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                         $rss->init();
 355                 }
 356
 357 //              print_r($rss);
 358
 359                 $feed = db_escape_string($feed);
 360
 361                 if (!$rss->error()) {
 362
 363                         // cache data for later
 364                         if (!$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
 365                                 $new_rss_hash = sha1($rss_data);
 366
 367                                 if ($new_rss_hash != $rss_hash && count($rss->get_items()) > 0 ) {
 368                                         _debug("saving $cache_filename", $debug_enabled);
 369                                         @file_put_contents($cache_filename, $feed_data);
 370                                 }
 371                         }
 372
 373                         // We use local pluginhost here because we need to load different per-user feed plugins
 374                         $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
 375
 376                         _debug("processing feed data...", $debug_enabled);
 377
 378 //                      db_query("BEGIN");
 379
 380                         if (DB_TYPE == "pgsql") {
 381                                 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
 382                         } else {
 383                                 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
 384                         }
 385
 386                         $result = db_query("SELECT title,site_url,owner_uid,favicon_avg_color,
 387                                 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
 388                                                 favicon_needs_check
 389                                 FROM ttrss_feeds WHERE id = '$feed'");
 390
 391                         $registered_title = db_fetch_result($result, 0, "title");
 392                         $orig_site_url = db_fetch_result($result, 0, "site_url");
 393                         $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
 394                                 "favicon_needs_check"));
 395                         $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
 396
 397                         $owner_uid = db_fetch_result($result, 0, "owner_uid");
 398
 399                         $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
 400
 401                         _debug("site_url: $site_url", $debug_enabled);
 402                         _debug("feed_title: " . $rss->get_title(), $debug_enabled);
 403
 404                         if ($favicon_needs_check || $force_refetch) {
 405
 406                                 /* terrible hack: if we crash on floicon shit here, we won't check
 407                                  * the icon avgcolor again (unless the icon got updated) */
 408
 409                                 $favicon_file = ICONS_DIR . "/$feed.ico";
 410                                 $favicon_modified = @filemtime($favicon_file);
 411
 412                                 _debug("checking favicon...", $debug_enabled);
 413
 414                                 check_feed_favicon($site_url, $feed);
 415                                 $favicon_modified_new = @filemtime($favicon_file);
 416
 417                                 if ($favicon_modified_new > $favicon_modified)
 418                                         $favicon_avg_color = '';
 419
 420                                 if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
 421                                                 require_once "colors.php";
 422
 423                                                 db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
 424                                                         id = '$feed'");
 425
 426                                                 $favicon_color = db_escape_string(
 427                                                         calculate_avg_color($favicon_file));
 428
 429                                                 $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
 430                                 } else if ($favicon_avg_color == 'fail') {
 431                                         _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
 432                                 }
 433
 434                                 db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
 435                                         $favicon_colorstring
 436                                         WHERE id = '$feed'");
 437                         }
 438
 439                         if (!$registered_title || $registered_title == "[Unknown]") {
 440
 441                                 $feed_title = db_escape_string($rss->get_title());
 442
 443                                 if ($feed_title) {
 444                                         _debug("registering title: $feed_title", $debug_enabled);
 445
 446                                         db_query("UPDATE ttrss_feeds SET
 447                                                 title = '$feed_title' WHERE id = '$feed'");
 448                                 }
 449                         }
 450
 451                         if ($site_url && $orig_site_url != $site_url) {
 452                                 db_query("UPDATE ttrss_feeds SET
 453                                         site_url = '$site_url' WHERE id = '$feed'");
 454                         }
 455
 456                         _debug("loading filters & labels...", $debug_enabled);
 457
 458                         $filters = load_filters($feed, $owner_uid);
 459                         $labels = get_all_labels($owner_uid);
 460
 461                         _debug("" . count($filters) . " filters loaded.", $debug_enabled);
 462
 463                         $items = $rss->get_items();
 464
 465                         if (!is_array($items)) {
 466                                 _debug("no articles found.", $debug_enabled);
 467
 468                                 db_query("UPDATE ttrss_feeds
 469                                         SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
 470
 471                                 return; // no articles
 472                         }
 473
 474                         if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
 475
 476                                 _debug("checking for PUSH hub...", $debug_enabled);
 477
 478                                 $feed_hub_url = false;
 479
 480                                 $links = $rss->get_links('hub');
 481
 482                                 if ($links && is_array($links)) {
 483                                         foreach ($links as $l) {
 484                                                 $feed_hub_url = $l;
 485                                                 break;
 486                                         }
 487                                 }
 488
 489                                 _debug("feed hub url: $feed_hub_url", $debug_enabled);
 490
 491                                 if ($feed_hub_url && function_exists('curl_init') &&
 492                                         !ini_get("open_basedir")) {
 493
 494                                         require_once 'lib/pubsubhubbub/subscriber.php';
 495
 496                                         $callback_url = get_self_url_prefix() .
