X-Git-Url: https://perl5.git.perl.org/perl5.git/blobdiff_plain/89ebb4a3f2a55825eeed13aaf58db5c73d2140ef..973dddac3cae262865053bf44d56f52beac46f92:/regexec.c diff --git a/regexec.c b/regexec.c index f254713..e058216 100644 --- a/regexec.c +++ b/regexec.c @@ -55,7 +55,6 @@ # define PERL_NO_GET_CONTEXT #endif -/*SUPPRESS 112*/ /* * pregcomp and pregexec -- regsub and regerror are not used in perl * @@ -98,7 +97,6 @@ #define RF_warned 2 /* warned about big count? */ #define RF_evaled 4 /* Did an EVAL with setting? */ #define RF_utf8 8 /* String contains multibyte chars? */ -#define RF_false 16 /* odd number of nested negatives */ #define UTF ((PL_reg_flags & RF_utf8) != 0) @@ -141,7 +139,12 @@ #define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim)) #define HOPMAYBE3c(pos,off,lim) ((char*)HOPMAYBE3(pos,off,lim)) -#define LOAD_UTF8_CHARCLASS(a,b) STMT_START { if (!CAT2(PL_utf8_,a)) { ENTER; save_re_context(); (void)CAT2(is_utf8_, a)((U8*)b); LEAVE; } } STMT_END +#define LOAD_UTF8_CHARCLASS(class,str) STMT_START { \ + if (!CAT2(PL_utf8_,class)) { bool ok; ENTER; save_re_context(); ok=CAT2(is_utf8_,class)((const U8*)str); assert(ok); LEAVE; } } STMT_END +#define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS(alnum,"a") +#define LOAD_UTF8_CHARCLASS_DIGIT() LOAD_UTF8_CHARCLASS(digit,"0") +#define LOAD_UTF8_CHARCLASS_SPACE() LOAD_UTF8_CHARCLASS(space," ") +#define LOAD_UTF8_CHARCLASS_MARK() LOAD_UTF8_CHARCLASS(mark, "\xcd\x86") /* for use after a quantifier and before an EXACT-like node -- japhy */ #define JUMPABLE(rn) ( \ @@ -175,9 +178,9 @@ static void restore_pos(pTHX_ void *arg); STATIC CHECKPOINT S_regcppush(pTHX_ I32 parenfloor) { - int retval = PL_savestack_ix; + const int retval = PL_savestack_ix; #define REGCP_PAREN_ELEMS 4 - int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS; + const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS; int p; if (paren_elems_to_push < 0) @@ -207,11 +210,11 @@ S_regcppush(pTHX_ I32 parenfloor) } /* These are needed since we do not localize EVAL nodes: */ -# define REGCP_SET(cp) DEBUG_r(PerlIO_printf(Perl_debug_log, \ +# define REGCP_SET(cp) DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, \ " Setting an EVAL scope, savestack=%"IVdf"\n", \ (IV)PL_savestack_ix)); cp = PL_savestack_ix -# define REGCP_UNWIND(cp) DEBUG_r(cp != PL_savestack_ix ? \ +# define REGCP_UNWIND(cp) DEBUG_EXECUTE_r(cp != PL_savestack_ix ? \ PerlIO_printf(Perl_debug_log, \ " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \ (IV)(cp), (IV)PL_savestack_ix) : 0); regcpblow(cp) @@ -222,7 +225,8 @@ S_regcppop(pTHX) I32 i; U32 paren = 0; char *input; - I32 tmps; + + GET_RE_DEBUG_FLAGS_DECL; /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */ i = SSPOPINT; @@ -236,13 +240,14 @@ S_regcppop(pTHX) /* Now restore the parentheses context. */ for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); i > 0; i -= REGCP_PAREN_ELEMS) { + I32 tmps; paren = (U32)SSPOPINT; PL_reg_start_tmp[paren] = (char *) SSPOPPTR; PL_regstartp[paren] = SSPOPINT; tmps = SSPOPINT; if (paren <= *PL_reglastparen) PL_regendp[paren] = tmps; - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n", (UV)paren, (IV)PL_regstartp[paren], @@ -251,7 +256,7 @@ S_regcppop(pTHX) (paren > *PL_reglastparen ? "(no)" : "")); ); } - DEBUG_r( + DEBUG_EXECUTE_r( if ((I32)(*PL_reglastparen + 1) <= PL_regnpar) { PerlIO_printf(Perl_debug_log, " restoring \\%"IVdf"..\\%"IVdf" to undef\n", @@ -281,7 +286,7 @@ S_regcppop(pTHX) STATIC char * S_regcp_set_to(pTHX_ I32 ss) { - I32 tmp = PL_savestack_ix; + const I32 tmp = PL_savestack_ix; PL_savestack_ix = ss; regcppop(); @@ -364,7 +369,7 @@ S_cache_re(pTHX_ regexp *prog) /* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */ -/* If SCREAM, then SvPVX(sv) should be compatible with strpos and strend. +/* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend. Otherwise, only SvCUR(sv) is used to get strbeg. */ /* XXXX We assume that strpos is strbeg unless sv. */ @@ -405,37 +410,40 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, register SV *check; char *strbeg; char *t; - int do_utf8 = sv ? SvUTF8(sv) : 0; /* if no sv we have to assume bytes */ + const int do_utf8 = sv ? SvUTF8(sv) : 0; /* if no sv we have to assume bytes */ I32 ml_anch; register char *other_last = Nullch; /* other substr checked before this */ char *check_at = Nullch; /* check substr found at this pos */ - I32 multiline = prog->reganch & PMf_MULTILINE; + const I32 multiline = prog->reganch & PMf_MULTILINE; #ifdef DEBUGGING - char *i_strpos = strpos; - SV *dsv = PERL_DEBUG_PAD_ZERO(0); + const char * const i_strpos = strpos; + SV * const dsv = PERL_DEBUG_PAD_ZERO(0); #endif + + GET_RE_DEBUG_FLAGS_DECL; + RX_MATCH_UTF8_set(prog,do_utf8); if (prog->reganch & ROPT_UTF8) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "UTF-8 regex...\n")); PL_reg_flags |= RF_utf8; } - DEBUG_r({ - char *s = PL_reg_match_utf8 ? + DEBUG_EXECUTE_r({ + const char *s = PL_reg_match_utf8 ? sv_uni_display(dsv, sv, 60, UNI_DISPLAY_REGEX) : strpos; - int len = PL_reg_match_utf8 ? + const int len = PL_reg_match_utf8 ? strlen(s) : strend - strpos; if (!PL_colorset) reginitcolors(); if (PL_reg_match_utf8) - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "UTF-8 target...\n")); PerlIO_printf(Perl_debug_log, - "%sGuessing start of match, REx%s `%s%.60s%s%s' against `%s%.*s%s%s'...\n", - PL_colors[4],PL_colors[5],PL_colors[0], + "%sGuessing start of match, REx%s \"%s%.60s%s%s\" against \"%s%.*s%s%s\"...\n", + PL_colors[4], PL_colors[5], PL_colors[0], prog->precomp, PL_colors[1], (strlen(prog->precomp) > 60 ? "..." : ""), @@ -448,7 +456,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, /* CHR_DIST() would be more correct here but it makes things slow. */ if (prog->minlen > strend - strpos) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too short... [re_intuit_start]\n")); goto fail; } @@ -464,7 +472,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, check = prog->check_substr; } if (check == &PL_sv_undef) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Non-utf string cannot match utf check string\n")); goto fail; } @@ -476,10 +484,10 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, if (!ml_anch) { if ( !(prog->reganch & (ROPT_ANCH_GPOS /* Checked by the caller */ | ROPT_IMPLICIT)) /* not a real BOL */ - /* SvCUR is not set on references: SvRV and SvPVX overlap */ + /* SvCUR is not set on references: SvRV and SvPVX_const overlap */ && sv && !SvROK(sv) && (strpos != strbeg)) { - DEBUG_r(PerlIO_printf(Perl_debug_log, "Not at start...\n")); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n")); goto fail; } if (prog->check_offset_min == prog->check_offset_max && @@ -493,22 +501,22 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, if ( strend - s > slen || strend - s < slen - 1 || (strend - s == slen && strend[-1] != '\n')) { - DEBUG_r(PerlIO_printf(Perl_debug_log, "String too long...\n")); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n")); goto fail_finish; } /* Now should match s[0..slen-2] */ slen--; - if (slen && (*SvPVX(check) != *s + if (slen && (*SvPVX_const(check) != *s || (slen > 1 - && memNE(SvPVX(check), s, slen)))) { + && memNE(SvPVX_const(check), s, slen)))) { report_neq: - DEBUG_r(PerlIO_printf(Perl_debug_log, "String not equal...