include/rssfuncs.php

   1 <?php
   2         define_default('DAEMON_UPDATE_LOGIN_LIMIT', 30);
   3         define_default('DAEMON_FEED_LIMIT', 500);
   4         define_default('DAEMON_SLEEP_INTERVAL', 120);
   5
   6         function update_feedbrowser_cache() {
   7
   8                 $result = db_query("SELECT feed_url, site_url, title, COUNT(id) AS subscribers
   9                         FROM ttrss_feeds WHERE (SELECT COUNT(id) = 0 FROM ttrss_feeds AS tf
  10                                 WHERE tf.feed_url = ttrss_feeds.feed_url
  11                                 AND (private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%'))
  12                                 GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
  13
  14                 db_query("BEGIN");
  15
  16                 db_query("DELETE FROM ttrss_feedbrowser_cache");
  17
  18                 $count = 0;
  19
  20                 while ($line = db_fetch_assoc($result)) {
  21                         $subscribers = db_escape_string($line["subscribers"]);
  22                         $feed_url = db_escape_string($line["feed_url"]);
  23                         $title = db_escape_string($line["title"]);
  24                         $site_url = db_escape_string($line["site_url"]);
  25
  26                         $tmp_result = db_query("SELECT subscribers FROM
  27                                 ttrss_feedbrowser_cache WHERE feed_url = '$feed_url'");
  28
  29                         if (db_num_rows($tmp_result) == 0) {
  30
  31                                 db_query("INSERT INTO ttrss_feedbrowser_cache
  32                                         (feed_url, site_url, title, subscribers) VALUES ('$feed_url',
  33                                                 '$site_url', '$title', '$subscribers')");
  34
  35                                 ++$count;
  36
  37                         }
  38
  39                 }
  40
  41                 db_query("COMMIT");
  42
  43                 return $count;
  44
  45         }
  46
  47
  48         /**
  49          * Update a feed batch.
  50          * Used by daemons to update n feeds by run.
  51          * Only update feed needing a update, and not being processed
  52          * by another process.
  53          *
  54          * @param mixed $link Database link
  55          * @param integer $limit Maximum number of feeds in update batch. Default to DAEMON_FEED_LIMIT.
  56          * @param boolean $from_http Set to true if you call this function from http to disable cli specific code.
  57          * @param boolean $debug Set to false to disable debug output. Default to true.
  58          * @return void
  59          */
  60         function update_daemon_common($limit = DAEMON_FEED_LIMIT, $from_http = false, $debug = true) {
  61                 // Process all other feeds using last_updated and interval parameters
  62
  63                 $schema_version = get_schema_version();
  64
  65                 if ($schema_version != SCHEMA_VERSION) {
  66                         die("Schema version is wrong, please upgrade the database.\n");
  67                 }
  68
  69                 define('PREFS_NO_CACHE', true);
  70
  71                 // Test if the user has loggued in recently. If not, it does not update its feeds.
  72                 if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
  73                         if (DB_TYPE == "pgsql") {
  74                                 $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
  75                         } else {
  76                                 $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
  77                         }
  78                 } else {
  79                         $login_thresh_qpart = "";
  80                 }
  81
  82                 // Test if the feed need a update (update interval exceded).
  83                 if (DB_TYPE == "pgsql") {
  84                         $update_limit_qpart = "AND ((
  85                                         ttrss_feeds.update_interval = 0
  86                                         AND ttrss_user_prefs.value != '-1'
  87                                         AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
  88                                 ) OR (
  89                                         ttrss_feeds.update_interval > 0
  90                                         AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
  91                                 ) OR ttrss_feeds.last_updated IS NULL
  92                                 OR last_updated = '1970-01-01 00:00:00')";
  93                 } else {
  94                         $update_limit_qpart = "AND ((
  95                                         ttrss_feeds.update_interval = 0
  96                                         AND ttrss_user_prefs.value != '-1'
  97                                         AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
  98                                 ) OR (
  99                                         ttrss_feeds.update_interval > 0
 100                                         AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
 101                                 ) OR ttrss_feeds.last_updated IS NULL
 102                                 OR last_updated = '1970-01-01 00:00:00')";
 103                 }
 104
 105                 // Test if feed is currently being updated by another process.
 106                 if (DB_TYPE == "pgsql") {
 107                         $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
 108                 } else {
 109                         $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 10 MINUTE))";
 110                 }
 111
 112                 // Test if there is a limit to number of updated feeds
 113                 $query_limit = "";
 114                 if($limit) $query_limit = sprintf("LIMIT %d", $limit);
 115
 116                 $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
 117                         FROM
 118                                 ttrss_feeds, ttrss_users, ttrss_user_prefs
 119                         WHERE
 120                                 ttrss_feeds.owner_uid = ttrss_users.id
 121                                 AND ttrss_users.id = ttrss_user_prefs.owner_uid
 122                                 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
 123                                 $login_thresh_qpart $update_limit_qpart
 124                                 $updstart_thresh_qpart
 125                                 ORDER BY last_updated $query_limit";
 126
 127                 // We search for feed needing update.
 128                 $result = db_query($query);
 129
 130                 if($debug) _debug(sprintf("Scheduled %d feeds to update...", db_num_rows($result)));
 131
 132                 // Here is a little cache magic in order to minimize risk of double feed updates.
 133                 $feeds_to_update = array();
 134                 while ($line = db_fetch_assoc($result)) {
 135                         array_push($feeds_to_update, db_escape_string($line['feed_url']));
 136                 }
 137
 138                 // We update the feed last update started date before anything else.
 139                 // There is no lag due to feed contents downloads
 140                 // It prevent an other process to update the same feed.
 141
 142                 if(count($feeds_to_update) > 0) {
 143                         $feeds_quoted = array();
 144
 145                         foreach ($feeds_to_update as $feed) {
 146                                 array_push($feeds_quoted, "'" . db_escape_string($feed) . "'");
 147                         }
 148
 149                         db_query(sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
 150                                 WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
 151                 }
 152
 153                 $nf = 0;
 154
 155                 // For each feed, we call the feed update function.
