]>
Commit | Line | Data |
---|---|---|
cd07592c AD |
1 | <?php |
2 | class FeedParser { | |
3 | private $doc; | |
4 | private $error; | |
54f23d38 | 5 | private $libxml_errors = array(); |
cd07592c AD |
6 | private $items; |
7 | private $link; | |
8 | private $title; | |
9 | private $type; | |
b9eee80e | 10 | private $xpath; |
cd07592c AD |
11 | |
12 | const FEED_RDF = 0; | |
13 | const FEED_RSS = 1; | |
14 | const FEED_ATOM = 2; | |
15 | ||
4ad04ee2 AD |
16 | function normalize_encoding($data) { |
17 | if (preg_match('/^(<\?xml[\t\n\r ].*?encoding[\t\n\r ]*=[\t\n\r ]*["\'])(.+?)(["\'].*?\?>)/s', $data, $matches) === 1) { | |
18 | $data = mb_convert_encoding($data, 'UTF-8', $matches[2]); | |
19 | ||
20 | $data = preg_replace('/^<\?xml[\t\n\r ].*?\?>/s', $matches[1] . "UTF-8" . $matches[3] , $data); | |
21 | } | |
22 | ||
23 | return $data; | |
24 | } | |
25 | ||
cd07592c AD |
26 | function __construct($data) { |
27 | libxml_use_internal_errors(true); | |
28 | libxml_clear_errors(); | |
29 | $this->doc = new DOMDocument(); | |
30 | $this->doc->loadXML($data); | |
d1f3fa97 | 31 | |
f612dbe8 AD |
32 | mb_substitute_character("none"); |
33 | ||
d1f3fa97 AD |
34 | $error = libxml_get_last_error(); |
35 | ||
4f00f55c | 36 | // libxml compiled without iconv? |
5fa36a64 | 37 | if ($error && $error->code == 32) { |
4ad04ee2 | 38 | $data = $this->normalize_encoding($data); |
4f00f55c | 39 | |
4ad04ee2 AD |
40 | if ($data) { |
41 | libxml_clear_errors(); | |
f612dbe8 | 42 | |
4ad04ee2 AD |
43 | $this->doc = new DOMDocument(); |
44 | $this->doc->loadXML($data); | |
f612dbe8 | 45 | |
4ad04ee2 | 46 | $error = libxml_get_last_error(); |
5fa36a64 | 47 | } |
4f00f55c AD |
48 | } |
49 | ||
50 | // some terrible invalid unicode entity? | |
5fa36a64 | 51 | if ($error) { |
4d49863f | 52 | foreach (libxml_get_errors() as $err) { |
5fa36a64 | 53 | if ($err->code == 9) { |
4ad04ee2 AD |
54 | // if the source feed is not in utf8, next conversion will fail |
55 | $data = $this->normalize_encoding($data); | |
56 | ||
5fa36a64 | 57 | // remove dangling bytes |
58 | $data = mb_convert_encoding($data, 'UTF-8', 'UTF-8'); | |
4d49863f | 59 | |
5fa36a64 | 60 | // apparently not all UTF-8 characters are valid for XML |
61 | $data = preg_replace('/[^\x{0009}\x{000a}\x{000d}\x{0020}-\x{D7FF}\x{E000}-\x{FFFD}]+/u', ' ', $data); | |
62 | ||
63 | if ($data) { | |
64 | libxml_clear_errors(); | |
65 | ||
66 | $this->doc = new DOMDocument(); | |
67 | $this->doc->loadXML($data); | |
68 | ||
69 | $error = libxml_get_last_error(); | |
70 | } | |
71 | break; | |
72 | } | |
f612dbe8 | 73 | } |
d1f3fa97 | 74 | } |
4d49863f | 75 | |
4d49863f AD |
76 | if ($error) { |
77 | foreach (libxml_get_errors() as $error) { | |
78 | if ($error->level == LIBXML_ERR_FATAL) { | |
54f23d38 | 79 | if(!isset($this->error)) //currently only the first error is reported |
80 | $this->error = $this->format_error($error); | |
81 | $this->libxml_errors [] = $this->format_error($error); | |
d3305ff8 | 82 | } |
4d49863f | 83 | } |
d3305ff8 | 84 | } |
cd07592c AD |
85 | libxml_clear_errors(); |
86 | ||
