X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regexec.c;h=625f8eb4dde684602860d53150fa64b03cac4a31;hb=7d3fb23018f73b213481a8b6b108e1dc03cefcff;hp=bdbdb5918c4fdb8310e1fe2b0a79f6d3ce568730;hpb=7ea3cd407b6ec2a3e424bdfbc486b6e01d6d28bd;p=p5sagit%2Fp5-mst-13.2.git diff --git a/regexec.c b/regexec.c index bdbdb59..625f8eb 100644 --- a/regexec.c +++ b/regexec.c @@ -38,11 +38,11 @@ # define Perl_re_intuit_start my_re_intuit_start /* *These* symbols are masked to allow static link. */ # define Perl_pregexec my_pregexec -# define Perl_reginitcolors my_reginitcolors +# define Perl_reginitcolors my_reginitcolors # define Perl_regclass_swash my_regclass_swash # define PERL_NO_GET_CONTEXT -#endif +#endif /*SUPPRESS 112*/ /* @@ -67,7 +67,7 @@ * **** Alterations to Henry's code are... **** - **** Copyright (c) 1991-2000, Larry Wall + **** Copyright (c) 1991-2001, Larry Wall **** **** You may distribute under the terms of either the GNU General Public **** License or the Artistic License, as specified in the README file. @@ -107,37 +107,53 @@ */ #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv)) -#define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b) +#define CHR_DIST(a,b) (DO_UTF8(PL_reg_sv) ? utf8_distance(a,b) : a - b) #define reghop_c(pos,off) ((char*)reghop((U8*)pos, off)) #define reghopmaybe_c(pos,off) ((char*)reghopmaybe((U8*)pos, off)) -#define HOP(pos,off) (UTF ? reghop((U8*)pos, off) : (U8*)(pos + off)) -#define HOPMAYBE(pos,off) (UTF ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off)) +#define HOP(pos,off) (DO_UTF8(PL_reg_sv) ? reghop((U8*)pos, off) : (U8*)(pos + off)) +#define HOPMAYBE(pos,off) (DO_UTF8(PL_reg_sv) ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off)) #define HOPc(pos,off) ((char*)HOP(pos,off)) #define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off)) -static void restore_pos(pTHXo_ void *arg); +#define reghop3_c(pos,off,lim) ((char*)reghop3((U8*)pos, off, (U8*)lim)) +#define reghopmaybe3_c(pos,off,lim) ((char*)reghopmaybe3((U8*)pos, off, (U8*)lim)) +#define HOP3(pos,off,lim) (DO_UTF8(PL_reg_sv) ? reghop3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off)) +#define HOPMAYBE3(pos,off,lim) (DO_UTF8(PL_reg_sv) ? reghopmaybe3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off)) +#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)) (void)CAT2(is_utf8_, a)((U8*)b); } STMT_END +static void restore_pos(pTHXo_ void *arg); STATIC CHECKPOINT S_regcppush(pTHX_ I32 parenfloor) { int retval = PL_savestack_ix; - int i = (PL_regsize - parenfloor) * 4; +#define REGCP_PAREN_ELEMS 4 + int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS; int p; - SSCHECK(i + 5); +#define REGCP_OTHER_ELEMS 5 + SSCHECK(paren_elems_to_push + REGCP_OTHER_ELEMS); for (p = PL_regsize; p > parenfloor; p--) { +/* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */ SSPUSHINT(PL_regendp[p]); SSPUSHINT(PL_regstartp[p]); SSPUSHPTR(PL_reg_start_tmp[p]); SSPUSHINT(p); } +/* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */ SSPUSHINT(PL_regsize); SSPUSHINT(*PL_reglastparen); SSPUSHPTR(PL_reginput); - SSPUSHINT(i + 3); - SSPUSHINT(SAVEt_REGCONTEXT); +#define REGCP_FRAME_ELEMS 2 +/* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and + * are needed for the regexp context stack bookkeeping. */ + SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); + SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */ + return retval; } @@ -154,16 +170,22 @@ S_regcppush(pTHX_ I32 parenfloor) STATIC char * S_regcppop(pTHX) { - I32 i = SSPOPINT; + I32 i; U32 paren = 0; char *input; I32 tmps; - assert(i == SAVEt_REGCONTEXT); + + /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */ i = SSPOPINT; + assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */ + i = SSPOPINT; /* Parentheses elements to pop. */ input = (char *) SSPOPPTR; *PL_reglastparen = SSPOPINT; PL_regsize = SSPOPINT; - for (i -= 3; i > 0; i -= 4) { + + /* Now restore the parentheses context. */ + for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); + i > 0; i -= REGCP_PAREN_ELEMS) { paren = (U32)SSPOPINT; PL_reg_start_tmp[paren] = (char *) SSPOPPTR; PL_regstartp[paren] = SSPOPINT; @@ -173,9 +195,9 @@ S_regcppop(pTHX) DEBUG_r( PerlIO_printf(Perl_debug_log, " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n", - (UV)paren, (IV)PL_regstartp[paren], + (UV)paren, (IV)PL_regstartp[paren], (IV)(PL_reg_start_tmp[paren] - PL_bostr), - (IV)PL_regendp[paren], + (IV)PL_regendp[paren], (paren > *PL_reglastparen ? "(no)" : "")); ); } @@ -260,7 +282,7 @@ Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *stren /* nosave: For optimizations. */ { return - regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, + regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, nosave ? 0 : REXEC_COPY_STR); } @@ -272,11 +294,11 @@ S_cache_re(pTHX_ regexp *prog) PL_regprogram = prog->program; #endif PL_regnpar = prog->nparens; - PL_regdata = prog->data; - PL_reg_re = prog; + PL_regdata = prog->data; + PL_reg_re = prog; } -/* +/* * Need to implement the following flags for reg_anch: * * USE_INTUIT_NOML - Useful to call re_intuit_start() first @@ -354,11 +376,15 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, (strend - strpos > 60 ? "..." : "")) ); - if (prog->minlen > strend - strpos) { + if (prog->reganch & ROPT_UTF8) + PL_reg_flags |= RF_utf8; + + if (prog->minlen > CHR_DIST((U8*)strend, (U8*)strpos)) { DEBUG_r(PerlIO_printf(Perl_debug_log, "String too short...\n")); goto fail; } strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos; + PL_regeol = strend; check = prog->check_substr; if (prog->reganch & ROPT_ANCH) { /* Match at beg-of-str or after \n */ ml_anch = !( (prog->reganch & ROPT_ANCH_SINGLE) @@ -373,16 +399,16 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, DEBUG_r(PerlIO_printf(Perl_debug_log, "Not at start...