X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regcomp.c;h=f9e253926b4482a2611b9357de3e3f1dfdb4a321;hb=baed7233d4dfe516b6be04fb05fade5080a282e0;hp=6815e13b2e8acf185a026ed7a24bcea62cb5b42e;hpb=35ff78560a01016ce2a3dffe29f18ce851bc0b90;p=p5sagit%2Fp5-mst-13.2.git diff --git a/regcomp.c b/regcomp.c index 6815e13..f9e2539 100644 --- a/regcomp.c +++ b/regcomp.c @@ -19,6 +19,16 @@ * with the POSIX routines of the same names. */ +#ifdef IN_XSUB_RE +/* We *really* need to overwrite these symbols: */ +# define Perl_pregcomp my_regcomp +# define Perl_regdump my_regdump +# define Perl_regprop my_regprop +/* *These* symbols are masked to allow static link. */ +# define Perl_pregfree my_regfree +# define Perl_regnext my_regnext +#endif + /*SUPPRESS 112*/ /* * pregcomp and pregexec -- regsub and regerror are not used in perl @@ -55,7 +65,10 @@ */ #include "EXTERN.h" #include "perl.h" -#include "INTERN.h" + +#ifndef IN_XSUB_RE +# include "INTERN.h" +#endif #define REG_COMP_C #include "regcomp.h" @@ -93,7 +106,7 @@ * Flags to be passed up and down. */ #define WORST 0 /* Worst case. */ -#define HASWIDTH 0x1 /* Known never to match null string. */ +#define HASWIDTH 0x1 /* Known to match non-null strings. */ #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */ #define SPSTART 0x4 /* Starts with * or +. */ #define TRYAGAIN 0x8 /* Weeded out a declaration. */ @@ -267,6 +280,10 @@ study_chunk(regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 /* Allow dumping */ n = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2; while (n <= stop) { + /* Purify reports a benign UMR here sometimes, because we + * don't initialize the OP() slot of a node when that node + * is occupied by just the trailing null of the string in + * an EXACT node */ if (regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) { OP(n) = OPTIMIZED; NEXT_OFF(n) = 0; @@ -421,7 +438,8 @@ study_chunk(regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 if (!scan) /* It was not CURLYX, but CURLY. */ scan = next; if (dowarn && (minnext + deltanext == 0) - && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))) + && !(data->flags & (SF_HAS_PAR|SF_IN_PAR)) + && maxcount <= 10000) /* Complement check for big count */ warn("Strange *+?{} on zero-length expression"); min += minnext * mincount; is_inf |= (maxcount == REG_INFTY && (minnext + deltanext) > 0 @@ -429,7 +447,6 @@ study_chunk(regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 delta += (minnext + deltanext) * maxcount - minnext * mincount; /* Try powerful optimization CURLYX => CURLYN. */ -#ifdef REGALIGN_STRUCT if ( OP(oscan) == CURLYX && data && data->flags & SF_IN_PAR && !(data->flags & SF_HAS_EVAL) @@ -461,16 +478,11 @@ study_chunk(regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */ #endif } -#endif nogo: /* Try optimization CURLYX => CURLYM. */ if ( OP(oscan) == CURLYX && data -#ifdef REGALIGN_STRUCT && !(data->flags & SF_HAS_PAR) -#else - && !(data->flags & (SF_HAS_PAR|SF_IN_PAR)) -#endif && !