]> git.wh0rd.org - tt-rss.git/blobdiff - update.php
pngcrush.sh
[tt-rss.git] / update.php
old mode 100755 (executable)
new mode 100644 (file)
index f23244d..17eafe1
 #!/usr/bin/env php
 <?php
-       set_include_path(get_include_path() . PATH_SEPARATOR .
-               dirname(__FILE__) . "/include");
+       set_include_path(dirname(__FILE__) ."/include" . PATH_SEPARATOR .
+               get_include_path());
 
        define('DISABLE_SESSIONS', true);
 
        chdir(dirname(__FILE__));
 
+       require_once "autoload.php";
        require_once "functions.php";
-       require_once "rssfuncs.php";
-       require_once "sanity_check.php";
        require_once "config.php";
+       require_once "sanity_check.php";
        require_once "db.php";
        require_once "db-prefs.php";
 
        if (!defined('PHP_EXECUTABLE'))
                define('PHP_EXECUTABLE', '/usr/bin/php');
 
-       $op = $argv;
+       $pdo = Db::pdo();
+
+       init_plugins();
+
+       $longopts = array("feeds",
+                       "feedbrowser",
+                       "daemon",
+                       "daemon-loop",
+                       "task:",
+                       "cleanup-tags",
+                       "quiet",
+                       "log:",
+                       "log-level:",
+                       "indexes",
+                       "pidlock:",
+                       "update-schema",
+                       "convert-filters",
+                       "force-update",
+                       "gen-search-idx",
+                       "list-plugins",
+                       "debug-feed:",
+                       "force-refetch",
+                       "force-rehash",
+                       "help");
+
+       foreach (PluginHost::getInstance()->get_commands() as $command => $data) {
+               array_push($longopts, $command . $data["suffix"]);
+       }
 
-       if (count($argv) == 1 || in_array("-help", $op) ) {
-               print "Tiny Tiny RSS data update script.\n\n";
-               print "Options:\n";
-               print "  -feeds              - update feeds\n";
-               print "  -feedbrowser        - update feedbrowser\n";
-               print "  -daemon             - start single-process update daemon\n";
-               print "  -cleanup-tags       - perform tags table maintenance\n";
-               print "  -get-feeds          - receive popular feeds from linked instances\n";
-               print "  -import USER FILE   - import articles from XML\n";
-               print "  -update-self        - update tt-rss installation to latest version\n";
-               print "  -quiet              - don't show messages\n";
-               print "  -indexes            - recreate missing schema indexes\n";
-               print "  -help               - show this help\n";
-               return;
+       $options = getopt("", $longopts);
+
+       if (!is_array($options)) {
+               die("error: getopt() failed. ".
+                       "Most probably you are using PHP CGI to run this script ".
+                       "instead of required PHP CLI. Check tt-rss wiki page on updating feeds for ".
+                       "additional information.\n");
        }
 
-       define('QUIET', in_array("-quiet", $op));
+       if (count($options) == 0 && !defined('STDIN')) {
+               ?> <html>
+               <head>
+               <title>Tiny Tiny RSS data update script.</title>
+               <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
+               </head>
 
-       if (!in_array("-daemon", $op)) {
-               $lock_filename = "update.lock";
-       } else {
-               $lock_filename = "update_daemon.lock";
+               <body>
+               <div class="floatingLogo"><img src="images/logo_small.png"></div>
+               <h1><?php echo __("Tiny Tiny RSS data update script.") ?></h1>
+
+               <?php print_error("Please run this script from the command line. Use option \"--help\" to display command help if this error is displayed erroneously."); ?>
+
+               </body></html>
+       <?php
+               exit;
        }
 