\n")); goto fail; } - if (prog->check_offset_min == prog->check_offset_max) { + if (prog->check_offset_min == prog->check_offset_max && + !(prog->reganch & ROPT_SANY_SEEN)) { /* Substring at constant offset from beg-of-str... */ I32 slen; - PL_regeol = strend; /* Used in HOP() */ - s = HOPc(strpos, prog->check_offset_min); + s = HOP3c(strpos, prog->check_offset_min, strend); if (SvTAIL(check)) { slen = SvCUR(check); /* >= 1 */ - if ( strend - s > slen || strend - s < slen - 1 + 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")); goto fail_finish; @@ -412,7 +438,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, if (!ml_anch) { I32 end = prog->check_offset_max + CHR_SVLEN(check) - (SvTAIL(check) != 0); - I32 eshift = strend - s - end; + I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end; if (end_shift < eshift) end_shift = eshift; @@ -443,16 +469,20 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, || ( BmRARE(check) == '\n' && (BmPREVIOUS(check) == SvCUR(check) - 1) && SvTAIL(check) )) - s = screaminstr(sv, check, + s = screaminstr(sv, check, start_shift + (s - strbeg), end_shift, pp, 0); else goto fail_finish; if (data) *data->scream_olds = s; } + else if (prog->reganch & ROPT_SANY_SEEN) + s = fbm_instr((U8*)(s + start_shift), + (U8*)(strend - end_shift), + check, PL_multiline ? FBMrf_MULTILINE : 0); else - s = fbm_instr((unsigned char*)s + start_shift, - (unsigned char*)strend - end_shift, + s = fbm_instr(HOP3(s, start_shift, strend), + HOP3(strend, -end_shift, strbeg), check, PL_multiline ? FBMrf_MULTILINE : 0); /* Update the count-of-usability, remove useless subpatterns, @@ -491,34 +521,33 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, if (check == prog->float_substr) { do_other_anchored: { - char *last = s - start_shift, *last1, *last2; + char *last = HOP3c(s, -start_shift, strbeg), *last1, *last2; char *s1 = s; - tmp = PL_bostr; t = s - prog->check_offset_max; if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ && (!(prog->reganch & ROPT_UTF8) - || (PL_bostr = strpos, /* Used in regcopmaybe() */ - (t = reghopmaybe_c(s, -(prog->check_offset_max))) + || ((t = reghopmaybe3_c(s, -(prog->check_offset_max), strpos)) && t > strpos))) /* EMPTY */; else t = strpos; - t += prog->anchored_offset; + t = HOP3c(t, prog->anchored_offset, strend); if (t < other_last) /* These positions already checked */ t = other_last; - PL_bostr = tmp; - last2 = last1 = strend - prog->minlen; + last2 = last1 = HOP3c(strend, -prog->minlen, strbeg); if (last < last1) last1 = last; /* XXXX It is not documented what units *_offsets are in. Assume bytes. */ /* On end-of-str: see comment below. */ s = fbm_instr((unsigned char*)t, - (unsigned char*)last1 + prog->anchored_offset - + SvCUR(prog->anchored_substr) - - (SvTAIL(prog->anchored_substr)!=0), - prog->anchored_substr, PL_multiline ? FBMrf_MULTILINE : 0); - DEBUG_r(PerlIO_printf(Perl_debug_log, "%s anchored substr `%s%.*s%s'%s", + HOP3(HOP3(last1, prog->anchored_offset, strend) + + SvCUR(prog->anchored_substr), + -(SvTAIL(prog->anchored_substr)!=0), strbeg), + prog->anchored_substr, + PL_multiline ? FBMrf_MULTILINE : 0); + DEBUG_r(PerlIO_printf(Perl_debug_log, + "%s anchored substr `%s%.*s%s'%s", (s ? "Found" : "Contradicts"), PL_colors[0], (int)(SvCUR(prog->anchored_substr) @@ -533,17 +562,16 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, } DEBUG_r(PerlIO_printf(Perl_debug_log, ", trying floating at offset %ld...\n", - (long)(s1 + 1 - i_strpos))); - PL_regeol = strend; /* Used in HOP() */ - other_last = last1 + prog->anchored_offset + 1; - s = HOPc(last, 1); + (long)(HOP3c(s1, 1, strend) - i_strpos))); + other_last = HOP3c(last1, prog->anchored_offset+1, strend); + s = HOP3c(last, 1, strend); goto restart; } else { DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", (long)(s - i_strpos))); - t = s - prog->anchored_offset; - other_last = s + 1; + t = HOP3c(s, -prog->anchored_offset, strbeg); + other_last = HOP3c(s, 1, strend); s = s1; if (t == strpos) goto try_at_start; @@ -555,11 +583,12 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, char *last, *last1; char *s1 = s; - t = s - start_shift; - last1 = last = strend - prog->minlen + prog->float_min_offset; - if (last - t > prog->float_max_offset) - last = t + prog->float_max_offset; - s = t + prog->float_min_offset; + t = HOP3c(s, -start_shift, strbeg); + last1 = last = + HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg); + if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset) + last = HOP3c(t, prog->float_max_offset, strend); + s = HOP3c(t, prog->float_min_offset, strend); if (s < other_last) s = other_last; /* XXXX It is not documented what units *_offsets are in. Assume bytes. */ @@ -586,15 +615,14 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, DEBUG_r(PerlIO_printf(Perl_debug_log, ", trying anchored starting at offset %ld...\n", (long)(s1 + 1 - i_strpos))); - other_last = last + 1; - PL_regeol = strend; /* Used in HOP() */ - s = HOPc(t, 1); + other_last = last; + s = HOP3c(t, 1, strend); goto restart; } else { DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", (long)(s - i_strpos))); - other_last = s + 1; + other_last = s; /* Fix this later. --Hugo */ s = s1; if (t == strpos) goto try_at_start; @@ -604,13 +632,10 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, } t = s - prog->check_offset_max; - tmp = PL_bostr; if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ && (!(prog->reganch & ROPT_UTF8) - || (PL_bostr = strpos, /* Used in regcopmaybe() */ - ((t = reghopmaybe_c(s, -(prog->check_offset_max))) - && t > strpos)))) { - PL_bostr = tmp; + || ((t = reghopmaybe3_c(s, -prog->check_offset_max, strpos)) + && t > strpos))) { /* Fixed substring is found far enough so that the match cannot start at strpos. */ try_at_offset: @@ -633,7 +658,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, Thus we can arrive here only if check substr is float. Redo checking for "other"=="fixed". */ - strpos = t + 1; + 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))); goto do_other_anchored; @@ -668,7 +693,6 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, ++BmUSEFUL(prog->check_substr); /* hooray/5 */ } else { - PL_bostr = tmp; /* The found string does not prohibit matching at strpos, - no optimization of calling REx engine can be performed, unless it was an MBOL and we are not after MBOL, @@ -721,13 +745,16 @@ 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 - ? STR_LEN(prog->regstclass) + ? CHR_DIST(str+STR_LEN(prog->regstclass), str) : 1); char *endpos = (prog->anchored_substr || ml_anch) - ? s + (prog->minlen? cl_l : 0) - : (prog->float_substr ? check_at - start_shift + cl_l - : strend) ; + ? HOP3c(s, (prog->minlen ? cl_l : 0), strend) + : (prog->float_substr + ? HOP3c(HOP3c(check_at, -start_shift, strbeg), + cl_l, strend) + : strend); char *startpos = strbeg; t = s; @@ -754,8 +781,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, if (prog->anchored_substr == check) { DEBUG_r( what = "anchored" ); hop_and_restart: - PL_regeol = strend; /* Used in HOP() */ - s = HOPc(t, 1); + 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, @@ -773,7 +799,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, if (t + start_shift >= check_at) /* Contradicts floating=check */ goto retry_floating_check; /* Recheck anchored substring, but not floating... */ - s = check_at; + s = check_at; if (!check) goto giveup; DEBUG_r( PerlIO_printf(Perl_debug_log, @@ -801,11 +827,11 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, goto hop_and_restart; } DEBUG_r( if (t != s) - PerlIO_printf(Perl_debug_log, + PerlIO_printf(Perl_debug_log, "By STCLASS: moving %ld --> %ld\n", (long)(t - i_strpos), (long)(s - i_strpos)); else - PerlIO_printf(Perl_debug_log, + PerlIO_printf(Perl_debug_log, "Does not contradict STCLASS...\n") ); } giveup: @@ -854,8 +880,14 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta case EXACTF: m = STRING(c); ln = STR_LEN(c); - c1 = *(U8*)m; - c2 = PL_fold[c1]; + if (UTF) { + c1 = to_utf8_lower((U8*)m); + c2 = to_utf8_upper((U8*)m); + } + else { + c1 = *(U8*)m; + c2 = PL_fold[c1]; + } goto do_exactf; case EXACTFL: m = STRING(c); @@ -867,27 +899,45 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if (norun && e < s) e = s; /* Due to minlen logic of intuit() */ - /* Here it is NOT UTF! */ - if (c1 == c2) { - while (s <= e) { - if ( *(U8*)s == c1 - && (ln == 1 || !(OP(c) == EXACTF - ? ibcmp(s, m, ln) - : ibcmp_locale(s, m, ln))) - && (norun || regtry(prog, s)) ) - goto got_it; - s++; - } - } else { - while (s <= e) { - if ( (*(U8*)s == c1 || *(U8*)s == c2) - && (ln == 1 || !(OP(c) == EXACTF - ? ibcmp(s, m, ln) - : ibcmp_locale(s, m, ln))) - && (norun || regtry(prog, s)) ) - goto got_it; - s++; - } + + if (do_utf8) { + STRLEN len; + if (c1 == c2) + while (s <= e) { + if ( utf8_to_uvchr((U8*)s, &len) == c1 + && regtry(prog, s) ) + goto got_it; + s += len; + } + else + while (s <= e) { + UV c = utf8_to_uvchr((U8*)s, &len); + if ( (c == c1 || c == c2) && regtry(prog, s) ) + goto got_it; + s += len; + } + } + else { + if (c1 == c2) + while (s <= e) { + if ( *(U8*)s == c1 + && (ln == 1 || !(OP(c) == EXACTF + ? ibcmp(s, m, ln) + : ibcmp_locale(s, m, ln))) + && (norun || regtry(prog, s)) ) + goto got_it; + s++; + } + else + while (s <= e) { + if ( (*(U8*)s == c1 || *(U8*)s == c2) + && (ln == 1 || !(OP(c) == EXACTF + ? ibcmp(s, m, ln) + : ibcmp_locale(s, m, ln))) + && (norun || regtry(prog, s)) ) + goto got_it; + s++; + } } break; case BOUNDL: @@ -898,12 +948,13 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if (s == startpos) tmp = '\n'; else { - U8 *r = reghop((U8*)s, -1); - - tmp = (I32)utf8_to_uv(r, s - (char*)r, 0, 0); + U8 *r = reghop3((U8*)s, -1, (U8*)startpos); + + tmp = (I32)utf8n_to_uvchr(r, s - (char*)r, 0, 0); } tmp = ((OP(c) == BOUND ? - isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0); + isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); + LOAD_UTF8_CHARCLASS(alnum,"a"); while (s < strend) { if (tmp == !(OP(c) == BOUND ? swash_fetch(PL_utf8_alnum, (U8*)s) : @@ -940,12 +991,13 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if (s == startpos) tmp = '\n'; else { - U8 *r = reghop((U8*)s, -1); - - tmp = (I32)utf8_to_uv(r, s - (char*)r, 0, 0); + U8 *r = reghop3((U8*)s, -1, (U8*)startpos); + + tmp = (I32)utf8n_to_uvchr(r, s - (char*)r, 0, 0); } tmp = ((OP(c) == NBOUND ? - isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0); + isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); + LOAD_UTF8_CHARCLASS(alnum,"a"); while (s < strend) { if (tmp == !