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         $basename = implode('/', array_slice(explode('/',$url),0,3)) . '/';
 266
 267         $doc = new DOMDocument();
 268         @$doc->loadHTML($s);
 269         $xpath = new DomXPath($doc);
 270
 271         $base = $xpath->query("//base");
 272         foreach ($base as $node) {
 273                 $attr = array();
 274
 275                 if ($node->attributes->length)
 276                         foreach ($node->attributes as $attribute)
 277                                 $attr[$attribute->name] = $attribute->value;
 278
 279                 if ($attr["href"] != "")
 280                         $basename = $attr["href"] ;
 281         }
 282
 283         $list = $xpath->query("//link");
 284         foreach ($list as $node) {
 285                 $attr = array();
 286
 287                 if ($node->attributes->length)
 288                         foreach ($node->attributes as $attribute)
 289                                 $attr[$attribute->name] = $attribute->value;
 290
 291                 if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/atom+xml"))
 292                         $ret["feed_atom"] = $attr["href"];
 293
 294                 if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/rss+xml"))
 295                         $ret["feed_rss"] = $attr["href"];
 296         }
 297
 298         // Drupal and perhaps others only provide relative URLs. Turn them into absolute.
 299
 300         if(x($ret,'feed_atom') && (! strstr($ret['feed_atom'],'://')))
 301                 $ret['feed_atom'] = $basename . $ret['feed_atom'];
 302         if(x($ret,'feed_rss') && (! strstr($ret['feed_rss'],'://')))
 303                 $ret['feed_rss'] = $basename . $ret['feed_rss'];
 304
 305         return $ret;
 306 }}
 307
 308
 309 /**
 310  *
 311  * Probe a network address to discover what kind of protocols we need to communicate with it.
 312  *
 313  * Warning: this function is a bit touchy and there are some subtle dependencies within the logic flow.
 314  * Edit with care.
 315  *
 316  */
 317
 318 /**
 319  *
 320  * PROBE_DIASPORA has a bias towards returning Diaspora information
 321  * while PROBE_NORMAL has a bias towards dfrn/zot - in the case where
 322  * an address (such as a Friendica address) supports more than one type
 323  * of network.
 324  *
 325  */
 326
 327
 328 define ( 'PROBE_NORMAL',   0);
 329 define ( 'PROBE_DIASPORA', 1);
 330
 331 function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
 332         require_once('include/email.php');
 333
 334         $result = array();
 335
 336         if(! $url)
 337                 return $result;
 338
 339         $result = Cache::get("probe_url:".$mode.":".$url);
 340         if (!is_null($result)) {
 341                 $result = unserialize($result);
 342                 return $result;
 343         }
 344
 345         $network = null;
 346         $diaspora = false;
 347         $diaspora_base = '';
 348         $diaspora_guid = '';
 349         $diaspora_key = '';
 350         $has_lrdd = false;
 351         $email_conversant = false;
 352         $connectornetworks = false;
 353         $appnet = false;
 354
 355         if (strpos($url,'twitter.com')) {
 356                 $connectornetworks = true;
 357                 $network = NETWORK_TWITTER;
 358         }
 359
 360         // Twitter is deactivated since twitter closed its old API
 361         //$twitter = ((strpos($url,'twitter.com') !== false) ? true : false);
 362         $lastfm  = ((strpos($url,'last.fm/user') !== false) ? true : false);
 363
 364         $at_addr = ((strpos($url,'@') !== false) ? true : false);
 365
 366         if((!$appnet) && (!$lastfm) && !$connectornetworks) {
 367
 368                 if(strpos($url,'mailto:') !== false && $at_addr) {
 369                         $url = str_replace('mailto:','',$url);
 370                         $links = array();
 371                 }
 372                 else
 373                         $links = lrdd($url);
 374
 375                 if(count($links)) {
 376                         $has_lrdd = true;
 377
 378                         logger('probe_url: found lrdd links: ' . print_r($links,true), LOGGER_DATA);
 379                         foreach($links as $link) {
 380                                 if($link['@attributes']['rel'] === NAMESPACE_ZOT)
 381                                         $zot = unamp($link['@attributes']['href']);
 382                                 if($link['@attributes']['rel'] === NAMESPACE_DFRN)
 383                                         $dfrn = unamp($link['@attributes']['href']);
 384                                 if($link['@attributes']['rel'] === 'salmon')
 385                                         $notify = unamp($link['@attributes']['href']);
 386                                 if($link['@attributes']['rel'] === NAMESPACE_FEED)
 387                                         $poll = unamp($link['@attributes']['href']);
 388                                 if($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard')
 389                                         $hcard = unamp($link['@attributes']['href']);
 390                                 if($link['@attributes']['rel'] === 'http://webfinger.net/rel/profile-page')
 391                                         $profile = unamp($link['@attributes']['href']);
