Upgrade Module-Metadata to 1.000011

9 files changed, 211 insertions, 52 deletions

diff --git a/MANIFEST b/MANIFEST
index 25f254612e..ed489cc005 100644
--- a/MANIFEST
+++ b/MANIFEST
@@ -1772,8 +1772,12 @@ cpan/Module-Load/t/to_load/Must/Be/Loaded.pm	Module::Load tests
 cpan/Module-Load/t/to_load/TestModule.pm	Module::Load tests
 cpan/Module-Load/t/to_load/ToBeLoaded		Module::Load tests
 cpan/Module-Metadata/lib/Module/Metadata.pm
+cpan/Module-Metadata/t/encoding.t
 cpan/Module-Metadata/t/lib/0_1/Foo.pm
 cpan/Module-Metadata/t/lib/0_2/Foo.pm
+cpan/Module-Metadata/t/lib/BOMTest/UTF16BE.pm
+cpan/Module-Metadata/t/lib/BOMTest/UTF16LE.pm
+cpan/Module-Metadata/t/lib/BOMTest/UTF8.pm
 cpan/Module-Metadata/t/lib/DistGen.pm
 cpan/Module-Metadata/t/lib/MBTest.pm
 cpan/Module-Metadata/t/lib/Tie/CPHash.pm
diff --git a/Porting/Maintainers.pl b/Porting/Maintainers.pl
index 4e6283cd49..ae977fee05 100755
--- a/Porting/Maintainers.pl
+++ b/Porting/Maintainers.pl
@@ -1289,7 +1289,7 @@ use File::Glob qw(:case);
 
     'Module::Metadata' => {
         'MAINTAINER'   => 'dagolden',
-        'DISTRIBUTION' => 'VPIT/Module-Metadata-1.000010.tar.gz',
+        'DISTRIBUTION' => 'APEIRON/Module-Metadata-1.000011.tar.gz',
         'FILES'        => q[cpan/Module-Metadata],
         'EXCLUDED'     => [
             qr{^maint},
diff --git a/cpan/Module-Metadata/lib/Module/Metadata.pm b/cpan/Module-Metadata/lib/Module/Metadata.pm
index ce21787193..e3c2504946 100644
--- a/cpan/Module-Metadata/lib/Module/Metadata.pm
+++ b/cpan/Module-Metadata/lib/Module/Metadata.pm
@@ -11,7 +11,7 @@ package Module::Metadata;
 
 use strict;
 use vars qw($VERSION);
-$VERSION = '1.000010';
+$VERSION = '1.000011';
 $VERSION = eval $VERSION;
 
 use Carp qw/croak/;
@@ -61,7 +61,6 @@ my $VERS_REGEXP = qr{ # match a VERSION definition
   =[^=~]  # = but not ==, nor =~
 }x;
 
-
 sub new_from_file {
   my $class    = shift;
   my $filename = File::Spec->rel2abs( shift );
@@ -440,9 +439,49 @@ sub _parse_file {
   my $fh = IO::File->new( $filename )
     or croak( "Can't open '$filename': $!" );
 
+  $self->_handle_bom($fh, $filename);
+
   $self->_parse_fh($fh);
 }
 
+# Look for a UTF-8/UTF-16BE/UTF-16LE BOM at the beginning of the stream.
+# If there's one, then skip it and set the :encoding layer appropriately.
+sub _handle_bom {
+  my ($self, $fh, $filename) = @_;
+
+  my $pos = $fh->getpos;
+  return unless defined $pos;
+
+  my $buf = ' ' x 2;
+  my $count = $fh->read( $buf, length $buf );
+  return unless defined $count and $count >= 2;
+
+  my $encoding;
+  if ( $buf eq "\x{FE}\x{FF}" ) {
+    $encoding = 'UTF-16BE';
+  } elsif ( $buf eq "\x{FF}\x{FE}" ) {
+    $encoding = 'UTF-16LE';
+  } elsif ( $buf eq "\x{EF}\x{BB}" ) {
+    $buf = ' ';
+    $count = $fh->read( $buf, length $buf );
+    if ( defined $count and $count >= 1 and $buf eq "\x{BF}" ) {
+      $encoding = 'UTF-8';
+    }
+  }
+
+  if ( defined $encoding ) {
+    if ( "$]" >= 5.008 ) {
+      # $fh->binmode requires perl 5.10
+      binmode( $fh, ":encoding($encoding)" );
+    }
+  } else {
+    $fh->setpos($pos)
+      or croak( sprintf "Can't reset position to the top of '$filename'" );
+  }
+
+  return $encoding;
+}
+
 sub _parse_fh {
   my ($self, $fh) = @_;
 