(OP(c) == NBOUND ? swash_fetch(PL_utf8_alnum, (U8*)s) : @@ -974,6 +1026,7 @@ 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"); while (s < strend) { if (swash_fetch(PL_utf8_alnum, (U8*)s)) { if (tmp && (norun || regtry(prog, s))) @@ -1031,6 +1084,7 @@ 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"); while (s < strend) { if (!swash_fetch(PL_utf8_alnum, (U8*)s)) { if (tmp && (norun || regtry(prog, s))) @@ -1088,6 +1142,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta break; case SPACE: if (do_utf8) { + LOAD_UTF8_CHARCLASS(space," "); while (s < strend) { if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s)) { if (tmp && (norun || regtry(prog, s))) @@ -1145,6 +1200,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta break; case NSPACE: if (do_utf8) { + LOAD_UTF8_CHARCLASS(space," "); while (s < strend) { if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s))) { if (tmp && (norun || regtry(prog, s))) @@ -1202,6 +1258,7 @@ 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"); while (s < strend) { if (swash_fetch(PL_utf8_digit,(U8*)s)) { if (tmp && (norun || regtry(prog, s))) @@ -1259,6 +1316,7 @@ 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"); while (s < strend) { if (!swash_fetch(PL_utf8_digit,(U8*)s)) { if (tmp && (norun || regtry(prog, s))) @@ -1346,12 +1404,13 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * I32 scream_pos = -1; /* Internal iterator of scream. */ char *scream_olds; SV* oreplsv = GvSV(PL_replgv); + bool do_utf8 = DO_UTF8(sv); PL_regcc = 0; cache_re(prog); #ifdef DEBUGGING - PL_regnarrate = PL_debug & 512; + PL_regnarrate = DEBUG_r_TEST; #endif /* Be paranoid... */ @@ -1361,12 +1420,23 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * } minlen = prog->minlen; - if (strend - startpos < minlen) goto phooey; + if (do_utf8) { + if (!(prog->reganch & ROPT_SANY_SEEN)) + if (utf8_distance((U8*)strend, (U8*)startpos) < minlen) goto phooey; + } + else { + if (strend - startpos < minlen) goto phooey; + } if (startpos == strbeg) /* is ^ valid at stringarg? */ PL_regprev = '\n'; else { - PL_regprev = (U32)stringarg[-1]; + if (prog->reganch & ROPT_UTF8 && do_utf8) { + U8 *s = reghop3((U8*)stringarg, -1, (U8*)strbeg); + PL_regprev = utf8n_to_uvchr(s, (U8*)stringarg - s, NULL, 0); + } + else + PL_regprev = (U32)stringarg[-1]; if (!PL_multiline && PL_regprev == '\n') PL_regprev = '\0'; /* force ^ to NOT match */ } @@ -1419,7 +1489,8 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * PL_reg_ganch = strbeg; } - if (!(flags & REXEC_CHECKED) && prog->check_substr != Nullsv) { + if (do_utf8 == (UTF!=0) && + !(flags & REXEC_CHECKED) && prog->check_substr != Nullsv) { re_scream_pos_data d; d.scream_olds = &scream_olds; @@ -1454,7 +1525,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * if (minlen) dontbother = minlen - 1; - end = HOPc(strend, -dontbother) - 1; + end = HOP3c(strend, -dontbother, strbeg) - 1; /* for multiline we only have to try after newlines */ if (prog->check_substr) { if (s == startpos) @@ -1492,7 +1563,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * } /* Messy cases: unanchored match. */ - if (prog->anchored_substr && prog->reganch & ROPT_SKIP) { + if (prog->anchored_substr && prog->reganch & ROPT_SKIP) { /* we have /x+whatever/ */ /* it must be a one character string (XXXX Except UTF?) */ char ch = SvPVX(prog->anchored_substr)[0]; @@ -1500,7 +1571,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * int did_match = 0; #endif - if (UTF) { + if (do_utf8) { while (s < strend) { if (*s == ch) { DEBUG_r( did_match = 1 ); @@ -1529,18 +1600,19 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * "Did not find anchored character...\n")); } /*SUPPRESS 560*/ - else if (prog->anchored_substr != Nullsv - || (prog->float_substr != Nullsv - && prog->float_max_offset < strend - s)) { - SV *must = prog->anchored_substr + else if (do_utf8 == (UTF!=0) && + (prog->anchored_substr != Nullsv + || (prog->float_substr != Nullsv + && prog->float_max_offset < strend - s))) { + SV *must = prog->anchored_substr ? prog->anchored_substr : prog->float_substr; - I32 back_max = + I32 back_max = prog->anchored_substr ? prog->anchored_offset : prog->float_max_offset; - I32 back_min = + I32 back_min = prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset; - char *last = HOPc(strend, /* Cannot start after this */ + char *last = HOP3c(strend, /* Cannot start after this */ -(I32)(CHR_SVLEN(must) - - (SvTAIL(must) != 0) + back_min)); + - (SvTAIL(must) != 0) + back_min), strbeg); char *last1; /* Last position checked before */ #ifdef DEBUGGING int did_match = 0; @@ -1557,11 +1629,11 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * dontbother = end_shift; strend = HOPc(strend, -dontbother); while ( (s <= last) && - ((flags & REXEC_SCREAM) - ? (s = screaminstr(sv, must, HOPc(s, back_min) - strbeg, + ((flags & REXEC_SCREAM) + ? (s = screaminstr(sv, must, HOP3c(s, back_min, strend) - strbeg, end_shift, &scream_pos, 0)) - : (s = fbm_instr((unsigned char*)HOP(s, back_min), - (unsigned char*)strend, must, + : (s = fbm_instr((unsigned char*)HOP3(s, back_min, strend), + (unsigned char*)strend, must, PL_multiline ? FBMrf_MULTILINE : 0))) ) { DEBUG_r( did_match = 1 ); if (HOPc(s, -back_max) > last1) { @@ -1574,7 +1646,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * last1 = HOPc(s, -back_min); s = t; } - if (UTF) { + if (do_utf8) { while (s <= last1) { if (regtry(prog, s)) goto got_it; @@ -1631,13 +1703,13 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * if (memEQ(strend - len + 1, little, len - 1)) last = strend - len + 1; else if (!PL_multiline) - last = memEQ(strend - len, little, len) + last = memEQ(strend - len, little, len) ? strend - len : Nullch; else goto find_last; } else { find_last: - if (len) + if (len) last = rninstr(s, strend, little, little + len); else last = strend; /* matching `$' */ @@ -1655,7 +1727,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * dontbother = minlen - 1; strend -= dontbother; /* this one's always in bytes! */ /* We don't know much -- general case. */ - if (UTF) { + if (do_utf8) { for (;;) { if (regtry(prog, s)) goto got_it; @@ -1706,7 +1778,7 @@ got_it: prog->sublen = PL_regeol - strbeg; /* strend may have been modified */ } } - + return 1; phooey: @@ -1755,7 +1827,7 @@ S_regtry(pTHX_ regexp *prog, char *startpos) DEFSV = PL_reg_sv; } - if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv) + if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv) && (mg = mg_find(PL_reg_sv, 'g')))) { /* prepare for quick setting of pos */ sv_magic(PL_reg_sv, (SV*)0, 'g', Nullch, 