include/Scrape.php

   1 <?php
   2
   3 require_once('library/HTML5/Parser.php');
   4 require_once('include/crypto.php');
   5
   6 if(! function_exists('scrape_dfrn')) {
   7 function scrape_dfrn($url, $dont_probe = false) {
   8
   9         $a = get_app();
  10
  11         $ret = array();
  12
  13         logger('scrape_dfrn: url=' . $url);
  14
  15         $s = fetch_url($url);
  16
  17         if(! $s)
  18                 return $ret;
  19
  20         if (!$dont_probe) {
  21                 $probe = probe_url($url);
  22
  23                 if (isset($probe["addr"]))
  24                         $ret["addr"] = $probe["addr"];
  25         }
  26
  27         $headers = $a->get_curl_headers();
  28         logger('scrape_dfrn: headers=' . $headers, LOGGER_DEBUG);
  29
  30
  31         $lines = explode("\n",$headers);
  32         if(count($lines)) {
  33                 foreach($lines as $line) {
  34                         // don't try and run feeds through the html5 parser
  35                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
  36                                 return ret;
  37                 }
  38         }
  39
  40         try {
  41                 $dom = HTML5_Parser::parse($s);
  42         } catch (DOMException $e) {
  43                 logger('scrape_dfrn: parse error: ' . $e);
  44         }
  45
  46         if(! $dom)
  47                 return $ret;
  48
  49         $items = $dom->getElementsByTagName('link');
  50
  51         // get DFRN link elements
  52
  53         foreach($items as $item) {
  54                 $x = $item->getAttribute('rel');
  55                 if(($x === 'alternate') && ($item->getAttribute('type') === 'application/atom+xml'))
  56                         $ret['feed_atom'] = $item->getAttribute('href');
  57                 if(substr($x,0,5) == "dfrn-") {
  58                         $ret[$x] = $item->getAttribute('href');
  59                 }
  60                 if($x === 'lrdd') {
  61                         $decoded = urldecode($item->getAttribute('href'));
  62                         if(preg_match('/acct:([^@]*)@/',$decoded,$matches))
  63                                 $ret['nick'] = $matches[1];
  64                 }
  65         }
  66
  67         // Pull out hCard profile elements
  68
  69         $largest_photo = 0;
  70
  71         $items = $dom->getElementsByTagName('*');
  72         foreach($items as $item) {
  73                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
  74                         $level2 = $item->getElementsByTagName('*');
  75                         foreach($level2 as $x) {
  76                                 if(attribute_contains($x->getAttribute('class'),'fn')) {
  77                                         $ret['fn'] = $x->textContent;
  78                                 }
  79                                 if((attribute_contains($x->getAttribute('class'),'photo'))
  80                                         || (attribute_contains($x->getAttribute('class'),'avatar'))) {
  81                                         $size = intval($x->getAttribute('width'));
  82                                         // dfrn prefers 175, so if we find this, we set largest_size so it can't be topped.
  83                                         if(($size > $largest_photo) || ($size == 175) || (! $largest_photo)) {
  84                                                 $ret['photo'] = $x->getAttribute('src');
  85                                                 $largest_photo = (($size == 175) ? 9999 : $size);
  86                                         }
  87                                 }
  88                                 if(attribute_contains($x->getAttribute('class'),'key')) {
  89                                         $ret['key'] = $x->textContent;
  90                                 }
  91                         }
  92                 }
  93         }
  94
  95         return $ret;
  96 }}
  97
  98
  99
 100
 101
 102
 103 if(! function_exists('validate_dfrn')) {
 104 function validate_dfrn($a) {
 105         $errors = 0;
 106         if(! x($a,'key'))
 107                 $errors ++;
 108         if(! x($a,'dfrn-request'))
 109                 $errors ++;
 110         if(! x($a,'dfrn-confirm'))
 111                 $errors ++;
 112         if(! x($a,'dfrn-notify'))
 113                 $errors ++;
 114         if(! x($a,'dfrn-poll'))
 115                 $errors ++;
 116         return $errors;
 117 }}
 118
 119 if(! function_exists('scrape_meta')) {
 120 function scrape_meta($url) {
 121
 122         $a = get_app();
 123
 124         $ret = array();
 125
 126         logger('scrape_meta: url=' . $url);
 127
 128         $s = fetch_url($url);
 129
 130         if(! $s)
 131                 return $ret;
 132
 133         $headers = $a->get_curl_headers();
 134         logger('scrape_meta: headers=' . $headers, LOGGER_DEBUG);
 135
 136         $lines = explode("\n",$headers);
