Mercurial > vim
annotate src/regexp_nfa.c @ 4579:7a2be4a39423 v7.3.1037
updated for version 7.3.1037
Problem: Look-behind matching is very slow on long lines.
Solution: Add a byte limit to how far back an attempt is made.
| author | Bram Moolenaar <bram@vim.org> |
|---|---|
| date | Wed, 29 May 2013 18:45:11 +0200 |
| parents | b22bff1a6af8 |
| children | 321cfbef9431 |
| 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 */ | |
| 120 NFA_FIRST_NL = NFA_ANY + ADD_NL, | |
| 121 NFA_LAST_NL = NFA_NUPPER + ADD_NL, | |
| 122 | |
| 123 /* Character classes [:alnum:] etc */ | |
| 124 NFA_CLASS_ALNUM, | |
| 125 NFA_CLASS_ALPHA, | |
| 126 NFA_CLASS_BLANK, | |
| 127 NFA_CLASS_CNTRL, | |
| 128 NFA_CLASS_DIGIT, | |
| 129 NFA_CLASS_GRAPH, | |
| 130 NFA_CLASS_LOWER, | |
| 131 NFA_CLASS_PRINT, | |
| 132 NFA_CLASS_PUNCT, | |
| 133 NFA_CLASS_SPACE, | |
| 134 NFA_CLASS_UPPER, | |
| 135 NFA_CLASS_XDIGIT, | |
| 136 NFA_CLASS_TAB, | |
| 137 NFA_CLASS_RETURN, | |
| 138 NFA_CLASS_BACKSPACE, | |
| 139 NFA_CLASS_ESCAPE | |
| 140 }; | |
| 141 | |
| 142 /* Keep in sync with classchars. */ | |
| 143 static int nfa_classcodes[] = { | |
| 144 NFA_ANY, NFA_IDENT, NFA_SIDENT, NFA_KWORD,NFA_SKWORD, | |
| 145 NFA_FNAME, NFA_SFNAME, NFA_PRINT, NFA_SPRINT, | |
| 146 NFA_WHITE, NFA_NWHITE, NFA_DIGIT, NFA_NDIGIT, | |
| 147 NFA_HEX, NFA_NHEX, NFA_OCTAL, NFA_NOCTAL, | |
| 148 NFA_WORD, NFA_NWORD, NFA_HEAD, NFA_NHEAD, | |
| 149 NFA_ALPHA, NFA_NALPHA, NFA_LOWER, NFA_NLOWER, | |
| 150 NFA_UPPER, NFA_NUPPER | |
| 151 }; | |
| 152 | |
| 153 static char_u e_misplaced[] = N_("E866: (NFA regexp) Misplaced %c"); | |
| 154 | |
| 155 /* | |
| 156 * NFA errors can be of 3 types: | |
| 157 * *** NFA runtime errors, when something unknown goes wrong. The NFA fails | |
| 158 * silently and revert the to backtracking engine. | |
| 159 * syntax_error = FALSE; | |
| 160 * *** Regexp syntax errors, when the input regexp is not syntactically correct. | |
| 161 * The NFA engine displays an error message, and nothing else happens. | |
| 162 * syntax_error = TRUE | |
| 163 * *** Unsupported features, when the input regexp uses an operator that is not | |
| 164 * implemented in the NFA. The NFA engine fails silently, and reverts to the | |
| 165 * old backtracking engine. | |
| 166 * syntax_error = FALSE | |
| 167 * "The NFA fails" means that "compiling the regexp with the NFA fails": | |
| 168 * nfa_regcomp() returns FAIL. | |
| 169 */ | |
| 170 static int syntax_error = FALSE; | |
| 171 | |
| 172 /* NFA regexp \ze operator encountered. */ | |
|
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
173 static int nfa_has_zend; |
| 4444 | 174 |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
175 /* 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
|
176 * the whole match (subexpr 0) is used. */ |
|
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
177 static int nfa_nsubexpr; |
|
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
178 |
| 4444 | 179 static int *post_start; /* holds the postfix form of r.e. */ |
| 180 static int *post_end; | |
| 181 static int *post_ptr; | |
| 182 | |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
183 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
|
184 * executing. */ |
| 4444 | 185 static int istate; /* Index in the state vector, used in new_state() */ |
| 186 | |
| 187 | |
| 188 static int nfa_regcomp_start __ARGS((char_u*expr, int re_flags)); | |
| 189 static int nfa_recognize_char_class __ARGS((char_u *start, char_u *end, int extra_newl)); | |
| 190 static int nfa_emit_equi_class __ARGS((int c, int neg)); | |
| 191 static int nfa_regatom __ARGS((void)); | |
| 192 static int nfa_regpiece __ARGS((void)); | |
| 193 static int nfa_regconcat __ARGS((void)); | |
| 194 static int nfa_regbranch __ARGS((void)); | |
| 195 static int nfa_reg __ARGS((int paren)); | |
| 196 #ifdef DEBUG | |
| 197 static void nfa_set_code __ARGS((int c)); | |
| 198 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
|
199 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
|
200 static void nfa_print_state2 __ARGS((FILE *debugf, nfa_state_T *state, garray_T *indent)); |
| 4444 | 201 static void nfa_dump __ARGS((nfa_regprog_T *prog)); |
| 202 #endif | |
| 203 static int *re2post __ARGS((void)); | |
| 204 static nfa_state_T *new_state __ARGS((int c, nfa_state_T *out, nfa_state_T *out1)); | |
| 205 static nfa_state_T *post2nfa __ARGS((int *postfix, int *end, int nfa_calc_size)); | |
| 206 static int check_char_class __ARGS((int class, int c)); | |
| 207 static void st_error __ARGS((int *postfix, int *end, int *p)); | |
| 208 static void nfa_save_listids __ARGS((nfa_state_T *start, int *list)); | |
| 209 static void nfa_restore_listids __ARGS((nfa_state_T *start, int *list)); | |
| 210 static void nfa_set_null_listids __ARGS((nfa_state_T *start)); | |
| 211 static void nfa_set_neg_listids __ARGS((nfa_state_T *start)); | |
| 212 static long nfa_regtry __ARGS((nfa_state_T *start, colnr_T col)); | |
| 213 static long nfa_regexec_both __ARGS((char_u *line, colnr_T col)); | |
| 214 static regprog_T *nfa_regcomp __ARGS((char_u *expr, int re_flags)); | |
| 215 static int nfa_regexec __ARGS((regmatch_T *rmp, char_u *line, colnr_T col)); | |
| 216 static long nfa_regexec_multi __ARGS((regmmatch_T *rmp, win_T *win, buf_T *buf, linenr_T lnum, colnr_T col, proftime_T *tm)); | |
| 217 | |
| 218 /* helper functions used when doing re2post() ... regatom() parsing */ | |
| 219 #define EMIT(c) do { \ | |
| 220 if (post_ptr >= post_end) \ | |
| 221 return FAIL; \ | |
| 222 *post_ptr++ = c; \ | |
| 223 } while (0) | |
| 224 | |
| 225 /* | |
| 226 * Initialize internal variables before NFA compilation. | |
| 227 * Return OK on success, FAIL otherwise. | |
| 228 */ | |
| 229 static int | |
| 230 nfa_regcomp_start(expr, re_flags) | |
| 231 char_u *expr; | |
| 232 int re_flags; /* see vim_regcomp() */ | |
| 233 { | |
| 4458 | 234 size_t postfix_size; |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
235 int nstate_max; |
| 4444 | 236 |
| 237 nstate = 0; | |
| 238 istate = 0; | |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
239 /* A reasonable estimation for maximum size */ |
| 4458 | 240 nstate_max = (int)(STRLEN(expr) + 1) * NFA_POSTFIX_MULTIPLIER; |
| 4444 | 241 |
| 4454 | 242 /* Some items blow up in size, such as [A-z]. Add more space for that. |
| 243 * TODO: some patterns may still fail. */ | |
| 4458 | 244 nstate_max += 1000; |
| 4454 | 245 |
| 246 /* Size for postfix representation of expr. */ | |
| 4444 | 247 postfix_size = sizeof(*post_start) * nstate_max; |
| 4454 | 248 |
| 4444 | 249 post_start = (int *)lalloc(postfix_size, TRUE); |
| 250 if (post_start == NULL) | |
| 251 return FAIL; | |
| 252 vim_memset(post_start, 0, postfix_size); | |
| 253 post_ptr = post_start; | |
| 4454 | 254 post_end = post_start + nstate_max; |
| 4444 | 255 nfa_has_zend = FALSE; |
| 256 | |
| 257 regcomp_start(expr, re_flags); | |
| 258 | |
| 259 return OK; | |
| 260 } | |
| 261 | |
| 262 /* | |
| 263 * Search between "start" and "end" and try to recognize a | |
| 264 * character class in expanded form. For example [0-9]. | |
| 265 * On success, return the id the character class to be emitted. | |
| 266 * On failure, return 0 (=FAIL) | |
| 267 * Start points to the first char of the range, while end should point | |
| 268 * to the closing brace. | |
| 269 */ | |
| 270 static int | |
| 271 nfa_recognize_char_class(start, end, extra_newl) | |
| 272 char_u *start; | |
| 273 char_u *end; | |
| 274 int extra_newl; | |
| 275 { | |
| 276 int i; | |
| 277 /* Each of these variables takes up a char in "config[]", | |
| 278 * in the order they are here. */ | |
| 279 int not = FALSE, af = FALSE, AF = FALSE, az = FALSE, AZ = FALSE, | |
| 280 o7 = FALSE, o9 = FALSE, underscore = FALSE, newl = FALSE; | |
| 281 char_u *p; | |
| 282 #define NCONFIGS 16 | |
| 283 int classid[NCONFIGS] = { | |
| 284 NFA_DIGIT, NFA_NDIGIT, NFA_HEX, NFA_NHEX, | |
| 285 NFA_OCTAL, NFA_NOCTAL, NFA_WORD, NFA_NWORD, | |
| 286 NFA_HEAD, NFA_NHEAD, NFA_ALPHA, NFA_NALPHA, | |
| 287 NFA_LOWER, NFA_NLOWER, NFA_UPPER, NFA_NUPPER | |
| 288 }; | |
| 4450 | 289 char_u myconfig[10]; |
| 4444 | 290 char_u config[NCONFIGS][9] = { |
| 291 "000000100", /* digit */ | |
| 292 "100000100", /* non digit */ | |
| 293 "011000100", /* hex-digit */ | |
| 294 "111000100", /* non hex-digit */ | |
| 295 "000001000", /* octal-digit */ | |
| 296 "100001000", /* [^0-7] */ | |
| 297 "000110110", /* [0-9A-Za-z_] */ | |
| 298 "100110110", /* [^0-9A-Za-z_] */ | |
| 299 "000110010", /* head of word */ | |
| 300 "100110010", /* not head of word */ | |
| 301 "000110000", /* alphabetic char a-z */ | |
| 302 "100110000", /* non alphabetic char */ | |
| 303 "000100000", /* lowercase letter */ | |
| 304 "100100000", /* non lowercase */ | |
| 305 "000010000", /* uppercase */ | |
| 306 "100010000" /* non uppercase */ | |
| 307 }; | |
| 308 | |
| 309 if (extra_newl == TRUE) | |
| 310 newl = TRUE; | |
| 311 | |
| 312 if (*end != ']') | |
| 313 return FAIL; | |
| 314 p = start; | |
| 315 if (*p == '^') | |
| 316 { | |
| 317 not = TRUE; | |
| 318 p ++; | |
| 319 } | |
| 320 | |
| 321 while (p < end) | |
| 322 { | |
| 323 if (p + 2 < end && *(p + 1) == '-') | |
| 324 { | |
| 325 switch (*p) | |
| 326 { | |
| 327 case '0': | |
| 328 if (*(p + 2) == '9') | |
| 329 { | |
| 330 o9 = TRUE; | |
| 331 break; | |
| 332 } | |
| 333 else | |
| 334 if (*(p + 2) == '7') | |
| 335 { | |
| 336 o7 = TRUE; | |
| 337 break; | |
| 338 } | |
| 339 case 'a': | |
| 340 if (*(p + 2) == 'z') | |
| 341 { | |
| 342 az = TRUE; | |
| 343 break; | |
| 344 } | |
| 345 else | |
| 346 if (*(p + 2) == 'f') | |
| 347 { | |
| 348 af = TRUE; | |
| 349 break; | |
| 350 } | |
| 351 case 'A': | |
| 352 if (*(p + 2) == 'Z') | |
| 353 { | |
| 354 AZ = TRUE; | |
| 355 break; | |
| 356 } | |
| 357 else | |
| 358 if (*(p + 2) == 'F') | |
| 359 { | |
| 360 AF = TRUE; | |
| 361 break; | |
| 362 } | |
| 363 /* FALLTHROUGH */ | |
| 364 default: | |
| 365 return FAIL; | |
| 366 } | |
| 367 p += 3; | |
| 368 } | |
| 369 else if (p + 1 < end && *p == '\\' && *(p + 1) == 'n') | |
| 370 { | |
| 371 newl = TRUE; | |
| 372 p += 2; | |
| 373 } | |
| 374 else if (*p == '_') | |
| 375 { | |
| 376 underscore = TRUE; | |
| 377 p ++; | |
| 378 } | |
| 379 else if (*p == '\n') | |
| 380 { | |
| 381 newl = TRUE; | |
| 382 p ++; | |
| 383 } | |
| 384 else | |
| 385 return FAIL; | |
| 386 } /* while (p < end) */ | |
| 387 | |
| 388 if (p != end) | |
| 389 return FAIL; | |
| 390 | |
| 391 /* build the config that represents the ranges we gathered */ | |
| 392 STRCPY(myconfig, "000000000"); | |
| 393 if (not == TRUE) | |
| 394 myconfig[0] = '1'; | |
| 395 if (af == TRUE) | |
| 396 myconfig[1] = '1'; | |
| 397 if (AF == TRUE) | |
| 398 myconfig[2] = '1'; | |
| 399 if (az == TRUE) | |
| 400 myconfig[3] = '1'; | |
| 401 if (AZ == TRUE) | |
| 402 myconfig[4] = '1'; | |
| 403 if (o7 == TRUE) | |
| 404 myconfig[5] = '1'; | |
| 405 if (o9 == TRUE) | |
| 406 myconfig[6] = '1'; | |
| 407 if (underscore == TRUE) | |
| 408 myconfig[7] = '1'; | |
| 409 if (newl == TRUE) | |
| 410 { | |
| 411 myconfig[8] = '1'; | |
| 412 extra_newl = ADD_NL; | |
| 413 } | |
| 414 /* try to recognize character classes */ | |
| 415 for (i = 0; i < NCONFIGS; i++) | |
| 4450 | 416 if (STRNCMP(myconfig, config[i], 8) == 0) |
| 4444 | 417 return classid[i] + extra_newl; |
| 418 | |
| 419 /* fallthrough => no success so far */ | |
| 420 return FAIL; | |
| 421 | |
| 422 #undef NCONFIGS | |
| 423 } | |
| 424 | |
| 425 /* | |
| 426 * Produce the bytes for equivalence class "c". | |
| 427 * Currently only handles latin1, latin9 and utf-8. | |
| 428 * Emits bytes in postfix notation: 'a,b,NFA_OR,c,NFA_OR' is | |
| 429 * equivalent to 'a OR b OR c' | |
| 430 * | |
| 431 * NOTE! When changing this function, also update reg_equi_class() | |
| 432 */ | |
| 433 static int | |
| 434 nfa_emit_equi_class(c, neg) | |
| 435 int c; | |
| 436 int neg; | |
| 437 { | |
| 438 int first = TRUE; | |
| 439 int glue = neg == TRUE ? NFA_CONCAT : NFA_OR; | |
| 440 #define EMIT2(c) \ | |
| 441 EMIT(c); \ | |
| 442 if (neg == TRUE) { \ | |
| 443 EMIT(NFA_NOT); \ | |
| 444 } \ | |
| 445 if (first == FALSE) \ | |
| 446 EMIT(glue); \ | |
| 447 else \ | |
| 448 first = FALSE; \ | |
| 449 | |
| 450 #ifdef FEAT_MBYTE | |
| 451 if (enc_utf8 || STRCMP(p_enc, "latin1") == 0 | |
| 452 || STRCMP(p_enc, "iso-8859-15") == 0) | |
| 453 #endif | |
| 454 { | |
| 455 switch (c) | |
| 456 { | |
| 457 case 'A': case '\300': case '\301': case '\302': | |
| 458 case '\303': case '\304': case '\305': | |
| 459 EMIT2('A'); EMIT2('\300'); EMIT2('\301'); | |
| 460 EMIT2('\302'); EMIT2('\303'); EMIT2('\304'); | |
| 461 EMIT2('\305'); | |
| 462 return OK; | |
| 463 | |
| 464 case 'C': case '\307': | |
| 465 EMIT2('C'); EMIT2('\307'); | |
| 466 return OK; | |
| 467 | |
| 468 case 'E': case '\310': case '\311': case '\312': case '\313': | |
| 469 EMIT2('E'); EMIT2('\310'); EMIT2('\311'); | |
| 470 EMIT2('\312'); EMIT2('\313'); | |
| 471 return OK; | |
| 472 | |
| 473 case 'I': case '\314': case '\315': case '\316': case '\317': | |
| 474 EMIT2('I'); EMIT2('\314'); EMIT2('\315'); | |
| 475 EMIT2('\316'); EMIT2('\317'); | |
| 476 return OK; | |
| 477 | |
| 478 case 'N': case '\321': | |
| 479 EMIT2('N'); EMIT2('\321'); | |
| 480 return OK; | |
| 481 | |
| 482 case 'O': case '\322': case '\323': case '\324': case '\325': | |
| 483 case '\326': | |
| 484 EMIT2('O'); EMIT2('\322'); EMIT2('\323'); | |
| 485 EMIT2('\324'); EMIT2('\325'); EMIT2('\326'); | |
| 486 return OK; | |
| 487 | |
| 488 case 'U': case '\331': case '\332': case '\333': case '\334': | |
| 489 EMIT2('U'); EMIT2('\331'); EMIT2('\332'); | |
| 490 EMIT2('\333'); EMIT2('\334'); | |
| 491 return OK; | |
| 492 | |
| 493 case 'Y': case '\335': | |
| 494 EMIT2('Y'); EMIT2('\335'); | |
| 495 return OK; | |
| 496 | |
| 497 case 'a': case '\340': case '\341': case '\342': | |
| 498 case '\343': case '\344': case '\345': | |
| 499 EMIT2('a'); EMIT2('\340'); EMIT2('\341'); | |
| 500 EMIT2('\342'); EMIT2('\343'); EMIT2('\344'); | |
| 501 EMIT2('\345'); | |
| 502 return OK; | |
| 503 | |
| 504 case 'c': case '\347': | |
| 505 EMIT2('c'); EMIT2('\347'); | |
| 506 return OK; | |
| 507 | |
| 508 case 'e': case '\350': case '\351': case '\352': case '\353': | |
| 509 EMIT2('e'); EMIT2('\350'); EMIT2('\351'); | |
| 510 EMIT2('\352'); EMIT2('\353'); | |
| 511 return OK; | |
| 512 | |
| 513 case 'i': case '\354': case '\355': case '\356': case '\357': | |
| 514 EMIT2('i'); EMIT2('\354'); EMIT2('\355'); | |
| 515 EMIT2('\356'); EMIT2('\357'); | |
| 516 return OK; | |
| 517 | |
| 518 case 'n': case '\361': | |
| 519 EMIT2('n'); EMIT2('\361'); | |
| 520 return OK; | |
| 521 | |
| 522 case 'o': case '\362': case '\363': case '\364': case '\365': | |
| 523 case '\366': | |
| 524 EMIT2('o'); EMIT2('\362'); EMIT2('\363'); | |
| 525 EMIT2('\364'); EMIT2('\365'); EMIT2('\366'); | |
| 526 return OK; | |
| 527 | |
| 528 case 'u': case '\371': case '\372': case '\373': case '\374': | |
| 529 EMIT2('u'); EMIT2('\371'); EMIT2('\372'); | |
| 530 EMIT2('\373'); EMIT2('\374'); | |
| 531 return OK; | |
| 532 | |
| 533 case 'y': case '\375': case '\377': | |
| 534 EMIT2('y'); EMIT2('\375'); EMIT2('\377'); | |
| 535 return OK; | |
| 536 | |
| 537 default: | |
| 538 return FAIL; | |
| 539 } | |
| 540 } | |
| 541 | |
| 542 EMIT(c); | |
| 543 return OK; | |
| 544 #undef EMIT2 | |
| 545 } | |
| 546 | |
| 547 /* | |
| 548 * Code to parse regular expression. | |
| 549 * | |
| 550 * We try to reuse parsing functions in regexp.c to | |
| 551 * minimize surprise and keep the syntax consistent. | |
| 552 */ | |
| 553 | |
| 554 /* | |
| 555 * Parse the lowest level. | |
| 556 * | |
| 557 * An atom can be one of a long list of items. Many atoms match one character | |
| 558 * in the text. It is often an ordinary character or a character class. | |
| 559 * Braces can be used to make a pattern into an atom. The "\z(\)" construct | |
| 560 * is only for syntax highlighting. | |
| 561 * | |
| 562 * atom ::= ordinary-atom | |
| 563 * or \( pattern \) | |
| 564 * or \%( pattern \) | |
| 565 * or \z( pattern \) | |
| 566 */ | |
| 567 static int | |
| 568 nfa_regatom() | |
| 569 { | |
| 570 int c; | |
| 571 int charclass; | |
| 572 int equiclass; | |
| 573 int collclass; | |
| 574 int got_coll_char; | |
| 575 char_u *p; | |
| 576 char_u *endp; | |
| 577 #ifdef FEAT_MBYTE | |
| 578 char_u *old_regparse = regparse; | |
| 579 #endif | |
| 580 int extra = 0; | |
| 581 int first; | |
| 582 int emit_range; | |
| 583 int negated; | |
| 584 int result; | |
| 585 int startc = -1; | |
| 586 int endc = -1; | |
| 587 int oldstartc = -1; | |
| 588 int cpo_lit; /* 'cpoptions' contains 'l' flag */ | |
| 589 int cpo_bsl; /* 'cpoptions' contains '\' flag */ | |
| 590 int glue; /* ID that will "glue" nodes together */ | |
| 591 | |
| 592 cpo_lit = vim_strchr(p_cpo, CPO_LITERAL) != NULL; | |
| 593 cpo_bsl = vim_strchr(p_cpo, CPO_BACKSL) != NULL; | |
| 594 | |
| 595 c = getchr(); | |
| 596 switch (c) | |
| 597 { | |
|
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
598 case NUL: |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
599 syntax_error = TRUE; |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
600 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
|
601 |
| 4444 | 602 case Magic('^'): |
| 603 EMIT(NFA_BOL); | |
| 604 break; | |
| 605 | |
| 606 case Magic('$'): | |
| 607 EMIT(NFA_EOL); | |
| 608 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 609 had_eol = TRUE; | |
| 610 #endif | |
| 611 break; | |
| 612 | |
| 613 case Magic('<'): | |
| 614 EMIT(NFA_BOW); | |
| 615 break; | |
| 616 | |
| 617 case Magic('>'): | |
| 618 EMIT(NFA_EOW); | |
| 619 break; | |
| 620 | |
| 621 case Magic('_'): | |
| 622 c = no_Magic(getchr()); | |
| 623 if (c == '^') /* "\_^" is start-of-line */ | |
| 624 { | |
| 625 EMIT(NFA_BOL); | |
| 626 break; | |
| 627 } | |
| 628 if (c == '$') /* "\_$" is end-of-line */ | |
| 629 { | |
| 630 EMIT(NFA_EOL); | |
| 631 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 632 had_eol = TRUE; | |
| 633 #endif | |
| 634 break; | |
| 635 } | |
| 636 | |
| 637 extra = ADD_NL; | |
| 638 | |
| 639 /* "\_[" is collection plus newline */ | |
| 640 if (c == '[') | |
|
4517
9a2183bd8295
updated for version 7.3.1006
Bram Moolenaar <bram@vim.org>
parents:
4515
diff
changeset
|
641 goto collection; |
| 4444 | 642 |
| 643 /* "\_x" is character class plus newline */ | |
| 644 /*FALLTHROUGH*/ | |
| 645 | |
| 646 /* | |
| 647 * Character classes. | |
| 648 */ | |
| 649 case Magic('.'): | |
| 650 case Magic('i'): | |
| 651 case Magic('I'): | |
| 652 case Magic('k'): | |
| 653 case Magic('K'): | |
| 654 case Magic('f'): | |
| 655 case Magic('F'): | |
| 656 case Magic('p'): | |
| 657 case Magic('P'): | |
| 658 case Magic('s'): | |
| 659 case Magic('S'): | |
| 660 case Magic('d'): | |
| 661 case Magic('D'): | |
| 662 case Magic('x'): | |
| 663 case Magic('X'): | |
| 664 case Magic('o'): | |
| 665 case Magic('O'): | |
| 666 case Magic('w'): | |
| 667 case Magic('W'): | |
| 668 case Magic('h'): | |
| 669 case Magic('H'): | |
| 670 case Magic('a'): | |
| 671 case Magic('A'): | |
| 672 case Magic('l'): | |
| 673 case Magic('L'): | |
| 674 case Magic('u'): | |
| 675 case Magic('U'): | |
| 676 p = vim_strchr(classchars, no_Magic(c)); | |
| 677 if (p == NULL) | |
| 678 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
679 EMSGN("INTERNAL: Unknown character class char: %ld", c); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
680 return FAIL; |
| 4444 | 681 } |
| 682 #ifdef FEAT_MBYTE | |
| 683 /* When '.' is followed by a composing char ignore the dot, so that | |
| 684 * the composing char is matched here. */ | |
| 685 if (enc_utf8 && c == Magic('.') && utf_iscomposing(peekchr())) | |
| 686 { | |
|
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
687 old_regparse = regparse; |
| 4444 | 688 c = getchr(); |
| 689 goto nfa_do_multibyte; | |
| 690 } | |
| 691 #endif | |
| 692 EMIT(nfa_classcodes[p - classchars]); | |
| 693 if (extra == ADD_NL) | |
| 694 { | |
| 695 EMIT(NFA_NEWL); | |
| 696 EMIT(NFA_OR); | |
| 697 regflags |= RF_HASNL; | |
| 698 } | |
| 699 break; | |
| 700 | |
| 701 case Magic('n'): | |
| 702 if (reg_string) | |
| 703 /* In a string "\n" matches a newline character. */ | |
| 704 EMIT(NL); | |
| 705 else | |
| 706 { | |
| 707 /* In buffer text "\n" matches the end of a line. */ | |
| 708 EMIT(NFA_NEWL); | |
| 709 regflags |= RF_HASNL; | |
| 710 } | |
| 711 break; | |
| 712 | |
| 713 case Magic('('): | |
| 714 if (nfa_reg(REG_PAREN) == FAIL) | |
| 715 return FAIL; /* cascaded error */ | |
| 716 break; | |
| 717 | |
| 718 case Magic('|'): | |
| 719 case Magic('&'): | |
| 720 case Magic(')'): | |
| 721 syntax_error = TRUE; | |
| 4450 | 722 EMSGN(_(e_misplaced), no_Magic(c)); |
| 4444 | 723 return FAIL; |
| 724 | |
| 725 case Magic('='): | |
| 726 case Magic('?'): | |
| 727 case Magic('+'): | |
| 728 case Magic('@'): | |
| 729 case Magic('*'): | |
| 730 case Magic('{'): | |
| 731 /* these should follow an atom, not form an atom */ | |
| 732 syntax_error = TRUE; | |
| 4450 | 733 EMSGN(_(e_misplaced), no_Magic(c)); |
| 4444 | 734 return FAIL; |
| 735 | |
| 736 case Magic('~'): /* previous substitute pattern */ | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
737 /* TODO: Not supported yet */ |
| 4444 | 738 return FAIL; |
| 739 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
740 case Magic('1'): EMIT(NFA_BACKREF1); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
