Mercurial > vim
annotate src/regexp_nfa.c @ 4615:5679b8ddd8cc v7.3.1055
updated for version 7.3.1055
Problem: Negated collection does not match newline.
Solution: Handle newline differently. (Hiroshi Shirosaki)
| author | Bram Moolenaar <bram@vim.org> |
|---|---|
| date | Thu, 30 May 2013 11:51:08 +0200 |
| parents | 321cfbef9431 |
| children | 857f6c53f117 |
| rev | line source |
|---|---|
| 4444 | 1 /* vi:set ts=8 sts=4 sw=4: |
| 2 * | |
| 3 * NFA regular expression implementation. | |
| 4 * | |
| 5 * This file is included in "regexp.c". | |
| 6 */ | |
| 7 | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
8 /* |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
9 * Logging of NFA engine. |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
10 * |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
11 * The NFA engine can write four log files: |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
12 * - Error log: Contains NFA engine's fatal errors. |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
13 * - Dump log: Contains compiled NFA state machine's information. |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
14 * - Run log: Contains information of matching procedure. |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
15 * - Debug log: Contains detailed information of matching procedure. Can be |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
16 * disabled by undefining NFA_REGEXP_DEBUG_LOG. |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
17 * The first one can also be used without debug mode. |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
18 * The last three are enabled when compiled as debug mode and individually |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
19 * disabled by commenting them out. |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
20 * The log files can get quite big! |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
21 * Do disable all of this when compiling Vim for debugging, undefine DEBUG in |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
22 * regexp.c |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
23 */ |
| 4444 | 24 #ifdef DEBUG |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
25 # define NFA_REGEXP_ERROR_LOG "nfa_regexp_error.log" |
| 4444 | 26 # define ENABLE_LOG |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
27 # define NFA_REGEXP_DUMP_LOG "nfa_regexp_dump.log" |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
28 # define NFA_REGEXP_RUN_LOG "nfa_regexp_run.log" |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
29 # define NFA_REGEXP_DEBUG_LOG "nfa_regexp_debug.log" |
| 4444 | 30 #endif |
| 31 | |
| 32 /* Upper limit allowed for {m,n} repetitions handled by NFA */ | |
| 33 #define NFA_BRACES_MAXLIMIT 10 | |
| 34 /* For allocating space for the postfix representation */ | |
| 35 #define NFA_POSTFIX_MULTIPLIER (NFA_BRACES_MAXLIMIT + 2)*2 | |
| 36 | |
| 37 enum | |
| 38 { | |
| 39 NFA_SPLIT = -1024, | |
| 40 NFA_MATCH, | |
| 41 NFA_SKIP_CHAR, /* matches a 0-length char */ | |
| 42 NFA_END_NEG_RANGE, /* Used when expanding [^ab] */ | |
| 43 | |
| 44 NFA_CONCAT, | |
| 45 NFA_OR, | |
| 46 NFA_STAR, | |
| 47 NFA_PLUS, | |
| 48 NFA_QUEST, | |
| 49 NFA_QUEST_NONGREEDY, /* Non-greedy version of \? */ | |
| 50 NFA_NOT, /* used for [^ab] negated char ranges */ | |
| 51 | |
| 52 NFA_BOL, /* ^ Begin line */ | |
| 53 NFA_EOL, /* $ End line */ | |
| 54 NFA_BOW, /* \< Begin word */ | |
| 55 NFA_EOW, /* \> End word */ | |
| 56 NFA_BOF, /* \%^ Begin file */ | |
| 57 NFA_EOF, /* \%$ End file */ | |
| 58 NFA_NEWL, | |
| 59 NFA_ZSTART, /* Used for \zs */ | |
| 60 NFA_ZEND, /* Used for \ze */ | |
| 61 NFA_NOPEN, /* Start of subexpression marked with \%( */ | |
| 62 NFA_NCLOSE, /* End of subexpr. marked with \%( ... \) */ | |
| 63 NFA_START_INVISIBLE, | |
| 64 NFA_END_INVISIBLE, | |
| 65 NFA_COMPOSING, /* Next nodes in NFA are part of the | |
| 66 composing multibyte char */ | |
| 67 NFA_END_COMPOSING, /* End of a composing char in the NFA */ | |
| 68 | |
| 69 /* The following are used only in the postfix form, not in the NFA */ | |
| 70 NFA_PREV_ATOM_NO_WIDTH, /* Used for \@= */ | |
| 71 NFA_PREV_ATOM_NO_WIDTH_NEG, /* Used for \@! */ | |
| 72 NFA_PREV_ATOM_JUST_BEFORE, /* Used for \@<= */ | |
| 73 NFA_PREV_ATOM_JUST_BEFORE_NEG, /* Used for \@<! */ | |
| 74 NFA_PREV_ATOM_LIKE_PATTERN, /* Used for \@> */ | |
| 75 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
76 NFA_BACKREF1, /* \1 */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
77 NFA_BACKREF2, /* \2 */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
78 NFA_BACKREF3, /* \3 */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
79 NFA_BACKREF4, /* \4 */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
80 NFA_BACKREF5, /* \5 */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
81 NFA_BACKREF6, /* \6 */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
82 NFA_BACKREF7, /* \7 */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
83 NFA_BACKREF8, /* \8 */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
84 NFA_BACKREF9, /* \9 */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
85 NFA_SKIP, /* Skip characters */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
86 |
| 4444 | 87 NFA_MOPEN, |
| 88 NFA_MCLOSE = NFA_MOPEN + NSUBEXP, | |
| 89 | |
| 90 /* NFA_FIRST_NL */ | |
| 91 NFA_ANY = NFA_MCLOSE + NSUBEXP, /* Match any one character. */ | |
| 92 NFA_ANYOF, /* Match any character in this string. */ | |
| 93 NFA_ANYBUT, /* Match any character not in this string. */ | |
| 94 NFA_IDENT, /* Match identifier char */ | |
| 95 NFA_SIDENT, /* Match identifier char but no digit */ | |
| 96 NFA_KWORD, /* Match keyword char */ | |
| 97 NFA_SKWORD, /* Match word char but no digit */ | |
| 98 NFA_FNAME, /* Match file name char */ | |
| 99 NFA_SFNAME, /* Match file name char but no digit */ | |
| 100 NFA_PRINT, /* Match printable char */ | |
| 101 NFA_SPRINT, /* Match printable char but no digit */ | |
| 102 NFA_WHITE, /* Match whitespace char */ | |
| 103 NFA_NWHITE, /* Match non-whitespace char */ | |
| 104 NFA_DIGIT, /* Match digit char */ | |
| 105 NFA_NDIGIT, /* Match non-digit char */ | |
| 106 NFA_HEX, /* Match hex char */ | |
| 107 NFA_NHEX, /* Match non-hex char */ | |
| 108 NFA_OCTAL, /* Match octal char */ | |
| 109 NFA_NOCTAL, /* Match non-octal char */ | |
| 110 NFA_WORD, /* Match word char */ | |
| 111 NFA_NWORD, /* Match non-word char */ | |
| 112 NFA_HEAD, /* Match head char */ | |
| 113 NFA_NHEAD, /* Match non-head char */ | |
| 114 NFA_ALPHA, /* Match alpha char */ | |
| 115 NFA_NALPHA, /* Match non-alpha char */ | |
| 116 NFA_LOWER, /* Match lowercase char */ | |
| 117 NFA_NLOWER, /* Match non-lowercase char */ | |
| 118 NFA_UPPER, /* Match uppercase char */ | |
| 119 NFA_NUPPER, /* Match non-uppercase char */ | |
|
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
120 |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
121 NFA_CURSOR, /* Match cursor pos */ |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
122 NFA_LNUM, /* Match line number */ |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
123 NFA_LNUM_GT, /* Match > line number */ |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
124 NFA_LNUM_LT, /* Match < line number */ |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
125 NFA_COL, /* Match cursor column */ |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
126 NFA_COL_GT, /* Match > cursor column */ |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
127 NFA_COL_LT, /* Match < cursor column */ |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
128 NFA_VCOL, /* Match cursor virtual column */ |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
129 NFA_VCOL_GT, /* Match > cursor virtual column */ |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
130 NFA_VCOL_LT, /* Match < cursor virtual column */ |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
131 |
| 4444 | 132 NFA_FIRST_NL = NFA_ANY + ADD_NL, |
| 133 NFA_LAST_NL = NFA_NUPPER + ADD_NL, | |
| 134 | |
| 135 /* Character classes [:alnum:] etc */ | |
| 136 NFA_CLASS_ALNUM, | |
| 137 NFA_CLASS_ALPHA, | |
| 138 NFA_CLASS_BLANK, | |
| 139 NFA_CLASS_CNTRL, | |
| 140 NFA_CLASS_DIGIT, | |
| 141 NFA_CLASS_GRAPH, | |
| 142 NFA_CLASS_LOWER, | |
| 143 NFA_CLASS_PRINT, | |
| 144 NFA_CLASS_PUNCT, | |
| 145 NFA_CLASS_SPACE, | |
| 146 NFA_CLASS_UPPER, | |
| 147 NFA_CLASS_XDIGIT, | |
| 148 NFA_CLASS_TAB, | |
| 149 NFA_CLASS_RETURN, | |
| 150 NFA_CLASS_BACKSPACE, | |
| 151 NFA_CLASS_ESCAPE | |
| 152 }; | |
| 153 | |
| 154 /* Keep in sync with classchars. */ | |
| 155 static int nfa_classcodes[] = { | |
| 156 NFA_ANY, NFA_IDENT, NFA_SIDENT, NFA_KWORD,NFA_SKWORD, | |
| 157 NFA_FNAME, NFA_SFNAME, NFA_PRINT, NFA_SPRINT, | |
| 158 NFA_WHITE, NFA_NWHITE, NFA_DIGIT, NFA_NDIGIT, | |
| 159 NFA_HEX, NFA_NHEX, NFA_OCTAL, NFA_NOCTAL, | |
| 160 NFA_WORD, NFA_NWORD, NFA_HEAD, NFA_NHEAD, | |
| 161 NFA_ALPHA, NFA_NALPHA, NFA_LOWER, NFA_NLOWER, | |
| 162 NFA_UPPER, NFA_NUPPER | |
| 163 }; | |
| 164 | |
| 165 static char_u e_misplaced[] = N_("E866: (NFA regexp) Misplaced %c"); | |
| 166 | |
| 167 /* | |
| 168 * NFA errors can be of 3 types: | |
| 169 * *** NFA runtime errors, when something unknown goes wrong. The NFA fails | |
| 170 * silently and revert the to backtracking engine. | |
| 171 * syntax_error = FALSE; | |
| 172 * *** Regexp syntax errors, when the input regexp is not syntactically correct. | |
| 173 * The NFA engine displays an error message, and nothing else happens. | |
| 174 * syntax_error = TRUE | |
| 175 * *** Unsupported features, when the input regexp uses an operator that is not | |
| 176 * implemented in the NFA. The NFA engine fails silently, and reverts to the | |
| 177 * old backtracking engine. | |
| 178 * syntax_error = FALSE | |
| 179 * "The NFA fails" means that "compiling the regexp with the NFA fails": | |
| 180 * nfa_regcomp() returns FAIL. | |
| 181 */ | |
| 182 static int syntax_error = FALSE; | |
| 183 | |
| 184 /* NFA regexp \ze operator encountered. */ | |
|
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
185 static int nfa_has_zend; |
| 4444 | 186 |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
187 /* 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
|
188 * the whole match (subexpr 0) is used. */ |
|
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
189 static int nfa_nsubexpr; |
|
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
190 |
| 4444 | 191 static int *post_start; /* holds the postfix form of r.e. */ |
| 192 static int *post_end; | |
| 193 static int *post_ptr; | |
| 194 | |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
195 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
|
196 * executing. */ |
| 4444 | 197 static int istate; /* Index in the state vector, used in new_state() */ |
| 198 | |
| 199 | |
| 200 static int nfa_regcomp_start __ARGS((char_u*expr, int re_flags)); | |
| 201 static int nfa_recognize_char_class __ARGS((char_u *start, char_u *end, int extra_newl)); | |
| 202 static int nfa_emit_equi_class __ARGS((int c, int neg)); | |
| 203 static int nfa_regatom __ARGS((void)); | |
| 204 static int nfa_regpiece __ARGS((void)); | |
| 205 static int nfa_regconcat __ARGS((void)); | |
| 206 static int nfa_regbranch __ARGS((void)); | |
| 207 static int nfa_reg __ARGS((int paren)); | |
| 208 #ifdef DEBUG | |
| 209 static void nfa_set_code __ARGS((int c)); | |
| 210 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
|
211 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
|
212 static void nfa_print_state2 __ARGS((FILE *debugf, nfa_state_T *state, garray_T *indent)); |
| 4444 | 213 static void nfa_dump __ARGS((nfa_regprog_T *prog)); |
| 214 #endif | |
| 215 static int *re2post __ARGS((void)); | |
| 216 static nfa_state_T *new_state __ARGS((int c, nfa_state_T *out, nfa_state_T *out1)); | |
| 217 static nfa_state_T *post2nfa __ARGS((int *postfix, int *end, int nfa_calc_size)); | |
| 218 static int check_char_class __ARGS((int class, int c)); | |
| 219 static void st_error __ARGS((int *postfix, int *end, int *p)); | |
|
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
220 static void nfa_set_neg_listids __ARGS((nfa_state_T *start)); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
221 static void nfa_set_null_listids __ARGS((nfa_state_T *start)); |
| 4444 | 222 static void nfa_save_listids __ARGS((nfa_state_T *start, int *list)); |
| 223 static void nfa_restore_listids __ARGS((nfa_state_T *start, int *list)); | |
|
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
224 static int nfa_re_num_cmp __ARGS((long_u val, int op, long_u pos)); |
| 4444 | 225 static long nfa_regtry __ARGS((nfa_state_T *start, colnr_T col)); |
| 226 static long nfa_regexec_both __ARGS((char_u *line, colnr_T col)); | |
| 227 static regprog_T *nfa_regcomp __ARGS((char_u *expr, int re_flags)); | |
| 228 static int nfa_regexec __ARGS((regmatch_T *rmp, char_u *line, colnr_T col)); | |
| 229 static long nfa_regexec_multi __ARGS((regmmatch_T *rmp, win_T *win, buf_T *buf, linenr_T lnum, colnr_T col, proftime_T *tm)); | |
| 230 | |
| 231 /* helper functions used when doing re2post() ... regatom() parsing */ | |
| 232 #define EMIT(c) do { \ | |
| 233 if (post_ptr >= post_end) \ | |
| 234 return FAIL; \ | |
| 235 *post_ptr++ = c; \ | |
| 236 } while (0) | |
| 237 | |
| 238 /* | |
| 239 * Initialize internal variables before NFA compilation. | |
| 240 * Return OK on success, FAIL otherwise. | |
| 241 */ | |
| 242 static int | |
| 243 nfa_regcomp_start(expr, re_flags) | |
| 244 char_u *expr; | |
| 245 int re_flags; /* see vim_regcomp() */ | |
| 246 { | |
| 4458 | 247 size_t postfix_size; |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
248 int nstate_max; |
| 4444 | 249 |
| 250 nstate = 0; | |
| 251 istate = 0; | |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
252 /* A reasonable estimation for maximum size */ |
| 4458 | 253 nstate_max = (int)(STRLEN(expr) + 1) * NFA_POSTFIX_MULTIPLIER; |
| 4444 | 254 |
| 4454 | 255 /* Some items blow up in size, such as [A-z]. Add more space for that. |
| 256 * TODO: some patterns may still fail. */ | |
| 4458 | 257 nstate_max += 1000; |
| 4454 | 258 |
| 259 /* Size for postfix representation of expr. */ | |
| 4444 | 260 postfix_size = sizeof(*post_start) * nstate_max; |
| 4454 | 261 |
| 4444 | 262 post_start = (int *)lalloc(postfix_size, TRUE); |
| 263 if (post_start == NULL) | |
| 264 return FAIL; | |
| 265 vim_memset(post_start, 0, postfix_size); | |
| 266 post_ptr = post_start; | |
| 4454 | 267 post_end = post_start + nstate_max; |
| 4444 | 268 nfa_has_zend = FALSE; |
| 269 | |
| 270 regcomp_start(expr, re_flags); | |
| 271 | |
| 272 return OK; | |
| 273 } | |
| 274 | |
| 275 /* | |
| 276 * Search between "start" and "end" and try to recognize a | |
| 277 * character class in expanded form. For example [0-9]. | |
| 278 * On success, return the id the character class to be emitted. | |
| 279 * On failure, return 0 (=FAIL) | |
| 280 * Start points to the first char of the range, while end should point | |
| 281 * to the closing brace. | |
| 282 */ | |
| 283 static int | |
| 284 nfa_recognize_char_class(start, end, extra_newl) | |
| 285 char_u *start; | |
| 286 char_u *end; | |
| 287 int extra_newl; | |
| 288 { | |
| 289 int i; | |
| 290 /* Each of these variables takes up a char in "config[]", | |
| 291 * in the order they are here. */ | |
| 292 int not = FALSE, af = FALSE, AF = FALSE, az = FALSE, AZ = FALSE, | |
| 293 o7 = FALSE, o9 = FALSE, underscore = FALSE, newl = FALSE; | |
| 294 char_u *p; | |
| 295 #define NCONFIGS 16 | |
| 296 int classid[NCONFIGS] = { | |
| 297 NFA_DIGIT, NFA_NDIGIT, NFA_HEX, NFA_NHEX, | |
| 298 NFA_OCTAL, NFA_NOCTAL, NFA_WORD, NFA_NWORD, | |
| 299 NFA_HEAD, NFA_NHEAD, NFA_ALPHA, NFA_NALPHA, | |
| 300 NFA_LOWER, NFA_NLOWER, NFA_UPPER, NFA_NUPPER | |
| 301 }; | |
| 4450 | 302 char_u myconfig[10]; |
| 4444 | 303 char_u config[NCONFIGS][9] = { |
| 304 "000000100", /* digit */ | |
| 305 "100000100", /* non digit */ | |
| 306 "011000100", /* hex-digit */ | |
| 307 "111000100", /* non hex-digit */ | |
| 308 "000001000", /* octal-digit */ | |
| 309 "100001000", /* [^0-7] */ | |
| 310 "000110110", /* [0-9A-Za-z_] */ | |
| 311 "100110110", /* [^0-9A-Za-z_] */ | |
| 312 "000110010", /* head of word */ | |
| 313 "100110010", /* not head of word */ | |
| 314 "000110000", /* alphabetic char a-z */ | |
| 315 "100110000", /* non alphabetic char */ | |
| 316 "000100000", /* lowercase letter */ | |
| 317 "100100000", /* non lowercase */ | |
| 318 "000010000", /* uppercase */ | |
| 319 "100010000" /* non uppercase */ | |
| 320 }; | |
| 321 | |
| 322 if (extra_newl == TRUE) | |
| 323 newl = TRUE; | |
| 324 | |
| 325 if (*end != ']') | |
| 326 return FAIL; | |
| 327 p = start; | |
| 328 if (*p == '^') | |
| 329 { | |
| 330 not = TRUE; | |
| 331 p ++; | |
| 332 } | |
| 333 | |
| 334 while (p < end) | |
| 335 { | |
| 336 if (p + 2 < end && *(p + 1) == '-') | |
| 337 { | |
| 338 switch (*p) | |
| 339 { | |
| 340 case '0': | |
| 341 if (*(p + 2) == '9') | |
| 342 { | |
| 343 o9 = TRUE; | |
| 344 break; | |
| 345 } | |
| 346 else | |
| 347 if (*(p + 2) == '7') | |
| 348 { | |
| 349 o7 = TRUE; | |
| 350 break; | |
| 351 } | |
| 352 case 'a': | |
| 353 if (*(p + 2) == 'z') | |
| 354 { | |
| 355 az = TRUE; | |
| 356 break; | |
| 357 } | |
| 358 else | |
| 359 if (*(p + 2) == 'f') | |
| 360 { | |
| 361 af = TRUE; | |
| 362 break; | |
| 363 } | |
| 364 case 'A': | |
| 365 if (*(p + 2) == 'Z') | |
| 366 { | |
| 367 AZ = TRUE; | |
| 368 break; | |
| 369 } | |
| 370 else | |
| 371 if (*(p + 2) == 'F') | |
| 372 { | |
| 373 AF = TRUE; | |
| 374 break; | |
| 375 } | |
| 376 /* FALLTHROUGH */ | |
| 377 default: | |
| 378 return FAIL; | |
| 379 } | |
| 380 p += 3; | |
| 381 } | |
| 382 else if (p + 1 < end && *p == '\\' && *(p + 1) == 'n') | |
| 383 { | |
| 384 newl = TRUE; | |
| 385 p += 2; | |
| 386 } | |
| 387 else if (*p == '_') | |
| 388 { | |
| 389 underscore = TRUE; | |
| 390 p ++; | |
| 391 } | |
| 392 else if (*p == '\n') | |
| 393 { | |
| 394 newl = TRUE; | |
| 395 p ++; | |
| 396 } | |
| 397 else | |
| 398 return FAIL; | |
| 399 } /* while (p < end) */ | |
| 400 | |
| 401 if (p != end) | |
| 402 return FAIL; | |
| 403 | |
| 404 /* build the config that represents the ranges we gathered */ | |
| 405 STRCPY(myconfig, "000000000"); | |
| 406 if (not == TRUE) | |
| 407 myconfig[0] = '1'; | |
| 408 if (af == TRUE) | |
| 409 myconfig[1] = '1'; | |
| 410 if (AF == TRUE) | |
| 411 myconfig[2] = '1'; | |
| 412 if (az == TRUE) | |
| 413 myconfig[3] = '1'; | |
| 414 if (AZ == TRUE) | |
| 415 myconfig[4] = '1'; | |
| 416 if (o7 == TRUE) | |
| 417 myconfig[5] = '1'; | |
| 418 if (o9 == TRUE) | |
| 419 myconfig[6] = '1'; | |
| 420 if (underscore == TRUE) | |
| 421 myconfig[7] = '1'; | |
| 422 if (newl == TRUE) | |
| 423 { | |
| 424 myconfig[8] = '1'; | |
| 425 extra_newl = ADD_NL; | |
| 426 } | |
| 427 /* try to recognize character classes */ | |
| 428 for (i = 0; i < NCONFIGS; i++) | |
| 4450 | 429 if (STRNCMP(myconfig, config[i], 8) == 0) |
| 4444 | 430 return classid[i] + extra_newl; |
| 431 | |
| 432 /* fallthrough => no success so far */ | |
| 433 return FAIL; | |
| 434 | |
| 435 #undef NCONFIGS | |
| 436 } | |
| 437 | |
| 438 /* | |
| 439 * Produce the bytes for equivalence class "c". | |
| 440 * Currently only handles latin1, latin9 and utf-8. | |
| 441 * Emits bytes in postfix notation: 'a,b,NFA_OR,c,NFA_OR' is | |
| 442 * equivalent to 'a OR b OR c' | |
| 443 * | |
| 444 * NOTE! When changing this function, also update reg_equi_class() | |
| 445 */ | |
| 446 static int | |
| 447 nfa_emit_equi_class(c, neg) | |
| 448 int c; | |
| 449 int neg; | |
| 450 { | |
| 451 int first = TRUE; | |
| 452 int glue = neg == TRUE ? NFA_CONCAT : NFA_OR; | |
| 453 #define EMIT2(c) \ | |
| 454 EMIT(c); \ | |
| 455 if (neg == TRUE) { \ | |
| 456 EMIT(NFA_NOT); \ | |
| 457 } \ | |
| 458 if (first == FALSE) \ | |
| 459 EMIT(glue); \ | |
| 460 else \ | |
| 461 first = FALSE; \ | |
| 462 | |
| 463 #ifdef FEAT_MBYTE | |
| 464 if (enc_utf8 || STRCMP(p_enc, "latin1") == 0 | |
| 465 || STRCMP(p_enc, "iso-8859-15") == 0) | |
| 466 #endif | |
| 467 { | |
| 468 switch (c) | |
| 469 { | |
| 470 case 'A': case '\300': case '\301': case '\302': | |
| 471 case '\303': case '\304': case '\305': | |
| 472 EMIT2('A'); EMIT2('\300'); EMIT2('\301'); | |
| 473 EMIT2('\302'); EMIT2('\303'); EMIT2('\304'); | |
| 474 EMIT2('\305'); | |
| 475 return OK; | |
| 476 | |
| 477 case 'C': case '\307': | |
| 478 EMIT2('C'); EMIT2('\307'); | |
| 479 return OK; | |
| 480 | |
| 481 case 'E': case '\310': case '\311': case '\312': case '\313': | |
| 482 EMIT2('E'); EMIT2('\310'); EMIT2('\311'); | |
| 483 EMIT2('\312'); EMIT2('\313'); | |
| 484 return OK; | |
| 485 | |
| 486 case 'I': case '\314': case '\315': case '\316': case '\317': | |
| 487 EMIT2('I'); EMIT2('\314'); EMIT2('\315'); | |
| 488 EMIT2('\316'); EMIT2('\317'); | |
| 489 return OK; | |
| 490 | |
| 491 case 'N': case '\321': | |
| 492 EMIT2('N'); EMIT2('\321'); | |
| 493 return OK; | |
| 494 | |
| 495 case 'O': case '\322': case '\323': case '\324': case '\325': | |
| 496 case '\326': | |
| 497 EMIT2('O'); EMIT2('\322'); EMIT2('\323'); | |
| 498 EMIT2('\324'); EMIT2('\325'); EMIT2('\326'); | |
| 499 return OK; | |
| 500 | |
| 501 case 'U': case '\331': case '\332': case '\333': case '\334': | |
| 502 EMIT2('U'); EMIT2('\331'); EMIT2('\332'); | |
| 503 EMIT2('\333'); EMIT2('\334'); | |
| 504 return OK; | |
| 505 | |
| 506 case 'Y': case '\335': | |
| 507 EMIT2('Y'); EMIT2('\335'); | |
| 508 return OK; | |
| 509 | |
| 510 case 'a': case '\340': case '\341': case '\342': | |
| 511 case '\343': case '\344': case '\345': | |
| 512 EMIT2('a'); EMIT2('\340'); EMIT2('\341'); | |
| 513 EMIT2('\342'); EMIT2('\343'); EMIT2('\344'); | |
| 514 EMIT2('\345'); | |
| 515 return OK; | |
| 516 | |
| 517 case 'c': case '\347': | |
| 518 EMIT2('c'); EMIT2('\347'); | |
| 519 return OK; | |
| 520 | |
| 521 case 'e': case '\350': case '\351': case '\352': case '\353': | |
| 522 EMIT2('e'); EMIT2('\350'); EMIT2('\351'); | |
| 523 EMIT2('\352'); EMIT2('\353'); | |
| 524 return OK; | |
| 525 | |
| 526 case 'i': case '\354': case '\355': case '\356': case '\357': | |
| 527 EMIT2('i'); EMIT2('\354'); EMIT2('\355'); | |
| 528 EMIT2('\356'); EMIT2('\357'); | |
| 529 return OK; | |
| 530 | |
| 531 case 'n': case '\361': | |
| 532 EMIT2('n'); EMIT2('\361'); | |
| 533 return OK; | |
| 534 | |
| 535 case 'o': case '\362': case '\363': case '\364': case '\365': | |
| 536 case '\366': | |
| 537 EMIT2('o'); EMIT2('\362'); EMIT2('\363'); | |
| 538 EMIT2('\364'); EMIT2('\365'); EMIT2('\366'); | |
| 539 return OK; | |
| 540 | |
| 541 case 'u': case '\371': case '\372': case '\373': case '\374': | |
| 542 EMIT2('u'); EMIT2('\371'); EMIT2('\372'); | |
| 543 EMIT2('\373'); EMIT2('\374'); | |
| 544 return OK; | |
| 545 | |
| 546 case 'y': case '\375': case '\377': | |
| 547 EMIT2('y'); EMIT2('\375'); EMIT2('\377'); | |
| 548 return OK; | |
| 549 | |
| 550 default: | |
| 551 return FAIL; | |
| 552 } | |
| 553 } | |
| 554 | |
| 555 EMIT(c); | |
| 556 return OK; | |
| 557 #undef EMIT2 | |
| 558 } | |
| 559 | |
| 560 /* | |
| 561 * Code to parse regular expression. | |
| 562 * | |
| 563 * We try to reuse parsing functions in regexp.c to | |
