Integrate perlio:

[p5sagit/p5-mst-13.2.git] / t / io / utf8.t

diff --git a/t/io/utf8.t b/t/io/utf8.t
index 38d7007..337bd52 100755 (executable)
--- a/t/io/utf8.t
+++ b/t/io/utf8.t
@@ -3,14 +3,16 @@
 BEGIN {
     chdir 't' if -d 't';
     @INC = '../lib';
-    unless (exists $open::layers{'perlio'}) {
+    unless (find PerlIO::Layer 'perlio') {
        print "1..0 # Skip: not perlio\n";
        exit 0;
     }
 }
 
+no utf8; # needed for use utf8 not griping about the raw octets
+
 $| = 1;
-print "1..25\n";
+print "1..31\n";
 
 open(F,"+>:utf8",'a');
 print F chr(0x100).'£';
@@ -30,13 +32,21 @@ print "not " unless getc(F) eq "\n";
 print "ok 5\n";
 seek(F,0,0);
 binmode(F,":bytes");
-print "not " unless getc(F) eq chr(0xc4);
+my $chr = chr(0xc4);
+if (ord('A') == 193) { $chr = chr(0x8c); } # EBCDIC
+print "not " unless getc(F) eq $chr;
 print "ok 6\n";
-print "not " unless getc(F) eq chr(0x80);
+$chr = chr(0x80);
+if (ord('A') == 193) { $chr = chr(0x41); } # EBCDIC
+print "not " unless getc(F) eq $chr;
 print "ok 7\n";
-print "not " unless getc(F) eq chr(0xc2);
+$chr = chr(0xc2);
+if (ord('A') == 193) { $chr = chr(0x80); } # EBCDIC
+print "not " unless getc(F) eq $chr;
 print "ok 8\n";
-print "not " unless getc(F) eq chr(0xa3);
+$chr = chr(0xa3);
+if (ord('A') == 193) { $chr = chr(0x44); } # EBCDIC
+print "not " unless getc(F) eq $chr;
 print "ok 9\n";
 print "not " unless getc(F) eq "\n";
 print "ok 10\n";
@@ -70,7 +80,9 @@ print "ok 14\n";
 open F, "a" or die $!; # Not UTF
 $x = <F>;
 chomp($x);
-print "not " unless $x eq chr(196).chr(172);
+$chr = chr(196).chr(172);
+if (ord('A') == 193) { $chr = chr(141).chr(83); } # EBCDIC
+print "not " unless $x eq $chr;
 print "ok 15\n";
 close F;
 
@@ -99,7 +111,7 @@ print "ok 18\n";
 }
 
 { my $x = tell(F);
-    { use bytes; $y += 3;}
+    { use bytes; if (ord('A')==193){$y += 2;}else{$y += 3;}} # EBCDIC ASCII
     print "not ($x,$y) " unless $x == $y;
     print "ok 19\n";
 }
@@ -109,7 +121,9 @@ close F;
 open F, "a" or die $!; # Not UTF
 $x = <F>;
 chomp($x);
-printf "not (%vd) ", $x unless $x eq v196.172.194.130;
+$chr = v196.172.194.130;
+if (ord('A') == 193) { $chr = v141.83.130; } # EBCDIC
+printf "not (%vd) ", $x unless $x eq $chr;
 print "ok 20\n";
 
 open F, "<:utf8", "a" or die $!;
@@ -121,8 +135,13 @@ print "ok 21\n";
 
 # Now let's make it suffer.
 open F, ">", "a" or die $!;
-eval { print F $a; };
-print "not " unless $@ and $@ =~ /Wide character in print/i;
+my $w;
+{
+    use warnings 'utf8';
+    local $SIG{__WARN__} = sub { $w = $_[0] };
+    print F $a;
+}
+print "not " if ($@ || $w !~ /Wide character in print/i);
 print "ok 22\n";
 }
 
@@ -135,7 +154,9 @@ close F;
 
 open F, "<", "a" or die $!;
 $x = <F>; chomp $x;
-print "not " unless $x eq v196.172.130;
+$chr = v196.172.130;
+if (ord('A') == 193) { $chr = v141.83.130; } # EBCDIC
+print "not " unless $x eq $chr;
 print "ok 23\n";
 
 # Right.
@@ -148,15 +169,103 @@ close F;
 
 open F, "<", "a" or die $!;
 $x = <F>; chomp $x;
-print "not " unless $x eq v196.172.130;
+print "not " unless $x eq $chr;
 print "ok 24\n";
 
 # Now we have a deformed file.
-open F, "<:utf8", "a" or die $!;
-$x = <F>; chomp $x;
-{ local $SIG{__WARN__} = sub { print "ok 25\n"; };
-eval { sprintf "%vd\n", $x; }
+
+if (ord('A') == 193) {
+    print "ok 25 # Skip: EBCDIC\n"; # EBCDIC doesn't complain
+} else {
+    open F, "<:utf8", "a" or die $!;
+    $x = <F>; chomp $x;
+    local $SIG{__WARN__} = sub { print "ok 25\n" };
+    eval { sprintf "%vd\n", $x };
 }
 
+close F;
 unlink('a');
 
+open F, ">:utf8", "a";
+@a = map { chr(1 << ($_ << 2)) } 0..5; # 0x1, 0x10, .., 0x100000
+unshift @a, chr(0); # ... and a null byte in front just for fun
+print F @a;
+close F;
+
+my $c;
+
+# read() should work on characters, not bytes
+open F, "<:utf8", "a";
+$a = 0;
+for (@a) {
+    unless (($c = read(F, $b, 1) == 1)  &&
+            length($b)           == 1  &&
+            ord($b)              == ord($_) &&
+            tell(F)              == ($a += bytes::length($b))) {
+        print '# ord($_)           == ', ord($_), "\n";
+        print '# ord($b)           == ', ord($b), "\n";
+        print '# length($b)        == ', length($b), "\n";
+        print '# bytes::length($b) == ', bytes::length($b), "\n";
+        print '# tell(F)           == ', tell(F), "\n";
+        print '# $a                == ', $a, "\n";
+        print '# $c                == ', $c, "\n";
+        print "not ";
+        last;
+    }
+}
+close F;
+print "ok 26\n";
+
+{
+    # Check that warnings are on on I/O, and that they can be muffled.
+
+    local $SIG{__WARN__} = sub { $@ = shift };
+
+    undef $@;
+    open F, ">a";
+    print F chr(0x100);
+    close(F);
+
+    print $@ =~ /Wide character in print/ ? "ok 27\n" : "not ok 27\n";
+
+    undef $@;
+    open F, ">:utf8", "a";
+    print F chr(0x100);
+    close(F);
+
+    print defined $@ ? "not ok 28\n" : "ok 28\n";
+
+    undef $@;
+    open F, ">a";
+    binmode(F, ":utf8");
+    print F chr(0x100);
+    close(F);
+
+    print defined $@ ? "not ok 29\n" : "ok 29\n";
+
+    no warnings 'utf8';
+
+    undef $@;
+    open F, ">a";
+    print F chr(0x100);
+    close(F);
+
+    print defined $@ ? "not ok 30\n" : "ok 30\n";
+
+    use warnings 'utf8';
+
+    undef $@;
+    open F, ">a";
+    print F chr(0x100);
+    close(F);
+
+    print $@ =~ /Wide character in print/ ? "ok 31\n" : "not ok 31\n";
+}
+
+# sysread() and syswrite() tested in lib/open.t since Fnctl is used
+
+END {
+    1 while unlink "a";
+    1 while unlink "b";
+}
+