\n")); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n")); goto fail_finish; } } - else if (*SvPVX(check) != *s + else if (*SvPVX_const(check) != *s || ((slen = SvCUR(check)) > 1 - && memNE(SvPVX(check), s, slen))) + && memNE(SvPVX_const(check), s, slen))) goto report_neq; goto success_at_start; } @@ -519,9 +527,9 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, end_shift = prog->minlen - start_shift - CHR_SVLEN(check) + (SvTAIL(check) != 0); if (!ml_anch) { - I32 end = prog->check_offset_max + CHR_SVLEN(check) + const I32 end = prog->check_offset_max + CHR_SVLEN(check) - (SvTAIL(check) != 0); - I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end; + const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end; if (end_shift < eshift) end_shift = eshift; @@ -546,7 +554,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, the "check" substring in the region corrected by start/end_shift. */ if (flags & REXEC_SCREAM) { I32 p = -1; /* Internal iterator of scream. */ - I32 *pp = data ? data->scream_pos : &p; + I32 * const pp = data ? data->scream_pos : &p; if (PL_screamfirst[BmRARE(check)] >= 0 || ( BmRARE(check) == '\n' @@ -558,7 +566,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, goto fail_finish; /* we may be pointing at the wrong string */ if (s && RX_MATCH_COPIED(prog)) - s = strbeg + (s - SvPVX(sv)); + s = strbeg + (s - SvPVX_const(sv)); if (data) *data->scream_olds = s; } @@ -574,12 +582,12 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, /* Update the count-of-usability, remove useless subpatterns, unshift s. */ - DEBUG_r(PerlIO_printf(Perl_debug_log, "%s %s substr `%s%.*s%s'%s%s", + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s %s substr \"%s%.*s%s\"%s%s", (s ? "Found" : "Did not find"), (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) ? "anchored" : "floating"), PL_colors[0], (int)(SvCUR(check) - (SvTAIL(check)!=0)), - SvPVX(check), + SvPVX_const(check), PL_colors[1], (SvTAIL(check) ? "$" : ""), (s ? " at offset " : "...\n") ) ); @@ -589,12 +597,12 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, check_at = s; /* Finish the diagnostic message */ - DEBUG_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) ); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) ); /* Got a candidate. Check MBOL anchoring, and the *other* substr. Start with the other substr. XXXX no SCREAM optimization yet - and a very coarse implementation - XXXX /ttx+/ results in anchored=`ttx', floating=`x'. floating will + XXXX /ttx+/ results in anchored="ttx", floating="x". floating will *always* match. Probably should be marked during compile... Probably it is right to do no SCREAM here... */ @@ -607,7 +615,8 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) { do_other_anchored: { - char *last = HOP3c(s, -start_shift, strbeg), *last1, *last2; + char * const last = HOP3c(s, -start_shift, strbeg); + char *last1, *last2; char *s1 = s; SV* must; @@ -630,7 +639,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; if (must == &PL_sv_undef) { s = (char*)NULL; - DEBUG_r(must = prog->anchored_utf8); /* for debug */ + DEBUG_EXECUTE_r(must = prog->anchored_utf8); /* for debug */ } else s = fbm_instr( @@ -640,21 +649,21 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, must, multiline ? FBMrf_MULTILINE : 0 ); - DEBUG_r(PerlIO_printf(Perl_debug_log, - "%s anchored substr `%s%.*s%s'%s", + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, + "%s anchored substr \"%s%.*s%s\"%s", (s ? "Found" : "Contradicts"), PL_colors[0], (int)(SvCUR(must) - (SvTAIL(must)!=0)), - SvPVX(must), + SvPVX_const(must), PL_colors[1], (SvTAIL(must) ? "$" : ""))); if (!s) { if (last1 >= last2) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, ", giving up...\n")); goto fail_finish; } - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, ", trying floating at offset %ld...\n", (long)(HOP3c(s1, 1, strend) - i_strpos))); other_last = HOP3c(last1, prog->anchored_offset+1, strend); @@ -662,7 +671,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, goto restart; } else { - DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", (long)(s - i_strpos))); t = HOP3c(s, -prog->anchored_offset, strbeg); other_last = HOP3c(s, 1, strend); @@ -693,26 +702,26 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, and end-of-str is not later than strend we are OK. */ if (must == &PL_sv_undef) { s = (char*)NULL; - DEBUG_r(must = prog->float_utf8); /* for debug message */ + DEBUG_EXECUTE_r(must = prog->float_utf8); /* for debug message */ } else s = fbm_instr((unsigned char*)s, (unsigned char*)last + SvCUR(must) - (SvTAIL(must)!=0), must, multiline ? FBMrf_MULTILINE : 0); - DEBUG_r(PerlIO_printf(Perl_debug_log, "%s floating substr `%s%.*s%s'%s", + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s floating substr \"%s%.*s%s\"%s", (s ? "Found" : "Contradicts"), PL_colors[0], (int)(SvCUR(must) - (SvTAIL(must)!=0)), - SvPVX(must), + SvPVX_const(must), PL_colors[1], (SvTAIL(must) ? "$" : ""))); if (!s) { if (last1 == last) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, ", giving up...\n")); goto fail_finish; } - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, ", trying anchored starting at offset %ld...\n", (long)(s1 + 1 - i_strpos))); other_last = last; @@ -720,7 +729,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, goto restart; } else { - DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", (long)(s - i_strpos))); other_last = s; /* Fix this later. --Hugo */ s = s1; @@ -759,34 +768,34 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, is float. Redo checking for "other"=="fixed". */ strpos = t + 1; - DEBUG_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n", - PL_colors[0],PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset))); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n", + PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset))); goto do_other_anchored; } /* We don't contradict the found floating substring. */ /* XXXX Why not check for STCLASS? */ s = t + 1; - DEBUG_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n", - PL_colors[0],PL_colors[1], (long)(s - i_strpos))); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n", + PL_colors[0], PL_colors[1], (long)(s - i_strpos))); goto set_useful; } /* Position contradicts check-string */ /* XXXX probably better to look for check-string than for "\n", so one should lower the limit for t? */ - DEBUG_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n", - PL_colors[0],PL_colors[1], (long)(t + 1 - i_strpos))); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n", + PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos))); other_last = strpos = s = t + 1; goto restart; } t++; } - DEBUG_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n", - PL_colors[0],PL_colors[1])); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n", + PL_colors[0], PL_colors[1])); goto fail_finish; } else { - DEBUG_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n", - PL_colors[0],PL_colors[1])); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n", + PL_colors[0], PL_colors[1])); } s = t; set_useful: @@ -808,9 +817,9 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, t = strpos; goto find_anchor; } - DEBUG_r( if (ml_anch) + DEBUG_EXECUTE_r( if (ml_anch) PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n", - (long)(strpos - i_strpos), PL_colors[0],PL_colors[1]); + (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]); ); success_at_start: if (!(prog->reganch & ROPT_NAUGHTY) /* XXXX If strpos moved? */ @@ -825,7 +834,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, ))) { /* If flags & SOMETHING - do not do it many times on the same match */ - DEBUG_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n")); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n")); SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr); if (do_utf8 ? prog->check_substr : prog->check_utf8) SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8); @@ -853,49 +862,48 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, regstclass does not come from lookahead... */ /* If regstclass takes bytelength more than 1: If charlength==1, OK. This leaves EXACTF only, which is dealt with in find_byclass(). */ - U8* str = (U8*)STRING(prog->regstclass); - int cl_l = (PL_regkind[(U8)OP(prog->regstclass)] == EXACT + const U8* const str = (U8*)STRING(prog->regstclass); + const int cl_l = (PL_regkind[(U8)OP(prog->regstclass)] == EXACT ? CHR_DIST(str+STR_LEN(prog->regstclass), str) : 1); - char *endpos = (prog->anchored_substr || prog->anchored_utf8 || ml_anch) + const char * const endpos = (prog->anchored_substr || prog->anchored_utf8 || ml_anch) ? HOP3c(s, (prog->minlen ? cl_l : 0), strend) : (prog->float_substr || prog->float_utf8 ? HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend) : strend); - char *startpos = strbeg; t = s; cache_re(prog); - s = find_byclass(prog, prog->regstclass, s, endpos, startpos, 1); + s = find_byclass(prog, prog->regstclass, s, endpos, 1); if (!s) { #ifdef DEBUGGING - char *what = 0; + const char *what = 0; #endif if (endpos == strend) { - DEBUG_r( PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "Could not match STCLASS...