 137         if(count($lines)) {
 138                 foreach($lines as $line) {
 139                         // don't try and run feeds through the html5 parser
 140                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
 141                                 return ret;
 142                 }
 143         }
 144
 145         try {
 146                 $dom = HTML5_Parser::parse($s);
 147         } catch (DOMException $e) {
 148                 logger('scrape_meta: parse error: ' . $e);
 149         }
 150
 151         if(! $dom)
 152                 return $ret;
 153
 154         $items = $dom->getElementsByTagName('meta');
 155
 156         // get DFRN link elements
 157
 158         foreach($items as $item) {
 159                 $x = $item->getAttribute('name');
 160                 if(substr($x,0,5) == "dfrn-")
 161                         $ret[$x] = $item->getAttribute('content');
 162         }
 163
 164         return $ret;
 165 }}
 166
 167
 168 if(! function_exists('scrape_vcard')) {
 169 function scrape_vcard($url) {
 170
 171         $a = get_app();
 172
 173         $ret = array();
 174
 175         logger('scrape_vcard: url=' . $url);
 176
 177         $s = fetch_url($url);
 178
 179         if(! $s)
 180                 return $ret;
 181
 182         $headers = $a->get_curl_headers();
 183         $lines = explode("\n",$headers);
 184         if(count($lines)) {
 185                 foreach($lines as $line) {
 186                         // don't try and run feeds through the html5 parser
 187                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
 188                                 return ret;
 189                 }
 190         }
 191
 192         try {
 193                 $dom = HTML5_Parser::parse($s);
 194         } catch (DOMException $e) {
 195                 logger('scrape_vcard: parse error: ' . $e);
 196         }
 197
 198         if(! $dom)
 199                 return $ret;
 200
 201         // Pull out hCard profile elements
 202
 203         $largest_photo = 0;
 204
 205         $items = $dom->getElementsByTagName('*');
 206         foreach($items as $item) {
 207                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
 208                         $level2 = $item->getElementsByTagName('*');
 209                         foreach($level2 as $x) {
 210                                 if(attribute_contains($x->getAttribute('class'),'fn'))
 211                                         $ret['fn'] = $x->textContent;
 212                                 if((attribute_contains($x->getAttribute('class'),'photo'))
 213                                         || (attribute_contains($x->getAttribute('class'),'avatar'))) {
 214                                         $size = intval($x->getAttribute('width'));
 215                                         if(($size > $largest_photo) || (! $largest_photo)) {
 216                                                 $ret['photo'] = $x->getAttribute('src');
 217                                                 $largest_photo = $size;
 218                                         }
 219                                 }
 220                                 if((attribute_contains($x->getAttribute('class'),'nickname'))
 221                                         || (attribute_contains($x->getAttribute('class'),'uid'))) {
 222                                         $ret['nick'] = $x->textContent;
 223                                 }
 224                         }
 225                 }
 226         }
 227
 228         return $ret;
 229 }}
 230
 231
 232 if(! function_exists('scrape_feed')) {
 233 function scrape_feed($url) {
 234
 235         $a = get_app();
 236
 237         $ret = array();
 238         $s = fetch_url($url);
 239
 240         $headers = $a->get_curl_headers();
 241         $code = $a->get_curl_code();
 242
 243         logger('scrape_feed: returns: ' . $code . ' headers=' . $headers, LOGGER_DEBUG);
 244
 245         if(! $s) {
 246                 logger('scrape_feed: no data returned for ' . $url);
 247                 return $ret;
 248         }
 249
 250
 251         $lines = explode("\n",$headers);
 252         if(count($lines)) {
 253                 foreach($lines as $line) {
 254                         if(stristr($line,'content-type:')) {
 255                                 if(stristr($line,'application/atom+xml') || stristr($s,'<feed')) {
 256                                         $ret['feed_atom'] = $url;
 257                                         return $ret;
 258                                 }
 259                                 if(stristr($line,'application/rss+xml') || stristr($s,'<rss')) {
 260                                         $ret['feed_rss'] = $url;
 261                                         return $ret;
 262                                 }
 263                         }
 264                 }
 265                 // perhaps an RSS version 1 feed with a generic or incorrect content-type?
