWL); | |
972 EMIT(NFA_OR); | |
973 } | |
974 regparse = endp; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
975 mb_ptr_adv(regparse); |
4444 | 976 return OK; |
977 } | |
978 /* | |
979 * Failed to recognize a character class. Use the simple | |
980 * version that turns [abc] into 'a' OR 'b' OR 'c' | |
981 */ | |
982 startc = endc = oldstartc = -1; | |
983 first = TRUE; /* Emitting first atom in this sequence? */ | |
984 negated = FALSE; | |
985 glue = NFA_OR; | |
986 if (*regparse == '^') /* negated range */ | |
987 { | |
988 negated = TRUE; | |
989 glue = NFA_CONCAT; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
990 mb_ptr_adv(regparse); |
4444 | 991 } |
992 if (*regparse == '-') | |
993 { | |
994 startc = '-'; | |
995 EMIT(startc); | |
996 TRY_NEG(); | |
997 EMIT_GLUE(); | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
998 mb_ptr_adv(regparse); |
4444 | 999 } |
1000 /* Emit the OR branches for each character in the [] */ | |
1001 emit_range = FALSE; | |
1002 while (regparse < endp) | |
1003 { | |
1004 oldstartc = startc; | |
1005 startc = -1; | |
1006 got_coll_char = FALSE; | |
1007 if (*regparse == '[') | |
1008 { | |
1009 /* Check for [: :], [= =], [. .] */ | |
1010 equiclass = collclass = 0; | |
1011 charclass = get_char_class(®parse); | |
1012 if (charclass == CLASS_NONE) | |
1013 { | |
1014 equiclass = get_equi_class(®parse); | |
1015 if (equiclass == 0) | |
1016 collclass = get_coll_element(®parse); | |
1017 } | |
1018 | |
1019 /* Character class like [:alpha:] */ | |
1020 if (charclass != CLASS_NONE) | |
1021 { | |
1022 switch (charclass) | |
1023 { | |
1024 case CLASS_ALNUM: | |
1025 EMIT(NFA_CLASS_ALNUM); | |
1026 break; | |
1027 case CLASS_ALPHA: | |
1028 EMIT(NFA_CLASS_ALPHA); | |
1029 break; | |
1030 case CLASS_BLANK: | |
1031 EMIT(NFA_CLASS_BLANK); | |
1032 break; | |
1033 case CLASS_CNTRL: | |
1034 EMIT(NFA_CLASS_CNTRL); | |
1035 break; | |
1036 case CLASS_DIGIT: | |
1037 EMIT(NFA_CLASS_DIGIT); | |
1038 break; | |
1039 case CLASS_GRAPH: | |
1040 EMIT(NFA_CLASS_GRAPH); | |
1041 break; | |
1042 case CLASS_LOWER: | |
1043 EMIT(NFA_CLASS_LOWER); | |
1044 break; | |
1045 case CLASS_PRINT: | |
1046 EMIT(NFA_CLASS_PRINT); | |
1047 break; | |
1048 case CLASS_PUNCT: | |
1049 EMIT(NFA_CLASS_PUNCT); | |
1050 break; | |
1051 case CLASS_SPACE: | |
1052 EMIT(NFA_CLASS_SPACE); | |
1053 break; | |
1054 case CLASS_UPPER: | |
1055 EMIT(NFA_CLASS_UPPER); | |
1056 break; | |
1057 case CLASS_XDIGIT: | |
1058 EMIT(NFA_CLASS_XDIGIT); | |
1059 break; | |
1060 case CLASS_TAB: | |
1061 EMIT(NFA_CLASS_TAB); | |
1062 break; | |
1063 case CLASS_RETURN: | |
1064 EMIT(NFA_CLASS_RETURN); | |
1065 break; | |
1066 case CLASS_BACKSPACE: | |
1067 EMIT(NFA_CLASS_BACKSPACE); | |
1068 break; | |
1069 case CLASS_ESCAPE: | |
1070 EMIT(NFA_CLASS_ESCAPE); | |
1071 break; | |
1072 } | |
1073 TRY_NEG(); | |
1074 EMIT_GLUE(); | |
1075 continue; | |
1076 } | |
1077 /* Try equivalence class [=a=] and the like */ | |
1078 if (equiclass != 0) | |
1079 { | |
1080 result = nfa_emit_equi_class(equiclass, negated); | |
1081 if (result == FAIL) | |
1082 { | |
1083 /* should never happen */ | |
1084 EMSG_RET_FAIL(_("E868: Error building NFA with equivalence class!")); | |
1085 } | |
1086 EMIT_GLUE(); | |
1087 continue; | |
1088 } | |
1089 /* Try collating class like [. .] */ | |
1090 if (collclass != 0) | |
1091 { | |
1092 startc = collclass; /* allow [.a.]-x as a range */ | |
1093 /* Will emit the proper atom at the end of the | |
1094 * while loop. */ | |
1095 } | |
1096 } | |
1097 /* Try a range like 'a-x' or '\t-z' */ | |
1098 if (*regparse == '-') | |
1099 { | |
1100 emit_range = TRUE; | |
1101 startc = oldstartc; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1102 mb_ptr_adv(regparse); |
4444 | 1103 continue; /* reading the end of the range */ |
1104 } | |
1105 | |
1106 /* Now handle simple and escaped characters. | |
1107 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim | |
1108 * accepts "\t", "\e", etc., but only when the 'l' flag in | |
1109 * 'cpoptions' is not included. | |
1110 * Posix doesn't recognize backslash at all. | |
1111 */ | |
1112 if (*regparse == '\\' | |
1113 && !cpo_bsl | |
1114 && regparse + 1 <= endp | |
1115 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL | |
1116 || (!cpo_lit | |
1117 && vim_strchr(REGEXP_ABBR, regparse[1]) | |
1118 != NULL) | |
1119 ) | |
1120 ) | |
1121 { | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1122 mb_ptr_adv(regparse); |
4444 | 1123 |
4507
9dbbddb2ed10
updated for version 7.3.1001
Bram Moolenaar <bram@vim.org>
parents:
4503
diff
changeset
|
1124 if (*regparse == 'n') |
4444 | 1125 startc = reg_string ? NL : NFA_NEWL; |
1126 else | |
1127 if (*regparse == 'd' | |
1128 || *regparse == 'o' | |
1129 || *regparse == 'x' | |
1130 || *regparse == 'u' | |
1131 || *regparse == 'U' | |
1132 ) | |
1133 { | |
1134 /* TODO(RE) This needs more testing */ | |
1135 startc = coll_get_char(); | |
1136 got_coll_char = TRUE; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1137 mb_ptr_back(old_regparse, regparse); |
4444 | 1138 } |
1139 else | |
1140 { | |
1141 /* \r,\t,\e,\b */ | |
1142 startc = backslash_trans(*regparse); | |
1143 } | |
1144 } | |
1145 | |
1146 /* Normal printable char */ | |
1147 if (startc == -1) | |
1148 #ifdef FEAT_MBYTE | |
1149 startc = (*mb_ptr2char)(regparse); | |
1150 #else | |
1151 startc = *regparse; | |
1152 #endif | |
1153 | |
1154 /* Previous char was '-', so this char is end of range. */ | |
1155 if (emit_range) | |
1156 { | |
1157 endc = startc; startc = oldstartc; | |
1158 if (startc > endc) | |
1159 EMSG_RET_FAIL(_(e_invrange)); | |
1160 #ifdef FEAT_MBYTE | |
1161 if (has_mbyte && ((*mb_char2len)(startc) > 1 | |
1162 || (*mb_char2len)(endc) > 1)) | |
1163 { | |
1164 if (endc > startc + 256) | |
1165 EMSG_RET_FAIL(_(e_invrange)); | |
1166 /* Emit the range. "startc" was already emitted, so | |
1167 * skip it. */ | |
1168 for (c = startc + 1; c <= endc; c++) | |
1169 { | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1170 EMIT(c); |
4444 | 1171 TRY_NEG(); |
1172 EMIT_GLUE(); | |
1173 } | |
1174 emit_range = FALSE; | |
1175 } | |
1176 else | |
1177 #endif | |
1178 { | |
1179 #ifdef EBCDIC | |
1180 int alpha_only = FALSE; | |
1181 | |
1182 /* for alphabetical range skip the gaps | |
1183 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */ | |
1184 if (isalpha(startc) && isalpha(endc)) | |
1185 alpha_only = TRUE; | |
1186 #endif | |
1187 /* Emit the range. "startc" was already emitted, so | |
1188 * skip it. */ | |
1189 for (c = startc + 1; c <= endc; c++) | |
1190 #ifdef EBCDIC | |
1191 if (!alpha_only || isalpha(startc)) | |
1192 #endif | |
1193 { | |
1194 EMIT(c); | |
1195 TRY_NEG(); | |
1196 EMIT_GLUE(); | |
1197 } | |
1198 emit_range = FALSE; | |
1199 } | |
1200 } | |
1201 else | |
1202 { | |
1203 /* | |
1204 * This char (startc) is not part of a range. Just | |
1205 * emit it. | |
1206 * | |
1207 * Normally, simply emit startc. But if we get char | |
1208 * code=0 from a collating char, then replace it with | |
1209 * 0x0a. | |
1210 * | |
1211 * This is needed to completely mimic the behaviour of | |
1212 * the backtracking engine. | |
1213 */ | |
1214 if (got_coll_char == TRUE && startc == 0) | |
1215 EMIT(0x0a); | |
1216 else | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1217 EMIT(startc); |
4444 | 1218 TRY_NEG(); |
1219 EMIT_GLUE(); | |
1220 } | |
1221 | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1222 mb_ptr_adv(regparse); |
4444 | 1223 } /* while (p < endp) */ |
1224 | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1225 mb_ptr_back(old_regparse, regparse); |
4444 | 1226 if (*regparse == '-') /* if last, '-' is just a char */ |
1227 { | |
1228 EMIT('-'); | |
1229 TRY_NEG(); | |
1230 EMIT_GLUE(); | |
1231 } | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1232 mb_ptr_adv(regparse); |
4444 | 1233 |
1234 /* skip the trailing ] */ | |
1235 regparse = endp; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1236 mb_ptr_adv(regparse); |
4444 | 1237 if (negated == TRUE) |
1238 { | |
1239 /* Mark end of negated char range */ | |
1240 EMIT(NFA_END_NEG_RANGE); | |
1241 EMIT(NFA_CONCAT); | |
1242 } | |
4615
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1243 |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1244 /* \_[] also matches \n but it's not negated */ |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1245 if (extra == ADD_NL) |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1246 { |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1247 EMIT(reg_string ? NL : NFA_NEWL); |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1248 EMIT(NFA_OR); |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1249 } |
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1250 |
4444 | 1251 return OK; |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1252 } /* if exists closing ] */ |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1253 |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1254 if (reg_strict) |
4444 | 1255 { |
1256 syntax_error = TRUE; | |
1257 EMSG_RET_FAIL(_(e_missingbracket)); | |
1258 } | |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1259 /* FALLTHROUGH */ |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1260 |
4444 | 1261 default: |
1262 { | |
1263 #ifdef FEAT_MBYTE | |
1264 int plen; | |
1265 | |
1266 nfa_do_multibyte: | |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1267 /* 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
|
1268 if (enc_utf8 && ((*mb_char2len)(c) |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1269 != (plen = (*mb_ptr2len)(old_regparse)) |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1270 || utf_iscomposing(c))) |
4444 | 1271 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
1272 int i = 0; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
1273 |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
1274 /* 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
|
1275 * or more composing characters. |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1276 * 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
|
1277 * 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
|
1278 * NFA_END_COMPOSING is the ). Note that right now we are |
4444 | 1279 * building the postfix form, not the NFA itself; |
1280 * 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
|
1281 * 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
|
1282 for (;;) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1283 { |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1284 EMIT(c); |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1285 if (i > 0) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1286 EMIT(NFA_CONCAT); |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1287 if ((i += utf_char2len(c)) >= plen) |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1288 break; |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1289 c = utf_ptr2char(old_regparse + i); |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1290 } |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1291 EMIT(NFA_COMPOSING); |
4444 | 1292 regparse = old_regparse + plen; |
1293 } | |
1294 else | |
1295 #endif | |
1296 { | |
1297 c = no_Magic(c); | |
1298 EMIT(c); | |
1299 } | |
1300 return OK; | |
1301 } | |
1302 } | |
1303 | |
1304 #undef TRY_NEG | |
1305 #undef EMIT_GLUE | |
1306 | |
1307 return OK; | |
1308 } | |
1309 | |
1310 /* | |
1311 * Parse something followed by possible [*+=]. | |
1312 * | |
1313 * A piece is an atom, possibly followed by a multi, an indication of how many | |
1314 * times the atom can be matched. Example: "a*" matches any sequence of "a" | |
1315 * characters: "", "a", "aa", etc. | |
1316 * | |
1317 * piece ::= atom | |
1318 * or atom multi | |
1319 */ | |
1320 static int | |
1321 nfa_regpiece() | |
1322 { | |
1323 int i; | |
1324 int op; | |
1325 int ret; | |
1326 long minval, maxval; | |
1327 int greedy = TRUE; /* Braces are prefixed with '-' ? */ | |
1328 char_u *old_regparse, *new_regparse; | |
1329 int c2; | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1330 int old_post_pos; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1331 int my_post_start; |
4444 | 1332 int old_regnpar; |
1333 int quest; | |
1334 | |
1335 /* Save the current position in the regexp, so that we can use it if | |
1336 * <atom>{m,n} is next. */ | |
1337 old_regparse = regparse; | |
1338 /* Save current number of open parenthesis, so we can use it if | |
1339 * <atom>{m,n} is next */ | |
1340 old_regnpar = regnpar; | |
1341 /* store current pos in the postfix form, for \{m,n} involving 0s */ | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1342 my_post_start = (int)(post_ptr - post_start); |
4444 | 1343 |
1344 ret = nfa_regatom(); | |
1345 if (ret == FAIL) | |
1346 return FAIL; /* cascaded error */ | |
1347 | |
1348 op = peekchr(); | |
1349 if (re_multi_type(op) == NOT_MULTI) | |
1350 return OK; | |
1351 | |
1352 skipchr(); | |
1353 switch (op) | |
1354 { | |
1355 case Magic('*'): | |
1356 EMIT(NFA_STAR); | |
1357 break; | |
1358 | |
1359 case Magic('+'): | |
1360 /* | |
1361 * Trick: Normally, (a*)\+ would match the whole input "aaa". The | |
1362 * first and only submatch would be "aaa". But the backtracking | |
1363 * engine interprets the plus as "try matching one more time", and | |
1364 * a* matches a second time at the end of the input, the empty | |
1365 * string. | |
1366 * The submatch will the empty string. | |
1367 * | |
1368 * In order to be consistent with the old engine, we disable | |
1369 * NFA_PLUS, and replace <atom>+ with <atom><atom>* | |
1370 */ | |
1371 /* EMIT(NFA_PLUS); */ | |
1372 regnpar = old_regnpar; | |
1373 regparse = old_regparse; | |
1374 curchr = -1; | |
1375 if (nfa_regatom() == FAIL) | |
1376 return FAIL; | |
1377 EMIT(NFA_STAR); | |
1378 EMIT(NFA_CONCAT); | |
1379 skipchr(); /* skip the \+ */ | |
1380 break; | |
1381 | |
1382 case Magic('@'): | |
1383 op = no_Magic(getchr()); | |
1384 switch(op) | |
1385 { | |
1386 case '=': | |
1387 EMIT(NFA_PREV_ATOM_NO_WIDTH); | |
1388 break; | |
4661
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
1389 case '!': |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
1390 EMIT(NFA_PREV_ATOM_NO_WIDTH_NEG); |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
1391 break; |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1392 case '0': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1393 case '1': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1394 case '2': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1395 case '3': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1396 case '4': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1397 case '5': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1398 case '6': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1399 case '7': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1400 case '8': |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1401 case '9': |
4444 | 1402 case '<': |
1403 case '>': | |
1404 /* Not supported yet */ | |
1405 return FAIL; | |
1406 default: | |
1407 syntax_error = TRUE; | |
4450 | 1408 EMSGN(_("E869: (NFA) Unknown operator '\\@%c'"), op); |
4444 | 1409 return FAIL; |
1410 } | |
1411 break; | |
1412 | |
1413 case Magic('?'): | |
1414 case Magic('='): | |
1415 EMIT(NFA_QUEST); | |
1416 break; | |
1417 | |
1418 case Magic('{'): | |
1419 /* a{2,5} will expand to 'aaa?a?a?' | |
1420 * a{-1,3} will expand to 'aa??a??', where ?? is the nongreedy | |
1421 * version of '?' | |
1422 * \v(ab){2,3} will expand to '(ab)(ab)(ab)?', where all the | |
1423 * parenthesis have the same id | |
1424 */ | |
1425 | |
1426 greedy = TRUE; | |
1427 c2 = peekchr(); | |
1428 if (c2 == '-' || c2 == Magic('-')) | |
1429 { | |
1430 skipchr(); | |
1431 greedy = FALSE; | |
1432 } | |
1433 if (!read_limits(&minval, &maxval)) | |
1434 { | |
1435 syntax_error = TRUE; | |
1436 EMSG_RET_FAIL(_("E870: (NFA regexp) Error reading repetition limits")); | |
1437 } | |
1438 /* <atom>{0,inf}, <atom>{0,} and <atom>{} are equivalent to | |
1439 * <atom>* */ | |
1440 if (minval == 0 && maxval == MAX_LIMIT && greedy) | |
1441 { | |
1442 EMIT(NFA_STAR); | |
1443 break; | |
1444 } | |
1445 | |
1446 if (maxval > NFA_BRACES_MAXLIMIT) | |
1447 { | |
1448 /* This would yield a huge automaton and use too much memory. | |
1449 * Revert to old engine */ | |
1450 return FAIL; | |
1451 } | |
1452 | |
1453 /* Special case: x{0} or x{-0} */ | |
1454 if (maxval == 0) | |
1455 { | |
1456 /* Ignore result of previous call to nfa_regatom() */ | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1457 post_ptr = post_start + my_post_start; |
4444 | 1458 /* NFA_SKIP_CHAR has 0-length and works everywhere */ |
1459 EMIT(NFA_SKIP_CHAR); | |
1460 return OK; | |
1461 } | |
1462 | |
1463 /* Ignore previous call to nfa_regatom() */ | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1464 post_ptr = post_start + my_post_start; |
4444 | 1465 /* Save pos after the repeated atom and the \{} */ |
1466 new_regparse = regparse; | |
1467 | |
1468 quest = (greedy == TRUE? NFA_QUEST : NFA_QUEST_NONGREEDY); | |
1469 for (i = 0; i < maxval; i++) | |
1470 { | |
1471 /* Goto beginning of the repeated atom */ | |
1472 regparse = old_regparse; | |
1473 curchr = -1; | |
1474 /* Restore count of parenthesis */ | |
1475 regnpar = old_regnpar; | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1476 old_post_pos = (int)(post_ptr - post_start); |
4444 | 1477 if (nfa_regatom() == FAIL) |
1478 return FAIL; | |
1479 /* after "minval" times, atoms are optional */ | |
1480 if (i + 1 > minval) | |
1481 EMIT(quest); | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1482 if (old_post_pos != my_post_start) |
4444 | 1483 EMIT(NFA_CONCAT); |
1484 } | |
1485 | |
1486 /* Go to just after the repeated atom and the \{} */ | |
1487 regparse = new_regparse; | |
1488 curchr = -1; | |
1489 | |
1490 break; | |
1491 | |
1492 | |
1493 default: | |
1494 break; | |
1495 } /* end switch */ | |
1496 | |
1497 if (re_multi_type(peekchr()) != NOT_MULTI) | |
1498 { | |
1499 /* Can't have a multi follow a multi. */ | |
1500 syntax_error = TRUE; | |
1501 EMSG_RET_FAIL(_("E871: (NFA regexp) Can't have a multi follow a multi !")); | |
1502 } | |
1503 | |
1504 return OK; | |
1505 } | |
1506 | |
1507 /* | |
1508 * Parse one or more pieces, concatenated. It matches a match for the | |
1509 * first piece, followed by a match for the second piece, etc. Example: | |
1510 * "f[0-9]b", first matches "f", then a digit and then "b". | |
1511 * | |
1512 * concat ::= piece | |
1513 * or piece piece | |
1514 * or piece piece piece | |
1515 * etc. | |
1516 */ | |
1517 static int | |
1518 nfa_regconcat() | |
1519 { | |
1520 int cont = TRUE; | |
1521 int first = TRUE; | |
1522 | |
1523 while (cont) | |
1524 { | |
1525 switch (peekchr()) | |
1526 { | |
1527 case NUL: | |
1528 case Magic('|'): | |
1529 case Magic('&'): | |
1530 case Magic(')'): | |
1531 cont = FALSE; | |
1532 break; | |
1533 | |
1534 case Magic('Z'): | |
1535 #ifdef FEAT_MBYTE | |
1536 regflags |= RF_ICOMBINE; | |
1537 #endif | |
1538 skipchr_keepstart(); | |
1539 break; | |
1540 case Magic('c'): | |
1541 regflags |= RF_ICASE; | |
1542 skipchr_keepstart(); | |
1543 break; | |
1544 case Magic('C'): | |
1545 regflags |= RF_NOICASE; | |
1546 skipchr_keepstart(); | |
1547 break; | |
1548 case Magic('v'): | |
1549 reg_magic = MAGIC_ALL; | |
1550 skipchr_keepstart(); | |
1551 curchr = -1; | |
1552 break; | |
1553 case Magic('m'): | |
1554 reg_magic = MAGIC_ON; | |
1555 skipchr_keepstart(); | |
1556 curchr = -1; | |
1557 break; | |
1558 case Magic('M'): | |
1559 reg_magic = MAGIC_OFF; | |
1560 skipchr_keepstart(); | |
1561 curchr = -1; | |
1562 break; | |
1563 case Magic('V'): | |
