Revision 112

Date:
2011/08/15 14:19:22
Author:
ahitrov
Revision Log:
Contenido::Parser object now can be made itself
Encoding patched
Files:

Legend:

 
Added
 
Removed
 
Modified

  • utf8/core/lib/Contenido/Parser.pm

     
    12 12 use Contenido::File::Scheme::FILE;
    13 13 use Contenido::Parser::Util;
    14 14
    15 sub new {
    16 my ($proto) = @_;
    17 my $class = ref($proto) || $proto;
    18 my $self = {};
    19 bless $self, $class;
    20
    21 return $self;
    22 }
    23
    24
    15 25 sub fetch {
    16 26 my ($self, $input, %opts) = @_;
    17 27
    … …  
    62 72 if ( $content ) {
    63 73 unless ( $encoding ) {
    64 74 $encoding = $self->__try_content_encoding( substr($content, 0, 350) );
    65 if ( $encoding && $encoding ne 'utf-8' ) {
    66 Encode::from_to($content, $encoding, 'utf-8');
    75 }
    76 if ( $encoding && $encoding ne 'utf-8' ) {
    77 warn "Encoding from $encoding\n..." if $DEBUG;
    78 Encode::from_to($content, $encoding, 'utf-8');
    79 if ( exists $self->{headers} ) {
    80 foreach my $header ( keys %{$self->{headers}} ) {
    81 if ( ref $self->{headers}{$header} eq 'ARRAY' ) {
    82 foreach my $val ( @{$self->{headers}{$header}} ) {
    83 Encode::from_to($val, $encoding, 'utf-8');
    84 }
    85 } else {
    86 Encode::from_to($self->{headers}{$header}, $encoding, 'utf-8');
    87 }
    88 }
    67 89 }
    68 90 }
    69 91 $self->{encoding} = $encoding;
    … …  
    92 114 my ($self, $input)= @_;
    93 115 if ( $input =~ /encoding[\ ]?=[\ ]?[\"\']?([a-z\-\d]+)/i ) {
    94 116 return lc($1);
    117 } elsif ( $input =~ /charset[\ ]?=[\ ]?[\"\']?([a-z\-\d]+)/i ) {
    118 return lc($1);
    95 119 } elsif ( $input =~ /(utf-8|windows-1251|koi8-r)/i ) {
    96 120 return lc($1);
    97 121 } else {