| 1 | /* regexp.h |
| 2 | * |
| 3 | * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003, |
| 4 | * by Larry Wall and others |
| 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 | * |
| 9 | */ |
| 10 | |
| 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 | |
| 18 | |
| 19 | struct regnode { |
| 20 | U8 flags; |
| 21 | U8 type; |
| 22 | U16 next_off; |
| 23 | }; |
| 24 | |
| 25 | typedef struct regnode regnode; |
| 26 | |
| 27 | struct reg_substr_data; |
| 28 | |
| 29 | struct reg_data; |
| 30 | |
| 31 | typedef struct regexp { |
| 32 | I32 *startp; |
| 33 | I32 *endp; |
| 34 | regnode *regstclass; |
| 35 | struct reg_substr_data *substrs; |
| 36 | char *precomp; /* pre-compilation regular expression */ |
| 37 | struct reg_data *data; /* Additional data. */ |
| 38 | char *subbeg; /* saved or original string |
| 39 | so \digit works forever. */ |
| 40 | #ifdef PERL_OLD_COPY_ON_WRITE |
| 41 | SV *saved_copy; /* If non-NULL, SV which is COW from original */ |
| 42 | #endif |
| 43 | U32 *offsets; /* offset annotations 20001228 MJD */ |
| 44 | I32 sublen; /* Length of string pointed by subbeg */ |
| 45 | I32 refcnt; |
| 46 | I32 minlen; /* mininum possible length of $& */ |
| 47 | I32 prelen; /* length of precomp */ |
| 48 | U32 nparens; /* number of parentheses */ |
| 49 | U32 lastparen; /* last paren matched */ |
| 50 | U32 lastcloseparen; /* last paren matched */ |
| 51 | U32 reganch; /* Internal use only + |
| 52 | Tainted information used by regexec? */ |
| 53 | regnode program[1]; /* Unwarranted chumminess with compiler. */ |
| 54 | } regexp; |
| 55 | |
| 56 | #define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS|ROPT_ANCH_SBOL) |
| 57 | #define ROPT_ANCH_SINGLE (ROPT_ANCH_SBOL|ROPT_ANCH_GPOS) |
| 58 | #define ROPT_ANCH_BOL 0x00001 |
| 59 | #define ROPT_ANCH_MBOL 0x00002 |
| 60 | #define ROPT_ANCH_SBOL 0x00004 |
| 61 | #define ROPT_ANCH_GPOS 0x00008 |
| 62 | #define ROPT_SKIP 0x00010 |
| 63 | #define ROPT_IMPLICIT 0x00020 /* Converted .* to ^.* */ |
| 64 | #define ROPT_NOSCAN 0x00040 /* Check-string always at start. */ |
| 65 | #define ROPT_GPOS_SEEN 0x00080 |
| 66 | #define ROPT_CHECK_ALL 0x00100 |
| 67 | #define ROPT_LOOKBEHIND_SEEN 0x00200 |
| 68 | #define ROPT_EVAL_SEEN 0x00400 |
| 69 | #define ROPT_CANY_SEEN 0x00800 |
| 70 | #define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */ |
| 71 | |
| 72 | /* 0xf800 of reganch is used by PMf_COMPILETIME */ |
| 73 | |
| 74 | #define ROPT_UTF8 0x10000 |
| 75 | #define ROPT_NAUGHTY 0x20000 /* how exponential is this pattern? */ |
| 76 | #define ROPT_COPY_DONE 0x40000 /* subbeg is a copy of the string */ |
| 77 | #define ROPT_TAINTED_SEEN 0x80000 |
| 78 | #define ROPT_MATCH_UTF8 0x10000000 /* subbeg is utf-8 */ |
| 79 | |
| 80 | #define RE_USE_INTUIT_NOML 0x0100000 /* Best to intuit before matching */ |
| 81 | #define RE_USE_INTUIT_ML 0x0200000 |
| 82 | #define REINT_AUTORITATIVE_NOML 0x0400000 /* Can trust a positive answer */ |
| 83 | #define REINT_AUTORITATIVE_ML 0x0800000 |
| 84 | #define REINT_ONCE_NOML 0x1000000 /* Intuit can succed once only. */ |
| 85 | #define REINT_ONCE_ML 0x2000000 |
| 86 | #define RE_INTUIT_ONECHAR 0x4000000 |
| 87 | #define RE_INTUIT_TAIL 0x8000000 |
| 88 | |
| 89 | #define RE_USE_INTUIT (RE_USE_INTUIT_NOML|RE_USE_INTUIT_ML) |
| 90 | #define REINT_AUTORITATIVE (REINT_AUTORITATIVE_NOML|REINT_AUTORITATIVE_ML) |
| 91 | #define REINT_ONCE (REINT_ONCE_NOML|REINT_ONCE_ML) |
| 92 | |
| 93 | #define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN) |