0); @@ -1926,31 +1998,36 @@ S_regmatch(pTHX_ regnode *prog) SV *prop = sv_newmortal(); int docolor = *PL_colors[0]; int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ - int l = (PL_regeol - locinput > taill ? taill : PL_regeol - locinput); + 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 position another one (pref_len - pref0_len chars), after the current position the third one. We assume that pref0_len <= pref_len, otherwise we decrease pref0_len. */ - int pref_len = (locinput - PL_bostr > (5 + taill) - l - ? (5 + taill) - l : locinput - PL_bostr); - int pref0_len = pref_len - (locinput - PL_reg_starttry); + int pref_len = (locinput - PL_bostr) > (5 + taill) - l + ? (5 + taill) - l : locinput - PL_bostr; + int pref0_len; + while (UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len))) + pref_len++; + pref0_len = pref_len - (locinput - PL_reg_starttry); if (l + pref_len < (5 + taill) && l < PL_regeol - locinput) - l = ( PL_regeol - locinput > (5 + taill) - pref_len + l = ( PL_regeol - locinput > (5 + taill) - pref_len ? (5 + taill) - pref_len : PL_regeol - locinput); + while (UTF8_IS_CONTINUATION(*(U8*)(locinput + l))) + l--; if (pref0_len < 0) pref0_len = 0; if (pref0_len > pref_len) pref0_len = pref_len; regprop(prop, scan); - PerlIO_printf(Perl_debug_log, + 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), - PL_colors[4], pref0_len, + (IV)(locinput - PL_bostr), + PL_colors[4], pref0_len, locinput - pref_len, PL_colors[5], - PL_colors[2], pref_len - pref0_len, + PL_colors[2], pref_len - pref0_len, locinput - pref_len + pref0_len, PL_colors[3], (docolor ? "" : "> <"), PL_colors[0], l, locinput, PL_colors[1], @@ -1968,7 +2045,7 @@ S_regmatch(pTHX_ regnode *prog) case BOL: if (locinput == PL_bostr ? PL_regprev == '\n' - : (PL_multiline && + : (PL_multiline && (nextchr || locinput < PL_regeol) && locinput[-1] == '\n') ) { /* regtill = regbol; */ @@ -2013,32 +2090,51 @@ S_regmatch(pTHX_ regnode *prog) sayNO; break; case SANY: - if (DO_UTF8(PL_reg_sv)) { - locinput += PL_utf8skip[nextchr]; - if (locinput > PL_regeol) - sayNO; - nextchr = UCHARAT(locinput); - break; - } if (!nextchr && locinput >= PL_regeol) sayNO; nextchr = UCHARAT(++locinput); break; case REG_ANY: - if (DO_UTF8(PL_reg_sv)) { + if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n') + sayNO; + if (do_utf8) { locinput += PL_utf8skip[nextchr]; if (locinput > PL_regeol) sayNO; nextchr = UCHARAT(locinput); - break; } - if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n') - sayNO; - nextchr = UCHARAT(++locinput); + else + nextchr = UCHARAT(++locinput); break; case EXACT: s = STRING(scan); ln = STR_LEN(scan); + if (do_utf8 != (UTF!=0)) { + char *l = locinput; + char *e = s + ln; + STRLEN len; + if (do_utf8) + while (s < e) { + if (l >= PL_regeol) + sayNO; + if (*((U8*)s) != utf8_to_uvchr((U8*)l, &len)) + sayNO; + s++; + l += len; + } + else + while (s < e) { + if (l >= PL_regeol) + sayNO; + if (*((U8*)l) != utf8_to_uvchr((U8*)s, &len)) + sayNO; + s += len; + l++; + } + locinput = l; + nextchr = UCHARAT(locinput); + break; + } /* Inline the first character, for speed. */ if (UCHARAT(s) != nextchr) sayNO; @@ -2056,21 +2152,19 @@ S_regmatch(pTHX_ regnode *prog) s = STRING(scan); ln = STR_LEN(scan); - if (UTF) { + if (do_utf8) { char *l = locinput; - char *e = s + ln; + char *e; + e = s + ln; c1 = OP(scan) == EXACTF; while (s < e) { - if (l >= PL_regeol) - sayNO; - if (utf8_to_uv((U8*)s, e - s, 0, 0) != - (c1 ? - toLOWER_utf8((U8*)l) : - toLOWER_LC_utf8((U8*)l))) - { + if (l >= PL_regeol) { sayNO; } - s += UTF8SKIP(s); + if ((UTF ? utf8n_to_uvchr((U8*)s, e - s, 0, 0) : *((U8*)s)) != + (c1 ? toLOWER_utf8((U8*)l) : toLOWER_LC_utf8((U8*)l))) + sayNO; + s += UTF ? UTF8SKIP(s) : 1; l += UTF8SKIP(l); } locinput = l; @@ -2140,6 +2234,7 @@ S_regmatch(pTHX_ regnode *prog) if (!nextchr && locinput >= PL_regeol) sayNO; if (do_utf8) { + LOAD_UTF8_CHARCLASS(alnum,"a"); if (OP(scan) == NALNUM ? swash_fetch(PL_utf8_alnum, (U8*)locinput) : isALNUM_LC_utf8((U8*)locinput)) @@ -2167,15 +2262,16 @@ S_regmatch(pTHX_ regnode *prog) ln = PL_regprev; else { U8 *r = reghop((U8*)locinput, -1); - - ln = utf8_to_uv(r, s - (char*)r, 0, 0); + + ln = utf8n_to_uvchr(r, s - (char*)r, 0, 0); } if (OP(scan) == BOUND || OP(scan) == NBOUND) { ln = isALNUM_uni(ln); + LOAD_UTF8_CHARCLASS(alnum,"a"); n = swash_fetch(PL_utf8_alnum, (U8*)locinput); } else { - ln = isALNUM_LC_uni(ln); + ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln)); n = isALNUM_LC_utf8((U8*)locinput); } } @@ -2201,8 +2297,9 @@ S_regmatch(pTHX_ regnode *prog) case SPACE: if (!nextchr) sayNO; - if (DO_UTF8(PL_reg_sv)) { - if (nextchr & 0x80) { + if (do_utf8) { + if (UTF8_IS_CONTINUED(nextchr)) { + LOAD_UTF8_CHARCLASS(space," "); if (!(OP(scan) == SPACE ? swash_fetch(PL_utf8_space, (U8*)locinput) : isSPACE_LC_utf8((U8*)locinput))) @@ -2231,7 +2328,8 @@ S_regmatch(pTHX_ regnode *prog) case NSPACE: if (!nextchr && locinput >= PL_regeol) sayNO; - if (DO_UTF8(PL_reg_sv)) { + if (do_utf8) { + LOAD_UTF8_CHARCLASS(space," "); if (OP(scan) == NSPACE ? swash_fetch(PL_utf8_space, (U8*)locinput) : isSPACE_LC_utf8((U8*)locinput)) @@ -2253,7 +2351,8 @@ S_regmatch(pTHX_ regnode *prog) case DIGIT: if (!nextchr) sayNO; - if (DO_UTF8(PL_reg_sv)) { + if (do_utf8) { + LOAD_UTF8_CHARCLASS(digit,"0"); if (!