This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
more POSIX tests, and more autoloading
[perl5.git] / regexp.h
... / ...
CommitLineData
1/* regexp.h
2 */
3
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
11
12struct regnode {
13 U8 flags;
14 U8 type;
15 U16 next_off;
16};
17
18typedef struct regnode regnode;
19
20struct reg_substr_data;
21
22struct reg_data;
23
24typedef struct regexp {
25 I32 *startp;
26 I32 *endp;
27 regnode *regstclass;
28 struct reg_substr_data *substrs;
29 char *precomp; /* pre-compilation regular expression */
30 struct reg_data *data; /* Additional data. */
31 char *subbeg; /* saved or original string
32 so \digit works forever. */
33 U32 *offsets; /* offset annotations 20001228 MJD */
34 I32 sublen; /* Length of string pointed by subbeg */
35 I32 refcnt;
36 I32 minlen; /* mininum possible length of $& */
37 I32 prelen; /* length of precomp */
38 U32 nparens; /* number of parentheses */
39 U32 lastparen; /* last paren matched */
40 U32 lastcloseparen; /* last paren matched */
41 U32 reganch; /* Internal use only +
42 Tainted information used by regexec? */
43 regnode program[1]; /* Unwarranted chumminess with compiler. */
44} regexp;
45
46#define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS|ROPT_ANCH_SBOL)
47#define ROPT_ANCH_SINGLE (ROPT_ANCH_SBOL|ROPT_ANCH_GPOS)
48#define ROPT_ANCH_BOL 0x00001
49#define ROPT_ANCH_MBOL 0x00002
50#define ROPT_ANCH_SBOL 0x00004
51#define ROPT_ANCH_GPOS 0x00008
52#define ROPT_SKIP 0x00010
53#define ROPT_IMPLICIT 0x00020 /* Converted .* to ^.* */
54#define ROPT_NOSCAN 0x00040 /* Check-string always at start. */
55#define ROPT_GPOS_SEEN 0x00080
56#define ROPT_CHECK_ALL 0x00100
57#define ROPT_LOOKBEHIND_SEEN 0x00200
58#define ROPT_EVAL_SEEN 0x00400
59#define ROPT_CANY_SEEN 0x00800
60#define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */
61
62/* 0xf800 of reganch is used by PMf_COMPILETIME */
63
64#define ROPT_UTF8 0x10000
65#define ROPT_NAUGHTY 0x20000 /* how exponential is this pattern? */
66#define ROPT_COPY_DONE 0x40000 /* subbeg is a copy of the string */
67#define ROPT_TAINTED_SEEN 0x80000
68
69#define RE_USE_INTUIT_NOML 0x0100000 /* Best to intuit before matching */
70#define RE_USE_INTUIT_ML 0x0200000
71#define REINT_AUTORITATIVE_NOML 0x0400000 /* Can trust a positive answer */
72#define REINT_AUTORITATIVE_ML 0x0800000
73#define REINT_ONCE_NOML 0x1000000 /* Intuit can succed once only. */
74#define REINT_ONCE_ML 0x2000000
75#define RE_INTUIT_ONECHAR 0x4000000
76#define RE_INTUIT_TAIL 0x8000000
77
78#define RE_USE_INTUIT (RE_USE_INTUIT_NOML|RE_USE_INTUIT_ML)
79#define REINT_AUTORITATIVE (REINT_AUTORITATIVE_NOML|REINT_AUTORITATIVE_ML)
80#define REINT_ONCE (REINT_ONCE_NOML|REINT_ONCE_ML)
81
82#define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN)
83#define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN)
84#define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN)
85#define RX_MATCH_TAINTED_set(prog, t) ((t) \
86 ? RX_MATCH_TAINTED_on(prog) \
87 : RX_MATCH_TAINTED_off(prog))
88
89#define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE)
90#define RX_MATCH_COPIED_on(prog) ((prog)->reganch |= ROPT_COPY_DONE)
91#define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE)
92#define RX_MATCH_COPIED_set(prog,t) ((t) \
93 ? RX_MATCH_COPIED_on(prog) \
94 : RX_MATCH_COPIED_off(prog))
95
96#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
97#define REXEC_CHECKED 0x02 /* check_substr already checked. */
98#define REXEC_SCREAM 0x04 /* use scream table. */
99#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
100#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
101#define REXEC_ML 0x20 /* $* was set. */
102
103#define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
104#define ReREFCNT_dec(re) CALLREGFREE(aTHX_ re)
105
106#define FBMcf_TAIL_DOLLAR 1
107#define FBMcf_TAIL_DOLLARM 2
108#define FBMcf_TAIL_Z 4
109#define FBMcf_TAIL_z 8
110#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
111
112#define FBMrf_MULTILINE 1
113
114struct re_scream_pos_data_s;