1564 reg_magic = MAGIC_NONE; | |
1565 skipchr_keepstart(); | |
1566 curchr = -1; | |
1567 break; | |
1568 | |
1569 default: | |
1570 if (nfa_regpiece() == FAIL) | |
1571 return FAIL; | |
1572 if (first == FALSE) | |
1573 EMIT(NFA_CONCAT); | |
1574 else | |
1575 first = FALSE; | |
1576 break; | |
1577 } | |
1578 } | |
1579 | |
1580 return OK; | |
1581 } | |
1582 | |
1583 /* | |
1584 * Parse a branch, one or more concats, separated by "\&". It matches the | |
1585 * last concat, but only if all the preceding concats also match at the same | |
1586 * position. Examples: | |
1587 * "foobeep\&..." matches "foo" in "foobeep". | |
1588 * ".*Peter\&.*Bob" matches in a line containing both "Peter" and "Bob" | |
1589 * | |
1590 * branch ::= concat | |
1591 * or concat \& concat | |
1592 * or concat \& concat \& concat | |
1593 * etc. | |
1594 */ | |
1595 static int | |
1596 nfa_regbranch() | |
1597 { | |
1598 int ch; | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1599 int old_post_pos; |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1600 |
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1601 old_post_pos = (int)(post_ptr - post_start); |
4444 | 1602 |
1603 /* First branch, possibly the only one */ | |
1604 if (nfa_regconcat() == FAIL) | |
1605 return FAIL; | |
1606 | |
1607 ch = peekchr(); | |
1608 /* Try next concats */ | |
1609 while (ch == Magic('&')) | |
1610 { | |
1611 skipchr(); | |
1612 EMIT(NFA_NOPEN); | |
1613 EMIT(NFA_PREV_ATOM_NO_WIDTH); | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1614 old_post_pos = (int)(post_ptr - post_start); |
4444 | 1615 if (nfa_regconcat() == FAIL) |
1616 return FAIL; | |
1617 /* if concat is empty, skip a input char. But do emit a node */ | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1618 if (old_post_pos == (int)(post_ptr - post_start)) |
4444 | 1619 EMIT(NFA_SKIP_CHAR); |
1620 EMIT(NFA_CONCAT); | |
1621 ch = peekchr(); | |
1622 } | |
1623 | |
1624 /* Even if a branch is empty, emit one node for it */ | |
4651
f10f63aaec5c
updated for version 7.3.1073
Bram Moolenaar <bram@vim.org>
parents:
4649
diff
changeset
|
1625 if (old_post_pos == (int)(post_ptr - post_start)) |
4444 | 1626 EMIT(NFA_SKIP_CHAR); |
1627 | |
1628 return OK; | |
1629 } | |
1630 | |
1631 /* | |
1632 * Parse a pattern, one or more branches, separated by "\|". It matches | |
1633 * anything that matches one of the branches. Example: "foo\|beep" matches | |
1634 * "foo" and matches "beep". If more than one branch matches, the first one | |
1635 * is used. | |
1636 * | |
1637 * pattern ::= branch | |
1638 * or branch \| branch | |
1639 * or branch \| branch \| branch | |
1640 * etc. | |
1641 */ | |
1642 static int | |
1643 nfa_reg(paren) | |
1644 int paren; /* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */ | |
1645 { | |
1646 int parno = 0; | |
1647 | |
1648 #ifdef FEAT_SYN_HL | |
1649 #endif | |
1650 if (paren == REG_PAREN) | |
1651 { | |
1652 if (regnpar >= NSUBEXP) /* Too many `(' */ | |
1653 { | |
1654 syntax_error = TRUE; | |
1655 EMSG_RET_FAIL(_("E872: (NFA regexp) Too many '('")); | |
1656 } | |
1657 parno = regnpar++; | |
1658 } | |
1659 | |
1660 if (nfa_regbranch() == FAIL) | |
1661 return FAIL; /* cascaded error */ | |
1662 | |
1663 while (peekchr() == Magic('|')) | |
1664 { | |
1665 skipchr(); | |
1666 if (nfa_regbranch() == FAIL) | |
1667 return FAIL; /* cascaded error */ | |
1668 EMIT(NFA_OR); | |
1669 } | |
1670 | |
1671 /* Check for proper termination. */ | |
1672 if (paren != REG_NOPAREN && getchr() != Magic(')')) | |
1673 { | |
1674 syntax_error = TRUE; | |
1675 if (paren == REG_NPAREN) | |
1676 EMSG2_RET_FAIL(_(e_unmatchedpp), reg_magic == MAGIC_ALL); | |
1677 else | |
1678 EMSG2_RET_FAIL(_(e_unmatchedp), reg_magic == MAGIC_ALL); | |
1679 } | |
1680 else if (paren == REG_NOPAREN && peekchr() != NUL) | |
1681 { | |
1682 syntax_error = TRUE; | |
1683 if (peekchr() == Magic(')')) | |
1684 EMSG2_RET_FAIL(_(e_unmatchedpar), reg_magic == MAGIC_ALL); | |
1685 else | |
1686 EMSG_RET_FAIL(_("E873: (NFA regexp) proper termination error")); | |
1687 } | |
1688 /* | |
1689 * Here we set the flag allowing back references to this set of | |
1690 * parentheses. | |
1691 */ | |
1692 if (paren == REG_PAREN) | |
1693 { | |
1694 had_endbrace[parno] = TRUE; /* have seen the close paren */ | |
1695 EMIT(NFA_MOPEN + parno); | |
1696 } | |
1697 | |
1698 return OK; | |
1699 } | |
1700 | |
1701 #ifdef DEBUG | |
1702 static char_u code[50]; | |
1703 | |
1704 static void | |
1705 nfa_set_code(c) | |
1706 int c; | |
1707 { | |
1708 int addnl = FALSE; | |
1709 | |
1710 if (c >= NFA_FIRST_NL && c <= NFA_LAST_NL) | |
1711 { | |
1712 addnl = TRUE; | |
1713 c -= ADD_NL; | |
1714 } | |
1715 | |
1716 STRCPY(code, ""); | |
1717 switch (c) | |
1718 { | |
1719 case NFA_MATCH: STRCPY(code, "NFA_MATCH "); break; | |
1720 case NFA_SPLIT: STRCPY(code, "NFA_SPLIT "); break; | |
1721 case NFA_CONCAT: STRCPY(code, "NFA_CONCAT "); break; | |
1722 case NFA_NEWL: STRCPY(code, "NFA_NEWL "); break; | |
1723 case NFA_ZSTART: STRCPY(code, "NFA_ZSTART"); break; | |
1724 case NFA_ZEND: STRCPY(code, "NFA_ZEND"); break; | |
1725 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1726 case NFA_BACKREF1: STRCPY(code, "NFA_BACKREF1"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1727 case NFA_BACKREF2: STRCPY(code, "NFA_BACKREF2"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1728 case NFA_BACKREF3: STRCPY(code, "NFA_BACKREF3"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1729 case NFA_BACKREF4: STRCPY(code, "NFA_BACKREF4"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1730 case NFA_BACKREF5: STRCPY(code, "NFA_BACKREF5"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1731 case NFA_BACKREF6: STRCPY(code, "NFA_BACKREF6"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1732 case NFA_BACKREF7: STRCPY(code, "NFA_BACKREF7"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1733 case NFA_BACKREF8: STRCPY(code, "NFA_BACKREF8"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1734 case NFA_BACKREF9: STRCPY(code, "NFA_BACKREF9"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1735 case NFA_SKIP: STRCPY(code, "NFA_SKIP"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1736 |
4444 | 1737 case NFA_PREV_ATOM_NO_WIDTH: |
1738 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
|
1739 case NFA_PREV_ATOM_NO_WIDTH_NEG: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
1740 STRCPY(code, "NFA_PREV_ATOM_NO_WIDTH_NEG"); break; |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
1741 case NFA_NOPEN: STRCPY(code, "NFA_NOPEN"); break; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
1742 case NFA_NCLOSE: STRCPY(code, "NFA_NCLOSE"); break; |
4444 | 1743 case NFA_START_INVISIBLE: STRCPY(code, "NFA_START_INVISIBLE"); break; |
1744 case NFA_END_INVISIBLE: STRCPY(code, "NFA_END_INVISIBLE"); break; | |
1745 | |
1746 case NFA_COMPOSING: STRCPY(code, "NFA_COMPOSING"); break; | |
1747 case NFA_END_COMPOSING: STRCPY(code, "NFA_END_COMPOSING"); break; | |
1748 | |
1749 case NFA_MOPEN + 0: | |
1750 case NFA_MOPEN + 1: | |
1751 case NFA_MOPEN + 2: | |
1752 case NFA_MOPEN + 3: | |
1753 case NFA_MOPEN + 4: | |
1754 case NFA_MOPEN + 5: | |
1755 case NFA_MOPEN + 6: | |
1756 case NFA_MOPEN + 7: | |
1757 case NFA_MOPEN + 8: | |
1758 case NFA_MOPEN + 9: | |
1759 STRCPY(code, "NFA_MOPEN(x)"); | |
1760 code[10] = c - NFA_MOPEN + '0'; | |
1761 break; | |
1762 case NFA_MCLOSE + 0: | |
1763 case NFA_MCLOSE + 1: | |
1764 case NFA_MCLOSE + 2: | |
1765 case NFA_MCLOSE + 3: | |
1766 case NFA_MCLOSE + 4: | |
1767 case NFA_MCLOSE + 5: | |
1768 case NFA_MCLOSE + 6: | |
1769 case NFA_MCLOSE + 7: | |
1770 case NFA_MCLOSE + 8: | |
1771 case NFA_MCLOSE + 9: | |
1772 STRCPY(code, "NFA_MCLOSE(x)"); | |
1773 code[11] = c - NFA_MCLOSE + '0'; | |
1774 break; | |
1775 case NFA_EOL: STRCPY(code, "NFA_EOL "); break; | |
1776 case NFA_BOL: STRCPY(code, "NFA_BOL "); break; | |
1777 case NFA_EOW: STRCPY(code, "NFA_EOW "); break; | |
1778 case NFA_BOW: STRCPY(code, "NFA_BOW "); break; | |
4671
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
1779 case NFA_EOF: STRCPY(code, "NFA_EOF "); break; |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
1780 case NFA_BOF: STRCPY(code, "NFA_BOF "); break; |
4444 | 1781 case NFA_STAR: STRCPY(code, "NFA_STAR "); break; |
1782 case NFA_PLUS: STRCPY(code, "NFA_PLUS "); break; | |
1783 case NFA_NOT: STRCPY(code, "NFA_NOT "); break; | |
1784 case NFA_SKIP_CHAR: STRCPY(code, "NFA_SKIP_CHAR"); break; | |
1785 case NFA_OR: STRCPY(code, "NFA_OR"); break; | |
1786 case NFA_QUEST: STRCPY(code, "NFA_QUEST"); break; | |
1787 case NFA_QUEST_NONGREEDY: STRCPY(code, "NFA_QUEST_NON_GREEDY"); break; | |
1788 case NFA_END_NEG_RANGE: STRCPY(code, "NFA_END_NEG_RANGE"); break; | |
1789 case NFA_CLASS_ALNUM: STRCPY(code, "NFA_CLASS_ALNUM"); break; | |
1790 case NFA_CLASS_ALPHA: STRCPY(code, "NFA_CLASS_ALPHA"); break; | |
1791 case NFA_CLASS_BLANK: STRCPY(code, "NFA_CLASS_BLANK"); break; | |
1792 case NFA_CLASS_CNTRL: STRCPY(code, "NFA_CLASS_CNTRL"); break; | |
1793 case NFA_CLASS_DIGIT: STRCPY(code, "NFA_CLASS_DIGIT"); break; | |
1794 case NFA_CLASS_GRAPH: STRCPY(code, "NFA_CLASS_GRAPH"); break; | |
1795 case NFA_CLASS_LOWER: STRCPY(code, "NFA_CLASS_LOWER"); break; | |
1796 case NFA_CLASS_PRINT: STRCPY(code, "NFA_CLASS_PRINT"); break; | |
1797 case NFA_CLASS_PUNCT: STRCPY(code, "NFA_CLASS_PUNCT"); break; | |
1798 case NFA_CLASS_SPACE: STRCPY(code, "NFA_CLASS_SPACE"); break; | |
1799 case NFA_CLASS_UPPER: STRCPY(code, "NFA_CLASS_UPPER"); break; | |
1800 case NFA_CLASS_XDIGIT: STRCPY(code, "NFA_CLASS_XDIGIT"); break; | |
1801 case NFA_CLASS_TAB: STRCPY(code, "NFA_CLASS_TAB"); break; | |
1802 case NFA_CLASS_RETURN: STRCPY(code, "NFA_CLASS_RETURN"); break; | |
1803 case NFA_CLASS_BACKSPACE: STRCPY(code, "NFA_CLASS_BACKSPACE"); break; | |
1804 case NFA_CLASS_ESCAPE: STRCPY(code, "NFA_CLASS_ESCAPE"); break; | |
1805 | |
1806 case NFA_ANY: STRCPY(code, "NFA_ANY"); break; | |
1807 case NFA_IDENT: STRCPY(code, "NFA_IDENT"); break; | |
1808 case NFA_SIDENT:STRCPY(code, "NFA_SIDENT"); break; | |
1809 case NFA_KWORD: STRCPY(code, "NFA_KWORD"); break; | |
1810 case NFA_SKWORD:STRCPY(code, "NFA_SKWORD"); break; | |
1811 case NFA_FNAME: STRCPY(code, "NFA_FNAME"); break; | |
1812 case NFA_SFNAME:STRCPY(code, "NFA_SFNAME"); break; | |
1813 case NFA_PRINT: STRCPY(code, "NFA_PRINT"); break; | |
1814 case NFA_SPRINT:STRCPY(code, "NFA_SPRINT"); break; | |
1815 case NFA_WHITE: STRCPY(code, "NFA_WHITE"); break; | |
1816 case NFA_NWHITE:STRCPY(code, "NFA_NWHITE"); break; | |
1817 case NFA_DIGIT: STRCPY(code, "NFA_DIGIT"); break; | |
1818 case NFA_NDIGIT:STRCPY(code, "NFA_NDIGIT"); break; | |
1819 case NFA_HEX: STRCPY(code, "NFA_HEX"); break; | |
1820 case NFA_NHEX: STRCPY(code, "NFA_NHEX"); break; | |
1821 case NFA_OCTAL: STRCPY(code, "NFA_OCTAL"); break; | |
1822 case NFA_NOCTAL:STRCPY(code, "NFA_NOCTAL"); break; | |
1823 case NFA_WORD: STRCPY(code, "NFA_WORD"); break; | |
1824 case NFA_NWORD: STRCPY(code, "NFA_NWORD"); break; | |
1825 case NFA_HEAD: STRCPY(code, "NFA_HEAD"); break; | |
1826 case NFA_NHEAD: STRCPY(code, "NFA_NHEAD"); break; | |
1827 case NFA_ALPHA: STRCPY(code, "NFA_ALPHA"); break; | |
1828 case NFA_NALPHA:STRCPY(code, "NFA_NALPHA"); break; | |
1829 case NFA_LOWER: STRCPY(code, "NFA_LOWER"); break; | |
1830 case NFA_NLOWER:STRCPY(code, "NFA_NLOWER"); break; | |
1831 case NFA_UPPER: STRCPY(code, "NFA_UPPER"); break; | |
1832 case NFA_NUPPER:STRCPY(code, "NFA_NUPPER"); break; | |
1833 | |
1834 default: | |
1835 STRCPY(code, "CHAR(x)"); | |
1836 code[5] = c; | |
1837 } | |
1838 | |
1839 if (addnl == TRUE) | |
1840 STRCAT(code, " + NEWLINE "); | |
1841 | |
1842 } | |
1843 | |
1844 #ifdef ENABLE_LOG | |
1845 static FILE *log_fd; | |
1846 | |
1847 /* | |
1848 * Print the postfix notation of the current regexp. | |
1849 */ | |
1850 static void | |
1851 nfa_postfix_dump(expr, retval) | |
1852 char_u *expr; | |
1853 int retval; | |
1854 { | |
1855 int *p; | |
1856 FILE *f; | |
1857 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
1858 f = fopen(NFA_REGEXP_DUMP_LOG, "a"); |
4444 | 1859 if (f != NULL) |
1860 { | |
1861 fprintf(f, "\n-------------------------\n"); | |
1862 if (retval == FAIL) | |
1863 fprintf(f, ">>> NFA engine failed ... \n"); | |
1864 else if (retval == OK) | |
1865 fprintf(f, ">>> NFA engine succeeded !\n"); | |
1866 fprintf(f, "Regexp: \"%s\"\nPostfix notation (char): \"", expr); | |
4464 | 1867 for (p = post_start; *p && p < post_end; p++) |
4444 | 1868 { |
1869 nfa_set_code(*p); | |
1870 fprintf(f, "%s, ", code); | |
1871 } | |
1872 fprintf(f, "\"\nPostfix notation (int): "); | |
4464 | 1873 for (p = post_start; *p && p < post_end; p++) |
4444 | 1874 fprintf(f, "%d ", *p); |
1875 fprintf(f, "\n\n"); | |
1876 fclose(f); | |
1877 } | |
1878 } | |
1879 | |
1880 /* | |
1881 * Print the NFA starting with a root node "state". | |
1882 */ | |
1883 static void | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1884 nfa_print_state(debugf, state) |
4444 | 1885 FILE *debugf; |
1886 nfa_state_T *state; | |
1887 { | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1888 garray_T indent; |
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 ga_init2(&indent, 1, 64); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1891 ga_append(&indent, '\0'); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1892 nfa_print_state2(debugf, state, &indent); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1893 ga_clear(&indent); |
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 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1896 static void |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1897 nfa_print_state2(debugf, state, indent) |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1898 FILE *debugf; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1899 nfa_state_T *state; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1900 garray_T *indent; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1901 { |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1902 char_u *p; |
4444 | 1903 |
1904 if (state == NULL) | |
1905 return; | |
1906 | |
1907 fprintf(debugf, "(%2d)", abs(state->id)); | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1908 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1909 /* Output indent */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1910 p = (char_u *)indent->ga_data; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1911 if (indent->ga_len >= 3) |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1912 { |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1913 int last = indent->ga_len - 3; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1914 char_u save[2]; |
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 STRNCPY(save, &p[last], 2); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1917 STRNCPY(&p[last], "+-", 2); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1918 fprintf(debugf, " %s", p); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1919 STRNCPY(&p[last], save, 2); |
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 else |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1922 fprintf(debugf, " %s", p); |
4444 | 1923 |
1924 nfa_set_code(state->c); | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1925 fprintf(debugf, "%s%s (%d) (id=%d)\n", |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1926 state->negated ? "NOT " : "", code, state->c, abs(state->id)); |
4444 | 1927 if (state->id < 0) |
1928 return; | |
1929 | |
1930 state->id = abs(state->id) * -1; | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1931 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1932 /* grow indent for state->out */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1933 indent->ga_len -= 1; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1934 if (state->out1) |
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1935 ga_concat(indent, (char_u *)"| "); |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1936 else |
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1937 ga_concat(indent, (char_u *)" "); |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1938 ga_append(indent, '\0'); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1939 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1940 nfa_print_state2(debugf, state->out, indent); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1941 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1942 /* replace last part of indent for state->out1 */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1943 indent->ga_len -= 3; |
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1944 ga_concat(indent, (char_u *)" "); |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1945 ga_append(indent, '\0'); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1946 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1947 nfa_print_state2(debugf, state->out1, indent); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1948 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1949 /* shrink indent */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1950 indent->ga_len -= 3; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1951 ga_append(indent, '\0'); |
4444 | 1952 } |
1953 | |
1954 /* | |
1955 * Print the NFA state machine. | |
1956 */ | |
1957 static void | |
1958 nfa_dump(prog) | |
1959 nfa_regprog_T *prog; | |
1960 { | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
1961 FILE *debugf = fopen(NFA_REGEXP_DUMP_LOG, "a"); |
4444 | 1962 |
1963 if (debugf != NULL) | |
1964 { | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1965 nfa_print_state(debugf, prog->start); |
4444 | 1966 fclose(debugf); |
1967 } | |
1968 } | |
1969 #endif /* ENABLE_LOG */ | |
1970 #endif /* DEBUG */ | |
1971 | |
1972 /* | |
1973 * Parse r.e. @expr and convert it into postfix form. | |
1974 * Return the postfix string on success, NULL otherwise. | |
1975 */ | |
1976 static int * | |
1977 re2post() | |
1978 { | |
1979 if (nfa_reg(REG_NOPAREN) == FAIL) | |
1980 return NULL; | |
1981 EMIT(NFA_MOPEN); | |
1982 return post_start; | |
1983 } | |
1984 | |
1985 /* NB. Some of the code below is inspired by Russ's. */ | |
1986 | |
1987 /* | |
1988 * Represents an NFA state plus zero or one or two arrows exiting. | |
1989 * if c == MATCH, no arrows out; matching state. | |
1990 * If c == SPLIT, unlabeled arrows to out and out1 (if != NULL). | |
1991 * If c < 256, labeled arrow with character c to out. | |
1992 */ | |
1993 | |
1994 static nfa_state_T *state_ptr; /* points to nfa_prog->state */ | |
1995 | |
1996 /* | |
1997 * Allocate and initialize nfa_state_T. | |
1998 */ | |
1999 static nfa_state_T * | |
2000 new_state(c, out, out1) | |
2001 int c; | |
2002 nfa_state_T *out; | |
2003 nfa_state_T *out1; | |
2004 { | |
2005 nfa_state_T *s; | |
2006 | |
2007 if (istate >= nstate) | |
2008 return NULL; | |
2009 | |
2010 s = &state_ptr[istate++]; | |
2011 | |
2012 s->c = c; | |
2013 s->out = out; | |
2014 s->out1 = out1; | |
2015 | |
2016 s->id = istate; | |
2017 s->lastlist = 0; | |
2018 s->negated = FALSE; | |
2019 | |
2020 return s; | |
2021 } | |
2022 | |
2023 /* | |
2024 * A partially built NFA without the matching state filled in. | |
2025 * Frag_T.start points at the start state. | |
2026 * Frag_T.out is a list of places that need to be set to the | |
2027 * next state for this fragment. | |
2028 */ | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2029 |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2030 /* 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
|
2031 * uninitialized, we use the pointers themselves |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2032 * as storage for the Ptrlists. */ |