741 case Magic('2'): EMIT(NFA_BACKREF2); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
742 case Magic('3'): EMIT(NFA_BACKREF3); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
743 case Magic('4'): EMIT(NFA_BACKREF4); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
744 case Magic('5'): EMIT(NFA_BACKREF5); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
745 case Magic('6'): EMIT(NFA_BACKREF6); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
746 case Magic('7'): EMIT(NFA_BACKREF7); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
747 case Magic('8'): EMIT(NFA_BACKREF8); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
748 case Magic('9'): EMIT(NFA_BACKREF9); break; |
| 4444 | 749 |
| 750 case Magic('z'): | |
| 751 c = no_Magic(getchr()); | |
| 752 switch (c) | |
| 753 { | |
| 754 case 's': | |
| 755 EMIT(NFA_ZSTART); | |
| 756 break; | |
| 757 case 'e': | |
| 758 EMIT(NFA_ZEND); | |
| 759 nfa_has_zend = TRUE; | |
|
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
760 break; |
| 4444 | 761 case '1': |
| 762 case '2': | |
| 763 case '3': | |
| 764 case '4': | |
| 765 case '5': | |
| 766 case '6': | |
| 767 case '7': | |
| 768 case '8': | |
| 769 case '9': | |
| 770 case '(': | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
771 /* TODO: \z1...\z9 and \z( not yet supported */ |
| 4444 | 772 return FAIL; |
| 773 default: | |
| 774 syntax_error = TRUE; | |
| 4450 | 775 EMSGN(_("E867: (NFA) Unknown operator '\\z%c'"), |
| 4444 | 776 no_Magic(c)); |
| 777 return FAIL; | |
| 778 } | |
| 779 break; | |
| 780 | |
| 781 case Magic('%'): | |
| 782 c = no_Magic(getchr()); | |
| 783 switch (c) | |
| 784 { | |
| 785 /* () without a back reference */ | |
| 786 case '(': | |
| 787 if (nfa_reg(REG_NPAREN) == FAIL) | |
| 788 return FAIL; | |
| 789 EMIT(NFA_NOPEN); | |
| 790 break; | |
| 791 | |
| 792 case 'd': /* %d123 decimal */ | |
| 793 case 'o': /* %o123 octal */ | |
| 794 case 'x': /* %xab hex 2 */ | |
| 795 case 'u': /* %uabcd hex 4 */ | |
| 796 case 'U': /* %U1234abcd hex 8 */ | |
|
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
797 { |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
798 int nr; |
|
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
799 |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
800 switch (c) |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
801 { |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
802 case 'd': nr = getdecchrs(); break; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
803 case 'o': nr = getoctchrs(); break; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
804 case 'x': nr = gethexchrs(2); break; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
805 case 'u': nr = gethexchrs(4); break; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
806 case 'U': nr = gethexchrs(8); break; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
807 default: nr = -1; break; |
|
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
808 } |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
809 |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
810 if (nr < 0) |
|
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
811 EMSG2_RET_FAIL( |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
812 _("E678: Invalid character after %s%%[dxouU]"), |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
813 reg_magic == MAGIC_ALL); |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
814 /* TODO: what if a composing character follows? */ |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
815 EMIT(nr); |
|
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
816 } |
| 4444 | 817 break; |
| 818 | |
| 819 /* Catch \%^ and \%$ regardless of where they appear in the | |
| 820 * pattern -- regardless of whether or not it makes sense. */ | |
| 821 case '^': | |
| 822 EMIT(NFA_BOF); | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
823 /* TODO: Not yet supported */ |
| 4444 | 824 return FAIL; |
| 825 break; | |
| 826 | |
| 827 case '$': | |
| 828 EMIT(NFA_EOF); | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
829 /* TODO: Not yet supported */ |
| 4444 | 830 return FAIL; |
| 831 break; | |
| 832 | |
| 833 case '#': | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
834 /* TODO: not supported yet */ |
| 4444 | 835 return FAIL; |
| 836 break; | |
| 837 | |
| 838 case 'V': | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
839 /* TODO: not supported yet */ |
| 4444 | 840 return FAIL; |
| 841 break; | |
| 842 | |
| 843 case '[': | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
844 /* TODO: \%[abc] not supported yet */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
845 return FAIL; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
846 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
847 case '0': |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
848 case '1': |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
849 case '2': |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
850 case '3': |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
851 case '4': |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
852 case '5': |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
853 case '6': |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
854 case '7': |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
855 case '8': |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
856 case '9': |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
857 case '<': |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
858 case '>': |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
859 case '\'': |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
860 /* TODO: not supported yet */ |
| 4444 | 861 return FAIL; |
| 862 | |
| 863 default: | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
864 syntax_error = TRUE; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
865 EMSGN(_("E867: (NFA) Unknown operator '\\%%%c'"), |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
866 no_Magic(c)); |
| 4444 | 867 return FAIL; |
| 868 } | |
| 869 break; | |
| 870 | |
| 871 case Magic('['): | |
|
4517
9a2183bd8295
updated for version 7.3.1006
Bram Moolenaar <bram@vim.org>
parents:
4515
diff
changeset
|
872 collection: |
| 4444 | 873 /* |
| 874 * Glue is emitted between several atoms from the []. | |
| 875 * It is either NFA_OR, or NFA_CONCAT. | |
| 876 * | |
| 877 * [abc] expands to 'a b NFA_OR c NFA_OR' (in postfix notation) | |
| 878 * [^abc] expands to 'a NFA_NOT b NFA_NOT NFA_CONCAT c NFA_NOT | |
| 879 * NFA_CONCAT NFA_END_NEG_RANGE NFA_CONCAT' (in postfix | |
| 880 * notation) | |
| 881 * | |
| 882 */ | |
| 883 | |
| 884 | |
| 885 /* Emit negation atoms, if needed. | |
| 886 * The CONCAT below merges the NOT with the previous node. */ | |
| 887 #define TRY_NEG() \ | |
| 888 if (negated == TRUE) \ | |
| 889 { \ | |
| 890 EMIT(NFA_NOT); \ | |
| 891 } | |
| 892 | |
| 893 /* Emit glue between important nodes : CONCAT or OR. */ | |
| 894 #define EMIT_GLUE() \ | |
| 895 if (first == FALSE) \ | |
| 896 EMIT(glue); \ | |
| 897 else \ | |
| 898 first = FALSE; | |
| 899 | |
| 900 p = regparse; | |
| 901 endp = skip_anyof(p); | |
| 902 if (*endp == ']') | |
| 903 { | |
| 904 /* | |
| 905 * Try to reverse engineer character classes. For example, | |
| 906 * recognize that [0-9] stands for \d and [A-Za-z_] with \h, | |
| 907 * and perform the necessary substitutions in the NFA. | |
| 908 */ | |
| 909 result = nfa_recognize_char_class(regparse, endp, | |
| 910 extra == ADD_NL); | |
| 911 if (result != FAIL) | |
| 912 { | |
| 913 if (result >= NFA_DIGIT && result <= NFA_NUPPER) | |
| 914 EMIT(result); | |
| 915 else /* must be char class + newline */ | |
| 916 { | |
| 917 EMIT(result - ADD_NL); | |
| 918 EMIT(NFA_NEWL); | |
| 919 EMIT(NFA_OR); | |
| 920 } | |
| 921 regparse = endp; | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
922 mb_ptr_adv(regparse); |
| 4444 | 923 return OK; |
| 924 } | |
| 925 /* | |
| 926 * Failed to recognize a character class. Use the simple | |
| 927 * version that turns [abc] into 'a' OR 'b' OR 'c' | |
| 928 */ | |
| 929 startc = endc = oldstartc = -1; | |
| 930 first = TRUE; /* Emitting first atom in this sequence? */ | |
| 931 negated = FALSE; | |
| 932 glue = NFA_OR; | |
| 933 if (*regparse == '^') /* negated range */ | |
| 934 { | |
| 935 negated = TRUE; | |
| 936 glue = NFA_CONCAT; | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
937 mb_ptr_adv(regparse); |
| 4444 | 938 } |
| 939 if (*regparse == '-') | |
| 940 { | |
| 941 startc = '-'; | |
| 942 EMIT(startc); | |
| 943 TRY_NEG(); | |
| 944 EMIT_GLUE(); | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
945 mb_ptr_adv(regparse); |
| 4444 | 946 } |
| 947 /* Emit the OR branches for each character in the [] */ | |
| 948 emit_range = FALSE; | |
| 949 while (regparse < endp) | |
| 950 { | |
| 951 oldstartc = startc; | |
| 952 startc = -1; | |
| 953 got_coll_char = FALSE; | |
| 954 if (*regparse == '[') | |
| 955 { | |
| 956 /* Check for [: :], [= =], [. .] */ | |
| 957 equiclass = collclass = 0; | |
| 958 charclass = get_char_class(®parse); | |
| 959 if (charclass == CLASS_NONE) | |
| 960 { | |
| 961 equiclass = get_equi_class(®parse); | |
| 962 if (equiclass == 0) | |
| 963 collclass = get_coll_element(®parse); | |
| 964 } | |
| 965 | |
| 966 /* Character class like [:alpha:] */ | |
| 967 if (charclass != CLASS_NONE) | |
| 968 { | |
| 969 switch (charclass) | |
| 970 { | |
| 971 case CLASS_ALNUM: | |
| 972 EMIT(NFA_CLASS_ALNUM); | |
| 973 break; | |
| 974 case CLASS_ALPHA: | |
| 975 EMIT(NFA_CLASS_ALPHA); | |
| 976 break; | |
| 977 case CLASS_BLANK: | |
| 978 EMIT(NFA_CLASS_BLANK); | |
| 979 break; | |
| 980 case CLASS_CNTRL: | |
| 981 EMIT(NFA_CLASS_CNTRL); | |
| 982 break; | |
| 983 case CLASS_DIGIT: | |
| 984 EMIT(NFA_CLASS_DIGIT); | |
| 985 break; | |
| 986 case CLASS_GRAPH: | |
| 987 EMIT(NFA_CLASS_GRAPH); | |
| 988 break; | |
| 989 case CLASS_LOWER: | |
| 990 EMIT(NFA_CLASS_LOWER); | |
| 991 break; | |
| 992 case CLASS_PRINT: | |
| 993 EMIT(NFA_CLASS_PRINT); | |
| 994 break; | |
| 995 case CLASS_PUNCT: | |
| 996 EMIT(NFA_CLASS_PUNCT); | |
| 997 break; | |
| 998 case CLASS_SPACE: | |
| 999 EMIT(NFA_CLASS_SPACE); | |
| 1000 break; | |
| 1001 case CLASS_UPPER: | |
| 1002 EMIT(NFA_CLASS_UPPER); | |
| 1003 break; | |
| 1004 case CLASS_XDIGIT: | |
| 1005 EMIT(NFA_CLASS_XDIGIT); | |
| 1006 break; | |
| 1007 case CLASS_TAB: | |
| 1008 EMIT(NFA_CLASS_TAB); | |
| 1009 break; | |
| 1010 case CLASS_RETURN: | |
| 1011 EMIT(NFA_CLASS_RETURN); | |
| 1012 break; | |
| 1013 case CLASS_BACKSPACE: | |
| 1014 EMIT(NFA_CLASS_BACKSPACE); | |
| 1015 break; | |
| 1016 case CLASS_ESCAPE: | |
| 1017 EMIT(NFA_CLASS_ESCAPE); | |
| 1018 break; | |
| 1019 } | |
| 1020 TRY_NEG(); | |
| 1021 EMIT_GLUE(); | |
| 1022 continue; | |
| 1023 } | |
| 1024 /* Try equivalence class [=a=] and the like */ | |
| 1025 if (equiclass != 0) | |
| 1026 { | |
| 1027 result = nfa_emit_equi_class(equiclass, negated); | |
| 1028 if (result == FAIL) | |
| 1029 { | |
| 1030 /* should never happen */ | |
| 1031 EMSG_RET_FAIL(_("E868: Error building NFA with equivalence class!")); | |
| 1032 } | |
| 1033 EMIT_GLUE(); | |
| 1034 continue; | |
| 1035 } | |
| 1036 /* Try collating class like [. .] */ | |
| 1037 if (collclass != 0) | |
| 1038 { | |
| 1039 startc = collclass; /* allow [.a.]-x as a range */ | |
| 1040 /* Will emit the proper atom at the end of the | |
| 1041 * while loop. */ | |
| 1042 } | |
| 1043 } | |
| 1044 /* Try a range like 'a-x' or '\t-z' */ | |
| 1045 if (*regparse == '-') | |
| 1046 { | |
| 1047 emit_range = TRUE; | |
| 1048 startc = oldstartc; | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1049 mb_ptr_adv(regparse); |
| 4444 | 1050 continue; /* reading the end of the range */ |
| 1051 } | |
| 1052 | |
| 1053 /* Now handle simple and escaped characters. | |
| 1054 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim | |
| 1055 * accepts "\t", "\e", etc., but only when the 'l' flag in | |
| 1056 * 'cpoptions' is not included. | |
| 1057 * Posix doesn't recognize backslash at all. | |
| 1058 */ | |
| 1059 if (*regparse == '\\' | |
| 1060 && !cpo_bsl | |
| 1061 && regparse + 1 <= endp | |
| 1062 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL | |
| 1063 || (!cpo_lit | |
| 1064 && vim_strchr(REGEXP_ABBR, regparse[1]) | |
| 1065 != NULL) | |
| 1066 ) | |
| 1067 ) | |
| 1068 { | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1069 mb_ptr_adv(regparse); |
| 4444 | 1070 |
|
4507
9dbbddb2ed10
updated for version 7.3.1001
Bram Moolenaar <bram@vim.org>
parents:
4503
diff
changeset
|
1071 if (*regparse == 'n') |
| 4444 | 1072 startc = reg_string ? NL : NFA_NEWL; |
| 1073 else | |
| 1074 if (*regparse == 'd' | |
| 1075 || *regparse == 'o' | |
| 1076 || *regparse == 'x' | |
| 1077 || *regparse == 'u' | |
| 1078 || *regparse == 'U' | |
| 1079 ) | |
| 1080 { | |
| 1081 /* TODO(RE) This needs more testing */ | |
| 1082 startc = coll_get_char(); | |
| 1083 got_coll_char = TRUE; | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1084 mb_ptr_back(old_regparse, regparse); |
| 4444 | 1085 } |
| 1086 else | |
| 1087 { | |
| 1088 /* \r,\t,\e,\b */ | |
| 1089 startc = backslash_trans(*regparse); | |
| 1090 } | |
| 1091 } | |
| 1092 | |
| 1093 /* Normal printable char */ | |
| 1094 if (startc == -1) | |
| 1095 #ifdef FEAT_MBYTE | |
| 1096 startc = (*mb_ptr2char)(regparse); | |
| 1097 #else | |
| 1098 startc = *regparse; | |
| 1099 #endif | |
| 1100 | |
| 1101 /* Previous char was '-', so this char is end of range. */ | |
| 1102 if (emit_range) | |
| 1103 { | |
| 1104 endc = startc; startc = oldstartc; | |
| 1105 if (startc > endc) | |
| 1106 EMSG_RET_FAIL(_(e_invrange)); | |
| 1107 #ifdef FEAT_MBYTE | |
| 1108 if (has_mbyte && ((*mb_char2len)(startc) > 1 | |
| 1109 || (*mb_char2len)(endc) > 1)) | |
| 1110 { | |
| 1111 if (endc > startc + 256) | |
| 1112 EMSG_RET_FAIL(_(e_invrange)); | |
| 1113 /* Emit the range. "startc" was already emitted, so | |
| 1114 * skip it. */ | |
| 1115 for (c = startc + 1; c <= endc; c++) | |
| 1116 { | |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1117 EMIT(c); |
| 4444 | 1118 TRY_NEG(); |
| 1119 EMIT_GLUE(); | |
| 1120 } | |
| 1121 emit_range = FALSE; | |
| 1122 } | |
| 1123 else | |
| 1124 #endif | |
| 1125 { | |
| 1126 #ifdef EBCDIC | |
| 1127 int alpha_only = FALSE; | |
| 1128 | |
| 1129 /* for alphabetical range skip the gaps | |
| 1130 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */ | |
| 1131 if (isalpha(startc) && isalpha(endc)) | |
| 1132 alpha_only = TRUE; | |
| 1133 #endif | |
| 1134 /* Emit the range. "startc" was already emitted, so | |
| 1135 * skip it. */ | |
| 1136 for (c = startc + 1; c <= endc; c++) | |
| 1137 #ifdef EBCDIC | |
| 1138 if (!alpha_only || isalpha(startc)) | |
| 1139 #endif | |
| 1140 { | |
| 1141 EMIT(c); | |
| 1142 TRY_NEG(); | |
| 1143 EMIT_GLUE(); | |
| 1144 } | |
| 1145 emit_range = FALSE; | |
| 1146 } | |
| 1147 } | |
| 1148 else | |
| 1149 { | |
| 1150 /* | |
| 1151 * This char (startc) is not part of a range. Just | |
| 1152 * emit it. | |
| 1153 * | |
| 1154 * Normally, simply emit startc. But if we get char | |
| 1155 * code=0 from a collating char, then replace it with | |
| 1156 * 0x0a. | |
| 1157 * | |
| 1158 * This is needed to completely mimic the behaviour of | |
| 1159 * the backtracking engine. | |
| 1160 */ | |
| 1161 if (got_coll_char == TRUE && startc == 0) | |
| 1162 EMIT(0x0a); | |
| 1163 else | |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1164 EMIT(startc); |
| 4444 | 1165 TRY_NEG(); |
| 1166 EMIT_GLUE(); | |
| 1167 } | |
| 1168 | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1169 mb_ptr_adv(regparse); |
| 4444 | 1170 } /* while (p < endp) */ |
| 1171 | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1172 mb_ptr_back(old_regparse, regparse); |
| 4444 | 1173 if (*regparse == '-') /* if last, '-' is just a char */ |
| 1174 { | |
| 1175 EMIT('-'); | |
| 1176 TRY_NEG(); | |
| 1177 EMIT_GLUE(); | |
| 1178 } | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1179 mb_ptr_adv(regparse); |
| 4444 | 1180 |
| 1181 if (extra == ADD_NL) /* \_[] also matches \n */ | |
| 1182 { | |
| 1183 EMIT(reg_string ? NL : NFA_NEWL); | |
| 1184 TRY_NEG(); | |
| 1185 EMIT_GLUE(); | |
| 1186 } | |
| 1187 | |
| 1188 /* skip the trailing ] */ | |
| 1189 regparse = endp; | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1190 mb_ptr_adv(regparse); |
| 4444 | 1191 if (negated == TRUE) |
| 1192 { | |
| 1193 /* Mark end of negated char range */ | |
| 1194 EMIT(NFA_END_NEG_RANGE); | |
| 1195 EMIT(NFA_CONCAT); | |
| 1196 } | |
| 1197 return OK; | |
|
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1198 } /* if exists closing ] */ |
|
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1199 |
|
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1200 if (reg_strict) |
| 4444 | 1201 { |
| 1202 syntax_error = TRUE; | |
| 1203 EMSG_RET_FAIL(_(e_missingbracket)); | |
| 1204 } | |
|
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1205 /* FALLTHROUGH */ |
|
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1206 |
| 4444 | 1207 default: |
| 1208 { | |
| 1209 #ifdef FEAT_MBYTE | |
| 1210 int plen; | |
| 1211 | |
| 1212 nfa_do_multibyte: | |
|
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1213 /* 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
|
1214 if (enc_utf8 && ((*mb_char2len)(c) |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1215 != (plen = (*mb_ptr2len)(old_regparse)) |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1216 || utf_iscomposing(c))) |
| 4444 | 1217 { |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
1218 int i = 0; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
1219 |
|
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
1220 /* 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
|
1221 * or more composing characters. |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1222 * 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
|
1223 * 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
|
1224 * NFA_END_COMPOSING is the ). Note that right now we are |
| 4444 | 1225 * building the postfix form, not the NFA itself; |
| 1226 * 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
|
1227 * 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
|
1228 for (;;) |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1229 { |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1230 EMIT(c); |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1231 if (i > 0) |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1232 EMIT(NFA_CONCAT); |
|
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1233 if ((i += utf_char2len(c)) >= plen) |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1234 break; |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1235 c = utf_ptr2char(old_regparse + i); |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1236 } |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1237 EMIT(NFA_COMPOSING); |
| 4444 | 1238 regparse = old_regparse + plen; |
| 1239 } | |
| 1240 else | |
| 1241 #endif | |
| 1242 { | |
| 1243 c = no_Magic(c); | |
| 1244 EMIT(c); | |
| 1245 } | |
| 1246 return OK; | |
| 1247 } | |
| 1248 } | |
| 1249 | |
| 1250 #undef TRY_NEG | |
| 1251 #undef EMIT_GLUE | |
| 1252 | |
| 1253 return OK; | |
| 1254 } | |
| 1255 | |
| 1256 /* | |
| 1257 * Parse something followed by possible [*+=]. | |
| 1258 * | |
| 1259 * A piece is an atom, possibly followed by a multi, an indication of how many | |
| 1260 * times the atom can be matched. Example: "a*" matches any sequence of "a" | |
| 1261 * characters: "", "a", "aa", etc. | |
| 1262 * | |
| 1263 * piece ::= atom | |
| 1264 * or atom multi | |
| 1265 */ | |
| 1266 static int | |
| 1267 nfa_regpiece() | |
| 1268 { | |
| 1269 int i; | |
| 1270 int op; | |
| 1271 int ret; | |
| 1272 long minval, maxval; | |
| 1273 int greedy = TRUE; /* Braces are prefixed with '-' ? */ | |
| 1274 char_u *old_regparse, *new_regparse; | |
| 1275 int c2; | |
| 1276 int *old_post_ptr, *my_post_start; | |
| 1277 int old_regnpar; | |
| 1278 int quest; | |
| 1279 | |
| 1280 /* Save the current position in the regexp, so that we can use it if | |
| 1281 * <atom>{m,n} is next. */ | |
| 1282 old_regparse = regparse; | |
| 1283 /* Save current number of open parenthesis, so we can use it if | |
| 1284 * <atom>{m,n} is next */ | |
| 1285 old_regnpar = regnpar; | |
| 1286 /* store current pos in the postfix form, for \{m,n} involving 0s */ | |
| 1287 my_post_start = post_ptr; | |
| 1288 | |
| 1289 ret = nfa_regatom(); | |
| 1290 if (ret == FAIL) | |
| 1291 return FAIL; /* cascaded error */ | |
| 1292 | |
| 1293 op = peekchr(); | |
| 1294 if (re_multi_type(op) == NOT_MULTI) | |
| 1295 return OK; | |
| 1296 | |
| 1297 skipchr(); | |
| 1298 switch (op) | |
| 1299 { | |
| 1300 case Magic('*'): | |
| 1301 EMIT(NFA_STAR); | |
| 1302 break; | |
| 1303 | |
| 1304 case Magic('+'): | |
| 1305 /* | |
| 1306 * Trick: Normally, (a*)\+ would match the whole input "aaa". The | |
| 1307 * first and only submatch would be "aaa". But the backtracking | |
| 1308 * engine interprets the plus as "try matching one more time", and | |
| 1309 * a* matches a second time at the end of the input, the empty | |
| 1310 * string. | |
| 1311 * The submatch will the empty string. | |
| 1312 * | |
| 1313 * In order to be consistent with the old engine, we disable | |
| 1314 * NFA_PLUS, and replace <atom>+ with <atom><atom>* | |
| 1315 */ | |
| 1316 /* EMIT(NFA_PLUS); */ | |
