| File: | build/source/llvm/lib/Support/regengine.inc |
| Warning: | line 839, column 35 Dereference of null pointer |
Press '?' to see keyboard shortcuts
Keyboard shortcuts:
| 1 | /*- | |||
| 2 | * This code is derived from OpenBSD's libc/regex, original license follows: | |||
| 3 | * | |||
| 4 | * Copyright (c) 1992, 1993, 1994 Henry Spencer. | |||
| 5 | * Copyright (c) 1992, 1993, 1994 | |||
| 6 | * The Regents of the University of California. All rights reserved. | |||
| 7 | * | |||
| 8 | * This code is derived from software contributed to Berkeley by | |||
| 9 | * Henry Spencer. | |||
| 10 | * | |||
| 11 | * Redistribution and use in source and binary forms, with or without | |||
| 12 | * modification, are permitted provided that the following conditions | |||
| 13 | * are met: | |||
| 14 | * 1. Redistributions of source code must retain the above copyright | |||
| 15 | * notice, this list of conditions and the following disclaimer. | |||
| 16 | * 2. Redistributions in binary form must reproduce the above copyright | |||
| 17 | * notice, this list of conditions and the following disclaimer in the | |||
| 18 | * documentation and/or other materials provided with the distribution. | |||
| 19 | * 3. Neither the name of the University nor the names of its contributors | |||
| 20 | * may be used to endorse or promote products derived from this software | |||
| 21 | * without specific prior written permission. | |||
| 22 | * | |||
| 23 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND | |||
| 24 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE | |||
| 25 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE | |||
| 26 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE | |||
| 27 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL | |||
| 28 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS | |||
| 29 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) | |||
| 30 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT | |||
| 31 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY | |||
| 32 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF | |||
| 33 | * SUCH DAMAGE. | |||
| 34 | * | |||
| 35 | * @(#)regexec.c 8.3 (Berkeley) 3/20/94 | |||
| 36 | */ | |||
| 37 | ||||
| 38 | /* | |||
| 39 | * the outer shell of llvm_regexec() | |||
| 40 | * | |||
| 41 | * This file includes engine.inc *twice*, after muchos fiddling with the | |||
| 42 | * macros that code uses. This lets the same code operate on two different | |||
| 43 | * representations for state sets. | |||
| 44 | */ | |||
| 45 | #include <sys/types.h> | |||
| 46 | #include <stdio.h> | |||
| 47 | #include <stdlib.h> | |||
| 48 | #include <string.h> | |||
| 49 | #include <limits.h> | |||
| 50 | #include <ctype.h> | |||
| 51 | #include "regex_impl.h" | |||
| 52 | ||||
| 53 | #include "regutils.h" | |||
| 54 | #include "regex2.h" | |||
| 55 | ||||
| 56 | /* macros for manipulating states, small version */ | |||
| 57 | /* FIXME: 'states' is assumed as 'long' on small version. */ | |||
| 58 | #define states1long long /* for later use in llvm_regexec() decision */ | |||
| 59 | #define stateschar * states1long | |||
| 60 | #define CLEAR(v)memset(v, 0, m->g->nstates) ((v) = 0) | |||
| 61 | #define SET0(v, n)((v)[n] = 0) ((v) &= ~((unsigned long)1 << (n))) | |||
| 62 | #define SET1(v, n)((v)[n] = 1) ((v) |= (unsigned long)1 << (n)) | |||
| 63 | #define ISSET(v, n)((v)[n]) (((v) & ((unsigned long)1 << (n))) != 0) | |||
| 64 | #define ASSIGN(d, s)memmove(d, s, m->g->nstates) ((d) = (s)) | |||
| 65 | #define EQ(a, b)(memcmp(a, b, m->g->nstates) == 0) ((a) == (b)) | |||
| 66 | #define STATEVARSlong vn; char *space long dummy /* dummy version */ | |||
| 67 | #define STATESETUP(m, n){ (m)->space = malloc((n)*(m)->g->nstates); if ((m)-> space == ((void*)0)) return(12); (m)->vn = 0; } /* nothing */ | |||
| 68 | #define STATETEARDOWN(m){ free((m)->space); } /* nothing */ | |||
| 69 | #define SETUP(v)((v) = &m->space[m->vn++ * m->g->nstates]) ((v) = 0) | |||
| 70 | #define onestatelong long | |||
| 71 | #define INIT(o, n)((o) = (n)) ((o) = (unsigned long)1 << (n)) | |||
| 72 | #define INC(o)((o)++) ((o) = (unsigned long)(o) << 1) | |||
| 73 | #define ISSTATEIN(v, o)((v)[o]) (((v) & (o)) != 0) | |||
| 74 | /* some abbreviations; note that some of these know variable names! */ | |||
| 75 | /* do "if I'm here, I can also be there" etc without branches */ | |||
| 76 | #define FWD(dst, src, n)((dst)[here+(n)] |= (src)[here]) ((dst) |= ((unsigned long)(src)&(here)) << (n)) | |||
| 77 | #define BACK(dst, src, n)((dst)[here-(n)] |= (src)[here]) ((dst) |= ((unsigned long)(src)&(here)) >> (n)) | |||
| 78 | #define ISSETBACK(v, n)((v)[here - (n)]) (((v) & ((unsigned long)here >> (n))) != 0) | |||
| 79 | /* function names */ | |||
| 80 | #define SNAMES /* engine.inc looks after details */ | |||
| 81 | ||||
| 82 | #include "regengine.inc" | |||
| 83 | ||||
| 84 | /* now undo things */ | |||
| 85 | #undef stateschar * | |||
| 86 | #undef CLEAR | |||
| 87 | #undef SET0 | |||
| 88 | #undef SET1 | |||
| 89 | #undef ISSET | |||
| 90 | #undef ASSIGN | |||
| 91 | #undef EQ | |||
| 92 | #undef STATEVARSlong vn; char *space | |||
| 93 | #undef STATESETUP | |||
| 94 | #undef STATETEARDOWN | |||
| 95 | #undef SETUP | |||
| 96 | #undef onestatelong | |||
| 97 | #undef INIT | |||
| 98 | #undef INC | |||
| 99 | #undef ISSTATEIN | |||
| 100 | #undef FWD | |||
| 101 | #undef BACK | |||
| 102 | #undef ISSETBACK | |||
| 103 | #undef SNAMES | |||
| 104 | ||||
| 105 | /* macros for manipulating states, large version */ | |||
| 106 | #define stateschar * char * | |||
| 107 | #define CLEAR(v)memset(v, 0, m->g->nstates) memset(v, 0, m->g->nstates) | |||
| 108 | #define SET0(v, n)((v)[n] = 0) ((v)[n] = 0) | |||
| 109 | #define SET1(v, n)((v)[n] = 1) ((v)[n] = 1) | |||
| 110 | #define ISSET(v, n)((v)[n]) ((v)[n]) | |||
| 111 | #define ASSIGN(d, s)memmove(d, s, m->g->nstates) memmove(d, s, m->g->nstates) | |||
| 112 | #define EQ(a, b)(memcmp(a, b, m->g->nstates) == 0) (memcmp(a, b, m->g->nstates) == 0) | |||
| 113 | #define STATEVARSlong vn; char *space long vn; char *space | |||
| 114 | #define STATESETUP(m, nv){ (m)->space = malloc((nv)*(m)->g->nstates); if ((m) ->space == ((void*)0)) return(12); (m)->vn = 0; } { (m)->space = malloc((nv)*(m)->g->nstates); \ | |||
| 115 | if ((m)->space == NULL((void*)0)) return(REG_ESPACE12); \ | |||
| 116 | (m)->vn = 0; } | |||
| 117 | #define STATETEARDOWN(m){ free((m)->space); } { free((m)->space); } | |||
| 118 | #define SETUP(v)((v) = &m->space[m->vn++ * m->g->nstates]) ((v) = &m->space[m->vn++ * m->g->nstates]) | |||
| 119 | #define onestatelong long | |||
| 120 | #define INIT(o, n)((o) = (n)) ((o) = (n)) | |||
| 121 | #define INC(o)((o)++) ((o)++) | |||
| 122 | #define ISSTATEIN(v, o)((v)[o]) ((v)[o]) | |||
| 123 | /* some abbreviations; note that some of these know variable names! */ | |||
| 124 | /* do "if I'm here, I can also be there" etc without branches */ | |||
| 125 | #define FWD(dst, src, n)((dst)[here+(n)] |= (src)[here]) ((dst)[here+(n)] |= (src)[here]) | |||
| 126 | #define BACK(dst, src, n)((dst)[here-(n)] |= (src)[here]) ((dst)[here-(n)] |= (src)[here]) | |||
| 127 | #define ISSETBACK(v, n)((v)[here - (n)]) ((v)[here - (n)]) | |||
| 128 | /* function names */ | |||
| 129 | #define LNAMES /* flag */ | |||
| 130 | ||||
| 131 | #include "regengine.inc" | |||
| 132 | ||||
| 133 | /* | |||
| 134 | - llvm_regexec - interface for matching | |||
