include/Scrape.php

   1 <?php
   2
   3 require_once('library/HTML5/Parser.php');
   4 require_once('include/crypto.php');
   5 require_once('include/feed.php');
   6
   7 if(! function_exists('scrape_dfrn')) {
   8 function scrape_dfrn($url, $dont_probe = false) {
   9
  10         $a = get_app();
  11
  12         $ret = array();
  13
  14         logger('scrape_dfrn: url=' . $url);
  15
  16         // Try to fetch the data from noscrape. This is faster than parsing the HTML
  17         $noscrape = str_replace("/hcard/", "/noscrape/", $url);
  18         $noscrapejson = fetch_url($noscrape);
  19         $noscrapedata = array();
  20         if ($noscrapejson) {
  21                 $noscrapedata = json_decode($noscrapejson, true);
  22
  23                 if (is_array($noscrapedata)) {
  24                         if ($noscrapedata["nick"] != "")
  25                                 return($noscrapedata);
  26                 } else
  27                         $noscrapedata = array();
  28         }
  29
  30         $s = fetch_url($url);
  31
  32         if(! $s)
  33                 return $ret;
  34
  35         if (!$dont_probe) {
  36                 $probe = probe_url($url);
  37
  38                 if (isset($probe["addr"]))
  39                         $ret["addr"] = $probe["addr"];
  40         }
  41
  42         $headers = $a->get_curl_headers();
  43         logger('scrape_dfrn: headers=' . $headers, LOGGER_DEBUG);
  44
  45
  46         $lines = explode("\n",$headers);
  47         if(count($lines)) {
  48                 foreach($lines as $line) {
  49                         // don't try and run feeds through the html5 parser
  50                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
  51                                 return ret;
  52                 }
  53         }
  54
  55         try {
  56                 $dom = HTML5_Parser::parse($s);
  57         } catch (DOMException $e) {
  58                 logger('scrape_dfrn: parse error: ' . $e);
  59         }
  60
  61         if(! $dom)
  62                 return $ret;
  63
  64         $items = $dom->getElementsByTagName('link');
  65
  66         // get DFRN link elements
  67
  68         foreach($items as $item) {
  69                 $x = $item->getAttribute('rel');
  70                 if(($x === 'alternate') && ($item->getAttribute('type') === 'application/atom+xml'))
  71                         $ret['feed_atom'] = $item->getAttribute('href');
  72                 if(substr($x,0,5) == "dfrn-") {
  73                         $ret[$x] = $item->getAttribute('href');
  74                 }
  75                 if($x === 'lrdd') {
  76                         $decoded = urldecode($item->getAttribute('href'));
  77                         if(preg_match('/acct:([^@]*)@/',$decoded,$matches))
  78                                 $ret['nick'] = $matches[1];
  79                 }
  80         }
  81
  82         // Pull out hCard profile elements
  83
  84         $largest_photo = 0;
  85
  86         $items = $dom->getElementsByTagName('*');
  87         foreach($items as $item) {
  88                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
  89                         $level2 = $item->getElementsByTagName('*');
  90                         foreach($level2 as $x) {
  91                                 if(attribute_contains($x->getAttribute('class'),'fn')) {
  92                                         $ret['fn'] = $x->textContent;
  93                                 }
  94                                 if((attribute_contains($x->getAttribute('class'),'photo'))
  95                                         || (attribute_contains($x->getAttribute('class'),'avatar'))) {
  96                                         $size = intval($x->getAttribute('width'));
  97                                         // dfrn prefers 175, so if we find this, we set largest_size so it can't be topped.
