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 typedef struct thread thread_T;
|
|
33
|
|
34 /*
|
7
|
35 * Structure returned by vim_regcomp() to pass on to vim_regexec().
|
4444
|
36 * This is the general structure. For the actual matcher, two specific
|
|
37 * structures are used. See code below.
|
|
38 */
|
|
39 typedef struct regprog
|
|
40 {
|
|
41 regengine_T *engine;
|
|
42 unsigned regflags;
|
|
43 } regprog_T;
|
|
44
|
|
45 /*
|
|
46 * Structure used by the back track matcher.
|
7
|
47 * These fields are only to be used in regexp.c!
|
4444
|
48 * See regexp.c for an explanation.
|
7
|
49 */
|
|
50 typedef struct
|
|
51 {
|
4444
|
52 /* These two members implement regprog_T */
|
|
53 regengine_T *engine;
|
|
54 unsigned regflags;
|
|
55
|
7
|
56 int regstart;
|
|
57 char_u reganch;
|
|
58 char_u *regmust;
|
|
59 int regmlen;
|
4444
|
60 char_u reghasz;
|
|
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 visits;
|
|
77 thread_T *lastthread;
|
|
78 int negated;
|
|
79 };
|
|
80
|
|
81 /*
|
|
82 * Structure used by the NFA matcher.
|
|
83 */
|
|
84 typedef struct
|
|
85 {
|
|
86 /* These two members implement regprog_T */
|
|
87 regengine_T *engine;
|
7
|
88 unsigned regflags;
|
4444
|
89
|
|
90 regprog_T regprog;
|
|
91 nfa_state_T *start;
|
|
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 */
|