-       $lock_handle = make_lockfile($lock_filename);
-       $must_exit = false;
+       if (count($options) == 0 || isset($options["help"]) ) {
+               print "Tiny Tiny RSS data update script.\n\n";
+               print "Options:\n";
+               print "  --feeds              - update feeds\n";
+               print "  --feedbrowser        - update feedbrowser\n";
+               print "  --daemon             - start single-process update daemon\n";
+               print "  --task N             - create lockfile using this task id\n";
+               print "  --cleanup-tags       - perform tags table maintenance\n";
+               print "  --quiet              - don't output messages to stdout\n";
+               print "  --log FILE           - log messages to FILE\n";
+               print "  --log-level N        - log verbosity level\n";
+               print "  --indexes            - recreate missing schema indexes\n";
+               print "  --update-schema      - update database schema\n";
+               print "  --gen-search-idx     - generate basic PostgreSQL fulltext search index\n";
+               print "  --convert-filters    - convert type1 filters to type2\n";
+               print "  --force-update       - force update of all feeds\n";
+               print "  --list-plugins       - list all available plugins\n";
+               print "  --debug-feed N       - perform debug update of feed N\n";
+               print "  --force-refetch      - debug update: force refetch feed data\n";
+               print "  --force-rehash       - debug update: force rehash articles\n";
+               print "  --help               - show this help\n";
+               print "Plugin options:\n";
+
+               foreach (PluginHost::getInstance()->get_commands() as $command => $data) {
+                       $args = $data['arghelp'];
+                       printf(" --%-19s - %s\n", "$command $args", $data["description"]);
+               }
 
-       // Try to lock a file in order to avoid concurrent update.
-       if (!$lock_handle) {
-               die("error: Can't create lockfile ($lock_filename). ".
-                       "Maybe another update process is already running.\n");
+               return;
        }
 
-       // Create a database connection.
-       $link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);
+       if (!isset($options['daemon'])) {
+               require_once "errorhandler.php";
+       }
 
-       init_connection($link);
+       if (!isset($options['update-schema'])) {
+               $schema_version = get_schema_version();
 
-       if (in_array("-feeds", $op)) {
-               // Update all feeds needing a update.
-               update_daemon_common($link);
+               if ($schema_version != SCHEMA_VERSION) {
+                       die("Schema version is wrong, please upgrade the database.\n");
+               }
+       }
 
-               // Update feedbrowser
-               $count = update_feedbrowser_cache($link);
-               _debug("Feedbrowser updated, $count feeds processed.");
+       Debug::set_enabled(true);
+       Debug::set_quiet(isset($options['quiet']));
 
-               // Purge orphans and cleanup tags
-               purge_orphans($link, true);
+       if (isset($options["log-level"])) {
+           Debug::set_loglevel((int)$options["log-level"]);
+    }
 
-               $rc = cleanup_tags($link, 14, 50000);
-               _debug("Cleaned $rc cached tags.");
+       if (isset($options["log"])) {
+               Debug::set_logfile($options["log"]);
+        Debug::log("Logging to " . $options["log"]);
+    }
 
-               get_linked_feeds($link);
+       if (!isset($options["daemon"])) {
+               $lock_filename = "update.lock";
+       } else {
+               $lock_filename = "update_daemon.lock";
        }
 
-       if (in_array("-feedbrowser", $op)) {
-               $count = update_feedbrowser_cache($link);
-               print "Finished, $count feeds processed.\n";
+       if (isset($options["task"])) {
+               Debug::log("Using task id " . $options["task"]);
+               $lock_filename = $lock_filename . "-task_" . $options["task"];
        }
 
-       if (in_array("-daemon", $op)) {
-               $op = array_diff($op, array("-daemon"));
-               while (true) {
-                       passthru(PHP_EXECUTABLE . " " . implode(' ', $op) . " -daemon-loop");
-                       _debug("Sleeping for " . DAEMON_SLEEP_INTERVAL . " seconds...");
-                       sleep(DAEMON_SLEEP_INTERVAL);
-               }
+       if (isset($options["pidlock"])) {
+               $my_pid = $options["pidlock"];
+               $lock_filename = "update_daemon-$my_pid.lock";
+
        }
 
-       if (in_array("-daemon-loop", $op)) {
-               if (!make_stampfile('update_daemon.stamp')) {
-                       die("error: unable to create stampfile\n");
-               }
+       Debug::log("Lock: $lock_filename");
 
-               // Call to the feed batch update function
-               // or regenerate feedbrowser cache
+       $lock_handle = make_lockfile($lock_filename);
+       $must_exit = false;
 
-               if (rand(0,100) > 30) {
-                       update_daemon_common($link);
-               } else {
-                       $count = update_feedbrowser_cache($link);
-                       _debug("Feedbrowser updated, $count feeds processed.");
+       if (isset($options["task"]) && isset($options["pidlock"])) {
+               $waits = $options["task"] * 5;
+               Debug::log("Waiting before update ($waits)");
+               sleep($waits);
+       }
 