\n") ); goto fail; } - DEBUG_r( PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "This position contradicts STCLASS...\n") ); if ((prog->reganch & ROPT_ANCH) && !ml_anch) goto fail; /* Contradict one of substrings */ if (prog->anchored_substr || prog->anchored_utf8) { if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) { - DEBUG_r( what = "anchored" ); + DEBUG_EXECUTE_r( what = "anchored" ); hop_and_restart: s = HOP3c(t, 1, strend); if (s + start_shift + end_shift > strend) { /* XXXX Should be taken into account earlier? */ - DEBUG_r( PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "Could not match STCLASS...\n") ); goto fail; } if (!check) goto giveup; - DEBUG_r( PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "Looking for %s substr starting at offset %ld...\n", what, (long)(s + start_shift - i_strpos)) ); goto restart; @@ -907,7 +915,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, s = check_at; if (!check) goto giveup; - DEBUG_r( PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "Looking for anchored substr starting at offset %ld...\n", (long)(other_last - i_strpos)) ); goto do_other_anchored; @@ -918,9 +926,9 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, s = t = t + 1; if (!check) goto giveup; - DEBUG_r( PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "Looking for /%s^%s/m starting at offset %ld...\n", - PL_colors[0],PL_colors[1], (long)(t - i_strpos)) ); + PL_colors[0], PL_colors[1], (long)(t - i_strpos)) ); goto try_at_offset; } if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */ @@ -928,23 +936,23 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, /* Check is floating subtring. */ retry_floating_check: t = check_at - start_shift; - DEBUG_r( what = "floating" ); + DEBUG_EXECUTE_r( what = "floating" ); goto hop_and_restart; } if (t != s) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "By STCLASS: moving %ld --> %ld\n", (long)(t - i_strpos), (long)(s - i_strpos)) ); } else { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Does not contradict STCLASS...\n"); ); } } giveup: - DEBUG_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n", + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n", PL_colors[4], (check ? "Guessed" : "Giving up"), PL_colors[5], (long)(s - i_strpos)) ); return s; @@ -953,16 +961,17 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, if (prog->check_substr || prog->check_utf8) /* could be removed already */ BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */ fail: - DEBUG_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n", - PL_colors[4],PL_colors[5])); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n", + PL_colors[4], PL_colors[5])); return Nullch; } /* We know what class REx starts with. Try to find this position... */ STATIC char * -S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *startpos, I32 norun) +S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, const char *strend, I32 norun) { - I32 doevery = (prog->reganch & ROPT_SKIP) == 0; + dVAR; + const I32 doevery = (prog->reganch & ROPT_SKIP) == 0; char *m; STRLEN ln; STRLEN lnc; @@ -971,7 +980,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta unsigned int c2; char *e; register I32 tmp = 1; /* Scratch variable? */ - register bool do_utf8 = PL_reg_match_utf8; + register const bool do_utf8 = PL_reg_match_utf8; /* We know what class it must start with. */ switch (OP(c)) { @@ -1030,14 +1039,15 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta U8 *sm = (U8 *) m; U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; + const U32 uniflags = ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY; to_utf8_lower((U8*)m, tmpbuf1, &ulen1); to_utf8_upper((U8*)m, tmpbuf2, &ulen2); c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE, - 0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY); + 0, uniflags); c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE, - 0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY); + 0, uniflags); lnc = 0; while (sm < ((U8 *) m + ln)) { lnc++; @@ -1075,16 +1085,14 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if (do_utf8) { UV c, f; U8 tmpbuf [UTF8_MAXBYTES+1]; - U8 foldbuf[UTF8_MAXBYTES_CASE+1]; STRLEN len, foldlen; - + const U32 uniflags = ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY; if (c1 == c2) { /* Upper and lower of 1st char are equal - * probably not a "letter". */ while (s <= e) { c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, - ckWARN(WARN_UTF8) ? - 0 : UTF8_ALLOW_ANY); + uniflags); if ( c == c1 && (ln == len || ibcmp_utf8(s, (char **)0, 0, do_utf8, @@ -1092,6 +1100,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta && (norun || regtry(prog, s)) ) goto got_it; else { + U8 foldbuf[UTF8_MAXBYTES_CASE+1]; uvchr_to_utf8(tmpbuf, c); f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); if ( f != c @@ -1110,8 +1119,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta else { while (s <= e) { c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, - ckWARN(WARN_UTF8) ? - 0 : UTF8_ALLOW_ANY); + uniflags); /* Handle some of the three Greek sigmas cases. * Note that not all the possible combinations @@ -1131,6 +1139,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta && (norun || regtry(prog, s)) ) goto got_it; else { + U8 foldbuf[UTF8_MAXBYTES_CASE+1]; uvchr_to_utf8(tmpbuf, c); f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); if ( f != c @@ -1184,7 +1193,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } tmp = ((OP(c) == BOUND ? isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); - LOAD_UTF8_CHARCLASS(alnum,"a"); + LOAD_UTF8_CHARCLASS_ALNUM(); while (s + (uskip = UTF8SKIP(s)) <= strend) { if (tmp == !(OP(c) == BOUND ? swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : @@ -1227,7 +1236,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } tmp = ((OP(c) == NBOUND ? isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); - LOAD_UTF8_CHARCLASS(alnum,"a"); + LOAD_UTF8_CHARCLASS_ALNUM(); while (s + (uskip = UTF8SKIP(s)) <= strend) { if (tmp == !