| 564 * minimize surprise and keep the syntax consistent. | |
| 565 */ | |
| 566 | |
| 567 /* | |
| 568 * Parse the lowest level. | |
| 569 * | |
| 570 * An atom can be one of a long list of items. Many atoms match one character | |
| 571 * in the text. It is often an ordinary character or a character class. | |
| 572 * Braces can be used to make a pattern into an atom. The "\z(\)" construct | |
| 573 * is only for syntax highlighting. | |
| 574 * | |
| 575 * atom ::= ordinary-atom | |
| 576 * or \( pattern \) | |
| 577 * or \%( pattern \) | |
| 578 * or \z( pattern \) | |
| 579 */ | |
| 580 static int | |
| 581 nfa_regatom() | |
| 582 { | |
| 583 int c; | |
| 584 int charclass; | |
| 585 int equiclass; | |
| 586 int collclass; | |
| 587 int got_coll_char; | |
| 588 char_u *p; | |
| 589 char_u *endp; | |
| 590 #ifdef FEAT_MBYTE | |
| 591 char_u *old_regparse = regparse; | |
| 592 #endif | |
| 593 int extra = 0; | |
| 594 int first; | |
| 595 int emit_range; | |
| 596 int negated; | |
| 597 int result; | |
| 598 int startc = -1; | |
| 599 int endc = -1; | |
| 600 int oldstartc = -1; | |
| 601 int cpo_lit; /* 'cpoptions' contains 'l' flag */ | |
| 602 int cpo_bsl; /* 'cpoptions' contains '\' flag */ | |
| 603 int glue; /* ID that will "glue" nodes together */ | |
| 604 | |
| 605 cpo_lit = vim_strchr(p_cpo, CPO_LITERAL) != NULL; | |
| 606 cpo_bsl = vim_strchr(p_cpo, CPO_BACKSL) != NULL; | |
| 607 | |
| 608 c = getchr(); | |
| 609 switch (c) | |
| 610 { | |
|
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
611 case NUL: |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
612 syntax_error = TRUE; |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
613 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
|
614 |
| 4444 | 615 case Magic('^'): |
| 616 EMIT(NFA_BOL); | |
| 617 break; | |
| 618 | |
| 619 case Magic('$'): | |
| 620 EMIT(NFA_EOL); | |
| 621 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 622 had_eol = TRUE; | |
| 623 #endif | |
| 624 break; | |
| 625 | |
| 626 case Magic('<'): | |
| 627 EMIT(NFA_BOW); | |
| 628 break; | |
| 629 | |
| 630 case Magic('>'): | |
| 631 EMIT(NFA_EOW); | |
| 632 break; | |
| 633 | |
| 634 case Magic('_'): | |
| 635 c = no_Magic(getchr()); | |
| 636 if (c == '^') /* "\_^" is start-of-line */ | |
| 637 { | |
| 638 EMIT(NFA_BOL); | |
| 639 break; | |
| 640 } | |
| 641 if (c == '$') /* "\_$" is end-of-line */ | |
| 642 { | |
| 643 EMIT(NFA_EOL); | |
| 644 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 645 had_eol = TRUE; | |
| 646 #endif | |
| 647 break; | |
| 648 } | |
| 649 | |
| 650 extra = ADD_NL; | |
| 651 | |
| 652 /* "\_[" is collection plus newline */ | |
| 653 if (c == '[') | |
|
4517
9a2183bd8295
updated for version 7.3.1006
Bram Moolenaar <bram@vim.org>
parents:
4515
diff
changeset
|
654 goto collection; |
| 4444 | 655 |
| 656 /* "\_x" is character class plus newline */ | |
| 657 /*FALLTHROUGH*/ | |
| 658 | |
| 659 /* | |
| 660 * Character classes. | |
| 661 */ | |
| 662 case Magic('.'): | |
| 663 case Magic('i'): | |
| 664 case Magic('I'): | |
| 665 case Magic('k'): | |
| 666 case Magic('K'): | |
| 667 case Magic('f'): | |
| 668 case Magic('F'): | |
| 669 case Magic('p'): | |
| 670 case Magic('P'): | |
| 671 case Magic('s'): | |
| 672 case Magic('S'): | |
| 673 case Magic('d'): | |
| 674 case Magic('D'): | |
| 675 case Magic('x'): | |
| 676 case Magic('X'): | |
| 677 case Magic('o'): | |
| 678 case Magic('O'): | |
| 679 case Magic('w'): | |
| 680 case Magic('W'): | |
| 681 case Magic('h'): | |
| 682 case Magic('H'): | |
| 683 case Magic('a'): | |
| 684 case Magic('A'): | |
| 685 case Magic('l'): | |
| 686 case Magic('L'): | |
| 687 case Magic('u'): | |
| 688 case Magic('U'): | |
| 689 p = vim_strchr(classchars, no_Magic(c)); | |
| 690 if (p == NULL) | |
| 691 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
692 EMSGN("INTERNAL: Unknown character class char: %ld", c); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
693 return FAIL; |
| 4444 | 694 } |
| 695 #ifdef FEAT_MBYTE | |
| 696 /* When '.' is followed by a composing char ignore the dot, so that | |
| 697 * the composing char is matched here. */ | |
| 698 if (enc_utf8 && c == Magic('.') && utf_iscomposing(peekchr())) | |
| 699 { | |
|
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
700 old_regparse = regparse; |
| 4444 | 701 c = getchr(); |
| 702 goto nfa_do_multibyte; | |
| 703 } | |
| 704 #endif | |
| 705 EMIT(nfa_classcodes[p - classchars]); | |
| 706 if (extra == ADD_NL) | |
| 707 { | |
| 708 EMIT(NFA_NEWL); | |
| 709 EMIT(NFA_OR); | |
| 710 regflags |= RF_HASNL; | |
| 711 } | |
| 712 break; | |
| 713 | |
| 714 case Magic('n'): | |
| 715 if (reg_string) | |
| 716 /* In a string "\n" matches a newline character. */ | |
| 717 EMIT(NL); | |
| 718 else | |
| 719 { | |
| 720 /* In buffer text "\n" matches the end of a line. */ | |
| 721 EMIT(NFA_NEWL); | |
| 722 regflags |= RF_HASNL; | |
| 723 } | |
| 724 break; | |
| 725 | |
| 726 case Magic('('): | |
| 727 if (nfa_reg(REG_PAREN) == FAIL) | |
| 728 return FAIL; /* cascaded error */ | |
| 729 break; | |
| 730 | |
| 731 case Magic('|'): | |
| 732 case Magic('&'): | |
| 733 case Magic(')'): | |
| 734 syntax_error = TRUE; | |
| 4450 | 735 EMSGN(_(e_misplaced), no_Magic(c)); |
| 4444 | 736 return FAIL; |
| 737 | |
| 738 case Magic('='): | |
| 739 case Magic('?'): | |
| 740 case Magic('+'): | |
| 741 case Magic('@'): | |
| 742 case Magic('*'): | |
| 743 case Magic('{'): | |
| 744 /* these should follow an atom, not form an atom */ | |
| 745 syntax_error = TRUE; | |
| 4450 | 746 EMSGN(_(e_misplaced), no_Magic(c)); |
| 4444 | 747 return FAIL; |
| 748 | |
| 749 case Magic('~'): /* previous substitute pattern */ | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
750 /* TODO: Not supported yet */ |
| 4444 | 751 return FAIL; |
| 752 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
753 case Magic('1'): EMIT(NFA_BACKREF1); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
754 case Magic('2'): EMIT(NFA_BACKREF2); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
755 case Magic('3'): EMIT(NFA_BACKREF3); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
756 case Magic('4'): EMIT(NFA_BACKREF4); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
757 case Magic('5'): EMIT(NFA_BACKREF5); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
758 case Magic('6'): EMIT(NFA_BACKREF6); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
759 case Magic('7'): EMIT(NFA_BACKREF7); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
760 case Magic('8'): EMIT(NFA_BACKREF8); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
761 case Magic('9'): EMIT(NFA_BACKREF9); break; |
| 4444 | 762 |
| 763 case Magic('z'): | |
| 764 c = no_Magic(getchr()); | |
| 765 switch (c) | |
| 766 { | |
| 767 case 's': | |
| 768 EMIT(NFA_ZSTART); | |
| 769 break; | |
| 770 case 'e': | |
| 771 EMIT(NFA_ZEND); | |
| 772 nfa_has_zend = TRUE; | |
|
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
773 break; |
| 4444 | 774 case '1': |
| 775 case '2': | |
| 776 case '3': | |
| 777 case '4': | |
| 778 case '5': | |
| 779 case '6': | |
| 780 case '7': | |
| 781 case '8': | |
| 782 case '9': | |
| 783 case '(': | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
784 /* TODO: \z1...\z9 and \z( not yet supported */ |
| 4444 | 785 return FAIL; |
| 786 default: | |
| 787 syntax_error = TRUE; | |
| 4450 | 788 EMSGN(_("E867: (NFA) Unknown operator '\\z%c'"), |
| 4444 | 789 no_Magic(c)); |
| 790 return FAIL; | |
| 791 } | |
| 792 break; | |
| 793 | |
| 794 case Magic('%'): | |
| 795 c = no_Magic(getchr()); | |
| 796 switch (c) | |
| 797 { | |
| 798 /* () without a back reference */ | |
| 799 case '(': | |
| 800 if (nfa_reg(REG_NPAREN) == FAIL) | |
| 801 return FAIL; | |
| 802 EMIT(NFA_NOPEN); | |
| 803 break; | |
| 804 | |
| 805 case 'd': /* %d123 decimal */ | |
| 806 case 'o': /* %o123 octal */ | |
| 807 case 'x': /* %xab hex 2 */ | |
| 808 case 'u': /* %uabcd hex 4 */ | |
| 809 case 'U': /* %U1234abcd hex 8 */ | |
|
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
810 { |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
811 int nr; |
|
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
812 |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
813 switch (c) |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
814 { |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
815 case 'd': nr = getdecchrs(); break; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
816 case 'o': nr = getoctchrs(); break; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
817 case 'x': nr = gethexchrs(2); break; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
818 case 'u': nr = gethexchrs(4); break; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
819 case 'U': nr = gethexchrs(8); break; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
820 default: nr = -1; break; |
|
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
821 } |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
822 |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
823 if (nr < 0) |
|
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
824 EMSG2_RET_FAIL( |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
825 _("E678: Invalid character after %s%%[dxouU]"), |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
826 reg_magic == MAGIC_ALL); |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
827 /* TODO: what if a composing character follows? */ |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
828 EMIT(nr); |
|
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
829 } |
| 4444 | 830 break; |
| 831 | |
| 832 /* Catch \%^ and \%$ regardless of where they appear in the | |
| 833 * pattern -- regardless of whether or not it makes sense. */ | |
| 834 case '^': | |
| 835 EMIT(NFA_BOF); | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
836 /* TODO: Not yet supported */ |
| 4444 | 837 return FAIL; |
| 838 break; | |
| 839 | |
| 840 case '$': | |
| 841 EMIT(NFA_EOF); | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
842 /* TODO: Not yet supported */ |
| 4444 | 843 return FAIL; |
| 844 break; | |
| 845 | |
| 846 case '#': | |
|
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
847 EMIT(NFA_CURSOR); |
| 4444 | 848 break; |
| 849 | |
| 850 case 'V': | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
851 /* TODO: not supported yet */ |
| 4444 | 852 return FAIL; |
| 853 break; | |
| 854 | |
| 855 case '[': | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
856 /* TODO: \%[abc] not supported yet */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
857 return FAIL; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
858 |
| 4444 | 859 default: |
|
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
860 { |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
861 long_u n = 0; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
862 int cmp = c; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
863 |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
864 if (c == '<' || c == '>') |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
865 c = getchr(); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
866 while (VIM_ISDIGIT(c)) |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
867 { |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
868 n = n * 10 + (c - '0'); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
869 c = getchr(); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
870 } |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
871 if (c == 'l' || c == 'c' || c == 'v') |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
872 { |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
873 EMIT(n); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
874 if (c == 'l') |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
875 EMIT(cmp == '<' ? NFA_LNUM_LT : |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
876 cmp == '>' ? NFA_LNUM_GT : NFA_LNUM); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
877 else if (c == 'c') |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
878 EMIT(cmp == '<' ? NFA_COL_LT : |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
879 cmp == '>' ? NFA_COL_GT : NFA_COL); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
880 else |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
881 EMIT(cmp == '<' ? NFA_VCOL_LT : |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
882 cmp == '>' ? NFA_VCOL_GT : NFA_VCOL); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
883 break; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
884 } |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
885 else if (c == '\'') |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
886 /* TODO: \%'m not supported yet */ |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
887 return FAIL; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
888 } |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
889 syntax_error = TRUE; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
890 EMSGN(_("E867: (NFA) Unknown operator '\\%%%c'"), |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
891 no_Magic(c)); |
| 4444 | 892 return FAIL; |
| 893 } | |
| 894 break; | |
| 895 | |
| 896 case Magic('['): | |
|
4517
9a2183bd8295
updated for version 7.3.1006
Bram Moolenaar <bram@vim.org>
parents:
4515
diff
changeset
|
897 collection: |
| 4444 | 898 /* |
| 899 * Glue is emitted between several atoms from the []. | |
| 900 * It is either NFA_OR, or NFA_CONCAT. | |
| 901 * | |
| 902 * [abc] expands to 'a b NFA_OR c NFA_OR' (in postfix notation) | |
| 903 * [^abc] expands to 'a NFA_NOT b NFA_NOT NFA_CONCAT c NFA_NOT | |
| 904 * NFA_CONCAT NFA_END_NEG_RANGE NFA_CONCAT' (in postfix | |
| 905 * notation) | |
| 906 * | |
| 907 */ | |
| 908 | |
| 909 | |
| 910 /* Emit negation atoms, if needed. | |
| 911 * The CONCAT below merges the NOT with the previous node. */ | |
| 912 #define TRY_NEG() \ | |
| 913 if (negated == TRUE) \ | |
| 914 { \ | |
| 915 EMIT(NFA_NOT); \ | |
| 916 } | |
| 917 | |
| 918 /* Emit glue between important nodes : CONCAT or OR. */ | |
| 919 #define EMIT_GLUE() \ | |
| 920 if (first == FALSE) \ | |
| 921 EMIT(glue); \ | |
| 922 else \ | |
| 923 first = FALSE; | |
| 924 | |
| 925 p = regparse; | |
| 926 endp = skip_anyof(p); | |
| 927 if (*endp == ']') | |
| 928 { | |
| 929 /* | |
| 930 * Try to reverse engineer character classes. For example, | |
| 931 * recognize that [0-9] stands for \d and [A-Za-z_] with \h, | |
| 932 * and perform the necessary substitutions in the NFA. | |
| 933 */ | |
| 934 result = nfa_recognize_char_class(regparse, endp, | |
| 935 extra == ADD_NL); | |
| 936 if (result != FAIL) | |
| 937 { | |
| 938 if (result >= NFA_DIGIT && result <= NFA_NUPPER) | |
| 939 EMIT(result); | |
| 940 else /* must be char class + newline */ | |
| 941 { | |
| 942 EMIT(result - ADD_NL); | |
| 943 EMIT(NFA_NEWL); | |
| 944 EMIT(NFA_OR); | |
| 945 } | |
| 946 regparse = endp; | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
947 mb_ptr_adv(regparse); |
| 4444 | 948 return OK; |
| 949 } | |
| 950 /* | |
| 951 * Failed to recognize a character class. Use the simple | |
| 952 * version that turns [abc] into 'a' OR 'b' OR 'c' | |
| 953 */ | |
| 954 startc = endc = oldstartc = -1; | |
| 955 first = TRUE; /* Emitting first atom in this sequence? */ | |
| 956 negated = FALSE; | |
| 957 glue = NFA_OR; | |
| 958 if (*regparse == '^') /* negated range */ | |
| 959 { | |
| 960 negated = TRUE; | |
| 961 glue = NFA_CONCAT; | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
962 mb_ptr_adv(regparse); |
| 4444 | 963 } |
| 964 if (*regparse == '-') | |
| 965 { | |
| 966 startc = '-'; | |
| 967 EMIT(startc); | |
| 968 TRY_NEG(); | |
| 969 EMIT_GLUE(); | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
970 mb_ptr_adv(regparse); |
| 4444 | 971 } |
| 972 /* Emit the OR branches for each character in the [] */ | |
| 973 emit_range = FALSE; | |
| 974 while (regparse < endp) | |
| 975 { | |
| 976 oldstartc = startc; | |
| 977 startc = -1; | |
| 978 got_coll_char = FALSE; | |
| 979 if (*regparse == '[') | |
| 980 { | |
| 981 /* Check for [: :], [= =], [. .] */ | |
| 982 equiclass = collclass = 0; | |
| 983 charclass = get_char_class(®parse); | |
| 984 if (charclass == CLASS_NONE) | |
| 985 { | |
| 986 equiclass = get_equi_class(®parse); | |
| 987 if (equiclass == 0) | |
| 988 collclass = get_coll_element(®parse); | |
| 989 } | |
| 990 | |
| 991 /* Character class like [:alpha:] */ | |
| 992 if (charclass != CLASS_NONE) | |
| 993 { | |
| 994 switch (charclass) | |
| 995 { | |
| 996 case CLASS_ALNUM: | |
| 997 EMIT(NFA_CLASS_ALNUM); | |
| 998 break; | |
| 999 case CLASS_ALPHA: | |
| 1000 EMIT(NFA_CLASS_ALPHA); | |
| 1001 break; | |
| 1002 case CLASS_BLANK: | |
| 1003 EMIT(NFA_CLASS_BLANK); | |
| 1004 break; | |
| 1005 case CLASS_CNTRL: | |
| 1006 EMIT(NFA_CLASS_CNTRL); | |
| 1007 break; | |
| 1008 case CLASS_DIGIT: | |
| 1009 EMIT(NFA_CLASS_DIGIT); | |
| 1010 break; | |
| 1011 case CLASS_GRAPH: | |
| 1012 EMIT(NFA_CLASS_GRAPH); | |
| 1013 break; | |
| 1014 case CLASS_LOWER: | |
| 1015 EMIT(NFA_CLASS_LOWER); | |
| 1016 break; | |
| 1017 case CLASS_PRINT: | |
| 1018 EMIT(NFA_CLASS_PRINT); | |
| 1019 break; | |
| 1020 case CLASS_PUNCT: | |
| 1021 EMIT(NFA_CLASS_PUNCT); | |
| 1022 break; | |
| 1023 case CLASS_SPACE: | |
| 1024 EMIT(NFA_CLASS_SPACE); | |
| 1025 break; | |
| 1026 case CLASS_UPPER: | |
| 1027 EMIT(NFA_CLASS_UPPER); | |
| 1028 break; | |
| 1029 case CLASS_XDIGIT: | |
| 1030 EMIT(NFA_CLASS_XDIGIT); | |
| 1031 break; | |
| 1032 case CLASS_TAB: | |
| 1033 EMIT(NFA_CLASS_TAB); | |
| 1034 break; | |
| 1035 case CLASS_RETURN: | |
| 1036 EMIT(NFA_CLASS_RETURN); | |
| 1037 break; | |
| 1038 case CLASS_BACKSPACE: | |
| 1039 EMIT(NFA_CLASS_BACKSPACE); | |
| 1040 break; | |
| 1041 case CLASS_ESCAPE: | |
| 1042 EMIT(NFA_CLASS_ESCAPE); | |
| 1043 break; | |
| 1044 } | |
| 1045 TRY_NEG(); | |
| 1046 EMIT_GLUE(); | |
| 1047 continue; | |
| 1048 } | |
| 1049 /* Try equivalence class [=a=] and the like */ | |
| 1050 if (equiclass != 0) | |
| 1051 { | |
| 1052 result = nfa_emit_equi_class(equiclass, negated); | |
| 1053 if (result == FAIL) | |
| 1054 { | |
| 1055 /* should never happen */ | |
| 1056 EMSG_RET_FAIL(_("E868: Error building NFA with equivalence class!")); | |
| 1057 } | |
| 1058 EMIT_GLUE(); | |
| 1059 continue; | |
| 1060 } | |
| 1061 /* Try collating class like [. .] */ | |
| 1062 if (collclass != 0) | |
| 1063 { | |
| 1064 startc = collclass; /* allow [.a.]-x as a range */ | |
| 1065 /* Will emit the proper atom at the end of the | |
| 1066 * while loop. */ | |
| 1067 } | |
| 1068 } | |
| 1069 /* Try a range like 'a-x' or '\t-z' */ | |
| 1070 if (*regparse == '-') | |
| 1071 { | |
| 1072 emit_range = TRUE; | |
| 1073 startc = oldstartc; | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1074 mb_ptr_adv(regparse); |
| 4444 | 1075 continue; /* reading the end of the range */ |
| 1076 } | |
| 1077 | |
| 1078 /* Now handle simple and escaped characters. | |
| 1079 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim | |
| 1080 * accepts "\t", "\e", etc., but only when the 'l' flag in | |
| 1081 * 'cpoptions' is not included. | |
| 1082 * Posix doesn't recognize backslash at all. | |
| 1083 */ | |
| 1084 if (*regparse == '\\' | |
| 1085 && !cpo_bsl | |
| 1086 && regparse + 1 <= endp | |
| 1087 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL | |
| 1088 || (!cpo_lit | |
| 1089 && vim_strchr(REGEXP_ABBR, regparse[1]) | |
| 1090 != NULL) | |
| 1091 ) | |
| 1092 ) | |
| 1093 { | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1094 mb_ptr_adv(regparse); |
| 4444 | 1095 |
|
4507
9dbbddb2ed10
updated for version 7.3.1001
Bram Moolenaar <bram@vim.org>
parents:
4503
diff
changeset
|
1096 if (*regparse == 'n') |
| 4444 | 1097 startc = reg_string ? NL : NFA_NEWL; |
| 1098 else | |
| 1099 if (*regparse == 'd' | |
| 1100 || *regparse == 'o' | |
| 1101 || *regparse == 'x' | |
| 1102 || *regparse == 'u' | |
| 1103 || *regparse == 'U' | |
| 1104 ) | |
| 1105 { | |
| 1106 /* TODO(RE) This needs more testing */ | |
| 1107 startc = coll_get_char(); | |
| 1108 got_coll_char = TRUE; | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1109 mb_ptr_back(old_regparse, regparse); |
| 4444 | 1110 } |
| 1111 else | |
| 1112 { | |
| 1113 /* \r,\t,\e,\b */ | |
| 1114 startc = backslash_trans(*regparse); | |
| 1115 } | |
| 1116 } | |
| 1117 | |
| 1118 /* Normal printable char */ | |
| 1119 if (startc == -1) | |
| 1120 #ifdef FEAT_MBYTE | |
| 1121 startc = (*mb_ptr2char)(regparse); | |
| 1122 #else | |
| 1123 startc = *regparse; | |
| 1124 #endif | |
| 1125 | |
| 1126 /* Previous char was '-', so this char is end of range. */ | |
| 1127 if (emit_range) | |
| 1128 { | |
| 1129 endc = startc; startc = oldstartc; | |
| 1130 if (startc > endc) | |
| 1131 EMSG_RET_FAIL(_(e_invrange)); | |
| 1132 #ifdef FEAT_MBYTE | |
| 1133 if (has_mbyte && ((*mb_char2len)(startc) > 1 | |
| 1134 || (*mb_char2len)(endc) > 1)) | |
| 1135 { | |
| 1136 if (endc > startc + 256) | |
| 1137 EMSG_RET_FAIL(_(e_invrange)); | |
| 1138 /* Emit the range. "startc" was already emitted, so | |
| 1139 * skip it. */ | |
| 1140 for (c = startc + 1; c <= endc; c++) | |
| 1141 { | |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1142 EMIT(c); |
| 4444 | 1143 TRY_NEG(); |
| 1144 EMIT_GLUE(); | |
| 1145 } | |
| 1146 emit_range = FALSE; | |
| 1147 } | |
| 1148 else | |
| 1149 #endif | |
| 1150 { | |
| 1151 #ifdef EBCDIC | |
| 1152 int alpha_only = FALSE; | |
| 1153 | |
| 1154 /* for alphabetical range skip the gaps | |
| 1155 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */ | |
| 1156 if (isalpha(startc) && isalpha(endc)) | |
| 1157 alpha_only = TRUE; | |
| 1158 #endif | |
| 1159 /* Emit the range. "startc" was already emitted, so | |
| 1160 * skip it. */ | |
| 1161 for (c = startc + 1; c <= endc; c++) | |
| 1162 #ifdef EBCDIC | |
| 1163 if (!alpha_only || isalpha(startc)) | |
| 1164 #endif | |
| 1165 { | |
| 1166 EMIT(c); | |
| 1167 TRY_NEG(); | |
| 1168 EMIT_GLUE(); | |
| 1169 } | |
| 1170 emit_range = FALSE; | |
| 1171 } | |
| 1172 } | |
| 1173 else | |
| 1174 { | |
| 1175 /* | |
| 1176 * This char (startc) is not part of a range. Just | |
| 1177 * emit it. | |
| 1178 * | |
| 1179 * Normally, simply emit startc. But if we get char | |
| 1180 * code=0 from a collating char, then replace it with | |
| 1181 * 0x0a. | |
| 1182 * | |
| 1183 * This is needed to completely mimic the behaviour of | |
| 1184 * the backtracking engine. | |
| 1185 */ | |
| 1186 if (got_coll_char == TRUE && startc == 0) | |
| 1187 EMIT(0x0a); | |
| 1188 else | |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1189 EMIT(startc); |
| 4444 | 1190 TRY_NEG(); |
| 1191 EMIT_GLUE(); | |
| 1192 } | |
| 1193 | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1194 mb_ptr_adv(regparse); |
| 4444 | 1195 } /* while (p < endp) */ |
| 1196 | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1197 mb_ptr_back(old_regparse, regparse); |
| 4444 | 1198 if (*regparse == '-') /* if last, '-' is just a char */ |
| 1199 { | |
| 1200 EMIT('-'); | |
| 1201 TRY_NEG(); | |
| 1202 EMIT_GLUE(); | |
| 1203 } | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1204 mb_ptr_adv(regparse); |
| 4444 | 1205 |
| 1206 /* skip the trailing ] */ | |
| 1207 regparse = endp; | |
|
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1208 mb_ptr_adv(regparse); |
