feedwordpress / feedwordpress_parser.class.php

feedwordpress_parser.class.php in FeedWordPress 2011.1018, at feedwordpress_parser.class.php

204 lines 5.2 KB

Raw Download Zip

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