 266                 if(stristr($s,'</item>')) {
 267                         $ret['feed_rss'] = $url;
 268                         return $ret;
 269                 }
 270         }
 271
 272         $basename = implode('/', array_slice(explode('/',$url),0,3)) . '/';
 273
 274         $doc = new DOMDocument();
 275         @$doc->loadHTML($s);
 276         $xpath = new DomXPath($doc);
 277
 278         $base = $xpath->query("//base");
 279         foreach ($base as $node) {
 280                 $attr = array();
 281
 282                 if ($node->attributes->length)
 283                         foreach ($node->attributes as $attribute)
 284                                 $attr[$attribute->name] = $attribute->value;
 285
 286                 if ($attr["href"] != "")
 287                         $basename = $attr["href"] ;
 288         }
 289
 290         $list = $xpath->query("//link");
 291         foreach ($list as $node) {
 292                 $attr = array();
 293
 294                 if ($node->attributes->length)
 295                         foreach ($node->attributes as $attribute)
 296                                 $attr[$attribute->name] = $attribute->value;
 297
 298                 if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/atom+xml"))
 299                         $ret["feed_atom"] = $attr["href"];
 300
 301                 if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/rss+xml"))
 302                         $ret["feed_rss"] = $attr["href"];
 303         }
 304
 305         // Drupal and perhaps others only provide relative URLs. Turn them into absolute.
 306
 307         if(x($ret,'feed_atom') && (! strstr($ret['feed_atom'],'://')))
 308                 $ret['feed_atom'] = $basename . $ret['feed_atom'];
 309         if(x($ret,'feed_rss') && (! strstr($ret['feed_rss'],'://')))
 310                 $ret['feed_rss'] = $basename . $ret['feed_rss'];
 311
 312         return $ret;
 313 }}
 314
 315
 316 /**
 317  *
 318  * Probe a network address to discover what kind of protocols we need to communicate with it.
 319  *
 320  * Warning: this function is a bit touchy and there are some subtle dependencies within the logic flow.
 321  * Edit with care.
 322  *
 323  */
 324
 325 /**
 326  *
 327  * PROBE_DIASPORA has a bias towards returning Diaspora information
 328  * while PROBE_NORMAL has a bias towards dfrn/zot - in the case where
 329  * an address (such as a Friendica address) supports more than one type
 330  * of network.
 331  *
 332  */
 333
 334
 335 define ( 'PROBE_NORMAL',   0);
 336 define ( 'PROBE_DIASPORA', 1);
 337
 338 function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
 339         require_once('include/email.php');
 340
 341         $result = array();
 342
 343         if(! $url)
 344                 return $result;
 345
 346         $result = Cache::get("probe_url:".$mode.":".$url);
 347         if (!is_null($result)) {
 348                 $result = unserialize($result);
 349                 return $result;
 350         }
 351
 352         $network = null;
 353         $diaspora = false;
 354         $diaspora_base = '';
 355         $diaspora_guid = '';
 356         $diaspora_key = '';
 357         $has_lrdd = false;
 358         $email_conversant = false;
 359         $connectornetworks = false;
 360         $appnet = false;
 361
 362         if (strpos($url,'twitter.com')) {
 363                 $connectornetworks = true;
 364                 $network = NETWORK_TWITTER;
 365         }
 366
 367         // Twitter is deactivated since twitter closed its old API
 368         //$twitter = ((strpos($url,'twitter.com') !== false) ? true : false);
 369         $lastfm  = ((strpos($url,'last.fm/user') !== false) ? true : false);
 370
 371         $at_addr = ((strpos($url,'@') !== false) ? true : false);
 372
 373         if((!$appnet) && (!$lastfm) && !$connectornetworks) {
 374
 375                 if(strpos($url,'mailto:') !== false && $at_addr) {
 376                         $url = str_replace('mailto:','',$url);
 377                         $links = array();
 378                 }
 379                 else
 380                         $links = lrdd($url);
 381
 382                 if(count($links)) {
 383                         $has_lrdd = true;
 384
 385                         logger('probe_url: found lrdd links: ' . print_r($links,true), LOGGER_DATA);
 386                         foreach($links as $link) {
 387                                 if($link['@attributes']['rel'] === NAMESPACE_ZOT)
 388                                         $zot = unamp($link['@attributes']['href']);
 389                                 if($link['@attributes']['rel'] === NAMESPACE_DFRN)
 390                                         $dfrn = unamp($link['@attributes']['href']);
 391                                 if($link['@attributes']['rel'] === 'salmon')
 392                                         $notify = unamp($link['@attributes']['href']);
 393                                 if($link['@attributes']['rel'] === NAMESPACE_FEED)
 394                                         $poll = unamp($link['@attributes']['href']);
 395                                 if($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard')
 396                                         $hcard = unamp($link['@attributes']['href']);
 397                                 if($link['@attributes']['rel'] === 'http://webfinger.net/rel/profile-page')
 398                                         $profile = unamp($link['@attributes']['href']);
 399                                 if($link['@attributes']['rel'] === 'http://portablecontacts.net/spec/1.0')
 400                                         $poco = unamp($link['@attributes']['href']);
 401                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/seed_location') {
