X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=src%2FNetwork%2FProbe.php;h=3f10895c3c5c8874054066735ef4b259061da1ab;hb=6a376c29d85cb3e3b5ccf99bc604da472a7191c7;hp=00c38a4c12c16b383234beb6583bcea71c532c80;hpb=cc7444d84db2bc729c40be2190dbfbe3ca7725ba;p=friendica.git diff --git a/src/Network/Probe.php b/src/Network/Probe.php index 00c38a4c12..3f10895c3c 100644 --- a/src/Network/Probe.php +++ b/src/Network/Probe.php @@ -17,7 +17,6 @@ use Friendica\Core\Logger; use Friendica\Core\Protocol; use Friendica\Core\System; use Friendica\Database\DBA; -use Friendica\Model\Contact; use Friendica\Model\Profile; use Friendica\Protocol\ActivityPub; use Friendica\Protocol\Email; @@ -46,9 +45,10 @@ class Probe */ private static function rearrangeData($data) { - $fields = ["name", "nick", "guid", "url", "addr", "alias", - "photo", "community", "keywords", "location", "about", + $fields = ["name", "nick", "guid", "url", "addr", "alias", "photo", "account-type", + "community", "keywords", "location", "about", "gender", "hide", "batch", "notify", "poll", "request", "confirm", "poco", + "following", "followers", "inbox", "outbox", "sharedinbox", "priority", "network", "pubkey", "baseurl"]; $newdata = []; @@ -198,7 +198,7 @@ class Probe $links = self::lrdd($webbie); Logger::log('webfingerDfrn: '.$webbie.':'.print_r($links, true), Logger::DATA); - if (count($links)) { + if (!empty($links) && is_array($links)) { foreach ($links as $link) { if ($link['@attributes']['rel'] === NAMESPACE_DFRN) { $profile_link = $link['@attributes']['href']; @@ -348,8 +348,11 @@ class Probe if (!self::$istimeout) { $ap_profile = ActivityPub::probeProfile($uri); - if (!empty($ap_profile) && empty($network) && (defaults($data, 'network', '') != Protocol::DFRN)) { + if (empty($data) || (!empty($ap_profile) && empty($network) && (defaults($data, 'network', '') != Protocol::DFRN))) { $data = $ap_profile; + } elseif (!empty($ap_profile)) { + $ap_profile['batch'] = ''; + $data = array_merge($ap_profile, $data); } } else { Logger::notice('Time out detected. AP will not be probed.', ['uri' => $uri]); @@ -391,122 +394,82 @@ class Probe $data['network'] = Protocol::PHANTOM; } + if (!isset($data['hide']) && in_array($data['network'], Protocol::FEDERATED)) { + $data['hide'] = self::getHideStatus($data['url']); + } + $data = self::rearrangeData($data); // Only store into the cache if the value seems to be valid if (!in_array($data['network'], [Protocol::PHANTOM, Protocol::MAIL])) { Cache::set('Probe::uri:' . $network . ':' . $uri, $data, Cache::DAY); + } - /// @todo temporary fix - we need a real contact update function that updates only changing fields - /// The biggest problem is the avatar picture that could have a reduced image size. - /// It should only be updated if the existing picture isn't existing anymore. - /// We only update the contact when it is no probing for a specific network. - if (($data['network'] != Protocol::FEED) - && ($network == '') - && $data['name'] - && $data['nick'] - && $data['url'] - && $data['addr'] - && $data['poll'] - ) { - $fields = [ - 'name' => $data['name'], - 'nick' => $data['nick'], - 'url' => $data['url'], - 'addr' => $data['addr'], - 'photo' => $data['photo'], - 'keywords' => $data['keywords'], - 'location' => $data['location'], - 'about' => $data['about'], - 'notify' => $data['notify'], - 'network' => $data['network'], - 'server_url' => $data['baseurl'] - ]; - - // This doesn't cover the case when a community isn't a community anymore - if (!empty($data['community']) && $data['community']) { - $fields['community'] = $data['community']; - $fields['contact-type'] = Contact::TYPE_COMMUNITY; - } + return $data; + } - $fieldnames = []; - foreach ($fields as $key => $val) { - if (empty($val)) { - unset($fields[$key]); - } else { - $fieldnames[] = $key; - } - } + /** + * Fetches the "hide" status from the profile + * + * @param string $url URL of the profile + * + * @return boolean "hide" status + */ + private static function getHideStatus($url) + { + $curlResult = Network::curl($url); + if (!$curlResult->isSuccess()) { + return false; + } - $fields['updated'] = DateTimeFormat::utcNow(); + // If the file is too large then exit + if (defaults($curlResult->getInfo(), 'download_content_length', 0) > 1000000) { + return false; + } - $condition = ['nurl' => Strings::normaliseLink($data['url'])]; + // If it isn't a HTML file then exit + if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) { + return false; + } - $old_fields = DBA::selectFirst('gcontact', $fieldnames, $condition); + $body = $curlResult->getBody(); - // When the gcontact doesn't exist, the value "true" will trigger an insert. - // In difference to the public contacts we want to have every contact - // in the world in our global contacts. - if (!