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'),'uid'))
  94                                         $ret['guid'] = $x->textContent;
  95                                 if(attribute_contains($x->getAttribute('class'),'nickname'))
  96                                         $ret['nickname'] = $x->textContent;
  97                                 if(attribute_contains($x->getAttribute('class'),'fn'))
  98                                         $ret['fn'] = $x->textContent;
  99                                 if(attribute_contains($x->getAttribute('class'),'searchable'))
 100                                         $ret['searchable'] = $x->textContent;
 101                                 if(attribute_contains($x->getAttribute('class'),'key'))
 102                                         $ret['key'] = $x->textContent;
 103                                 if(attribute_contains($x->getAttribute('class'),'url'))
 104                                         $ret['url'] = $x->textContent;
 105                                 if((attribute_contains($x->getAttribute('class'),'photo'))
 106                                         || (attribute_contains($x->getAttribute('class'),'avatar'))) {
 107                                         $size = intval($x->getAttribute('width'));
 108                                         // dfrn prefers 175, so if we find this, we set largest_size so it can't be topped.
 109                                         if(($size > $largest_photo) || ($size == 175) || (! $largest_photo)) {
 110                                                 $ret['photo'] = $x->getAttribute('src');
 111                                                 $largest_photo = (($size == 175) ? 9999 : $size);
 112                                         }
 113                                 }
 114                         }
 115                 }
 116         }
 117         return array_merge($ret, $noscrapedata);
 118 }}
 119
 120
 121
 122
 123
 124
 125 if(! function_exists('validate_dfrn')) {
 126 function validate_dfrn($a) {
 127         $errors = 0;
 128         if(! x($a,'key'))
 129                 $errors ++;
 130         if(! x($a,'dfrn-request'))
 131                 $errors ++;
 132         if(! x($a,'dfrn-confirm'))
 133                 $errors ++;
 134         if(! x($a,'dfrn-notify'))
 135                 $errors ++;
 136         if(! x($a,'dfrn-poll'))
 137                 $errors ++;
 138         return $errors;
 139 }}
 140
 141 if(! function_exists('scrape_meta')) {
 142 function scrape_meta($url) {
 143
 144         $a = get_app();
 145
 146         $ret = array();
 147
 148         logger('scrape_meta: url=' . $url);
 149
 150         $s = fetch_url($url);
 151
 152         if(! $s)
 153                 return $ret;
 154
 155         $headers = $a->get_curl_headers();
 156         logger('scrape_meta: headers=' . $headers, LOGGER_DEBUG);
 157
 158         $lines = explode("\n",$headers);
 159         if(count($lines)) {
 160                 foreach($lines as $line) {
 161                         // don't try and run feeds through the html5 parser
 162                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
 163                                 return ret;
 164                 }
 165         }
 166
 167         try {
 168                 $dom = HTML5_Parser::parse($s);
 169         } catch (DOMException $e) {
 170                 logger('scrape_meta: parse error: ' . $e);
 171         }
 172
 173         if(! $dom)
 174                 return $ret;
 175
 176         $items = $dom->getElementsByTagName('meta');
 177
 178         // get DFRN link elements
 179
 180         foreach($items as $item) {
 181                 $x = $item->getAttribute('name');
 182                 if(substr($x,0,5) == "dfrn-")
 183                         $ret[$x] = $item->getAttribute('content');
 184         }
 185
 186         return $ret;
 187 }}
 188
 189
 190 if(! function_exists('scrape_vcard')) {
 191 function scrape_vcard($url) {
 192
 193         $a = get_app();
 194
 195         $ret = array();
 196
 197         logger('scrape_vcard: url=' . $url);
 198
 199         $s = fetch_url($url);
 200
 201         if(! $s)
 202                 return $ret;
 203
 204         $headers = $a->get_curl_headers();
 205         $lines = explode("\n",$headers);
 206         if(count($lines)) {
 207                 foreach($lines as $line) {
 208                         // don't try and run feeds through the html5 parser
 209                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
 210                                 return ret;
 211                 }
 212         }
 213
 214         try {
 215                 $dom = HTML5_Parser::parse($s);
 216         } catch (DOMException $e) {
 217                 logger('scrape_vcard: parse error: ' . $e);
 218         }
 219
 220         if(! $dom)
 221                 return $ret;
 222
 223         // Pull out hCard profile elements
 224
 225         $largest_photo = 0;
 226
 227         $items = $dom->getElementsByTagName('*');
 228         foreach($items as $item) {
 229                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