(OP(scan) == DIGIT ? swash_fetch(PL_utf8_digit, (U8*)locinput) : isDIGIT_LC_utf8((U8*)locinput))) @@ -2275,7 +2374,8 @@ S_regmatch(pTHX_ regnode *prog) case NDIGIT: if (!nextchr && locinput >= PL_regeol) sayNO; - if (DO_UTF8(PL_reg_sv)) { + if (do_utf8) { + LOAD_UTF8_CHARCLASS(digit,"0"); if (OP(scan) == NDIGIT ? swash_fetch(PL_utf8_digit, (U8*)locinput) : isDIGIT_LC_utf8((U8*)locinput)) @@ -2292,6 +2392,7 @@ S_regmatch(pTHX_ regnode *prog) nextchr = UCHARAT(++locinput); break; case CLUMP: + LOAD_UTF8_CHARCLASS(mark,"~"); if (locinput >= PL_regeol || swash_fetch(PL_utf8_mark,(U8*)locinput)) sayNO; locinput += PL_utf8skip[nextchr]; @@ -2315,7 +2416,7 @@ S_regmatch(pTHX_ regnode *prog) break; s = PL_bostr + ln; - if (UTF && OP(scan) != REF) { /* REF can do byte comparison */ + if (do_utf8 && OP(scan) != REF) { /* REF can do byte comparison */ char *l = locinput; char *e = PL_bostr + PL_regendp[n]; /* @@ -2379,7 +2480,7 @@ S_regmatch(pTHX_ regnode *prog) COP *ocurcop = PL_curcop; SV **ocurpad = PL_curpad; SV *ret; - + 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)) ); @@ -2390,7 +2491,7 @@ S_regmatch(pTHX_ regnode *prog) SPAGAIN; ret = POPs; PUTBACK; - + PL_op = oop; PL_curpad = ocurpad; PL_curcop = ocurcop; @@ -2419,10 +2520,9 @@ S_regmatch(pTHX_ regnode *prog) I32 osize = PL_regsize; I32 onpar = PL_regnpar; - pm.op_pmflags = 0; - pm.op_pmdynflags = (UTF||DO_UTF8(ret) ? PMdf_UTF8 : 0); + Zero(&pm, 1, PMOP); re = CALLREGCOMP(aTHX_ t, t + len, &pm); - if (!(SvFLAGS(ret) + if (!(SvFLAGS(ret) & (SVs_TEMP | SVs_PADTMP | SVf_READONLY))) sv_magic(ret,(SV*)ReREFCNT_inc(re),'r',0,0); PL_regprecomp = oprecomp; @@ -2430,7 +2530,7 @@ S_regmatch(pTHX_ regnode *prog) PL_regnpar = onpar; } DEBUG_r( - PerlIO_printf(Perl_debug_log, + PerlIO_printf(Perl_debug_log, "Entering embedded `%s%.60s%s%s'\n", PL_colors[0], re->precomp, @@ -2443,7 +2543,7 @@ S_regmatch(pTHX_ regnode *prog) state.re = PL_reg_re; PL_regcc = 0; - + cp = regcppush(0); /* Save *all* the positions. */ REGCP_SET(lastcp); cache_re(re); @@ -2532,7 +2632,7 @@ S_regmatch(pTHX_ regnode *prog) 1) After matching X, regnode for CURLYX is processed; - 2) This regnode creates infoblock on the stack, and calls + 2) This regnode creates infoblock on the stack, and calls regmatch() recursively with the starting point at WHILEM node; 3) Each hit of WHILEM node tries to match A and Z (in the order @@ -2553,7 +2653,7 @@ S_regmatch(pTHX_ regnode *prog) and whatever it mentions via ->next, and additional attached trees corresponding to temporarily unset infoblocks as in "5" above. - In the following picture infoblocks for outer loop of + In the following picture infoblocks for outer loop of (Y(A)*?Z)*?T are denoted O, for inner I. NULL starting block is denoted by x. The matched string is YAAZYAZT. Temporarily postponed infoblocks are drawn below the "reset" infoblock. @@ -2646,10 +2746,10 @@ S_regmatch(pTHX_ regnode *prog) PL_reginput = locinput; DEBUG_r( - PerlIO_printf(Perl_debug_log, - "%*s %ld out of %ld..%ld cc=%lx\n", + PerlIO_printf(Perl_debug_log, + "%*s %ld out of %ld..%ld cc=%lx\n", REPORT_CODE_OFF+PL_regindent*2, "", - (long)n, (long)cc->min, + (long)n, (long)cc->min, (long)cc->max, (long)cc) ); @@ -2748,7 +2848,7 @@ S_regmatch(pTHX_ regnode *prog) PL_regcc = cc; if (n >= cc->max) { /* Maximum greed exceeded? */ - if (ckWARN(WARN_REGEXP) && n >= REG_INFTY + if (ckWARN(WARN_REGEXP) && n >= REG_INFTY && !(PL_reg_flags & RF_warned)) { PL_reg_flags |= RF_warned; Perl_warner(aTHX_ WARN_REGEXP, "%s limit (%d) exceeded", @@ -2800,7 +2900,7 @@ S_regmatch(pTHX_ regnode *prog) REPORT_CODE_OFF+PL_regindent*2, "") ); } - if (ckWARN(WARN_REGEXP) && n >= REG_INFTY + if (ckWARN(WARN_REGEXP) && n >= REG_INFTY && !(PL_reg_flags & RF_warned)) { PL_reg_flags |= RF_warned; Perl_warner(aTHX_ WARN_REGEXP, "%s limit (%d) exceeded", @@ -2822,13 +2922,13 @@ S_regmatch(pTHX_ regnode *prog) sayNO; } /* NOT REACHED */ - case BRANCHJ: + case BRANCHJ: next = scan + ARG(scan); if (next == scan) next = NULL; inner = NEXTOPER(NEXTOPER(scan)); goto do_branch; - case BRANCH: + case BRANCH: inner = NEXTOPER(scan); do_branch: { @@ -2871,7 +2971,7 @@ S_regmatch(pTHX_ regnode *prog) { I32 l = 0; CHECKPOINT lastcp; - + /* We suppose that the next guy does not need backtracking: in particular, it is of constant length, and has no parenths to influence future backrefs. */ @@ -3035,13 +3135,23 @@ S_regmatch(pTHX_ regnode *prog) * when we know what character comes next. */ if (PL_regkind[(U8)OP(next)] == EXACT) { - c1 = (U8)*STRING(next); - if (OP(next) == EXACTF) - c2 = PL_fold[c1]; - else if (OP(next) == EXACTFL) - c2 = PL_fold_locale[c1]; - else - c2 = c1; + U8 *s = (U8*)STRING(next); + if (!UTF) { + c2 = c1 = *s; + if (OP(next) == EXACTF) + c2 = PL_fold[c1]; + else if (OP(next) == EXACTFL) + c2 = PL_fold_locale[c1]; + } + else { /* UTF */ + if (OP(next) == EXACTF) { + c1 = to_utf8_lower(s); + c2 = to_utf8_upper(s); + } + else { + c2 = c1 = utf8_to_uvchr(s, NULL); + } + } } else c1 = c2 = -1000; @@ -3054,29 +3164,65 @@ S_regmatch(pTHX_ regnode *prog) locinput = PL_reginput; REGCP_SET(lastcp); if (c1 != -1000) { - char *e = locinput + n - ln; /* Should not check after this */ + char *e; /* Should not check after this */ char *old = locinput; - if (e >= PL_regeol || (n == REG_INFTY)) + if (n == REG_INFTY) { e = PL_regeol - 1; + if (do_utf8) + while (UTF8_IS_CONTINUATION(*(U8*)e)) + e--; + } + else if (do_utf8) { + int m = n - ln; + for (e = locinput; + m >0 && e + UTF8SKIP(e) <= PL_regeol; m--) + e += UTF8SKIP(e); + } + else { + e = locinput + n - ln; + if (e >= PL_regeol) + e = PL_regeol - 1; + } while (1) { + int count; /* Find place 'next' could work */ - if (c1 == c2) { - while (locinput <= e && *locinput != c1) - locinput++; - } else { - while (locinput <= e - && *locinput != c1 - && *locinput != c2) - locinput++; + if (!do_utf8) { + if (c1 == c2) { + while (locinput <= e && *locinput != c1) + locinput++; + } else { + while (locinput <= e + && *locinput != c1 + && *locinput != c2) + locinput++; + } + count = locinput - old; + } + else { + STRLEN len; + if (c1 == c2) { + for (count = 0; + locinput <= e && + utf8_to_uvchr((U8*)locinput, &len) != c1; + count++) + locinput += len; + + } else { + for (count = 0; locinput <= e; count++) { + UV c = utf8_to_uvchr((U8*)locinput, &len); + if (c == c1 || c == c2) + break; + locinput += len; + } + } } - if (locinput > e) + if (locinput > e) sayNO; /* PL_reginput == old now */ if (locinput != old) { ln = 1; /* Did some */ - if (regrepeat(scan, locinput - old) < - locinput - old) + if (regrepeat(scan, count) < count) sayNO; } /* PL_reginput == locinput now */ @@ -3084,15 +3230,30 @@ S_regmatch(pTHX_ regnode *prog) PL_reginput = locinput; /* Could be reset... */ REGCP_UNWIND(lastcp); /* Couldn't or didn't -- move forward. */ - old = locinput++; + old = locinput; + if (do_utf8) + locinput += UTF8SKIP(locinput); + else + locinput++; } } else while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */ + UV c; + if (c1 != -1000) { + if (do_utf8) + c = utf8_to_uvchr((U8*)PL_reginput, NULL); + else + c = UCHARAT(PL_reginput); + /* If it could work, try it. */ + if (c == c1 || c == c2) + { + TRYPAREN(paren, n, PL_reginput); + REGCP_UNWIND(lastcp); + } + } /* If it could work, try it. */ - if (c1 == -1000 || - UCHARAT(PL_reginput) == c1 || - UCHARAT(PL_reginput) == c2) + else if (c1 == -1000) { TRYPAREN(paren, n, PL_reginput); REGCP_UNWIND(lastcp); @@ -3122,11 +3283,16 @@ S_regmatch(pTHX_ regnode *prog) } REGCP_SET(lastcp); if (paren) { + UV c; while (n >= ln) { + if (c1 != -1000) { + if (do_utf8) + c = utf8_to_uvchr((U8*)PL_reginput, NULL); + else + c = UCHARAT(PL_reginput); + } /* If it could work, try it. */ - if (c1 == -1000 || - UCHARAT(PL_reginput) == c1 || - UCHARAT(PL_reginput) == c2) + if (c1 == -1000 || c == c1 || c == c2) { TRYPAREN(paren, n, PL_reginput); REGCP_UNWIND(lastcp); @@ -3137,11 +3303,16 @@ S_regmatch(pTHX_ regnode *prog) } } else { + UV c; while (n >= ln) { + if (c1 != -1000) { + if (do_utf8) + c = utf8_to_uvchr((U8*)PL_reginput, NULL); + else + c = UCHARAT(PL_reginput); + } /* If it could work, try it. */ - if (c1 == -1000 || - UCHARAT(PL_reginput) == c1 || - UCHARAT(PL_reginput) == c2) + if (c1 == -1000 || c == c1 || c == c2) { TRYPAREN(paren, n, PL_reginput); REGCP_UNWIND(lastcp); @@ -3206,7 +3377,7 @@ S_regmatch(pTHX_ regnode *prog) case SUSPEND: n = 1; PL_reginput = locinput; - goto do_ifmatch; + goto do_ifmatch; case UNLESSM: n = 0; if (scan->flags) { @@ -3219,7 +3390,7 @@ S_regmatch(pTHX_ regnode *prog) PL_reginput = s; } else { - if (locinput < PL_bostr + scan->flags) + if (locinput < PL_bostr + scan->flags) goto say_yes; PL_reginput = locinput - scan->flags; goto do_ifmatch; @@ -3240,7 +3411,7 @@ S_regmatch(pTHX_ regnode *prog) PL_reginput = s; } else { - if (locinput < PL_bostr + scan->flags) + if (locinput < PL_bostr + scan->flags) goto say_no; PL_reginput = locinput - scan->flags; goto do_ifmatch; @@ -3333,14 +3504,14 @@ do_no: { re_unwind_branch_t *uwb = &(uw->branch); I32 lastparen = uwb->lastparen; - + REGCP_UNWIND(uwb->lastcp); for (n = *PL_reglastparen; n > lastparen; n--) PL_regendp[n] = -1; *PL_reglastparen = n; scan = next = uwb->next; - if ( !scan || - OP(scan) != (uwb->type == RE_UNWIND_BRANCH + if ( !scan || + OP(scan) != (uwb->type == RE_UNWIND_BRANCH ? BRANCH : BRANCHJ) ) { /* Failure */ unwind = uwb->prev; #ifdef DEBUGGING @@ -3401,9 +3572,9 @@ S_regrepeat(pTHX_ regnode *p, I32 max) loceol = scan + max; switch (OP(p)) { case REG_ANY: - if (DO_UTF8(PL_reg_sv)) { + if (do_utf8) { loceol = PL_regeol; - while (scan < loceol && *scan != '\n') { + while (scan < loceol && hardcount < max && *scan != '\n') { scan += UTF8SKIP(scan); hardcount++; } @@ -3413,15 +3584,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) } break; case SANY: - if (DO_UTF8(PL_reg_sv)) { - loceol = PL_regeol; - while (scan < loceol) { - scan += UTF8SKIP(scan); - hardcount++; - } - } else { - scan = loceol; - } + scan = loceol; break; case EXACT: /* length of string is 1 */ c = (U8)*STRING(p); @@ -3444,7 +3607,8 @@ S_regrepeat(pTHX_ regnode *p, I32 max) case ANYOF: if (do_utf8) { loceol = PL_regeol; - while (scan < loceol && reginclass(p, (U8*)scan, do_utf8)) { + while (hardcount < max && scan < loceol && + reginclass(p, (U8*)scan, do_utf8)) { scan += UTF8SKIP(scan); hardcount++; } @@ -3454,9 +3618,11 @@ S_regrepeat(pTHX_ regnode *p, I32 max) } break; case ALNUM: - if (DO_UTF8(PL_reg_sv)) { + if (do_utf8) { loceol = PL_regeol; - while (scan < loceol && swash_fetch(PL_utf8_alnum, (U8*)scan)) { + LOAD_UTF8_CHARCLASS(alnum,"a"); + while (hardcount < max && scan < loceol && + swash_fetch(PL_utf8_alnum, (U8*)scan)) { scan += UTF8SKIP(scan); hardcount++; } @@ -3467,9 +3633,10 @@ S_regrepeat(pTHX_ regnode *p, I32 max) break; case ALNUML: PL_reg_flags |= RF_tainted; - if (DO_UTF8(PL_reg_sv)) { + if (do_utf8) { loceol = PL_regeol; - while (scan < loceol && isALNUM_LC_utf8((U8*)scan)) { + while (hardcount < max && scan < loceol && + isALNUM_LC_utf8((U8*)scan)) { scan += UTF8SKIP(scan); hardcount++; } @@ -3479,9 +3646,11 @@ S_regrepeat(pTHX_ regnode *p, I32 max) } break; case NALNUM: - if (DO_UTF8(PL_reg_sv)) { + if (do_utf8) { loceol = PL_regeol; - while (scan < loceol && !swash_fetch(PL_utf8_alnum, (U8*)scan)) { + LOAD_UTF8_CHARCLASS(alnum,"a"); + while (hardcount < max && scan < loceol && + !swash_fetch(PL_utf8_alnum, (U8*)scan)) { scan += UTF8SKIP(scan); hardcount++; } @@ -3492,9 +3661,10 @@ S_regrepeat(pTHX_ regnode *p, I32 max) break; case NALNUML: PL_reg_flags |= RF_tainted; - if (DO_UTF8(PL_reg_sv)) { + if (do_utf8) { loceol = PL_regeol; - while (scan < loceol && !isALNUM_LC_utf8((U8*)scan)) { + while (hardcount < max && scan < loceol && + !isALNUM_LC_utf8((U8*)scan)) { scan += UTF8SKIP(scan); hardcount++; } @@ -3504,9 +3674,10 @@ S_regrepeat(pTHX_ regnode *p, I32 max) } break; case SPACE: - if (DO_UTF8(PL_reg_sv)) { + if (do_utf8) { loceol = PL_regeol; - while (scan < loceol && + LOAD_UTF8_CHARCLASS(space," "); + while (hardcount < max && scan < loceol && (*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { scan += UTF8SKIP(scan); hardcount++; @@ -3518,9 +3689,9 @@ S_regrepeat(pTHX_ regnode *p, I32 max) break; case SPACEL: PL_reg_flags |= RF_tainted; - if (DO_UTF8(PL_reg_sv)) { + if (do_utf8) { loceol = PL_regeol; - while (scan < loceol && + while (hardcount < max && scan < loceol && (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { scan += UTF8SKIP(scan); hardcount++; @@ -3531,9 +3702,10 @@ S_regrepeat(pTHX_ regnode *p, I32 max) } break; case NSPACE: - if (DO_UTF8(PL_reg_sv)) { + if (do_utf8) { loceol = PL_regeol; - while (scan < loceol && + LOAD_UTF8_CHARCLASS(space," "); + while (hardcount < max && scan < loceol && !