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($noscrapedata);
  26                         else
  27                                 unset($noscrapedata["nick"]);
  28                 } else
  29                         $noscrapedata = array();
  30         }
  31
  32         $s = fetch_url($url);
  33
  34         if (!$s)
  35                 return $ret;
  36
  37         if (!$dont_probe) {
  38                 $probe = probe_url($url);
  39
  40                 if (isset($probe["addr"]))
  41                         $ret["addr"] = $probe["addr"];
  42         }
  43
  44         $headers = $a->get_curl_headers();
  45         logger('scrape_dfrn: headers=' . $headers, LOGGER_DEBUG);
  46
  47
  48         $lines = explode("\n",$headers);
  49         if(count($lines)) {
  50                 foreach($lines as $line) {
  51                         // don't try and run feeds through the html5 parser
  52                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
  53                                 return ret;
  54                 }
  55         }
  56
  57         try {
  58                 $dom = HTML5_Parser::parse($s);
  59         } catch (DOMException $e) {
  60                 logger('scrape_dfrn: parse error: ' . $e);
  61         }
  62
  63         if(! $dom)
  64                 return $ret;
  65
  66         $items = $dom->getElementsByTagName('link');
  67
  68         // get DFRN link elements
  69
  70         foreach($items as $item) {
  71                 $x = $item->getAttribute('rel');
  72                 if(($x === 'alternate') && ($item->getAttribute('type') === 'application/atom+xml'))
  73                         $ret['feed_atom'] = $item->getAttribute('href');
  74                 if(substr($x,0,5) == "dfrn-") {
  75                         $ret[$x] = $item->getAttribute('href');
  76                 }
  77                 if($x === 'lrdd') {
  78                         $decoded = urldecode($item->getAttribute('href'));
  79                         if(preg_match('/acct:([^@]*)@/',$decoded,$matches))
  80                                 $ret['nick'] = $matches[1];
  81                 }
  82         }
  83
  84         // Pull out hCard profile elements
  85
  86         $largest_photo = 0;
  87
  88         $items = $dom->getElementsByTagName('*');
  89         foreach($items as $item) {
  90                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
  91                         $level2 = $item->getElementsByTagName('*');
  92                         foreach($level2 as $x) {
  93                                 if(attribute_contains($x->getAttribute('class'),'fn')) {
  94                                         $ret['fn'] = $x->textContent;
  95                                 }
  96                                 if((attribute_contains($x->getAttribute('class'),'photo'))
  97                                         || (attribute_contains($x->getAttribute('class'),'avatar'))) {
  98                                         $size = intval($x->getAttribute('width'));
  99                                         // dfrn prefers 175, so if we find this, we set largest_size so it can't be topped.
 100                                         if(($size > $largest_photo) || ($size == 175) || (! $largest_photo)) {
 101                                                 $ret['photo'] = $x->getAttribute('src');
 102                                                 $largest_photo = (($size == 175) ? 9999 : $size);
 103                                         }
 104                                 }
 105                                 if(attribute_contains($x->getAttribute('class'),'key')) {
 106                                         $ret['key'] = $x->textContent;
 107                                 }
 108                         }
 109                 }
 110         }
 111         return array_merge($ret, $noscrapedata);
 112 }}
 113
 114
 115
 116
 117
 118
 119 if(! function_exists('validate_dfrn')) {
 120 function validate_dfrn($a) {
 121         $errors = 0;
 122         if(! x($a,'key'))
 123                 $errors ++;
 124         if(! x($a,'dfrn-request'))
 125                 $errors ++;
 126         if(! x($a,'dfrn-confirm'))
 127                 $errors ++;
 128         if(! x($a,'dfrn-notify'))
 129                 $errors ++;
 130         if(! x($a,'dfrn-poll'))
 131                 $errors ++;
 132         return $errors;
 133 }}
 134
 135 if(! function_exists('scrape_meta')) {
 136 function scrape_meta($url) {
 137
 138         $a = get_app();
 139
 140         $ret = array();
 141
 142         logger('scrape_meta: url=' . $url);
 143
 144         $s = fetch_url($url);
 145
 146         if(! $s)
 147                 return $ret;
 148
 149         $headers = $a->get_curl_headers();
 150         logger('scrape_meta: headers=' . $headers, LOGGER_DEBUG);
 151
 152         $lines = explode("\n",$headers);
 153         if(count($lines)) {
 154                 foreach($lines as $line) {
