Change 14001 by jhi@alpha on 2002/01/01 17:29:05
Better support for multicharacter foldings.
Now all but two of the CaseFold.txt cases work-- but
only when the target string is single-character, some
regex maloptimization probably at work.
Affected files ...
.... //depot/perl/regexec.c#233 edit
Differences ...
==== //depot/perl/regexec.c#233 (text) ====
Index: perl/regexec.c
--- perl/regexec.c.~1~ Tue Jan 1 10:45:06 2002
+++ perl/regexec.c Tue Jan 1 10:45:06 2002
@@ -960,7 +960,7 @@
c1 = *(U8*)m;
c2 = PL_fold_locale[c1];
do_exactf:
- e = strend - ln;
+ e = do_utf8 ? s + ln - 1 : strend - ln;
if (norun && e < s)
e = s; /* Due to minlen logic of intuit() */
@@ -2406,31 +2406,37 @@
if (do_utf8 && UTF) {
/* Both the target and the pattern are utf8. */
+ U8 lfoldbuf[UTF8_MAXLEN_FOLD+1], *lf;
+ U8 sfoldbuf[UTF8_MAXLEN_FOLD+1], *sf;
+ STRLEN lfoldlen, sfoldlen;
+ STRLEN llen = 0;
+ STRLEN slen = 0;
+
while (s < e) {
- if (l >= PL_regeol)
- sayNO;
- if (UTF8SKIP(s) != UTF8SKIP(l) ||
- memNE(s, (char*)l, UTF8SKIP(s))) {
- U8 lfoldbuf[UTF8_MAXLEN_FOLD+1];
- STRLEN lfoldlen;
+ /* Fold them and walk them characterwise. */
- /* Try one of them folded. */
-
+ if (llen == 0) {
to_utf8_fold((U8*)l, lfoldbuf, &lfoldlen);
- if (UTF8SKIP(s) != lfoldlen ||
- memNE(s, (char*)lfoldbuf, lfoldlen)) {
- U8 sfoldbuf[UTF8_MAXLEN_FOLD+1];
- STRLEN sfoldlen;
+ lf = lfoldbuf;
+ llen = lfoldlen;
+ }
- /* Try both of them folded. */
+ if (slen == 0) {
+ to_utf8_fold((U8*)s, sfoldbuf, &sfoldlen);
+ sf = sfoldbuf;
+ slen = sfoldlen;
+ }
- to_utf8_fold((U8*)s, sfoldbuf, &sfoldlen);
- if (sfoldlen != lfoldlen ||
- memNE((char*)sfoldbuf,
- (char*)lfoldbuf, lfoldlen))
- sayNO;
- }
+ while (llen && slen) {
+ if (UTF8SKIP(lf) != UTF8SKIP(sf) ||
+ memNE((char*)lf, (char*)sf, UTF8SKIP(lf)))
+ sayNO;
+ llen -= UTF8SKIP(lf);
+ lf += UTF8SKIP(lf);
+ slen -= UTF8SKIP(sf);
+ sf += UTF8SKIP(sf);
}
+
l += UTF8SKIP(l);
s += UTF8SKIP(s);
}
End of Patch.
