Make print, syswrite, send, readline, getc honour utf8-ness of PerlIO.

(sysread, recv and write i.e. formats still to do...)
Allow :utf8 or :bytes in PerlIO_apply_layers() so that
   open($fh,">:utf8","name")
etc. work. - "applying" those just sets/clears the UTF8 bit of the top layer,
so no extra overhead is involved.
Tweak t/comp/require.t to add a 'use bytes' to permit its dubious writing of BOM
to a non-utf8 stream.
Add initial io/utf8.t
Fix SvPVutf8() - sv_2pv() was not expecting to be called with something
that was already SvPOK() - (we just fossiked with SvUTF8 bit). Fix that
and also just use the SvPV macro in sv_2pvutf8() to avoid the issue/overhead.

p4raw-id: //depot/perlio@8054

2 files changed, 53 insertions, 1 deletions

diff --git a/t/comp/require.t b/t/comp/require.t
index eaea3ad5f6..e634532275 100755
--- a/t/comp/require.t
+++ b/t/comp/require.t
@@ -21,6 +21,7 @@ sub write_file {
     my $f = shift;
     open(REQ,">$f") or die "Can't write '$f': $!";
     binmode REQ;
+    use bytes;
     print REQ @_;
     close REQ;
 }
@@ -132,7 +133,7 @@ $i++; do_require(qq(${utf8}print "ok $i\n"; 1;\n));
 
 sub bytes_to_utf16 {
     my $utf16 = pack("$_[0]*", unpack("C*", $_[1]));
-    return @_ == 3 && $_[2] ? pack("$_[0]", 0xFEFF) . $utf16 : $utf16; 
+    return @_ == 3 && $_[2] ? pack("$_[0]", 0xFEFF) . $utf16 : $utf16;
 }
 
 $i++; do_require(bytes_to_utf16('n', qq(print "ok $i\\n"; 1;\n), 1)); # BE
diff --git a/t/io/utf8.t b/t/io/utf8.t
new file mode 100755
index 0000000000..1e47c33d3f
--- /dev/null
+++ b/t/io/utf8.t
@@ -0,0 +1,51 @@
+#!./perl
+
+BEGIN {
+    chdir 't' if -d 't';
+    @INC = '../lib';
+    require Config; import Config;
+    unless ($Config{'useperlio'}) {
+	print "1..0 # Skip: not perlio\n";
+	exit 0;
+    }
+}
+
+$| = 1;
+print "1..11\n";
+
+open(F,"+>:utf8",'a');
+print F chr(0x100).'£';
+print '#'.tell(F)."\n";
+print "not " unless tell(F) == 4;
+print "ok 1\n";
+print F "\n";
+print '#'.tell(F)."\n";
+print "not " unless tell(F) >= 5;
+print "ok 2\n";
+seek(F,0,0);
+print "not " unless getc(F) eq chr(0x100);
+print "ok 3\n";
+print "not " unless getc(F) eq "£";
+print "ok 4\n";
+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);
+print "ok 6\n";
+print "not " unless getc(F) eq chr(0x80);
+print "ok 7\n";
+print "not " unless getc(F) eq chr(0xc2);
+print "ok 8\n";
+print "not " unless getc(F) eq chr(0xa3);
+print "ok 9\n";
+print "not " unless getc(F) eq "\n";
+print "ok 10\n";
+seek(F,0,0);
+binmode(F,":utf8");
+print "not " unless scalar(<F>) eq "\x{100}£\n";
+print "ok 11\n";
+close(F);
+
+# unlink('a');
+