| 135 | * | |||
| 136 | * We put this here so we can exploit knowledge of the state representation | |||
| 137 | * when choosing which matcher to call. Also, by this point the matchers | |||
| 138 | * have been prototyped. | |||
| 139 | */ | |||
| 140 | int /* 0 success, REG_NOMATCH failure */ | |||
| 141 | llvm_regexec(const llvm_regex_t *preg, const char *string, size_t nmatch, | |||
| 142 | llvm_regmatch_t pmatch[], int eflags) | |||
| 143 | { | |||
| 144 | struct re_guts *g = preg->re_g; | |||
| 145 | #ifdef REDEBUG | |||
| 146 | # define GOODFLAGS(f)((f)&(00001|00002|00004)) (f) | |||
| 147 | #else | |||
| 148 | # define GOODFLAGS(f)((f)&(00001|00002|00004)) ((f)&(REG_NOTBOL00001|REG_NOTEOL00002|REG_STARTEND00004)) | |||
| 149 | #endif | |||
| 150 | ||||
| 151 | if (preg->re_magic != MAGIC1((('r'^0200)<<8) | 'e') || g->magic != MAGIC2((('R'^0200)<<8)|'E')) | |||
| ||||
| 152 | return(REG_BADPAT2); | |||
| 153 | assert(!(g->iflags®EX_BAD))((void) (0)); | |||
| 154 | if (g->iflags®EX_BAD04) /* backstop for no-debug case */ | |||
| 155 | return(REG_BADPAT2); | |||
| 156 | eflags = GOODFLAGS(eflags)((eflags)&(00001|00002|00004)); | |||
| 157 | ||||
| 158 | if (g->nstates <= (long)(CHAR_BIT8*sizeof(states1long)) && !(eflags®_LARGE01000)) | |||
| 159 | return(smatcher(g, string, nmatch, pmatch, eflags)); | |||
| 160 | else | |||
| 161 | return(lmatcher(g, string, nmatch, pmatch, eflags)); | |||
| 162 | } |
| 1 | /*- | ||||
| 2 | * This code is derived from OpenBSD's libc/regex, original license follows: | ||||
| 3 | * | ||||
| 4 | * Copyright (c) 1992, 1993, 1994 Henry Spencer. | ||||
| 5 | * Copyright (c) 1992, 1993, 1994 | ||||
| 6 | * The Regents of the University of California. All rights reserved. | ||||
| 7 | * | ||||
| 8 | * This code is derived from software contributed to Berkeley by | ||||
| 9 | * Henry Spencer. | ||||
| 10 | * | ||||
| 11 | * Redistribution and use in source and binary forms, with or without | ||||
| 12 | * modification, are permitted provided that the following conditions | ||||
| 13 | * are met: | ||||
| 14 | * 1. Redistributions of source code must retain the above copyright | ||||
| 15 | * notice, this list of conditions and the following disclaimer. | ||||
| 16 | * 2. Redistributions in binary form must reproduce the above copyright | ||||
| 17 | * notice, this list of conditions and the following disclaimer in the | ||||
| 18 | * documentation and/or other materials provided with the distribution. | ||||
| 19 | * 3. Neither the name of the University nor the names of its contributors | ||||
| 20 | * may be used to endorse or promote products derived from this software | ||||
| 21 | * without specific prior written permission. | ||||
| 22 | * | ||||
| 23 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND | ||||
| 24 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE | ||||
| 25 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE | ||||
| 26 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE | ||||
| 27 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL | ||||
| 28 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS | ||||
| 29 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) | ||||
| 30 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT | ||||
| 31 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY | ||||
| 32 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF | ||||
| 33 | * SUCH DAMAGE. | ||||
| 34 | * | ||||
| 35 | * @(#)engine.c 8.5 (Berkeley) 3/20/94 | ||||
| 36 | */ | ||||
| 37 | |||||
| 38 | /* | ||||
| 39 | * The matching engine and friends. This file is #included by regexec.c | ||||
| 40 | * after suitable #defines of a variety of macros used herein, so that | ||||
| 41 | * different state representations can be used without duplicating masses | ||||
| 42 | * of code. | ||||
| 43 | */ | ||||
| 44 | |||||
| 45 | #ifdef SNAMES | ||||
| 46 | #define matcher smatcher | ||||
| 47 | #define fast sfast | ||||
| 48 | #define slow sslow | ||||
| 49 | #define dissect sdissect | ||||
| 50 | #define backref sbackref | ||||
| 51 | #define step sstep | ||||
| 52 | #define print sprint | ||||
| 53 | #define at sat | ||||
| 54 | #define match smat | ||||
| 55 | #define nope snope | ||||
| 56 | #define step_back sstep_back | ||||
| 57 | #endif | ||||
| 58 | #ifdef LNAMES | ||||
| 59 | #define matcher lmatcher | ||||
| 60 | #define fast lfast | ||||
| 61 | #define slow lslow | ||||
| 62 | #define dissect ldissect | ||||
| 63 | #define backref lbackref | ||||
| 64 | #define step lstep | ||||
| 65 | #define print lprint | ||||
| 66 | #define at lat | ||||
| 67 | #define match lmat | ||||
| 68 | #define nope lnope | ||||
| 69 | #define step_back lstep_back | ||||
| 70 | #endif | ||||
| 71 | |||||
| 72 | /* another structure passed up and down to avoid zillions of parameters */ | ||||
| 73 | struct match { | ||||
| 74 | struct re_guts *g; | ||||
| 75 | int eflags; | ||||
| 76 | llvm_regmatch_t *pmatch; /* [nsub+1] (0 element unused) */ | ||||
| 77 | const char *offp; /* offsets work from here */ | ||||
| 78 | const char *beginp; /* start of string -- virtual NUL precedes */ | ||||
| 79 | const char *endp; /* end of string -- virtual NUL here */ | ||||
| 80 | const char *coldp; /* can be no match starting before here */ | ||||
| 81 | const char **lastpos; /* [nplus+1] */ | ||||
| 82 | STATEVARSlong vn; char *space; | ||||
| 83 | stateschar * st; /* current states */ | ||||
| 84 | stateschar * fresh; /* states for a fresh start */ | ||||
| 85 | stateschar * tmp; /* temporary */ | ||||
| 86 | stateschar * empty; /* empty set of states */ | ||||
| 87 | }; | ||||
| 88 | |||||
| 89 | static int matcher(struct re_guts *, const char *, size_t, | ||||
| 90 | llvm_regmatch_t[], int); | ||||
| 91 | static const char *dissect(struct match *, const char *, const char *, sopno, | ||||
| 92 | sopno); | ||||
| 93 | static const char *backref(struct match *, const char *, const char *, sopno, | ||||
| 94 | sopno, sopno, int); | ||||
| 95 | static const char *fast(struct match *, const char *, const char *, sopno, sopno); | ||||
| 96 | static const char *slow(struct match *, const char *, const char *, sopno, sopno); | ||||
| 97 | static stateschar * step(struct re_guts *, sopno, sopno, stateschar *, int, stateschar *); | ||||
| 98 | #define MAX_RECURSION100 100 | ||||
| 99 | #define BOL((127 +1)+1) (OUT(127 +1)+1) | ||||
| 100 | #define EOL(((127 +1)+1)+1) (BOL((127 +1)+1)+1) | ||||
| 101 | #define BOLEOL(((127 +1)+1)+2) (BOL((127 +1)+1)+2) | ||||
| 102 | #define NOTHING(((127 +1)+1)+3) (BOL((127 +1)+1)+3) | ||||
| 103 | #define BOW(((127 +1)+1)+4) (BOL((127 +1)+1)+4) | ||||
| 104 | #define EOW(((127 +1)+1)+5) (BOL((127 +1)+1)+5) | ||||
| 105 | #define CODEMAX(((127 +1)+1)+5) (BOL((127 +1)+1)+5) /* highest code used */ | ||||
| 106 | #define NONCHAR(c)((c) > 127) ((c) > CHAR_MAX127) | ||||
| 107 | #define NNONCHAR((((127 +1)+1)+5)-127) (CODEMAX(((127 +1)+1)+5)-CHAR_MAX127) | ||||
| 108 | #ifdef REDEBUG | ||||
| 109 | static void print(struct match *, const char *, stateschar *, int, FILE *); | ||||
| 110 | #endif | ||||
| 111 | #ifdef REDEBUG | ||||
| 112 | static void at( | ||||
| 113 | struct match *, const char *, const char *, const char *, sopno, sopno); | ||||
| 114 | #endif | ||||
| 115 | #ifdef REDEBUG | ||||
| 116 | static char *pchar(int); | ||||
| 117 | #endif | ||||
| 118 | |||||
| 119 | #ifdef REDEBUG | ||||
| 120 | #define SP(t, s, c) print(m, t, s, c, stdoutstdout) | ||||