 155                         // don't try and run feeds through the html5 parser
 156                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
 157                                 return ret;
 158                 }
 159         }
 160
 161         try {
 162                 $dom = HTML5_Parser::parse($s);
 163         } catch (DOMException $e) {
 164                 logger('scrape_meta: parse error: ' . $e);
 165         }
 166
 167         if(! $dom)
 168                 return $ret;
 169
 170         $items = $dom->getElementsByTagName('meta');
 171
 172         // get DFRN link elements
 173
 174         foreach($items as $item) {
 175                 $x = $item->getAttribute('name');
 176                 if(substr($x,0,5) == "dfrn-")
 177                         $ret[$x] = $item->getAttribute('content');
 178         }
 179
 180         return $ret;
 181 }}
 182
 183
 184 if(! function_exists('scrape_vcard')) {
 185 function scrape_vcard($url) {
 186
 187         $a = get_app();
 188
 189         $ret = array();
 190
 191         logger('scrape_vcard: url=' . $url);
 192
 193         $s = fetch_url($url);
 194
 195         if(! $s)
 196                 return $ret;
 197
 198         $headers = $a->get_curl_headers();
 199         $lines = explode("\n",$headers);
 200         if(count($lines)) {
 201                 foreach($lines as $line) {
 202                         // don't try and run feeds through the html5 parser
 203                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
 204                                 return ret;
 205                 }
 206         }
 207
 208         try {
 209                 $dom = HTML5_Parser::parse($s);
 210         } catch (DOMException $e) {
 211                 logger('scrape_vcard: parse error: ' . $e);
 212         }
 213
 214         if(! $dom)
 215                 return $ret;
 216
 217         // Pull out hCard profile elements
 218
 219         $largest_photo = 0;
 220
 221         $items = $dom->getElementsByTagName('*');
 222         foreach($items as $item) {
 223                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
 224                         $level2 = $item->getElementsByTagName('*');
 225                         foreach($level2 as $x) {
 226                                 if(attribute_contains($x->getAttribute('class'),'fn'))
 227                                         $ret['fn'] = $x->textContent;
 228                                 if((attribute_contains($x->getAttribute('class'),'photo'))
 229                                         || (attribute_contains($x->getAttribute('class'),'avatar'))) {
 230                                         $size = intval($x->getAttribute('width'));
 231                                         if(($size > $largest_photo) || (! $largest_photo)) {
 232                                                 $ret['photo'] = $x->getAttribute('src');
 233                                                 $largest_photo = $size;
 234                                         }
 235                                 }
 236                                 if((attribute_contains($x->getAttribute('class'),'nickname'))
 237                                         || (attribute_contains($x->getAttribute('class'),'uid'))) {
 238                                         $ret['nick'] = $x->textContent;
 239                                 }
 240                         }
 241                 }
 242         }
 243
 244         return $ret;
 245 }}
 246
 247
 248 if(! function_exists('scrape_feed')) {
 249 function scrape_feed($url) {
 250
 251         $a = get_app();
 252
 253         $ret = array();
 254         $cookiejar = tempnam(get_temppath(), 'cookiejar-scrape-feed-');
 255         $s = fetch_url($url, false, $redirects, 0, Null, $cookiejar);
 256         unlink($cookiejar);
 257
 258         $headers = $a->get_curl_headers();
 259         $code = $a->get_curl_code();
 260
 261         logger('scrape_feed: returns: ' . $code . ' headers=' . $headers, LOGGER_DEBUG);
 262
 263         if(! $s) {
 264                 logger('scrape_feed: no data returned for ' . $url);
 265                 return $ret;
 266         }
 267
 268
 269         $lines = explode("\n",$headers);
 270         if(count($lines)) {
 271                 foreach($lines as $line) {
 272                         if(stristr($line,'content-type:')) {
 273                                 if(stristr($line,'application/atom+xml') || stristr($s,'<feed')) {
 274                                         $ret['feed_atom'] = $url;
 275                                         return $ret;
 276                                 }
 277                                 if(stristr($line,'application/rss+xml') || stristr($s,'<rss')) {
 278                                         $ret['feed_rss'] = $url;
 279                                         return $ret;
 280                                 }
 281                         }
 282                 }
 283                 // perhaps an RSS version 1 feed with a generic or incorrect content-type?
