1N/A#!./perl
1N/A
1N/ABEGIN {
1N/A chdir 't' if -d 't';
1N/A @INC = '../lib';
1N/A unless (find PerlIO::Layer 'perlio') {
1N/A print "1..0 # Skip: not perlio\n";
1N/A exit 0;
1N/A }
1N/A}
1N/A
1N/Ano utf8; # needed for use utf8 not griping about the raw octets
1N/A
1N/Arequire "./test.pl";
1N/A
1N/Aplan(tests => 53);
1N/A
1N/A$| = 1;
1N/A
1N/Aopen(F,"+>:utf8",'a');
1N/Aprint F chr(0x100).'�';
1N/Aok( tell(F) == 4, tell(F) );
1N/Aprint F "\n";
1N/Aok( tell(F) >= 5, tell(F) );
1N/Aseek(F,0,0);
1N/Aok( getc(F) eq chr(0x100) );
1N/Aok( getc(F) eq "�" );
1N/Aok( getc(F) eq "\n" );
1N/Aseek(F,0,0);
1N/Abinmode(F,":bytes");
1N/Amy $chr = chr(0xc4);
1N/Aif (ord('A') == 193) { $chr = chr(0x8c); } # EBCDIC
1N/Aok( getc(F) eq $chr );
1N/A$chr = chr(0x80);
1N/Aif (ord('A') == 193) { $chr = chr(0x41); } # EBCDIC
1N/Aok( getc(F) eq $chr );
1N/A$chr = chr(0xc2);
1N/Aif (ord('A') == 193) { $chr = chr(0x80); } # EBCDIC
1N/Aok( getc(F) eq $chr );
1N/A$chr = chr(0xa3);
1N/Aif (ord('A') == 193) { $chr = chr(0x44); } # EBCDIC
1N/Aok( getc(F) eq $chr );
1N/Aok( getc(F) eq "\n" );
1N/Aseek(F,0,0);
1N/Abinmode(F,":utf8");
1N/Aok( scalar(<F>) eq "\x{100}�\n" );
1N/Aseek(F,0,0);
1N/A$buf = chr(0x200);
1N/A$count = read(F,$buf,2,1);
1N/Aok( $count == 2 );
1N/Aok( $buf eq "\x{200}\x{100}�" );
1N/Aclose(F);
1N/A
1N/A{
1N/A $a = chr(300); # This *is* UTF-encoded
1N/A $b = chr(130); # This is not.
1N/A
1N/A open F, ">:utf8", 'a' or die $!;
1N/A print F $a,"\n";
1N/A close F;
1N/A
1N/A open F, "<:utf8", 'a' or die $!;
1N/A $x = <F>;
1N/A chomp($x);
1N/A ok( $x eq chr(300) );
1N/A
1N/A open F, "a" or die $!; # Not UTF
1N/A binmode(F, ":bytes");
1N/A $x = <F>;
1N/A chomp($x);
1N/A $chr = chr(196).chr(172);
1N/A if (ord('A') == 193) { $chr = chr(141).chr(83); } # EBCDIC
1N/A ok( $x eq $chr );
1N/A close F;
1N/A
1N/A open F, ">:utf8", 'a' or die $!;
1N/A binmode(F); # we write a "\n" and then tell() - avoid CRLF issues.
1N/A binmode(F,":utf8"); # turn UTF-8-ness back on
1N/A print F $a;
1N/A my $y;
1N/A { my $x = tell(F);
1N/A { use bytes; $y = length($a);}
1N/A ok( $x == $y );
1N/A }
1N/A
1N/A { # Check byte length of $b
1N/A use bytes; my $y = length($b);
1N/A ok( $y == 1 );
1N/A }
1N/A
1N/A print F $b,"\n"; # Don't upgrades $b
1N/A
1N/A { # Check byte length of $b
1N/A use bytes; my $y = length($b);
1N/A ok( $y == 1 );
1N/A }
1N/A
1N/A {
1N/A my $x = tell(F);
1N/A { use bytes; if (ord('A')==193){$y += 2;}else{$y += 3;}} # EBCDIC ASCII
1N/A ok( $x == $y );
1N/A }
1N/A
1N/A close F;
1N/A
1N/A open F, "a" or die $!; # Not UTF
1N/A binmode(F, ":bytes");
1N/A $x = <F>;
1N/A chomp($x);
1N/A $chr = v196.172.194.130;
1N/A if (ord('A') == 193) { $chr = v141.83.130; } # EBCDIC
1N/A ok( $x eq $chr, sprintf('(%vd)', $x) );
1N/A
1N/A open F, "<:utf8", "a" or die $!;
1N/A $x = <F>;
1N/A chomp($x);
1N/A close F;
1N/A ok( $x eq chr(300).chr(130), sprintf('(%vd)', $x) );
1N/A
1N/A open F, ">", "a" or die $!;
1N/A if (${^OPEN} =~ /:utf8/) {
1N/A binmode(F, ":bytes:");
1N/A }
1N/A
1N/A # Now let's make it suffer.