 392                                 if($link['@attributes']['rel'] === 'http://portablecontacts.net/spec/1.0')
 393                                         $poco = unamp($link['@attributes']['href']);
 394                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/seed_location') {
 395                                         $diaspora_base = unamp($link['@attributes']['href']);
 396                                         $diaspora = true;
 397                                 }
 398                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/guid') {
 399                                         $diaspora_guid = unamp($link['@attributes']['href']);
 400                                         $diaspora = true;
 401                                 }
 402                                 if($link['@attributes']['rel'] === 'diaspora-public-key') {
 403                                         $diaspora_key = base64_decode(unamp($link['@attributes']['href']));
 404                                         if(strstr($diaspora_key,'RSA '))
 405                                                 $pubkey = rsatopem($diaspora_key);
 406                                         else
 407                                                 $pubkey = $diaspora_key;
 408                                         $diaspora = true;
 409                                 }
 410                                 if(($link['@attributes']['rel'] === 'http://ostatus.org/schema/1.0/subscribe') AND ($mode == PROBE_NORMAL)) {
 411                                         $diaspora = false;
 412                                 }
 413                         }
 414
 415                         // Status.Net can have more than one profile URL. We need to match the profile URL
 416                         // to a contact on incoming messages to prevent spam, and we won't know which one
 417                         // to match. So in case of two, one of them is stored as an alias. Only store URL's
 418                         // and not webfinger user@host aliases. If they've got more than two non-email style
 419                         // aliases, let's hope we're lucky and get one that matches the feed author-uri because
 420                         // otherwise we're screwed.
 421
 422                         foreach($links as $link) {
 423                                 if($link['@attributes']['rel'] === 'alias') {
 424                                         if(strpos($link['@attributes']['href'],'@') === false) {
 425                                                 if(isset($profile)) {
 426                                                         if($link['@attributes']['href'] !== $profile)
 427                                                                 $alias = unamp($link['@attributes']['href']);
 428                                                 }
 429                                                 else
 430                                                         $profile = unamp($link['@attributes']['href']);
 431                                         }
 432                                 }
 433                         }
 434
 435                         // If the profile is different from the url then the url is abviously an alias
 436                         if (($alias == "") AND ($profile != "") AND !$at_addr AND (normalise_link($profile) != normalise_link($url)))
 437                                 $alias = $url;
 438                 }
 439                 elseif($mode == PROBE_NORMAL) {
 440
 441                         // Check email
 442
 443                         $orig_url = $url;
 444                         if((strpos($orig_url,'@')) && validate_email($orig_url)) {
 445                                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1",
 446                                         intval(local_user())
 447                                 );
 448                                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1",
 449                                         intval(local_user())
 450                                 );
 451                                 if(count($x) && count($r)) {
 452                                         $mailbox = construct_mailbox_name($r[0]);
 453                                         $password = '';
 454                                         openssl_private_decrypt(hex2bin($r[0]['pass']),$password,$x[0]['prvkey']);
 455                                         $mbox = email_connect($mailbox,$r[0]['user'],$password);
 456                                         if(! $mbox)
 457                                                 logger('probe_url: email_connect failed.');
 458                                         unset($password);
 459                                 }
 460                                 if($mbox) {
 461                                         $msgs = email_poll($mbox,$orig_url);
 462                                         logger('probe_url: searching ' . $orig_url . ', ' . count($msgs) . ' messages found.', LOGGER_DEBUG);
 463                                         if(count($msgs)) {
 464                                                 $addr = $orig_url;
 465                                                 $network = NETWORK_MAIL;
 466                                                 $name = substr($url,0,strpos($url,'@'));
 467                                                 $phost = substr($url,strpos($url,'@')+1);