 284                 if(stristr($s,'</item>')) {
 285                         $ret['feed_rss'] = $url;
 286                         return $ret;
 287                 }
 288         }
 289
 290         $basename = implode('/', array_slice(explode('/',$url),0,3)) . '/';
 291
 292         $doc = new DOMDocument();
 293         @$doc->loadHTML($s);
 294         $xpath = new DomXPath($doc);
 295
 296         $base = $xpath->query("//base");
 297         foreach ($base as $node) {
 298                 $attr = array();
 299
 300                 if ($node->attributes->length)
 301                         foreach ($node->attributes as $attribute)
 302                                 $attr[$attribute->name] = $attribute->value;
 303
 304                 if ($attr["href"] != "")
 305                         $basename = $attr["href"] ;
 306         }
 307
 308         $list = $xpath->query("//link");
 309         foreach ($list as $node) {
 310                 $attr = array();
 311
 312                 if ($node->attributes->length)
 313                         foreach ($node->attributes as $attribute)
 314                                 $attr[$attribute->name] = $attribute->value;
 315
 316                 if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/atom+xml"))
 317                         $ret["feed_atom"] = $attr["href"];
 318
 319                 if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/rss+xml"))
 320                         $ret["feed_rss"] = $attr["href"];
 321         }
 322
 323         // Drupal and perhaps others only provide relative URLs. Turn them into absolute.
 324
 325         if(x($ret,'feed_atom') && (! strstr($ret['feed_atom'],'://')))
 326                 $ret['feed_atom'] = $basename . $ret['feed_atom'];
 327         if(x($ret,'feed_rss') && (! strstr($ret['feed_rss'],'://')))
 328                 $ret['feed_rss'] = $basename . $ret['feed_rss'];
 329
 330         return $ret;
 331 }}
 332
 333
 334 /**
 335  *
 336  * Probe a network address to discover what kind of protocols we need to communicate with it.
 337  *
 338  * Warning: this function is a bit touchy and there are some subtle dependencies within the logic flow.
 339  * Edit with care.
 340  *
 341  */
 342
 343 /**
 344  *
 345  * PROBE_DIASPORA has a bias towards returning Diaspora information
 346  * while PROBE_NORMAL has a bias towards dfrn/zot - in the case where
 347  * an address (such as a Friendica address) supports more than one type
 348  * of network.
 349  *
 350  */
 351
 352
 353 define ( 'PROBE_NORMAL',   0);
 354 define ( 'PROBE_DIASPORA', 1);
 355
 356 function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
 357         require_once('include/email.php');
 358
 359         $result = array();
 360
 361         if (!$url)
 362                 return $result;
 363
 364         $result = Cache::get("probe_url:".$mode.":".$url);
 365         if (!is_null($result)) {
 366                 $result = unserialize($result);
 367                 return $result;
 368         }
 369
 370         $original_url = $url;
 371         $network = null;
 372         $diaspora = false;
 373         $diaspora_base = '';
 374         $diaspora_guid = '';
 375         $diaspora_key = '';
 376         $has_lrdd = false;
 377         $email_conversant = false;
 378         $connectornetworks = false;
 379         $appnet = false;
 380
 381         if (strpos($url,'twitter.com')) {
 382                 $connectornetworks = true;
 383                 $network = NETWORK_TWITTER;
 384         }
 385
 386         $lastfm  = ((strpos($url,'last.fm/user') !== false) ? true : false);
 387
 388         $at_addr = ((strpos($url,'@') !== false) ? true : false);
 389
 390         if((!$appnet) && (!$lastfm) && !$connectornetworks) {
 391
 392                 if(strpos($url,'mailto:') !== false && $at_addr) {
 393                         $url = str_replace('mailto:','',$url);
 394                         $links = array();
 395                 }
 396                 else
 397                         $links = lrdd($url);
 398
 399                 if ((count($links) == 0) AND strstr($url, "/index.php")) {
 400                         $url = str_replace("/index.php", "", $url);
 401                         $links = lrdd($url);
 402                 }
 403
 404                 if (count($links)) {
 405                         $has_lrdd = true;
 406
 407                         logger('probe_url: found lrdd links: ' . print_r($links,true), LOGGER_DATA);
 408                         foreach($links as $link) {
 409                                 if($link['@attributes']['rel'] === NAMESPACE_ZOT)
 410                                         $zot = unamp($link['@attributes']['href']);
 411                                 if($link['@attributes']['rel'] === NAMESPACE_DFRN)
 412                                         $dfrn = unamp($link['@attributes']['href']);
 413                                 if($link['@attributes']['rel'] === 'salmon')
 414                                         $notify = unamp($link['@attributes']['href']);
 415                                 if($link['@attributes']['rel'] === NAMESPACE_FEED)
 416                                         $poll = unamp($link['@attributes']['href']);
 417                                 if($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard')
 418                                         $hcard = unamp($link['@attributes']['href']);
 419                                 if($link['@attributes']['rel'] === 'http://webfinger.net/rel/profile-page')
 420                                         $profile = unamp($link['@attributes']['href']);
