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         $s = fetch_url($url);
  16
  17         if(! $s)
  18                 return $ret;
  19
  20         $probe = probe_url($url);
  21
  22         if (isset($probe["addr"]))
  23                 $ret["addr"] = $probe["addr"];
  24
  25         $headers = $a->get_curl_headers();
  26         logger('scrape_dfrn: headers=' . $headers, LOGGER_DEBUG);
  27
  28
  29         $lines = explode("\n",$headers);
  30         if(count($lines)) {
  31                 foreach($lines as $line) {
  32                         // don't try and run feeds through the html5 parser
  33                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
  34                                 return ret;
  35                 }
  36         }
  37
  38         try {
  39                 $dom = HTML5_Parser::parse($s);
  40         } catch (DOMException $e) {
  41                 logger('scrape_dfrn: parse error: ' . $e);
  42         }
  43
  44         if(! $dom)
  45                 return $ret;
  46
  47         $items = $dom->getElementsByTagName('link');
  48
  49         // get DFRN link elements
  50
  51         foreach($items as $item) {
  52                 $x = $item->getAttribute('rel');
  53                 if(($x === 'alternate') && ($item->getAttribute('type') === 'application/atom+xml'))
  54                         $ret['feed_atom'] = $item->getAttribute('href');
  55                 if(substr($x,0,5) == "dfrn-") {
  56                         $ret[$x] = $item->getAttribute('href');
  57                 }
  58                 if($x === 'lrdd') {
  59                         $decoded = urldecode($item->getAttribute('href'));
  60                         if(preg_match('/acct:([^@]*)@/',$decoded,$matches))
  61                                 $ret['nick'] = $matches[1];
  62                 }
  63         }
  64
  65         // Pull out hCard profile elements
  66
  67         $largest_photo = 0;
  68
  69         $items = $dom->getElementsByTagName('*');
  70         foreach($items as $item) {
  71                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
  72                         $level2 = $item->getElementsByTagName('*');
  73                         foreach($level2 as $x) {
  74                                 if(attribute_contains($x->getAttribute('class'),'fn')) {
  75                                         $ret['fn'] = $x->textContent;
  76                                 }
  77                                 if((attribute_contains($x->getAttribute('class'),'photo'))
  78                                         || (attribute_contains($x->getAttribute('class'),'avatar'))) {
  79                                         $size = intval($x->getAttribute('width'));
  80                                         // dfrn prefers 175, so if we find this, we set largest_size so it can't be topped.
  81                                         if(($size > $largest_photo) || ($size == 175) || (! $largest_photo)) {
  82                                                 $ret['photo'] = $x->getAttribute('src');
  83                                                 $largest_photo = (($size == 175) ? 9999 : $size);
  84                                         }
  85                                 }
  86                                 if(attribute_contains($x->getAttribute('class'),'key')) {
  87                                         $ret['key'] = $x->textContent;
  88                                 }
  89                         }
  90                 }
  91         }
  92
  93         return $ret;
  94 }}
  95
  96
  97
  98
  99
 100
 101 if(! function_exists('validate_dfrn')) {
 102 function validate_dfrn($a) {
 103         $errors = 0;
 104         if(! x($a,'key'))
 105                 $errors ++;
 106         if(! x($a,'dfrn-request'))
 107                 $errors ++;
 108         if(! x($a,'dfrn-confirm'))
 109                 $errors ++;
 110         if(! x($a,'dfrn-notify'))
 111                 $errors ++;
 112         if(! x($a,'dfrn-poll'))
 113                 $errors ++;
 114         return $errors;
 115 }}
 116
 117 if(! function_exists('scrape_meta')) {
 118 function scrape_meta($url) {
 119
 120         $a = get_app();
 121
 122         $ret = array();
 123
 124         logger('scrape_meta: url=' . $url);
 125
 126         $s = fetch_url($url);
 127
 128         if(! $s)
 129                 return $ret;
 130
 131         $headers = $a->get_curl_headers();
 132         logger('scrape_meta: headers=' . $headers, LOGGER_DEBUG);
 133
 134         $lines = explode("\n",$headers);
 135         if(count($lines)) {
 136                 foreach($lines as $line) {
 137                         // don't try and run feeds through the html5 parser
