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, $dont_probe = false) {
   8
   9         $a = get_app();
  10
  11         $ret = array();
  12
  13         logger('scrape_dfrn: url=' . $url);
  14
  15         // Try to fetch the data from noscrape. This is faster than parsing the HTML
  16         $noscrape = str_replace("/hcard/", "/noscrape/", $url);
  17         $noscrapejson = fetch_url($noscrape);
  18         $noscrapedata = array();
  19         if ($noscrapejson) {
  20                 $noscrapedata = json_decode($noscrapejson, true);
  21
  22                 if (is_array($noscrapedata)) {
  23                         if ($noscrapedata["nick"] != "")
  24                                 return($noscrapedata);
  25                 } else
  26                         $noscrapedata = array();
  27         }
  28
  29         $s = fetch_url($url);
  30
  31         if(! $s)
  32                 return $ret;
  33
  34         if (!$dont_probe) {
  35                 $probe = probe_url($url);
  36
  37                 if (isset($probe["addr"]))
  38                         $ret["addr"] = $probe["addr"];
  39         }
  40
  41         $headers = $a->get_curl_headers();
  42         logger('scrape_dfrn: headers=' . $headers, LOGGER_DEBUG);
  43
  44
  45         $lines = explode("\n",$headers);
  46         if(count($lines)) {
  47                 foreach($lines as $line) {
  48                         // don't try and run feeds through the html5 parser
  49                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
  50                                 return ret;
  51                 }
  52         }
  53
  54         try {
  55                 $dom = HTML5_Parser::parse($s);
  56         } catch (DOMException $e) {
  57                 logger('scrape_dfrn: parse error: ' . $e);
  58         }
  59
  60         if(! $dom)
  61                 return $ret;
  62
  63         $items = $dom->getElementsByTagName('link');
  64
  65         // get DFRN link elements
  66
  67         foreach($items as $item) {
  68                 $x = $item->getAttribute('rel');
  69                 if(($x === 'alternate') && ($item->getAttribute('type') === 'application/atom+xml'))
  70                         $ret['feed_atom'] = $item->getAttribute('href');
  71                 if(substr($x,0,5) == "dfrn-") {
  72                         $ret[$x] = $item->getAttribute('href');
  73                 }
  74                 if($x === 'lrdd') {
  75                         $decoded = urldecode($item->getAttribute('href'));
  76                         if(preg_match('/acct:([^@]*)@/',$decoded,$matches))
  77                                 $ret['nick'] = $matches[1];
  78                 }
  79         }
  80
  81         // Pull out hCard profile elements
  82
  83         $largest_photo = 0;
  84
  85         $items = $dom->getElementsByTagName('*');
  86         foreach($items as $item) {
  87                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
  88                         $level2 = $item->getElementsByTagName('*');
  89                         foreach($level2 as $x) {
  90                                 if(attribute_contains($x->getAttribute('class'),'fn')) {
  91                                         $ret['fn'] = $x->textContent;
  92                                 }
  93                                 if((attribute_contains($x->getAttribute('class'),'photo'))
  94                                         || (attribute_contains($x->getAttribute('class'),'avatar'))) {
  95                                         $size = intval($x->getAttribute('width'));
  96                                         // dfrn prefers 175, so if we find this, we set largest_size so it can't be topped.
  97                                         if(($size > $largest_photo) || ($size == 175) || (! $largest_photo)) {
  98                                                 $ret['photo'] = $x->getAttribute('src');
  99                                                 $largest_photo = (($size == 175) ? 9999 : $size);
 100                                         }
 101                                 }
 102                                 if(attribute_contains($x->getAttribute('class'),'key')) {
 103                                         $ret['key'] = $x->textContent;
 104                                 }
 105                         }
 106                 }
 107         }
 108         return array_merge($ret, $noscrapedata);
 109 }}
 110
 111
 112
 113
 114
 115
 116 if(! function_exists('validate_dfrn')) {
 117 function validate_dfrn($a) {
 118         $errors = 0;
 119         if(! x($a,'key'))
 120                 $errors ++;
 121         if(! x($a,'dfrn-request'))
 122                 $errors ++;
 123         if(! x($a,'dfrn-confirm'))
 124                 $errors ++;
 125         if(! x($a,'dfrn-notify'))
 126                 $errors ++;
 127         if(! x($a,'dfrn-poll'))
 128                 $errors ++;
 129         return $errors;
 130 }}
 131
 132 if(! function_exists('scrape_meta')) {
 133 function scrape_meta($url) {
 134
 135         $a = get_app();
 136
 137         $ret = array();
 138
 139         logger('scrape_meta: url=' . $url);
 140
 141         $s = fetch_url($url);
 142
 143         if(! $s)
 144                 return $ret;
 145
