]> git.mxchange.org Git - friendica.git/blob - include/Scrape.php
Merge branch 'master', remote-tracking branch 'remotes/upstream/master'
[friendica.git] / include / Scrape.php
1 <?php
2
3 require_once('library/HTML5/Parser.php');
4 require_once('include/crypto.php');
5
6 if(! function_exists('scrape_dfrn')) {
7 function scrape_dfrn($url) {
8
9         $a = get_app();
10
11         $ret = array();
12
13         logger('scrape_dfrn: url=' . $url);
14
15         $s = fetch_url($url);
16
17         if(! $s) 
18                 return $ret;
19
20         $headers = $a->get_curl_headers();
21         logger('scrape_dfrn: headers=' . $headers, LOGGER_DEBUG);
22
23
24         $lines = explode("\n",$headers);
25         if(count($lines)) {
26                 foreach($lines as $line) {                              
27                         // don't try and run feeds through the html5 parser
28                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
29                                 return ret;
30                 }
31         }
32
33         try {
34                 $dom = HTML5_Parser::parse($s);
35         } catch (DOMException $e) {
36                 logger('scrape_dfrn: parse error: ' . $e);
37         }
38
39         if(! $dom)
40                 return $ret;
41
42         $items = $dom->getElementsByTagName('link');
43
44         // get DFRN link elements
45
46         foreach($items as $item) {
47                 $x = $item->getAttribute('rel');
48                 if(($x === 'alternate') && ($item->getAttribute('type') === 'application/atom+xml'))
49                         $ret['feed_atom'] = $item->getAttribute('href');
50                 if(substr($x,0,5) == "dfrn-") {
51                         $ret[$x] = $item->getAttribute('href');
52                 }
53                 if($x === 'lrdd') {
54                         $decoded = urldecode($item->getAttribute('href'));
55                         if(preg_match('/acct:([^@]*)@/',$decoded,$matches))
56                                 $ret['nick'] = $matches[1];
57                 }
58         }
59
60         // Pull out hCard profile elements
61
62         $largest_photo = 0;
63
64         $items = $dom->getElementsByTagName('*');
65         foreach($items as $item) {
66                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
67                         $level2 = $item->getElementsByTagName('*');
68                         foreach($level2 as $x) {
69                                 if(attribute_contains($x->getAttribute('class'),'fn')) {
70                                         $ret['fn'] = $x->textContent;
71                                 }
72                                 if((attribute_contains($x->getAttribute('class'),'photo'))
73                                         || (attribute_contains($x->getAttribute('class'),'avatar'))) {
74                                         $size = intval($x->getAttribute('width'));
75                                         // dfrn prefers 175, so if we find this, we set largest_size so it can't be topped.
76                                         if(($size > $largest_photo) || ($size == 175) || (! $largest_photo)) {
77                                                 $ret['photo'] = $x->getAttribute('src');
78                                                 $largest_photo = (($size == 175) ? 9999 : $size);
79                                         }
80                                 }
81                                 if(attribute_contains($x->getAttribute('class'),'key')) {
82                                         $ret['key'] = $x->textContent;
83                                 }
84                         }
85                 }
86         }
87
88         return $ret;
89 }}
90
91
92
93
94
95
96 if(! function_exists('validate_dfrn')) {
97 function validate_dfrn($a) {
98         $errors = 0;
99         if(! x($a,'key'))
100                 $errors ++;
101         if(! x($a,'dfrn-request'))
102                 $errors ++;
103         if(! x($a,'dfrn-confirm'))
104                 $errors ++;
105         if(! x($a,'dfrn-notify'))
106                 $errors ++;
107         if(! x($a,'dfrn-poll'))
108                 $errors ++;
109         return $errors;
110 }}
111
112 if(! function_exists('scrape_meta')) {
113 function scrape_meta($url) {
114
115         $a = get_app();
116
117         $ret = array();
118
119         logger('scrape_meta: url=' . $url);
120
121         $s = fetch_url($url);
122
123         if(! $s) 
124                 return $ret;
125
126         $headers = $a->get_curl_headers();
127         logger('scrape_meta: headers=' . $headers, LOGGER_DEBUG);
128
129         $lines = explode("\n",$headers);
130         if(count($lines)) {
131                 foreach($lines as $line) {                              
132                         // don't try and run feeds through the html5 parser
133                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
134                                 return ret;
135                 }
136         }
137
138         try {
139                 $dom = HTML5_Parser::parse($s);
140         } catch (DOMException $e) {
141                 logger('scrape_meta: parse error: ' . $e);
142         }
143
144         if(! $dom)
145                 return $ret;
146
147         $items = $dom->getElementsByTagName('meta');