 156                 foreach ($feeds_to_update as $feed) {
 157                         if($debug) _debug("Base feed: $feed");
 158
 159                         //update_rss_feed($line["id"], true);
 160
 161                         // since we have the data cached, we can deal with other feeds with the same url
 162
 163                         $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
 164                         FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
 165                                 ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
 166                                 ttrss_users.id = ttrss_user_prefs.owner_uid AND
 167                                 ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
 168                                 feed_url = '".db_escape_string($feed)."' AND
 169                                 (ttrss_feeds.update_interval > 0 OR
 170                                         ttrss_user_prefs.value != '-1')
 171                                 $login_thresh_qpart
 172                         ORDER BY ttrss_feeds.id $query_limit");
 173
 174                         if (db_num_rows($tmp_result) > 0) {
 175                                 while ($tline = db_fetch_assoc($tmp_result)) {
 176                                         if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
 177                                         update_rss_feed($tline["id"], true);
 178                                         ++$nf;
 179                                 }
 180                         }
 181                 }
 182
 183                 require_once "digest.php";
 184
 185                 // Send feed digests by email if needed.
 186                 send_headlines_digests($debug);
 187
 188                 return $nf;
 189
 190         } // function update_daemon_common
 191
 192         // ignore_daemon is not used
 193         function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false) {
 194
 195                 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
 196
 197                 _debug("start", $debug_enabled);
 198
 199                 $result = db_query("SELECT id,update_interval,auth_login,
 200                         feed_url,auth_pass,cache_images,last_updated,
 201                         mark_unread_on_update, owner_uid,
 202                         pubsub_state, auth_pass_encrypted,
 203                         (SELECT max(date_entered) FROM
 204                                 ttrss_entries, ttrss_user_entries where ref_id = id AND feed_id = '$feed') AS last_article_timestamp
 205                         FROM ttrss_feeds WHERE id = '$feed'");
 206
 207                 if (db_num_rows($result) == 0) {
 208                         _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
 209                         return false;
 210                 }
 211
 212                 $last_updated = db_fetch_result($result, 0, "last_updated");
 213                 $last_article_timestamp = @strtotime(db_fetch_result($result, 0, "last_article_timestamp"));
 214
 215                 if (defined('_DISABLE_HTTP_304'))
 216                         $last_article_timestamp = 0;
 217
 218                 $owner_uid = db_fetch_result($result, 0, "owner_uid");
 219                 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
 220                         0, "mark_unread_on_update"));
 221                 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
 222                 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
 223                         0, "auth_pass_encrypted"));
 224
 225                 db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
 226                         WHERE id = '$feed'");
 227
 228                 $auth_login = db_fetch_result($result, 0, "auth_login");
 229                 $auth_pass = db_fetch_result($result, 0, "auth_pass");
 230
 231                 if ($auth_pass_encrypted) {
 232                         require_once "crypt.php";
 233                         $auth_pass = decrypt_string($auth_pass);
 234                 }
 235
 236                 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
 237                 $fetch_url = db_fetch_result($result, 0, "feed_url");
 238
 239                 $feed = db_escape_string($feed);
 240
 241                 $date_feed_processed = date('Y-m-d H:i');
 242
 243                 $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
 244
 245                 $pluginhost = new PluginHost();
 246                 $pluginhost->set_debug($debug_enabled);
 247                 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
 248
 249                 $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
 250                 $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
 251                 $pluginhost->load_data();
 252
 253                 $rss = false;
 254                 $rss_hash = false;
 255
 256                 $force_refetch = isset($_REQUEST["force_refetch"]);
 257
 258                 if (file_exists($cache_filename) &&
 259                         is_readable($cache_filename) &&
 260                         !$auth_login && !$auth_pass &&
 261                         filemtime($cache_filename) > time() - 30) {
 262
 263                         _debug("using local cache.", $debug_enabled);
 264
 265                         @$feed_data = file_get_contents($cache_filename);
 266
 267                         if ($feed_data) {
 268                                 $rss_hash = sha1($feed_data);
 269                         }
 270
 271                 } else {
 272                         _debug("local cache will not be used for this feed", $debug_enabled);
 273                 }
 274
 275                 if (!$rss) {
 276
 277                         foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
 278                                 $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed);
 279                         }
 280
 281                         if (!$feed_data) {
 282                                 _debug("fetching [$fetch_url]...", $debug_enabled);
 283                                 _debug("If-Modified-Since: ".gmdate('D, d M Y H:i:s \G\M\T', $last_article_timestamp), $debug_enabled);
 284
 285                                 $feed_data = fetch_file_contents($fetch_url, false,
 286                                         $auth_login, $auth_pass, false,
 287                                         $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
 288                                         $force_refetch ? 0 : $last_article_timestamp);
 289
 290                                 global $fetch_curl_used;
 291
 292                                 if (!$fetch_curl_used) {
 293                                         $tmp = @gzdecode($feed_data);
 294
 295                                         if ($tmp) $feed_data = $tmp;
 296                                 }
 297
 298                                 $feed_data = trim($feed_data);
 299
 300                                 _debug("fetch done.", $debug_enabled);
 301
 302                                 /* if ($feed_data) {
 303                                         $error = verify_feed_xml($feed_data);
 304
 305                                         if ($error) {
 306                                                 _debug("error verifying XML, code: " . $error->code, $debug_enabled);
 307
 308                                                 if ($error->code == 26) {
 309                                                         _debug("got error 26, trying to decode entities...", $debug_enabled);
 310
 311                                                         $feed_data = html_entity_decode($feed_data, ENT_COMPAT, 'UTF-8');
 312
 313                                                         $error = verify_feed_xml($feed_data);
 314
 315                                                         if ($error) $feed_data = '';
 316                                                 }
 317                                         }
 318                                 } */
 319                         }
 320
 321                         if (!$feed_data) {
 322                                 global $fetch_last_error;
 323                                 global $fetch_last_error_code;
 324
 325                                 _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
 326
 327                                 $error_escaped = '';
 328
 329                                 // If-Modified-Since
 330                                 if ($fetch_last_error_code != 304) {