 497                                                 "/public.php?op=pubsub&id=$feed";
 498
 499                                         $s = new Subscriber($feed_hub_url, $callback_url);
 500
 501                                         $rc = $s->subscribe($fetch_url);
 502
 503                                         _debug("feed hub url found, subscribe request sent.", $debug_enabled);
 504
 505                                         db_query("UPDATE ttrss_feeds SET pubsub_state = 1
 506                                                 WHERE id = '$feed'");
 507                                 }
 508                         }
 509
 510                         _debug("processing articles...", $debug_enabled);
 511
 512                         foreach ($items as $item) {
 513                                 if ($_REQUEST['xdebug'] == 3) {
 514                                         print_r($item);
 515                                 }
 516
 517                                 $entry_guid = $item->get_id();
 518                                 if (!$entry_guid) $entry_guid = $item->get_link();
 519                                 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
 520
 521                                 _debug("f_guid $entry_guid", $debug_enabled);
 522
 523                                 if (!$entry_guid) continue;
 524
 525                                 $entry_guid = "$owner_uid,$entry_guid";
 526
 527                                 $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
 528
 529                                 _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
 530
 531                                 $entry_timestamp = "";
 532
 533                                 $entry_timestamp = $item->get_date();
 534
 535                                 _debug("orig date: " . $item->get_date(), $debug_enabled);
 536
 537                                 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
 538                                         $entry_timestamp = time();
 539                                         $no_orig_date = 'true';
 540                                 } else {
 541                                         $no_orig_date = 'false';
 542                                 }
 543
 544                                 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
 545
 546                                 _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
 547
 548 //                              $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
 549 //                              $entry_title = decode_numeric_entities($entry_title);
 550                                 $entry_title = $item->get_title();
 551
 552                                 $entry_link = rewrite_relative_url($site_url, $item->get_link());
 553
 554                                 _debug("title $entry_title", $debug_enabled);
 555                                 _debug("link $entry_link", $debug_enabled);
 556
 557                                 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
 558
 559                                 $entry_content = $item->get_content();
 560                                 if (!$entry_content) $entry_content = $item->get_description();
 561
 562                                 if ($_REQUEST["xdebug"] == 2) {
 563                                         print "content: ";
 564                                         print $entry_content;
 565                                         print "\n";
 566                                 }
 567
 568                                 $entry_comments = $item->get_comments_url();
 569                                 $entry_author = $item->get_author();
 570
 571                                 $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
 572
 573                                 $entry_comments = db_escape_string(mb_substr(trim($entry_comments), 0, 245));
 574                                 $entry_author = db_escape_string(mb_substr(trim($entry_author), 0, 245));
 575
 576                                 $num_comments = (int) $item->get_comments_count();
 577
 578                                 _debug("author $entry_author", $debug_enabled);
 579                                 _debug("num_comments: $num_comments", $debug_enabled);
 580                                 _debug("looking for tags...", $debug_enabled);
 581
 582                                 // parse <category> entries into tags
 583
 584                                 $additional_tags = array();
 585
 586                                 $additional_tags_src = $item->get_categories();
 587
 588                                 if (is_array($additional_tags_src)) {
 589                                         foreach ($additional_tags_src as $tobj) {
 590                                                 array_push($additional_tags, $tobj);
 591                                         }
 592                                 }
 593
 594                                 $entry_tags = array_unique($additional_tags);
 595
 596                                 for ($i = 0; $i < count($entry_tags); $i++)
 597                                         $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
 598
 599                                 _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
 600
 601                                 _debug("done collecting data.", $debug_enabled);
 602
 603                                 // TODO: less memory-hungry implementation
 604
 605                                 _debug("applying plugin filters..", $debug_enabled);
 606
 607                                 // FIXME not sure if owner_uid is a good idea here, we may have a base entry without user entry (?)
