Changeset 3521:b8a16a351d6d for inc

Timestamp:

01/20/17 14:14:22 (9 years ago)

Author:

franck <carnet.franck.paul@…>

Branch:

default

Message:

backlinks:

• retrieving ping URLs, let trackback first, then pingback, then finally webmention
• get source post content to compose webmention excerpt and retrieve title
• use source post title as blog name if this one is unknown (Anonymous blog is used if neither title nor blog name are known)

File:

• inc/core/class.dc.trackback.php (modified) (7 diffs)

inc/core/class.dc.trackback.php

@@ -356 +356 @@
      /**
      Receives a webmention and insert it as a comment of given post.
-     
+
      NB: plugin Fair Trackback check source content to find url.
 
@@ -387 +387 @@
 
                # Create a comment for received webmention
-               // maybe better to try to fetch a title from the source ?...
-               $title = __('A website mention this entry.');
-               $comment = '';
-               $excerpt = sprintf('<a href="%s" rel="nofollow">%s</a>',$from_url,$from_url);
+               $remote_content = $this->getRemoteContent($from_url);
+
+               # We want a title...
+               if (!preg_match('!<title>([^<].*?)</title>!mis',$remote_content,$m)) {
+                    throw new Exception(__('Where\'s your title?'), 0);
+               }
+               $title = trim(html::clean($m[1]));
+               $title = html::decodeEntities($title);
+               $title = html::escapeHTML($title);
+               $title = text::cutString($title,60);
+
+               preg_match('!<body[^>]*?>(.*)?</body>!msi',$remote_content,$m);
+               $source = $m[1];
+               $source = preg_replace('![\r\n\s]+!ms',' ',$source);
+               $source = preg_replace( "/<\/*(h\d|p|th|td|li|dt|dd|pre|caption|input|textarea|button)[^>]*>/","\n\n",$source );
+               $source = strip_tags($source,'<a>');
+               $source = explode("\n\n",$source);
+
+               $excerpt = '';
+               foreach ($source as $line) {
+                    if (strpos($line, $to_url) !== false) {
+                         if (preg_match("!<a[^>]+?".$to_url."[^>]*>([^>]+?)</a>!",$line,$m)) {
+                              $excerpt = strip_tags($line);
+                              break;
+                         }
+                    }
+               }
+               if ($excerpt) {
+                    $excerpt = '(&#8230;) '.text::cutString(html::escapeHTML($excerpt),200).' (&#8230;)';
+               }
+               else {
+                    $excerpt = '(&#8230;)';
+               }
+
                $this->addBacklink($post_id,$from_url,'',$title,$excerpt,$comment);
 
@@ -443 +473 @@
      {
           if (empty($blog_name)) {
-               $blog_name = 'Anonymous blog';
+               // Let use title as text link for this backlink
+               $blog_name = ($title ?: 'Anonymous blog');
           }
 
@@ -522 +553 @@
 
      /**
-     Retreive local post from a given IRL
+     Retreive local post from a given URL
 
      @param    to_url         <b>string</b>       Target URL
@@ -692 +723 @@
           }
 
-          # If we've got a X-Pingback header and it's a valid URL, it will be enough
-          if ($pb_url && filter_var($pb_url,FILTER_VALIDATE_URL) && preg_match('!^https?:!',$pb_url)) {
-               return $pb_url.'|'.$url;
-          }
-
-          # No X-Pingback header. A link rel=pingback, maybe ?
-          $pattern_pingback = '!<link rel="pingback" href="(.*?)"( /)?>!msi';
-
-          if (preg_match($pattern_pingback,$page_content,$m)) {
-               $pb_url = $m[1];
-               if (filter_var($pb_url,FILTER_VALIDATE_URL) && preg_match('!^https?:!',$pb_url)) {
-                    return $pb_url.'|'.$url;
-               }
-          }
-
-          # No pingback ? OK, let's check for a trackback data chunk...
+          # Let's check for an elderly trackback data chunk...
           $pattern_rdf =
           '/<rdf:RDF.*?>.*?'.
@@ -730 +746 @@
           }
 
+          # No trackback ? OK, let see if we've got a X-Pingback header and it's a valid URL, it will be enough
+          if ($pb_url && filter_var($pb_url,FILTER_VALIDATE_URL) && preg_match('!^https?:!',$pb_url)) {
+               return $pb_url.'|'.$url;
+          }
+
+          # No X-Pingback header. A link rel=pingback, maybe ?
+          $pattern_pingback = '!<link rel="pingback" href="(.*?)"( /)?>!msi';
+
+          if (preg_match($pattern_pingback,$page_content,$m)) {
+               $pb_url = $m[1];
+               if (filter_var($pb_url,FILTER_VALIDATE_URL) && preg_match('!^https?:!',$pb_url)) {
+                    return $pb_url.'|'.$url;
+               }
+          }
+
           # Nothing, let's try webmention. Only support x/html content
           if ($wm_url) {
@@ -792 +823 @@
      public static function checkURLs($from_url,$to_url)
      {
-          if (!(filter_var($from_url, FILTER_VALIDATE_URL) && preg_match('!^https?://!',$from_url))) {
+          if (!(filter_var($from_url,FILTER_VALIDATE_URL) && preg_match('!^https?://!',$from_url))) {
                throw new Exception(__('No valid source URL provided? Try again!'), 0);
           }
 
-          if (!(filter_var($to_url, FILTER_VALIDATE_URL) && preg_match('!^https?://!',$to_url))) {
+          if (!(filter_var($to_url,FILTER_VALIDATE_URL) && preg_match('!^https?://!',$to_url))) {
                throw new Exception(__('No valid target URL provided? Try again!'), 0);
           }