feedwordpress / feedwordpress_parser.class.php

feedwordpress_parser.class.php in FeedWordPress 2014.0805, at feedwordpress_parser.class.php

245 lines 6.6 KB

Raw Download Zip

1	<?php
2	class FeedWordPress_Parser extends SimplePie_Parser {
3	function reset_parser (&$xml) {
4	xml_parser_free($xml);
5
6	$xml = xml_parser_create_ns($this->encoding, $this->separator);
7	xml_parser_set_option($xml, XML_OPTION_SKIP_WHITE, 1);
8	xml_parser_set_option($xml, XML_OPTION_CASE_FOLDING, 0);
9	xml_set_object($xml, $this);
10	xml_set_character_data_handler($xml, 'cdata');
11	xml_set_element_handler($xml, 'tag_open', 'tag_close');
12	xml_set_start_namespace_decl_handler($xml, 'start_xmlns');
13	}
14
15	function parse (&$data, $encoding) {
16	$data = apply_filters('feedwordpress_parser_parse', $data, $encoding, $this);
17
18	// Use UTF-8 if we get passed US-ASCII, as every US-ASCII character is a UTF-8 character
19	if (strtoupper($encoding) === 'US-ASCII')
20	{
21	$this->encoding = 'UTF-8';
22	}
23	else
24	{
25	$this->encoding = $encoding;
26	}
27
28	// Strip BOM:
29	// UTF-32 Big Endian BOM
30	if (substr($data, 0, 4) === "\x00\x00\xFE\xFF")
31	{
32	$data = substr($data, 4);
33	}
34	// UTF-32 Little Endian BOM
35	elseif (substr($data, 0, 4) === "\xFF\xFE\x00\x00")
36	{
37	$data = substr($data, 4);
38	}
39	// UTF-16 Big Endian BOM
40	elseif (substr($data, 0, 2) === "\xFE\xFF")
41	{
42	$data = substr($data, 2);
43	}
44	// UTF-16 Little Endian BOM
45	elseif (substr($data, 0, 2) === "\xFF\xFE")
46	{
47	$data = substr($data, 2);
48	}
49	// UTF-8 BOM
50	elseif (substr($data, 0, 3) === "\xEF\xBB\xBF")
51	{
52	$data = substr($data, 3);
53	}
54
55	if (substr($data, 0, 5) === '<?xml' && strspn(substr($data, 5, 1), "\x09\x0A\x0D\x20") && ($pos = strpos($data, '?>')) !== false)
56	{
57	$declaration = new SimplePie_XML_Declaration_Parser(substr($data, 5, $pos - 5));
58	if ($declaration->parse())
59	{
60	$data = substr($data, $pos + 2);
61	$data = '<?xml version="' . $declaration->version . '" encoding="' . $encoding . '" standalone="' . (($declaration->standalone) ? 'yes' : 'no') . '"?>' . $data;
62	}
63	else
64	{
65	$this->error_string = 'SimplePie bug! Please report this!';
66	return false;
67	}
68	}
69
70	$return = true;
71
72	static $xml_is_sane = null;
73	if ($xml_is_sane === null)
74	{
75	$parser_check = xml_parser_create();
76	xml_parse_into_struct($parser_check, '<foo>&</foo>', $values);
77	xml_parser_free($parser_check);
78	$xml_is_sane = isset($values[0]['value']);
79	}
80
81	// Create the parser
82	if ($xml_is_sane)
83	{
84	$xml = xml_parser_create_ns($this->encoding, $this->separator);
85	xml_parser_set_option($xml, XML_OPTION_SKIP_WHITE, 1);
86	xml_parser_set_option($xml, XML_OPTION_CASE_FOLDING, 0);
87	xml_set_object($xml, $this);
88	xml_set_character_data_handler($xml, 'cdata');
89	xml_set_element_handler($xml, 'tag_open', 'tag_close');
90	xml_set_start_namespace_decl_handler($xml, 'start_xmlns');
91
92	// Parse!
93	$parseResults = xml_parse($xml, $data, true);
94
95	$endOfJunk = strpos($data, '<?xml');
96	if (!$parseResults and $endOfJunk > 0) :
97	// There is some junk before the feed prolog. Try to get rid of it.
98	$newData = substr($data, $endOfJunk);
99	$newData = trim($newData);
100	$this->reset_parser($xml);
101
102	$parseResults = xml_parse($xml, $newData, true);
103	endif;
104
105	if (!$parseResults)
106	{
107	if (class_exists('DOMDocument')) :
108	libxml_use_internal_errors(true);
109	$doc = new DOMDocument;