 608                                 $result = db_query("SELECT plugin_data,title,content,link,tag_cache,author FROM ttrss_entries, ttrss_user_entries
 609                                         WHERE ref_id = id AND (guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed') AND owner_uid = $owner_uid");
 610
 611                                 if (db_num_rows($result) != 0) {
 612                                         $entry_plugin_data = db_fetch_result($result, 0, "plugin_data");
 613                                         $stored_article = array("title" => db_fetch_result($result, 0, "title"),
 614                                                 "content" => db_fetch_result($result, 0, "content"),
 615                                                 "link" => db_fetch_result($result, 0, "link"),
 616                                                 "tags" => explode(",", db_fetch_result($result, 0, "tag_cache")),
 617                                                 "author" => db_fetch_result($result, 0, "author"));
 618                                 } else {
 619                                         $entry_plugin_data = "";
 620                                         $stored_article = array();
 621                                 }
 622
 623                                 $article = array("owner_uid" => $owner_uid, // read only
 624                                         "guid" => $entry_guid, // read only
 625                                         "title" => $entry_title,
 626                                         "content" => $entry_content,
 627                                         "link" => $entry_link,
 628                                         "tags" => $entry_tags,
 629                                         "plugin_data" => $entry_plugin_data,
 630                                         "author" => $entry_author,
 631                                         "stored" => $stored_article);
 632
 633                                 foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
 634                                         $article = $plugin->hook_article_filter($article);
 635                                 }
 636
 637                                 $entry_tags = $article["tags"];
 638                                 $entry_guid = db_escape_string($entry_guid);
 639                                 $entry_title = db_escape_string($article["title"]);
 640                                 $entry_author = db_escape_string($article["author"]);
 641                                 $entry_link = db_escape_string($article["link"]);
 642                                 $entry_plugin_data = db_escape_string($article["plugin_data"]);
 643                                 $entry_content = $article["content"]; // escaped below
 644
 645
 646                                 _debug("plugin data: $entry_plugin_data", $debug_enabled);
 647
 648                                 if ($cache_images && is_writable(CACHE_DIR . '/images'))
 649                                         cache_images($entry_content, $site_url, $debug_enabled);
 650
 651                                 $entry_content = db_escape_string($entry_content, false);
 652
 653                                 $content_hash = "SHA1:" . sha1($entry_content);
 654
 655                                 db_query("BEGIN");
 656
 657                                 $result = db_query("SELECT id FROM      ttrss_entries
 658                                         WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
 659
 660                                 if (db_num_rows($result) == 0) {
 661
 662                                         _debug("base guid [$entry_guid] not found", $debug_enabled);
 663
 664                                         // base post entry does not exist, create it
 665
 666                                         $result = db_query(
 667                                                 "INSERT INTO ttrss_entries
 668                                                         (title,
 669                                                         guid,
 670                                                         link,
 671                                                         updated,
 672                                                         content,
 673                                                         content_hash,
 674                                                         no_orig_date,
 675                                                         date_updated,
 676                                                         date_entered,
 677                                                         comments,
 678                                                         num_comments,
 679                                                         plugin_data,
 680                                                         author)
 681                                                 VALUES
 682                                                         ('$entry_title',
 683                                                         '$entry_guid_hashed',
 684                                                         '$entry_link',
 685                                                         '$entry_timestamp_fmt',
 686                                                         '$entry_content',
 687                                                         '$content_hash',
 688                                                         $no_orig_date,
 689                                                         NOW(),
 690                                                         '$date_feed_processed',
 691                                                         '$entry_comments',
 692                                                         '$num_comments',
 693                                                         '$entry_plugin_data',
 694                                                         '$entry_author')");
 695
 696                                         $article_labels = array();
 697
 698                                 } else {
 699                                         // we keep encountering the entry in feeds, so we need to
 700                                         // update date_updated column so that we don't get horrible
 701                                         // dupes when the entry gets purged and reinserted again e.g.