  98                                         if(($size > $largest_photo) || ($size == 175) || (! $largest_photo)) {
  99                                                 $ret['photo'] = $x->getAttribute('src');
 100                                                 $largest_photo = (($size == 175) ? 9999 : $size);
 101                                         }
 102                                 }
 103                                 if(attribute_contains($x->getAttribute('class'),'key')) {
 104                                         $ret['key'] = $x->textContent;
 105                                 }
 106                         }
 107                 }
 108         }
 109         return array_merge($ret, $noscrapedata);
 110 }}
 111
 112
 113
 114
 115
 116
 117 if(! function_exists('validate_dfrn')) {
 118 function validate_dfrn($a) {
 119         $errors = 0;
 120         if(! x($a,'key'))
 121                 $errors ++;
 122         if(! x($a,'dfrn-request'))
 123                 $errors ++;
 124         if(! x($a,'dfrn-confirm'))
 125                 $errors ++;
 126         if(! x($a,'dfrn-notify'))
 127                 $errors ++;
 128         if(! x($a,'dfrn-poll'))
 129                 $errors ++;
 130         return $errors;
 131 }}
 132
 133 if(! function_exists('scrape_meta')) {
 134 function scrape_meta($url) {
 135
 136         $a = get_app();
 137
 138         $ret = array();
 139
 140         logger('scrape_meta: url=' . $url);
 141
 142         $s = fetch_url($url);
 143
 144         if(! $s)
 145                 return $ret;
 146
 147         $headers = $a->get_curl_headers();
 148         logger('scrape_meta: headers=' . $headers, LOGGER_DEBUG);
 149
 150         $lines = explode("\n",$headers);
 151         if(count($lines)) {
 152                 foreach($lines as $line) {
 153                         // don't try and run feeds through the html5 parser
 154                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
 155                                 return ret;
 156                 }
 157         }
 158
 159         try {
 160                 $dom = HTML5_Parser::parse($s);
 161         } catch (DOMException $e) {
 162                 logger('scrape_meta: parse error: ' . $e);
 163         }
 164
 165         if(! $dom)
 166                 return $ret;
 167
 168         $items = $dom->getElementsByTagName('meta');
 169
 170         // get DFRN link elements
 171
 172         foreach($items as $item) {
 173                 $x = $item->getAttribute('name');
 174                 if(substr($x,0,5) == "dfrn-")
 175                         $ret[$x] = $item->getAttribute('content');
 176         }
 177
 178         return $ret;
 179 }}
 180
 181
 182 if(! function_exists('scrape_vcard')) {
 183 function scrape_vcard($url) {
 184
 185         $a = get_app();
 186
 187         $ret = array();
 188
 189         logger('scrape_vcard: url=' . $url);
 190
 191         $s = fetch_url($url);
 192
 193         if(! $s)
 194                 return $ret;
 195
 196         $headers = $a->get_curl_headers();
 197         $lines = explode("\n",$headers);
 198         if(count($lines)) {
 199                 foreach($lines as $line) {
 200                         // don't try and run feeds through the html5 parser
 201                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
 202                                 return ret;
 203                 }
 204         }
 205
 206         try {
 207                 $dom = HTML5_Parser::parse($s);
 208         } catch (DOMException $e) {
 209                 logger('scrape_vcard: parse error: ' . $e);
 210         }
 211
 212         if(! $dom)
 213                 return $ret;
 214
 215         // Pull out hCard profile elements
 216
 217         $largest_photo = 0;
 218
 219         $items = $dom->getElementsByTagName('*');
 220         foreach($items as $item) {
 221                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
 222                         $level2 = $item->getElementsByTagName('*');
 223                         foreach($level2 as $x) {
 224                                 if(attribute_contains($x->getAttribute('class'),'fn'))
 225                                         $ret['fn'] = $x->textContent;
 226                                 if((attribute_contains($x->getAttribute('class'),'photo'))
 227                                         || (attribute_contains($x->getAttribute('class'),'avatar'))) {
 228                                         $size = intval($x->getAttribute('width'));
 229                                         if(($size > $largest_photo) || (! $largest_photo)) {
 230                                                 $ret['photo'] = $x->getAttribute('src');
 231                                                 $largest_photo = $size;
 232                                         }
 233                                 }
 234                                 if((attribute_contains($x->getAttribute('class'),'nickname'))
 235                                         || (attribute_contains($x->getAttribute('class'),'uid'))) {
 236                                         $ret['nick'] = $x->textContent;
 237                                 }
 238                         }
 239                 }
 240         }
 241
 242         return $ret;
 243 }}
 244
 245
 246 if(! function_exists('scrape_feed')) {
 247 function scrape_feed($url) {
 248
 249         $a = get_app();
 250
 251         $ret = array();
 252         $cookiejar = tempnam(get_temppath(), 'cookiejar-scrape-feed-');
 253         $s = fetch_url($url, false, $redirects, 0, Null, $cookiejar);
 254         unlink($cookiejar);
 255
 256         $headers = $a->get_curl_headers();
 257         $code = $a->get_curl_code();
 258
 259         logger('scrape_feed: returns: ' . $code . ' headers=' . $headers, LOGGER_DEBUG);
 260
 261         if(! $s) {
 262                 logger('scrape_feed: no data returned for ' . $url);
 263                 return $ret;
 264         }
 265
 266
 267         $lines = explode("\n",$headers);
 268         if(count($lines)) {
 269                 foreach($lines as $line) {
 270                         if(stristr($line,'content-type:')) {
 271                                 if(stristr($line,'application/atom+xml') || stristr($s,'<feed')) {
 272                                         $ret['feed_atom'] = $url;
 273                                         return $ret;
 274                                 }
 275                                 if(stristr($line,'application/rss+xml') || stristr($s,'<rss')) {
 276                                         $ret['feed_rss'] = $url;
 277                                         return $ret;
 278                                 }
 279                         }
 280                 }
 281                 // perhaps an RSS version 1 feed with a generic or incorrect content-type?
