|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/* Copyright (C) 2011-2012 Laurent Destailleur <[email protected]> |
|
4
|
|
|
* Copyright (C) 2024 MDW <[email protected]> |
|
5
|
|
|
* Copyright (C) 2024 Rafael San José <[email protected]> |
|
6
|
|
|
* |
|
7
|
|
|
* This program is free software; you can redistribute it and/or modify |
|
8
|
|
|
* it under the terms of the GNU General Public License as published by |
|
9
|
|
|
* the Free Software Foundation; either version 3 of the License, or |
|
10
|
|
|
* (at your option) any later version. |
|
11
|
|
|
* |
|
12
|
|
|
* This program is distributed in the hope that it will be useful, |
|
13
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
14
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
|
15
|
|
|
* GNU General Public License for more details. |
|
16
|
|
|
* |
|
17
|
|
|
* You should have received a copy of the GNU General Public License |
|
18
|
|
|
* along with this program. If not, see <https://www.gnu.org/licenses/>. |
|
19
|
|
|
*/ |
|
20
|
|
|
|
|
21
|
|
|
namespace Dolibarr\Code\Core\Classes; |
|
22
|
|
|
|
|
23
|
|
|
use DoliDB; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* \file htdocs/core/class/rssparser.class.php |
|
27
|
|
|
* \ingroup core |
|
28
|
|
|
* \brief File of class to parse RSS feeds |
|
29
|
|
|
*/ |
|
30
|
|
|
|
|
31
|
|
|
// @phan-file-suppress PhanPluginPHPDocInWrongComment |
|
32
|
|
|
|
|
33
|
|
|
/** |
|
34
|
|
|
* Class to parse RSS files |
|
35
|
|
|
*/ |
|
36
|
|
|
class RssParser |
|
37
|
|
|
{ |
|
38
|
|
|
/** |
|
39
|
|
|
* @var DoliDB Database handler. |
|
40
|
|
|
*/ |
|
41
|
|
|
public $db; |
|
42
|
|
|
|
|
43
|
|
|
/** |
|
44
|
|
|
* @var string Error code (or message) |
|
45
|
|
|
*/ |
|
46
|
|
|
public $error = ''; |
|
47
|
|
|
|
|
48
|
|
|
public $feed_version; |
|
49
|
|
|
|
|
50
|
|
|
private $_format = ''; |
|
51
|
|
|
private $_urlRSS; |
|
52
|
|
|
private $_language; |
|
53
|
|
|
private $_generator; |
|
54
|
|
|
private $_copyright; |
|
55
|
|
|
private $_lastbuilddate; |
|
56
|
|
|
private $_imageurl; |
|
57
|
|
|
private $_link; |
|
58
|
|
|
private $_title; |
|
59
|
|
|
private $_description; |
|
60
|
|
|
private $_lastfetchdate; // Last successful fetch |
|
61
|
|
|
private $_rssarray = array(); |
|
62
|
|
|
|
|
63
|
|
|
private $current_namespace; |
|
64
|
|
|
public $items = array(); |
|
65
|
|
|
public $current_item = array(); |
|
66
|
|
|
public $channel = array(); |
|
67
|
|
|
public $textinput = array(); |
|
68
|
|
|
public $image = array(); |
|
69
|
|
|
|
|
70
|
|
|
private $initem; |
|
71
|
|
|
private $intextinput; |
|
72
|
|
|
private $incontent; |
|
73
|
|
|
private $inimage; |
|
74
|
|
|
private $inchannel; |
|
75
|
|
|
|
|
76
|
|
|
// For parsing with xmlparser |
|
77
|
|
|
public $stack = array(); // parser stack |
|
78
|
|
|
private $_CONTENT_CONSTRUCTS = array('content', 'summary', 'info', 'title', 'tagline', 'copyright'); |
|
79
|
|
|
|
|
80
|
|
|
|
|
81
|
|
|
/** |
|
82
|
|
|
* Constructor |
|
83
|
|
|
* |
|
84
|
|
|
* @param DoliDB $db Database handler |
|
85
|
|
|
*/ |
|
86
|
|
|
public function __construct($db) |
|
87
|
|
|
{ |
|
88
|
|
|
$this->db = $db; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* getFormat |
|
93
|
|
|
* |
|
94
|
|
|
* @return string |
|
95
|
|
|
*/ |
|
96
|
|
|
public function getFormat() |
|
97
|
|
|
{ |
|
98
|
|
|
return $this->_format; |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
/** |
|
102
|
|
|
* getUrlRss |
|
103
|
|
|
* |
|
104
|
|
|
* @return string |
|
105
|
|
|
*/ |
|
106
|
|
|
public function getUrlRss() |
|
107
|
|
|
{ |
|
108
|
|
|
return $this->_urlRSS; |
|
109
|
|
|
} |
|
110
|
|
|
/** |
|
111
|
|
|
* getLanguage |
|
112
|
|
|
* |
|
113
|
|
|
* @return string |
|
114
|
|
|
*/ |
|
115
|
|
|
public function getLanguage() |
|
116
|
|
|
{ |
|
117
|
|
|
return $this->_language; |
|
118
|
|
|
} |
|
119
|
|
|
/** |
|
120
|
|
|
* getGenerator |
|
121
|
|
|
* |
|
122
|
|
|
* @return string |
|
123
|
|
|
*/ |
|
124
|
|
|
public function getGenerator() |
|
125
|
|
|
{ |
|
126
|
|
|
return $this->_generator; |
|
127
|
|
|
} |
|
128
|
|
|
/** |
|
129
|
|
|
* getCopyright |
|
130
|
|
|
* |
|
131
|
|
|
* @return string |
|
132
|
|
|
*/ |
|
133
|
|
|
public function getCopyright() |
|
134
|
|
|
{ |
|
135
|
|
|
return $this->_copyright; |
|
136
|
|
|
} |
|
137
|