110	$doc->loadHTML('<html>'.$data.'</html>');
111	///DBG/ echo "<plaintext>";
112	///DBG/ $dd = $doc->getElementsByTagName('html');
113	///DBG/ for ($i = 0; $i < $dd->length; $i++) :
114	///DBG/ var_dump($dd->item($i)->nodeName);
115	///DBG/ endfor;
116	///DBG/ var_dump($doc);
117	libxml_use_internal_errors(false);
118	///DBG/ die;
119	endif;
120
121	$this->error_code = xml_get_error_code($xml);
122	$this->error_string = xml_error_string($this->error_code);
123	///DBG/ echo "WOOGA WOOGA"; var_dump($this->error_string); die;
124	$return = false;
125	}
126
127	$this->current_line = xml_get_current_line_number($xml);
128	$this->current_column = xml_get_current_column_number($xml);
129	$this->current_byte = xml_get_current_byte_index($xml);
130	xml_parser_free($xml);
131
132	return $return;
133	}
134	else
135	{
136	libxml_clear_errors();
137	$xml = new XMLReader();
138	$xml->xml($data);
139	while (@$xml->read())
140	{
141	switch ($xml->nodeType)
142	{
143
144	case constant('XMLReader::END_ELEMENT'):
145	if ($xml->namespaceURI !== '')
146	{
147	$tagName = "{$xml->namespaceURI}{$this->separator}{$xml->localName}";
148	}
149	else
150	{
151	$tagName = $xml->localName;
152	}
153	$this->tag_close(null, $tagName);
154	break;
155	case constant('XMLReader::ELEMENT'):
156	$empty = $xml->isEmptyElement;
157	if ($xml->namespaceURI !== '')
158	{
159	$tagName = "{$xml->namespaceURI}{$this->separator}{$xml->localName}";
160	}
161	else
162	{
163	$tagName = $xml->localName;
164	}
165	$attributes = array();
166	while ($xml->moveToNextAttribute())
167	{
168	if ($xml->namespaceURI !== '')
169	{
170	$attrName = "{$xml->namespaceURI}{$this->separator}{$xml->localName}";
171	}
172	else
173	{
174	$attrName = $xml->localName;
175	}
176	$attributes[$attrName] = $xml->value;
177	}
178
179	foreach ($attributes as $attr => $value) :
180	list($ns, $local) = $this->split_ns($attr);
181	if ($ns=='http://www.w3.org/2000/xmlns/') :
182	if ('xmlns' == $local) : $local = false; endif;
183	$this->start_xmlns(null, $local, $value);
184	endif;
185	endforeach;
186
187	$this->tag_open(null, $tagName, $attributes);
188	if ($empty)
189	{
190	$this->tag_close(null, $tagName);
191	}
192	break;
193	case constant('XMLReader::TEXT'):
194
195	case constant('XMLReader::CDATA'):
196	$this->cdata(null, $xml->value);
197	break;
198	}
199	}
200	if ($error = libxml_get_last_error())
201	{
202	$this->error_code = $error->code;
203	$this->error_string = $error->message;
204	$this->current_line = $error->line;
205	$this->current_column = $error->column;
206	return false;
207	}
208	else
209	{
210	return true;
211	}
212	}
213	} /* FeedWordPress_Parser::parse() */
214
215	var $xmlns_stack = array();
216	var $xmlns_current = array();
217	function tag_open ($parser, $tag, $attributes) {
218	$ret = parent::tag_open($parser, $tag, $attributes);
219	if ($this->current_xhtml_construct < 0) :
220	$this->data['xmlns'] = $this->xmlns_current;
221	$this->xmlns_stack[] = $this->xmlns_current;
222	endif;
223	return $ret;
224	}
225
226	function tag_close($parser, $tag) {
227	if ($this->current_xhtml_construct < 0) :
228	$this->xmlns_current = array_pop($this->xmlns_stack);
229	endif;
230	$ret = parent::tag_close($parser, $tag);
231	return $ret;
232	}
233
234	function start_xmlns ($parser, $prefix, $uri) {
235	if (!$prefix) :
236	$prefix = '';
237	endif;
238	if ($this->current_xhtml_construct < 0) :
239	$this->xmlns_current[$prefix] = $uri;
240	endif;
241	return true;
242	} /* FeedWordPress_Parser::start_xmlns() */
243	}
244
245