 331                                         $error_escaped = db_escape_string($fetch_last_error);
 332                                 } else {
 333                                         _debug("source claims data not modified, nothing to do.", $debug_enabled);
 334                                 }
 335
 336                                 db_query(
 337                                         "UPDATE ttrss_feeds SET last_error = '$error_escaped',
 338                                                 last_updated = NOW() WHERE id = '$feed'");
 339
 340                                 return;
 341                         }
 342                 }
 343
 344                 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
 345                         $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
 346                 }
 347
 348                 // set last update to now so if anything *simplepie* crashes later we won't be
 349                 // continuously failing on the same feed
 350                 //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
 351
 352                 if (!$rss) {
 353                         $rss = new FeedParser($feed_data);
 354                         $rss->init();
 355                 }
 356
 357                 require_once "lib/languagedetect/LanguageDetect.php";
 358
 359                 $lang = new Text_LanguageDetect();
 360                 $lang->setNameMode(2);
 361
 362 //              print_r($rss);
 363
 364                 $feed = db_escape_string($feed);
 365
 366                 if (!$rss->error()) {
 367
 368                         // cache data for later
 369                         if (!$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
 370                                 $new_rss_hash = sha1($rss_data);
 371
 372                                 if ($new_rss_hash != $rss_hash && count($rss->get_items()) > 0 ) {
 373                                         _debug("saving $cache_filename", $debug_enabled);
 374                                         @file_put_contents($cache_filename, $feed_data);
 375                                 }
 376                         }
 377
 378                         // We use local pluginhost here because we need to load different per-user feed plugins
 379                         $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
 380
 381                         _debug("processing feed data...", $debug_enabled);
 382
 383 //                      db_query("BEGIN");
 384
 385                         if (DB_TYPE == "pgsql") {
 386                                 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
 387                         } else {
 388                                 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
 389                         }
 390
 391                         $result = db_query("SELECT title,site_url,owner_uid,favicon_avg_color,
 392                                 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
 393                                                 favicon_needs_check
 394                                 FROM ttrss_feeds WHERE id = '$feed'");
 395
 396                         $registered_title = db_fetch_result($result, 0, "title");
 397                         $orig_site_url = db_fetch_result($result, 0, "site_url");
 398                         $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
 399                                 "favicon_needs_check"));
 400                         $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
 401
 402                         $owner_uid = db_fetch_result($result, 0, "owner_uid");
 403
 404                         $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
 405
 406                         _debug("site_url: $site_url", $debug_enabled);
 407                         _debug("feed_title: " . $rss->get_title(), $debug_enabled);
 408
 409                         if ($favicon_needs_check || $force_refetch) {
 410
 411                                 /* terrible hack: if we crash on floicon shit here, we won't check
 412                                  * the icon avgcolor again (unless the icon got updated) */
 413
 414                                 $favicon_file = ICONS_DIR . "/$feed.ico";
 415                                 $favicon_modified = @filemtime($favicon_file);
 416
 417                                 _debug("checking favicon...", $debug_enabled);
 418
 419                                 check_feed_favicon($site_url, $feed);
 420                                 $favicon_modified_new = @filemtime($favicon_file);
 421
 422                                 if ($favicon_modified_new > $favicon_modified)
 423                                         $favicon_avg_color = '';
 424
 425                                 if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
 426                                                 require_once "colors.php";
 427
 428                                                 db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
 429                                                         id = '$feed'");
 430
 431                                                 $favicon_color = db_escape_string(
 432                                                         calculate_avg_color($favicon_file));
 433
 434                                                 $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
 435                                 } else if ($favicon_avg_color == 'fail') {
 436                                         _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
 437                                 }
 438
 439                                 db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
 440                                         $favicon_colorstring
 441                                         WHERE id = '$feed'");
 442                         }
 443
 444                         if (!$registered_title || $registered_title == "[Unknown]") {
 445
 446                                 $feed_title = db_escape_string($rss->get_title());
 447
 448                                 if ($feed_title) {
 449                                         _debug("registering title: $feed_title", $debug_enabled);
 450
 451                                         db_query("UPDATE ttrss_feeds SET
 452                                                 title = '$feed_title' WHERE id = '$feed'");
 453                                 }
 454                         }
 455
 456                         if ($site_url && $orig_site_url != $site_url) {
 457                                 db_query("UPDATE ttrss_feeds SET
 458                                         site_url = '$site_url' WHERE id = '$feed'");
 459                         }
 460
 461                         _debug("loading filters & labels...", $debug_enabled);
 462
 463                         $filters = load_filters($feed, $owner_uid);
 464                         $labels = get_all_labels($owner_uid);
 465
 466                         _debug("" . count($filters) . " filters loaded.", $debug_enabled);
 467
 468                         $items = $rss->get_items();
 469
 470                         if (!is_array($items)) {
 471                                 _debug("no articles found.", $debug_enabled);
 472
 473                                 db_query("UPDATE ttrss_feeds
 474                                         SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
 475
 476                                 return; // no articles
 477                         }
 478
 479                         if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
 480
 481                                 _debug("checking for PUSH hub...", $debug_enabled);
 482
 483                                 $feed_hub_url = false;
 484
 485                                 $links = $rss->get_links('hub');
 486
 487                                 if ($links && is_array($links)) {
 488                                         foreach ($links as $l) {
 489                                                 $feed_hub_url = $l;
 490                                                 break;
 491                                         }
 492                                 }
 493
 494                                 _debug("feed hub url: $feed_hub_url", $debug_enabled);
 495
 496                                 if ($feed_hub_url && function_exists('curl_init') &&
 497                                         !ini_get("open_basedir")) {
 498
 499                                         require_once 'lib/pubsubhubbub/subscriber.php';
 500
 501                                         $callback_url = get_self_url_prefix() .