 282                 if(stristr($s,'</item>')) {
 283                         $ret['feed_rss'] = $url;
 284                         return $ret;
 285                 }
 286         }
 287
 288         $basename = implode('/', array_slice(explode('/',$url),0,3)) . '/';
 289
 290         $doc = new DOMDocument();
 291         @$doc->loadHTML($s);
 292         $xpath = new DomXPath($doc);
 293
 294         $base = $xpath->query("//base");
 295         foreach ($base as $node) {
 296                 $attr = array();
 297
 298                 if ($node->attributes->length)
 299                         foreach ($node->attributes as $attribute)
 300                                 $attr[$attribute->name] = $attribute->value;
 301
 302                 if ($attr["href"] != "")
 303                         $basename = $attr["href"] ;
 304         }
 305
 306         $list = $xpath->query("//link");
 307         foreach ($list as $node) {
 308                 $attr = array();
 309
 310                 if ($node->attributes->length)
 311                         foreach ($node->attributes as $attribute)
 312                                 $attr[$attribute->name] = $attribute->value;
 313
 314                 if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/atom+xml"))
 315                         $ret["feed_atom"] = $attr["href"];
 316
 317                 if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/rss+xml"))
 318                         $ret["feed_rss"] = $attr["href"];
 319         }
 320
 321         // Drupal and perhaps others only provide relative URLs. Turn them into absolute.
 322
 323         if(x($ret,'feed_atom') && (! strstr($ret['feed_atom'],'://')))
 324                 $ret['feed_atom'] = $basename . $ret['feed_atom'];
 325         if(x($ret,'feed_rss') && (! strstr($ret['feed_rss'],'://')))
 326                 $ret['feed_rss'] = $basename . $ret['feed_rss'];
 327
 328         return $ret;
 329 }}
 330
 331
 332 /**
 333  *
 334  * Probe a network address to discover what kind of protocols we need to communicate with it.
 335  *
 336  * Warning: this function is a bit touchy and there are some subtle dependencies within the logic flow.
 337  * Edit with care.
 338  *
 339  */
 340
 341 /**
 342  *
 343  * PROBE_DIASPORA has a bias towards returning Diaspora information
 344  * while PROBE_NORMAL has a bias towards dfrn/zot - in the case where
 345  * an address (such as a Friendica address) supports more than one type
 346  * of network.
 347  *
 348  */
 349
 350
 351 define ( 'PROBE_NORMAL',   0);
 352 define ( 'PROBE_DIASPORA', 1);
 353
 354 function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
 355         require_once('include/email.php');
 356
 357         $result = array();
 358
 359         if (!$url)
 360                 return $result;
 361
 362         $result = Cache::get("probe_url:".$mode.":".$url);
 363         if (!is_null($result)) {
 364                 $result = unserialize($result);
 365                 return $result;
 366         }
 367
 368         $original_url = $url;
 369         $network = null;
 370         $diaspora = false;
 371         $diaspora_base = '';
 372         $diaspora_guid = '';
 373         $diaspora_key = '';
 374         $has_lrdd = false;
 375         $email_conversant = false;
 376         $connectornetworks = false;
 377         $appnet = false;
 378
 379         if (strpos($url,'twitter.com')) {
 380                 $connectornetworks = true;
 381                 $network = NETWORK_TWITTER;
 382         }
 383
 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($links) == 0) AND strstr($url, "/index.php")) {
 398                         $url = str_replace("/index.php", "", $url);
 399                         $links = lrdd($url);
 400                 }
 401
 402                 if (count($links)) {
 403                         $has_lrdd = true;
 404
 405                         logger('probe_url: found lrdd links: ' . print_r($links,true), LOGGER_DATA);
 406                         foreach($links as $link) {
 407                                 if($link['@attributes']['rel'] === NAMESPACE_ZOT)
 408                                         $zot = unamp($link['@attributes']['href']);
 409                                 if($link['@attributes']['rel'] === NAMESPACE_DFRN)
 410                                         $dfrn = unamp($link['@attributes']['href']);
 411                                 if($link['@attributes']['rel'] === 'salmon')
 412                                         $notify = unamp($link['@attributes']['href']);
 413                                 if($link['@attributes']['rel'] === NAMESPACE_FEED)
 414                                         $poll = unamp($link['@attributes']['href']);
 415                                 if($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard')
 416                                         $hcard = unamp($link['@attributes']['href']);
 417                                 if($link['@attributes']['rel'] === 'http://webfinger.net/rel/profile-page')
 418                                         $profile = unamp($link['@attributes']['href']);
 419                                 if($link['@attributes']['rel'] === 'http://portablecontacts.net/spec/1.0')
 420                                         $poco = unamp($link['@attributes']['href']);
 421                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/seed_location') {