 138                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
 139                                 return ret;
 140                 }
 141         }
 142
 143         try {
 144                 $dom = HTML5_Parser::parse($s);
 145         } catch (DOMException $e) {
 146                 logger('scrape_meta: parse error: ' . $e);
 147         }
 148
 149         if(! $dom)
 150                 return $ret;
 151
 152         $items = $dom->getElementsByTagName('meta');
 153
 154         // get DFRN link elements
 155
 156         foreach($items as $item) {
 157                 $x = $item->getAttribute('name');
 158                 if(substr($x,0,5) == "dfrn-")
 159                         $ret[$x] = $item->getAttribute('content');
 160         }
 161
 162         return $ret;
 163 }}
 164
 165
 166 if(! function_exists('scrape_vcard')) {
 167 function scrape_vcard($url) {
 168
 169         $a = get_app();
 170
 171         $ret = array();
 172
 173         logger('scrape_vcard: url=' . $url);
 174
 175         $s = fetch_url($url);
 176
 177         if(! $s)
 178                 return $ret;
 179
 180         $headers = $a->get_curl_headers();
 181         $lines = explode("\n",$headers);
 182         if(count($lines)) {
 183                 foreach($lines as $line) {
 184                         // don't try and run feeds through the html5 parser
 185                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
 186                                 return ret;
 187                 }
 188         }
 189
 190         try {
 191                 $dom = HTML5_Parser::parse($s);
 192         } catch (DOMException $e) {
 193                 logger('scrape_vcard: parse error: ' . $e);
 194         }
 195
 196         if(! $dom)
 197                 return $ret;
 198
 199         // Pull out hCard profile elements
 200
 201         $largest_photo = 0;
 202
 203         $items = $dom->getElementsByTagName('*');
 204         foreach($items as $item) {
 205                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
 206                         $level2 = $item->getElementsByTagName('*');
 207                         foreach($level2 as $x) {
 208                                 if(attribute_contains($x->getAttribute('class'),'fn'))
 209                                         $ret['fn'] = $x->textContent;
 210                                 if((attribute_contains($x->getAttribute('class'),'photo'))
 211                                         || (attribute_contains($x->getAttribute('class'),'avatar'))) {
 212                                         $size = intval($x->getAttribute('width'));
 213                                         if(($size > $largest_photo) || (! $largest_photo)) {
 214                                                 $ret['photo'] = $x->getAttribute('src');
 215                                                 $largest_photo = $size;
 216                                         }
 217                                 }
 218                                 if((attribute_contains($x->getAttribute('class'),'nickname'))
 219                                         || (attribute_contains($x->getAttribute('class'),'uid'))) {
 220                                         $ret['nick'] = $x->textContent;
 221                                 }
 222                         }
 223                 }
 224         }
 225
 226         return $ret;
 227 }}
 228
 229
 230 if(! function_exists('scrape_feed')) {
 231 function scrape_feed($url) {
 232
 233         $a = get_app();
 234
 235         $ret = array();
 236         $s = fetch_url($url);
 237
 238         $headers = $a->get_curl_headers();
 239         $code = $a->get_curl_code();
 240
 241         logger('scrape_feed: returns: ' . $code . ' headers=' . $headers, LOGGER_DEBUG);
 242
 243         if(! $s) {
 244                 logger('scrape_feed: no data returned for ' . $url);
 245                 return $ret;
 246         }
 247
 248
 249         $lines = explode("\n",$headers);
 250         if(count($lines)) {
 251                 foreach($lines as $line) {
 252                         if(stristr($line,'content-type:')) {
 253                                 if(stristr($line,'application/atom+xml') || stristr($s,'<feed')) {
 254                                         $ret['feed_atom'] = $url;
 255                                         return $ret;
 256                                 }
 257                                 if(stristr($line,'application/rss+xml') || stristr($s,'<rss')) {
 258                                         $ret['feed_rss'] = $url;
 259                                         return $ret;
 260                                 }
 261                         }
 262                 }
 263                 // perhaps an RSS version 1 feed with a generic or incorrect content-type?