|
|
/** |
|
138
|
|
|
* getLastBuildDate |
|
139
|
|
|
* |
|
140
|
|
|
* @return string |
|
141
|
|
|
*/ |
|
142
|
|
|
public function getLastBuildDate() |
|
143
|
|
|
{ |
|
144
|
|
|
return $this->_lastbuilddate; |
|
145
|
|
|
} |
|
146
|
|
|
/** |
|
147
|
|
|
* getImageUrl |
|
148
|
|
|
* |
|
149
|
|
|
* @return string |
|
150
|
|
|
*/ |
|
151
|
|
|
public function getImageUrl() |
|
152
|
|
|
{ |
|
153
|
|
|
return $this->_imageurl; |
|
154
|
|
|
} |
|
155
|
|
|
/** |
|
156
|
|
|
* getLink |
|
157
|
|
|
* |
|
158
|
|
|
* @return string |
|
159
|
|
|
*/ |
|
160
|
|
|
public function getLink() |
|
161
|
|
|
{ |
|
162
|
|
|
return $this->_link; |
|
163
|
|
|
} |
|
164
|
|
|
/** |
|
165
|
|
|
* getTitle |
|
166
|
|
|
* |
|
167
|
|
|
* @return string |
|
168
|
|
|
*/ |
|
169
|
|
|
public function getTitle() |
|
170
|
|
|
{ |
|
171
|
|
|
return $this->_title; |
|
172
|
|
|
} |
|
173
|
|
|
/** |
|
174
|
|
|
* getDescription |
|
175
|
|
|
* |
|
176
|
|
|
* @return string |
|
177
|
|
|
*/ |
|
178
|
|
|
public function getDescription() |
|
179
|
|
|
{ |
|
180
|
|
|
return $this->_description; |
|
181
|
|
|
} |
|
182
|
|
|
/** |
|
183
|
|
|
* getLastFetchDate |
|
184
|
|
|
* |
|
185
|
|
|
* @return int |
|
186
|
|
|
*/ |
|
187
|
|
|
public function getLastFetchDate() |
|
188
|
|
|
{ |
|
189
|
|
|
return $this->_lastfetchdate; |
|
190
|
|
|
} |
|
191
|
|
|
/** |
|
192
|
|
|
* getItems |
|
193
|
|
|
* |
|
194
|
|
|
* @return array |
|
195
|
|
|
*/ |
|
196
|
|
|
public function getItems() |
|
197
|
|
|
{ |
|
198
|
|
|
return $this->_rssarray; |
|
199
|
|
|
} |
|
200
|
|
|
|
|
201
|
|
|
/** |
|
202
|
|
|
* Parse rss URL |
|
203
|
|
|
* |
|
204
|
|
|
* @param string $urlRSS Url to parse |
|
205
|
|
|
* @param int $maxNb Max nb of records to get (0 for no limit) |
|
206
|
|
|
* @param int $cachedelay 0=No cache, nb of seconds we accept cache files (cachedir must also be defined) |
|
207
|
|
|
* @param string $cachedir Directory where to save cache file (For example $conf->externalrss->dir_temp) |
|
208
|
|
|
* @return int Return integer <0 if KO, >0 if OK |
|
209
|
|
|
*/ |
|
210
|
|
|
public function parser($urlRSS, $maxNb = 0, $cachedelay = 60, $cachedir = '') |
|
211
|
|
|
{ |
|
212
|
|
|
include_once DOL_DOCUMENT_ROOT . '/core/lib/files.lib.php'; |
|
213
|
|
|
include_once DOL_DOCUMENT_ROOT . '/core/lib/geturl.lib.php'; |
|
214
|
|
|
|
|
215
|
|
|
$rss = ''; |
|
216
|
|
|
$str = ''; // This will contain content of feed |
|
217
|
|
|
|
|
218
|
|
|
// Check parameters |
|
219
|
|
|
if (!dol_is_url($urlRSS)) { |
|
220
|
|
|
$this->error = "ErrorBadUrl"; |
|
221
|
|
|
return -1; |
|
222
|
|
|
} |
|
223
|
|
|
|
|
224
|
|
|
$this->_urlRSS = $urlRSS; |
|
225
|
|
|
$newpathofdestfile = $cachedir . '/' . dol_hash($this->_urlRSS, 3); // Force md5 hash (does not contain special chars) |
|
226
|
|
|
$newmask = '0644'; |
|
227
|
|
|
|
|
228
|
|
|
//dol_syslog("RssParser::parser parse url=".$urlRSS." => cache file=".$newpathofdestfile); |
|
229
|
|
|
$nowgmt = dol_now(); |
|
230
|
|
|
|
|
231
|
|
|
// Search into cache |
|
232
|
|
|
$foundintocache = 0; |
|
233
|
|
|
if ($cachedelay > 0 && $cachedir) { |
|
234
|
|
|
$filedate = dol_filemtime($newpathofdestfile); |
|
235
|
|
|
if ($filedate >= ($nowgmt - $cachedelay)) { |
|
236
|
|
|
//dol_syslog("RssParser::parser cache file ".$newpathofdestfile." is not older than now - cachedelay (".$nowgmt." - ".$cachedelay.") so we use it."); |
|
237
|
|
|
$foundintocache = 1; |
|
238
|
|
|
|
|
239
|
|
|
$this->_lastfetchdate = $filedate; |
|
240
|
|
|
} else { |
|
241
|
|
|
dol_syslog(get_class($this) . "::parser cache file " . $newpathofdestfile . " is not found or older than now - cachedelay (" . $nowgmt . " - " . $cachedelay . ") so we can't use it."); |
|
242
|
|
|
} |
|
243
|
|
|
} |
|
244
|
|
|
|
|
245
|
|
|
// Load file into $str |
|
246
|
|
|
if ($foundintocache) { // Cache file found and is not too old |
|
247
|
|
|
$str = file_get_contents($newpathofdestfile); |
|
248
|
|
|
} else { |
|
249
|
|
|
try { |
|
250
|
|
|
$result = getURLContent($this->_urlRSS, 'GET', '', 1, array(), array('http', 'https'), 0); |
|
251
|
|
|
|
|
252
|
|
|
if (!empty($result['content'])) { |
|
253
|
|
|
$str = $result['content']; |
|
254
|
|
|
} elseif (!empty($result['curl_error_msg'])) { |
|
255
|
|
|
$this->error = 'Error retrieving URL ' . $this->_urlRSS . ' - ' . $result['curl_error_msg']; |
