]> git.mxchange.org Git - friendica.git/blobdiff - include/Scrape.php
how the heck did favicon get into the repo? Gone now...
[friendica.git] / include / Scrape.php
index a6bb5f728c6d659be732ca65b7ab66439b552671..698ec9cf04c4e182ecefd98752b7d645a48215ed 100644 (file)
@@ -241,6 +241,17 @@ function scrape_feed($url) {
                }
        }
 
+
+       $head = $dom->getElementsByTagName('base');
+       if($head) {
+               foreach($head as $head0) {
+                       $basename = $head0->getAttribute('href');
+                       break;
+               }
+       }
+       if(! $basename)
+               $basename = substr($url,0,strrpos($url,'/')) . '/';
+
        $items = $dom->getElementsByTagName('link');
 
        // get Atom/RSS link elements, take the first one of either.
@@ -259,6 +270,13 @@ function scrape_feed($url) {
                }       
        }
 
+       // Drupal and perhaps others only provide relative URL's. Turn them into absolute.
+
+       if(x($ret,'feed_atom') && (! strstr($ret['feed_atom'],'://')))
+               $ret['feed_atom'] = $basename . $ret['feed_atom'];
+       if(x($ret,'feed_rss') && (! strstr($ret['feed_rss'],'://')))
+               $ret['feed_rss'] = $basename . $ret['feed_rss'];
+
        return $ret;
 }}
 
@@ -274,10 +292,13 @@ function probe_url($url) {
        $diaspora = false;      
        $email_conversant = false;
 
-       if($url) {
+       $twitter = ((strpos($url,'twitter.com') !== false) ? true : false);
+
+       if(! $twitter) {
                $links = lrdd($url);
 
                if(count($links)) {
+                       logger('probe_url: found lrdd links: ' . print_r($links,true), LOGGER_DATA);
                        foreach($links as $link) {
                                if($link['@attributes']['rel'] === NAMESPACE_DFRN)
                                        $dfrn = unamp($link['@attributes']['href']);
@@ -345,8 +366,11 @@ function probe_url($url) {
                                                $poll = 'email ' . random_string();
                                                $priority = 0;
                                                $x = email_msg_meta($mbox,$msgs[0]);
-                                               $adr = imap_rfc822_parse_adrlist($x->from,'');
-                                               if(strlen($adr[0]->personal))
+                                               if(stristr($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(isset($adr) && strlen($adr[0]->personal))
                                                        $vcard['fn'] = notags($adr[0]->personal);
                                        }
                                        imap_close($mbox);
@@ -391,76 +415,101 @@ function probe_url($url) {
                                $profile = $url;
                }
 
+               if($twitter) {          
+                       logger('twitter: setup');
+                       $tid = basename($url);
+                       $tapi = 'https://api.twitter.com/1/statuses/user_timeline.rss';
+                       if(intval($tid))
+                               $poll = $tapi . '?user_id=' . $tid;
+                       else
+                               $poll = $tapi . '?screen_name=' . $tid;
+                       $profile = 'http://twitter.com/#!/' . $tid;
+               }
+
                if(! x($vcard,'fn'))
                        if(x($vcard,'nick'))
                                $vcard['fn'] = $vcard['nick'];
 
-               if((! isset($vcard)) && (! $poll)) {
+       
+               if(((! isset($vcard)) && (! $poll)) || ($twitter)) {
 