| 121 | #define AT(t, p1, p2, s1, s2) at(m, t, p1, p2, s1, s2) | ||||
| 122 | #define NOTE(str) { if (m->eflags®_TRACE00400) (void)printf("=%s\n", (str))__printf_chk (2 - 1, "=%s\n", (str)); } | ||||
| 123 | static int nope = 0; | ||||
| 124 | #else | ||||
| 125 | #define SP(t, s, c) /* nothing */ | ||||
| 126 | #define AT(t, p1, p2, s1, s2) /* nothing */ | ||||
| 127 | #define NOTE(s) /* nothing */ | ||||
| 128 | #endif | ||||
| 129 | |||||
| 130 | /* | ||||
| 131 | - matcher - the actual matching engine | ||||
| 132 | */ | ||||
| 133 | static int /* 0 success, REG_NOMATCH failure */ | ||||
| 134 | matcher(struct re_guts *g, const char *string, size_t nmatch, | ||||
| 135 | llvm_regmatch_t pmatch[], | ||||
| 136 | int eflags) | ||||
| 137 | { | ||||
| 138 | const char *endp; | ||||
| 139 | size_t i; | ||||
| 140 | struct match mv; | ||||
| 141 | struct match *m = &mv; | ||||
| 142 | const char *dp; | ||||
| 143 | const sopno gf = g->firststate+1; /* +1 for OEND */ | ||||
| 144 | const sopno gl = g->laststate; | ||||
| 145 | const char *start; | ||||
| 146 | const char *stop; | ||||
| 147 | |||||
| 148 | /* simplify the situation where possible */ | ||||
| 149 | if (g->cflags®_NOSUB0004) | ||||
| 150 | nmatch = 0; | ||||
| 151 | if (eflags®_STARTEND00004) { | ||||
| 152 | start = string + pmatch[0].rm_so; | ||||
| 153 | stop = string + pmatch[0].rm_eo; | ||||
| 154 | } else { | ||||
| 155 | start = string; | ||||
| 156 | stop = start + strlen(start); | ||||
| 157 | } | ||||
| 158 | if (stop < start) | ||||
| 159 | return(REG_INVARG16); | ||||
| 160 | |||||
| 161 | /* prescreening; this does wonders for this rather slow code */ | ||||
| 162 | if (g->must != NULL((void*)0)) { | ||||
| 163 | for (dp = start; dp < stop; dp++) | ||||
| 164 | if (*dp == g->must[0] && stop - dp >= g->mlen && | ||||
| 165 | memcmp(dp, g->must, (size_t)g->mlen) == 0) | ||||
| 166 | break; | ||||
| 167 | if (dp == stop) /* we didn't find g->must */ | ||||
| 168 | return(REG_NOMATCH1); | ||||
| 169 | } | ||||
| 170 | |||||
| 171 | /* match struct setup */ | ||||
| 172 | m->g = g; | ||||
| 173 | m->eflags = eflags; | ||||
| 174 | m->pmatch = NULL((void*)0); | ||||
| 175 | m->lastpos = NULL((void*)0); | ||||
| 176 | m->offp = string; | ||||
| 177 | m->beginp = start; | ||||
| 178 | m->endp = stop; | ||||
| 179 | STATESETUP(m, 4){ (m)->space = malloc((4)*(m)->g->nstates); if ((m)-> space == ((void*)0)) return(12); (m)->vn = 0; }; | ||||
| 180 | SETUP(m->st)((m->st) = &m->space[m->vn++ * m->g->nstates ]); | ||||
| 181 | SETUP(m->fresh)((m->fresh) = &m->space[m->vn++ * m->g->nstates ]); | ||||
| 182 | SETUP(m->tmp)((m->tmp) = &m->space[m->vn++ * m->g->nstates ]); | ||||
| 183 | SETUP(m->empty)((m->empty) = &m->space[m->vn++ * m->g->nstates ]); | ||||
| 184 | CLEAR(m->empty)memset(m->empty, 0, m->g->nstates); | ||||
| 185 | |||||
| 186 | /* this loop does only one repetition except for backrefs */ | ||||
| 187 | for (;;) { | ||||
| 188 | endp = fast(m, start, stop, gf, gl); | ||||
| 189 | if (endp
| ||||
| 190 | free(m->pmatch); | ||||
| 191 | free((void*)m->lastpos); | ||||
| 192 | STATETEARDOWN(m){ free((m)->space); }; | ||||
| 193 | return(REG_NOMATCH1); | ||||
| 194 | } | ||||
| 195 | if (nmatch == 0 && !g->backrefs) | ||||
| 196 | break; /* no further info needed */ | ||||
| 197 | |||||
| 198 | /* where? */ | ||||
| 199 | assert(m->coldp != NULL)((void) (0)); | ||||
| 200 | for (;;) { | ||||
| 201 | NOTE("finding start"); | ||||
| 202 | endp = slow(m, m->coldp, stop, gf, gl); | ||||
| 203 | if (endp != NULL((void*)0)) | ||||
| 204 | break; | ||||
| 205 | assert(m->coldp < m->endp)((void) (0)); | ||||
| 206 | m->coldp++; | ||||
| 207 | } | ||||
| 208 | if (nmatch == 1 && !g->backrefs) | ||||
| 209 | break; /* no further info needed */ | ||||
| 210 | |||||
| 211 | /* oh my, they want the subexpressions... */ | ||||
| 212 | if (m->pmatch == NULL((void*)0)) | ||||
| 213 | m->pmatch = (llvm_regmatch_t *)malloc((m->g->nsub + 1) * | ||||
| 214 | sizeof(llvm_regmatch_t)); | ||||
| 215 | if (m->pmatch == NULL((void*)0)) { | ||||
| 216 | STATETEARDOWN(m){ free((m)->space); }; | ||||
| 217 | return(REG_ESPACE12); | ||||
| 218 | } | ||||
| 219 | for (i = 1; i <= m->g->nsub; i++) | ||||
| 220 | m->pmatch[i].rm_so = m->pmatch[i].rm_eo = -1; | ||||
| 221 | if (!g->backrefs && !(m->eflags®_BACKR02000)) { | ||||
| 222 | NOTE("dissecting"); | ||||
| 223 | dp = dissect(m, m->coldp, endp, gf, gl); | ||||
| 224 | } else { | ||||
| 225 | if (g->nplus > 0 && m->lastpos == NULL((void*)0)) | ||||
| 226 | m->lastpos = (const char **)malloc((g->nplus+1) * | ||||
| 227 | sizeof(char *)); | ||||
| 228 | if (g->nplus > 0 && m->lastpos == NULL((void*)0)) { | ||||
| 229 | free(m->pmatch); | ||||
| 230 | STATETEARDOWN(m){ free((m)->space); }; | ||||
| 231 | return(REG_ESPACE12); | ||||
| 232 | } | ||||
| 233 | NOTE("backref dissect"); | ||||
| 234 | dp = backref(m, m->coldp, endp, gf, gl, (sopno)0, 0); | ||||
| 235 | } | ||||
| 236 | if (dp != NULL((void*)0)) | ||||
| 237 | break; | ||||
| 238 | |||||
| 239 | /* uh-oh... we couldn't find a subexpression-level match */ | ||||
| 240 | assert(g->backrefs)((void) (0)); /* must be back references doing it */ | ||||
| 241 | assert(g->nplus == 0 || m->lastpos != NULL)((void) (0)); | ||||
| 242 | for (;;) { | ||||
| 243 | if (dp != NULL((void*)0) || endp <= m->coldp) | ||||
| 244 | break; /* defeat */ | ||||
| 245 | NOTE("backoff"); | ||||
| 246 | endp = slow(m, m->coldp, endp-1, gf, gl); | ||||
| 247 | if (endp == NULL((void*)0)) | ||||
| 248 | break; /* defeat */ | ||||
| 249 | /* try it on a shorter possibility */ | ||||
| 250 | #ifndef NDEBUG | ||||
| 251 | for (i = 1; i <= m->g->nsub; i++) { | ||||
| 252 | assert(m->pmatch[i].rm_so == -1)((void) (0)); | ||||
| 253 | assert(m->pmatch[i].rm_eo == -1)((void) (0)); | ||||
| 254 | } | ||||
| 255 | #endif | ||||
| 256 | NOTE("backoff dissect"); | ||||
| 257 | dp = backref(m, m->coldp, endp, gf, gl, (sopno)0, 0); | ||||
| 258 | } | ||||
| 259 | assert(dp == NULL || dp == endp)((void) (0)); | ||||
| 260 | if (dp != NULL((void*)0)) /* found a shorter one */ | ||||
| 261 | break; | ||||
| 262 | |||||
| 263 | /* despite initial appearances, there is no match here */ | ||||
| 264 | NOTE("false alarm"); | ||||
| 265 | if (m->coldp == stop) | ||||
| 266 | break; | ||||
| 267 | start = m->coldp + 1; /* recycle starting later */ | ||||
| 268 | } | ||||
| 269 | |||||
| 270 | /* fill in the details if requested */ | ||||
| 271 | if (nmatch > 0) { | ||||
| 272 | pmatch[0].rm_so = m->coldp - m->offp; | ||||
| 273 | pmatch[0].rm_eo = endp - m->offp; | ||||
| 274 | } | ||||
| 275 | if (nmatch > 1) { | ||||
| 276 | assert(m->pmatch != NULL)((void) (0)); | ||||
| 277 | for (i = 1; i < nmatch; i++) | ||||
| 278 | if (i <= m->g->nsub) | ||||
| 279 | pmatch[i] = m->pmatch[i]; | ||||
| 280 | else { | ||||
| 281 | pmatch[i].rm_so = -1; | ||||
| 282 | pmatch[i].rm_eo = -1; | ||||
| 283 | } | ||||
| 284 | } | ||||
| 285 | |||||
| 286 | if (m->pmatch != NULL((void*)0)) | ||||
| 287 | free((char *)m->pmatch); | ||||
| 288 | if (m->lastpos != NULL((void*)0)) | ||||
| 289 | free((char *)m->lastpos); | ||||
| 290 | STATETEARDOWN(m){ free((m)->space); }; | ||||
| 291 | return(0); | ||||
| 292 | } | ||||
| 293 | |||||
| 294 | /* Step back from "stop" to a position where the strip startst..stopst might | ||||
| 295 | * match. This can always conservatively return "stop - 1", but may return an | ||||
