This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
create utfperl branch
[perl5.git] / regexp.h
CommitLineData
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
12struct regnode {
13 U8 flags;
14 U8 type;
15 U16 next_off;
16};
17
18typedef struct regnode regnode;
19
c31fac66
GS
20struct reg_data {
21 U32 count;
22 U8 *what;
23 void* data[1];
24};
25
2779dcf1
IZ
26struct reg_substr_datum {
27 I32 min_offset;
28 I32 max_offset;
29 SV *substr;
30};
31
32struct reg_substr_data {
33 struct reg_substr_datum data[3]; /* Actual array */
34};
35
378cc40b 36typedef struct regexp {
c277df42 37 I32 refcnt;
fe14fcc3
LW
38 char **startp;
39 char **endp;
c277df42 40 regnode *regstclass;
79072805
LW
41 I32 minlen; /* mininum possible length of $& */
42 I32 prelen; /* length of precomp */
a0d0e21e
LW
43 U32 nparens; /* number of parentheses */
44 U32 lastparen; /* last paren matched */
378cc40b
LW
45 char *precomp; /* pre-compilation regular expression */
46 char *subbase; /* saved string so \digit works forever */
9ef589d8 47 char *subbeg; /* same, but not responsible for allocation */
00bf170e 48 char *subend; /* end of subbase */
a0d0e21e 49 U16 naughty; /* how exponential is this pattern? */
c277df42
IZ
50 U16 reganch; /* Internal use only +
51 Tainted information used by regexec? */
2779dcf1 52#if 0
c277df42
IZ
53 SV *anchored_substr; /* Substring at fixed position wrt start. */
54 I32 anchored_offset; /* Position of it. */
55 SV *float_substr; /* Substring at variable position wrt start. */
56 I32 float_min_offset; /* Minimal position of it. */
57 I32 float_max_offset; /* Maximal position of it. */
58 SV *check_substr; /* Substring to check before matching. */
59 I32 check_offset_min; /* Offset of the above. */
60 I32 check_offset_max; /* Offset of the above. */
2779dcf1
IZ
61#else
62 struct reg_substr_data *substrs;
63#endif
c277df42
IZ
64 struct reg_data *data; /* Additional data. */
65 regnode program[1]; /* Unwarranted chumminess with compiler. */
378cc40b
LW
66} regexp;
67
2779dcf1
IZ
68#define anchored_substr substrs->data[0].substr
69#define anchored_offset substrs->data[0].min_offset
70#define float_substr substrs->data[1].substr
71#define float_min_offset substrs->data[1].min_offset
72#define float_max_offset substrs->data[1].max_offset
73#define check_substr substrs->data[2].substr
74#define check_offset_min substrs->data[2].min_offset
75#define check_offset_max substrs->data[2].max_offset
76
c277df42
IZ
77#define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS)
78#define ROPT_ANCH_SINGLE (ROPT_ANCH_BOL|ROPT_ANCH_GPOS)
79#define ROPT_ANCH_BOL 1
80#define ROPT_ANCH_MBOL 2
81#define ROPT_ANCH_GPOS 4
82#define ROPT_SKIP 8
83#define ROPT_IMPLICIT 0x10 /* Converted .* to ^.* */
84#define ROPT_NOSCAN 0x20 /* Check-string always at start. */
85#define ROPT_GPOS_SEEN 0x40
86#define ROPT_CHECK_ALL 0x80
87#define ROPT_LOOKBEHIND_SEEN 0x100
ce862d02 88#define ROPT_EVAL_SEEN 0x200
8782bef2
GB
89#define ROPT_TAINTED_SEEN 0x400
90/* 0xf800 of reganch is used by PMf_COMPILETIME */
c277df42
IZ
91
92#define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN)
72311751
GS
93#define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN)
94#define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN)
95#define RX_MATCH_TAINTED_set(prog, t) ((t) \
96 ? RX_MATCH_TAINTED_on(prog) \
97 : RX_MATCH_TAINTED_off(prog))
c277df42
IZ
98
99#define REXEC_COPY_STR 1 /* Need to copy the string. */
100#define REXEC_CHECKED 2 /* check_substr already checked. */
101
102#define ReREFCNT_inc(re) ((re && re->refcnt++), re)
103#define ReREFCNT_dec(re) pregfree(re)