(OP(c) == NBOUND ? swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : @@ -1256,7 +1265,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta break; case ALNUM: if (do_utf8) { - LOAD_UTF8_CHARCLASS(alnum,"a"); + LOAD_UTF8_CHARCLASS_ALNUM(); while (s + (uskip = UTF8SKIP(s)) <= strend) { if (swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) { if (tmp && (norun || regtry(prog, s))) @@ -1314,7 +1323,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta break; case NALNUM: if (do_utf8) { - LOAD_UTF8_CHARCLASS(alnum,"a"); + LOAD_UTF8_CHARCLASS_ALNUM(); while (s + (uskip = UTF8SKIP(s)) <= strend) { if (!swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) { if (tmp && (norun || regtry(prog, s))) @@ -1372,7 +1381,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta break; case SPACE: if (do_utf8) { - LOAD_UTF8_CHARCLASS(space," "); + LOAD_UTF8_CHARCLASS_SPACE(); while (s + (uskip = UTF8SKIP(s)) <= strend) { if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)) { if (tmp && (norun || regtry(prog, s))) @@ -1430,7 +1439,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta break; case NSPACE: if (do_utf8) { - LOAD_UTF8_CHARCLASS(space," "); + LOAD_UTF8_CHARCLASS_SPACE(); while (s + (uskip = UTF8SKIP(s)) <= strend) { if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8))) { if (tmp && (norun || regtry(prog, s))) @@ -1488,7 +1497,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta break; case DIGIT: if (do_utf8) { - LOAD_UTF8_CHARCLASS(digit,"0"); + LOAD_UTF8_CHARCLASS_DIGIT(); while (s + (uskip = UTF8SKIP(s)) <= strend) { if (swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) { if (tmp && (norun || regtry(prog, s))) @@ -1546,7 +1555,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta break; case NDIGIT: if (do_utf8) { - LOAD_UTF8_CHARCLASS(digit,"0"); + LOAD_UTF8_CHARCLASS_DIGIT(); while (s + (uskip = UTF8SKIP(s)) <= strend) { if (!swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) { if (tmp && (norun || regtry(prog, s))) @@ -1628,18 +1637,20 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * register char *startpos = stringarg; I32 minlen; /* must match at least this many chars */ I32 dontbother = 0; /* how many characters not to try at end */ - /* I32 start_shift = 0; */ /* Offset of the start to find - constant substr. */ /* CC */ I32 end_shift = 0; /* Same for the end. */ /* CC */ I32 scream_pos = -1; /* Internal iterator of scream. */ char *scream_olds; SV* oreplsv = GvSV(PL_replgv); - bool do_utf8 = DO_UTF8(sv); - I32 multiline = prog->reganch & PMf_MULTILINE; + const bool do_utf8 = DO_UTF8(sv); + const I32 multiline = prog->reganch & PMf_MULTILINE; #ifdef DEBUGGING SV *dsv0 = PERL_DEBUG_PAD_ZERO(0); SV *dsv1 = PERL_DEBUG_PAD_ZERO(1); #endif + + GET_RE_DEBUG_FLAGS_DECL; + + PERL_UNUSED_ARG(data); RX_MATCH_UTF8_set(prog,do_utf8); PL_regcc = 0; @@ -1657,7 +1668,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * minlen = prog->minlen; if (strend - startpos < minlen) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too short [regexec_flags]...\n")); goto phooey; } @@ -1718,25 +1729,25 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * d.scream_pos = &scream_pos; s = re_intuit_start(prog, sv, s, strend, flags, &d); if (!s) { - DEBUG_r(PerlIO_printf(Perl_debug_log, "Not present...\n")); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n")); goto phooey; /* not present */ } } - DEBUG_r({ - char *s0 = UTF ? - pv_uni_display(dsv0, (U8*)prog->precomp, prog->prelen, 60, - UNI_DISPLAY_REGEX) : - prog->precomp; - int len0 = UTF ? SvCUR(dsv0) : prog->prelen; - char *s1 = do_utf8 ? sv_uni_display(dsv1, sv, 60, + DEBUG_EXECUTE_r({ + const char * const s0 = UTF + ? pv_uni_display(dsv0, (U8*)prog->precomp, prog->prelen, 60, + UNI_DISPLAY_REGEX) + : prog->precomp; + const int len0 = UTF ? SvCUR(dsv0) : prog->prelen; + const char * const s1 = do_utf8 ? sv_uni_display(dsv1, sv, 60, UNI_DISPLAY_REGEX) : startpos; - int len1 = do_utf8 ? SvCUR(dsv1) : strend - startpos; + const int len1 = do_utf8 ? SvCUR(dsv1) : strend - startpos; if (!PL_colorset) reginitcolors(); PerlIO_printf(Perl_debug_log, - "%sMatching REx%s `%s%*.*s%s%s' against `%s%.*s%s%s'\n", - PL_colors[4],PL_colors[5],PL_colors[0], + "%sMatching REx%s \"%s%*.*s%s%s\" against \"%s%.*s%s%s\"\n", + PL_colors[4], PL_colors[5], PL_colors[0], len0, len0, s0, PL_colors[1], len0 > 60 ? "..." : "", @@ -1806,12 +1817,12 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * #endif if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); - ch = SvPVX(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0]; + ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0]; if (do_utf8) { while (s < strend) { if (*s == ch) { - DEBUG_r( did_match = 1 ); + DEBUG_EXECUTE_r( did_match = 1 ); if (regtry(prog, s)) goto got_it; s += UTF8SKIP(s); while (s < strend && *s == ch) @@ -1823,7 +1834,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * else { while (s < strend) { if (*s == ch) { - DEBUG_r( did_match = 1 ); + DEBUG_EXECUTE_r( did_match = 1 ); if (regtry(prog, s)) goto got_it; s++; while (s < strend && *s == ch) @@ -1832,12 +1843,11 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * s++; } } - DEBUG_r(if (!did_match) + DEBUG_EXECUTE_r(if (!did_match) PerlIO_printf(Perl_debug_log, "Did not find anchored character...\n") ); } - /*SUPPRESS 560*/ else if (prog->anchored_substr != Nullsv || prog->anchored_utf8 != Nullsv || ((prog->float_substr != Nullsv || prog->float_utf8 != Nullsv) @@ -1875,7 +1885,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * else last1 = s - 1; /* bogus */ - /* XXXX check_substr already used to find `s', can optimize if + /* XXXX check_substr already used to find "s", can optimize if check_substr==must. */ scream_pos = -1; dontbother = end_shift; @@ -1889,8 +1899,8 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * multiline ? FBMrf_MULTILINE : 0))) ) { /* we may be pointing at the wrong string */ if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog)) - s = strbeg + (s - SvPVX(sv)); - DEBUG_r( did_match = 1 ); + s = strbeg + (s - SvPVX_const(sv)); + DEBUG_EXECUTE_r( did_match = 1 ); if (HOPc(s, -back_max) > last1) { last1 = HOPc(s, -back_min); s = HOPc(s, -back_max); @@ -1916,14 +1926,14 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * } } } - DEBUG_r(if (!did_match) + DEBUG_EXECUTE_r(if (!did_match) PerlIO_printf(Perl_debug_log, - "Did not find %s substr `%s%.*s%s'%s...\n", + "Did not find %s substr \"%s%.*s%s\"%s...\n", ((must == prog->anchored_substr || must == prog->anchored_utf8) ? "anchored" : "floating"), PL_colors[0], (int)(SvCUR(must) - (SvTAIL(must)!=0)), - SvPVX(must), + SvPVX_const(must), PL_colors[1], (SvTAIL(must) ? "$" : "")) ); goto phooey; @@ -1935,30 +1945,30 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * if (PL_regkind[op] != EXACT && op != CANY) strend = HOPc(strend, -(minlen - 1)); } - DEBUG_r({ + DEBUG_EXECUTE_r({ SV *prop = sv_newmortal(); - char *s0; - char *s1; + const char *s0; + const char *s1; int len0; int len1; regprop(prop, c); s0 = UTF ? - pv_uni_display(dsv0, (U8*)SvPVX(prop), SvCUR(prop), 60, + pv_uni_display(dsv0, (U8*)SvPVX_const(prop), SvCUR(prop), 60, UNI_DISPLAY_REGEX) : - SvPVX(prop); + SvPVX_const(prop); len0 = UTF ? SvCUR(dsv0) : SvCUR(prop); s1 = UTF ? sv_uni_display(dsv1, sv, 60, UNI_DISPLAY_REGEX) : s; len1 = UTF ? SvCUR(dsv1) : strend - s; PerlIO_printf(Perl_debug_log, - "Matching stclass `%*.*s' against `%*.*s'\n", + "Matching stclass \"%*.*s\" against \"%*.*s\"\n", len0, len0, s0, len1, len1, s1); }); - if (find_byclass(prog, c, s, strend, startpos, 0)) + if (find_byclass(prog, c, s, strend, 0)) goto got_it; - DEBUG_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass...\n")); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass...\n")); } else { dontbother = 0; @@ -1978,11 +1988,11 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * last = scream_olds; /* Only one occurrence. */ /* we may be pointing at the wrong string */ else if (RX_MATCH_COPIED(prog)) - s = strbeg + (s - SvPVX(sv)); + s = strbeg + (s - SvPVX_const(sv)); } else { STRLEN len; - char *little = SvPV(float_real, len); + const char * const little = SvPV_const(float_real, len); if (SvTAIL(float_real)) { if (memEQ(strend - len + 1, little, len - 1)) @@ -1997,13 +2007,13 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * if (len) last = rninstr(s, strend, little, little + len); else - last = strend; /* matching `$' */ + last = strend; /* matching "$" */ } } if (last == NULL) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sCan't trim the tail, match fails (should not happen)%s\n", - PL_colors[4],PL_colors[5])); + PL_colors[4], PL_colors[5])); goto phooey; /* Should not happen! */ } dontbother = strend - last + prog->float_min_offset; @@ -2049,7 +2059,7 @@ got_it: RX_MATCH_COPY_FREE(prog); if (flags & REXEC_COPY_STR) { I32 i = PL_regeol - startpos + (stringarg - strbeg); -#ifdef PERL_COPY_ON_WRITE +#ifdef PERL_OLD_COPY_ON_WRITE if ((SvIsCOW(sv) || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) { if (DEBUG_C_TEST) { @@ -2058,7 +2068,7 @@ got_it: (int) SvTYPE(sv)); } prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv); - prog->subbeg = SvPVX(prog->saved_copy); + prog->subbeg = (char *)SvPVX_const(prog->saved_copy); assert (SvPOKp(prog->saved_copy)); } else #endif @@ -2078,8 +2088,8 @@ got_it: return 1; phooey: - DEBUG_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n", - PL_colors[4],PL_colors[5])); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n", + PL_colors[4], PL_colors[5])); if (PL_reg_eval_set) restore_pos(aTHX_ 0); return 0; @@ -2095,6 +2105,7 @@ S_regtry(pTHX_ regexp *prog, char *startpos) register I32 *sp; register I32 *ep; CHECKPOINT lastcp; + GET_RE_DEBUG_FLAGS_DECL; #ifdef DEBUGGING PL_regindent = 0; /* XXXX Not good when matches are reenterable... */ @@ -2103,7 +2114,7 @@ S_regtry(pTHX_ regexp *prog, char *startpos) MAGIC *mg; PL_reg_eval_set = RS_init; - DEBUG_r(DEBUG_s( + DEBUG_EXECUTE_r(DEBUG_s( PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n", (IV)(PL_stack_sp - PL_stack_base)); )); @@ -2135,7 +2146,7 @@ S_regtry(pTHX_ regexp *prog, char *startpos) SAVEDESTRUCTOR_X(restore_pos, 0); } if (!PL_reg_curpm) { - Newz(22,PL_reg_curpm, 1, PMOP); + Newxz(PL_reg_curpm, 1, PMOP); #ifdef USE_ITHREADS { SV* repointer = newSViv(0); @@ -2156,7 +2167,7 @@ S_regtry(pTHX_ regexp *prog, char *startpos) $` inside (?{}) could fail... */ PL_reg_oldsaved = prog->subbeg; PL_reg_oldsavedlen = prog->sublen; -#ifdef PERL_COPY_ON_WRITE +#ifdef PERL_OLD_COPY_ON_WRITE PL_nrs = prog->saved_copy; #endif RX_MATCH_COPIED_off(prog); @@ -2175,13 +2186,13 @@ S_regtry(pTHX_ regexp *prog, char *startpos) prog->lastparen = 0; prog->lastcloseparen = 0; PL_regsize = 0; - DEBUG_r(PL_reg_starttry = startpos); + DEBUG_EXECUTE_r(PL_reg_starttry = startpos); if (PL_reg_start_tmpl <= prog->nparens) { PL_reg_start_tmpl = prog->nparens*3/2 + 3; if(PL_reg_start_tmp) Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); else - New(22,PL_reg_start_tmp, PL_reg_start_tmpl, char*); + Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*); } /* XXXX What this code is doing here?!!! There should be no need @@ -2256,7 +2267,105 @@ typedef union re_unwind_t { #define sayNO_SILENT goto do_no #define saySAME(x) if (x) goto yes; else goto no -#define REPORT_CODE_OFF 24 +#define POSCACHE_SUCCESS 0 /* caching success rather than failure */ +#define POSCACHE_SEEN 1 /* we know what we're caching */ +#define POSCACHE_START 2 /* the real cache: this bit maps to pos 0 */ +#define CACHEsayYES STMT_START { \ + if (cache_offset | cache_bit) { \ + if (!(PL_reg_poscache[0] & (1<states[ state ].wordnum ) { \ + if ( !accepted ) { \ + ENTER; \ + SAVETMPS; \ + bufflen = TRIE_INITAL_ACCEPT_BUFFLEN ; \ + sv_accept_buff=NEWSV( 1234, \ + bufflen * sizeof(reg_trie_accepted) - 1 ); \ + SvCUR_set( sv_accept_buff, sizeof(reg_trie_accepted) ); \ + SvPOK_on( sv_accept_buff ); \ + sv_2mortal( sv_accept_buff ); \ + accept_buff = (reg_trie_accepted*)SvPV_nolen( sv_accept_buff );\ + } else { \ + if ( accepted >= bufflen ) { \ + bufflen *= 2; \ + accept_buff =(reg_trie_accepted*)SvGROW( sv_accept_buff, \ + bufflen * sizeof(reg_trie_accepted) ); \ + } \ + SvCUR_set( sv_accept_buff,SvCUR( sv_accept_buff ) \ + + sizeof( reg_trie_accepted ) ); \ + } \ + accept_buff[ accepted ].wordnum = trie->states[ state ].wordnum; \ + accept_buff[ accepted ].endpos = uc; \ + ++accepted; \ + } } STMT_END + +#define TRIE_HANDLE_CHAR STMT_START { \ + if ( uvc < 256 ) { \ + charid = trie->charmap[ uvc ]; \ + } else { \ + charid = 0; \ + if( trie->widecharmap ) { \ + SV** svpp = (SV**)NULL; \ + svpp = hv_fetch( trie->widecharmap, (char*)&uvc, \ + sizeof( UV ), 0 ); \ + if ( svpp ) { \ + charid = (U16)SvIV( *svpp ); \ + } \ + } \ + } \ + if ( charid && \ + ( base + charid > trie->uniquecharcount ) && \ + ( base + charid - 1 - trie->uniquecharcount < trie->lasttrans) && \ + trie->trans[ base + charid - 1 - trie->uniquecharcount ].check == state ) \ + { \ + state = trie->trans[ base + charid - 1 - trie->uniquecharcount ].next; \ + } else { \ + state = 0; \ + } \ + uc += len; \ + } STMT_END /* - regmatch - main matching routine @@ -2275,6 +2384,7 @@ typedef union re_unwind_t { STATIC I32 /* 0 failure, 1 success */ S_regmatch(pTHX_ regnode *prog) { + dVAR; register regnode *scan; /* Current node. */ regnode *next; /* Next node. */ regnode *inner; /* Next node in internal branch. */ @@ -2287,29 +2397,42 @@ S_regmatch(pTHX_ regnode *prog) register I32 c1 = 0, c2 = 0, paren; /* case fold search, parenth */ int minmod = 0, sw = 0, logical = 0; I32 unwind = 0; + + /* used by the trie code */ + SV *sv_accept_buff = 0; /* accepting states we have traversed */ + reg_trie_accepted *accept_buff = 0; /* "" */ + reg_trie_data *trie; /* what trie are we using right now */ + U32 accepted = 0; /* how many accepting states we have seen*/ + #if 0 I32 firstcp = PL_savestack_ix; #endif - register bool do_utf8 = PL_reg_match_utf8; + register const bool do_utf8 = PL_reg_match_utf8; #ifdef DEBUGGING SV *dsv0 = PERL_DEBUG_PAD_ZERO(0); SV *dsv1 = PERL_DEBUG_PAD_ZERO(1); SV *dsv2 = PERL_DEBUG_PAD_ZERO(2); + + SV *re_debug_flags = NULL; #endif + U32 uniflags = ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY; + + GET_RE_DEBUG_FLAGS; #ifdef DEBUGGING PL_regindent++; #endif + /* Note that nextchr is a byte even in UTF */ nextchr = UCHARAT(locinput); scan = prog; while (scan != NULL) { - DEBUG_r( { + DEBUG_EXECUTE_r( { SV *prop = sv_newmortal(); - int docolor = *PL_colors[0]; - int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ + const int docolor = *PL_colors[0]; + const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ int l = (PL_regeol - locinput) > taill ? taill : (PL_regeol - locinput); /* The part of the string before starttry has one color (pref0_len chars), between starttry and current @@ -2335,22 +2458,22 @@ S_regmatch(pTHX_ regnode *prog) pref0_len = pref_len; regprop(prop, scan); { - char *s0 = + const char * const s0 = do_utf8 && OP(scan) != CANY ? pv_uni_display(dsv0, (U8*)(locinput - pref_len), pref0_len, 60, UNI_DISPLAY_REGEX) : locinput - pref_len; - int len0 = do_utf8 ? strlen(s0) : pref0_len; - char *s1 = do_utf8 && OP(scan) != CANY ? + const int len0 = do_utf8 ? strlen(s0) : pref0_len; + const char * const s1 = do_utf8 && OP(scan) != CANY ? pv_uni_display(dsv1, (U8*)(locinput - pref_len + pref0_len), pref_len - pref0_len, 60, UNI_DISPLAY_REGEX) : locinput - pref_len + pref0_len; - int len1 = do_utf8 ? strlen(s1) : pref_len - pref0_len; - char *s2 = do_utf8 && OP(scan) != CANY ? + const int len1 = do_utf8 ? strlen(s1) : pref_len - pref0_len; + const char * const s2 = do_utf8 && OP(scan) != CANY ? pv_uni_display(dsv2, (U8*)locinput, PL_regeol - locinput, 60, UNI_DISPLAY_REGEX) : locinput; - int len2 = do_utf8 ? strlen(s2) : l; + const int len2 = do_utf8 ? strlen(s2) : l; PerlIO_printf(Perl_debug_log, "%4"IVdf" <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3"IVdf":%*s%s\n", (IV)(locinput - PL_bostr), @@ -2367,7 +2490,7 @@ S_regmatch(pTHX_ regnode *prog) 15 - l - pref_len + 1, "", (IV)(scan - PL_regprogram), PL_regindent*2, "", - SvPVX(prop)); + SvPVX_const(prop)); } }); @@ -2444,24 +2567,245 @@ S_regmatch(pTHX_ regnode *prog) else nextchr = UCHARAT(++locinput); break; + + + + /* + traverse the TRIE keeping track of all accepting states + we transition through until we get to a failing node. + + we use two slightly different pieces of code to handle + the traversal depending on whether its case sensitive or + not. we reuse the accept code however. (this should probably + be turned into a macro.) + + */ + case TRIEF: + case TRIEFL: + { + U8 *uc = ( U8* )locinput; + U32 state = 1; + U16 charid = 0; + U32 base = 0; + UV uvc = 0; + STRLEN len = 0; + STRLEN foldlen = 0; + U8 *uscan = (U8*)NULL; + STRLEN bufflen=0; + accepted = 0; + + trie = (reg_trie_data*)PL_regdata->data[ ARG( scan ) ]; + + while ( state && uc <= (U8*)PL_regeol ) { + + TRIE_CHECK_STATE_IS_ACCEPTING; + + base = trie->states[ state ].trans.base; + + DEBUG_TRIE_EXECUTE_r( + PerlIO_printf( Perl_debug_log, + "%*s %sState: %4"UVxf", Base: %4"UVxf", Accepted: %4"UVxf" ", + REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4], + (UV)state, (UV)base, (UV)accepted ); + ); + + if ( base ) { + + if ( do_utf8 ) { + if ( foldlen>0 ) { + uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); + foldlen -= len; + uscan += len; + len=0; + } else { + U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; + uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); + uvc = to_uni_fold( uvc, foldbuf, &foldlen ); + foldlen -= UNISKIP( uvc ); + uscan = foldbuf + UNISKIP( uvc ); + } + } else { + uvc = (UV)*uc; + len = 1; + } + + TRIE_HANDLE_CHAR; + + } else { + state = 0; + } + DEBUG_TRIE_EXECUTE_r( + PerlIO_printf( Perl_debug_log, + "Charid:%3x CV:%4"UVxf" After State: %4"UVxf"%s\n", + charid, uvc, (UV)state, PL_colors[5] ); + ); + } + if ( !accepted ) { + sayNO; + } else { + goto TrieAccept; + } + } + /* unreached codepoint: we jump into the middle of the next case + from previous if blocks */ + case TRIE: + { + U8 *uc = (U8*)locinput; + U32 state = 1; + U16 charid = 0; + U32 base = 0; + UV uvc = 0; + STRLEN len = 0; + STRLEN bufflen = 0; + accepted = 0; + + trie = (reg_trie_data*)PL_regdata->data[ ARG( scan ) ]; + + while ( state && uc <= (U8*)PL_regeol ) { + + TRIE_CHECK_STATE_IS_ACCEPTING; + + base = trie->states[ state ].trans.base; + + DEBUG_TRIE_EXECUTE_r( + PerlIO_printf( Perl_debug_log, + "%*s %sState: %4"UVxf", Base: %4"UVxf", Accepted: %4"UVxf" ", + REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4], + (UV)state, (UV)base, (UV)accepted ); + ); + + if ( base ) { + + if ( do_utf8 ) { + uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); + } else { + uvc = (U32)*uc; + len = 1; + } + + TRIE_HANDLE_CHAR; + + } else { + state = 0; + } + DEBUG_TRIE_EXECUTE_r( + PerlIO_printf( Perl_debug_log, + "Charid:%3x CV:%4"UVxf" After State: %4"UVxf"%s\n", + charid, uvc, (UV)state, PL_colors[5] ); + ); + } + if ( !accepted ) { + sayNO; + } + } + + + /* + There was at least one accepting state that we + transitioned through. Presumably the number of accepting + states is going to be low, typically one or two. So we + simply scan through to find the one with lowest wordnum. + Once we find it, we swap the last state into its place + and decrement the size. We then try to match the rest of + the pattern at the point where the word ends, if we + succeed then we end the loop, otherwise the loop + eventually terminates once all of the accepting states + have been tried. + */ + TrieAccept: + { + int gotit = 0; + + if ( accepted == 1 ) { + DEBUG_EXECUTE_r({ + SV **tmp = av_fetch( trie->words, accept_buff[ 0 ].wordnum-1, 0 ); + PerlIO_printf( Perl_debug_log, + "%*s %sonly one match : #%d <%s>%s\n", + REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4], + accept_buff[ 0 ].wordnum, + tmp ? SvPV_nolen_const( *tmp ) : "not compiled under -Dr", + PL_colors[5] ); + }); + PL_reginput = (char *)accept_buff[ 0 ].endpos; + /* in this case we free tmps/leave before we call regmatch + as we wont be using accept_buff again. */ + FREETMPS; + LEAVE; + gotit = regmatch( scan + NEXT_OFF( scan ) ); + } else { + DEBUG_EXECUTE_r( + PerlIO_printf( Perl_debug_log,"%*s %sgot %"IVdf" possible matches%s\n", + REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4], (IV)accepted, + PL_colors[5] ); + ); + while ( !gotit && accepted-- ) { + U32 best = 0; + U32 cur; + for( cur = 1 ; cur <= accepted ; cur++ ) { + DEBUG_TRIE_EXECUTE_r( + PerlIO_printf( Perl_debug_log, + "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n", + REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4], + (IV)best, accept_buff[ best ].wordnum, (IV)cur, + accept_buff[ cur ].wordnum, PL_colors[5] ); + ); + + if ( accept_buff[ cur ].wordnum < accept_buff[ best ].wordnum ) + best = cur; + } + DEBUG_EXECUTE_r({ + SV **tmp = av_fetch( trie->words, accept_buff[ best ].wordnum - 1, 0 ); + PerlIO_printf( Perl_debug_log, "%*s %strying alternation #%d <%s> at 0x%p%s\n", + REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4], + accept_buff[best].wordnum, + tmp ? SvPV_nolen_const( *tmp ) : "not compiled under -Dr",scan, + PL_colors[5] ); + }); + if ( best= PL_regeol) sayNO; if (NATIVE_TO_UNI(*(U8*)s) != utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen, - ckWARN(WARN_UTF8) ? - 0 : UTF8_ALLOW_ANY)) + uniflags)) sayNO; l += ulen; s ++; @@ -2470,12 +2814,12 @@ S_regmatch(pTHX_ regnode *prog) else { /* The target is not utf8, the pattern is utf8. */ while (s < e) { + STRLEN ulen; if (l >= PL_regeol) sayNO; if (NATIVE_TO_UNI(*((U8*)l)) != utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen, - ckWARN(WARN_UTF8) ? - 0 : UTF8_ALLOW_ANY)) + uniflags)) sayNO; s += ulen; l ++; @@ -2584,7 +2928,7 @@ S_regmatch(pTHX_ regnode *prog) if (!nextchr) sayNO; if (do_utf8) { - LOAD_UTF8_CHARCLASS(alnum,"a"); + LOAD_UTF8_CHARCLASS_ALNUM(); if (!(OP(scan) == ALNUM ? swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) : isALNUM_LC_utf8((U8*)locinput))) @@ -2607,7 +2951,7 @@ S_regmatch(pTHX_ regnode *prog) if (!nextchr && locinput >= PL_regeol) sayNO; if (do_utf8) { - LOAD_UTF8_CHARCLASS(alnum,"a"); + LOAD_UTF8_CHARCLASS_ALNUM(); if (OP(scan) == NALNUM ? swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) : isALNUM_LC_utf8((U8*)locinput)) @@ -2634,13 +2978,13 @@ S_regmatch(pTHX_ regnode *prog) if (locinput == PL_bostr) ln = '\n'; else { - U8 *r = reghop3((U8*)locinput, -1, (U8*)PL_bostr); + const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr); ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, 0); } if (OP(scan) == BOUND || OP(scan) == NBOUND) { ln = isALNUM_uni(ln); - LOAD_UTF8_CHARCLASS(alnum,"a"); + LOAD_UTF8_CHARCLASS_ALNUM(); n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8); } else { @@ -2672,7 +3016,7 @@ S_regmatch(pTHX_ regnode *prog) sayNO; if (do_utf8) { if (UTF8_IS_CONTINUED(nextchr)) { - LOAD_UTF8_CHARCLASS(space," "); + LOAD_UTF8_CHARCLASS_SPACE(); if (!(OP(scan) == SPACE ? swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) : isSPACE_LC_utf8((U8*)locinput))) @@ -2702,7 +3046,7 @@ S_regmatch(pTHX_ regnode *prog) if (!nextchr && locinput >= PL_regeol) sayNO; if (do_utf8) { - LOAD_UTF8_CHARCLASS(space," "); + LOAD_UTF8_CHARCLASS_SPACE(); if (OP(scan) == NSPACE ? swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) : isSPACE_LC_utf8((U8*)locinput)) @@ -2725,7 +3069,7 @@ S_regmatch(pTHX_ regnode *prog) if (!nextchr) sayNO; if (do_utf8) { - LOAD_UTF8_CHARCLASS(digit,"0"); + LOAD_UTF8_CHARCLASS_DIGIT(); if (!