| 296 | * earlier position if matches at later positions are impossible. */ | ||||
| 297 | static const char * | ||||
| 298 | step_back(struct re_guts *g, const char *start, const char *stop, sopno startst, | ||||
| 299 | sopno stopst) | ||||
| 300 | { | ||||
| 301 | /* Always step back at least one character. */ | ||||
| 302 | assert(stop > start)((void) (0)); | ||||
| 303 | const char *res = stop - 1; | ||||
| 304 | |||||
| 305 | /* Check whether the strip startst..stropst starts with a fixed character, | ||||
| 306 | * ignoring any closing parentheses. If not, return a conservative result. */ | ||||
| 307 | for (;;) { | ||||
| 308 | if (startst >= stopst) | ||||
| 309 | return res; | ||||
| 310 | if (OP(g->strip[startst])((g->strip[startst])&0xf8000000LU) != ORPAREN(14LU<<((unsigned)27))) | ||||
| 311 | break; | ||||
| 312 | startst++; | ||||
| 313 | } | ||||
| 314 | if (OP(g->strip[startst])((g->strip[startst])&0xf8000000LU) != OCHAR(2LU<<((unsigned)27))) | ||||
| 315 | return res; | ||||
| 316 | |||||
| 317 | /* Find the character that starts the following match. */ | ||||
| 318 | char ch = OPND(g->strip[startst])((g->strip[startst])&0x07ffffffLU); | ||||
| 319 | for (; res != start; --res) { | ||||
| 320 | if (*res == ch) { | ||||
| 321 | /* Try to check the next fixed character as well. */ | ||||
| 322 | sopno nextst = startst + 1; | ||||
| 323 | const char *next = res + 1; | ||||
| 324 | if (nextst >= stopst || OP(g->strip[nextst])((g->strip[nextst])&0xf8000000LU) != OCHAR(2LU<<((unsigned)27)) || next >= stop || | ||||
| 325 | *next == (char)OPND(g->strip[nextst])((g->strip[nextst])&0x07ffffffLU)) | ||||
| 326 | break; | ||||
| 327 | } | ||||
| 328 | } | ||||
| 329 | return res; | ||||
| 330 | } | ||||
| 331 | |||||
| 332 | /* | ||||
| 333 | - dissect - figure out what matched what, no back references | ||||
| 334 | */ | ||||
| 335 | static const char * /* == stop (success) always */ | ||||
| 336 | dissect(struct match *m, const char *start, const char *stop, sopno startst, | ||||
| 337 | sopno stopst) | ||||
| 338 | { | ||||
| 339 | int i; | ||||
| 340 | sopno ss; /* start sop of current subRE */ | ||||
| 341 | sopno es; /* end sop of current subRE */ | ||||
| 342 | const char *sp; /* start of string matched by it */ | ||||
| 343 | const char *stp; /* string matched by it cannot pass here */ | ||||
| 344 | const char *rest; /* start of rest of string */ | ||||
| 345 | const char *tail; /* string unmatched by rest of RE */ | ||||
| 346 | sopno ssub; /* start sop of subsubRE */ | ||||
| 347 | sopno esub; /* end sop of subsubRE */ | ||||
| 348 | const char *ssp; /* start of string matched by subsubRE */ | ||||
| 349 | const char *sep; /* end of string matched by subsubRE */ | ||||
| 350 | const char *oldssp; /* previous ssp */ | ||||
| 351 | |||||
| 352 | AT("diss", start, stop, startst, stopst); | ||||
| 353 | sp = start; | ||||
| 354 | for (ss = startst; ss < stopst; ss = es) { | ||||
| 355 | /* identify end of subRE */ | ||||
| 356 | es = ss; | ||||
| 357 | switch (OP(m->g->strip[es])((m->g->strip[es])&0xf8000000LU)) { | ||||
| 358 | case OPLUS_(9LU<<((unsigned)27)): | ||||
| 359 | case OQUEST_(11LU<<((unsigned)27)): | ||||
| 360 | es += OPND(m->g->strip[es])((m->g->strip[es])&0x07ffffffLU); | ||||
| 361 | break; | ||||
| 362 | case OCH_(15LU<<((unsigned)27)): | ||||
| 363 | while (OP(m->g->strip[es])((m->g->strip[es])&0xf8000000LU) != O_CH(18LU<<((unsigned)27))) | ||||
| 364 | es += OPND(m->g->strip[es])((m->g->strip[es])&0x07ffffffLU); | ||||
| 365 | break; | ||||
| 366 | } | ||||
| 367 | es++; | ||||
| 368 | |||||
| 369 | /* figure out what it matched */ | ||||
| 370 | switch (OP(m->g->strip[ss])((m->g->strip[ss])&0xf8000000LU)) { | ||||
| 371 | case OEND(1LU<<((unsigned)27)): | ||||
| 372 | assert(nope)((void) (0)); | ||||
| 373 | break; | ||||
| 374 | case OCHAR(2LU<<((unsigned)27)): | ||||
| 375 | sp++; | ||||
| 376 | break; | ||||
| 377 | case OBOL(3LU<<((unsigned)27)): | ||||
| 378 | case OEOL(4LU<<((unsigned)27)): | ||||
| 379 | case OBOW(19LU<<((unsigned)27)): | ||||
| 380 | case OEOW(20LU<<((unsigned)27)): | ||||
| 381 | break; | ||||
| 382 | case OANY(5LU<<((unsigned)27)): | ||||
| 383 | case OANYOF(6LU<<((unsigned)27)): | ||||
| 384 | sp++; | ||||
| 385 | break; | ||||
| 386 | case OBACK_(7LU<<((unsigned)27)): | ||||
| 387 | case O_BACK(8LU<<((unsigned)27)): | ||||
| 388 | assert(nope)((void) (0)); | ||||
| 389 | break; | ||||
| 390 | /* cases where length of match is hard to find */ | ||||
| 391 | case OQUEST_(11LU<<((unsigned)27)): | ||||
| 392 | stp = stop; | ||||
| 393 | for (;;) { | ||||
| 394 | /* how long could this one be? */ | ||||
| 395 | rest = slow(m, sp, stp, ss, es); | ||||
| 396 | assert(rest != NULL)((void) (0)); /* it did match */ | ||||
| 397 | /* could the rest match the rest? */ | ||||
| 398 | tail = slow(m, rest, stop, es, stopst); | ||||
| 399 | if (tail == stop) | ||||
| 400 | break; /* yes! */ | ||||
| 401 | /* no -- try a shorter match for this one */ | ||||
| 402 | stp = step_back(m->g, sp, rest, es, stopst); | ||||
| 403 | assert(stp >= sp)((void) (0)); /* it did work */ | ||||
| 404 | } | ||||
| 405 | ssub = ss + 1; | ||||
| 406 | esub = es - 1; | ||||
| 407 | /* did innards match? */ | ||||
| 408 | if (slow(m, sp, rest, ssub, esub) != NULL((void*)0)) { | ||||
| 409 | const char *dp = dissect(m, sp, rest, ssub, esub); | ||||
| 410 | (void)dp; /* avoid warning if assertions off */ | ||||
| 411 | assert(dp == rest)((void) (0)); | ||||
| 412 | } else /* no */ | ||||
| 413 | assert(sp == rest)((void) (0)); | ||||
| 414 | sp = rest; | ||||
| 415 | break; | ||||
| 416 | case OPLUS_(9LU<<((unsigned)27)): | ||||
| 417 | stp = stop; | ||||
| 418 | for (;;) { | ||||
| 419 | /* how long could this one be? */ | ||||
| 420 | rest = slow(m, sp, stp, ss, es); | ||||
| 421 | assert(rest != NULL)((void) (0)); /* it did match */ | ||||
| 422 | /* could the rest match the rest? */ | ||||
| 423 | tail = slow(m, rest, stop, es, stopst); | ||||
| 424 | if (tail == stop) | ||||
| 425 | break; /* yes! */ | ||||
| 426 | /* no -- try a shorter match for this one */ | ||||
| 427 | stp = step_back(m->g, sp, rest, es, stopst); | ||||
| 428 | assert(stp >= sp)((void) (0)); /* it did work */ | ||||
| 429 | } | ||||
| 430 | ssub = ss + 1; | ||||
| 431 | esub = es - 1; | ||||
| 432 | ssp = sp; | ||||
| 433 | oldssp = ssp; | ||||
| 434 | for (;;) { /* find last match of innards */ | ||||
| 435 | sep = slow(m, ssp, rest, ssub, esub); | ||||
| 436 | if (sep == NULL((void*)0) || sep == ssp) | ||||
| 437 | break; /* failed or matched null */ | ||||
| 438 | oldssp = ssp; /* on to next try */ | ||||
| 439 | ssp = sep; | ||||
| 440 | } | ||||
| 441 | if (sep == NULL((void*)0)) { | ||||
| 442 | /* last successful match */ | ||||
| 443 | sep = ssp; | ||||
| 444 | ssp = oldssp; | ||||
| 445 | } | ||||
| 446 | assert(sep == rest)((void) (0)); /* must exhaust substring */ | ||||
| 447 | assert(slow(m, ssp, sep, ssub, esub) == rest)((void) (0)); | ||||
| 448 | { | ||||
| 449 | const char *dp = dissect(m, ssp, sep, ssub, esub); | ||||
| 450 | (void)dp; /* avoid warning if assertions off */ | ||||
| 451 | assert(dp == sep)((void) (0)); | ||||
| 452 | } | ||||
| 453 | sp = rest; | ||||
| 454 | break; | ||||
| 455 | case OCH_(15LU<<((unsigned)27)): | ||||
| 456 | stp = stop; | ||||
| 457 | for (;;) { | ||||
| 458 | /* how long could this one be? */ | ||||
| 459 | rest = slow(m, sp, stp, ss, es); | ||||
| 460 | assert(rest != NULL)((void) (0)); /* it did match */ | ||||