| 1317 regnpar = old_regnpar; | |
| 1318 regparse = old_regparse; | |
| 1319 curchr = -1; | |
| 1320 if (nfa_regatom() == FAIL) | |
| 1321 return FAIL; | |
| 1322 EMIT(NFA_STAR); | |
| 1323 EMIT(NFA_CONCAT); | |
| 1324 skipchr(); /* skip the \+ */ | |
| 1325 break; | |
| 1326 | |
| 1327 case Magic('@'): | |
| 1328 op = no_Magic(getchr()); | |
| 1329 switch(op) | |
| 1330 { | |
| 1331 case '=': | |
| 1332 EMIT(NFA_PREV_ATOM_NO_WIDTH); | |
| 1333 break; | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1334 case '0': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1335 case '1': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1336 case '2': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1337 case '3': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1338 case '4': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1339 case '5': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1340 case '6': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1341 case '7': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1342 case '8': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1343 case '9': |
| 4444 | 1344 case '!': |
| 1345 case '<': | |
| 1346 case '>': | |
| 1347 /* Not supported yet */ | |
| 1348 return FAIL; | |
| 1349 default: | |
| 1350 syntax_error = TRUE; | |
| 4450 | 1351 EMSGN(_("E869: (NFA) Unknown operator '\\@%c'"), op); |
| 4444 | 1352 return FAIL; |
| 1353 } | |
| 1354 break; | |
| 1355 | |
| 1356 case Magic('?'): | |
| 1357 case Magic('='): | |
| 1358 EMIT(NFA_QUEST); | |
| 1359 break; | |
| 1360 | |
| 1361 case Magic('{'): | |
| 1362 /* a{2,5} will expand to 'aaa?a?a?' | |
| 1363 * a{-1,3} will expand to 'aa??a??', where ?? is the nongreedy | |
| 1364 * version of '?' | |
| 1365 * \v(ab){2,3} will expand to '(ab)(ab)(ab)?', where all the | |
| 1366 * parenthesis have the same id | |
| 1367 */ | |
| 1368 | |
| 1369 greedy = TRUE; | |
| 1370 c2 = peekchr(); | |
| 1371 if (c2 == '-' || c2 == Magic('-')) | |
| 1372 { | |
| 1373 skipchr(); | |
| 1374 greedy = FALSE; | |
| 1375 } | |
| 1376 if (!read_limits(&minval, &maxval)) | |
| 1377 { | |
| 1378 syntax_error = TRUE; | |
| 1379 EMSG_RET_FAIL(_("E870: (NFA regexp) Error reading repetition limits")); | |
| 1380 } | |
| 1381 /* <atom>{0,inf}, <atom>{0,} and <atom>{} are equivalent to | |
| 1382 * <atom>* */ | |
| 1383 if (minval == 0 && maxval == MAX_LIMIT && greedy) | |
| 1384 { | |
| 1385 EMIT(NFA_STAR); | |
| 1386 break; | |
| 1387 } | |
| 1388 | |
| 1389 if (maxval > NFA_BRACES_MAXLIMIT) | |
| 1390 { | |
| 1391 /* This would yield a huge automaton and use too much memory. | |
| 1392 * Revert to old engine */ | |
| 1393 return FAIL; | |
| 1394 } | |
| 1395 | |
| 1396 /* Special case: x{0} or x{-0} */ | |
| 1397 if (maxval == 0) | |
| 1398 { | |
| 1399 /* Ignore result of previous call to nfa_regatom() */ | |
| 1400 post_ptr = my_post_start; | |
| 1401 /* NFA_SKIP_CHAR has 0-length and works everywhere */ | |
| 1402 EMIT(NFA_SKIP_CHAR); | |
| 1403 return OK; | |
| 1404 } | |
| 1405 | |
| 1406 /* Ignore previous call to nfa_regatom() */ | |
| 1407 post_ptr = my_post_start; | |
| 1408 /* Save pos after the repeated atom and the \{} */ | |
| 1409 new_regparse = regparse; | |
| 1410 | |
| 1411 quest = (greedy == TRUE? NFA_QUEST : NFA_QUEST_NONGREEDY); | |
| 1412 for (i = 0; i < maxval; i++) | |
| 1413 { | |
| 1414 /* Goto beginning of the repeated atom */ | |
| 1415 regparse = old_regparse; | |
| 1416 curchr = -1; | |
| 1417 /* Restore count of parenthesis */ | |
| 1418 regnpar = old_regnpar; | |
| 1419 old_post_ptr = post_ptr; | |
| 1420 if (nfa_regatom() == FAIL) | |
| 1421 return FAIL; | |
| 1422 /* after "minval" times, atoms are optional */ | |
| 1423 if (i + 1 > minval) | |
| 1424 EMIT(quest); | |
| 1425 if (old_post_ptr != my_post_start) | |
| 1426 EMIT(NFA_CONCAT); | |
| 1427 } | |
| 1428 | |
| 1429 /* Go to just after the repeated atom and the \{} */ | |
| 1430 regparse = new_regparse; | |
| 1431 curchr = -1; | |
| 1432 | |
| 1433 break; | |
| 1434 | |
| 1435 | |
| 1436 default: | |
| 1437 break; | |
| 1438 } /* end switch */ | |
| 1439 | |
| 1440 if (re_multi_type(peekchr()) != NOT_MULTI) | |
| 1441 { | |
| 1442 /* Can't have a multi follow a multi. */ | |
| 1443 syntax_error = TRUE; | |
| 1444 EMSG_RET_FAIL(_("E871: (NFA regexp) Can't have a multi follow a multi !")); | |
| 1445 } | |
| 1446 | |
| 1447 return OK; | |
| 1448 } | |
| 1449 | |
| 1450 /* | |
| 1451 * Parse one or more pieces, concatenated. It matches a match for the | |
| 1452 * first piece, followed by a match for the second piece, etc. Example: | |
| 1453 * "f[0-9]b", first matches "f", then a digit and then "b". | |
| 1454 * | |
| 1455 * concat ::= piece | |
| 1456 * or piece piece | |
| 1457 * or piece piece piece | |
| 1458 * etc. | |
| 1459 */ | |
| 1460 static int | |
| 1461 nfa_regconcat() | |
| 1462 { | |
| 1463 int cont = TRUE; | |
| 1464 int first = TRUE; | |
| 1465 | |
| 1466 while (cont) | |
| 1467 { | |
| 1468 switch (peekchr()) | |
| 1469 { | |
| 1470 case NUL: | |
| 1471 case Magic('|'): | |
| 1472 case Magic('&'): | |
| 1473 case Magic(')'): | |
| 1474 cont = FALSE; | |
| 1475 break; | |
| 1476 | |
| 1477 case Magic('Z'): | |
| 1478 #ifdef FEAT_MBYTE | |
| 1479 regflags |= RF_ICOMBINE; | |
| 1480 #endif | |
| 1481 skipchr_keepstart(); | |
| 1482 break; | |
| 1483 case Magic('c'): | |
| 1484 regflags |= RF_ICASE; | |
| 1485 skipchr_keepstart(); | |
| 1486 break; | |
| 1487 case Magic('C'): | |
| 1488 regflags |= RF_NOICASE; | |
| 1489 skipchr_keepstart(); | |
| 1490 break; | |
| 1491 case Magic('v'): | |
| 1492 reg_magic = MAGIC_ALL; | |
| 1493 skipchr_keepstart(); | |
| 1494 curchr = -1; | |
| 1495 break; | |
| 1496 case Magic('m'): | |
| 1497 reg_magic = MAGIC_ON; | |
| 1498 skipchr_keepstart(); | |
| 1499 curchr = -1; | |
| 1500 break; | |
| 1501 case Magic('M'): | |
| 1502 reg_magic = MAGIC_OFF; | |
| 1503 skipchr_keepstart(); | |
| 1504 curchr = -1; | |
| 1505 break; | |
| 1506 case Magic('V'): | |
| 1507 reg_magic = MAGIC_NONE; | |
| 1508 skipchr_keepstart(); | |
| 1509 curchr = -1; | |
| 1510 break; | |
| 1511 | |
| 1512 default: | |
| 1513 if (nfa_regpiece() == FAIL) | |
| 1514 return FAIL; | |
| 1515 if (first == FALSE) | |
| 1516 EMIT(NFA_CONCAT); | |
| 1517 else | |
| 1518 first = FALSE; | |
| 1519 break; | |
| 1520 } | |
| 1521 } | |
| 1522 | |
| 1523 return OK; | |
| 1524 } | |
| 1525 | |
| 1526 /* | |
| 1527 * Parse a branch, one or more concats, separated by "\&". It matches the | |
| 1528 * last concat, but only if all the preceding concats also match at the same | |
| 1529 * position. Examples: | |
| 1530 * "foobeep\&..." matches "foo" in "foobeep". | |
| 1531 * ".*Peter\&.*Bob" matches in a line containing both "Peter" and "Bob" | |
| 1532 * | |
| 1533 * branch ::= concat | |
| 1534 * or concat \& concat | |
| 1535 * or concat \& concat \& concat | |
| 1536 * etc. | |
| 1537 */ | |
| 1538 static int | |
| 1539 nfa_regbranch() | |
| 1540 { | |
| 1541 int ch; | |
| 1542 int *old_post_ptr; | |
| 1543 | |
| 1544 old_post_ptr = post_ptr; | |
| 1545 | |
| 1546 /* First branch, possibly the only one */ | |
| 1547 if (nfa_regconcat() == FAIL) | |
| 1548 return FAIL; | |
| 1549 | |
| 1550 ch = peekchr(); | |
| 1551 /* Try next concats */ | |
| 1552 while (ch == Magic('&')) | |
| 1553 { | |
| 1554 skipchr(); | |
| 1555 EMIT(NFA_NOPEN); | |
| 1556 EMIT(NFA_PREV_ATOM_NO_WIDTH); | |
| 1557 old_post_ptr = post_ptr; | |
| 1558 if (nfa_regconcat() == FAIL) | |
| 1559 return FAIL; | |
| 1560 /* if concat is empty, skip a input char. But do emit a node */ | |
| 1561 if (old_post_ptr == post_ptr) | |
| 1562 EMIT(NFA_SKIP_CHAR); | |
| 1563 EMIT(NFA_CONCAT); | |
| 1564 ch = peekchr(); | |
| 1565 } | |
| 1566 | |
| 1567 /* Even if a branch is empty, emit one node for it */ | |
| 1568 if (old_post_ptr == post_ptr) | |
| 1569 EMIT(NFA_SKIP_CHAR); | |
| 1570 | |
| 1571 return OK; | |
| 1572 } | |
| 1573 | |
| 1574 /* | |
| 1575 * Parse a pattern, one or more branches, separated by "\|". It matches | |
| 1576 * anything that matches one of the branches. Example: "foo\|beep" matches | |
| 1577 * "foo" and matches "beep". If more than one branch matches, the first one | |
| 1578 * is used. | |
| 1579 * | |
| 1580 * pattern ::= branch | |
| 1581 * or branch \| branch | |
| 1582 * or branch \| branch \| branch | |
| 1583 * etc. | |
| 1584 */ | |
| 1585 static int | |
| 1586 nfa_reg(paren) | |
| 1587 int paren; /* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */ | |
| 1588 { | |
| 1589 int parno = 0; | |
| 1590 | |
| 1591 #ifdef FEAT_SYN_HL | |
| 1592 #endif | |
| 1593 if (paren == REG_PAREN) | |
| 1594 { | |
| 1595 if (regnpar >= NSUBEXP) /* Too many `(' */ | |
| 1596 { | |
| 1597 syntax_error = TRUE; | |
| 1598 EMSG_RET_FAIL(_("E872: (NFA regexp) Too many '('")); | |
| 1599 } | |
| 1600 parno = regnpar++; | |
| 1601 } | |
| 1602 | |
| 1603 if (nfa_regbranch() == FAIL) | |
| 1604 return FAIL; /* cascaded error */ | |
| 1605 | |
| 1606 while (peekchr() == Magic('|')) | |
| 1607 { | |
| 1608 skipchr(); | |
| 1609 if (nfa_regbranch() == FAIL) | |
| 1610 return FAIL; /* cascaded error */ | |
| 1611 EMIT(NFA_OR); | |
| 1612 } | |
| 1613 | |
| 1614 /* Check for proper termination. */ | |
| 1615 if (paren != REG_NOPAREN && getchr() != Magic(')')) | |
| 1616 { | |
| 1617 syntax_error = TRUE; | |
| 1618 if (paren == REG_NPAREN) | |
| 1619 EMSG2_RET_FAIL(_(e_unmatchedpp), reg_magic == MAGIC_ALL); | |
| 1620 else | |
| 1621 EMSG2_RET_FAIL(_(e_unmatchedp), reg_magic == MAGIC_ALL); | |
| 1622 } | |
| 1623 else if (paren == REG_NOPAREN && peekchr() != NUL) | |
| 1624 { | |
| 1625 syntax_error = TRUE; | |
| 1626 if (peekchr() == Magic(')')) | |
| 1627 EMSG2_RET_FAIL(_(e_unmatchedpar), reg_magic == MAGIC_ALL); | |
| 1628 else | |
| 1629 EMSG_RET_FAIL(_("E873: (NFA regexp) proper termination error")); | |
| 1630 } | |
| 1631 /* | |
| 1632 * Here we set the flag allowing back references to this set of | |
| 1633 * parentheses. | |
| 1634 */ | |
| 1635 if (paren == REG_PAREN) | |
| 1636 { | |
| 1637 had_endbrace[parno] = TRUE; /* have seen the close paren */ | |
| 1638 EMIT(NFA_MOPEN + parno); | |
| 1639 } | |
| 1640 | |
| 1641 return OK; | |
| 1642 } | |
| 1643 | |
| 1644 #ifdef DEBUG | |
| 1645 static char_u code[50]; | |
| 1646 | |
| 1647 static void | |
| 1648 nfa_set_code(c) | |
| 1649 int c; | |
| 1650 { | |
| 1651 int addnl = FALSE; | |
| 1652 | |
| 1653 if (c >= NFA_FIRST_NL && c <= NFA_LAST_NL) | |
| 1654 { | |
| 1655 addnl = TRUE; | |
| 1656 c -= ADD_NL; | |
| 1657 } | |
| 1658 | |
| 1659 STRCPY(code, ""); | |
| 1660 switch (c) | |
| 1661 { | |
| 1662 case NFA_MATCH: STRCPY(code, "NFA_MATCH "); break; | |
| 1663 case NFA_SPLIT: STRCPY(code, "NFA_SPLIT "); break; | |
| 1664 case NFA_CONCAT: STRCPY(code, "NFA_CONCAT "); break; | |
| 1665 case NFA_NEWL: STRCPY(code, "NFA_NEWL "); break; | |
| 1666 case NFA_ZSTART: STRCPY(code, "NFA_ZSTART"); break; | |
| 1667 case NFA_ZEND: STRCPY(code, "NFA_ZEND"); break; | |
| 1668 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1669 case NFA_BACKREF1: STRCPY(code, "NFA_BACKREF1"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1670 case NFA_BACKREF2: STRCPY(code, "NFA_BACKREF2"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1671 case NFA_BACKREF3: STRCPY(code, "NFA_BACKREF3"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1672 case NFA_BACKREF4: STRCPY(code, "NFA_BACKREF4"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1673 case NFA_BACKREF5: STRCPY(code, "NFA_BACKREF5"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1674 case NFA_BACKREF6: STRCPY(code, "NFA_BACKREF6"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1675 case NFA_BACKREF7: STRCPY(code, "NFA_BACKREF7"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1676 case NFA_BACKREF8: STRCPY(code, "NFA_BACKREF8"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1677 case NFA_BACKREF9: STRCPY(code, "NFA_BACKREF9"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1678 case NFA_SKIP: STRCPY(code, "NFA_SKIP"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1679 |
| 4444 | 1680 case NFA_PREV_ATOM_NO_WIDTH: |
| 1681 STRCPY(code, "NFA_PREV_ATOM_NO_WIDTH"); break; | |
| 1682 case NFA_NOPEN: STRCPY(code, "NFA_MOPEN_INVISIBLE"); break; | |
| 1683 case NFA_NCLOSE: STRCPY(code, "NFA_MCLOSE_INVISIBLE"); break; | |
| 1684 case NFA_START_INVISIBLE: STRCPY(code, "NFA_START_INVISIBLE"); break; | |
| 1685 case NFA_END_INVISIBLE: STRCPY(code, "NFA_END_INVISIBLE"); break; | |
| 1686 | |
| 1687 case NFA_COMPOSING: STRCPY(code, "NFA_COMPOSING"); break; | |
| 1688 case NFA_END_COMPOSING: STRCPY(code, "NFA_END_COMPOSING"); break; | |
| 1689 | |
| 1690 case NFA_MOPEN + 0: | |
| 1691 case NFA_MOPEN + 1: | |
| 1692 case NFA_MOPEN + 2: | |
| 1693 case NFA_MOPEN + 3: | |
| 1694 case NFA_MOPEN + 4: | |
| 1695 case NFA_MOPEN + 5: | |
| 1696 case NFA_MOPEN + 6: | |
| 1697 case NFA_MOPEN + 7: | |
| 1698 case NFA_MOPEN + 8: | |
| 1699 case NFA_MOPEN + 9: | |
| 1700 STRCPY(code, "NFA_MOPEN(x)"); | |
| 1701 code[10] = c - NFA_MOPEN + '0'; | |
| 1702 break; | |
| 1703 case NFA_MCLOSE + 0: | |
| 1704 case NFA_MCLOSE + 1: | |
| 1705 case NFA_MCLOSE + 2: | |
| 1706 case NFA_MCLOSE + 3: | |
| 1707 case NFA_MCLOSE + 4: | |
| 1708 case NFA_MCLOSE + 5: | |
| 1709 case NFA_MCLOSE + 6: | |
| 1710 case NFA_MCLOSE + 7: | |
| 1711 case NFA_MCLOSE + 8: | |
| 1712 case NFA_MCLOSE + 9: | |
| 1713 STRCPY(code, "NFA_MCLOSE(x)"); | |
| 1714 code[11] = c - NFA_MCLOSE + '0'; | |
| 1715 break; | |
| 1716 case NFA_EOL: STRCPY(code, "NFA_EOL "); break; | |
| 1717 case NFA_BOL: STRCPY(code, "NFA_BOL "); break; | |
| 1718 case NFA_EOW: STRCPY(code, "NFA_EOW "); break; | |
| 1719 case NFA_BOW: STRCPY(code, "NFA_BOW "); break; | |
| 1720 case NFA_STAR: STRCPY(code, "NFA_STAR "); break; | |
| 1721 case NFA_PLUS: STRCPY(code, "NFA_PLUS "); break; | |
| 1722 case NFA_NOT: STRCPY(code, "NFA_NOT "); break; | |
| 1723 case NFA_SKIP_CHAR: STRCPY(code, "NFA_SKIP_CHAR"); break; | |
| 1724 case NFA_OR: STRCPY(code, "NFA_OR"); break; | |
| 1725 case NFA_QUEST: STRCPY(code, "NFA_QUEST"); break; | |
| 1726 case NFA_QUEST_NONGREEDY: STRCPY(code, "NFA_QUEST_NON_GREEDY"); break; | |
| 1727 case NFA_END_NEG_RANGE: STRCPY(code, "NFA_END_NEG_RANGE"); break; | |
| 1728 case NFA_CLASS_ALNUM: STRCPY(code, "NFA_CLASS_ALNUM"); break; | |
| 1729 case NFA_CLASS_ALPHA: STRCPY(code, "NFA_CLASS_ALPHA"); break; | |
| 1730 case NFA_CLASS_BLANK: STRCPY(code, "NFA_CLASS_BLANK"); break; | |
| 1731 case NFA_CLASS_CNTRL: STRCPY(code, "NFA_CLASS_CNTRL"); break; | |
| 1732 case NFA_CLASS_DIGIT: STRCPY(code, "NFA_CLASS_DIGIT"); break; | |
| 1733 case NFA_CLASS_GRAPH: STRCPY(code, "NFA_CLASS_GRAPH"); break; | |
| 1734 case NFA_CLASS_LOWER: STRCPY(code, "NFA_CLASS_LOWER"); break; | |
| 1735 case NFA_CLASS_PRINT: STRCPY(code, "NFA_CLASS_PRINT"); break; | |
| 1736 case NFA_CLASS_PUNCT: STRCPY(code, "NFA_CLASS_PUNCT"); break; | |
| 1737 case NFA_CLASS_SPACE: STRCPY(code, "NFA_CLASS_SPACE"); break; | |
| 1738 case NFA_CLASS_UPPER: STRCPY(code, "NFA_CLASS_UPPER"); break; | |
| 1739 case NFA_CLASS_XDIGIT: STRCPY(code, "NFA_CLASS_XDIGIT"); break; | |
| 1740 case NFA_CLASS_TAB: STRCPY(code, "NFA_CLASS_TAB"); break; | |
| 1741 case NFA_CLASS_RETURN: STRCPY(code, "NFA_CLASS_RETURN"); break; | |
| 1742 case NFA_CLASS_BACKSPACE: STRCPY(code, "NFA_CLASS_BACKSPACE"); break; | |
| 1743 case NFA_CLASS_ESCAPE: STRCPY(code, "NFA_CLASS_ESCAPE"); break; | |
| 1744 | |
| 1745 case NFA_ANY: STRCPY(code, "NFA_ANY"); break; | |
| 1746 case NFA_IDENT: STRCPY(code, "NFA_IDENT"); break; | |
| 1747 case NFA_SIDENT:STRCPY(code, "NFA_SIDENT"); break; | |
| 1748 case NFA_KWORD: STRCPY(code, "NFA_KWORD"); break; | |
| 1749 case NFA_SKWORD:STRCPY(code, "NFA_SKWORD"); break; | |
| 1750 case NFA_FNAME: STRCPY(code, "NFA_FNAME"); break; | |
| 1751 case NFA_SFNAME:STRCPY(code, "NFA_SFNAME"); break; | |
| 1752 case NFA_PRINT: STRCPY(code, "NFA_PRINT"); break; | |
| 1753 case NFA_SPRINT:STRCPY(code, "NFA_SPRINT"); break; | |
| 1754 case NFA_WHITE: STRCPY(code, "NFA_WHITE"); break; | |
| 1755 case NFA_NWHITE:STRCPY(code, "NFA_NWHITE"); break; | |
| 1756 case NFA_DIGIT: STRCPY(code, "NFA_DIGIT"); break; | |
| 1757 case NFA_NDIGIT:STRCPY(code, "NFA_NDIGIT"); break; | |
| 1758 case NFA_HEX: STRCPY(code, "NFA_HEX"); break; | |
| 1759 case NFA_NHEX: STRCPY(code, "NFA_NHEX"); break; | |
| 1760 case NFA_OCTAL: STRCPY(code, "NFA_OCTAL"); break; | |
| 1761 case NFA_NOCTAL:STRCPY(code, "NFA_NOCTAL"); break; | |
| 1762 case NFA_WORD: STRCPY(code, "NFA_WORD"); break; | |
| 1763 case NFA_NWORD: STRCPY(code, "NFA_NWORD"); break; | |
| 1764 case NFA_HEAD: STRCPY(code, "NFA_HEAD"); break; | |
| 1765 case NFA_NHEAD: STRCPY(code, "NFA_NHEAD"); break; | |
| 1766 case NFA_ALPHA: STRCPY(code, "NFA_ALPHA"); break; | |
| 1767 case NFA_NALPHA:STRCPY(code, "NFA_NALPHA"); break; | |
| 1768 case NFA_LOWER: STRCPY(code, "NFA_LOWER"); break; | |
| 1769 case NFA_NLOWER:STRCPY(code, "NFA_NLOWER"); break; | |
| 1770 case NFA_UPPER: STRCPY(code, "NFA_UPPER"); break; | |
| 1771 case NFA_NUPPER:STRCPY(code, "NFA_NUPPER"); break; | |
| 1772 | |
| 1773 default: | |
| 1774 STRCPY(code, "CHAR(x)"); | |
| 1775 code[5] = c; | |
| 1776 } | |
| 1777 | |
| 1778 if (addnl == TRUE) | |
| 1779 STRCAT(code, " + NEWLINE "); | |
| 1780 | |
| 1781 } | |
| 1782 | |
| 1783 #ifdef ENABLE_LOG | |
| 1784 static FILE *log_fd; | |
| 1785 | |
| 1786 /* | |
| 1787 * Print the postfix notation of the current regexp. | |
| 1788 */ | |
| 1789 static void | |
| 1790 nfa_postfix_dump(expr, retval) | |
| 1791 char_u *expr; | |
| 1792 int retval; | |
| 1793 { | |
| 1794 int *p; | |
| 1795 FILE *f; | |
| 1796 | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
1797 f = fopen(NFA_REGEXP_DUMP_LOG, "a"); |
| 4444 | 1798 if (f != NULL) |
| 1799 { | |
| 1800 fprintf(f, "\n-------------------------\n"); | |
| 1801 if (retval == FAIL) | |
| 1802 fprintf(f, ">>> NFA engine failed ... \n"); | |
| 1803 else if (retval == OK) | |
| 1804 fprintf(f, ">>> NFA engine succeeded !\n"); | |
| 1805 fprintf(f, "Regexp: \"%s\"\nPostfix notation (char): \"", expr); | |
| 4464 | 1806 for (p = post_start; *p && p < post_end; p++) |
| 4444 | 1807 { |
| 1808 nfa_set_code(*p); | |
| 1809 fprintf(f, "%s, ", code); | |
| 1810 } | |
| 1811 fprintf(f, "\"\nPostfix notation (int): "); | |
| 4464 | 1812 for (p = post_start; *p && p < post_end; p++) |
| 4444 | 1813 fprintf(f, "%d ", *p); |
| 1814 fprintf(f, "\n\n"); | |
| 1815 fclose(f); | |
| 1816 } | |
| 1817 } | |
| 1818 | |
| 1819 /* | |
| 1820 * Print the NFA starting with a root node "state". | |
| 1821 */ | |
| 1822 static void | |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1823 nfa_print_state(debugf, state) |
| 4444 | 1824 FILE *debugf; |
| 1825 nfa_state_T *state; | |
| 1826 { | |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1827 garray_T indent; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1828 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1829 ga_init2(&indent, 1, 64); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1830 ga_append(&indent, '\0'); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1831 nfa_print_state2(debugf, state, &indent); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1832 ga_clear(&indent); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1833 } |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1834 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1835 static void |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1836 nfa_print_state2(debugf, state, indent) |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1837 FILE *debugf; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1838 nfa_state_T *state; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1839 garray_T *indent; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1840 { |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1841 char_u *p; |
| 4444 | 1842 |
| 1843 if (state == NULL) | |
| 1844 return; | |
| 1845 | |
| 1846 fprintf(debugf, "(%2d)", abs(state->id)); | |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1847 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1848 /* Output indent */ |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1849 p = (char_u *)indent->ga_data; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1850 if (indent->ga_len >= 3) |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1851 { |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1852 int last = indent->ga_len - 3; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1853 char_u save[2]; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1854 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1855 STRNCPY(save, &p[last], 2); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1856 STRNCPY(&p[last], "+-", 2); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1857 fprintf(debugf, " %s", p); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1858 STRNCPY(&p[last], save, 2); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1859 } |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1860 else |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1861 fprintf(debugf, " %s", p); |
| 4444 | 1862 |
| 1863 nfa_set_code(state->c); | |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1864 fprintf(debugf, "%s%s (%d) (id=%d)\n", |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1865 state->negated ? "NOT " : "", code, state->c, abs(state->id)); |
| 4444 | 1866 if (state->id < 0) |
| 1867 return; | |
| 1868 | |
