split() utf8 fixes. Should fix both 20001014.001 and 20000426.003.

The problem was that rx->minlen was in chars while pp_split()
thought it would be in bytes.

p4raw-id: //depot/perl@7234

1 files changed, 35 insertions, 1 deletions

diff --git a/t/pragma/utf8.t b/t/pragma/utf8.t
index 953064c7be..51c084cffa 100755
--- a/t/pragma/utf8.t
+++ b/t/pragma/utf8.t
@@ -10,7 +10,7 @@ BEGIN {
     }
 }
 
-print "1..82\n";
+print "1..87\n";
 
 my $test = 1;
 
@@ -437,3 +437,37 @@ sub nok_bytes {
     print "ok $test\n";
     $test++;
 }
+
+{
+    # bug id 20000426.003
+
+    use utf8;
+
+    my $s = "\x20\x40\x{80}\x{100}\x{80}\x40\x20";
+
+    my ($a, $b, $c) = split(/\x40/, $s);
+    print "not "
+	unless $a eq "\x20" && $b eq "\x{80}\x{100}\x{80}" && $c eq $a;
+    print "ok $test\n";
+    $test++;
+
+    my ($a, $b) = split(/\x{100}/, $s);
+    print "not " unless $a eq "\x20\x40\x{80}" && $b eq "\x{80}\x40\x20";
+    print "ok $test\n";
+    $test++;
+
+    my ($a, $b) = split(/\x{80}\x{100}\x{80}/, $s);
+    print "not " unless $a eq "\x20\x40" && $b eq "\x40\x20";
+    print "ok $test\n";
+    $test++;
+
+    my ($a, $b) = split(/\x40\x{80}/, $s);
+    print "not " unless $a eq "\x20" && $b eq "\x{100}\x{80}\x40\x20";
+    print "ok $test\n";
+    $test++;
+
+    my ($a, $b, $c) = split(/[\x40\x{80}]+/, $s);
+    print "not " unless $a eq "\x20" && $b eq "\x{100}" && $c eq "\x20";
+    print "ok $test\n";
+    $test++;
+}