output to perl scalars must be in octets, apparently; warn if the scalar is improperl...

[debbugs.git] / Debbugs / Common.pm

diff --git a/Debbugs/Common.pm b/Debbugs/Common.pm
index 283f33ef58a6b517d62ebf362f0ccfa9fefc3b86..cf53b07e0e35344c35823496315585cd958df8ca 100644 (file)
--- a/Debbugs/Common.pm
+++ b/Debbugs/Common.pm
@@ -55,7 +55,7 @@ BEGIN{
                     lock   => [qw(filelock unfilelock lockpid)],
                    );
      @EXPORT_OK = ();
-     Exporter::export_ok_tags(qw(lock quit date util misc));
+     Exporter::export_ok_tags(keys %EXPORT_TAGS);
      $EXPORT_TAGS{all} = [@EXPORT_OK];
 }
 
@@ -70,10 +70,12 @@ use IO::Scalar;
 use Debbugs::MIME qw(decode_rfc1522);
 use Mail::Address;
 use Cwd qw(cwd);
+use Storable qw(dclone);
 
 use Params::Validate qw(validate_with :types);
 
 use Fcntl qw(:DEFAULT :flock);
+use Encode qw(is_utf8 decode_utf8);
 
 our $DEBUG_FH = \*STDERR if not defined $DEBUG_FH;
 
@@ -429,6 +431,7 @@ sub __add_to_hash {
     $type //= 'address';
     my $fh = IO::File->new($fn,'r') or
        die "Unable to open $fn for reading: $!";
+    binmode($fh,':encoding(UTF-8)');
     while (<$fh>) {
        chomp;
        next unless m/^(\S+)\s+(\S.*\S)\s*$/;
@@ -814,6 +817,10 @@ Will carp if given a scalar which isn't a scalarref or a glob (or
 globref), and return /dev/null. May return undef if IO::Scalar or
 IO::File fails. (Check $!)
 
+The scalar will fill with octets, not perl's internal encoding, so you
+must use decode_utf8() after on the scalar, and encode_utf8() on it
+before. This appears to be a bug in the underlying modules.
+
 =cut
 
 sub globify_scalar {
@@ -823,7 +830,12 @@ sub globify_scalar {
          if (defined ref($scalar)) {
               if (ref($scalar) eq 'SCALAR' and
                   not UNIVERSAL::isa($scalar,'GLOB')) {
-                   return IO::Scalar->new($scalar);
+                   if (is_utf8(${$scalar})) {
+                       ${$scalar} = decode_utf8(${$scalar});
+                       carp(q(\$scalar must not be in perl's internal encoding));
+                   }
+                   open $handle, '>:scalar:utf8', $scalar;
+                   return $handle;
               }
               else {
                    return $scalar;
@@ -836,7 +848,7 @@ sub globify_scalar {
               carp "Given a non-scalar reference, non-glob to globify_scalar; returning /dev/null handle";
          }
      }
-     return IO::File->new('/dev/null','w');
+     return IO::File->new('/dev/null','>:encoding(UTF-8)');
 }
 
 =head2 cleanup_eval_fail()
@@ -870,7 +882,7 @@ sub cleanup_eval_fail {
     # ditch the "at foo/bar/baz.pm line 5"
     $error =~ s/\sat\s\S+\sline\s\d+//;
     # ditch croak messages
-    $error =~ s/^\s{3,}.+//g;
+    $error =~ s/^\t+.+\n?//g;
     # ditch trailing multiple periods in case there was a cascade of
     # die messages.
     $error =~ s/\.+$/\./;
@@ -894,6 +906,7 @@ sub hash_slice(\%@) {
     return map {exists $hashref->{$_}?($_,$hashref->{$_}):()} @keys;
 }
 
+
 1;
 
 __END__