 230                         $level2 = $item->getElementsByTagName('*');
 231                         foreach($level2 as $x) {
 232                                 if(attribute_contains($x->getAttribute('class'),'fn'))
 233                                         $ret['fn'] = $x->textContent;
 234                                 if((attribute_contains($x->getAttribute('class'),'photo'))
 235                                         || (attribute_contains($x->getAttribute('class'),'avatar'))) {
 236                                         $size = intval($x->getAttribute('width'));
 237                                         if(($size > $largest_photo) || (! $largest_photo)) {
 238                                                 $ret['photo'] = $x->getAttribute('src');
 239                                                 $largest_photo = $size;
 240                                         }
 241                                 }
 242                                 if((attribute_contains($x->getAttribute('class'),'nickname'))
 243                                         || (attribute_contains($x->getAttribute('class'),'uid'))) {
 244                                         $ret['nick'] = $x->textContent;
 245                                 }
 246                         }
 247                 }
 248         }
 249
 250         return $ret;
 251 }}
 252
 253
 254 if(! function_exists('scrape_feed')) {
 255 function scrape_feed($url) {
 256
 257         $a = get_app();
 258
 259         $ret = array();
 260         $cookiejar = tempnam(get_temppath(), 'cookiejar-scrape-feed-');
 261         $s = fetch_url($url, false, $redirects, 0, Null, $cookiejar);
 262         unlink($cookiejar);
 263
 264         $headers = $a->get_curl_headers();
 265         $code = $a->get_curl_code();
 266
 267         logger('scrape_feed: returns: ' . $code . ' headers=' . $headers, LOGGER_DEBUG);
 268
 269         if(! $s) {
 270                 logger('scrape_feed: no data returned for ' . $url);
 271                 return $ret;
 272         }
 273
 274
 275         $lines = explode("\n",$headers);
 276         if(count($lines)) {
 277                 foreach($lines as $line) {
 278                         if(stristr($line,'content-type:')) {
 279                                 if(stristr($line,'application/atom+xml') || stristr($s,'<feed')) {
 280                                         $ret['feed_atom'] = $url;
 281                                         return $ret;
 282                                 }
 283                                 if(stristr($line,'application/rss+xml') || stristr($s,'<rss')) {
 284                                         $ret['feed_rss'] = $url;
 285                                         return $ret;
 286                                 }
 287                         }
 288                 }
 289                 // perhaps an RSS version 1 feed with a generic or incorrect content-type?
 290                 if(stristr($s,'</item>')) {
 291                         $ret['feed_rss'] = $url;
 292                         return $ret;
 293                 }
 294         }
 295
 296         $basename = implode('/', array_slice(explode('/',$url),0,3)) . '/';
 297
 298         $doc = new DOMDocument();
 299         @$doc->loadHTML($s);
 300         $xpath = new DomXPath($doc);
 301
 302         $base = $xpath->query("//base");
 303         foreach ($base as $node) {
 304                 $attr = array();
 305
 306                 if ($node->attributes->length)
 307                         foreach ($node->attributes as $attribute)
 308                                 $attr[$attribute->name] = $attribute->value;
 309
 310                 if ($attr["href"] != "")
 311                         $basename = $attr["href"] ;
 312         }
 313
 314         $list = $xpath->query("//link");
 315         foreach ($list as $node) {
 316                 $attr = array();
 317
 318                 if ($node->attributes->length)
 319                         foreach ($node->attributes as $attribute)
 320                                 $attr[$attribute->name] = $attribute->value;
 321
 322                 if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/atom+xml"))
 323                         $ret["feed_atom"] = $attr["href"];
 324
 325                 if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/rss+xml"))
 326                         $ret["feed_rss"] = $attr["href"];
 327         }
 328
 329         // Drupal and perhaps others only provide relative URLs. Turn them into absolute.
 330
 331         if(x($ret,'feed_atom') && (! strstr($ret['feed_atom'],'://')))
 332                 $ret['feed_atom'] = $basename . $ret['feed_atom'];
 333         if(x($ret,'feed_rss') && (! strstr($ret['feed_rss'],'://')))
 334                 $ret['feed_rss'] = $basename . $ret['feed_rss'];
 335
 336         return $ret;
 337 }}
 338
 339
 340 /**
 341  *
 342  * Probe a network address to discover what kind of protocols we need to communicate with it.
 343  *
 344  * Warning: this function is a bit touchy and there are some subtle dependencies within the logic flow.
 345  * Edit with care.