| 461 | /* could the rest match the rest? */ | ||||
| 462 | tail = slow(m, rest, stop, es, stopst); | ||||
| 463 | if (tail == stop) | ||||
| 464 | break; /* yes! */ | ||||
| 465 | /* no -- try a shorter match for this one */ | ||||
| 466 | stp = rest - 1; | ||||
| 467 | assert(stp >= sp)((void) (0)); /* it did work */ | ||||
| 468 | } | ||||
| 469 | ssub = ss + 1; | ||||
| 470 | esub = ss + OPND(m->g->strip[ss])((m->g->strip[ss])&0x07ffffffLU) - 1; | ||||
| 471 | assert(OP(m->g->strip[esub]) == OOR1)((void) (0)); | ||||
| 472 | for (;;) { /* find first matching branch */ | ||||
| 473 | if (slow(m, sp, rest, ssub, esub) == rest) | ||||
| 474 | break; /* it matched all of it */ | ||||
| 475 | /* that one missed, try next one */ | ||||
| 476 | assert(OP(m->g->strip[esub]) == OOR1)((void) (0)); | ||||
| 477 | esub++; | ||||
| 478 | assert(OP(m->g->strip[esub]) == OOR2)((void) (0)); | ||||
| 479 | ssub = esub + 1; | ||||
| 480 | esub += OPND(m->g->strip[esub])((m->g->strip[esub])&0x07ffffffLU); | ||||
| 481 | if (OP(m->g->strip[esub])((m->g->strip[esub])&0xf8000000LU) == OOR2(17LU<<((unsigned)27))) | ||||
| 482 | esub--; | ||||
| 483 | else | ||||
| 484 | assert(OP(m->g->strip[esub]) == O_CH)((void) (0)); | ||||
| 485 | } | ||||
| 486 | { | ||||
| 487 | const char *dp = dissect(m, sp, rest, ssub, esub); | ||||
| 488 | (void)dp; /* avoid warning if assertions off */ | ||||
| 489 | assert(dp == rest)((void) (0)); | ||||
| 490 | } | ||||
| 491 | sp = rest; | ||||
| 492 | break; | ||||
| 493 | case O_PLUS(10LU<<((unsigned)27)): | ||||
| 494 | case O_QUEST(12LU<<((unsigned)27)): | ||||
| 495 | case OOR1(16LU<<((unsigned)27)): | ||||
| 496 | case OOR2(17LU<<((unsigned)27)): | ||||
| 497 | case O_CH(18LU<<((unsigned)27)): | ||||
| 498 | assert(nope)((void) (0)); | ||||
| 499 | break; | ||||
| 500 | case OLPAREN(13LU<<((unsigned)27)): | ||||
| 501 | i = OPND(m->g->strip[ss])((m->g->strip[ss])&0x07ffffffLU); | ||||
| 502 | assert(0 < i && i <= m->g->nsub)((void) (0)); | ||||
| 503 | m->pmatch[i].rm_so = sp - m->offp; | ||||
| 504 | break; | ||||
| 505 | case ORPAREN(14LU<<((unsigned)27)): | ||||
| 506 | i = OPND(m->g->strip[ss])((m->g->strip[ss])&0x07ffffffLU); | ||||
| 507 | assert(0 < i && i <= m->g->nsub)((void) (0)); | ||||
| 508 | m->pmatch[i].rm_eo = sp - m->offp; | ||||
| 509 | break; | ||||
| 510 | default: /* uh oh */ | ||||
| 511 | assert(nope)((void) (0)); | ||||
| 512 | break; | ||||
| 513 | } | ||||
| 514 | } | ||||
| 515 | |||||
| 516 | assert(sp == stop)((void) (0)); | ||||
| 517 | return(sp); | ||||
| 518 | } | ||||
| 519 | |||||
| 520 | /* | ||||
| 521 | - backref - figure out what matched what, figuring in back references | ||||
| 522 | */ | ||||
| 523 | static const char * /* == stop (success) or NULL (failure) */ | ||||
| 524 | backref(struct match *m, const char *start, const char *stop, sopno startst, | ||||
| 525 | sopno stopst, sopno lev, int rec) /* PLUS nesting level */ | ||||
| 526 | { | ||||
| 527 | int i; | ||||
| 528 | sopno ss; /* start sop of current subRE */ | ||||
| 529 | const char *sp; /* start of string matched by it */ | ||||
| 530 | sopno ssub; /* start sop of subsubRE */ | ||||
| 531 | sopno esub; /* end sop of subsubRE */ | ||||
| 532 | const char *ssp; /* start of string matched by subsubRE */ | ||||
| 533 | const char *dp; | ||||
| 534 | size_t len; | ||||
| 535 | int hard; | ||||
| 536 | sop s; | ||||
| 537 | llvm_regoff_t offsave; | ||||
| 538 | cset *cs; | ||||
| 539 | |||||
| 540 | AT("back", start, stop, startst, stopst); | ||||
| 541 | sp = start; | ||||
| 542 | |||||
| 543 | /* get as far as we can with easy stuff */ | ||||
| 544 | hard = 0; | ||||
| 545 | for (ss = startst; !hard && ss < stopst; ss++) | ||||
| 546 | switch (OP(s = m->g->strip[ss])((s = m->g->strip[ss])&0xf8000000LU)) { | ||||
| 547 | case OCHAR(2LU<<((unsigned)27)): | ||||
| 548 | if (sp == stop || *sp++ != (char)OPND(s)((s)&0x07ffffffLU)) | ||||
| 549 | return(NULL((void*)0)); | ||||
| 550 | break; | ||||
| 551 | case OANY(5LU<<((unsigned)27)): | ||||
| 552 | if (sp == stop) | ||||
| 553 | return(NULL((void*)0)); | ||||
| 554 | sp++; | ||||
| 555 | break; | ||||
| 556 | case OANYOF(6LU<<((unsigned)27)): | ||||
| 557 | cs = &m->g->sets[OPND(s)((s)&0x07ffffffLU)]; | ||||
| 558 | if (sp == stop || !CHIN(cs, *sp++)((cs)->ptr[(uch)(*sp++)] & (cs)->mask)) | ||||
| 559 | return(NULL((void*)0)); | ||||
| 560 | break; | ||||
| 561 | case OBOL(3LU<<((unsigned)27)): | ||||
| 562 | if ( (sp == m->beginp && !(m->eflags®_NOTBOL00001)) || | ||||
| 563 | (sp < m->endp && *(sp-1) == '\n' && | ||||
| 564 | (m->g->cflags®_NEWLINE0010)) ) | ||||
| 565 | { /* yes */ } | ||||
| 566 | else | ||||
| 567 | return(NULL((void*)0)); | ||||
| 568 | break; | ||||
| 569 | case OEOL(4LU<<((unsigned)27)): | ||||
| 570 | if ( (sp == m->endp && !(m->eflags®_NOTEOL00002)) || | ||||
| 571 | (sp < m->endp && *sp == '\n' && | ||||
| 572 | (m->g->cflags®_NEWLINE0010)) ) | ||||
| 573 | { /* yes */ } | ||||
| 574 | else | ||||
| 575 | return(NULL((void*)0)); | ||||
| 576 | break; | ||||
| 577 | case OBOW(19LU<<((unsigned)27)): | ||||
| 578 | if (( (sp == m->beginp && !(m->eflags®_NOTBOL00001)) || | ||||
| 579 | (sp < m->endp && *(sp-1) == '\n' && | ||||
| 580 | (m->g->cflags®_NEWLINE0010)) || | ||||
| 581 | (sp > m->beginp && | ||||
| 582 | !ISWORD(*(sp-1))(((*__ctype_b_loc ())[(int) ((*(sp-1)&0xff))] & (unsigned short int) _ISalnum) || (*(sp-1)) == '_')) ) && | ||||
| 583 | (sp < m->endp && ISWORD(*sp)(((*__ctype_b_loc ())[(int) ((*sp&0xff))] & (unsigned short int) _ISalnum) || (*sp) == '_')) ) | ||||
| 584 | { /* yes */ } | ||||
| 585 | else | ||||
| 586 | return(NULL((void*)0)); | ||||
| 587 | break; | ||||
| 588 | case OEOW(20LU<<((unsigned)27)): | ||||
| 589 | if (( (sp == m->endp && !(m->eflags®_NOTEOL00002)) || | ||||
| 590 | (sp < m->endp && *sp == '\n' && | ||||
| 591 | (m->g->cflags®_NEWLINE0010)) || | ||||
| 592 | (sp < m->endp && !ISWORD(*sp)(((*__ctype_b_loc ())[(int) ((*sp&0xff))] & (unsigned short int) _ISalnum) || (*sp) == '_')) ) && | ||||
| 593 | (sp > m->beginp && ISWORD(*(sp-1))(((*__ctype_b_loc ())[(int) ((*(sp-1)&0xff))] & (unsigned short int) _ISalnum) || (*(sp-1)) == '_')) ) | ||||
| 594 | { /* yes */ } | ||||
| 595 | else | ||||
| 596 | return(NULL((void*)0)); | ||||
| 597 | break; | ||||
| 598 | case O_QUEST(12LU<<((unsigned)27)): | ||||
| 599 | case O_CH(18LU<<((unsigned)27)): | ||||
| 600 | break; | ||||
| 601 | case OOR1(16LU<<((unsigned)27)): /* matches null but needs to skip */ | ||||
| 602 | ss++; | ||||
| 603 | s = m->g->strip[ss]; | ||||
| 604 | do { | ||||
| 605 | assert(OP(s) == OOR2)((void) (0)); | ||||
| 606 | ss += OPND(s)((s)&0x07ffffffLU); | ||||
| 607 | } while (OP(s = m->g->strip[ss])((s = m->g->strip[ss])&0xf8000000LU) != O_CH(18LU<<((unsigned)27))); | ||||
| 608 | /* note that the ss++ gets us past the O_CH */ | ||||
| 609 | break; | ||||
| 610 | default: /* have to make a choice */ | ||||
| 611 | hard = 1; | ||||
| 612 | break; | ||||
| 613 | } | ||||
| 614 | if (!hard) { /* that was it! */ | ||||
| 615 | if (sp != stop) | ||||
| 616 | return(NULL((void*)0)); | ||||
| 617 | return(sp); | ||||
| 618 | } | ||||
| 619 | ss--; /* adjust for the for's final increment */ | ||||
| 620 | |||||
| 621 | /* the hard stuff */ | ||||
| 622 | AT("hard", sp, stop, ss, stopst); | ||||
| 623 | s = m->g->strip[ss]; | ||||
| 624 | switch (OP(s)((s)&0xf8000000LU)) { | ||||
| 625 | case OBACK_(7LU<<((unsigned)27)): /* the vilest depths */ | ||||