 146         $headers = $a->get_curl_headers();
 147         logger('scrape_meta: headers=' . $headers, LOGGER_DEBUG);
 148
 149         $lines = explode("\n",$headers);
 150         if(count($lines)) {
 151                 foreach($lines as $line) {
 152                         // don't try and run feeds through the html5 parser
 153                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
 154                                 return ret;
 155                 }
 156         }
 157
 158         try {
 159                 $dom = HTML5_Parser::parse($s);
 160         } catch (DOMException $e) {
 161                 logger('scrape_meta: parse error: ' . $e);
 162         }
 163
 164         if(! $dom)
 165                 return $ret;
 166
 167         $items = $dom->getElementsByTagName('meta');
 168
 169         // get DFRN link elements
 170
 171         foreach($items as $item) {
 172                 $x = $item->getAttribute('name');
 173                 if(substr($x,0,5) == "dfrn-")
 174                         $ret[$x] = $item->getAttribute('content');
 175         }
 176
 177         return $ret;
 178 }}
 179
 180
 181 if(! function_exists('scrape_vcard')) {
 182 function scrape_vcard($url) {
 183
 184         $a = get_app();
 185
 186         $ret = array();
 187
 188         logger('scrape_vcard: url=' . $url);
 189
 190         $s = fetch_url($url);
 191
 192         if(! $s)
 193                 return $ret;
 194
 195         $headers = $a->get_curl_headers();
 196         $lines = explode("\n",$headers);
 197         if(count($lines)) {
 198                 foreach($lines as $line) {
 199                         // don't try and run feeds through the html5 parser
 200                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
 201                                 return ret;
 202                 }
 203         }
 204
 205         try {
 206                 $dom = HTML5_Parser::parse($s);
 207         } catch (DOMException $e) {
 208                 logger('scrape_vcard: parse error: ' . $e);
 209         }
 210
 211         if(! $dom)
 212                 return $ret;
 213
 214         // Pull out hCard profile elements
 215
 216         $largest_photo = 0;
 217
 218         $items = $dom->getElementsByTagName('*');
 219         foreach($items as $item) {
 220                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
 221                         $level2 = $item->getElementsByTagName('*');
 222                         foreach($level2 as $x) {
 223                                 if(attribute_contains($x->getAttribute('class'),'fn'))
 224                                         $ret['fn'] = $x->textContent;
 225                                 if((attribute_contains($x->getAttribute('class'),'photo'))
 226                                         || (attribute_contains($x->getAttribute('class'),'avatar'))) {
 227                                         $size = intval($x->getAttribute('width'));
 228                                         if(($size > $largest_photo) || (! $largest_photo)) {
 229                                                 $ret['photo'] = $x->getAttribute('src');
 230                                                 $largest_photo = $size;
 231                                         }
 232                                 }
 233                                 if((attribute_contains($x->getAttribute('class'),'nickname'))
 234                                         || (attribute_contains($x->getAttribute('class'),'uid'))) {
 235                                         $ret['nick'] = $x->textContent;
 236                                 }
 237                         }
 238                 }
 239         }
 240
 241         return $ret;
 242 }}
 243
 244
 245 if(! function_exists('scrape_feed')) {
 246 function scrape_feed($url) {
 247
 248         $a = get_app();
 249
 250         $ret = array();
 251         $cookiejar = tempnam(get_temppath(), 'cookiejar-scrape-feed-');
 252         $s = fetch_url($url, false, $redirects, 0, Null, $cookiejar);
 253         unlink($cookiejar);
 254
 255         $headers = $a->get_curl_headers();
 256         $code = $a->get_curl_code();
 257
 258         logger('scrape_feed: returns: ' . $code . ' headers=' . $headers, LOGGER_DEBUG);
 259
 260         if(! $s) {
 261                 logger('scrape_feed: no data returned for ' . $url);
 262                 return $ret;
 263         }
 264
 265
 266         $lines = explode("\n",$headers);
 267         if(count($lines)) {
 268                 foreach($lines as $line) {
 269                         if(stristr($line,'content-type:')) {
 270                                 if(stristr($line,'application/atom+xml') || stristr($s,'<feed')) {
 271                                         $ret['feed_atom'] = $url;
 272                                         return $ret;
 273                                 }
 274                                 if(stristr($line,'application/rss+xml') || stristr($s,'<rss')) {
 275                                         $ret['feed_rss'] = $url;
 276                                         return $ret;
 277                                 }
 278                         }
 279                 }
 280                 // perhaps an RSS version 1 feed with a generic or incorrect content-type?