| 1869 state->id = abs(state->id) * -1; | |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1870 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1871 /* grow indent for state->out */ |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1872 indent->ga_len -= 1; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1873 if (state->out1) |
|
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1874 ga_concat(indent, (char_u *)"| "); |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1875 else |
|
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1876 ga_concat(indent, (char_u *)" "); |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1877 ga_append(indent, '\0'); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1878 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1879 nfa_print_state2(debugf, state->out, indent); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1880 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1881 /* replace last part of indent for state->out1 */ |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1882 indent->ga_len -= 3; |
|
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1883 ga_concat(indent, (char_u *)" "); |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1884 ga_append(indent, '\0'); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1885 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1886 nfa_print_state2(debugf, state->out1, indent); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1887 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1888 /* shrink indent */ |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1889 indent->ga_len -= 3; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1890 ga_append(indent, '\0'); |
| 4444 | 1891 } |
| 1892 | |
| 1893 /* | |
| 1894 * Print the NFA state machine. | |
| 1895 */ | |
| 1896 static void | |
| 1897 nfa_dump(prog) | |
| 1898 nfa_regprog_T *prog; | |
| 1899 { | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
1900 FILE *debugf = fopen(NFA_REGEXP_DUMP_LOG, "a"); |
| 4444 | 1901 |
| 1902 if (debugf != NULL) | |
| 1903 { | |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1904 nfa_print_state(debugf, prog->start); |
| 4444 | 1905 fclose(debugf); |
| 1906 } | |
| 1907 } | |
| 1908 #endif /* ENABLE_LOG */ | |
| 1909 #endif /* DEBUG */ | |
| 1910 | |
| 1911 /* | |
| 1912 * Parse r.e. @expr and convert it into postfix form. | |
| 1913 * Return the postfix string on success, NULL otherwise. | |
| 1914 */ | |
| 1915 static int * | |
| 1916 re2post() | |
| 1917 { | |
| 1918 if (nfa_reg(REG_NOPAREN) == FAIL) | |
| 1919 return NULL; | |
| 1920 EMIT(NFA_MOPEN); | |
| 1921 return post_start; | |
| 1922 } | |
| 1923 | |
| 1924 /* NB. Some of the code below is inspired by Russ's. */ | |
| 1925 | |
| 1926 /* | |
| 1927 * Represents an NFA state plus zero or one or two arrows exiting. | |
| 1928 * if c == MATCH, no arrows out; matching state. | |
| 1929 * If c == SPLIT, unlabeled arrows to out and out1 (if != NULL). | |
| 1930 * If c < 256, labeled arrow with character c to out. | |
| 1931 */ | |
| 1932 | |
| 1933 static nfa_state_T *state_ptr; /* points to nfa_prog->state */ | |
| 1934 | |
| 1935 /* | |
| 1936 * Allocate and initialize nfa_state_T. | |
| 1937 */ | |
| 1938 static nfa_state_T * | |
| 1939 new_state(c, out, out1) | |
| 1940 int c; | |
| 1941 nfa_state_T *out; | |
| 1942 nfa_state_T *out1; | |
| 1943 { | |
| 1944 nfa_state_T *s; | |
| 1945 | |
| 1946 if (istate >= nstate) | |
| 1947 return NULL; | |
| 1948 | |
| 1949 s = &state_ptr[istate++]; | |
| 1950 | |
| 1951 s->c = c; | |
| 1952 s->out = out; | |
| 1953 s->out1 = out1; | |
| 1954 | |
| 1955 s->id = istate; | |
| 1956 s->lastlist = 0; | |
| 1957 s->negated = FALSE; | |
| 1958 | |
| 1959 return s; | |
| 1960 } | |
| 1961 | |
| 1962 /* | |
| 1963 * A partially built NFA without the matching state filled in. | |
| 1964 * Frag_T.start points at the start state. | |
| 1965 * Frag_T.out is a list of places that need to be set to the | |
| 1966 * next state for this fragment. | |
| 1967 */ | |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1968 |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1969 /* 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
|
1970 * uninitialized, we use the pointers themselves |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1971 * as storage for the Ptrlists. */ |
| 4444 | 1972 typedef union Ptrlist Ptrlist; |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1973 union Ptrlist |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1974 { |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1975 Ptrlist *next; |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1976 nfa_state_T *s; |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1977 }; |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1978 |
| 4444 | 1979 struct Frag |
| 1980 { | |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1981 nfa_state_T *start; |
| 4444 | 1982 Ptrlist *out; |
| 1983 }; | |
| 1984 typedef struct Frag Frag_T; | |
| 1985 | |
| 1986 static Frag_T frag __ARGS((nfa_state_T *start, Ptrlist *out)); | |
| 1987 static Ptrlist *list1 __ARGS((nfa_state_T **outp)); | |
| 1988 static void patch __ARGS((Ptrlist *l, nfa_state_T *s)); | |
| 1989 static Ptrlist *append __ARGS((Ptrlist *l1, Ptrlist *l2)); | |
| 1990 static void st_push __ARGS((Frag_T s, Frag_T **p, Frag_T *stack_end)); | |
| 1991 static Frag_T st_pop __ARGS((Frag_T **p, Frag_T *stack)); | |
| 1992 | |
| 1993 /* | |
| 4456 | 1994 * Initialize a Frag_T struct and return it. |
| 4444 | 1995 */ |
| 1996 static Frag_T | |
| 1997 frag(start, out) | |
| 1998 nfa_state_T *start; | |
| 1999 Ptrlist *out; | |
| 2000 { | |
| 4456 | 2001 Frag_T n; |
| 2002 | |
| 2003 n.start = start; | |
| 2004 n.out = out; | |
| 4444 | 2005 return n; |
| 2006 } | |
| 2007 | |
| 2008 /* | |
| 2009 * Create singleton list containing just outp. | |
| 2010 */ | |
| 2011 static Ptrlist * | |
| 2012 list1(outp) | |
| 2013 nfa_state_T **outp; | |
| 2014 { | |
| 2015 Ptrlist *l; | |
| 2016 | |
| 2017 l = (Ptrlist *)outp; | |
| 2018 l->next = NULL; | |
| 2019 return l; | |
| 2020 } | |
| 2021 | |
| 2022 /* | |
| 2023 * Patch the list of states at out to point to start. | |
| 2024 */ | |
| 2025 static void | |
| 2026 patch(l, s) | |
| 2027 Ptrlist *l; | |
| 2028 nfa_state_T *s; | |
| 2029 { | |
| 2030 Ptrlist *next; | |
| 2031 | |
| 2032 for (; l; l = next) | |
| 2033 { | |
| 2034 next = l->next; | |
| 2035 l->s = s; | |
| 2036 } | |
| 2037 } | |
| 2038 | |
| 2039 | |
| 2040 /* | |
| 2041 * Join the two lists l1 and l2, returning the combination. | |
| 2042 */ | |
| 2043 static Ptrlist * | |
| 2044 append(l1, l2) | |
| 2045 Ptrlist *l1; | |
| 2046 Ptrlist *l2; | |
| 2047 { | |
| 2048 Ptrlist *oldl1; | |
| 2049 | |
| 2050 oldl1 = l1; | |
| 2051 while (l1->next) | |
| 2052 l1 = l1->next; | |
| 2053 l1->next = l2; | |
| 2054 return oldl1; | |
| 2055 } | |
| 2056 | |
| 2057 /* | |
| 2058 * Stack used for transforming postfix form into NFA. | |
| 2059 */ | |
| 2060 static Frag_T empty; | |
| 2061 | |
| 2062 static void | |
| 2063 st_error(postfix, end, p) | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2064 int *postfix UNUSED; |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2065 int *end UNUSED; |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2066 int *p UNUSED; |
| 4444 | 2067 { |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2068 #ifdef NFA_REGEXP_ERROR_LOG |
| 4444 | 2069 FILE *df; |
| 2070 int *p2; | |
| 2071 | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2072 df = fopen(NFA_REGEXP_ERROR_LOG, "a"); |
| 4444 | 2073 if (df) |
| 2074 { | |
| 2075 fprintf(df, "Error popping the stack!\n"); | |
| 2076 #ifdef DEBUG | |
| 2077 fprintf(df, "Current regexp is \"%s\"\n", nfa_regengine.expr); | |
| 2078 #endif | |
| 2079 fprintf(df, "Postfix form is: "); | |
| 2080 #ifdef DEBUG | |
| 2081 for (p2 = postfix; p2 < end; p2++) | |
| 2082 { | |
| 2083 nfa_set_code(*p2); | |
| 2084 fprintf(df, "%s, ", code); | |
| 2085 } | |
| 2086 nfa_set_code(*p); | |
| 2087 fprintf(df, "\nCurrent position is: "); | |
| 2088 for (p2 = postfix; p2 <= p; p2 ++) | |
| 2089 { | |
| 2090 nfa_set_code(*p2); | |
| 2091 fprintf(df, "%s, ", code); | |
| 2092 } | |
| 2093 #else | |
| 2094 for (p2 = postfix; p2 < end; p2++) | |
| 2095 { | |
| 2096 fprintf(df, "%d, ", *p2); | |
| 2097 } | |
| 2098 fprintf(df, "\nCurrent position is: "); | |
| 2099 for (p2 = postfix; p2 <= p; p2 ++) | |
| 2100 { | |
| 2101 fprintf(df, "%d, ", *p2); | |
| 2102 } | |
| 2103 #endif | |
| 2104 fprintf(df, "\n--------------------------\n"); | |
| 2105 fclose(df); | |
| 2106 } | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2107 #endif |
| 4444 | 2108 EMSG(_("E874: (NFA) Could not pop the stack !")); |
| 2109 } | |
| 2110 | |
| 2111 /* | |
| 2112 * Push an item onto the stack. | |
| 2113 */ | |
| 2114 static void | |
| 2115 st_push(s, p, stack_end) | |
| 2116 Frag_T s; | |
| 2117 Frag_T **p; | |
| 2118 Frag_T *stack_end; | |
| 2119 { | |
| 2120 Frag_T *stackp = *p; | |
| 2121 | |
| 2122 if (stackp >= stack_end) | |
| 2123 return; | |
| 2124 *stackp = s; | |
| 2125 *p = *p + 1; | |
| 2126 } | |
| 2127 | |
| 2128 /* | |
| 2129 * Pop an item from the stack. | |
| 2130 */ | |
| 2131 static Frag_T | |
| 2132 st_pop(p, stack) | |
| 2133 Frag_T **p; | |
| 2134 Frag_T *stack; | |
| 2135 { | |
| 2136 Frag_T *stackp; | |
| 2137 | |
| 2138 *p = *p - 1; | |
| 2139 stackp = *p; | |
| 2140 if (stackp < stack) | |
| 2141 return empty; | |
| 2142 return **p; | |
| 2143 } | |
| 2144 | |
| 2145 /* | |
| 2146 * Convert a postfix form into its equivalent NFA. | |
| 2147 * Return the NFA start state on success, NULL otherwise. | |
| 2148 */ | |
| 2149 static nfa_state_T * | |
| 2150 post2nfa(postfix, end, nfa_calc_size) | |
| 2151 int *postfix; | |
| 2152 int *end; | |
| 2153 int nfa_calc_size; | |
| 2154 { | |
| 2155 int *p; | |
| 2156 int mopen; | |
| 2157 int mclose; | |
| 2158 Frag_T *stack = NULL; | |
| 2159 Frag_T *stackp = NULL; | |
| 2160 Frag_T *stack_end = NULL; | |
| 2161 Frag_T e1; | |
| 2162 Frag_T e2; | |
| 2163 Frag_T e; | |
| 2164 nfa_state_T *s; | |
| 2165 nfa_state_T *s1; | |
| 2166 nfa_state_T *matchstate; | |
| 4484 | 2167 nfa_state_T *ret = NULL; |
| 4444 | 2168 |
| 2169 if (postfix == NULL) | |
| 2170 return NULL; | |
| 2171 | |
| 4456 | 2172 #define PUSH(s) st_push((s), &stackp, stack_end) |
| 4444 | 2173 #define POP() st_pop(&stackp, stack); \ |
| 2174 if (stackp < stack) \ | |
| 2175 { \ | |
| 2176 st_error(postfix, end, p); \ | |
| 2177 return NULL; \ | |
| 2178 } | |
| 2179 | |
| 2180 if (nfa_calc_size == FALSE) | |
| 2181 { | |
| 2182 /* Allocate space for the stack. Max states on the stack : nstate */ | |
| 4462 | 2183 stack = (Frag_T *) lalloc((nstate + 1) * sizeof(Frag_T), TRUE); |
| 4444 | 2184 stackp = stack; |
| 4462 | 2185 stack_end = stack + (nstate + 1); |
| 4444 | 2186 } |
| 2187 | |
| 2188 for (p = postfix; p < end; ++p) | |
| 2189 { | |
| 2190 switch (*p) | |
| 2191 { | |
| 2192 case NFA_CONCAT: | |
| 2193 /* Catenation. | |
| 2194 * Pay attention: this operator does not exist | |
| 2195 * in the r.e. itself (it is implicit, really). | |
| 2196 * It is added when r.e. is translated to postfix | |
| 2197 * form in re2post(). | |
| 2198 * | |
| 2199 * No new state added here. */ | |
| 2200 if (nfa_calc_size == TRUE) | |
| 2201 { | |
| 4458 | 2202 /* nstate += 0; */ |
| 4444 | 2203 break; |
| 2204 } | |
| 2205 e2 = POP(); | |
| 2206 e1 = POP(); | |
| 2207 patch(e1.out, e2.start); | |
| 2208 PUSH(frag(e1.start, e2.out)); | |
| 2209 break; | |
| 2210 | |
| 2211 case NFA_NOT: | |
| 2212 /* Negation of a character */ | |
| 2213 if (nfa_calc_size == TRUE) | |
| 2214 { | |
| 4458 | 2215 /* nstate += 0; */ |
| 4444 | 2216 break; |
| 2217 } | |
| 2218 e1 = POP(); | |
| 2219 e1.start->negated = TRUE; | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2220 #ifdef FEAT_MBYTE |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2221 if (e1.start->c == NFA_COMPOSING) |
| 4444 | 2222 e1.start->out1->negated = TRUE; |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2223 #endif |
| 4444 | 2224 PUSH(e1); |
| 2225 break; | |
| 2226 | |
| 2227 case NFA_OR: | |
| 2228 /* Alternation */ | |
| 2229 if (nfa_calc_size == TRUE) | |
| 2230 { | |
| 4458 | 2231 nstate++; |
| 4444 | 2232 break; |
| 2233 } | |
| 2234 e2 = POP(); | |
| 2235 e1 = POP(); | |
| 2236 s = new_state(NFA_SPLIT, e1.start, e2.start); | |
| 2237 if (s == NULL) | |
| 4484 | 2238 goto theend; |
| 4444 | 2239 PUSH(frag(s, append(e1.out, e2.out))); |
| 2240 break; | |
| 2241 | |
| 2242 case NFA_STAR: | |
| 2243 /* Zero or more */ | |
| 2244 if (nfa_calc_size == TRUE) | |
| 2245 { | |
| 4458 | 2246 nstate++; |
| 4444 | 2247 break; |
| 2248 } | |
| 2249 e = POP(); | |
| 2250 s = new_state(NFA_SPLIT, e.start, NULL); | |
| 2251 if (s == NULL) | |
| 4484 | 2252 goto theend; |
| 4444 | 2253 patch(e.out, s); |
| 2254 PUSH(frag(s, list1(&s->out1))); | |
| 2255 break; | |
| 2256 | |
| 2257 case NFA_QUEST: | |
| 2258 /* one or zero atoms=> greedy match */ | |
| 2259 if (nfa_calc_size == TRUE) | |
| 2260 { | |
| 4458 | 2261 nstate++; |
| 4444 | 2262 break; |
| 2263 } | |
| 2264 e = POP(); | |
| 2265 s = new_state(NFA_SPLIT, e.start, NULL); | |
| 2266 if (s == NULL) | |
| 4484 | 2267 goto theend; |
| 4444 | 2268 PUSH(frag(s, append(e.out, list1(&s->out1)))); |
| 2269 break; | |
| 2270 | |
| 2271 case NFA_QUEST_NONGREEDY: | |
| 2272 /* zero or one atoms => non-greedy match */ | |
| 2273 if (nfa_calc_size == TRUE) | |
| 2274 { | |
| 4458 | 2275 nstate++; |
| 4444 | 2276 break; |
| 2277 } | |
| 2278 e = POP(); | |
| 2279 s = new_state(NFA_SPLIT, NULL, e.start); | |
| 2280 if (s == NULL) | |
| 4484 | 2281 goto theend; |
| 4444 | 2282 PUSH(frag(s, append(e.out, list1(&s->out)))); |
| 2283 break; | |
| 2284 | |
| 2285 case NFA_PLUS: | |
| 2286 /* One or more */ | |
| 2287 if (nfa_calc_size == TRUE) | |
| 2288 { | |
| 4458 | 2289 nstate++; |
| 4444 | 2290 break; |
| 2291 } | |
| 2292 e = POP(); | |
| 2293 s = new_state(NFA_SPLIT, e.start, NULL); | |
| 2294 if (s == NULL) | |
| 4484 | 2295 goto theend; |
| 4444 | 2296 patch(e.out, s); |
| 2297 PUSH(frag(e.start, list1(&s->out1))); | |
| 2298 break; | |
| 2299 | |
| 2300 case NFA_SKIP_CHAR: | |
| 2301 /* Symbol of 0-length, Used in a repetition | |
| 2302 * with max/min count of 0 */ | |
| 2303 if (nfa_calc_size == TRUE) | |
| 2304 { | |
| 4458 | 2305 nstate++; |
| 4444 | 2306 break; |
| 2307 } | |
| 2308 s = new_state(NFA_SKIP_CHAR, NULL, NULL); | |
| 2309 if (s == NULL) | |
| 4484 | 2310 goto theend; |
| 4444 | 2311 PUSH(frag(s, list1(&s->out))); |
| 2312 break; | |
| 2313 | |
| 2314 case NFA_PREV_ATOM_NO_WIDTH: | |
| 2315 /* The \@= operator: match the preceding atom with 0 width. | |
| 2316 * Surrounds the preceding atom with START_INVISIBLE and | |
| 2317 * END_INVISIBLE, similarly to MOPEN. | |
| 2318 */ | |
| 2319 /* TODO: Maybe this drops the speed? */ | |
| 4484 | 2320 goto theend; |
| 4444 | 2321 |
| 2322 if (nfa_calc_size == TRUE) | |
| 2323 { | |
| 2324 nstate += 2; | |
| 2325 break; | |
| 2326 } | |
| 2327 e = POP(); | |
| 2328 s1 = new_state(NFA_END_INVISIBLE, NULL, NULL); | |
| 2329 if (s1 == NULL) | |
| 4484 | 2330 goto theend; |
| 4444 | 2331 patch(e.out, s1); |
| 2332 | |
| 2333 s = new_state(NFA_START_INVISIBLE, e.start, s1); | |
| 2334 if (s == NULL) | |
| 4484 | 2335 goto theend; |
| 4444 | 2336 PUSH(frag(s, list1(&s1->out))); |
| 2337 break; | |
| 2338 | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2339 #ifdef FEAT_MBYTE |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2340 case NFA_COMPOSING: /* char with composing char */ |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2341 #if 0 |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2342 /* TODO */ |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2343 if (regflags & RF_ICOMBINE) |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2344 { |
|
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
2345 /* use the base character only */ |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2346 } |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2347 #endif |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2348 /* FALLTHROUGH */ |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2349 #endif |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2350 |
| 4444 | 2351 case NFA_MOPEN + 0: /* Submatch */ |
| 2352 case NFA_MOPEN + 1: | |
| 2353 case NFA_MOPEN + 2: | |
| 2354 case NFA_MOPEN + 3: | |
| 2355 case NFA_MOPEN + 4: | |
| 2356 case NFA_MOPEN + 5: | |
| 2357 case NFA_MOPEN + 6: | |
| 2358 case NFA_MOPEN + 7: | |
| 2359 case NFA_MOPEN + 8: | |
| 2360 case NFA_MOPEN + 9: | |
| 2361 case NFA_NOPEN: /* \%( "Invisible Submatch" */ | |
| 2362 if (nfa_calc_size == TRUE) | |
| 2363 { | |
| 2364 nstate += 2; | |
| 2365 break; | |
| 2366 } | |
| 2367 | |
| 2368 mopen = *p; | |
| 2369 switch (*p) | |
| 2370 { | |
| 2371 case NFA_NOPEN: | |
| 2372 mclose = NFA_NCLOSE; | |
| 2373 break; | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2374 #ifdef FEAT_MBYTE |
| 4444 | 2375 case NFA_COMPOSING: |
| 2376 mclose = NFA_END_COMPOSING; | |
| 2377 break; | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2378 #endif |
| 4444 | 2379 default: |
| 2380 /* NFA_MOPEN(0) ... NFA_MOPEN(9) */ | |
| 2381 mclose = *p + NSUBEXP; | |
| 2382 break; | |
| 2383 } | |
| 2384 | |
| 2385 /* Allow "NFA_MOPEN" as a valid postfix representation for | |
| 2386 * the empty regexp "". In this case, the NFA will be | |
| 2387 * NFA_MOPEN -> NFA_MCLOSE. Note that this also allows | |
| 2388 * empty groups of parenthesis, and empty mbyte chars */ | |
| 2389 if (stackp == stack) | |
| 2390 { | |
| 2391 s = new_state(mopen, NULL, NULL); | |
| 2392 if (s == NULL) | |
| 4484 | 2393 goto theend; |
| 4444 | 2394 s1 = new_state(mclose, NULL, NULL); |
| 2395 if (s1 == NULL) | |
| 4484 | 2396 goto theend; |
| 4444 | 2397 patch(list1(&s->out), s1); |
| 2398 PUSH(frag(s, list1(&s1->out))); | |
| 2399 break; | |
| 2400 } | |
| 2401 | |
| 2402 /* At least one node was emitted before NFA_MOPEN, so | |
| 2403 * at least one node will be between NFA_MOPEN and NFA_MCLOSE */ | |
| 2404 e = POP(); | |
| 2405 s = new_state(mopen, e.start, NULL); /* `(' */ | |
| 2406 if (s == NULL) | |
| 4484 | 2407 goto theend; |
| 4444 | 2408 |
| 2409 s1 = new_state(mclose, NULL, NULL); /* `)' */ | |
| 2410 if (s1 == NULL) | |
| 4484 | 2411 goto theend; |
| 4444 | 2412 patch(e.out, s1); |
| 2413 | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2414 #ifdef FEAT_MBYTE |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2415 if (mopen == NFA_COMPOSING) |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2416 /* COMPOSING->out1 = END_COMPOSING */ |
| 4444 | 2417 patch(list1(&s->out1), s1); |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2418 #endif |
| 4444 | 2419 |
| 2420 PUSH(frag(s, list1(&s1->out))); | |
| 2421 break; | |
| 2422 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2423 case NFA_BACKREF1: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2424 case NFA_BACKREF2: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2425 case NFA_BACKREF3: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2426 case NFA_BACKREF4: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2427 case NFA_BACKREF5: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2428 case NFA_BACKREF6: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2429 case NFA_BACKREF7: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2430 case NFA_BACKREF8: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2431 case NFA_BACKREF9: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2432 if (nfa_calc_size == TRUE) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2433 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2434 nstate += 2; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2435 break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2436 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2437 s = new_state(*p, NULL, NULL); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2438 if (s == NULL) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2439 goto theend; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2440 s1 = new_state(NFA_SKIP, NULL, NULL); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2441 if (s1 == NULL) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2442 goto theend; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2443 patch(list1(&s->out), s1); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2444 PUSH(frag(s, list1(&s1->out))); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2445 break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2446 |
| 4444 | 2447 case NFA_ZSTART: |
| 2448 case NFA_ZEND: | |
| 2449 default: | |
| 2450 /* Operands */ | |
| 2451 if (nfa_calc_size == TRUE) | |
| 2452 { | |
| 4458 | 2453 nstate++; |
| 4444 | 2454 break; |
| 2455 } | |
| 2456 s = new_state(*p, NULL, NULL); | |
| 2457 if (s == NULL) | |
| 4484 | 2458 goto theend; |
| 4444 | 2459 PUSH(frag(s, list1(&s->out))); |
| 2460 break; | |
| 2461 | |
| 2462 } /* switch(*p) */ | |
| 2463 | |
| 2464 } /* for(p = postfix; *p; ++p) */ | |
| 2465 | |
| 2466 if (nfa_calc_size == TRUE) | |
| 2467 { | |
| 4458 | 2468 nstate++; |
| 4484 | 2469 goto theend; /* Return value when counting size is ignored anyway */ |
| 4444 | 2470 } |
| 2471 | |
| 2472 e = POP(); | |
| 2473 if (stackp != stack) | |
| 2474 EMSG_RET_NULL(_("E875: (NFA regexp) (While converting from postfix to NFA), too many states left on stack")); | |