-                       purge_orphans($link, true);
+       // Try to lock a file in order to avoid concurrent update.
+       if (!$lock_handle) {
+               die("error: Can't create lockfile ($lock_filename). ".
+                       "Maybe another update process is already running.\n");
+       }
 
-                       $rc = cleanup_tags($link, 14, 50000);
+       if (isset($options["force-update"])) {
+               Debug::log("marking all feeds as needing update...");
 
-                       _debug("Cleaned $rc cached tags.");
+               $pdo->query( "UPDATE ttrss_feeds SET
+          last_update_started = '1970-01-01', last_updated = '1970-01-01'");
+       }
 
-                       get_linked_feeds($link);
-               }
+       if (isset($options["feeds"])) {
+               RSSUtils::update_daemon_common();
+               RSSUtils::housekeeping_common(true);
 
+               PluginHost::getInstance()->run_hooks(PluginHost::HOOK_UPDATE_TASK, "hook_update_task", $op);
        }
 
-       if (in_array("-cleanup-tags", $op)) {
-               $rc = cleanup_tags($link, 14, 50000);
-               _debug("$rc tags deleted.\n");
+       if (isset($options["feedbrowser"])) {
+               $count = RSSUtils::update_feedbrowser_cache();
+               print "Finished, $count feeds processed.\n";
        }
 
-       if (in_array("-get-feeds", $op)) {
-               get_linked_feeds($link);
-       }
+       if (isset($options["daemon"])) {
+               while (true) {
+                       $quiet = (isset($options["quiet"])) ? "--quiet" : "";
+            $log = isset($options['log']) ? '--log '.$options['log'] : '';
+            $log_level = isset($options['log-level']) ? '--log-level '.$options['log-level'] : '';
 
-       if (in_array("-import",$op)) {
-               $username = $argv[count($argv) - 2];
-               $filename = $argv[count($argv) - 1];
+                       passthru(PHP_EXECUTABLE . " " . $argv[0] ." --daemon-loop $quiet $log $log_level");
 
-               if (!$username) {
-                       print "error: please specify username.\n";
-                       return;
-               }
+                       // let's enforce a minimum spawn interval as to not forkbomb the host
+                       $spawn_interval = max(60, DAEMON_SLEEP_INTERVAL);
 
-               if (!is_file($filename)) {
-                       print "error: input filename ($filename) doesn't exist.\n";
-                       return;
+                       Debug::log("Sleeping for $spawn_interval seconds...");
+                       sleep($spawn_interval);
                }
+       }
 
-               _debug("importing $filename for user $username...\n");
-
-               $result = db_query($link, "SELECT id FROM ttrss_users WHERE login = '$username'");
-
-               if (db_num_rows($result) == 0) {
-                       print "error: could not find user $username.\n";
-                       return;
+       if (isset($options["daemon-loop"])) {
+               if (!make_stampfile('update_daemon.stamp')) {
+                       Debug::log("warning: unable to create stampfile\n");
                }
 
-               $owner_uid = db_fetch_result($result, 0, "id");
+               RSSUtils::update_daemon_common(isset($options["pidlock"]) ? 50 : DAEMON_FEED_LIMIT);
 
-               perform_data_import($link, $filename, $owner_uid);
+               if (!isset($options["pidlock"]) || $options["task"] == 0)
+                       RSSUtils::housekeeping_common(true);
 
+               PluginHost::getInstance()->run_hooks(PluginHost::HOOK_UPDATE_TASK, "hook_update_task", $op);
        }
 
-       if (in_array("-indexes", $op)) {
-               _debug("PLEASE BACKUP YOUR DATABASE BEFORE PROCEEDING!");
-               _debug("Type 'yes' to continue.");
+       if (isset($options["cleanup-tags"])) {
+               $rc = cleanup_tags( 14, 50000);
+               Debug::log("$rc tags deleted.\n");
+       }
+
+       if (isset($options["indexes"])) {
+               Debug::log("PLEASE BACKUP YOUR DATABASE BEFORE PROCEEDING!");
+               Debug::log("Type 'yes' to continue.");
 
                if (read_stdin() != 'yes')
                        exit;
 