(OP(scan) == DIGIT ? swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) : isDIGIT_LC_utf8((U8*)locinput))) @@ -2748,7 +3092,7 @@ S_regmatch(pTHX_ regnode *prog) if (!nextchr && locinput >= PL_regeol) sayNO; if (do_utf8) { - LOAD_UTF8_CHARCLASS(digit,"0"); + LOAD_UTF8_CHARCLASS_DIGIT(); if (OP(scan) == NDIGIT ? swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) : isDIGIT_LC_utf8((U8*)locinput)) @@ -2768,7 +3112,7 @@ S_regmatch(pTHX_ regnode *prog) if (locinput >= PL_regeol) sayNO; if (do_utf8) { - LOAD_UTF8_CHARCLASS(mark,"~"); + LOAD_UTF8_CHARCLASS_MARK(); if (swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) sayNO; locinput += PL_utf8skip[nextchr]; @@ -2798,17 +3142,18 @@ S_regmatch(pTHX_ regnode *prog) s = PL_bostr + ln; if (do_utf8 && OP(scan) != REF) { /* REF can do byte comparison */ char *l = locinput; - char *e = PL_bostr + PL_regendp[n]; + const char *e = PL_bostr + PL_regendp[n]; /* * Note that we can't do the "other character" lookup trick as * in the 8-bit case (no pun intended) because in Unicode we * have to map both upper and title case to lower case. */ if (OP(scan) == REFF) { - STRLEN ulen1, ulen2; - U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; - U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; while (s < e) { + STRLEN ulen1, ulen2; + U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; + U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; + if (l >= PL_regeol) sayNO; toLOWER_utf8((U8*)s, tmpbuf1, &ulen1); @@ -2859,7 +3204,7 @@ S_regmatch(pTHX_ regnode *prog) n = ARG(scan); PL_op = (OP_4tree*)PL_regdata->data[n]; - DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) ); + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) ); PAD_SAVE_LOCAL(old_comppad, (PAD*)PL_regdata->data[n + 2]); PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr; @@ -2902,15 +3247,15 @@ S_regmatch(pTHX_ regnode *prog) } else { STRLEN len; - char *t = SvPV(ret, len); + const char *t = SvPV_const(ret, len); PMOP pm; - char *oprecomp = PL_regprecomp; - I32 osize = PL_regsize; - I32 onpar = PL_regnpar; + char * const oprecomp = PL_regprecomp; + const I32 osize = PL_regsize; + const I32 onpar = PL_regnpar; Zero(&pm, 1, PMOP); if (DO_UTF8(ret)) pm.op_pmdynflags |= PMdf_DYN_UTF8; - re = CALLREGCOMP(aTHX_ t, t + len, &pm); + re = CALLREGCOMP(aTHX_ (char*)t, (char*)t + len, &pm); if (!(SvFLAGS(ret) & (SVs_TEMP | SVs_PADTMP | SVf_READONLY | SVs_GMG))) @@ -2920,9 +3265,9 @@ S_regmatch(pTHX_ regnode *prog) PL_regsize = osize; PL_regnpar = onpar; } - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, - "Entering embedded `%s%.60s%s%s'\n", + "Entering embedded \"%s%.60s%s%s\"\n", PL_colors[0], re->precomp, PL_colors[1], @@ -3142,11 +3487,12 @@ S_regmatch(pTHX_ regnode *prog) CHECKPOINT cp, lastcp; CURCUR* cc = PL_regcc; char *lastloc = cc->lastloc; /* Detection of 0-len. */ + I32 cache_offset = 0, cache_bit = 0; n = cc->cur + 1; /* how many we know we matched */ PL_reginput = locinput; - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s %ld out of %ld..%ld cc=%"UVxf"\n", REPORT_CODE_OFF+PL_regindent*2, "", @@ -3160,7 +3506,7 @@ S_regmatch(pTHX_ regnode *prog) PL_regcc = cc->oldcc; if (PL_regcc) ln = PL_regcc->cur; - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s empty match detected, try continuation...\n", REPORT_CODE_OFF+PL_regindent*2, "") @@ -3194,7 +3540,7 @@ S_regmatch(pTHX_ regnode *prog) PL_reg_leftiter = PL_reg_maxiter; } if (PL_reg_leftiter-- == 0) { - I32 size = (PL_reg_maxiter + 7)/8; + const I32 size = (PL_reg_maxiter + 7 + POSCACHE_START)/8; if (PL_reg_poscache) { if ((I32)PL_reg_poscache_size < size) { Renew(PL_reg_poscache, size, char); @@ -3204,32 +3550,35 @@ S_regmatch(pTHX_ regnode *prog) } else { PL_reg_poscache_size = size; - Newz(29, PL_reg_poscache, size, char); + Newxz(PL_reg_poscache, size, char); } - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%sDetected a super-linear match, switching on caching%s...\n", PL_colors[4], PL_colors[5]) ); } if (PL_reg_leftiter < 0) { - I32 o = locinput - PL_bostr, b; - - o = (scan->flags & 0xf) - 1 + o * (scan->flags>>4); - b = o % 8; - o /= 8; - if (PL_reg_poscache[o] & (1<flags & 0xf) - 1 + POSCACHE_START + + cache_offset * (scan->flags>>4); + cache_bit = cache_offset % 8; + cache_offset /= 8; + if (PL_reg_poscache[cache_offset] & (1<next)) { regcpblow(cp); - sayYES; /* All done. */ + CACHEsayYES; /* All done. */ } REGCP_UNWIND(lastcp); regcppop(); @@ -3259,10 +3608,10 @@ S_regmatch(pTHX_ regnode *prog) "Complex regular subexpression recursion", REG_INFTY - 1); } - sayNO; + CACHEsayNO; } - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s trying longer...\n", REPORT_CODE_OFF+PL_regindent*2, "") @@ -3275,13 +3624,13 @@ S_regmatch(pTHX_ regnode *prog) REGCP_SET(lastcp); if (regmatch(cc->scan)) { regcpblow(cp); - sayYES; + CACHEsayYES; } REGCP_UNWIND(lastcp); regcppop(); cc->cur = n - 1; cc->lastloc = lastloc; - sayNO; + CACHEsayNO; } /* Prefer scan over next for maximal matching. */ @@ -3293,12 +3642,12 @@ S_regmatch(pTHX_ regnode *prog) REGCP_SET(lastcp); if (regmatch(cc->scan)) { regcpblow(cp); - sayYES; + CACHEsayYES; } REGCP_UNWIND(lastcp); regcppop(); /* Restore some previous $s? */ PL_reginput = locinput; - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s failed, try continuation...\n", REPORT_CODE_OFF+PL_regindent*2, "") @@ -3317,13 +3666,13 @@ S_regmatch(pTHX_ regnode *prog) if (PL_regcc) ln = PL_regcc->cur; if (regmatch(cc->next)) - sayYES; + CACHEsayYES; if (PL_regcc) PL_regcc->cur = ln; PL_regcc = cc; cc->cur = n - 1; cc->lastloc = lastloc; - sayNO; + CACHEsayNO; } /* NOT REACHED */ case BRANCHJ: @@ -3340,7 +3689,7 @@ S_regmatch(pTHX_ regnode *prog) if (OP(next) != c1) /* No choice. */ next = inner; /* Avoid recursion. */ else { - I32 lastparen = *PL_reglastparen; + const I32 lastparen = *PL_reglastparen; I32 unwind1; re_unwind_branch_t *uw; @@ -3452,7 +3801,7 @@ S_regmatch(pTHX_ regnode *prog) else { n = regrepeat_hard(scan, n, &l); locinput = PL_reginput; - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s matched %"IVdf" times, len=%"IVdf"...\n", (int)(REPORT_CODE_OFF+PL_regindent*2), "", @@ -3491,7 +3840,7 @@ S_regmatch(pTHX_ regnode *prog) UCHARAT(PL_reginput) == c1 || UCHARAT(PL_reginput) == c2) { - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s trying tail with n=%"IVdf"...\n", (int)(REPORT_CODE_OFF+PL_regindent*2), "", (IV)n) @@ -3587,16 +3936,13 @@ S_regmatch(pTHX_ regnode *prog) to_utf8_upper((U8*)s, tmpbuf2, &ulen2); c1 = utf8n_to_uvuni(tmpbuf1, UTF8_MAXBYTES, 0, - ckWARN(WARN_UTF8) ? - 0 : UTF8_ALLOW_ANY); + uniflags); c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXBYTES, 0, - ckWARN(WARN_UTF8) ? - 0 : UTF8_ALLOW_ANY); + uniflags); } else { c2 = c1 = utf8n_to_uvchr(s, UTF8_MAXBYTES, 0, - ckWARN(WARN_UTF8) ? - 0 : UTF8_ALLOW_ANY); + uniflags); } } } @@ -3650,26 +3996,25 @@ S_regmatch(pTHX_ regnode *prog) count = locinput - old; } else { - STRLEN len; if (c1 == c2) { + STRLEN len; /* count initialised to * utf8_distance(old, locinput) */ while (locinput <= e && utf8n_to_uvchr((U8*)locinput, UTF8_MAXBYTES, &len, - ckWARN(WARN_UTF8) ? - 0 : UTF8_ALLOW_ANY) != (UV)c1) { + uniflags) != (UV)c1) { locinput += len; count++; } } else { + STRLEN len; /* count initialised to * utf8_distance(old, locinput) */ while (locinput <= e) { UV c = utf8n_to_uvchr((U8*)locinput, UTF8_MAXBYTES, &len, - ckWARN(WARN_UTF8) ? - 0 : UTF8_ALLOW_ANY); + uniflags); if (c == (UV)c1 || c == (UV)c2) break; locinput += len; @@ -3705,8 +4050,7 @@ S_regmatch(pTHX_ regnode *prog) if (do_utf8) c = utf8n_to_uvchr((U8*)PL_reginput, UTF8_MAXBYTES, 0, - ckWARN(WARN_UTF8) ? - 0 : UTF8_ALLOW_ANY); + uniflags); else c = UCHARAT(PL_reginput); /* If it could work, try it. */ @@ -3755,8 +4099,7 @@ S_regmatch(pTHX_ regnode *prog) if (do_utf8) c = utf8n_to_uvchr((U8*)PL_reginput, UTF8_MAXBYTES, 0, - ckWARN(WARN_UTF8) ? - 0 : UTF8_ALLOW_ANY); + uniflags); else c = UCHARAT(PL_reginput); } @@ -3778,8 +4121,7 @@ S_regmatch(pTHX_ regnode *prog) if (do_utf8) c = utf8n_to_uvchr((U8*)PL_reginput, UTF8_MAXBYTES, 0, - ckWARN(WARN_UTF8) ? - 0 : UTF8_ALLOW_ANY); + uniflags); else c = UCHARAT(PL_reginput); } @@ -3825,7 +4167,7 @@ S_regmatch(pTHX_ regnode *prog) PL_reg_re = re; cache_re(re); - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s continuation failed...