@@ -456,20 +495,21 @@ sub _parse_fh {
     my $line_num = $.;
 
     chomp( $line );
-    next if $line =~ /^\s*#/;
 
+    # From toke.c : any line that begins by "=X", where X is an alphabetic
+    # character, introduces a POD segment.
     my $is_cut;
-    if ( $line =~ /^=(.{0,3})/ ) {
-      $is_cut = $1 eq 'cut';
+    if ( $line =~ /^=([a-zA-Z].*)/ ) {
+      my $cmd = $1;
+      # Then it goes back to Perl code for "=cutX" where X is a non-alphabetic
+      # character (which includes the newline, but here we chomped it away).
+      $is_cut = $cmd =~ /^cut(?:[^a-zA-Z]|$)/;
       $in_pod = !$is_cut;
     }
 
-    # Would be nice if we could also check $in_string or something too
-    last if !$in_pod && $line =~ /^__(?:DATA|END)__$/;
-
-    if ( $in_pod || $is_cut ) {
+    if ( $in_pod ) {
 
-      if ( $line =~ /^=head\d\s+(.+)\s*$/ ) {
+      if ( $line =~ /^=head[1-4]\s+(.+)\s*$/ ) {
 	push( @pod, $1 );
 	if ( $self->{collect_pod} && length( $pod_data ) ) {
           $pod{$pod_sect} = $pod_data;
@@ -477,16 +517,26 @@ sub _parse_fh {
         }
 	$pod_sect = $1;
 
-
       } elsif ( $self->{collect_pod} ) {
 	$pod_data .= "$line\n";
 
       }
 
-    } else {
+    } elsif ( $is_cut ) {
 
+      if ( $self->{collect_pod} && length( $pod_data ) ) {
+        $pod{$pod_sect} = $pod_data;
+        $pod_data = '';
+      }
       $pod_sect = '';
-      $pod_data = '';
+
+    } else {
+
+      # Skip comments in code
+      next if $line =~ /^\s*#/;
+
+      # Would be nice if we could also check $in_string or something too
+      last if $line =~ /^__(?:DATA|END)__$/;
 
       # parse $line to see if it's a $VERSION declaration
       my( $vers_sig, $vers_fullname, $vers_pkg ) =
@@ -497,7 +547,7 @@ sub _parse_fh {
       if ( $line =~ /$PKG_REGEXP/o ) {
         $pkg = $1;
         push( @pkgs, $pkg ) unless grep( $pkg eq $_, @pkgs );
-        $vers{$pkg} = (defined $2 ? $2 : undef)  unless exists( $vers{$pkg} );
+        $vers{$pkg} = $2 unless exists( $vers{$pkg} );
         $need_vers = defined $2 ? 0 : 1;
 
       # VERSION defined with full package spec, i.e. $Module::VERSION
@@ -508,14 +558,6 @@ sub _parse_fh {
 	unless ( defined $vers{$vers_pkg} && length $vers{$vers_pkg} ) {
 	  $vers{$vers_pkg} =
 	    $self->_evaluate_version_line( $vers_sig, $vers_fullname, $line );
-	} else {
-	  # Warn unless the user is using the "$VERSION = eval
-	  # $VERSION" idiom (though there are probably other idioms
-	  # that we should watch out for...)
-	  warn <<"EOM" unless $line =~ /=\s*eval/;
-Package '$vers_pkg' already declared with version '$vers{$vers_pkg}',
-ignoring subsequent declaration on line $line_num.
-EOM
 	}
 
       # first non-comment line in undeclared package main is VERSION
@@ -541,12 +583,7 @@ EOM
 
 	unless ( defined $vers{$pkg} && length $vers{$pkg} ) {
 	  $vers{$pkg} = $v;
-	} else {
-	  warn <<"EOM";
-Package '$pkg' already declared with version '$vers{$pkg}'
-ignoring new version '$v' on line $line_num.
-EOM
-	}
+	} 
 
       }
 
@@ -738,27 +775,43 @@ without executing unsafe code.
 
 =item C<< new_from_file($filename, collect_pod => 1) >>
 
-Construct a C<Module::Metadata> object given the path to a file. Takes an
-optional argument C<collect_pod> which is a boolean that determines whether POD
-data is collected and stored for reference. POD data is not collected by
-default. POD headings are always collected.  Returns undef if the filename
-does not exist.
+Constructs a C<Module::Metadata> object given the path to a file.  Returns
+undef if the filename does not exist.
+
+C<collect_pod> is a optional boolean argument that determines whether POD
+data is collected and stored for reference.  POD data is not collected by
+default.  POD headings are always collected.
+
+If the file begins by an UTF-8, UTF-16BE or UTF-16LE byte-order mark, then
+it is skipped before processing, and the content of the file is also decoded
+appropriately starting from perl 5.8.
 