 346  *
 347  */
 348
 349 /**
 350  *
 351  * PROBE_DIASPORA has a bias towards returning Diaspora information
 352  * while PROBE_NORMAL has a bias towards dfrn/zot - in the case where
 353  * an address (such as a Friendica address) supports more than one type
 354  * of network.
 355  *
 356  */
 357
 358
 359 define ( 'PROBE_NORMAL',   0);
 360 define ( 'PROBE_DIASPORA', 1);
 361
 362 function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
 363         require_once('include/email.php');
 364
 365         $result = array();
 366
 367         if (!$url)
 368                 return $result;
 369
 370         $result = Cache::get("probe_url:".$mode.":".$url);
 371         if (!is_null($result)) {
 372                 $result = unserialize($result);
 373                 return $result;
 374         }
 375
 376         $original_url = $url;
 377         $network = null;
 378         $diaspora = false;
 379         $diaspora_base = '';
 380         $diaspora_guid = '';
 381         $diaspora_key = '';
 382         $has_lrdd = false;
 383         $email_conversant = false;
 384         $connectornetworks = false;
 385         $appnet = false;
 386
 387         if (strpos($url,'twitter.com')) {
 388                 $connectornetworks = true;
 389                 $network = NETWORK_TWITTER;
 390         }
 391
 392         $lastfm  = ((strpos($url,'last.fm/user') !== false) ? true : false);
 393
 394         $at_addr = ((strpos($url,'@') !== false) ? true : false);
 395
 396         if((!$appnet) && (!$lastfm) && !$connectornetworks) {
 397
 398                 if(strpos($url,'mailto:') !== false && $at_addr) {
 399                         $url = str_replace('mailto:','',$url);
 400                         $links = array();
 401                 }
 402                 else
 403                         $links = lrdd($url);
 404
 405                 if ((count($links) == 0) AND strstr($url, "/index.php")) {
 406                         $url = str_replace("/index.php", "", $url);
 407                         $links = lrdd($url);
 408                 }
 409
 410                 if (count($links)) {
 411                         $has_lrdd = true;
 412
 413                         logger('probe_url: found lrdd links: ' . print_r($links,true), LOGGER_DATA);
 414                         foreach($links as $link) {
 415                                 if($link['@attributes']['rel'] === NAMESPACE_ZOT)
 416                                         $zot = unamp($link['@attributes']['href']);
 417                                 if($link['@attributes']['rel'] === NAMESPACE_DFRN)
 418                                         $dfrn = unamp($link['@attributes']['href']);
 419                                 if($link['@attributes']['rel'] === 'salmon')
 420                                         $notify = unamp($link['@attributes']['href']);
 421                                 if($link['@attributes']['rel'] === NAMESPACE_FEED)
 422                                         $poll = unamp($link['@attributes']['href']);
 423                                 if($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard')
 424                                         $hcard = unamp($link['@attributes']['href']);
 425                                 if($link['@attributes']['rel'] === 'http://webfinger.net/rel/profile-page')
 426                                         $profile = unamp($link['@attributes']['href']);
 427                                 if($link['@attributes']['rel'] === 'http://portablecontacts.net/spec/1.0')
 428                                         $poco = unamp($link['@attributes']['href']);
 429                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/seed_location') {
 430                                         $diaspora_base = unamp($link['@attributes']['href']);
 431                                         $diaspora = true;
 432                                 }
 433                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/guid') {
 434                                         $diaspora_guid = unamp($link['@attributes']['href']);
 435                                         $diaspora = true;
 436                                 }
 437                                 if($link['@attributes']['rel'] === 'diaspora-public-key') {
 438                                         $diaspora_key = base64_decode(unamp($link['@attributes']['href']));
 439                                         if(strstr($diaspora_key,'RSA '))
 440                                                 $pubkey = rsatopem($diaspora_key);
 441                                         else
 442                                                 $pubkey = $diaspora_key;
 443                                         $diaspora = true;
 444                                 }
 445                                 if(($link['@attributes']['rel'] === 'http://ostatus.org/schema/1.0/subscribe') AND ($mode == PROBE_NORMAL)) {
 446                                         $diaspora = false;
 447                                 }
 448                         }
 449
 450                         // Status.Net can have more than one profile URL. We need to match the profile URL
 451                         // to a contact on incoming messages to prevent spam, and we won't know which one
 452                         // to match. So in case of two, one of them is stored as an alias. Only store URL's
 453                         // and not webfinger user@host aliases. If they've got more than two non-email style
 454                         // aliases, let's hope we're lucky and get one that matches the feed author-uri because
 455                         // otherwise we're screwed.