 468                                                 $profile = 'http://' . $phost;
 469                                                 // fix nick character range
 470                                                 $vcard = array('fn' => $name, 'nick' => $name, 'photo' => avatar_img($url));
 471                                                 $notify = 'smtp ' . random_string();
 472                                                 $poll = 'email ' . random_string();
 473                                                 $priority = 0;
 474                                                 $x = email_msg_meta($mbox,$msgs[0]);
 475                                                 if(stristr($x[0]->from,$orig_url))
 476                                                         $adr = imap_rfc822_parse_adrlist($x[0]->from,'');
 477                                                 elseif(stristr($x[0]->to,$orig_url))
 478                                                         $adr = imap_rfc822_parse_adrlist($x[0]->to,'');
 479                                                 if(isset($adr)) {
 480                                                         foreach($adr as $feadr) {
 481                                                                 if((strcasecmp($feadr->mailbox,$name) == 0)
 482                                                                         &&(strcasecmp($feadr->host,$phost) == 0)
 483                                                                         && (strlen($feadr->personal))) {
 484
 485                                                                         $personal = imap_mime_header_decode($feadr->personal);
 486                                                                         $vcard['fn'] = "";
 487                                                                         foreach($personal as $perspart)
 488                                                                                 if ($perspart->charset != "default")
 489                                                                                         $vcard['fn'] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
 490                                                                                 else
 491                                                                                         $vcard['fn'] .= $perspart->text;
 492
 493                                                                         $vcard['fn'] = notags($vcard['fn']);
 494                                                                 }
 495                                                         }
 496                                                 }
 497                                         }
 498                                         imap_close($mbox);
 499                                 }
 500                         }
 501                 }
 502         }
 503
 504         if($mode == PROBE_NORMAL) {
 505
 506                 if(strlen($zot)) {
 507                         $s = fetch_url($zot);
 508                         if($s) {
 509                                 $j = json_decode($s);
 510                                 if($j) {
 511                                         $network = NETWORK_ZOT;
 512                                         $vcard   = array(
 513                                                 'fn'    => $j->fullname,
 514                                                 'nick'  => $j->nickname,
 515                                                 'photo' => $j->photo
 516                                         );
 517                                         $profile  = $j->url;
 518                                         $notify   = $j->post;
 519                                         $pubkey   = $j->pubkey;
 520                                         $poll     = 'N/A';
 521                                 }
 522                         }
 523                 }
 524
 525
 526                 if(strlen($dfrn)) {
 527                         $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn);
 528                         if(is_array($ret) && x($ret,'dfrn-request')) {
 529                                 $network = NETWORK_DFRN;
 530                                 $request = $ret['dfrn-request'];
 531                                 $confirm = $ret['dfrn-confirm'];
 532                                 $notify  = $ret['dfrn-notify'];
 533                                 $poll    = $ret['dfrn-poll'];
 534
 535                                 $vcard = array();
 536                                 $vcard['fn'] = $ret['fn'];
 537                                 $vcard['nick'] = $ret['nick'];
 538                                 $vcard['photo'] = $ret['photo'];
 539                         }
 540                 }
 541         }
 542
 543         if($diaspora && $diaspora_base && $diaspora_guid) {
 544                 if($mode == PROBE_DIASPORA || ! $notify) {
 545                         $notify = $diaspora_base . 'receive/users/' . $diaspora_guid;
 546                         $batch  = $diaspora_base . 'receive/public' ;
 547                 }
 548                 if(strpos($url,'@'))
 549                         $addr = str_replace('acct:', '', $url);
 550         }
 551
 552         if($network !== NETWORK_ZOT && $network !== NETWORK_DFRN && $network !== NETWORK_MAIL) {
 553                 if($diaspora)
