Mercurial > vim
annotate src/regexp.h @ 6212:c9f5516b7967
Added tag v7-4-441 for changeset e754b23b7d1c
| author | Bram Moolenaar <bram@vim.org> |
|---|---|
| date | Tue, 09 Sep 2014 18:45:50 +0200 |
| parents | 0ea551fa607d |
| children | adfbffe1e642 |
| rev | line source |
|---|---|
| 7 | 1 /* vi:set ts=8 sts=4 sw=4: |
| 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 | |
| 30 typedef struct regengine regengine_T; | |
| 31 | |
| 32 /* | |
| 7 | 33 * Structure returned by vim_regcomp() to pass on to vim_regexec(). |
| 4444 | 34 * This is the general structure. For the actual matcher, two specific |
| 35 * structures are used. See code below. | |
| 36 */ | |
| 37 typedef struct regprog | |
| 38 { | |
| 39 regengine_T *engine; | |
| 40 unsigned regflags; | |
| 41 } regprog_T; | |
| 42 | |
| 43 /* | |
| 44 * Structure used by the back track matcher. | |
| 7 | 45 * These fields are only to be used in regexp.c! |
| 4444 | 46 * See regexp.c for an explanation. |
| 7 | 47 */ |
| 48 typedef struct | |
| 49 { | |
| 4444 | 50 /* These two members implement regprog_T */ |
| 51 regengine_T *engine; | |
| 52 unsigned regflags; | |
| 53 | |
| 7 | 54 int regstart; |
| 55 char_u reganch; | |
| 56 char_u *regmust; | |
| 57 int regmlen; | |
|
4686
8db697ae406a
updated for version 7.3.1090
Bram Moolenaar <bram@vim.org>
parents:
4647
diff
changeset
|
58 #ifdef FEAT_SYN_HL |
| 4444 | 59 char_u reghasz; |
|
4686
8db697ae406a
updated for version 7.3.1090
Bram Moolenaar <bram@vim.org>
parents:
4647
diff
changeset
|
60 #endif |
| 4444 | 61 char_u program[1]; /* actually longer.. */ |
| 62 } bt_regprog_T; | |
| 63 | |
| 64 /* | |
| 65 * Structure representing a NFA state. | |
| 66 * A NFA state may have no outgoing edge, when it is a NFA_MATCH state. | |
| 67 */ | |
| 68 typedef struct nfa_state nfa_state_T; | |
| 69 struct nfa_state | |
| 70 { | |
| 71 int c; | |
| 72 nfa_state_T *out; | |
| 73 nfa_state_T *out1; | |
| 74 int id; | |
|
4718
ec72bb4a0fc2
updated for version 7.3.1106
Bram Moolenaar <bram@vim.org>
parents:
4690
diff
changeset
|
75 int lastlist[2]; /* 0: normal, 1: recursive */ |
|
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
76 int val; |
| 4444 | 77 }; |
| 78 | |
| 79 /* | |
| 80 * Structure used by the NFA matcher. | |
| 81 */ | |
| 82 typedef struct | |
| 83 { | |
| 84 /* These two members implement regprog_T */ | |
| 85 regengine_T *engine; | |
| 7 | 86 unsigned regflags; |
| 4444 | 87 |
|
4690
9d97a0c045ef
updated for version 7.3.1092
Bram Moolenaar <bram@vim.org>
parents:
4686
diff
changeset
|
88 nfa_state_T *start; /* points into state[] */ |
|
4772
03375ccf28a2
updated for version 7.3.1133
Bram Moolenaar <bram@vim.org>
parents:
4718
diff
changeset
|
89 |
|
03375ccf28a2
updated for version 7.3.1133
Bram Moolenaar <bram@vim.org>
parents:
4718
diff
changeset
|
90 int reganch; /* pattern starts with ^ */ |
|
03375ccf28a2
updated for version 7.3.1133
Bram Moolenaar <bram@vim.org>
parents:
4718
diff
changeset
|
91 int regstart; /* char at start of pattern */ |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4785
diff
changeset
|
92 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
|
93 |
|
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