 264                 if(stristr($s,'</item>')) {
 265                         $ret['feed_rss'] = $url;
 266                         return $ret;
 267                 }
 268         }
 269
 270         $basename = implode('/', array_slice(explode('/',$url),0,3)) . '/';
 271
 272         $doc = new DOMDocument();
 273         @$doc->loadHTML($s);
 274         $xpath = new DomXPath($doc);
 275
 276         $base = $xpath->query("//base");
 277         foreach ($base as $node) {
 278                 $attr = array();
 279
 280                 if ($node->attributes->length)
 281                         foreach ($node->attributes as $attribute)
 282                                 $attr[$attribute->name] = $attribute->value;
 283
 284                 if ($attr["href"] != "")
 285                         $basename = $attr["href"] ;
 286         }
 287
 288         $list = $xpath->query("//link");
 289         foreach ($list as $node) {
 290                 $attr = array();
 291
 292                 if ($node->attributes->length)
 293                         foreach ($node->attributes as $attribute)
 294                                 $attr[$attribute->name] = $attribute->value;
 295
 296                 if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/atom+xml"))
 297                         $ret["feed_atom"] = $attr["href"];
 298
 299                 if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/rss+xml"))
 300                         $ret["feed_rss"] = $attr["href"];
 301         }
 302
 303         // Drupal and perhaps others only provide relative URLs. Turn them into absolute.
 304
 305         if(x($ret,'feed_atom') && (! strstr($ret['feed_atom'],'://')))
 306                 $ret['feed_atom'] = $basename . $ret['feed_atom'];
 307         if(x($ret,'feed_rss') && (! strstr($ret['feed_rss'],'://')))
 308                 $ret['feed_rss'] = $basename . $ret['feed_rss'];
 309
 310         return $ret;
 311 }}
 312
 313
 314 /**
 315  *
 316  * Probe a network address to discover what kind of protocols we need to communicate with it.
 317  *
 318  * Warning: this function is a bit touchy and there are some subtle dependencies within the logic flow.
 319  * Edit with care.
 320  *
 321  */
 322
 323 /**
 324  *
 325  * PROBE_DIASPORA has a bias towards returning Diaspora information
 326  * while PROBE_NORMAL has a bias towards dfrn/zot - in the case where
 327  * an address (such as a Friendica address) supports more than one type
 328  * of network.
 329  *
 330  */
 331
 332
 333 define ( 'PROBE_NORMAL',   0);
 334 define ( 'PROBE_DIASPORA', 1);
 335
 336 function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
 337         require_once('include/email.php');
 338
 339         $result = array();
 340
 341         if(! $url)
 342                 return $result;
 343
 344         $result = Cache::get("probe_url:".$mode.":".$url);
 345         if (!is_null($result)) {
 346                 $result = unserialize($result);
 347                 return $result;
 348         }
 349
 350         $network = null;
 351         $diaspora = false;
 352         $diaspora_base = '';
 353         $diaspora_guid = '';
 354         $diaspora_key = '';
 355         $has_lrdd = false;
 356         $email_conversant = false;
 357         $connectornetworks = false;
 358         $appnet = false;
 359
 360         if (strpos($url,'twitter.com')) {
 361                 $connectornetworks = true;
 362                 $network = NETWORK_TWITTER;
 363         }
 364
 365         // Twitter is deactivated since twitter closed its old API
 366         //$twitter = ((strpos($url,'twitter.com') !== false) ? true : false);
 367         $lastfm  = ((strpos($url,'last.fm/user') !== false) ? true : false);
 368
 369         $at_addr = ((strpos($url,'@') !== false) ? true : false);
 370
 371         if((!$appnet) && (!$lastfm) && !$connectornetworks) {
 372
 373                 if(strpos($url,'mailto:') !== false && $at_addr) {
 374                         $url = str_replace('mailto:','',$url);
 375                         $links = array();
 376                 }
 377                 else
 378                         $links = lrdd($url);
 379
 380                 if(count($links)) {
 381                         $has_lrdd = true;
 382
 383                         logger('probe_url: found lrdd links: ' . print_r($links,true), LOGGER_DATA);
 384                         foreach($links as $link) {
 385                                 if($link['@attributes']['rel'] === NAMESPACE_ZOT)
 386                                         $zot = unamp($link['@attributes']['href']);
 387                                 if($link['@attributes']['rel'] === NAMESPACE_DFRN)
 388                                         $dfrn = unamp($link['@attributes']['href']);
 389                                 if($link['@attributes']['rel'] === 'salmon')
 390                                         $notify = unamp($link['@attributes']['href']);
 391                                 if($link['@attributes']['rel'] === NAMESPACE_FEED)
 392                                         $poll = unamp($link['@attributes']['href']);
 393                                 if($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard')
 394                                         $hcard = unamp($link['@attributes']['href']);
 395                                 if($link['@attributes']['rel'] === 'http://webfinger.net/rel/profile-page')