|
256
|
|
|
return -1; |
|
257
|
|
|
} |
|
258
|
|
|
} catch (Exception $e) { |
|
|
|
|
|
|
259
|
|
|
$this->error = 'Error retrieving URL ' . $this->_urlRSS . ' - ' . $e->getMessage(); |
|
260
|
|
|
return -2; |
|
261
|
|
|
} |
|
262
|
|
|
} |
|
263
|
|
|
|
|
264
|
|
|
if ($str !== false) { |
|
265
|
|
|
// Convert $str into xml |
|
266
|
|
|
if (getDolGlobalString('EXTERNALRSS_USE_SIMPLEXML')) { |
|
267
|
|
|
//print 'xx'.LIBXML_NOCDATA; |
|
268
|
|
|
libxml_use_internal_errors(false); |
|
269
|
|
|
if (LIBXML_VERSION < 20900) { |
|
270
|
|
|
// Avoid load of external entities (security problem). |
|
271
|
|
|
// Required only if LIBXML_VERSION < 20900 |
|
272
|
|
|
// @phan-suppress-next-line PhanDeprecatedFunctionInternal |
|
273
|
|
|
libxml_disable_entity_loader(true); |
|
274
|
|
|
} |
|
275
|
|
|
|
|
276
|
|
|
$rss = simplexml_load_string($str, "SimpleXMLElement", LIBXML_NOCDATA); |
|
277
|
|
|
} else { |
|
278
|
|
|
if (!function_exists('xml_parser_create')) { |
|
279
|
|
|
$this->error = 'Function xml_parser_create are not supported by your PHP'; |
|
280
|
|
|
return -1; |
|
281
|
|
|
} |
|
282
|
|
|
|
|
283
|
|
|
try { |
|
284
|
|
|
// @phan-suppress-next-line PhanTypeMismatchArgumentInternalProbablyReal |
|
285
|
|
|
$xmlparser = xml_parser_create(null); |
|
286
|
|
|
|
|
287
|
|
|
xml_parser_set_option($xmlparser, XML_OPTION_CASE_FOLDING, 0); |
|
288
|
|
|
xml_parser_set_option($xmlparser, XML_OPTION_SKIP_WHITE, 1); |
|
289
|
|
|
xml_parser_set_option($xmlparser, XML_OPTION_TARGET_ENCODING, "UTF-8"); |
|
290
|
|
|
//xml_set_external_entity_ref_handler($xmlparser, 'extEntHandler'); // Seems to have no effect even when function extEntHandler exists. |
|
291
|
|
|
|
|
292
|
|
|
if (!is_resource($xmlparser) && !is_object($xmlparser)) { |
|
293
|
|
|
$this->error = "ErrorFailedToCreateParser"; |
|
294
|
|
|
return -1; |
|
295
|
|
|
} |
|
296
|
|
|
|
|
297
|
|
|
xml_set_object($xmlparser, $this); |
|
298
|
|
|
// @phan-suppress-next-line PhanUndeclaredFunctionInCallable |
|
299
|
|
|
xml_set_element_handler($xmlparser, 'feed_start_element', 'feed_end_element'); |
|
300
|
|
|
// @phan-suppress-next-line PhanUndeclaredFunctionInCallable |
|
301
|
|
|
xml_set_character_data_handler($xmlparser, 'feed_cdata'); |
|
302
|
|
|
|
|
303
|
|
|
$status = xml_parse($xmlparser, $str, false); |
|
304
|
|
|
|
|
305
|
|
|
xml_parser_free($xmlparser); |
|
306
|
|
|
|
|
307
|
|
|
$rss = $this; |
|
308
|
|
|
//var_dump($status.' '.$rss->_format);exit; |
|
309
|
|
|
} catch (Exception $e) { |
|
310
|
|
|
$rss = null; |
|
311
|
|
|
} |
|
312
|
|
|
} |
|
313
|
|
|
} |
|
314
|
|
|
|
|
315
|
|
|
// If $rss loaded |
|
316
|
|
|
if ($rss) { |
|
317
|
|
|
// Save file into cache |
|
318
|
|
|
if (empty($foundintocache) && $cachedir) { |
|
319
|
|
|
dol_syslog(get_class($this) . "::parser cache file " . $newpathofdestfile . " is saved onto disk."); |
|
320
|
|
|
if (!dol_is_dir($cachedir)) { |
|
321
|
|
|
dol_mkdir($cachedir); |
|
322
|
|
|
} |
|
323
|
|
|
$fp = fopen($newpathofdestfile, 'w'); |
|
324
|
|
|
if ($fp) { |
|
325
|
|
|
fwrite($fp, $str); |
|
326
|
|
|
fclose($fp); |
|
327
|
|
|
dolChmod($newpathofdestfile); |
|
328
|
|
|
|
|
329
|
|
|
$this->_lastfetchdate = $nowgmt; |
|
330
|
|
|
} else { |
|
331
|
|
|
print 'Error, failed to open file ' . $newpathofdestfile . ' for write'; |
|
332
|
|
|
} |
|
333
|
|
|
} |
|
334
|
|
|
|
|
335
|
|
|
unset($str); // Free memory |
|
336
|
|
|
|
|
337
|
|
|
if (empty($rss->_format)) { // If format not detected automatically |
|
338
|
|
|
$rss->_format = 'rss'; |
|
339
|
|
|
if (empty($rss->channel)) { |
|
340
|
|
|
$rss->_format = 'atom'; |
|
341
|
|
|
} |
|
342
|
|
|
} |
|
343
|
|
|
|
|
344
|
|
|
$items = array(); |
|
345
|
|
|
|
|
346
|
|
|
// Save description entries |
|
347
|
|
|
if ($rss->_format == 'rss') { |
|
348
|
|
|
//var_dump($rss); |
|
349
|
|
|
if (getDolGlobalString('EXTERNALRSS_USE_SIMPLEXML')) { |
|
350
|
|
|
if (!empty($rss->channel->language)) { |
|
351
|
|
|
$this->_language = sanitizeVal((string) $rss->channel->language); |
|
352
|
|
|
} |
|
353
|
|
|
if (!empty($rss->channel->generator)) { |
|
354
|
|
|
$this->_generator = sanitizeVal((string) $rss->channel->generator); |
|
355
|
|
|
} |
|
356
|
|
|
if (!empty($rss->channel->copyright)) { |
|
357
|
|
|