 456
 457                         $backup_alias = "";
 458
 459                         foreach($links as $link) {
 460                                 if($link['@attributes']['rel'] === 'alias') {
 461                                         if(strpos($link['@attributes']['href'],'@') === false) {
 462                                                 if(isset($profile)) {
 463                                                         $alias_url = $link['@attributes']['href'];
 464
 465                                                         if(($alias_url !== $profile) AND ($backup_alias == "") AND
 466                                                                 ($alias_url !== str_replace("/index.php", "", $profile)))
 467                                                                 $backup_alias = $alias_url;
 468
 469                                                         if(($alias_url !== $profile) AND !strstr($alias_url, "index.php") AND
 470                                                                 ($alias_url !== str_replace("/index.php", "", $profile)))
 471                                                                 $alias = $alias_url;
 472                                                 }
 473                                                 else
 474                                                         $profile = unamp($link['@attributes']['href']);
 475                                         }
 476                                 }
 477                         }
 478
 479                         if ($alias == "")
 480                                 $alias = $backup_alias;
 481
 482                         // If the profile is different from the url then the url is abviously an alias
 483                         if (($alias == "") AND ($profile != "") AND !$at_addr AND (normalise_link($profile) != normalise_link($url)))
 484                                 $alias = $url;
 485                 }
 486                 elseif($mode == PROBE_NORMAL) {
 487
 488                         // Check email
 489
 490                         $orig_url = $url;
 491                         if((strpos($orig_url,'@')) && validate_email($orig_url)) {
 492                                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1",
 493                                         intval(local_user())
 494                                 );
 495                                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1",
 496                                         intval(local_user())
 497                                 );
 498                                 if(count($x) && count($r)) {
 499                                         $mailbox = construct_mailbox_name($r[0]);
 500                                         $password = '';
 501                                         openssl_private_decrypt(hex2bin($r[0]['pass']),$password,$x[0]['prvkey']);
 502                                         $mbox = email_connect($mailbox,$r[0]['user'],$password);
 503                                         if(! $mbox)
 504                                                 logger('probe_url: email_connect failed.');
 505                                         unset($password);
 506                                 }
 507                                 if($mbox) {
 508                                         $msgs = email_poll($mbox,$orig_url);
 509                                         logger('probe_url: searching ' . $orig_url . ', ' . count($msgs) . ' messages found.', LOGGER_DEBUG);
 510                                         if(count($msgs)) {
 511                                                 $addr = $orig_url;
 512                                                 $network = NETWORK_MAIL;
 513                                                 $name = substr($url,0,strpos($url,'@'));
 514                                                 $phost = substr($url,strpos($url,'@')+1);
 515                                                 $profile = 'http://' . $phost;
 516                                                 // fix nick character range
 517                                                 $vcard = array('fn' => $name, 'nick' => $name, 'photo' => avatar_img($url));
 518                                                 $notify = 'smtp ' . random_string();
 519                                                 $poll = 'email ' . random_string();
 520                                                 $priority = 0;
 521                                                 $x = email_msg_meta($mbox,$msgs[0]);
 522                                                 if(stristr($x[0]->from,$orig_url))
 523                                                         $adr = imap_rfc822_parse_adrlist($x[0]->from,'');
 524                                                 elseif(stristr($x[0]->to,$orig_url))
 525                                                         $adr = imap_rfc822_parse_adrlist($x[0]->to,'');
 526                                                 if(isset($adr)) {
 527                                                         foreach($adr as $feadr) {
 528                                                                 if((strcasecmp($feadr->mailbox,$name) == 0)
 529                                                                         &&(strcasecmp($feadr->host,$phost) == 0)