4444 | 2033 typedef union Ptrlist Ptrlist; |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2034 union Ptrlist |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2035 { |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2036 Ptrlist *next; |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2037 nfa_state_T *s; |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2038 }; |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2039 |
4444 | 2040 struct Frag |
2041 { | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2042 nfa_state_T *start; |
4444 | 2043 Ptrlist *out; |
2044 }; | |
2045 typedef struct Frag Frag_T; | |
2046 | |
2047 static Frag_T frag __ARGS((nfa_state_T *start, Ptrlist *out)); | |
2048 static Ptrlist *list1 __ARGS((nfa_state_T **outp)); | |
2049 static void patch __ARGS((Ptrlist *l, nfa_state_T *s)); | |
2050 static Ptrlist *append __ARGS((Ptrlist *l1, Ptrlist *l2)); | |
2051 static void st_push __ARGS((Frag_T s, Frag_T **p, Frag_T *stack_end)); | |
2052 static Frag_T st_pop __ARGS((Frag_T **p, Frag_T *stack)); | |
2053 | |
2054 /* | |
4456 | 2055 * Initialize a Frag_T struct and return it. |
4444 | 2056 */ |
2057 static Frag_T | |
2058 frag(start, out) | |
2059 nfa_state_T *start; | |
2060 Ptrlist *out; | |
2061 { | |
4456 | 2062 Frag_T n; |
2063 | |
2064 n.start = start; | |
2065 n.out = out; | |
4444 | 2066 return n; |
2067 } | |
2068 | |
2069 /* | |
2070 * Create singleton list containing just outp. | |
2071 */ | |
2072 static Ptrlist * | |
2073 list1(outp) | |
2074 nfa_state_T **outp; | |
2075 { | |
2076 Ptrlist *l; | |
2077 | |
2078 l = (Ptrlist *)outp; | |
2079 l->next = NULL; | |
2080 return l; | |
2081 } | |
2082 | |
2083 /* | |
2084 * Patch the list of states at out to point to start. | |
2085 */ | |
2086 static void | |
2087 patch(l, s) | |
2088 Ptrlist *l; | |
2089 nfa_state_T *s; | |
2090 { | |
2091 Ptrlist *next; | |
2092 | |
2093 for (; l; l = next) | |
2094 { | |
2095 next = l->next; | |
2096 l->s = s; | |
2097 } | |
2098 } | |
2099 | |
2100 | |
2101 /* | |
2102 * Join the two lists l1 and l2, returning the combination. | |
2103 */ | |
2104 static Ptrlist * | |
2105 append(l1, l2) | |
2106 Ptrlist *l1; | |
2107 Ptrlist *l2; | |
2108 { | |
2109 Ptrlist *oldl1; | |
2110 | |
2111 oldl1 = l1; | |
2112 while (l1->next) | |
2113 l1 = l1->next; | |
2114 l1->next = l2; | |
2115 return oldl1; | |
2116 } | |
2117 | |
2118 /* | |
2119 * Stack used for transforming postfix form into NFA. | |
2120 */ | |
2121 static Frag_T empty; | |
2122 | |
2123 static void | |
2124 st_error(postfix, end, p) | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2125 int *postfix UNUSED; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2126 int *end UNUSED; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2127 int *p UNUSED; |
4444 | 2128 { |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2129 #ifdef NFA_REGEXP_ERROR_LOG |
4444 | 2130 FILE *df; |
2131 int *p2; | |
2132 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2133 df = fopen(NFA_REGEXP_ERROR_LOG, "a"); |
4444 | 2134 if (df) |
2135 { | |
2136 fprintf(df, "Error popping the stack!\n"); | |
2137 #ifdef DEBUG | |
2138 fprintf(df, "Current regexp is \"%s\"\n", nfa_regengine.expr); | |
2139 #endif | |
2140 fprintf(df, "Postfix form is: "); | |
2141 #ifdef DEBUG | |
2142 for (p2 = postfix; p2 < end; p2++) | |
2143 { | |
2144 nfa_set_code(*p2); | |
2145 fprintf(df, "%s, ", code); | |
2146 } | |
2147 nfa_set_code(*p); | |
2148 fprintf(df, "\nCurrent position is: "); | |
2149 for (p2 = postfix; p2 <= p; p2 ++) | |
2150 { | |
2151 nfa_set_code(*p2); | |
2152 fprintf(df, "%s, ", code); | |
2153 } | |
2154 #else | |
2155 for (p2 = postfix; p2 < end; p2++) | |
2156 { | |
2157 fprintf(df, "%d, ", *p2); | |
2158 } | |
2159 fprintf(df, "\nCurrent position is: "); | |
2160 for (p2 = postfix; p2 <= p; p2 ++) | |
2161 { | |
2162 fprintf(df, "%d, ", *p2); | |
2163 } | |
2164 #endif | |
2165 fprintf(df, "\n--------------------------\n"); | |
2166 fclose(df); | |
2167 } | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2168 #endif |
4444 | 2169 EMSG(_("E874: (NFA) Could not pop the stack !")); |
2170 } | |
2171 | |
2172 /* | |
2173 * Push an item onto the stack. | |
2174 */ | |
2175 static void | |
2176 st_push(s, p, stack_end) | |
2177 Frag_T s; | |
2178 Frag_T **p; | |
2179 Frag_T *stack_end; | |
2180 { | |
2181 Frag_T *stackp = *p; | |
2182 | |
2183 if (stackp >= stack_end) | |
2184 return; | |
2185 *stackp = s; | |
2186 *p = *p + 1; | |
2187 } | |
2188 | |
2189 /* | |
2190 * Pop an item from the stack. | |
2191 */ | |
2192 static Frag_T | |
2193 st_pop(p, stack) | |
2194 Frag_T **p; | |
2195 Frag_T *stack; | |
2196 { | |
2197 Frag_T *stackp; | |
2198 | |
2199 *p = *p - 1; | |
2200 stackp = *p; | |
2201 if (stackp < stack) | |
2202 return empty; | |
2203 return **p; | |
2204 } | |
2205 | |
2206 /* | |
2207 * Convert a postfix form into its equivalent NFA. | |
2208 * Return the NFA start state on success, NULL otherwise. | |
2209 */ | |
2210 static nfa_state_T * | |
2211 post2nfa(postfix, end, nfa_calc_size) | |
2212 int *postfix; | |
2213 int *end; | |
2214 int nfa_calc_size; | |
2215 { | |
2216 int *p; | |
2217 int mopen; | |
2218 int mclose; | |
2219 Frag_T *stack = NULL; | |
2220 Frag_T *stackp = NULL; | |
2221 Frag_T *stack_end = NULL; | |
2222 Frag_T e1; | |
2223 Frag_T e2; | |
2224 Frag_T e; | |
2225 nfa_state_T *s; | |
2226 nfa_state_T *s1; | |
2227 nfa_state_T *matchstate; | |
4484 | 2228 nfa_state_T *ret = NULL; |
4444 | 2229 |
2230 if (postfix == NULL) | |
2231 return NULL; | |
2232 | |
4456 | 2233 #define PUSH(s) st_push((s), &stackp, stack_end) |
4444 | 2234 #define POP() st_pop(&stackp, stack); \ |
2235 if (stackp < stack) \ | |
2236 { \ | |
2237 st_error(postfix, end, p); \ | |
2238 return NULL; \ | |
2239 } | |
2240 | |
2241 if (nfa_calc_size == FALSE) | |
2242 { | |
2243 /* Allocate space for the stack. Max states on the stack : nstate */ | |
4462 | 2244 stack = (Frag_T *) lalloc((nstate + 1) * sizeof(Frag_T), TRUE); |
4444 | 2245 stackp = stack; |
4462 | 2246 stack_end = stack + (nstate + 1); |
4444 | 2247 } |
2248 | |
2249 for (p = postfix; p < end; ++p) | |
2250 { | |
2251 switch (*p) | |
2252 { | |
2253 case NFA_CONCAT: | |
2254 /* Catenation. | |
2255 * Pay attention: this operator does not exist | |
2256 * in the r.e. itself (it is implicit, really). | |
2257 * It is added when r.e. is translated to postfix | |
2258 * form in re2post(). | |
2259 * | |
2260 * No new state added here. */ | |
2261 if (nfa_calc_size == TRUE) | |
2262 { | |
4458 | 2263 /* nstate += 0; */ |
4444 | 2264 break; |
2265 } | |
2266 e2 = POP(); | |
2267 e1 = POP(); | |
2268 patch(e1.out, e2.start); | |
2269 PUSH(frag(e1.start, e2.out)); | |
2270 break; | |
2271 | |
2272 case NFA_NOT: | |
2273 /* Negation of a character */ | |
2274 if (nfa_calc_size == TRUE) | |
2275 { | |
4458 | 2276 /* nstate += 0; */ |
4444 | 2277 break; |
2278 } | |
2279 e1 = POP(); | |
2280 e1.start->negated = TRUE; | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2281 #ifdef FEAT_MBYTE |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2282 if (e1.start->c == NFA_COMPOSING) |
4444 | 2283 e1.start->out1->negated = TRUE; |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2284 #endif |
4444 | 2285 PUSH(e1); |
2286 break; | |
2287 | |
2288 case NFA_OR: | |
2289 /* Alternation */ | |
2290 if (nfa_calc_size == TRUE) | |
2291 { | |
4458 | 2292 nstate++; |
4444 | 2293 break; |
2294 } | |
2295 e2 = POP(); | |
2296 e1 = POP(); | |
2297 s = new_state(NFA_SPLIT, e1.start, e2.start); | |
2298 if (s == NULL) | |
4484 | 2299 goto theend; |
4444 | 2300 PUSH(frag(s, append(e1.out, e2.out))); |
2301 break; | |
2302 | |
2303 case NFA_STAR: | |
2304 /* Zero or more */ | |
2305 if (nfa_calc_size == TRUE) | |
2306 { | |
4458 | 2307 nstate++; |
4444 | 2308 break; |
2309 } | |
2310 e = POP(); | |
2311 s = new_state(NFA_SPLIT, e.start, NULL); | |
2312 if (s == NULL) | |
4484 | 2313 goto theend; |
4444 | 2314 patch(e.out, s); |
2315 PUSH(frag(s, list1(&s->out1))); | |
2316 break; | |
2317 | |
2318 case NFA_QUEST: | |
2319 /* one or zero atoms=> greedy match */ | |
2320 if (nfa_calc_size == TRUE) | |
2321 { | |
4458 | 2322 nstate++; |
4444 | 2323 break; |
2324 } | |
2325 e = POP(); | |
2326 s = new_state(NFA_SPLIT, e.start, NULL); | |
2327 if (s == NULL) | |
4484 | 2328 goto theend; |
4444 | 2329 PUSH(frag(s, append(e.out, list1(&s->out1)))); |
2330 break; | |
2331 | |
2332 case NFA_QUEST_NONGREEDY: | |
2333 /* zero or one atoms => non-greedy match */ | |
2334 if (nfa_calc_size == TRUE) | |
2335 { | |
4458 | 2336 nstate++; |
4444 | 2337 break; |
2338 } | |
2339 e = POP(); | |
2340 s = new_state(NFA_SPLIT, NULL, e.start); | |
2341 if (s == NULL) | |
4484 | 2342 goto theend; |
4444 | 2343 PUSH(frag(s, append(e.out, list1(&s->out)))); |
2344 break; | |
2345 | |
2346 case NFA_PLUS: | |
2347 /* One or more */ | |
2348 if (nfa_calc_size == TRUE) | |
2349 { | |
4458 | 2350 nstate++; |
4444 | 2351 break; |
2352 } | |
2353 e = POP(); | |
2354 s = new_state(NFA_SPLIT, e.start, NULL); | |
2355 if (s == NULL) | |
4484 | 2356 goto theend; |
4444 | 2357 patch(e.out, s); |
2358 PUSH(frag(e.start, list1(&s->out1))); | |
2359 break; | |
2360 | |
2361 case NFA_SKIP_CHAR: | |
2362 /* Symbol of 0-length, Used in a repetition | |
2363 * with max/min count of 0 */ | |
2364 if (nfa_calc_size == TRUE) | |
2365 { | |
4458 | 2366 nstate++; |
4444 | 2367 break; |
2368 } | |
2369 s = new_state(NFA_SKIP_CHAR, NULL, NULL); | |
2370 if (s == NULL) | |
4484 | 2371 goto theend; |
4444 | 2372 PUSH(frag(s, list1(&s->out))); |
2373 break; | |
2374 | |
2375 case NFA_PREV_ATOM_NO_WIDTH: | |
4661
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
2376 case NFA_PREV_ATOM_NO_WIDTH_NEG: |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2377 /* The \@= operator: match the preceding atom with zero width. |
4661
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
2378 * The \@! operator: no match for the preceding atom. |
4444 | 2379 * Surrounds the preceding atom with START_INVISIBLE and |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2380 * END_INVISIBLE, similarly to MOPEN. */ |
4444 | 2381 |
2382 if (nfa_calc_size == TRUE) | |
2383 { | |
2384 nstate += 2; | |
2385 break; | |
2386 } | |
2387 e = POP(); | |
2388 s1 = new_state(NFA_END_INVISIBLE, NULL, NULL); | |
2389 if (s1 == NULL) | |
4484 | 2390 goto theend; |
4444 | 2391 patch(e.out, s1); |
2392 | |
2393 s = new_state(NFA_START_INVISIBLE, e.start, s1); | |
2394 if (s == NULL) | |
4484 | 2395 goto theend; |
4661
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
2396 if (*p == NFA_PREV_ATOM_NO_WIDTH_NEG) |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
2397 { |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
2398 s->negated = TRUE; |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
2399 s1->negated = TRUE; |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
2400 } |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
2401 |
4444 | 2402 PUSH(frag(s, list1(&s1->out))); |
2403 break; | |
2404 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2405 #ifdef FEAT_MBYTE |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2406 case NFA_COMPOSING: /* char with composing char */ |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2407 #if 0 |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2408 /* TODO */ |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2409 if (regflags & RF_ICOMBINE) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2410 { |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
2411 /* use the base character only */ |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2412 } |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2413 #endif |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2414 /* FALLTHROUGH */ |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2415 #endif |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2416 |
4444 | 2417 case NFA_MOPEN + 0: /* Submatch */ |
2418 case NFA_MOPEN + 1: | |
2419 case NFA_MOPEN + 2: | |
2420 case NFA_MOPEN + 3: | |
2421 case NFA_MOPEN + 4: | |
2422 case NFA_MOPEN + 5: | |
2423 case NFA_MOPEN + 6: | |
2424 case NFA_MOPEN + 7: | |
2425 case NFA_MOPEN + 8: | |
2426 case NFA_MOPEN + 9: | |
2427 case NFA_NOPEN: /* \%( "Invisible Submatch" */ | |
2428 if (nfa_calc_size == TRUE) | |
2429 { | |
2430 nstate += 2; | |
2431 break; | |
2432 } | |
2433 | |
2434 mopen = *p; | |
2435 switch (*p) | |
2436 { | |
2437 case NFA_NOPEN: | |
2438 mclose = NFA_NCLOSE; | |
2439 break; | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2440 #ifdef FEAT_MBYTE |
4444 | 2441 case NFA_COMPOSING: |
2442 mclose = NFA_END_COMPOSING; | |
2443 break; | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2444 #endif |
4444 | 2445 default: |
2446 /* NFA_MOPEN(0) ... NFA_MOPEN(9) */ | |
2447 mclose = *p + NSUBEXP; | |
2448 break; | |
2449 } | |
2450 | |
2451 /* Allow "NFA_MOPEN" as a valid postfix representation for | |
2452 * the empty regexp "". In this case, the NFA will be | |
2453 * NFA_MOPEN -> NFA_MCLOSE. Note that this also allows | |
2454 * empty groups of parenthesis, and empty mbyte chars */ | |
2455 if (stackp == stack) | |
2456 { | |
2457 s = new_state(mopen, NULL, NULL); | |
2458 if (s == NULL) | |
4484 | 2459 goto theend; |
4444 | 2460 s1 = new_state(mclose, NULL, NULL); |
2461 if (s1 == NULL) | |
4484 | 2462 goto theend; |
4444 | 2463 patch(list1(&s->out), s1); |
2464 PUSH(frag(s, list1(&s1->out))); | |
2465 break; | |
2466 } | |
2467 | |
2468 /* At least one node was emitted before NFA_MOPEN, so | |
2469 * at least one node will be between NFA_MOPEN and NFA_MCLOSE */ | |
2470 e = POP(); | |
2471 s = new_state(mopen, e.start, NULL); /* `(' */ | |
2472 if (s == NULL) | |
4484 | 2473 goto theend; |
4444 | 2474 |
2475 s1 = new_state(mclose, NULL, NULL); /* `)' */ | |
2476 if (s1 == NULL) | |
4484 | 2477 goto theend; |
4444 | 2478 patch(e.out, s1); |
2479 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2480 #ifdef FEAT_MBYTE |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2481 if (mopen == NFA_COMPOSING) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2482 /* COMPOSING->out1 = END_COMPOSING */ |
4444 | 2483 patch(list1(&s->out1), s1); |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2484 #endif |
4444 | 2485 |
2486 PUSH(frag(s, list1(&s1->out))); | |
2487 break; | |
2488 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2489 case NFA_BACKREF1: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2490 case NFA_BACKREF2: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2491 case NFA_BACKREF3: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2492 case NFA_BACKREF4: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2493 case NFA_BACKREF5: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2494 case NFA_BACKREF6: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2495 case NFA_BACKREF7: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2496 case NFA_BACKREF8: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2497 case NFA_BACKREF9: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2498 if (nfa_calc_size == TRUE) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2499 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2500 nstate += 2; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2501 break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2502 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2503 s = new_state(*p, NULL, NULL); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2504 if (s == NULL) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2505 goto theend; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2506 s1 = new_state(NFA_SKIP, NULL, NULL); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2507 if (s1 == NULL) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2508 goto theend; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2509 patch(list1(&s->out), s1); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2510 PUSH(frag(s, list1(&s1->out))); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2511 break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2512 |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2513 case NFA_LNUM: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2514 case NFA_LNUM_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2515 case NFA_LNUM_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2516 case NFA_VCOL: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2517 case NFA_VCOL_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2518 case NFA_VCOL_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2519 case NFA_COL: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2520 case NFA_COL_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2521 case NFA_COL_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2522 if (nfa_calc_size == TRUE) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2523 { |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2524 nstate += 1; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2525 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2526 } |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2527 e1 = POP(); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2528 s = new_state(*p, NULL, NULL); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2529 if (s == NULL) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2530 goto theend; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2531 s->val = e1.start->c; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2532 PUSH(frag(s, list1(&s->out))); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2533 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2534 |
4444 | 2535 case NFA_ZSTART: |
2536 case NFA_ZEND: | |
2537 default: | |
2538 /* Operands */ | |
2539 if (nfa_calc_size == TRUE) | |
2540 { | |
4458 | 2541 nstate++; |
4444 | 2542 break; |
2543 } | |
2544 s = new_state(*p, NULL, NULL); | |
2545 if (s == NULL) | |
4484 | 2546 goto theend; |
4444 | 2547 PUSH(frag(s, list1(&s->out))); |
2548 break; | |
2549 | |
2550 } /* switch(*p) */ | |
2551 | |
2552 } /* for(p = postfix; *p; ++p) */ | |
2553 | |
2554 if (nfa_calc_size == TRUE) | |
2555 { | |
4458 | 2556 nstate++; |
4484 | 2557 goto theend; /* Return value when counting size is ignored anyway */ |
4444 | 2558 } |
2559 | |
2560 e = POP(); | |
2561 if (stackp != stack) | |
2562 EMSG_RET_NULL(_("E875: (NFA regexp) (While converting from postfix to NFA), too many states left on stack")); | |
2563 | |
2564 if (istate >= nstate) | |
2565 EMSG_RET_NULL(_("E876: (NFA regexp) Not enough space to store the whole NFA ")); | |
2566 | |
2567 matchstate = &state_ptr[istate++]; /* the match state */ | |
2568 matchstate->c = NFA_MATCH; | |
2569 matchstate->out = matchstate->out1 = NULL; | |
2570 | |
2571 patch(e.out, matchstate); | |
4484 | 2572 ret = e.start; |
2573 | |
2574 theend: | |
2575 vim_free(stack); | |
2576 return ret; | |
4444 | 2577 |
2578 #undef POP1 | |
2579 #undef PUSH1 | |
2580 #undef POP2 | |
2581 #undef PUSH2 | |
2582 #undef POP | |
2583 #undef PUSH | |
2584 } | |
2585 | |
2586 /**************************************************************** | |
2587 * NFA execution code. | |
2588 ****************************************************************/ | |
2589 | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2590 typedef struct |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2591 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2592 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
|
2593 |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2594 /* 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
|