148
149         // get DFRN link elements
150
151         foreach($items as $item) {
152                 $x = $item->getAttribute('name');
153                 if(substr($x,0,5) == "dfrn-")
154                         $ret[$x] = $item->getAttribute('content');
155         }
156
157         return $ret;
158 }}
159
160
161 if(! function_exists('scrape_vcard')) {
162 function scrape_vcard($url) {
163
164         $a = get_app();
165
166         $ret = array();
167
168         logger('scrape_vcard: url=' . $url);
169
170         $s = fetch_url($url);
171
172         if(! $s) 
173                 return $ret;
174
175         $headers = $a->get_curl_headers();
176         $lines = explode("\n",$headers);
177         if(count($lines)) {
178                 foreach($lines as $line) {                              
179                         // don't try and run feeds through the html5 parser
180                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
181                                 return ret;
182                 }
183         }
184
185         try {
186                 $dom = HTML5_Parser::parse($s);
187         } catch (DOMException $e) {
188                 logger('scrape_vcard: parse error: ' . $e);
189         }
190
191         if(! $dom)
192                 return $ret;
193
194         // Pull out hCard profile elements
195
196         $largest_photo = 0;
197
198         $items = $dom->getElementsByTagName('*');
199         foreach($items as $item) {
200                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
201                         $level2 = $item->getElementsByTagName('*');
202                         foreach($level2 as $x) {
203                                 if(attribute_contains($x->getAttribute('class'),'fn'))
204                                         $ret['fn'] = $x->textContent;
205                                 if((attribute_contains($x->getAttribute('class'),'photo'))
206                                         || (attribute_contains($x->getAttribute('class'),'avatar'))) {
207                                         $size = intval($x->getAttribute('width'));
208                                         if(($size > $largest_photo) || (! $largest_photo)) {
209                                                 $ret['photo'] = $x->getAttribute('src');
210                                                 $largest_photo = $size;
211                                         }
212                                 }
213                                 if((attribute_contains($x->getAttribute('class'),'nickname'))
214                                         || (attribute_contains($x->getAttribute('class'),'uid'))) {
215                                         $ret['nick'] = $x->textContent;
216                                 }
217                         }
218                 }
219         }
220
221         return $ret;
222 }}
223
224
225 if(! function_exists('scrape_feed')) {
226 function scrape_feed($url) {
227
228         $a = get_app();
229
230         $ret = array();
231         $s = fetch_url($url);
232
233         $headers = $a->get_curl_headers();
234         $code = $a->get_curl_code();
235
236         logger('scrape_feed: returns: ' . $code . ' headers=' . $headers, LOGGER_DEBUG);
237
238         if(! $s) {
239                 logger('scrape_feed: no data returned for ' . $url); 
240                 return $ret;
241         }
242
243
244         $lines = explode("\n",$headers);
245         if(count($lines)) {
246                 foreach($lines as $line) {                              
247                         if(stristr($line,'content-type:')) {
248                                 if(stristr($line,'application/atom+xml') || stristr($s,'<feed')) {
249                                         $ret['feed_atom'] = $url;
250                                         return $ret;
251                                 }
252                                 if(stristr($line,'application/rss+xml') || stristr($s,'<rss')) {
253                                         $ret['feed_rss'] = $url;
254                                         return $ret;
255                                 }
256                         }
257                 }
258                 // perhaps an RSS version 1 feed with a generic or incorrect content-type?
259                 if(stristr($s,'</item>')) {
260                         $ret['feed_rss'] = $url;
261                         return $ret;
262                 }
263         }
264
265         try {
266                 $dom = HTML5_Parser::parse($s);
267         } catch (DOMException $e) {
268                 logger('scrape_feed: parse error: ' . $e);
269         }
270
271         if(! $dom) {
272                 logger('scrape_feed: failed to parse.');
273                 return $ret;
274         }
275
276
277         $head = $dom->getElementsByTagName('base');
278         if($head) {
279                 foreach($head as $head0) {
280                         $basename = $head0->getAttribute('href');
281                         break;
282                 }
283         }
284         if(! $basename)
285                 $basename = substr($url,0,strrpos($url,'/')) . '/';
286
287         $items = $dom->getElementsByTagName('link');
288
289         // get Atom/RSS link elements, take the first one of either.
