[perl #23769] Unicode regex broken on simple example

regrepeat() did not work right for UTF-8(ed Latin-1)
in the EXACT case, which made the \x{a0}+ fail.

p4raw-id: //depot/perl@21158

diff --git a/regexec.c b/regexec.c
index 464ceaf..d2e9c66 100644 (file)
--- a/regexec.c
+++ b/regexec.c
@@ -4065,10 +4065,16 @@ S_regrepeat(pTHX_ regnode *p, I32 max)
     case CANY:
        scan = loceol;
        break;
-    case EXACT:                /* length of string is 1 */
-       c = (U8)*STRING(p);
-       while (scan < loceol && UCHARAT(scan) == c)
-           scan++;
+    case EXACT:
+        if (do_utf8) {
+            c = (U8)*STRING(p);
+            while (scan < loceol && utf8_to_uvuni((U8*)scan, 0) == c)
+                 scan += UTF8SKIP(scan);
+       } else {                /* length of string is 1 */
+            c = (U8)*STRING(p);
+            while (scan < loceol && UCHARAT(scan) == c)
+                 scan++;
+       }
        break;
     case EXACTF:       /* length of string is 1 */
        c = (U8)*STRING(p);

diff --git a/t/op/pat.t b/t/op/pat.t
index 27262bd..54f67fc 100755 (executable)
--- a/t/op/pat.t
+++ b/t/op/pat.t
@@ -6,7 +6,7 @@
 
 $| = 1;
 
-print "1..1012\n";
+print "1..1015\n";
 
 BEGIN {
     chdir 't' if -d 't';
@@ -3212,5 +3212,16 @@ ok("  \x{10428}" =~ qr/\x{10400}/i,
 ok("  \x{1E01}x" =~ qr/\x{1E00}X/i,
    "<20030808193656.5109.1@llama.ni-s.u-net.com>");
 
-# last test 1012
+{
+    # [perl #23769] Unicode regex broken on simple example
+    # regrepeat() didn't handle UTF-8 EXACT case right.
+
+    my $s = "\x{a0}\x{a0}\x{a0}\x{100}"; chop $s;
+
+    ok($s =~ /\x{a0}/,       "[perl #23769]");
+    ok($s =~ /\x{a0}+/,      "[perl #23769]");
+    ok($s =~ /\x{a0}\x{a0}/, "[perl #23769]");
+}
+
+# last test 1015