 554                         $network = NETWORK_DIASPORA;
 555                 elseif($has_lrdd AND ($notify))
 556                         $network  = NETWORK_OSTATUS;
 557
 558                 if(strpos($url,'@'))
 559                         $addr = str_replace('acct:', '', $url);
 560
 561                 $priority = 0;
 562
 563                 if($hcard && ! $vcard) {
 564                         $vcard = scrape_vcard($hcard);
 565
 566                         // Google doesn't use absolute url in profile photos
 567
 568                         if((x($vcard,'photo')) && substr($vcard['photo'],0,1) == '/') {
 569                                 $h = @parse_url($hcard);
 570                                 if($h)
 571                                         $vcard['photo'] = $h['scheme'] . '://' . $h['host'] . $vcard['photo'];
 572                         }
 573
 574                         logger('probe_url: scrape_vcard: ' . print_r($vcard,true), LOGGER_DATA);
 575                 }
 576
 577                 if($diaspora && $addr) {
 578                         // Diaspora returns the name as the nick. As the nick will never be updated,
 579                         // let's use the Diaspora nickname (the first part of the handle) as the nick instead
 580                         $addr_parts = explode('@', $addr);
 581                         $vcard['nick'] = $addr_parts[0];
 582                 }
 583
 584                 /* if($twitter) {
 585                         logger('twitter: setup');
 586                         $tid = basename($url);
 587                         $tapi = 'https://api.twitter.com/1/statuses/user_timeline.rss';
 588                         if(intval($tid))
 589                                 $poll = $tapi . '?user_id=' . $tid;
 590                         else
 591                                 $poll = $tapi . '?screen_name=' . $tid;
 592                         $profile = 'http://twitter.com/#!/' . $tid;
 593                         //$vcard['photo'] = 'https://api.twitter.com/1/users/profile_image/' . $tid;
 594                         $vcard['photo'] = 'https://api.twitter.com/1/users/profile_image?screen_name=' . $tid . '&size=bigger';
 595                         $vcard['nick'] = $tid;
 596                         $vcard['fn'] = $tid;
 597                 } */
 598
 599                 if($lastfm) {
 600                         $profile = $url;
 601                         $poll = str_replace(array('www.','last.fm/'),array('','ws.audioscrobbler.com/1.0/'),$url) . '/recenttracks.rss';
 602                         $vcard['nick'] = basename($url);
 603                         $vcard['fn'] = $vcard['nick'] . t(' on Last.fm');
 604                         $network = NETWORK_FEED;
 605                 }
 606
 607                 if(! x($vcard,'fn'))
 608                         if(x($vcard,'nick'))
 609                                 $vcard['fn'] = $vcard['nick'];
 610
 611                 $check_feed = false;
 612
 613                 if(stristr($url,'tumblr.com') && (! stristr($url,'/rss'))) {
 614                         $poll = $url . '/rss';
 615                         $check_feed = true;
 616                         // Will leave it to others to figure out how to grab the avatar, which is on the $url page in the open graph meta links
 617                 }
 618
 619                 if($appnet || ! $poll)
 620                         $check_feed = true;
 621                 if((! isset($vcard)) || (! x($vcard,'fn')) || (! $profile))
 622                         $check_feed = true;
 623                 if(($at_addr) && (! count($links)))
 624                         $check_feed = false;
 625
 626                 if ($connectornetworks)
 627                         $check_feed = false;
 628
 629                 if($check_feed) {
 630
 631                         $feedret = scrape_feed(($poll) ? $poll : $url);
 632
 633                         logger('probe_url: scrape_feed ' . (($poll)? $poll : $url) . ' returns: ' . print_r($feedret,true), LOGGER_DATA);
 634                         if(count($feedret) && ($feedret['feed_atom'] || $feedret['feed_rss'])) {
 635                                 $poll = ((x($feedret,'feed_atom')) ? unamp($feedret['feed_atom']) : unamp($feedret['feed_rss']));
 636                                 if(! x($vcard))
 637                                         $vcard = array();
 638                         }
 639
 640                         if(x($feedret,'photo') && (! x($vcard,'photo')))
 641                                 $vcard['photo'] = $feedret['photo'];
 642                         require_once('library/simplepie/simplepie.inc');
 643                         $feed = new SimplePie();
 644                         $xml = fetch_url($poll);
 645
 646                         logger('probe_url: fetch feed: ' . $poll . ' returns: ' . $xml, LOGGER_DATA);
 647                         $a = get_app();
 648
 649                         logger('probe_url: scrape_feed: headers: ' . $a->get_curl_headers(), LOGGER_DATA);
 650
 651                         // Don't try and parse an empty string
 652                         $feed->set_raw_data(($xml) ? $xml : '<?xml version="1.0" encoding="utf-8" ?><xml></xml>');
 653
 654                         $feed->init();
 655                         if($feed->error()) {
 656                                 logger('probe_url: scrape_feed: Error parsing XML: ' . $feed->error());