1N/A my $w;
1N/A {
1N/A use warnings 'utf8';
1N/A local $SIG{__WARN__} = sub { $w = $_[0] };
1N/A print F $a;
1N/A ok( !($@ || $w !~ /Wide character in print/i) );
1N/A }
1N/A}
1N/A
1N/A# Hm. Time to get more evil.
1N/Aopen F, ">:utf8", "a" or die $!;
1N/Aprint F $a;
1N/Abinmode(F, ":bytes");
1N/Aprint F chr(130)."\n";
1N/Aclose F;
1N/A
1N/Aopen F, "<", "a" or die $!;
1N/Abinmode(F, ":bytes");
1N/A$x = <F>; chomp $x;
1N/A$chr = v196.172.130;
1N/Aif (ord('A') == 193) { $chr = v141.83.130; } # EBCDIC
1N/Aok( $x eq $chr );
1N/A
1N/A# Right.
1N/Aopen F, ">:utf8", "a" or die $!;
1N/Aprint F $a;
1N/Aclose F;
1N/Aopen F, ">>", "a" or die $!;
1N/Aprint F chr(130)."\n";
1N/Aclose F;
1N/A
1N/Aopen F, "<", "a" or die $!;
1N/A$x = <F>; chomp $x;
1N/Aok( $x eq $chr );
1N/A
1N/A# Now we have a deformed file.
1N/A
1N/ASKIP: {
1N/A if (ord('A') == 193) {
1N/A skip( "EBCDIC doesn't complain" );
1N/A } else {
1N/A open F, "<:utf8", "a" or die $!;
1N/A $x = <F>; chomp $x;
1N/A local $SIG{__WARN__} = sub { ok( 1 ) };
1N/A eval { sprintf "%vd\n", $x };
1N/A }
1N/A}
1N/A
1N/Aclose F;
1N/Aunlink('a');
1N/A
1N/Aopen F, ">:utf8", "a";
1N/A@a = map { chr(1 << ($_ << 2)) } 0..5; # 0x1, 0x10, .., 0x100000
1N/Aunshift @a, chr(0); # ... and a null byte in front just for fun
1N/Aprint F @a;
1N/Aclose F;
1N/A
1N/Amy $c;
1N/A
1N/A# read() should work on characters, not bytes
1N/Aopen F, "<:utf8", "a";
1N/A$a = 0;
1N/Afor (@a) {
1N/A unless (($c = read(F, $b, 1) == 1) &&
1N/A length($b) == 1 &&
1N/A ord($b) == ord($_) &&
1N/A tell(F) == ($a += bytes::length($b))) {
1N/A print '# ord($_) == ', ord($_), "\n";
1N/A print '# ord($b) == ', ord($b), "\n";
1N/A print '# length($b) == ', length($b), "\n";
1N/A print '# bytes::length($b) == ', bytes::length($b), "\n";
1N/A print '# tell(F) == ', tell(F), "\n";
1N/A print '# $a == ', $a, "\n";
1N/A print '# $c == ', $c, "\n";
1N/A print "not ";
1N/A last;
1N/A }
1N/A}
1N/Aclose F;
1N/Aok( 1 );
1N/A
1N/A{
1N/A # Check that warnings are on on I/O, and that they can be muffled.