 281                 if(stristr($s,'</item>')) {
 282                         $ret['feed_rss'] = $url;
 283                         return $ret;
 284                 }
 285         }
 286
 287         $basename = implode('/', array_slice(explode('/',$url),0,3)) . '/';
 288
 289         $doc = new DOMDocument();
 290         @$doc->loadHTML($s);
 291         $xpath = new DomXPath($doc);
 292
 293         $base = $xpath->query("//base");
 294         foreach ($base as $node) {
 295                 $attr = array();
 296
 297                 if ($node->attributes->length)
 298                         foreach ($node->attributes as $attribute)
 299                                 $attr[$attribute->name] = $attribute->value;
 300
 301                 if ($attr["href"] != "")
 302                         $basename = $attr["href"] ;
 303         }
 304
 305         $list = $xpath->query("//link");
 306         foreach ($list as $node) {
 307                 $attr = array();
 308
 309                 if ($node->attributes->length)
 310                         foreach ($node->attributes as $attribute)
 311                                 $attr[$attribute->name] = $attribute->value;
 312
 313                 if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/atom+xml"))
 314                         $ret["feed_atom"] = $attr["href"];
 315
 316                 if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/rss+xml"))
 317                         $ret["feed_rss"] = $attr["href"];
 318         }
 319
 320         // Drupal and perhaps others only provide relative URLs. Turn them into absolute.
 321
 322         if(x($ret,'feed_atom') && (! strstr($ret['feed_atom'],'://')))
 323                 $ret['feed_atom'] = $basename . $ret['feed_atom'];
 324         if(x($ret,'feed_rss') && (! strstr($ret['feed_rss'],'://')))
 325                 $ret['feed_rss'] = $basename . $ret['feed_rss'];
 326
 327         return $ret;
 328 }}
 329
 330
 331 /**
 332  *
 333  * Probe a network address to discover what kind of protocols we need to communicate with it.
 334  *
 335  * Warning: this function is a bit touchy and there are some subtle dependencies within the logic flow.
 336  * Edit with care.
 337  *
 338  */
 339
 340 /**
 341  *
 342  * PROBE_DIASPORA has a bias towards returning Diaspora information
 343  * while PROBE_NORMAL has a bias towards dfrn/zot - in the case where
 344  * an address (such as a Friendica address) supports more than one type
 345  * of network.
 346  *
 347  */
 348
 349
 350 define ( 'PROBE_NORMAL',   0);
 351 define ( 'PROBE_DIASPORA', 1);
 352
 353 function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
 354         require_once('include/email.php');
 355
 356         $result = array();
 357
 358         if(! $url)
 359                 return $result;
 360
 361         $result = Cache::get("probe_url:".$mode.":".$url);
 362         if (!is_null($result)) {
 363                 $result = unserialize($result);
 364                 return $result;
 365         }
 366
 367         $network = null;
 368         $diaspora = false;
 369         $diaspora_base = '';
 370         $diaspora_guid = '';
 371         $diaspora_key = '';
 372         $has_lrdd = false;
 373         $email_conversant = false;
 374         $connectornetworks = false;
 375         $appnet = false;
 376
 377         if (strpos($url,'twitter.com')) {
 378                 $connectornetworks = true;
 379                 $network = NETWORK_TWITTER;
 380         }
 381
 382         // Twitter is deactivated since twitter closed its old API
 383         //$twitter = ((strpos($url,'twitter.com') !== false) ? true : false);
 384         $lastfm  = ((strpos($url,'last.fm/user') !== false) ? true : false);
 385
 386         $at_addr = ((strpos($url,'@') !== false) ? true : false);
 387
 388         if((!$appnet) && (!$lastfm) && !$connectornetworks) {
 389
 390                 if(strpos($url,'mailto:') !== false && $at_addr) {
 391                         $url = str_replace('mailto:','',$url);
 392                         $links = array();
 393                 }
 394                 else
 395                         $links = lrdd($url);
 396
 397                 if(count($links)) {
 398                         $has_lrdd = true;
 399
 400                         logger('probe_url: found lrdd links: ' . print_r($links,true), LOGGER_DATA);
 401                         foreach($links as $link) {
 402                                 if($link['@attributes']['rel'] === NAMESPACE_ZOT)
 403                                         $zot = unamp($link['@attributes']['href']);
 404                                 if($link['@attributes']['rel'] === NAMESPACE_DFRN)
 405                                         $dfrn = unamp($link['@attributes']['href']);
 406                                 if($link['@attributes']['rel'] === 'salmon')
 407                                         $notify = unamp($link['@attributes']['href']);
 408                                 if($link['@attributes']['rel'] === NAMESPACE_FEED)
 409                                         $poll = unamp($link['@attributes']['href']);
 410                                 if($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard')
 411                                         $hcard = unamp($link['@attributes']['href']);
 412                                 if($link['@attributes']['rel'] === 'http://webfinger.net/rel/profile-page')
 413                                         $profile = unamp($link['@attributes']['href']);
 414                                 if($link['@attributes']['rel'] === 'http://portablecontacts.net/spec/1.0')
 415                                         $poco = unamp($link['@attributes']['href']);
 416                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/seed_location') {
 417                                         $diaspora_base = unamp($link['@attributes']['href']);
 418                                         $diaspora = true;
 419                                 }
 420                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/guid') {
 421                                         $diaspora_guid = unamp($link['@attributes']['href']);
 422                                         $diaspora = true;
 423                                 }
 424                                 if($link['@attributes']['rel'] === 'diaspora-public-key') {
 425                                         $diaspora_key = base64_decode(unamp($link['@attributes']['href']));
 426                                         if(strstr($diaspora_key,'RSA '))
 427                                                 $pubkey = rsatopem($diaspora_key);
 428                                         else
 429                                                 $pubkey = $diaspora_key;
 430                                         $diaspora = true;
 431                                 }
 432                                 if(($link['@attributes']['rel'] === 'http://ostatus.org/schema/1.0/subscribe') AND ($mode == PROBE_NORMAL)) {
 433                                         $diaspora = false;
 434                                 }
 435                         }
 436
 437                         // Status.Net can have more than one profile URL. We need to match the profile URL
 438                         // to a contact on incoming messages to prevent spam, and we won't know which one
 439                         // to match. So in case of two, one of them is stored as an alias. Only store URL's
 440                         // and not webfinger user@host aliases. If they've got more than two non-email style
 441                         // aliases, let's hope we're lucky and get one that matches the feed author-uri because
 442                         // otherwise we're screwed.