(data->flags & SF_HAS_EVAL) && !deltanext ) { /* XXXX How to optimize if data == 0? */ @@ -483,7 +495,6 @@ study_chunk(regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 && (OP(nxt2) != WHILEM)) nxt = nxt2; OP(nxt2) = SUCCEED; /* Whas WHILEM */ -#ifdef REGALIGN_STRUCT /* Need to optimize away parenths. */ if (data->flags & SF_IN_PAR) { /* Set the parenth number. */ @@ -519,7 +530,6 @@ study_chunk(regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 study_chunk(&nxt1, &deltanext, nxt, NULL, 0); } else oscan->flags = 0; -#endif } if (data && fl & (SF_HAS_PAR|SF_IN_PAR)) pars++; @@ -573,13 +583,11 @@ study_chunk(regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 if (data && (fl & SF_HAS_EVAL)) data->flags |= SF_HAS_EVAL; optimize_curly_tail: -#ifdef REGALIGN if (OP(oscan) != CURLYX) { while (regkind[(U8)OP(next = regnext(oscan))] == NOTHING && NEXT_OFF(next)) NEXT_OFF(oscan) += NEXT_OFF(next); } -#endif continue; default: /* REF only? */ if (flags & SCF_DO_SUBSTR) { @@ -624,11 +632,9 @@ study_chunk(regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 } else if (OP(scan) == OPEN) { pars++; } else if (OP(scan) == CLOSE && ARG(scan) == is_par) { -#ifdef REGALIGN_STRUCT next = regnext(scan); if ( next && (OP(next) != WHILEM) && next < last) -#endif is_par = 0; /* Disable optimization */ } else if (OP(scan) == EVAL) { if (data) @@ -658,20 +664,20 @@ study_chunk(regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 STATIC I32 add_data(I32 n, char *s) { - if (rx->data) { - Renewc(rx->data, - sizeof(*rx->data) + sizeof(void*) * (rx->data->count + n - 1), + if (regcomp_rx->data) { + Renewc(regcomp_rx->data, + sizeof(*regcomp_rx->data) + sizeof(void*) * (regcomp_rx->data->count + n - 1), char, struct reg_data); - Renew(rx->data->what, rx->data->count + n, U8); - rx->data->count += n; + Renew(regcomp_rx->data->what, regcomp_rx->data->count + n, U8); + regcomp_rx->data->count += n; } else { - Newc(1207, rx->data, sizeof(*rx->data) + sizeof(void*) * (n - 1), + Newc(1207, regcomp_rx->data, sizeof(*regcomp_rx->data) + sizeof(void*) * (n - 1), char, struct reg_data); - New(1208, rx->data->what, n, U8); - rx->data->count = n; + New(1208, regcomp_rx->data->what, n, U8); + regcomp_rx->data->count = n; } - Copy(s, rx->data->what + rx->data->count - n, n, U8); - return rx->data->count - n; + Copy(s, regcomp_rx->data->what + regcomp_rx->data->count - n, n, U8); + return regcomp_rx->data->count - n; } /* @@ -717,7 +723,7 @@ pregcomp(char *exp, char *xend, PMOP *pm) extralen = 0; /* First pass: determine size, legality. */ - regparse = exp; + regcomp_parse = exp; regxend = xend; regnaughty = 0; regnpar = 1; @@ -758,15 +764,10 @@ pregcomp(char *exp, char *xend, PMOP *pm) /* Small enough for pointer-storage convention? If extralen==0, this means that we will not need long jumps. */ -#ifndef REGALIGN_STRUCT - if (regsize >= 0x10000L && extralen) - FAIL("regexp too big"); -#else if (regsize >= 0x10000L && extralen) regsize += extralen; else extralen = 0; -#endif /* Allocate space and initialize. */ Newc(1001, r, sizeof(regexp) + (unsigned)regsize * sizeof(regnode), @@ -777,10 +778,11 @@ pregcomp(char *exp, char *xend, PMOP *pm) r->prelen = xend - exp; r->precomp = regprecomp; r->subbeg = r->subbase = NULL; - rx = r; + r->nparens = regnpar - 1; /* set early to validate backrefs */ + regcomp_rx = r; /* Second pass: emit code. */ - regparse = exp; + regcomp_parse = exp; regxend = xend; regnaughty = 0; regnpar = 1; @@ -948,12 +950,13 @@ pregcomp(char *exp, char *xend, PMOP *pm) r->check_substr = r->anchored_substr = r->float_substr = Nullsv; } - r->nparens = regnpar - 1; r->minlen = minlen; if (regseen & REG_SEEN_GPOS) r->reganch |= ROPT_GPOS_SEEN; if (regseen & REG_SEEN_LOOKBEHIND) r->reganch |= ROPT_LOOKBEHIND_SEEN; + if (regseen & REG_SEEN_EVAL) + r->reganch |= ROPT_EVAL_SEEN; Newz(1002, r->startp, regnpar, char*); Newz(1002, r->endp, regnpar, char*); DEBUG_r(regdump(r)); @@ -981,25 +984,25 @@ reg(I32 paren, I32 *flagp) I32 flags, oregflags = regflags, have_branch = 0, open = 0; char c; - *flagp = HASWIDTH; /* Tentatively. */ + *flagp = 0; /* Tentatively. */ /* Make an OPEN node, if parenthesized. */ if (paren) { - if (*regparse == '?') { - regparse++; - paren = *regparse++; + if (*regcomp_parse == '?') { + U16 posflags = 0, negflags = 0; + U16 *flagsp = &posflags; + + regcomp_parse++; + paren = *regcomp_parse++; ret = NULL; /* For look-ahead/behind. */ switch (paren) { case '<': -#ifndef REGALIGN_STRUCT - FAIL("lookbehind non-implemented without REGALIGN_STRUCT"); -#endif regseen |= REG_SEEN_LOOKBEHIND; - if (*regparse == '!') + if (*regcomp_parse == '!') paren = ','; - if (*regparse != '=' && *regparse != '!') + if (*regcomp_parse != '=' && *regcomp_parse != '!') goto unknown; - regparse++; + regcomp_parse++; case '=': case '!': seen_zerolen++; @@ -1011,9 +1014,9 @@ reg(I32 paren, I32 *flagp) FAIL2("Sequence (?%c...) not implemented", (int)paren); break; case '#': - while (*regparse && *regparse != ')') - regparse++; - if (*regparse != ')') + while (*regcomp_parse && *regcomp_parse != ')') + regcomp_parse++; + if (*regcomp_parse != ')') FAIL("Sequence (?#... not terminated"); nextchar(); *flagp = TRYAGAIN; @@ -1023,38 +1026,46 @@ reg(I32 paren, I32 *flagp) dTHR; I32 count = 1, n = 0; char c; - char *s = regparse; + char *s = regcomp_parse; SV *sv; OP_4tree *sop, *rop; seen_zerolen++; - while (count && (c = *regparse)) { - if (c == '\\' && regparse[1]) - regparse++; + regseen |= REG_SEEN_EVAL; + while (count && (c = *regcomp_parse)) { + if (c == '\\' && regcomp_parse[1]) + regcomp_parse++; else if (c == '{') count++; else if (c == '}') count--; - regparse++; + regcomp_parse++; } - if (*regparse != ')') + if (*regcomp_parse != ')') FAIL("Sequence (?{...}) not terminated or not {}-balanced"); if (!SIZE_ONLY) { AV *av; - if (regparse - 1 - s) - sv = newSVpv(s, regparse - 1 - s); + if (regcomp_parse - 1 - s) + sv = newSVpv(s, regcomp_parse - 1 - s); else sv = newSVpv("", 0); rop = sv_compile_2op(sv, &sop, "re", &av); n = add_data(3, "nso"); - rx->data->data[n] = (void*)rop; - rx->data->data[n+1] = (void*)av; - rx->data->data[n+2] = (void*)sop; + regcomp_rx->data->data[n] = (void*)rop; + regcomp_rx->data->data[n+1] = (void*)av; + regcomp_rx->data->data[n+2] = (void*)sop; SvREFCNT_dec(sv); } else { /* First pass */ + if (curcop == &compiling) { + if (!(hints & HINT_RE_EVAL)) + FAIL("Eval-group not allowed, use re 'eval'"); + } + else { + FAIL("Eval-group not allowed at run time"); + } if (tainted) FAIL("Eval-group in insecure regular expression"); } @@ -1064,21 +1075,21 @@ reg(I32 paren, I32 *flagp) } case '(': { - if (regparse[0] == '?') { - if (regparse[1] == '=' || regparse[1] == '!' - || regparse[1] == '<' - || regparse[1] == '{') { /* Lookahead or eval. */ + if (regcomp_parse[0] == '?') { + if (regcomp_parse[1] == '=' || regcomp_parse[1] == '!' + || regcomp_parse[1] == '<' + || regcomp_parse[1] == '{') { /* Lookahead or eval. */ I32 flag; ret = reg_node(LOGICAL); regtail(ret, reg(1, &flag)); goto insert_if; } - } else if (regparse[0] >= '1' && regparse[0] <= '9' ) { - parno = atoi(regparse++); + } else if (regcomp_parse[0] >= '1' && regcomp_parse[0] <= '9' ) { + parno = atoi(regcomp_parse++); - while (isDIGIT(*regparse)) - regparse++; + while (isDIGIT(*regcomp_parse)) + regcomp_parse++; ret = reganode(GROUPP, parno); if ((c = *nextchar()) != ')') FAIL2("Switch (?