 422                                         $diaspora_base = unamp($link['@attributes']['href']);
 423                                         $diaspora = true;
 424                                 }
 425                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/guid') {
 426                                         $diaspora_guid = unamp($link['@attributes']['href']);
 427                                         $diaspora = true;
 428                                 }
 429                                 if($link['@attributes']['rel'] === 'diaspora-public-key') {
 430                                         $diaspora_key = base64_decode(unamp($link['@attributes']['href']));
 431                                         if(strstr($diaspora_key,'RSA '))
 432                                                 $pubkey = rsatopem($diaspora_key);
 433                                         else
 434                                                 $pubkey = $diaspora_key;
 435                                         $diaspora = true;
 436                                 }
 437                                 if(($link['@attributes']['rel'] === 'http://ostatus.org/schema/1.0/subscribe') AND ($mode == PROBE_NORMAL)) {
 438                                         $diaspora = false;
 439                                 }
 440                         }
 441
 442                         // Status.Net can have more than one profile URL. We need to match the profile URL
 443                         // to a contact on incoming messages to prevent spam, and we won't know which one
 444                         // to match. So in case of two, one of them is stored as an alias. Only store URL's
 445                         // and not webfinger user@host aliases. If they've got more than two non-email style
 446                         // aliases, let's hope we're lucky and get one that matches the feed author-uri because
 447                         // otherwise we're screwed.
 448
 449                         $backup_alias = "";
 450
 451                         foreach($links as $link) {
 452                                 if($link['@attributes']['rel'] === 'alias') {
 453                                         if(strpos($link['@attributes']['href'],'@') === false) {
 454                                                 if(isset($profile)) {
 455                                                         $alias_url = $link['@attributes']['href'];
 456
 457                                                         if(($alias_url !== $profile) AND ($backup_alias == "") AND
 458                                                                 ($alias_url !== str_replace("/index.php", "", $profile)))
 459                                                                 $backup_alias = $alias_url;
 460
 461                                                         if(($alias_url !== $profile) AND !strstr($alias_url, "index.php") AND
 462                                                                 ($alias_url !== str_replace("/index.php", "", $profile)))
 463                                                                 $alias = $alias_url;
 464                                                 }
 465                                                 else
 466                                                         $profile = unamp($link['@attributes']['href']);
 467                                         }
 468                                 }
 469                         }
 470
 471                         if ($alias == "")
 472                                 $alias = $backup_alias;
 473
 474                         // If the profile is different from the url then the url is abviously an alias
 475                         if (($alias == "") AND ($profile != "") AND !$at_addr AND (normalise_link($profile) != normalise_link($url)))
 476                                 $alias = $url;
 477                 }
 478                 elseif($mode == PROBE_NORMAL) {
 479
 480                         // Check email
 481
 482                         $orig_url = $url;
 483                         if((strpos($orig_url,'@')) && validate_email($orig_url)) {
 484                                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1",
 485                                         intval(local_user())
 486                                 );
 487                                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1",
 488                                         intval(local_user())
 489                                 );
 490                                 if(count($x) && count($r)) {
 491                                         $mailbox = construct_mailbox_name($r[0]);
 492                                         $password = '';
 493                                         openssl_private_decrypt(hex2bin($r[0]['pass']),$password,$x[0]['prvkey']);
 494                                         $mbox = email_connect($mailbox,$r[0]['user'],$password);
 495                                         if(! $mbox)
 496                                                 logger('probe_url: email_connect failed.');
 497                                         unset($password);
 498                                 }
 499                                 if($mbox) {
 500                                         $msgs = email_poll($mbox,$orig_url);