 421                                 if($link['@attributes']['rel'] === 'http://portablecontacts.net/spec/1.0')
 422                                         $poco = unamp($link['@attributes']['href']);
 423                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/seed_location') {
 424                                         $diaspora_base = unamp($link['@attributes']['href']);
 425                                         $diaspora = true;
 426                                 }
 427                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/guid') {
 428                                         $diaspora_guid = unamp($link['@attributes']['href']);
 429                                         $diaspora = true;
 430                                 }
 431                                 if($link['@attributes']['rel'] === 'diaspora-public-key') {
 432                                         $diaspora_key = base64_decode(unamp($link['@attributes']['href']));
 433                                         if(strstr($diaspora_key,'RSA '))
 434                                                 $pubkey = rsatopem($diaspora_key);
 435                                         else
 436                                                 $pubkey = $diaspora_key;
 437                                         $diaspora = true;
 438                                 }
 439                                 if(($link['@attributes']['rel'] === 'http://ostatus.org/schema/1.0/subscribe') AND ($mode == PROBE_NORMAL)) {
 440                                         $diaspora = false;
 441                                 }
 442                         }
 443
 444                         // Status.Net can have more than one profile URL. We need to match the profile URL
 445                         // to a contact on incoming messages to prevent spam, and we won't know which one
 446                         // to match. So in case of two, one of them is stored as an alias. Only store URL's
 447                         // and not webfinger user@host aliases. If they've got more than two non-email style
 448                         // aliases, let's hope we're lucky and get one that matches the feed author-uri because
 449                         // otherwise we're screwed.
 450
 451                         $backup_alias = "";
 452
 453                         foreach($links as $link) {
 454                                 if($link['@attributes']['rel'] === 'alias') {
 455                                         if(strpos($link['@attributes']['href'],'@') === false) {
 456                                                 if(isset($profile)) {
 457                                                         $alias_url = $link['@attributes']['href'];
 458
 459                                                         if(($alias_url !== $profile) AND ($backup_alias == "") AND
 460                                                                 ($alias_url !== str_replace("/index.php", "", $profile)))
 461                                                                 $backup_alias = $alias_url;
 462
 463                                                         if(($alias_url !== $profile) AND !strstr($alias_url, "index.php") AND
 464                                                                 ($alias_url !== str_replace("/index.php", "", $profile)))
 465                                                                 $alias = $alias_url;
 466                                                 }
 467                                                 else
 468                                                         $profile = unamp($link['@attributes']['href']);
 469                                         }
 470                                 }
 471                         }
 472
 473                         if ($alias == "")
 474                                 $alias = $backup_alias;
 475
 476                         // If the profile is different from the url then the url is abviously an alias
 477                         if (($alias == "") AND ($profile != "") AND !$at_addr AND (normalise_link($profile) != normalise_link($url)))
 478                                 $alias = $url;
 479                 }
 480                 elseif($mode == PROBE_NORMAL) {
 481
 482                         // Check email
 483
 484                         $orig_url = $url;
 485                         if((strpos($orig_url,'@')) && validate_email($orig_url)) {
 486                                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1",
 487                                         intval(local_user())
 488                                 );
 489                                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1",
 490                                         intval(local_user())
 491                                 );
 492                                 if(count($x) && count($r)) {
 493                                         $mailbox = construct_mailbox_name($r[0]);
 494                                         $password = '';
 495                                         openssl_private_decrypt(hex2bin($r[0]['pass']),$password,$x[0]['prvkey']);
 496                                         $mbox = email_connect($mailbox,$r[0]['user'],$password);
 497                                         if(! $mbox)
 498                                                 logger('probe_url: email_connect failed.');
 499                                         unset($password);
 500                                 }
 501                                 if($mbox) {