 530                                                                         && (strlen($feadr->personal))) {
 531
 532                                                                         $personal = imap_mime_header_decode($feadr->personal);
 533                                                                         $vcard['fn'] = "";
 534                                                                         foreach($personal as $perspart)
 535                                                                                 if ($perspart->charset != "default")
 536                                                                                         $vcard['fn'] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
 537                                                                                 else
 538                                                                                         $vcard['fn'] .= $perspart->text;
 539
 540                                                                         $vcard['fn'] = notags($vcard['fn']);
 541                                                                 }
 542                                                         }
 543                                                 }
 544                                         }
 545                                         imap_close($mbox);
 546                                 }
 547                         }
 548                 }
 549         }
 550
 551         if($mode == PROBE_NORMAL) {
 552
 553                 if(strlen($zot)) {
 554                         $s = fetch_url($zot);
 555                         if($s) {
 556                                 $j = json_decode($s);
 557                                 if($j) {
 558                                         $network = NETWORK_ZOT;
 559                                         $vcard   = array(
 560                                                 'fn'    => $j->fullname,
 561                                                 'nick'  => $j->nickname,
 562                                                 'photo' => $j->photo
 563                                         );
 564                                         $profile  = $j->url;
 565                                         $notify   = $j->post;
 566                                         $pubkey   = $j->pubkey;
 567                                         $poll     = 'N/A';
 568                                 }
 569                         }
 570                 }
 571
 572
 573                 if(strlen($dfrn)) {
 574                         $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn, true);
 575                         if(is_array($ret) && x($ret,'dfrn-request')) {
 576                                 $network = NETWORK_DFRN;
 577                                 $request = $ret['dfrn-request'];
 578                                 $confirm = $ret['dfrn-confirm'];
 579                                 $notify  = $ret['dfrn-notify'];
 580                                 $poll    = $ret['dfrn-poll'];
 581
 582                                 $vcard = array();
 583                                 $vcard['fn'] = $ret['fn'];
 584                                 $vcard['nick'] = $ret['nick'];
 585                                 $vcard['photo'] = $ret['photo'];
 586                         }
 587                 }
 588         }
 589
 590         // Scrape the public key from the hcard.
 591         // Diaspora will remove it from the webfinger somewhere in the future.
 592         if (($hcard != "") AND ($pubkey == "")) {
 593                 $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn, true);
 594                 if (isset($ret["guid"]))
 595                         $diaspora_guid = $ret["guid"];
 596                 if (isset($ret["url"]))
 597                         $diaspora_base = $ret["url"];
 598
 599                 if (isset($ret["key"])) {
 600                         $hcard_key = $ret["key"];
 601                         if(strstr($hcard_key,'RSA '))
 602                                 $pubkey = rsatopem($hcard_key);
 603                         else
 604                                 $pubkey = $hcard_key;
 605                 }
 606         }
 607
 608         if(($network == "") && $diaspora_base && $diaspora_guid) {
 609                 $diaspora_notify = $diaspora_base.'receive/users/'.$diaspora_guid;
 610                 $diaspora = true;
 611
 612                 if($mode == PROBE_DIASPORA || !$notify || ($notify == $diaspora_notify)) {
 613                         $notify = $diaspora_notify;
 614                         $batch  = $diaspora_base . 'receive/public' ;
 615                 }
 616                 if(strpos($url,'@'))
 617                         $addr = str_replace('acct:', '', $url);
 618         }
 619
 620         if($network !== NETWORK_ZOT && $network !== NETWORK_DFRN && $network !== NETWORK_MAIL) {
 621                 if($diaspora)
 622                         $network = NETWORK_DIASPORA;
 623                 elseif($has_lrdd AND ($notify))
 624                         $network  = NETWORK_OSTATUS;
 625
 626                 if(strpos($url,'@'))
 627                         $addr = str_replace('acct:', '', $url);
 628
 629                 $priority = 0;
 630
 631                 if($hcard && ! $vcard) {
 632                         $vcard = scrape_vcard($hcard);
 633