| 94 | #define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN) |
| 95 | #define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN) |
| 96 | #define RX_MATCH_TAINTED_set(prog, t) ((t) \ |
| 97 | ? RX_MATCH_TAINTED_on(prog) \ |
| 98 | : RX_MATCH_TAINTED_off(prog)) |
| 99 | |
| 100 | #define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE) |
| 101 | #define RX_MATCH_COPIED_on(prog) ((prog)->reganch |= ROPT_COPY_DONE) |
| 102 | #define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE) |
| 103 | #define RX_MATCH_COPIED_set(prog,t) ((t) \ |
| 104 | ? RX_MATCH_COPIED_on(prog) \ |
| 105 | : RX_MATCH_COPIED_off(prog)) |
| 106 | |
| 107 | #ifdef PERL_OLD_COPY_ON_WRITE |
| 108 | #define RX_MATCH_COPY_FREE(rx) \ |
| 109 | STMT_START {if (rx->saved_copy) { \ |
| 110 | SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \ |
| 111 | } \ |
| 112 | if (RX_MATCH_COPIED(rx)) { \ |
| 113 | Safefree(rx->subbeg); \ |
| 114 | RX_MATCH_COPIED_off(rx); \ |
| 115 | }} STMT_END |
| 116 | #else |
| 117 | #define RX_MATCH_COPY_FREE(rx) \ |
| 118 | STMT_START {if (RX_MATCH_COPIED(rx)) { \ |
| 119 | Safefree(rx->subbeg); \ |
| 120 | RX_MATCH_COPIED_off(rx); \ |
| 121 | }} STMT_END |
| 122 | #endif |
| 123 | |
| 124 | #define RX_MATCH_UTF8(prog) ((prog)->reganch & ROPT_MATCH_UTF8) |
| 125 | #define RX_MATCH_UTF8_on(prog) ((prog)->reganch |= ROPT_MATCH_UTF8) |
| 126 | #define RX_MATCH_UTF8_off(prog) ((prog)->reganch &= ~ROPT_MATCH_UTF8) |
| 127 | #define RX_MATCH_UTF8_set(prog, t) ((t) \ |
| 128 | ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ |
| 129 | : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) |
| 130 | |
| 131 | #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ |
| 132 | #define REXEC_CHECKED 0x02 /* check_substr already checked. */ |
| 133 | #define REXEC_SCREAM 0x04 /* use scream table. */ |
| 134 | #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ |
| 135 | #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ |
| 136 | |
| 137 | #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re) |
| 138 | #define ReREFCNT_dec(re) CALLREGFREE(aTHX_ re) |
| 139 | |
| 140 | #define FBMcf_TAIL_DOLLAR 1 |
| 141 | #define FBMcf_TAIL_DOLLARM 2 |
| 142 | #define FBMcf_TAIL_Z 4 |
| 143 | #define FBMcf_TAIL_z 8 |
| 144 | #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) |
| 145 | |
| 146 | #define FBMrf_MULTILINE 1 |
| 147 | |
| 148 | struct re_scream_pos_data_s; |
| 149 | |
| 150 | /* an accepting state/position*/ |
| 151 | struct _reg_trie_accepted { |
| 152 | U8 *endpos; |
| 153 | U16 wordnum; |
| 154 | }; |
| 155 | typedef struct _reg_trie_accepted reg_trie_accepted; |
| 156 | |
| 157 | |
| 158 | /* structures for holding and saving the state maintained by regmatch() */ |
| 159 | |
| 160 | typedef I32 CHECKPOINT; |
| 161 | |
| 162 | /* Current curly descriptor */ |
| 163 | typedef struct curcur CURCUR; |
| 164 | struct curcur { |
| 165 | int parenfloor; /* how far back to strip paren data */ |
| 166 | int cur; /* how many instances of scan we've matched */ |
| 167 | int min; /* the minimal number of scans to match */ |
| 168 | int max; /* the maximal number of scans to match */ |
| 169 | int minmod; /* whether to work our way up or down */ |
| 170 | regnode * scan; /* the thing to match */ |
| 171 | regnode * next; /* what has to match after it */ |
| 172 | char * lastloc; /* where we started matching this scan */ |
| 173 | CURCUR * oldcc; /* current curly before we started this one */ |
| 174 | }; |
| 175 | |
| 176 | typedef struct re_cc_state |
| 177 | { |
| 178 | I32 ss; |
| 179 | regnode *node; |