290
291         if($items) {
292                 foreach($items as $item) {
293                         $x = $item->getAttribute('rel');
294                         if(($x === 'alternate') && ($item->getAttribute('type') === 'application/atom+xml')) {
295                                 if(! x($ret,'feed_atom'))
296                                         $ret['feed_atom'] = $item->getAttribute('href');
297                         }
298                         if(($x === 'alternate') && ($item->getAttribute('type') === 'application/rss+xml')) {
299                                 if(! x($ret,'feed_rss'))
300                                         $ret['feed_rss'] = $item->getAttribute('href');
301                         }
302                 }       
303         }
304
305         // Drupal and perhaps others only provide relative URL's. Turn them into absolute.
306
307         if(x($ret,'feed_atom') && (! strstr($ret['feed_atom'],'://')))
308                 $ret['feed_atom'] = $basename . $ret['feed_atom'];
309         if(x($ret,'feed_rss') && (! strstr($ret['feed_rss'],'://')))
310                 $ret['feed_rss'] = $basename . $ret['feed_rss'];
311
312         return $ret;
313 }}
314
315
316 /**
317  *
318  * Probe a network address to discover what kind of protocols we need to communicate with it.
319  *
320  * Warning: this function is a bit touchy and there are some subtle dependencies within the logic flow.
321  * Edit with care.
322  *
323  */
324
325 /**
326  *
327  * PROBE_DIASPORA has a bias towards returning Diaspora information
328  * while PROBE_NORMAL has a bias towards dfrn/zot - in the case where
329  * an address (such as a Friendica address) supports more than one type
330  * of network. 
331  *
332  */
333
334
335 define ( 'PROBE_NORMAL',   0);
336 define ( 'PROBE_DIASPORA', 1);
337
338 function probe_url($url, $mode = PROBE_NORMAL) {
339         require_once('include/email.php');
340
341         $result = array();
342
343         if(! $url)
344                 return $result;
345
346         $network = null;
347         $diaspora = false;
348         $diaspora_base = '';
349         $diaspora_guid = '';    
350         $diaspora_key = '';
351         $has_lrdd = false;
352         $email_conversant = false;
353
354         $twitter = ((strpos($url,'twitter.com') !== false) ? true : false);
355
356         $at_addr = ((strpos($url,'@') !== false) ? true : false);
357
358         if(! $twitter) {
359
360                 if(strpos($url,'mailto:') !== false && $at_addr) {
361                         $url = str_replace('mailto:','',$url);
362                         $links = array();
363                 }
364                 else
365                         $links = lrdd($url);
366
367                 if(count($links)) {
368                         $has_lrdd = true;
369
370                         logger('probe_url: found lrdd links: ' . print_r($links,true), LOGGER_DATA);
371                         foreach($links as $link) {
372                                 if($link['@attributes']['rel'] === NAMESPACE_ZOT)
373                                         $zot = unamp($link['@attributes']['href']);
374                                 if($link['@attributes']['rel'] === NAMESPACE_DFRN)
375                                         $dfrn = unamp($link['@attributes']['href']);
376                                 if($link['@attributes']['rel'] === 'salmon')
377                                         $notify = unamp($link['@attributes']['href']);
378                                 if($link['@attributes']['rel'] === NAMESPACE_FEED)
379                                         $poll = unamp($link['@attributes']['href']);
380                                 if($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard')
381                                         $hcard = unamp($link['@attributes']['href']);
382                                 if($link['@attributes']['rel'] === 'http://webfinger.net/rel/profile-page')
383                                         $profile = unamp($link['@attributes']['href']);
384                                 if($link['@attributes']['rel'] === 'http://portablecontacts.net/spec/1.0')
385                                         $poco = unamp($link['@attributes']['href']);
386                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/seed_location') {
387                                         $diaspora_base = unamp($link['@attributes']['href']);
388                                         $diaspora = true;
389                                 }
390                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/guid') {
391                                         $diaspora_guid = unamp($link['@attributes']['href']);
392                                         $diaspora = true;
393                                 }
394                                 if($link['@attributes']['rel'] === 'diaspora-public-key') {
395                                         $diaspora_key = base64_decode(unamp($link['@attributes']['href']));
396                                         $pubkey = rsatopem($diaspora_key);
397                                         $diaspora = true;
398                                 }
399                         }
400
401                         // Status.Net can have more than one profile URL. We need to match the profile URL
402                         // to a contact on incoming messages to prevent spam, and we won't know which one
403                         // to match. So in case of two, one of them is stored as an alias. Only store URL's
404                         // and not webfinger user@host aliases. If they've got more than two non-email style
405                         // aliases, let's hope we're lucky and get one that matches the feed author-uri because 
406                         // otherwise we're screwed.