(number%c not recognized", c); @@ -1108,22 +1119,35 @@ reg(I32 paren, I32 *flagp) regtail(ret, ender); return ret; } else { - FAIL2("Unknown condition for (?(%.2s", regparse); + FAIL2("Unknown condition for (?(%.2s", regcomp_parse); } } case 0: FAIL("Sequence (? incomplete"); break; default: - --regparse; - while (*regparse && strchr("iogcmsx", *regparse)) { - if (*regparse != 'o') - pmflag(®flags, *regparse); - ++regparse; + --regcomp_parse; + parse_flags: + while (*regcomp_parse && strchr("iogcmsx", *regcomp_parse)) { + if (*regcomp_parse != 'o') + pmflag(flagsp, *regcomp_parse); + ++regcomp_parse; } + if (*regcomp_parse == '-') { + flagsp = &negflags; + ++regcomp_parse; + goto parse_flags; + } + regflags |= posflags; + regflags &= ~negflags; + if (*regcomp_parse == ':') { + regcomp_parse++; + paren = ':'; + break; + } unknown: - if (*regparse != ')') - FAIL2("Sequence (?%c...) not recognized", *regparse); + if (*regcomp_parse != ')') + FAIL2("Sequence (?%c...) not recognized", *regcomp_parse); nextchar(); *flagp = TRYAGAIN; return NULL; @@ -1142,7 +1166,7 @@ reg(I32 paren, I32 *flagp) br = regbranch(&flags, 1); if (br == NULL) return(NULL); - if (*regparse == '|') { + if (*regcomp_parse == '|') { if (!SIZE_ONLY && extralen) { reginsert(BRANCHJ, br); } else @@ -1157,11 +1181,11 @@ reg(I32 paren, I32 *flagp) regtail(ret, br); /* OPEN -> first. */ } else if (paren != '?') /* Not Conditional */ ret = br; - if (!(flags&HASWIDTH)) - *flagp &= ~HASWIDTH; + if (flags&HASWIDTH) + *flagp |= HASWIDTH; *flagp |= flags&SPSTART; lastbr = br; - while (*regparse == '|') { + while (*regcomp_parse == '|') { if (!SIZE_ONLY && extralen) { ender = reganode(LONGJMP,0); regtail(NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */ @@ -1174,8 +1198,8 @@ reg(I32 paren, I32 *flagp) return(NULL); regtail(lastbr, br); /* BRANCH -> BRANCH. */ lastbr = br; - if (!(flags&HASWIDTH)) - *flagp &= ~HASWIDTH; + if (flags&HASWIDTH) + *flagp |= HASWIDTH; *flagp |= flags&SPSTART; } @@ -1189,12 +1213,13 @@ reg(I32 paren, I32 *flagp) ender = reganode(CLOSE, parno); break; case '<': - case '>': case ',': case '=': case '!': - ender = reg_node(SUCCEED); *flagp &= ~HASWIDTH; + /* FALL THROUGH */ + case '>': + ender = reg_node(SUCCEED); break; case 0: ender = reg_node(END); @@ -1221,18 +1246,16 @@ reg(I32 paren, I32 *flagp) if (paren == '>') node = SUSPEND, flag = 0; reginsert(node,ret); -#ifdef REGALIGN_STRUCT ret->flags = flag; -#endif regtail(ret, reg_node(TAIL)); } } /* Check for proper termination. */ - if (paren && (regparse >= regxend || *nextchar() != ')')) { + if (paren && (regcomp_parse >= regxend || *nextchar() != ')')) { FAIL("unmatched () in regexp"); - } else if (!paren && regparse < regxend) { - if (*regparse == ')') { + } else if (!paren && regcomp_parse < regxend) { + if (*regcomp_parse == ')') { FAIL("unmatched () in regexp"); } else FAIL("junk on end of regexp"); /* "Can't happen". */ @@ -1272,9 +1295,9 @@ regbranch(I32 *flagp, I32 first) *flagp = WORST; /* Tentatively. */ - regparse--; + regcomp_parse--; nextchar(); - while (regparse < regxend && *regparse != '|' && *regparse != ')') { + while (regcomp_parse < regxend && *regcomp_parse != '|' && *regcomp_parse != ')') { flags &= ~TRYAGAIN; latest = regpiece(&flags); if (latest == NULL) { @@ -1321,7 +1344,7 @@ regpiece(I32 *flagp) register char op; register char *next; I32 flags; - char *origparse = regparse; + char *origparse = regcomp_parse; char *maxpos; I32 min; I32 max = REG_INFTY; @@ -1333,10 +1356,10 @@ regpiece(I32 *flagp) return(NULL); } - op = *regparse; + op = *regcomp_parse; - if (op == '{' && regcurly(regparse)) { - next = regparse + 1; + if (op == '{' && regcurly(regcomp_parse)) { + next = regcomp_parse + 1; maxpos = Nullch; while (isDIGIT(*next) || *next == ',') { if (*next == ',') { @@ -1350,18 +1373,18 @@ regpiece(I32 *flagp) if (*next == '}') { /* got one */ if (!maxpos) maxpos = next; - regparse++; - min = atoi(regparse); + regcomp_parse++; + min = atoi(regcomp_parse); if (*maxpos == ',') maxpos++; else - maxpos = regparse; + maxpos = regcomp_parse; max = atoi(maxpos); if (!max && *maxpos != '0') max = REG_INFTY; /* meaning "infinity" */ else if (max >= REG_INFTY) FAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1); - regparse = next; + regcomp_parse = next; nextchar(); do_curly: @@ -1384,12 +1407,12 @@ regpiece(I32 *flagp) if (SIZE_ONLY) extralen += 3; } -#ifdef REGALIGN_STRUCT ret->flags = 0; -#endif if (min > 0) - *flagp = (WORST|HASWIDTH); + *flagp = WORST; + if (max > 0) + *flagp |= HASWIDTH; if (max && max < min) FAIL("Can't do {n,m} with n > m"); if (!SIZE_ONLY) { @@ -1413,13 +1436,11 @@ regpiece(I32 *flagp) nextchar(); - *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH); + *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH); if (op == '*' && (flags&SIMPLE)) { reginsert(STAR, ret); -#ifdef REGALIGN_STRUCT ret->flags = 0; -#endif regnaughty += 4; } else if (op == '*') { @@ -1427,9 +1448,7 @@ regpiece(I32 *flagp) goto do_curly; } else if (op == '+' && (flags&SIMPLE)) { reginsert(PLUS, ret); -#ifdef REGALIGN_STRUCT ret->flags = 0; -#endif regnaughty += 3; } else if (op == '+') { @@ -1442,19 +1461,15 @@ regpiece(I32 *flagp) nest_check: if (dowarn && !SIZE_ONLY && !(flags&HASWIDTH) && max > 10000) { warn("%.*s matches null string many times", - regparse - origparse, origparse); + regcomp_parse - origparse, origparse); } - if (*regparse == '?') { + if (*regcomp_parse == '?') { nextchar(); reginsert(MINMOD, ret); -#ifdef REGALIGN regtail(ret, ret + NODE_STEP_REGNODE); -#else - regtail(ret, ret + 3); -#endif } - if (ISMULT2(regparse)) + if (ISMULT2(regcomp_parse)) FAIL("nested *?+ in regexp"); return(ret); @@ -1479,7 +1494,7 @@ regatom(I32 *flagp) *flagp = WORST; /* Tentatively. */ tryagain: - switch (*regparse) { + switch (*regcomp_parse) { case '^': seen_zerolen++; nextchar(); @@ -1491,7 +1506,7 @@ tryagain: ret = reg_node(BOL); break; case '$': - if (regparse[1]) + if (regcomp_parse[1]) seen_zerolen++; nextchar(); if (regflags & PMf_MULTILINE) @@ -1511,7 +1526,7 @@ tryagain: *flagp |= HASWIDTH|SIMPLE; break; case '[': - regparse++; + regcomp_parse++; ret = regclass(); *flagp |= HASWIDTH|SIMPLE; break; @@ -1531,12 +1546,12 @@ tryagain: *flagp |= TRYAGAIN; return NULL; } - FAIL2("internal urp in regexp at /%s/", regparse); + FAIL2("internal urp in regexp at /%s/", regcomp_parse); /* Supposed to be caught earlier. */ break; case '{': - if (!regcurly(regparse)) { - regparse++; + if (!regcurly(regcomp_parse)) { + regcomp_parse++; goto defchar; } /* FALL THROUGH */ @@ -1546,7 +1561,7 @@ tryagain: FAIL("?