 501                                         logger('probe_url: searching ' . $orig_url . ', ' . count($msgs) . ' messages found.', LOGGER_DEBUG);
 502                                         if(count($msgs)) {
 503                                                 $addr = $orig_url;
 504                                                 $network = NETWORK_MAIL;
 505                                                 $name = substr($url,0,strpos($url,'@'));
 506                                                 $phost = substr($url,strpos($url,'@')+1);
 507                                                 $profile = 'http://' . $phost;
 508                                                 // fix nick character range
 509                                                 $vcard = array('fn' => $name, 'nick' => $name, 'photo' => avatar_img($url));
 510                                                 $notify = 'smtp ' . random_string();
 511                                                 $poll = 'email ' . random_string();
 512                                                 $priority = 0;
 513                                                 $x = email_msg_meta($mbox,$msgs[0]);
 514                                                 if(stristr($x[0]->from,$orig_url))
 515                                                         $adr = imap_rfc822_parse_adrlist($x[0]->from,'');
 516                                                 elseif(stristr($x[0]->to,$orig_url))
 517                                                         $adr = imap_rfc822_parse_adrlist($x[0]->to,'');
 518                                                 if(isset($adr)) {
 519                                                         foreach($adr as $feadr) {
 520                                                                 if((strcasecmp($feadr->mailbox,$name) == 0)
 521                                                                         &&(strcasecmp($feadr->host,$phost) == 0)
 522                                                                         && (strlen($feadr->personal))) {
 523
 524                                                                         $personal = imap_mime_header_decode($feadr->personal);
 525                                                                         $vcard['fn'] = "";
 526                                                                         foreach($personal as $perspart)
 527                                                                                 if ($perspart->charset != "default")
 528                                                                                         $vcard['fn'] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
 529                                                                                 else
 530                                                                                         $vcard['fn'] .= $perspart->text;
 531
 532                                                                         $vcard['fn'] = notags($vcard['fn']);
 533                                                                 }
 534                                                         }
 535                                                 }
 536                                         }
 537                                         imap_close($mbox);
 538                                 }
 539                         }
 540                 }
 541         }
 542
 543         if($mode == PROBE_NORMAL) {
 544
 545                 if(strlen($zot)) {
 546                         $s = fetch_url($zot);
 547                         if($s) {
 548                                 $j = json_decode($s);
 549                                 if($j) {
 550                                         $network = NETWORK_ZOT;
 551                                         $vcard   = array(
 552                                                 'fn'    => $j->fullname,
 553                                                 'nick'  => $j->nickname,
 554                                                 'photo' => $j->photo
 555                                         );
 556                                         $profile  = $j->url;
 557                                         $notify   = $j->post;
 558                                         $pubkey   = $j->pubkey;
 559                                         $poll     = 'N/A';
 560                                 }
 561                         }
 562                 }
 563
 564
 565                 if(strlen($dfrn)) {
 566                         $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn, true);
 567                         if(is_array($ret) && x($ret,'dfrn-request')) {
 568                                 $network = NETWORK_DFRN;
 569                                 $request = $ret['dfrn-request'];
 570                                 $confirm = $ret['dfrn-confirm'];
 571                                 $notify  = $ret['dfrn-notify'];
 572                                 $poll    = $ret['dfrn-poll'];
 573
 574                                 $vcard = array();
 575                                 $vcard['fn'] = $ret['fn'];
 576                                 $vcard['nick'] = $ret['nick'];
 577                                 $vcard['photo'] = $ret['photo'];
 578                         }
 579                 }
 580         }
 581
 582         // Scrape the public key from the hcard.
 583         // Diaspora will remove it from the webfinger somewhere in the future.