407
408                         foreach($links as $link) {
409                                 if($link['@attributes']['rel'] === 'alias') {
410                                         if(strpos($link['@attributes']['href'],'@') === false) {
411                                                 if(isset($profile)) {
412                                                         if($link['@attributes']['href'] !== $profile)
413                                                                 $alias = unamp($link['@attributes']['href']);
414                                                 }
415                                                 else
416                                                         $profile = unamp($link['@attributes']['href']);
417                                         }
418                                 }
419                         }
420                 }
421                 elseif($mode == PROBE_NORMAL) {
422
423                         // Check email
424
425                         $orig_url = $url;
426                         if((strpos($orig_url,'@')) && validate_email($orig_url)) {
427                                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1",
428                                         intval(local_user())
429                                 );
430                                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1",
431                                         intval(local_user())
432                                 );
433                                 if(count($x) && count($r)) {
434                                     $mailbox = construct_mailbox_name($r[0]);
435                                         $password = '';
436                                         openssl_private_decrypt(hex2bin($r[0]['pass']),$password,$x[0]['prvkey']);
437                                         $mbox = email_connect($mailbox,$r[0]['user'],$password);
438                                         unset($password);
439                                 }
440                                 if($mbox) {
441                                         $msgs = email_poll($mbox,$orig_url);
442                                         if(count($msgs)) {
443                                                 $addr = $orig_url;
444                                                 $network = NETWORK_MAIL;
445                                                 $name = substr($url,0,strpos($url,'@'));
446                                                 $phost = substr($url,strpos($url,'@')+1);
447                                                 $profile = 'http://' . $phost;
448                                                 // fix nick character range
449                                                 $vcard = array('fn' => $name, 'nick' => $name, 'photo' => avatar_img($url));
450                                                 $notify = 'smtp ' . random_string();
451                                                 $poll = 'email ' . random_string();
452                                                 $priority = 0;
453                                                 $x = email_msg_meta($mbox,$msgs[0]);
454                                                 if(stristr($x->from,$orig_url))
455                                                         $adr = imap_rfc822_parse_adrlist($x->from,'');
456                                                 elseif(stristr($x->to,$orig_url))
457                                                         $adr = imap_rfc822_parse_adrlist($x->to,'');
458                                                 if(isset($adr)) {
459                                                         foreach($adr as $feadr) {
460                                                                 if((strcasecmp($feadr->mailbox,$name) == 0)
461                                                                         &&(strcasecmp($feadr->host,$phost) == 0)
462                                                                         && (strlen($feadr->personal))) {
463
464                                                                         $personal = imap_mime_header_decode($feadr->personal);
465                                                                         $vcard['fn'] = "";
466                                                                         foreach($personal as $perspart)
467                                                                                 if ($perspart->charset != "default")
468                                                                                         $vcard['fn'] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
469                                                                                 else
470                                                                                         $vcard['fn'] .= $perspart->text;
471
472                                                                         $vcard['fn'] = notags($vcard['fn']);
473                                                                 }
474                                                         }
475                                                 }
476                                         }
477                                         imap_close($mbox);
478                                 }
479                         }
480                 }
481         }       
482
483         if($mode == PROBE_NORMAL) {
484                 if(strlen($zot)) {
485                         $s = fetch_url($zot);
486                         if($s) {
487                                 $j = json_decode($s);
488                                 if($j) {
489                                         $network = NETWORK_ZOT;
490                                         $vcard   = array(
491                                                 'fn'    => $j->fullname, 
492                                                 'nick'  => $j->nickname, 
493                                                 'photo' => $j->photo
494                                         );
495                                         $profile  = $j->url;
