fix various html parse errors

[friendica.git] / include / Scrape.php

diff --git a/include/Scrape.php b/include/Scrape.php
index 58468a40d021ab6de2f008bd704cffbbe1233cfa..f5b15c82fbeb7eafef5c0dc58a2e1502f59db89e 100644 (file)
--- a/include/Scrape.php
+++ b/include/Scrape.php
@@ -30,8 +30,11 @@ function scrape_dfrn($url) {
                }
        }
 
-
-       $dom = HTML5_Parser::parse($s);
+       try {
+               $dom = HTML5_Parser::parse($s);
+       } catch (DOMException $e) {
+               logger('scrape_dfrn: parse error: ' . $e);
+       }
 
        if(! $dom)
                return $ret;
@@ -132,9 +135,11 @@ function scrape_meta($url) {
                }
        }
 
-
-
-       $dom = HTML5_Parser::parse($s);
+       try {
+               $dom = HTML5_Parser::parse($s);
+       } catch (DOMException $e) {
+               logger('scrape_meta: parse error: ' . $e);
+       }
 
        if(! $dom)
                return $ret;
@@ -177,7 +182,11 @@ function scrape_vcard($url) {
                }
        }
 
-       $dom = HTML5_Parser::parse($s);
+       try {
+               $dom = HTML5_Parser::parse($s);
+       } catch (DOMException $e) {
+               logger('scrape_vcard: parse error: ' . $e);
+       }
 
        if(! $dom)
                return $ret;
@@ -243,26 +252,16 @@ function scrape_feed($url) {
                }
        }
 
-       $dom = HTML5_Parser::parse($s);
+       try {
+               $dom = HTML5_Parser::parse($s);
+       } catch (DOMException $e) {
+               logger('scrape_feed: parse error: ' . $e);
+       }
 
        if(! $dom)
                return $ret;
 
 
-       $items = $dom->getElementsByTagName('img');
-
-       // get img elements (twitter)
-
-       if($items) {
-               foreach($items as $item) {
-                       $x = $item->getAttribute('id');
-                       if($x === 'profile-image') {
-                               $ret['photo'] = $item->getAttribute('src');
-                       }
-               }
-       }
-
-
        $head = $dom->getElementsByTagName('base');
        if($head) {
                foreach($head as $head0) {
@@ -496,8 +495,10 @@ function probe_url($url, $mode = PROBE_NORMAL) {
        }
 
        if($diaspora && $diaspora_base && $diaspora_guid) {
-               if($mode == PROBE_DIASPORA || ! $notify)
+               if($mode == PROBE_DIASPORA || ! $notify) {
                        $notify = $diaspora_base . 'receive/users/' . $diaspora_guid;
+                       $batch  = $diaspora_base . 'receive/public' ;
+               }
                if(strpos($url,'@'))
                        $addr = str_replace('acct:', '', $url);
        }                       
@@ -532,6 +533,9 @@ function probe_url($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['nick'] = $tid;
+                       $vcard['fn'] = $tid . '@twitter';
                }
 
                if(! x($vcard,'fn'))
@@ -542,7 +546,7 @@ function probe_url($url, $mode = PROBE_NORMAL) {
 
                if($twitter || ! $poll)
                        $check_feed = true;
-               if((! isset($vcard)) || (! $profile))
+               if((! isset($vcard)) || (! x($vcard,'fn')) || (! $profile))
                        $check_feed = true;
                if(($at_addr) && (! count($links)))
                        $check_feed = false;
@@ -675,6 +679,7 @@ function probe_url($url, $mode = PROBE_NORMAL) {
        $result['nick'] = $vcard['nick'];
        $result['url'] = $profile;
        $result['addr'] = $addr;
+       $result['batch'] = $batch;
        $result['notify'] = $notify;
        $result['poll'] = $poll;
        $result['request'] = $request;