 702                                         // in the case of SLOW SLOW OMG SLOW updating feeds
 703
 704                                         $base_entry_id = db_fetch_result($result, 0, "id");
 705
 706                                         db_query("UPDATE ttrss_entries SET date_updated = NOW()
 707                                                 WHERE id = '$base_entry_id'");
 708
 709                                         $article_labels = get_article_labels($base_entry_id, $owner_uid);
 710                                 }
 711
 712                                 // now it should exist, if not - bad luck then
 713
 714                                 $result = db_query("SELECT
 715                                                 id,content_hash,no_orig_date,title,plugin_data,guid,
 716                                                 ".SUBSTRING_FOR_DATE."(date_updated,1,19) as date_updated,
 717                                                 ".SUBSTRING_FOR_DATE."(updated,1,19) as updated,
 718                                                 num_comments
 719                                         FROM
 720                                                 ttrss_entries
 721                                         WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
 722
 723                                 $entry_ref_id = 0;
 724                                 $entry_int_id = 0;
 725
 726                                 if (db_num_rows($result) == 1) {
 727
 728                                         _debug("base guid found, checking for user record", $debug_enabled);
 729
 730                                         // this will be used below in update handler
 731                                         $orig_content_hash = db_fetch_result($result, 0, "content_hash");
 732                                         $orig_title = db_fetch_result($result, 0, "title");
 733                                         $orig_num_comments = db_fetch_result($result, 0, "num_comments");
 734                                         $orig_date_updated = strtotime(db_fetch_result($result,
 735                                                 0, "date_updated"));
 736                                         $orig_plugin_data = db_fetch_result($result, 0, "plugin_data");
 737
 738                                         $ref_id = db_fetch_result($result, 0, "id");
 739                                         $entry_ref_id = $ref_id;
 740
 741                                         /* $stored_guid = db_fetch_result($result, 0, "guid");
 742                                         if ($stored_guid != $entry_guid_hashed) {
 743                                                 if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
 744
 745                                                 db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
 746                                                         id = '$ref_id'");
 747                                         } */
 748
 749                                         // check for user post link to main table
 750
 751                                         // do we allow duplicate posts with same GUID in different feeds?
 752                                         if (get_pref("ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
 753                                                 $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
 754                                         } else {
 755                                                 $dupcheck_qpart = "";
 756                                         }
 757
 758                                         /* Collect article tags here so we could filter by them: */
 759
 760                                         $article_filters = get_article_filters($filters, $entry_title,
 761                                                 $entry_content, $entry_link, $entry_timestamp, $entry_author,
 762                                                 $entry_tags);
 763
 764                                         if ($debug_enabled) {
 765                                                 _debug("article filters: ", $debug_enabled);
 766                                                 if (count($article_filters) != 0) {
 767                                                         print_r($article_filters);
 768                                                 }
 769                                         }
 770
 771                                         if (find_article_filter($article_filters, "filter")) {
 772                                                 db_query("COMMIT"); // close transaction in progress
 773                                                 continue;
 774                                         }
 775
 776                                         $score = calculate_article_score($article_filters);
 777
 778                                         _debug("initial score: $score", $debug_enabled);
 779
 780                                         $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
 781                                                         ref_id = '$ref_id' AND owner_uid = '$owner_uid'
 782                                                         $dupcheck_qpart";
 783
 784 //                                      if ($_REQUEST["xdebug"]) print "$query\n";
 785
 786                                         $result = db_query($query);
 787
 788                                         // okay it doesn't exist - create user entry
 789                                         if (db_num_rows($result) == 0) {
 790
 791                                                 _debug("user record not found, creating...", $debug_enabled);
 792
 793                                                 if ($score >= -500 && !find_article_filter($article_filters, 'catchup')) {
 794                                                         $unread = 'true';
 795                                                         $last_read_qpart = 'NULL';
 796                                                 } else {
 797                                                         $unread = 'false';
 798                                                         $last_read_qpart = 'NOW()';
 799                                                 }
 800
 801                                                 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
 802                                                         $marked = 'true';
 803                                                 } else {
 804                                                         $marked = 'false';
 805                                                 }
 806
 807                                                 if (find_article_filter($article_filters, 'publish')) {
 808                                                         $published = 'true';
 809                                                 } else {
 810                                                         $published = 'false';
 811                                                 }
 812
 813                                                 // N-grams
 814
 815                                                 if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
 816
 817                                                         $result = db_query("SELECT COUNT(*) AS similar FROM
 818                                                                         ttrss_entries,ttrss_user_entries
 819                                                                 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
 820                                                                         AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
 821                                                                         AND owner_uid = $owner_uid");
 822
 823                                                         $ngram_similar = db_fetch_result($result, 0, "similar");
 824
 825                                                         _debug("N-gram similar results: $ngram_similar", $debug_enabled);
 826
 827                                                         if ($ngram_similar > 0) {
 828                                                                 $unread = 'false';
 829                                                         }
 830                                                 }
 831
 832                                                 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
 833                                                 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
 834
 835                                                 $result = db_query(
 836                                                         "INSERT INTO ttrss_user_entries
 837                                                                 (ref_id, owner_uid, feed_id, unread, last_read, marked,
 838                                                                 published, score, tag_cache, label_cache, uuid,
 839                                                                 last_marked, last_published)
 840                                                         VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
 841                                                                 $last_read_qpart, $marked, $published, '$score', '', '',
 842                                                                 '', $last_marked, $last_published)");
 843
 844                                                 if (PUBSUBHUBBUB_HUB && $published == 'true') {
 845                                                         $rss_link = get_self_url_prefix() .