| 626 | i = OPND(s)((s)&0x07ffffffLU); | ||||
| 627 | assert(0 < i && i <= m->g->nsub)((void) (0)); | ||||
| 628 | if (m->pmatch[i].rm_eo == -1) | ||||
| 629 | return(NULL((void*)0)); | ||||
| 630 | assert(m->pmatch[i].rm_so != -1)((void) (0)); | ||||
| 631 | len = m->pmatch[i].rm_eo - m->pmatch[i].rm_so; | ||||
| 632 | if (len == 0 && rec++ > MAX_RECURSION100) | ||||
| 633 | return(NULL((void*)0)); | ||||
| 634 | assert(stop - m->beginp >= len)((void) (0)); | ||||
| 635 | if (sp > stop - len) | ||||
| 636 | return(NULL((void*)0)); /* not enough left to match */ | ||||
| 637 | ssp = m->offp + m->pmatch[i].rm_so; | ||||
| 638 | if (memcmp(sp, ssp, len) != 0) | ||||
| 639 | return(NULL((void*)0)); | ||||
| 640 | while (m->g->strip[ss] != SOP(O_BACK, i)(((8LU<<((unsigned)27)))|(i))) | ||||
| 641 | ss++; | ||||
| 642 | return(backref(m, sp+len, stop, ss+1, stopst, lev, rec)); | ||||
| 643 | break; | ||||
| 644 | case OQUEST_(11LU<<((unsigned)27)): /* to null or not */ | ||||
| 645 | dp = backref(m, sp, stop, ss+1, stopst, lev, rec); | ||||
| 646 | if (dp != NULL((void*)0)) | ||||
| 647 | return(dp); /* not */ | ||||
| 648 | return(backref(m, sp, stop, ss+OPND(s)((s)&0x07ffffffLU)+1, stopst, lev, rec)); | ||||
| 649 | break; | ||||
| 650 | case OPLUS_(9LU<<((unsigned)27)): | ||||
| 651 | assert(m->lastpos != NULL)((void) (0)); | ||||
| 652 | assert(lev+1 <= m->g->nplus)((void) (0)); | ||||
| 653 | m->lastpos[lev+1] = sp; | ||||
| 654 | return(backref(m, sp, stop, ss+1, stopst, lev+1, rec)); | ||||
| 655 | break; | ||||
| 656 | case O_PLUS(10LU<<((unsigned)27)): | ||||
| 657 | if (sp == m->lastpos[lev]) /* last pass matched null */ | ||||
| 658 | return(backref(m, sp, stop, ss+1, stopst, lev-1, rec)); | ||||
| 659 | /* try another pass */ | ||||
| 660 | m->lastpos[lev] = sp; | ||||
| 661 | dp = backref(m, sp, stop, ss-OPND(s)((s)&0x07ffffffLU)+1, stopst, lev, rec); | ||||
| 662 | if (dp == NULL((void*)0)) | ||||
| 663 | return(backref(m, sp, stop, ss+1, stopst, lev-1, rec)); | ||||
| 664 | else | ||||
| 665 | return(dp); | ||||
| 666 | break; | ||||
| 667 | case OCH_(15LU<<((unsigned)27)): /* find the right one, if any */ | ||||
| 668 | ssub = ss + 1; | ||||
| 669 | esub = ss + OPND(s)((s)&0x07ffffffLU) - 1; | ||||
| 670 | assert(OP(m->g->strip[esub]) == OOR1)((void) (0)); | ||||
| 671 | for (;;) { /* find first matching branch */ | ||||
| 672 | dp = backref(m, sp, stop, ssub, stopst, lev, rec); | ||||
| 673 | if (dp != NULL((void*)0)) | ||||
| 674 | return(dp); | ||||
| 675 | /* that one missed, try next one */ | ||||
| 676 | if (OP(m->g->strip[esub])((m->g->strip[esub])&0xf8000000LU) == O_CH(18LU<<((unsigned)27))) | ||||
| 677 | return(NULL((void*)0)); /* there is none */ | ||||
| 678 | esub++; | ||||
| 679 | assert(OP(m->g->strip[esub]) == OOR2)((void) (0)); | ||||
| 680 | ssub = esub + 1; | ||||
| 681 | esub += OPND(m->g->strip[esub])((m->g->strip[esub])&0x07ffffffLU); | ||||
| 682 | if (OP(m->g->strip[esub])((m->g->strip[esub])&0xf8000000LU) == OOR2(17LU<<((unsigned)27))) | ||||
| 683 | esub--; | ||||
| 684 | else | ||||
| 685 | assert(OP(m->g->strip[esub]) == O_CH)((void) (0)); | ||||
| 686 | } | ||||
| 687 | break; | ||||
| 688 | case OLPAREN(13LU<<((unsigned)27)): /* must undo assignment if rest fails */ | ||||
| 689 | i = OPND(s)((s)&0x07ffffffLU); | ||||
| 690 | assert(0 < i && i <= m->g->nsub)((void) (0)); | ||||
| 691 | offsave = m->pmatch[i].rm_so; | ||||
| 692 | m->pmatch[i].rm_so = sp - m->offp; | ||||
| 693 | dp = backref(m, sp, stop, ss+1, stopst, lev, rec); | ||||
| 694 | if (dp != NULL((void*)0)) | ||||
| 695 | return(dp); | ||||
| 696 | m->pmatch[i].rm_so = offsave; | ||||
| 697 | return(NULL((void*)0)); | ||||
| 698 | break; | ||||
| 699 | case ORPAREN(14LU<<((unsigned)27)): /* must undo assignment if rest fails */ | ||||
| 700 | i = OPND(s)((s)&0x07ffffffLU); | ||||
| 701 | assert(0 < i && i <= m->g->nsub)((void) (0)); | ||||
| 702 | offsave = m->pmatch[i].rm_eo; | ||||
| 703 | m->pmatch[i].rm_eo = sp - m->offp; | ||||
| 704 | dp = backref(m, sp, stop, ss+1, stopst, lev, rec); | ||||
| 705 | if (dp != NULL((void*)0)) | ||||
| 706 | return(dp); | ||||
| 707 | m->pmatch[i].rm_eo = offsave; | ||||
| 708 | return(NULL((void*)0)); | ||||
| 709 | break; | ||||
| 710 | default: /* uh oh */ | ||||
| 711 | assert(nope)((void) (0)); | ||||
| 712 | break; | ||||
| 713 | } | ||||
| 714 | |||||
| 715 | /* "can't happen" */ | ||||
| 716 | assert(nope)((void) (0)); | ||||
| 717 | /* NOTREACHED */ | ||||
| 718 | return NULL((void*)0); | ||||
| 719 | } | ||||
| 720 | |||||
| 721 | /* | ||||
| 722 | - fast - step through the string at top speed | ||||
| 723 | */ | ||||
| 724 | static const char * /* where tentative match ended, or NULL */ | ||||
| 725 | fast(struct match *m, const char *start, const char *stop, sopno startst, | ||||
| 726 | sopno stopst) | ||||
| 727 | { | ||||
| 728 | stateschar * st = m->st; | ||||
| 729 | stateschar * fresh = m->fresh; | ||||
| 730 | stateschar * tmp = m->tmp; | ||||
| 731 | const char *p = start; | ||||
| 732 | int c = (start
| ||||
| 733 | int lastc; /* previous c */ | ||||
| 734 | int flagch; | ||||
| 735 | int i; | ||||
| 736 | const char *coldp; /* last p after which no match was underway */ | ||||
| 737 | |||||
| 738 | CLEAR(st)memset(st, 0, m->g->nstates); | ||||
| 739 | SET1(st, startst)((st)[startst] = 1); | ||||
| 740 | st = step(m->g, startst, stopst, st, NOTHING(((127 +1)+1)+3), st); | ||||
| 741 | ASSIGN(fresh, st)memmove(fresh, st, m->g->nstates); | ||||
| 742 | SP("start", st, *p); | ||||
| 743 | coldp = NULL((void*)0); | ||||
| 744 | for (;;) { | ||||
| 745 | /* next character */ | ||||
| 746 | lastc = c; | ||||
| 747 | c = (p == m->endp) ? OUT(127 +1) : *p; | ||||
| 748 | if (EQ(st, fresh)(memcmp(st, fresh, m->g->nstates) == 0)) | ||||
| 749 | coldp = p; | ||||
| 750 | |||||
| 751 | /* is there an EOL and/or BOL between lastc and c? */ | ||||
| 752 | flagch = '\0'; | ||||
| 753 | i = 0; | ||||
| 754 | if ( (lastc == '\n' && m->g->cflags®_NEWLINE0010) || | ||||
| 755 | (lastc == OUT(127 +1) && !(m->eflags®_NOTBOL00001)) ) { | ||||
| 756 | flagch = BOL((127 +1)+1); | ||||
| 757 | i = m->g->nbol; | ||||
| 758 | } | ||||
| 759 | if ( (c == '\n' && m->g->cflags®_NEWLINE0010) || | ||||
| 760 | (c == OUT(127 +1) && !(m->eflags®_NOTEOL00002)) ) { | ||||
| 761 | flagch = (flagch == BOL((127 +1)+1)) ? BOLEOL(((127 +1)+1)+2) : EOL(((127 +1)+1)+1); | ||||
| 762 | i += m->g->neol; | ||||
| 763 | } | ||||
| 764 | if (i
| ||||
| 765 | for (; i > 0; i--) | ||||
| 766 | st = step(m->g, startst, stopst, st, flagch, st); | ||||
| 767 | SP("boleol", st, c); | ||||
| 768 | } | ||||
| 769 | |||||
| 770 | /* how about a word boundary? */ | ||||
| 771 | if ( (flagch == BOL((127 +1)+1) || (lastc != OUT(127 +1) && !ISWORD(lastc)(((*__ctype_b_loc ())[(int) ((lastc&0xff))] & (unsigned short int) _ISalnum) || (lastc) == '_'))) && | ||||
| 772 | (c != OUT(127 +1) && ISWORD(c)(((*__ctype_b_loc ())[(int) ((c&0xff))] & (unsigned short int) _ISalnum) || (c) == '_')) ) { | ||||
| 773 | flagch = BOW(((127 +1)+1)+4); | ||||
| 774 | } | ||||
| 775 | if ( (lastc != OUT(127 +1) && ISWORD(lastc)(((*__ctype_b_loc ())[(int) ((lastc&0xff))] & (unsigned short int) _ISalnum) || (lastc) == '_')) && | ||||
| 776 | (flagch == EOL(((127 +1)+1)+1) || (c != OUT(127 +1) && !ISWORD(c)(((*__ctype_b_loc ())[(int) ((c&0xff))] & (unsigned short int) _ISalnum) || (c) == '_'))) ) { | ||||
| 777 | flagch = EOW(((127 +1)+1)+5); | ||||
