Mercurial > vim
annotate src/regexp.h @ 31965:b6a1e17f049d
Added tag v9.0.1314 for changeset c0a9bc376b54d4a62cd485c983ec34fe03fd7352
| author | Bram Moolenaar <Bram@vim.org> |
|---|---|
| date | Thu, 16 Feb 2023 16:15:05 +0100 |
| parents | 7fb4e244b16e |
| children | 90063f44c99a |
| rev | line source |
|---|---|
|
10042
4aead6a9b7a9
commit https://github.com/vim/vim/commit/edf3f97ae2af024708ebb4ac614227327033ca47
Christian Brabandt <cb@256bit.org>
parents:
6328
diff
changeset
|
1 /* vi:set ts=8 sts=4 sw=4 noet: |
| 7 | 2 * |
| 3 * NOTICE NOTICE NOTICE NOTICE NOTICE NOTICE NOTICE NOTICE NOTICE NOTICE | |
| 4 * | |
| 5 * This is NOT the original regular expression code as written by Henry | |
| 6 * Spencer. This code has been modified specifically for use with Vim, and | |
| 7 * should not be used apart from compiling Vim. If you want a good regular | |
| 8 * expression library, get the original code. | |
| 9 * | |
| 10 * NOTICE NOTICE NOTICE NOTICE NOTICE NOTICE NOTICE NOTICE NOTICE NOTICE | |
| 11 */ | |
| 12 | |
| 13 #ifndef _REGEXP_H | |
| 14 #define _REGEXP_H | |
| 15 | |
| 16 /* | |
| 17 * The number of sub-matches is limited to 10. | |
| 18 * The first one (index 0) is the whole match, referenced with "\0". | |
| 19 * The second one (index 1) is the first sub-match, referenced with "\1". | |
| 20 * This goes up to the tenth (index 9), referenced with "\9". | |
| 21 */ | |
| 22 #define NSUBEXP 10 | |
| 23 | |
| 24 /* | |
| 4444 | 25 * In the NFA engine: how many braces are allowed. |
| 26 * TODO(RE): Use dynamic memory allocation instead of static, like here | |
| 27 */ | |
| 28 #define NFA_MAX_BRACES 20 | |
| 29 | |
| 6328 | 30 /* |
| 31 * In the NFA engine: how many states are allowed | |
| 32 */ | |
| 33 #define NFA_MAX_STATES 100000 | |
|
28226
89c181c99e23
patch 8.2.4639: not sufficient parenthesis in preprocessor macros
Bram Moolenaar <Bram@vim.org>
parents:
18753
diff
changeset
|
34 #define NFA_TOO_EXPENSIVE (-1) |
| 6328 | 35 |
|
18753
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
36 // Which regexp engine to use? Needed for vim_regcomp(). |
|
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
37 // Must match with 'regexpengine'. |
| 6328 | 38 #define AUTOMATIC_ENGINE 0 |
| 39 #define BACKTRACKING_ENGINE 1 | |
| 40 #define NFA_ENGINE 2 | |
| 41 | |
| 4444 | 42 typedef struct regengine regengine_T; |
| 43 | |
| 44 /* | |
| 7 | 45 * Structure returned by vim_regcomp() to pass on to vim_regexec(). |
| 4444 | 46 * This is the general structure. For the actual matcher, two specific |
| 47 * structures are used. See code below. | |
| 48 */ | |
| 49 typedef struct regprog | |
| 50 { | |
| 51 regengine_T *engine; | |
| 52 unsigned regflags; | |
|
14354
ffd834f893aa
patch 8.1.0192: executing regexp recursively fails with a crash
Christian Brabandt <cb@256bit.org>
parents:
11521
diff
changeset
|
53 unsigned re_engine; // automatic, backtracking or nfa engine |
|
ffd834f893aa
patch 8.1.0192: executing regexp recursively fails with a crash
Christian Brabandt <cb@256bit.org>
parents:
11521
diff
changeset
|
54 unsigned re_flags; // second argument for vim_regcomp() |
|
ffd834f893aa
patch 8.1.0192: executing regexp recursively fails with a crash
Christian Brabandt <cb@256bit.org>
parents:
11521
diff
changeset
|
55 int re_in_use; // prog is being executed |
| 4444 | 56 } regprog_T; |
| 57 | |
| 58 /* | |
| 59 * Structure used by the back track matcher. | |
| 7 | 60 * These fields are only to be used in regexp.c! |
| 4444 | 61 * See regexp.c for an explanation. |
| 7 | 62 */ |
| 63 typedef struct | |
| 64 { | |
|
18753
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
65 // These four members implement regprog_T |
| 4444 | 66 regengine_T *engine; |
| 67 unsigned regflags; | |
| 6328 | 68 unsigned re_engine; |
|
14354
ffd834f893aa
patch 8.1.0192: executing regexp recursively fails with a crash
Christian Brabandt <cb@256bit.org>
parents:
11521
diff
changeset
|
69 unsigned re_flags; |
|
ffd834f893aa
patch 8.1.0192: executing regexp recursively fails with a crash
Christian Brabandt <cb@256bit.org>
parents:
11521
diff
changeset
|
70 int re_in_use; |
| 4444 | 71 |