 502                                                 "/public.php?op=pubsub&id=$feed";
 503
 504                                         $s = new Subscriber($feed_hub_url, $callback_url);
 505
 506                                         $rc = $s->subscribe($fetch_url);
 507
 508                                         _debug("feed hub url found, subscribe request sent.", $debug_enabled);
 509
 510                                         db_query("UPDATE ttrss_feeds SET pubsub_state = 1
 511                                                 WHERE id = '$feed'");
 512                                 }
 513                         }
 514
 515                         _debug("processing articles...", $debug_enabled);
 516
 517                         foreach ($items as $item) {
 518                                 if ($_REQUEST['xdebug'] == 3) {
 519                                         print_r($item);
 520                                 }
 521
 522                                 $entry_guid = $item->get_id();
 523                                 if (!$entry_guid) $entry_guid = $item->get_link();
 524                                 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
 525
 526                                 _debug("f_guid $entry_guid", $debug_enabled);
 527
 528                                 if (!$entry_guid) continue;
 529
 530                                 $entry_guid = "$owner_uid,$entry_guid";
 531
 532                                 $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
 533
 534                                 _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
 535
 536                                 $entry_timestamp = "";
 537
 538                                 $entry_timestamp = $item->get_date();
 539
 540                                 _debug("orig date: " . $item->get_date(), $debug_enabled);
 541
 542                                 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
 543                                         $entry_timestamp = time();
 544                                         $no_orig_date = 'true';
 545                                 } else {
 546                                         $no_orig_date = 'false';
 547                                 }
 548
 549                                 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
 550
 551                                 _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
 552
 553 //                              $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
 554 //                              $entry_title = decode_numeric_entities($entry_title);
 555                                 $entry_title = $item->get_title();
 556
 557                                 $entry_link = rewrite_relative_url($site_url, $item->get_link());
 558
 559                                 _debug("title $entry_title", $debug_enabled);
 560                                 _debug("link $entry_link", $debug_enabled);
 561
 562                                 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
 563
 564                                 $entry_content = $item->get_content();
 565                                 if (!$entry_content) $entry_content = $item->get_description();
 566
 567                                 if ($_REQUEST["xdebug"] == 2) {
 568                                         print "content: ";
 569                                         print $entry_content;
 570                                         print "\n";
 571                                 }
 572
 573                                 $entry_language = $lang->detect($entry_content, 1);
 574
 575                                 if (count($entry_language) > 0) {
 576                                         $entry_language = array_keys($entry_language);
 577                                         $entry_language = db_escape_string(substr($entry_language[0], 0, 2));
 578
 579                                         _debug("detected language: $entry_language", $debug_enabled);
 580                                 } else{
 581                                         $entry_language = "";
 582                                 }
 583
 584                                 $entry_comments = $item->get_comments_url();
 585                                 $entry_author = $item->get_author();
 586
 587                                 $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
 588
 589                                 $entry_comments = db_escape_string(mb_substr(trim($entry_comments), 0, 245));
 590                                 $entry_author = db_escape_string(mb_substr(trim($entry_author), 0, 245));
 591
 592                                 $num_comments = (int) $item->get_comments_count();
 593
 594                                 _debug("author $entry_author", $debug_enabled);
 595                                 _debug("num_comments: $num_comments", $debug_enabled);
 596                                 _debug("looking for tags...", $debug_enabled);
 597
 598                                 // parse <category> entries into tags
 599
 600                                 $additional_tags = array();
 601
 602                                 $additional_tags_src = $item->get_categories();
 603
 604                                 if (is_array($additional_tags_src)) {
 605                                         foreach ($additional_tags_src as $tobj) {
 606                                                 array_push($additional_tags, $tobj);
 607                                         }
 608                                 }
 609
 610                                 $entry_tags = array_unique($additional_tags);
 611
 612                                 for ($i = 0; $i < count($entry_tags); $i++)
 613                                         $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
 614
 615                                 _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
 616
 617                                 _debug("done collecting data.", $debug_enabled);
 618
 619                                 // TODO: less memory-hungry implementation
 620
 621                                 _debug("applying plugin filters..", $debug_enabled);
 622
 623                                 // FIXME not sure if owner_uid is a good idea here, we may have a base entry without user entry (?)
 624                                 $result = db_query("SELECT plugin_data,title,content,link,tag_cache,author FROM ttrss_entries, ttrss_user_entries
 625                                         WHERE ref_id = id AND (guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed') AND owner_uid = $owner_uid");
 626
 627                                 if (db_num_rows($result) != 0) {
 628                                         $entry_plugin_data = db_fetch_result($result, 0, "plugin_data");
 629                                         $stored_article = array("title" => db_fetch_result($result, 0, "title"),
 630                                                 "content" => db_fetch_result($result, 0, "content"),
 631                                                 "link" => db_fetch_result($result, 0, "link"),
 632                                                 "tags" => explode(",", db_fetch_result($result, 0, "tag_cache")),
 633                                                 "author" => db_fetch_result($result, 0, "author"));
 634                                 } else {
 635                                         $entry_plugin_data = "";
 636                                         $stored_article = array();
 637                                 }
 638
 639                                 $article = array("owner_uid" => $owner_uid, // read only
 640                                         "guid" => $entry_guid, // read only
 641                                         "title" => $entry_title,
 642                                         "content" => $entry_content,
 643                                         "link" => $entry_link,
 644                                         "tags" => $entry_tags,
 645                                         "plugin_data" => $entry_plugin_data,
 646                                         "author" => $entry_author,
 647                                         "stored" => $stored_article);
 648
 649                                 foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
 650                                         $article = $plugin->hook_article_filter($article);
 651                                 }
 652
 653                                 $entry_tags = $article["tags"];
 654                                 $entry_guid = db_escape_string($entry_guid);
 655                                 $entry_title = db_escape_string($article["title"]);
 656                                 $entry_author = db_escape_string($article["author"]);
 657                                 $entry_link = db_escape_string($article["link"]);
 658                                 $entry_plugin_data = db_escape_string($article["plugin_data"]);
 659                                 $entry_content = $article["content"]; // escaped below
 660
 661
 662                                 _debug("plugin data: $entry_plugin_data", $debug_enabled);
 663
 664                                 if ($cache_images && is_writable(CACHE_DIR . '/images'))
 665                                         cache_images($entry_content, $site_url, $debug_enabled);
 666
 667                                 $entry_content = db_escape_string($entry_content, false);
 668
 669                                 $content_hash = "SHA1:" . sha1($entry_content);
 670
 671                                 db_query("BEGIN");
 672
 673                                 $result = db_query("SELECT id FROM      ttrss_entries
 674                                         WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
 675
 676                                 if (db_num_rows($result) == 0) {
 677
 678                                         _debug("base guid [$entry_guid] not found", $debug_enabled);
 679
 680                                         // base post entry does not exist, create it
 681
 682                                         $result = db_query(
 683                                                 "INSERT INTO ttrss_entries
 684                                                         (title,
 685                                                         guid,
 686                                                         link,
 687                                                         updated,
 688                                                         content,
 689                                                         content_hash,
 690                                                         no_orig_date,
 691                                                         date_updated,
 692                                                         date_entered,
 693                                                         comments,
 694                                                         num_comments,
 695                                                         plugin_data,
 696                                                         lang,
 697                                                         author)
 698                                                 VALUES
 699                                                         ('$entry_title',
 700                                                         '$entry_guid_hashed',
 701                                                         '$entry_link',
 702                                                         '$entry_timestamp_fmt',
 703                                                         '$entry_content',
 704                                                         '$content_hash',
 705                                                         $no_orig_date,
 706                                                         NOW(),
 707                                                         '$date_feed_processed',
 708                                                         '$entry_comments',
 709                                                         '$num_comments',
 710                                                         '$entry_plugin_data',
 711                                                         '$entry_language',
 712                                                         '$entry_author')");
 713
 714                                         $article_labels = array();
 715
 716                                 } else {
 717                                         // we keep encountering the entry in feeds, so we need to
 718                                         // update date_updated column so that we don't get horrible
 719                                         // dupes when the entry gets purged and reinserted again e.g.