$this->_copyright = sanitizeVal((string) $rss->channel->copyright); |
|
358
|
|
|
} |
|
359
|
|
|
if (!empty($rss->channel->lastbuilddate)) { |
|
360
|
|
|
$this->_lastbuilddate = sanitizeVal((string) $rss->channel->lastbuilddate); |
|
361
|
|
|
} |
|
362
|
|
|
if (!empty($rss->channel->image->url[0])) { |
|
363
|
|
|
$this->_imageurl = sanitizeVal((string) $rss->channel->image->url[0]); |
|
364
|
|
|
} |
|
365
|
|
|
if (!empty($rss->channel->link)) { |
|
366
|
|
|
$this->_link = sanitizeVal((string) $rss->channel->link); |
|
367
|
|
|
} |
|
368
|
|
|
if (!empty($rss->channel->title)) { |
|
369
|
|
|
$this->_title = sanitizeVal((string) $rss->channel->title); |
|
370
|
|
|
} |
|
371
|
|
|
if (!empty($rss->channel->description)) { |
|
372
|
|
|
$this->_description = sanitizeVal((string) $rss->channel->description); |
|
373
|
|
|
} |
|
374
|
|
|
} else { |
|
375
|
|
|
//var_dump($rss->channel); |
|
376
|
|
|
if (!empty($rss->channel['language'])) { |
|
377
|
|
|
$this->_language = sanitizeVal((string) $rss->channel['language']); |
|
378
|
|
|
} |
|
379
|
|
|
if (!empty($rss->channel['generator'])) { |
|
380
|
|
|
$this->_generator = sanitizeVal((string) $rss->channel['generator']); |
|
381
|
|
|
} |
|
382
|
|
|
if (!empty($rss->channel['copyright'])) { |
|
383
|
|
|
$this->_copyright = sanitizeVal((string) $rss->channel['copyright']); |
|
384
|
|
|
} |
|
385
|
|
|
if (!empty($rss->channel['lastbuilddate'])) { |
|
386
|
|
|
$this->_lastbuilddate = sanitizeVal((string) $rss->channel['lastbuilddate']); |
|
387
|
|
|
} |
|
388
|
|
|
if (!empty($rss->image['url'])) { |
|
389
|
|
|
$this->_imageurl = sanitizeVal((string) $rss->image['url']); |
|
390
|
|
|
} |
|
391
|
|
|
if (!empty($rss->channel['link'])) { |
|
392
|
|
|
$this->_link = sanitizeVal((string) $rss->channel['link']); |
|
393
|
|
|
} |
|
394
|
|
|
if (!empty($rss->channel['title'])) { |
|
395
|
|
|
$this->_title = sanitizeVal((string) $rss->channel['title']); |
|
396
|
|
|
} |
|
397
|
|
|
if (!empty($rss->channel['description'])) { |
|
398
|
|
|
$this->_description = sanitizeVal((string) $rss->channel['description']); |
|
399
|
|
|
} |
|
400
|
|
|
} |
|
401
|
|
|
|
|
402
|
|
|
if (getDolGlobalString('EXTERNALRSS_USE_SIMPLEXML')) { |
|
403
|
|
|
$items = $rss->channel->item; // With simplexml |
|
404
|
|
|
} else { |
|
405
|
|
|
$items = $rss->items; // With xmlparse |
|
406
|
|
|
} |
|
407
|
|
|
//var_dump($items);exit; |
|
408
|
|
|
} elseif ($rss->_format == 'atom') { |
|
409
|
|
|
//var_dump($rss); |
|
410
|
|
|
if (getDolGlobalString('EXTERNALRSS_USE_SIMPLEXML')) { |
|
411
|
|
|
if (!empty($rss->generator)) { |
|
|
|
|
|
|
412
|
|
|
$this->_generator = sanitizeVal((string) $rss->generator); |
|
413
|
|
|
} |
|
414
|
|
|
if (!empty($rss->lastbuilddate)) { |
|
|
|
|
|
|
415
|
|
|
$this->_lastbuilddate = sanitizeVal((string) $rss->modified); |
|
|
|
|
|
|
416
|
|
|
} |
|
417
|
|
|
if (!empty($rss->link->href)) { |
|
|
|
|
|
|
418
|
|
|
$this->_link = sanitizeVal((string) $rss->link->href); |
|
419
|
|
|
} |
|
420
|
|
|
if (!empty($rss->title)) { |
|
|
|
|
|
|
421
|
|
|
$this->_title = sanitizeVal((string) $rss->title); |
|
422
|
|
|
} |
|
423
|
|
|
if (!empty($rss->description)) { |
|
|
|
|
|
|
424
|
|
|
$this->_description = sanitizeVal((string) $rss->description); |
|
425
|
|
|
} |
|
426
|
|
|
} else { |
|
427
|
|
|
//if (!empty($rss->channel['rss_language'])) $this->_language = (string) $rss->channel['rss_language']; |
|
428
|
|
|
if (!empty($rss->channel['generator'])) { |
|
429
|
|
|
$this->_generator = sanitizeVal((string) $rss->channel['generator']); |
|
430
|
|
|
} |
|
431
|
|
|
//if (!empty($rss->channel['rss_copyright'])) $this->_copyright = (string) $rss->channel['rss_copyright']; |
|
432
|
|
|
if (!empty($rss->channel['modified'])) { |
|
433
|
|
|
$this->_lastbuilddate = sanitizeVal((string) $rss->channel['modified']); |
|
434
|
|
|
} |
|
435
|
|
|
//if (!empty($rss->image['rss_url'])) $this->_imageurl = (string) $rss->image['rss_url']; |
|
436
|
|
|
if (!empty($rss->channel['link'])) { |
|
437
|
|
|
$this->_link = sanitizeVal((string) $rss->channel['link']); |
|
438
|
|
|
} |
|
439
|
|
|
if (!empty($rss->channel['title'])) { |
|
440
|
|
|
$this->_title = sanitizeVal((string) $rss->channel['title']); |
|
441
|
|
|
} |
|
442
|
|
|
//if (!empty($rss->channel['rss_description'])) $this->_description = (string) $rss->channel['rss_description']; |
