# http://rt.perl.org/rt3//Public/Bug/Display.html?id=69973 # http://perl5.git.perl.org/perl.git/commitdiff_plain/0abd0d78a73da1c4d13b1c700526b7e5d03b32d4?hp=fe88edf0c4ada4230b84bdb5417029b8f766694a diff --git a/ext/re/t/regop.t b/ext/re/t/regop.t index 9118bf6..46e6ec0 100644 --- ext/re/t/regop.t +++ ext/re/t/regop.t @@ -231,12 +231,12 @@ anchored "ABC" at 0 #Freeing REx: "(\\.COM|\\.EXE|\\.BAT|\\.CMD|\\.VBS|\\.VBE|\\.JS|\\.JSE|\\."...... %MATCHED% floating ""$ at 3..4 (checking floating) -1:1[1] 3:2[1] 5:2[64] 45:83[1] 47:84[1] 48:85[0] -stclass EXACTF <.> minlen 3 -Found floating substr ""$ at offset 30... -Does not contradict STCLASS... -Guessed: match at offset 26 -Matching stclass EXACTF <.> against ".exe" +#1:1[1] 3:2[1] 5:2[64] 45:83[1] 47:84[1] 48:85[0] +#stclass EXACTF <.> minlen 3 +#Found floating substr ""$ at offset 30... +#Does not contradict STCLASS... +#Guessed: match at offset 26 +#Matching stclass EXACTF <.> against ".exe" --- #Compiling REx "[q]" #size 12 nodes Got 100 bytes for offset annotations. diff --git a/regcomp.c b/regcomp.c index 6e9fa26..eb5f12f 100644 --- regcomp.c +++ regcomp.c @@ -2833,13 +2833,18 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp, } } else { /* - Currently we assume that the trie can handle unicode and ascii - matches fold cased matches. If this proves true then the following - define will prevent tries in this situation. - - #define TRIE_TYPE_IS_SAFE (UTF || optype==EXACT) -*/ + Currently we do not believe that the trie logic can + handle case insensitive matching properly when the + pattern is not unicode (thus forcing unicode semantics). + + If/when this is fixed the following define can be swapped + in below to fully enable trie logic. + #define TRIE_TYPE_IS_SAFE 1 + +*/ +#define TRIE_TYPE_IS_SAFE (UTF || optype==EXACT) + if ( last && TRIE_TYPE_IS_SAFE ) { make_trie( pRExC_state, startbranch, first, cur, tail, count, diff --git a/regexec.c b/regexec.c index 402ede3..ec09c28 100644 --- regexec.c +++ regexec.c @@ -1105,16 +1105,15 @@ Perl_re_intuit_start(pTHX_ REGEXP * const rx, SV *sv, char *strpos, #define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uscan, len, \ uvc, charid, foldlen, foldbuf, uniflags) STMT_START { \ - UV uvc_unfolded = 0; \ switch (trie_type) { \ case trie_utf8_fold: \ if ( foldlen>0 ) { \ - uvc_unfolded = uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \ + uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \ foldlen -= len; \ uscan += len; \ len=0; \ } else { \ - uvc_unfolded = uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ + uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \ foldlen -= UNISKIP( uvc ); \ uscan = foldbuf + UNISKIP( uvc ); \ @@ -1140,7 +1139,6 @@ uvc, charid, foldlen, foldbuf, uniflags) STMT_START { \ uvc = (UV)*uc; \ len = 1; \ } \ - \ if (uvc < 256) { \ charid = trie->charmap[ uvc ]; \ } \ @@ -1153,9 +1151,6 @@ uvc, charid, foldlen, foldbuf, uniflags) STMT_START { \ charid = (U16)SvIV(*svpp); \ } \ } \ - if (!charid && trie_type == trie_utf8_fold && !UTF) { \ - charid = trie->charmap[uvc_unfolded]; \ - } \ } STMT_END #define REXEC_FBC_EXACTISH_CHECK(CoNd) \