Update Parse::CPAN::Meta to 0.04_01

5 files changed, 46 insertions, 8 deletions

diff --git a/lib/Parse/CPAN/Meta.pm b/lib/Parse/CPAN/Meta.pm
index 8d5e29b5ef..6a80681794 100644
--- a/lib/Parse/CPAN/Meta.pm
+++ b/lib/Parse/CPAN/Meta.pm
@@ -5,7 +5,7 @@ use Carp 'croak';
 BEGIN {
 	require 5.004;
 	require Exporter;
-	$Parse::CPAN::Meta::VERSION   = '0.04';
+	$Parse::CPAN::Meta::VERSION   = '0.04_01';
 	@Parse::CPAN::Meta::ISA       = qw{ Exporter      };
 	@Parse::CPAN::Meta::EXPORT_OK = qw{ Load LoadFile };
 }
@@ -25,7 +25,19 @@ my %UNESCAPES = (
 );
 
 
-
+my %BOM = (                                                       
+	"\357\273\277" => 'UTF-8',                                    
+	"\376\377"     => 'UTF-16BE',                                 
+	"\377\376"     => 'UTF-16LE',                                 
+	"\0\0\376\377" => 'UTF-32BE',                                 
+	"\377\376\0\0" => 'UTF-32LE'                                  
+);                                                                
+                                                                  
+sub BOM_MIN_LENGTH () { 2 }                                       
+sub BOM_MAX_LENGTH () { 4 }                                       
+sub HAVE_UTF8      () { $] >= 5.007003 }                          
+                                                                  
+BEGIN { require utf8 if HAVE_UTF8 }
 
 
 #####################################################################
@@ -53,17 +65,32 @@ sub LoadFile ($) {
 # Parse a document from a string.
 # Doing checks on $_[0] prevents us having to do a string copy.
 sub Load ($) {
-	unless ( defined $_[0] ) {
+
+	my $str = $_[0];
+
+	# Handle special cases
+	foreach my $length ( BOM_MIN_LENGTH .. BOM_MAX_LENGTH ) {
+		if ( my $enc = $BOM{substr($str, 0, $length)} ) {
+			croak("Stream has a non UTF-8 BOM") unless $enc eq 'UTF-8';
+			substr($str, 0, $length) = ''; # strip UTF-8 bom if found, we'll just ignore it
+		}
+	}
+
+	if ( HAVE_UTF8 ) {
+		utf8::decode($str); # try to decode as utf8
+	}
+
+	unless ( defined $str ) {
 		croak("Did not provide a string to Load");
 	}
-	return () unless length $_[0];
-	unless ( $_[0] =~ /[\012\015]+$/ ) {
+	return() unless length $str;
+	unless ( $str =~ /[\012\015]+$/ ) {
 		croak("Stream does not end with newline character");
 	}
 
 	# Split the file into lines
 	my @lines = grep { ! /^\s*(?:\#.*)?$/ }
-	            split /(?:\015{1,2}\012|\015|\012)/, shift;
+	            split /(?:\015{1,2}\012|\015|\012)/, $str;
 
 	# A nibbling parser
 	my @documents = ();
diff --git a/lib/Parse/CPAN/Meta/Changes b/lib/Parse/CPAN/Meta/Changes
index 9bd147c7f8..5a166519a1 100644
--- a/lib/Parse/CPAN/Meta/Changes
+++ b/lib/Parse/CPAN/Meta/Changes
@@ -1,5 +1,8 @@
 Revision history for Perl extension Parse-CPAN-Meta
 
+0.04_01 Tue  3 Mar 2009
+	- Merge Unicode / BOM fixes from YAML::Tiny
+
 0.04 Wed  7 Jan 2009
 	- Matching changes in YAML::Tiny 1.36
 	- Fixing missing feature reported by H.Merijn Brand
diff --git a/lib/Parse/CPAN/Meta/t/11_meta_yml.t b/lib/Parse/CPAN/Meta/t/11_meta_yml.t
index 9b789f04c8..c5a493467d 100644
--- a/lib/Parse/CPAN/Meta/t/11_meta_yml.t
+++ b/lib/Parse/CPAN/Meta/t/11_meta_yml.t
@@ -95,7 +95,7 @@ generated_by: ExtUtils::MakeMaker version 6.30
 END_YAML
 	[ {
 		name              => 'ITS-SIN-FIDS-Content-XML',
-		version           => 0.01,
+		version           => "0.01",
 		version_from      => 'lib/ITS/SIN/FIDS/Content/XML.pm',
 		installdirs       => 'site',
 		requires          => {
diff --git a/lib/Parse/CPAN/Meta/t/15_multibyte.t b/lib/Parse/CPAN/Meta/t/15_multibyte.t
index c5e0d67147..a67c0f989a 100644
--- a/lib/Parse/CPAN/Meta/t/15_multibyte.t
+++ b/lib/Parse/CPAN/Meta/t/15_multibyte.t
@@ -20,7 +20,7 @@ BEGIN {
 
 use File::Spec::Functions ':ALL';
 use Parse::CPAN::Meta::Test;
-use Test::More tests(0, 1, 3);
+use Test::More tests(0, 1, 4);
 
 
 
@@ -46,3 +46,11 @@ SKIP: {
 		'XSLoader'   => 0,
 	}, 'build_requires ok' );
 }
+
+SKIP: {
+	skip "no utf8 support", 1 unless Parse::CPAN::Meta::HAVE_UTF8();
+	ok( utf8::is_utf8($yaml[0]->{author}), "utf8 decoded" );
+}
+
+exit(0);
+
diff --git a/lib/Parse/CPAN/Meta/t/data/utf_16_le_bom.yml b/lib/Parse/CPAN/Meta/t/data/utf_16_le_bom.yml
new file mode 100644
index 0000000000..b9230ebb5a
--- /dev/null
+++ b/lib/Parse/CPAN/Meta/t/data/utf_16_le_bom.yml