|
443
|
|
|
|
|
444
|
|
|
if (!empty($rss->channel)) { |
|
445
|
|
|
$this->_imageurl = sanitizeVal($this->getAtomImageUrl($rss->channel)); |
|
446
|
|
|
} |
|
447
|
|
|
} |
|
448
|
|
|
if (getDolGlobalString('EXTERNALRSS_USE_SIMPLEXML')) { |
|
449
|
|
|
$tmprss = xml2php($rss); |
|
450
|
|
|
$items = $tmprss['entry']; |
|
451
|
|
|
} else { |
|
452
|
|
|
// With simplexml |
|
453
|
|
|
$items = $rss->items; // With xmlparse |
|
454
|
|
|
} |
|
455
|
|
|
//var_dump($items);exit; |
|
456
|
|
|
} |
|
457
|
|
|
|
|
458
|
|
|
$i = 0; |
|
459
|
|
|
|
|
460
|
|
|
// Loop on each record |
|
461
|
|
|
if (is_array($items)) { |
|
462
|
|
|
foreach ($items as $item) { |
|
463
|
|
|
//var_dump($item);exit; |
|
464
|
|
|
if ($rss->_format == 'rss') { |
|
465
|
|
|
if (getDolGlobalString('EXTERNALRSS_USE_SIMPLEXML')) { |
|
466
|
|
|
$itemLink = sanitizeVal((string) $item->link); |
|
467
|
|
|
$itemTitle = sanitizeVal((string) $item->title); |
|
468
|
|
|
$itemDescription = sanitizeVal((string) $item->description); |
|
469
|
|
|
$itemPubDate = sanitizeVal((string) $item->pubDate); |
|
470
|
|
|
$itemId = ''; |
|
471
|
|
|
$itemAuthor = ''; |
|
472
|
|
|
} else { |
|
473
|
|
|
$itemLink = sanitizeVal((string) $item['link']); |
|
474
|
|
|
$itemTitle = sanitizeVal((string) $item['title']); |
|
475
|
|
|
$itemDescription = sanitizeVal((string) $item['description']); |
|
476
|
|
|
$itemPubDate = sanitizeVal((string) $item['pubdate']); |
|
477
|
|
|
$itemId = sanitizeVal((string) $item['guid']); |
|
478
|
|
|
$itemAuthor = sanitizeVal((string) ($item['author'] ?? '')); |
|
479
|
|
|
} |
|
480
|
|
|
|
|
481
|
|
|
// Loop on each category |
|
482
|
|
|
$itemCategory = array(); |
|
483
|
|
|
if (!empty($item->category) && is_array($item->category)) { |
|
484
|
|
|
foreach ($item->category as $cat) { |
|
485
|
|
|
$itemCategory[] = (string) $cat; |
|
486
|
|
|
} |
|
487
|
|
|
} |
|
488
|
|
|
} elseif ($rss->_format == 'atom') { |
|
489
|
|
|
$itemLink = (isset($item['link']) ? sanitizeVal((string) $item['link']) : ''); |
|
490
|
|
|
$itemTitle = sanitizeVal((string) $item['title']); |
|
491
|
|
|
$itemDescription = sanitizeVal($this->getAtomItemDescription($item)); |
|
492
|
|
|
$itemPubDate = sanitizeVal((string) $item['created']); |
|
493
|
|
|
$itemId = sanitizeVal((string) $item['id']); |
|
494
|
|
|
$itemAuthor = sanitizeVal((string) ($item['author'] ? $item['author'] : $item['author_name'])); |
|
495
|
|
|
$itemCategory = array(); |
|
496
|
|
|
} else { |
|
497
|
|
|
$itemLink = ''; |
|
498
|
|
|
$itemTitle = ''; |
|
499
|
|
|
$itemDescription = ''; |
|
500
|
|
|
$itemPubDate = ''; |
|
501
|
|
|
$itemId = ''; |
|
502
|
|
|
$itemAuthor = ''; |
|
503
|
|
|
$itemCategory = array(); |
|
504
|
|
|
print 'ErrorBadFeedFormat'; |
|
505
|
|
|
} |
|
506
|
|
|
|
|
507
|
|
|
// Add record to result array |
|
508
|
|
|
$this->_rssarray[$i] = array( |
|
509
|
|
|
'link' => $itemLink, |
|
510
|
|
|
'title' => $itemTitle, |
|
511
|
|
|
'description' => $itemDescription, |
|
512
|
|
|
'pubDate' => $itemPubDate, |
|
513
|
|
|
'category' => $itemCategory, |
|
514
|
|
|
'id' => $itemId, |
|
515
|
|
|
'author' => $itemAuthor |
|
516
|
|
|
); |
|
517
|
|
|
//var_dump($this->_rssarray); |
|
518
|
|
|
|
|
519
|
|
|
$i++; |
|
520
|
|
|
|
|
521
|
|
|
if ($i > $maxNb) { |
|
522
|
|
|
break; // We get all records we want |
|
523
|
|
|
} |
|
524
|
|
|
} |
|
525
|
|
|
} |
|
526
|
|
|
|
|
527
|
|
|
return 1; |
|
528
|
|
|
} else { |
|
529
|
|
|
$this->error = 'ErrorFailedToLoadRSSFile'; |
|
530
|
|
|
return -1; |
|
531
|
|
|
} |
|
532
|
|
|
} |
|
533
|
|
|
|
|
534
|
|
|
|
|
535
|
|
|
|
|
536
|
|
|
// phpcs:disable PEAR.NamingConventions.ValidFunctionName.ScopeNotCamelCaps |
|
537
|
|
|
/** |
|
538
|
|
|
* Triggered when opened tag is found |
|
539
|
|
|
* |
|
540
|
|
|
* @param string $p Start |
|
541
|
|
|
* @param string $element Tag |
|
542
|
|
|
* @param array $attrs Attributes of tags |
|
543
|
|
|
* @return void |
|
544
|
|
|
*/ |
|
545
|
|
|
public function feed_start_element($p, $element, $attrs) |
|
546
|
|
|
{ |
|
547
|
|
|
// phpcs:enable |
|
548
|
|
|
$el = $element = strtolower($element); |
|
549
|
|
|
$attrs = array_change_key_case($attrs, CASE_LOWER); |
|
550
|
|
|
|
|
551
|
|
|
// check for a namespace, and split if found |
|
552
|
|
|
$ns = false; |
