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 | ||
cad2e5aa | 20 | struct reg_substr_data; |
2779dcf1 | 21 | |
0ee3c8d0 JH |
22 | struct reg_data; |
23 | ||
378cc40b | 24 | typedef struct regexp { |
cf93c79d IZ |
25 | I32 *startp; |
26 | I32 *endp; | |
c277df42 | 27 | regnode *regstclass; |
2779dcf1 | 28 | struct reg_substr_data *substrs; |
cf93c79d | 29 | char *precomp; /* pre-compilation regular expression */ |
c277df42 | 30 | struct reg_data *data; /* Additional data. */ |
cf93c79d IZ |
31 | char *subbeg; /* saved or original string |
32 | so \digit works forever. */ | |
fac92740 | 33 | U32 *offsets; /* offset annotations 20001228 MJD */ |
cf93c79d IZ |
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 */ | |
a01268b5 | 40 | U32 lastcloseparen; /* last paren matched */ |
cf93c79d IZ |
41 | U32 reganch; /* Internal use only + |
42 | Tainted information used by regexec? */ | |
c277df42 | 43 | regnode program[1]; /* Unwarranted chumminess with compiler. */ |
378cc40b LW |
44 | } regexp; |
45 | ||
cad2e5aa JH |
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) | |
a0ed51b3 LW |
48 | #define ROPT_ANCH_BOL 0x00001 |
49 | #define ROPT_ANCH_MBOL 0x00002 | |
cad2e5aa JH |
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 | |
3baa4c62 | 59 | #define ROPT_SANY_SEEN 0x00800 |
f33976b4 | 60 | #define ROPT_CANY_SEEN 0x01000 |
a0ed51b3 | 61 | |
8782bef2 | 62 | /* 0xf800 of reganch is used by PMf_COMPILETIME */ |
c277df42 | 63 | |
a0ed51b3 LW |
64 | #define ROPT_UTF8 0x10000 |
65 | #define ROPT_NAUGHTY 0x20000 /* how exponential is this pattern? */ | |
cf93c79d | 66 | #define ROPT_COPY_DONE 0x40000 /* subbeg is a copy of the string */ |
bc9a0d2c | 67 | #define ROPT_TAINTED_SEEN 0x80000 |
a0ed51b3 | 68 | |
cad2e5aa JH |
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 | ||
c277df42 | 82 | #define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN) |
72311751 GS |
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)) | |
c277df42 | 88 | |
cf93c79d IZ |
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 | ||
cad2e5aa JH |
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. */ | |
cf93c79d | 100 | #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ |
cad2e5aa | 101 | #define REXEC_ML 0x20 /* $* was set. */ |
c277df42 | 102 | |
155aba94 | 103 | #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re) |
cad2e5aa | 104 | #define ReREFCNT_dec(re) CALLREGFREE(aTHX_ re) |
cf93c79d IZ |
105 | |
106 | #define FBMcf_TAIL_DOLLAR 1 | |
cad2e5aa JH |
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) | |
cf93c79d IZ |
111 | |
112 | #define FBMrf_MULTILINE 1 | |
cad2e5aa JH |
113 | |
114 | struct re_scream_pos_data_s; |