87 | $this->items = array(); | |
88 | } | |
89 | ||
90 | function init() { | |
91 | $root = $this->doc->firstChild; | |
04d2f9c8 AD |
92 | $xpath = new DOMXPath($this->doc); |
93 | $xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom'); | |
a3b9fd12 | 94 | $xpath->registerNamespace('atom03', 'http://purl.org/atom/ns#'); |
4c00e15b | 95 | $xpath->registerNamespace('media', 'http://search.yahoo.com/mrss/'); |
852d4ac8 | 96 | $xpath->registerNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#'); |
d4992d6b AD |
97 | $xpath->registerNamespace('slash', 'http://purl.org/rss/1.0/modules/slash/'); |
98 | $xpath->registerNamespace('dc', 'http://purl.org/dc/elements/1.1/'); | |
8a95d630 | 99 | $xpath->registerNamespace('content', 'http://purl.org/rss/1.0/modules/content/'); |
d71ac5d3 | 100 | $xpath->registerNamespace('thread', 'http://purl.org/syndication/thread/1.0'); |
d4992d6b | 101 | |
b9eee80e | 102 | $this->xpath = $xpath; |
04d2f9c8 | 103 | |
a9000b03 | 104 | $root = $xpath->query("(//atom03:feed|//atom:feed|//channel|//rdf:rdf|//rdf:RDF)"); |
cd07592c | 105 | |
2c6f3c2e | 106 | if ($root && $root->length > 0) { |
a9000b03 AD |
107 | $root = $root->item(0); |
108 | ||
109 | if ($root) { | |
110 | switch (mb_strtolower($root->tagName)) { | |
111 | case "rdf:rdf": | |
112 | $this->type = $this::FEED_RDF; | |
113 | break; | |
114 | case "channel": | |
115 | $this->type = $this::FEED_RSS; | |
116 | break; | |
117 | case "feed": | |
118 | $this->type = $this::FEED_ATOM; | |
119 | break; | |
120 | default: | |
121 | if( !isset($this->error) ){ | |
122 | $this->error = "Unknown/unsupported feed type"; | |
123 | } | |
124 | return; | |
349c4229 | 125 | } |
cd07592c AD |
126 | } |
127 | ||
cd07592c AD |
128 | switch ($this->type) { |
129 | case $this::FEED_ATOM: | |
cd07592c AD |
130 | |
131 | $title = $xpath->query("//atom:feed/atom:title")->item(0); | |
132 | ||
a3b9fd12 AD |
133 | if (!$title) |
134 | $title = $xpath->query("//atom03:feed/atom03:title")->item(0); | |
135 | ||
136 | ||
cd07592c AD |
137 | if ($title) { |
138 | $this->title = $title->nodeValue; | |
139 | } | |
140 | ||
141 | $link = $xpath->query("//atom:feed/atom:link[not(@rel)]")->item(0); | |
142 | ||
a3b9fd12 AD |
143 | if (!$link) |
144 | $link = $xpath->query("//atom03:feed/atom03:link[not(@rel)]")->item(0); | |
145 | ||
146 | ||
cd07592c AD |
147 | if ($link && $link->hasAttributes()) { |
148 | $this->link = $link->getAttribute("href"); | |
149 | } | |
150 | ||
151 | $articles = $xpath->query("//atom:entry"); | |
152 | ||
a3b9fd12 AD |
153 | if (!$articles || $articles->length == 0) |
154 | $articles = $xpath->query("//atom03:entry"); | |
155 | ||
cd07592c | 156 | foreach ($articles as $article) { |
4c00e15b | 157 | array_push($this->items, new FeedItem_Atom($article, $this->doc, $this->xpath)); |
cd07592c AD |
158 | } |
159 | ||
cd07592c AD |
160 | break; |
161 | case $this::FEED_RSS: | |
04d2f9c8 AD |
162 | $title = $xpath->query("//channel/title")->item(0); |
163 | ||
164 | if ($title) { | |
165 | $this->title = $title->nodeValue; | |
166 | } | |
167 | ||