2595 union |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2596 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2597 struct multipos |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2598 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2599 lpos_T start; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2600 lpos_T end; |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2601 } multi[NSUBEXP]; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2602 struct linepos |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2603 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2604 char_u *start; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2605 char_u *end; |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2606 } line[NSUBEXP]; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2607 } list; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2608 } regsub_T; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2609 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2610 /* 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
|
2611 typedef struct |
4444 | 2612 { |
2613 nfa_state_T *state; | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2614 int count; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2615 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
|
2616 } nfa_thread_T; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2617 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2618 /* nfa_list_T contains the alternative NFA execution states. */ |
4444 | 2619 typedef struct |
2620 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2621 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
|
2622 int n; /* nr of states currently in "t" */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2623 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
|
2624 int id; /* ID of the list */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2625 } nfa_list_T; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2626 |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2627 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2628 static void |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2629 log_subexpr(sub) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2630 regsub_T *sub; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2631 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2632 int j; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2633 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2634 for (j = 0; j < sub->in_use; j++) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2635 if (REG_MULTI) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2636 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
|
2637 j, |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2638 sub->list.multi[j].start.col, |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2639 (int)sub->list.multi[j].start.lnum, |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2640 sub->list.multi[j].end.col, |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2641 (int)sub->list.multi[j].end.lnum); |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2642 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2643 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
|
2644 j, |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2645 (char *)sub->list.line[j].start, |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2646 (char *)sub->list.line[j].end); |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2647 fprintf(log_fd, "\n"); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2648 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2649 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2650 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2651 /* 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
|
2652 static int nfa_match; |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2653 |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2654 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
|
2655 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
|
2656 static void addstate_here __ARGS((nfa_list_T *l, nfa_state_T *state, regsub_T *sub, int *ip)); |
4444 | 2657 |
4647
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 * 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
|
2660 */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2661 static int |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2662 sub_equal(sub1, sub2) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2663 regsub_T *sub1; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2664 regsub_T *sub2; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2665 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2666 int i; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2667 int todo; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2668 linenr_T s1, e1; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2669 linenr_T s2, e2; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2670 char_u *sp1, *ep1; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2671 char_u *sp2, *ep2; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2672 |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2673 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
|
2674 if (REG_MULTI) |
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 for (i = 0; i < todo; ++i) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2677 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2678 if (i < sub1->in_use) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2679 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2680 s1 = sub1->list.multi[i].start.lnum; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2681 e1 = sub1->list.multi[i].end.lnum; |
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 else |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2684 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2685 s1 = 0; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2686 e1 = 0; |
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 if (i < sub2->in_use) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2689 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2690 s2 = sub2->list.multi[i].start.lnum; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2691 e2 = sub2->list.multi[i].end.lnum; |
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 else |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2694 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2695 s2 = 0; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2696 e2 = 0; |
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 if (s1 != s2 || e1 != e2) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2699 return FALSE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2700 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
|
2701 != sub2->list.multi[i].start.col) |
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 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
|
2704 != sub2->list.multi[i].end.col) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2705 return FALSE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2706 } |
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 else |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2709 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2710 for (i = 0; i < todo; ++i) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2711 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2712 if (i < sub1->in_use) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2713 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2714 sp1 = sub1->list.line[i].start; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2715 ep1 = sub1->list.line[i].end; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2716 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2717 else |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2718 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2719 sp1 = NULL; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2720 ep1 = NULL; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2721 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2722 if (i < sub2->in_use) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2723 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2724 sp2 = sub2->list.line[i].start; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2725 ep2 = sub2->list.line[i].end; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2726 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2727 else |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2728 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2729 sp2 = NULL; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2730 ep2 = NULL; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2731 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2732 if (sp1 != sp2 || ep1 != ep2) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2733 return FALSE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2734 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2735 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2736 |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2737 return TRUE; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2738 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2739 |
4444 | 2740 static void |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2741 addstate(l, state, sub, off) |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2742 nfa_list_T *l; /* runtime state list */ |
4444 | 2743 nfa_state_T *state; /* state to update */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2744 regsub_T *sub; /* pointers to subexpressions */ |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2745 int off; /* byte offset, when -1 go to next line */ |
4444 | 2746 { |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2747 int subidx; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2748 nfa_thread_T *thread; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2749 lpos_T save_lpos; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2750 int save_in_use; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2751 char_u *save_ptr; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2752 int i; |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2753 #ifdef ENABLE_LOG |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2754 int did_print = FALSE; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2755 #endif |
4444 | 2756 |
2757 if (l == NULL || state == NULL) | |
2758 return; | |
2759 | |
2760 switch (state->c) | |
2761 { | |
2762 case NFA_SPLIT: | |
2763 case NFA_NOT: | |
2764 case NFA_NOPEN: | |
2765 case NFA_NCLOSE: | |
2766 case NFA_MCLOSE: | |
2767 case NFA_MCLOSE + 1: | |
2768 case NFA_MCLOSE + 2: | |
2769 case NFA_MCLOSE + 3: | |
2770 case NFA_MCLOSE + 4: | |
2771 case NFA_MCLOSE + 5: | |
2772 case NFA_MCLOSE + 6: | |
2773 case NFA_MCLOSE + 7: | |
2774 case NFA_MCLOSE + 8: | |
2775 case NFA_MCLOSE + 9: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2776 /* 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
|
2777 * "out1" may be added below. */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2778 break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2779 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2780 case NFA_MOPEN: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2781 case NFA_MOPEN + 1: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2782 case NFA_MOPEN + 2: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2783 case NFA_MOPEN + 3: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2784 case NFA_MOPEN + 4: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2785 case NFA_MOPEN + 5: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2786 case NFA_MOPEN + 6: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2787 case NFA_MOPEN + 7: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2788 case NFA_MOPEN + 8: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2789 case NFA_MOPEN + 9: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2790 /* 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
|
2791 * 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
|
2792 if (state->lastlist == l->id) |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2793 goto skip_add; |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2794 state->lastlist = l->id; |
4444 | 2795 break; |
2796 | |
2797 default: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2798 if (state->lastlist == l->id) |
4444 | 2799 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2800 /* 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
|
2801 * 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
|
2802 if (!nfa_has_backref) |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2803 { |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2804 skip_add: |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2805 #ifdef ENABLE_LOG |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2806 nfa_set_code(state->c); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2807 fprintf(log_fd, "> Not adding state %d to list %d. char %d: %s\n", |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2808 abs(state->id), l->id, state->c, code); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2809 #endif |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2810 return; |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2811 } |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2812 |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2813 /* 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
|
2814 * positions. */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2815 for (i = 0; i < l->n; ++i) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2816 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2817 thread = &l->t[i]; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2818 if (thread->state->id == state->id |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2819 && sub_equal(&thread->sub, sub)) |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2820 goto skip_add; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2821 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2822 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2823 |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2824 /* 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
|
2825 * lot) bigger */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2826 if (nfa_has_backref && l->n == l->len) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2827 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2828 int newlen = l->len * 3 / 2 + 50; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2829 |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2830 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
|
2831 l->len = newlen; |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2832 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2833 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2834 /* add the state to the list */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2835 state->lastlist = l->id; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2836 thread = &l->t[l->n++]; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2837 thread->state = state; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2838 thread->sub.in_use = sub->in_use; |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2839 if (sub->in_use > 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2840 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2841 /* Copy the match start and end positions. */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2842 if (REG_MULTI) |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2843 mch_memmove(&thread->sub.list.multi[0], |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2844 &sub->list.multi[0], |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2845 sizeof(struct multipos) * sub->in_use); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2846 else |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
2847 mch_memmove(&thread->sub.list.line[0], |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2848 &sub->list.line[0], |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2849 sizeof(struct linepos) * sub->in_use); |
4444 | 2850 } |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2851 #ifdef ENABLE_LOG |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2852 { |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2853 int col; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2854 |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2855 if (thread->sub.in_use <= 0) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2856 col = -1; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2857 else if (REG_MULTI) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2858 col = thread->sub.list.multi[0].start.col; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2859 else |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2860 col = (int)(thread->sub.list.line[0].start - regline); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2861 nfa_set_code(state->c); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2862 fprintf(log_fd, "> Adding state %d to list %d. char %d: %s (start col %d)\n", |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2863 abs(state->id), l->id, state->c, code, col); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2864 did_print = TRUE; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2865 } |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2866 #endif |
4444 | 2867 } |
2868 | |
2869 #ifdef ENABLE_LOG | |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2870 if (!did_print) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2871 { |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2872 int col; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2873 |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2874 if (sub->in_use <= 0) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2875 col = -1; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2876 else if (REG_MULTI) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2877 col = sub->list.multi[0].start.col; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2878 else |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2879 col = (int)(sub->list.line[0].start - regline); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2880 nfa_set_code(state->c); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2881 fprintf(log_fd, "> Processing state %d for list %d. char %d: %s (start col %d)\n", |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2882 abs(state->id), l->id, state->c, code, col); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
2883 } |
4444 | 2884 #endif |
2885 switch (state->c) | |
2886 { | |
2887 case NFA_MATCH: | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2888 nfa_match = TRUE; |
4444 | 2889 break; |
2890 | |
2891 case NFA_SPLIT: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2892 addstate(l, state->out, sub, off); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2893 addstate(l, state->out1, sub, off); |
4444 | 2894 break; |
2895 | |
2896 #if 0 | |
2897 case NFA_END_NEG_RANGE: | |
2898 /* Nothing to handle here. nfa_regmatch() will take care of it */ | |
2899 break; | |
2900 | |
2901 case NFA_NOT: | |
2902 EMSG(_("E999: (NFA regexp internal error) Should not process NOT node !")); | |
2903 #ifdef ENABLE_LOG | |
2904 fprintf(f, "\n\n>>> E999: Added state NFA_NOT to a list ... Something went wrong ! Why wasn't it processed already? \n\n"); | |
2905 #endif | |
2906 break; | |
2907 | |
2908 case NFA_COMPOSING: | |
2909 /* nfa_regmatch() will match all the bytes of this composing char. */ | |
