]> git.wh0rd.org - tt-rss.git/blobdiff - include/rssfuncs.php
move counter stuff to a separate class
[tt-rss.git] / include / rssfuncs.php
index c03e6681e0e1b9c2128f265e2aa06948f16eb7a9..33b6ff6e4018f1df17948ce158ecb36c2c37ec85 100644 (file)
@@ -4,6 +4,8 @@
        define_default('DAEMON_SLEEP_INTERVAL', 120);
        define_default('_MIN_CACHE_FILE_SIZE', 1024);
 
+       // TODO: this needs to be removed from global namespace into classes/RSS.php or something
+
        function calculate_article_hash($article, $pluginhost) {
                $tmp = "";
 
 
        }
 
-
-       /**
-        * Update a feed batch.
-        * Used by daemons to update n feeds by run.
-        * Only update feed needing a update, and not being processed
-        * by another process.
-        *
-        * @param mixed $link Database link
-        * @param integer $limit Maximum number of feeds in update batch. Default to DAEMON_FEED_LIMIT.
-        * @param boolean $from_http Set to true if you call this function from http to disable cli specific code.
-        * @param boolean $debug Set to false to disable debug output. Default to true.
-        * @return void
-        */
-       function update_daemon_common($limit = DAEMON_FEED_LIMIT, $from_http = false, $debug = true) {
+       function update_daemon_common($limit = DAEMON_FEED_LIMIT, $debug = true) {
                // Process all other feeds using last_updated and interval parameters
 
                $schema_version = get_schema_version();
                        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"]);
 
                                                array_push($batch_owners, $tline["owner_uid"]);
 
                                        $fstarted = microtime(true);
-                                       $rss = update_rss_feed($tline["id"], true, false);
+                                       update_rss_feed($tline["id"], true, false);
                                        _debug_suppress(false);
 
                                        _debug(sprintf("    %.4f (sec)", microtime(true) - $fstarted));
                        housekeeping_user($owner_uid);
                }
 
-               require_once "digest.php";
-
                // Send feed digests by email if needed.
-               send_headlines_digests($debug);
+               Digest::send_headlines_digests($debug);
 
                return $nf;
 
-       } // function update_daemon_common
+       }
 
        // this is used when subscribing
        function set_basic_feed_info($feed) {
                }
        }
 
-       // ignore_daemon is not used
-       function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
+       /**
+         * @SuppressWarnings(PHPMD.UnusedFormalParameter)
+       */
+       function update_rss_feed($feed, $no_cache = false) {
 
                $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
 
                $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
                $pluginhost->load_data();
 
-               if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
-                       _debug("using previously initialized parser object");
-               } else {
-                       $rss_hash = false;
+               $rss_hash = false;
 
-                       $force_refetch = isset($_REQUEST["force_refetch"]);
-                       $feed_data = "";
+               $force_refetch = isset($_REQUEST["force_refetch"]);
+               $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, 0, $auth_login, $auth_pass);
-                       }
+               foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
+                       $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, 0, $auth_login, $auth_pass);
+               }
 
-                       // try cache
-                       if (!$feed_data &&
-                               file_exists($cache_filename) &&
-                               is_readable($cache_filename) &&
-                               !$auth_login && !$auth_pass &&
-                               filemtime($cache_filename) > time() - 30) {
+               // try cache
+               if (!$feed_data &&
+                       file_exists($cache_filename) &&
+                       is_readable($cache_filename) &&
+                       !$auth_login && !$auth_pass &&
+                       filemtime($cache_filename) > time() - 30) {
 
-                               _debug("using local cache [$cache_filename].", $debug_enabled);
+                       _debug("using local cache [$cache_filename].", $debug_enabled);
 
-                               @$feed_data = file_get_contents($cache_filename);
+                       @$feed_data = file_get_contents($cache_filename);
 
-                               if ($feed_data) {
-                                       $rss_hash = sha1($feed_data);
-                               }
-
-                       } else {
-                               _debug("local cache will not be used for this feed", $debug_enabled);
+                       if ($feed_data) {
+                               $rss_hash = sha1($feed_data);
                        }
 
-                       // fetch feed from source
-                       if (!$feed_data) {
-                               _debug("fetching [$fetch_url]...", $debug_enabled);
+               } else {
+                       _debug("local cache will not be used for this feed", $debug_enabled);
+               }
 
-                               if (ini_get("open_basedir") && function_exists("curl_init")) {
-                                       _debug("not using CURL due to open_basedir restrictions");
-                               }
+               // fetch feed from source
+               if (!$feed_data) {
+                       _debug("fetching [$fetch_url]...", $debug_enabled);
 