| 4444 | 1209 if (negated == TRUE) |
| 1210 { | |
| 1211 /* Mark end of negated char range */ | |
| 1212 EMIT(NFA_END_NEG_RANGE); | |
| 1213 EMIT(NFA_CONCAT); | |
| 1214 } | |
|
4615
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1215 |
|
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1216 /* \_[] also matches \n but it's not negated */ |
|
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1217 if (extra == ADD_NL) |
|
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1218 { |
|
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1219 EMIT(reg_string ? NL : NFA_NEWL); |
|
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1220 EMIT(NFA_OR); |
|
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1221 } |
|
5679b8ddd8cc
updated for version 7.3.1055
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
1222 |
| 4444 | 1223 return OK; |
|
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1224 } /* if exists closing ] */ |
|
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1225 |
|
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1226 if (reg_strict) |
| 4444 | 1227 { |
| 1228 syntax_error = TRUE; | |
| 1229 EMSG_RET_FAIL(_(e_missingbracket)); | |
| 1230 } | |
|
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1231 /* FALLTHROUGH */ |
|
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1232 |
| 4444 | 1233 default: |
| 1234 { | |
| 1235 #ifdef FEAT_MBYTE | |
| 1236 int plen; | |
| 1237 | |
| 1238 nfa_do_multibyte: | |
|
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1239 /* 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
|
1240 if (enc_utf8 && ((*mb_char2len)(c) |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1241 != (plen = (*mb_ptr2len)(old_regparse)) |
|
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1242 || utf_iscomposing(c))) |
| 4444 | 1243 { |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
1244 int i = 0; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
1245 |
|
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
1246 /* 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
|
1247 * or more composing characters. |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1248 * 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
|
1249 * 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
|
1250 * NFA_END_COMPOSING is the ). Note that right now we are |
| 4444 | 1251 * building the postfix form, not the NFA itself; |
| 1252 * 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
|
1253 * 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
|
1254 for (;;) |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1255 { |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1256 EMIT(c); |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1257 if (i > 0) |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1258 EMIT(NFA_CONCAT); |
|
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1259 if ((i += utf_char2len(c)) >= plen) |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1260 break; |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1261 c = utf_ptr2char(old_regparse + i); |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1262 } |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1263 EMIT(NFA_COMPOSING); |
| 4444 | 1264 regparse = old_regparse + plen; |
| 1265 } | |
| 1266 else | |
| 1267 #endif | |
| 1268 { | |
| 1269 c = no_Magic(c); | |
| 1270 EMIT(c); | |
| 1271 } | |
| 1272 return OK; | |
| 1273 } | |
| 1274 } | |
| 1275 | |
| 1276 #undef TRY_NEG | |
| 1277 #undef EMIT_GLUE | |
| 1278 | |
| 1279 return OK; | |
| 1280 } | |
| 1281 | |
| 1282 /* | |
| 1283 * Parse something followed by possible [*+=]. | |
| 1284 * | |
| 1285 * A piece is an atom, possibly followed by a multi, an indication of how many | |
| 1286 * times the atom can be matched. Example: "a*" matches any sequence of "a" | |
| 1287 * characters: "", "a", "aa", etc. | |
| 1288 * | |
| 1289 * piece ::= atom | |
| 1290 * or atom multi | |
| 1291 */ | |
| 1292 static int | |
| 1293 nfa_regpiece() | |
| 1294 { | |
| 1295 int i; | |
| 1296 int op; | |
| 1297 int ret; | |
| 1298 long minval, maxval; | |
| 1299 int greedy = TRUE; /* Braces are prefixed with '-' ? */ | |
| 1300 char_u *old_regparse, *new_regparse; | |
| 1301 int c2; | |
| 1302 int *old_post_ptr, *my_post_start; | |
| 1303 int old_regnpar; | |
| 1304 int quest; | |
| 1305 | |
| 1306 /* Save the current position in the regexp, so that we can use it if | |
| 1307 * <atom>{m,n} is next. */ | |
| 1308 old_regparse = regparse; | |
| 1309 /* Save current number of open parenthesis, so we can use it if | |
| 1310 * <atom>{m,n} is next */ | |
| 1311 old_regnpar = regnpar; | |
| 1312 /* store current pos in the postfix form, for \{m,n} involving 0s */ | |
| 1313 my_post_start = post_ptr; | |
| 1314 | |
| 1315 ret = nfa_regatom(); | |
| 1316 if (ret == FAIL) | |
| 1317 return FAIL; /* cascaded error */ | |
| 1318 | |
| 1319 op = peekchr(); | |
| 1320 if (re_multi_type(op) == NOT_MULTI) | |
| 1321 return OK; | |
| 1322 | |
| 1323 skipchr(); | |
| 1324 switch (op) | |
| 1325 { | |
| 1326 case Magic('*'): | |
| 1327 EMIT(NFA_STAR); | |
| 1328 break; | |
| 1329 | |
| 1330 case Magic('+'): | |
| 1331 /* | |
| 1332 * Trick: Normally, (a*)\+ would match the whole input "aaa". The | |
| 1333 * first and only submatch would be "aaa". But the backtracking | |
| 1334 * engine interprets the plus as "try matching one more time", and | |
| 1335 * a* matches a second time at the end of the input, the empty | |
| 1336 * string. | |
| 1337 * The submatch will the empty string. | |
| 1338 * | |
| 1339 * In order to be consistent with the old engine, we disable | |
| 1340 * NFA_PLUS, and replace <atom>+ with <atom><atom>* | |
| 1341 */ | |
| 1342 /* EMIT(NFA_PLUS); */ | |
| 1343 regnpar = old_regnpar; | |
| 1344 regparse = old_regparse; | |
| 1345 curchr = -1; | |
| 1346 if (nfa_regatom() == FAIL) | |
| 1347 return FAIL; | |
| 1348 EMIT(NFA_STAR); | |
| 1349 EMIT(NFA_CONCAT); | |
| 1350 skipchr(); /* skip the \+ */ | |
| 1351 break; | |
| 1352 | |
| 1353 case Magic('@'): | |
| 1354 op = no_Magic(getchr()); | |
| 1355 switch(op) | |
| 1356 { | |
| 1357 case '=': | |
| 1358 EMIT(NFA_PREV_ATOM_NO_WIDTH); | |
| 1359 break; | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1360 case '0': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1361 case '1': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1362 case '2': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1363 case '3': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1364 case '4': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1365 case '5': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1366 case '6': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1367 case '7': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1368 case '8': |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
1369 case '9': |
| 4444 | 1370 case '!': |
| 1371 case '<': | |
| 1372 case '>': | |
| 1373 /* Not supported yet */ | |
| 1374 return FAIL; | |
| 1375 default: | |
| 1376 syntax_error = TRUE; | |
| 4450 | 1377 EMSGN(_("E869: (NFA) Unknown operator '\\@%c'"), op); |
| 4444 | 1378 return FAIL; |
| 1379 } | |
| 1380 break; | |
| 1381 | |
| 1382 case Magic('?'): | |
| 1383 case Magic('='): | |
| 1384 EMIT(NFA_QUEST); | |
| 1385 break; | |
| 1386 | |
| 1387 case Magic('{'): | |
| 1388 /* a{2,5} will expand to 'aaa?a?a?' | |
| 1389 * a{-1,3} will expand to 'aa??a??', where ?? is the nongreedy | |
| 1390 * version of '?' | |
| 1391 * \v(ab){2,3} will expand to '(ab)(ab)(ab)?', where all the | |
| 1392 * parenthesis have the same id | |
| 1393 */ | |
| 1394 | |
| 1395 greedy = TRUE; | |
| 1396 c2 = peekchr(); | |
| 1397 if (c2 == '-' || c2 == Magic('-')) | |
| 1398 { | |
| 1399 skipchr(); | |
| 1400 greedy = FALSE; | |
| 1401 } | |
| 1402 if (!read_limits(&minval, &maxval)) | |
| 1403 { | |
| 1404 syntax_error = TRUE; | |
| 1405 EMSG_RET_FAIL(_("E870: (NFA regexp) Error reading repetition limits")); | |
| 1406 } | |
| 1407 /* <atom>{0,inf}, <atom>{0,} and <atom>{} are equivalent to | |
| 1408 * <atom>* */ | |
| 1409 if (minval == 0 && maxval == MAX_LIMIT && greedy) | |
| 1410 { | |
| 1411 EMIT(NFA_STAR); | |
| 1412 break; | |
| 1413 } | |
| 1414 | |
| 1415 if (maxval > NFA_BRACES_MAXLIMIT) | |
| 1416 { | |
| 1417 /* This would yield a huge automaton and use too much memory. | |
| 1418 * Revert to old engine */ | |
| 1419 return FAIL; | |
| 1420 } | |
| 1421 | |
| 1422 /* Special case: x{0} or x{-0} */ | |
| 1423 if (maxval == 0) | |
| 1424 { | |
| 1425 /* Ignore result of previous call to nfa_regatom() */ | |
| 1426 post_ptr = my_post_start; | |
| 1427 /* NFA_SKIP_CHAR has 0-length and works everywhere */ | |
| 1428 EMIT(NFA_SKIP_CHAR); | |
| 1429 return OK; | |
| 1430 } | |
| 1431 | |
| 1432 /* Ignore previous call to nfa_regatom() */ | |
| 1433 post_ptr = my_post_start; | |
| 1434 /* Save pos after the repeated atom and the \{} */ | |
| 1435 new_regparse = regparse; | |
| 1436 | |
| 1437 quest = (greedy == TRUE? NFA_QUEST : NFA_QUEST_NONGREEDY); | |
| 1438 for (i = 0; i < maxval; i++) | |
| 1439 { | |
| 1440 /* Goto beginning of the repeated atom */ | |
| 1441 regparse = old_regparse; | |
| 1442 curchr = -1; | |
| 1443 /* Restore count of parenthesis */ | |
| 1444 regnpar = old_regnpar; | |
| 1445 old_post_ptr = post_ptr; | |
| 1446 if (nfa_regatom() == FAIL) | |
| 1447 return FAIL; | |
| 1448 /* after "minval" times, atoms are optional */ | |
| 1449 if (i + 1 > minval) | |
| 1450 EMIT(quest); | |
| 1451 if (old_post_ptr != my_post_start) | |
| 1452 EMIT(NFA_CONCAT); | |
| 1453 } | |
| 1454 | |
| 1455 /* Go to just after the repeated atom and the \{} */ | |
| 1456 regparse = new_regparse; | |
| 1457 curchr = -1; | |
| 1458 | |
| 1459 break; | |
| 1460 | |
| 1461 | |
| 1462 default: | |
| 1463 break; | |
| 1464 } /* end switch */ | |
| 1465 | |
| 1466 if (re_multi_type(peekchr()) != NOT_MULTI) | |
| 1467 { | |
| 1468 /* Can't have a multi follow a multi. */ | |
| 1469 syntax_error = TRUE; | |
| 1470 EMSG_RET_FAIL(_("E871: (NFA regexp) Can't have a multi follow a multi !")); | |
| 1471 } | |
| 1472 | |
| 1473 return OK; | |
| 1474 } | |
| 1475 | |
| 1476 /* | |
| 1477 * Parse one or more pieces, concatenated. It matches a match for the | |
| 1478 * first piece, followed by a match for the second piece, etc. Example: | |
| 1479 * "f[0-9]b", first matches "f", then a digit and then "b". | |
| 1480 * | |
| 1481 * concat ::= piece | |
| 1482 * or piece piece | |
| 1483 * or piece piece piece | |
| 1484 * etc. | |
| 1485 */ | |
| 1486 static int | |
| 1487 nfa_regconcat() | |
| 1488 { | |
| 1489 int cont = TRUE; | |
| 1490 int first = TRUE; | |
| 1491 | |
| 1492 while (cont) | |
| 1493 { | |
| 1494 switch (peekchr()) | |
| 1495 { | |
| 1496 case NUL: | |
| 1497 case Magic('|'): | |
| 1498 case Magic('&'): | |
| 1499 case Magic(')'): | |
| 1500 cont = FALSE; | |
| 1501 break; | |
| 1502 | |
| 1503 case Magic('Z'): | |
| 1504 #ifdef FEAT_MBYTE | |
| 1505 regflags |= RF_ICOMBINE; | |
| 1506 #endif | |
| 1507 skipchr_keepstart(); | |
| 1508 break; | |
| 1509 case Magic('c'): | |
| 1510 regflags |= RF_ICASE; | |
| 1511 skipchr_keepstart(); | |
| 1512 break; | |
| 1513 case Magic('C'): | |
| 1514 regflags |= RF_NOICASE; | |
| 1515 skipchr_keepstart(); | |
| 1516 break; | |
| 1517 case Magic('v'): | |
| 1518 reg_magic = MAGIC_ALL; | |
| 1519 skipchr_keepstart(); | |
| 1520 curchr = -1; | |
| 1521 break; | |
| 1522 case Magic('m'): | |
| 1523 reg_magic = MAGIC_ON; | |
| 1524 skipchr_keepstart(); | |
| 1525 curchr = -1; | |
| 1526 break; | |
| 1527 case Magic('M'): | |
| 1528 reg_magic = MAGIC_OFF; | |
| 1529 skipchr_keepstart(); | |
| 1530 curchr = -1; | |
| 1531 break; | |
| 1532 case Magic('V'): | |
| 1533 reg_magic = MAGIC_NONE; | |
| 1534 skipchr_keepstart(); | |
| 1535 curchr = -1; | |
| 1536 break; | |
| 1537 | |
| 1538 default: | |
| 1539 if (nfa_regpiece() == FAIL) | |
| 1540 return FAIL; | |
| 1541 if (first == FALSE) | |
| 1542 EMIT(NFA_CONCAT); | |
| 1543 else | |
| 1544 first = FALSE; | |
| 1545 break; | |
| 1546 } | |
| 1547 } | |
| 1548 | |
| 1549 return OK; | |
| 1550 } | |
| 1551 | |
| 1552 /* | |
| 1553 * Parse a branch, one or more concats, separated by "\&". It matches the | |
| 1554 * last concat, but only if all the preceding concats also match at the same | |
| 1555 * position. Examples: | |
| 1556 * "foobeep\&..." matches "foo" in "foobeep". | |
| 1557 * ".*Peter\&.*Bob" matches in a line containing both "Peter" and "Bob" | |
| 1558 * | |
| 1559 * branch ::= concat | |
| 1560 * or concat \& concat | |
| 1561 * or concat \& concat \& concat | |
| 1562 * etc. | |
| 1563 */ | |
| 1564 static int | |
| 1565 nfa_regbranch() | |
| 1566 { | |
| 1567 int ch; | |
| 1568 int *old_post_ptr; | |
| 1569 | |
| 1570 old_post_ptr = post_ptr; | |
| 1571 | |
| 1572 /* First branch, possibly the only one */ | |
| 1573 if (nfa_regconcat() == FAIL) | |
| 1574 return FAIL; | |
| 1575 | |
| 1576 ch = peekchr(); | |
| 1577 /* Try next concats */ | |
| 1578 while (ch == Magic('&')) | |
| 1579 { | |
| 1580 skipchr(); | |
| 1581 EMIT(NFA_NOPEN); | |
| 1582 EMIT(NFA_PREV_ATOM_NO_WIDTH); | |
| 1583 old_post_ptr = post_ptr; | |
| 1584 if (nfa_regconcat() == FAIL) | |
| 1585 return FAIL; | |
| 1586 /* if concat is empty, skip a input char. But do emit a node */ | |
| 1587 if (old_post_ptr == post_ptr) | |
| 1588 EMIT(NFA_SKIP_CHAR); | |
| 1589 EMIT(NFA_CONCAT); | |
| 1590 ch = peekchr(); | |
| 1591 } | |
| 1592 | |
| 1593 /* Even if a branch is empty, emit one node for it */ | |
| 1594 if (old_post_ptr == post_ptr) | |
| 1595 EMIT(NFA_SKIP_CHAR); | |
| 1596 | |
| 1597 return OK; | |
| 1598 } | |
| 1599 | |
| 1600 /* | |
| 1601 * Parse a pattern, one or more branches, separated by "\|". It matches | |
| 1602 * anything that matches one of the branches. Example: "foo\|beep" matches | |
| 1603 * "foo" and matches "beep". If more than one branch matches, the first one | |
| 1604 * is used. | |
| 1605 * | |
| 1606 * pattern ::= branch | |
| 1607 * or branch \| branch | |
| 1608 * or branch \| branch \| branch | |
| 1609 * etc. | |
| 1610 */ | |
| 1611 static int | |
| 1612 nfa_reg(paren) | |
| 1613 int paren; /* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */ | |
| 1614 { | |
| 1615 int parno = 0; | |
| 1616 | |
| 1617 #ifdef FEAT_SYN_HL | |
| 1618 #endif | |
| 1619 if (paren == REG_PAREN) | |
| 1620 { | |
| 1621 if (regnpar >= NSUBEXP) /* Too many `(' */ | |
| 1622 { | |
| 1623 syntax_error = TRUE; | |
| 1624 EMSG_RET_FAIL(_("E872: (NFA regexp) Too many '('")); | |
| 1625 } | |
| 1626 parno = regnpar++; | |
| 1627 } | |
| 1628 | |
| 1629 if (nfa_regbranch() == FAIL) | |
| 1630 return FAIL; /* cascaded error */ | |
| 1631 | |
| 1632 while (peekchr() == Magic('|')) | |
| 1633 { | |
| 1634 skipchr(); | |
| 1635 if (nfa_regbranch() == FAIL) | |
| 1636 return FAIL; /* cascaded error */ | |
| 1637 EMIT(NFA_OR); | |
| 1638 } | |
| 1639 | |
| 1640 /* Check for proper termination. */ | |
| 1641 if (paren != REG_NOPAREN && getchr() != Magic(')')) | |
| 1642 { | |
| 1643 syntax_error = TRUE; | |
| 1644 if (paren == REG_NPAREN) | |
| 1645 EMSG2_RET_FAIL(_(e_unmatchedpp), reg_magic == MAGIC_ALL); | |
| 1646 else | |
| 1647 EMSG2_RET_FAIL(_(e_unmatchedp), reg_magic == MAGIC_ALL); | |
| 1648 } | |
| 1649 else if (paren == REG_NOPAREN && peekchr() != NUL) | |
| 1650 { | |
| 1651 syntax_error = TRUE; | |
| 1652 if (peekchr() == Magic(')')) | |
| 1653 EMSG2_RET_FAIL(_(e_unmatchedpar), reg_magic == MAGIC_ALL); | |
| 1654 else | |
| 1655 EMSG_RET_FAIL(_("E873: (NFA regexp) proper termination error")); | |
| 1656 } | |
| 1657 /* | |
| 1658 * Here we set the flag allowing back references to this set of | |
| 1659 * parentheses. | |
| 1660 */ | |
| 1661 if (paren == REG_PAREN) | |
| 1662 { | |
| 1663 had_endbrace[parno] = TRUE; /* have seen the close paren */ | |
| 1664 EMIT(NFA_MOPEN + parno); | |
| 1665 } | |
| 1666 | |
| 1667 return OK; | |
| 1668 } | |
| 1669 | |
| 1670 #ifdef DEBUG | |
| 1671 static char_u code[50]; | |
| 1672 | |
| 1673 static void | |
| 1674 nfa_set_code(c) | |
| 1675 int c; | |
| 1676 { | |
| 1677 int addnl = FALSE; | |
| 1678 | |
| 1679 if (c >= NFA_FIRST_NL && c <= NFA_LAST_NL) | |
| 1680 { | |
| 1681 addnl = TRUE; | |
| 1682 c -= ADD_NL; | |
| 1683 } | |
| 1684 | |
| 1685 STRCPY(code, ""); | |
| 1686 switch (c) | |
| 1687 { | |
| 1688 case NFA_MATCH: STRCPY(code, "NFA_MATCH "); break; | |
| 1689 case NFA_SPLIT: STRCPY(code, "NFA_SPLIT "); break; | |
| 1690 case NFA_CONCAT: STRCPY(code, "NFA_CONCAT "); break; | |
| 1691 case NFA_NEWL: STRCPY(code, "NFA_NEWL "); break; | |
| 1692 case NFA_ZSTART: STRCPY(code, "NFA_ZSTART"); break; | |
| 1693 case NFA_ZEND: STRCPY(code, "NFA_ZEND"); break; | |
| 1694 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1695 case NFA_BACKREF1: STRCPY(code, "NFA_BACKREF1"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1696 case NFA_BACKREF2: STRCPY(code, "NFA_BACKREF2"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1697 case NFA_BACKREF3: STRCPY(code, "NFA_BACKREF3"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1698 case NFA_BACKREF4: STRCPY(code, "NFA_BACKREF4"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1699 case NFA_BACKREF5: STRCPY(code, "NFA_BACKREF5"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1700 case NFA_BACKREF6: STRCPY(code, "NFA_BACKREF6"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1701 case NFA_BACKREF7: STRCPY(code, "NFA_BACKREF7"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1702 case NFA_BACKREF8: STRCPY(code, "NFA_BACKREF8"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1703 case NFA_BACKREF9: STRCPY(code, "NFA_BACKREF9"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1704 case NFA_SKIP: STRCPY(code, "NFA_SKIP"); break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1705 |
| 4444 | 1706 case NFA_PREV_ATOM_NO_WIDTH: |
| 1707 STRCPY(code, "NFA_PREV_ATOM_NO_WIDTH"); break; | |
|
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
1708 case NFA_PREV_ATOM_NO_WIDTH_NEG: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
1709 STRCPY(code, "NFA_PREV_ATOM_NO_WIDTH_NEG"); break; |
| 4444 | 1710 case NFA_NOPEN: STRCPY(code, "NFA_MOPEN_INVISIBLE"); break; |
| 1711 case NFA_NCLOSE: STRCPY(code, "NFA_MCLOSE_INVISIBLE"); break; | |
| 1712 case NFA_START_INVISIBLE: STRCPY(code, "NFA_START_INVISIBLE"); break; | |
| 1713 case NFA_END_INVISIBLE: STRCPY(code, "NFA_END_INVISIBLE"); break; | |
| 1714 | |
| 1715 case NFA_COMPOSING: STRCPY(code, "NFA_COMPOSING"); break; | |
| 1716 case NFA_END_COMPOSING: STRCPY(code, "NFA_END_COMPOSING"); break; | |
| 1717 | |
| 1718 case NFA_MOPEN + 0: | |
| 1719 case NFA_MOPEN + 1: | |
| 1720 case NFA_MOPEN + 2: | |
| 1721 case NFA_MOPEN + 3: | |
| 1722 case NFA_MOPEN + 4: | |
| 1723 case NFA_MOPEN + 5: | |
| 1724 case NFA_MOPEN + 6: | |
| 1725 case NFA_MOPEN + 7: | |
| 1726 case NFA_MOPEN + 8: | |
| 1727 case NFA_MOPEN + 9: | |
| 1728 STRCPY(code, "NFA_MOPEN(x)"); | |
| 1729 code[10] = c - NFA_MOPEN + '0'; | |
| 1730 break; | |
| 1731 case NFA_MCLOSE + 0: | |
| 1732 case NFA_MCLOSE + 1: | |
| 1733 case NFA_MCLOSE + 2: | |
| 1734 case NFA_MCLOSE + 3: | |
| 1735 case NFA_MCLOSE + 4: | |
| 1736 case NFA_MCLOSE + 5: | |
| 1737 case NFA_MCLOSE + 6: | |
| 1738 case NFA_MCLOSE + 7: | |
| 1739 case NFA_MCLOSE + 8: | |
| 1740 case NFA_MCLOSE + 9: | |
| 1741 STRCPY(code, "NFA_MCLOSE(x)"); | |
| 1742 code[11] = c - NFA_MCLOSE + '0'; | |
| 1743 break; | |
| 1744 case NFA_EOL: STRCPY(code, "NFA_EOL "); break; | |
| 1745 case NFA_BOL: STRCPY(code, "NFA_BOL "); break; | |
| 1746 case NFA_EOW: STRCPY(code, "NFA_EOW "); break; | |
| 1747 case NFA_BOW: STRCPY(code, "NFA_BOW "); break; | |
| 1748 case NFA_STAR: STRCPY(code, "NFA_STAR "); break; | |
| 1749 case NFA_PLUS: STRCPY(code, "NFA_PLUS "); break; | |
| 1750 case NFA_NOT: STRCPY(code, "NFA_NOT "); break; | |
| 1751 case NFA_SKIP_CHAR: STRCPY(code, "NFA_SKIP_CHAR"); break; | |
| 1752 case NFA_OR: STRCPY(code, "NFA_OR"); break; | |
| 1753 case NFA_QUEST: STRCPY(code, "NFA_QUEST"); break; | |
| 1754 case NFA_QUEST_NONGREEDY: STRCPY(code, "NFA_QUEST_NON_GREEDY"); break; | |
| 1755 case NFA_END_NEG_RANGE: STRCPY(code, "NFA_END_NEG_RANGE"); break; | |
| 1756 case NFA_CLASS_ALNUM: STRCPY(code, "NFA_CLASS_ALNUM"); break; | |
| 1757 case NFA_CLASS_ALPHA: STRCPY(code, "NFA_CLASS_ALPHA"); break; | |
| 1758 case NFA_CLASS_BLANK: STRCPY(code, "NFA_CLASS_BLANK"); break; | |
| 1759 case NFA_CLASS_CNTRL: STRCPY(code, "NFA_CLASS_CNTRL"); break; | |
| 1760 case NFA_CLASS_DIGIT: STRCPY(code, "NFA_CLASS_DIGIT"); break; | |
| 1761 case NFA_CLASS_GRAPH: STRCPY(code, "NFA_CLASS_GRAPH"); break; | |
| 1762 case NFA_CLASS_LOWER: STRCPY(code, "NFA_CLASS_LOWER"); break; | |
| 1763 case NFA_CLASS_PRINT: STRCPY(code, "NFA_CLASS_PRINT"); break; | |
| 1764 case NFA_CLASS_PUNCT: STRCPY(code, "NFA_CLASS_PUNCT"); break; | |
| 1765 case NFA_CLASS_SPACE: STRCPY(code, "NFA_CLASS_SPACE"); break; | |
| 1766 case NFA_CLASS_UPPER: STRCPY(code, "NFA_CLASS_UPPER"); break; | |
| 1767 case NFA_CLASS_XDIGIT: STRCPY(code, "NFA_CLASS_XDIGIT"); break; | |
| 1768 case NFA_CLASS_TAB: STRCPY(code, "NFA_CLASS_TAB"); break; | |
| 1769 case NFA_CLASS_RETURN: STRCPY(code, "NFA_CLASS_RETURN"); break; | |
| 1770 case NFA_CLASS_BACKSPACE: STRCPY(code, "NFA_CLASS_BACKSPACE"); break; | |
| 1771 case NFA_CLASS_ESCAPE: STRCPY(code, "NFA_CLASS_ESCAPE"); break; | |
| 1772 | |
| 1773 case NFA_ANY: STRCPY(code, "NFA_ANY"); break; | |
| 1774 case NFA_IDENT: STRCPY(code, "NFA_IDENT"); break; | |
| 1775 case NFA_SIDENT:STRCPY(code, "NFA_SIDENT"); break; | |
| 1776 case NFA_KWORD: STRCPY(code, "NFA_KWORD"); break; | |
| 1777 case NFA_SKWORD:STRCPY(code, "NFA_SKWORD"); break; | |
| 1778 case NFA_FNAME: STRCPY(code, "NFA_FNAME"); break; | |
| 1779 case NFA_SFNAME:STRCPY(code, "NFA_SFNAME"); break; | |
| 1780 case NFA_PRINT: STRCPY(code, "NFA_PRINT"); break; | |
| 1781 case NFA_SPRINT:STRCPY(code, "NFA_SPRINT"); break; | |
| 1782 case NFA_WHITE: STRCPY(code, "NFA_WHITE"); break; | |
| 1783 case NFA_NWHITE:STRCPY(code, "NFA_NWHITE"); break; | |
| 1784 case NFA_DIGIT: STRCPY(code, "NFA_DIGIT"); break; | |
| 1785 case NFA_NDIGIT:STRCPY(code, "NFA_NDIGIT"); break; | |
| 1786 case NFA_HEX: STRCPY(code, "NFA_HEX"); break; | |
| 1787 case NFA_NHEX: STRCPY(code, "NFA_NHEX"); break; | |
| 1788 case NFA_OCTAL: STRCPY(code, "NFA_OCTAL"); break; | |
| 1789 case NFA_NOCTAL:STRCPY(code, "NFA_NOCTAL"); break; | |
| 1790 case NFA_WORD: STRCPY(code, "NFA_WORD"); break; | |
| 1791 case NFA_NWORD: STRCPY(code, "NFA_NWORD"); break; | |
| 1792 case NFA_HEAD: STRCPY(code, "NFA_HEAD"); break; | |
| 1793 case NFA_NHEAD: STRCPY(code, "NFA_NHEAD"); break; | |
| 1794 case NFA_ALPHA: STRCPY(code, "NFA_ALPHA"); break; | |
| 1795 case NFA_NALPHA:STRCPY(code, "NFA_NALPHA"); break; | |
| 1796 case NFA_LOWER: STRCPY(code, "NFA_LOWER"); break; | |
| 1797 case NFA_NLOWER:STRCPY(code, "NFA_NLOWER"); break; | |
| 1798 case NFA_UPPER: STRCPY(code, "NFA_UPPER"); break; | |
| 1799 case NFA_NUPPER:STRCPY(code, "NFA_NUPPER"); break; | |
| 1800 | |
| 1801 default: | |
| 1802 STRCPY(code, "CHAR(x)"); | |
| 1803 code[5] = c; | |
| 1804 } | |
| 1805 | |
| 1806 if (addnl == TRUE) | |
| 1807 STRCAT(code, " + NEWLINE "); | |
| 1808 | |
| 1809 } | |
| 1810 | |
| 1811 #ifdef ENABLE_LOG | |
| 1812 static FILE *log_fd; | |
