X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=include%2FScrape.php;h=6ee3dabfca88c11c42576e2a5c75c5194a96283e;hb=143e1a4a836e1fe4d05ab5d1c6008aa0198d2db5;hp=ce18bb10333f5c5dddf573c609263512716a28e3;hpb=c42d6b133e8e8bec59e5c0869d1dc8d46a7af706;p=friendica.git diff --git a/include/Scrape.php b/include/Scrape.php index ce18bb1033..6ee3dabfca 100644 --- a/include/Scrape.php +++ b/include/Scrape.php @@ -4,7 +4,7 @@ require_once('library/HTML5/Parser.php'); require_once('include/crypto.php'); if(! function_exists('scrape_dfrn')) { -function scrape_dfrn($url) { +function scrape_dfrn($url, $dont_probe = false) { $a = get_app(); @@ -17,6 +17,13 @@ function scrape_dfrn($url) { if(! $s) return $ret; + if (!$dont_probe) { + $probe = probe_url($url); + + if (isset($probe["addr"])) + $ret["addr"] = $probe["addr"]; + } + $headers = $a->get_curl_headers(); logger('scrape_dfrn: headers=' . $headers, LOGGER_DEBUG); @@ -249,7 +256,7 @@ function scrape_feed($url) { $ret['feed_atom'] = $url; return $ret; } - if(stristr($line,'application/rss+xml') || stristr($s,'loadHTML($s); + $xpath = new DomXPath($doc); + $base = $xpath->query("//base"); + foreach ($base as $node) { + $attr = array(); - $head = $dom->getElementsByTagName('base'); - if($head) { - foreach($head as $head0) { - $basename = $head0->getAttribute('href'); - break; - } + if ($node->attributes->length) + foreach ($node->attributes as $attribute) + $attr[$attribute->name] = $attribute->value; + + if ($attr["href"] != "") + $basename = $attr["href"] ; } - if(! $basename) - $basename = implode('/', array_slice(explode('/',$url),0,3)) . '/'; - $items = $dom->getElementsByTagName('link'); + $list = $xpath->query("//link"); + foreach ($list as $node) { + $attr = array(); - // get Atom/RSS link elements, take the first one of either. + if ($node->attributes->length) + foreach ($node->attributes as $attribute) + $attr[$attribute->name] = $attribute->value; - if($items) { - foreach($items as $item) { - $x = $item->getAttribute('rel'); - if(($x === 'alternate') && ($item->getAttribute('type') === 'application/atom+xml')) { - if(! x($ret,'feed_atom')) - $ret['feed_atom'] = $item->getAttribute('href'); - } - if(($x === 'alternate') && ($item->getAttribute('type') === 'application/rss+xml')) { - if(! x($ret,'feed_rss')) - $ret['feed_rss'] = $item->getAttribute('href'); - } - } + if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/atom+xml")) + $ret["feed_atom"] = $attr["href"]; + + if (($attr["rel"] == "alternate") AND ($attr["type"] == "application/rss+xml")) + $ret["feed_rss"] = $attr["href"]; } - // Drupal and perhaps others only provide relative URL's. Turn them into absolute. + // Drupal and perhaps others only provide relative URLs. Turn them into absolute. if(x($ret,'feed_atom') && (! strstr($ret['feed_atom'],'://'))) $ret['feed_atom'] = $basename . $ret['feed_atom']; @@ -327,7 +327,7 @@ function scrape_feed($url) { * PROBE_DIASPORA has a bias towards returning Diaspora information * while PROBE_NORMAL has a bias towards dfrn/zot - in the case where * an address (such as a Friendica address) supports more than one type - * of network. + * of network. * */ @@ -335,7 +335,7 @@ function scrape_feed($url) { define ( 'PROBE_NORMAL', 0); define ( 'PROBE_DIASPORA', 1); -function probe_url($url, $mode = PROBE_NORMAL) { +function probe_url($url, $mode = PROBE_NORMAL, $level = 1) { require_once('include/email.php'); $result = array(); @@ -414,7 +414,7 @@ function probe_url($url, $mode = PROBE_NORMAL) { $pubkey = $diaspora_key; $diaspora = true; } - if($link['@attributes']['rel'] === 'http://ostatus.org/schema/1.0/subscribe') { + if(($link['@attributes']['rel'] === 'http://ostatus.org/schema/1.0/subscribe') AND ($mode == PROBE_NORMAL)) { $diaspora = false; } } @@ -509,6 +509,7 @@ function probe_url($url, $mode = PROBE_NORMAL) { } if($mode == PROBE_NORMAL) { + if(strlen($zot)) { $s = fetch_url($zot); if($s) { @@ -528,8 +529,9 @@ function probe_url($url, $mode = PROBE_NORMAL) { } } + if(strlen($dfrn)) { - $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn); + $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn, true); if(is_array($ret) && x($ret,'dfrn-request')) { $network = NETWORK_DFRN; $request = $ret['dfrn-request']; @@ -557,7 +559,7 @@ function probe_url($url, $mode = PROBE_NORMAL) { if($network !== NETWORK_ZOT && $network !== NETWORK_DFRN && $network !== NETWORK_MAIL) { if($diaspora) $network = NETWORK_DIASPORA; - elseif($has_lrdd) + elseif($has_lrdd AND ($notify)) $network = NETWORK_OSTATUS; if(strpos($url,'@')) @@ -634,6 +636,7 @@ function probe_url($url, $mode = PROBE_NORMAL) { if($check_feed) { $feedret = scrape_feed(($poll) ? $poll : $url); + logger('probe_url: scrape_feed ' . (($poll)? $poll : $url) . ' returns: ' . print_r($feedret,true), LOGGER_DATA); if(count($feedret) && ($feedret['feed_atom'] || $feedret['feed_rss'])) { $poll = ((x($feedret,'feed_atom')) ? unamp($feedret['feed_atom']) : unamp($feedret['feed_rss'])); @@ -653,12 +656,13 @@ function probe_url($url, $mode = PROBE_NORMAL) { logger('probe_url: scrape_feed: headers: ' . $a->get_curl_headers(), LOGGER_DATA); // Don't try and parse an empty string - $feed->set_raw_data(($xml) ? $xml : ''); + $feed->set_raw_data(($xml) ? $xml : ''); $feed->init(); - if($feed->error()) + if($feed->error()) { logger('probe_url: scrape_feed: Error parsing XML: ' . $feed->error()); - + $network = NETWORK_PHANTOM; + } if(! x($vcard,'photo')) $vcard['photo'] = $feed->get_image_url(); @@ -670,6 +674,7 @@ function probe_url($url, $mode = PROBE_NORMAL) { $vcard['fn'] = trim(unxmlify($author->get_email())); if(strpos($vcard['fn'],'@') !== false) $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@')); + $email = unxmlify($author->get_email()); if(! $profile && $author->get_link()) $profile = trim(unxmlify($author->get_link())); @@ -681,6 +686,15 @@ function probe_url($url, $mode = PROBE_NORMAL) { $vcard['photo'] = $elems['link'][0]['attribs']['']['href']; } } + // Fetch fullname via poco:displayName + $pocotags = $feed->get_feed_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'author'); + if ($pocotags) { + $elems = $pocotags[0]['child']['http://portablecontacts.net/spec/1.0']; + if (isset($elems["displayName"])) + $vcard['fn'] = $elems["displayName"][0]["data"]; + if (isset($elems["preferredUsername"])) + $vcard['nick'] = $elems["preferredUsername"][0]["data"]; + } } else { $item = $feed->get_item(0); @@ -712,6 +726,45 @@ function probe_url($url, $mode = PROBE_NORMAL) { } } + // Workaround for misconfigured Friendica servers + if (($network == "") AND (strstr($url, "/profile/"))) { + $noscrape = str_replace("/profile/", "/noscrape/", $url); + $noscrapejson = fetch_url($noscrape); + if ($noscrapejson) { + + $network = NETWORK_DFRN; + + $poco = str_replace("/profile/", "/poco/", $url); + + $noscrapedata = json_decode($noscrapejson, true); + + if (isset($noscrapedata["addr"])) + $addr = $noscrapedata["addr"]; + + if (isset($noscrapedata["fn"])) + $vcard["fn"] = $noscrapedata["fn"]; + + if (isset($noscrapedata["key"])) + $pubkey = $noscrapedata["key"]; + + if (isset($noscrapedata["photo"])) + $vcard["photo"] = $noscrapedata["photo"]; + + if (isset($noscrapedata["dfrn-request"])) + $request = $noscrapedata["dfrn-request"]; + + if (isset($noscrapedata["dfrn-confirm"])) + $confirm = $noscrapedata["dfrn-confirm"]; + + if (isset($noscrapedata["dfrn-notify"])) + $notify = $noscrapedata["dfrn-notify"]; + + if (isset($noscrapedata["dfrn-poll"])) + $poll = $noscrapedata["dfrn-poll"]; + + } + } + if((! $vcard['photo']) && strlen($email)) $vcard['photo'] = avatar_img($email); if($poll === $profile) @@ -757,21 +810,24 @@ function probe_url($url, $mode = PROBE_NORMAL) { $vcard['fn'] = $url; if (($notify != "") AND ($poll != "")) { - $baseurl = matching($notify, $poll); + $baseurl = matching(normalise_link($notify), normalise_link($poll)); - $baseurl2 = matching($baseurl, $profile); + $baseurl2 = matching($baseurl, normalise_link($profile)); if ($baseurl2 != "") $baseurl = $baseurl2; } if (($baseurl == "") AND ($notify != "")) - $baseurl = matching($profile, $notify); + $baseurl = matching(normalise_link($profile), normalise_link($notify)); if (($baseurl == "") AND ($poll != "")) - $baseurl = matching($profile, $poll); + $baseurl = matching(normalise_link($profile), normalise_link($poll)); $baseurl = rtrim($baseurl, "/"); + if(strpos($url,'@') AND ($addr == "") AND ($network == NETWORK_DFRN)) + $addr = str_replace('acct:', '', $url); + $vcard['fn'] = notags($vcard['fn']); $vcard['nick'] = str_replace(' ','',notags($vcard['nick'])); @@ -794,16 +850,28 @@ function probe_url($url, $mode = PROBE_NORMAL) { logger('probe_url: ' . print_r($result,true), LOGGER_DEBUG); - // Trying if it maybe a diaspora account - if (($result['network'] == NETWORK_FEED) OR ($result['addr'] == "")) { - require_once('include/bbcode.php'); - $address = GetProfileUsername($url, "", true); - $result2 = probe_url($address, $mode); - if ($result2['network'] != "") - $result = $result2; + if ($level == 1) { + // Trying if it maybe a diaspora account + if (($result['network'] == NETWORK_FEED) OR ($result['addr'] == "")) { + require_once('include/bbcode.php'); + $address = GetProfileUsername($url, "", true); + $result2 = probe_url($address, $mode, ++$level); + if ($result2['network'] != "") + $result = $result2; + } + + // Maybe it's some non standard GNU Social installation (Single user, subfolder or no uri rewrite) + if (($result['network'] == NETWORK_FEED) AND ($result['baseurl'] != "") AND ($result['nick'] != "")) { + $addr = $result['nick'].'@'.str_replace("http://", "", $result['baseurl']); + $result2 = probe_url($addr, $mode, ++$level); + if (($result2['network'] != "") AND ($result2['network'] != NETWORK_FEED)) + $result = $result2; + } } - Cache::set("probe_url:".$mode.":".$url,serialize($result)); + // Only store into the cache if the value seems to be valid + if ($result['network'] != NETWORK_PHANTOM) + Cache::set("probe_url:".$mode.":".$url,serialize($result), CACHE_DAY); return $result; }