| 2475 | |
| 2476 if (istate >= nstate) | |
| 2477 EMSG_RET_NULL(_("E876: (NFA regexp) Not enough space to store the whole NFA ")); | |
| 2478 | |
| 2479 matchstate = &state_ptr[istate++]; /* the match state */ | |
| 2480 matchstate->c = NFA_MATCH; | |
| 2481 matchstate->out = matchstate->out1 = NULL; | |
| 2482 | |
| 2483 patch(e.out, matchstate); | |
| 4484 | 2484 ret = e.start; |
| 2485 | |
| 2486 theend: | |
| 2487 vim_free(stack); | |
| 2488 return ret; | |
| 4444 | 2489 |
| 2490 #undef POP1 | |
| 2491 #undef PUSH1 | |
| 2492 #undef POP2 | |
| 2493 #undef PUSH2 | |
| 2494 #undef POP | |
| 2495 #undef PUSH | |
| 2496 } | |
| 2497 | |
| 2498 /**************************************************************** | |
| 2499 * NFA execution code. | |
| 2500 ****************************************************************/ | |
| 2501 | |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2502 typedef struct |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2503 { |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2504 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
|
2505 |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2506 /* 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
|
2507 union |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2508 { |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2509 struct multipos |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2510 { |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2511 lpos_T start; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2512 lpos_T end; |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2513 } multi[NSUBEXP]; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2514 struct linepos |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2515 { |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2516 char_u *start; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2517 char_u *end; |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2518 } line[NSUBEXP]; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2519 } list; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2520 } regsub_T; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2521 |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2522 /* 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
|
2523 typedef struct |
| 4444 | 2524 { |
| 2525 nfa_state_T *state; | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2526 int count; |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2527 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
|
2528 } nfa_thread_T; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2529 |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2530 /* nfa_list_T contains the alternative NFA execution states. */ |
| 4444 | 2531 typedef struct |
| 2532 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2533 nfa_thread_T *t; /* allocated array of states */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2534 int n; /* nr of states in "t" */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2535 int id; /* ID of the list */ |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2536 } nfa_list_T; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2537 |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2538 #ifdef ENABLE_LOG |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2539 static void |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2540 log_subexpr(sub) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2541 regsub_T *sub; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2542 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2543 int j; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2544 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2545 for (j = 0; j < sub->in_use; j++) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2546 if (REG_MULTI) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2547 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
|
2548 j, |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2549 sub->list.multi[j].start.col, |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2550 (int)sub->list.multi[j].start.lnum, |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2551 sub->list.multi[j].end.col, |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2552 (int)sub->list.multi[j].end.lnum); |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2553 else |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2554 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
|
2555 j, |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2556 (char *)sub->list.line[j].start, |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2557 (char *)sub->list.line[j].end); |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2558 fprintf(log_fd, "\n"); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2559 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2560 #endif |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2561 |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2562 /* 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
|
2563 static int nfa_match; |
|
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2564 |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2565 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
|
2566 static void addstate_here __ARGS((nfa_list_T *l, nfa_state_T *state, regsub_T *sub, int *ip)); |
| 4444 | 2567 |
| 2568 static void | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2569 addstate(l, state, sub, off) |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2570 nfa_list_T *l; /* runtime state list */ |
| 4444 | 2571 nfa_state_T *state; /* state to update */ |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2572 regsub_T *sub; /* pointers to subexpressions */ |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2573 int off; /* byte offset, when -1 go to next line */ |
| 4444 | 2574 { |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2575 int subidx; |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2576 nfa_thread_T *lastthread; |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2577 lpos_T save_lpos; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2578 int save_in_use; |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2579 char_u *save_ptr; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2580 int i; |
| 4444 | 2581 |
| 2582 if (l == NULL || state == NULL) | |
| 2583 return; | |
| 2584 | |
| 2585 switch (state->c) | |
| 2586 { | |
| 2587 case NFA_SPLIT: | |
| 2588 case NFA_NOT: | |
| 2589 case NFA_NOPEN: | |
| 2590 case NFA_NCLOSE: | |
| 2591 case NFA_MCLOSE: | |
| 2592 case NFA_MCLOSE + 1: | |
| 2593 case NFA_MCLOSE + 2: | |
| 2594 case NFA_MCLOSE + 3: | |
| 2595 case NFA_MCLOSE + 4: | |
| 2596 case NFA_MCLOSE + 5: | |
| 2597 case NFA_MCLOSE + 6: | |
| 2598 case NFA_MCLOSE + 7: | |
| 2599 case NFA_MCLOSE + 8: | |
| 2600 case NFA_MCLOSE + 9: | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2601 /* 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
|
2602 * "out1" may be added below. */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2603 break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2604 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2605 case NFA_MOPEN: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2606 case NFA_MOPEN + 1: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2607 case NFA_MOPEN + 2: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2608 case NFA_MOPEN + 3: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2609 case NFA_MOPEN + 4: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2610 case NFA_MOPEN + 5: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2611 case NFA_MOPEN + 6: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2612 case NFA_MOPEN + 7: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2613 case NFA_MOPEN + 8: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2614 case NFA_MOPEN + 9: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2615 /* 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
|
2616 * 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
|
2617 if (state->lastlist == l->id) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2618 return; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2619 state->lastlist = l->id; |
| 4444 | 2620 break; |
| 2621 | |
| 2622 default: | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2623 if (state->lastlist == l->id) |
| 4444 | 2624 { |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2625 /* 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
|
2626 * unless it is an MOPEN that is used for a backreference. */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2627 return; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2628 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2629 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2630 /* add the state to the list */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2631 state->lastlist = l->id; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2632 lastthread = &l->t[l->n++]; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2633 lastthread->state = state; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2634 lastthread->sub.in_use = sub->in_use; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2635 if (sub->in_use > 0) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2636 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2637 /* Copy the match start and end positions. */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2638 if (REG_MULTI) |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2639 mch_memmove(&lastthread->sub.list.multi[0], |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2640 &sub->list.multi[0], |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2641 sizeof(struct multipos) * sub->in_use); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2642 else |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2643 mch_memmove(&lastthread->sub.list.line[0], |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2644 &sub->list.line[0], |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2645 sizeof(struct linepos) * sub->in_use); |
| 4444 | 2646 } |
| 2647 } | |
| 2648 | |
| 2649 #ifdef ENABLE_LOG | |
| 2650 nfa_set_code(state->c); | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2651 fprintf(log_fd, "> Adding state %d to list. Character %d: %s\n", |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2652 abs(state->id), state->c, code); |
| 4444 | 2653 #endif |
| 2654 switch (state->c) | |
| 2655 { | |
| 2656 case NFA_MATCH: | |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2657 nfa_match = TRUE; |
| 4444 | 2658 break; |
| 2659 | |
| 2660 case NFA_SPLIT: | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2661 addstate(l, state->out, sub, off); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2662 addstate(l, state->out1, sub, off); |
| 4444 | 2663 break; |
| 2664 | |
| 2665 #if 0 | |
| 2666 case NFA_END_NEG_RANGE: | |
| 2667 /* Nothing to handle here. nfa_regmatch() will take care of it */ | |
| 2668 break; | |
| 2669 | |
| 2670 case NFA_NOT: | |
| 2671 EMSG(_("E999: (NFA regexp internal error) Should not process NOT node !")); | |
| 2672 #ifdef ENABLE_LOG | |
| 2673 fprintf(f, "\n\n>>> E999: Added state NFA_NOT to a list ... Something went wrong ! Why wasn't it processed already? \n\n"); | |
| 2674 #endif | |
| 2675 break; | |
| 2676 | |
| 2677 case NFA_COMPOSING: | |
| 2678 /* nfa_regmatch() will match all the bytes of this composing char. */ | |
| 2679 break; | |
| 2680 #endif | |
| 2681 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2682 case NFA_SKIP_CHAR: |
| 4444 | 2683 case NFA_NOPEN: |
| 2684 case NFA_NCLOSE: | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2685 addstate(l, state->out, sub, off); |
| 4444 | 2686 break; |
| 2687 | |
| 2688 /* If this state is reached, then a recursive call of nfa_regmatch() | |
| 2689 * succeeded. the next call saves the found submatches in the | |
| 2690 * first state after the "invisible" branch. */ | |
| 2691 #if 0 | |
| 2692 case NFA_END_INVISIBLE: | |
| 2693 break; | |
| 2694 #endif | |
| 2695 | |
| 2696 case NFA_MOPEN + 0: | |
| 2697 case NFA_MOPEN + 1: | |
| 2698 case NFA_MOPEN + 2: | |
| 2699 case NFA_MOPEN + 3: | |
| 2700 case NFA_MOPEN + 4: | |
| 2701 case NFA_MOPEN + 5: | |
| 2702 case NFA_MOPEN + 6: | |
| 2703 case NFA_MOPEN + 7: | |
| 2704 case NFA_MOPEN + 8: | |
| 2705 case NFA_MOPEN + 9: | |
| 2706 case NFA_ZSTART: | |
| 2707 if (state->c == NFA_ZSTART) | |
| 2708 subidx = 0; | |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2709 else |
|
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2710 subidx = state->c - NFA_MOPEN; |
| 4444 | 2711 |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2712 /* 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
|
2713 * restore it when it was in use. Otherwise fill any gap. */ |
| 4444 | 2714 if (REG_MULTI) |
| 2715 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2716 if (subidx < sub->in_use) |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2717 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2718 save_lpos = sub->list.multi[subidx].start; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2719 save_in_use = -1; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2720 } |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2721 else |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2722 { |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2723 save_in_use = sub->in_use; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2724 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
|
2725 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2726 sub->list.multi[i].start.lnum = -1; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2727 sub->list.multi[i].end.lnum = -1; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2728 } |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2729 sub->in_use = subidx + 1; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2730 } |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2731 if (off == -1) |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2732 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2733 sub->list.multi[subidx].start.lnum = reglnum + 1; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2734 sub->list.multi[subidx].start.col = 0; |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2735 } |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2736 else |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2737 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2738 sub->list.multi[subidx].start.lnum = reglnum; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2739 sub->list.multi[subidx].start.col = |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2740 (colnr_T)(reginput - regline + off); |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2741 } |
| 4444 | 2742 } |
| 2743 else | |
| 2744 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2745 if (subidx < sub->in_use) |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2746 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2747 save_ptr = sub->list.line[subidx].start; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2748 save_in_use = -1; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2749 } |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2750 else |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2751 { |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2752 save_in_use = sub->in_use; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2753 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
|
2754 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2755 sub->list.line[i].start = NULL; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2756 sub->list.line[i].end = NULL; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2757 } |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2758 sub->in_use = subidx + 1; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2759 } |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2760 sub->list.line[subidx].start = reginput + off; |
| 4444 | 2761 } |
| 2762 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2763 addstate(l, state->out, sub, off); |
| 4444 | 2764 |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2765 if (save_in_use == -1) |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2766 { |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2767 if (REG_MULTI) |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2768 sub->list.multi[subidx].start = save_lpos; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2769 else |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2770 sub->list.line[subidx].start = save_ptr; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2771 } |
| 4444 | 2772 else |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2773 sub->in_use = save_in_use; |
| 4444 | 2774 break; |
| 2775 | |
| 2776 case NFA_MCLOSE + 0: | |
|
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
2777 if (nfa_has_zend) |
| 4444 | 2778 { |
|
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
2779 /* 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
|
2780 * 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
|
2781 addstate(l, state->out, sub, off); |
| 4444 | 2782 break; |
| 2783 } | |
| 2784 case NFA_MCLOSE + 1: | |
| 2785 case NFA_MCLOSE + 2: | |
| 2786 case NFA_MCLOSE + 3: | |
| 2787 case NFA_MCLOSE + 4: | |
| 2788 case NFA_MCLOSE + 5: | |
| 2789 case NFA_MCLOSE + 6: | |
| 2790 case NFA_MCLOSE + 7: | |
| 2791 case NFA_MCLOSE + 8: | |
| 2792 case NFA_MCLOSE + 9: | |
| 2793 case NFA_ZEND: | |
| 2794 if (state->c == NFA_ZEND) | |
| 2795 subidx = 0; | |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2796 else |
|
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2797 subidx = state->c - NFA_MCLOSE; |
| 4444 | 2798 |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2799 /* 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
|
2800 * has done that. */ |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2801 save_in_use = sub->in_use; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2802 if (sub->in_use <= subidx) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2803 sub->in_use = subidx + 1; |
| 4444 | 2804 if (REG_MULTI) |
| 2805 { | |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2806 save_lpos = sub->list.multi[subidx].end; |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2807 if (off == -1) |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2808 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2809 sub->list.multi[subidx].end.lnum = reglnum + 1; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2810 sub->list.multi[subidx].end.col = 0; |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2811 } |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2812 else |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2813 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2814 sub->list.multi[subidx].end.lnum = reglnum; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2815 sub->list.multi[subidx].end.col = |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2816 (colnr_T)(reginput - regline + off); |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2817 } |
| 4444 | 2818 } |
| 2819 else | |
| 2820 { | |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2821 save_ptr = sub->list.line[subidx].end; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2822 sub->list.line[subidx].end = reginput + off; |
| 4444 | 2823 } |
| 2824 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2825 addstate(l, state->out, sub, off); |
| 4444 | 2826 |
| 2827 if (REG_MULTI) | |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2828 sub->list.multi[subidx].end = save_lpos; |
| 4444 | 2829 else |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2830 sub->list.line[subidx].end = save_ptr; |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2831 sub->in_use = save_in_use; |
| 4444 | 2832 break; |
| 2833 } | |
| 2834 } | |
| 2835 | |
| 2836 /* | |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2837 * 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
|
2838 * 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
|
2839 * 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
|
2840 * matters for alternatives. |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2841 */ |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2842 static void |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2843 addstate_here(l, state, sub, ip) |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2844 nfa_list_T *l; /* runtime state list */ |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2845 nfa_state_T *state; /* state to update */ |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2846 regsub_T *sub; /* pointers to subexpressions */ |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2847 int *ip; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2848 { |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2849 int tlen = l->n; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2850 int count; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2851 int i = *ip; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2852 |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2853 /* 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
|
