feedwordpress / feedwordpress_parser.class.php

feedwordpress_parser.class.php in FeedWordPress 2012.1212, at feedwordpress_parser.class.php

221 lines 5.8 KB

Raw Download Zip

1	<?php
2	class FeedWordPress_Parser extends SimplePie_Parser {
3	function parse (&$data, $encoding) {
4	$data = apply_filters('feedwordpress_parser_parse', $data, $encoding, $this);
5
6	// Use UTF-8 if we get passed US-ASCII, as every US-ASCII character is a UTF-8 character
7	if (strtoupper($encoding) === 'US-ASCII')
8	{
9	$this->encoding = 'UTF-8';
10	}
11	else
12	{
13	$this->encoding = $encoding;
14	}
15
16	// Strip BOM:
17	// UTF-32 Big Endian BOM
18	if (substr($data, 0, 4) === "\x00\x00\xFE\xFF")
19	{
20	$data = substr($data, 4);
21	}
22	// UTF-32 Little Endian BOM
23	elseif (substr($data, 0, 4) === "\xFF\xFE\x00\x00")
24	{
25	$data = substr($data, 4);
26	}
27	// UTF-16 Big Endian BOM
28	elseif (substr($data, 0, 2) === "\xFE\xFF")
29	{
30	$data = substr($data, 2);
31	}
32	// UTF-16 Little Endian BOM
33	elseif (substr($data, 0, 2) === "\xFF\xFE")
34	{
35	$data = substr($data, 2);
36	}
37	// UTF-8 BOM
38	elseif (substr($data, 0, 3) === "\xEF\xBB\xBF")
39	{
40	$data = substr($data, 3);
41	}
42
43	if (substr($data, 0, 5) === '<?xml' && strspn(substr($data, 5, 1), "\x09\x0A\x0D\x20") && ($pos = strpos($data, '?>')) !== false)
44	{
45	$declaration = new SimplePie_XML_Declaration_Parser(substr($data, 5, $pos - 5));
46	if ($declaration->parse())
47	{
48	$data = substr($data, $pos + 2);
49	$data = '<?xml version="' . $declaration->version . '" encoding="' . $encoding . '" standalone="' . (($declaration->standalone) ? 'yes' : 'no') . '"?>' . $data;
50	}
51	else
52	{
53	$this->error_string = 'SimplePie bug! Please report this!';
54	return false;
55	}
56	}
57
58	$return = true;
59
60	static $xml_is_sane = null;
61	if ($xml_is_sane === null)
62	{
63	$parser_check = xml_parser_create();
64	xml_parse_into_struct($parser_check, '<foo>&</foo>', $values);
65	xml_parser_free($parser_check);
66	$xml_is_sane = isset($values[0]['value']);
67	}
68
69	// Create the parser
70	if ($xml_is_sane)
71	{
72	$xml = xml_parser_create_ns($this->encoding, $this->separator);
73	xml_parser_set_option($xml, XML_OPTION_SKIP_WHITE, 1);
74	xml_parser_set_option($xml, XML_OPTION_CASE_FOLDING, 0);
75	xml_set_object($xml, $this);
76	xml_set_character_data_handler($xml, 'cdata');
77	xml_set_element_handler($xml, 'tag_open', 'tag_close');
78	xml_set_start_namespace_decl_handler($xml, 'start_xmlns');
79
80	// Parse!
81	if (!xml_parse($xml, $data, true))
82	{
83	if (class_exists('DOMDocument')) :
84	libxml_use_internal_errors(true);
85	$doc = new DOMDocument;
86	$doc->loadHTML('<html>'.$data.'</html>');
87	///DBG/ echo "<plaintext>";
88	///DBG/ $dd = $doc->getElementsByTagName('html');
89	///DBG/ for ($i = 0; $i < $dd->length; $i++) :
90	///DBG/ var_dump($dd->item($i)->nodeName);
91	///DBG/ endfor;
92	///DBG/ var_dump($doc);
93	libxml_use_internal_errors(false);
94	///DBG/ die;
95	endif;
96
97	$this->error_code = xml_get_error_code($xml);
98	$this->error_string = xml_error_string($this->error_code);