 584         if (($hcard != "") AND ($pubkey == "")) {
 585                 $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn, true);
 586                 if (isset($ret["key"])) {
 587                         $hcard_key = $ret["key"];
 588                         if(strstr($hcard_key,'RSA '))
 589                                 $pubkey = rsatopem($hcard_key);
 590                         else
 591                                 $pubkey = $hcard_key;
 592                 }
 593         }
 594         if($diaspora && $diaspora_base && $diaspora_guid) {
 595                 $diaspora_notify = $diaspora_base.'receive/users/'.$diaspora_guid;
 596
 597                 if($mode == PROBE_DIASPORA || ! $notify || ($notify == $diaspora_notify)) {
 598                         $notify = $diaspora_notify;
 599                         $batch  = $diaspora_base . 'receive/public' ;
 600                 }
 601                 if(strpos($url,'@'))
 602                         $addr = str_replace('acct:', '', $url);
 603         }
 604
 605         if($network !== NETWORK_ZOT && $network !== NETWORK_DFRN && $network !== NETWORK_MAIL) {
 606                 if($diaspora)
 607                         $network = NETWORK_DIASPORA;
 608                 elseif($has_lrdd AND ($notify))
 609                         $network  = NETWORK_OSTATUS;
 610
 611                 if(strpos($url,'@'))
 612                         $addr = str_replace('acct:', '', $url);
 613
 614                 $priority = 0;
 615
 616                 if($hcard && ! $vcard) {
 617                         $vcard = scrape_vcard($hcard);
 618
 619                         // Google doesn't use absolute url in profile photos
 620
 621                         if((x($vcard,'photo')) && substr($vcard['photo'],0,1) == '/') {
 622                                 $h = @parse_url($hcard);
 623                                 if($h)
 624                                         $vcard['photo'] = $h['scheme'] . '://' . $h['host'] . $vcard['photo'];
 625                         }
 626
 627                         logger('probe_url: scrape_vcard: ' . print_r($vcard,true), LOGGER_DATA);
 628                 }
 629
 630                 if($diaspora && $addr) {
 631                         // Diaspora returns the name as the nick. As the nick will never be updated,
 632                         // let's use the Diaspora nickname (the first part of the handle) as the nick instead
 633                         $addr_parts = explode('@', $addr);
 634                         $vcard['nick'] = $addr_parts[0];
 635                 }
 636
 637                 if($lastfm) {
 638                         $profile = $url;
 639                         $poll = str_replace(array('www.','last.fm/'),array('','ws.audioscrobbler.com/1.0/'),$url) . '/recenttracks.rss';
 640                         $vcard['nick'] = basename($url);
 641                         $vcard['fn'] = $vcard['nick'] . t(' on Last.fm');
 642                         $network = NETWORK_FEED;
 643                 }
 644
 645                 if(! x($vcard,'fn'))
 646                         if(x($vcard,'nick'))
 647                                 $vcard['fn'] = $vcard['nick'];
 648
 649                 $check_feed = false;
 650
 651                 if(stristr($url,'tumblr.com') && (! stristr($url,'/rss'))) {
 652                         $poll = $url . '/rss';
 653                         $check_feed = true;
 654                         // Will leave it to others to figure out how to grab the avatar, which is on the $url page in the open graph meta links
 655                 }
 656
 657                 if($appnet || ! $poll)
 658                         $check_feed = true;
 659                 if((! isset($vcard)) || (! x($vcard,'fn')) || (! $profile))
 660                         $check_feed = true;
 661                 if(($at_addr) && (! count($links)))
 662                         $check_feed = false;
 663
 664                 if ($connectornetworks)
 665                         $check_feed = false;
 666
 667                 if($check_feed) {
 668
 669                         $feedret = scrape_feed(($poll) ? $poll : $url);
 670
 671                         logger('probe_url: scrape_feed ' . (($poll)? $poll : $url) . ' returns: ' . print_r($feedret,true), LOGGER_DATA);
 672                         if(count($feedret) && ($feedret['feed_atom'] || $feedret['feed_rss'])) {
 673                                 $poll = ((x($feedret,'feed_atom')) ? unamp($feedret['feed_atom']) : unamp($feedret['feed_rss']));
 674                                 if(! x($vcard))
 675                                         $vcard = array();
 676                         }
 677
 678                         if(x($feedret,'photo') && (! x($vcard,'photo')))
 679                                 $vcard['photo'] = $feedret['photo'];
 680
 681                         $cookiejar = tempnam(get_temppath(), 'cookiejar-scrape-feed-');
 682                         $xml = fetch_url($poll, false, $redirects, 0, Null, $cookiejar);
 683                         unlink($cookiejar);
 684
 685                         logger('probe_url: fetch feed: ' . $poll . ' returns: ' . $xml, LOGGER_DATA);
 686
 687                         if ($xml == "") {
 688                                 logger("scrape_feed: XML is empty for feed ".$poll);
 689                                 $network = NETWORK_PHANTOM;