| 7 | 72 int regstart; |
| 73 char_u reganch; | |
| 74 char_u *regmust; | |
| 75 int regmlen; | |
|
4686
8db697ae406a
updated for version 7.3.1090
Bram Moolenaar <bram@vim.org>
parents:
4647
diff
changeset
|
76 #ifdef FEAT_SYN_HL |
| 4444 | 77 char_u reghasz; |
|
4686
8db697ae406a
updated for version 7.3.1090
Bram Moolenaar <bram@vim.org>
parents:
4647
diff
changeset
|
78 #endif |
|
18753
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
79 char_u program[1]; // actually longer.. |
| 4444 | 80 } bt_regprog_T; |
| 81 | |
| 82 /* | |
| 83 * Structure representing a NFA state. | |
|
16378
3d6b282e2d6e
patch 8.1.1194: typos and small problems in source files
Bram Moolenaar <Bram@vim.org>
parents:
14354
diff
changeset
|
84 * An NFA state may have no outgoing edge, when it is a NFA_MATCH state. |
| 4444 | 85 */ |
| 86 typedef struct nfa_state nfa_state_T; | |
| 87 struct nfa_state | |
| 88 { | |
| 89 int c; | |
| 90 nfa_state_T *out; | |
| 91 nfa_state_T *out1; | |
| 92 int id; | |
|
18753
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
93 int lastlist[2]; // 0: normal, 1: recursive |
|
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
94 int val; |
| 4444 | 95 }; |
| 96 | |
| 97 /* | |
| 98 * Structure used by the NFA matcher. | |
| 99 */ | |
| 100 typedef struct | |
| 101 { | |
|
18753
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
102 // These three members implement regprog_T |
| 4444 | 103 regengine_T *engine; |
| 7 | 104 unsigned regflags; |
| 6328 | 105 unsigned re_engine; |
|
14354
ffd834f893aa
patch 8.1.0192: executing regexp recursively fails with a crash
Christian Brabandt <cb@256bit.org>
parents:
11521
diff
changeset
|
106 unsigned re_flags; |
|
ffd834f893aa
patch 8.1.0192: executing regexp recursively fails with a crash
Christian Brabandt <cb@256bit.org>
parents:
11521
diff
changeset
|
107 int re_in_use; |
| 4444 | 108 |
|
18753
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
109 nfa_state_T *start; // points into state[] |
|
4772
03375ccf28a2
updated for version 7.3.1133
Bram Moolenaar <bram@vim.org>
parents:
4718
diff
changeset
|
110 |
|
18753
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
111 int reganch; // pattern starts with ^ |
|
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
112 int regstart; // char at start of pattern |
|
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
113 char_u *match_text; // plain text to match with |
|
4772
03375ccf28a2
updated for version 7.3.1133
Bram Moolenaar <bram@vim.org>
parents:
4718
diff
changeset
|
114 |
|
18753
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
115 int has_zend; // pattern contains \ze |
|
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
116 int has_backref; // pattern contains \1 .. \9 |
|
4686
8db697ae406a
updated for version 7.3.1090
Bram Moolenaar <bram@vim.org>
parents:
4647
diff
changeset
|
117 #ifdef FEAT_SYN_HL |
|
8db697ae406a
updated for version 7.3.1090
Bram Moolenaar <bram@vim.org>
parents:
4647
diff
changeset
|
118 int reghasz; |
|
8db697ae406a
updated for version 7.3.1090
Bram Moolenaar <bram@vim.org>
parents:
4647
diff
changeset
|
119 #endif |
|
4690
9d97a0c045ef
updated for version 7.3.1092
Bram Moolenaar <bram@vim.org>
parents:
4686
diff
changeset
|
120 char_u *pattern; |
|
18753
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
121 int nsubexp; // number of () |
| 4444 | 122 int nstate; |
|
18753
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
123 nfa_state_T state[1]; // actually longer.. |
| 4444 | 124 } nfa_regprog_T; |
| 7 | 125 |
| 126 /* | |
| 127 * Structure to be used for single-line matching. | |
| 128 * Sub-match "no" starts at "startp[no]" and ends just before "endp[no]". | |
| 129 * When there is no match, the pointer is NULL. | |
| 130 */ | |
| 131 typedef struct | |
| 132 { | |
| 133 regprog_T *regprog; | |
| 134 char_u *startp[NSUBEXP]; | |
| 135 char_u *endp[NSUBEXP]; | |
|
31235
7fb4e244b16e
patch 9.0.0951: trying every character position for a match is inefficient
Bram Moolenaar <Bram@vim.org>
parents:
29071
diff
changeset
|
136 |
|
7fb4e244b16e
patch 9.0.0951: trying every character position for a match is inefficient