| 1813 | |
| 1814 /* | |
| 1815 * Print the postfix notation of the current regexp. | |
| 1816 */ | |
| 1817 static void | |
| 1818 nfa_postfix_dump(expr, retval) | |
| 1819 char_u *expr; | |
| 1820 int retval; | |
| 1821 { | |
| 1822 int *p; | |
| 1823 FILE *f; | |
| 1824 | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
1825 f = fopen(NFA_REGEXP_DUMP_LOG, "a"); |
| 4444 | 1826 if (f != NULL) |
| 1827 { | |
| 1828 fprintf(f, "\n-------------------------\n"); | |
| 1829 if (retval == FAIL) | |
| 1830 fprintf(f, ">>> NFA engine failed ... \n"); | |
| 1831 else if (retval == OK) | |
| 1832 fprintf(f, ">>> NFA engine succeeded !\n"); | |
| 1833 fprintf(f, "Regexp: \"%s\"\nPostfix notation (char): \"", expr); | |
| 4464 | 1834 for (p = post_start; *p && p < post_end; p++) |
| 4444 | 1835 { |
| 1836 nfa_set_code(*p); | |
| 1837 fprintf(f, "%s, ", code); | |
| 1838 } | |
| 1839 fprintf(f, "\"\nPostfix notation (int): "); | |
| 4464 | 1840 for (p = post_start; *p && p < post_end; p++) |
| 4444 | 1841 fprintf(f, "%d ", *p); |
| 1842 fprintf(f, "\n\n"); | |
| 1843 fclose(f); | |
| 1844 } | |
| 1845 } | |
| 1846 | |
| 1847 /* | |
| 1848 * Print the NFA starting with a root node "state". | |
| 1849 */ | |
| 1850 static void | |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1851 nfa_print_state(debugf, state) |
| 4444 | 1852 FILE *debugf; |
| 1853 nfa_state_T *state; | |
| 1854 { | |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1855 garray_T indent; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1856 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1857 ga_init2(&indent, 1, 64); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1858 ga_append(&indent, '\0'); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1859 nfa_print_state2(debugf, state, &indent); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1860 ga_clear(&indent); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1861 } |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1862 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1863 static void |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1864 nfa_print_state2(debugf, state, indent) |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1865 FILE *debugf; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1866 nfa_state_T *state; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1867 garray_T *indent; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1868 { |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1869 char_u *p; |
| 4444 | 1870 |
| 1871 if (state == NULL) | |
| 1872 return; | |
| 1873 | |
| 1874 fprintf(debugf, "(%2d)", abs(state->id)); | |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1875 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1876 /* Output indent */ |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1877 p = (char_u *)indent->ga_data; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1878 if (indent->ga_len >= 3) |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1879 { |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1880 int last = indent->ga_len - 3; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1881 char_u save[2]; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1882 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1883 STRNCPY(save, &p[last], 2); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1884 STRNCPY(&p[last], "+-", 2); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1885 fprintf(debugf, " %s", p); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1886 STRNCPY(&p[last], save, 2); |
|
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 else |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1889 fprintf(debugf, " %s", p); |
| 4444 | 1890 |
| 1891 nfa_set_code(state->c); | |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1892 fprintf(debugf, "%s%s (%d) (id=%d)\n", |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1893 state->negated ? "NOT " : "", code, state->c, abs(state->id)); |
| 4444 | 1894 if (state->id < 0) |
| 1895 return; | |
| 1896 | |
| 1897 state->id = abs(state->id) * -1; | |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1898 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1899 /* grow indent for state->out */ |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1900 indent->ga_len -= 1; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1901 if (state->out1) |
|
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1902 ga_concat(indent, (char_u *)"| "); |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1903 else |
|
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1904 ga_concat(indent, (char_u *)" "); |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1905 ga_append(indent, '\0'); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1906 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1907 nfa_print_state2(debugf, state->out, indent); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1908 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1909 /* replace last part of indent for state->out1 */ |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1910 indent->ga_len -= 3; |
|
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1911 ga_concat(indent, (char_u *)" "); |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1912 ga_append(indent, '\0'); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1913 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1914 nfa_print_state2(debugf, state->out1, indent); |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1915 |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1916 /* shrink indent */ |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1917 indent->ga_len -= 3; |
|
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1918 ga_append(indent, '\0'); |
| 4444 | 1919 } |
| 1920 | |
| 1921 /* | |
| 1922 * Print the NFA state machine. | |
| 1923 */ | |
| 1924 static void | |
| 1925 nfa_dump(prog) | |
| 1926 nfa_regprog_T *prog; | |
| 1927 { | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
1928 FILE *debugf = fopen(NFA_REGEXP_DUMP_LOG, "a"); |
| 4444 | 1929 |
| 1930 if (debugf != NULL) | |
| 1931 { | |
|
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1932 nfa_print_state(debugf, prog->start); |
| 4444 | 1933 fclose(debugf); |
| 1934 } | |
| 1935 } | |
| 1936 #endif /* ENABLE_LOG */ | |
| 1937 #endif /* DEBUG */ | |
| 1938 | |
| 1939 /* | |
| 1940 * Parse r.e. @expr and convert it into postfix form. | |
| 1941 * Return the postfix string on success, NULL otherwise. | |
| 1942 */ | |
| 1943 static int * | |
| 1944 re2post() | |
| 1945 { | |
| 1946 if (nfa_reg(REG_NOPAREN) == FAIL) | |
| 1947 return NULL; | |
| 1948 EMIT(NFA_MOPEN); | |
| 1949 return post_start; | |
| 1950 } | |
| 1951 | |
| 1952 /* NB. Some of the code below is inspired by Russ's. */ | |
| 1953 | |
| 1954 /* | |
| 1955 * Represents an NFA state plus zero or one or two arrows exiting. | |
| 1956 * if c == MATCH, no arrows out; matching state. | |
| 1957 * If c == SPLIT, unlabeled arrows to out and out1 (if != NULL). | |
| 1958 * If c < 256, labeled arrow with character c to out. | |
| 1959 */ | |
| 1960 | |
| 1961 static nfa_state_T *state_ptr; /* points to nfa_prog->state */ | |
| 1962 | |
| 1963 /* | |
| 1964 * Allocate and initialize nfa_state_T. | |
| 1965 */ | |
| 1966 static nfa_state_T * | |
| 1967 new_state(c, out, out1) | |
| 1968 int c; | |
| 1969 nfa_state_T *out; | |
| 1970 nfa_state_T *out1; | |
| 1971 { | |
| 1972 nfa_state_T *s; | |
| 1973 | |
| 1974 if (istate >= nstate) | |
| 1975 return NULL; | |
| 1976 | |
| 1977 s = &state_ptr[istate++]; | |
| 1978 | |
| 1979 s->c = c; | |
| 1980 s->out = out; | |
| 1981 s->out1 = out1; | |
| 1982 | |
| 1983 s->id = istate; | |
| 1984 s->lastlist = 0; | |
| 1985 s->negated = FALSE; | |
| 1986 | |
| 1987 return s; | |
| 1988 } | |
| 1989 | |
| 1990 /* | |
| 1991 * A partially built NFA without the matching state filled in. | |
| 1992 * Frag_T.start points at the start state. | |
| 1993 * Frag_T.out is a list of places that need to be set to the | |
| 1994 * next state for this fragment. | |
| 1995 */ | |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1996 |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1997 /* 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
|
1998 * uninitialized, we use the pointers themselves |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1999 * as storage for the Ptrlists. */ |
| 4444 | 2000 typedef union Ptrlist Ptrlist; |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2001 union Ptrlist |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2002 { |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2003 Ptrlist *next; |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2004 nfa_state_T *s; |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2005 }; |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2006 |
| 4444 | 2007 struct Frag |
| 2008 { | |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
2009 nfa_state_T *start; |
| 4444 | 2010 Ptrlist *out; |
| 2011 }; | |
| 2012 typedef struct Frag Frag_T; | |
| 2013 | |
| 2014 static Frag_T frag __ARGS((nfa_state_T *start, Ptrlist *out)); | |
| 2015 static Ptrlist *list1 __ARGS((nfa_state_T **outp)); | |
| 2016 static void patch __ARGS((Ptrlist *l, nfa_state_T *s)); | |
| 2017 static Ptrlist *append __ARGS((Ptrlist *l1, Ptrlist *l2)); | |
| 2018 static void st_push __ARGS((Frag_T s, Frag_T **p, Frag_T *stack_end)); | |
| 2019 static Frag_T st_pop __ARGS((Frag_T **p, Frag_T *stack)); | |
| 2020 | |
| 2021 /* | |
| 4456 | 2022 * Initialize a Frag_T struct and return it. |
| 4444 | 2023 */ |
| 2024 static Frag_T | |
| 2025 frag(start, out) | |
| 2026 nfa_state_T *start; | |
| 2027 Ptrlist *out; | |
| 2028 { | |
| 4456 | 2029 Frag_T n; |
| 2030 | |
| 2031 n.start = start; | |
| 2032 n.out = out; | |
| 4444 | 2033 return n; |
| 2034 } | |
| 2035 | |
| 2036 /* | |
| 2037 * Create singleton list containing just outp. | |
| 2038 */ | |
| 2039 static Ptrlist * | |
| 2040 list1(outp) | |
| 2041 nfa_state_T **outp; | |
| 2042 { | |
| 2043 Ptrlist *l; | |
| 2044 | |
| 2045 l = (Ptrlist *)outp; | |
| 2046 l->next = NULL; | |
| 2047 return l; | |
| 2048 } | |
| 2049 | |
| 2050 /* | |
| 2051 * Patch the list of states at out to point to start. | |
| 2052 */ | |
| 2053 static void | |
| 2054 patch(l, s) | |
| 2055 Ptrlist *l; | |
| 2056 nfa_state_T *s; | |
| 2057 { | |
| 2058 Ptrlist *next; | |
| 2059 | |
| 2060 for (; l; l = next) | |
| 2061 { | |
| 2062 next = l->next; | |
| 2063 l->s = s; | |
| 2064 } | |
| 2065 } | |
| 2066 | |
| 2067 | |
| 2068 /* | |
| 2069 * Join the two lists l1 and l2, returning the combination. | |
| 2070 */ | |
| 2071 static Ptrlist * | |
| 2072 append(l1, l2) | |
| 2073 Ptrlist *l1; | |
| 2074 Ptrlist *l2; | |
| 2075 { | |
| 2076 Ptrlist *oldl1; | |
| 2077 | |
| 2078 oldl1 = l1; | |
| 2079 while (l1->next) | |
| 2080 l1 = l1->next; | |
| 2081 l1->next = l2; | |
| 2082 return oldl1; | |
| 2083 } | |
| 2084 | |
| 2085 /* | |
| 2086 * Stack used for transforming postfix form into NFA. | |
| 2087 */ | |
| 2088 static Frag_T empty; | |
| 2089 | |
| 2090 static void | |
| 2091 st_error(postfix, end, p) | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2092 int *postfix UNUSED; |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2093 int *end UNUSED; |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2094 int *p UNUSED; |
| 4444 | 2095 { |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2096 #ifdef NFA_REGEXP_ERROR_LOG |
| 4444 | 2097 FILE *df; |
| 2098 int *p2; | |
| 2099 | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2100 df = fopen(NFA_REGEXP_ERROR_LOG, "a"); |
| 4444 | 2101 if (df) |
| 2102 { | |
| 2103 fprintf(df, "Error popping the stack!\n"); | |
| 2104 #ifdef DEBUG | |
| 2105 fprintf(df, "Current regexp is \"%s\"\n", nfa_regengine.expr); | |
| 2106 #endif | |
| 2107 fprintf(df, "Postfix form is: "); | |
| 2108 #ifdef DEBUG | |
| 2109 for (p2 = postfix; p2 < end; p2++) | |
| 2110 { | |
| 2111 nfa_set_code(*p2); | |
| 2112 fprintf(df, "%s, ", code); | |
| 2113 } | |
| 2114 nfa_set_code(*p); | |
| 2115 fprintf(df, "\nCurrent position is: "); | |
| 2116 for (p2 = postfix; p2 <= p; p2 ++) | |
| 2117 { | |
| 2118 nfa_set_code(*p2); | |
| 2119 fprintf(df, "%s, ", code); | |
| 2120 } | |
| 2121 #else | |
| 2122 for (p2 = postfix; p2 < end; p2++) | |
| 2123 { | |
| 2124 fprintf(df, "%d, ", *p2); | |
| 2125 } | |
| 2126 fprintf(df, "\nCurrent position is: "); | |
| 2127 for (p2 = postfix; p2 <= p; p2 ++) | |
| 2128 { | |
| 2129 fprintf(df, "%d, ", *p2); | |
| 2130 } | |
| 2131 #endif | |
| 2132 fprintf(df, "\n--------------------------\n"); | |
| 2133 fclose(df); | |
| 2134 } | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2135 #endif |
| 4444 | 2136 EMSG(_("E874: (NFA) Could not pop the stack !")); |
| 2137 } | |
| 2138 | |
| 2139 /* | |
| 2140 * Push an item onto the stack. | |
| 2141 */ | |
| 2142 static void | |
| 2143 st_push(s, p, stack_end) | |
| 2144 Frag_T s; | |
| 2145 Frag_T **p; | |
| 2146 Frag_T *stack_end; | |
| 2147 { | |
| 2148 Frag_T *stackp = *p; | |
| 2149 | |
| 2150 if (stackp >= stack_end) | |
| 2151 return; | |
| 2152 *stackp = s; | |
| 2153 *p = *p + 1; | |
| 2154 } | |
| 2155 | |
| 2156 /* | |
| 2157 * Pop an item from the stack. | |
| 2158 */ | |
| 2159 static Frag_T | |
| 2160 st_pop(p, stack) | |
| 2161 Frag_T **p; | |
| 2162 Frag_T *stack; | |
| 2163 { | |
| 2164 Frag_T *stackp; | |
| 2165 | |
| 2166 *p = *p - 1; | |
| 2167 stackp = *p; | |
| 2168 if (stackp < stack) | |
| 2169 return empty; | |
| 2170 return **p; | |
| 2171 } | |
| 2172 | |
| 2173 /* | |
| 2174 * Convert a postfix form into its equivalent NFA. | |
| 2175 * Return the NFA start state on success, NULL otherwise. | |
| 2176 */ | |
| 2177 static nfa_state_T * | |
| 2178 post2nfa(postfix, end, nfa_calc_size) | |
| 2179 int *postfix; | |
| 2180 int *end; | |
| 2181 int nfa_calc_size; | |
| 2182 { | |
| 2183 int *p; | |
| 2184 int mopen; | |
| 2185 int mclose; | |
| 2186 Frag_T *stack = NULL; | |
| 2187 Frag_T *stackp = NULL; | |
| 2188 Frag_T *stack_end = NULL; | |
| 2189 Frag_T e1; | |
| 2190 Frag_T e2; | |
| 2191 Frag_T e; | |
| 2192 nfa_state_T *s; | |
| 2193 nfa_state_T *s1; | |
| 2194 nfa_state_T *matchstate; | |
| 4484 | 2195 nfa_state_T *ret = NULL; |
| 4444 | 2196 |
| 2197 if (postfix == NULL) | |
| 2198 return NULL; | |
| 2199 | |
| 4456 | 2200 #define PUSH(s) st_push((s), &stackp, stack_end) |
| 4444 | 2201 #define POP() st_pop(&stackp, stack); \ |
| 2202 if (stackp < stack) \ | |
| 2203 { \ | |
| 2204 st_error(postfix, end, p); \ | |
| 2205 return NULL; \ | |
| 2206 } | |
| 2207 | |
| 2208 if (nfa_calc_size == FALSE) | |
| 2209 { | |
| 2210 /* Allocate space for the stack. Max states on the stack : nstate */ | |
| 4462 | 2211 stack = (Frag_T *) lalloc((nstate + 1) * sizeof(Frag_T), TRUE); |
| 4444 | 2212 stackp = stack; |
| 4462 | 2213 stack_end = stack + (nstate + 1); |
| 4444 | 2214 } |
| 2215 | |
| 2216 for (p = postfix; p < end; ++p) | |
| 2217 { | |
| 2218 switch (*p) | |
| 2219 { | |
| 2220 case NFA_CONCAT: | |
| 2221 /* Catenation. | |
| 2222 * Pay attention: this operator does not exist | |
| 2223 * in the r.e. itself (it is implicit, really). | |
| 2224 * It is added when r.e. is translated to postfix | |
| 2225 * form in re2post(). | |
| 2226 * | |
| 2227 * No new state added here. */ | |
| 2228 if (nfa_calc_size == TRUE) | |
| 2229 { | |
| 4458 | 2230 /* nstate += 0; */ |
| 4444 | 2231 break; |
| 2232 } | |
| 2233 e2 = POP(); | |
| 2234 e1 = POP(); | |
| 2235 patch(e1.out, e2.start); | |
| 2236 PUSH(frag(e1.start, e2.out)); | |
| 2237 break; | |
| 2238 | |
| 2239 case NFA_NOT: | |
| 2240 /* Negation of a character */ | |
| 2241 if (nfa_calc_size == TRUE) | |
| 2242 { | |
| 4458 | 2243 /* nstate += 0; */ |
| 4444 | 2244 break; |
| 2245 } | |
| 2246 e1 = POP(); | |
| 2247 e1.start->negated = TRUE; | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2248 #ifdef FEAT_MBYTE |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2249 if (e1.start->c == NFA_COMPOSING) |
| 4444 | 2250 e1.start->out1->negated = TRUE; |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2251 #endif |
| 4444 | 2252 PUSH(e1); |
| 2253 break; | |
| 2254 | |
| 2255 case NFA_OR: | |
| 2256 /* Alternation */ | |
| 2257 if (nfa_calc_size == TRUE) | |
| 2258 { | |
| 4458 | 2259 nstate++; |
| 4444 | 2260 break; |
| 2261 } | |
| 2262 e2 = POP(); | |
| 2263 e1 = POP(); | |
| 2264 s = new_state(NFA_SPLIT, e1.start, e2.start); | |
| 2265 if (s == NULL) | |
| 4484 | 2266 goto theend; |
| 4444 | 2267 PUSH(frag(s, append(e1.out, e2.out))); |
| 2268 break; | |
| 2269 | |
| 2270 case NFA_STAR: | |
| 2271 /* Zero or more */ | |
| 2272 if (nfa_calc_size == TRUE) | |
| 2273 { | |
| 4458 | 2274 nstate++; |
| 4444 | 2275 break; |
| 2276 } | |
| 2277 e = POP(); | |
| 2278 s = new_state(NFA_SPLIT, e.start, NULL); | |
| 2279 if (s == NULL) | |
| 4484 | 2280 goto theend; |
| 4444 | 2281 patch(e.out, s); |
| 2282 PUSH(frag(s, list1(&s->out1))); | |
| 2283 break; | |
| 2284 | |
| 2285 case NFA_QUEST: | |
| 2286 /* one or zero atoms=> greedy match */ | |
| 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 PUSH(frag(s, append(e.out, list1(&s->out1)))); |
| 2297 break; | |
| 2298 | |
| 2299 case NFA_QUEST_NONGREEDY: | |
| 2300 /* zero or one atoms => non-greedy match */ | |
| 2301 if (nfa_calc_size == TRUE) | |
| 2302 { | |
| 4458 | 2303 nstate++; |
| 4444 | 2304 break; |
| 2305 } | |
| 2306 e = POP(); | |
| 2307 s = new_state(NFA_SPLIT, NULL, e.start); | |
| 2308 if (s == NULL) | |
| 4484 | 2309 goto theend; |
| 4444 | 2310 PUSH(frag(s, append(e.out, list1(&s->out)))); |
| 2311 break; | |
| 2312 | |
| 2313 case NFA_PLUS: | |
| 2314 /* One or more */ | |
| 2315 if (nfa_calc_size == TRUE) | |
| 2316 { | |
| 4458 | 2317 nstate++; |
| 4444 | 2318 break; |
| 2319 } | |
| 2320 e = POP(); | |
| 2321 s = new_state(NFA_SPLIT, e.start, NULL); | |
| 2322 if (s == NULL) | |
| 4484 | 2323 goto theend; |
| 4444 | 2324 patch(e.out, s); |
| 2325 PUSH(frag(e.start, list1(&s->out1))); | |
| 2326 break; | |
| 2327 | |
| 2328 case NFA_SKIP_CHAR: | |
| 2329 /* Symbol of 0-length, Used in a repetition | |
| 2330 * with max/min count of 0 */ | |
| 2331 if (nfa_calc_size == TRUE) | |
| 2332 { | |
| 4458 | 2333 nstate++; |
| 4444 | 2334 break; |
| 2335 } | |
| 2336 s = new_state(NFA_SKIP_CHAR, NULL, NULL); | |
| 2337 if (s == NULL) | |
| 4484 | 2338 goto theend; |
| 4444 | 2339 PUSH(frag(s, list1(&s->out))); |
| 2340 break; | |
| 2341 | |
| 2342 case NFA_PREV_ATOM_NO_WIDTH: | |
| 2343 /* The \@= operator: match the preceding atom with 0 width. | |
| 2344 * Surrounds the preceding atom with START_INVISIBLE and | |
| 2345 * END_INVISIBLE, similarly to MOPEN. | |
| 2346 */ | |
| 2347 /* TODO: Maybe this drops the speed? */ | |
| 4484 | 2348 goto theend; |
| 4444 | 2349 |
| 2350 if (nfa_calc_size == TRUE) | |
| 2351 { | |
| 2352 nstate += 2; | |
| 2353 break; | |
| 2354 } | |
| 2355 e = POP(); | |
| 2356 s1 = new_state(NFA_END_INVISIBLE, NULL, NULL); | |
| 2357 if (s1 == NULL) | |
| 4484 | 2358 goto theend; |
| 4444 | 2359 patch(e.out, s1); |
| 2360 | |
| 2361 s = new_state(NFA_START_INVISIBLE, e.start, s1); | |
| 2362 if (s == NULL) | |
| 4484 | 2363 goto theend; |
| 4444 | 2364 PUSH(frag(s, list1(&s1->out))); |
| 2365 break; | |
| 2366 | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2367 #ifdef FEAT_MBYTE |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2368 case NFA_COMPOSING: /* char with composing char */ |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2369 #if 0 |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2370 /* TODO */ |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2371 if (regflags & RF_ICOMBINE) |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2372 { |
|
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
2373 /* use the base character only */ |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2374 } |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2375 #endif |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2376 /* FALLTHROUGH */ |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2377 #endif |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2378 |
| 4444 | 2379 case NFA_MOPEN + 0: /* Submatch */ |
| 2380 case NFA_MOPEN + 1: | |
| 2381 case NFA_MOPEN + 2: | |
| 2382 case NFA_MOPEN + 3: | |
| 2383 case NFA_MOPEN + 4: | |
| 2384 case NFA_MOPEN + 5: | |
| 2385 case NFA_MOPEN + 6: | |
| 2386 case NFA_MOPEN + 7: | |
| 2387 case NFA_MOPEN + 8: | |
| 2388 case NFA_MOPEN + 9: | |
| 2389 case NFA_NOPEN: /* \%( "Invisible Submatch" */ | |
| 2390 if (nfa_calc_size == TRUE) | |
| 2391 { | |
| 2392 nstate += 2; | |
| 2393 break; | |
| 2394 } | |
| 2395 | |
| 2396 mopen = *p; | |
| 2397 switch (*p) | |
| 2398 { | |
| 2399 case NFA_NOPEN: | |
| 2400 mclose = NFA_NCLOSE; | |
| 2401 break; | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2402 #ifdef FEAT_MBYTE |
| 4444 | 2403 case NFA_COMPOSING: |
| 2404 mclose = NFA_END_COMPOSING; | |
| 2405 break; | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2406 #endif |
| 4444 | 2407 default: |
| 2408 /* NFA_MOPEN(0) ... NFA_MOPEN(9) */ | |
| 2409 mclose = *p + NSUBEXP; | |
| 2410 break; | |
| 2411 } | |
| 2412 | |
| 2413 /* Allow "NFA_MOPEN" as a valid postfix representation for | |
| 2414 * the empty regexp "". In this case, the NFA will be | |
| 2415 * NFA_MOPEN -> NFA_MCLOSE. Note that this also allows | |
| 2416 * empty groups of parenthesis, and empty mbyte chars */ | |
| 2417 if (stackp == stack) | |
| 2418 { | |
| 2419 s = new_state(mopen, NULL, NULL); | |
| 2420 if (s == NULL) | |
| 4484 | 2421 goto theend; |
| 4444 | 2422 s1 = new_state(mclose, NULL, NULL); |
| 2423 if (s1 == NULL) | |
| 4484 | 2424 goto theend; |
| 4444 | 2425 patch(list1(&s->out), s1); |
| 2426 PUSH(frag(s, list1(&s1->out))); | |
| 2427 break; | |
| 2428 } | |
| 2429 | |
| 2430 /* At least one node was emitted before NFA_MOPEN, so | |
| 2431 * at least one node will be between NFA_MOPEN and NFA_MCLOSE */ | |
| 2432 e = POP(); | |
| 2433 s = new_state(mopen, e.start, NULL); /* `(' */ | |
| 2434 if (s == NULL) | |
| 4484 | 2435 goto theend; |
| 4444 | 2436 |
| 2437 s1 = new_state(mclose, NULL, NULL); /* `)' */ | |
| 2438 if (s1 == NULL) | |
| 4484 | 2439 goto theend; |
| 4444 | 2440 patch(e.out, s1); |
| 2441 | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2442 #ifdef FEAT_MBYTE |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2443 if (mopen == NFA_COMPOSING) |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2444 /* COMPOSING->out1 = END_COMPOSING */ |
| 4444 | 2445 patch(list1(&s->out1), s1); |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2446 #endif |
| 4444 | 2447 |
| 2448 PUSH(frag(s, list1(&s1->out))); | |
| 2449 break; | |