\n", REPORT_CODE_OFF+PL_regindent*2, "") @@ -3833,7 +4175,7 @@ S_regmatch(pTHX_ regnode *prog) sayNO_SILENT; } if (locinput < PL_regtill) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n", PL_colors[4], (long)(locinput - PL_reg_starttry), @@ -3860,7 +4202,6 @@ S_regmatch(pTHX_ regnode *prog) } else PL_reginput = locinput; - PL_reg_flags ^= RF_false; goto do_ifmatch; case IFMATCH: n = 1; @@ -3876,8 +4217,6 @@ S_regmatch(pTHX_ regnode *prog) do_ifmatch: inner = NEXTOPER(NEXTOPER(scan)); if (regmatch(inner) != n) { - if (n == 0) - PL_reg_flags ^= RF_false; say_no: if (logical) { logical = 0; @@ -3887,8 +4226,6 @@ S_regmatch(pTHX_ regnode *prog) else sayNO; } - if (n == 0) - PL_reg_flags ^= RF_false; say_yes: if (logical) { logical = 0; @@ -3923,15 +4260,15 @@ S_regmatch(pTHX_ regnode *prog) sayNO; yes_loud: - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s %scould match...%s\n", - REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4],PL_colors[5]) + REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4], PL_colors[5]) ); goto yes; yes_final: - DEBUG_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n", - PL_colors[4],PL_colors[5])); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n", + PL_colors[4], PL_colors[5])); yes: #ifdef DEBUGGING PL_regindent--; @@ -3944,10 +4281,10 @@ yes: return 1; no: - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s %sfailed...%s\n", - REPORT_CODE_OFF+PL_regindent*2, "",PL_colors[4],PL_colors[5]) + REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4], PL_colors[5]) ); goto do_no; no_final: @@ -3960,7 +4297,7 @@ do_no: case RE_UNWIND_BRANCHJ: { re_unwind_branch_t *uwb = &(uw->branch); - I32 lastparen = uwb->lastparen; + const I32 lastparen = uwb->lastparen; REGCP_UNWIND(uwb->lastcp); for (n = *PL_reglastparen; n > lastparen; n--) @@ -3977,7 +4314,6 @@ do_no: goto do_no; } /* Have more choice yet. Reuse the same uwb. */ - /*SUPPRESS 560*/ if ((n = (uwb->type == RE_UNWIND_BRANCH ? NEXT_OFF(next) : ARG(next)))) next += n; @@ -4016,8 +4352,9 @@ do_no: * rather than incrementing count on every character. [Er, except utf8.]] */ STATIC I32 -S_regrepeat(pTHX_ regnode *p, I32 max) +S_regrepeat(pTHX_ const regnode *p, I32 max) { + dVAR; register char *scan; register I32 c; register char *loceol = PL_regeol; @@ -4090,7 +4427,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) case ALNUM: if (do_utf8) { loceol = PL_regeol; - LOAD_UTF8_CHARCLASS(alnum,"a"); + LOAD_UTF8_CHARCLASS_ALNUM(); while (hardcount < max && scan < loceol && swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { scan += UTF8SKIP(scan); @@ -4118,7 +4455,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) case NALNUM: if (do_utf8) { loceol = PL_regeol; - LOAD_UTF8_CHARCLASS(alnum,"a"); + LOAD_UTF8_CHARCLASS_ALNUM(); while (hardcount < max && scan < loceol && !swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { scan += UTF8SKIP(scan); @@ -4146,7 +4483,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) case SPACE: if (do_utf8) { loceol = PL_regeol; - LOAD_UTF8_CHARCLASS(space," "); + LOAD_UTF8_CHARCLASS_SPACE(); while (hardcount < max && scan < loceol && (*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { @@ -4175,7 +4512,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) case NSPACE: if (do_utf8) { loceol = PL_regeol; - LOAD_UTF8_CHARCLASS(space," "); + LOAD_UTF8_CHARCLASS_SPACE(); while (hardcount < max && scan < loceol && !(*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { @@ -4204,7 +4541,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) case DIGIT: if (do_utf8) { loceol = PL_regeol; - LOAD_UTF8_CHARCLASS(digit,"0"); + LOAD_UTF8_CHARCLASS_DIGIT(); while (hardcount < max && scan < loceol && swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { scan += UTF8SKIP(scan); @@ -4218,7 +4555,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) case NDIGIT: if (do_utf8) { loceol = PL_regeol; - LOAD_UTF8_CHARCLASS(digit,"0"); + LOAD_UTF8_CHARCLASS_DIGIT(); while (hardcount < max && scan < loceol && !swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { scan += UTF8SKIP(scan); @@ -4239,14 +4576,16 @@ S_regrepeat(pTHX_ regnode *p, I32 max) c = scan - PL_reginput; PL_reginput = scan; - DEBUG_r( - { + DEBUG_r({ + SV *re_debug_flags = NULL; SV *prop = sv_newmortal(); - + GET_RE_DEBUG_FLAGS; + DEBUG_EXECUTE_r({ regprop(prop, p); PerlIO_printf(Perl_debug_log, "%*s %s can match %"IVdf" times out of %"IVdf"...\n", - REPORT_CODE_OFF+1, "", SvPVX(prop),(IV)c,(IV)max); + REPORT_CODE_OFF+1, "", SvPVX_const(prop),(IV)c,(IV)max); + }); }); return(c); @@ -4309,19 +4648,19 @@ S_regrepeat_hard(pTHX_ regnode *p, I32 max, I32 *lp) */ SV * -Perl_regclass_swash(pTHX_ register regnode* node, bool doinit, SV** listsvp, SV **altsvp) +Perl_regclass_swash(pTHX_ register const regnode* node, bool doinit, SV** listsvp, SV **altsvp) { SV *sw = NULL; SV *si = NULL; SV *alt = NULL; if (PL_regdata && PL_regdata->count) { - U32 n = ARG(node); + const U32 n = ARG(node); if (PL_regdata->what[n] == 's') { - SV *rv = (SV*)PL_regdata->data[n]; - AV *av = (AV*)SvRV((SV*)rv); - SV **ary = AvARRAY(av); + SV * const rv = (SV*)PL_regdata->data[n]; + AV * const av = (AV*)SvRV((SV*)rv); + SV **const ary = AvARRAY(av); SV **a, **b; /* See the end of regcomp.c:S_reglass() for @@ -4361,9 +4700,10 @@ Perl_regclass_swash(pTHX_ register regnode* node, bool doinit, SV** listsvp, SV */ STATIC bool -S_reginclass(pTHX_ register regnode *n, register U8* p, STRLEN* lenp, register bool do_utf8) +S_reginclass(pTHX_ register const regnode *n, register const U8* p, STRLEN* lenp, register bool do_utf8) { - char flags = ANYOF_FLAGS(n); + dVAR; + const char flags = ANYOF_FLAGS(n); bool match = FALSE; UV c = *p; STRLEN len = 0; @@ -4385,7 +4725,7 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, STRLEN* lenp, register b match = TRUE; if (!match) { AV *av; - SV *sw = regclass_swash(n, TRUE, 0, (SV**)&av); + SV * const sw = regclass_swash(n, TRUE, 0, (SV**)&av); if (sw) { if (swash_fetch(sw, p, do_utf8)) @@ -4393,11 +4733,10 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, STRLEN* lenp, register b else if (flags & ANYOF_FOLD) { if (!match && lenp && av) { I32 i; - for (i = 0; i <= av_len(av); i++) { - SV* sv = *av_fetch(av, i, FALSE); + SV* const sv = *av_fetch(av, i, FALSE); STRLEN len; - char *s = SvPV(sv, len); + const char * const s = SvPV_const(sv, len); if (len <= plen && memEQ(s, (char*)p, len)) { *lenp = len; @@ -4548,11 +4887,12 @@ S_reghopmaybe3(pTHX_ U8* s, I32 off, U8* lim) static void restore_pos(pTHX_ void *arg) { + PERL_UNUSED_ARG(arg); if (PL_reg_eval_set) { if (PL_reg_oldsaved) { PL_reg_re->subbeg = PL_reg_oldsaved; PL_reg_re->sublen = PL_reg_oldsavedlen; -#ifdef PERL_COPY_ON_WRITE +#ifdef PERL_OLD_COPY_ON_WRITE PL_reg_re->saved_copy = PL_nrs; #endif RX_MATCH_COPIED_on(PL_reg_re); @@ -4566,8 +4906,8 @@ restore_pos(pTHX_ void *arg) STATIC void S_to_utf8_substr(pTHX_ register regexp *prog) { - SV* sv; if (prog->float_substr && !prog->float_utf8) { + SV* sv; prog->float_utf8 = sv = newSVsv(prog->float_substr); sv_utf8_upgrade(sv); if (SvTAIL(prog->float_substr)) @@ -4576,6 +4916,7 @@ S_to_utf8_substr(pTHX_ register regexp *prog) prog->check_utf8 = sv; } if (prog->anchored_substr && !prog->anchored_utf8) { + SV* sv; prog->anchored_utf8 = sv = newSVsv(prog->anchored_substr); sv_utf8_upgrade(sv); if (SvTAIL(prog->anchored_substr)) @@ -4588,8 +4929,8 @@ S_to_utf8_substr(pTHX_ register regexp *prog) STATIC void S_to_byte_substr(pTHX_ register regexp *prog) { - SV* sv; if (prog->float_utf8 && !prog->float_substr) { + SV* sv; prog->float_substr = sv = newSVsv(prog->float_utf8); if (sv_utf8_downgrade(sv, TRUE)) { if (SvTAIL(prog->float_utf8)) @@ -4602,6 +4943,7 @@ S_to_byte_substr(pTHX_ register regexp *prog) prog->check_substr = sv; } if (prog->anchored_utf8 && !prog->anchored_substr) { + SV* sv; prog->anchored_substr = sv = newSVsv(prog->anchored_utf8); if (sv_utf8_downgrade(sv, TRUE)) { if (SvTAIL(prog->anchored_utf8)) @@ -4614,3 +4956,13 @@ S_to_byte_substr(pTHX_ register regexp *prog) prog->check_substr = sv; } } + +/* + * Local variables: + * c-indentation-style: bsd + * c-basic-offset: 4 + * indent-tabs-mode: t + * End: + * + * ex: set ts=8 sts=4 sw=4 noet: + */