-                       $ret = scrape_feed($url);
-                       logger('probe_url: scrape_feed returns: ' . print_r($ret,true), LOGGER_DATA);
-                       if(count($ret) && ($ret['feed_atom'] || $ret['feed_rss'])) {
-                               $poll = ((x($ret,'feed_atom')) ? unamp($ret['feed_atom']) : unamp($ret['feed_rss']));
+                       $feedret = scrape_feed($url);
+                       logger('probe_url: scrape_feed 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']));
                                $vcard = array();
-                                       if(x($ret,'photo'))
-                                               $vcard['photo'] = $ret['photo'];
-                               require_once('simplepie/simplepie.inc');
-                           $feed = new SimplePie();
-                               $xml = fetch_url($poll);
-
-                       $feed->set_raw_data($xml);
-
-                           $feed->init();
-
-                               if(! x($vcard,'photo'))
-                                       $vcard['photo'] = $feed->get_image_url();
-                               $author = $feed->get_author();
-                               if($author) {                   
-                                       $vcard['fn'] = unxmlify(trim($author->get_name()));
-                                       if(! $vcard['fn'])
-                                               $vcard['fn'] = trim(unxmlify($author->get_email()));
-                                       if(strpos($vcard['fn'],'@') !== false)
-                                               $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
-                                       $vcard['nick'] = strtolower(notags(unxmlify($vcard['fn'])));
-                                       if(strpos($vcard['nick'],' '))
-                                               $vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' ')));
-                                       $email = unxmlify($author->get_email());
-                               }
-                               else {
-                                       $item = $feed->get_item(0);
-                                       if($item) {
-                                               $author = $item->get_author();
-                                               if($author) {                   
-                                                       $vcard['fn'] = trim(unxmlify($author->get_name()));
-                                                       if(! $vcard['fn'])
-                                                               $vcard['fn'] = trim(unxmlify($author->get_email()));
-                                                       if(strpos($vcard['fn'],'@') !== false)
-                                                               $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
-                                                       $vcard['nick'] = strtolower(unxmlify($vcard['fn']));
-                                                       if(strpos($vcard['nick'],' '))
-                                                               $vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' ')));
-                                                       $email = unxmlify($author->get_email());
-                                               }
-                                               if(! $vcard['photo']) {
-                                                       $rawmedia = $item->get_item_tags('http://search.yahoo.com/mrss/','thumbnail');
-                                                       if($rawmedia && $rawmedia[0]['attribs']['']['url'])
-                                                               $vcard['photo'] = unxmlify($rawmedia[0]['attribs']['']['url']);
-                                               }
+                       }
+
+                       if(x($feedret,'photo'))
+                               $vcard['photo'] = $feedret['photo'];
+                       require_once('library/simplepie/simplepie.inc');
+                   $feed = new SimplePie();
+                       $xml = fetch_url($poll);
+
+                       logger('probe_url: fetch feed: ' . $poll . ' returns: ' . $xml, LOGGER_DATA);
+                       $a = get_app();
+
+                       logger('probe_url: scrape_feed: headers: ' . $a->get_curl_headers(), $LOGGER_DATA);
+
+                       $feed->set_raw_data($xml);
+
+                   $feed->init();
+                       if($feed->error())
+                               logger('probe_url: scrape_feed: Error parsing XML: ' . $feed->error());
+
+                       if(! x($vcard,'photo'))
+                               $vcard['photo'] = $feed->get_image_url();
+                       $author = $feed->get_author();
+                       if($author) {                   
+                               $vcard['fn'] = unxmlify(trim($author->get_name()));
+                               if(! $vcard['fn'])
+                                       $vcard['fn'] = trim(unxmlify($author->get_email()));
+                               if(strpos($vcard['fn'],'@') !== false)
+                                       $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
+                               $email = unxmlify($author->get_email());
+                       }
+                       else {
+                               $item = $feed->get_item(0);
+                               if($item) {
+                                       $author = $item->get_author();
+                                       if($author) {                   
+                                               $vcard['fn'] = trim(unxmlify($author->get_name()));
+                                               if(! $vcard['fn'])
+                                                       $vcard['fn'] = trim(unxmlify($author->get_email()));
+                                               if(strpos($vcard['fn'],'@') !== false)
+                                                       $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
+                                               $email = unxmlify($author->get_email());
+                                       }
+                                       if(! $vcard['photo']) {
+                                               $rawmedia = $item->get_item_tags('http://search.yahoo.com/mrss/','thumbnail');
+                                               if($rawmedia && $rawmedia[0]['attribs']['']['url'])
+                                                       $vcard['photo'] = unxmlify($rawmedia[0]['attribs']['']['url']);
                                        }
                                }
-                               if((! $vcard['photo']) && strlen($email))
-                                       $vcard['photo'] = gravatar_img($email);
-                               if($poll === $profile)
-                                       $lnk = $feed->get_permalink();
-                               if(isset($lnk) && strlen($lnk))
-                                       $profile = $lnk;        
-                               if(! (x($vcard,'fn')))
-                                       $vcard['fn'] = notags($feed->get_title());
-                               if(! (x($vcard,'fn')))
-                                       $vcard['fn'] = notags($feed->get_description());
-                               $network = 'feed';
-                               $priority = 2;
                        }
+                       if((! $vcard['photo']) && strlen($email))
+                               $vcard['photo'] = gravatar_img($email);
+                       if($poll === $profile)
+                               $lnk = $feed->get_permalink();
+                       if(isset($lnk) && strlen($lnk))
+                               $profile = $lnk;        
+
+                       if(! (x($vcard,'fn')))
+                               $vcard['fn'] = notags($feed->get_title());
+                       if(! (x($vcard,'fn')))
+                               $vcard['fn'] = notags($feed->get_description());
+
+                       if(strpos($vcard['fn'],'Twitter / ') !== false) {
+                               $vcard['fn'] = substr($vcard['fn'],strpos($vcard['fn'],'/')+1);
+                               $vcard['fn'] = trim($vcard['fn']);
+                       }
+                       if(! x($vcard,'nick')) {
+                               $vcard['nick'] = strtolower(notags(unxmlify($vcard['fn'])));
+                               if(strpos($vcard['nick'],' '))
+                                       $vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' ')));
+                       }
+                       $network = 'feed';
+                       $priority = 2;
                }
        }