|
553
|
|
|
if (strpos($element, ':')) { |
|
554
|
|
|
list($ns, $el) = explode(':', $element, 2); |
|
555
|
|
|
} |
|
556
|
|
|
if ($ns and $ns != 'rdf') { |
|
557
|
|
|
$this->current_namespace = $ns; |
|
558
|
|
|
} |
|
559
|
|
|
|
|
560
|
|
|
// if feed type isn't set, then this is first element of feed identify feed from root element |
|
561
|
|
|
if (empty($this->_format)) { |
|
562
|
|
|
if ($el == 'rdf') { |
|
563
|
|
|
$this->_format = 'rss'; |
|
564
|
|
|
$this->feed_version = '1.0'; |
|
565
|
|
|
} elseif ($el == 'rss') { |
|
566
|
|
|
$this->_format = 'rss'; |
|
567
|
|
|
$this->feed_version = $attrs['version']; |
|
568
|
|
|
} elseif ($el == 'feed') { |
|
569
|
|
|
$this->_format = 'atom'; |
|
570
|
|
|
$this->feed_version = $attrs['version']; |
|
571
|
|
|
$this->inchannel = true; |
|
572
|
|
|
} |
|
573
|
|
|
return; |
|
574
|
|
|
} |
|
575
|
|
|
|
|
576
|
|
|
if ($el == 'channel') { |
|
577
|
|
|
$this->inchannel = true; |
|
578
|
|
|
} elseif ($el == 'item' || $el == 'entry') { |
|
579
|
|
|
$this->initem = true; |
|
580
|
|
|
if (isset($attrs['rdf:about'])) { |
|
581
|
|
|
$this->current_item['about'] = $attrs['rdf:about']; |
|
582
|
|
|
} |
|
583
|
|
|
} elseif ($this->_format == 'rss' && $this->current_namespace == '' && $el == 'textinput') { |
|
584
|
|
|
// if we're in the default namespace of an RSS feed, |
|
585
|
|
|
// record textinput or image fields |
|
586
|
|
|
$this->intextinput = true; |
|
587
|
|
|
} elseif ($this->_format == 'rss' && $this->current_namespace == '' && $el == 'image') { |
|
588
|
|
|
$this->inimage = true; |
|
589
|
|
|
} elseif ($this->_format == 'atom' && in_array($el, $this->_CONTENT_CONSTRUCTS)) { |
|
590
|
|
|
// handle atom content constructs |
|
591
|
|
|
// avoid clashing w/ RSS mod_content |
|
592
|
|
|
if ($el == 'content') { |
|
593
|
|
|
$el = 'atom_content'; |
|
594
|
|
|
} |
|
595
|
|
|
|
|
596
|
|
|
$this->incontent = $el; |
|
597
|
|
|
} elseif ($this->_format == 'atom' && $this->incontent) { |
|
598
|
|
|
// if inside an Atom content construct (e.g. content or summary) field treat tags as text |
|
599
|
|
|
// if tags are inlined, then flatten |
|
600
|
|
|
$attrs_str = implode(' ', array_map('rss_map_attrs', array_keys($attrs), array_values($attrs))); |
|
601
|
|
|
|
|
602
|
|
|
$this->append_content("<$element $attrs_str>"); |
|
603
|
|
|
|
|
604
|
|
|
array_unshift($this->stack, $el); |
|
605
|
|
|
} elseif ($this->_format == 'atom' && $el == 'link') { |
|
606
|
|
|
// Atom support many links per containing element. |
|
607
|
|
|
// Magpie treats link elements of type rel='alternate' |
|
608
|
|
|
// as being equivalent to RSS's simple link element. |
|
609
|
|
|
if (isset($attrs['rel']) && $attrs['rel'] == 'alternate') { |
|
610
|
|
|
$link_el = 'link'; |
|
611
|
|
|
} elseif (!isset($attrs['rel'])) { |
|
612
|
|
|
$link_el = 'link'; |
|
613
|
|
|
} else { |
|
614
|
|
|
$link_el = 'link_' . $attrs['rel']; |
|
615
|
|
|
} |
|
616
|
|
|
|
|
617
|
|
|
$this->append($link_el, $attrs['href']); |
|
618
|
|
|
} else { |
|
619
|
|
|
// set stack[0] to current element |
|
620
|
|
|
array_unshift($this->stack, $el); |
|
621
|
|
|
} |
|
622
|
|
|
} |
|
623
|
|
|
|
|
624
|
|
|
|
|
625
|
|
|
// phpcs:disable PEAR.NamingConventions.ValidFunctionName.ScopeNotCamelCaps |
|
626
|
|
|
/** |
|
627
|
|
|
* Triggered when CDATA is found |
|
628
|
|
|
* |
|
629
|
|
|
* @param string $p P |
|
630
|
|
|
* @param string $text Tag |
|
631
|
|
|
* @return void |
|
632
|
|
|
*/ |
|
633
|
|
|
public function feed_cdata($p, $text) |
|
634
|
|
|
{ |
|
635
|
|
|
// phpcs:enable |
|
636
|
|
|
if ($this->_format == 'atom' and $this->incontent) { |
|
637
|
|
|
$this->append_content($text); |
|
638
|
|
|
} else { |
|
639
|
|
|
$current_el = implode('_', array_reverse($this->stack)); |
|
640
|
|
|
$this->append($current_el, $text); |
|
641
|
|
|
} |
|
642
|
|
|
} |
|
643
|
|
|
|
|
644
|
|
|
// phpcs:disable PEAR.NamingConventions.ValidFunctionName.ScopeNotCamelCaps |
|
645
|
|
|
/** |
|
646
|
|
|
* Triggered when closed tag is found |
|
647
|
|
|
* |
|
648
|
|
|
* @param string $p P |
|
649
|
|
|
* @param string $el Tag |
|
650
|
|
|
* @return void |
|
651
|
|
|
*/ |
|
652
|
|
|
public function feed_end_element($p, $el) |
|
653
|
|
|
{ |
|
654
|
|
|
// phpcs:enable |
|
655
|
|
|
$el = strtolower($el); |
|
656
|
|
|
|
|
657
