Mercurial > vim
annotate src/regexp.h @ 4686:8db697ae406a v7.3.1090
updated for version 7.3.1090
Problem: New regexp engine does not support \z1 .. \z9 and \z(.
Solution: Implement the syntax submatches.
author | Bram Moolenaar <bram@vim.org> |
---|---|
date | Sat, 01 Jun 2013 23:02:54 +0200 |
parents | 857f6c53f117 |
children | 9d97a0c045ef |
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; | |
75 int lastlist; | |
76 int negated; | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
77 int val; |
4444 | 78 }; |
79 | |
80 /* | |
81 * Structure used by the NFA matcher. | |
82 */ | |
83 typedef struct | |
84 { | |
85 /* These two members implement regprog_T */ | |
86 regengine_T *engine; | |
7 | 87 unsigned regflags; |
4444 | 88 |
89 regprog_T regprog; | |
90 nfa_state_T *start; | |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
91 int has_zend; /* pattern contains \ze */ |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
92 int has_backref; /* pattern contains \1 .. \9 */ |
4686
8db697ae406a
updated for version 7.3.1090
Bram Moolenaar <bram@vim.org>
parents:
4647
diff
changeset
|
93 #ifdef FEAT_SYN_HL |
8db697ae406a
updated for version 7.3.1090
Bram Moolenaar <bram@vim.org>
parents:
4647
diff
changeset
|
94 int reghasz; |
8db697ae406a
updated for version 7.3.1090
Bram Moolenaar <bram@vim.org>
parents:
4647
diff
changeset
|
95 #endif |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
96 int nsubexp; /* number of () */ |
4444 | 97 int nstate; |
98 nfa_state_T state[0]; /* actually longer.. */ | |
99 } nfa_regprog_T; | |
7 | 100 |
101 /* | |
102 * Structure to be used for single-line matching. | |
103 * Sub-match "no" starts at "startp[no]" and ends just before "endp[no]". | |
104 * When there is no match, the pointer is NULL. | |
105 */ | |
106 typedef struct | |
107 { | |
108 regprog_T *regprog; | |
109 char_u *startp[NSUBEXP]; | |
110 char_u *endp[NSUBEXP]; | |
111 int rm_ic; | |
112 } regmatch_T; | |
113 | |
114 /* | |
115 * Structure to be used for multi-line matching. | |
116 * Sub-match "no" starts in line "startpos[no].lnum" column "startpos[no].col" | |
117 * and ends in line "endpos[no].lnum" just before column "endpos[no].col". | |
118 * The line numbers are relative to the first line, thus startpos[0].lnum is | |
119 * always 0. | |
120 * When there is no match, the line number is -1. | |
121 */ | |
122 typedef struct | |
123 { | |
124 regprog_T *regprog; | |
125 lpos_T startpos[NSUBEXP]; | |
126 lpos_T endpos[NSUBEXP]; | |
127 int rmm_ic; | |
418 | 128 colnr_T rmm_maxcol; /* when not zero: maximum column */ |
7 | 129 } regmmatch_T; |
130 | |
131 /* | |
132 * Structure used to store external references: "\z\(\)" to "\z\1". | |
133 * Use a reference count to avoid the need to copy this around. When it goes | |
134 * from 1 to zero the matches need to be freed. | |
135 */ | |
136 typedef struct | |
137 { | |
138 short refcnt; | |
139 char_u *matches[NSUBEXP]; | |
140 } reg_extmatch_T; | |
141 | |
4444 | 142 struct regengine |
143 { | |
144 regprog_T *(*regcomp)(char_u*, int); | |
145 int (*regexec)(regmatch_T*, char_u*, colnr_T); | |
146 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
147 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
148 int (*regexec_nl)(regmatch_T*, char_u*, colnr_T); | |
149 #endif | |
150 long (*regexec_multi)(regmmatch_T*, win_T*, buf_T*, linenr_T, colnr_T, proftime_T*); | |
151 #ifdef DEBUG | |
152 char_u *expr; | |
153 #endif | |
154 }; | |
155 | |
7 | 156 #endif /* _REGEXP_H */ |