 634                         // Google doesn't use absolute url in profile photos
 635
 636                         if((x($vcard,'photo')) && substr($vcard['photo'],0,1) == '/') {
 637                                 $h = @parse_url($hcard);
 638                                 if($h)
 639                                         $vcard['photo'] = $h['scheme'] . '://' . $h['host'] . $vcard['photo'];
 640                         }
 641
 642                         logger('probe_url: scrape_vcard: ' . print_r($vcard,true), LOGGER_DATA);
 643                 }
 644
 645                 if($diaspora && $addr) {
 646                         // Diaspora returns the name as the nick. As the nick will never be updated,
 647                         // let's use the Diaspora nickname (the first part of the handle) as the nick instead
 648                         $addr_parts = explode('@', $addr);
 649                         $vcard['nick'] = $addr_parts[0];
 650                 }
 651
 652                 if($lastfm) {
 653                         $profile = $url;
 654                         $poll = str_replace(array('www.','last.fm/'),array('','ws.audioscrobbler.com/1.0/'),$url) . '/recenttracks.rss';
 655                         $vcard['nick'] = basename($url);
 656                         $vcard['fn'] = $vcard['nick'] . t(' on Last.fm');
 657                         $network = NETWORK_FEED;
 658                 }
 659
 660                 if(! x($vcard,'fn'))
 661                         if(x($vcard,'nick'))
 662                                 $vcard['fn'] = $vcard['nick'];
 663
 664                 $check_feed = false;
 665
 666                 if(stristr($url,'tumblr.com') && (! stristr($url,'/rss'))) {
 667                         $poll = $url . '/rss';
 668                         $check_feed = true;
 669                         // Will leave it to others to figure out how to grab the avatar, which is on the $url page in the open graph meta links
 670                 }
 671
 672                 if($appnet || ! $poll)
 673                         $check_feed = true;
 674                 if((! isset($vcard)) || (! x($vcard,'fn')) || (! $profile))
 675                         $check_feed = true;
 676                 if(($at_addr) && (! count($links)))
 677                         $check_feed = false;
 678
 679                 if ($connectornetworks)
 680                         $check_feed = false;
 681
 682                 if($check_feed) {
 683
 684                         $feedret = scrape_feed(($poll) ? $poll : $url);
 685
 686                         logger('probe_url: scrape_feed ' . (($poll)? $poll : $url) . ' returns: ' . print_r($feedret,true), LOGGER_DATA);
 687                         if(count($feedret) && ($feedret['feed_atom'] || $feedret['feed_rss'])) {
 688                                 $poll = ((x($feedret,'feed_atom')) ? unamp($feedret['feed_atom']) : unamp($feedret['feed_rss']));
 689                                 if(! x($vcard))
 690                                         $vcard = array();
 691                         }
 692
 693                         if(x($feedret,'photo') && (! x($vcard,'photo')))
 694                                 $vcard['photo'] = $feedret['photo'];
 695
 696                         $cookiejar = tempnam(get_temppath(), 'cookiejar-scrape-feed-');
 697                         $xml = fetch_url($poll, false, $redirects, 0, Null, $cookiejar);
 698                         unlink($cookiejar);
 699
 700                         logger('probe_url: fetch feed: ' . $poll . ' returns: ' . $xml, LOGGER_DATA);
 701
 702                         if ($xml == "") {
 703                                 logger("scrape_feed: XML is empty for feed ".$poll);
 704                                 $network = NETWORK_PHANTOM;
 705                         } else {
 706                                 $data = feed_import($xml,$dummy1,$dummy2, $dummy3, true);
 707
 708                                 if (!is_array($data)) {
 709                                         logger("scrape_feed: This doesn't seem to be a feed: ".$poll);
 710                                         $network = NETWORK_PHANTOM;
 711                                 } else {
 712                                         if (($vcard["photo"] == "") AND ($data["header"]["author-avatar"] != ""))
 713                                                 $vcard["photo"] = $data["header"]["author-avatar"];
 714
 715                                         if (($vcard["fn"] == "") AND ($data["header"]["author-name"] != ""))
 716                                                 $vcard["fn"] = $data["header"]["author-name"];
 717
 718                                         if (($vcard["nick"] == "") AND ($data["header"]["author-nick"] != ""))
 719                                                 $vcard["nick"] = $data["header"]["author-nick"];
 720
 721                                         if ($network == NETWORK_OSTATUS) {
 722                                                 if ($data["header"]["author-id"] != "")
 723                                                         $alias = $data["header"]["author-id"];
 724
 725                                                 if ($data["header"]["author-link"] != "")
 726                                                         $profile = $data["header"]["author-link"];
 727