 657                                 $network = NETWORK_PHANTOM;
 658                         }
 659
 660                         if(! x($vcard,'photo'))
 661                                 $vcard['photo'] = $feed->get_image_url();
 662                         $author = $feed->get_author();
 663
 664                         if($author) {
 665                                 $vcard['fn'] = unxmlify(trim($author->get_name()));
 666                                 if(! $vcard['fn'])
 667                                         $vcard['fn'] = trim(unxmlify($author->get_email()));
 668                                 if(strpos($vcard['fn'],'@') !== false)
 669                                         $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
 670
 671                                 $email = unxmlify($author->get_email());
 672                                 if(! $profile && $author->get_link())
 673                                         $profile = trim(unxmlify($author->get_link()));
 674                                 if(! $vcard['photo']) {
 675                                         $rawtags = $feed->get_feed_tags( SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
 676                                         if($rawtags) {
 677                                                 $elems = $rawtags[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10];
 678                                                 if((x($elems,'link')) && ($elems['link'][0]['attribs']['']['rel'] === 'photo'))
 679                                                         $vcard['photo'] = $elems['link'][0]['attribs']['']['href'];
 680                                         }
 681                                 }
 682                                 // Fetch fullname via poco:displayName
 683                                 $pocotags = $feed->get_feed_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
 684                                 if ($pocotags) {
 685                                         $elems = $pocotags[0]['child']['http://portablecontacts.net/spec/1.0'];
 686                                         if (isset($elems["displayName"]))
 687                                                 $vcard['fn'] = $elems["displayName"][0]["data"];
 688                                         if (isset($elems["preferredUsername"]))
 689                                                 $vcard['nick'] = $elems["preferredUsername"][0]["data"];
 690                                 }
 691                         }
 692                         else {
 693                                 $item = $feed->get_item(0);
 694                                 if($item) {
 695                                         $author = $item->get_author();
 696                                         if($author) {
 697                                                 $vcard['fn'] = trim(unxmlify($author->get_name()));
 698                                                 if(! $vcard['fn'])
 699                                                         $vcard['fn'] = trim(unxmlify($author->get_email()));
 700                                                 if(strpos($vcard['fn'],'@') !== false)
 701                                                         $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
 702                                                 $email = unxmlify($author->get_email());
 703                                                 if(! $profile && $author->get_link())
 704                                                         $profile = trim(unxmlify($author->get_link()));
 705                                         }
 706                                         if(! $vcard['photo']) {
 707                                                 $rawmedia = $item->get_item_tags('http://search.yahoo.com/mrss/','thumbnail');
 708                                                 if($rawmedia && $rawmedia[0]['attribs']['']['url'])
 709                                                         $vcard['photo'] = unxmlify($rawmedia[0]['attribs']['']['url']);
 710                                         }
 711                                         if(! $vcard['photo']) {
 712                                                 $rawtags = $item->get_item_tags( SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
 713                                                 if($rawtags) {
 714                                                         $elems = $rawtags[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10];
 715                                                         if((x($elems,'link')) && ($elems['link'][0]['attribs']['']['rel'] === 'photo'))
 716                                                                 $vcard['photo'] = $elems['link'][0]['attribs']['']['href'];
 717                                                 }
 718                                         }
 719                                 }
 720                         }
 721
 722                         if((! $vcard['photo']) && strlen($email))
 723                                 $vcard['photo'] = avatar_img($email);
 724                         if($poll === $profile)
 725                                 $lnk = $feed->get_permalink();
 726                         if(isset($lnk) && strlen($lnk))
 727                                 $profile = $lnk;
 728
 729                         if(! $network) {