496                                         $notify   = $j->post;
497                                         $pubkey   = $j->pubkey;
498                                         $poll     = 'N/A';
499                                 }
500                         }
501                 }
502
503                 if(strlen($dfrn)) {
504                         $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn);
505                         if(is_array($ret) && x($ret,'dfrn-request')) {
506                                 $network = NETWORK_DFRN;
507                                 $request = $ret['dfrn-request'];
508                                 $confirm = $ret['dfrn-confirm'];
509                                 $notify  = $ret['dfrn-notify'];
510                                 $poll    = $ret['dfrn-poll'];
511
512                                 $vcard = array();
513                                 $vcard['fn'] = $ret['fn'];
514                                 $vcard['nick'] = $ret['nick'];
515                                 $vcard['photo'] = $ret['photo'];
516                         }
517                 }
518         }
519
520         if($diaspora && $diaspora_base && $diaspora_guid) {
521                 if($mode == PROBE_DIASPORA || ! $notify) {
522                         $notify = $diaspora_base . 'receive/users/' . $diaspora_guid;
523                         $batch  = $diaspora_base . 'receive/public' ;
524                 }
525                 if(strpos($url,'@'))
526                         $addr = str_replace('acct:', '', $url);
527         }                       
528
529         if($network !== NETWORK_ZOT && $network !== NETWORK_DFRN && $network !== NETWORK_MAIL) {
530                 if($diaspora)
531                         $network = NETWORK_DIASPORA;
532                 elseif($has_lrdd)
533                         $network  = NETWORK_OSTATUS;
534                 $priority = 0;
535
536                 if($hcard && ! $vcard) {
537                         $vcard = scrape_vcard($hcard);
538
539                         // Google doesn't use absolute url in profile photos
540         
541                         if((x($vcard,'photo')) && substr($vcard['photo'],0,1) == '/') {
542                                 $h = @parse_url($hcard);
543                                 if($h)
544                                         $vcard['photo'] = $h['scheme'] . '://' . $h['host'] . $vcard['photo'];
545                         }
546                 
547                         logger('probe_url: scrape_vcard: ' . print_r($vcard,true), LOGGER_DATA);
548                 }
549
550                 if($twitter) {          
551                         logger('twitter: setup');
552                         $tid = basename($url);
553                         $tapi = 'https://api.twitter.com/1/statuses/user_timeline.rss';
554                         if(intval($tid))
555                                 $poll = $tapi . '?user_id=' . $tid;
556                         else
557                                 $poll = $tapi . '?screen_name=' . $tid;
558                         $profile = 'http://twitter.com/#!/' . $tid;
559                         $vcard['photo'] = 'https://api.twitter.com/1/users/profile_image/' . $tid;
560                         $vcard['nick'] = $tid;
561                         $vcard['fn'] = $tid . '@twitter';
562                 }
563
564                 if(! x($vcard,'fn'))
565                         if(x($vcard,'nick'))
566                                 $vcard['fn'] = $vcard['nick'];
567
568                 $check_feed = false;
569
570                 if($twitter || ! $poll)
571                         $check_feed = true;
572                 if((! isset($vcard)) || (! x($vcard,'fn')) || (! $profile))
573                         $check_feed = true;
574                 if(($at_addr) && (! count($links)))
575                         $check_feed = false;
576
577                 if($check_feed) {
578
579                         $feedret = scrape_feed(($poll) ? $poll : $url);
580                         logger('probe_url: scrape_feed ' . (($poll)? $poll : $url) . ' returns: ' . print_r($feedret,true), LOGGER_DATA);
581                         if(count($feedret) && ($feedret['feed_atom'] || $feedret['feed_rss'])) {
582                                 $poll = ((x($feedret,'feed_atom')) ? unamp($feedret['feed_atom']) : unamp($feedret['feed_rss']));
583                                 if(! x($vcard)) 
584                                         $vcard = array();
585                         }
586
587                         if(x($feedret,'photo') && (! x($vcard,'photo')))
588                                 $vcard['photo'] = $feedret['photo'];
589                         require_once('library/simplepie/simplepie.inc');
590                     $feed = new SimplePie();
591                         $xml = fetch_url($poll);
592
593                         logger('probe_url: fetch feed: ' . $poll . ' returns: ' . $xml, LOGGER_DATA);
594                         $a = get_app();
595
596                         logger('probe_url: scrape_feed: headers: ' . $a->get_curl_headers(), LOGGER_DATA);
597
598                         $feed->set_raw_data($xml);
599
600                     $feed->init();
601                         if($feed->error())
602                                 logger('probe_url: scrape_feed: Error parsing XML: ' . $feed->error());
603
604
605                         if(! x($vcard,'photo'))
606                                 $vcard['photo'] = $feed->get_image_url();
607                         $author = $feed->get_author();
608
609                         if($author) {                   
610                                 $vcard['fn'] = unxmlify(trim($author->get_name()));
611                                 if(! $vcard['fn'])
612                                         $vcard['fn'] = trim(unxmlify($author->get_email()));
613                                 if(strpos($vcard['fn'],'@') !== false)
614                                         $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
615                                 $email = unxmlify($author->get_email());