 443
 444                         foreach($links as $link) {
 445                                 if($link['@attributes']['rel'] === 'alias') {
 446                                         if(strpos($link['@attributes']['href'],'@') === false) {
 447                                                 if(isset($profile)) {
 448                                                         if($link['@attributes']['href'] !== $profile)
 449                                                                 $alias = unamp($link['@attributes']['href']);
 450                                                 }
 451                                                 else
 452                                                         $profile = unamp($link['@attributes']['href']);
 453                                         }
 454                                 }
 455                         }
 456
 457                         // If the profile is different from the url then the url is abviously an alias
 458                         if (($alias == "") AND ($profile != "") AND !$at_addr AND (normalise_link($profile) != normalise_link($url)))
 459                                 $alias = $url;
 460                 }
 461                 elseif($mode == PROBE_NORMAL) {
 462
 463                         // Check email
 464
 465                         $orig_url = $url;
 466                         if((strpos($orig_url,'@')) && validate_email($orig_url)) {
 467                                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1",
 468                                         intval(local_user())
 469                                 );
 470                                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1",
 471                                         intval(local_user())
 472                                 );
 473                                 if(count($x) && count($r)) {
 474                                         $mailbox = construct_mailbox_name($r[0]);
 475                                         $password = '';
 476                                         openssl_private_decrypt(hex2bin($r[0]['pass']),$password,$x[0]['prvkey']);
 477                                         $mbox = email_connect($mailbox,$r[0]['user'],$password);
 478                                         if(! $mbox)
 479                                                 logger('probe_url: email_connect failed.');
 480                                         unset($password);
 481                                 }
 482                                 if($mbox) {
 483                                         $msgs = email_poll($mbox,$orig_url);
 484                                         logger('probe_url: searching ' . $orig_url . ', ' . count($msgs) . ' messages found.', LOGGER_DEBUG);
 485                                         if(count($msgs)) {
 486                                                 $addr = $orig_url;
 487                                                 $network = NETWORK_MAIL;
 488                                                 $name = substr($url,0,strpos($url,'@'));
 489                                                 $phost = substr($url,strpos($url,'@')+1);
 490                                                 $profile = 'http://' . $phost;
 491                                                 // fix nick character range
 492                                                 $vcard = array('fn' => $name, 'nick' => $name, 'photo' => avatar_img($url));
 493                                                 $notify = 'smtp ' . random_string();
 494                                                 $poll = 'email ' . random_string();
 495                                                 $priority = 0;
 496                                                 $x = email_msg_meta($mbox,$msgs[0]);
 497                                                 if(stristr($x[0]->from,$orig_url))
 498                                                         $adr = imap_rfc822_parse_adrlist($x[0]->from,'');
 499                                                 elseif(stristr($x[0]->to,$orig_url))
 500                                                         $adr = imap_rfc822_parse_adrlist($x[0]->to,'');
 501                                                 if(isset($adr)) {
 502                                                         foreach($adr as $feadr) {
 503                                                                 if((strcasecmp($feadr->mailbox,$name) == 0)
 504                                                                         &&(strcasecmp($feadr->host,$phost) == 0)
 505                                                                         && (strlen($feadr->personal))) {
 506
 507                                                                         $personal = imap_mime_header_decode($feadr->personal);
 508                                                                         $vcard['fn'] = "";
 509                                                                         foreach($personal as $perspart)
 510                                                                                 if ($perspart->charset != "default")
 511                                                                                         $vcard['fn'] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
 512                                                                                 else
 513                                                                                         $vcard['fn'] .= $perspart->text;
 514
 515                                                                         $vcard['fn'] = notags($vcard['fn']);
 516                                                                 }
 517                                                         }