2854 addstate(l, state, sub, 0); |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2855 |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2856 /* 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
|
2857 if (i + 1 == tlen) |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2858 return; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2859 |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2860 /* 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
|
2861 count = l->n - tlen; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2862 if (count > 1) |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2863 { |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2864 /* 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
|
2865 * end to the current position */ |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2866 mch_memmove(&(l->t[i + count]), |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2867 &(l->t[i + 1]), |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2868 sizeof(nfa_thread_T) * (l->n - i - 1)); |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2869 mch_memmove(&(l->t[i]), |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2870 &(l->t[l->n - 1]), |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2871 sizeof(nfa_thread_T) * count); |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2872 } |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2873 else |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2874 { |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2875 /* overwrite the current state */ |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2876 l->t[i] = l->t[l->n - 1]; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2877 } |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2878 --l->n; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2879 *ip = i - 1; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2880 } |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2881 |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2882 /* |
| 4444 | 2883 * Check character class "class" against current character c. |
| 2884 */ | |
| 2885 static int | |
| 2886 check_char_class(class, c) | |
| 2887 int class; | |
| 2888 int c; | |
| 2889 { | |
| 2890 switch (class) | |
| 2891 { | |
| 2892 case NFA_CLASS_ALNUM: | |
| 4464 | 2893 if (c >= 1 && c <= 255 && isalnum(c)) |
| 4444 | 2894 return OK; |
| 2895 break; | |
| 2896 case NFA_CLASS_ALPHA: | |
| 4464 | 2897 if (c >= 1 && c <= 255 && isalpha(c)) |
| 4444 | 2898 return OK; |
| 2899 break; | |
| 2900 case NFA_CLASS_BLANK: | |
| 2901 if (c == ' ' || c == '\t') | |
| 2902 return OK; | |
| 2903 break; | |
| 2904 case NFA_CLASS_CNTRL: | |
| 4464 | 2905 if (c >= 1 && c <= 255 && iscntrl(c)) |
| 4444 | 2906 return OK; |
| 2907 break; | |
| 2908 case NFA_CLASS_DIGIT: | |
| 2909 if (VIM_ISDIGIT(c)) | |
| 2910 return OK; | |
| 2911 break; | |
| 2912 case NFA_CLASS_GRAPH: | |
| 4464 | 2913 if (c >= 1 && c <= 255 && isgraph(c)) |
| 4444 | 2914 return OK; |
| 2915 break; | |
| 2916 case NFA_CLASS_LOWER: | |
| 2917 if (MB_ISLOWER(c)) | |
| 2918 return OK; | |
| 2919 break; | |
| 2920 case NFA_CLASS_PRINT: | |
| 2921 if (vim_isprintc(c)) | |
| 2922 return OK; | |
| 2923 break; | |
| 2924 case NFA_CLASS_PUNCT: | |
| 4464 | 2925 if (c >= 1 && c <= 255 && ispunct(c)) |
| 4444 | 2926 return OK; |
| 2927 break; | |
| 2928 case NFA_CLASS_SPACE: | |
| 2929 if ((c >=9 && c <= 13) || (c == ' ')) | |
| 2930 return OK; | |
| 2931 break; | |
| 2932 case NFA_CLASS_UPPER: | |
| 2933 if (MB_ISUPPER(c)) | |
| 2934 return OK; | |
| 2935 break; | |
| 2936 case NFA_CLASS_XDIGIT: | |
| 2937 if (vim_isxdigit(c)) | |
| 2938 return OK; | |
| 2939 break; | |
| 2940 case NFA_CLASS_TAB: | |
| 2941 if (c == '\t') | |
| 2942 return OK; | |
| 2943 break; | |
| 2944 case NFA_CLASS_RETURN: | |
| 2945 if (c == '\r') | |
| 2946 return OK; | |
| 2947 break; | |
| 2948 case NFA_CLASS_BACKSPACE: | |
| 2949 if (c == '\b') | |
| 2950 return OK; | |
| 2951 break; | |
| 2952 case NFA_CLASS_ESCAPE: | |
| 2953 if (c == '\033') | |
| 2954 return OK; | |
| 2955 break; | |
| 2956 | |
| 2957 default: | |
| 2958 /* should not be here :P */ | |
| 2959 EMSG_RET_FAIL(_("E877: (NFA regexp) Invalid character class ")); | |
| 2960 } | |
| 2961 return FAIL; | |
| 2962 } | |
| 2963 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2964 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
|
2965 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2966 /* |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2967 * Check for a match with subexpression "subidx". |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2968 * return TRUE if it matches. |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2969 */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2970 static int |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2971 match_backref(sub, subidx, bytelen) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2972 regsub_T *sub; /* pointers to subexpressions */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2973 int subidx; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2974 int *bytelen; /* out: length of match in bytes */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2975 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2976 int len; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2977 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2978 if (sub->in_use <= subidx) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2979 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2980 retempty: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2981 /* backref was not set, match an empty string */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2982 *bytelen = 0; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2983 return TRUE; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2984 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2985 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2986 if (REG_MULTI) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2987 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2988 if (sub->list.multi[subidx].start.lnum < 0 |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2989 || sub->list.multi[subidx].end.lnum < 0) |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2990 goto retempty; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2991 /* TODO: line breaks */ |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2992 len = sub->list.multi[subidx].end.col |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2993 - sub->list.multi[subidx].start.col; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2994 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
|
2995 reginput, &len) == 0) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2996 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2997 *bytelen = len; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2998 return TRUE; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2999 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3000 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3001 else |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3002 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3003 if (sub->list.line[subidx].start == NULL |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3004 || sub->list.line[subidx].end == NULL) |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3005 goto retempty; |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3006 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
|
3007 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
|
3008 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3009 *bytelen = len; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3010 return TRUE; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3011 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3012 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3013 return FALSE; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3014 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3015 |
| 4444 | 3016 /* |
| 3017 * Set all NFA nodes' list ID equal to -1. | |
| 3018 */ | |
| 3019 static void | |
| 3020 nfa_set_neg_listids(start) | |
| 3021 nfa_state_T *start; | |
| 3022 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3023 if (start != NULL && start->lastlist >= 0) |
| 4444 | 3024 { |
| 3025 start->lastlist = -1; | |
| 3026 nfa_set_neg_listids(start->out); | |
| 3027 nfa_set_neg_listids(start->out1); | |
| 3028 } | |
| 3029 } | |
| 3030 | |
| 3031 /* | |
| 3032 * Set all NFA nodes' list ID equal to 0. | |
| 3033 */ | |
| 3034 static void | |
| 3035 nfa_set_null_listids(start) | |
| 3036 nfa_state_T *start; | |
| 3037 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3038 if (start != NULL && start->lastlist == -1) |
| 4444 | 3039 { |
| 3040 start->lastlist = 0; | |
| 3041 nfa_set_null_listids(start->out); | |
| 3042 nfa_set_null_listids(start->out1); | |
| 3043 } | |
| 3044 } | |
| 3045 | |
| 3046 /* | |
| 3047 * Save list IDs for all NFA states in "list". | |
| 3048 */ | |
| 3049 static void | |
| 3050 nfa_save_listids(start, list) | |
| 3051 nfa_state_T *start; | |
| 3052 int *list; | |
| 3053 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3054 if (start != NULL && start->lastlist != -1) |
| 4444 | 3055 { |
| 3056 list[abs(start->id)] = start->lastlist; | |
| 3057 start->lastlist = -1; | |
| 3058 nfa_save_listids(start->out, list); | |
| 3059 nfa_save_listids(start->out1, list); | |
| 3060 } | |
| 3061 } | |
| 3062 | |
| 3063 /* | |
| 3064 * Restore list IDs from "list" to all NFA states. | |
| 3065 */ | |
| 3066 static void | |
| 3067 nfa_restore_listids(start, list) | |
| 3068 nfa_state_T *start; | |
| 3069 int *list; | |
| 3070 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3071 if (start != NULL && start->lastlist == -1) |
| 4444 | 3072 { |
| 3073 start->lastlist = list[abs(start->id)]; | |
| 3074 nfa_restore_listids(start->out, list); | |
| 3075 nfa_restore_listids(start->out1, list); | |
| 3076 } | |
| 3077 } | |
| 3078 | |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3079 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
|
3080 |
| 4444 | 3081 /* |
| 3082 * Main matching routine. | |
| 3083 * | |
| 3084 * Run NFA to determine whether it matches reginput. | |
| 3085 * | |
| 3086 * Return TRUE if there is a match, FALSE otherwise. | |
| 3087 * Note: Caller must ensure that: start != NULL. | |
| 3088 */ | |
| 3089 static int | |
| 3090 nfa_regmatch(start, submatch, m) | |
| 3091 nfa_state_T *start; | |
| 3092 regsub_T *submatch; | |
| 3093 regsub_T *m; | |
| 3094 { | |
| 3095 int result; | |
| 3096 int size = 0; | |
| 3097 int flag = 0; | |
| 3098 int old_reglnum = -1; | |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3099 int go_to_nextline = FALSE; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3100 nfa_thread_T *t; |
| 4444 | 3101 char_u *old_reginput = NULL; |
| 3102 char_u *old_regline = NULL; | |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3103 nfa_list_T list[3]; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3104 nfa_list_T *listtbl[2][2]; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3105 nfa_list_T *ll; |
| 4444 | 3106 int listid = 1; |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3107 int listidx; |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3108 nfa_list_T *thislist; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3109 nfa_list_T *nextlist; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3110 nfa_list_T *neglist; |
| 4444 | 3111 int *listids = NULL; |
| 4460 | 3112 #ifdef NFA_REGEXP_DEBUG_LOG |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3113 FILE *debug = fopen(NFA_REGEXP_DEBUG_LOG, "a"); |
| 4444 | 3114 |
| 3115 if (debug == NULL) | |
| 3116 { | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3117 EMSG2(_("(NFA) COULD NOT OPEN %s !"), NFA_REGEXP_DEBUG_LOG); |
| 4444 | 3118 return FALSE; |
| 3119 } | |
| 3120 #endif | |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3121 nfa_match = FALSE; |
| 4444 | 3122 |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3123 /* Allocate memory for the lists of nodes. */ |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3124 size = (nstate + 1) * sizeof(nfa_thread_T); |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3125 list[0].t = (nfa_thread_T *)lalloc(size, TRUE); |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3126 list[1].t = (nfa_thread_T *)lalloc(size, TRUE); |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3127 list[2].t = (nfa_thread_T *)lalloc(size, TRUE); |
| 4444 | 3128 if (list[0].t == NULL || list[1].t == NULL || list[2].t == NULL) |
| 3129 goto theend; | |
| 3130 vim_memset(list[0].t, 0, size); | |
| 3131 vim_memset(list[1].t, 0, size); | |
| 3132 vim_memset(list[2].t, 0, size); | |
| 3133 | |
| 3134 #ifdef ENABLE_LOG | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3135 log_fd = fopen(NFA_REGEXP_RUN_LOG, "a"); |
| 4444 | 3136 if (log_fd != NULL) |
| 3137 { | |
| 3138 fprintf(log_fd, "**********************************\n"); | |
| 3139 nfa_set_code(start->c); | |
| 3140 fprintf(log_fd, " RUNNING nfa_regmatch() starting with state %d, code %s\n", | |
| 3141 abs(start->id), code); | |
| 3142 fprintf(log_fd, "**********************************\n"); | |
| 3143 } | |
| 3144 else | |
| 3145 { | |
| 3146 EMSG(_("Could not open temporary log file for writing, displaying on stderr ... ")); | |
| 3147 log_fd = stderr; | |
| 3148 } | |
| 3149 #endif | |
| 3150 | |
| 3151 thislist = &list[0]; | |
| 3152 thislist->n = 0; | |
| 3153 nextlist = &list[1]; | |
| 3154 nextlist->n = 0; | |
| 3155 neglist = &list[2]; | |
| 3156 neglist->n = 0; | |
| 3157 #ifdef ENABLE_LOG | |
| 3158 fprintf(log_fd, "(---) STARTSTATE\n"); | |
| 3159 #endif | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3160 thislist->id = listid; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3161 addstate(thislist, start, m, 0); |
| 4444 | 3162 |
| 3163 /* There are two cases when the NFA advances: 1. input char matches the | |
| 3164 * NFA node and 2. input char does not match the NFA node, but the next | |
| 3165 * node is NFA_NOT. The following macro calls addstate() according to | |
| 3166 * these rules. It is used A LOT, so use the "listtbl" table for speed */ | |
| 3167 listtbl[0][0] = NULL; | |
| 3168 listtbl[0][1] = neglist; | |
| 3169 listtbl[1][0] = nextlist; | |
| 3170 listtbl[1][1] = NULL; | |
| 3171 #define ADD_POS_NEG_STATE(node) \ | |
| 3172 ll = listtbl[result ? 1 : 0][node->negated]; \ | |
| 3173 if (ll != NULL) \ | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3174 addstate(ll, node->out , &t->sub, clen); |
| 4444 | 3175 |
| 3176 | |
| 3177 /* | |
| 3178 * Run for each character. | |
| 3179 */ | |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3180 for (;;) |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3181 { |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3182 int curc; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3183 int clen; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3184 |
| 4444 | 3185 #ifdef FEAT_MBYTE |
| 3186 if (has_mbyte) | |
| 3187 { | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3188 curc = (*mb_ptr2char)(reginput); |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3189 clen = (*mb_ptr2len)(reginput); |
| 4444 | 3190 } |
| 3191 else | |
| 3192 #endif | |
| 3193 { | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3194 curc = *reginput; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3195 clen = 1; |
| 4444 | 3196 } |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3197 if (curc == NUL) |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3198 { |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3199 clen = 0; |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3200 go_to_nextline = FALSE; |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3201 } |
| 4444 | 3202 |
| 3203 /* swap lists */ | |
| 3204 thislist = &list[flag]; | |
| 3205 nextlist = &list[flag ^= 1]; | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3206 nextlist->n = 0; /* clear nextlist */ |
| 4444 | 3207 listtbl[1][0] = nextlist; |
| 3208 ++listid; | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3209 thislist->id = listid; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3210 nextlist->id = listid + 1; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3211 neglist->id = listid + 1; |
| 4444 | 3212 |
| 3213 #ifdef ENABLE_LOG | |
| 3214 fprintf(log_fd, "------------------------------------------\n"); | |
| 3215 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
|
3216 fprintf(log_fd, ">>> Advanced one character ... Current char is %c (code %d) \n", curc, (int)curc); |
| 4444 | 3217 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
|
3218 { |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3219 int i; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3220 |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3221 for (i = 0; i < thislist->n; i++) |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3222 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
|
3223 } |
| 4444 | 3224 fprintf(log_fd, "\n"); |
| 3225 #endif | |
| 3226 | |
| 4460 | 3227 #ifdef NFA_REGEXP_DEBUG_LOG |
| 4444 | 3228 fprintf(debug, "\n-------------------\n"); |
| 3229 #endif | |
| 4480 | 3230 /* |
| 3231 * If the state lists are empty we can stop. | |
| 3232 */ | |
| 3233 if (thislist->n == 0 && neglist->n == 0) | |
| 3234 break; | |
| 4444 | 3235 |
| 3236 /* compute nextlist */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3237 for (listidx = 0; listidx < thislist->n || neglist->n > 0; ++listidx) |
| 4444 | 3238 { |
| 3239 if (neglist->n > 0) | |
| 3240 { | |
| 3241 t = &neglist->t[0]; | |
| 4468 | 3242 neglist->n--; |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3243 listidx--; |
| 4444 | 3244 } |
| 3245 else | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3246 t = &thislist->t[listidx]; |
| 4444 | 3247 |
| 4460 | 3248 #ifdef NFA_REGEXP_DEBUG_LOG |
| 4444 | 3249 nfa_set_code(t->state->c); |
| 3250 fprintf(debug, "%s, ", code); | |
| 3251 #endif | |
| 3252 #ifdef ENABLE_LOG | |
| 3253 nfa_set_code(t->state->c); | |
| 3254 fprintf(log_fd, "(%d) %s, code %d ... \n", abs(t->state->id), | |
| 3255 code, (int)t->state->c); | |
| 3256 #endif | |
| 3257 | |
| 3258 /* | |
| 3259 * Handle the possible codes of the current state. | |
| 3260 * The most important is NFA_MATCH. | |
| 3261 */ | |
| 3262 switch (t->state->c) | |
| 3263 { | |
| 3264 case NFA_MATCH: | |
|
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3265 { |
|
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3266 int j; |
|
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3267 |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3268 nfa_match = TRUE; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3269 submatch->in_use = t->sub.in_use; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3270 if (REG_MULTI) |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3271 for (j = 0; j < submatch->in_use; j++) |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3272 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3273 submatch->list.multi[j].start = |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3274 t->sub.list.multi[j].start; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3275 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
|
3276 } |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3277 else |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3278 for (j = 0; j < submatch->in_use; j++) |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3279 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3280 submatch->list.line[j].start = |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3281 t->sub.list.line[j].start; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3282 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
|
3283 } |
| 4444 | 3284 #ifdef ENABLE_LOG |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3285 log_subexpr(&t->sub); |
| 4444 | 3286 #endif |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3287 /* 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
|
3288 * 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
|
3289 * 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
|
3290 * correct. */ |
|
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3291 if (nextlist->n == 0 && neglist->n == 0) |
|
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3292 clen = 0; |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3293 goto nextchar; |
|
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3294 } |
| 4444 | 3295 |
| 3296 case NFA_END_INVISIBLE: | |
| 3297 /* This is only encountered after a NFA_START_INVISIBLE node. | |
| 3298 * They surround a zero-width group, used with "\@=" and "\&". | |
| 3299 * If we got here, it means that the current "invisible" group | |
| 3300 * finished successfully, so return control to the parent | |
| 3301 * nfa_regmatch(). Submatches are stored in *m, and used in | |
| 3302 * the parent call. */ | |
| 3303 if (start->c == NFA_MOPEN + 0) | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3304 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
| 4444 | 3305 else |
| 3306 { | |
| 3307 *m = t->sub; | |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3308 nfa_match = TRUE; |
| 4444 | 3309 } |
| 3310 break; | |
| 3311 | |
| 3312 case NFA_START_INVISIBLE: | |
| 3313 /* Save global variables, and call nfa_regmatch() to check if | |