 =item C<< new_from_handle($handle, $filename, collect_pod => 1) >>
 
 This works just like C<new_from_file>, except that a handle can be provided
-as the first argument.  Note that there is no validation to confirm that the
-handle is a handle or something that can act like one.  Passing something that
-isn't a handle will cause a exception when trying to read from it.  The
-C<filename> argument is mandatory or undef will be returned.
+as the first argument.
+
+Note that there is no validation to confirm that the handle is a handle or
+something that can act like one.  Passing something that isn't a handle will
+cause a exception when trying to read from it.  The C<filename> argument is
+mandatory or undef will be returned.
+
+You are responsible for setting the decoding layers on C<$handle> if
+required.
 
 =item C<< new_from_module($module, collect_pod => 1, inc => \@dirs) >>
 
-Construct a C<Module::Metadata> object given a module or package name. In addition
-to accepting the C<collect_pod> argument as described above, this
-method accepts a C<inc> argument which is a reference to an array of
-of directories to search for the module. If none are given, the
-default is @INC.  Returns undef if the module cannot be found.
+Constructs a C<Module::Metadata> object given a module or package name.
+Returns undef if the module cannot be found.
+
+In addition to accepting the C<collect_pod> argument as described above,
+this method accepts a C<inc> argument which is a reference to an array of
+directories to search for the module.  If none are given, the default is
+@INC.
+
+If the file that contains the module begins by an UTF-8, UTF-16BE or
+UTF-16LE byte-order mark, then it is skipped before processing, and the
+content of the file is also decoded appropriately starting from perl 5.8.
 
 =item C<< find_module_by_name($module, \@dirs) >>
 
@@ -905,7 +958,7 @@ Original code from Module::Build::ModuleInfo by Ken Williams
 Released as Module::Metadata by Matt S Trout (mst) <mst@shadowcat.co.uk> with
 assistance from David Golden (xdg) <dagolden@cpan.org>.
 
-=head1 COPYRIGHT
+=head1 COPYRIGHT & LICENSE
 
 Original code Copyright (c) 2001-2011 Ken Williams.
 Additional code Copyright (c) 2010-2011 Matt Trout and David Golden.
diff --git a/cpan/Module-Metadata/t/encoding.t b/cpan/Module-Metadata/t/encoding.t
new file mode 100644
index 0000000000..a0970e051c
--- /dev/null
+++ b/cpan/Module-Metadata/t/encoding.t
@@ -0,0 +1,30 @@
+#!perl
+
+use strict;
+use File::Spec;
+use Test::More;
+
+use Module::Metadata;
+
+if ("$]" < 5.008_003) {
+  plan skip_all => 'Encoding test needs at least perl 5.8.3';
+}
+
+my %versions = (
+ UTF8    => 3,
+ UTF16BE => 4,
+ UTF16LE => 5,
+);
+
+plan tests => 4 * scalar(keys %versions);
+
+for my $enc (sort keys %versions) {
+  my $pkg  = "BOMTest::$enc";
+  my $vers = $versions{$enc};
+  my $pm   = File::Spec->catfile(qw<t lib BOMTest> => "$enc.pm");
+  my $info = Module::Metadata->new_from_file($pm);
+  is( $info->name, $pkg, "$enc: default package was found" );
+  is( $info->version, $vers, "$enc: version for default package" );
+  is( $info->version('Heart'), '1', 'version for ASCII package' );
+  is( $info->version("C\x{153}ur"), '2', 'version for Unicode package' );
+}
diff --git a/cpan/Module-Metadata/t/lib/BOMTest/UTF16BE.pm b/cpan/Module-Metadata/t/lib/BOMTest/UTF16BE.pm
new file mode 100644
index 0000000000..17c6a4ad65
--- /dev/null
+++ b/cpan/Module-Metadata/t/lib/BOMTest/UTF16BE.pm
diff --git a/cpan/Module-Metadata/t/lib/BOMTest/UTF16LE.pm b/cpan/Module-Metadata/t/lib/BOMTest/UTF16LE.pm
new file mode 100644
index 0000000000..a46de6de4e
--- /dev/null
+++ b/cpan/Module-Metadata/t/lib/BOMTest/UTF16LE.pm
diff --git a/cpan/Module-Metadata/t/lib/BOMTest/UTF8.pm b/cpan/Module-Metadata/t/lib/BOMTest/UTF8.pm
new file mode 100644
index 0000000000..9062ac6075
--- /dev/null
+++ b/cpan/Module-Metadata/t/lib/BOMTest/UTF8.pm
@@ -0,0 +1,13 @@
+package Heart;
+
+our $VERSION = 1;
+
+package BOMTest::UTF8;
+
+our $VERSION = 3;
+
+package Cœur;
+
+our $VERSION = 2;
+
+1;
diff --git a/cpan/Module-Metadata/t/metadata.t b/cpan/Module-Metadata/t/metadata.t
index 2b2d58ec0d..b7adb1ec4b 100644
--- a/cpan/Module-Metadata/t/metadata.t
+++ b/cpan/Module-Metadata/t/metadata.t
@@ -28,6 +28,15 @@ package Simple;
 our $VERSION;
 $VERSION = '1.23';
 ---
+  '1.23' => <<'---', # commented & defined on same line
+package Simple;
+our $VERSION = '1.23'; # our $VERSION = '4.56';
+---
+  '1.23' => <<'---', # commented & defined on separate lines
+package Simple;
+# our $VERSION = '4.56';
+our $VERSION = '1.23';
+---
   '1.23' => <<'---', # use vars
 package Simple;
 use vars qw( $VERSION );
@@ -203,7 +212,7 @@ package Simple v1.2.3_4 {
 );
 my %modules = reverse @modules;
 