 396                                         $profile = unamp($link['@attributes']['href']);
 397                                 if($link['@attributes']['rel'] === 'http://portablecontacts.net/spec/1.0')
 398                                         $poco = unamp($link['@attributes']['href']);
 399                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/seed_location') {
 400                                         $diaspora_base = unamp($link['@attributes']['href']);
 401                                         $diaspora = true;
 402                                 }
 403                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/guid') {
 404                                         $diaspora_guid = unamp($link['@attributes']['href']);
 405                                         $diaspora = true;
 406                                 }
 407                                 if($link['@attributes']['rel'] === 'diaspora-public-key') {
 408                                         $diaspora_key = base64_decode(unamp($link['@attributes']['href']));
 409                                         if(strstr($diaspora_key,'RSA '))
 410                                                 $pubkey = rsatopem($diaspora_key);
 411                                         else
 412                                                 $pubkey = $diaspora_key;
 413                                         $diaspora = true;
 414                                 }
 415                                 if(($link['@attributes']['rel'] === 'http://ostatus.org/schema/1.0/subscribe') AND ($mode == PROBE_NORMAL)) {
 416                                         $diaspora = false;
 417                                 }
 418                         }
 419
 420                         // Status.Net can have more than one profile URL. We need to match the profile URL
 421                         // to a contact on incoming messages to prevent spam, and we won't know which one
 422                         // to match. So in case of two, one of them is stored as an alias. Only store URL's
 423                         // and not webfinger user@host aliases. If they've got more than two non-email style
 424                         // aliases, let's hope we're lucky and get one that matches the feed author-uri because
 425                         // otherwise we're screwed.
 426
 427                         foreach($links as $link) {
 428                                 if($link['@attributes']['rel'] === 'alias') {
 429                                         if(strpos($link['@attributes']['href'],'@') === false) {
 430                                                 if(isset($profile)) {
 431                                                         if($link['@attributes']['href'] !== $profile)
 432                                                                 $alias = unamp($link['@attributes']['href']);
 433                                                 }
 434                                                 else
 435                                                         $profile = unamp($link['@attributes']['href']);
 436                                         }
 437                                 }
 438                         }
 439
 440                         // If the profile is different from the url then the url is abviously an alias
 441                         if (($alias == "") AND ($profile != "") AND !$at_addr AND (normalise_link($profile) != normalise_link($url)))
 442                                 $alias = $url;
 443                 }
 444                 elseif($mode == PROBE_NORMAL) {
 445
 446                         // Check email
 447
 448                         $orig_url = $url;
 449                         if((strpos($orig_url,'@')) && validate_email($orig_url)) {
 450                                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1",
 451                                         intval(local_user())
 452                                 );
 453                                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1",
 454                                         intval(local_user())
 455                                 );
 456                                 if(count($x) && count($r)) {
 457                                         $mailbox = construct_mailbox_name($r[0]);
 458                                         $password = '';
 459                                         openssl_private_decrypt(hex2bin($r[0]['pass']),$password,$x[0]['prvkey']);
 460                                         $mbox = email_connect($mailbox,$r[0]['user'],$password);
 461                                         if(! $mbox)
 462                                                 logger('probe_url: email_connect failed.');
 463                                         unset($password);
 464                                 }
 465                                 if($mbox) {
 466                                         $msgs = email_poll($mbox,$orig_url);
 467                                         logger('probe_url: searching ' . $orig_url . ', ' . count($msgs) . ' messages found.', LOGGER_DEBUG);
 468                                         if(count($msgs)) {
 469                                                 $addr = $orig_url;
 470                                                 $network = NETWORK_MAIL;
 471                                                 $name = substr($url,0,strpos($url,'@'));