-                               $feed_data = fetch_file_contents($fetch_url, false,
-                                       $auth_login, $auth_pass, false,
-                                       $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
-                                       0);
+                       if (ini_get("open_basedir") && function_exists("curl_init")) {
+                               _debug("not using CURL due to open_basedir restrictions");
+                       }
 
-                               global $fetch_curl_used;
+                       $feed_data = fetch_file_contents($fetch_url, false,
+                               $auth_login, $auth_pass, false,
+                               $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
+                               0);
 
-                               if (!$fetch_curl_used) {
-                                       $tmp = @gzdecode($feed_data);
+                       global $fetch_curl_used;
 
-                                       if ($tmp) $feed_data = $tmp;
-                               }
+                       if (!$fetch_curl_used) {
+                               $tmp = @gzdecode($feed_data);
 
-                               $feed_data = trim($feed_data);
+                               if ($tmp) $feed_data = $tmp;
+                       }
 
-                               _debug("fetch done.", $debug_enabled);
+                       $feed_data = trim($feed_data);
 
-                               // 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);
+                       _debug("fetch done.", $debug_enabled);
 
-                                       if ($new_rss_hash != $rss_hash) {
-                                               _debug("saving $cache_filename", $debug_enabled);
-                                               @file_put_contents($cache_filename, $feed_data);
-                                       }
+                       // cache vanilla feed data for re-use
+                       if ($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) {
-                               global $fetch_last_error;
-                               global $fetch_last_error_code;
+               if (!$feed_data) {
+                       global $fetch_last_error;
+                       global $fetch_last_error_code;
 
-                               _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
+                       _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
 
-                               $error_escaped = '';
+                       $error_escaped = '';
 
-                               // If-Modified-Since
-                               if ($fetch_last_error_code != 304) {
-                                       $error_escaped = db_escape_string($fetch_last_error);
-                               } else {
-                                       _debug("source claims data not modified, nothing to do.", $debug_enabled);
-                               }
+                       // If-Modified-Since
+                       if ($fetch_last_error_code != 304) {
+                               $error_escaped = db_escape_string($fetch_last_error);
+                       } else {
+                               _debug("source claims data not modified, nothing to do.", $debug_enabled);
+                       }
 
-                               db_query(
-                                       "UPDATE ttrss_feeds SET last_error = '$error_escaped',
-                                               last_updated = NOW() WHERE id = '$feed'");
+                       db_query(
+                               "UPDATE ttrss_feeds SET last_error = '$error_escaped',
+                                       last_updated = NOW() WHERE id = '$feed'");
 
-                               return;
-                       }
+                       return;
                }
 
                foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
                        $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
                }
 
-               // set last update to now so if anything *simplepie* crashes later we won't be
-               // continuously failing on the same feed
-               //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
-
-               if (!$rss) {
-                       $rss = new FeedParser($feed_data);
-                       $rss->init();
-               }
-
-//             print_r($rss);
+               $rss = new FeedParser($feed_data);
+               $rss->init();
 
                $feed = db_escape_string($feed);
 
                                if (db_num_rows($result) != 0) {
                                        $base_entry_id = db_fetch_result($result, 0, "id");
                                        $entry_stored_hash = db_fetch_result($result, 0, "content_hash");
-                                       $article_labels = get_article_labels($base_entry_id, $owner_uid);
+                                       $article_labels = Article::get_article_labels($base_entry_id, $owner_uid);
                                        $entry_language = db_fetch_result($result, 0, "lang");
 
-                                       $existing_tags = get_article_tags($base_entry_id, $owner_uid);
+                                       $existing_tags = Article::get_article_tags($base_entry_id, $owner_uid);
                                        $entry_tags = array_unique(array_merge($entry_tags, $existing_tags));
 
                                } else {
                                $matched_rules = array();
 
                                $article_filters = get_article_filters($filters, $article["title"],
-                                       $article["content"], $article["link"], 0, $article["author"],
+                                       $article["content"], $article["link"], $article["author"],
                                        $article["tags"], $matched_rules);
 
                                if ($debug_enabled) {
 
                                if (db_num_rows($result) == 0) {
 
-                                       _debug("base guid [$entry_guid] not found", $debug_enabled);
+                                       _debug("base guid [$entry_guid or $entry_guid_hashed] not found, creating...", $debug_enabled);
 
                                        // base post entry does not exist, create it
 
-                                       $result = db_query(
+                                       db_query(
                                                "INSERT INTO ttrss_entries
                                                        (title,
                                                        guid,
                                                        '$entry_language',
                                                        '$entry_author')");
 
-                               } else {
-                                       $base_entry_id = db_fetch_result($result, 0, "id");
                                }
 
                                // now it should exist, if not - bad luck then
                                _debug("assigning labels [other]...", $debug_enabled);
 