| 2450 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2451 case NFA_BACKREF1: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2452 case NFA_BACKREF2: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2453 case NFA_BACKREF3: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2454 case NFA_BACKREF4: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2455 case NFA_BACKREF5: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2456 case NFA_BACKREF6: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2457 case NFA_BACKREF7: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2458 case NFA_BACKREF8: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2459 case NFA_BACKREF9: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2460 if (nfa_calc_size == TRUE) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2461 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2462 nstate += 2; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2463 break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2464 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2465 s = new_state(*p, NULL, NULL); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2466 if (s == NULL) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2467 goto theend; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2468 s1 = new_state(NFA_SKIP, NULL, NULL); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2469 if (s1 == NULL) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2470 goto theend; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2471 patch(list1(&s->out), s1); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2472 PUSH(frag(s, list1(&s1->out))); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2473 break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2474 |
|
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2475 case NFA_LNUM: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2476 case NFA_LNUM_GT: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2477 case NFA_LNUM_LT: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2478 case NFA_VCOL: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2479 case NFA_VCOL_GT: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2480 case NFA_VCOL_LT: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2481 case NFA_COL: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2482 case NFA_COL_GT: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2483 case NFA_COL_LT: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2484 if (nfa_calc_size == TRUE) |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2485 { |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2486 nstate += 1; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2487 break; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2488 } |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2489 e1 = POP(); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2490 s = new_state(*p, NULL, NULL); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2491 if (s == NULL) |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2492 goto theend; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2493 s->val = e1.start->c; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2494 PUSH(frag(s, list1(&s->out))); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2495 break; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2496 |
| 4444 | 2497 case NFA_ZSTART: |
| 2498 case NFA_ZEND: | |
| 2499 default: | |
| 2500 /* Operands */ | |
| 2501 if (nfa_calc_size == TRUE) | |
| 2502 { | |
| 4458 | 2503 nstate++; |
| 4444 | 2504 break; |
| 2505 } | |
| 2506 s = new_state(*p, NULL, NULL); | |
| 2507 if (s == NULL) | |
| 4484 | 2508 goto theend; |
| 4444 | 2509 PUSH(frag(s, list1(&s->out))); |
| 2510 break; | |
| 2511 | |
| 2512 } /* switch(*p) */ | |
| 2513 | |
| 2514 } /* for(p = postfix; *p; ++p) */ | |
| 2515 | |
| 2516 if (nfa_calc_size == TRUE) | |
| 2517 { | |
| 4458 | 2518 nstate++; |
| 4484 | 2519 goto theend; /* Return value when counting size is ignored anyway */ |
| 4444 | 2520 } |
| 2521 | |
| 2522 e = POP(); | |
| 2523 if (stackp != stack) | |
| 2524 EMSG_RET_NULL(_("E875: (NFA regexp) (While converting from postfix to NFA), too many states left on stack")); | |
| 2525 | |
| 2526 if (istate >= nstate) | |
| 2527 EMSG_RET_NULL(_("E876: (NFA regexp) Not enough space to store the whole NFA ")); | |
| 2528 | |
| 2529 matchstate = &state_ptr[istate++]; /* the match state */ | |
| 2530 matchstate->c = NFA_MATCH; | |
| 2531 matchstate->out = matchstate->out1 = NULL; | |
| 2532 | |
| 2533 patch(e.out, matchstate); | |
| 4484 | 2534 ret = e.start; |
| 2535 | |
| 2536 theend: | |
| 2537 vim_free(stack); | |
| 2538 return ret; | |
| 4444 | 2539 |
| 2540 #undef POP1 | |
| 2541 #undef PUSH1 | |
| 2542 #undef POP2 | |
| 2543 #undef PUSH2 | |
| 2544 #undef POP | |
| 2545 #undef PUSH | |
| 2546 } | |
| 2547 | |
| 2548 /**************************************************************** | |
| 2549 * NFA execution code. | |
| 2550 ****************************************************************/ | |
| 2551 | |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2552 typedef struct |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2553 { |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2554 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
|
2555 |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2556 /* 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
|
2557 union |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2558 { |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2559 struct multipos |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2560 { |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2561 lpos_T start; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2562 lpos_T end; |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2563 } multi[NSUBEXP]; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2564 struct linepos |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2565 { |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2566 char_u *start; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2567 char_u *end; |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2568 } line[NSUBEXP]; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2569 } list; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2570 } regsub_T; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2571 |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2572 /* 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
|
2573 typedef struct |
| 4444 | 2574 { |
| 2575 nfa_state_T *state; | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2576 int count; |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2577 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
|
2578 } nfa_thread_T; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2579 |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2580 /* nfa_list_T contains the alternative NFA execution states. */ |
| 4444 | 2581 typedef struct |
| 2582 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2583 nfa_thread_T *t; /* allocated array of states */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2584 int n; /* nr of states in "t" */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2585 int id; /* ID of the list */ |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2586 } nfa_list_T; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2587 |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2588 #ifdef ENABLE_LOG |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2589 static void |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2590 log_subexpr(sub) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2591 regsub_T *sub; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2592 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2593 int j; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2594 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2595 for (j = 0; j < sub->in_use; j++) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2596 if (REG_MULTI) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2597 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
|
2598 j, |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2599 sub->list.multi[j].start.col, |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2600 (int)sub->list.multi[j].start.lnum, |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2601 sub->list.multi[j].end.col, |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2602 (int)sub->list.multi[j].end.lnum); |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2603 else |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2604 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
|
2605 j, |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2606 (char *)sub->list.line[j].start, |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2607 (char *)sub->list.line[j].end); |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2608 fprintf(log_fd, "\n"); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2609 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2610 #endif |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2611 |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2612 /* 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
|
2613 static int nfa_match; |
|
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2614 |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2615 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
|
2616 static void addstate_here __ARGS((nfa_list_T *l, nfa_state_T *state, regsub_T *sub, int *ip)); |
| 4444 | 2617 |
| 2618 static void | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2619 addstate(l, state, sub, off) |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2620 nfa_list_T *l; /* runtime state list */ |
| 4444 | 2621 nfa_state_T *state; /* state to update */ |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2622 regsub_T *sub; /* pointers to subexpressions */ |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2623 int off; /* byte offset, when -1 go to next line */ |
| 4444 | 2624 { |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2625 int subidx; |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2626 nfa_thread_T *lastthread; |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2627 lpos_T save_lpos; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2628 int save_in_use; |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2629 char_u *save_ptr; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2630 int i; |
| 4444 | 2631 |
| 2632 if (l == NULL || state == NULL) | |
| 2633 return; | |
| 2634 | |
| 2635 switch (state->c) | |
| 2636 { | |
| 2637 case NFA_SPLIT: | |
| 2638 case NFA_NOT: | |
| 2639 case NFA_NOPEN: | |
| 2640 case NFA_NCLOSE: | |
| 2641 case NFA_MCLOSE: | |
| 2642 case NFA_MCLOSE + 1: | |
| 2643 case NFA_MCLOSE + 2: | |
| 2644 case NFA_MCLOSE + 3: | |
| 2645 case NFA_MCLOSE + 4: | |
| 2646 case NFA_MCLOSE + 5: | |
| 2647 case NFA_MCLOSE + 6: | |
| 2648 case NFA_MCLOSE + 7: | |
| 2649 case NFA_MCLOSE + 8: | |
| 2650 case NFA_MCLOSE + 9: | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2651 /* 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
|
2652 * "out1" may be added below. */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2653 break; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2654 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2655 case NFA_MOPEN: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2656 case NFA_MOPEN + 1: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2657 case NFA_MOPEN + 2: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2658 case NFA_MOPEN + 3: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2659 case NFA_MOPEN + 4: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2660 case NFA_MOPEN + 5: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2661 case NFA_MOPEN + 6: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2662 case NFA_MOPEN + 7: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2663 case NFA_MOPEN + 8: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2664 case NFA_MOPEN + 9: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2665 /* 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
|
2666 * 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
|
2667 if (state->lastlist == l->id) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2668 return; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2669 state->lastlist = l->id; |
| 4444 | 2670 break; |
| 2671 | |
| 2672 default: | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2673 if (state->lastlist == l->id) |
| 4444 | 2674 { |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2675 /* 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
|
2676 * 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
|
2677 return; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2678 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2679 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2680 /* add the state to the list */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2681 state->lastlist = l->id; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2682 lastthread = &l->t[l->n++]; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2683 lastthread->state = state; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2684 lastthread->sub.in_use = sub->in_use; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2685 if (sub->in_use > 0) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2686 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2687 /* Copy the match start and end positions. */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2688 if (REG_MULTI) |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2689 mch_memmove(&lastthread->sub.list.multi[0], |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2690 &sub->list.multi[0], |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2691 sizeof(struct multipos) * sub->in_use); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2692 else |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2693 mch_memmove(&lastthread->sub.list.line[0], |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2694 &sub->list.line[0], |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2695 sizeof(struct linepos) * sub->in_use); |
| 4444 | 2696 } |
| 2697 } | |
| 2698 | |
| 2699 #ifdef ENABLE_LOG | |
| 2700 nfa_set_code(state->c); | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2701 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
|
2702 abs(state->id), state->c, code); |
| 4444 | 2703 #endif |
| 2704 switch (state->c) | |
| 2705 { | |
| 2706 case NFA_MATCH: | |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2707 nfa_match = TRUE; |
| 4444 | 2708 break; |
| 2709 | |
| 2710 case NFA_SPLIT: | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2711 addstate(l, state->out, sub, off); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2712 addstate(l, state->out1, sub, off); |
| 4444 | 2713 break; |
| 2714 | |
| 2715 #if 0 | |
| 2716 case NFA_END_NEG_RANGE: | |
| 2717 /* Nothing to handle here. nfa_regmatch() will take care of it */ | |
| 2718 break; | |
| 2719 | |
| 2720 case NFA_NOT: | |
| 2721 EMSG(_("E999: (NFA regexp internal error) Should not process NOT node !")); | |
| 2722 #ifdef ENABLE_LOG | |
| 2723 fprintf(f, "\n\n>>> E999: Added state NFA_NOT to a list ... Something went wrong ! Why wasn't it processed already? \n\n"); | |
| 2724 #endif | |
| 2725 break; | |
| 2726 | |
| 2727 case NFA_COMPOSING: | |
| 2728 /* nfa_regmatch() will match all the bytes of this composing char. */ | |
| 2729 break; | |
| 2730 #endif | |
| 2731 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2732 case NFA_SKIP_CHAR: |
| 4444 | 2733 case NFA_NOPEN: |
| 2734 case NFA_NCLOSE: | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2735 addstate(l, state->out, sub, off); |
| 4444 | 2736 break; |
| 2737 | |
| 2738 /* If this state is reached, then a recursive call of nfa_regmatch() | |
| 2739 * succeeded. the next call saves the found submatches in the | |
| 2740 * first state after the "invisible" branch. */ | |
| 2741 #if 0 | |
| 2742 case NFA_END_INVISIBLE: | |
| 2743 break; | |
| 2744 #endif | |
| 2745 | |
| 2746 case NFA_MOPEN + 0: | |
| 2747 case NFA_MOPEN + 1: | |
| 2748 case NFA_MOPEN + 2: | |
| 2749 case NFA_MOPEN + 3: | |
| 2750 case NFA_MOPEN + 4: | |
| 2751 case NFA_MOPEN + 5: | |
| 2752 case NFA_MOPEN + 6: | |
| 2753 case NFA_MOPEN + 7: | |
| 2754 case NFA_MOPEN + 8: | |
| 2755 case NFA_MOPEN + 9: | |
| 2756 case NFA_ZSTART: | |
| 2757 if (state->c == NFA_ZSTART) | |
| 2758 subidx = 0; | |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2759 else |
|
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2760 subidx = state->c - NFA_MOPEN; |
| 4444 | 2761 |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2762 /* 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
|
2763 * restore it when it was in use. Otherwise fill any gap. */ |
| 4444 | 2764 if (REG_MULTI) |
| 2765 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2766 if (subidx < sub->in_use) |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2767 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2768 save_lpos = sub->list.multi[subidx].start; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2769 save_in_use = -1; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2770 } |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2771 else |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2772 { |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2773 save_in_use = sub->in_use; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2774 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
|
2775 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2776 sub->list.multi[i].start.lnum = -1; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2777 sub->list.multi[i].end.lnum = -1; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2778 } |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2779 sub->in_use = subidx + 1; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2780 } |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2781 if (off == -1) |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2782 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2783 sub->list.multi[subidx].start.lnum = reglnum + 1; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2784 sub->list.multi[subidx].start.col = 0; |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2785 } |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2786 else |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2787 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2788 sub->list.multi[subidx].start.lnum = reglnum; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2789 sub->list.multi[subidx].start.col = |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2790 (colnr_T)(reginput - regline + off); |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2791 } |
| 4444 | 2792 } |
| 2793 else | |
| 2794 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2795 if (subidx < sub->in_use) |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2796 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2797 save_ptr = sub->list.line[subidx].start; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2798 save_in_use = -1; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2799 } |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2800 else |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2801 { |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2802 save_in_use = sub->in_use; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2803 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
|
2804 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2805 sub->list.line[i].start = NULL; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2806 sub->list.line[i].end = NULL; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2807 } |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2808 sub->in_use = subidx + 1; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2809 } |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2810 sub->list.line[subidx].start = reginput + off; |
| 4444 | 2811 } |
| 2812 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2813 addstate(l, state->out, sub, off); |
| 4444 | 2814 |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2815 if (save_in_use == -1) |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2816 { |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2817 if (REG_MULTI) |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2818 sub->list.multi[subidx].start = save_lpos; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2819 else |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2820 sub->list.line[subidx].start = save_ptr; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2821 } |
| 4444 | 2822 else |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2823 sub->in_use = save_in_use; |
| 4444 | 2824 break; |
| 2825 | |
| 2826 case NFA_MCLOSE + 0: | |
|
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
2827 if (nfa_has_zend) |
| 4444 | 2828 { |
|
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
2829 /* 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
|
2830 * 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
|
2831 addstate(l, state->out, sub, off); |
| 4444 | 2832 break; |
| 2833 } | |
| 2834 case NFA_MCLOSE + 1: | |
| 2835 case NFA_MCLOSE + 2: | |
| 2836 case NFA_MCLOSE + 3: | |
| 2837 case NFA_MCLOSE + 4: | |
| 2838 case NFA_MCLOSE + 5: | |
| 2839 case NFA_MCLOSE + 6: | |
| 2840 case NFA_MCLOSE + 7: | |
| 2841 case NFA_MCLOSE + 8: | |
| 2842 case NFA_MCLOSE + 9: | |
| 2843 case NFA_ZEND: | |
| 2844 if (state->c == NFA_ZEND) | |
| 2845 subidx = 0; | |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2846 else |
|
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2847 subidx = state->c - NFA_MCLOSE; |
| 4444 | 2848 |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2849 /* 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
|
2850 * has done that. */ |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2851 save_in_use = sub->in_use; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2852 if (sub->in_use <= subidx) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2853 sub->in_use = subidx + 1; |
| 4444 | 2854 if (REG_MULTI) |
| 2855 { | |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2856 save_lpos = sub->list.multi[subidx].end; |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2857 if (off == -1) |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2858 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2859 sub->list.multi[subidx].end.lnum = reglnum + 1; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2860 sub->list.multi[subidx].end.col = 0; |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2861 } |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2862 else |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2863 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2864 sub->list.multi[subidx].end.lnum = reglnum; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2865 sub->list.multi[subidx].end.col = |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2866 (colnr_T)(reginput - regline + off); |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2867 } |
| 4444 | 2868 } |
| 2869 else | |
| 2870 { | |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2871 save_ptr = sub->list.line[subidx].end; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2872 sub->list.line[subidx].end = reginput + off; |
| 4444 | 2873 } |
| 2874 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2875 addstate(l, state->out, sub, off); |
| 4444 | 2876 |