 730                                 $network = NETWORK_FEED;
 731                                 // If it is a feed, don't take the author name as feed name
 732                                 unset($vcard['fn']);
 733                         }
 734                         if(! (x($vcard,'fn')))
 735                                 $vcard['fn'] = notags($feed->get_title());
 736                         if(! (x($vcard,'fn')))
 737                                 $vcard['fn'] = notags($feed->get_description());
 738
 739                         if(strpos($vcard['fn'],'Twitter / ') !== false) {
 740                                 $vcard['fn'] = substr($vcard['fn'],strpos($vcard['fn'],'/')+1);
 741                                 $vcard['fn'] = trim($vcard['fn']);
 742                         }
 743                         if(! x($vcard,'nick')) {
 744                                 $vcard['nick'] = strtolower(notags(unxmlify($vcard['fn'])));
 745                                 if(strpos($vcard['nick'],' '))
 746                                         $vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' ')));
 747                         }
 748                         if(! $priority)
 749                                 $priority = 2;
 750                 }
 751         }
 752
 753         if(! x($vcard,'photo')) {
 754                 $a = get_app();
 755                 $vcard['photo'] = $a->get_baseurl() . '/images/person-175.jpg' ;
 756         }
 757
 758         if(! $profile)
 759                 $profile = $url;
 760
 761         // No human could be associated with this link, use the URL as the contact name
 762
 763         if(($network === NETWORK_FEED) && ($poll) && (! x($vcard,'fn')))
 764                 $vcard['fn'] = $url;
 765
 766         if (($notify != "") AND ($poll != "")) {
 767                 $baseurl = matching(normalise_link($notify), normalise_link($poll));
 768
 769                 $baseurl2 = matching($baseurl, normalise_link($profile));
 770                 if ($baseurl2 != "")
 771                         $baseurl = $baseurl2;
 772         }
 773
 774         if (($baseurl == "") AND ($notify != ""))
 775                 $baseurl = matching(normalise_link($profile), normalise_link($notify));
 776
 777         if (($baseurl == "") AND ($poll != ""))
 778                 $baseurl = matching(normalise_link($profile), normalise_link($poll));
 779
 780         $baseurl = rtrim($baseurl, "/");
 781
 782         if(strpos($url,'@') AND ($addr == "") AND ($network == NETWORK_DFRN))
 783                 $addr = str_replace('acct:', '', $url);
 784
 785         $vcard['fn'] = notags($vcard['fn']);
 786         $vcard['nick'] = str_replace(' ','',notags($vcard['nick']));
 787
 788         $result['name'] = $vcard['fn'];
 789         $result['nick'] = $vcard['nick'];
 790         $result['url'] = $profile;
 791         $result['addr'] = $addr;
 792         $result['batch'] = $batch;
 793         $result['notify'] = $notify;
 794         $result['poll'] = $poll;
 795         $result['request'] = $request;
 796         $result['confirm'] = $confirm;
 797         $result['poco'] = $poco;
 798         $result['photo'] = $vcard['photo'];
 799         $result['priority'] = $priority;
 800         $result['network'] = $network;
 801         $result['alias'] = $alias;
 802         $result['pubkey'] = $pubkey;
 803         $result['baseurl'] = $baseurl;
 804
 805         logger('probe_url: ' . print_r($result,true), LOGGER_DEBUG);
 806
 807         if ($level == 1) {
 808                 // Trying if it maybe a diaspora account
 809                 if (($result['network'] == NETWORK_FEED) OR ($result['addr'] == "")) {
 810                         require_once('include/bbcode.php');
 811                         $address = GetProfileUsername($url, "", true);
 812                         $result2 = probe_url($address, $mode, ++$level);
 813                         if ($result2['network'] != "")
 814                                 $result = $result2;
 815                 }
 816
 817                 // Maybe it's some non standard GNU Social installation (Single user, subfolder or no uri rewrite)
 818                 if (($result['network'] == NETWORK_FEED) AND ($result['baseurl'] != "") AND ($result['nick'] != "")) {
 819                         $addr = $result['nick'].'@'.str_replace("http://", "", $result['baseurl']);
 820                         $result2 = probe_url($addr, $mode, ++$level);
 821                         if (($result2['network'] != "") AND ($result2['network'] != NETWORK_FEED))
 822                                 $result = $result2;
 823                 }
 824         }
 825
 826         // Only store into the cache if the value seems to be valid
 827         if ($result['network'] != NETWORK_FEED)
 828                 Cache::set("probe_url:".$mode.":".$url,serialize($result), CACHE_DAY);
 829
 830         return $result;
 831 }
 832
 833 function matching($part1, $part2) {
 834         $len = min(strlen($part1), strlen($part2));
 835
 836         $match = "";
 837         $matching = true;
 838         $i = 0;
 839         while (($i <= $len) AND $matching) {
 840                 if (substr($part1, $i, 1) == substr($part2, $i, 1))
 841                         $match .= substr($part1, $i, 1);
 842                 else
 843                         $matching = false;
 844
 845                 $i++;
 846         }
 847         return($match);
 848 }