It could be possible for the case-insensitive

Unicode-aware string comparison to wander off
to the la-la land.

p4raw-id: //depot/perl@13669

diff --git a/embed.h b/embed.h
index 9523ef5..ea261bf 100644 (file)
--- a/embed.h
+++ b/embed.h
@@ -1782,7 +1782,7 @@
 #define hv_undef(a)            Perl_hv_undef(aTHX_ a)
 #define ibcmp(a,b,c)           Perl_ibcmp(aTHX_ a,b,c)
 #define ibcmp_locale(a,b,c)    Perl_ibcmp_locale(aTHX_ a,b,c)
-#define ibcmp_utf8(a,b,c,d,e)  Perl_ibcmp_utf8(aTHX_ a,b,c,d,e)
+#define ibcmp_utf8(a,b,c,d,e,f)        Perl_ibcmp_utf8(aTHX_ a,b,c,d,e,f)
 #define ingroup(a,b)           Perl_ingroup(aTHX_ a,b)
 #define init_argv_symbols(a,b) Perl_init_argv_symbols(aTHX_ a,b)
 #define init_debugger()                Perl_init_debugger(aTHX)

diff --git a/embed.pl b/embed.pl
index 639ba6c..91c2ac0 100755 (executable)
--- a/embed.pl
+++ b/embed.pl
@@ -1333,7 +1333,7 @@ Apd       |HE*    |hv_store_ent   |HV* tb|SV* key|SV* val|U32 hash
 Apd    |void   |hv_undef       |HV* tb
 Ap     |I32    |ibcmp          |const char* a|const char* b|I32 len
 Ap     |I32    |ibcmp_locale   |const char* a|const char* b|I32 len
-Apd    |I32    |ibcmp_utf8     |const char* a|bool ua|const char* b|bool ub|I32 len
+Apd    |I32    |ibcmp_utf8     |const char* a|bool ua|I32 len1|const char* b|bool ub|I32 len2
 p      |bool   |ingroup        |Gid_t testgid|Uid_t effective
 p      |void   |init_argv_symbols|int|char **
 p      |void   |init_debugger

diff --git a/pod/perlapi.pod b/pod/perlapi.pod
index 6ac32f4..7bdf75c 100644 (file)
--- a/pod/perlapi.pod
+++ b/pod/perlapi.pod
@@ -1113,14 +1113,13 @@ Found in file hv.c
 Return true if the strings s1 and s2 differ case-insensitively, false
 if not (if they are equal case-insensitively).  If u1 is true, the
 string s1 is assumed to be in UTF-8-encoded Unicode.  If u2 is true,
-the string s2 is assumed to be in UTF-8-encoded Unicode.  (If both u1
-and u2 are false, ibcmp() is called.)
+the string s2 is assumed to be in UTF-8-encoded Unicode.
 
 For case-insensitiveness, the "casefolding" of Unicode is used
 instead of upper/lowercasing both the characters, see
 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
 
-       I32     ibcmp_utf8(const char* a, bool ua, const char* b, bool ub, I32 len)
+       I32     ibcmp_utf8(const char* a, bool ua, I32 len1, const char* b, bool ub, I32 len2)
 
 =for hackers
 Found in file utf8.c

diff --git a/pod/perlunicode.pod b/pod/perlunicode.pod
index b1ffed5..e707dc6 100644 (file)
--- a/pod/perlunicode.pod
+++ b/pod/perlunicode.pod
@@ -885,9 +885,9 @@ code points).
 
 =item *
 
-ibcmp_utf8(s1, u1, s2, u2, len) can be used to compare two strings
-case-insensitively in Unicode.  (For case-sensitive comparisons you
-can just use memEQ() and memNE() as usual.)
+ibcmp_utf8(s1, u1, len1, s2, u2, len2) can be used to compare two
+strings case-insensitively in Unicode.  (For case-sensitive
+comparisons you can just use memEQ() and memNE() as usual.)
 
 =back
 

diff --git a/proto.h b/proto.h
index 5c9a514..f9161bb 100644 (file)
--- a/proto.h
+++ b/proto.h
@@ -315,7 +315,7 @@ PERL_CALLCONV HE*   Perl_hv_store_ent(pTHX_ HV* tb, SV* key, SV* val, U32 hash);
 PERL_CALLCONV void     Perl_hv_undef(pTHX_ HV* tb);
 PERL_CALLCONV I32      Perl_ibcmp(pTHX_ const char* a, const char* b, I32 len);
 PERL_CALLCONV I32      Perl_ibcmp_locale(pTHX_ const char* a, const char* b, I32 len);
-PERL_CALLCONV I32      Perl_ibcmp_utf8(pTHX_ const char* a, bool ua, const char* b, bool ub, I32 len);
+PERL_CALLCONV I32      Perl_ibcmp_utf8(pTHX_ const char* a, bool ua, I32 len1, const char* b, bool ub, I32 len2);
 PERL_CALLCONV bool     Perl_ingroup(pTHX_ Gid_t testgid, Uid_t effective);
 PERL_CALLCONV void     Perl_init_argv_symbols(pTHX_ int, char **);
 PERL_CALLCONV void     Perl_init_debugger(pTHX);