| 2877 if (REG_MULTI) | |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2878 sub->list.multi[subidx].end = save_lpos; |
| 4444 | 2879 else |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
2880 sub->list.line[subidx].end = save_ptr; |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2881 sub->in_use = save_in_use; |
| 4444 | 2882 break; |
| 2883 } | |
| 2884 } | |
| 2885 | |
| 2886 /* | |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2887 * 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
|
2888 * 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
|
2889 * 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
|
2890 * matters for alternatives. |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2891 */ |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2892 static void |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2893 addstate_here(l, state, sub, ip) |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2894 nfa_list_T *l; /* runtime state list */ |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2895 nfa_state_T *state; /* state to update */ |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2896 regsub_T *sub; /* pointers to subexpressions */ |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2897 int *ip; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2898 { |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2899 int tlen = l->n; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2900 int count; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2901 int i = *ip; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2902 |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2903 /* 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
|
2904 addstate(l, state, sub, 0); |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2905 |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2906 /* 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
|
2907 if (i + 1 == tlen) |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2908 return; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2909 |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2910 /* 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
|
2911 count = l->n - tlen; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2912 if (count > 1) |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2913 { |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2914 /* 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
|
2915 * end to the current position */ |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2916 mch_memmove(&(l->t[i + count]), |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2917 &(l->t[i + 1]), |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2918 sizeof(nfa_thread_T) * (l->n - i - 1)); |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2919 mch_memmove(&(l->t[i]), |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2920 &(l->t[l->n - 1]), |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2921 sizeof(nfa_thread_T) * count); |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2922 } |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2923 else |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2924 { |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2925 /* overwrite the current state */ |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2926 l->t[i] = l->t[l->n - 1]; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2927 } |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2928 --l->n; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2929 *ip = i - 1; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2930 } |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2931 |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2932 /* |
| 4444 | 2933 * Check character class "class" against current character c. |
| 2934 */ | |
| 2935 static int | |
| 2936 check_char_class(class, c) | |
| 2937 int class; | |
| 2938 int c; | |
| 2939 { | |
| 2940 switch (class) | |
| 2941 { | |
| 2942 case NFA_CLASS_ALNUM: | |
| 4464 | 2943 if (c >= 1 && c <= 255 && isalnum(c)) |
| 4444 | 2944 return OK; |
| 2945 break; | |
| 2946 case NFA_CLASS_ALPHA: | |
| 4464 | 2947 if (c >= 1 && c <= 255 && isalpha(c)) |
| 4444 | 2948 return OK; |
| 2949 break; | |
| 2950 case NFA_CLASS_BLANK: | |
| 2951 if (c == ' ' || c == '\t') | |
| 2952 return OK; | |
| 2953 break; | |
| 2954 case NFA_CLASS_CNTRL: | |
| 4464 | 2955 if (c >= 1 && c <= 255 && iscntrl(c)) |
| 4444 | 2956 return OK; |
| 2957 break; | |
| 2958 case NFA_CLASS_DIGIT: | |
| 2959 if (VIM_ISDIGIT(c)) | |
| 2960 return OK; | |
| 2961 break; | |
| 2962 case NFA_CLASS_GRAPH: | |
| 4464 | 2963 if (c >= 1 && c <= 255 && isgraph(c)) |
| 4444 | 2964 return OK; |
| 2965 break; | |
| 2966 case NFA_CLASS_LOWER: | |
| 2967 if (MB_ISLOWER(c)) | |
| 2968 return OK; | |
| 2969 break; | |
| 2970 case NFA_CLASS_PRINT: | |
| 2971 if (vim_isprintc(c)) | |
| 2972 return OK; | |
| 2973 break; | |
| 2974 case NFA_CLASS_PUNCT: | |
| 4464 | 2975 if (c >= 1 && c <= 255 && ispunct(c)) |
| 4444 | 2976 return OK; |
| 2977 break; | |
| 2978 case NFA_CLASS_SPACE: | |
| 2979 if ((c >=9 && c <= 13) || (c == ' ')) | |
| 2980 return OK; | |
| 2981 break; | |
| 2982 case NFA_CLASS_UPPER: | |
| 2983 if (MB_ISUPPER(c)) | |
| 2984 return OK; | |
| 2985 break; | |
| 2986 case NFA_CLASS_XDIGIT: | |
| 2987 if (vim_isxdigit(c)) | |
| 2988 return OK; | |
| 2989 break; | |
| 2990 case NFA_CLASS_TAB: | |
| 2991 if (c == '\t') | |
| 2992 return OK; | |
| 2993 break; | |
| 2994 case NFA_CLASS_RETURN: | |
| 2995 if (c == '\r') | |
| 2996 return OK; | |
| 2997 break; | |
| 2998 case NFA_CLASS_BACKSPACE: | |
| 2999 if (c == '\b') | |
| 3000 return OK; | |
| 3001 break; | |
| 3002 case NFA_CLASS_ESCAPE: | |
| 3003 if (c == '\033') | |
| 3004 return OK; | |
| 3005 break; | |
| 3006 | |
| 3007 default: | |
| 3008 /* should not be here :P */ | |
| 3009 EMSG_RET_FAIL(_("E877: (NFA regexp) Invalid character class ")); | |
| 3010 } | |
| 3011 return FAIL; | |
| 3012 } | |
| 3013 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3014 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
|
3015 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3016 /* |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3017 * Check for a match with subexpression "subidx". |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3018 * return TRUE if it matches. |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3019 */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3020 static int |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3021 match_backref(sub, subidx, bytelen) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3022 regsub_T *sub; /* pointers to subexpressions */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3023 int subidx; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3024 int *bytelen; /* out: length of match in bytes */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3025 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3026 int len; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3027 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3028 if (sub->in_use <= subidx) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3029 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3030 retempty: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3031 /* backref was not set, match an empty string */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3032 *bytelen = 0; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3033 return TRUE; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3034 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3035 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3036 if (REG_MULTI) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3037 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3038 if (sub->list.multi[subidx].start.lnum < 0 |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3039 || sub->list.multi[subidx].end.lnum < 0) |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3040 goto retempty; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3041 /* TODO: line breaks */ |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3042 len = sub->list.multi[subidx].end.col |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3043 - sub->list.multi[subidx].start.col; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3044 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
|
3045 reginput, &len) == 0) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3046 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3047 *bytelen = len; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3048 return TRUE; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3049 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3050 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3051 else |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3052 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3053 if (sub->list.line[subidx].start == NULL |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3054 || sub->list.line[subidx].end == NULL) |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3055 goto retempty; |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3056 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
|
3057 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
|
3058 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3059 *bytelen = len; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3060 return TRUE; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3061 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3062 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3063 return FALSE; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3064 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3065 |
| 4444 | 3066 /* |
| 3067 * Set all NFA nodes' list ID equal to -1. | |
| 3068 */ | |
| 3069 static void | |
| 3070 nfa_set_neg_listids(start) | |
| 3071 nfa_state_T *start; | |
| 3072 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3073 if (start != NULL && start->lastlist >= 0) |
| 4444 | 3074 { |
| 3075 start->lastlist = -1; | |
| 3076 nfa_set_neg_listids(start->out); | |
| 3077 nfa_set_neg_listids(start->out1); | |
| 3078 } | |
| 3079 } | |
| 3080 | |
| 3081 /* | |
| 3082 * Set all NFA nodes' list ID equal to 0. | |
| 3083 */ | |
| 3084 static void | |
| 3085 nfa_set_null_listids(start) | |
| 3086 nfa_state_T *start; | |
| 3087 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3088 if (start != NULL && start->lastlist == -1) |
| 4444 | 3089 { |
| 3090 start->lastlist = 0; | |
| 3091 nfa_set_null_listids(start->out); | |
| 3092 nfa_set_null_listids(start->out1); | |
| 3093 } | |
| 3094 } | |
| 3095 | |
| 3096 /* | |
| 3097 * Save list IDs for all NFA states in "list". | |
| 3098 */ | |
| 3099 static void | |
| 3100 nfa_save_listids(start, list) | |
| 3101 nfa_state_T *start; | |
| 3102 int *list; | |
| 3103 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3104 if (start != NULL && start->lastlist != -1) |
| 4444 | 3105 { |
| 3106 list[abs(start->id)] = start->lastlist; | |
| 3107 start->lastlist = -1; | |
| 3108 nfa_save_listids(start->out, list); | |
| 3109 nfa_save_listids(start->out1, list); | |
| 3110 } | |
| 3111 } | |
| 3112 | |
| 3113 /* | |
| 3114 * Restore list IDs from "list" to all NFA states. | |
| 3115 */ | |
| 3116 static void | |
| 3117 nfa_restore_listids(start, list) | |
| 3118 nfa_state_T *start; | |
| 3119 int *list; | |
| 3120 { | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3121 if (start != NULL && start->lastlist == -1) |
| 4444 | 3122 { |
| 3123 start->lastlist = list[abs(start->id)]; | |
| 3124 nfa_restore_listids(start->out, list); | |
| 3125 nfa_restore_listids(start->out1, list); | |
| 3126 } | |
| 3127 } | |
| 3128 | |
|
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3129 static int |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3130 nfa_re_num_cmp(val, op, pos) |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3131 long_u val; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3132 int op; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3133 long_u pos; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3134 { |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3135 if (op == 1) return pos > val; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3136 if (op == 2) return pos < val; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3137 return val == pos; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3138 } |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3139 |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3140 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
|
3141 |
| 4444 | 3142 /* |
| 3143 * Main matching routine. | |
| 3144 * | |
| 3145 * Run NFA to determine whether it matches reginput. | |
| 3146 * | |
| 3147 * Return TRUE if there is a match, FALSE otherwise. | |
| 3148 * Note: Caller must ensure that: start != NULL. | |
| 3149 */ | |
| 3150 static int | |
| 3151 nfa_regmatch(start, submatch, m) | |
| 3152 nfa_state_T *start; | |
| 3153 regsub_T *submatch; | |
| 3154 regsub_T *m; | |
| 3155 { | |
| 3156 int result; | |
| 3157 int size = 0; | |
| 3158 int flag = 0; | |
| 3159 int old_reglnum = -1; | |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3160 int go_to_nextline = FALSE; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3161 nfa_thread_T *t; |
| 4444 | 3162 char_u *old_reginput = NULL; |
| 3163 char_u *old_regline = NULL; | |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3164 nfa_list_T list[3]; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3165 nfa_list_T *listtbl[2][2]; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3166 nfa_list_T *ll; |
| 4444 | 3167 int listid = 1; |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3168 int listidx; |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3169 nfa_list_T *thislist; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3170 nfa_list_T *nextlist; |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3171 nfa_list_T *neglist; |
| 4444 | 3172 int *listids = NULL; |
| 4460 | 3173 #ifdef NFA_REGEXP_DEBUG_LOG |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3174 FILE *debug = fopen(NFA_REGEXP_DEBUG_LOG, "a"); |
| 4444 | 3175 |
| 3176 if (debug == NULL) | |
| 3177 { | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3178 EMSG2(_("(NFA) COULD NOT OPEN %s !"), NFA_REGEXP_DEBUG_LOG); |
| 4444 | 3179 return FALSE; |
| 3180 } | |
| 3181 #endif | |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3182 nfa_match = FALSE; |
| 4444 | 3183 |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3184 /* Allocate memory for the lists of nodes. */ |
|
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3185 size = (nstate + 1) * sizeof(nfa_thread_T); |
|
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3186 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
|
3187 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
|
3188 list[2].t = (nfa_thread_T *)lalloc(size, TRUE); |
| 4444 | 3189 if (list[0].t == NULL || list[1].t == NULL || list[2].t == NULL) |
| 3190 goto theend; | |
| 3191 vim_memset(list[0].t, 0, size); | |
| 3192 vim_memset(list[1].t, 0, size); | |
| 3193 vim_memset(list[2].t, 0, size); | |
| 3194 | |
| 3195 #ifdef ENABLE_LOG | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3196 log_fd = fopen(NFA_REGEXP_RUN_LOG, "a"); |
| 4444 | 3197 if (log_fd != NULL) |
| 3198 { | |
| 3199 fprintf(log_fd, "**********************************\n"); | |
| 3200 nfa_set_code(start->c); | |
| 3201 fprintf(log_fd, " RUNNING nfa_regmatch() starting with state %d, code %s\n", | |
| 3202 abs(start->id), code); | |
| 3203 fprintf(log_fd, "**********************************\n"); | |
| 3204 } | |
| 3205 else | |
| 3206 { | |
| 3207 EMSG(_("Could not open temporary log file for writing, displaying on stderr ... ")); | |
| 3208 log_fd = stderr; | |
| 3209 } | |
| 3210 #endif | |
| 3211 | |
| 3212 thislist = &list[0]; | |
| 3213 thislist->n = 0; | |
| 3214 nextlist = &list[1]; | |
| 3215 nextlist->n = 0; | |
| 3216 neglist = &list[2]; | |
| 3217 neglist->n = 0; | |
| 3218 #ifdef ENABLE_LOG | |
| 3219 fprintf(log_fd, "(---) STARTSTATE\n"); | |
| 3220 #endif | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3221 thislist->id = listid; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3222 addstate(thislist, start, m, 0); |
| 4444 | 3223 |
| 3224 /* There are two cases when the NFA advances: 1. input char matches the | |
| 3225 * NFA node and 2. input char does not match the NFA node, but the next | |
| 3226 * node is NFA_NOT. The following macro calls addstate() according to | |
| 3227 * these rules. It is used A LOT, so use the "listtbl" table for speed */ | |
| 3228 listtbl[0][0] = NULL; | |
| 3229 listtbl[0][1] = neglist; | |
| 3230 listtbl[1][0] = nextlist; | |
| 3231 listtbl[1][1] = NULL; | |
| 3232 #define ADD_POS_NEG_STATE(node) \ | |
| 3233 ll = listtbl[result ? 1 : 0][node->negated]; \ | |
| 3234 if (ll != NULL) \ | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3235 addstate(ll, node->out , &t->sub, clen); |
| 4444 | 3236 |
| 3237 | |
| 3238 /* | |
| 3239 * Run for each character. | |
| 3240 */ | |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3241 for (;;) |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3242 { |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3243 int curc; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3244 int clen; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3245 |
| 4444 | 3246 #ifdef FEAT_MBYTE |
| 3247 if (has_mbyte) | |
| 3248 { | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3249 curc = (*mb_ptr2char)(reginput); |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3250 clen = (*mb_ptr2len)(reginput); |
| 4444 | 3251 } |
| 3252 else | |
| 3253 #endif | |
| 3254 { | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3255 curc = *reginput; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3256 clen = 1; |
| 4444 | 3257 } |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3258 if (curc == NUL) |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3259 { |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3260 clen = 0; |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3261 go_to_nextline = FALSE; |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3262 } |
| 4444 | 3263 |
| 3264 /* swap lists */ | |
| 3265 thislist = &list[flag]; | |
| 3266 nextlist = &list[flag ^= 1]; | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3267 nextlist->n = 0; /* clear nextlist */ |
| 4444 | 3268 listtbl[1][0] = nextlist; |
| 3269 ++listid; | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3270 thislist->id = listid; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3271 nextlist->id = listid + 1; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3272 neglist->id = listid + 1; |
| 4444 | 3273 |
| 3274 #ifdef ENABLE_LOG | |
| 3275 fprintf(log_fd, "------------------------------------------\n"); | |
| 3276 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
|
3277 fprintf(log_fd, ">>> Advanced one character ... Current char is %c (code %d) \n", curc, (int)curc); |
| 4444 | 3278 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
|
3279 { |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3280 int i; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3281 |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3282 for (i = 0; i < thislist->n; i++) |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3283 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
|
3284 } |
| 4444 | 3285 fprintf(log_fd, "\n"); |
| 3286 #endif | |
| 3287 | |
| 4460 | 3288 #ifdef NFA_REGEXP_DEBUG_LOG |
| 4444 | 3289 fprintf(debug, "\n-------------------\n"); |
| 3290 #endif | |
| 4480 | 3291 /* |
| 3292 * If the state lists are empty we can stop. | |
| 3293 */ | |
| 3294 if (thislist->n == 0 && neglist->n == 0) | |
| 3295 break; | |
| 4444 | 3296 |
| 3297 /* compute nextlist */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3298 for (listidx = 0; listidx < thislist->n || neglist->n > 0; ++listidx) |
| 4444 | 3299 { |
| 3300 if (neglist->n > 0) | |
| 3301 { | |
| 3302 t = &neglist->t[0]; | |
| 4468 | 3303 neglist->n--; |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3304 listidx--; |
| 4444 | 3305 } |
| 3306 else | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3307 t = &thislist->t[listidx]; |
| 4444 | 3308 |
| 4460 | 3309 #ifdef NFA_REGEXP_DEBUG_LOG |
| 4444 | 3310 nfa_set_code(t->state->c); |
| 3311 fprintf(debug, "%s, ", code); | |
| 3312 #endif | |
| 3313 #ifdef ENABLE_LOG | |
| 3314 nfa_set_code(t->state->c); | |
| 3315 fprintf(log_fd, "(%d) %s, code %d ... \n", abs(t->state->id), | |
| 3316 code, (int)t->state->c); | |
| 3317 #endif | |
| 3318 | |
| 3319 /* | |
| 3320 * Handle the possible codes of the current state. | |
| 3321 * The most important is NFA_MATCH. | |
| 3322 */ | |
| 3323 switch (t->state->c) | |
| 3324 { | |
| 3325 case NFA_MATCH: | |
|
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3326 { |
|
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3327 int j; |
|
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3328 |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3329 nfa_match = TRUE; |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3330 submatch->in_use = t->sub.in_use; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3331 if (REG_MULTI) |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3332 for (j = 0; j < submatch->in_use; j++) |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3333 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3334 submatch->list.multi[j].start = |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3335 t->sub.list.multi[j].start; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3336 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
|
3337 } |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3338 else |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3339 for (j = 0; j < submatch->in_use; j++) |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3340 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3341 submatch->list.line[j].start = |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3342 t->sub.list.line[j].start; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3343 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
|
3344 } |
| 4444 | 3345 #ifdef ENABLE_LOG |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3346 log_subexpr(&t->sub); |
| 4444 | 3347 #endif |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3348 /* 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
|
3349 * 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
|
3350 * 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
|
3351 * correct. */ |
|
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3352 if (nextlist->n == 0 && neglist->n == 0) |
|
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3353 clen = 0; |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3354 goto nextchar; |
|
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3355 } |
| 4444 | 3356 |
| 3357 case NFA_END_INVISIBLE: | |
| 3358 /* This is only encountered after a NFA_START_INVISIBLE node. | |
| 3359 * They surround a zero-width group, used with "\@=" and "\&". | |