| 3314 * the current concat matches at this position. The concat | |
| 3315 * ends with the node NFA_END_INVISIBLE */ | |
| 3316 old_reginput = reginput; | |
| 3317 old_regline = regline; | |
| 3318 old_reglnum = reglnum; | |
| 3319 if (listids == NULL) | |
| 3320 { | |
| 3321 listids = (int *) lalloc(sizeof(int) * nstate, TRUE); | |
| 3322 if (listids == NULL) | |
| 3323 { | |
| 3324 EMSG(_("E878: (NFA) Could not allocate memory for branch traversal!")); | |
| 3325 return 0; | |
| 3326 } | |
| 3327 } | |
| 3328 #ifdef ENABLE_LOG | |
| 3329 if (log_fd != stderr) | |
| 3330 fclose(log_fd); | |
| 3331 log_fd = NULL; | |
| 3332 #endif | |
| 3333 /* Have to clear the listid field of the NFA nodes, so that | |
| 3334 * nfa_regmatch() and addstate() can run properly after | |
| 3335 * recursion. */ | |
| 3336 nfa_save_listids(start, listids); | |
| 3337 nfa_set_null_listids(start); | |
| 3338 result = nfa_regmatch(t->state->out, submatch, m); | |
| 3339 nfa_set_neg_listids(start); | |
| 3340 nfa_restore_listids(start, listids); | |
| 3341 | |
| 3342 #ifdef ENABLE_LOG | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3343 log_fd = fopen(NFA_REGEXP_RUN_LOG, "a"); |
| 4444 | 3344 if (log_fd != NULL) |
| 3345 { | |
| 3346 fprintf(log_fd, "****************************\n"); | |
| 3347 fprintf(log_fd, "FINISHED RUNNING nfa_regmatch() recursively\n"); | |
| 3348 fprintf(log_fd, "MATCH = %s\n", result == TRUE ? "OK" : "FALSE"); | |
| 3349 fprintf(log_fd, "****************************\n"); | |
| 3350 } | |
| 3351 else | |
| 3352 { | |
| 3353 EMSG(_("Could not open temporary log file for writing, displaying on stderr ... ")); | |
| 3354 log_fd = stderr; | |
| 3355 } | |
| 3356 #endif | |
| 3357 if (result == TRUE) | |
| 3358 { | |
|
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3359 int j; |
|
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3360 |
| 4444 | 3361 /* Restore position in input text */ |
| 3362 reginput = old_reginput; | |
| 3363 regline = old_regline; | |
| 3364 reglnum = old_reglnum; | |
| 3365 /* Copy submatch info from the recursive call */ | |
| 3366 if (REG_MULTI) | |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3367 for (j = 1; j < m->in_use; j++) |
| 4444 | 3368 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3369 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
|
3370 t->sub.list.multi[j].end = m->list.multi[j].end; |
| 4444 | 3371 } |
| 3372 else | |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3373 for (j = 1; j < m->in_use; j++) |
| 4444 | 3374 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3375 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
|
3376 t->sub.list.line[j].end = m->list.line[j].end; |
| 4444 | 3377 } |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3378 t->sub.in_use = m->in_use; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3379 |
| 4444 | 3380 /* t->state->out1 is the corresponding END_INVISIBLE node */ |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3381 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
|
3382 &listidx); |
| 4444 | 3383 } |
| 3384 else | |
| 3385 { | |
| 3386 /* continue with next input char */ | |
| 3387 reginput = old_reginput; | |
| 3388 } | |
| 3389 break; | |
| 3390 | |
| 3391 case NFA_BOL: | |
| 3392 if (reginput == regline) | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3393 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
| 4444 | 3394 break; |
| 3395 | |
| 3396 case NFA_EOL: | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3397 if (curc == NUL) |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3398 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
| 4444 | 3399 break; |
| 3400 | |
| 3401 case NFA_BOW: | |
| 3402 { | |
| 3403 int bow = TRUE; | |
| 3404 | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3405 if (curc == NUL) |
| 4444 | 3406 bow = FALSE; |
| 3407 #ifdef FEAT_MBYTE | |
| 3408 else if (has_mbyte) | |
| 3409 { | |
| 3410 int this_class; | |
| 3411 | |
| 3412 /* Get class of current and previous char (if it exists). */ | |
| 4503 | 3413 this_class = mb_get_class_buf(reginput, reg_buf); |
| 4444 | 3414 if (this_class <= 1) |
| 3415 bow = FALSE; | |
| 3416 else if (reg_prev_class() == this_class) | |
| 3417 bow = FALSE; | |
| 3418 } | |
| 3419 #endif | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3420 else if (!vim_iswordc_buf(curc, reg_buf) |
| 4503 | 3421 || (reginput > regline |
| 3422 && vim_iswordc_buf(reginput[-1], reg_buf))) | |
| 4444 | 3423 bow = FALSE; |
| 3424 if (bow) | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3425 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
| 4444 | 3426 break; |
| 3427 } | |
| 3428 | |
| 3429 case NFA_EOW: | |
| 3430 { | |
| 3431 int eow = TRUE; | |
| 3432 | |
| 3433 if (reginput == regline) | |
| 3434 eow = FALSE; | |
| 3435 #ifdef FEAT_MBYTE | |
| 3436 else if (has_mbyte) | |
| 3437 { | |
| 3438 int this_class, prev_class; | |
| 3439 | |
| 3440 /* Get class of current and previous char (if it exists). */ | |
| 4503 | 3441 this_class = mb_get_class_buf(reginput, reg_buf); |
| 4444 | 3442 prev_class = reg_prev_class(); |
| 3443 if (this_class == prev_class | |
| 3444 || prev_class == 0 || prev_class == 1) | |
| 3445 eow = FALSE; | |
| 3446 } | |
| 3447 #endif | |
| 4503 | 3448 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
|
3449 || (reginput[0] != NUL |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3450 && vim_iswordc_buf(curc, reg_buf))) |
| 4444 | 3451 eow = FALSE; |
| 3452 if (eow) | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3453 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
| 4444 | 3454 break; |
| 3455 } | |
| 3456 | |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3457 #ifdef FEAT_MBYTE |
| 4444 | 3458 case NFA_COMPOSING: |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3459 { |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3460 int mc = curc; |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3461 int len = 0; |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3462 nfa_state_T *end; |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3463 nfa_state_T *sta; |
|
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3464 int cchars[MAX_MCO]; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3465 int ccount = 0; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3466 int j; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3467 |
| 4444 | 3468 sta = t->state->out; |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3469 len = 0; |
|
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3470 if (utf_iscomposing(sta->c)) |
|
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3471 { |
|
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3472 /* Only match composing character(s), ignore base |
|
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3473 * character. Used for ".{composing}" and "{composing}" |
|
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3474 * (no preceding character). */ |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3475 len += mb_char2len(mc); |
|
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3476 } |
|
4551
f0c37f5cf4ea
updated for version 7.3.1023
Bram Moolenaar <bram@vim.org>
parents:
4549
diff
changeset
|
3477 if (ireg_icombine && len == 0) |
| 4444 | 3478 { |
|
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3479 /* If \Z was present, then ignore composing characters. |
|
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3480 * 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
|
3481 /* TODO: How about negated? */ |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3482 if (len == 0 && sta->c != curc) |
|
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3483 result = FAIL; |
|
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3484 else |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3485 result = OK; |
|
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3486 while (sta->c != NFA_END_COMPOSING) |
|
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3487 sta = sta->out; |
| 4444 | 3488 } |
|
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3489 |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3490 /* Check base character matches first, unless ignored. */ |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3491 else if (len > 0 || mc == sta->c) |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3492 { |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3493 if (len == 0) |
|
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3494 { |
|
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3495 len += mb_char2len(mc); |
|
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3496 sta = sta->out; |
|
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3497 } |
| 4444 | 3498 |
|
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3499 /* 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
|
3500 * Get them into cchars[] first. */ |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3501 while (len < clen) |
|
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3502 { |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3503 mc = mb_ptr2char(reginput + len); |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3504 cchars[ccount++] = mc; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3505 len += mb_char2len(mc); |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3506 if (ccount == MAX_MCO) |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3507 break; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3508 } |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3509 |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3510 /* 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
|
3511 * 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
|
3512 * composing chars are matched. */ |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3513 result = OK; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3514 while (sta->c != NFA_END_COMPOSING) |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3515 { |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3516 for (j = 0; j < ccount; ++j) |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3517 if (cchars[j] == sta->c) |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3518 break; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3519 if (j == ccount) |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3520 { |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3521 result = FAIL; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3522 break; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3523 } |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3524 sta = sta->out; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3525 } |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3526 } |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3527 else |
|
4525
36ddcf4cecbc
updated for version 7.3.1010
Bram Moolenaar <bram@vim.org>
parents:
4517
diff
changeset
|
3528 result = FAIL; |
|
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3529 |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3530 end = t->state->out1; /* NFA_END_COMPOSING */ |
| 4444 | 3531 ADD_POS_NEG_STATE(end); |
| 3532 break; | |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3533 } |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3534 #endif |
| 4444 | 3535 |
| 3536 case NFA_NEWL: | |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3537 if (curc == NUL && !reg_line_lbr && REG_MULTI |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3538 && reglnum <= reg_maxline) |
| 4444 | 3539 { |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3540 go_to_nextline = TRUE; |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3541 /* 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
|
3542 * at the start of the next line. */ |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3543 addstate(nextlist, t->state->out, &t->sub, -1); |
| 4444 | 3544 } |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3545 else if (curc == '\n' && reg_line_lbr) |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3546 { |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3547 /* 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
|
3548 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
|
3549 } |
| 4444 | 3550 break; |
| 3551 | |
| 3552 case NFA_CLASS_ALNUM: | |
| 3553 case NFA_CLASS_ALPHA: | |
| 3554 case NFA_CLASS_BLANK: | |
| 3555 case NFA_CLASS_CNTRL: | |
| 3556 case NFA_CLASS_DIGIT: | |
| 3557 case NFA_CLASS_GRAPH: | |
| 3558 case NFA_CLASS_LOWER: | |
| 3559 case NFA_CLASS_PRINT: | |
| 3560 case NFA_CLASS_PUNCT: | |
| 3561 case NFA_CLASS_SPACE: | |
| 3562 case NFA_CLASS_UPPER: | |
| 3563 case NFA_CLASS_XDIGIT: | |
| 3564 case NFA_CLASS_TAB: | |
| 3565 case NFA_CLASS_RETURN: | |
| 3566 case NFA_CLASS_BACKSPACE: | |
| 3567 case NFA_CLASS_ESCAPE: | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3568 result = check_char_class(t->state->c, curc); |
| 4444 | 3569 ADD_POS_NEG_STATE(t->state); |
| 3570 break; | |
| 3571 | |
| 3572 case NFA_END_NEG_RANGE: | |
| 3573 /* This follows a series of negated nodes, like: | |
| 3574 * 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
|
3575 if (curc > 0) |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3576 addstate(nextlist, t->state->out, &t->sub, clen); |
| 4444 | 3577 break; |
| 3578 | |
| 3579 case NFA_ANY: | |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3580 /* 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
|
3581 if (curc > 0) |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3582 addstate(nextlist, t->state->out, &t->sub, clen); |
| 4444 | 3583 break; |
| 3584 | |
| 3585 /* | |
| 3586 * Character classes like \a for alpha, \d for digit etc. | |
| 3587 */ | |
| 3588 case NFA_IDENT: /* \i */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3589 result = vim_isIDc(curc); |
| 4444 | 3590 ADD_POS_NEG_STATE(t->state); |
| 3591 break; | |
| 3592 | |
| 3593 case NFA_SIDENT: /* \I */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3594 result = !VIM_ISDIGIT(curc) && vim_isIDc(curc); |
| 4444 | 3595 ADD_POS_NEG_STATE(t->state); |
| 3596 break; | |
| 3597 | |
| 3598 case NFA_KWORD: /* \k */ | |
| 4503 | 3599 result = vim_iswordp_buf(reginput, reg_buf); |
| 4444 | 3600 ADD_POS_NEG_STATE(t->state); |
| 3601 break; | |
| 3602 | |
| 3603 case NFA_SKWORD: /* \K */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3604 result = !VIM_ISDIGIT(curc) |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3605 && vim_iswordp_buf(reginput, reg_buf); |
| 4444 | 3606 ADD_POS_NEG_STATE(t->state); |
| 3607 break; | |
| 3608 | |
| 3609 case NFA_FNAME: /* \f */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3610 result = vim_isfilec(curc); |
| 4444 | 3611 ADD_POS_NEG_STATE(t->state); |
| 3612 break; | |
| 3613 | |
| 3614 case NFA_SFNAME: /* \F */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3615 result = !VIM_ISDIGIT(curc) && vim_isfilec(curc); |
| 4444 | 3616 ADD_POS_NEG_STATE(t->state); |
| 3617 break; | |
| 3618 | |
| 3619 case NFA_PRINT: /* \p */ | |
| 4470 | 3620 result = ptr2cells(reginput) == 1; |
| 4444 | 3621 ADD_POS_NEG_STATE(t->state); |
| 3622 break; | |
| 3623 | |
| 3624 case NFA_SPRINT: /* \P */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3625 result = !VIM_ISDIGIT(curc) && ptr2cells(reginput) == 1; |
| 4444 | 3626 ADD_POS_NEG_STATE(t->state); |
| 3627 break; | |
| 3628 | |
| 3629 case NFA_WHITE: /* \s */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3630 result = vim_iswhite(curc); |
| 4444 | 3631 ADD_POS_NEG_STATE(t->state); |
| 3632 break; | |
| 3633 | |
| 3634 case NFA_NWHITE: /* \S */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3635 result = curc != NUL && !vim_iswhite(curc); |
| 4444 | 3636 ADD_POS_NEG_STATE(t->state); |
| 3637 break; | |
| 3638 | |
| 3639 case NFA_DIGIT: /* \d */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3640 result = ri_digit(curc); |
| 4444 | 3641 ADD_POS_NEG_STATE(t->state); |
| 3642 break; | |
| 3643 | |
| 3644 case NFA_NDIGIT: /* \D */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3645 result = curc != NUL && !ri_digit(curc); |
| 4444 | 3646 ADD_POS_NEG_STATE(t->state); |
| 3647 break; | |
| 3648 | |
| 3649 case NFA_HEX: /* \x */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3650 result = ri_hex(curc); |
| 4444 | 3651 ADD_POS_NEG_STATE(t->state); |
| 3652 break; | |
| 3653 | |
| 3654 case NFA_NHEX: /* \X */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3655 result = curc != NUL && !ri_hex(curc); |
| 4444 | 3656 ADD_POS_NEG_STATE(t->state); |
| 3657 break; | |
| 3658 | |
| 3659 case NFA_OCTAL: /* \o */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3660 result = ri_octal(curc); |
| 4444 | 3661 ADD_POS_NEG_STATE(t->state); |
| 3662 break; | |
| 3663 | |
| 3664 case NFA_NOCTAL: /* \O */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3665 result = curc != NUL && !ri_octal(curc); |
| 4444 | 3666 ADD_POS_NEG_STATE(t->state); |
| 3667 break; | |
| 3668 | |
| 3669 case NFA_WORD: /* \w */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3670 result = ri_word(curc); |
| 4444 | 3671 ADD_POS_NEG_STATE(t->state); |
| 3672 break; | |
| 3673 | |
| 3674 case NFA_NWORD: /* \W */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3675 result = curc != NUL && !ri_word(curc); |
| 4444 | 3676 ADD_POS_NEG_STATE(t->state); |
| 3677 break; | |
| 3678 | |
| 3679 case NFA_HEAD: /* \h */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3680 result = ri_head(curc); |
| 4444 | 3681 ADD_POS_NEG_STATE(t->state); |
| 3682 break; | |
| 3683 | |
| 3684 case NFA_NHEAD: /* \H */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3685 result = curc != NUL && !ri_head(curc); |
| 4444 | 3686 ADD_POS_NEG_STATE(t->state); |
| 3687 break; | |
| 3688 | |
| 3689 case NFA_ALPHA: /* \a */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3690 result = ri_alpha(curc); |
| 4444 | 3691 ADD_POS_NEG_STATE(t->state); |
| 3692 break; | |
| 3693 | |
| 3694 case NFA_NALPHA: /* \A */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3695 result = curc != NUL && !ri_alpha(curc); |
| 4444 | 3696 ADD_POS_NEG_STATE(t->state); |
| 3697 break; | |
| 3698 | |
| 3699 case NFA_LOWER: /* \l */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3700 result = ri_lower(curc); |
| 4444 | 3701 ADD_POS_NEG_STATE(t->state); |
| 3702 break; | |
| 3703 | |
| 3704 case NFA_NLOWER: /* \L */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3705 result = curc != NUL && !ri_lower(curc); |
| 4444 | 3706 ADD_POS_NEG_STATE(t->state); |
| 3707 break; | |
| 3708 | |
| 3709 case NFA_UPPER: /* \u */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3710 result = ri_upper(curc); |
| 4444 | 3711 ADD_POS_NEG_STATE(t->state); |
| 3712 break; | |
| 3713 | |
| 3714 case NFA_NUPPER: /* \U */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3715 result = curc != NUL && !ri_upper(curc); |
| 4444 | 3716 ADD_POS_NEG_STATE(t->state); |
| 3717 break; | |
| 3718 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3719 case NFA_BACKREF1: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3720 case NFA_BACKREF2: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3721 case NFA_BACKREF3: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3722 case NFA_BACKREF4: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3723 case NFA_BACKREF5: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3724 case NFA_BACKREF6: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3725 case NFA_BACKREF7: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3726 case NFA_BACKREF8: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3727 case NFA_BACKREF9: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3728 /* \1 .. \9 */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3729 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3730 int subidx = t->state->c - NFA_BACKREF1 + 1; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3731 int bytelen; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3732 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3733 result = match_backref(&t->sub, subidx, &bytelen); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3734 if (result) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3735 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3736 if (bytelen == 0) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3737 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3738 /* 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
|
3739 * be used next */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3740 addstate_here(thislist, t->state->out, &t->sub, |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3741 &listidx); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3742 thislist->t[listidx + 1].count = 0; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3743 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3744 else if (bytelen <= clen) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3745 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3746 /* match current character, jump ahead to out of |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3747 * NFA_SKIP */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3748 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
|
