Merge pull request #1044 from annando/master

[friendica.git] / include / Scrape.php

diff --git a/include/Scrape.php b/include/Scrape.php
index 7eaac3b44924669981b072bfe7347bf6c9f1b170..a585047ebdcdc2d98d0cf9890ae5e6de6b9e2a9e 100644 (file)
--- a/include/Scrape.php
+++ b/include/Scrape.php
@@ -14,7 +14,7 @@ function scrape_dfrn($url) {
 
        $s = fetch_url($url);
 
-       if(! $s) 
+       if(! $s)
                return $ret;
 
        $headers = $a->get_curl_headers();
@@ -23,7 +23,7 @@ function scrape_dfrn($url) {
 
        $lines = explode("\n",$headers);
        if(count($lines)) {
-               foreach($lines as $line) {                              
+               foreach($lines as $line) {
                        // don't try and run feeds through the html5 parser
                        if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
                                return ret;
@@ -346,12 +346,13 @@ 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);
+       // Twitter is deactivated since twitter closed its old API
+       //$twitter = ((strpos($url,'twitter.com') !== false) ? true : false);
        $lastfm  = ((strpos($url,'last.fm/user') !== false) ? true : false);
 
        $at_addr = ((strpos($url,'@') !== false) ? true : false);
@@ -485,7 +486,7 @@ function probe_url($url, $mode = PROBE_NORMAL) {
                                }
                        }
                }
-       }       
+       }
 
        if($mode == PROBE_NORMAL) {
                if(strlen($zot)) {
@@ -531,7 +532,7 @@ 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)
@@ -544,13 +545,13 @@ 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);
                }
 
@@ -561,7 +562,7 @@ function probe_url($url, $mode = PROBE_NORMAL) {
                        $vcard['nick'] = $addr_parts[0];
                }
 
-               if($twitter) {          
+               /* if($twitter) {
                        logger('twitter: setup');
                        $tid = basename($url);
                        $tapi = 'https://api.twitter.com/1/statuses/user_timeline.rss';
@@ -574,7 +575,7 @@ function probe_url($url, $mode = PROBE_NORMAL) {
                        $vcard['photo'] = 'https://api.twitter.com/1/users/profile_image?screen_name=' . $tid . '&size=bigger';
                        $vcard['nick'] = $tid;
                        $vcard['fn'] = $tid;
-               }
+               } */
 
                if($lastfm) {
                        $profile = $url;
@@ -590,6 +591,12 @@ function probe_url($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')) || (! $profile))
@@ -603,7 +610,7 @@ function probe_url($url, $mode = PROBE_NORMAL) {
                        logger('probe_url: scrape_feed ' . (($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']));
-                               if(! x($vcard)) 
+                               if(! x($vcard))
                                        $vcard = array();
                        }
 
@@ -618,7 +625,8 @@ 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 : '<?xml version="1.0" encoding="utf-8" ?><xml></xml>');
 
                    $feed->init();
                        if($feed->error())