99	///DBG/ echo "WOOGA WOOGA"; var_dump($this->error_string); die;
100	$return = false;
101	}
102
103	$this->current_line = xml_get_current_line_number($xml);
104	$this->current_column = xml_get_current_column_number($xml);
105	$this->current_byte = xml_get_current_byte_index($xml);
106	xml_parser_free($xml);
107
108	return $return;
109	}
110	else
111	{
112	libxml_clear_errors();
113	$xml = new XMLReader();
114	$xml->xml($data);
115	while (@$xml->read())
116	{
117	switch ($xml->nodeType)
118	{
119
120	case constant('XMLReader::END_ELEMENT'):
121	if ($xml->namespaceURI !== '')
122	{
123	$tagName = "{$xml->namespaceURI}{$this->separator}{$xml->localName}";
124	}
125	else
126	{
127	$tagName = $xml->localName;
128	}
129	$this->tag_close(null, $tagName);
130	break;
131	case constant('XMLReader::ELEMENT'):
132	$empty = $xml->isEmptyElement;
133	if ($xml->namespaceURI !== '')
134	{
135	$tagName = "{$xml->namespaceURI}{$this->separator}{$xml->localName}";
136	}
137	else
138	{
139	$tagName = $xml->localName;
140	}
141	$attributes = array();
142	while ($xml->moveToNextAttribute())
143	{
144	if ($xml->namespaceURI !== '')
145	{
146	$attrName = "{$xml->namespaceURI}{$this->separator}{$xml->localName}";
147	}
148	else
149	{
150	$attrName = $xml->localName;
151	}
152	$attributes[$attrName] = $xml->value;
153	}
154
155	foreach ($attributes as $attr => $value) :
156	list($ns, $local) = $this->split_ns($attr);
157	if ($ns=='http://www.w3.org/2000/xmlns/') :
158	if ('xmlns' == $local) : $local = false; endif;
159	$this->start_xmlns(null, $local, $value);
160	endif;
161	endforeach;
162
163	$this->tag_open(null, $tagName, $attributes);
164	if ($empty)
165	{
166	$this->tag_close(null, $tagName);
167	}
168	break;
169	case constant('XMLReader::TEXT'):
170
171	case constant('XMLReader::CDATA'):
172	$this->cdata(null, $xml->value);
173	break;
174	}
175	}
176	if ($error = libxml_get_last_error())
177	{
178	$this->error_code = $error->code;
179	$this->error_string = $error->message;
180	$this->current_line = $error->line;
181	$this->current_column = $error->column;
182	return false;
183	}
184	else
185	{
186	return true;
187	}
188	}
189	} /* FeedWordPress_Parser::parse() */
190
191	var $xmlns_stack = array();
192	var $xmlns_current = array();
193	function tag_open ($parser, $tag, $attributes) {
194	$ret = parent::tag_open($parser, $tag, $attributes);
195	if ($this->current_xhtml_construct < 0) :
196	$this->data['xmlns'] = $this->xmlns_current;
197	$this->xmlns_stack[] = $this->xmlns_current;
198	endif;
199	return $ret;
200	}
201
202	function tag_close($parser, $tag) {
203	if ($this->current_xhtml_construct < 0) :
204	$this->xmlns_current = array_pop($this->xmlns_stack);
205	endif;
206	$ret = parent::tag_close($parser, $tag);
207	return $ret;
208	}
209
210	function start_xmlns ($parser, $prefix, $uri) {
211	if (!$prefix) :
212	$prefix = '';
213	endif;
214	if ($this->current_xhtml_construct < 0) :
215	$this->xmlns_current[$prefix] = $uri;
216	endif;
217	return true;
218	} /* FeedWordPress_Parser::start_xmlns() */
219	}
220
221