2910 break; | |
2911 #endif | |
2912 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2913 case NFA_SKIP_CHAR: |
4444 | 2914 case NFA_NOPEN: |
2915 case NFA_NCLOSE: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2916 addstate(l, state->out, sub, off); |
4444 | 2917 break; |
2918 | |
2919 case NFA_MOPEN + 0: | |
2920 case NFA_MOPEN + 1: | |
2921 case NFA_MOPEN + 2: | |
2922 case NFA_MOPEN + 3: | |
2923 case NFA_MOPEN + 4: | |
2924 case NFA_MOPEN + 5: | |
2925 case NFA_MOPEN + 6: | |
2926 case NFA_MOPEN + 7: | |
2927 case NFA_MOPEN + 8: | |
2928 case NFA_MOPEN + 9: | |
2929 case NFA_ZSTART: | |
2930 if (state->c == NFA_ZSTART) | |
2931 subidx = 0; | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2932 else |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2933 subidx = state->c - NFA_MOPEN; |
4444 | 2934 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2935 /* 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
|
2936 * restore it when it was in use. Otherwise fill any gap. */ |
4649
9f502d8d5165
updated for version 7.3.1072
Bram Moolenaar <bram@vim.org>
parents:
4647
diff
changeset
|
2937 save_ptr = NULL; |
4444 | 2938 if (REG_MULTI) |
2939 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2940 if (subidx < sub->in_use) |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2941 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2942 save_lpos = sub->list.multi[subidx].start; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2943 save_in_use = -1; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2944 } |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2945 else |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2946 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2947 save_in_use = sub->in_use; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2948 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
|
2949 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2950 sub->list.multi[i].start.lnum = -1; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2951 sub->list.multi[i].end.lnum = -1; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2952 } |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2953 sub->in_use = subidx + 1; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2954 } |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2955 if (off == -1) |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2956 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2957 sub->list.multi[subidx].start.lnum = reglnum + 1; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2958 sub->list.multi[subidx].start.col = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2959 } |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2960 else |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2961 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2962 sub->list.multi[subidx].start.lnum = reglnum; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2963 sub->list.multi[subidx].start.col = |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2964 (colnr_T)(reginput - regline + off); |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2965 } |
4444 | 2966 } |
2967 else | |
2968 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2969 if (subidx < sub->in_use) |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2970 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2971 save_ptr = sub->list.line[subidx].start; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2972 save_in_use = -1; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2973 } |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2974 else |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2975 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2976 save_in_use = sub->in_use; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2977 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
|
2978 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2979 sub->list.line[i].start = NULL; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2980 sub->list.line[i].end = NULL; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2981 } |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2982 sub->in_use = subidx + 1; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2983 } |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2984 sub->list.line[subidx].start = 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 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2989 if (save_in_use == -1) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2990 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2991 if (REG_MULTI) |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2992 sub->list.multi[subidx].start = save_lpos; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2993 else |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2994 sub->list.line[subidx].start = save_ptr; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2995 } |
4444 | 2996 else |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2997 sub->in_use = save_in_use; |
4444 | 2998 break; |
2999 | |
3000 case NFA_MCLOSE + 0: | |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3001 if (nfa_has_zend) |
4444 | 3002 { |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
3003 /* 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
|
3004 * 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
|
3005 addstate(l, state->out, sub, off); |
4444 | 3006 break; |
3007 } | |
3008 case NFA_MCLOSE + 1: | |
3009 case NFA_MCLOSE + 2: | |
3010 case NFA_MCLOSE + 3: | |
3011 case NFA_MCLOSE + 4: | |
3012 case NFA_MCLOSE + 5: | |
3013 case NFA_MCLOSE + 6: | |
3014 case NFA_MCLOSE + 7: | |
3015 case NFA_MCLOSE + 8: | |
3016 case NFA_MCLOSE + 9: | |
3017 case NFA_ZEND: | |
3018 if (state->c == NFA_ZEND) | |
3019 subidx = 0; | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3020 else |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3021 subidx = state->c - NFA_MCLOSE; |
4444 | 3022 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3023 /* 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
|
3024 * has done that. */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3025 save_in_use = sub->in_use; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3026 if (sub->in_use <= subidx) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3027 sub->in_use = subidx + 1; |
4444 | 3028 if (REG_MULTI) |
3029 { | |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3030 save_lpos = sub->list.multi[subidx].end; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3031 if (off == -1) |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3032 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3033 sub->list.multi[subidx].end.lnum = reglnum + 1; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3034 sub->list.multi[subidx].end.col = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3035 } |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3036 else |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3037 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3038 sub->list.multi[subidx].end.lnum = reglnum; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3039 sub->list.multi[subidx].end.col = |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3040 (colnr_T)(reginput - regline + off); |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3041 } |
4444 | 3042 } |
3043 else | |
3044 { | |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3045 save_ptr = sub->list.line[subidx].end; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3046 sub->list.line[subidx].end = reginput + off; |
4444 | 3047 } |
3048 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3049 addstate(l, state->out, sub, off); |
4444 | 3050 |
3051 if (REG_MULTI) | |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3052 sub->list.multi[subidx].end = save_lpos; |
4444 | 3053 else |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3054 sub->list.line[subidx].end = save_ptr; |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3055 sub->in_use = save_in_use; |
4444 | 3056 break; |
3057 } | |
3058 } | |
3059 | |
3060 /* | |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3061 * 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
|
3062 * 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
|
3063 * 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
|
3064 * matters for alternatives. |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3065 */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3066 static void |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3067 addstate_here(l, state, sub, ip) |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3068 nfa_list_T *l; /* runtime state list */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3069 nfa_state_T *state; /* state to update */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3070 regsub_T *sub; /* pointers to subexpressions */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3071 int *ip; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3072 { |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3073 int tlen = l->n; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3074 int count; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3075 int i = *ip; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3076 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3077 /* 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
|
3078 addstate(l, state, sub, 0); |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3079 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3080 /* 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
|
3081 if (i + 1 == tlen) |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3082 return; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3083 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3084 /* 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
|
3085 count = l->n - tlen; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3086 if (count == 1) |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3087 { |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3088 /* overwrite the current state */ |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3089 l->t[i] = l->t[l->n - 1]; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3090 } |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3091 else if (count > 1) |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3092 { |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3093 /* 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
|
3094 * end to the current position */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3095 mch_memmove(&(l->t[i + count]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3096 &(l->t[i + 1]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3097 sizeof(nfa_thread_T) * (l->n - i - 1)); |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3098 mch_memmove(&(l->t[i]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3099 &(l->t[l->n - 1]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3100 sizeof(nfa_thread_T) * count); |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3101 } |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3102 --l->n; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3103 *ip = i - 1; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3104 } |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3105 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3106 /* |
4444 | 3107 * Check character class "class" against current character c. |
3108 */ | |
3109 static int | |
3110 check_char_class(class, c) | |
3111 int class; | |
3112 int c; | |
3113 { | |
3114 switch (class) | |
3115 { | |
3116 case NFA_CLASS_ALNUM: | |
4464 | 3117 if (c >= 1 && c <= 255 && isalnum(c)) |
4444 | 3118 return OK; |
3119 break; | |
3120 case NFA_CLASS_ALPHA: | |
4464 | 3121 if (c >= 1 && c <= 255 && isalpha(c)) |
4444 | 3122 return OK; |
3123 break; | |
3124 case NFA_CLASS_BLANK: | |
3125 if (c == ' ' || c == '\t') | |
3126 return OK; | |
3127 break; | |
3128 case NFA_CLASS_CNTRL: | |
4464 | 3129 if (c >= 1 && c <= 255 && iscntrl(c)) |
4444 | 3130 return OK; |
3131 break; | |
3132 case NFA_CLASS_DIGIT: | |
3133 if (VIM_ISDIGIT(c)) | |
3134 return OK; | |
3135 break; | |
3136 case NFA_CLASS_GRAPH: | |
4464 | 3137 if (c >= 1 && c <= 255 && isgraph(c)) |
4444 | 3138 return OK; |
3139 break; | |
3140 case NFA_CLASS_LOWER: | |
3141 if (MB_ISLOWER(c)) | |
3142 return OK; | |
3143 break; | |
3144 case NFA_CLASS_PRINT: | |
3145 if (vim_isprintc(c)) | |
3146 return OK; | |
3147 break; | |
3148 case NFA_CLASS_PUNCT: | |
4464 | 3149 if (c >= 1 && c <= 255 && ispunct(c)) |
4444 | 3150 return OK; |
3151 break; | |
3152 case NFA_CLASS_SPACE: | |
3153 if ((c >=9 && c <= 13) || (c == ' ')) | |
3154 return OK; | |
3155 break; | |
3156 case NFA_CLASS_UPPER: | |
3157 if (MB_ISUPPER(c)) | |
3158 return OK; | |
3159 break; | |
3160 case NFA_CLASS_XDIGIT: | |
3161 if (vim_isxdigit(c)) | |
3162 return OK; | |
3163 break; | |
3164 case NFA_CLASS_TAB: | |
3165 if (c == '\t') | |
3166 return OK; | |
3167 break; | |
3168 case NFA_CLASS_RETURN: | |
3169 if (c == '\r') | |
3170 return OK; | |
3171 break; | |
3172 case NFA_CLASS_BACKSPACE: | |
3173 if (c == '\b') | |
3174 return OK; | |
3175 break; | |
3176 case NFA_CLASS_ESCAPE: | |
3177 if (c == '\033') | |
3178 return OK; | |
3179 break; | |
3180 | |
3181 default: | |
3182 /* should not be here :P */ | |
3183 EMSG_RET_FAIL(_("E877: (NFA regexp) Invalid character class ")); | |
3184 } | |
3185 return FAIL; | |
3186 } | |
3187 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3188 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
|
3189 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3190 /* |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3191 * Check for a match with subexpression "subidx". |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3192 * return TRUE if it matches. |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3193 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3194 static int |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3195 match_backref(sub, subidx, bytelen) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3196 regsub_T *sub; /* pointers to subexpressions */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3197 int subidx; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3198 int *bytelen; /* out: length of match in bytes */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3199 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3200 int len; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3201 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3202 if (sub->in_use <= subidx) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3203 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3204 retempty: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3205 /* backref was not set, match an empty string */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3206 *bytelen = 0; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3207 return TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3208 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3209 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3210 if (REG_MULTI) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3211 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3212 if (sub->list.multi[subidx].start.lnum < 0 |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3213 || sub->list.multi[subidx].end.lnum < 0) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3214 goto retempty; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3215 /* TODO: line breaks */ |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3216 len = sub->list.multi[subidx].end.col |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3217 - sub->list.multi[subidx].start.col; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3218 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
|
3219 reginput, &len) == 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3220 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3221 *bytelen = len; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3222 return TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3223 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3224 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3225 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3226 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3227 if (sub->list.line[subidx].start == NULL |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3228 || sub->list.line[subidx].end == NULL) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3229 goto retempty; |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3230 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
|
3231 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
|
3232 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3233 *bytelen = len; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3234 return TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3235 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3236 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3237 return FALSE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3238 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3239 |
4444 | 3240 /* |
3241 * Set all NFA nodes' list ID equal to -1. | |
3242 */ | |
3243 static void | |
3244 nfa_set_neg_listids(start) | |
3245 nfa_state_T *start; | |
3246 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3247 if (start != NULL && start->lastlist >= 0) |
4444 | 3248 { |
3249 start->lastlist = -1; | |
3250 nfa_set_neg_listids(start->out); | |
3251 nfa_set_neg_listids(start->out1); | |
3252 } | |
3253 } | |
3254 | |
3255 /* | |
3256 * Set all NFA nodes' list ID equal to 0. | |
3257 */ | |
3258 static void | |
3259 nfa_set_null_listids(start) | |
3260 nfa_state_T *start; | |
3261 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3262 if (start != NULL && start->lastlist == -1) |
4444 | 3263 { |
3264 start->lastlist = 0; | |
3265 nfa_set_null_listids(start->out); | |
3266 nfa_set_null_listids(start->out1); | |
3267 } | |
3268 } | |
3269 | |
3270 /* | |
3271 * Save list IDs for all NFA states in "list". | |
3272 */ | |
3273 static void | |
3274 nfa_save_listids(start, list) | |
3275 nfa_state_T *start; | |
3276 int *list; | |
3277 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3278 if (start != NULL && start->lastlist != -1) |
4444 | 3279 { |
3280 list[abs(start->id)] = start->lastlist; | |
3281 start->lastlist = -1; | |
3282 nfa_save_listids(start->out, list); | |
3283 nfa_save_listids(start->out1, list); | |
3284 } | |
3285 } | |
3286 | |
3287 /* | |
3288 * Restore list IDs from "list" to all NFA states. | |
3289 */ | |
3290 static void | |
3291 nfa_restore_listids(start, list) | |
3292 nfa_state_T *start; | |
3293 int *list; | |
3294 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3295 if (start != NULL && start->lastlist == -1) |
4444 | 3296 { |
3297 start->lastlist = list[abs(start->id)]; | |
3298 nfa_restore_listids(start->out, list); | |
3299 nfa_restore_listids(start->out1, list); | |
3300 } | |
3301 } | |
3302 | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3303 static int |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3304 nfa_re_num_cmp(val, op, pos) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3305 long_u val; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3306 int op; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3307 long_u pos; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3308 { |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3309 if (op == 1) return pos > val; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3310 if (op == 2) return pos < val; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3311 return val == pos; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3312 } |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3313 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3314 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