Bram Moolenaar <Bram@vim.org>
parents:
29071
diff
changeset
|
137 colnr_T rm_matchcol; // match start without "\zs" |
| 7 | 138 int rm_ic; |
| 139 } regmatch_T; | |
| 140 | |
| 141 /* | |
| 142 * Structure to be used for multi-line matching. | |
| 143 * Sub-match "no" starts in line "startpos[no].lnum" column "startpos[no].col" | |
| 144 * and ends in line "endpos[no].lnum" just before column "endpos[no].col". | |
| 145 * The line numbers are relative to the first line, thus startpos[0].lnum is | |
| 146 * always 0. | |
| 147 * When there is no match, the line number is -1. | |
| 148 */ | |
| 149 typedef struct | |
| 150 { | |
| 151 regprog_T *regprog; | |
| 152 lpos_T startpos[NSUBEXP]; | |
| 153 lpos_T endpos[NSUBEXP]; | |
|
31235
7fb4e244b16e
patch 9.0.0951: trying every character position for a match is inefficient
Bram Moolenaar <Bram@vim.org>
parents:
29071
diff
changeset
|
154 |
|
7fb4e244b16e
patch 9.0.0951: trying every character position for a match is inefficient
Bram Moolenaar <Bram@vim.org>
parents:
29071
diff
changeset
|
155 colnr_T rmm_matchcol; // match start without "\zs" |
| 7 | 156 int rmm_ic; |
|
18753
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
157 colnr_T rmm_maxcol; // when not zero: maximum column |
| 7 | 158 } regmmatch_T; |
| 159 | |
| 160 /* | |
| 161 * Structure used to store external references: "\z\(\)" to "\z\1". | |
| 162 * Use a reference count to avoid the need to copy this around. When it goes | |
| 163 * from 1 to zero the matches need to be freed. | |
| 164 */ | |
| 165 typedef struct | |
| 166 { | |
| 167 short refcnt; | |
| 168 char_u *matches[NSUBEXP]; | |
| 169 } reg_extmatch_T; | |
| 170 | |
| 4444 | 171 struct regengine |
| 172 { | |
|
28568
53f25723f378
patch 8.2.4808: unused item in engine struct
Bram Moolenaar <Bram@vim.org>
parents:
28226
diff
changeset
|
173 // bt_regcomp or nfa_regcomp |
| 4444 | 174 regprog_T *(*regcomp)(char_u*, int); |
|
28568
53f25723f378
patch 8.2.4808: unused item in engine struct
Bram Moolenaar <Bram@vim.org>
parents:
28226
diff
changeset
|
175 // bt_regfree or nfa_regfree |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4785
diff
changeset
|
176 void (*regfree)(regprog_T *); |
|
28568
53f25723f378
patch 8.2.4808: unused item in engine struct
Bram Moolenaar <Bram@vim.org>
parents:
28226
diff
changeset
|
177 // bt_regexec_nl or nfa_regexec_nl |
|
11521
578df034735d
patch 8.0.0643: when a pattern search is slow Vim becomes unusable
Christian Brabandt <cb@256bit.org>
parents:
10042
diff
changeset
|
178 int (*regexec_nl)(regmatch_T *, char_u *, colnr_T, int); |
|
28568
53f25723f378
patch 8.2.4808: unused item in engine struct
Bram Moolenaar <Bram@vim.org>
parents:
28226
diff
changeset
|
179 // bt_regexec_mult or nfa_regexec_mult |
|
29071
b90bca860b5a
patch 8.2.5057: using gettimeofday() for timeout is very inefficient
Bram Moolenaar <Bram@vim.org>
parents:
29048
diff
changeset
|
180 long (*regexec_multi)(regmmatch_T *, win_T *, buf_T *, linenr_T, colnr_T, int *); |
|
28568
53f25723f378
patch 8.2.4808: unused item in engine struct
Bram Moolenaar <Bram@vim.org>
parents:
28226
diff
changeset
|
181 //char_u *expr; |
| 4444 | 182 }; |
| 183 | |
|
29048
c98fc7a4dde4
patch 8.2.5046: vim_regsub() can overwrite the destination
Bram Moolenaar <Bram@vim.org>
parents:
28568
diff
changeset
|
184 // Flags used by vim_regsub() and vim_regsub_both() |
|
c98fc7a4dde4
patch 8.2.5046: vim_regsub() can overwrite the destination
Bram Moolenaar <Bram@vim.org>
parents:
28568
diff
changeset
|
185 #define REGSUB_COPY 1 |
|
c98fc7a4dde4
patch 8.2.5046: vim_regsub() can overwrite the destination
Bram Moolenaar <Bram@vim.org>
parents:
28568
diff
changeset
|
186 #define REGSUB_MAGIC 2 |
|
c98fc7a4dde4
patch 8.2.5046: vim_regsub() can overwrite the destination
Bram Moolenaar <Bram@vim.org>
parents:
28568
diff
changeset
|
187 #define REGSUB_BACKSLASH 4 |
|
c98fc7a4dde4
patch 8.2.5046: vim_regsub() can overwrite the destination
Bram Moolenaar <Bram@vim.org>
parents:
28568
diff
changeset
|
188 |
|
18753
6e3dc2d630c2
patch 8.1.2366: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
16378
diff
changeset
|
189 #endif // _REGEXP_H |