168 | $link = $xpath->query("//channel/link")->item(0); | |
169 | ||
1874c8d6 AD |
170 | if ($link) { |
171 | if ($link->getAttribute("href")) | |
172 | $this->link = $link->getAttribute("href"); | |
173 | else if ($link->nodeValue) | |
174 | $this->link = $link->nodeValue; | |
04d2f9c8 AD |
175 | } |
176 | ||
177 | $articles = $xpath->query("//channel/item"); | |
178 | ||
179 | foreach ($articles as $article) { | |
4c00e15b | 180 | array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath)); |
04d2f9c8 AD |
181 | } |
182 | ||
cd07592c | 183 | break; |
852d4ac8 AD |
184 | case $this::FEED_RDF: |
185 | $xpath->registerNamespace('rssfake', 'http://purl.org/rss/1.0/'); | |
186 | ||
187 | $title = $xpath->query("//rssfake:channel/rssfake:title")->item(0); | |
188 | ||
189 | if ($title) { | |
190 | $this->title = $title->nodeValue; | |
191 | } | |
192 | ||
193 | $link = $xpath->query("//rssfake:channel/rssfake:link")->item(0); | |
194 | ||
195 | if ($link) { | |
196 | $this->link = $link->nodeValue; | |
197 | } | |
198 | ||
199 | $articles = $xpath->query("//rssfake:item"); | |
200 | ||
201 | foreach ($articles as $article) { | |
202 | array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath)); | |
203 | } | |
204 | ||
205 | break; | |
206 | ||
cd07592c | 207 | } |
852d4ac8 | 208 | } else { |
349c4229 | 209 | if( !isset($this->error) ){ |
210 | $this->error = "Unknown/unsupported feed type"; | |
211 | } | |
852d4ac8 | 212 | return; |
cd07592c AD |
213 | } |
214 | } | |
215 | ||
216 | function format_error($error) { | |
217 | if ($error) { | |
218 | return sprintf("LibXML error %s at line %d (column %d): %s", | |
219 | $error->code, $error->line, $error->column, | |
220 | $error->message); | |
221 | } else { | |
222 | return ""; | |
223 | } | |
224 | } | |
225 | ||
226 | function error() { | |
227 | return $this->error; | |
228 | } | |
229 | ||
54f23d38 | 230 | function errors() { |
231 | return $this->libxml_errors; | |
232 | } | |
233 | ||
cd07592c AD |
234 | function get_link() { |
235 | return $this->link; | |
236 | } | |
237 | ||
238 | function get_title() { | |
239 | return $this->title; | |
240 | } | |
241 | ||
242 | function get_items() { | |
243 | return $this->items; | |
244 | } | |
245 | ||
b9eee80e AD |
246 | function get_links($rel) { |
247 | $rv = array(); | |
248 | ||
249 | switch ($this->type) { | |
250 | case $this::FEED_ATOM: | |
251 | $links = $this->xpath->query("//atom:feed/atom:link"); | |
252 | ||
253 | foreach ($links as $link) { | |
254 | if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) { | |
255 | array_push($rv, $link->getAttribute('href')); | |
256 | } | |
257 | } | |
258 | break; | |
259 | case $this::FEED_RSS: | |
f17c3ee2 AD |
260 | $links = $this->xpath->query("//atom:link"); |
261 | ||
3c8060ac AD |
262 | foreach ($links as $link) { |
263 | if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) { | |
264 | array_push($rv, $link->getAttribute('href')); | |
265 | } | |
b9eee80e AD |
266 | } |
267 | break; | |
268 | } | |
269 | ||
270 | return $rv; | |
271 | } | |
cd07592c | 272 | } ?> |