-plan tests => 51 + 2 * keys( %modules );
+plan tests => 54 + 2 * keys( %modules );
 
 require_ok('Module::Metadata');
 
@@ -420,6 +429,9 @@ Simple - It's easy.
 
 Simple Simon
 
+You can find me on the IRC channel
+#simon on irc.perl.org.
+
 =cut
 ---
 $dist->regen;
@@ -459,13 +471,59 @@ is( $pm_info->pod('NAME'), undef,
 $pm_info = Module::Metadata->new_from_module(
              $dist->name, inc => [ 'lib', @INC ], collect_pod => 1 );
 
-my $name = $pm_info->pod('NAME');
-if ( $name ) {
-  $name =~ s/^\s+//;
-  $name =~ s/\s+$//;
+{
+  my %pod;
+  for my $section (qw(NAME AUTHOR)) {
+    my $content = $pm_info->pod( $section );
+    if ( $content ) {
+      $content =~ s/^\s+//;
+      $content =~ s/\s+$//;
+    }
+    $pod{$section} = $content;
+  }
+  my %expected = (
+    NAME   => q|Simple - It's easy.|,
+    AUTHOR => <<'EXPECTED'
+Simple Simon
+
+You can find me on the IRC channel
+#simon on irc.perl.org.
+EXPECTED
+  );
+  for my $text (values %expected) {
+    $text =~ s/^\s+//;
+    $text =~ s/\s+$//;
+  }
+  is( $pod{NAME},   $expected{NAME},   'collected NAME pod section' );
+  is( $pod{AUTHOR}, $expected{AUTHOR}, 'collected AUTHOR pod section' );
 }
-is( $name, q|Simple - It's easy.|, 'collected pod section' );
 
+{
+  # test things that look like POD, but aren't
+$dist->change_file( 'lib/Simple.pm', <<'---' );
+package Simple;
+
+=YES THIS STARTS POD
+
+our $VERSION = '999';
+
+=cute
+
+our $VERSION = '666';
+
+=cut
+
+*foo
+=*no_this_does_not_start_pod;
+
+our $VERSION = '1.23';
+
+---
+  $dist->regen;
+  $pm_info = Module::Metadata->new_from_file('lib/Simple.pm');
+  is( $pm_info->name, 'Simple', 'found default package' );
+  is( $pm_info->version, '1.23', 'version for default package' );
+}
 
 {
   # Make sure processing stops after __DATA__
diff --git a/pod/perldelta.pod b/pod/perldelta.pod
index 101b6029c8..4854d0e41c 100644
--- a/pod/perldelta.pod
+++ b/pod/perldelta.pod
@@ -201,9 +201,10 @@ Various enhancements include the new use of Module::Metadata.
 
 =item *
 
-L<Module::Metadata> has been upgraded from version 1.000009 to 1.000010.  The
+L<Module::Metadata> has been upgraded from version 1.000009 to 1.000011.  The
 creation of a Module::Metadata object for a typical module file has been sped
-up by about 40%.
+up by about 40%, and some spurious warnings about C<$VERSION>s have been
+suppressed.
 
 =item *