 472                                                 $phost = substr($url,strpos($url,'@')+1);
 473                                                 $profile = 'http://' . $phost;
 474                                                 // fix nick character range
 475                                                 $vcard = array('fn' => $name, 'nick' => $name, 'photo' => avatar_img($url));
 476                                                 $notify = 'smtp ' . random_string();
 477                                                 $poll = 'email ' . random_string();
 478                                                 $priority = 0;
 479                                                 $x = email_msg_meta($mbox,$msgs[0]);
 480                                                 if(stristr($x[0]->from,$orig_url))
 481                                                         $adr = imap_rfc822_parse_adrlist($x[0]->from,'');
 482                                                 elseif(stristr($x[0]->to,$orig_url))
 483                                                         $adr = imap_rfc822_parse_adrlist($x[0]->to,'');
 484                                                 if(isset($adr)) {
 485                                                         foreach($adr as $feadr) {
 486                                                                 if((strcasecmp($feadr->mailbox,$name) == 0)
 487                                                                         &&(strcasecmp($feadr->host,$phost) == 0)
 488                                                                         && (strlen($feadr->personal))) {
 489
 490                                                                         $personal = imap_mime_header_decode($feadr->personal);
 491                                                                         $vcard['fn'] = "";
 492                                                                         foreach($personal as $perspart)
 493                                                                                 if ($perspart->charset != "default")
 494                                                                                         $vcard['fn'] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
 495                                                                                 else
 496                                                                                         $vcard['fn'] .= $perspart->text;
 497
 498                                                                         $vcard['fn'] = notags($vcard['fn']);
 499                                                                 }
 500                                                         }
 501                                                 }
 502                                         }
 503                                         imap_close($mbox);
 504                                 }
 505                         }
 506                 }
 507         }
 508
 509         if($mode == PROBE_NORMAL) {
 510
 511                 if(strlen($zot)) {
 512                         $s = fetch_url($zot);
 513                         if($s) {
 514                                 $j = json_decode($s);
 515                                 if($j) {
 516                                         $network = NETWORK_ZOT;
 517                                         $vcard   = array(
 518                                                 'fn'    => $j->fullname,
 519                                                 'nick'  => $j->nickname,
 520                                                 'photo' => $j->photo
 521                                         );
 522                                         $profile  = $j->url;
 523                                         $notify   = $j->post;
 524                                         $pubkey   = $j->pubkey;
 525                                         $poll     = 'N/A';
 526                                 }
 527                         }
 528                 }
 529
 530
 531                 if(strlen($dfrn)) {
 532                         $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn, true);
 533                         if(is_array($ret) && x($ret,'dfrn-request')) {
 534                                 $network = NETWORK_DFRN;
 535                                 $request = $ret['dfrn-request'];
 536                                 $confirm = $ret['dfrn-confirm'];
 537                                 $notify  = $ret['dfrn-notify'];
 538                                 $poll    = $ret['dfrn-poll'];
 539
 540                                 $vcard = array();
 541                                 $vcard['fn'] = $ret['fn'];
 542                                 $vcard['nick'] = $ret['nick'];
 543                                 $vcard['photo'] = $ret['photo'];
 544                         }
 545                 }
 546         }
 547
 548         if($diaspora && $diaspora_base && $diaspora_guid) {
 549                 if($mode == PROBE_DIASPORA || ! $notify) {
 550                         $notify = $diaspora_base . 'receive/users/' . $diaspora_guid;
 551                         $batch  = $diaspora_base . 'receive/public' ;
 552                 }
 553                 if(strpos($url,'@'))
 554                         $addr = str_replace('acct:', '', $url);
 555         }
 556
 557         if($network !== NETWORK_ZOT && $network !== NETWORK_DFRN && $network !== NETWORK_MAIL) {