 402                                         $diaspora_base = unamp($link['@attributes']['href']);
 403                                         $diaspora = true;
 404                                 }
 405                                 if($link['@attributes']['rel'] === 'http://joindiaspora.com/guid') {
 406                                         $diaspora_guid = unamp($link['@attributes']['href']);
 407                                         $diaspora = true;
 408                                 }
 409                                 if($link['@attributes']['rel'] === 'diaspora-public-key') {
 410                                         $diaspora_key = base64_decode(unamp($link['@attributes']['href']));
 411                                         if(strstr($diaspora_key,'RSA '))
 412                                                 $pubkey = rsatopem($diaspora_key);
 413                                         else
 414                                                 $pubkey = $diaspora_key;
 415                                         $diaspora = true;
 416                                 }
 417                                 if(($link['@attributes']['rel'] === 'http://ostatus.org/schema/1.0/subscribe') AND ($mode == PROBE_NORMAL)) {
 418                                         $diaspora = false;
 419                                 }
 420                         }
 421
 422                         // Status.Net can have more than one profile URL. We need to match the profile URL
 423                         // to a contact on incoming messages to prevent spam, and we won't know which one
 424                         // to match. So in case of two, one of them is stored as an alias. Only store URL's
 425                         // and not webfinger user@host aliases. If they've got more than two non-email style
 426                         // aliases, let's hope we're lucky and get one that matches the feed author-uri because
 427                         // otherwise we're screwed.
 428
 429                         foreach($links as $link) {
 430                                 if($link['@attributes']['rel'] === 'alias') {
 431                                         if(strpos($link['@attributes']['href'],'@') === false) {
 432                                                 if(isset($profile)) {
 433                                                         if($link['@attributes']['href'] !== $profile)
 434                                                                 $alias = unamp($link['@attributes']['href']);
 435                                                 }
 436                                                 else
 437                                                         $profile = unamp($link['@attributes']['href']);
 438                                         }
 439                                 }
 440                         }
 441
 442                         // If the profile is different from the url then the url is abviously an alias
 443                         if (($alias == "") AND ($profile != "") AND !$at_addr AND (normalise_link($profile) != normalise_link($url)))
 444                                 $alias = $url;
 445                 }
 446                 elseif($mode == PROBE_NORMAL) {
 447
 448                         // Check email
 449
 450                         $orig_url = $url;
 451                         if((strpos($orig_url,'@')) && validate_email($orig_url)) {
 452                                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1",
 453                                         intval(local_user())
 454                                 );
 455                                 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1",
 456                                         intval(local_user())
 457                                 );
 458                                 if(count($x) && count($r)) {
 459                                         $mailbox = construct_mailbox_name($r[0]);
 460                                         $password = '';
 461                                         openssl_private_decrypt(hex2bin($r[0]['pass']),$password,$x[0]['prvkey']);
 462                                         $mbox = email_connect($mailbox,$r[0]['user'],$password);
 463                                         if(! $mbox)
 464                                                 logger('probe_url: email_connect failed.');
 465                                         unset($password);
 466                                 }
 467                                 if($mbox) {
 468                                         $msgs = email_poll($mbox,$orig_url);
 469                                         logger('probe_url: searching ' . $orig_url . ', ' . count($msgs) . ' messages found.', LOGGER_DEBUG);
 470                                         if(count($msgs)) {
 471                                                 $addr = $orig_url;
 472                                                 $network = NETWORK_MAIL;
 473                                                 $name = substr($url,0,strpos($url,'@'));
 474                                                 $phost = substr($url,strpos($url,'@')+1);
 475                                                 $profile = 'http://' . $phost;
 476                                                 // fix nick character range
 477                                                 $vcard = array('fn' => $name, 'nick' => $name, 'photo' => avatar_img($url));
 478                                                 $notify = 'smtp ' . random_string();
 479                                                 $poll = 'email ' . random_string();
 480                                                 $priority = 0;
 481                                                 $x = email_msg_meta($mbox,$msgs[0]);