 728                                         } elseif(!$profile AND ($data["header"]["author-link"] != "") AND !in_array($network, array("", NETWORK_FEED)))
 729                                                 $profile = $data["header"]["author-link"];
 730                                 }
 731                         }
 732
 733                         // Workaround for misconfigured Friendica servers
 734                         if (($network == "") AND (strstr($url, "/profile/"))) {
 735                                 $noscrape = str_replace("/profile/", "/noscrape/", $url);
 736                                 $noscrapejson = fetch_url($noscrape);
 737                                 if ($noscrapejson) {
 738
 739                                         $network = NETWORK_DFRN;
 740
 741                                         $poco = str_replace("/profile/", "/poco/", $url);
 742
 743                                         $noscrapedata = json_decode($noscrapejson, true);
 744
 745                                         if (isset($noscrapedata["addr"]))
 746                                                 $addr = $noscrapedata["addr"];
 747
 748                                         if (isset($noscrapedata["fn"]))
 749                                                 $vcard["fn"] = $noscrapedata["fn"];
 750
 751                                         if (isset($noscrapedata["key"]))
 752                                                 $pubkey = $noscrapedata["key"];
 753
 754                                         if (isset($noscrapedata["photo"]))
 755                                                 $vcard["photo"] = $noscrapedata["photo"];
 756
 757                                         if (isset($noscrapedata["dfrn-request"]))
 758                                                 $request = $noscrapedata["dfrn-request"];
 759
 760                                         if (isset($noscrapedata["dfrn-confirm"]))
 761                                                 $confirm = $noscrapedata["dfrn-confirm"];
 762
 763                                         if (isset($noscrapedata["dfrn-notify"]))
 764                                                 $notify = $noscrapedata["dfrn-notify"];
 765
 766                                         if (isset($noscrapedata["dfrn-poll"]))
 767                                                 $poll = $noscrapedata["dfrn-poll"];
 768
 769                                 }
 770                         }
 771
 772                         if(! $network)
 773                                 $network = NETWORK_FEED;
 774
 775                         if(! x($vcard,'nick')) {
 776                                 $vcard['nick'] = strtolower(notags(unxmlify($vcard['fn'])));
 777                                 if(strpos($vcard['nick'],' '))
 778                                         $vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' ')));
 779                         }
 780                         if(! $priority)
 781                                 $priority = 2;
 782                 }
 783         }
 784
 785         if(! x($vcard,'photo')) {
 786                 $a = get_app();
 787                 $vcard['photo'] = App::get_baseurl() . '/images/person-175.jpg' ;
 788         }
 789
 790         if(! $profile)
 791                 $profile = $url;
 792
 793         // No human could be associated with this link, use the URL as the contact name
 794
 795         if(($network === NETWORK_FEED) && ($poll) && (! x($vcard,'fn')))
 796                 $vcard['fn'] = $url;
 797
 798         if ($diaspora_base != "")
 799                 $baseurl = $diaspora_base;
 800
 801         if (($baseurl == "") AND ($notify != "") AND ($poll != "")) {
 802                 $baseurl = matching_url(normalise_link($notify), normalise_link($poll));
 803
 804                 $baseurl2 = matching_url($baseurl, normalise_link($profile));
 805                 if ($baseurl2 != "")
 806                         $baseurl = $baseurl2;
 807         }
 808
 809         if (($baseurl == "") AND ($notify != ""))
 810                 $baseurl = matching_url(normalise_link($profile), normalise_link($notify));
 811
 812         if (($baseurl == "") AND ($poll != ""))
 813                 $baseurl = matching_url(normalise_link($profile), normalise_link($poll));
 814
 815         if (substr($baseurl, -10) == "/index.php")
 816                 $baseurl = str_replace("/index.php", "", $baseurl);
 817
 818         if ($network == "")
 819                 $network = NETWORK_PHANTOM;
 820
 821         $baseurl = rtrim($baseurl, "/");
 822
 823         if(strpos($url,'@') AND ($addr == "") AND ($network == NETWORK_DFRN))
 824                 $addr = str_replace('acct:', '', $url);
 825
 826         $vcard['fn'] = notags($vcard['fn']);
 827         $vcard['nick'] = str_replace(' ','',notags($vcard['nick']));
 828
 829         $result['name'] = $vcard['fn'];
 830         $result['nick'] = $vcard['nick'];
 831         $result['guid'] = $diaspora_guid;
 832         $result['url'] = $profile;
 833         $result['addr'] = $addr;
 834         $result['batch'] = $batch;
 835         $result['notify'] = $notify;
 836         $result['poll'] = $poll;
 837         $result['request'] = $request;
 838         $result['confirm'] = $confirm;
 839         $result['poco'] = $poco;
 840         $result['photo'] = $vcard['photo'];
 841         $result['priority'] = $priority;
 842         $result['network'] = $network;
