3 # This is free software, you may use it and distribute it under the same terms as
6 # Copyright 2001-2003 AxKit.com Ltd., 2002-2006 Christian Glahn, 2006-2009 Petr Pajas
10 package XML::LibXML::SAX::Parser;
14 use vars qw($VERSION @ISA);
17 use XML::LibXML::Common qw(:libxml);
19 use XML::SAX::DocumentLocator;
21 $VERSION = "2.0207"; # VERSION TEMPLATE: DO NOT CHANGE
22 @ISA = ('XML::SAX::Base');
25 return $XML::LibXML::__threads_shared ? 0 : 1;
28 sub _parse_characterstream {
29 my ($self, $fh, $options) = @_;
30 die "parsing a characterstream is not supported at this time";
33 sub _parse_bytestream {
34 my ($self, $fh, $options) = @_;
35 my $parser = XML::LibXML->new();
36 my $doc = exists($options->{Source}{SystemId}) ? $parser->parse_fh($fh, $options->{Source}{SystemId}) : $parser->parse_fh($fh);
37 $self->generate($doc);
41 my ($self, $str, $options) = @_;
42 my $parser = XML::LibXML->new();
43 my $doc = exists($options->{Source}{SystemId}) ? $parser->parse_string($str, $options->{Source}{SystemId}) : $parser->parse_string($str);
44 $self->generate($doc);
48 my ($self, $sysid, $options) = @_;
49 my $parser = XML::LibXML->new();
50 my $doc = $parser->parse_file($sysid);
51 $self->generate($doc);
58 my $doc = $node->ownerDocument();
60 # precompute some DocumentLocator values
67 my $dtd = defined $doc ? $doc->externalSubset() : undef;
69 $locator{PublicId} = $dtd->publicId();
70 $locator{SystemId} = $dtd->systemId();
73 $locator{Encoding} = $doc->encoding();
74 $locator{XMLVersion} = $doc->version();
76 $self->set_document_locator(
77 XML::SAX::DocumentLocator->new(
78 sub { $locator{PublicId} },
79 sub { $locator{SystemId} },
80 sub { defined($self->{current_node}) ? $self->{current_node}->line_number() : undef },
82 sub { $locator{Encoding} },
83 sub { $locator{XMLVersion} },
88 if ( $node->nodeType() == XML_DOCUMENT_NODE
89 || $node->nodeType == XML_HTML_DOCUMENT_NODE ) {
90 $self->start_document({});
91 $self->xml_decl({Version => $node->getVersion, Encoding => $node->getEncoding});
92 $self->process_node($node);
93 $self->end_document({});
98 my ($self, $node) = @_;
100 local $self->{current_node} = $node;
102 my $node_type = $node->nodeType();
103 if ($node_type == XML_COMMENT_NODE) {
104 $self->comment( { Data => $node->getData } );
106 elsif ($node_type == XML_TEXT_NODE
107 || $node_type == XML_CDATA_SECTION_NODE) {
108 # warn($node->getData . "\n");
109 $self->characters( { Data => $node->nodeValue } );
111 elsif ($node_type == XML_ELEMENT_NODE) {
112 # warn("<" . $node->getName . ">\n");
113 $self->process_element($node);
114 # warn("</" . $node->getName . ">\n");
116 elsif ($node_type == XML_ENTITY_REF_NODE) {
117 foreach my $kid ($node->childNodes) {
118 # warn("child of entity ref: " . $kid->getType() . " called: " . $kid->getName . "\n");
119 $self->process_node($kid);
122 elsif ($node_type == XML_DOCUMENT_NODE
123 || $node_type == XML_HTML_DOCUMENT_NODE
124 || $node_type == XML_DOCUMENT_FRAG_NODE) {
125 # sometimes it is just useful to generate SAX events from
126 # a document fragment (very good with filters).
127 foreach my $kid ($node->childNodes) {
128 $self->process_node($kid);
131 elsif ($node_type == XML_PI_NODE) {
132 $self->processing_instruction( { Target => $node->getName, Data => $node->getData } );
134 elsif ($node_type == XML_COMMENT_NODE) {
135 $self->comment( { Data => $node->getData } );
137 elsif ( $node_type == XML_XINCLUDE_START
138 || $node_type == XML_XINCLUDE_END ) {
140 # i may want to handle this one day, dunno yet
142 elsif ($node_type == XML_DTD_NODE ) {
144 # i will support DTDs, but had no time yet.
147 # warn("unsupported node type: $node_type");
152 sub process_element {
153 my ($self, $element) = @_;
156 my @ns_maps = $element->getNamespaces;
158 foreach my $ns (@ns_maps) {
159 $self->start_prefix_mapping(
161 NamespaceURI => $ns->href,
162 Prefix => ( defined $ns->localname ? $ns->localname : ''),
167 foreach my $attr ($element->attributes) {
169 # warn("Attr: $attr -> ", $attr->getName, " = ", $attr->getData, "\n");
170 # this isa dump thing...
171 if ($attr->isa('XML::LibXML::Namespace')) {
172 # TODO This needs fixing modulo agreeing on what
173 # is the right thing to do here.
174 unless ( defined $attr->name ) {
175 ## It's an atter like "xmlns='foo'"
176 $attribs->{"{}xmlns"} =
179 LocalName => "xmlns",
181 Value => $attr->href,
186 my $prefix = "xmlns";
187 my $localname = $attr->localname;
188 my $key = "{http://www.w3.org/2000/xmlns/}";
191 if ( defined $localname ) {
193 $name.= ":".$localname;
199 Value => $attr->href,
200 NamespaceURI => "http://www.w3.org/2000/xmlns/",
202 LocalName => $localname,
207 my $ns = $attr->namespaceURI;
209 $ns = '' unless defined $ns;
210 $key = "{$ns}".$attr->localname;
211 ## Not sure why, but $attr->name is coming through stripped
212 ## of its prefix, so we need to hand-assemble a real name.
213 my $name = $attr->name;
214 $name = "" unless defined $name;
216 my $prefix = $attr->prefix;
217 $prefix = "" unless defined $prefix;
218 $name = "$prefix:$name"
219 if index( $name, ":" ) < 0 && length $prefix;
224 Value => $attr->value,
227 LocalName => $attr->localname,
231 # warn("Attr made: ", Dumper($attribs->{$key}), "\n");
235 Name => $element->nodeName,
236 Attributes => $attribs,
237 NamespaceURI => $element->namespaceURI,
238 Prefix => $element->prefix || "",
239 LocalName => $element->localname,
242 $self->start_element($node);
244 foreach my $child ($element->childNodes) {
245 $self->process_node($child);
248 my $end_node = { %$node };
250 delete $end_node->{Attributes};
252 $self->end_element($end_node);
254 foreach my $ns (@ns_maps) {
255 $self->end_prefix_mapping(
257 NamespaceURI => $ns->href,
258 Prefix => ( defined $ns->localname ? $ns->localname : ''),