|
|
|
if ($el == 'item' or $el == 'entry') { |
|
658
|
|
|
$this->items[] = $this->current_item; |
|
659
|
|
|
$this->current_item = array(); |
|
660
|
|
|
$this->initem = false; |
|
661
|
|
|
} elseif ($this->_format == 'rss' and $this->current_namespace == '' and $el == 'textinput') { |
|
662
|
|
|
$this->intextinput = false; |
|
663
|
|
|
} elseif ($this->_format == 'rss' and $this->current_namespace == '' and $el == 'image') { |
|
664
|
|
|
$this->inimage = false; |
|
665
|
|
|
} elseif ($this->_format == 'atom' and in_array($el, $this->_CONTENT_CONSTRUCTS)) { |
|
666
|
|
|
$this->incontent = false; |
|
667
|
|
|
} elseif ($el == 'channel' or $el == 'feed') { |
|
668
|
|
|
$this->inchannel = false; |
|
669
|
|
|
} elseif ($this->_format == 'atom' and $this->incontent) { |
|
670
|
|
|
// balance tags properly |
|
671
|
|
|
// note: i don't think this is actually necessary |
|
672
|
|
|
if ($this->stack[0] == $el) { |
|
673
|
|
|
$this->append_content("</$el>"); |
|
674
|
|
|
} else { |
|
675
|
|
|
$this->append_content("<$el />"); |
|
676
|
|
|
} |
|
677
|
|
|
|
|
678
|
|
|
array_shift($this->stack); |
|
679
|
|
|
} else { |
|
680
|
|
|
array_shift($this->stack); |
|
681
|
|
|
} |
|
682
|
|
|
|
|
683
|
|
|
$this->current_namespace = false; |
|
684
|
|
|
} |
|
685
|
|
|
|
|
686
|
|
|
|
|
687
|
|
|
/** |
|
688
|
|
|
* To concat 2 strings with no warning if an operand is not defined |
|
689
|
|
|
* |
|
690
|
|
|
* @param string $str1 Str1 |
|
691
|
|
|
* @param string $str2 Str2 |
|
692
|
|
|
* @return string String cancatenated |
|
693
|
|
|
*/ |
|
694
|
|
|
public function concat(&$str1, $str2 = "") |
|
695
|
|
|
{ |
|
696
|
|
|
if (!isset($str1)) { |
|
697
|
|
|
$str1 = ""; |
|
698
|
|
|
} |
|
699
|
|
|
$str1 .= $str2; |
|
700
|
|
|
return $str1; |
|
701
|
|
|
} |
|
702
|
|
|
|
|
703
|
|
|
// phpcs:disable PEAR.NamingConventions.ValidFunctionName.ScopeNotCamelCaps |
|
704
|
|
|
/** |
|
705
|
|
|
* Enter description here ... |
|
706
|
|
|
* |
|
707
|
|
|
* @param string $text Text |
|
708
|
|
|
* @return void |
|
709
|
|
|
*/ |
|
710
|
|
|
public function append_content($text) |
|
711
|
|
|
{ |
|
712
|
|
|
// phpcs:enable |
|
713
|
|
|
if (!empty($this->initem)) { |
|
714
|
|
|
$this->concat($this->current_item[$this->incontent], $text); |
|
715
|
|
|
} elseif (!empty($this->inchannel)) { |
|
716
|
|
|
$this->concat($this->channel[$this->incontent], $text); |
|
717
|
|
|
} |
|
718
|
|
|
} |
|
719
|
|
|
|
|
720
|
|
|
/** |
|
721
|
|
|
* smart append - field and namespace aware |
|
722
|
|
|
* |
|
723
|
|
|
* @param string $el El |
|
724
|
|
|
* @param string $text Text |
|
725
|
|
|
* @return void |
|
726
|
|
|
*/ |
|
727
|
|
|
public function append($el, $text) |
|
728
|
|
|
{ |
|
729
|
|
|
if (!$el) { |
|
730
|
|
|
return; |
|
731
|
|
|
} |
|
732
|
|
|
if (!empty($this->current_namespace)) { |
|
733
|
|
|
if (!empty($this->initem)) { |
|
734
|
|
|
$this->concat($this->current_item[$this->current_namespace][$el], $text); |
|
735
|
|
|
} elseif (!empty($this->inchannel)) { |
|
736
|
|
|
$this->concat($this->channel[$this->current_namespace][$el], $text); |
|
737
|
|
|
} elseif (!empty($this->intextinput)) { |
|
738
|
|
|
$this->concat($this->textinput[$this->current_namespace][$el], $text); |
|
739
|
|
|
} elseif (!empty($this->inimage)) { |
|
740
|
|
|
$this->concat($this->image[$this->current_namespace][$el], $text); |
|
741
|
|
|
} |
|
742
|
|
|
} else { |
|
743
|
|
|
if (!empty($this->initem)) { |
|
744
|
|
|
$this->concat($this->current_item[$el], $text); |
|
745
|
|
|
} elseif (!empty($this->intextinput)) { |
|
746
|
|
|
$this->concat($this->textinput[$el], $text); |
|
747
|
|
|
} elseif (!empty($this->inimage)) { |
|
748
|
|
|
$this->concat($this->image[$el], $text); |
|
749
|
|
|
} elseif (!empty($this->inchannel)) { |
|
750
|
|
|
$this->concat($this->channel[$el], $text); |
|
751
|
|
|
} |
|
752
|
|
|
} |
|
753
|
|
|
} |
|
754
|
|
|
|
|
755
|
|
|
/** |
|
756
|
|
|
* Return a description/summary for one item from a ATOM feed |
|
757
|
|
|
* |
|
758
|
|
|
* @param array $item A parsed item of a ATOM feed |
|
759
|
|
|
* @param int $maxlength (optional) The maximum length for the description |
|
760
|
|
|
* @return string A summary description |
|
761
|
|
|
*/ |
|
762
|
|
|
private function getAtomItemDescription(array $item, $maxlength = 500) |
|