 518                                                 }
 519                                         }
 520                                         imap_close($mbox);
 521                                 }
 522                         }
 523                 }
 524         }
 525
 526         if($mode == PROBE_NORMAL) {
 527
 528                 if(strlen($zot)) {
 529                         $s = fetch_url($zot);
 530                         if($s) {
 531                                 $j = json_decode($s);
 532                                 if($j) {
 533                                         $network = NETWORK_ZOT;
 534                                         $vcard   = array(
 535                                                 'fn'    => $j->fullname,
 536                                                 'nick'  => $j->nickname,
 537                                                 'photo' => $j->photo
 538                                         );
 539                                         $profile  = $j->url;
 540                                         $notify   = $j->post;
 541                                         $pubkey   = $j->pubkey;
 542                                         $poll     = 'N/A';
 543                                 }
 544                         }
 545                 }
 546
 547
 548                 if(strlen($dfrn)) {
 549                         $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn, true);
 550                         if(is_array($ret) && x($ret,'dfrn-request')) {
 551                                 $network = NETWORK_DFRN;
 552                                 $request = $ret['dfrn-request'];
 553                                 $confirm = $ret['dfrn-confirm'];
 554                                 $notify  = $ret['dfrn-notify'];
 555                                 $poll    = $ret['dfrn-poll'];
 556
 557                                 $vcard = array();
 558                                 $vcard['fn'] = $ret['fn'];
 559                                 $vcard['nick'] = $ret['nick'];
 560                                 $vcard['photo'] = $ret['photo'];
 561                         }
 562                 }
 563         }
 564
 565         // Scrape the public key from the hcard.
 566         // Diaspora will remove it from the webfinger somewhere in the future.
 567         if (($hcard != "") AND ($pubkey == "")) {
 568                 $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn, true);
 569                 if (isset($ret["key"])) {
 570                         $hcard_key = $ret["key"];
 571                         if(strstr($hcard_key,'RSA '))
 572                                 $pubkey = rsatopem($hcard_key);
 573                         else
 574                                 $pubkey = $hcard_key;
 575                 }
 576         }
 577         if($diaspora && $diaspora_base && $diaspora_guid) {
 578                 $diaspora_notify = $diaspora_base.'receive/users/'.$diaspora_guid;
 579
 580                 if($mode == PROBE_DIASPORA || ! $notify || ($notify == $diaspora_notify)) {
 581                         $notify = $diaspora_notify;
 582                         $batch  = $diaspora_base . 'receive/public' ;
 583                 }
 584                 if(strpos($url,'@'))
 585                         $addr = str_replace('acct:', '', $url);
 586         }
 587
 588         if($network !== NETWORK_ZOT && $network !== NETWORK_DFRN && $network !== NETWORK_MAIL) {
 589                 if($diaspora)
 590                         $network = NETWORK_DIASPORA;
 591                 elseif($has_lrdd AND ($notify))
 592                         $network  = NETWORK_OSTATUS;
 593
 594                 if(strpos($url,'@'))
 595                         $addr = str_replace('acct:', '', $url);
 596
 597                 $priority = 0;
 598
 599                 if($hcard && ! $vcard) {
 600                         $vcard = scrape_vcard($hcard);
 601
 602                         // Google doesn't use absolute url in profile photos
 603
 604                         if((x($vcard,'photo')) && substr($vcard['photo'],0,1) == '/') {
 605                                 $h = @parse_url($hcard);
 606                                 if($h)
 607                                         $vcard['photo'] = $h['scheme'] . '://' . $h['host'] . $vcard['photo'];
 608                         }
 609
 610                         logger('probe_url: scrape_vcard: ' . print_r($vcard,true), LOGGER_DATA);
 611                 }
 612
 613                 if($diaspora && $addr) {
 614                         // Diaspora returns the name as the nick. As the nick will never be updated,
 615                         // let's use the Diaspora nickname (the first part of the handle) as the nick instead
 616                         $addr_parts = explode('@', $addr);
 617                         $vcard['nick'] = $addr_parts[0];
 618                 }
 619
 620                 /* if($twitter) {
 621                         logger('twitter: setup');
 622                         $tid = basename($url);
 623                         $tapi = 'https://api.twitter.com/1/statuses/user_timeline.rss';
 624                         if(intval($tid))
 625                                 $poll = $tapi . '?user_id=' . $tid;
 626                         else
 627                                 $poll = $tapi . '?screen_name=' . $tid;
 628                         $profile = 'http://twitter.com/#!/' . $tid;
 629                         //$vcard['photo'] = 'https://api.twitter.com/1/users/profile_image/' . $tid;
 630                         $vcard['photo'] = 'https://api.twitter.com/1/users/profile_image?screen_name=' . $tid . '&size=bigger';
 631                         $vcard['nick'] = $tid;
 632                         $vcard['fn'] = $tid;