 843         $result['alias'] = $alias;
 844         $result['pubkey'] = $pubkey;
 845         $result['baseurl'] = $baseurl;
 846
 847         logger('probe_url: ' . print_r($result,true), LOGGER_DEBUG);
 848
 849         if ($level == 1) {
 850                 // Trying if it maybe a diaspora account
 851                 if (($result['network'] == NETWORK_FEED) OR ($result['addr'] == "")) {
 852                         require_once('include/bbcode.php');
 853                         $address = GetProfileUsername($url, "", true);
 854                         $result2 = probe_url($address, $mode, ++$level);
 855                         if (!in_array($result2['network'], array("", NETWORK_PHANTOM, NETWORK_FEED)))
 856                                 $result = $result2;
 857                 }
 858
 859                 // Maybe it's some non standard GNU Social installation (Single user, subfolder or no uri rewrite)
 860                 if (($result['network'] == NETWORK_FEED) AND ($result['baseurl'] != "") AND ($result['nick'] != "")) {
 861                         $addr = $result['nick'].'@'.str_replace("http://", "", $result['baseurl']);
 862                         $result2 = probe_url($addr, $mode, ++$level);
 863                         if (!in_array($result2['network'], array("", NETWORK_PHANTOM, NETWORK_FEED)))
 864                                 $result = $result2;
 865                 }
 866
 867                 // Quickfix for Hubzilla systems with enabled OStatus plugin
 868                 if (($result['network'] == NETWORK_DIASPORA) AND ($result["batch"] == "")) {
 869                         $result2 = probe_url($url, PROBE_DIASPORA, ++$level);
 870                         if ($result2['network'] == NETWORK_DIASPORA) {
 871                                 $addr = $result["addr"];
 872                                 $result = $result2;
 873
 874                                 if (($result["addr"] == "") AND ($addr != ""))
 875                                         $result["addr"] = $addr;
 876                         }
 877                 }
 878         }
 879
 880         // Only store into the cache if the value seems to be valid
 881         if ($result['network'] != NETWORK_PHANTOM) {
 882                 Cache::set("probe_url:".$mode.":".$original_url,serialize($result), CACHE_DAY);
 883
 884                 /// @todo temporary fix - we need a real contact update function that updates only changing fields
 885                 /// The biggest problem is the avatar picture that could have a reduced image size.
 886                 /// It should only be updated if the existing picture isn't existing anymore.
 887                 if (($result['network'] != NETWORK_FEED) AND ($mode == PROBE_NORMAL) AND
 888                         $result["name"] AND $result["nick"] AND $result["url"] AND $result["addr"] AND $result["poll"])
 889                         q("UPDATE `contact` SET `name` = '%s', `nick` = '%s', `url` = '%s', `addr` = '%s',
 890                                         `notify` = '%s', `poll` = '%s', `alias` = '%s', `success_update` = '%s'
 891                                 WHERE `nurl` = '%s' AND NOT `self` AND `uid` = 0",
 892                                 dbesc($result["name"]),
 893                                 dbesc($result["nick"]),
 894                                 dbesc($result["url"]),
 895                                 dbesc($result["addr"]),
 896                                 dbesc($result["notify"]),
 897                                 dbesc($result["poll"]),
 898                                 dbesc($result["alias"]),
 899                                 dbesc(datetime_convert()),
 900                                 dbesc(normalise_link($result['url']))
 901                 );
 902         }
 903
 904         return $result;
 905 }
 906
 907 /**
 908  * @brief Find the matching part between two url
 909  *
 910  * @param string $url1
 911  * @param string $url2
 912  * @return string The matching part
 913  */
 914 function matching_url($url1, $url2) {
 915
 916         if (($url1 == "") OR ($url2 == ""))
 917                 return "";
 918
 919         $url1 = normalise_link($url1);
 920         $url2 = normalise_link($url2);
 921
 922         $parts1 = parse_url($url1);
 923         $parts2 = parse_url($url2);
 924
 925         if (!isset($parts1["host"]) OR !isset($parts2["host"]))
 926                 return "";
 927
 928         if ($parts1["scheme"] != $parts2["scheme"])
 929                 return "";
 930
 931         if ($parts1["host"] != $parts2["host"])
 932                 return "";
 933
 934         if ($parts1["port"] != $parts2["port"])
 935                 return "";
 936
 937         $match = $parts1["scheme"]."://".$parts1["host"];
 938
 939         if ($parts1["port"])
 940                 $match .= ":".$parts1["port"];
 941
 942         $pathparts1 = explode("/", $parts1["path"]);
 943         $pathparts2 = explode("/", $parts2["path"]);
 944
 945         $i = 0;
 946         $path = "";
 947         do {
 948                 $path1 = $pathparts1[$i];
 949                 $path2 = $pathparts2[$i];
 950
 951                 if ($path1 == $path2)
 952                         $path .= $path1."/";
 953
 954         } while (($path1 == $path2) AND ($i++ <= count($pathparts1)));
 955
 956         $match .= $path;
 957
 958         return normalise_link($match);
 959 }