]> git.wh0rd.org - tt-rss.git/blobdiff - include/rssfuncs.php
enable caching of media in article enclosures
[tt-rss.git] / include / rssfuncs.php
index fbb26ef9086cfe5411347e6453d1d248a1b0eda0..c03e6681e0e1b9c2128f265e2aa06948f16eb7a9 100644 (file)
@@ -2,7 +2,7 @@
        define_default('DAEMON_UPDATE_LOGIN_LIMIT', 30);
        define_default('DAEMON_FEED_LIMIT', 500);
        define_default('DAEMON_SLEEP_INTERVAL', 120);
-       define_default('_MIN_CACHE_IMAGE_SIZE', 1024);
+       define_default('_MIN_CACHE_FILE_SIZE', 1024);
 
        function calculate_article_hash($article, $pluginhost) {
                $tmp = "";
@@ -23,9 +23,8 @@
        function update_feedbrowser_cache() {
 
                $result = db_query("SELECT feed_url, site_url, title, COUNT(id) AS subscribers
-                       FROM ttrss_feeds WHERE (SELECT COUNT(id) = 0 FROM ttrss_feeds AS tf
-                               WHERE tf.feed_url = ttrss_feeds.feed_url
-                               AND (private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%'))
+                       FROM ttrss_feeds WHERE feed_url NOT IN (SELECT feed_url FROM ttrss_feeds
+                               WHERE private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%')
                                GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
 
                db_query("BEGIN");
                                        $callback_url = get_self_url_prefix() .
                                                "/public.php?op=pubsub&id=$feed";
 
-                                       $s = new Subscriber($feed_hub_url, $callback_url);
+                                       $s = new Pubsubhubbub\Subscriber\Subscriber($feed_hub_url, $callback_url);
 
                                        $rc = $s->subscribe($feed_self_url);
 
                                _debug("force catchup: $entry_force_catchup");
 
                                if ($cache_images && is_writable(CACHE_DIR . '/images'))
-                                       cache_images($entry_content, $site_url, $debug_enabled);
+                                       cache_media($entry_content, $site_url, $debug_enabled);
 
                                $entry_content = db_escape_string($entry_content, false);
 
                                                                "/public.php?op=rss&id=-2&key=" .
                                                                get_feed_access_key(-2, false, $owner_uid);
 
-                                                       $p = new Publisher(PUBSUBHUBBUB_HUB);
+                                                       $p = new pubsubhubbub\publisher\Publisher(PUBSUBHUBBUB_HUB);
 
                                                        /* $pubsub_result = */ $p->publish_update($rss_link);
                                                }
                                        }
                                }
 
+                               if ($cache_images && is_writable(CACHE_DIR . '/images'))
+                                       cache_enclosures($enclosures, $site_url, $debug_enabled);
+
                                if ($debug_enabled) {
                                        _debug("article enclosures:", $debug_enabled);
                                        print_r($enclosures);
                return $rss;
        }
 
-       function cache_images($html, $site_url, $debug) {
+       function cache_enclosures($enclosures, $site_url, $debug) {
+               foreach ($enclosures as $enc) {
+
+                       if (preg_match("/(image|audio|video)/", $enc[1])) {
+
+                               $src = rewrite_relative_url($site_url, $enc[0]);
+
+                               $local_filename = CACHE_DIR . "/images/" . sha1($src);
+
+                               if ($debug) _debug("cache_enclosures: downloading: $src to $local_filename");
+
+                               if (!file_exists($local_filename)) {
+                                       $file_content = fetch_file_contents($src);
+
+                                       if ($file_content && strlen($file_content) > _MIN_CACHE_FILE_SIZE) {
+                                               file_put_contents($local_filename, $file_content);
+                                       }
+                               } else {
+                                       touch($local_filename);
+                               }
+                       }
+               }
+       }
+
+       function cache_media($html, $site_url, $debug) {
                libxml_use_internal_errors(true);
 
                $charset_hack = '<head>
                $doc->loadHTML($charset_hack . $html);
                $xpath = new DOMXPath($doc);
 
-               $entries = $xpath->query('(//img[@src])');
+               $entries = $xpath->query('(//img[@src])|(//video/source[@src])|(//audio/source[@src])');
 
                foreach ($entries as $entry) {
-                       if ($entry->hasAttribute('src')) {
+                       if ($entry->hasAttribute('src') && strpos($entry->getAttribute('src'), "data:") !== 0) {
                                $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
 
-                               $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
+                               $local_filename = CACHE_DIR . "/images/" . sha1($src);
 
-                               if ($debug) _debug("cache_images: downloading: $src to $local_filename");
+                               if ($debug) _debug("cache_media: downloading: $src to $local_filename");
 
                                if (!file_exists($local_filename)) {
                                        $file_content = fetch_file_contents($src);
 
-                                       if ($file_content && strlen($file_content) > _MIN_CACHE_IMAGE_SIZE) {
+                                       if ($file_content && strlen($file_content) > _MIN_CACHE_FILE_SIZE) {
                                                file_put_contents($local_filename, $file_content);
                                        }
                                } else {