Mercurial > vim
annotate src/regexp.h @ 4659:80b0081824fa v7.3.1077
updated for version 7.3.1077
Problem: Python: Allocating dict the wrong way, causing a crash.
Solution: Use py_dict_alloc(). Fix some exception problems. (ZyX)
author | Bram Moolenaar <bram@vim.org> |
---|---|
date | Thu, 30 May 2013 22:06:33 +0200 |
parents | 857f6c53f117 |
children | 8db697ae406a |
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; | |
4444 | 58 char_u reghasz; |
59 char_u program[1]; /* actually longer.. */ | |
60 } bt_regprog_T; | |
61 | |
62 /* | |
63 * Structure representing a NFA state. | |
64 * A NFA state may have no outgoing edge, when it is a NFA_MATCH state. | |
65 */ | |
66 typedef struct nfa_state nfa_state_T; | |
67 struct nfa_state | |
68 { | |
69 int c; | |
70 nfa_state_T *out; | |
71 nfa_state_T *out1; | |
72 int id; | |
73 int lastlist; | |
74 int negated; | |
4583
321cfbef9431
updated for version 7.3.1039
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
75 int val; |
4444 | 76 }; |
77 | |
78 /* | |
79 * Structure used by the NFA matcher. | |
80 */ | |
81 typedef struct | |
82 { | |
83 /* These two members implement regprog_T */ | |
84 regengine_T *engine; | |
7 | 85 unsigned regflags; |
4444 | 86 |
87 regprog_T regprog; | |
88 nfa_state_T *start; | |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
89 int has_zend; /* pattern contains \ze */ |
4647
857f6c53f117
updated for version 7.3.1071
Bram Moolenaar <bram@vim.org>
parents:
4583
diff
changeset
|
90 int has_backref; /* pattern contains \1 .. \9 */ |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
91 int nsubexp; /* number of () */ |
4444 | 92 int nstate; |
93 nfa_state_T state[0]; /* actually longer.. */ | |
94 } nfa_regprog_T; | |
7 | 95 |
96 /* | |
97 * Structure to be used for single-line matching. | |
98 * Sub-match "no" starts at "startp[no]" and ends just before "endp[no]". | |
99 * When there is no match, the pointer is NULL. | |
100 */ | |
101 typedef struct | |
102 { | |
103 regprog_T *regprog; | |
104 char_u *startp[NSUBEXP]; | |
105 char_u *endp[NSUBEXP]; | |
106 int rm_ic; | |
107 } regmatch_T; | |
108 | |
109 /* | |
110 * Structure to be used for multi-line matching. | |
111 * Sub-match "no" starts in line "startpos[no].lnum" column "startpos[no].col" | |
112 * and ends in line "endpos[no].lnum" just before column "endpos[no].col". | |
113 * The line numbers are relative to the first line, thus startpos[0].lnum is | |
114 * always 0. | |
115 * When there is no match, the line number is -1. | |
116 */ | |
117 typedef struct | |
118 { | |
119 regprog_T *regprog; | |
120 lpos_T startpos[NSUBEXP]; | |
121 lpos_T endpos[NSUBEXP]; | |
122 int rmm_ic; | |
418 | 123 colnr_T rmm_maxcol; /* when not zero: maximum column */ |
7 | 124 } regmmatch_T; |
125 | |
126 /* | |
127 * Structure used to store external references: "\z\(\)" to "\z\1". | |
128 * Use a reference count to avoid the need to copy this around. When it goes | |
129 * from 1 to zero the matches need to be freed. | |
130 */ | |
131 typedef struct | |
132 { | |
133 short refcnt; | |
134 char_u *matches[NSUBEXP]; | |
135 } reg_extmatch_T; | |
136 | |
4444 | 137 struct regengine |
138 { | |
139 regprog_T *(*regcomp)(char_u*, int); | |
140 int (*regexec)(regmatch_T*, char_u*, colnr_T); | |
141 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
142 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
143 int (*regexec_nl)(regmatch_T*, char_u*, colnr_T); | |
144 #endif | |
145 long (*regexec_multi)(regmmatch_T*, win_T*, buf_T*, linenr_T, colnr_T, proftime_T*); | |
146 #ifdef DEBUG | |
147 char_u *expr; | |
148 #endif | |
149 }; | |
150 | |
7 | 151 #endif /* _REGEXP_H */ |