 502                                         $msgs = email_poll($mbox,$orig_url);
 503                                         logger('probe_url: searching ' . $orig_url . ', ' . count($msgs) . ' messages found.', LOGGER_DEBUG);
 504                                         if(count($msgs)) {
 505                                                 $addr = $orig_url;
 506                                                 $network = NETWORK_MAIL;
 507                                                 $name = substr($url,0,strpos($url,'@'));
 508                                                 $phost = substr($url,strpos($url,'@')+1);
 509                                                 $profile = 'http://' . $phost;
 510                                                 // fix nick character range
 511                                                 $vcard = array('fn' => $name, 'nick' => $name, 'photo' => avatar_img($url));
 512                                                 $notify = 'smtp ' . random_string();
 513                                                 $poll = 'email ' . random_string();
 514                                                 $priority = 0;
 515                                                 $x = email_msg_meta($mbox,$msgs[0]);
 516                                                 if(stristr($x[0]->from,$orig_url))
 517                                                         $adr = imap_rfc822_parse_adrlist($x[0]->from,'');
 518                                                 elseif(stristr($x[0]->to,$orig_url))
 519                                                         $adr = imap_rfc822_parse_adrlist($x[0]->to,'');
 520                                                 if(isset($adr)) {
 521                                                         foreach($adr as $feadr) {
 522                                                                 if((strcasecmp($feadr->mailbox,$name) == 0)
 523                                                                         &&(strcasecmp($feadr->host,$phost) == 0)
 524                                                                         && (strlen($feadr->personal))) {
 525
 526                                                                         $personal = imap_mime_header_decode($feadr->personal);
 527                                                                         $vcard['fn'] = "";
 528                                                                         foreach($personal as $perspart)
 529                                                                                 if ($perspart->charset != "default")
 530                                                                                         $vcard['fn'] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
 531                                                                                 else
 532                                                                                         $vcard['fn'] .= $perspart->text;
 533
 534                                                                         $vcard['fn'] = notags($vcard['fn']);
 535                                                                 }
 536                                                         }
 537                                                 }
 538                                         }
 539                                         imap_close($mbox);
 540                                 }
 541                         }
 542                 }
 543         }
 544
 545         if($mode == PROBE_NORMAL) {
 546
 547                 if(strlen($zot)) {
 548                         $s = fetch_url($zot);
 549                         if($s) {
 550                                 $j = json_decode($s);
 551                                 if($j) {
 552                                         $network = NETWORK_ZOT;
 553                                         $vcard   = array(
 554                                                 'fn'    => $j->fullname,
 555                                                 'nick'  => $j->nickname,
 556                                                 'photo' => $j->photo
 557                                         );
 558                                         $profile  = $j->url;
 559                                         $notify   = $j->post;
 560                                         $pubkey   = $j->pubkey;
 561                                         $poll     = 'N/A';
 562                                 }
 563                         }
 564                 }
 565
 566
 567                 if(strlen($dfrn)) {
 568                         $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn, true);
 569                         if(is_array($ret) && x($ret,'dfrn-request')) {
 570                                 $network = NETWORK_DFRN;
 571                                 $request = $ret['dfrn-request'];
 572                                 $confirm = $ret['dfrn-confirm'];
 573                                 $notify  = $ret['dfrn-notify'];
 574                                 $poll    = $ret['dfrn-poll'];
 575
 576                                 $vcard = array();
 577                                 $vcard['fn'] = $ret['fn'];
 578                                 $vcard['nick'] = $ret['nick'];
 579                                 $vcard['photo'] = $ret['photo'];
 580                         }
 581                 }
 582         }
 583
 584         // Scrape the public key from the hcard.
 585         // Diaspora will remove it from the webfinger somewhere in the future.
 586         if (($hcard != "") AND ($pubkey == "")) {