616                                 if(! $profile && $author->get_link())
617                                         $profile = trim(unxmlify($author->get_link()));
618                                 if(! $vcard['photo']) {
619                                         $rawtags = $feed->get_feed_tags( SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
620                                 if($rawtags) {
621                                                 $elems = $rawtags[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10];
622                                                 if((x($elems,'link')) && ($elems['link'][0]['attribs']['']['rel'] === 'photo'))
623                                                         $vcard['photo'] = $elems['link'][0]['attribs']['']['href'];
624                                 }
625                                 }
626                         }
627                         else {
628                                 $item = $feed->get_item(0);
629                                 if($item) {
630                                         $author = $item->get_author();
631                                         if($author) {                   
632                                                 $vcard['fn'] = trim(unxmlify($author->get_name()));
633                                                 if(! $vcard['fn'])
634                                                         $vcard['fn'] = trim(unxmlify($author->get_email()));
635                                                 if(strpos($vcard['fn'],'@') !== false)
636                                                         $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
637                                                 $email = unxmlify($author->get_email());
638                                                 if(! $profile && $author->get_link())
639                                                         $profile = trim(unxmlify($author->get_link()));
640                                         }
641                                         if(! $vcard['photo']) {
642                                                 $rawmedia = $item->get_item_tags('http://search.yahoo.com/mrss/','thumbnail');
643                                                 if($rawmedia && $rawmedia[0]['attribs']['']['url'])
644                                                         $vcard['photo'] = unxmlify($rawmedia[0]['attribs']['']['url']);
645                                         }
646                                         if(! $vcard['photo']) {
647                                                 $rawtags = $item->get_item_tags( SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
648                                         if($rawtags) {
649                                                         $elems = $rawtags[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10];
650                                                         if((x($elems,'link')) && ($elems['link'][0]['attribs']['']['rel'] === 'photo'))
651                                                                 $vcard['photo'] = $elems['link'][0]['attribs']['']['href'];
652                                         }
653                                         }
654                                 }
655                         }
656
657                         if((! $vcard['photo']) && strlen($email))
658                                 $vcard['photo'] = avatar_img($email);
659                         if($poll === $profile)
660                                 $lnk = $feed->get_permalink();
661                         if(isset($lnk) && strlen($lnk))
662                                 $profile = $lnk;        
663
664                         if(! (x($vcard,'fn')))
665                                 $vcard['fn'] = notags($feed->get_title());
666                         if(! (x($vcard,'fn')))
667                                 $vcard['fn'] = notags($feed->get_description());
668
669                         if(strpos($vcard['fn'],'Twitter / ') !== false) {
670                                 $vcard['fn'] = substr($vcard['fn'],strpos($vcard['fn'],'/')+1);
671                                 $vcard['fn'] = trim($vcard['fn']);
672                         }
673                         if(! x($vcard,'nick')) {
674                                 $vcard['nick'] = strtolower(notags(unxmlify($vcard['fn'])));
675                                 if(strpos($vcard['nick'],' '))
676                                         $vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' ')));
677                         }
678                         if(! $network)
679                                 $network = NETWORK_FEED;
680                         if(! $priority)
681                                 $priority = 2;
682                 }
683         }
684
685         if(! x($vcard,'photo')) {
686                 $a = get_app();
687                 $vcard['photo'] = $a->get_baseurl() . '/images/person-175.jpg' ; 
688         }
689
690         if(! $profile)
691                 $profile = $url;
692
693         // No human could be associated with this link, use the URL as the contact name
694
695         if(($network === NETWORK_FEED) && ($poll) && (! x($vcard,'fn')))
696                 $vcard['fn'] = $url;
697
698         $vcard['fn'] = notags($vcard['fn']);
699         $vcard['nick'] = str_replace(' ','',notags($vcard['nick']));
700                 
701         $result['name'] = $vcard['fn'];
702         $result['nick'] = $vcard['nick'];
703         $result['url'] = $profile;
704         $result['addr'] = $addr;
705         $result['batch'] = $batch;
706         $result['notify'] = $notify;
707         $result['poll'] = $poll;
708         $result['request'] = $request;
709         $result['confirm'] = $confirm;
710         $result['poco'] = $poco;
711         $result['photo'] = $vcard['photo'];
712         $result['priority'] = $priority;
713         $result['network'] = $network;
714         $result['alias'] = $alias;
715         $result['pubkey'] = $pubkey;
716
717         logger('probe_url: ' . print_r($result,true), LOGGER_DEBUG);
718
719         return $result;
720 }