 720                                         // in the case of SLOW SLOW OMG SLOW updating feeds
 721
 722                                         $base_entry_id = db_fetch_result($result, 0, "id");
 723
 724                                         db_query("UPDATE ttrss_entries SET date_updated = NOW()
 725                                                 WHERE id = '$base_entry_id'");
 726
 727                                         $article_labels = get_article_labels($base_entry_id, $owner_uid);
 728                                 }
 729
 730                                 // now it should exist, if not - bad luck then
 731
 732                                 $result = db_query("SELECT
 733                                                 id,content_hash,no_orig_date,title,plugin_data,guid,
 734                                                 ".SUBSTRING_FOR_DATE."(date_updated,1,19) as date_updated,
 735                                                 ".SUBSTRING_FOR_DATE."(updated,1,19) as updated,
 736                                                 num_comments
 737                                         FROM
 738                                                 ttrss_entries
 739                                         WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
 740
 741                                 $entry_ref_id = 0;
 742                                 $entry_int_id = 0;
 743
 744                                 if (db_num_rows($result) == 1) {
 745
 746                                         _debug("base guid found, checking for user record", $debug_enabled);
 747
 748                                         // this will be used below in update handler
 749                                         $orig_content_hash = db_fetch_result($result, 0, "content_hash");
 750                                         $orig_title = db_fetch_result($result, 0, "title");
 751                                         $orig_num_comments = db_fetch_result($result, 0, "num_comments");
 752                                         $orig_date_updated = strtotime(db_fetch_result($result,
 753                                                 0, "date_updated"));
 754                                         $orig_plugin_data = db_fetch_result($result, 0, "plugin_data");
 755
 756                                         $ref_id = db_fetch_result($result, 0, "id");
 757                                         $entry_ref_id = $ref_id;
 758
 759                                         /* $stored_guid = db_fetch_result($result, 0, "guid");
 760                                         if ($stored_guid != $entry_guid_hashed) {
 761                                                 if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
 762
 763                                                 db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
 764                                                         id = '$ref_id'");
 765                                         } */
 766
 767                                         // check for user post link to main table
 768
 769                                         // do we allow duplicate posts with same GUID in different feeds?
 770                                         if (get_pref("ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
 771                                                 $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
 772                                         } else {
 773                                                 $dupcheck_qpart = "";
 774                                         }
 775
 776                                         /* Collect article tags here so we could filter by them: */
 777
 778                                         $article_filters = get_article_filters($filters, $entry_title,
 779                                                 $entry_content, $entry_link, $entry_timestamp, $entry_author,
 780                                                 $entry_tags);
 781
 782                                         if ($debug_enabled) {
 783                                                 _debug("article filters: ", $debug_enabled);
 784                                                 if (count($article_filters) != 0) {
 785                                                         print_r($article_filters);
 786                                                 }
 787                                         }
 788
 789                                         if (find_article_filter($article_filters, "filter")) {
 790                                                 db_query("COMMIT"); // close transaction in progress
 791                                                 continue;
 792                                         }
 793
 794                                         $score = calculate_article_score($article_filters);
 795
 796                                         _debug("initial score: $score", $debug_enabled);
 797
 798                                         $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
 799                                                         ref_id = '$ref_id' AND owner_uid = '$owner_uid'
 800                                                         $dupcheck_qpart";
 801
 802 //                                      if ($_REQUEST["xdebug"]) print "$query\n";
 803
 804                                         $result = db_query($query);
 805
 806                                         // okay it doesn't exist - create user entry
 807                                         if (db_num_rows($result) == 0) {
 808
 809                                                 _debug("user record not found, creating...", $debug_enabled);
 810
 811                                                 if ($score >= -500 && !find_article_filter($article_filters, 'catchup')) {
 812                                                         $unread = 'true';
 813                                                         $last_read_qpart = 'NULL';
 814                                                 } else {
 815                                                         $unread = 'false';
 816                                                         $last_read_qpart = 'NOW()';
 817                                                 }
 818
 819                                                 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
 820                                                         $marked = 'true';
 821                                                 } else {
 822                                                         $marked = 'false';
 823                                                 }
 824
 825                                                 if (find_article_filter($article_filters, 'publish')) {
 826                                                         $published = 'true';
 827                                                 } else {
 828                                                         $published = 'false';
 829                                                 }
 830
 831                                                 // N-grams
 832
 833                                                 if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
 834
 835                                                         $result = db_query("SELECT COUNT(*) AS similar FROM
 836                                                                         ttrss_entries,ttrss_user_entries
 837                                                                 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
 838                                                                         AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
 839                                                                         AND owner_uid = $owner_uid");
 840
 841                                                         $ngram_similar = db_fetch_result($result, 0, "similar");
 842
 843                                                         _debug("N-gram similar results: $ngram_similar", $debug_enabled);
 844
 845                                                         if ($ngram_similar > 0) {
 846                                                                 $unread = 'false';
 847                                                         }
 848                                                 }
 849
 850                                                 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
 851                                                 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
 852
 853                                                 $result = db_query(
 854                                                         "INSERT INTO ttrss_user_entries
 855                                                                 (ref_id, owner_uid, feed_id, unread, last_read, marked,
 856                                                                 published, score, tag_cache, label_cache, uuid,
 857                                                                 last_marked, last_published)
 858                                                         VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
 859                                                                 $last_read_qpart, $marked, $published, '$score', '', '',
 860                                                                 '', $last_marked, $last_published)");
 861
 862                                                 if (PUBSUBHUBBUB_HUB && $published == 'true') {
 863                                                         $rss_link = get_self_url_prefix() .