 482                                                 if(stristr($x[0]->from,$orig_url))
 483                                                         $adr = imap_rfc822_parse_adrlist($x[0]->from,'');
 484                                                 elseif(stristr($x[0]->to,$orig_url))
 485                                                         $adr = imap_rfc822_parse_adrlist($x[0]->to,'');
 486                                                 if(isset($adr)) {
 487                                                         foreach($adr as $feadr) {
 488                                                                 if((strcasecmp($feadr->mailbox,$name) == 0)
 489                                                                         &&(strcasecmp($feadr->host,$phost) == 0)
 490                                                                         && (strlen($feadr->personal))) {
 491
 492                                                                         $personal = imap_mime_header_decode($feadr->personal);
 493                                                                         $vcard['fn'] = "";
 494                                                                         foreach($personal as $perspart)
 495                                                                                 if ($perspart->charset != "default")
 496                                                                                         $vcard['fn'] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
 497                                                                                 else
 498                                                                                         $vcard['fn'] .= $perspart->text;
 499
 500                                                                         $vcard['fn'] = notags($vcard['fn']);
 501                                                                 }
 502                                                         }
 503                                                 }
 504                                         }
 505                                         imap_close($mbox);
 506                                 }
 507                         }
 508                 }
 509         }
 510
 511         if($mode == PROBE_NORMAL) {
 512
 513                 if(strlen($zot)) {
 514                         $s = fetch_url($zot);
 515                         if($s) {
 516                                 $j = json_decode($s);
 517                                 if($j) {
 518                                         $network = NETWORK_ZOT;
 519                                         $vcard   = array(
 520                                                 'fn'    => $j->fullname,
 521                                                 'nick'  => $j->nickname,
 522                                                 'photo' => $j->photo
 523                                         );
 524                                         $profile  = $j->url;
 525                                         $notify   = $j->post;
 526                                         $pubkey   = $j->pubkey;
 527                                         $poll     = 'N/A';
 528                                 }
 529                         }
 530                 }
 531
 532
 533                 if(strlen($dfrn)) {
 534                         $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn, true);
 535                         if(is_array($ret) && x($ret,'dfrn-request')) {
 536                                 $network = NETWORK_DFRN;
 537                                 $request = $ret['dfrn-request'];
 538                                 $confirm = $ret['dfrn-confirm'];
 539                                 $notify  = $ret['dfrn-notify'];
 540                                 $poll    = $ret['dfrn-poll'];
 541
 542                                 $vcard = array();
 543                                 $vcard['fn'] = $ret['fn'];
 544                                 $vcard['nick'] = $ret['nick'];
 545                                 $vcard['photo'] = $ret['photo'];
 546                         }
 547                 }
 548         }
 549
 550         if($diaspora && $diaspora_base && $diaspora_guid) {
 551                 if($mode == PROBE_DIASPORA || ! $notify) {
 552                         $notify = $diaspora_base . 'receive/users/' . $diaspora_guid;
 553                         $batch  = $diaspora_base . 'receive/public' ;
 554                 }
 555                 if(strpos($url,'@'))
 556                         $addr = str_replace('acct:', '', $url);
 557         }
 558
 559         if($network !== NETWORK_ZOT && $network !== NETWORK_DFRN && $network !== NETWORK_MAIL) {
 560                 if($diaspora)
 561                         $network = NETWORK_DIASPORA;
 562                 elseif($has_lrdd AND ($notify))
 563                         $network  = NETWORK_OSTATUS;
 564
 565                 if(strpos($url,'@'))
 566                         $addr = str_replace('acct:', '', $url);
 567
 568                 $priority = 0;
 569
 570                 if($hcard && ! $vcard) {
 571                         $vcard = scrape_vcard($hcard);
 572
 573                         // Google doesn't use absolute url in profile photos
 574
 575                         if((x($vcard,'photo')) && substr($vcard['photo'],0,1) == '/') {
 576                                 $h = @parse_url($hcard);
 577                                 if($h)
 578                                         $vcard['photo'] = $h['scheme'] . '://' . $h['host'] . $vcard['photo'];
 579                         }
 580
 581                         logger('probe_url: scrape_vcard: ' . print_r($vcard,true), LOGGER_DATA);
 582                 }
 583