| 778 | } | ||||
| 779 | if (flagch == BOW(((127 +1)+1)+4) || flagch == EOW(((127 +1)+1)+5)) { | ||||
| 780 | st = step(m->g, startst, stopst, st, flagch, st); | ||||
| 781 | SP("boweow", st, c); | ||||
| 782 | } | ||||
| 783 | |||||
| 784 | /* are we done? */ | ||||
| 785 | if (ISSET(st, stopst)((st)[stopst]) || p == stop) | ||||
| 786 | break; /* NOTE BREAK OUT */ | ||||
| 787 | |||||
| 788 | /* no, we must deal with this character */ | ||||
| 789 | ASSIGN(tmp, st)memmove(tmp, st, m->g->nstates); | ||||
| 790 | ASSIGN(st, fresh)memmove(st, fresh, m->g->nstates); | ||||
| 791 | assert(c != OUT)((void) (0)); | ||||
| 792 | st = step(m->g, startst, stopst, tmp, c, st); | ||||
| 793 | SP("aft", st, c); | ||||
| 794 | assert(EQ(step(m->g, startst, stopst, st, NOTHING, st), st))((void) (0)); | ||||
| 795 | p++; | ||||
| 796 | } | ||||
| 797 | |||||
| 798 | assert(coldp != NULL)((void) (0)); | ||||
| 799 | m->coldp = coldp; | ||||
| 800 | if (ISSET(st, stopst)((st)[stopst])) | ||||
| 801 | return(p+1); | ||||
| 802 | else | ||||
| 803 | return(NULL((void*)0)); | ||||
| 804 | } | ||||
| 805 | |||||
| 806 | /* | ||||
| 807 | - slow - step through the string more deliberately | ||||
| 808 | */ | ||||
| 809 | static const char * /* where it ended */ | ||||
| 810 | slow(struct match *m, const char *start, const char *stop, sopno startst, | ||||
| 811 | sopno stopst) | ||||
| 812 | { | ||||
| 813 | /* Quickly skip over fixed character matches at the start. */ | ||||
| 814 | const char *p = start; | ||||
| 815 | for (; startst < stopst; ++startst) { | ||||
| 816 | int hard = 0; | ||||
| 817 | sop s = m->g->strip[startst]; | ||||
| 818 | switch (OP(s)((s)&0xf8000000LU)) { | ||||
| 819 | case OLPAREN(13LU<<((unsigned)27)): | ||||
| 820 | case ORPAREN(14LU<<((unsigned)27)): | ||||
| 821 | /* Not relevant here. */ | ||||
| 822 | break; | ||||
| 823 | case OCHAR(2LU<<((unsigned)27)): | ||||
| 824 | if (p == stop || *p != (char)OPND(s)((s)&0x07ffffffLU)) | ||||
| 825 | return NULL((void*)0); | ||||
| 826 | ++p; | ||||
| 827 | break; | ||||
| 828 | default: | ||||
| 829 | hard = 1; | ||||
| 830 | break; | ||||
| 831 | } | ||||
| 832 | if (hard) | ||||
| 833 | break; | ||||
| 834 | } | ||||
| 835 | |||||
| 836 | stateschar * st = m->st; | ||||
| 837 | stateschar * empty = m->empty; | ||||
| 838 | stateschar * tmp = m->tmp; | ||||
| 839 | int c = (p
| ||||
| |||||
| 840 | int lastc; /* previous c */ | ||||
| 841 | int flagch; | ||||
| 842 | int i; | ||||
| 843 | const char *matchp; /* last p at which a match ended */ | ||||
| 844 | |||||
| 845 | AT("slow", start, stop, startst, stopst); | ||||
| 846 | CLEAR(st)memset(st, 0, m->g->nstates); | ||||
| 847 | SET1(st, startst)((st)[startst] = 1); | ||||
| 848 | SP("sstart", st, *p); | ||||
| 849 | st = step(m->g, startst, stopst, st, NOTHING(((127 +1)+1)+3), st); | ||||
| 850 | matchp = NULL((void*)0); | ||||
| 851 | for (;;) { | ||||
| 852 | /* next character */ | ||||
| 853 | lastc = c; | ||||
| 854 | c = (p == m->endp) ? OUT(127 +1) : *p; | ||||
| 855 | |||||
| 856 | /* is there an EOL and/or BOL between lastc and c? */ | ||||
| 857 | flagch = '\0'; | ||||
| 858 | i = 0; | ||||
| 859 | if ( (lastc == '\n' && m->g->cflags®_NEWLINE0010) || | ||||
| 860 | (lastc == OUT(127 +1) && !(m->eflags®_NOTBOL00001)) ) { | ||||
| 861 | flagch = BOL((127 +1)+1); | ||||
| 862 | i = m->g->nbol; | ||||
| 863 | } | ||||
| 864 | if ( (c == '\n' && m->g->cflags®_NEWLINE0010) || | ||||
| 865 | (c == OUT(127 +1) && !(m->eflags®_NOTEOL00002)) ) { | ||||
| 866 | flagch = (flagch == BOL((127 +1)+1)) ? BOLEOL(((127 +1)+1)+2) : EOL(((127 +1)+1)+1); | ||||
| 867 | i += m->g->neol; | ||||
| 868 | } | ||||
| 869 | if (i != 0) { | ||||
| 870 | for (; i > 0; i--) | ||||
| 871 | st = step(m->g, startst, stopst, st, flagch, st); | ||||
| 872 | SP("sboleol", st, c); | ||||
| 873 | } | ||||
| 874 | |||||
| 875 | /* how about a word boundary? */ | ||||
| 876 | if ( (flagch == BOL((127 +1)+1) || (lastc != OUT(127 +1) && !ISWORD(lastc)(((*__ctype_b_loc ())[(int) ((lastc&0xff))] & (unsigned short int) _ISalnum) || (lastc) == '_'))) && | ||||
| 877 | (c != OUT(127 +1) && ISWORD(c)(((*__ctype_b_loc ())[(int) ((c&0xff))] & (unsigned short int) _ISalnum) || (c) == '_')) ) { | ||||
| 878 | flagch = BOW(((127 +1)+1)+4); | ||||
| 879 | } | ||||
| 880 | if ( (lastc != OUT(127 +1) && ISWORD(lastc)(((*__ctype_b_loc ())[(int) ((lastc&0xff))] & (unsigned short int) _ISalnum) || (lastc) == '_')) && | ||||
| 881 | (flagch == EOL(((127 +1)+1)+1) || (c != OUT(127 +1) && !ISWORD(c)(((*__ctype_b_loc ())[(int) ((c&0xff))] & (unsigned short int) _ISalnum) || (c) == '_'))) ) { | ||||
| 882 | flagch = EOW(((127 +1)+1)+5); | ||||
| 883 | } | ||||
| 884 | if (flagch == BOW(((127 +1)+1)+4) || flagch == EOW(((127 +1)+1)+5)) { | ||||
| 885 | st = step(m->g, startst, stopst, st, flagch, st); | ||||
| 886 | SP("sboweow", st, c); | ||||
| 887 | } | ||||
| 888 | |||||
| 889 | /* are we done? */ | ||||
| 890 | if (ISSET(st, stopst)((st)[stopst])) | ||||
| 891 | matchp = p; | ||||
| 892 | if (EQ(st, empty)(memcmp(st, empty, m->g->nstates) == 0) || p == stop) | ||||
| 893 | break; /* NOTE BREAK OUT */ | ||||
| 894 | |||||
| 895 | /* no, we must deal with this character */ | ||||
| 896 | ASSIGN(tmp, st)memmove(tmp, st, m->g->nstates); | ||||
| 897 | ASSIGN(st, empty)memmove(st, empty, m->g->nstates); | ||||
| 898 | assert(c != OUT)((void) (0)); | ||||
| 899 | st = step(m->g, startst, stopst, tmp, c, st); | ||||
| 900 | SP("saft", st, c); | ||||
| 901 | assert(EQ(step(m->g, startst, stopst, st, NOTHING, st), st))((void) (0)); | ||||
| 902 | p++; | ||||
| 903 | } | ||||
| 904 | |||||
| 905 | return(matchp); | ||||
| 906 | } | ||||
| 907 | |||||
| 908 | |||||
| 909 | /* | ||||
| 910 | - step - map set of states reachable before char to set reachable after | ||||
| 911 | */ | ||||
| 912 | static stateschar * | ||||
| 913 | step(struct re_guts *g, | ||||
| 914 | sopno start, /* start state within strip */ | ||||
| 915 | sopno stop, /* state after stop state within strip */ | ||||
| 916 | stateschar * bef, /* states reachable before */ | ||||
| 917 | int ch, /* character or NONCHAR code */ | ||||
| 918 | stateschar * aft) /* states already known reachable after */ | ||||
| 919 | { | ||||
| 920 | cset *cs; | ||||
| 921 | sop s; | ||||
| 922 | sopno pc; | ||||
| 923 | onestatelong here; /* note, macros know this name */ | ||||
| 924 | sopno look; | ||||
| 925 | int i; | ||||
| 926 | |||||
| 927 | for (pc = start, INIT(here, pc)((here) = (pc)); pc != stop; pc++, INC(here)((here)++)) { | ||||
| 928 | s = g->strip[pc]; | ||||
| 929 | switch (OP(s)((s)&0xf8000000LU)) { | ||||
| 930 | case OEND(1LU<<((unsigned)27)): | ||||
| 931 | assert(pc == stop-1)((void) (0)); | ||||
| 932 | break; | ||||
| 933 | case OCHAR(2LU<<((unsigned)27)): | ||||
| 934 | /* only characters can match */ | ||||
| 935 | assert(!NONCHAR(ch) || ch != (char)OPND(s))((void) (0)); | ||||
| 936 | if (ch == (char)OPND(s)((s)&0x07ffffffLU)) | ||||
| 937 | FWD(aft, bef, 1)((aft)[here+(1)] |= (bef)[here]); | ||||
| 938 | break; | ||||
| 939 | case OBOL(3LU<<((unsigned)27)): | ||||
| 940 | if (ch == BOL((127 +1)+1) || ch == BOLEOL(((127 +1)+1)+2)) | ||||
| 941 | FWD(aft, bef, 1)((aft)[here+(1)] |= (bef)[here]); | ||||
| 942 | break; | ||||
| 943 | case OEOL(4LU<<((unsigned)27)): | ||||
| 944 | if (ch == EOL(((127 +1)+1)+1) || ch == BOLEOL(((127 +1)+1)+2)) | ||||
| 945 | FWD(aft, bef, 1)((aft)[here+(1)] |= (bef)[here]); | ||||
