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