 864                                                                 "/public.php?op=rss&id=-2&key=" .
 865                                                                 get_feed_access_key(-2, false, $owner_uid);
 866
 867                                                         $p = new Publisher(PUBSUBHUBBUB_HUB);
 868
 869                                                         $pubsub_result = $p->publish_update($rss_link);
 870                                                 }
 871
 872                                                 $result = db_query(
 873                                                         "SELECT int_id FROM ttrss_user_entries WHERE
 874                                                                 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
 875                                                                 feed_id = '$feed' LIMIT 1");
 876
 877                                                 if (db_num_rows($result) == 1) {
 878                                                         $entry_int_id = db_fetch_result($result, 0, "int_id");
 879                                                 }
 880                                         } else {
 881                                                 _debug("user record FOUND", $debug_enabled);
 882
 883                                                 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
 884                                                 $entry_int_id = db_fetch_result($result, 0, "int_id");
 885                                         }
 886
 887                                         _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
 888
 889                                         $post_needs_update = false;
 890                                         $update_insignificant = false;
 891
 892                                         if ($orig_num_comments != $num_comments) {
 893                                                 $post_needs_update = true;
 894                                                 $update_insignificant = true;
 895                                         }
 896
 897                                         if ($entry_plugin_data != $orig_plugin_data) {
 898                                                 $post_needs_update = true;
 899                                                 $update_insignificant = true;
 900                                         }
 901
 902                                         if ($content_hash != $orig_content_hash) {
 903                                                 $post_needs_update = true;
 904                                                 $update_insignificant = false;
 905                                         }
 906
 907                                         if (db_escape_string($orig_title) != $entry_title) {
 908                                                 $post_needs_update = true;
 909                                                 $update_insignificant = false;
 910                                         }
 911
 912                                         // if post needs update, update it and mark all user entries
 913                                         // linking to this post as updated
 914                                         if ($post_needs_update) {
 915
 916                                                 if (defined('DAEMON_EXTENDED_DEBUG')) {
 917                                                         _debug("post $entry_guid_hashed needs update...", $debug_enabled);
 918                                                 }
 919
 920 //                                              print "<!-- post $orig_title needs update : $post_needs_update -->";
 921
 922                                                 db_query("UPDATE ttrss_entries
 923                                                         SET title = '$entry_title', content = '$entry_content',
 924                                                                 content_hash = '$content_hash',
 925                                                                 updated = '$entry_timestamp_fmt',
 926                                                                 num_comments = '$num_comments',
 927                                                                 plugin_data = '$entry_plugin_data'
 928                                                         WHERE id = '$ref_id'");
 929
 930                                                 if (!$update_insignificant) {
 931                                                         if ($mark_unread_on_update) {
 932                                                                 db_query("UPDATE ttrss_user_entries
 933                                                                         SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
 934                                                         }
 935                                                 }
 936                                         }
 937                                 }
 938
 939                                 db_query("COMMIT");
 940
 941                                 _debug("assigning labels...", $debug_enabled);
 942
 943                                 assign_article_to_label_filters($entry_ref_id, $article_filters,
 944                                         $owner_uid, $article_labels);
 945
 946                                 _debug("looking for enclosures...", $debug_enabled);
 947
 948                                 // enclosures
 949
 950                                 $enclosures = array();
 951
 952                                 $encs = $item->get_enclosures();
 953
 954                                 if (is_array($encs)) {
 955                                         foreach ($encs as $e) {
 956                                                 $e_item = array(
 957                                                         $e->link, $e->type, $e->length);
 958                                                 array_push($enclosures, $e_item);
 959                                         }
 960                                 }
 961
 962                                 if ($debug_enabled) {
 963                                         _debug("article enclosures:", $debug_enabled);
 964                                         print_r($enclosures);
 965                                 }
 966
 967                                 db_query("BEGIN");
 968
 969                                 foreach ($enclosures as $enc) {
 970                                         $enc_url = db_escape_string($enc[0]);
 971                                         $enc_type = db_escape_string($enc[1]);
 972                                         $enc_dur = db_escape_string($enc[2]);
 973
 974                                         $result = db_query("SELECT id FROM ttrss_enclosures
 975                                                 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
 976
 977                                         if (db_num_rows($result) == 0) {
 978                                                 db_query("INSERT INTO ttrss_enclosures
 979                                                         (content_url, content_type, title, duration, post_id) VALUES
 980                                                         ('$enc_url', '$enc_type', '', '$enc_dur', '$entry_ref_id')");
 981                                         }
 982                                 }
 983
 984                                 db_query("COMMIT");
 985
 986                                 // check for manual tags (we have to do it here since they're loaded from filters)
 987
 988                                 foreach ($article_filters as $f) {