| 3360 * If we got here, it means that the current "invisible" group | |
| 3361 * finished successfully, so return control to the parent | |
| 3362 * nfa_regmatch(). Submatches are stored in *m, and used in | |
| 3363 * the parent call. */ | |
| 3364 if (start->c == NFA_MOPEN + 0) | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3365 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
| 4444 | 3366 else |
| 3367 { | |
| 3368 *m = t->sub; | |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3369 nfa_match = TRUE; |
| 4444 | 3370 } |
| 3371 break; | |
| 3372 | |
| 3373 case NFA_START_INVISIBLE: | |
| 3374 /* Save global variables, and call nfa_regmatch() to check if | |
| 3375 * the current concat matches at this position. The concat | |
| 3376 * ends with the node NFA_END_INVISIBLE */ | |
| 3377 old_reginput = reginput; | |
| 3378 old_regline = regline; | |
| 3379 old_reglnum = reglnum; | |
| 3380 if (listids == NULL) | |
| 3381 { | |
| 3382 listids = (int *) lalloc(sizeof(int) * nstate, TRUE); | |
| 3383 if (listids == NULL) | |
| 3384 { | |
| 3385 EMSG(_("E878: (NFA) Could not allocate memory for branch traversal!")); | |
| 3386 return 0; | |
| 3387 } | |
| 3388 } | |
| 3389 #ifdef ENABLE_LOG | |
| 3390 if (log_fd != stderr) | |
| 3391 fclose(log_fd); | |
| 3392 log_fd = NULL; | |
| 3393 #endif | |
| 3394 /* Have to clear the listid field of the NFA nodes, so that | |
| 3395 * nfa_regmatch() and addstate() can run properly after | |
| 3396 * recursion. */ | |
| 3397 nfa_save_listids(start, listids); | |
| 3398 nfa_set_null_listids(start); | |
| 3399 result = nfa_regmatch(t->state->out, submatch, m); | |
| 3400 nfa_set_neg_listids(start); | |
| 3401 nfa_restore_listids(start, listids); | |
| 3402 | |
| 3403 #ifdef ENABLE_LOG | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3404 log_fd = fopen(NFA_REGEXP_RUN_LOG, "a"); |
| 4444 | 3405 if (log_fd != NULL) |
| 3406 { | |
| 3407 fprintf(log_fd, "****************************\n"); | |
| 3408 fprintf(log_fd, "FINISHED RUNNING nfa_regmatch() recursively\n"); | |
| 3409 fprintf(log_fd, "MATCH = %s\n", result == TRUE ? "OK" : "FALSE"); | |
| 3410 fprintf(log_fd, "****************************\n"); | |
| 3411 } | |
| 3412 else | |
| 3413 { | |
| 3414 EMSG(_("Could not open temporary log file for writing, displaying on stderr ... ")); | |
| 3415 log_fd = stderr; | |
| 3416 } | |
| 3417 #endif | |
| 3418 if (result == TRUE) | |
| 3419 { | |
|
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3420 int j; |
|
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3421 |
| 4444 | 3422 /* Restore position in input text */ |
| 3423 reginput = old_reginput; | |
| 3424 regline = old_regline; | |
| 3425 reglnum = old_reglnum; | |
| 3426 /* Copy submatch info from the recursive call */ | |
| 3427 if (REG_MULTI) | |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3428 for (j = 1; j < m->in_use; j++) |
| 4444 | 3429 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3430 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
|
3431 t->sub.list.multi[j].end = m->list.multi[j].end; |
| 4444 | 3432 } |
| 3433 else | |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3434 for (j = 1; j < m->in_use; j++) |
| 4444 | 3435 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
3436 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
|
3437 t->sub.list.line[j].end = m->list.line[j].end; |
| 4444 | 3438 } |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3439 t->sub.in_use = m->in_use; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3440 |
| 4444 | 3441 /* 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
|
3442 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
|
3443 &listidx); |
| 4444 | 3444 } |
| 3445 else | |
| 3446 { | |
| 3447 /* continue with next input char */ | |
| 3448 reginput = old_reginput; | |
| 3449 } | |
| 3450 break; | |
| 3451 | |
| 3452 case NFA_BOL: | |
| 3453 if (reginput == regline) | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3454 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
| 4444 | 3455 break; |
| 3456 | |
| 3457 case NFA_EOL: | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3458 if (curc == NUL) |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3459 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
| 4444 | 3460 break; |
| 3461 | |
| 3462 case NFA_BOW: | |
| 3463 { | |
| 3464 int bow = TRUE; | |
| 3465 | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3466 if (curc == NUL) |
| 4444 | 3467 bow = FALSE; |
| 3468 #ifdef FEAT_MBYTE | |
| 3469 else if (has_mbyte) | |
| 3470 { | |
| 3471 int this_class; | |
| 3472 | |
| 3473 /* Get class of current and previous char (if it exists). */ | |
| 4503 | 3474 this_class = mb_get_class_buf(reginput, reg_buf); |
| 4444 | 3475 if (this_class <= 1) |
| 3476 bow = FALSE; | |
| 3477 else if (reg_prev_class() == this_class) | |
| 3478 bow = FALSE; | |
| 3479 } | |
| 3480 #endif | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3481 else if (!vim_iswordc_buf(curc, reg_buf) |
| 4503 | 3482 || (reginput > regline |
| 3483 && vim_iswordc_buf(reginput[-1], reg_buf))) | |
| 4444 | 3484 bow = FALSE; |
| 3485 if (bow) | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3486 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
| 4444 | 3487 break; |
| 3488 } | |
| 3489 | |
| 3490 case NFA_EOW: | |
| 3491 { | |
| 3492 int eow = TRUE; | |
| 3493 | |
| 3494 if (reginput == regline) | |
| 3495 eow = FALSE; | |
| 3496 #ifdef FEAT_MBYTE | |
| 3497 else if (has_mbyte) | |
| 3498 { | |
| 3499 int this_class, prev_class; | |
| 3500 | |
| 3501 /* Get class of current and previous char (if it exists). */ | |
| 4503 | 3502 this_class = mb_get_class_buf(reginput, reg_buf); |
| 4444 | 3503 prev_class = reg_prev_class(); |
| 3504 if (this_class == prev_class | |
| 3505 || prev_class == 0 || prev_class == 1) | |
| 3506 eow = FALSE; | |
| 3507 } | |
| 3508 #endif | |
| 4503 | 3509 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
|
3510 || (reginput[0] != NUL |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3511 && vim_iswordc_buf(curc, reg_buf))) |
| 4444 | 3512 eow = FALSE; |
| 3513 if (eow) | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3514 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
| 4444 | 3515 break; |
| 3516 } | |
| 3517 | |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3518 #ifdef FEAT_MBYTE |
| 4444 | 3519 case NFA_COMPOSING: |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3520 { |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3521 int mc = curc; |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3522 int len = 0; |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3523 nfa_state_T *end; |
|
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3524 nfa_state_T *sta; |
|
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3525 int cchars[MAX_MCO]; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3526 int ccount = 0; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3527 int j; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3528 |
| 4444 | 3529 sta = t->state->out; |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3530 len = 0; |
|
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3531 if (utf_iscomposing(sta->c)) |
|
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3532 { |
|
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3533 /* Only match composing character(s), ignore base |
|
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3534 * character. Used for ".{composing}" and "{composing}" |
|
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3535 * (no preceding character). */ |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3536 len += mb_char2len(mc); |
|
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3537 } |
|
4551
f0c37f5cf4ea
updated for version 7.3.1023
Bram Moolenaar <bram@vim.org>
parents:
4549
diff
changeset
|
3538 if (ireg_icombine && len == 0) |
| 4444 | 3539 { |
|
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3540 /* If \Z was present, then ignore composing characters. |
|
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3541 * 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
|
3542 /* TODO: How about negated? */ |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3543 if (len == 0 && sta->c != curc) |
|
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3544 result = FAIL; |
|
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3545 else |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3546 result = OK; |
|
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3547 while (sta->c != NFA_END_COMPOSING) |
|
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3548 sta = sta->out; |
| 4444 | 3549 } |
|
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3550 |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3551 /* Check base character matches first, unless ignored. */ |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3552 else if (len > 0 || mc == sta->c) |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3553 { |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3554 if (len == 0) |
|
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3555 { |
|
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3556 len += mb_char2len(mc); |
|
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3557 sta = sta->out; |
|
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3558 } |
| 4444 | 3559 |
|
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3560 /* 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
|
3561 * Get them into cchars[] first. */ |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3562 while (len < clen) |
|
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3563 { |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3564 mc = mb_ptr2char(reginput + len); |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3565 cchars[ccount++] = mc; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3566 len += mb_char2len(mc); |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3567 if (ccount == MAX_MCO) |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3568 break; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3569 } |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3570 |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3571 /* 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
|
3572 * 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
|
3573 * composing chars are matched. */ |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3574 result = OK; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3575 while (sta->c != NFA_END_COMPOSING) |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3576 { |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3577 for (j = 0; j < ccount; ++j) |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3578 if (cchars[j] == sta->c) |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3579 break; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3580 if (j == ccount) |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3581 { |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3582 result = FAIL; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3583 break; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3584 } |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3585 sta = sta->out; |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3586 } |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3587 } |
|
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3588 else |
|
4525
36ddcf4cecbc
updated for version 7.3.1010
Bram Moolenaar <bram@vim.org>
parents:
4517
diff
changeset
|
3589 result = FAIL; |
|
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3590 |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3591 end = t->state->out1; /* NFA_END_COMPOSING */ |
| 4444 | 3592 ADD_POS_NEG_STATE(end); |
| 3593 break; | |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3594 } |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3595 #endif |
| 4444 | 3596 |
| 3597 case NFA_NEWL: | |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3598 if (curc == NUL && !reg_line_lbr && REG_MULTI |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3599 && reglnum <= reg_maxline) |
| 4444 | 3600 { |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3601 go_to_nextline = TRUE; |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3602 /* 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
|
3603 * at the start of the next line. */ |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3604 addstate(nextlist, t->state->out, &t->sub, -1); |
| 4444 | 3605 } |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3606 else if (curc == '\n' && reg_line_lbr) |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3607 { |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3608 /* 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
|
3609 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
|
3610 } |
| 4444 | 3611 break; |
| 3612 | |
| 3613 case NFA_CLASS_ALNUM: | |
| 3614 case NFA_CLASS_ALPHA: | |
| 3615 case NFA_CLASS_BLANK: | |
| 3616 case NFA_CLASS_CNTRL: | |
| 3617 case NFA_CLASS_DIGIT: | |
| 3618 case NFA_CLASS_GRAPH: | |
| 3619 case NFA_CLASS_LOWER: | |
| 3620 case NFA_CLASS_PRINT: | |
| 3621 case NFA_CLASS_PUNCT: | |
| 3622 case NFA_CLASS_SPACE: | |
| 3623 case NFA_CLASS_UPPER: | |
| 3624 case NFA_CLASS_XDIGIT: | |
| 3625 case NFA_CLASS_TAB: | |
| 3626 case NFA_CLASS_RETURN: | |
| 3627 case NFA_CLASS_BACKSPACE: | |
| 3628 case NFA_CLASS_ESCAPE: | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3629 result = check_char_class(t->state->c, curc); |
| 4444 | 3630 ADD_POS_NEG_STATE(t->state); |
| 3631 break; | |
| 3632 | |
| 3633 case NFA_END_NEG_RANGE: | |
| 3634 /* This follows a series of negated nodes, like: | |
| 3635 * 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
|
3636 if (curc > 0) |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3637 addstate(nextlist, t->state->out, &t->sub, clen); |
| 4444 | 3638 break; |
| 3639 | |
| 3640 case NFA_ANY: | |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3641 /* 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
|
3642 if (curc > 0) |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3643 addstate(nextlist, t->state->out, &t->sub, clen); |
| 4444 | 3644 break; |
| 3645 | |
| 3646 /* | |
| 3647 * Character classes like \a for alpha, \d for digit etc. | |
| 3648 */ | |
| 3649 case NFA_IDENT: /* \i */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3650 result = vim_isIDc(curc); |
| 4444 | 3651 ADD_POS_NEG_STATE(t->state); |
| 3652 break; | |
| 3653 | |
| 3654 case NFA_SIDENT: /* \I */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3655 result = !VIM_ISDIGIT(curc) && vim_isIDc(curc); |
| 4444 | 3656 ADD_POS_NEG_STATE(t->state); |
| 3657 break; | |
| 3658 | |
| 3659 case NFA_KWORD: /* \k */ | |
| 4503 | 3660 result = vim_iswordp_buf(reginput, reg_buf); |
| 4444 | 3661 ADD_POS_NEG_STATE(t->state); |
| 3662 break; | |
| 3663 | |
| 3664 case NFA_SKWORD: /* \K */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3665 result = !VIM_ISDIGIT(curc) |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3666 && vim_iswordp_buf(reginput, reg_buf); |
| 4444 | 3667 ADD_POS_NEG_STATE(t->state); |
| 3668 break; | |
| 3669 | |
| 3670 case NFA_FNAME: /* \f */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3671 result = vim_isfilec(curc); |
| 4444 | 3672 ADD_POS_NEG_STATE(t->state); |
| 3673 break; | |
| 3674 | |
| 3675 case NFA_SFNAME: /* \F */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3676 result = !VIM_ISDIGIT(curc) && vim_isfilec(curc); |
| 4444 | 3677 ADD_POS_NEG_STATE(t->state); |
| 3678 break; | |
| 3679 | |
| 3680 case NFA_PRINT: /* \p */ | |
| 4470 | 3681 result = ptr2cells(reginput) == 1; |
| 4444 | 3682 ADD_POS_NEG_STATE(t->state); |
| 3683 break; | |
| 3684 | |
| 3685 case NFA_SPRINT: /* \P */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3686 result = !VIM_ISDIGIT(curc) && ptr2cells(reginput) == 1; |
| 4444 | 3687 ADD_POS_NEG_STATE(t->state); |
| 3688 break; | |
| 3689 | |
| 3690 case NFA_WHITE: /* \s */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3691 result = vim_iswhite(curc); |
| 4444 | 3692 ADD_POS_NEG_STATE(t->state); |
| 3693 break; | |
| 3694 | |
| 3695 case NFA_NWHITE: /* \S */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3696 result = curc != NUL && !vim_iswhite(curc); |
| 4444 | 3697 ADD_POS_NEG_STATE(t->state); |
| 3698 break; | |
| 3699 | |
| 3700 case NFA_DIGIT: /* \d */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3701 result = ri_digit(curc); |
| 4444 | 3702 ADD_POS_NEG_STATE(t->state); |
| 3703 break; | |
| 3704 | |
| 3705 case NFA_NDIGIT: /* \D */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3706 result = curc != NUL && !ri_digit(curc); |
| 4444 | 3707 ADD_POS_NEG_STATE(t->state); |
| 3708 break; | |
| 3709 | |
| 3710 case NFA_HEX: /* \x */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3711 result = ri_hex(curc); |
| 4444 | 3712 ADD_POS_NEG_STATE(t->state); |
| 3713 break; | |
| 3714 | |
| 3715 case NFA_NHEX: /* \X */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3716 result = curc != NUL && !ri_hex(curc); |
| 4444 | 3717 ADD_POS_NEG_STATE(t->state); |
| 3718 break; | |
| 3719 | |
| 3720 case NFA_OCTAL: /* \o */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3721 result = ri_octal(curc); |
| 4444 | 3722 ADD_POS_NEG_STATE(t->state); |
| 3723 break; | |
| 3724 | |
| 3725 case NFA_NOCTAL: /* \O */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3726 result = curc != NUL && !ri_octal(curc); |
| 4444 | 3727 ADD_POS_NEG_STATE(t->state); |
| 3728 break; | |
| 3729 | |
| 3730 case NFA_WORD: /* \w */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3731 result = ri_word(curc); |
| 4444 | 3732 ADD_POS_NEG_STATE(t->state); |
| 3733 break; | |
| 3734 | |
| 3735 case NFA_NWORD: /* \W */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3736 result = curc != NUL && !ri_word(curc); |
| 4444 | 3737 ADD_POS_NEG_STATE(t->state); |
| 3738 break; | |
| 3739 | |
| 3740 case NFA_HEAD: /* \h */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3741 result = ri_head(curc); |
| 4444 | 3742 ADD_POS_NEG_STATE(t->state); |
| 3743 break; | |
| 3744 | |
| 3745 case NFA_NHEAD: /* \H */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3746 result = curc != NUL && !ri_head(curc); |
| 4444 | 3747 ADD_POS_NEG_STATE(t->state); |
| 3748 break; | |
| 3749 | |
| 3750 case NFA_ALPHA: /* \a */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3751 result = ri_alpha(curc); |
| 4444 | 3752 ADD_POS_NEG_STATE(t->state); |
| 3753 break; | |
| 3754 | |
| 3755 case NFA_NALPHA: /* \A */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3756 result = curc != NUL && !ri_alpha(curc); |
| 4444 | 3757 ADD_POS_NEG_STATE(t->state); |
| 3758 break; | |
| 3759 | |
| 3760 case NFA_LOWER: /* \l */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3761 result = ri_lower(curc); |
| 4444 | 3762 ADD_POS_NEG_STATE(t->state); |
| 3763 break; | |
| 3764 | |
| 3765 case NFA_NLOWER: /* \L */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3766 result = curc != NUL && !ri_lower(curc); |
| 4444 | 3767 ADD_POS_NEG_STATE(t->state); |
| 3768 break; | |
| 3769 | |
| 3770 case NFA_UPPER: /* \u */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3771 result = ri_upper(curc); |
| 4444 | 3772 ADD_POS_NEG_STATE(t->state); |
| 3773 break; | |
| 3774 | |
| 3775 case NFA_NUPPER: /* \U */ | |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3776 result = curc != NUL && !ri_upper(curc); |
| 4444 | 3777 ADD_POS_NEG_STATE(t->state); |
| 3778 break; | |
| 3779 | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3780 case NFA_BACKREF1: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3781 case NFA_BACKREF2: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3782 case NFA_BACKREF3: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3783 case NFA_BACKREF4: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3784 case NFA_BACKREF5: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3785 case NFA_BACKREF6: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3786 case NFA_BACKREF7: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3787 case NFA_BACKREF8: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3788 case NFA_BACKREF9: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3789 /* \1 .. \9 */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3790 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3791 int subidx = t->state->c - NFA_BACKREF1 + 1; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3792 int bytelen; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3793 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3794 result = match_backref(&t->sub, subidx, &bytelen); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3795 if (result) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3796 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3797 if (bytelen == 0) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3798 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3799 /* 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
|
3800 * be used next */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3801 addstate_here(thislist, t->state->out, &t->sub, |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3802 &listidx); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3803 thislist->t[listidx + 1].count = 0; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3804 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3805 else if (bytelen <= clen) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3806 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3807 /* match current character, jump ahead to out of |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3808 * NFA_SKIP */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3809 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
|
3810 #ifdef ENABLE_LOG |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3811 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3812 #endif |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3813 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3814 else |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3815 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3816 /* skip ofer the matched characters, set character |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3817 * count in NFA_SKIP */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3818 addstate(nextlist, t->state->out, &t->sub, bytelen); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3819 nextlist->t[nextlist->n - 1].count = bytelen - clen; |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3820 #ifdef ENABLE_LOG |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3821 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3822 #endif |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3823 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3824 |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3825 } |
