Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* regexp.h |
d6376244 | 2 | * |
e6906430 JH |
3 | * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003, |
4 | * by Larry Wall and others | |
d6376244 JH |
5 | * |
6 | * You may distribute under the terms of either the GNU General Public | |
7 | * License or the Artistic License, as specified in the README file. | |
8 | * | |
a0d0e21e LW |
9 | */ |
10 | ||
378cc40b LW |
11 | /* |
12 | * Definitions etc. for regexp(3) routines. | |
13 | * | |
14 | * Caveat: this is V8 regexp(3) [actually, a reimplementation thereof], | |
15 | * not the System V one. | |
16 | */ | |
17 | ||
378cc40b | 18 | |
c277df42 IZ |
19 | struct regnode { |
20 | U8 flags; | |
21 | U8 type; | |
22 | U16 next_off; | |
23 | }; | |
24 | ||
25 | typedef struct regnode regnode; | |
26 | ||
cad2e5aa | 27 | struct reg_substr_data; |
2779dcf1 | 28 | |
0ee3c8d0 JH |
29 | struct reg_data; |
30 | ||
378cc40b | 31 | typedef struct regexp { |
cf93c79d IZ |
32 | I32 *startp; |
33 | I32 *endp; | |
c277df42 | 34 | regnode *regstclass; |
2779dcf1 | 35 | struct reg_substr_data *substrs; |
cf93c79d | 36 | char *precomp; /* pre-compilation regular expression */ |
c277df42 | 37 | struct reg_data *data; /* Additional data. */ |
cf93c79d IZ |
38 | char *subbeg; /* saved or original string |
39 | so \digit works forever. */ | |
fac92740 | 40 | U32 *offsets; /* offset annotations 20001228 MJD */ |
cf93c79d IZ |
41 | I32 sublen; /* Length of string pointed by subbeg */ |
42 | I32 refcnt; | |
43 | I32 minlen; /* mininum possible length of $& */ | |
44 | I32 prelen; /* length of precomp */ | |
45 | U32 nparens; /* number of parentheses */ | |
46 | U32 lastparen; /* last paren matched */ | |
a01268b5 | 47 | U32 lastcloseparen; /* last paren matched */ |
cf93c79d IZ |
48 | U32 reganch; /* Internal use only + |
49 | Tainted information used by regexec? */ | |
c277df42 | 50 | regnode program[1]; /* Unwarranted chumminess with compiler. */ |
378cc40b LW |
51 | } regexp; |
52 | ||
cad2e5aa JH |
53 | #define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS|ROPT_ANCH_SBOL) |
54 | #define ROPT_ANCH_SINGLE (ROPT_ANCH_SBOL|ROPT_ANCH_GPOS) | |
a0ed51b3 LW |
55 | #define ROPT_ANCH_BOL 0x00001 |
56 | #define ROPT_ANCH_MBOL 0x00002 | |
cad2e5aa JH |
57 | #define ROPT_ANCH_SBOL 0x00004 |
58 | #define ROPT_ANCH_GPOS 0x00008 | |
59 | #define ROPT_SKIP 0x00010 | |
60 | #define ROPT_IMPLICIT 0x00020 /* Converted .* to ^.* */ | |
61 | #define ROPT_NOSCAN 0x00040 /* Check-string always at start. */ | |
62 | #define ROPT_GPOS_SEEN 0x00080 | |
63 | #define ROPT_CHECK_ALL 0x00100 | |
64 | #define ROPT_LOOKBEHIND_SEEN 0x00200 | |
65 | #define ROPT_EVAL_SEEN 0x00400 | |
cce850e4 JH |
66 | #define ROPT_CANY_SEEN 0x00800 |
67 | #define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */ | |
a0ed51b3 | 68 | |
8782bef2 | 69 | /* 0xf800 of reganch is used by PMf_COMPILETIME */ |
c277df42 | 70 | |
a0ed51b3 LW |
71 | #define ROPT_UTF8 0x10000 |
72 | #define ROPT_NAUGHTY 0x20000 /* how exponential is this pattern? */ | |
cf93c79d | 73 | #define ROPT_COPY_DONE 0x40000 /* subbeg is a copy of the string */ |
bc9a0d2c | 74 | #define ROPT_TAINTED_SEEN 0x80000 |
2ea5368e | 75 | #define ROPT_MATCH_UTF8 0x10000000 /* subbeg is utf-8 */ |
a0ed51b3 | 76 | |
cad2e5aa JH |
77 | #define RE_USE_INTUIT_NOML 0x0100000 /* Best to intuit before matching */ |
78 | #define RE_USE_INTUIT_ML 0x0200000 | |
79 | #define REINT_AUTORITATIVE_NOML 0x0400000 /* Can trust a positive answer */ | |
80 | #define REINT_AUTORITATIVE_ML 0x0800000 | |
81 | #define REINT_ONCE_NOML 0x1000000 /* Intuit can succed once only. */ | |
82 | #define REINT_ONCE_ML 0x2000000 | |
83 | #define RE_INTUIT_ONECHAR 0x4000000 | |
84 | #define RE_INTUIT_TAIL 0x8000000 | |
85 | ||
86 | #define RE_USE_INTUIT (RE_USE_INTUIT_NOML|RE_USE_INTUIT_ML) | |
87 | #define REINT_AUTORITATIVE (REINT_AUTORITATIVE_NOML|REINT_AUTORITATIVE_ML) | |
88 | #define REINT_ONCE (REINT_ONCE_NOML|REINT_ONCE_ML) | |
89 | ||
c277df42 | 90 | #define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN) |
72311751 GS |
91 | #define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN) |
92 | #define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN) | |
93 | #define RX_MATCH_TAINTED_set(prog, t) ((t) \ | |
94 | ? RX_MATCH_TAINTED_on(prog) \ | |
95 | : RX_MATCH_TAINTED_off(prog)) | |
c277df42 | 96 | |
cf93c79d IZ |