763
|
|
|
{ |
|
764
|
|
|
$result = ""; |
|
765
|
|
|
|
|
766
|
|
|
if (isset($item['summary'])) { |
|
767
|
|
|
$result = $item['summary']; |
|
768
|
|
|
} elseif (isset($item['atom_content'])) { |
|
769
|
|
|
$result = $item['atom_content']; |
|
770
|
|
|
} |
|
771
|
|
|
|
|
772
|
|
|
// remove all HTML elements that can possible break the maximum size of a tooltip, |
|
773
|
|
|
// like headings, image, video etc. and allow only simple style elements |
|
774
|
|
|
$result = strip_tags($result, "<br><p><ul><ol><li>"); |
|
775
|
|
|
|
|
776
|
|
|
$result = str_replace("\n", "", $result); |
|
777
|
|
|
|
|
778
|
|
|
if (strlen($result) > $maxlength) { |
|
779
|
|
|
$result = substr($result, 0, $maxlength); |
|
780
|
|
|
$result .= "..."; |
|
781
|
|
|
} |
|
782
|
|
|
|
|
783
|
|
|
return $result; |
|
784
|
|
|
} |
|
785
|
|
|
|
|
786
|
|
|
/** |
|
787
|
|
|
* Return a URL to a image of the given ATOM feed |
|
788
|
|
|
* |
|
789
|
|
|
* @param array $feed The ATOM feed that possible contain a link to a logo or icon |
|
790
|
|
|
* @return string A URL to a image from a ATOM feed when found, otherwise a empty string |
|
791
|
|
|
*/ |
|
792
|
|
|
private function getAtomImageUrl(array $feed) |
|
793
|
|
|
{ |
|
794
|
|
|
if (isset($feed['icon'])) { |
|
795
|
|
|
return $feed['logo']; |
|
796
|
|
|
} |
|
797
|
|
|
|
|
798
|
|
|
if (isset($feed['icon'])) { |
|
799
|
|
|
return $feed['logo']; |
|
800
|
|
|
} |
|
801
|
|
|
|
|
802
|
|
|
if (isset($feed['webfeeds:logo'])) { |
|
803
|
|
|
return $feed['webfeeds:logo']; |
|
804
|
|
|
} |
|
805
|
|
|
|
|
806
|
|
|
if (isset($feed['webfeeds:icon'])) { |
|
807
|
|
|
return $feed['webfeeds:icon']; |
|
808
|
|
|
} |
|
809
|
|
|
|
|
810
|
|
|
if (isset($feed['webfeeds:wordmark'])) { |
|
811
|
|
|
return $feed['webfeeds:wordmark']; |
|
812
|
|
|
} |
|
813
|
|
|
|
|
814
|
|
|
return ""; |
|
815
|
|
|
} |
|
816
|
|
|
} |
|
817
|
|
|
|
|
818
|
|
|
/* |
|
819
|
|
|
* A method for the xml_set_external_entity_ref_handler() |
|
820
|
|
|
* |
|
821
|
|
|
* @param XMLParser $parser |
|
822
|
|
|
* @param string $ent |
|
823
|
|
|
* @param string|false $base |
|
824
|
|
|
* @param string $sysID |
|
825
|
|
|
* @param string|false $pubID |
|
826
|
|
|
* @return bool |
|
827
|
|
|
function extEntHandler($parser, $ent, $base, $sysID, $pubID) { |
|
828
|
|
|
print 'extEntHandler ran'; |
|
829
|
|
|
return true; |
|
830
|
|
|
} |
|
831
|
|
|
*/ |
|
832
|
|
|
|
|
833
|
|
|
/** |
|
834
|
|
|
* Function to convert an XML object into an array |
|
835
|
|
|
* |
|
836
|
|
|
* @param string $k Key |
|
837
|
|
|
* @param string $v Value |
|
838
|
|
|
* @return string |
|
839
|
|
|
*/ |
|
840
|
|
|
function rss_map_attrs($k, $v) |
|
841
|
|
|
{ |
|
842
|
|
|
return "$k=\"$v\""; |
|
843
|
|
|
} |
|
844
|
|
|
|
|
845
|
|
|
/** |
|
846
|
|
|
* Function to convert an XML object into an array |
|
847
|
|
|
* |
|
848
|
|
|
* @param SimpleXMLElement $xml Xml |
|
|
|
|
|
|
849
|
|
|
* @return array|string |
|
850
|
|
|
*/ |
|
851
|
|
|
function xml2php($xml) |
|
852
|
|
|
{ |
|
853
|
|
|
$threads = 0; |
|
854
|
|
|
$tab = false; |
|
855
|
|
|
$array = array(); |
|
856
|
|
|
foreach ($xml->children() as $key => $value) { |
|
857
|
|
|
$child = xml2php($value); |
|
858
|
|
|
|
|
859
|
|
|
//To deal with the attributes |
|
860
|
|
|
foreach ($value->attributes() as $ak => $av) { |
|
861
|
|
|
$child[$ak] = (string) $av; |
|
862
|
|
|
} |
|
863
|
|
|
|
|
864
|
|
|
//Let see if the new child is not in the array |
|
865
|
|
|
if ($tab === false && in_array($key, array_keys($array))) { |
|
866
|
|
|
//If this element is already in the array we will create an indexed array |
|
867
|
|
|
$tmp = $array[$key]; |
|
868
|
|
|
$array[$key] = null; |
|
869
|
|
|
$array[$key][] = $tmp; |
|
870
|
|
|
$array[$key][] = $child; |
|
871
|
|
|
$tab = true; |
|
872
|
|
|
} elseif ($tab === true) { |
|
873
|
|
|
//Add an element in an existing array |
|
874
|
|
|
$array[$key][] = $child; |
|
875
|
|
|
} else { |
|
876
|
|
|
//Add a simple element |
|
877
|
|
|
$array[$key] = $child; |
|
878
|
|
|
} |
|
879
|
|
|
|
|
880
|
|
|
$threads++; |
|
881
|
|
|
} |
|
882
|
|
|
|
|
883
|
|
|
|
|
884
|
|
|
if ($threads == 0) { |
|
885
|
|
|
return (string) $xml; |
|
886
|
|
|
} |
|
887
|
|
|
|
|
888
|
|
|
return $array; |
|
889
|
|
|
} |
|
890
|
|
|
|
alxarafe /
alixar
Loading history...