]> git.wh0rd.org - tt-rss.git/blobdiff - include/rssfuncs.php
update stored article after filter has finished processing to prevent plugins from...
[tt-rss.git] / include / rssfuncs.php
index c8aa7dc23786c0ed368be73b4a68314a761f0ee3..f657f8413fffe379bda3d14385ebc57ea02b453d 100644 (file)
@@ -2,6 +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);
 
        function update_feedbrowser_cache() {
 
@@ -79,7 +80,7 @@
                        $login_thresh_qpart = "";
                }
 
-               // Test if the feed need a update (update interval exceded).
+               // Test if the feed need a update (update interval exceeded).
                if (DB_TYPE == "pgsql") {
                        $update_limit_qpart = "AND ((
                                        ttrss_feeds.update_interval = 0
                                ) OR (
                                        ttrss_feeds.update_interval > 0
                                        AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
-                               ) OR ttrss_feeds.last_updated IS NULL
-                               OR last_updated = '1970-01-01 00:00:00')";
+                               ) OR (ttrss_feeds.last_updated IS NULL
+                                       AND ttrss_user_prefs.value != '-1')
+                               OR (last_updated = '1970-01-01 00:00:00'
+                                       AND ttrss_user_prefs.value != '-1'))";
                } else {
                        $update_limit_qpart = "AND ((
                                        ttrss_feeds.update_interval = 0
                                ) OR (
                                        ttrss_feeds.update_interval > 0
                                        AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
-                               ) OR ttrss_feeds.last_updated IS NULL
-                               OR last_updated = '1970-01-01 00:00:00')";
+                               ) OR (ttrss_feeds.last_updated IS NULL
+                                       AND ttrss_user_prefs.value != '-1')
+                               OR (last_updated = '1970-01-01 00:00:00'
+                                       AND ttrss_user_prefs.value != '-1'))";
                }
 
                // Test if feed is currently being updated by another process.
                                ttrss_feeds, ttrss_users, ttrss_user_prefs
                        WHERE
                                ttrss_feeds.owner_uid = ttrss_users.id
+                               AND ttrss_user_prefs.profile IS NULL
                                AND ttrss_users.id = ttrss_user_prefs.owner_uid
                                AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
                                $login_thresh_qpart $update_limit_qpart
                                ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
                                ttrss_users.id = ttrss_user_prefs.owner_uid AND
                                ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
+                               ttrss_user_prefs.profile IS NULL AND
                                feed_url = '".db_escape_string($feed)."' AND
                                (ttrss_feeds.update_interval > 0 OR
                                        ttrss_user_prefs.value != '-1')
                        ORDER BY ttrss_feeds.id $query_limit");
 
                        if (db_num_rows($tmp_result) > 0) {
+                               $rss = false;
+
                                while ($tline = db_fetch_assoc($tmp_result)) {
                                        if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
-                                       update_rss_feed($tline["id"], true);
+
+                                       $rss = update_rss_feed($tline["id"], true, false);
                                        _debug_suppress(false);
                                        ++$nf;
                                }
        } // function update_daemon_common
 
        // ignore_daemon is not used
-       function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false) {
+       function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
 
                $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
 
                _debug("start", $debug_enabled);
 
                $result = db_query("SELECT id,update_interval,auth_login,
-                       feed_url,auth_pass,cache_images,last_updated,
+                       feed_url,auth_pass,cache_images,
                        mark_unread_on_update, owner_uid,
                        pubsub_state, auth_pass_encrypted,
                        (SELECT max(date_entered) FROM
                        return false;
                }
 
-               $last_updated = db_fetch_result($result, 0, "last_updated");
                $last_article_timestamp = @strtotime(db_fetch_result($result, 0, "last_article_timestamp"));
 
                if (defined('_DISABLE_HTTP_304'))
                $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
                $pluginhost->load_data();
 
-               $rss = false;
-               $rss_hash = false;
-
-               $force_refetch = isset($_REQUEST["force_refetch"]);
+               if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
+                       _debug("using previously initialized parser object");
+               } else {
+                       $rss_hash = false;
 
-               if (file_exists($cache_filename) &&
-                       is_readable($cache_filename) &&
-                       !$auth_login && !$auth_pass &&
-                       filemtime($cache_filename) > time() - 30) {
+                       $force_refetch = isset($_REQUEST["force_refetch"]);
 
-                       _debug("using local cache.", $debug_enabled);
+                       foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
+                               $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, $last_article_timestamp, $auth_login, $auth_pass);
+                       }
 