 690                         } else {
 691                                 $data = feed_import($xml,$dummy1,$dummy2, $dummy3, true);
 692
 693                                 if (!is_array($data)) {
 694                                         logger("scrape_feed: This doesn't seem to be a feed: ".$poll);
 695                                         $network = NETWORK_PHANTOM;
 696                                 } else {
 697                                         if (($vcard["photo"] == "") AND ($data["header"]["author-avatar"] != ""))
 698                                                 $vcard["photo"] = $data["header"]["author-avatar"];
 699
 700                                         if (($vcard["fn"] == "") AND ($data["header"]["author-name"] != ""))
 701                                                 $vcard["fn"] = $data["header"]["author-name"];
 702
 703                                         if (($vcard["nick"] == "") AND ($data["header"]["author-nick"] != ""))
 704                                                 $vcard["nick"] = $data["header"]["author-nick"];
 705
 706                                         if(!$profile AND ($data["header"]["author-link"] != "") AND !in_array($network, array("", NETWORK_FEED)))
 707                                                 $profile = $data["header"]["author-link"];
 708                                 }
 709                         }
 710
 711                         // Workaround for misconfigured Friendica servers
 712                         if (($network == "") AND (strstr($url, "/profile/"))) {
 713                                 $noscrape = str_replace("/profile/", "/noscrape/", $url);
 714                                 $noscrapejson = fetch_url($noscrape);
 715                                 if ($noscrapejson) {
 716
 717                                         $network = NETWORK_DFRN;
 718
 719                                         $poco = str_replace("/profile/", "/poco/", $url);
 720
 721                                         $noscrapedata = json_decode($noscrapejson, true);
 722
 723                                         if (isset($noscrapedata["addr"]))
 724                                                 $addr = $noscrapedata["addr"];
 725
 726                                         if (isset($noscrapedata["fn"]))
 727                                                 $vcard["fn"] = $noscrapedata["fn"];
 728
 729                                         if (isset($noscrapedata["key"]))
 730                                                 $pubkey = $noscrapedata["key"];
 731
 732                                         if (isset($noscrapedata["photo"]))
 733                                                 $vcard["photo"] = $noscrapedata["photo"];
 734
 735                                         if (isset($noscrapedata["dfrn-request"]))
 736                                                 $request = $noscrapedata["dfrn-request"];
 737
 738                                         if (isset($noscrapedata["dfrn-confirm"]))
 739                                                 $confirm = $noscrapedata["dfrn-confirm"];
 740
 741                                         if (isset($noscrapedata["dfrn-notify"]))
 742                                                 $notify = $noscrapedata["dfrn-notify"];
 743
 744                                         if (isset($noscrapedata["dfrn-poll"]))
 745                                                 $poll = $noscrapedata["dfrn-poll"];
 746
 747                                 }
 748                         }
 749
 750                         if(! $network)
 751                                 $network = NETWORK_FEED;
 752
 753                         if(! x($vcard,'nick')) {
 754                                 $vcard['nick'] = strtolower(notags(unxmlify($vcard['fn'])));
 755                                 if(strpos($vcard['nick'],' '))
 756                                         $vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' ')));
 757                         }
 758                         if(! $priority)
 759                                 $priority = 2;
 760                 }
 761         }
 762
 763         if(! x($vcard,'photo')) {
 764                 $a = get_app();
 765                 $vcard['photo'] = App::get_baseurl() . '/images/person-175.jpg' ;
 766         }
 767
 768         if(! $profile)
 769                 $profile = $url;
 770
 771         // No human could be associated with this link, use the URL as the contact name
 772
 773         if(($network === NETWORK_FEED) && ($poll) && (! x($vcard,'fn')))
 774                 $vcard['fn'] = $url;
 775
 776         if (($notify != "") AND ($poll != "")) {
 777                 $baseurl = matching_url(normalise_link($notify), normalise_link($poll));
 778
 779                 $baseurl2 = matching_url($baseurl, normalise_link($profile));
 780                 if ($baseurl2 != "")
 781                         $baseurl = $baseurl2;
 782         }
 783
 784         if (($baseurl == "") AND ($notify != ""))
 785                 $baseurl = matching_url(normalise_link($profile), normalise_link($notify));
 786
 787         if (($baseurl == "") AND ($poll != ""))
 788                 $baseurl = matching_url(normalise_link($profile), normalise_link($poll));
 789
 790         if (substr($baseurl, -10) == "/index.php")