 633                 } */
 634
 635                 if($lastfm) {
 636                         $profile = $url;
 637                         $poll = str_replace(array('www.','last.fm/'),array('','ws.audioscrobbler.com/1.0/'),$url) . '/recenttracks.rss';
 638                         $vcard['nick'] = basename($url);
 639                         $vcard['fn'] = $vcard['nick'] . t(' on Last.fm');
 640                         $network = NETWORK_FEED;
 641                 }
 642
 643                 if(! x($vcard,'fn'))
 644                         if(x($vcard,'nick'))
 645                                 $vcard['fn'] = $vcard['nick'];
 646
 647                 $check_feed = false;
 648
 649                 if(stristr($url,'tumblr.com') && (! stristr($url,'/rss'))) {
 650                         $poll = $url . '/rss';
 651                         $check_feed = true;
 652                         // Will leave it to others to figure out how to grab the avatar, which is on the $url page in the open graph meta links
 653                 }
 654
 655                 if($appnet || ! $poll)
 656                         $check_feed = true;
 657                 if((! isset($vcard)) || (! x($vcard,'fn')) || (! $profile))
 658                         $check_feed = true;
 659                 if(($at_addr) && (! count($links)))
 660                         $check_feed = false;
 661
 662                 if ($connectornetworks)
 663                         $check_feed = false;
 664
 665                 if($check_feed) {
 666
 667                         $feedret = scrape_feed(($poll) ? $poll : $url);
 668
 669                         logger('probe_url: scrape_feed ' . (($poll)? $poll : $url) . ' returns: ' . print_r($feedret,true), LOGGER_DATA);
 670                         if(count($feedret) && ($feedret['feed_atom'] || $feedret['feed_rss'])) {
 671                                 $poll = ((x($feedret,'feed_atom')) ? unamp($feedret['feed_atom']) : unamp($feedret['feed_rss']));
 672                                 if(! x($vcard))
 673                                         $vcard = array();
 674                         }
 675
 676                         if(x($feedret,'photo') && (! x($vcard,'photo')))
 677                                 $vcard['photo'] = $feedret['photo'];
 678                         require_once('library/simplepie/simplepie.inc');
 679                         $feed = new SimplePie();
 680                         $cookiejar = tempnam(get_temppath(), 'cookiejar-scrape-feed-');
 681                         $xml = fetch_url($poll, false, $redirects, 0, Null, $cookiejar);
 682                         unlink($cookiejar);
 683
 684                         logger('probe_url: fetch feed: ' . $poll . ' returns: ' . $xml, LOGGER_DATA);
 685                         $a = get_app();
 686
 687                         logger('probe_url: scrape_feed: headers: ' . $a->get_curl_headers(), LOGGER_DATA);
 688
 689                         // Don't try and parse an empty string
 690                         $feed->set_raw_data(($xml) ? $xml : '<?xml version="1.0" encoding="utf-8" ?><xml></xml>');
 691
 692                         $feed->init();
 693                         if($feed->error()) {
 694                                 logger('probe_url: scrape_feed: Error parsing XML: ' . $feed->error());
 695                                 $network = NETWORK_PHANTOM;
 696                         }
 697
 698                         if(! x($vcard,'photo'))
 699                                 $vcard['photo'] = $feed->get_image_url();
 700                         $author = $feed->get_author();
 701
 702                         if($author) {
 703                                 $vcard['fn'] = unxmlify(trim($author->get_name()));
 704                                 if(! $vcard['fn'])
 705                                         $vcard['fn'] = trim(unxmlify($author->get_email()));
 706                                 if(strpos($vcard['fn'],'@') !== false)
 707                                         $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
 708
 709                                 $email = unxmlify($author->get_email());
 710                                 if(! $profile && $author->get_link())
 711                                         $profile = trim(unxmlify($author->get_link()));
 712                                 if(! $vcard['photo']) {
 713                                         $rawtags = $feed->get_feed_tags( SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
 714                                         if($rawtags) {
 715                                                 $elems = $rawtags[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10];
 716                                                 if((x($elems,'link')) && ($elems['link'][0]['attribs']['']['rel'] === 'photo'))
 717                                                         $vcard['photo'] = $elems['link'][0]['attribs']['']['href'];
 718                                         }
 719                                 }
 720                                 // Fetch fullname via poco:displayName
 721                                 $pocotags = $feed->get_feed_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
 722                                 if ($pocotags) {
 723                                         $elems = $pocotags[0]['child']['http://portablecontacts.net/spec/1.0'];
 724                                         if (isset($elems["displayName"]))
 725                                                 $vcard['fn'] = $elems["displayName"][0]["data"];
 726                                         if (isset($elems["preferredUsername"]))
 727                                                 $vcard['nick'] = $elems["preferredUsername"][0]["data"];