 587                 $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn, true);
 588                 if (isset($ret["key"])) {
 589                         $hcard_key = $ret["key"];
 590                         if(strstr($hcard_key,'RSA '))
 591                                 $pubkey = rsatopem($hcard_key);
 592                         else
 593                                 $pubkey = $hcard_key;
 594                 }
 595         }
 596         if($diaspora && $diaspora_base && $diaspora_guid) {
 597                 $diaspora_notify = $diaspora_base.'receive/users/'.$diaspora_guid;
 598
 599                 if($mode == PROBE_DIASPORA || ! $notify || ($notify == $diaspora_notify)) {
 600                         $notify = $diaspora_notify;
 601                         $batch  = $diaspora_base . 'receive/public' ;
 602                 }
 603                 if(strpos($url,'@'))
 604                         $addr = str_replace('acct:', '', $url);
 605         }
 606
 607         if($network !== NETWORK_ZOT && $network !== NETWORK_DFRN && $network !== NETWORK_MAIL) {
 608                 if($diaspora)
 609                         $network = NETWORK_DIASPORA;
 610                 elseif($has_lrdd AND ($notify))
 611                         $network  = NETWORK_OSTATUS;
 612
 613                 if(strpos($url,'@'))
 614                         $addr = str_replace('acct:', '', $url);
 615
 616                 $priority = 0;
 617
 618                 if($hcard && ! $vcard) {
 619                         $vcard = scrape_vcard($hcard);
 620
 621                         // Google doesn't use absolute url in profile photos
 622
 623                         if((x($vcard,'photo')) && substr($vcard['photo'],0,1) == '/') {
 624                                 $h = @parse_url($hcard);
 625                                 if($h)
 626                                         $vcard['photo'] = $h['scheme'] . '://' . $h['host'] . $vcard['photo'];
 627                         }
 628
 629                         logger('probe_url: scrape_vcard: ' . print_r($vcard,true), LOGGER_DATA);
 630                 }
 631
 632                 if($diaspora && $addr) {
 633                         // Diaspora returns the name as the nick. As the nick will never be updated,
 634                         // let's use the Diaspora nickname (the first part of the handle) as the nick instead
 635                         $addr_parts = explode('@', $addr);
 636                         $vcard['nick'] = $addr_parts[0];
 637                 }
 638
 639                 if($lastfm) {
 640                         $profile = $url;
 641                         $poll = str_replace(array('www.','last.fm/'),array('','ws.audioscrobbler.com/1.0/'),$url) . '/recenttracks.rss';
 642                         $vcard['nick'] = basename($url);
 643                         $vcard['fn'] = $vcard['nick'] . t(' on Last.fm');
 644                         $network = NETWORK_FEED;
 645                 }
 646
 647                 if(! x($vcard,'fn'))
 648                         if(x($vcard,'nick'))
 649                                 $vcard['fn'] = $vcard['nick'];
 650
 651                 $check_feed = false;
 652
 653                 if(stristr($url,'tumblr.com') && (! stristr($url,'/rss'))) {
 654                         $poll = $url . '/rss';
 655                         $check_feed = true;
 656                         // Will leave it to others to figure out how to grab the avatar, which is on the $url page in the open graph meta links
 657                 }
 658
 659                 if($appnet || ! $poll)
 660                         $check_feed = true;
 661                 if((! isset($vcard)) || (! x($vcard,'fn')) || (! $profile))
 662                         $check_feed = true;
 663                 if(($at_addr) && (! count($links)))
 664                         $check_feed = false;
 665
 666                 if ($connectornetworks)
 667                         $check_feed = false;
 668
 669                 if($check_feed) {
 670
 671                         $feedret = scrape_feed(($poll) ? $poll : $url);
 672
 673                         logger('probe_url: scrape_feed ' . (($poll)? $poll : $url) . ' returns: ' . print_r($feedret,true), LOGGER_DATA);
 674                         if(count($feedret) && ($feedret['feed_atom'] || $feedret['feed_rss'])) {
 675                                 $poll = ((x($feedret,'feed_atom')) ? unamp($feedret['feed_atom']) : unamp($feedret['feed_rss']));
 676                                 if(! x($vcard))
 677                                         $vcard = array();
 678                         }
 679
 680                         if(x($feedret,'photo') && (! x($vcard,'photo')))
 681                                 $vcard['photo'] = $feedret['photo'];
 682
 683                         $cookiejar = tempnam(get_temppath(), 'cookiejar-scrape-feed-');
 684                         $xml = fetch_url($poll, false, $redirects, 0, Null, $cookiejar);
 685                         unlink($cookiejar);
 686
 687                         logger('probe_url: fetch feed: ' . $poll . ' returns: ' . $xml, LOGGER_DATA);
 688
 689                         if ($xml == "") {
 690                                 logger("scrape_feed: XML is empty for feed ".$poll);
 691                                 $network = NETWORK_PHANTOM;
 692                         } else {
 693                                 $data = feed_import($xml,$dummy1,$dummy2, $dummy3, true);
 694
 695                                 if (!is_array($data)) {
 696                                         logger("scrape_feed: This doesn't seem to be a feed: ".$poll);