-                       @$feed_data = file_get_contents($cache_filename);
+                       // try cache
+                       if (!$feed_data &&
+                               file_exists($cache_filename) &&
+                               is_readable($cache_filename) &&
+                               !$auth_login && !$auth_pass &&
+                               filemtime($cache_filename) > time() - 30) {
 
-                       if ($feed_data) {
-                               $rss_hash = sha1($feed_data);
-                       }
+                               _debug("using local cache [$cache_filename].", $debug_enabled);
 
-               } else {
-                       _debug("local cache will not be used for this feed", $debug_enabled);
-               }
+                               @$feed_data = file_get_contents($cache_filename);
 
-               if (!$rss) {
+                               if ($feed_data) {
+                                       $rss_hash = sha1($feed_data);
+                               }
 
-                       foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
-                               $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed);
+                       } else {
+                               _debug("local cache will not be used for this feed", $debug_enabled);
                        }
 
+                       // fetch feed from source
                        if (!$feed_data) {
                                _debug("fetching [$fetch_url]...", $debug_enabled);
                                _debug("If-Modified-Since: ".gmdate('D, d M Y H:i:s \G\M\T', $last_article_timestamp), $debug_enabled);
                                                }
                                        }
                                } */
+
+                               // cache vanilla feed data for re-use
+                               if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
+                                       $new_rss_hash = sha1($feed_data);
+
+                                       if ($new_rss_hash != $rss_hash) {
+                                               _debug("saving $cache_filename", $debug_enabled);
+                                               @file_put_contents($cache_filename, $feed_data);
+                                       }
+                               }
                        }
 
                        if (!$feed_data) {
                        $rss->init();
                }
 
-               require_once "lib/languagedetect/LanguageDetect.php";
+               if (DETECT_ARTICLE_LANGUAGE) {
+                       require_once "lib/languagedetect/LanguageDetect.php";
 
-               $lang = new Text_LanguageDetect();
-               $lang->setNameMode(2);
+                       $lang = new Text_LanguageDetect();
+                       $lang->setNameMode(2);
+               }
 
 //             print_r($rss);
 
 
                if (!$rss->error()) {
 
-                       // cache data for later
-                       if (!$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
-                               $new_rss_hash = sha1($rss_data);
-
-                               if ($new_rss_hash != $rss_hash && count($rss->get_items()) > 0 ) {
-                                       _debug("saving $cache_filename", $debug_enabled);
-                                       @file_put_contents($cache_filename, $feed_data);
-                               }
-                       }
-
                        // We use local pluginhost here because we need to load different per-user feed plugins
                        $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
 
 
                                _debug("feed hub url: $feed_hub_url", $debug_enabled);
 
-                               if ($feed_hub_url && function_exists('curl_init') &&
+                               $feed_self_url = $fetch_url;
+
+                               $links = $rss->get_links('self');
+
+                               if ($links && is_array($links)) {
+                                       foreach ($links as $l) {
+                                               $feed_self_url = $l;
+                                               break;
+                                       }
+                               }
+
+                               _debug("feed self url = $feed_self_url");
+
+                               if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
                                        !ini_get("open_basedir")) {
 
                                        require_once 'lib/pubsubhubbub/subscriber.php';
 
                                        $s = new Subscriber($feed_hub_url, $callback_url);
 
-                                       $rc = $s->subscribe($fetch_url);
+                                       $rc = $s->subscribe($feed_self_url);
 
-                                       _debug("feed hub url found, subscribe request sent.", $debug_enabled);
+                                       _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
 
                                        db_query("UPDATE ttrss_feeds SET pubsub_state = 1
                                                WHERE id = '$feed'");
 
                                if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
                                        $entry_timestamp = time();
-                                       $no_orig_date = 'true';
-                               } else {
-                                       $no_orig_date = 'false';
                                }
 
                                $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
                                        print "\n";
                                }
 
-                               $entry_language = $lang->detect($entry_content, 1);
+                               $entry_language = "";
+
+                               if (DETECT_ARTICLE_LANGUAGE) {
+                                       $entry_language = $lang->detect($entry_title . " " . $entry_content, 1);
+
+                                       if (count($entry_language) > 0) {
+                                               $entry_language = array_keys($entry_language);
 
-                               if (count($entry_language) > 0) {
-                                       $entry_language = array_keys($entry_language);
-                                       $entry_language = db_escape_string(substr($entry_language[0], 0, 2));
+                                               // the fuck?
+                                               if (is_array($entry_language))
+                                                       $entry_language = "";
+                                               else
+                                                       $entry_language = db_escape_string(substr($entry_language[0], 0, 2));
 
