Commit | Line | Data |
---|---|---|
a0d0e21e LW |
1 | /* regexp.h |
2 | */ | |
3 | ||
378cc40b LW |
4 | /* |
5 | * Definitions etc. for regexp(3) routines. | |
6 | * | |
7 | * Caveat: this is V8 regexp(3) [actually, a reimplementation thereof], | |
8 | * not the System V one. | |
9 | */ | |
10 | ||
378cc40b | 11 | |
c277df42 IZ |
12 | struct regnode { |
13 | U8 flags; | |
14 | U8 type; | |
15 | U16 next_off; | |
16 | }; | |
17 | ||
18 | typedef struct regnode regnode; | |
19 | ||
c31fac66 GS |
20 | struct reg_data { |
21 | U32 count; | |
22 | U8 *what; | |
23 | void* data[1]; | |
24 | }; | |
25 | ||
378cc40b | 26 | typedef struct regexp { |
c277df42 | 27 | I32 refcnt; |
fe14fcc3 LW |
28 | char **startp; |
29 | char **endp; | |
c277df42 | 30 | regnode *regstclass; |
79072805 LW |
31 | I32 minlen; /* mininum possible length of $& */ |
32 | I32 prelen; /* length of precomp */ | |
a0d0e21e LW |
33 | U32 nparens; /* number of parentheses */ |
34 | U32 lastparen; /* last paren matched */ | |
378cc40b LW |
35 | char *precomp; /* pre-compilation regular expression */ |
36 | char *subbase; /* saved string so \digit works forever */ | |
9ef589d8 | 37 | char *subbeg; /* same, but not responsible for allocation */ |
00bf170e | 38 | char *subend; /* end of subbase */ |
a0d0e21e | 39 | U16 naughty; /* how exponential is this pattern? */ |
c277df42 IZ |
40 | U16 reganch; /* Internal use only + |
41 | Tainted information used by regexec? */ | |
42 | SV *anchored_substr; /* Substring at fixed position wrt start. */ | |
43 | I32 anchored_offset; /* Position of it. */ | |
44 | SV *float_substr; /* Substring at variable position wrt start. */ | |
45 | I32 float_min_offset; /* Minimal position of it. */ | |
46 | I32 float_max_offset; /* Maximal position of it. */ | |
47 | SV *check_substr; /* Substring to check before matching. */ | |
48 | I32 check_offset_min; /* Offset of the above. */ | |
49 | I32 check_offset_max; /* Offset of the above. */ | |
50 | struct reg_data *data; /* Additional data. */ | |
51 | regnode program[1]; /* Unwarranted chumminess with compiler. */ | |
378cc40b LW |
52 | } regexp; |
53 | ||
c277df42 IZ |
54 | #define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS) |
55 | #define ROPT_ANCH_SINGLE (ROPT_ANCH_BOL|ROPT_ANCH_GPOS) | |
56 | #define ROPT_ANCH_BOL 1 | |
57 | #define ROPT_ANCH_MBOL 2 | |
58 | #define ROPT_ANCH_GPOS 4 | |
59 | #define ROPT_SKIP 8 | |
60 | #define ROPT_IMPLICIT 0x10 /* Converted .* to ^.* */ | |
61 | #define ROPT_NOSCAN 0x20 /* Check-string always at start. */ | |
62 | #define ROPT_GPOS_SEEN 0x40 | |
63 | #define ROPT_CHECK_ALL 0x80 | |
64 | #define ROPT_LOOKBEHIND_SEEN 0x100 | |
65 | ||
66 | #define ROPT_TAINTED_SEEN 0x8000 | |
67 | ||
68 | #define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN) | |
69 | #define RX_MATCH_TAINTED_SET(prog, t) ((t) \ | |
70 | ? ((prog)->reganch |= ROPT_TAINTED_SEEN) \ | |
71 | : ((prog)->reganch &= ~ROPT_TAINTED_SEEN)) | |
72 | ||
73 | #define REXEC_COPY_STR 1 /* Need to copy the string. */ | |
74 | #define REXEC_CHECKED 2 /* check_substr already checked. */ | |
75 | ||
76 | #define ReREFCNT_inc(re) ((re && re->refcnt++), re) | |
77 | #define ReREFCNT_dec(re) pregfree(re) |