 697                                         $network = NETWORK_PHANTOM;
 698                                 } else {
 699                                         if (($vcard["photo"] == "") AND ($data["header"]["author-avatar"] != ""))
 700                                                 $vcard["photo"] = $data["header"]["author-avatar"];
 701
 702                                         if (($vcard["fn"] == "") AND ($data["header"]["author-name"] != ""))
 703                                                 $vcard["fn"] = $data["header"]["author-name"];
 704
 705                                         if (($vcard["nick"] == "") AND ($data["header"]["author-nick"] != ""))
 706                                                 $vcard["nick"] = $data["header"]["author-nick"];
 707
 708                                         if ($network == NETWORK_OSTATUS) {
 709                                                 if ($data["header"]["author-id"] != "")
 710                                                         $alias = $data["header"]["author-id"];
 711
 712                                                 if ($data["header"]["author-link"] != "")
 713                                                         $profile = $data["header"]["author-link"];
 714
 715                                         } elseif(!$profile AND ($data["header"]["author-link"] != "") AND !in_array($network, array("", NETWORK_FEED)))
 716                                                 $profile = $data["header"]["author-link"];
 717                                 }
 718                         }
 719
 720                         // Workaround for misconfigured Friendica servers
 721                         if (($network == "") AND (strstr($url, "/profile/"))) {
 722                                 $noscrape = str_replace("/profile/", "/noscrape/", $url);
 723                                 $noscrapejson = fetch_url($noscrape);
 724                                 if ($noscrapejson) {
 725
 726                                         $network = NETWORK_DFRN;
 727
 728                                         $poco = str_replace("/profile/", "/poco/", $url);
 729
 730                                         $noscrapedata = json_decode($noscrapejson, true);
 731
 732                                         if (isset($noscrapedata["addr"]))
 733                                                 $addr = $noscrapedata["addr"];
 734
 735                                         if (isset($noscrapedata["fn"]))
 736                                                 $vcard["fn"] = $noscrapedata["fn"];
 737
 738                                         if (isset($noscrapedata["key"]))
 739                                                 $pubkey = $noscrapedata["key"];
 740
 741                                         if (isset($noscrapedata["photo"]))
 742                                                 $vcard["photo"] = $noscrapedata["photo"];
 743
 744                                         if (isset($noscrapedata["dfrn-request"]))
 745                                                 $request = $noscrapedata["dfrn-request"];
 746
 747                                         if (isset($noscrapedata["dfrn-confirm"]))
 748                                                 $confirm = $noscrapedata["dfrn-confirm"];
 749
 750                                         if (isset($noscrapedata["dfrn-notify"]))
 751                                                 $notify = $noscrapedata["dfrn-notify"];
 752
 753                                         if (isset($noscrapedata["dfrn-poll"]))
 754                                                 $poll = $noscrapedata["dfrn-poll"];
 755
 756                                 }
 757                         }
 758
 759                         if(! $network)
 760                                 $network = NETWORK_FEED;
 761
 762                         if(! x($vcard,'nick')) {
 763                                 $vcard['nick'] = strtolower(notags(unxmlify($vcard['fn'])));
 764                                 if(strpos($vcard['nick'],' '))
 765                                         $vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' ')));
 766                         }
 767                         if(! $priority)
 768                                 $priority = 2;
 769                 }
 770         }
 771
 772         if(! x($vcard,'photo')) {
 773                 $a = get_app();
 774                 $vcard['photo'] = App::get_baseurl() . '/images/person-175.jpg' ;
 775         }
 776
 777         if(! $profile)
 778                 $profile = $url;
 779
 780         // No human could be associated with this link, use the URL as the contact name
 781
 782         if(($network === NETWORK_FEED) && ($poll) && (! x($vcard,'fn')))
 783                 $vcard['fn'] = $url;
 784
 785         if (($notify != "") AND ($poll != "")) {
 786                 $baseurl = matching_url(normalise_link($notify), normalise_link($poll));
 787
 788                 $baseurl2 = matching_url($baseurl, normalise_link($profile));
 789                 if ($baseurl2 != "")
 790                         $baseurl = $baseurl2;
 791         }
 792
 793         if (($baseurl == "") AND ($notify != ""))
 794                 $baseurl = matching_url(normalise_link($profile), normalise_link($notify));
 795
 796         if (($baseurl == "") AND ($poll != ""))
 797                 $baseurl = matching_url(normalise_link($profile), normalise_link($poll));
 798
 799         if (substr($baseurl, -10) == "/index.php")
 800                 $baseurl = str_replace("/index.php", "", $baseurl);