diff --git a/regexec.c b/regexec.c
index 0d97be3..4b073d2 100644 (file)
--- a/regexec.c
+++ b/regexec.c
@@ -971,7 +971,8 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
                    while (s <= e) {
                        if ( utf8_to_uvchr((U8*)s, &len) == c1
                             && (ln == 1 ||
-                                ibcmp_utf8(s, do_utf8, m, UTF, ln)) )
+                                ibcmp_utf8(s, do_utf8,  strend - s,
+                                           m, UTF, ln)) )
                            goto got_it;
                        s += len;
                    }
@@ -980,7 +981,8 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
                        UV c = utf8_to_uvchr((U8*)s, &len);
                        if ( (c == c1 || c == c2)
                             && (ln == 1 ||
-                                ibcmp_utf8(s, do_utf8, m, UTF, ln)) )
+                                ibcmp_utf8(s, do_utf8, strend - s,
+                                           m, UTF, ln)) )
                            goto got_it;
                        s += len;
                    }

diff --git a/utf8.c b/utf8.c
index 1a024b7..747b78c 100644 (file)
--- a/utf8.c
+++ b/utf8.c
@@ -1579,8 +1579,7 @@ Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags)
 Return true if the strings s1 and s2 differ case-insensitively, false
 if not (if they are equal case-insensitively).  If u1 is true, the
 string s1 is assumed to be in UTF-8-encoded Unicode.  If u2 is true,
-the string s2 is assumed to be in UTF-8-encoded Unicode.  (If both u1
-and u2 are false, ibcmp() is called.)
+the string s2 is assumed to be in UTF-8-encoded Unicode.
 
 For case-insensitiveness, the "casefolding" of Unicode is used
 instead of upper/lowercasing both the characters, see
@@ -1588,50 +1587,52 @@ http://www.unicode.org/unicode/reports/tr21/ (Case Mappings).
 
 =cut */
 I32
-Perl_ibcmp_utf8(pTHX_ const char *s1, bool u1, const char *s2, bool u2, register I32 len)
-{
-     if (u1 || u2) {
-         register U8 *a = (U8*)s1;
-         register U8 *b = (U8*)s2;
-         STRLEN la, lb;
-         UV ca, cb;
-         STRLEN ulen1, ulen2;
-         U8 tmpbuf1[UTF8_MAXLEN*3+1];
-         U8 tmpbuf2[UTF8_MAXLEN*3+1];
-         
-         while (len) {
+Perl_ibcmp_utf8(pTHX_ const char *s1, bool u1, register I32 len1, const char *s2, bool u2, register I32 len2)
+{
+     register U8 *a  = (U8*)s1;
+     register U8 *b  = (U8*)s2;
+     register U8 *ae = b + len1;
+     register U8 *be = b + len2;
+     STRLEN la, lb;
+     UV ca, cb;
+     STRLEN ulen1, ulen2;
+     U8 tmpbuf1[UTF8_MAXLEN*3+1];
+     U8 tmpbuf2[UTF8_MAXLEN*3+1];
+     
+     while (a < ae && b < be) {
+         if (u1) {
+              if (a + UTF8SKIP(a) > ae)
+                   break;
+              ca = utf8_to_uvchr((U8*)a, &la);
+         } else {
+              ca = *a;
+              la = 1;
+         }
+         if (u2) {
+              if (b + UTF8SKIP(b) > be)
+                   break;
+              cb = utf8_to_uvchr((U8*)b, &lb);
+         } else {
+              cb = *b;
+              lb = 1;
+         }
+         if (ca != cb) {
               if (u1)
-                   ca = utf8_to_uvchr((U8*)a, &la);
-              else {
-                   ca = *a;
-                   la = 1;
-              }
+                   to_uni_fold(NATIVE_TO_UNI(ca), tmpbuf1, &ulen1);
+              else
+                   ulen1 = 1;
               if (u2)
-                   cb = utf8_to_uvchr((U8*)b, &lb);
-              else {
-                   cb = *b;
-                   lb = 1;
-              }
-              if (ca != cb) {
-                   if (u1)
-                        to_uni_fold(NATIVE_TO_UNI(ca), tmpbuf1, &ulen1);
-                   else
-                        ulen1 = 1;
-                   if (u2)
-                        to_uni_fold(NATIVE_TO_UNI(cb), tmpbuf2, &ulen2);
-                   else
-                        ulen2 = 1;
-                   if (ulen1 != ulen2
-                       || (ulen1 == 1 && PL_fold[ca] != PL_fold[cb])
-                       || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1))
-                        return 1;
-              }
-              a += la;
-              b += lb;
+                   to_uni_fold(NATIVE_TO_UNI(cb), tmpbuf2, &ulen2);
+              else
+                   ulen2 = 1;
+              if (ulen1 != ulen2
+                  || (ulen1 == 1 && PL_fold[ca] != PL_fold[cb])
+                  || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1))
+                   return 1; /* mismatch */
          }
-         return 0;
+         a += la;
+         b += lb;
      }
-     else
-         return ibcmp(s1, s2, len);
+     return a == ae && b == be ? 0 : 1; /* 0 match, 1 mismatch */
 }