 584                 if($diaspora && $addr) {
 585                         // Diaspora returns the name as the nick. As the nick will never be updated,
 586                         // let's use the Diaspora nickname (the first part of the handle) as the nick instead
 587                         $addr_parts = explode('@', $addr);
 588                         $vcard['nick'] = $addr_parts[0];
 589                 }
 590
 591                 /* if($twitter) {
 592                         logger('twitter: setup');
 593                         $tid = basename($url);
 594                         $tapi = 'https://api.twitter.com/1/statuses/user_timeline.rss';
 595                         if(intval($tid))
 596                                 $poll = $tapi . '?user_id=' . $tid;
 597                         else
 598                                 $poll = $tapi . '?screen_name=' . $tid;
 599                         $profile = 'http://twitter.com/#!/' . $tid;
 600                         //$vcard['photo'] = 'https://api.twitter.com/1/users/profile_image/' . $tid;
 601                         $vcard['photo'] = 'https://api.twitter.com/1/users/profile_image?screen_name=' . $tid . '&size=bigger';
 602                         $vcard['nick'] = $tid;
 603                         $vcard['fn'] = $tid;
 604                 } */
 605
 606                 if($lastfm) {
 607                         $profile = $url;
 608                         $poll = str_replace(array('www.','last.fm/'),array('','ws.audioscrobbler.com/1.0/'),$url) . '/recenttracks.rss';
 609                         $vcard['nick'] = basename($url);
 610                         $vcard['fn'] = $vcard['nick'] . t(' on Last.fm');
 611                         $network = NETWORK_FEED;
 612                 }
 613
 614                 if(! x($vcard,'fn'))
 615                         if(x($vcard,'nick'))
 616                                 $vcard['fn'] = $vcard['nick'];
 617
 618                 $check_feed = false;
 619
 620                 if(stristr($url,'tumblr.com') && (! stristr($url,'/rss'))) {
 621                         $poll = $url . '/rss';
 622                         $check_feed = true;
 623                         // Will leave it to others to figure out how to grab the avatar, which is on the $url page in the open graph meta links
 624                 }
 625
 626                 if($appnet || ! $poll)
 627                         $check_feed = true;
 628                 if((! isset($vcard)) || (! x($vcard,'fn')) || (! $profile))
 629                         $check_feed = true;
 630                 if(($at_addr) && (! count($links)))
 631                         $check_feed = false;
 632
 633                 if ($connectornetworks)
 634                         $check_feed = false;
 635
 636                 if($check_feed) {
 637
 638                         $feedret = scrape_feed(($poll) ? $poll : $url);
 639
 640                         logger('probe_url: scrape_feed ' . (($poll)? $poll : $url) . ' returns: ' . print_r($feedret,true), LOGGER_DATA);
 641                         if(count($feedret) && ($feedret['feed_atom'] || $feedret['feed_rss'])) {
 642                                 $poll = ((x($feedret,'feed_atom')) ? unamp($feedret['feed_atom']) : unamp($feedret['feed_rss']));
 643                                 if(! x($vcard))
 644                                         $vcard = array();
 645                         }
 646
 647                         if(x($feedret,'photo') && (! x($vcard,'photo')))
 648                                 $vcard['photo'] = $feedret['photo'];
 649                         require_once('library/simplepie/simplepie.inc');
 650                         $feed = new SimplePie();
 651                         $xml = fetch_url($poll);
 652
 653                         logger('probe_url: fetch feed: ' . $poll . ' returns: ' . $xml, LOGGER_DATA);
 654                         $a = get_app();
 655
 656                         logger('probe_url: scrape_feed: headers: ' . $a->get_curl_headers(), LOGGER_DATA);
 657
 658                         // Don't try and parse an empty string
 659                         $feed->set_raw_data(($xml) ? $xml : '<?xml version="1.0" encoding="utf-8" ?><xml></xml>');
 660
 661                         $feed->init();
 662                         if($feed->error()) {
 663                                 logger('probe_url: scrape_feed: Error parsing XML: ' . $feed->error());
 664                                 $network = NETWORK_PHANTOM;
 665                         }
 666
 667                         if(! x($vcard,'photo'))
 668                                 $vcard['photo'] = $feed->get_image_url();
 669                         $author = $feed->get_author();
 670
 671                         if($author) {
 672                                 $vcard['fn'] = unxmlify(trim($author->get_name()));
 673                                 if(! $vcard['fn'])
 674                                         $vcard['fn'] = trim(unxmlify($author->get_email()));
 675                                 if(strpos($vcard['fn'],'@') !== false)
 676                                         $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
 677
 678                                 $email = unxmlify($author->get_email());
 679                                 if(! $profile && $author->get_link())
 680                                         $profile = trim(unxmlify($author->get_link()));
 681                                 if(! $vcard['photo']) {
 682                                         $rawtags = $feed->get_feed_tags( SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
 683                                         if($rawtags) {
 684                                                 $elems = $rawtags[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10];