 801
 802         $baseurl = rtrim($baseurl, "/");
 803
 804         if(strpos($url,'@') AND ($addr == "") AND ($network == NETWORK_DFRN))
 805                 $addr = str_replace('acct:', '', $url);
 806
 807         $vcard['fn'] = notags($vcard['fn']);
 808         $vcard['nick'] = str_replace(' ','',notags($vcard['nick']));
 809
 810         $result['name'] = $vcard['fn'];
 811         $result['nick'] = $vcard['nick'];
 812         $result['url'] = $profile;
 813         $result['addr'] = $addr;
 814         $result['batch'] = $batch;
 815         $result['notify'] = $notify;
 816         $result['poll'] = $poll;
 817         $result['request'] = $request;
 818         $result['confirm'] = $confirm;
 819         $result['poco'] = $poco;
 820         $result['photo'] = $vcard['photo'];
 821         $result['priority'] = $priority;
 822         $result['network'] = $network;
 823         $result['alias'] = $alias;
 824         $result['pubkey'] = $pubkey;
 825         $result['baseurl'] = $baseurl;
 826
 827         logger('probe_url: ' . print_r($result,true), LOGGER_DEBUG);
 828
 829         if ($level == 1) {
 830                 // Trying if it maybe a diaspora account
 831                 if (($result['network'] == NETWORK_FEED) OR ($result['addr'] == "")) {
 832                         require_once('include/bbcode.php');
 833                         $address = GetProfileUsername($url, "", true);
 834                         $result2 = probe_url($address, $mode, ++$level);
 835                         if ($result2['network'] != "")
 836                                 $result = $result2;
 837                 }
 838
 839                 // Maybe it's some non standard GNU Social installation (Single user, subfolder or no uri rewrite)
 840                 if (($result['network'] == NETWORK_FEED) AND ($result['baseurl'] != "") AND ($result['nick'] != "")) {
 841                         $addr = $result['nick'].'@'.str_replace("http://", "", $result['baseurl']);
 842                         $result2 = probe_url($addr, $mode, ++$level);
 843                         if (($result2['network'] != "") AND ($result2['network'] != NETWORK_FEED))
 844                                 $result = $result2;
 845                 }
 846         }
 847
 848         // Only store into the cache if the value seems to be valid
 849         if (!in_array($result['network'], array(NETWORK_PHANTOM, ""))) {
 850                 Cache::set("probe_url:".$mode.":".$original_url,serialize($result), CACHE_DAY);
 851
 852                 /// @todo temporary fix - we need a real contact update function that updates only changing fields
 853                 /// The biggest problem is the avatar picture that could have a reduced image size.
 854                 /// It should only be updated if the existing picture isn't existing anymore.
 855                 if (($result['network'] != NETWORK_FEED) AND ($mode == PROBE_NORMAL) AND
 856                         $result["name"] AND $result["nick"] AND $result["url"] AND $result["addr"] AND $result["poll"])
 857                         q("UPDATE `contact` SET `name` = '%s', `nick` = '%s', `url` = '%s', `addr` = '%s',
 858                                         `notify` = '%s', `poll` = '%s', `alias` = '%s', `success_update` = '%s'
 859                                 WHERE `nurl` = '%s' AND NOT `self` AND `uid` = 0",
 860                                 dbesc($result["name"]),
 861                                 dbesc($result["nick"]),
 862                                 dbesc($result["url"]),
 863                                 dbesc($result["addr"]),
 864                                 dbesc($result["notify"]),
 865                                 dbesc($result["poll"]),
 866                                 dbesc($result["alias"]),
 867                                 dbesc(datetime_convert()),
 868                                 dbesc(normalise_link($result['url']))
 869                 );
 870         }
 871
 872         return $result;
 873 }
 874
 875 /**
 876  * @brief Find the matching part between two url
 877  *
 878  * @param string $url1
 879  * @param string $url2
 880  * @return string The matching part
 881  */
 882 function matching_url($url1, $url2) {
 883
 884         if (($url1 == "") OR ($url2 == ""))
 885                 return "";
 886
 887         $url1 = normalise_link($url1);
 888         $url2 = normalise_link($url2);
 889
 890         $parts1 = parse_url($url1);
 891         $parts2 = parse_url($url2);
 892
 893         if (!isset($parts1["host"]) OR !isset($parts2["host"]))
 894                 return "";
 895
 896         if ($parts1["scheme"] != $parts2["scheme"])
 897                 return "";
 898
 899         if ($parts1["host"] != $parts2["host"])
 900                 return "";
 901
 902         if ($parts1["port"] != $parts2["port"])
 903                 return "";
 904
 905         $match = $parts1["scheme"]."://".$parts1["host"];
 906
 907         if ($parts1["port"])
 908                 $match .= ":".$parts1["port"];
 909
 910         $pathparts1 = explode("/", $parts1["path"]);
 911         $pathparts2 = explode("/", $parts2["path"]);
 912
 913         $i = 0;
 914         $path = "";
 915         do {
 916                 $path1 = $pathparts1[$i];
 917                 $path2 = $pathparts2[$i];
 918
 919                 if ($path1 == $path2)
 920                         $path .= $path1."/";
 921
 922         } while (($path1 == $path2) AND ($i++ <= count($pathparts1)));
 923
 924         $match .= $path;
 925
 926         return normalise_link($match);
 927 }