 989                                         if ($f["type"] == "tag") {
 990
 991                                                 $manual_tags = trim_array(explode(",", $f["param"]));
 992
 993                                                 foreach ($manual_tags as $tag) {
 994                                                         if (tag_is_valid($tag)) {
 995                                                                 array_push($entry_tags, $tag);
 996                                                         }
 997                                                 }
 998                                         }
 999                                 }
1000
1001                                 // Skip boring tags
1002
1003                                 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
1004                                         'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1005
1006                                 $filtered_tags = array();
1007                                 $tags_to_cache = array();
1008
1009                                 if ($entry_tags && is_array($entry_tags)) {
1010                                         foreach ($entry_tags as $tag) {
1011                                                 if (array_search($tag, $boring_tags) === false) {
1012                                                         array_push($filtered_tags, $tag);
1013                                                 }
1014                                         }
1015                                 }
1016
1017                                 $filtered_tags = array_unique($filtered_tags);
1018
1019                                 if ($debug_enabled) {
1020                                         _debug("filtered article tags:", $debug_enabled);
1021                                         print_r($filtered_tags);
1022                                 }
1023
1024                                 // Save article tags in the database
1025
1026                                 if (count($filtered_tags) > 0) {
1027
1028                                         db_query("BEGIN");
1029
1030                                         foreach ($filtered_tags as $tag) {
1031
1032                                                 $tag = sanitize_tag($tag);
1033                                                 $tag = db_escape_string($tag);
1034
1035                                                 if (!tag_is_valid($tag)) continue;
1036
1037                                                 $result = db_query("SELECT id FROM ttrss_tags
1038                                                         WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1039                                                         owner_uid = '$owner_uid' LIMIT 1");
1040
1041                                                         if ($result && db_num_rows($result) == 0) {
1042
1043                                                                 db_query("INSERT INTO ttrss_tags
1044                                                                         (owner_uid,tag_name,post_int_id)
1045                                                                         VALUES ('$owner_uid','$tag', '$entry_int_id')");
1046                                                         }
1047
1048                                                 array_push($tags_to_cache, $tag);
1049                                         }
1050
1051                                         /* update the cache */
1052
1053                                         $tags_to_cache = array_unique($tags_to_cache);
1054
1055                                         $tags_str = db_escape_string(join(",", $tags_to_cache));
1056
1057                                         db_query("UPDATE ttrss_user_entries
1058                                                 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1059                                                 AND owner_uid = $owner_uid");
1060
1061                                         db_query("COMMIT");
1062                                 }
1063
1064                                 if (get_pref("AUTO_ASSIGN_LABELS", $owner_uid, false)) {
1065                                         _debug("auto-assigning labels...", $debug_enabled);
1066
1067                                         foreach ($labels as $label) {
1068                                                 $caption = preg_quote($label["caption"]);
1069
1070                                                 if ($caption && preg_match("/\b$caption\b/i", "$tags_str " . strip_tags($entry_content) . " $entry_title")) {
1071                                                         if (!labels_contains_caption($article_labels, $caption)) {
1072                                                                 label_add_article($entry_ref_id, $caption, $owner_uid);
1073                                                         }
1074                                                 }
1075                                         }
1076                                 }
1077
1078                                 _debug("article processed", $debug_enabled);
1079                         }
1080
1081                         _debug("purging feed...", $debug_enabled);
1082
1083                         purge_feed($feed, 0, $debug_enabled);
1084
1085                         db_query("UPDATE ttrss_feeds
1086                                 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
1087
1088 //                      db_query("COMMIT");
1089
1090                 } else {
1091
1092                         $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
1093
1094                         _debug("error fetching feed: $error_msg", $debug_enabled);
1095
1096                         db_query(
1097                                 "UPDATE ttrss_feeds SET last_error = '$error_msg',
1098                                         last_updated = NOW() WHERE id = '$feed'");
1099                 }
1100
1101                 unset($rss);
1102
1103                 _debug("done", $debug_enabled);
1104         }
1105
1106         function cache_images($html, $site_url, $debug) {
1107                 $cache_dir = CACHE_DIR . "/images";
1108
1109                 libxml_use_internal_errors(true);
1110
1111                 $charset_hack = '<head>
1112                         <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1113                 </head>';
1114
1115                 $doc = new DOMDocument();
1116                 $doc->loadHTML($charset_hack . $html);
1117                 $xpath = new DOMXPath($doc);
1118
1119                 $entries = $xpath->query('(//img[@src])');
1120
1121                 foreach ($entries as $entry) {
1122                         if ($entry->hasAttribute('src')) {
1123                                 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1124
1125                                 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1126
1127                                 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1128
1129                                 if (!file_exists($local_filename)) {
1130                                         $file_content = fetch_file_contents($src);
1131
1132                                         if ($file_content && strlen($file_content) > 1024) {
1133                                                 file_put_contents($local_filename, $file_content);
1134                                         }
1135                                 }
1136
1137                                 /* if (file_exists($local_filename)) {
1138                                         $entry->setAttribute('src', SELF_URL_PATH . '/image.php?url=' .