| 4482 | 3826 break; |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3827 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3828 case NFA_SKIP: |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3829 /* charater of previous matching \1 .. \9 */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3830 if (t->count - clen <= 0) |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3831 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3832 /* end of match, go to what follows */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3833 addstate(nextlist, t->state->out, &t->sub, clen); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3834 #ifdef ENABLE_LOG |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3835 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3836 #endif |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3837 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3838 else |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3839 { |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3840 /* add state again with decremented count */ |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3841 addstate(nextlist, t->state, &t->sub, 0); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3842 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
|
3843 #ifdef ENABLE_LOG |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3844 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3845 #endif |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3846 } |
|
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3847 break; |
| 4482 | 3848 |
| 3849 case NFA_SKIP_CHAR: | |
| 3850 case NFA_ZSTART: | |
|
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
3851 case NFA_ZEND: |
| 4482 | 3852 /* TODO: should not happen? */ |
| 3853 break; | |
| 3854 | |
|
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3855 case NFA_LNUM: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3856 case NFA_LNUM_GT: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3857 case NFA_LNUM_LT: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3858 result = (REG_MULTI && |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3859 nfa_re_num_cmp(t->state->val, t->state->c - NFA_LNUM, |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3860 (long_u)(reglnum + reg_firstlnum))); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3861 if (result) |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3862 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3863 break; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3864 |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3865 case NFA_COL: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3866 case NFA_COL_GT: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3867 case NFA_COL_LT: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3868 result = nfa_re_num_cmp(t->state->val, t->state->c - NFA_COL, |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3869 (long_u)(reginput - regline) + 1); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3870 if (result) |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3871 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3872 break; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3873 |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3874 case NFA_VCOL: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3875 case NFA_VCOL_GT: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3876 case NFA_VCOL_LT: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3877 result = nfa_re_num_cmp(t->state->val, t->state->c - NFA_VCOL, |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3878 (long_u)win_linetabsize( |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3879 reg_win == NULL ? curwin : reg_win, |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3880 regline, (colnr_T)(reginput - regline)) + 1); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3881 if (result) |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3882 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3883 break; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3884 |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3885 case NFA_CURSOR: |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3886 result = (reg_win != NULL |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3887 && (reglnum + reg_firstlnum == reg_win->w_cursor.lnum) |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3888 && ((colnr_T)(reginput - regline) |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3889 == reg_win->w_cursor.col)); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3890 if (result) |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3891 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3892 break; |
|
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3893 |
| 4444 | 3894 default: /* regular character */ |
|
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3895 { |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3896 int c = t->state->c; |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3897 |
| 4482 | 3898 /* TODO: put this in #ifdef later */ |
|
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3899 if (c < -256) |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3900 EMSGN("INTERNAL: Negative state char: %ld", c); |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3901 if (is_Magic(c)) |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3902 c = un_Magic(c); |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3903 result = (c == curc); |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3904 |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3905 if (!result && ireg_ic) |
|
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3906 result = MB_TOLOWER(c) == MB_TOLOWER(curc); |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3907 #ifdef FEAT_MBYTE |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3908 /* 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
|
3909 * 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
|
3910 * character uses NFA_COMPOSING above. */ |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3911 if (result && enc_utf8 && !ireg_icombine |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3912 && clen != utf_char2len(curc)) |
|
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3913 result = FALSE; |
|
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3914 #endif |
| 4444 | 3915 ADD_POS_NEG_STATE(t->state); |
| 3916 break; | |
|
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3917 } |
| 4444 | 3918 } |
| 3919 | |
| 3920 } /* for (thislist = thislist; thislist->state; thislist++) */ | |
| 3921 | |
|
4557
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
3922 /* 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
|
3923 * start state to the list of states. |
|
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
3924 * 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
|
3925 * matters! |
|
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
3926 * 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
|
3927 * 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
|
3928 * 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
|
3929 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
|
3930 && reglnum == 0 && clen != 0 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
3931 && (ireg_maxcol == 0 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4577
diff
changeset
|
3932 || (colnr_T)(reginput - regline) < ireg_maxcol)) |
| 4444 | 3933 { |
| 3934 #ifdef ENABLE_LOG | |
| 3935 fprintf(log_fd, "(---) STARTSTATE\n"); | |
| 3936 #endif | |
|
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3937 addstate(nextlist, start, m, clen); |
| 4444 | 3938 } |
| 3939 | |
| 3940 #ifdef ENABLE_LOG | |
| 3941 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
|
3942 { |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3943 int i; |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3944 |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3945 for (i = 0; i < thislist->n; i++) |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3946 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
|
3947 } |
| 4444 | 3948 fprintf(log_fd, "\n"); |
| 3949 #endif | |
| 3950 | |
| 3951 nextchar: | |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3952 /* 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
|
3953 * finish. */ |
|
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3954 if (clen != 0) |
|
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3955 reginput += clen; |
|
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3956 else if (go_to_nextline) |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3957 reg_nextline(); |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3958 else |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3959 break; |
|
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3960 } |
| 4444 | 3961 |
| 3962 #ifdef ENABLE_LOG | |
| 3963 if (log_fd != stderr) | |
| 3964 fclose(log_fd); | |
| 3965 log_fd = NULL; | |
| 3966 #endif | |
| 3967 | |
| 3968 theend: | |
| 3969 /* Free memory */ | |
| 3970 vim_free(list[0].t); | |
| 3971 vim_free(list[1].t); | |
| 3972 vim_free(list[2].t); | |
| 3973 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
|
3974 vim_free(listids); |
| 4444 | 3975 #undef ADD_POS_NEG_STATE |
| 4460 | 3976 #ifdef NFA_REGEXP_DEBUG_LOG |
| 4444 | 3977 fclose(debug); |
| 3978 #endif | |
| 3979 | |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3980 return nfa_match; |
| 4444 | 3981 } |
| 3982 | |
| 3983 /* | |
| 3984 * Try match of "prog" with at regline["col"]. | |
| 3985 * Returns 0 for failure, number of lines contained in the match otherwise. | |
| 3986 */ | |
| 3987 static long | |
| 3988 nfa_regtry(start, col) | |
| 3989 nfa_state_T *start; | |
| 3990 colnr_T col; | |
| 3991 { | |
| 3992 int i; | |
| 3993 regsub_T sub, m; | |
| 3994 #ifdef ENABLE_LOG | |
| 3995 FILE *f; | |
| 3996 #endif | |
| 3997 | |
| 3998 reginput = regline + col; | |
| 3999 need_clear_subexpr = TRUE; | |
| 4000 | |
| 4001 #ifdef ENABLE_LOG | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
4002 f = fopen(NFA_REGEXP_RUN_LOG, "a"); |
| 4444 | 4003 if (f != NULL) |
| 4004 { | |
| 4005 fprintf(f, "\n\n\n\n\n\n\t\t=======================================================\n"); | |
| 4006 fprintf(f, " =======================================================\n"); | |
| 4007 #ifdef DEBUG | |
| 4008 fprintf(f, "\tRegexp is \"%s\"\n", nfa_regengine.expr); | |
| 4009 #endif | |
| 4010 fprintf(f, "\tInput text is \"%s\" \n", reginput); | |
| 4011 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
|
4012 nfa_print_state(f, start); |
| 4444 | 4013 fprintf(f, "\n\n"); |
| 4014 fclose(f); | |
| 4015 } | |
| 4016 else | |
| 4017 EMSG(_("Could not open temporary log file for writing ")); | |
| 4018 #endif | |
| 4019 | |
| 4020 if (REG_MULTI) | |
| 4021 { | |
| 4022 /* Use 0xff to set lnum to -1 */ | |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4023 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
|
4024 vim_memset(m.list.multi, 0xff, sizeof(struct multipos) * nfa_nsubexpr); |
| 4444 | 4025 } |
| 4026 else | |
| 4027 { | |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4028 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
|
4029 vim_memset(m.list.line, 0, sizeof(struct linepos) * nfa_nsubexpr); |
| 4444 | 4030 } |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4031 sub.in_use = 0; |
|
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4032 m.in_use = 0; |
| 4444 | 4033 |
| 4034 if (nfa_regmatch(start, &sub, &m) == FALSE) | |
| 4035 return 0; | |
| 4036 | |
| 4037 cleanup_subexpr(); | |
| 4038 if (REG_MULTI) | |
| 4039 { | |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4040 for (i = 0; i < sub.in_use; i++) |
| 4444 | 4041 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4042 reg_startpos[i] = sub.list.multi[i].start; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4043 reg_endpos[i] = sub.list.multi[i].end; |
| 4444 | 4044 } |
| 4045 | |
| 4046 if (reg_startpos[0].lnum < 0) | |
| 4047 { | |
| 4048 reg_startpos[0].lnum = 0; | |
| 4049 reg_startpos[0].col = col; | |
| 4050 } | |
| 4051 if (reg_endpos[0].lnum < 0) | |
| 4052 { | |
|
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
4053 /* pattern has a \ze but it didn't match, use current end */ |
| 4444 | 4054 reg_endpos[0].lnum = reglnum; |
| 4055 reg_endpos[0].col = (int)(reginput - regline); | |
| 4056 } | |
| 4057 else | |
| 4058 /* Use line number of "\ze". */ | |
| 4059 reglnum = reg_endpos[0].lnum; | |
| 4060 } | |
| 4061 else | |
| 4062 { | |
|
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
4063 for (i = 0; i < sub.in_use; i++) |
| 4444 | 4064 { |
|
4577
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4065 reg_startp[i] = sub.list.line[i].start; |
|
b22bff1a6af8
updated for version 7.3.1036
Bram Moolenaar <bram@vim.org>
parents:
4573
diff
changeset
|
4066 reg_endp[i] = sub.list.line[i].end; |
| 4444 | 4067 } |
| 4068 | |
| 4069 if (reg_startp[0] == NULL) | |
| 4070 reg_startp[0] = regline + col; | |
| 4071 if (reg_endp[0] == NULL) | |
| 4072 reg_endp[0] = reginput; | |
| 4073 } | |
| 4074 | |
| 4075 return 1 + reglnum; | |
| 4076 } | |
| 4077 | |
| 4078 /* | |
| 4079 * Match a regexp against a string ("line" points to the string) or multiple | |
| 4080 * lines ("line" is NULL, use reg_getline()). | |
| 4081 * | |
| 4082 * Returns 0 for failure, number of lines contained in the match otherwise. | |
| 4083 */ | |
| 4084 static long | |
| 4085 nfa_regexec_both(line, col) | |
| 4086 char_u *line; | |
| 4087 colnr_T col; /* column to start looking for match */ | |
| 4088 { | |
| 4089 nfa_regprog_T *prog; | |
| 4090 long retval = 0L; | |
| 4091 int i; | |
| 4092 | |
| 4093 if (REG_MULTI) | |
| 4094 { | |
| 4095 prog = (nfa_regprog_T *)reg_mmatch->regprog; | |
| 4096 line = reg_getline((linenr_T)0); /* relative to the cursor */ | |
| 4097 reg_startpos = reg_mmatch->startpos; | |
| 4098 reg_endpos = reg_mmatch->endpos; | |
| 4099 } | |
| 4100 else | |
| 4101 { | |
| 4102 prog = (nfa_regprog_T *)reg_match->regprog; | |
| 4103 reg_startp = reg_match->startp; | |
| 4104 reg_endp = reg_match->endp; | |
| 4105 } | |
| 4106 | |
| 4107 /* Be paranoid... */ | |
| 4108 if (prog == NULL || line == NULL) | |
| 4109 { | |
| 4110 EMSG(_(e_null)); | |
| 4111 goto theend; | |
| 4112 } | |
| 4113 | |
| 4114 /* If the start column is past the maximum column: no need to try. */ | |
| 4115 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
| 4116 goto theend; | |
| 4117 | |
| 4118 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */ | |
| 4119 if (prog->regflags & RF_ICASE) | |
| 4120 ireg_ic = TRUE; | |
| 4121 else if (prog->regflags & RF_NOICASE) | |
| 4122 ireg_ic = FALSE; | |
| 4123 | |
| 4124 #ifdef FEAT_MBYTE | |
| 4125 /* If pattern contains "\Z" overrule value of ireg_icombine */ | |
| 4126 if (prog->regflags & RF_ICOMBINE) | |
| 4127 ireg_icombine = TRUE; | |
| 4128 #endif | |
| 4129 | |
| 4130 regline = line; | |
| 4131 reglnum = 0; /* relative to line */ | |
| 4132 | |
|
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4133 nfa_has_zend = prog->has_zend; |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4134 nfa_nsubexpr = prog->nsubexp; |
|
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4135 |
| 4444 | 4136 nstate = prog->nstate; |
| 4137 for (i = 0; i < nstate; ++i) | |
| 4138 { | |
| 4139 prog->state[i].id = i; | |
| 4140 prog->state[i].lastlist = 0; | |
| 4141 } | |
| 4142 | |
| 4143 retval = nfa_regtry(prog->start, col); | |
| 4144 | |
| 4145 theend: | |
| 4146 return retval; | |
| 4147 } | |
| 4148 | |
| 4149 /* | |
| 4150 * Compile a regular expression into internal code for the NFA matcher. | |
| 4151 * Returns the program in allocated space. Returns NULL for an error. | |
| 4152 */ | |
| 4153 static regprog_T * | |
| 4154 nfa_regcomp(expr, re_flags) | |
| 4155 char_u *expr; | |
| 4156 int re_flags; | |
| 4157 { | |
|
4541
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4158 nfa_regprog_T *prog = NULL; |
| 4458 | 4159 size_t prog_size; |
| 4444 | 4160 int *postfix; |
| 4161 | |
| 4162 if (expr == NULL) | |
| 4163 return NULL; | |
| 4164 | |
| 4165 #ifdef DEBUG | |
| 4166 nfa_regengine.expr = expr; | |
| 4167 #endif | |
| 4168 | |
| 4169 init_class_tab(); | |
| 4170 | |
| 4171 if (nfa_regcomp_start(expr, re_flags) == FAIL) | |
| 4172 return NULL; | |
| 4173 | |
| 4174 /* 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
|
4175 * (and count its size). */ |
| 4444 | 4176 postfix = re2post(); |
| 4177 if (postfix == NULL) | |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4178 { |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4179 /* TODO: only give this error for debugging? */ |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4180 if (post_ptr >= post_end) |
|
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4181 EMSGN("Internal error: estimated max number of states insufficient: %ld", post_end - post_start); |
| 4444 | 4182 goto fail; /* Cascaded (syntax?) error */ |
|
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4183 } |
| 4444 | 4184 |
| 4185 /* | |
| 4186 * In order to build the NFA, we parse the input regexp twice: | |
| 4187 * 1. first pass to count size (so we can allocate space) | |
| 4188 * 2. second to emit code | |
| 4189 */ | |
| 4190 #ifdef ENABLE_LOG | |
| 4191 { | |
|
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
4192 FILE *f = fopen(NFA_REGEXP_RUN_LOG, "a"); |
| 4444 | 4193 |
| 4194 if (f != NULL) | |
| 4195 { | |
| 4196 fprintf(f, "\n*****************************\n\n\n\n\tCompiling regexp \"%s\" ... hold on !\n", expr); | |
| 4197 fclose(f); | |
| 4198 } | |
| 4199 } | |
| 4200 #endif | |
| 4201 | |
| 4202 /* | |
| 4203 * PASS 1 | |
| 4204 * Count number of NFA states in "nstate". Do not build the NFA. | |
| 4205 */ | |
| 4206 post2nfa(postfix, post_ptr, TRUE); | |
|
4541
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4207 |
|
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4208 /* Space for compiled regexp */ |
|
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4209 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
|
4210 prog = (nfa_regprog_T *)lalloc(prog_size, TRUE); |
|
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4211 if (prog == NULL) |
|
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4212 goto fail; |
|
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4213 vim_memset(prog, 0, prog_size); |
| 4444 | 4214 state_ptr = prog->state; |
| 4215 | |
| 4216 /* | |
| 4217 * PASS 2 | |
| 4218 * Build the NFA | |
| 4219 */ | |
| 4220 prog->start = post2nfa(postfix, post_ptr, FALSE); | |
| 4221 if (prog->start == NULL) | |
| 4222 goto fail; | |
| 4223 | |
| 4224 prog->regflags = regflags; | |
| 4225 prog->engine = &nfa_regengine; | |
| 4226 prog->nstate = nstate; | |
|
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4227 prog->has_zend = nfa_has_zend; |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4228 prog->nsubexp = regnpar; |
| 4444 | 4229 #ifdef ENABLE_LOG |
| 4230 nfa_postfix_dump(expr, OK); | |
| 4231 nfa_dump(prog); | |
| 4232 #endif | |
| 4233 | |
| 4234 out: | |
| 4235 vim_free(post_start); | |
| 4236 post_start = post_ptr = post_end = NULL; | |
| 4237 state_ptr = NULL; | |
| 4238 return (regprog_T *)prog; | |
| 4239 | |
| 4240 fail: | |
| 4241 vim_free(prog); | |
| 4242 prog = NULL; | |
| 4243 #ifdef ENABLE_LOG | |
| 4244 nfa_postfix_dump(expr, FAIL); | |
| 4245 #endif | |
| 4246 #ifdef DEBUG | |
| 4247 nfa_regengine.expr = NULL; | |
| 4248 #endif | |
| 4249 goto out; | |
| 4250 } | |
| 4251 | |
| 4252 | |
| 4253 /* | |
| 4254 * Match a regexp against a string. | |
| 4255 * "rmp->regprog" is a compiled regexp as returned by nfa_regcomp(). | |
| 4256 * Uses curbuf for line count and 'iskeyword'. | |
| 4257 * | |
| 4258 * Return TRUE if there is a match, FALSE if not. | |
| 4259 */ | |
| 4260 static int | |
| 4261 nfa_regexec(rmp, line, col) | |
| 4262 regmatch_T *rmp; | |
| 4263 char_u *line; /* string to match against */ | |
| 4264 colnr_T col; /* column to start looking for match */ | |
| 4265 { | |
| 4266 reg_match = rmp; | |
| 4267 reg_mmatch = NULL; | |
| 4268 reg_maxline = 0; | |
| 4269 reg_line_lbr = FALSE; | |
| 4270 reg_buf = curbuf; | |
| 4271 reg_win = NULL; | |
| 4272 ireg_ic = rmp->rm_ic; | |
| 4273 #ifdef FEAT_MBYTE | |
| 4274 ireg_icombine = FALSE; | |
| 4275 #endif | |
| 4276 ireg_maxcol = 0; | |
| 4277 return (nfa_regexec_both(line, col) != 0); | |
| 4278 } | |
| 4279 | |
| 4280 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
| 4281 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
| 4282 | |
| 4283 static int nfa_regexec_nl __ARGS((regmatch_T *rmp, char_u *line, colnr_T col)); | |
| 4284 | |
| 4285 /* | |
| 4286 * Like nfa_regexec(), but consider a "\n" in "line" to be a line break. | |
| 4287 */ | |
| 4288 static int | |
| 4289 nfa_regexec_nl(rmp, line, col) | |
| 4290 regmatch_T *rmp; | |
| 4291 char_u *line; /* string to match against */ | |
| 4292 colnr_T col; /* column to start looking for match */ | |
| 4293 { | |
| 4294 reg_match = rmp; | |
| 4295 reg_mmatch = NULL; | |
| 4296 reg_maxline = 0; | |
| 4297 reg_line_lbr = TRUE; | |
| 4298 reg_buf = curbuf; | |
| 4299 reg_win = NULL; | |
| 4300 ireg_ic = rmp->rm_ic; | |
| 4301 #ifdef FEAT_MBYTE | |
| 4302 ireg_icombine = FALSE; | |
| 4303 #endif | |
| 4304 ireg_maxcol = 0; | |
| 4305 return (nfa_regexec_both(line, col) != 0); | |
| 4306 } | |
| 4307 #endif | |
| 4308 | |
| 4309 | |
| 4310 /* | |
| 4311 * Match a regexp against multiple lines. | |
| 4312 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 4313 * Uses curbuf for line count and 'iskeyword'. | |
| 4314 * | |
| 4315 * Return zero if there is no match. Return number of lines contained in the | |
| 4316 * match otherwise. | |
| 4317 * | |
| 4318 * Note: the body is the same as bt_regexec() except for nfa_regexec_both() | |
| 4319 * | |
| 4320 * ! Also NOTE : match may actually be in another line. e.g.: | |
| 4321 * when r.e. is \nc, cursor is at 'a' and the text buffer looks like | |
| 4322 * | |
| 4323 * +-------------------------+ | |
| 4324 * |a | | |
| 4325 * |b | | |
| 4326 * |c | | |
| 4327 * | | | |
| 4328 * +-------------------------+ | |
| 4329 * | |
| 4330 * then nfa_regexec_multi() returns 3. while the original | |
| 4331 * vim_regexec_multi() returns 0 and a second call at line 2 will return 2. | |
| 4332 * | |
| 4333 * FIXME if this behavior is not compatible. | |
| 4334 */ | |
| 4335 static long | |
| 4336 nfa_regexec_multi(rmp, win, buf, lnum, col, tm) | |
| 4337 regmmatch_T *rmp; | |
| 4338 win_T *win; /* window in which to search or NULL */ | |
| 4339 buf_T *buf; /* buffer in which to search */ | |
| 4340 linenr_T lnum; /* nr of line to start looking for match */ | |
| 4341 colnr_T col; /* column to start looking for match */ | |
| 4342 proftime_T *tm UNUSED; /* timeout limit or NULL */ | |
| 4343 { | |
| 4344 reg_match = NULL; | |
| 4345 reg_mmatch = rmp; | |
| 4346 reg_buf = buf; | |
| 4347 reg_win = win; | |
| 4348 reg_firstlnum = lnum; | |
| 4349 reg_maxline = reg_buf->b_ml.ml_line_count - lnum; | |
| 4350 reg_line_lbr = FALSE; | |
| 4351 ireg_ic = rmp->rmm_ic; | |
| 4352 #ifdef FEAT_MBYTE | |
| 4353 ireg_icombine = FALSE; | |
| 4354 #endif | |
| 4355 ireg_maxcol = rmp->rmm_maxcol; | |
| 4356 | |
| 4503 | 4357 return nfa_regexec_both(NULL, col); |
| 4444 | 4358 } |
| 4359 | |
| 4360 #ifdef DEBUG | |
| 4361 # undef ENABLE_LOG | |
| 4362 #endif |