X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=include%2FScrape.php;h=1df8964f7cccffc825327b628a08633c29914a58;hb=159a7bda85d1514ed3a8d67cea7deca6ed96b437;hp=b20d7d6046dec03ce419b54b34e5cc1ae660da9b;hpb=9663299da1be50eef4327778fae619fc6f9edef4;p=friendica.git diff --git a/include/Scrape.php b/include/Scrape.php index b20d7d6046..1df8964f7c 100644 --- a/include/Scrape.php +++ b/include/Scrape.php @@ -282,7 +282,7 @@ function scrape_feed($url) { } } if(! $basename) - $basename = substr($url,0,strrpos($url,'/')) . '/'; + $basename = implode('/', array_slice(explode('/',$url),0,3)) . '/'; $items = $dom->getElementsByTagName('link'); @@ -346,16 +346,17 @@ function probe_url($url, $mode = PROBE_NORMAL) { $network = null; $diaspora = false; $diaspora_base = ''; - $diaspora_guid = ''; + $diaspora_guid = ''; $diaspora_key = ''; $has_lrdd = false; $email_conversant = false; $twitter = ((strpos($url,'twitter.com') !== false) ? true : false); + $lastfm = ((strpos($url,'last.fm/user') !== false) ? true : false); $at_addr = ((strpos($url,'@') !== false) ? true : false); - if(! $twitter) { + if((! $twitter) && (! $lastfm)) { if(strpos($url,'mailto:') !== false && $at_addr) { $url = str_replace('mailto:','',$url); @@ -393,7 +394,10 @@ function probe_url($url, $mode = PROBE_NORMAL) { } if($link['@attributes']['rel'] === 'diaspora-public-key') { $diaspora_key = base64_decode(unamp($link['@attributes']['href'])); - $pubkey = rsatopem($diaspora_key); + if(strstr($diaspora_key,'RSA ')) + $pubkey = rsatopem($diaspora_key); + else + $pubkey = $diaspora_key; $diaspora = true; } } @@ -431,14 +435,17 @@ function probe_url($url, $mode = PROBE_NORMAL) { intval(local_user()) ); if(count($x) && count($r)) { - $mailbox = construct_mailbox_name($r[0]); + $mailbox = construct_mailbox_name($r[0]); $password = ''; openssl_private_decrypt(hex2bin($r[0]['pass']),$password,$x[0]['prvkey']); $mbox = email_connect($mailbox,$r[0]['user'],$password); + if(! $mbox) + logger('probe_url: email_connect failed.'); unset($password); } if($mbox) { $msgs = email_poll($mbox,$orig_url); + logger('probe_url: searching ' . $orig_url . ', ' . count($msgs) . ' messages found.', LOGGER_DEBUG); if(count($msgs)) { $addr = $orig_url; $network = NETWORK_MAIL; @@ -451,10 +458,10 @@ function probe_url($url, $mode = PROBE_NORMAL) { $poll = 'email ' . random_string(); $priority = 0; $x = email_msg_meta($mbox,$msgs[0]); - if(stristr($x->from,$orig_url)) - $adr = imap_rfc822_parse_adrlist($x->from,''); - elseif(stristr($x->to,$orig_url)) - $adr = imap_rfc822_parse_adrlist($x->to,''); + if(stristr($x[0]->from,$orig_url)) + $adr = imap_rfc822_parse_adrlist($x[0]->from,''); + elseif(stristr($x[0]->to,$orig_url)) + $adr = imap_rfc822_parse_adrlist($x[0]->to,''); if(isset($adr)) { foreach($adr as $feadr) { if((strcasecmp($feadr->mailbox,$name) == 0) @@ -478,7 +485,7 @@ function probe_url($url, $mode = PROBE_NORMAL) { } } } - } + } if($mode == PROBE_NORMAL) { if(strlen($zot)) { @@ -524,7 +531,7 @@ function probe_url($url, $mode = PROBE_NORMAL) { } if(strpos($url,'@')) $addr = str_replace('acct:', '', $url); - } + } if($network !== NETWORK_ZOT && $network !== NETWORK_DFRN && $network !== NETWORK_MAIL) { if($diaspora) @@ -537,17 +544,24 @@ function probe_url($url, $mode = PROBE_NORMAL) { $vcard = scrape_vcard($hcard); // Google doesn't use absolute url in profile photos - + if((x($vcard,'photo')) && substr($vcard['photo'],0,1) == '/') { $h = @parse_url($hcard); if($h) $vcard['photo'] = $h['scheme'] . '://' . $h['host'] . $vcard['photo']; } - + logger('probe_url: scrape_vcard: ' . print_r($vcard,true), LOGGER_DATA); } - if($twitter) { + if($diaspora && $addr) { + // Diaspora returns the name as the nick. As the nick will never be updated, + // let's use the Diaspora nickname (the first part of the handle) as the nick instead + $addr_parts = explode('@', $addr); + $vcard['nick'] = $addr_parts[0]; + } + + if($twitter) { logger('twitter: setup'); $tid = basename($url); $tapi = 'https://api.twitter.com/1/statuses/user_timeline.rss'; @@ -556,9 +570,18 @@ function probe_url($url, $mode = PROBE_NORMAL) { else $poll = $tapi . '?screen_name=' . $tid; $profile = 'http://twitter.com/#!/' . $tid; - $vcard['photo'] = 'https://api.twitter.com/1/users/profile_image/' . $tid; + //$vcard['photo'] = 'https://api.twitter.com/1/users/profile_image/' . $tid; + $vcard['photo'] = 'https://api.twitter.com/1/users/profile_image?screen_name=' . $tid . '&size=bigger'; $vcard['nick'] = $tid; - $vcard['fn'] = $tid . '@twitter'; + $vcard['fn'] = $tid; + } + + if($lastfm) { + $profile = $url; + $poll = str_replace(array('www.','last.fm/'),array('','ws.audioscrobbler.com/1.0/'),$url) . '/recenttracks.rss'; + $vcard['nick'] = basename($url); + $vcard['fn'] = $vcard['nick'] . t(' on Last.fm'); + $network = NETWORK_FEED; } if(! x($vcard,'fn')) @@ -567,6 +590,12 @@ function probe_url($url, $mode = PROBE_NORMAL) { $check_feed = false; + if(stristr($url,'tumblr.com') && (! stristr($url,'/rss'))) { + $poll = $url . '/rss'; + $check_feed = true; + // Will leave it to others to figure out how to grab the avatar, which is on the $url page in the open graph meta links + } + if($twitter || ! $poll) $check_feed = true; if((! isset($vcard)) || (! x($vcard,'fn')) || (! $profile)) @@ -595,7 +624,8 @@ function probe_url($url, $mode = PROBE_NORMAL) { logger('probe_url: scrape_feed: headers: ' . $a->get_curl_headers(), LOGGER_DATA); - $feed->set_raw_data($xml); + // Don't try and parse an empty string + $feed->set_raw_data(($xml) ? $xml : ''); $feed->init(); if($feed->error())