97 | #define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE) |
98 | #define RX_MATCH_COPIED_on(prog) ((prog)->reganch |= ROPT_COPY_DONE) | |
99 | #define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE) | |
100 | #define RX_MATCH_COPIED_set(prog,t) ((t) \ | |
101 | ? RX_MATCH_COPIED_on(prog) \ | |
102 | : RX_MATCH_COPIED_off(prog)) | |
103 | ||
2ea5368e JH |
104 | #define RX_MATCH_UTF8(prog) ((prog)->reganch & ROPT_MATCH_UTF8) |
105 | #define RX_MATCH_UTF8_on(prog) ((prog)->reganch |= ROPT_MATCH_UTF8) | |
106 | #define RX_MATCH_UTF8_off(prog) ((prog)->reganch &= ~ROPT_MATCH_UTF8) | |
107 | #define RX_MATCH_UTF8_set(prog, t) ((t) \ | |
108 | ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ | |
109 | : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) | |
110 | ||
cad2e5aa JH |
111 | #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ |
112 | #define REXEC_CHECKED 0x02 /* check_substr already checked. */ | |
113 | #define REXEC_SCREAM 0x04 /* use scream table. */ | |
114 | #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ | |
cf93c79d | 115 | #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ |
cad2e5aa | 116 | #define REXEC_ML 0x20 /* $* was set. */ |
c277df42 | 117 | |
155aba94 | 118 | #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re) |
cad2e5aa | 119 | #define ReREFCNT_dec(re) CALLREGFREE(aTHX_ re) |
cf93c79d IZ |
120 | |
121 | #define FBMcf_TAIL_DOLLAR 1 | |
cad2e5aa JH |
122 | #define FBMcf_TAIL_DOLLARM 2 |
123 | #define FBMcf_TAIL_Z 4 | |
124 | #define FBMcf_TAIL_z 8 | |
125 | #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) | |
cf93c79d IZ |
126 | |
127 | #define FBMrf_MULTILINE 1 | |
cad2e5aa JH |
128 | |
129 | struct re_scream_pos_data_s; | |
4c49a5e0 NC |
130 | |
131 | struct re_save_state { | |
132 | U32 re_state_reg_flags; /* from regexec.c */ | |
133 | char *re_state_bostr; | |
134 | char *re_state_reginput; /* String-input pointer. */ | |
135 | char *re_state_regbol; /* Beginning of input, for ^ check. */ | |
136 | char *re_state_regeol; /* End of input, for $ check. */ | |
137 | I32 *re_state_regstartp; /* Pointer to startp array. */ | |
138 | I32 *re_state_regendp; /* Ditto for endp. */ | |
139 | U32 *re_state_reglastparen; /* Similarly for lastparen. */ | |
140 | U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */ | |
141 | char *re_state_regtill; /* How far we are required to go. */ | |
142 | char **re_state_reg_start_tmp; /* from regexec.c */ | |
143 | U32 re_state_reg_start_tmpl; /* from regexec.c */ | |
144 | I32 re_state_reg_eval_set; /* from regexec.c */ | |
145 | I32 re_state_regnarrate; /* from regexec.c */ | |
146 | int re_state_regindent; /* from regexec.c */ | |
147 | struct re_cc_state *re_state_reg_call_cc; /* from regexec.c */ | |
148 | regexp *re_state_reg_re; /* from regexec.c */ | |
149 | char *re_state_reg_ganch; /* from regexec.c */ | |
150 | SV *re_state_reg_sv; /* from regexec.c */ | |
151 | bool re_state_reg_match_utf8; /* from regexec.c */ | |
152 | MAGIC *re_state_reg_magic; /* from regexec.c */ | |
153 | I32 re_state_reg_oldpos; /* from regexec.c */ | |
154 | PMOP *re_state_reg_oldcurpm; /* from regexec.c */ | |
155 | PMOP *re_state_reg_curpm; /* from regexec.c */ | |
156 | char *re_state_reg_oldsaved; /* old saved substr during match */ | |
157 | STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */ | |
158 | I32 re_state_reg_maxiter; /* max wait until caching pos */ | |
159 | I32 re_state_reg_leftiter; /* wait until caching pos */ | |
160 | char *re_state_reg_poscache; /* cache of pos of WHILEM */ | |
161 | STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */ | |
162 | I32 re_state_regsize; /* from regexec.c */ | |
163 | char *re_state_reg_starttry; /* from regexec.c */ | |
164 | ||
165 | struct reg_data *re_state_regdata; /* from regexec.c renamed was data */ | |
166 | regnode *re_state_regprogram; /* from regexec.c */ | |
167 | struct curcur *re_state_regcc; /* from regexec.c */ | |
168 | char *re_state_regprecomp; /* uncompiled string. */ | |
169 | I32 re_state_regnpar; /* () count. */ | |
170 | }; | |
171 | ||
172 | #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \ | |
173 | (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack))) | |
174 | /* | |
175 | * Local variables: | |
176 | * c-indentation-style: bsd | |
177 | * c-basic-offset: 4 | |
178 | * indent-tabs-mode: t | |
179 | * End: | |
180 | * | |
181 | * ex: set ts=8 sts=4 sw=4 noet: | |
182 | */ |