1139                                                 base64_encode($src));
1140                                 } */
1141                         }
1142                 }
1143
1144                 //$node = $doc->getElementsByTagName('body')->item(0);
1145                 //return $doc->saveXML($node);
1146         }
1147
1148         function expire_error_log($debug) {
1149                 if ($debug) _debug("Removing old error log entries...");
1150
1151                 if (DB_TYPE == "pgsql") {
1152                         db_query("DELETE FROM ttrss_error_log
1153                                 WHERE created_at < NOW() - INTERVAL '7 days'");
1154                 } else {
1155                         db_query("DELETE FROM ttrss_error_log
1156                                 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1157                 }
1158
1159         }
1160
1161         function expire_lock_files($debug) {
1162                 //if ($debug) _debug("Removing old lock files...");
1163
1164                 $num_deleted = 0;
1165
1166                 if (is_writable(LOCK_DIRECTORY)) {
1167                         $files = glob(LOCK_DIRECTORY . "/*.lock");
1168
1169                         if ($files) {
1170                                 foreach ($files as $file) {
1171                                         if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
1172                                                 unlink($file);
1173                                                 ++$num_deleted;
1174                                         }
1175                                 }
1176                         }
1177                 }
1178
1179                 if ($debug) _debug("Removed $num_deleted old lock files.");
1180         }
1181
1182         function expire_cached_files($debug) {
1183                 foreach (array("simplepie", "images", "export", "upload") as $dir) {
1184                         $cache_dir = CACHE_DIR . "/$dir";
1185
1186 //                      if ($debug) _debug("Expiring $cache_dir");
1187
1188                         $num_deleted = 0;
1189
1190                         if (is_writable($cache_dir)) {
1191                                 $files = glob("$cache_dir/*");
1192
1193                                 if ($files) {
1194                                         foreach ($files as $file) {
1195                                                 if (time() - filemtime($file) > 86400*7) {
1196                                                         unlink($file);
1197
1198                                                         ++$num_deleted;
1199                                                 }
1200                                         }
1201                                 }
1202                         }
1203
1204                         if ($debug) _debug("$cache_dir: removed $num_deleted files.");
1205                 }
1206         }
1207
1208         /**
1209         * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1210         * Returns the url query as associative array
1211         *
1212         * @param    string    query
1213         * @return    array    params
1214         */
1215         function convertUrlQuery($query) {
1216                 $queryParts = explode('&', $query);
1217
1218                 $params = array();
1219
1220                 foreach ($queryParts as $param) {
1221                         $item = explode('=', $param);
1222                         $params[$item[0]] = $item[1];
1223                 }
1224
1225                 return $params;
1226         }
1227
1228         function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
1229                 $matches = array();
1230
1231                 foreach ($filters as $filter) {
1232                         $match_any_rule = $filter["match_any_rule"];
1233                         $inverse = $filter["inverse"];
1234                         $filter_match = false;
1235
1236                         foreach ($filter["rules"] as $rule) {
1237                                 $match = false;
1238                                 $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
1239                                 $rule_inverse = $rule["inverse"];
1240
1241                                 if (!$reg_exp)
1242                                         continue;
1243
1244                                 switch ($rule["type"]) {
1245                                 case "title":
1246                                         $match = @preg_match("/$reg_exp/i", $title);
1247                                         break;
1248                                 case "content":
1249                                         // we don't need to deal with multiline regexps
1250                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1251
1252                                         $match = @preg_match("/$reg_exp/i", $content);
1253                                         break;
1254                                 case "both":
1255                                         // we don't need to deal with multiline regexps
1256                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1257
1258                                         $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
1259                                         break;
1260                                 case "link":
1261                                         $match = @preg_match("/$reg_exp/i", $link);
1262                                         break;
1263                                 case "author":
1264                                         $match = @preg_match("/$reg_exp/i", $author);
1265                                         break;
1266                                 case "tag":
1267                                         foreach ($tags as $tag) {
1268                                                 if (@preg_match("/$reg_exp/i", $tag)) {
1269                                                         $match = true;
1270                                                         break;
1271                                                 }
1272                                         }
1273                                         break;
1274                                 }
1275
1276                                 if ($rule_inverse) $match = !$match;
1277
1278                                 if ($match_any_rule) {
1279                                         if ($match) {
1280                                                 $filter_match = true;
1281                                                 break;
1282                                         }
1283                                 } else {
1284                                         $filter_match = $match;
1285                                         if (!$match) {
1286                                                 break;
1287                                         }
1288                                 }
1289                         }
1290
1291                         if ($inverse) $filter_match = !$filter_match;
1292
1293                         if ($filter_match) {
1294                                 foreach ($filter["actions"] AS $action) {
1295                                         array_push($matches, $action);
1296
1297                                         // if Stop action encountered, perform no further processing
1298                                         if ($action["type"] == "stop") return $matches;
1299                                 }
1300                         }
1301                 }
1302
1303                 return $matches;
1304         }
1305
1306         function find_article_filter($filters, $filter_name) {
1307                 foreach ($filters as $f) {
1308                         if ($f["type"] == $filter_name) {
1309                                 return $f;
1310                         };
1311                 }
1312                 return false;
1313         }
1314
1315         function find_article_filters($filters, $filter_name) {
1316                 $results = array();
1317
1318                 foreach ($filters as $f) {
1319                         if ($f["type"] == $filter_name) {
1320                                 array_push($results, $f);
1321                         };
1322                 }
1323                 return $results;
1324         }
1325
1326         function calculate_article_score($filters) {
1327                 $score = 0;
1328
1329                 foreach ($filters as $f) {
1330                         if ($f["type"] == "score") {
1331                                 $score += $f["param"];
1332                         };
1333                 }
1334                 return $score;
1335         }
1336
1337         function labels_contains_caption($labels, $caption) {
1338                 foreach ($labels as $label) {
1339                         if ($label[1] == $caption) {
1340                                 return true;
1341                         }
1342                 }
1343
1344                 return false;
1345         }
1346
1347         function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
1348                 foreach ($filters as $f) {
1349                         if ($f["type"] == "label") {
1350                                 if (!labels_contains_caption($article_labels, $f["param"])) {
1351                                         label_add_article($id, $f["param"], $owner_uid);
1352                                 }
1353                         }
1354                 }
1355         }
1356
1357         function make_guid_from_title($title) {
1358                 return preg_replace("/[ \"\',.:;]/", "-",
1359                         mb_strtolower(strip_tags($title), 'utf-8'));
1360         }
1361
1362         /* function verify_feed_xml($feed_data) {
1363                 libxml_use_internal_errors(true);
1364                 $doc = new DOMDocument();
1365                 $doc->loadXML($feed_data);
1366                 $error = libxml_get_last_error();
1367                 libxml_clear_errors();
1368                 return $error;
1369         } */
1370
1371         function housekeeping_common($debug) {
1372                 expire_cached_files($debug);
1373                 expire_lock_files($debug);
1374                 expire_error_log($debug);
1375
1376                 $count = update_feedbrowser_cache();
1377                 _debug("Feedbrowser updated, $count feeds processed.");
1378
1379                 purge_orphans( true);
1380                 $rc = cleanup_tags( 14, 50000);
1381
1382                 _debug("Cleaned $rc cached tags.");
1383
1384                 PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
1385
1386         }
1387 ?>