+*{} follows nothing in regexp"); break; case '\\': - switch (*++regparse) { + switch (*++regcomp_parse) { case 'A': seen_zerolen++; ret = reg_node(SBOL); @@ -1564,6 +1579,12 @@ tryagain: *flagp |= SIMPLE; nextchar(); break; + case 'z': + ret = reg_node(EOS); + *flagp |= SIMPLE; + seen_zerolen++; /* Do not optimize RE away */ + nextchar(); + break; case 'w': ret = reg_node((regflags & PMf_LOCALE) ? ALNUML : ALNUM); *flagp |= HASWIDTH|SIMPLE; @@ -1619,25 +1640,27 @@ tryagain: case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': { - I32 num = atoi(regparse); + I32 num = atoi(regcomp_parse); if (num > 9 && num >= regnpar) goto defchar; else { + if (!SIZE_ONLY && num > regcomp_rx->nparens) + FAIL("reference to nonexistent group"); regsawback = 1; ret = reganode((regflags & PMf_FOLD) ? ((regflags & PMf_LOCALE) ? REFFL : REFF) : REF, num); *flagp |= HASWIDTH; - while (isDIGIT(*regparse)) - regparse++; - regparse--; + while (isDIGIT(*regcomp_parse)) + regcomp_parse++; + regcomp_parse--; nextchar(); } } break; case '\0': - if (regparse >= regxend) + if (regcomp_parse >= regxend) FAIL("trailing \\ in regexp"); /* FALL THROUGH */ default: @@ -1647,8 +1670,8 @@ tryagain: case '#': if (regflags & PMf_EXTENDED) { - while (regparse < regxend && *regparse != '\n') regparse++; - if (regparse < regxend) + while (regcomp_parse < regxend && *regcomp_parse != '\n') regcomp_parse++; + if (regcomp_parse < regxend) goto tryagain; } /* FALL THROUGH */ @@ -1660,7 +1683,7 @@ tryagain: char *oldp, *s; I32 numlen; - regparse++; + regcomp_parse++; defchar: ret = reg_node((regflags & PMf_FOLD) @@ -1668,7 +1691,7 @@ tryagain: : EXACT); s = (char *) OPERAND(ret); regc(0, s++); /* save spot for len */ - for (len = 0, p = regparse - 1; + for (len = 0, p = regcomp_parse - 1; len < 127 && p < regxend; len++) { @@ -1690,6 +1713,7 @@ tryagain: case 'A': case 'G': case 'Z': + case 'z': case 'w': case 'W': case 'b': @@ -1772,7 +1796,7 @@ tryagain: regc(ender, s++); } loopdone: - regparse = p - 1; + regcomp_parse = p - 1; nextchar(); if (len < 0) FAIL("internal disaster in regexp"); @@ -1784,9 +1808,7 @@ tryagain: *OPERAND(ret) = len; regc('\0', s++); if (SIZE_ONLY) { -#ifdef REGALIGN_STRUCT regsize += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode); -#endif } else { regcode += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode); } @@ -1829,9 +1851,9 @@ regclass(void) ret = reg_node(ANYOF); for (Class = 0; Class < 33; Class++) regc(0, s++); - if (*regparse == '^') { /* Complement of range. */ + if (*regcomp_parse == '^') { /* Complement of range. */ regnaughty++; - regparse++; + regcomp_parse++; if (!SIZE_ONLY) *opnd |= ANYOF_INVERT; } @@ -1844,37 +1866,37 @@ regclass(void) } else { regsize += ANY_SKIP; } - if (*regparse == ']' || *regparse == '-') + if (*regcomp_parse == ']' || *regcomp_parse == '-') goto skipcond; /* allow 1st char to be ] or - */ - while (regparse < regxend && *regparse != ']') { + while (regcomp_parse < regxend && *regcomp_parse != ']') { skipcond: - Class = UCHARAT(regparse++); - if (Class == '[' && regparse + 1 < regxend && + Class = UCHARAT(regcomp_parse++); + if (Class == '[' && regcomp_parse + 1 < regxend && /* I smell either [: or [= or [. -- POSIX has been here, right? */ - (*regparse == ':' || *regparse == '=' || *regparse == '.')) { - char posixccc = *regparse; - char* posixccs = regparse++; + (*regcomp_parse == ':' || *regcomp_parse == '=' || *regcomp_parse == '.')) { + char posixccc = *regcomp_parse; + char* posixccs = regcomp_parse++; - while (regparse < regxend && *regparse != posixccc) - regparse++; - if (regparse == regxend) + while (regcomp_parse < regxend && *regcomp_parse != posixccc) + regcomp_parse++; + if (regcomp_parse == regxend) /* Grandfather lone [:, [=, [. */ - regparse = posixccs; + regcomp_parse = posixccs; else { - regparse++; /* skip over the posixccc */ - if (*regparse == ']') { + regcomp_parse++; /* skip over the posixccc */ + if (*regcomp_parse == ']') { /* Not Implemented Yet. * (POSIX Extended Character Classes, that is) * The text between e.g. [: and :] would start - * at posixccs + 1 and stop at regparse - 2. */ + * at posixccs + 1 and stop at regcomp_parse - 2. */ if (dowarn && !SIZE_ONLY) warn("Character class syntax [%c %c] is reserved for future extensions", posixccc, posixccc); - regparse++; /* skip over the ending ] */ + regcomp_parse++; /* skip over the ending ] */ } } } if (Class == '\\') { - Class = UCHARAT(regparse++); + Class = UCHARAT(regcomp_parse++); switch (Class) { case 'w': if (!SIZE_ONLY) { @@ -1962,17 +1984,17 @@ regclass(void) Class = '\007'; break; case 'x': - Class = scan_hex(regparse, 2, &numlen); - regparse += numlen; + Class = scan_hex(regcomp_parse, 2, &numlen); + regcomp_parse += numlen; break; case 'c': - Class = UCHARAT(regparse++); + Class = UCHARAT(regcomp_parse++); Class = toCTRL(Class); break; case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': - Class = scan_oct(--regparse, 3, &numlen); - regparse += numlen; + Class = scan_oct(--regcomp_parse, 3, &numlen); + regcomp_parse += numlen; break; } } @@ -1983,9 +2005,9 @@ regclass(void) } else { lastclass = Class; - if (*regparse == '-' && regparse+1 < regxend && - regparse[1] != ']') { - regparse++; + if (*regcomp_parse == '-' && regcomp_parse+1 < regxend && + regcomp_parse[1] != ']') { + regcomp_parse++; range = 1; continue; /* do it next time */ } @@ -1996,7 +2018,7 @@ regclass(void) } lastclass = Class; } - if (*regparse != ']') + if (*regcomp_parse != ']') FAIL("unmatched [] in regexp"); nextchar(); /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */ @@ -2021,25 +2043,25 @@ regclass(void) STATIC char* nextchar(void) { - char* retval = regparse++; + char* retval = regcomp_parse++; for (;;) { - if (*regparse == '(' && regparse[1] == '?' && - regparse[2] == '#') { - while (*regparse && *regparse != ')') - regparse++; - regparse++; + if (*regcomp_parse == '(' && regcomp_parse[1] == '?' && + regcomp_parse[2] == '#') { + while (*regcomp_parse && *regcomp_parse != ')') + regcomp_parse++; + regcomp_parse++; continue; } if (regflags & PMf_EXTENDED) { - if (isSPACE(*regparse)) { - regparse++; + if (isSPACE(*regcomp_parse)) { + regcomp_parse++; continue; } - else if (*regparse == '#') { - while (*regparse && *regparse != '\n') - regparse++; - regparse++; + else if (*regcomp_parse == '#') { + while (*regcomp_parse && *regcomp_parse != '\n') + regcomp_parse++; + regcomp_parse++; continue; } } @@ -2059,11 +2081,7 @@ reg_node(U8 op) ret = regcode; if (SIZE_ONLY) { SIZE_ALIGN(regsize); -#ifdef REGALIGN_STRUCT regsize += 1; -#else - regsize += 3; -#endif return(ret); } @@ -2087,11 +2105,7 @@ reganode(U8 op, U32 arg) ret = regcode; if (SIZE_ONLY) { SIZE_ALIGN(regsize); -#ifdef REGALIGN regsize += 2; -#else - regsize += 5; -#endif return(ret); } @@ -2143,9 +2157,6 @@ reginsert(U8 op, regnode *opnd) src = NEXTOPER(place); FILL_ADVANCE_NODE(place, op); Zero(src, offset, regnode); -#if defined(REGALIGN) && !defined(REGALIGN_STRUCT) - src[offset + 1] = '\177'; -#endif } /* @@ -2170,27 +2181,11 @@ regtail(regnode *p, regnode *val) scan = temp; } -#ifdef REGALIGN -# ifdef REGALIGN_STRUCT if (reg_off_by_arg[OP(scan)]) { ARG_SET(scan, val - scan); } else { NEXT_OFF(scan) = val - scan; } -# else - offset = val - scan; -# ifndef lint - *(short*)(scan+1) = offset; -# endif -#endif -#else - if (OP(scan) == BACK) - offset = scan - val; - else - offset = val - scan; - *(scan+1) = (offset>>8)&0377; - *(scan+2) = offset&0377; -#endif } /* @@ -2292,7 +2287,7 @@ dumpuntil(regnode *start, regnode *node, regnode *last, SV* sv, I32 l) else if (op == WHILEM) l--; } -#endif /* DEBUGGING */ +#endif /* DEBUGGING */ return node; } @@ -2354,8 +2349,10 @@ regdump(regexp *r) if (r->reganch & ROPT_IMPLICIT) PerlIO_printf(Perl_debug_log, "implicit "); PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen); + if (r->reganch & ROPT_EVAL_SEEN) + PerlIO_printf(Perl_debug_log, "with eval "); PerlIO_printf(Perl_debug_log, "\n"); -#endif /* DEBUGGING */ +#endif /* DEBUGGING */ } /* @@ -2381,6 +2378,9 @@ regprop(SV *sv, regnode *o) case EOL: p = "EOL"; break; + case EOS: + p = "EOS"; + break; case MEOL: p = "MEOL"; break; @@ -2436,18 +2436,10 @@ regprop(SV *sv, regnode *o) sv_catpvf(sv, "CURLY {%d,%d}", ARG1(o), ARG2(o)); break; case CURLYM: -#ifdef REGALIGN_STRUCT sv_catpvf(sv, "CURLYM[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o)); -#else - sv_catpvf(sv, "CURLYM {%d,%d}", ARG1(o), ARG2(o)); -#endif break; case CURLYN: -#ifdef REGALIGN_STRUCT sv_catpvf(sv, "CURLYN[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o)); -#else - sv_catpvf(sv, "CURLYN {%d,%d}", ARG1(o), ARG2(o)); -#endif break; case CURLYX: sv_catpvf(sv, "CURLYX {%d,%d}", ARG1(o), ARG2(o)); @@ -2481,18 +2473,10 @@ regprop(SV *sv, regnode *o) p = "GPOS"; break; case UNLESSM: -#ifdef REGALIGN_STRUCT sv_catpvf(sv, "UNLESSM[-%d]", o->flags); -#else - p = "UNLESSM"; -#endif break; case IFMATCH: -#ifdef REGALIGN_STRUCT sv_catpvf(sv, "IFMATCH[-%d]", o->flags); -#else - p = "IFMATCH"; -#endif break; case SUCCEED: p = "SUCCEED"; @@ -2562,7 +2546,7 @@ regprop(SV *sv, regnode *o) } if (p) sv_catpv(sv, p); -#endif /* DEBUGGING */ +#endif /* DEBUGGING */ } void @@ -2623,27 +2607,11 @@ regnext(register regnode *p) if (offset == 0) return(NULL); -#ifdef REGALIGN return(p+offset); -#else - if (OP(p) == BACK) - return(p-offset); - else - return(p+offset); -#endif } -#ifdef I_STDARG STATIC void re_croak2(const char* pat1,const char* pat2,...) -#else -/*VARARGS0*/ -static void -re_croak2(const char* pat1,const char* pat2, va_alist) - const char* pat1; - const char* pat2; - va_dcl -#endif { va_list args; STRLEN l1 = strlen(pat1); @@ -2657,13 +2625,9 @@ re_croak2(const char* pat1,const char* pat2, va_alist) l2 = 510 - l1; Copy(pat1, buf, l1 , char); Copy(pat2, buf + l1, l2 , char); - buf[l1 + l2 + 1] = '\n'; - buf[l1 + l2 + 2] = '\0'; -#ifdef I_STDARG + buf[l1 + l2] = '\n'; + buf[l1 + l2 + 1] = '\0'; va_start(args, pat2); -#else - va_start(args); -#endif message = mess(buf, &args); va_end(args); l1 = strlen(message);