 791                 $baseurl = str_replace("/index.php", "", $baseurl);
 792
 793         $baseurl = rtrim($baseurl, "/");
 794
 795         if(strpos($url,'@') AND ($addr == "") AND ($network == NETWORK_DFRN))
 796                 $addr = str_replace('acct:', '', $url);
 797
 798         $vcard['fn'] = notags($vcard['fn']);
 799         $vcard['nick'] = str_replace(' ','',notags($vcard['nick']));
 800
 801         $result['name'] = $vcard['fn'];
 802         $result['nick'] = $vcard['nick'];
 803         $result['url'] = $profile;
 804         $result['addr'] = $addr;
 805         $result['batch'] = $batch;
 806         $result['notify'] = $notify;
 807         $result['poll'] = $poll;
 808         $result['request'] = $request;
 809         $result['confirm'] = $confirm;
 810         $result['poco'] = $poco;
 811         $result['photo'] = $vcard['photo'];
 812         $result['priority'] = $priority;
 813         $result['network'] = $network;
 814         $result['alias'] = $alias;
 815         $result['pubkey'] = $pubkey;
 816         $result['baseurl'] = $baseurl;
 817
 818         logger('probe_url: ' . print_r($result,true), LOGGER_DEBUG);
 819
 820         if ($level == 1) {
 821                 // Trying if it maybe a diaspora account
 822                 if (($result['network'] == NETWORK_FEED) OR ($result['addr'] == "")) {
 823                         require_once('include/bbcode.php');
 824                         $address = GetProfileUsername($url, "", true);
 825                         $result2 = probe_url($address, $mode, ++$level);
 826                         if ($result2['network'] != "")
 827                                 $result = $result2;
 828                 }
 829
 830                 // Maybe it's some non standard GNU Social installation (Single user, subfolder or no uri rewrite)
 831                 if (($result['network'] == NETWORK_FEED) AND ($result['baseurl'] != "") AND ($result['nick'] != "")) {
 832                         $addr = $result['nick'].'@'.str_replace("http://", "", $result['baseurl']);
 833                         $result2 = probe_url($addr, $mode, ++$level);
 834                         if (($result2['network'] != "") AND ($result2['network'] != NETWORK_FEED))
 835                                 $result = $result2;
 836                 }
 837         }
 838
 839         // Only store into the cache if the value seems to be valid
 840         if ($result['network'] != NETWORK_PHANTOM) {
 841                 Cache::set("probe_url:".$mode.":".$original_url,serialize($result), CACHE_DAY);
 842
 843                 /// @todo temporary fix - we need a real contact update function that updates only changing fields
 844                 /// The biggest problem is the avatar picture that could have a reduced image size.
 845                 /// It should only be updated if the existing picture isn't existing anymore.
 846                 if (($result['network'] != NETWORK_FEED) AND $result["addr"] AND $result["name"] AND $result["nick"])
 847                         q("UPDATE `contact` SET `addr` = '%s', `alias` = '%s', `name` = '%s', `nick` = '%s',
 848                                 `name-date` = '%s', `uri-date` = '%s' WHERE `nurl` = '%s' AND NOT `self` AND `uid` = 0",
 849                                 dbesc($result["addr"]),
 850                                 dbesc($result["alias"]),
 851                                 dbesc($result["name"]),
 852                                 dbesc($result["nick"]),
 853                                 dbesc(datetime_convert()),
 854                                 dbesc(datetime_convert()),
 855                                 dbesc(normalise_link($result['url']))
 856                 );
 857         }
 858
 859         return $result;
 860 }
 861
 862 /**
 863  * @brief Find the matching part between two url
 864  *
 865  * @param string $url1
 866  * @param string $url2
 867  * @return string The matching part
 868  */
 869 function matching_url($url1, $url2) {
 870
 871         if (($url1 == "") OR ($url2 == ""))
 872                 return "";
 873
 874         $url1 = normalise_link($url1);
 875         $url2 = normalise_link($url2);
 876
 877         $parts1 = parse_url($url1);
 878         $parts2 = parse_url($url2);
 879
 880         if (!isset($parts1["host"]) OR !isset($parts2["host"]))
 881                 return "";
 882
 883         if ($parts1["scheme"] != $parts2["scheme"])
 884                 return "";
 885
 886         if ($parts1["host"] != $parts2["host"])
 887                 return "";
 888
 889         if ($parts1["port"] != $parts2["port"])
 890                 return "";
 891
 892         $match = $parts1["scheme"]."://".$parts1["host"];
 893
 894         if ($parts1["port"])
 895                 $match .= ":".$parts1["port"];
 896
 897         $pathparts1 = explode("/", $parts1["path"]);
 898         $pathparts2 = explode("/", $parts2["path"]);
 899
 900         $i = 0;
 901         $path = "";
 902         do {
 903                 $path1 = $pathparts1[$i];
 904                 $path2 = $pathparts2[$i];
 905
 906                 if ($path1 == $path2)
 907                         $path .= $path1."/";
 908
 909         } while (($path1 == $path2) AND ($i++ <= count($pathparts1)));
 910
 911         $match .= $path;
 912
 913         return normalise_link($match);
 914 }