94 int has_zend; /* pattern contains \ze */ |
|
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
95 int has_backref; /* pattern contains \1 .. \9 */ |
|
4686
8db697ae406a
updated for version 7.3.1090
Bram Moolenaar <bram@vim.org>
parents:
4647
diff
changeset
|
96 #ifdef FEAT_SYN_HL |
|
8db697ae406a
updated for version 7.3.1090
Bram Moolenaar <bram@vim.org>
parents:
4647
diff
changeset
|
97 int reghasz; |
|
8db697ae406a
updated for version 7.3.1090
Bram Moolenaar <bram@vim.org>
parents:
4647
diff
changeset
|
98 #endif |
|
4690
9d97a0c045ef
updated for version 7.3.1092
Bram Moolenaar <bram@vim.org>
parents:
4686
diff
changeset
|
99 #ifdef DEBUG |
|
9d97a0c045ef
updated for version 7.3.1092
Bram Moolenaar <bram@vim.org>
parents:
4686
diff
changeset
|
100 char_u *pattern; |
|
9d97a0c045ef
updated for version 7.3.1092
Bram Moolenaar <bram@vim.org>
parents:
4686
diff
changeset
|
101 #endif |
|
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
102 int nsubexp; /* number of () */ |
| 4444 | 103 int nstate; |
|
4837
05b8436873d4
updated for version 7.3.1165
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
104 nfa_state_T state[1]; /* actually longer.. */ |
| 4444 | 105 } nfa_regprog_T; |
| 7 | 106 |
| 107 /* | |
| 108 * Structure to be used for single-line matching. | |
| 109 * Sub-match "no" starts at "startp[no]" and ends just before "endp[no]". | |
| 110 * When there is no match, the pointer is NULL. | |
| 111 */ | |
| 112 typedef struct | |
| 113 { | |
| 114 regprog_T *regprog; | |
| 115 char_u *startp[NSUBEXP]; | |
| 116 char_u *endp[NSUBEXP]; | |
| 117 int rm_ic; | |
| 118 } regmatch_T; | |
| 119 | |
| 120 /* | |
| 121 * Structure to be used for multi-line matching. | |
| 122 * Sub-match "no" starts in line "startpos[no].lnum" column "startpos[no].col" | |
| 123 * and ends in line "endpos[no].lnum" just before column "endpos[no].col". | |
| 124 * The line numbers are relative to the first line, thus startpos[0].lnum is | |
| 125 * always 0. | |
| 126 * When there is no match, the line number is -1. | |
| 127 */ | |
| 128 typedef struct | |
| 129 { | |
| 130 regprog_T *regprog; | |
| 131 lpos_T startpos[NSUBEXP]; | |
| 132 lpos_T endpos[NSUBEXP]; | |
| 133 int rmm_ic; | |
| 418 | 134 colnr_T rmm_maxcol; /* when not zero: maximum column */ |
| 7 | 135 } regmmatch_T; |
| 136 | |
| 137 /* | |
| 138 * Structure used to store external references: "\z\(\)" to "\z\1". | |
| 139 * Use a reference count to avoid the need to copy this around. When it goes | |
| 140 * from 1 to zero the matches need to be freed. | |
| 141 */ | |
| 142 typedef struct | |
| 143 { | |
| 144 short refcnt; | |
| 145 char_u *matches[NSUBEXP]; | |
| 146 } reg_extmatch_T; | |
| 147 | |
| 4444 | 148 struct regengine |
| 149 { | |
| 150 regprog_T *(*regcomp)(char_u*, int); | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4785
diff
changeset
|
151 void (*regfree)(regprog_T *); |
| 5838 | 152 int (*regexec_nl)(regmatch_T*, char_u*, colnr_T, int); |
| 4444 | 153 long (*regexec_multi)(regmmatch_T*, win_T*, buf_T*, linenr_T, colnr_T, proftime_T*); |
| 154 #ifdef DEBUG | |
| 155 char_u *expr; | |
| 156 #endif | |
| 157 }; | |
| 158 | |
| 7 | 159 #endif /* _REGEXP_H */ |