[perl #23769] Unicode regex broken on simple example

[perl5.git] / t / op / pat.t

diff --git a/t/op/pat.t b/t/op/pat.t
index 7be0afc..54f67fc 100755 (executable)
--- a/t/op/pat.t
+++ b/t/op/pat.t
@@ -6,7 +6,7 @@
 
 $| = 1;
 
-print "1..1010\n";
+print "1..1015\n";
 
 BEGIN {
     chdir 't' if -d 't';
@@ -1365,10 +1365,10 @@ print "ok 247\n";
     print "ok 263\n";
 }
 
-{
+SKIP: {
     my $test = 264; # till 575
 
-    use charnames ':full';
+    use charnames ":full";
 
     # This is far from complete testing, there are dozens of character
     # classes in Unicode.  The mixing of literals and \N{...} is
@@ -3196,11 +3196,32 @@ ok("abcde" eq "$`", '# TODO #19049 - global match not setting $`');
 
 ok("123\x{100}" =~ /^.*1.*23\x{100}$/, 'uft8 + multiple floating substr');
 
+# LATIN SMALL/CAPITAL LETTER A WITH MACRON
 ok("  \x{101}" =~ qr/\x{100}/i,
    "<20030808193656.5109.1@llama.ni-s.u-net.com>");
 
+# LATIN SMALL/CAPITAL LETTER A WITH RING BELOW
 ok("  \x{1E01}" =~ qr/\x{1E00}/i,
    "<20030808193656.5109.1@llama.ni-s.u-net.com>");
 
-# last test 1010
+# DESERET SMALL/CAPITAL LETTER LONG I
+ok("  \x{10428}" =~ qr/\x{10400}/i,
+   "<20030808193656.5109.1@llama.ni-s.u-net.com>");
+
+# LATIN SMALL/CAPITAL LETTER A WITH RING BELOW + 'X'
+ok("  \x{1E01}x" =~ qr/\x{1E00}X/i,
+   "<20030808193656.5109.1@llama.ni-s.u-net.com>");
+
+{
+    # [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