| 180 | struct re_cc_state *prev; |
| 181 | CURCUR *cc; |
| 182 | regexp *re; |
| 183 | } re_cc_state; |
| 184 | |
| 185 | |
| 186 | |
| 187 | typedef enum { |
| 188 | resume_TRIE1, |
| 189 | resume_TRIE2, |
| 190 | resume_CURLYX, |
| 191 | resume_WHILEM1, |
| 192 | resume_WHILEM2, |
| 193 | resume_WHILEM3, |
| 194 | resume_WHILEM4, |
| 195 | resume_WHILEM5, |
| 196 | resume_WHILEM6, |
| 197 | resume_CURLYM1, |
| 198 | resume_CURLYM2, |
| 199 | resume_CURLYM3, |
| 200 | resume_CURLYM4, |
| 201 | resume_IFMATCH, |
| 202 | resume_PLUS1, |
| 203 | resume_PLUS2, |
| 204 | resume_PLUS3, |
| 205 | resume_PLUS4, |
| 206 | resume_END |
| 207 | } regmatch_resume_states; |
| 208 | |
| 209 | |
| 210 | typedef struct { |
| 211 | |
| 212 | /* these vars contain state that needs to be maintained |
| 213 | * across the main while loop ... */ |
| 214 | |
| 215 | regmatch_resume_states resume_state; /* where to jump to on return */ |
| 216 | regnode *scan; /* Current node. */ |
| 217 | regnode *next; /* Next node. */ |
| 218 | bool minmod; /* the next "{n.m}" is a "{n,m}?" */ |
| 219 | bool sw; /* the condition value in (?(cond)a|b) */ |
| 220 | int logical; |
| 221 | I32 unwind; /* savestack index of current unwind block */ |
| 222 | CURCUR *cc; /* current innermost curly struct */ |
| 223 | char *locinput; |
| 224 | |
| 225 | /* ... while the rest of these are local to an individual branch */ |
| 226 | |
| 227 | I32 n; /* no or next */ |
| 228 | I32 ln; /* len or last */ |
| 229 | |
| 230 | union { |
| 231 | struct { |
| 232 | reg_trie_accepted *accept_buff; |
| 233 | U32 accepted; /* how many accepting states we have seen */ |
| 234 | } trie; |
| 235 | |
| 236 | struct { |
| 237 | CHECKPOINT cp; /* remember current savestack indexes */ |
| 238 | CHECKPOINT lastcp; |
| 239 | } eval; |
| 240 | |
| 241 | struct { |
| 242 | CHECKPOINT cp; /* remember current savestack indexes */ |
| 243 | CURCUR *savecc; |
| 244 | } curlyx; |
| 245 | |
| 246 | struct { |
| 247 | CHECKPOINT cp; /* remember current savestack indexes */ |
| 248 | CHECKPOINT lastcp; |
| 249 | CURCUR *savecc; |
| 250 | char *lastloc; /* Detection of 0-len. */ |
| 251 | I32 cache_offset; |
| 252 | I32 cache_bit; |
| 253 | } whilem; |
| 254 | |
| 255 | struct { |
| 256 | I32 paren; |
| 257 | I32 c1, c2; /* case fold search */ |
| 258 | CHECKPOINT lastcp; |
| 259 | I32 l; |
| 260 | I32 matches; |
| 261 | I32 maxwanted; |
| 262 | } curlym; |
| 263 | |
| 264 | struct { |
| 265 | I32 paren; |
| 266 | CHECKPOINT lastcp; |
| 267 | I32 c1, c2; /* case fold search */ |
| 268 | char *e; |
| 269 | char *old; |
| 270 | int count; |
| 271 | } plus; /* and CURLYN/CURLY/STAR */ |
| 272 | |
| 273 | struct { |
| 274 | CHECKPOINT cp; /* remember current savestack indexes */ |
| 275 | CHECKPOINT lastcp; |
| 276 | CURCUR *savecc; |
| 277 | re_cc_state *cur_call_cc; |
| 278 | regexp *end_re; |
| 279 | } end; |
| 280 | }; |
| 281 | |
| 282 | re_cc_state *reg_call_cc; /* saved value of PL_reg_call_cc */ |
| 283 | } regmatch_state; |
| 284 | |
| 285 | /* how many regmatch_state structs to allocate as a single slab. |
| 286 | * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev |
| 287 | * pointers, plus 1 for any mythical malloc overhead. */ |
| 288 | |
| 289 | #define PERL_REGMATCH_SLAB_SLOTS \ |
| 290 | ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state)) |
| 291 | |
| 292 | typedef struct regmatch_slab { |
| 293 | regmatch_state states[PERL_REGMATCH_SLAB_SLOTS]; |
| 294 | struct regmatch_slab *prev, *next; |
| 295 | } regmatch_slab; |