$old_fields) { - $old_fields = true; + $doc = new DOMDocument(); + @$doc->loadHTML($body); - // These values have to be set only on insert - $fields['photo'] = $data['photo']; - $fields['created'] = DateTimeFormat::utcNow(); - } + $xpath = new DOMXPath($doc); - DBA::update('gcontact', $fields, $condition, $old_fields); - - $fields = [ - 'name' => $data['name'], - 'nick' => $data['nick'], - 'url' => $data['url'], - 'addr' => $data['addr'], - 'alias' => $data['alias'], - 'keywords' => $data['keywords'], - 'location' => $data['location'], - 'about' => $data['about'], - 'batch' => $data['batch'], - 'notify' => $data['notify'], - 'poll' => $data['poll'], - 'request' => $data['request'], - 'confirm' => $data['confirm'], - 'poco' => $data['poco'], - 'network' => $data['network'], - 'pubkey' => $data['pubkey'], - 'priority' => $data['priority'], - 'writable' => true, - 'rel' => Contact::SHARING - ]; - - $fieldnames = []; - - foreach ($fields as $key => $val) { - if (empty($val)) { - unset($fields[$key]); - } else { - $fieldnames[] = $key; - } + $list = $xpath->query('//meta[@name]'); + foreach ($list as $node) { + $meta_tag = []; + if ($node->attributes->length) { + foreach ($node->attributes as $attribute) { + $meta_tag[$attribute->name] = $attribute->value; } + } - $condition = ['nurl' => Strings::normaliseLink($data['url']), 'self' => false, 'uid' => 0]; - - // "$old_fields" will return a "false" when the contact doesn't exist. - // This won't trigger an insert. This is intended, since we only need - // public contacts for everyone we store items from. - // We don't need to store every contact on the planet. - $old_fields = DBA::selectFirst('contact', $fieldnames, $condition); + if (empty($meta_tag['content'])) { + continue; + } - $fields['name-date'] = DateTimeFormat::utcNow(); - $fields['uri-date'] = DateTimeFormat::utcNow(); - $fields['success_update'] = DateTimeFormat::utcNow(); + $content = strtolower(trim($meta_tag['content'])); - DBA::update('contact', $fields, $condition, $old_fields); + switch (strtolower(trim($meta_tag['name']))) { + case 'dfrn-global-visibility': + if ($content == 'false') { + return true; + } + break; + case 'robots': + if (strpos($content, 'noindex') !== false) { + return true; + } + break; } } - return $data; + return false; } /** @@ -576,7 +539,7 @@ class Probe } if ($host == 'twitter.com') { - return ["network" => Protocol::TWITTER]; + return self::twitter($uri); } $lrdd = self::hostMeta($host); @@ -617,7 +580,7 @@ class Probe $nick = substr($uri, 0, strpos($uri, '@')); if (strpos($uri, '@twitter.com')) { - return ["network" => Protocol::TWITTER]; + return self::twitter($uri); } $lrdd = self::hostMeta($host); @@ -845,7 +808,7 @@ class Probe } if (!empty($json["tags"])) { - $keywords = implode(" ", $json["tags"]); + $keywords = implode(", ", $json["tags"]); if ($keywords != "") { $data["keywords"] = $keywords; } @@ -860,6 +823,10 @@ class Probe $data["about"] = $json["about"]; } + if (!empty($json["gender"])) { + $data["gender"] = $json["gender"]; + } + if (!empty($json["key"])) { $data["pubkey"] = $json["key"]; } @@ -884,6 +851,12 @@ class Probe $data["poll"] = $json["dfrn-poll"]; } + if (isset($json["hide"])) { + $data["hide"] = (bool)$json["hide"]; + } else { + $data["hide"] = false; + } + return $data; } @@ -1506,6 +1479,66 @@ class Probe return $data; } + /** + * @brief Check for twitter contact + * + * @param string $uri + * + * @return array twitter data + */ + private static function twitter($uri) + { + if (preg_match('=(.*)@twitter.com=i', $uri, $matches)) { + $nick = $matches[1]; + } elseif (preg_match('=https?://twitter.com/(.*)=i', $uri, $matches)) { + $nick = $matches[1]; + } else { + return []; + } + + $data = []; + $data['url'] = 'https://twitter.com/' . $nick; + $data['addr'] = $nick . '@twitter.com'; + $data['nick'] = $data['name'] = $nick; + $data['network'] = Protocol::TWITTER; + $data['baseurl'] = 'https://twitter.com'; + + $curlResult = Network::curl($data['url'], false); + if (!$curlResult->isSuccess()) { + return []; + } + + $body = $curlResult->getBody(); + $doc = new DOMDocument(); + @$doc->loadHTML($body); + $xpath = new DOMXPath($doc); + + $list = $xpath->query('//img[@class]'); + foreach ($list as $node) { + $img_attr = []; + if ($node->attributes->length) { + foreach ($node->attributes as $attribute) { + $img_attr[$attribute->name] = $attribute->value; + } + } + + if (empty($img_attr['class'])) { + continue; + } + + if (strpos($img_attr['class'], 'ProfileAvatar-image') !== false) { + if (!empty($img_attr['src'])) { + $data['photo'] = $img_attr['src']; + } + if (!empty($img_attr['alt'])) { + $data['name'] = $img_attr['alt']; + } + } + } + + return $data; + } + /** * @brief Check page for feed link *