 728                                 }
 729                         }
 730                         else {
 731                                 $item = $feed->get_item(0);
 732                                 if($item) {
 733                                         $author = $item->get_author();
 734                                         if($author) {
 735                                                 $vcard['fn'] = trim(unxmlify($author->get_name()));
 736                                                 if(! $vcard['fn'])
 737                                                         $vcard['fn'] = trim(unxmlify($author->get_email()));
 738                                                 if(strpos($vcard['fn'],'@') !== false)
 739                                                         $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
 740                                                 $email = unxmlify($author->get_email());
 741                                                 if(! $profile && $author->get_link())
 742                                                         $profile = trim(unxmlify($author->get_link()));
 743                                         }
 744                                         if(! $vcard['photo']) {
 745                                                 $rawmedia = $item->get_item_tags('http://search.yahoo.com/mrss/','thumbnail');
 746                                                 if($rawmedia && $rawmedia[0]['attribs']['']['url'])
 747                                                         $vcard['photo'] = unxmlify($rawmedia[0]['attribs']['']['url']);
 748                                         }
 749                                         if(! $vcard['photo']) {
 750                                                 $rawtags = $item->get_item_tags( SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
 751                                                 if($rawtags) {
 752                                                         $elems = $rawtags[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10];
 753                                                         if((x($elems,'link')) && ($elems['link'][0]['attribs']['']['rel'] === 'photo'))
 754                                                                 $vcard['photo'] = $elems['link'][0]['attribs']['']['href'];
 755                                                 }
 756                                         }
 757                                 }
 758                         }
 759
 760                         // Workaround for misconfigured Friendica servers
 761                         if (($network == "") AND (strstr($url, "/profile/"))) {
 762                                 $noscrape = str_replace("/profile/", "/noscrape/", $url);
 763                                 $noscrapejson = fetch_url($noscrape);
 764                                 if ($noscrapejson) {
 765
 766                                         $network = NETWORK_DFRN;
 767
 768                                         $poco = str_replace("/profile/", "/poco/", $url);
 769
 770                                         $noscrapedata = json_decode($noscrapejson, true);
 771
 772                                         if (isset($noscrapedata["addr"]))
 773                                                 $addr = $noscrapedata["addr"];
 774
 775                                         if (isset($noscrapedata["fn"]))
 776                                                 $vcard["fn"] = $noscrapedata["fn"];
 777
 778                                         if (isset($noscrapedata["key"]))
 779                                                 $pubkey = $noscrapedata["key"];
 780
 781                                         if (isset($noscrapedata["photo"]))
 782                                                 $vcard["photo"] = $noscrapedata["photo"];
 783
 784                                         if (isset($noscrapedata["dfrn-request"]))
 785                                                 $request = $noscrapedata["dfrn-request"];
 786
 787                                         if (isset($noscrapedata["dfrn-confirm"]))
 788                                                 $confirm = $noscrapedata["dfrn-confirm"];
 789
 790                                         if (isset($noscrapedata["dfrn-notify"]))
 791                                                 $notify = $noscrapedata["dfrn-notify"];
 792
 793                                         if (isset($noscrapedata["dfrn-poll"]))
 794                                                 $poll = $noscrapedata["dfrn-poll"];
 795
 796                                 }
 797                         }
 798
 799                         if((! $vcard['photo']) && strlen($email))
 800                                 $vcard['photo'] = avatar_img($email);
 801                         if($poll === $profile)
 802                                 $lnk = $feed->get_permalink();
 803                         if(isset($lnk) && strlen($lnk))
 804                                 $profile = $lnk;
 805
 806                         if(! $network) {
 807                                 $network = NETWORK_FEED;
 808                                 // If it is a feed, don't take the author name as feed name
 809                                 unset($vcard['fn']);
 810                         }
 811                         if(! (x($vcard,'fn')))
 812                                 $vcard['fn'] = notags($feed->get_title());
 813                         if(! (x($vcard,'fn')))
 814                                 $vcard['fn'] = notags($feed->get_description());
 815
 816                         if(strpos($vcard['fn'],'Twitter / ') !== false) {
 817                                 $vcard['fn'] = substr($vcard['fn'],strpos($vcard['fn'],'/')+1);