|
3315 |
4444 | 3316 /* |
3317 * Main matching routine. | |
3318 * | |
3319 * Run NFA to determine whether it matches reginput. | |
3320 * | |
3321 * Return TRUE if there is a match, FALSE otherwise. | |
3322 * Note: Caller must ensure that: start != NULL. | |
3323 */ | |
3324 static int | |
3325 nfa_regmatch(start, submatch, m) | |
3326 nfa_state_T *start; | |
3327 regsub_T *submatch; | |
3328 regsub_T *m; | |
3329 { | |
3330 int result; | |
3331 int size = 0; | |
3332 int flag = 0; | |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3333 int go_to_nextline = FALSE; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3334 nfa_thread_T *t; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3335 nfa_list_T list[3]; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3336 nfa_list_T *listtbl[2][2]; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3337 nfa_list_T *ll; |
4444 | 3338 int listid = 1; |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3339 int listidx; |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3340 nfa_list_T *thislist; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3341 nfa_list_T *nextlist; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3342 nfa_list_T *neglist; |
4444 | 3343 int *listids = NULL; |
4460 | 3344 #ifdef NFA_REGEXP_DEBUG_LOG |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3345 FILE *debug = fopen(NFA_REGEXP_DEBUG_LOG, "a"); |
4444 | 3346 |
3347 if (debug == NULL) | |
3348 { | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3349 EMSG2(_("(NFA) COULD NOT OPEN %s !"), NFA_REGEXP_DEBUG_LOG); |
4444 | 3350 return FALSE; |
3351 } | |
3352 #endif | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3353 nfa_match = FALSE; |
4444 | 3354 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3355 /* Allocate memory for the lists of nodes. */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3356 size = (nstate + 1) * sizeof(nfa_thread_T); |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3357 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
|
3358 list[0].len = nstate + 1; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3359 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
|
3360 list[1].len = nstate + 1; |
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
3361 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
|
3362 list[2].len = nstate + 1; |
4444 | 3363 if (list[0].t == NULL || list[1].t == NULL || list[2].t == NULL) |
3364 goto theend; | |
3365 | |
3366 #ifdef ENABLE_LOG | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3367 log_fd = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 3368 if (log_fd != NULL) |
3369 { | |
3370 fprintf(log_fd, "**********************************\n"); | |
3371 nfa_set_code(start->c); | |
3372 fprintf(log_fd, " RUNNING nfa_regmatch() starting with state %d, code %s\n", | |
3373 abs(start->id), code); | |
3374 fprintf(log_fd, "**********************************\n"); | |
3375 } | |
3376 else | |
3377 { | |
3378 EMSG(_("Could not open temporary log file for writing, displaying on stderr ... ")); | |
3379 log_fd = stderr; | |
3380 } | |
3381 #endif | |
3382 | |
3383 thislist = &list[0]; | |
3384 thislist->n = 0; | |
3385 nextlist = &list[1]; | |
3386 nextlist->n = 0; | |
3387 neglist = &list[2]; | |
3388 neglist->n = 0; | |
3389 #ifdef ENABLE_LOG | |
3390 fprintf(log_fd, "(---) STARTSTATE\n"); | |
3391 #endif | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3392 thislist->id = listid; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3393 addstate(thislist, start, m, 0); |
4444 | 3394 |
3395 /* There are two cases when the NFA advances: 1. input char matches the | |
3396 * NFA node and 2. input char does not match the NFA node, but the next | |
3397 * node is NFA_NOT. The following macro calls addstate() according to | |
3398 * these rules. It is used A LOT, so use the "listtbl" table for speed */ | |
3399 listtbl[0][0] = NULL; | |
3400 listtbl[0][1] = neglist; | |
3401 listtbl[1][0] = nextlist; | |
3402 listtbl[1][1] = NULL; | |
3403 #define ADD_POS_NEG_STATE(node) \ | |
3404 ll = listtbl[result ? 1 : 0][node->negated]; \ | |
3405 if (ll != NULL) \ | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3406 addstate(ll, node->out , &t->sub, clen); |
4444 | 3407 |
3408 | |
3409 /* | |
3410 * Run for each character. | |
3411 */ | |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3412 for (;;) |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3413 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3414 int curc; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3415 int clen; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3416 |
4444 | 3417 #ifdef FEAT_MBYTE |
3418 if (has_mbyte) | |
3419 { | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3420 curc = (*mb_ptr2char)(reginput); |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3421 clen = (*mb_ptr2len)(reginput); |
4444 | 3422 } |
3423 else | |
3424 #endif | |
3425 { | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3426 curc = *reginput; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3427 clen = 1; |
4444 | 3428 } |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3429 if (curc == NUL) |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3430 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3431 clen = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3432 go_to_nextline = FALSE; |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3433 } |
4444 | 3434 |
3435 /* swap lists */ | |
3436 thislist = &list[flag]; | |
3437 nextlist = &list[flag ^= 1]; | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3438 nextlist->n = 0; /* clear nextlist */ |
4444 | 3439 listtbl[1][0] = nextlist; |
3440 ++listid; | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3441 thislist->id = listid; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3442 nextlist->id = listid + 1; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3443 neglist->id = listid + 1; |
4444 | 3444 |
3445 #ifdef ENABLE_LOG | |
3446 fprintf(log_fd, "------------------------------------------\n"); | |
3447 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
|
3448 fprintf(log_fd, ">>> Advanced one character ... Current char is %c (code %d) \n", curc, (int)curc); |
4444 | 3449 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
|
3450 { |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3451 int i; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3452 |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3453 for (i = 0; i < thislist->n; i++) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3454 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
|
3455 } |
4444 | 3456 fprintf(log_fd, "\n"); |
3457 #endif | |
3458 | |
4460 | 3459 #ifdef NFA_REGEXP_DEBUG_LOG |
4444 | 3460 fprintf(debug, "\n-------------------\n"); |
3461 #endif | |
4480 | 3462 /* |
3463 * If the state lists are empty we can stop. | |
3464 */ | |
3465 if (thislist->n == 0 && neglist->n == 0) | |
3466 break; | |
4444 | 3467 |
3468 /* compute nextlist */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3469 for (listidx = 0; listidx < thislist->n || neglist->n > 0; ++listidx) |
4444 | 3470 { |
3471 if (neglist->n > 0) | |
3472 { | |
3473 t = &neglist->t[0]; | |
4468 | 3474 neglist->n--; |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3475 listidx--; |
4444 | 3476 } |
3477 else | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3478 t = &thislist->t[listidx]; |
4444 | 3479 |
4460 | 3480 #ifdef NFA_REGEXP_DEBUG_LOG |
4444 | 3481 nfa_set_code(t->state->c); |
3482 fprintf(debug, "%s, ", code); | |
3483 #endif | |
3484 #ifdef ENABLE_LOG | |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3485 { |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3486 int col; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3487 |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3488 if (t->sub.in_use <= 0) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3489 col = -1; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3490 else if (REG_MULTI) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3491 col = t->sub.list.multi[0].start.col; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3492 else |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3493 col = (int)(t->sub.list.line[0].start - regline); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3494 nfa_set_code(t->state->c); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3495 fprintf(log_fd, "(%d) char %d %s (start col %d) ... \n", |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3496 abs(t->state->id), (int)t->state->c, code, col); |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3497 } |
4444 | 3498 #endif |
3499 | |
3500 /* | |
3501 * Handle the possible codes of the current state. | |
3502 * The most important is NFA_MATCH. | |
3503 */ | |
3504 switch (t->state->c) | |
3505 { | |
3506 case NFA_MATCH: | |
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3507 { |
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3508 int j; |
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3509 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3510 nfa_match = TRUE; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3511 submatch->in_use = t->sub.in_use; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3512 if (REG_MULTI) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3513 for (j = 0; j < submatch->in_use; j++) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3514 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3515 submatch->list.multi[j].start = |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3516 t->sub.list.multi[j].start; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3517 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
|
3518 } |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3519 else |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3520 for (j = 0; j < submatch->in_use; j++) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3521 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3522 submatch->list.line[j].start = |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3523 t->sub.list.line[j].start; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3524 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
|
3525 } |
4444 | 3526 #ifdef ENABLE_LOG |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3527 log_subexpr(&t->sub); |
4444 | 3528 #endif |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3529 /* 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
|
3530 * 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
|
3531 * 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
|
3532 * correct. */ |
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3533 if (nextlist->n == 0 && neglist->n == 0) |
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3534 clen = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3535 goto nextchar; |
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3536 } |
4444 | 3537 |
3538 case NFA_END_INVISIBLE: | |
3539 /* This is only encountered after a NFA_START_INVISIBLE node. | |
3540 * They surround a zero-width group, used with "\@=" and "\&". | |
3541 * If we got here, it means that the current "invisible" group | |
3542 * finished successfully, so return control to the parent | |
3543 * nfa_regmatch(). Submatches are stored in *m, and used in | |
3544 * the parent call. */ | |
3545 if (start->c == NFA_MOPEN + 0) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3546 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3547 else |
3548 { | |
4661
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
3549 /* do not set submatches for \@! */ |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
3550 if (!t->state->negated) |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
3551 /* TODO: only copy positions in use. */ |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
3552 *m = t->sub; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3553 nfa_match = TRUE; |
4444 | 3554 } |
3555 break; | |
3556 | |
3557 case NFA_START_INVISIBLE: | |
4669
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3558 { |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3559 char_u *save_reginput = reginput; |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3560 char_u *save_regline = regline; |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3561 int save_reglnum = reglnum; |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3562 int save_nfa_match = nfa_match; |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3563 |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3564 /* Call nfa_regmatch() to check if the current concat matches |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3565 * at this position. The concat ends with the node |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3566 * NFA_END_INVISIBLE */ |
4444 | 3567 if (listids == NULL) |
3568 { | |
4669
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3569 listids = (int *)lalloc(sizeof(int) * nstate, TRUE); |
4444 | 3570 if (listids == NULL) |
3571 { | |
3572 EMSG(_("E878: (NFA) Could not allocate memory for branch traversal!")); | |
3573 return 0; | |
3574 } | |
3575 } | |
3576 #ifdef ENABLE_LOG | |
3577 if (log_fd != stderr) | |
3578 fclose(log_fd); | |
3579 log_fd = NULL; | |
3580 #endif | |
3581 /* Have to clear the listid field of the NFA nodes, so that | |
3582 * nfa_regmatch() and addstate() can run properly after | |
3583 * recursion. */ | |
3584 nfa_save_listids(start, listids); | |
3585 nfa_set_null_listids(start); | |
3586 result = nfa_regmatch(t->state->out, submatch, m); | |
3587 nfa_set_neg_listids(start); | |
3588 nfa_restore_listids(start, listids); | |
4669
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3589 |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3590 /* restore position in input text */ |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3591 reginput = save_reginput; |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3592 regline = save_regline; |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3593 reglnum = save_reglnum; |
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3594 nfa_match = save_nfa_match; |
4444 | 3595 |
3596 #ifdef ENABLE_LOG | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3597 log_fd = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 3598 if (log_fd != NULL) |
3599 { | |
3600 fprintf(log_fd, "****************************\n"); | |
3601 fprintf(log_fd, "FINISHED RUNNING nfa_regmatch() recursively\n"); | |
3602 fprintf(log_fd, "MATCH = %s\n", result == TRUE ? "OK" : "FALSE"); | |
3603 fprintf(log_fd, "****************************\n"); | |
3604 } | |
3605 else | |
3606 { | |
3607 EMSG(_("Could not open temporary log file for writing, displaying on stderr ... ")); | |
3608 log_fd = stderr; | |
3609 } | |
3610 #endif | |
4661
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
3611 /* for \@! it is a match when result is FALSE */ |
0dce3d812e7a
updated for version 7.3.1078
Bram Moolenaar <bram@vim.org>
parents:
4657
diff
changeset
|
3612 if (result != t->state->negated) |
4444 | 3613 { |
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3614 int j; |
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3615 |
4444 | 3616 /* Copy submatch info from the recursive call */ |
3617 if (REG_MULTI) | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3618 for (j = 1; j < m->in_use; j++) |
4444 | 3619 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3620 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
|
3621 t->sub.list.multi[j].end = m->list.multi[j].end; |
4444 | 3622 } |
3623 else | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3624 for (j = 1; j < m->in_use; j++) |
4444 | 3625 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3626 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
|
3627 t->sub.list.line[j].end = m->list.line[j].end; |
4444 | 3628 } |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3629 if (m->in_use > t->sub.in_use) |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3630 t->sub.in_use = m->in_use; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3631 |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3632 /* t->state->out1 is the corresponding END_INVISIBLE node; |
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
3633 * Add it to the current list (zero-width match). */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3634 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
|
3635 &listidx); |
4444 | 3636 } |
3637 break; | |
4669
2d15582e132f
updated for version 7.3.1082
Bram Moolenaar <bram@vim.org>
parents:
4667
diff
changeset
|
3638 } |
4444 | 3639 |
3640 case NFA_BOL: | |
3641 if (reginput == regline) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3642 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3643 break; |
3644 | |
3645 case NFA_EOL: | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3646 if (curc == NUL) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3647 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3648 break; |
3649 | |
3650 case NFA_BOW: | |
3651 { | |
3652 int bow = TRUE; | |
3653 | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3654 if (curc == NUL) |
4444 | 3655 bow = FALSE; |
3656 #ifdef FEAT_MBYTE | |
3657 else if (has_mbyte) | |
3658 { | |
3659 int this_class; | |
3660 | |
3661 /* Get class of current and previous char (if it exists). */ | |
4503 | 3662 this_class = mb_get_class_buf(reginput, reg_buf); |
4444 | 3663 if (this_class <= 1) |
3664 bow = FALSE; | |
3665 else if (reg_prev_class() == this_class) | |
3666 bow = FALSE; | |
3667 } | |
3668 #endif | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3669 else if (!vim_iswordc_buf(curc, reg_buf) |
4503 | 3670 || (reginput > regline |
3671 && vim_iswordc_buf(reginput[-1], reg_buf))) | |
4444 | 3672 bow = FALSE; |
3673 if (bow) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3674 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3675 break; |
3676 } | |
3677 | |
3678 case NFA_EOW: | |
3679 { | |
3680 int eow = TRUE; | |
3681 | |
3682 if (reginput == regline) | |
3683 eow = FALSE; | |
3684 #ifdef FEAT_MBYTE | |
3685 else if (has_mbyte) | |
3686 { | |
3687 int this_class, prev_class; | |
3688 | |
3689 /* Get class of current and previous char (if it exists). */ | |
4503 | 3690 this_class = mb_get_class_buf(reginput, reg_buf); |
4444 | 3691 prev_class = reg_prev_class(); |
3692 if (this_class == prev_class | |
3693 || prev_class == 0 || prev_class == 1) | |
3694 eow = FALSE; | |
3695 } | |
3696 #endif | |
4503 | 3697 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
|
3698 || (reginput[0] != NUL |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3699 && vim_iswordc_buf(curc, reg_buf))) |
4444 | 3700 eow = FALSE; |
3701 if (eow) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3702 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3703 break; |
3704 } | |
3705 | |
4671
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3706 case NFA_BOF: |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3707 if (reglnum == 0 && reginput == regline |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3708 && (!REG_MULTI || reg_firstlnum == 1)) |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3709 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3710 break; |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3711 |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3712 case NFA_EOF: |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3713 if (reglnum == reg_maxline && curc == NUL) |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3714 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3715 break; |