3749 #ifdef ENABLE_LOG |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3750 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3751 #endif |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3752 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3753 else |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3754 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3755 /* skip ofer the matched characters, set character |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3756 * count in NFA_SKIP */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3757 addstate(nextlist, t->state->out, &t->sub, bytelen); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3758 nextlist->t[nextlist->n - 1].count = bytelen - clen; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3759 #ifdef ENABLE_LOG |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3760 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3761 #endif |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3762 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3763 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3764 } |
| 4482 | 3765 break; |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3766 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3767 case NFA_SKIP: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3768 /* charater of previous matching \1 .. \9 */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3769 if (t->count - clen <= 0) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3770 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3771 /* end of match, go to what follows */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3772 addstate(nextlist, t->state->out, &t->sub, clen); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3773 #ifdef ENABLE_LOG |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3774 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3775 #endif |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3776 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3777 else |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3778 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3779 /* add state again with decremented count */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3780 addstate(nextlist, t->state, &t->sub, 0); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3781 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
|
3782 #ifdef ENABLE_LOG |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3783 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3784 #endif |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3785 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3786 break; |
| 4482 | 3787 |
| 3788 case NFA_SKIP_CHAR: | |
| 3789 case NFA_ZSTART: | |
|
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
3790 case NFA_ZEND: |
| 4482 | 3791 /* TODO: should not happen? */ |
| 3792 break; | |
| 3793 | |
| 4444 | 3794 default: /* regular character */ |
|
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3795 { |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3796 int c = t->state->c; |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3797 |
| 4482 | 3798 /* TODO: put this in #ifdef later */ |
|
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3799 if (c < -256) |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3800 EMSGN("INTERNAL: Negative state char: %ld", c); |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3801 if (is_Magic(c)) |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3802 c = un_Magic(c); |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3803 result = (c == curc); |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3804 |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3805 if (!result && ireg_ic) |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3806 result = MB_TOLOWER(c) == MB_TOLOWER(curc); |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3807 #ifdef FEAT_MBYTE |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3808 /* 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
|
3809 * 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
|
3810 * character uses NFA_COMPOSING above. */ |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3811 if (result && enc_utf8 && !ireg_icombine |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3812 && clen != utf_char2len(curc)) |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3813 result = FALSE; |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3814 #endif |
| 4444 | 3815 ADD_POS_NEG_STATE(t->state); |
| 3816 break; | |
|
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3817 } |
| 4444 | 3818 } |
| 3819 | |
| 3820 } /* for (thislist = thislist; thislist->state; thislist++) */ | |
| 3821 | |
|
4557
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
3822 /* 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
|
3823 * start state to the list of states. |
|
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
3824 * 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
|
3825 * matters! |
|
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
3826 * 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
|
3827 * 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
|
3828 * 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
|
3829 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
|
3830 && reglnum == 0 && clen != 0 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
3831 && (ireg_maxcol == 0 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
3832 || (colnr_T)(reginput - regline) < ireg_maxcol)) |
| 4444 | 3833 { |
| 3834 #ifdef ENABLE_LOG | |
| 3835 fprintf(log_fd, "(---) STARTSTATE\n"); | |
| 3836 #endif | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3837 addstate(nextlist, start, m, clen); |
| 4444 | 3838 } |
| 3839 | |
| 3840 #ifdef ENABLE_LOG | |
| 3841 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
|
3842 { |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3843 int i; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3844 |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3845 for (i = 0; i < thislist->n; i++) |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3846 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
|
3847 } |
| 4444 | 3848 fprintf(log_fd, "\n"); |
| 3849 #endif | |
| 3850 | |
| 3851 nextchar: | |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3852 /* 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
|
3853 * finish. */ |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3854 if (clen != 0) |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3855 reginput += clen; |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3856 else if (go_to_nextline) |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3857 reg_nextline(); |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3858 else |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3859 break; |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3860 } |
| 4444 | 3861 |
| 3862 #ifdef ENABLE_LOG | |
| 3863 if (log_fd != stderr) | |
| 3864 fclose(log_fd); | |
| 3865 log_fd = NULL; | |
| 3866 #endif | |
| 3867 | |
| 3868 theend: | |
| 3869 /* Free memory */ | |
| 3870 vim_free(list[0].t); | |
| 3871 vim_free(list[1].t); | |
| 3872 vim_free(list[2].t); | |
| 3873 list[0].t = list[1].t = list[2].t = NULL; | |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3874 vim_free(listids); |
| 4444 | 3875 #undef ADD_POS_NEG_STATE |
| 4460 | 3876 #ifdef NFA_REGEXP_DEBUG_LOG |
| 4444 | 3877 fclose(debug); |
| 3878 #endif | |
| 3879 | |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3880 return nfa_match; |
| 4444 | 3881 } |
| 3882 | |
| 3883 /* | |
| 3884 * Try match of "prog" with at regline["col"]. | |
| 3885 * Returns 0 for failure, number of lines contained in the match otherwise. | |
| 3886 */ | |
| 3887 static long | |
| 3888 nfa_regtry(start, col) | |
| 3889 nfa_state_T *start; | |
| 3890 colnr_T col; | |
| 3891 { | |
| 3892 int i; | |
| 3893 regsub_T sub, m; | |
| 3894 #ifdef ENABLE_LOG | |
| 3895 FILE *f; | |
| 3896 #endif | |
| 3897 | |
| 3898 reginput = regline + col; | |
| 3899 need_clear_subexpr = TRUE; | |
| 3900 | |
| 3901 #ifdef ENABLE_LOG | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3902 f = fopen(NFA_REGEXP_RUN_LOG, "a"); |
| 4444 | 3903 if (f != NULL) |
| 3904 { | |
| 3905 fprintf(f, "\n\n\n\n\n\n\t\t=======================================================\n"); | |
| 3906 fprintf(f, " =======================================================\n"); | |
| 3907 #ifdef DEBUG | |
| 3908 fprintf(f, "\tRegexp is \"%s\"\n", nfa_regengine.expr); | |
| 3909 #endif | |
| 3910 fprintf(f, "\tInput text is \"%s\" \n", reginput); | |
| 3911 fprintf(f, " =======================================================\n\n\n\n\n\n\n"); | |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
3912 nfa_print_state(f, start); |
| 4444 | 3913 fprintf(f, "\n\n"); |
| 3914 fclose(f); | |
| 3915 } | |
| 3916 else | |
| 3917 EMSG(_("Could not open temporary log file for writing ")); | |
| 3918 #endif | |
| 3919 | |
| 3920 if (REG_MULTI) | |
| 3921 { | |
| 3922 /* Use 0xff to set lnum to -1 */ | |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3923 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
|
3924 vim_memset(m.list.multi, 0xff, sizeof(struct multipos) * nfa_nsubexpr); |
| 4444 | 3925 } |
| 3926 else | |
| 3927 { | |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3928 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
|
3929 vim_memset(m.list.line, 0, sizeof(struct linepos) * nfa_nsubexpr); |
| 4444 | 3930 } |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3931 sub.in_use = 0; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3932 m.in_use = 0; |
| 4444 | 3933 |
| 3934 if (nfa_regmatch(start, &sub, &m) == FALSE) | |
| 3935 return 0; | |
| 3936 | |
| 3937 cleanup_subexpr(); | |
| 3938 if (REG_MULTI) | |
| 3939 { | |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3940 for (i = 0; i < sub.in_use; i++) |
| 4444 | 3941 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3942 reg_startpos[i] = sub.list.multi[i].start; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3943 reg_endpos[i] = sub.list.multi[i].end; |
| 4444 | 3944 } |
| 3945 | |
| 3946 if (reg_startpos[0].lnum < 0) | |
| 3947 { | |
| 3948 reg_startpos[0].lnum = 0; | |
| 3949 reg_startpos[0].col = col; | |
| 3950 } | |
| 3951 if (reg_endpos[0].lnum < 0) | |
| 3952 { | |
|
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
3953 /* pattern has a \ze but it didn't match, use current end */ |
| 4444 | 3954 reg_endpos[0].lnum = reglnum; |
| 3955 reg_endpos[0].col = (int)(reginput - regline); | |
| 3956 } | |
| 3957 else | |
| 3958 /* Use line number of "\ze". */ | |
| 3959 reglnum = reg_endpos[0].lnum; | |
| 3960 } | |
| 3961 else | |
| 3962 { | |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3963 for (i = 0; i < sub.in_use; i++) |
| 4444 | 3964 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3965 reg_startp[i] = sub.list.line[i].start; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3966 reg_endp[i] = sub.list.line[i].end; |
| 4444 | 3967 } |
| 3968 | |
| 3969 if (reg_startp[0] == NULL) | |
| 3970 reg_startp[0] = regline + col; | |
| 3971 if (reg_endp[0] == NULL) | |
| 3972 reg_endp[0] = reginput; | |
| 3973 } | |
| 3974 | |
| 3975 return 1 + reglnum; | |
| 3976 } | |
| 3977 | |
| 3978 /* | |
| 3979 * Match a regexp against a string ("line" points to the string) or multiple | |
| 3980 * lines ("line" is NULL, use reg_getline()). | |
| 3981 * | |
| 3982 * Returns 0 for failure, number of lines contained in the match otherwise. | |
| 3983 */ | |
| 3984 static long | |
| 3985 nfa_regexec_both(line, col) | |
| 3986 char_u *line; | |
| 3987 colnr_T col; /* column to start looking for match */ | |
| 3988 { | |
| 3989 nfa_regprog_T *prog; | |
| 3990 long retval = 0L; | |
| 3991 int i; | |
| 3992 | |
| 3993 if (REG_MULTI) | |
| 3994 { | |
| 3995 prog = (nfa_regprog_T *)reg_mmatch->regprog; | |
| 3996 line = reg_getline((linenr_T)0); /* relative to the cursor */ | |
| 3997 reg_startpos = reg_mmatch->startpos; | |
| 3998 reg_endpos = reg_mmatch->endpos; | |
| 3999 } | |
| 4000 else | |
| 4001 { | |
| 4002 prog = (nfa_regprog_T *)reg_match->regprog; | |
| 4003 reg_startp = reg_match->startp; | |
| 4004 reg_endp = reg_match->endp; | |
| 4005 } | |
| 4006 | |
| 4007 /* Be paranoid... */ | |
| 4008 if (prog == NULL || line == NULL) | |
| 4009 { | |
| 4010 EMSG(_(e_null)); | |
| 4011 goto theend; | |
| 4012 } | |
| 4013 | |
| 4014 /* If the start column is past the maximum column: no need to try. */ | |
| 4015 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
| 4016 goto theend; | |
| 4017 | |
| 4018 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */ | |
| 4019 if (prog->regflags & RF_ICASE) | |
| 4020 ireg_ic = TRUE; | |
| 4021 else if (prog->regflags & RF_NOICASE) | |
| 4022 ireg_ic = FALSE; | |
| 4023 | |
| 4024 #ifdef FEAT_MBYTE | |
| 4025 /* If pattern contains "\Z" overrule value of ireg_icombine */ | |
| 4026 if (prog->regflags & RF_ICOMBINE) | |
| 4027 ireg_icombine = TRUE; | |
| 4028 #endif | |
| 4029 | |
| 4030 regline = line; | |
| 4031 reglnum = 0; /* relative to line */ | |
| 4032 | |
|
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4033 nfa_has_zend = prog->has_zend; |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4034 nfa_nsubexpr = prog->nsubexp; |
|
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4035 |
| 4444 | 4036 nstate = prog->nstate; |
| 4037 for (i = 0; i < nstate; ++i) | |
| 4038 { | |
| 4039 prog->state[i].id = i; | |
| 4040 prog->state[i].lastlist = 0; | |
| 4041 } | |
| 4042 | |
| 4043 retval = nfa_regtry(prog->start, col); | |
| 4044 | |
| 4045 theend: | |
| 4046 return retval; | |
| 4047 } | |
| 4048 | |
| 4049 /* | |
| 4050 * Compile a regular expression into internal code for the NFA matcher. | |
| 4051 * Returns the program in allocated space. Returns NULL for an error. | |
| 4052 */ | |
| 4053 static regprog_T * | |
| 4054 nfa_regcomp(expr, re_flags) | |
| 4055 char_u *expr; | |
| 4056 int re_flags; | |
| 4057 { | |
|
4541
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4058 nfa_regprog_T *prog = NULL; |
| 4458 | 4059 size_t prog_size; |
| 4444 | 4060 int *postfix; |
| 4061 | |
| 4062 if (expr == NULL) | |
| 4063 return NULL; | |
| 4064 | |
| 4065 #ifdef DEBUG | |
| 4066 nfa_regengine.expr = expr; | |
| 4067 #endif | |
| 4068 | |
| 4069 init_class_tab(); | |
| 4070 | |
| 4071 if (nfa_regcomp_start(expr, re_flags) == FAIL) | |
| 4072 return NULL; | |
| 4073 | |
| 4074 /* 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
|
4075 * (and count its size). */ |
| 4444 | 4076 postfix = re2post(); |
| 4077 if (postfix == NULL) | |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4078 { |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4079 /* TODO: only give this error for debugging? */ |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4080 if (post_ptr >= post_end) |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4081 EMSGN("Internal error: estimated max number of states insufficient: %ld", post_end - post_start); |
| 4444 | 4082 goto fail; /* Cascaded (syntax?) error */ |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4083 } |
| 4444 | 4084 |
| 4085 /* | |
| 4086 * In order to build the NFA, we parse the input regexp twice: | |
| 4087 * 1. first pass to count size (so we can allocate space) | |
| 4088 * 2. second to emit code | |
| 4089 */ | |
| 4090 #ifdef ENABLE_LOG | |
| 4091 { | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
4092 FILE *f = fopen(NFA_REGEXP_RUN_LOG, "a"); |
| 4444 | 4093 |
| 4094 if (f != NULL) | |
| 4095 { | |
| 4096 fprintf(f, "\n*****************************\n\n\n\n\tCompiling regexp \"%s\" ... hold on !\n", expr); | |
| 4097 fclose(f); | |
| 4098 } | |
| 4099 } | |
| 4100 #endif | |
| 4101 | |
| 4102 /* | |
| 4103 * PASS 1 | |
| 4104 * Count number of NFA states in "nstate". Do not build the NFA. | |
| 4105 */ | |
| 4106 post2nfa(postfix, post_ptr, TRUE); | |
|
4541
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4107 |
|
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4108 /* Space for compiled regexp */ |
|
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4109 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
|
4110 prog = (nfa_regprog_T *)lalloc(prog_size, TRUE); |
|
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4111 if (prog == NULL) |
|
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4112 goto fail; |
|
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4113 vim_memset(prog, 0, prog_size); |
| 4444 | 4114 state_ptr = prog->state; |
| 4115 | |
| 4116 /* | |
| 4117 * PASS 2 | |
| 4118 * Build the NFA | |
| 4119 */ | |
| 4120 prog->start = post2nfa(postfix, post_ptr, FALSE); | |
| 4121 if (prog->start == NULL) | |
| 4122 goto fail; | |
| 4123 | |
| 4124 prog->regflags = regflags; | |
| 4125 prog->engine = &nfa_regengine; | |
| 4126 prog->nstate = nstate; | |
|
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4127 prog->has_zend = nfa_has_zend; |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4128 prog->nsubexp = regnpar; |
| 4444 | 4129 #ifdef ENABLE_LOG |
| 4130 nfa_postfix_dump(expr, OK); | |
| 4131 nfa_dump(prog); | |
| 4132 #endif | |
| 4133 | |
| 4134 out: | |
| 4135 vim_free(post_start); | |
| 4136 post_start = post_ptr = post_end = NULL; | |
| 4137 state_ptr = NULL; | |
| 4138 return (regprog_T *)prog; | |
| 4139 | |
| 4140 fail: | |
| 4141 vim_free(prog); | |
| 4142 prog = NULL; | |
| 4143 #ifdef ENABLE_LOG | |
| 4144 nfa_postfix_dump(expr, FAIL); | |
| 4145 #endif | |
| 4146 #ifdef DEBUG | |
| 4147 nfa_regengine.expr = NULL; | |
| 4148 #endif | |
| 4149 goto out; | |
| 4150 } | |
| 4151 | |
| 4152 | |
| 4153 /* | |
| 4154 * Match a regexp against a string. | |
| 4155 * "rmp->regprog" is a compiled regexp as returned by nfa_regcomp(). | |
| 4156 * Uses curbuf for line count and 'iskeyword'. | |
| 4157 * | |
| 4158 * Return TRUE if there is a match, FALSE if not. | |
| 4159 */ | |
| 4160 static int | |
| 4161 nfa_regexec(rmp, line, col) | |
| 4162 regmatch_T *rmp; | |
| 4163 char_u *line; /* string to match against */ | |
| 4164 colnr_T col; /* column to start looking for match */ | |
| 4165 { | |
| 4166 reg_match = rmp; | |
| 4167 reg_mmatch = NULL; | |
| 4168 reg_maxline = 0; | |
| 4169 reg_line_lbr = FALSE; | |
| 4170 reg_buf = curbuf; | |
| 4171 reg_win = NULL; | |
| 4172 ireg_ic = rmp->rm_ic; | |
| 4173 #ifdef FEAT_MBYTE | |
| 4174 ireg_icombine = FALSE; | |
| 4175 #endif | |
| 4176 ireg_maxcol = 0; | |
| 4177 return (nfa_regexec_both(line, col) != 0); | |
| 4178 } | |
| 4179 | |
| 4180 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
| 4181 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
| 4182 | |
| 4183 static int nfa_regexec_nl __ARGS((regmatch_T *rmp, char_u *line, colnr_T col)); | |
| 4184 | |
| 4185 /* | |
| 4186 * Like nfa_regexec(), but consider a "\n" in "line" to be a line break. | |
| 4187 */ | |
| 4188 static int | |
| 4189 nfa_regexec_nl(rmp, line, col) | |
| 4190 regmatch_T *rmp; | |
| 4191 char_u *line; /* string to match against */ | |
| 4192 colnr_T col; /* column to start looking for match */ | |
| 4193 { | |
| 4194 reg_match = rmp; | |
| 4195 reg_mmatch = NULL; | |
| 4196 reg_maxline = 0; | |
| 4197 reg_line_lbr = TRUE; | |
| 4198 reg_buf = curbuf; | |
| 4199 reg_win = NULL; | |
| 4200 ireg_ic = rmp->rm_ic; | |
| 4201 #ifdef FEAT_MBYTE | |
| 4202 ireg_icombine = FALSE; | |
| 4203 #endif | |
| 4204 ireg_maxcol = 0; | |
| 4205 return (nfa_regexec_both(line, col) != 0); | |
| 4206 } | |
| 4207 #endif | |
| 4208 | |
| 4209 | |
| 4210 /* | |
| 4211 * Match a regexp against multiple lines. | |
| 4212 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 4213 * Uses curbuf for line count and 'iskeyword'. | |
| 4214 * | |
| 4215 * Return zero if there is no match. Return number of lines contained in the | |
| 4216 * match otherwise. | |
| 4217 * | |
| 4218 * Note: the body is the same as bt_regexec() except for nfa_regexec_both() | |
| 4219 * | |
| 4220 * ! Also NOTE : match may actually be in another line. e.g.: | |
| 4221 * when r.e. is \nc, cursor is at 'a' and the text buffer looks like | |
| 4222 * | |
| 4223 * +-------------------------+ | |
| 4224 * |a | | |
| 4225 * |b | | |
| 4226 * |c | | |
| 4227 * | | | |
| 4228 * +-------------------------+ | |
| 4229 * | |
| 4230 * then nfa_regexec_multi() returns 3. while the original | |
| 4231 * vim_regexec_multi() returns 0 and a second call at line 2 will return 2. | |
| 4232 * | |
| 4233 * FIXME if this behavior is not compatible. | |
| 4234 */ | |
| 4235 static long | |
| 4236 nfa_regexec_multi(rmp, win, buf, lnum, col, tm) | |
| 4237 regmmatch_T *rmp; | |
| 4238 win_T *win; /* window in which to search or NULL */ | |
| 4239 buf_T *buf; /* buffer in which to search */ | |
| 4240 linenr_T lnum; /* nr of line to start looking for match */ | |
| 4241 colnr_T col; /* column to start looking for match */ | |
| 4242 proftime_T *tm UNUSED; /* timeout limit or NULL */ | |
| 4243 { | |
| 4244 reg_match = NULL; | |
| 4245 reg_mmatch = rmp; | |
| 4246 reg_buf = buf; | |
| 4247 reg_win = win; | |
| 4248 reg_firstlnum = lnum; | |
| 4249 reg_maxline = reg_buf->b_ml.ml_line_count - lnum; | |
| 4250 reg_line_lbr = FALSE; | |
| 4251 ireg_ic = rmp->rmm_ic; | |
| 4252 #ifdef FEAT_MBYTE | |
| 4253 ireg_icombine = FALSE; | |
| 4254 #endif | |
| 4255 ireg_maxcol = rmp->rmm_maxcol; | |
| 4256 | |
| 4503 | 4257 return nfa_regexec_both(NULL, col); |
| 4444 | 4258 } |
| 4259 | |
| 4260 #ifdef DEBUG | |
| 4261 # undef ENABLE_LOG | |
| 4262 #endif |