| 946 | break; | ||||
| 947 | case OBOW(19LU<<((unsigned)27)): | ||||
| 948 | if (ch == BOW(((127 +1)+1)+4)) | ||||
| 949 | FWD(aft, bef, 1)((aft)[here+(1)] |= (bef)[here]); | ||||
| 950 | break; | ||||
| 951 | case OEOW(20LU<<((unsigned)27)): | ||||
| 952 | if (ch == EOW(((127 +1)+1)+5)) | ||||
| 953 | FWD(aft, bef, 1)((aft)[here+(1)] |= (bef)[here]); | ||||
| 954 | break; | ||||
| 955 | case OANY(5LU<<((unsigned)27)): | ||||
| 956 | if (!NONCHAR(ch)((ch) > 127)) | ||||
| 957 | FWD(aft, bef, 1)((aft)[here+(1)] |= (bef)[here]); | ||||
| 958 | break; | ||||
| 959 | case OANYOF(6LU<<((unsigned)27)): | ||||
| 960 | cs = &g->sets[OPND(s)((s)&0x07ffffffLU)]; | ||||
| 961 | if (!NONCHAR(ch)((ch) > 127) && CHIN(cs, ch)((cs)->ptr[(uch)(ch)] & (cs)->mask)) | ||||
| 962 | FWD(aft, bef, 1)((aft)[here+(1)] |= (bef)[here]); | ||||
| 963 | break; | ||||
| 964 | case OBACK_(7LU<<((unsigned)27)): /* ignored here */ | ||||
| 965 | case O_BACK(8LU<<((unsigned)27)): | ||||
| 966 | FWD(aft, aft, 1)((aft)[here+(1)] |= (aft)[here]); | ||||
| 967 | break; | ||||
| 968 | case OPLUS_(9LU<<((unsigned)27)): /* forward, this is just an empty */ | ||||
| 969 | FWD(aft, aft, 1)((aft)[here+(1)] |= (aft)[here]); | ||||
| 970 | break; | ||||
| 971 | case O_PLUS(10LU<<((unsigned)27)): /* both forward and back */ | ||||
| 972 | FWD(aft, aft, 1)((aft)[here+(1)] |= (aft)[here]); | ||||
| 973 | i = ISSETBACK(aft, OPND(s))((aft)[here - (((s)&0x07ffffffLU))]); | ||||
| 974 | BACK(aft, aft, OPND(s))((aft)[here-(((s)&0x07ffffffLU))] |= (aft)[here]); | ||||
| 975 | if (!i && ISSETBACK(aft, OPND(s))((aft)[here - (((s)&0x07ffffffLU))])) { | ||||
| 976 | /* oho, must reconsider loop body */ | ||||
| 977 | pc -= OPND(s)((s)&0x07ffffffLU) + 1; | ||||
| 978 | INIT(here, pc)((here) = (pc)); | ||||
| 979 | } | ||||
| 980 | break; | ||||
| 981 | case OQUEST_(11LU<<((unsigned)27)): /* two branches, both forward */ | ||||
| 982 | FWD(aft, aft, 1)((aft)[here+(1)] |= (aft)[here]); | ||||
| 983 | FWD(aft, aft, OPND(s))((aft)[here+(((s)&0x07ffffffLU))] |= (aft)[here]); | ||||
| 984 | break; | ||||
| 985 | case O_QUEST(12LU<<((unsigned)27)): /* just an empty */ | ||||
| 986 | FWD(aft, aft, 1)((aft)[here+(1)] |= (aft)[here]); | ||||
| 987 | break; | ||||
| 988 | case OLPAREN(13LU<<((unsigned)27)): /* not significant here */ | ||||
| 989 | case ORPAREN(14LU<<((unsigned)27)): | ||||
| 990 | FWD(aft, aft, 1)((aft)[here+(1)] |= (aft)[here]); | ||||
| 991 | break; | ||||
| 992 | case OCH_(15LU<<((unsigned)27)): /* mark the first two branches */ | ||||
| 993 | FWD(aft, aft, 1)((aft)[here+(1)] |= (aft)[here]); | ||||
| 994 | assert(OP(g->strip[pc+OPND(s)]) == OOR2)((void) (0)); | ||||
| 995 | FWD(aft, aft, OPND(s))((aft)[here+(((s)&0x07ffffffLU))] |= (aft)[here]); | ||||
| 996 | break; | ||||
| 997 | case OOR1(16LU<<((unsigned)27)): /* done a branch, find the O_CH */ | ||||
| 998 | if (ISSTATEIN(aft, here)((aft)[here])) { | ||||
| 999 | for (look = 1; | ||||
| 1000 | OP(s = g->strip[pc+look])((s = g->strip[pc+look])&0xf8000000LU) != O_CH(18LU<<((unsigned)27)); | ||||
| 1001 | look += OPND(s)((s)&0x07ffffffLU)) | ||||
| 1002 | assert(OP(s) == OOR2)((void) (0)); | ||||
| 1003 | FWD(aft, aft, look)((aft)[here+(look)] |= (aft)[here]); | ||||
| 1004 | } | ||||
| 1005 | break; | ||||
| 1006 | case OOR2(17LU<<((unsigned)27)): /* propagate OCH_'s marking */ | ||||
| 1007 | FWD(aft, aft, 1)((aft)[here+(1)] |= (aft)[here]); | ||||
| 1008 | if (OP(g->strip[pc+OPND(s)])((g->strip[pc+((s)&0x07ffffffLU)])&0xf8000000LU) != O_CH(18LU<<((unsigned)27))) { | ||||
| 1009 | assert(OP(g->strip[pc+OPND(s)]) == OOR2)((void) (0)); | ||||
| 1010 | FWD(aft, aft, OPND(s))((aft)[here+(((s)&0x07ffffffLU))] |= (aft)[here]); | ||||
| 1011 | } | ||||
| 1012 | break; | ||||
| 1013 | case O_CH(18LU<<((unsigned)27)): /* just empty */ | ||||
| 1014 | FWD(aft, aft, 1)((aft)[here+(1)] |= (aft)[here]); | ||||
| 1015 | break; | ||||
| 1016 | default: /* ooooops... */ | ||||
| 1017 | assert(nope)((void) (0)); | ||||
| 1018 | break; | ||||
| 1019 | } | ||||
| 1020 | } | ||||
| 1021 | |||||
| 1022 | return(aft); | ||||
| 1023 | } | ||||
| 1024 | |||||
| 1025 | #ifdef REDEBUG | ||||
| 1026 | /* | ||||
| 1027 | - print - print a set of states | ||||
| 1028 | */ | ||||
| 1029 | static void | ||||
| 1030 | print(struct match *m, const char *caption, stateschar * st, int ch, FILE *d) | ||||
| 1031 | { | ||||
| 1032 | struct re_guts *g = m->g; | ||||
| 1033 | int i; | ||||
| 1034 | int first = 1; | ||||
| 1035 | |||||
| 1036 | if (!(m->eflags®_TRACE00400)) | ||||
| 1037 | return; | ||||
| 1038 | |||||
| 1039 | (void)fprintf(d, "%s", caption)__fprintf_chk (d, 2 - 1, "%s", caption); | ||||
| 1040 | if (ch != '\0') | ||||
| 1041 | (void)fprintf(d, " %s", pchar(ch))__fprintf_chk (d, 2 - 1, " %s", pchar(ch)); | ||||
| 1042 | for (i = 0; i < g->nstates; i++) | ||||
| 1043 | if (ISSET(st, i)((st)[i])) { | ||||
| 1044 | (void)fprintf(d, "%s%d", (first) ? "\t" : ", ", i)__fprintf_chk (d, 2 - 1, "%s%d", (first) ? "\t" : ", ", i); | ||||
| 1045 | first = 0; | ||||
| 1046 | } | ||||
| 1047 | (void)fprintf(d, "\n")__fprintf_chk (d, 2 - 1, "\n"); | ||||
| 1048 | } | ||||
| 1049 | |||||
| 1050 | /* | ||||
| 1051 | - at - print current situation | ||||
| 1052 | */ | ||||
| 1053 | static void | ||||
| 1054 | at(struct match *m, const char *title, const char *start, const char *stop, | ||||
| 1055 | sopno startst, sopno stopst) | ||||
| 1056 | { | ||||
| 1057 | if (!(m->eflags®_TRACE00400)) | ||||
| 1058 | return; | ||||
| 1059 | |||||
| 1060 | (void)printf("%s %s-", title, pchar(*start))__printf_chk (2 - 1, "%s %s-", title, pchar(*start)); | ||||
| 1061 | (void)printf("%s ", pchar(*stop))__printf_chk (2 - 1, "%s ", pchar(*stop)); | ||||
| 1062 | (void)printf("%ld-%ld\n", (long)startst, (long)stopst)__printf_chk (2 - 1, "%ld-%ld\n", (long)startst, (long)stopst ); | ||||
| 1063 | } | ||||
| 1064 | |||||
| 1065 | #ifndef PCHARDONE | ||||
| 1066 | #define PCHARDONE /* never again */ | ||||
| 1067 | /* | ||||
| 1068 | - pchar - make a character printable | ||||
| 1069 | * | ||||
| 1070 | * Is this identical to regchar() over in debug.c? Well, yes. But a | ||||
| 1071 | * duplicate here avoids having a debugging-capable regexec.o tied to | ||||
| 1072 | * a matching debug.o, and this is convenient. It all disappears in | ||||
| 1073 | * the non-debug compilation anyway, so it doesn't matter much. | ||||
| 1074 | */ | ||||
| 1075 | static char * /* -> representation */ | ||||
| 1076 | pchar(int ch) | ||||
| 1077 | { | ||||
| 1078 | static char pbuf[10]; | ||||
| 1079 | |||||
| 1080 | if (isprint(ch)((*__ctype_b_loc ())[(int) ((ch))] & (unsigned short int) _ISprint) || ch == ' ') | ||||
| 1081 | (void)snprintf(pbuf, sizeof pbuf, "%c", ch)__builtin___snprintf_chk (pbuf, sizeof pbuf, 2 - 1, __builtin_object_size (pbuf, 2 > 1), "%c", ch); | ||||
| 1082 | else | ||||
| 1083 | (void)snprintf(pbuf, sizeof pbuf, "\\%o", ch)__builtin___snprintf_chk (pbuf, sizeof pbuf, 2 - 1, __builtin_object_size (pbuf, 2 > 1), "\\%o", ch); | ||||
| 1084 | return(pbuf); | ||||
| 1085 | } | ||||
| 1086 | #endif | ||||
| 1087 | #endif | ||||
| 1088 | |||||
| 1089 | #undef matcher | ||||
| 1090 | #undef fast | ||||
| 1091 | #undef slow | ||||
| 1092 | #undef dissect | ||||
| 1093 | #undef backref | ||||
| 1094 | #undef step | ||||
| 1095 | #undef print | ||||
| 1096 | #undef at | ||||
| 1097 | #undef match | ||||
| 1098 | #undef nope | ||||
| 1099 | #undef step_back |