 685                                                 if((x($elems,'link')) && ($elems['link'][0]['attribs']['']['rel'] === 'photo'))
 686                                                         $vcard['photo'] = $elems['link'][0]['attribs']['']['href'];
 687                                         }
 688                                 }
 689                                 // Fetch fullname via poco:displayName
 690                                 $pocotags = $feed->get_feed_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
 691                                 if ($pocotags) {
 692                                         $elems = $pocotags[0]['child']['http://portablecontacts.net/spec/1.0'];
 693                                         if (isset($elems["displayName"]))
 694                                                 $vcard['fn'] = $elems["displayName"][0]["data"];
 695                                         if (isset($elems["preferredUsername"]))
 696                                                 $vcard['nick'] = $elems["preferredUsername"][0]["data"];
 697                                 }
 698                         }
 699                         else {
 700                                 $item = $feed->get_item(0);
 701                                 if($item) {
 702                                         $author = $item->get_author();
 703                                         if($author) {
 704                                                 $vcard['fn'] = trim(unxmlify($author->get_name()));
 705                                                 if(! $vcard['fn'])
 706                                                         $vcard['fn'] = trim(unxmlify($author->get_email()));
 707                                                 if(strpos($vcard['fn'],'@') !== false)
 708                                                         $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
 709                                                 $email = unxmlify($author->get_email());
 710                                                 if(! $profile && $author->get_link())
 711                                                         $profile = trim(unxmlify($author->get_link()));
 712                                         }
 713                                         if(! $vcard['photo']) {
 714                                                 $rawmedia = $item->get_item_tags('http://search.yahoo.com/mrss/','thumbnail');
 715                                                 if($rawmedia && $rawmedia[0]['attribs']['']['url'])
 716                                                         $vcard['photo'] = unxmlify($rawmedia[0]['attribs']['']['url']);
 717                                         }
 718                                         if(! $vcard['photo']) {
 719                                                 $rawtags = $item->get_item_tags( SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
 720                                                 if($rawtags) {
 721                                                         $elems = $rawtags[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10];
 722                                                         if((x($elems,'link')) && ($elems['link'][0]['attribs']['']['rel'] === 'photo'))
 723                                                                 $vcard['photo'] = $elems['link'][0]['attribs']['']['href'];
 724                                                 }
 725                                         }
 726                                 }
 727                         }
 728
 729                         // Workaround for misconfigured Friendica servers
 730                         if (($network == "") AND (strstr($url, "/profile/"))) {
 731                                 $noscrape = str_replace("/profile/", "/noscrape/", $url);
 732                                 $noscrapejson = fetch_url($noscrape);
 733                                 if ($noscrapejson) {
 734
 735                                         $network = NETWORK_DFRN;
 736
 737                                         $poco = str_replace("/profile/", "/poco/", $url);
 738
 739                                         $noscrapedata = json_decode($noscrapejson, true);
 740
 741                                         if (isset($noscrapedata["addr"]))
 742                                                 $addr = $noscrapedata["addr"];
 743
 744                                         if (isset($noscrapedata["fn"]))
 745                                                 $vcard["fn"] = $noscrapedata["fn"];
 746
 747                                         if (isset($noscrapedata["key"]))
 748                                                 $pubkey = $noscrapedata["key"];
 749
 750                                         if (isset($noscrapedata["photo"]))
 751                                                 $vcard["photo"] = $noscrapedata["photo"];
 752
 753                                         if (isset($noscrapedata["dfrn-request"]))
 754                                                 $request = $noscrapedata["dfrn-request"];
 755
 756                                         if (isset($noscrapedata["dfrn-confirm"]))
 757                                                 $confirm = $noscrapedata["dfrn-confirm"];
 758
 759                                         if (isset($noscrapedata["dfrn-notify"]))
 760                                                 $notify = $noscrapedata["dfrn-notify"];
 761
 762                                         if (isset($noscrapedata["dfrn-poll"]))
 763                                                 $poll = $noscrapedata["dfrn-poll"];
 764
 765                                 }
 766                         }
 767
 768                         if((! $vcard['photo']) && strlen($email))