 558                 if($diaspora)
 559                         $network = NETWORK_DIASPORA;
 560                 elseif($has_lrdd AND ($notify))
 561                         $network  = NETWORK_OSTATUS;
 562
 563                 if(strpos($url,'@'))
 564                         $addr = str_replace('acct:', '', $url);
 565
 566                 $priority = 0;
 567
 568                 if($hcard && ! $vcard) {
 569                         $vcard = scrape_vcard($hcard);
 570
 571                         // Google doesn't use absolute url in profile photos
 572
 573                         if((x($vcard,'photo')) && substr($vcard['photo'],0,1) == '/') {
 574                                 $h = @parse_url($hcard);
 575                                 if($h)
 576                                         $vcard['photo'] = $h['scheme'] . '://' . $h['host'] . $vcard['photo'];
 577                         }
 578
 579                         logger('probe_url: scrape_vcard: ' . print_r($vcard,true), LOGGER_DATA);
 580                 }
 581
 582                 if($diaspora && $addr) {
 583                         // Diaspora returns the name as the nick. As the nick will never be updated,
 584                         // let's use the Diaspora nickname (the first part of the handle) as the nick instead
 585                         $addr_parts = explode('@', $addr);
 586                         $vcard['nick'] = $addr_parts[0];
 587                 }
 588
 589                 /* if($twitter) {
 590                         logger('twitter: setup');
 591                         $tid = basename($url);
 592                         $tapi = 'https://api.twitter.com/1/statuses/user_timeline.rss';
 593                         if(intval($tid))
 594                                 $poll = $tapi . '?user_id=' . $tid;
 595                         else
 596                                 $poll = $tapi . '?screen_name=' . $tid;
 597                         $profile = 'http://twitter.com/#!/' . $tid;
 598                         //$vcard['photo'] = 'https://api.twitter.com/1/users/profile_image/' . $tid;
 599                         $vcard['photo'] = 'https://api.twitter.com/1/users/profile_image?screen_name=' . $tid . '&size=bigger';
 600                         $vcard['nick'] = $tid;
 601                         $vcard['fn'] = $tid;
 602                 } */
 603
 604                 if($lastfm) {
 605                         $profile = $url;
 606                         $poll = str_replace(array('www.','last.fm/'),array('','ws.audioscrobbler.com/1.0/'),$url) . '/recenttracks.rss';
 607                         $vcard['nick'] = basename($url);
 608                         $vcard['fn'] = $vcard['nick'] . t(' on Last.fm');
 609                         $network = NETWORK_FEED;
 610                 }
 611
 612                 if(! x($vcard,'fn'))
 613                         if(x($vcard,'nick'))
 614                                 $vcard['fn'] = $vcard['nick'];
 615
 616                 $check_feed = false;
 617
 618                 if(stristr($url,'tumblr.com') && (! stristr($url,'/rss'))) {
 619                         $poll = $url . '/rss';
 620                         $check_feed = true;
 621                         // Will leave it to others to figure out how to grab the avatar, which is on the $url page in the open graph meta links
 622                 }
 623
 624                 if($appnet || ! $poll)
 625                         $check_feed = true;
 626                 if((! isset($vcard)) || (! x($vcard,'fn')) || (! $profile))
 627                         $check_feed = true;
 628                 if(($at_addr) && (! count($links)))
 629                         $check_feed = false;
 630
 631                 if ($connectornetworks)
 632                         $check_feed = false;
 633
 634                 if($check_feed) {
 635
 636                         $feedret = scrape_feed(($poll) ? $poll : $url);
 637
 638                         logger('probe_url: scrape_feed ' . (($poll)? $poll : $url) . ' returns: ' . print_r($feedret,true), LOGGER_DATA);
 639                         if(count($feedret) && ($feedret['feed_atom'] || $feedret['feed_rss'])) {
 640                                 $poll = ((x($feedret,'feed_atom')) ? unamp($feedret['feed_atom']) : unamp($feedret['feed_rss']));
 641                                 if(! x($vcard))
 642                                         $vcard = array();
 643                         }
 644
 645                         if(x($feedret,'photo') && (! x($vcard,'photo')))
 646                                 $vcard['photo'] = $feedret['photo'];
 647                         require_once('library/simplepie/simplepie.inc');
 648                         $feed = new SimplePie();
 649                         $xml = fetch_url($poll);
 650
 651                         logger('probe_url: fetch feed: ' . $poll . ' returns: ' . $xml, LOGGER_DATA);
 652                         $a = get_app();
 653
 654                         logger('probe_url: scrape_feed: headers: ' . $a->get_curl_headers(), LOGGER_DATA);
 655
 656                         // Don't try and parse an empty string
 657                         $feed->set_raw_data(($xml) ? $xml : '<?xml version="1.0" encoding="utf-8" ?><xml></xml>');
 658
 659                         $feed->init();
 660                         if($feed->error()) {
 661                                 logger('probe_url: scrape_feed: Error parsing XML: ' . $feed->error());