-               _debug("clearing existing indexes...");
+               Debug::log("clearing existing indexes...");
 
                if (DB_TYPE == "pgsql") {
-                       $result = db_query($link, "SELECT relname FROM
+                       $sth = $pdo->query( "SELECT relname FROM
                                pg_catalog.pg_class WHERE relname LIKE 'ttrss_%'
                                        AND relname NOT LIKE '%_pkey'
                                AND relkind = 'i'");
                } else {
-                       $result = db_query($link, "SELECT index_name,table_name FROM
+                       $sth = $pdo->query( "SELECT index_name,table_name FROM
                                information_schema.statistics WHERE index_name LIKE 'ttrss_%'");
                }
 
-               while ($line = db_fetch_assoc($result)) {
+               while ($line = $sth->fetch()) {
                        if (DB_TYPE == "pgsql") {
                                $statement = "DROP INDEX " . $line["relname"];
-                               _debug($statement);
+                               Debug::log($statement);
                        } else {
                                $statement = "ALTER TABLE ".
                                        $line['table_name']." DROP INDEX ".$line['index_name'];
-                               _debug($statement);
+                               Debug::log($statement);
                        }
-                       db_query($link, $statement, false);
+                       $pdo->query($statement);
                }
 
-               _debug("reading indexes from schema for: " . DB_TYPE);
+               Debug::log("reading indexes from schema for: " . DB_TYPE);
 
                $fp = fopen("schema/ttrss_schema_" . DB_TYPE . ".sql", "r");
                if ($fp) {
 
                                        $statement = "CREATE INDEX $index ON $table";
 
-                                       _debug($statement);
-                                       db_query($link, $statement);
+                                       Debug::log($statement);
+                                       $pdo->query($statement);
                                }
                        }
                        fclose($fp);
                } else {
-                       _debug("unable to open schema file.");
+                       Debug::log("unable to open schema file.");
                }
-               _debug("all done.");
+               Debug::log("all done.");
        }
 
-       if (in_array("-update-self", $op)) {
-               _debug("Warning: self-updating is experimental. Use at your own risk.");
-               _debug("Please backup your tt-rss directory before continuing. Your database will not be modified.");
-               _debug("Type 'yes' to continue.");
+       if (isset($options["convert-filters"])) {
+               Debug::log("WARNING: this will remove all existing type2 filters.");
+               Debug::log("Type 'yes' to continue.");
 
                if (read_stdin() != 'yes')
                        exit;
 
-               $work_dir = dirname(__FILE__);
-               $parent_dir = dirname($work_dir);
+               Debug::log("converting filters...");
 
-               if (!is_writable($work_dir) && !is_writable("$parent_dir")) {
-                       _debug("Both current and parent directories should be writable as current user.");
-                       exit;
-               }
+               $pdo->query("DELETE FROM ttrss_filters2");
 
-               if (!is_writable(sys_get_temp_dir())) {
-                       _debug("System temporary directory should be writable as current user.");
-                       exit;
-               }
+               $res = $pdo->query("SELECT * FROM ttrss_filters ORDER BY id");
 
-               _debug("Checking for tar...");
+               while ($line = $res->fetch()) {
+                       $owner_uid = $line["owner_uid"];
 
-               $system_rc = 0;
-               system("tar --version >/dev/null", $system_rc);
+                       // date filters are removed
+                       if ($line["filter_type"] != 5) {
+                               $filter = array();
 
-               if ($system_rc != 0) {
-                       _debug("Could not run tar executable (RC=$system_rc).");
-                       exit;
-               }
-
-               _debug("Checking for latest version...");
+                               if (sql_bool_to_bool($line["cat_filter"])) {
+                                       $feed_id = "CAT:" . (int)$line["cat_id"];
+                               } else {
+                                       $feed_id = (int)$line["feed_id"];
+                               }
 
-               $version_info = json_decode(fetch_file_contents("http://tt-rss.org/version.php"),
-                       true);
+                               $filter["enabled"] = $line["enabled"] ? "on" : "off";
+                               $filter["rule"] = array(
+                                       json_encode(array(
+                                               "reg_exp" => $line["reg_exp"],
+                                               "feed_id" => $feed_id,
+                                               "filter_type" => $line["filter_type"])));
 