-                                       _debug("detected language: $entry_language", $debug_enabled);
-                               } else{
-                                       $entry_language = "";
+                                               _debug("detected language: $entry_language", $debug_enabled);
+                                       } else {
+                                               $entry_language = "";
+                                       }
                                }
 
                                $entry_comments = $item->get_comments_url();
 
                                foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
                                        $article = $plugin->hook_article_filter($article);
+
+                                       $article["stored"] = array("title" => $article["title"],
+                                               "content" => $article["content"],
+                                               "link" => $article["link"],
+                                               "tags" => $article["tags"],
+                                               "author" => $article["author"]);
                                }
 
                                $entry_tags = $article["tags"];
                                $entry_plugin_data = db_escape_string($article["plugin_data"]);
                                $entry_content = $article["content"]; // escaped below
 
-
                                _debug("plugin data: $entry_plugin_data", $debug_enabled);
 
                                if ($cache_images && is_writable(CACHE_DIR . '/images'))
                                                        '$entry_timestamp_fmt',
                                                        '$entry_content',
                                                        '$content_hash',
-                                                       $no_orig_date,
+                                                       false,
                                                        NOW(),
                                                        '$date_feed_processed',
                                                        '$entry_comments',
                                // now it should exist, if not - bad luck then
 
                                $result = db_query("SELECT
-                                               id,content_hash,no_orig_date,title,plugin_data,guid,
-                                               ".SUBSTRING_FOR_DATE."(date_updated,1,19) as date_updated,
-                                               ".SUBSTRING_FOR_DATE."(updated,1,19) as updated,
+                                               id,content_hash,title,plugin_data,guid,
                                                num_comments
                                        FROM
                                                ttrss_entries
                                        $orig_content_hash = db_fetch_result($result, 0, "content_hash");
                                        $orig_title = db_fetch_result($result, 0, "title");
                                        $orig_num_comments = db_fetch_result($result, 0, "num_comments");
-                                       $orig_date_updated = strtotime(db_fetch_result($result,
-                                               0, "date_updated"));
                                        $orig_plugin_data = db_fetch_result($result, 0, "plugin_data");
 
                                        $ref_id = db_fetch_result($result, 0, "id");
 
                                                        $p = new Publisher(PUBSUBHUBBUB_HUB);
 
-                                                       $pubsub_result = $p->publish_update($rss_link);
+                                                       /* $pubsub_result = */ $p->publish_update($rss_link);
                                                }
 
                                                $result = db_query(
                                if (is_array($encs)) {
                                        foreach ($encs as $e) {
                                                $e_item = array(
-                                                       $e->link, $e->type, $e->length, $e->title);
+                                                       $e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
                                                array_push($enclosures, $e_item);
                                        }
                                }
                                        $enc_type = db_escape_string($enc[1]);
                                        $enc_dur = db_escape_string($enc[2]);
                                        $enc_title = db_escape_string($enc[3]);
+                                       $enc_width = intval($enc[4]);
+                                       $enc_height = intval($enc[5]);
 
                                        $result = db_query("SELECT id FROM ttrss_enclosures
                                                WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
 
                                        if (db_num_rows($result) == 0) {
                                                db_query("INSERT INTO ttrss_enclosures
-                                                       (content_url, content_type, title, duration, post_id) VALUES
-                                                       ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id')");
+                                                       (content_url, content_type, title, duration, post_id, width, height) VALUES
+                                                       ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
                                        }
                                }
 
 
                        $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
 
-                       _debug("error fetching feed: $error_msg", $debug_enabled);
+                       _debug("fetch error: $error_msg", $debug_enabled);
+
+                       if (count($rss->errors()) > 1) {
+                               foreach ($rss->errors() as $error) {
+                                       _debug("+ $error");
+                               }
+                       }
 
                        db_query(
                                "UPDATE ttrss_feeds SET last_error = '$error_msg',
-                                       last_updated = NOW() WHERE id = '$feed'");
-               }
+                               last_updated = NOW() WHERE id = '$feed'");
 
-               unset($rss);
+                       unset($rss);
+               }
 
                _debug("done", $debug_enabled);
+
+               return $rss;
        }
 
        function cache_images($html, $site_url, $debug) {
-               $cache_dir = CACHE_DIR . "/images";
-
                libxml_use_internal_errors(true);
 
                $charset_hack = '<head>
                                if (!file_exists($local_filename)) {
                                        $file_content = fetch_file_contents($src);
 
-                                       if ($file_content && strlen($file_content) > 1024) {
+                                       if ($file_content && strlen($file_content) > _MIN_CACHE_IMAGE_SIZE) {
                                                file_put_contents($local_filename, $file_content);
                                        }
                                }