(*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { scan += UTF8SKIP(scan); hardcount++; @@ -3545,9 +3717,9 @@ S_regrepeat(pTHX_ regnode *p, I32 max) } case NSPACEL: PL_reg_flags |= RF_tainted; - if (DO_UTF8(PL_reg_sv)) { + if (do_utf8) { loceol = PL_regeol; - while (scan < loceol && + while (hardcount < max && scan < loceol && !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { scan += UTF8SKIP(scan); hardcount++; @@ -3558,9 +3730,11 @@ S_regrepeat(pTHX_ regnode *p, I32 max) } break; case DIGIT: - if (DO_UTF8(PL_reg_sv)) { + if (do_utf8) { loceol = PL_regeol; - while (scan < loceol && swash_fetch(PL_utf8_digit,(U8*)scan)) { + LOAD_UTF8_CHARCLASS(digit,"0"); + while (hardcount < max && scan < loceol && + swash_fetch(PL_utf8_digit,(U8*)scan)) { scan += UTF8SKIP(scan); hardcount++; } @@ -3570,9 +3744,11 @@ S_regrepeat(pTHX_ regnode *p, I32 max) } break; case NDIGIT: - if (DO_UTF8(PL_reg_sv)) { + if (do_utf8) { loceol = PL_regeol; - while (scan < loceol && !swash_fetch(PL_utf8_digit,(U8*)scan)) { + LOAD_UTF8_CHARCLASS(digit,"0"); + while (hardcount < max && scan < loceol && + !swash_fetch(PL_utf8_digit,(U8*)scan)) { scan += UTF8SKIP(scan); hardcount++; } @@ -3591,22 +3767,22 @@ S_regrepeat(pTHX_ regnode *p, I32 max) c = scan - PL_reginput; PL_reginput = scan; - DEBUG_r( + DEBUG_r( { SV *prop = sv_newmortal(); regprop(prop, p); - PerlIO_printf(Perl_debug_log, - "%*s %s can match %"IVdf" times out of %"IVdf"...\n", + 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); }); - + return(c); } /* - regrepeat_hard - repeatedly match something, report total lenth and length - * + * * The repeater is supposed to have constant length. */ @@ -3623,7 +3799,7 @@ S_regrepeat_hard(pTHX_ regnode *p, I32 max, I32 *lp) return 0; start = PL_reginput; - if (UTF) { + if (DO_UTF8(PL_reg_sv)) { while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) { if (!count++) { l = 0; @@ -3652,7 +3828,7 @@ S_regrepeat_hard(pTHX_ regnode *p, I32 max, I32 *lp) } if (!res) PL_reginput = scan; - + return count; } @@ -3673,10 +3849,10 @@ Perl_regclass_swash(pTHX_ register regnode* node, bool doinit, SV** initsvp) SV *rv = (SV*)PL_regdata->data[n]; AV *av = (AV*)SvRV((SV*)rv); SV **a; - + si = *av_fetch(av, 0, FALSE); a = av_fetch(av, 1, FALSE); - + if (a) sw = *a; else if (si && doinit) { @@ -3701,16 +3877,21 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, register bool do_utf8) { char flags = ANYOF_FLAGS(n); bool match = FALSE; + UV c; + STRLEN len; + + if (do_utf8) + c = utf8_to_uvchr(p, &len); + else + c = *p; if (do_utf8 || (flags & ANYOF_UNICODE)) { if (do_utf8 && !ANYOF_RUNTIME(n)) { - STRLEN len; - UV c = utf8_to_uv_simple(p, &len); - if (len != (STRLEN)-1 && c < 256 && ANYOF_BITMAP_TEST(n, c)) match = TRUE; } - + if (do_utf8 && flags & ANYOF_UNICODE_ALL && c >= 256) + match = TRUE; if (!match) { SV *sw = regclass_swash(n, TRUE, 0); @@ -3719,22 +3900,20 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, register bool do_utf8) match = TRUE; else if (flags & ANYOF_FOLD) { U8 tmpbuf[UTF8_MAXLEN+1]; - + if (flags & ANYOF_LOCALE) { PL_reg_flags |= RF_tainted; - uv_to_utf8(tmpbuf, toLOWER_LC_utf8(p)); + uvchr_to_utf8(tmpbuf, toLOWER_LC_utf8(p)); } - else - uv_to_utf8(tmpbuf, toLOWER_utf8(p)); + else + uvchr_to_utf8(tmpbuf, toLOWER_utf8(p)); if (swash_fetch(sw, tmpbuf)) match = TRUE; } } } } - else { - U8 c = *p; - + if (!match && c < 256) { if (ANYOF_BITMAP_TEST(n, c)) match = TRUE; else if (flags & ANYOF_FOLD) { @@ -3795,19 +3974,25 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, register bool do_utf8) STATIC U8 * S_reghop(pTHX_ U8 *s, I32 off) -{ +{ + return S_reghop3(aTHX_ s, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr)); +} + +STATIC U8 * +S_reghop3(pTHX_ U8 *s, I32 off, U8* lim) +{ if (off >= 0) { - while (off-- && s < (U8*)PL_regeol) { + while (off-- && s < lim) { /* XXX could check well-formedness here */ s += UTF8SKIP(s); } } else { while (off++) { - if (s > (U8*)PL_bostr) { + if (s > lim) { s--; if (UTF8_IS_CONTINUED(*s)) { - while (s > (U8*)PL_bostr && UTF8_IS_CONTINUATION(*s)) + while (s > (U8*)lim && UTF8_IS_CONTINUATION(*s)) s--; } /* XXX could check well-formedness here */ @@ -3818,10 +4003,16 @@ S_reghop(pTHX_ U8 *s, I32 off) } STATIC U8 * -S_reghopmaybe(pTHX_ U8* s, I32 off) +S_reghopmaybe(pTHX_ U8 *s, I32 off) +{ + return S_reghopmaybe3(aTHX_ s, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr)); +} + +STATIC U8 * +S_reghopmaybe3(pTHX_ U8* s, I32 off, U8* lim) { if (off >= 0) { - while (off-- && s < (U8*)PL_regeol) { + while (off-- && s < lim) { /* XXX could check well-formedness here */ s += UTF8SKIP(s); } @@ -3830,10 +4021,10 @@ S_reghopmaybe(pTHX_ U8* s, I32 off) } else { while (off++) { - if (s > (U8*)PL_bostr) { + if (s > lim) { s--; if (UTF8_IS_CONTINUED(*s)) { - while (s > (U8*)PL_bostr && UTF8_IS_CONTINUATION(*s)) + while (s > (U8*)lim && UTF8_IS_CONTINUATION(*s)) s--; } /* XXX could check well-formedness here */