 662                                 $network = NETWORK_PHANTOM;
 663                         }
 664
 665                         if(! x($vcard,'photo'))
 666                                 $vcard['photo'] = $feed->get_image_url();
 667                         $author = $feed->get_author();
 668
 669                         if($author) {
 670                                 $vcard['fn'] = unxmlify(trim($author->get_name()));
 671                                 if(! $vcard['fn'])
 672                                         $vcard['fn'] = trim(unxmlify($author->get_email()));
 673                                 if(strpos($vcard['fn'],'@') !== false)
 674                                         $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
 675
 676                                 $email = unxmlify($author->get_email());
 677                                 if(! $profile && $author->get_link())
 678                                         $profile = trim(unxmlify($author->get_link()));
 679                                 if(! $vcard['photo']) {
 680                                         $rawtags = $feed->get_feed_tags( SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
 681                                         if($rawtags) {
 682                                                 $elems = $rawtags[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10];
 683                                                 if((x($elems,'link')) && ($elems['link'][0]['attribs']['']['rel'] === 'photo'))
 684                                                         $vcard['photo'] = $elems['link'][0]['attribs']['']['href'];
 685                                         }
 686                                 }
 687                                 // Fetch fullname via poco:displayName
 688                                 $pocotags = $feed->get_feed_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
 689                                 if ($pocotags) {
 690                                         $elems = $pocotags[0]['child']['http://portablecontacts.net/spec/1.0'];
 691                                         if (isset($elems["displayName"]))
 692                                                 $vcard['fn'] = $elems["displayName"][0]["data"];
 693                                         if (isset($elems["preferredUsername"]))
 694                                                 $vcard['nick'] = $elems["preferredUsername"][0]["data"];
 695                                 }
 696                         }
 697                         else {
 698                                 $item = $feed->get_item(0);
 699                                 if($item) {
 700                                         $author = $item->get_author();
 701                                         if($author) {
 702                                                 $vcard['fn'] = trim(unxmlify($author->get_name()));
 703                                                 if(! $vcard['fn'])
 704                                                         $vcard['fn'] = trim(unxmlify($author->get_email()));
 705                                                 if(strpos($vcard['fn'],'@') !== false)
 706                                                         $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
 707                                                 $email = unxmlify($author->get_email());
 708                                                 if(! $profile && $author->get_link())
 709                                                         $profile = trim(unxmlify($author->get_link()));
 710                                         }
 711                                         if(! $vcard['photo']) {
 712                                                 $rawmedia = $item->get_item_tags('http://search.yahoo.com/mrss/','thumbnail');
 713                                                 if($rawmedia && $rawmedia[0]['attribs']['']['url'])
 714                                                         $vcard['photo'] = unxmlify($rawmedia[0]['attribs']['']['url']);
 715                                         }
 716                                         if(! $vcard['photo']) {
 717                                                 $rawtags = $item->get_item_tags( SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
 718                                                 if($rawtags) {
 719                                                         $elems = $rawtags[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10];
 720                                                         if((x($elems,'link')) && ($elems['link'][0]['attribs']['']['rel'] === 'photo'))
 721                                                                 $vcard['photo'] = $elems['link'][0]['attribs']['']['href'];
 722                                                 }
 723                                         }
 724                                 }
 725                         }
 726
 727                         if((! $vcard['photo']) && strlen($email))
 728                                 $vcard['photo'] = avatar_img($email);
 729                         if($poll === $profile)
 730                                 $lnk = $feed->get_permalink();
 731                         if(isset($lnk) && strlen($lnk))
 732                                 $profile = $lnk;
 733
 734                         if(! $network) {