b3c59716e700
updated for version 7.3.1083
Bram Moolenaar <bram@vim.org>
parents:
4669
diff
changeset
|
3716 |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3717 #ifdef FEAT_MBYTE |
4444 | 3718 case NFA_COMPOSING: |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3719 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3720 int mc = curc; |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3721 int len = 0; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3722 nfa_state_T *end; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3723 nfa_state_T *sta; |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3724 int cchars[MAX_MCO]; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3725 int ccount = 0; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3726 int j; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3727 |
4444 | 3728 sta = t->state->out; |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3729 len = 0; |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3730 if (utf_iscomposing(sta->c)) |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3731 { |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3732 /* Only match composing character(s), ignore base |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3733 * character. Used for ".{composing}" and "{composing}" |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3734 * (no preceding character). */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3735 len += mb_char2len(mc); |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3736 } |
4551
f0c37f5cf4ea
updated for version 7.3.1023
Bram Moolenaar <bram@vim.org>
parents:
4549
diff
changeset
|
3737 if (ireg_icombine && len == 0) |
4444 | 3738 { |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3739 /* If \Z was present, then ignore composing characters. |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3740 * 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
|
3741 /* TODO: How about negated? */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3742 if (len == 0 && sta->c != curc) |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3743 result = FAIL; |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3744 else |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3745 result = OK; |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3746 while (sta->c != NFA_END_COMPOSING) |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3747 sta = sta->out; |
4444 | 3748 } |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3749 |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3750 /* Check base character matches first, unless ignored. */ |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3751 else if (len > 0 || mc == sta->c) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3752 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3753 if (len == 0) |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3754 { |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3755 len += mb_char2len(mc); |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3756 sta = sta->out; |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3757 } |
4444 | 3758 |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3759 /* 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
|
3760 * Get them into cchars[] first. */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3761 while (len < clen) |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3762 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3763 mc = mb_ptr2char(reginput + len); |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3764 cchars[ccount++] = mc; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3765 len += mb_char2len(mc); |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3766 if (ccount == MAX_MCO) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3767 break; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3768 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3769 |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3770 /* 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
|
3771 * 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
|
3772 * composing chars are matched. */ |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3773 result = OK; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3774 while (sta->c != NFA_END_COMPOSING) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3775 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3776 for (j = 0; j < ccount; ++j) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3777 if (cchars[j] == sta->c) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3778 break; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3779 if (j == ccount) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3780 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3781 result = FAIL; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3782 break; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3783 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3784 sta = sta->out; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3785 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3786 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3787 else |
4525
36ddcf4cecbc
updated for version 7.3.1010
Bram Moolenaar <bram@vim.org>
parents:
4517
diff
changeset
|
3788 result = FAIL; |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3789 |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3790 end = t->state->out1; /* NFA_END_COMPOSING */ |
4444 | 3791 ADD_POS_NEG_STATE(end); |
3792 break; | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3793 } |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3794 #endif |
4444 | 3795 |
3796 case NFA_NEWL: | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3797 if (curc == NUL && !reg_line_lbr && REG_MULTI |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3798 && reglnum <= reg_maxline) |
4444 | 3799 { |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3800 go_to_nextline = TRUE; |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3801 /* 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
|
3802 * at the start of the next line. */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3803 addstate(nextlist, t->state->out, &t->sub, -1); |
4444 | 3804 } |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3805 else if (curc == '\n' && reg_line_lbr) |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3806 { |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3807 /* 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
|
3808 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
|
3809 } |
4444 | 3810 break; |
3811 | |
3812 case NFA_CLASS_ALNUM: | |
3813 case NFA_CLASS_ALPHA: | |
3814 case NFA_CLASS_BLANK: | |
3815 case NFA_CLASS_CNTRL: | |
3816 case NFA_CLASS_DIGIT: | |
3817 case NFA_CLASS_GRAPH: | |
3818 case NFA_CLASS_LOWER: | |
3819 case NFA_CLASS_PRINT: | |
3820 case NFA_CLASS_PUNCT: | |
3821 case NFA_CLASS_SPACE: | |
3822 case NFA_CLASS_UPPER: | |
3823 case NFA_CLASS_XDIGIT: | |
3824 case NFA_CLASS_TAB: | |
3825 case NFA_CLASS_RETURN: | |
3826 case NFA_CLASS_BACKSPACE: | |
3827 case NFA_CLASS_ESCAPE: | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3828 result = check_char_class(t->state->c, curc); |
4444 | 3829 ADD_POS_NEG_STATE(t->state); |
3830 break; | |
3831 | |
3832 case NFA_END_NEG_RANGE: | |
3833 /* This follows a series of negated nodes, like: | |
3834 * 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
|
3835 if (curc > 0) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3836 addstate(nextlist, t->state->out, &t->sub, clen); |
4444 | 3837 break; |
3838 | |
3839 case NFA_ANY: | |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3840 /* 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
|
3841 if (curc > 0) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3842 addstate(nextlist, t->state->out, &t->sub, clen); |
4444 | 3843 break; |
3844 | |
3845 /* | |
3846 * Character classes like \a for alpha, \d for digit etc. | |
3847 */ | |
3848 case NFA_IDENT: /* \i */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3849 result = vim_isIDc(curc); |
4444 | 3850 ADD_POS_NEG_STATE(t->state); |
3851 break; | |
3852 | |
3853 case NFA_SIDENT: /* \I */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3854 result = !VIM_ISDIGIT(curc) && vim_isIDc(curc); |
4444 | 3855 ADD_POS_NEG_STATE(t->state); |
3856 break; | |
3857 | |
3858 case NFA_KWORD: /* \k */ | |
4503 | 3859 result = vim_iswordp_buf(reginput, reg_buf); |
4444 | 3860 ADD_POS_NEG_STATE(t->state); |
3861 break; | |
3862 | |
3863 case NFA_SKWORD: /* \K */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3864 result = !VIM_ISDIGIT(curc) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3865 && vim_iswordp_buf(reginput, reg_buf); |
4444 | 3866 ADD_POS_NEG_STATE(t->state); |
3867 break; | |
3868 | |
3869 case NFA_FNAME: /* \f */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3870 result = vim_isfilec(curc); |
4444 | 3871 ADD_POS_NEG_STATE(t->state); |
3872 break; | |
3873 | |
3874 case NFA_SFNAME: /* \F */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3875 result = !VIM_ISDIGIT(curc) && vim_isfilec(curc); |
4444 | 3876 ADD_POS_NEG_STATE(t->state); |
3877 break; | |
3878 | |
3879 case NFA_PRINT: /* \p */ | |
4470 | 3880 result = ptr2cells(reginput) == 1; |
4444 | 3881 ADD_POS_NEG_STATE(t->state); |
3882 break; | |
3883 | |
3884 case NFA_SPRINT: /* \P */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3885 result = !VIM_ISDIGIT(curc) && ptr2cells(reginput) == 1; |
4444 | 3886 ADD_POS_NEG_STATE(t->state); |
3887 break; | |
3888 | |
3889 case NFA_WHITE: /* \s */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3890 result = vim_iswhite(curc); |
4444 | 3891 ADD_POS_NEG_STATE(t->state); |
3892 break; | |
3893 | |
3894 case NFA_NWHITE: /* \S */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3895 result = curc != NUL && !vim_iswhite(curc); |
4444 | 3896 ADD_POS_NEG_STATE(t->state); |
3897 break; | |
3898 | |
3899 case NFA_DIGIT: /* \d */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3900 result = ri_digit(curc); |
4444 | 3901 ADD_POS_NEG_STATE(t->state); |
3902 break; | |
3903 | |
3904 case NFA_NDIGIT: /* \D */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3905 result = curc != NUL && !ri_digit(curc); |
4444 | 3906 ADD_POS_NEG_STATE(t->state); |
3907 break; | |
3908 | |
3909 case NFA_HEX: /* \x */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3910 result = ri_hex(curc); |
4444 | 3911 ADD_POS_NEG_STATE(t->state); |
3912 break; | |
3913 | |
3914 case NFA_NHEX: /* \X */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3915 result = curc != NUL && !ri_hex(curc); |
4444 | 3916 ADD_POS_NEG_STATE(t->state); |
3917 break; | |
3918 | |
3919 case NFA_OCTAL: /* \o */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3920 result = ri_octal(curc); |
4444 | 3921 ADD_POS_NEG_STATE(t->state); |
3922 break; | |
3923 | |
3924 case NFA_NOCTAL: /* \O */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3925 result = curc != NUL && !ri_octal(curc); |
4444 | 3926 ADD_POS_NEG_STATE(t->state); |
3927 break; | |
3928 | |
3929 case NFA_WORD: /* \w */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3930 result = ri_word(curc); |
4444 | 3931 ADD_POS_NEG_STATE(t->state); |
3932 break; | |
3933 | |
3934 case NFA_NWORD: /* \W */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3935 result = curc != NUL && !ri_word(curc); |
4444 | 3936 ADD_POS_NEG_STATE(t->state); |
3937 break; | |
3938 | |
3939 case NFA_HEAD: /* \h */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3940 result = ri_head(curc); |
4444 | 3941 ADD_POS_NEG_STATE(t->state); |
3942 break; | |
3943 | |
3944 case NFA_NHEAD: /* \H */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3945 result = curc != NUL && !ri_head(curc); |
4444 | 3946 ADD_POS_NEG_STATE(t->state); |
3947 break; | |
3948 | |
3949 case NFA_ALPHA: /* \a */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3950 result = ri_alpha(curc); |
4444 | 3951 ADD_POS_NEG_STATE(t->state); |
3952 break; | |
3953 | |
3954 case NFA_NALPHA: /* \A */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3955 result = curc != NUL && !ri_alpha(curc); |
4444 | 3956 ADD_POS_NEG_STATE(t->state); |
3957 break; | |
3958 | |
3959 case NFA_LOWER: /* \l */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3960 result = ri_lower(curc); |
4444 | 3961 ADD_POS_NEG_STATE(t->state); |
3962 break; | |
3963 | |
3964 case NFA_NLOWER: /* \L */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3965 result = curc != NUL && !ri_lower(curc); |
4444 | 3966 ADD_POS_NEG_STATE(t->state); |
3967 break; | |
3968 | |
3969 case NFA_UPPER: /* \u */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3970 result = ri_upper(curc); |
4444 | 3971 ADD_POS_NEG_STATE(t->state); |
3972 break; | |
3973 | |
3974 case NFA_NUPPER: /* \U */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3975 result = curc != NUL && !ri_upper(curc); |
4444 | 3976 ADD_POS_NEG_STATE(t->state); |
3977 break; | |
3978 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3979 case NFA_BACKREF1: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3980 case NFA_BACKREF2: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3981 case NFA_BACKREF3: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3982 case NFA_BACKREF4: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3983 case NFA_BACKREF5: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3984 case NFA_BACKREF6: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3985 case NFA_BACKREF7: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3986 case NFA_BACKREF8: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3987 case NFA_BACKREF9: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3988 /* \1 .. \9 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3989 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3990 int subidx = t->state->c - NFA_BACKREF1 + 1; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3991 int bytelen; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3992 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3993 result = match_backref(&t->sub, subidx, &bytelen); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3994 if (result) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3995 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3996 if (bytelen == 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3997 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3998 /* 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
|
3999 * be used next */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4000 addstate_here(thislist, t->state->out, &t->sub, |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4001 &listidx); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4002 thislist->t[listidx + 1].count = 0; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4003 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4004 else if (bytelen <= clen) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4005 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4006 /* match current character, jump ahead to out of |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4007 * NFA_SKIP */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4008 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
|
4009 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4010 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4011 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4012 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4013 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4014 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4015 /* skip ofer the matched characters, set character |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4016 * count in NFA_SKIP */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4017 addstate(nextlist, t->state->out, &t->sub, bytelen); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4018 nextlist->t[nextlist->n - 1].count = bytelen - clen; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4019 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4020 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4021 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4022 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4023 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4024 } |
4482 | 4025 break; |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4026 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4027 case NFA_SKIP: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4028 /* charater of previous matching \1 .. \9 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4029 if (t->count - clen <= 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4030 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4031 /* end of match, go to what follows */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4032 addstate(nextlist, t->state->out, &t->sub, clen); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4033 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4034 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4035 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4036 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4037 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4038 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4039 /* add state again with decremented count */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4040 addstate(nextlist, t->state, &t->sub, 0); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4041 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
|
4042 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4043 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4044 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4045 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4046 break; |
4482 | 4047 |
4048 case NFA_SKIP_CHAR: | |
4049 case NFA_ZSTART: | |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
4050 case NFA_ZEND: |
4482 | 4051 /* TODO: should not happen? */ |
4052 break; | |
4053 | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4054 case NFA_LNUM: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4055 case NFA_LNUM_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4056 case NFA_LNUM_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4057 result = (REG_MULTI && |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4058 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
|
4059 (long_u)(reglnum + reg_firstlnum))); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4060 if (result) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4061 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
|
4062 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4063 |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4064 case NFA_COL: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4065 case NFA_COL_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4066 case NFA_COL_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4067 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
|
4068 (long_u)(reginput - regline) + 1); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4069 if (result) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4070 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
|
