more enotify

[friendica.git] / include / Scrape.php

diff --git a/include/Scrape.php b/include/Scrape.php
index 1c9fe7d45434d7d7a92b1183e3f7d511235e73b8..e42d22afc64c9e0a792a90bc8abcd0a6c7a8930d 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 $head0) {
@@ -332,10 +331,12 @@ function probe_url($url, $mode = PROBE_NORMAL) {
        if(! $url)
                return $result;
 
+       $network = null;
        $diaspora = false;
        $diaspora_base = '';
        $diaspora_guid = '';    
        $diaspora_key = '';
+       $has_lrdd = false;
        $email_conversant = false;
 
        $twitter = ((strpos($url,'twitter.com') !== false) ? true : false);
@@ -352,6 +353,8 @@ function probe_url($url, $mode = PROBE_NORMAL) {
                        $links = lrdd($url);
 
                if(count($links)) {
+                       $has_lrdd = true;
+
                        logger('probe_url: found lrdd links: ' . print_r($links,true), LOGGER_DATA);
                        foreach($links as $link) {
                                if($link['@attributes']['rel'] === NAMESPACE_ZOT)
@@ -366,6 +369,8 @@ function probe_url($url, $mode = PROBE_NORMAL) {
                                        $hcard = unamp($link['@attributes']['href']);
                                if($link['@attributes']['rel'] === 'http://webfinger.net/rel/profile-page')
                                        $profile = unamp($link['@attributes']['href']);
+                               if($link['@attributes']['rel'] === 'http://portablecontacts.net/spec/1.0')
+                                       $poco = unamp($link['@attributes']['href']);
                                if($link['@attributes']['rel'] === 'http://joindiaspora.com/seed_location') {
                                        $diaspora_base = unamp($link['@attributes']['href']);
                                        $diaspora = true;
@@ -426,7 +431,8 @@ function probe_url($url, $mode = PROBE_NORMAL) {
                                                $addr = $orig_url;
                                                $network = NETWORK_MAIL;
                                                $name = substr($url,0,strpos($url,'@'));
-                                               $profile = 'http://' . substr($url,strpos($url,'@')+1);
+                                               $phost = substr($url,strpos($url,'@')+1);
+                                               $profile = 'http://' . $phost;
                                                // fix nick character range
                                                $vcard = array('fn' => $name, 'nick' => $name, 'photo' => gravatar_img($url));
                                                $notify = 'smtp ' . random_string();
@@ -437,8 +443,15 @@ function probe_url($url, $mode = PROBE_NORMAL) {
                                                        $adr = imap_rfc822_parse_adrlist($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);
+                                               if(isset($adr)) {
+                                                       foreach($adr as $feadr) {
+                                                               if((strcasecmp($feadr->mailbox,$name) == 0) 
+                                                                       &&(strcasecmp($feadr->host,$phost) == 0) 
+                                                                       && (strlen($feadr->personal))) {
+                                                                       $vcard['fn'] = notags($feadr->personal);
+                                                               }
+                                                       }
+                                               }
                                        }
                                        imap_close($mbox);
                                }
@@ -467,7 +480,7 @@ function probe_url($url, $mode = PROBE_NORMAL) {
                }
 
                if(strlen($dfrn)) {
-                       $ret = scrape_dfrn($dfrn);
+                       $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn);
                        if(is_array($ret) && x($ret,'dfrn-request')) {
                                $network = NETWORK_DFRN;
                                $request = $ret['dfrn-request'];
@@ -484,8 +497,10 @@ function probe_url($url, $mode = PROBE_NORMAL) {
        }
 
        if($diaspora && $diaspora_base && $diaspora_guid) {
-               if($mode == PROBE_DIASPORA || ! $notify)
-                       $notify = $diaspora_base . 'receive/post/' . $diaspora_guid;
+               if($mode == PROBE_DIASPORA || ! $notify) {
+                       $notify = $diaspora_base . 'receive/users/' . $diaspora_guid;
+                       $batch  = $diaspora_base . 'receive/public' ;
+               }
                if(strpos($url,'@'))
                        $addr = str_replace('acct:', '', $url);
        }                       
@@ -493,7 +508,7 @@ function probe_url($url, $mode = PROBE_NORMAL) {
        if($network !== NETWORK_ZOT && $network !== NETWORK_DFRN && $network !== NETWORK_MAIL) {
                if($diaspora)
                        $network = NETWORK_DIASPORA;
-               else
+               elseif($has_lrdd)
                        $network  = NETWORK_OSTATUS;
                $priority = 0;
 
@@ -520,6 +535,9 @@ 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($vcard,'fn'))
@@ -530,7 +548,7 @@ function probe_url($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)))
                        $check_feed = false;
@@ -637,7 +655,7 @@ function probe_url($url, $mode = PROBE_NORMAL) {
                                        $vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' ')));
                        }
                        if(! $network)
-                               $network = 'feed';
+                               $network = NETWORK_FEED;
                        if(! $priority)
                                $priority = 2;
                }
@@ -651,18 +669,24 @@ function probe_url($url, $mode = PROBE_NORMAL) {
        if(! $profile)
                $profile = $url;
 
-       $vcard['fn'] = notags($vcard['fn']);
-       $vcard['nick'] = str_replace(' ','',notags($vcard['nick']));
+       // No human could be associated with this link, use the URL as the contact name
 
+       if(($network === NETWORK_FEED) && ($poll) && (! x($vcard,'fn')))
+               $vcard['fn'] = $url;
 
+       $vcard['fn'] = notags($vcard['fn']);
+       $vcard['nick'] = str_replace(' ','',notags($vcard['nick']));
+               
        $result['name'] = $vcard['fn'];
        $result['nick'] = $vcard['nick'];
        $result['url'] = $profile;
        $result['addr'] = $addr;
+       $result['batch'] = $batch;
        $result['notify'] = $notify;
        $result['poll'] = $poll;
        $result['request'] = $request;
        $result['confirm'] = $confirm;
+       $result['poco'] = $poco;
        $result['photo'] = $vcard['photo'];
        $result['priority'] = $priority;
        $result['network'] = $network;