-               if (!is_array($version_info)) {
-                       _debug("Unable to fetch version information.");
-                       exit;
-               }
+                               $filter["action"] = array(
+                                       json_encode(array(
+                                               "action_id" => $line["action_id"],
+                                               "action_param_label" => $line["action_param"],
+                                               "action_param" => $line["action_param"])));
 
-               $target_version = $version_info["version"];
-               $target_dir = "$parent_dir/tt-rss-$target_version";
+                               // Oh god it's full of hacks
 
-               _debug("Target version: $target_version");
+                               $_REQUEST = $filter;
+                               $_SESSION["uid"] = $owner_uid;
 
-               if (version_compare(VERSION, $target_version) != -1 && !in_array("-force", $op)) {
-                       _debug("You are on latest version. Update not needed.");
-                       exit;
-               }
-               if (file_exists($target_dir)) {
-                       _debug("Target directory $target_dir already exists.");
-                       exit;
+                               $filters = new Pref_Filters($_REQUEST);
+                               $filters->add();
+                       }
                }
 
-               _debug("Downloading checksums...");
-               $md5sum_data = fetch_file_contents("http://tt-rss.org/download/md5sum.txt");
+       }
 
-               if (!$md5sum_data) {
-                       _debug("Could not download checksums.");
-                       exit;
-               }
+       if (isset($options["update-schema"])) {
+               Debug::log("checking for updates (" . DB_TYPE . ")...");
 
-               $md5sum_data = explode("\n", $md5sum_data);
+               $updater = new DbUpdater(Db::pdo(), DB_TYPE, SCHEMA_VERSION);
 
-               $tarball_url = "http://tt-rss.org/download/tt-rss-$target_version.tar.gz";
-               $data = fetch_file_contents($tarball_url);
+               if ($updater->isUpdateRequired()) {
+                       Debug::log("schema update required, version " . $updater->getSchemaVersion() . " to " . SCHEMA_VERSION);
+                       Debug::log("WARNING: please backup your database before continuing.");
+                       Debug::log("Type 'yes' to continue.");
 
-               if (!$data) {
-                       _debug("Could not download distribution tarball ($tarball_url).");
-                       exit;
-               }
+                       if (read_stdin() != 'yes')
+                               exit;
 
-               _debug("Verifying tarball checksum...");
+                       for ($i = $updater->getSchemaVersion() + 1; $i <= SCHEMA_VERSION; $i++) {
+                               Debug::log("performing update up to version $i...");
 
-               $target_md5sum = false;
+                               $result = $updater->performUpdateTo($i, false);
 
-               foreach ($md5sum_data as $line) {
-                       $pair = explode("  ", $line);
+                               Debug::log($result ? "OK!" : "FAILED!");
+
+                               if (!$result) return;
 
-                       if ($pair[1] == "tt-rss-$target_version.tar.gz") {
-                               $target_md5sum = $pair[0];
-                               break;
                        }
+               } else {
+                       Debug::log("update not required.");
                }
 
-               if (!$target_md5sum) {
-                       _debug("Unable to locate checksum for target version.");
-                       exit;
-               }
+       }
 
-               $test_md5sum = md5($data);
+       if (isset($options["gen-search-idx"])) {
+               echo "Generating search index (stemming set to English)...\n";
 
-               if ($test_md5sum != $target_md5sum) {
-                       _debug("Downloaded checksum doesn't match (got $test_md5sum, expected $target_md5sum).");
-                       exit;
-               }
+               $res = $pdo->query("SELECT COUNT(id) AS count FROM ttrss_entries WHERE tsvector_combined IS NULL");
+               $row = $res->fetch();
+               $count = $row['count'];
 
-               $tmp_file = tempnam(sys_get_temp_dir(), 'tt-rss');
-               _debug("Saving download to $tmp_file");
+               print "Articles to process: $count.\n";
 
-               if (!file_put_contents($tmp_file, $data)) {
-                       _debug("Unable to save download.");
-                       exit;
-               }
+               $limit = 500;
+               $processed = 0;
 
-               if (!chdir($parent_dir)) {
-                       _debug("Unable to change into parent directory.");
-                       exit;
-               }
+               $sth = $pdo->prepare("SELECT id, title, content FROM ttrss_entries WHERE
+          tsvector_combined IS NULL ORDER BY id LIMIT ?");
+               $sth->execute([$limit]);
 
