Re: [ID 20020630.002] utf8 regex only matches 32k

Message-Id: <200207011228.g61CS4T06766@crypt.compulink.co.uk>
Date: Mon, 01 Jul 2002 13:28:04 +0100

p4raw-id: //depot/perl@17390

diff --git a/regexec.c b/regexec.c
index 793201c..bdc05af 100644 (file)
--- a/regexec.c
+++ b/regexec.c
@@ -3990,7 +3990,9 @@ S_regrepeat(pTHX_ regnode *p, I32 max)
     register bool do_utf8 = PL_reg_match_utf8;
 
     scan = PL_reginput;
-    if (max != REG_INFTY && max < loceol - scan)
+    if (max == REG_INFTY)
+       max = I32_MAX;
+    else if (max < loceol - scan)
       loceol = scan + max;
     switch (OP(p)) {
     case REG_ANY:

diff --git a/t/op/pat.t b/t/op/pat.t
index 5681d6a..767c069 100755 (executable)
--- a/t/op/pat.t
+++ b/t/op/pat.t
@@ -6,7 +6,7 @@
 
 $| = 1;
 
-print "1..910\n";
+print "1..922\n";
 
 BEGIN {
     chdir 't' if -d 't';
@@ -2884,3 +2884,21 @@ EOF
 print "d" =~ /\p{InConsonant}/ ? "ok $test\n" : "not ok $test\n"; $test++;
 print "e" =~ /\P{InConsonant}/ ? "ok $test\n" : "not ok $test\n"; $test++;
 
+{
+    print "# [ID 20020630.002] utf8 regex only matches 32k\n";
+    $test = 911;
+    for ([ 'byte', "\x{ff}" ], [ 'utf8', "\x{1ff}" ]) {
+       my($type, $char) = @$_;
+       for my $len (32000, 32768, 33000) {
+           my $s = $char . "f" x $len;
+           my $r = $s =~ /$char([f]*)/gc;
+           print $r ? "ok $test\n" : "not ok $test\t# <$type x $len> fail\n";
+           ++$test;
+           print +(!$r or pos($s) == $len + 1) ? "ok $test\n"
+               : "not ok $test\t# <$type x $len> pos @{[ pos($s) ]}\n";
+           ++$test;
+       }
+    }
+}
+
+$test = 923;