 769                                 $vcard['photo'] = avatar_img($email);
 770                         if($poll === $profile)
 771                                 $lnk = $feed->get_permalink();
 772                         if(isset($lnk) && strlen($lnk))
 773                                 $profile = $lnk;
 774
 775                         if(! $network) {
 776                                 $network = NETWORK_FEED;
 777                                 // If it is a feed, don't take the author name as feed name
 778                                 unset($vcard['fn']);
 779                         }
 780                         if(! (x($vcard,'fn')))
 781                                 $vcard['fn'] = notags($feed->get_title());
 782                         if(! (x($vcard,'fn')))
 783                                 $vcard['fn'] = notags($feed->get_description());
 784
 785                         if(strpos($vcard['fn'],'Twitter / ') !== false) {
 786                                 $vcard['fn'] = substr($vcard['fn'],strpos($vcard['fn'],'/')+1);
 787                                 $vcard['fn'] = trim($vcard['fn']);
 788                         }
 789                         if(! x($vcard,'nick')) {
 790                                 $vcard['nick'] = strtolower(notags(unxmlify($vcard['fn'])));
 791                                 if(strpos($vcard['nick'],' '))
 792                                         $vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' ')));
 793                         }
 794                         if(! $priority)
 795                                 $priority = 2;
 796                 }
 797         }
 798
 799         if(! x($vcard,'photo')) {
 800                 $a = get_app();
 801                 $vcard['photo'] = $a->get_baseurl() . '/images/person-175.jpg' ;
 802         }
 803
 804         if(! $profile)
 805                 $profile = $url;
 806
 807         // No human could be associated with this link, use the URL as the contact name
 808
 809         if(($network === NETWORK_FEED) && ($poll) && (! x($vcard,'fn')))
 810                 $vcard['fn'] = $url;
 811
 812         if (($notify != "") AND ($poll != "")) {
 813                 $baseurl = matching(normalise_link($notify), normalise_link($poll));
 814
 815                 $baseurl2 = matching($baseurl, normalise_link($profile));
 816                 if ($baseurl2 != "")
 817                         $baseurl = $baseurl2;
 818         }
 819
 820         if (($baseurl == "") AND ($notify != ""))
 821                 $baseurl = matching(normalise_link($profile), normalise_link($notify));
 822
 823         if (($baseurl == "") AND ($poll != ""))
 824                 $baseurl = matching(normalise_link($profile), normalise_link($poll));
 825
 826         $baseurl = rtrim($baseurl, "/");
 827
 828         if(strpos($url,'@') AND ($addr == "") AND ($network == NETWORK_DFRN))
 829                 $addr = str_replace('acct:', '', $url);
 830
 831         $vcard['fn'] = notags($vcard['fn']);
 832         $vcard['nick'] = str_replace(' ','',notags($vcard['nick']));
 833
 834         $result['name'] = $vcard['fn'];
 835         $result['nick'] = $vcard['nick'];
 836         $result['url'] = $profile;
 837         $result['addr'] = $addr;
 838         $result['batch'] = $batch;
 839         $result['notify'] = $notify;
 840         $result['poll'] = $poll;
 841         $result['request'] = $request;
 842         $result['confirm'] = $confirm;
 843         $result['poco'] = $poco;
 844         $result['photo'] = $vcard['photo'];
 845         $result['priority'] = $priority;
 846         $result['network'] = $network;
 847         $result['alias'] = $alias;
 848         $result['pubkey'] = $pubkey;
 849         $result['baseurl'] = $baseurl;
 850
 851         logger('probe_url: ' . print_r($result,true), LOGGER_DEBUG);
 852
 853         if ($level == 1) {
 854                 // Trying if it maybe a diaspora account
 855                 if (($result['network'] == NETWORK_FEED) OR ($result['addr'] == "")) {
 856                         require_once('include/bbcode.php');
 857                         $address = GetProfileUsername($url, "", true);
 858                         $result2 = probe_url($address, $mode, ++$level);
 859                         if ($result2['network'] != "")
 860                                 $result = $result2;
 861                 }
 862
 863                 // Maybe it's some non standard GNU Social installation (Single user, subfolder or no uri rewrite)
 864                 if (($result['network'] == NETWORK_FEED) AND ($result['baseurl'] != "") AND ($result['nick'] != "")) {
 865                         $addr = $result['nick'].'@'.str_replace("http://", "", $result['baseurl']);
 866                         $result2 = probe_url($addr, $mode, ++$level);
 867                         if (($result2['network'] != "") AND ($result2['network'] != NETWORK_FEED))
 868                                 $result = $result2;
 869                 }
 870         }
 871
 872         // Only store into the cache if the value seems to be valid
 873         if ($result['network'] != NETWORK_PHANTOM)
 874                 Cache::set("probe_url:".$mode.":".$url,serialize($result), CACHE_DAY);
 875
 876         return $result;
 877 }
 878
 879 function matching($part1, $part2) {
 880         $len = min(strlen($part1), strlen($part2));
 881
 882         $match = "";
 883         $matching = true;
 884         $i = 0;
 885         while (($i <= $len) AND $matching) {
 886                 if (substr($part1, $i, 1) == substr($part2, $i, 1))
 887                         $match .= substr($part1, $i, 1);
 888                 else
 889                         $matching = false;
 890
 891                 $i++;
 892         }
 893         return($match);
 894 }