-               $old_dir = tmpdirname($parent_dir, "tt-rss-old");
+               $usth = $pdo->prepare("UPDATE ttrss_entries
+          SET tsvector_combined = to_tsvector('english', ?) WHERE id = ?");
 
-               _debug("Renaming current directory to ".basename($old_dir));
-               if (!rename($work_dir, $old_dir)) {
-                       _debug("Unable to rename current directory.");
-                       exit;
-               }
+               while (true) {
 
-               _debug("Extracting tarball...");
-               system("tar zxf $tmp_file", $system_rc);
+                       while ($line = $sth->fetch()) {
+                               $tsvector_combined = mb_substr(strip_tags($line["title"] . " " . $line["content"]), 0, 1000000);
 
-               if ($system_rc != 0) {
-                       _debug("Error while extracting tarball (RC=$system_rc).");
-                       exit;
-               }
+                               $usth->execute([$tsvector_combined, $line['id']]);
 
-               _debug("Renaming target directory...");
-               if (!rename($target_dir, $work_dir)) {
-                       _debug("Unable to rename target directory.");
-                       exit;
-               }
+                               $processed++;
+                       }
 
-               chdir($work_dir);
+                       print "Processed $processed articles...\n";
 
-               _debug("Copying config.php...");
-               if (!copy("$old_dir/config.php", "$work_dir/config.php")) {
-                       _debug("Unable to copy config.php to $work_dir.");
-                       exit;
+                       if ($processed < $limit) {
+                               echo "All done.\n";
+                               break;
+                       }
                }
+       }
+
+       if (isset($options["list-plugins"])) {
+               $tmppluginhost = new PluginHost();
+               $tmppluginhost->load_all($tmppluginhost::KIND_ALL, false);
+               $enabled = array_map("trim", explode(",", PLUGINS));
 
-               _debug("Cleaning up...");
-               unlink($tmp_file);
+               echo "List of all available plugins:\n";
 
-               _debug("Fixing permissions...");
+               foreach ($tmppluginhost->get_plugins() as $name => $plugin) {
+                       $about = $plugin->about();
 
-               $directories = array(
-                       CACHE_DIR,
-                       CACHE_DIR . "/htmlpurifier",
-                       CACHE_DIR . "/export",
-                       CACHE_DIR . "/images",
-                       CACHE_DIR . "/magpie",
-                       CACHE_DIR . "/simplepie",
-                       ICONS_DIR,
-                       LOCK_DIRECTORY);
+                       $status = $about[3] ? "system" : "user";
 
-               foreach ($directories as $dir) {
-                       _debug("-> $dir");
-                       chmod($dir, 0777);
+                       if (in_array($name, $enabled)) $name .= "*";
+
+                       printf("%-50s %-10s v%.2f (by %s)\n%s\n\n",
+                               $name, $status, $about[0], $about[2], $about[1]);
                }
 
-               _debug("Upgrade completed.");
-               _debug("Your old tt-rss directory is saved at $old_dir. ".
-                       "Please migrate locally modified files (if any) and remove it.");
-               _debug("You might need to re-enter current directory in shell to see new files.");
+               echo "Plugins marked by * are currently enabled for all users.\n";
+
        }
 
-       db_close($link);
+       if (isset($options["debug-feed"])) {
+               $feed = $options["debug-feed"];
+
+               if (isset($options["force-refetch"])) $_REQUEST["force_refetch"] = true;
+               if (isset($options["force-rehash"])) $_REQUEST["force_rehash"] = true;
 
-       if ($lock_handle != false) {
-               fclose($lock_handle);
+               Debug::set_loglevel(Debug::$LOG_EXTENDED);
+
+               $rc = RSSUtils::update_rss_feed($feed) != false ? 0 : 1;
+
+               exit($rc);
        }
 
+       PluginHost::getInstance()->run_commands($options);
+
        if (file_exists(LOCK_DIRECTORY . "/$lock_filename"))
+               if (strtoupper(substr(PHP_OS, 0, 3)) == 'WIN')
+                       fclose($lock_handle);
                unlink(LOCK_DIRECTORY . "/$lock_filename");
 ?>