                                foreach ($article_labels as $label) {
-                                       label_add_article($entry_ref_id, $label[1], $owner_uid);
+                                       Labels::add_article($entry_ref_id, $label[1], $owner_uid);
                                }
 
                                _debug("assigning labels [filters]...", $debug_enabled);
                                if (is_array($encs)) {
                                        foreach ($encs as $e) {
                                                $e_item = array(
-                                                       $e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
+                                                       rewrite_relative_url($site_url, $e->link),
+                                                       $e->type, $e->length, $e->title, $e->width, $e->height);
                                                array_push($enclosures, $e_item);
                                        }
                                }
                                last_updated = NOW() WHERE id = '$feed'");
 
                        unset($rss);
+                       return;
                }
 
                _debug("done", $debug_enabled);
 
-               return $rss;
+               return true;
        }
 
        function cache_enclosures($enclosures, $site_url, $debug) {
                return $params;
        }
 
-       function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags, &$matched_rules = false) {
+       function get_article_filters($filters, $title, $content, $link, $author, $tags, &$matched_rules = false) {
                $matches = array();
 
                foreach ($filters as $filter) {
                foreach ($filters as $f) {
                        if ($f["type"] == "label") {
                                if (!labels_contains_caption($article_labels, $f["param"])) {
-                                       label_add_article($id, $f["param"], $owner_uid);
+                                       Labels::add_article($id, $f["param"], $owner_uid);
                                }
                        }
                }
                        mb_strtolower(strip_tags($title), 'utf-8'));
        }
 
-       /* function verify_feed_xml($feed_data) {
-               libxml_use_internal_errors(true);
-               $doc = new DOMDocument();
-               $doc->loadXML($feed_data);
-               $error = libxml_get_last_error();
-               libxml_clear_errors();
-               return $error;
-       } */
-
        function cleanup_counters_cache($debug) {
                $result = db_query("DELETE FROM ttrss_counters_cache
                        WHERE feed_id > 0 AND
                                ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
                $crows = db_affected_rows($result);
 
-               _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
+               if ($debug) _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
        }
 
        function housekeeping_user($owner_uid) {
                $count = update_feedbrowser_cache();
                _debug("Feedbrowser updated, $count feeds processed.");
 
-               purge_orphans( true);
+               Article::purge_orphans( true);
                cleanup_counters_cache($debug);
 
                //$rc = cleanup_tags( 14, 50000);
 
                PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
        }
-?>
+
+       function check_feed_favicon($site_url, $feed) {
+       #               print "FAVICON [$site_url]: $favicon_url\n";
+
+               $icon_file = ICONS_DIR . "/$feed.ico";
+
+               if (!file_exists($icon_file)) {
+                       $favicon_url = get_favicon_url($site_url);
+
+                       if ($favicon_url) {
+                               // Limiting to "image" type misses those served with text/plain
+                               $contents = fetch_file_contents($favicon_url); // , "image");
+
+                               if ($contents) {
+                                       // Crude image type matching.
+                                       // Patterns gleaned from the file(1) source code.
+                                       if (preg_match('/^\x00\x00\x01\x00/', $contents)) {
+                                               // 0       string  \000\000\001\000        MS Windows icon resource
+                                               //error_log("check_feed_favicon: favicon_url=$favicon_url isa MS Windows icon resource");
+                                       }
+                                       elseif (preg_match('/^GIF8/', $contents)) {
+                                               // 0       string          GIF8            GIF image data
+                                               //error_log("check_feed_favicon: favicon_url=$favicon_url isa GIF image");
+                                       }
+                                       elseif (preg_match('/^\x89PNG\x0d\x0a\x1a\x0a/', $contents)) {
+                                               // 0       string          \x89PNG\x0d\x0a\x1a\x0a         PNG image data
+                                               //error_log("check_feed_favicon: favicon_url=$favicon_url isa PNG image");
+                                       }
+                                       elseif (preg_match('/^\xff\xd8/', $contents)) {
+                                               // 0       beshort         0xffd8          JPEG image data
+                                               //error_log("check_feed_favicon: favicon_url=$favicon_url isa JPG image");
+                                       }
+                                       else {
+                                               //error_log("check_feed_favicon: favicon_url=$favicon_url isa UNKNOWN type");
+                                               $contents = "";
+                                       }
+                               }
+
+                               if ($contents) {
+                                       $fp = @fopen($icon_file, "w");
+
+                                       if ($fp) {
+                                               fwrite($fp, $contents);
+                                               fclose($fp);
+                                               chmod($icon_file, 0644);
+                                       }
+                               }
+                       }
+                       return $icon_file;
+               }
+       }