4071 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4072 |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4073 case NFA_VCOL: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4074 case NFA_VCOL_GT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4075 case NFA_VCOL_LT: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4076 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
|
4077 (long_u)win_linetabsize( |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4078 reg_win == NULL ? curwin : reg_win, |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4079 regline, (colnr_T)(reginput - regline)) + 1); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4080 if (result) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4081 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
|
4082 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4083 |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4084 case NFA_CURSOR: |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4085 result = (reg_win != NULL |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4086 && (reglnum + reg_firstlnum == reg_win->w_cursor.lnum) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4087 && ((colnr_T)(reginput - regline) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4088 == reg_win->w_cursor.col)); |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4089 if (result) |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4090 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
|
4091 break; |
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
4092 |
4444 | 4093 default: /* regular character */ |
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4094 { |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4095 int c = t->state->c; |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4096 |
4482 | 4097 /* TODO: put this in #ifdef later */ |
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4098 if (c < -256) |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4099 EMSGN("INTERNAL: Negative state char: %ld", c); |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4100 if (is_Magic(c)) |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4101 c = un_Magic(c); |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4102 result = (c == curc); |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4103 |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4104 if (!result && ireg_ic) |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4105 result = MB_TOLOWER(c) == MB_TOLOWER(curc); |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4106 #ifdef FEAT_MBYTE |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4107 /* 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
|
4108 * 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
|
4109 * character uses NFA_COMPOSING above. */ |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4110 if (result && enc_utf8 && !ireg_icombine |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4111 && clen != utf_char2len(curc)) |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4112 result = FALSE; |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
4113 #endif |
4444 | 4114 ADD_POS_NEG_STATE(t->state); |
4115 break; | |
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
4116 } |
4444 | 4117 } |
4118 | |
4119 } /* for (thislist = thislist; thislist->state; thislist++) */ | |
4120 | |
4557
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
4121 /* 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
|
4122 * start state to the list of states. |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
4123 * 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
|
4124 * matters! |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
4125 * 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
|
4126 * 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
|
4127 * 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
|
4128 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
|
4129 && reglnum == 0 && clen != 0 |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
4130 && (ireg_maxcol == 0 |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
4131 || (colnr_T)(reginput - regline) < ireg_maxcol)) |
4444 | 4132 { |
4133 #ifdef ENABLE_LOG | |
4134 fprintf(log_fd, "(---) STARTSTATE\n"); | |
4135 #endif | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
4136 addstate(nextlist, start, m, clen); |
4444 | 4137 } |
4138 | |
4139 #ifdef ENABLE_LOG | |
4140 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
|
4141 { |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4142 int i; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4143 |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4144 for (i = 0; i < thislist->n; i++) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4145 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
|
4146 } |
4444 | 4147 fprintf(log_fd, "\n"); |
4148 #endif | |
4149 | |
4150 nextchar: | |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4151 /* 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
|
4152 * finish. */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4153 if (clen != 0) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
4154 reginput += clen; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4155 else if (go_to_nextline) |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4156 reg_nextline(); |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4157 else |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4158 break; |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
4159 } |
4444 | 4160 |
4161 #ifdef ENABLE_LOG | |
4162 if (log_fd != stderr) | |
4163 fclose(log_fd); | |
4164 log_fd = NULL; | |
4165 #endif | |
4166 | |
4167 theend: | |
4168 /* Free memory */ | |
4169 vim_free(list[0].t); | |
4170 vim_free(list[1].t); | |
4171 vim_free(list[2].t); | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4172 vim_free(listids); |
4444 | 4173 #undef ADD_POS_NEG_STATE |
4460 | 4174 #ifdef NFA_REGEXP_DEBUG_LOG |
4444 | 4175 fclose(debug); |
4176 #endif | |
4177 | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4178 return nfa_match; |
4444 | 4179 } |
4180 | |
4181 /* | |
4182 * Try match of "prog" with at regline["col"]. | |
4183 * Returns 0 for failure, number of lines contained in the match otherwise. | |
4184 */ | |
4185 static long | |
4186 nfa_regtry(start, col) | |
4187 nfa_state_T *start; | |
4188 colnr_T col; | |
4189 { | |
4190 int i; | |
4191 regsub_T sub, m; | |
4192 #ifdef ENABLE_LOG | |
4193 FILE *f; | |
4194 #endif | |
4195 | |
4196 reginput = regline + col; | |
4197 need_clear_subexpr = TRUE; | |
4198 | |
4199 #ifdef ENABLE_LOG | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
4200 f = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 4201 if (f != NULL) |
4202 { | |
4203 fprintf(f, "\n\n\n\n\n\n\t\t=======================================================\n"); | |
4204 fprintf(f, " =======================================================\n"); | |
4205 #ifdef DEBUG | |
4206 fprintf(f, "\tRegexp is \"%s\"\n", nfa_regengine.expr); | |
4207 #endif | |
4208 fprintf(f, "\tInput text is \"%s\" \n", reginput); | |
4657
93b7ed814bec
updated for version 7.3.1076
Bram Moolenaar <bram@vim.org>
parents:
4655
diff
changeset
|
4209 fprintf(f, " =======================================================\n\n"); |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
4210 nfa_print_state(f, start); |
4444 | 4211 fprintf(f, "\n\n"); |
4212 fclose(f); | |
4213 } | |
4214 else | |
4215 EMSG(_("Could not open temporary log file for writing ")); | |
4216 #endif | |
4217 | |
4218 if (REG_MULTI) | |
4219 { | |
4220 /* Use 0xff to set lnum to -1 */ | |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4221 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
|
4222 vim_memset(m.list.multi, 0xff, sizeof(struct multipos) * nfa_nsubexpr); |
4444 | 4223 } |
4224 else | |
4225 { | |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4226 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
|
4227 vim_memset(m.list.line, 0, sizeof(struct linepos) * nfa_nsubexpr); |
4444 | 4228 } |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4229 sub.in_use = 0; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4230 m.in_use = 0; |
4444 | 4231 |
4232 if (nfa_regmatch(start, &sub, &m) == FALSE) | |
4233 return 0; | |
4234 | |
4235 cleanup_subexpr(); | |
4236 if (REG_MULTI) | |
4237 { | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4238 for (i = 0; i < sub.in_use; i++) |
4444 | 4239 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4240 reg_startpos[i] = sub.list.multi[i].start; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4241 reg_endpos[i] = sub.list.multi[i].end; |
4444 | 4242 } |
4243 | |
4244 if (reg_startpos[0].lnum < 0) | |
4245 { | |
4246 reg_startpos[0].lnum = 0; | |
4247 reg_startpos[0].col = col; | |
4248 } | |
4249 if (reg_endpos[0].lnum < 0) | |
4250 { | |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
4251 /* pattern has a \ze but it didn't match, use current end */ |
4444 | 4252 reg_endpos[0].lnum = reglnum; |
4253 reg_endpos[0].col = (int)(reginput - regline); | |
4254 } | |
4255 else | |
4256 /* Use line number of "\ze". */ | |
4257 reglnum = reg_endpos[0].lnum; | |
4258 } | |
4259 else | |
4260 { | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4261 for (i = 0; i < sub.in_use; i++) |
4444 | 4262 { |
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4263 reg_startp[i] = sub.list.line[i].start; |
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4264 reg_endp[i] = sub.list.line[i].end; |
4444 | 4265 } |
4266 | |
4267 if (reg_startp[0] == NULL) | |
4268 reg_startp[0] = regline + col; | |
4269 if (reg_endp[0] == NULL) | |
4270 reg_endp[0] = reginput; | |
4271 } | |
4272 | |
4273 return 1 + reglnum; | |
4274 } | |
4275 | |
4276 /* | |
4277 * Match a regexp against a string ("line" points to the string) or multiple | |
4278 * lines ("line" is NULL, use reg_getline()). | |
4279 * | |
4280 * Returns 0 for failure, number of lines contained in the match otherwise. | |
4281 */ | |
4282 static long | |
4283 nfa_regexec_both(line, col) | |
4284 char_u *line; | |
4285 colnr_T col; /* column to start looking for match */ | |
4286 { | |
4287 nfa_regprog_T *prog; | |
4288 long retval = 0L; | |
4289 int i; | |
4290 | |
4291 if (REG_MULTI) | |
4292 { | |
4293 prog = (nfa_regprog_T *)reg_mmatch->regprog; | |
4294 line = reg_getline((linenr_T)0); /* relative to the cursor */ | |
4295 reg_startpos = reg_mmatch->startpos; | |
4296 reg_endpos = reg_mmatch->endpos; | |
4297 } | |
4298 else | |
4299 { | |
4300 prog = (nfa_regprog_T *)reg_match->regprog; | |
4301 reg_startp = reg_match->startp; | |
4302 reg_endp = reg_match->endp; | |
4303 } | |
4304 | |
4305 /* Be paranoid... */ | |
4306 if (prog == NULL || line == NULL) | |
4307 { | |
4308 EMSG(_(e_null)); | |
4309 goto theend; | |
4310 } | |
4311 | |
4312 /* If the start column is past the maximum column: no need to try. */ | |
4313 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
4314 goto theend; | |
4315 | |
4316 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */ | |
4317 if (prog->regflags & RF_ICASE) | |
4318 ireg_ic = TRUE; | |
4319 else if (prog->regflags & RF_NOICASE) | |
4320 ireg_ic = FALSE; | |
4321 | |
4322 #ifdef FEAT_MBYTE | |
4323 /* If pattern contains "\Z" overrule value of ireg_icombine */ | |
4324 if (prog->regflags & RF_ICOMBINE) | |
4325 ireg_icombine = TRUE; | |
4326 #endif | |
4327 | |
4328 regline = line; | |
4329 reglnum = 0; /* relative to line */ | |
4330 | |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4331 nfa_has_zend = prog->has_zend; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
4332 nfa_has_backref = prog->has_backref; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4333 nfa_nsubexpr = prog->nsubexp; |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4334 |
4444 | 4335 nstate = prog->nstate; |
4336 for (i = 0; i < nstate; ++i) | |
4337 { | |
4338 prog->state[i].id = i; | |
4339 prog->state[i].lastlist = 0; | |
4340 } | |
4341 | |
4342 retval = nfa_regtry(prog->start, col); | |
4343 | |
4344 theend: | |
4345 return retval; | |
4346 } | |
4347 | |
4348 /* | |
4349 * Compile a regular expression into internal code for the NFA matcher. | |
4350 * Returns the program in allocated space. Returns NULL for an error. | |
4351 */ | |
4352 static regprog_T * | |
4353 nfa_regcomp(expr, re_flags) | |
4354 char_u *expr; | |
4355 int re_flags; | |
4356 { | |
4541
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4357 nfa_regprog_T *prog = NULL; |
4458 | 4358 size_t prog_size; |
4444 | 4359 int *postfix; |
4360 | |
4361 if (expr == NULL) | |
4362 return NULL; | |
4363 | |
4364 #ifdef DEBUG | |
4365 nfa_regengine.expr = expr; | |
4366 #endif | |
4367 | |
4368 init_class_tab(); | |
4369 | |
4370 if (nfa_regcomp_start(expr, re_flags) == FAIL) | |
4371 return NULL; | |
4372 | |
4373 /* 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
|
4374 * (and count its size). */ |
4444 | 4375 postfix = re2post(); |
4376 if (postfix == NULL) | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4377 { |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4378 /* TODO: only give this error for debugging? */ |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4379 if (post_ptr >= post_end) |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4380 EMSGN("Internal error: estimated max number of states insufficient: %ld", post_end - post_start); |
4444 | 4381 goto fail; /* Cascaded (syntax?) error */ |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4382 } |
4444 | 4383 |
4384 /* | |
4385 * In order to build the NFA, we parse the input regexp twice: | |
4386 * 1. first pass to count size (so we can allocate space) | |
4387 * 2. second to emit code | |
4388 */ | |
4389 #ifdef ENABLE_LOG | |
4390 { | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
4391 FILE *f = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 4392 |
4393 if (f != NULL) | |
4394 { | |
4395 fprintf(f, "\n*****************************\n\n\n\n\tCompiling regexp \"%s\" ... hold on !\n", expr); | |
4396 fclose(f); | |
4397 } | |
4398 } | |
4399 #endif | |
4400 | |
4401 /* | |
4402 * PASS 1 | |
4403 * Count number of NFA states in "nstate". Do not build the NFA. | |
4404 */ | |
4405 post2nfa(postfix, post_ptr, TRUE); | |
4541
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4406 |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4407 /* Space for compiled regexp */ |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4408 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
|
4409 prog = (nfa_regprog_T *)lalloc(prog_size, TRUE); |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4410 if (prog == NULL) |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4411 goto fail; |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4412 vim_memset(prog, 0, prog_size); |
4444 | 4413 state_ptr = prog->state; |
4414 | |
4415 /* | |
4416 * PASS 2 | |
4417 * Build the NFA | |
4418 */ | |
4419 prog->start = post2nfa(postfix, post_ptr, FALSE); | |
4420 if (prog->start == NULL) | |
4421 goto fail; | |
4422 | |
4423 prog->regflags = regflags; | |
4424 prog->engine = &nfa_regengine; | |
4425 prog->nstate = nstate; | |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4426 prog->has_zend = nfa_has_zend; |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4615
diff
changeset
|
4427 prog->has_backref = nfa_has_backref; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4428 prog->nsubexp = regnpar; |
4444 | 4429 #ifdef ENABLE_LOG |
4430 nfa_postfix_dump(expr, OK); | |
4431 nfa_dump(prog); | |
4432 #endif | |
4433 | |
4434 out: | |
4435 vim_free(post_start); | |
4436 post_start = post_ptr = post_end = NULL; | |
4437 state_ptr = NULL; | |
4438 return (regprog_T *)prog; | |
4439 | |
4440 fail: | |
4441 vim_free(prog); | |
4442 prog = NULL; | |
4443 #ifdef ENABLE_LOG | |
4444 nfa_postfix_dump(expr, FAIL); | |
4445 #endif | |
4446 #ifdef DEBUG | |
4447 nfa_regengine.expr = NULL; | |
4448 #endif | |
4449 goto out; | |
4450 } | |
4451 | |
4452 | |
4453 /* | |
4454 * Match a regexp against a string. | |
4455 * "rmp->regprog" is a compiled regexp as returned by nfa_regcomp(). | |
4456 * Uses curbuf for line count and 'iskeyword'. | |
4457 * | |
4458 * Return TRUE if there is a match, FALSE if not. | |
4459 */ | |
4460 static int | |
4461 nfa_regexec(rmp, line, col) | |
4462 regmatch_T *rmp; | |
4463 char_u *line; /* string to match against */ | |
4464 colnr_T col; /* column to start looking for match */ | |
4465 { | |
4466 reg_match = rmp; | |
4467 reg_mmatch = NULL; | |
4468 reg_maxline = 0; | |
4469 reg_line_lbr = FALSE; | |
4470 reg_buf = curbuf; | |
4471 reg_win = NULL; | |
4472 ireg_ic = rmp->rm_ic; | |
4473 #ifdef FEAT_MBYTE | |
4474 ireg_icombine = FALSE; | |
4475 #endif | |
4476 ireg_maxcol = 0; | |
4477 return (nfa_regexec_both(line, col) != 0); | |
4478 } | |
4479 | |
4480 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
4481 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
4482 | |
4483 static int nfa_regexec_nl __ARGS((regmatch_T *rmp, char_u *line, colnr_T col)); | |
4484 | |
4485 /* | |
4486 * Like nfa_regexec(), but consider a "\n" in "line" to be a line break. | |
4487 */ | |
4488 static int | |
4489 nfa_regexec_nl(rmp, line, col) | |
4490 regmatch_T *rmp; | |
4491 char_u *line; /* string to match against */ | |
4492 colnr_T col; /* column to start looking for match */ | |
4493 { | |
4494 reg_match = rmp; | |
4495 reg_mmatch = NULL; | |
4496 reg_maxline = 0; | |
4497 reg_line_lbr = TRUE; | |
4498 reg_buf = curbuf; | |
4499 reg_win = NULL; | |
4500 ireg_ic = rmp->rm_ic; | |
4501 #ifdef FEAT_MBYTE | |
4502 ireg_icombine = FALSE; | |
4503 #endif | |
4504 ireg_maxcol = 0; | |
4505 return (nfa_regexec_both(line, col) != 0); | |
4506 } | |
4507 #endif | |
4508 | |
4509 | |
4510 /* | |
4511 * Match a regexp against multiple lines. | |
4512 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
4513 * Uses curbuf for line count and 'iskeyword'. | |
4514 * | |
4515 * Return zero if there is no match. Return number of lines contained in the | |
4516 * match otherwise. | |
4517 * | |
4518 * Note: the body is the same as bt_regexec() except for nfa_regexec_both() | |
4519 * | |
4520 * ! Also NOTE : match may actually be in another line. e.g.: | |
4521 * when r.e. is \nc, cursor is at 'a' and the text buffer looks like | |
4522 * | |
4523 * +-------------------------+ | |
4524 * |a | | |
4525 * |b | | |
4526 * |c | | |
4527 * | | | |
4528 * +-------------------------+ | |
4529 * | |
4530 * then nfa_regexec_multi() returns 3. while the original | |
4531 * vim_regexec_multi() returns 0 and a second call at line 2 will return 2. | |
4532 * | |
4533 * FIXME if this behavior is not compatible. | |
4534 */ | |
4535 static long | |
4536 nfa_regexec_multi(rmp, win, buf, lnum, col, tm) | |
4537 regmmatch_T *rmp; | |
4538 win_T *win; /* window in which to search or NULL */ | |
4539 buf_T *buf; /* buffer in which to search */ | |
4540 linenr_T lnum; /* nr of line to start looking for match */ | |
4541 colnr_T col; /* column to start looking for match */ | |
4542 proftime_T *tm UNUSED; /* timeout limit or NULL */ | |
4543 { | |
4544 reg_match = NULL; | |
4545 reg_mmatch = rmp; | |
4546 reg_buf = buf; | |
4547 reg_win = win; | |
4548 reg_firstlnum = lnum; | |
4549 reg_maxline = reg_buf->b_ml.ml_line_count - lnum; | |
4550 reg_line_lbr = FALSE; | |
4551 ireg_ic = rmp->rmm_ic; | |
4552 #ifdef FEAT_MBYTE | |
4553 ireg_icombine = FALSE; | |
4554 #endif | |
4555 ireg_maxcol = rmp->rmm_maxcol; | |
4556 | |
4503 | 4557 return nfa_regexec_both(NULL, col); |
4444 | 4558 } |
4559 | |
4560 #ifdef DEBUG | |
4561 # undef ENABLE_LOG | |
4562 #endif |