 818                                 $vcard['fn'] = trim($vcard['fn']);
 819                         }
 820                         if(! x($vcard,'nick')) {
 821                                 $vcard['nick'] = strtolower(notags(unxmlify($vcard['fn'])));
 822                                 if(strpos($vcard['nick'],' '))
 823                                         $vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' ')));
 824                         }
 825                         if(! $priority)
 826                                 $priority = 2;
 827                 }
 828         }
 829
 830         if(! x($vcard,'photo')) {
 831                 $a = get_app();
 832                 $vcard['photo'] = $a->get_baseurl() . '/images/person-175.jpg' ;
 833         }
 834
 835         if(! $profile)
 836                 $profile = $url;
 837
 838         // No human could be associated with this link, use the URL as the contact name
 839
 840         if(($network === NETWORK_FEED) && ($poll) && (! x($vcard,'fn')))
 841                 $vcard['fn'] = $url;
 842
 843         if (($notify != "") AND ($poll != "")) {
 844                 $baseurl = matching_url(normalise_link($notify), normalise_link($poll));
 845
 846                 $baseurl2 = matching_url($baseurl, normalise_link($profile));
 847                 if ($baseurl2 != "")
 848                         $baseurl = $baseurl2;
 849         }
 850
 851         if (($baseurl == "") AND ($notify != ""))
 852                 $baseurl = matching_url(normalise_link($profile), normalise_link($notify));
 853
 854         if (($baseurl == "") AND ($poll != ""))
 855                 $baseurl = matching_url(normalise_link($profile), normalise_link($poll));
 856
 857         $baseurl = rtrim($baseurl, "/");
 858
 859         if(strpos($url,'@') AND ($addr == "") AND ($network == NETWORK_DFRN))
 860                 $addr = str_replace('acct:', '', $url);
 861
 862         $vcard['fn'] = notags($vcard['fn']);
 863         $vcard['nick'] = str_replace(' ','',notags($vcard['nick']));
 864
 865         $result['name'] = $vcard['fn'];
 866         $result['nick'] = $vcard['nick'];
 867         $result['url'] = $profile;
 868         $result['addr'] = $addr;
 869         $result['batch'] = $batch;
 870         $result['notify'] = $notify;
 871         $result['poll'] = $poll;
 872         $result['request'] = $request;
 873         $result['confirm'] = $confirm;
 874         $result['poco'] = $poco;
 875         $result['photo'] = $vcard['photo'];
 876         $result['priority'] = $priority;
 877         $result['network'] = $network;
 878         $result['alias'] = $alias;
 879         $result['pubkey'] = $pubkey;
 880         $result['baseurl'] = $baseurl;
 881
 882         logger('probe_url: ' . print_r($result,true), LOGGER_DEBUG);
 883
 884         if ($level == 1) {
 885                 // Trying if it maybe a diaspora account
 886                 if (($result['network'] == NETWORK_FEED) OR ($result['addr'] == "")) {
 887                         require_once('include/bbcode.php');
 888                         $address = GetProfileUsername($url, "", true);
 889                         $result2 = probe_url($address, $mode, ++$level);
 890                         if ($result2['network'] != "")
 891                                 $result = $result2;
 892                 }
 893
 894                 // Maybe it's some non standard GNU Social installation (Single user, subfolder or no uri rewrite)
 895                 if (($result['network'] == NETWORK_FEED) AND ($result['baseurl'] != "") AND ($result['nick'] != "")) {
 896                         $addr = $result['nick'].'@'.str_replace("http://", "", $result['baseurl']);
 897                         $result2 = probe_url($addr, $mode, ++$level);
 898                         if (($result2['network'] != "") AND ($result2['network'] != NETWORK_FEED))
 899                                 $result = $result2;
 900                 }
 901         }
 902
 903         // Only store into the cache if the value seems to be valid
 904         if ($result['network'] != NETWORK_PHANTOM)
 905                 Cache::set("probe_url:".$mode.":".$url,serialize($result), CACHE_DAY);
 906
 907         return $result;
 908 }
 909
 910 /**
 911  * @brief Find the matching part between two url
 912  *
 913  * @param string $url1
 914  * @param string $url2
 915  * @return string The matching part
 916  */
 917 function matching_url($url1, $url2) {
 918
 919         if (($url1 == "") OR ($url2 == ""))
 920                 return "";
 921
 922         $url1 = normalise_link($url1);
 923         $url2 = normalise_link($url2);
 924
 925         $parts1 = parse_url($url1);
 926         $parts2 = parse_url($url2);
 927
 928         if (!isset($parts1["host"]) OR !isset($parts2["host"]))
 929                 return "";
 930
 931         if ($parts1["scheme"] != $parts2["scheme"])
 932                 return "";
 933
 934         if ($parts1["host"] != $parts2["host"])
 935                 return "";
 936
 937         if ($parts1["port"] != $parts2["port"])
 938                 return "";
 939
 940         $match = $parts1["scheme"]."://".$parts1["host"];
 941
 942         if ($parts1["port"])
 943                 $match .= ":".$parts1["port"];
 944
 945         $pathparts1 = explode("/", $parts1["path"]);
 946         $pathparts2 = explode("/", $parts2["path"]);
 947
 948         $i = 0;
 949         $path = "";
 950         do {
 951                 $path1 = $pathparts1[$i];
 952                 $path2 = $pathparts2[$i];
 953
 954                 if ($path1 == $path2)
 955                         $path .= $path1."/";
 956
 957         } while (($path1 == $path2) AND ($i++ <= count($pathparts1)));
 958
 959         $match .= $path;
 960
 961         return normalise_link($match);
 962 }