1N/A
1N/A local $SIG{__WARN__} = sub { $@ = shift };
1N/A
1N/A undef $@;
1N/A open F, ">a";
1N/A binmode(F, ":bytes");
1N/A print F chr(0x100);
1N/A close(F);
1N/A
1N/A like( $@, 'Wide character in print' );
1N/A
1N/A undef $@;
1N/A open F, ">:utf8", "a";
1N/A print F chr(0x100);
1N/A close(F);
1N/A
1N/A isnt( defined $@ );
1N/A
1N/A undef $@;
1N/A open F, ">a";
1N/A binmode(F, ":utf8");
1N/A print F chr(0x100);
1N/A close(F);
1N/A
1N/A isnt( defined $@ );
1N/A
1N/A no warnings 'utf8';
1N/A
1N/A undef $@;
1N/A open F, ">a";
1N/A print F chr(0x100);
1N/A close(F);
1N/A
1N/A isnt( defined $@ );
1N/A
1N/A use warnings 'utf8';
1N/A
1N/A undef $@;
1N/A open F, ">a";
1N/A binmode(F, ":bytes");
1N/A print F chr(0x100);
1N/A close(F);
1N/A
1N/A like( $@, 'Wide character in print' );
1N/A}
1N/A
1N/A{
1N/A open F, ">:bytes","a"; print F "\xde"; close F;
1N/A
1N/A open F, "<:bytes", "a";
1N/A my $b = chr 0x100;
1N/A $b .= <F>;
1N/A ok( $b eq chr(0x100).chr(0xde), "21395 '.= <>' utf8 vs. bytes" );
1N/A close F;
1N/A}
1N/A
1N/A{
1N/A open F, ">:utf8","a"; print F chr 0x100; close F;
1N/A
1N/A open F, "<:utf8", "a";
1N/A my $b = "\xde";
1N/A $b .= <F>;
1N/A ok( $b eq chr(0xde).chr(0x100), "21395 '.= <>' bytes vs. utf8" );
1N/A close F;
1N/A}
1N/A
1N/A{
1N/A my @a = ( [ 0x007F, "bytes" ],
1N/A [ 0x0080, "bytes" ],
1N/A [ 0x0080, "utf8" ],
1N/A [ 0x0100, "utf8" ] );
1N/A my $t = 34;
1N/A for my $u (@a) {
1N/A for my $v (@a) {
1N/A # print "# @$u - @$v\n";
1N/A open F, ">a";
1N/A binmode(F, ":" . $u->[1]);
1N/A print F chr($u->[0]);
1N/A close F;
1N/A
1N/A open F, "<a";
1N/A binmode(F, ":" . $u->[1]);
1N/A
1N/A my $s = chr($v->[0]);
1N/A utf8::upgrade($s) if $v->[1] eq "utf8";
1N/A
1N/A $s .= <F>;
1N/A ok( $s eq chr($v->[0]) . chr($u->[0]), 'rcatline utf8' );
1N/A close F;
1N/A $t++;
1N/A }
1N/A }
1N/A # last test here 49
1N/A}
1N/A
1N/A{
1N/A # [perl #23428] Somethings rotten in unicode semantics
1N/A open F, ">a";
1N/A binmode F, ":utf8";
1N/A syswrite(F, $a = chr(0x100));
1N/A close F;
1N/A is( ord($a), 0x100, '23428 syswrite should not downgrade scalar' );
1N/A like( $a, qr/^\w+/, '23428 syswrite should not downgrade scalar' );
1N/A}
1N/A
1N/A# sysread() and syswrite() tested in lib/open.t since Fcntl is used
1N/A
1N/A{
1N/A # <FH> on a :utf8 stream should complain immediately with -w
1N/A # if it finds bad UTF-8 (:encoding(utf8) works this way)
1N/A use warnings 'utf8';
1N/A undef $@;
1N/A local $SIG{__WARN__} = sub { $@ = shift };
1N/A open F, ">a";
1N/A binmode F;
1N/A my ($chrE4, $chrF6) = (chr(0xE4), chr(0xF6));
1N/A if (ord('A') == 193) # EBCDIC
1N/A { ($chrE4, $chrF6) = (chr(0x43), chr(0xEC)); }
1N/A print F "foo", $chrE4, "\n";
1N/A print F "foo", $chrF6, "\n";
1N/A close F;
1N/A open F, "<:utf8", "a";
1N/A undef $@;
1N/A my $line = <F>;
1N/A my ($chrE4, $chrF6) = ("E4", "F6");
1N/A if (ord('A') == 193) { ($chrE4, $chrF6) = ("43", "EC"); } # EBCDIC
1N/A like( $@, qr/utf8 "\\x$chrE4" does not map to Unicode .+ <F> line 1/,
1N/A "<:utf8 readline must warn about bad utf8");
1N/A undef $@;
1N/A $line .= <F>;
1N/A like( $@, qr/utf8 "\\x$chrF6" does not map to Unicode .+ <F> line 2/,
1N/A "<:utf8 rcatline must warn about bad utf8");
1N/A close F;
1N/A}
1N/A
1N/AEND {
1N/A 1 while unlink "a";
1N/A 1 while unlink "b";
1N/A}