 735                                 $network = NETWORK_FEED;
 736                                 // If it is a feed, don't take the author name as feed name
 737                                 unset($vcard['fn']);
 738                         }
 739                         if(! (x($vcard,'fn')))
 740                                 $vcard['fn'] = notags($feed->get_title());
 741                         if(! (x($vcard,'fn')))
 742                                 $vcard['fn'] = notags($feed->get_description());
 743
 744                         if(strpos($vcard['fn'],'Twitter / ') !== false) {
 745                                 $vcard['fn'] = substr($vcard['fn'],strpos($vcard['fn'],'/')+1);
 746                                 $vcard['fn'] = trim($vcard['fn']);
 747                         }
 748                         if(! x($vcard,'nick')) {
 749                                 $vcard['nick'] = strtolower(notags(unxmlify($vcard['fn'])));
 750                                 if(strpos($vcard['nick'],' '))
 751                                         $vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' ')));
 752                         }
 753                         if(! $priority)
 754                                 $priority = 2;
 755                 }
 756         }
 757
 758         if(! x($vcard,'photo')) {
 759                 $a = get_app();
 760                 $vcard['photo'] = $a->get_baseurl() . '/images/person-175.jpg' ;
 761         }
 762
 763         if(! $profile)
 764                 $profile = $url;
 765
 766         // No human could be associated with this link, use the URL as the contact name
 767
 768         if(($network === NETWORK_FEED) && ($poll) && (! x($vcard,'fn')))
 769                 $vcard['fn'] = $url;
 770
 771         if (($notify != "") AND ($poll != "")) {
 772                 $baseurl = matching(normalise_link($notify), normalise_link($poll));
 773
 774                 $baseurl2 = matching($baseurl, normalise_link($profile));
 775                 if ($baseurl2 != "")
 776                         $baseurl = $baseurl2;
 777         }
 778
 779         if (($baseurl == "") AND ($notify != ""))
 780                 $baseurl = matching(normalise_link($profile), normalise_link($notify));
 781
 782         if (($baseurl == "") AND ($poll != ""))
 783                 $baseurl = matching(normalise_link($profile), normalise_link($poll));
 784
 785         $baseurl = rtrim($baseurl, "/");
 786
 787         if(strpos($url,'@') AND ($addr == "") AND ($network == NETWORK_DFRN))
 788                 $addr = str_replace('acct:', '', $url);
 789
 790         $vcard['fn'] = notags($vcard['fn']);
 791         $vcard['nick'] = str_replace(' ','',notags($vcard['nick']));
 792
 793         $result['name'] = $vcard['fn'];
 794         $result['nick'] = $vcard['nick'];
 795         $result['url'] = $profile;
 796         $result['addr'] = $addr;
 797         $result['batch'] = $batch;
 798         $result['notify'] = $notify;
 799         $result['poll'] = $poll;
 800         $result['request'] = $request;
 801         $result['confirm'] = $confirm;
 802         $result['poco'] = $poco;
 803         $result['photo'] = $vcard['photo'];
 804         $result['priority'] = $priority;
 805         $result['network'] = $network;
 806         $result['alias'] = $alias;
 807         $result['pubkey'] = $pubkey;
 808         $result['baseurl'] = $baseurl;
 809
 810         logger('probe_url: ' . print_r($result,true), LOGGER_DEBUG);
 811
 812         if ($level == 1) {
 813                 // Trying if it maybe a diaspora account
 814                 if (($result['network'] == NETWORK_FEED) OR ($result['addr'] == "")) {
 815                         require_once('include/bbcode.php');
 816                         $address = GetProfileUsername($url, "", true);
 817                         $result2 = probe_url($address, $mode, ++$level);
 818                         if ($result2['network'] != "")
 819                                 $result = $result2;
 820                 }
 821
 822                 // Maybe it's some non standard GNU Social installation (Single user, subfolder or no uri rewrite)
 823                 if (($result['network'] == NETWORK_FEED) AND ($result['baseurl'] != "") AND ($result['nick'] != "")) {
 824                         $addr = $result['nick'].'@'.str_replace("http://", "", $result['baseurl']);
 825                         $result2 = probe_url($addr, $mode, ++$level);
 826                         if (($result2['network'] != "") AND ($result2['network'] != NETWORK_FEED))
 827                                 $result = $result2;
 828                 }
 829         }
 830
 831         // Only store into the cache if the value seems to be valid
 832         if ($result['network'] != NETWORK_FEED)
 833                 Cache::set("probe_url:".$mode.":".$url,serialize($result), CACHE_DAY);
 834
 835         return $result;
 836 }
 837
 838 function matching($part1, $part2) {
 839         $len = min(strlen($part1), strlen($part2));
 840
 841         $match = "";
 842         $matching = true;
 843         $i = 0;
 844         while (($i <= $len) AND $matching) {
 845                 if (substr($part1, $i, 1) == substr($part2, $i, 1))
 846                         $match .= substr($part1, $i, 1);
 847                 else
 848                         $matching = false;
 849
 850                 $i++;
 851         }
 852         return($match);
 853 }