lib/XML/LibXML/SAX.pm

   1 # $Id$
   2 #
   3 # This is free software, you may use it and distribute it under the same terms as
   4 # Perl itself.
   5 #
   6 # Copyright 2001-2003 AxKit.com Ltd., 2002-2006 Christian Glahn, 2006-2009 Petr Pajas
   7 #
   8 #
   9
  10 package XML::LibXML::SAX;
  11
  12 use strict;
  13 use warnings;
  14
  15 use vars qw($VERSION @ISA);
  16
  17 $VERSION = "2.0207"; # VERSION TEMPLATE: DO NOT CHANGE
  18
  19 use XML::LibXML;
  20 use XML::SAX::Base;
  21
  22 use parent qw(XML::SAX::Base);
  23
  24 use Carp;
  25 use IO::File;
  26
  27 sub CLONE_SKIP {
  28   return $XML::LibXML::__threads_shared ? 0 : 1;
  29 }
  30
  31 sub set_feature {
  32         my ($self, $feat, $val) = @_;
  33
  34         if ($feat eq 'http://xmlns.perl.org/sax/join-character-data') {
  35                 $self->{JOIN_CHARACTERS} = $val;
  36                 return 1;
  37         }
  38
  39         shift(@_);
  40         return $self->SUPER::set_feature(@_);
  41 }
  42
  43 sub _parse_characterstream {
  44     my ( $self, $fh ) = @_;
  45     # this my catch the xml decl, so the parser won't get confused about
  46     # a possibly wrong encoding.
  47     croak( "not implemented yet" );
  48 }
  49
  50 # See:
  51 # https://rt.cpan.org/Public/Bug/Display.html?id=132759
  52 sub _calc_new_XML_LibXML_parser_for_compatibility_with_XML_Simple_etc
  53 {
  54     return XML::LibXML->new( expand_entities => 1, );
  55 }
  56
  57 sub _parse_bytestream {
  58     my ( $self, $fh ) = @_;
  59     $self->{ParserOptions}{LibParser}      = $self->_calc_new_XML_LibXML_parser_for_compatibility_with_XML_Simple_etc() unless defined $self->{ParserOptions}{LibParser};
  60     $self->{ParserOptions}{ParseFunc}      = \&XML::LibXML::parse_fh;
  61     $self->{ParserOptions}{ParseFuncParam} = $fh;
  62     $self->_parse;
  63     return $self->end_document({});
  64 }
  65
  66 sub _parse_string {
  67     my ( $self, $string ) = @_;
  68     $self->{ParserOptions}{LibParser}      = $self->_calc_new_XML_LibXML_parser_for_compatibility_with_XML_Simple_etc() unless defined $self->{ParserOptions}{LibParser};
  69     $self->{ParserOptions}{ParseFunc}      = \&XML::LibXML::parse_string;
  70     $self->{ParserOptions}{ParseFuncParam} = $string;
  71     $self->_parse;
  72     return $self->end_document({});
  73 }
  74
  75 sub _parse_systemid {
  76     my $self = shift;
  77     $self->{ParserOptions}{LibParser}      = $self->_calc_new_XML_LibXML_parser_for_compatibility_with_XML_Simple_etc() unless defined $self->{ParserOptions}{LibParser};
  78     $self->{ParserOptions}{ParseFunc}      = \&XML::LibXML::parse_file;
  79     $self->{ParserOptions}{ParseFuncParam} = shift;
  80     $self->_parse;
  81     return $self->end_document({});
  82 }
  83
  84 sub parse_chunk {
  85     my ( $self, $chunk ) = @_;
  86     $self->{ParserOptions}{LibParser}      = $self->_calc_new_XML_LibXML_parser_for_compatibility_with_XML_Simple_etc() unless defined $self->{ParserOptions}{LibParser};
  87     $self->{ParserOptions}{ParseFunc}      = \&XML::LibXML::parse_xml_chunk;
  88     $self->{ParserOptions}{LibParser}->{IS_FILTER}=1; # a hack to prevent parse_xml_chunk from issuing end_document
  89     $self->{ParserOptions}{ParseFuncParam} = $chunk;
  90     $self->_parse;
  91     return;
  92 }
  93
  94 sub _parse {
  95     my $self = shift;
  96     my $args = bless $self->{ParserOptions}, ref($self);
  97
  98     if (defined($self->{JOIN_CHARACTERS})) {
  99         $args->{LibParser}->{JOIN_CHARACTERS} = $self->{JOIN_CHARACTERS};
 100     } else {
 101         $args->{LibParser}->{JOIN_CHARACTERS} = 0;
 102     }
 103
 104     $args->{LibParser}->set_handler( $self );
 105     eval {
 106       $args->{ParseFunc}->($args->{LibParser}, $args->{ParseFuncParam});
 107     };
 108
 109     if ( $args->{LibParser}->{SAX}->{State} == 1 ) {
 110         croak( "SAX Exception not implemented, yet; Data ended before document ended\n" );
 111     }
 112
 113     # break a possible circular reference
 114     $args->{LibParser}->set_handler( undef );
 115     if ( $@ ) {
 116         croak $@;
 117     }
 118     return;
 119 }
 120
 121 1;
 122