 846                                                                 "/public.php?op=rss&id=-2&key=" .
 847                                                                 get_feed_access_key(-2, false, $owner_uid);
 848
 849                                                         $p = new Publisher(PUBSUBHUBBUB_HUB);
 850
 851                                                         $pubsub_result = $p->publish_update($rss_link);
 852                                                 }
 853
 854                                                 $result = db_query(
 855                                                         "SELECT int_id FROM ttrss_user_entries WHERE
 856                                                                 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
 857                                                                 feed_id = '$feed' LIMIT 1");
 858
 859                                                 if (db_num_rows($result) == 1) {
 860                                                         $entry_int_id = db_fetch_result($result, 0, "int_id");
 861                                                 }
 862                                         } else {
 863                                                 _debug("user record FOUND", $debug_enabled);
 864
 865                                                 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
 866                                                 $entry_int_id = db_fetch_result($result, 0, "int_id");
 867                                         }
 868
 869                                         _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
 870
 871                                         $post_needs_update = false;
 872                                         $update_insignificant = false;
 873
 874                                         if ($orig_num_comments != $num_comments) {
 875                                                 $post_needs_update = true;
 876                                                 $update_insignificant = true;
 877                                         }
 878
 879                                         if ($entry_plugin_data != $orig_plugin_data) {
 880                                                 $post_needs_update = true;
 881                                                 $update_insignificant = true;
 882                                         }
 883
 884                                         if ($content_hash != $orig_content_hash) {
 885                                                 $post_needs_update = true;
 886                                                 $update_insignificant = false;
 887                                         }
 888
 889                                         if (db_escape_string($orig_title) != $entry_title) {
 890                                                 $post_needs_update = true;
 891                                                 $update_insignificant = false;
 892                                         }
 893
 894                                         // if post needs update, update it and mark all user entries
 895                                         // linking to this post as updated
 896                                         if ($post_needs_update) {
 897
 898                                                 if (defined('DAEMON_EXTENDED_DEBUG')) {
 899                                                         _debug("post $entry_guid_hashed needs update...", $debug_enabled);
 900                                                 }
 901
 902 //                                              print "<!-- post $orig_title needs update : $post_needs_update -->";
 903
 904                                                 db_query("UPDATE ttrss_entries
 905                                                         SET title = '$entry_title', content = '$entry_content',
 906                                                                 content_hash = '$content_hash',
 907                                                                 updated = '$entry_timestamp_fmt',
 908                                                                 num_comments = '$num_comments',
 909                                                                 plugin_data = '$entry_plugin_data'
 910                                                         WHERE id = '$ref_id'");
 911
 912                                                 if (!$update_insignificant) {
 913                                                         if ($mark_unread_on_update) {
 914                                                                 db_query("UPDATE ttrss_user_entries
 915                                                                         SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
 916                                                         }
 917                                                 }
 918                                         }
 919                                 }
 920
 921                                 db_query("COMMIT");
 922
 923                                 _debug("assigning labels...", $debug_enabled);
 924
 925                                 assign_article_to_label_filters($entry_ref_id, $article_filters,
 926                                         $owner_uid, $article_labels);
 927
 928                                 _debug("looking for enclosures...", $debug_enabled);
 929
 930                                 // enclosures
 931
 932                                 $enclosures = array();
 933
 934                                 $encs = $item->get_enclosures();
 935
 936                                 if (is_array($encs)) {
 937                                         foreach ($encs as $e) {
 938                                                 $e_item = array(
 939                                                         $e->link, $e->type, $e->length);
 940                                                 array_push($enclosures, $e_item);
 941                                         }
 942                                 }
 943
 944                                 if ($debug_enabled) {
 945                                         _debug("article enclosures:", $debug_enabled);
 946                                         print_r($enclosures);
 947                                 }
 948
 949                                 db_query("BEGIN");
 950
 951                                 foreach ($enclosures as $enc) {
 952                                         $enc_url = db_escape_string($enc[0]);
 953                                         $enc_type = db_escape_string($enc[1]);
 954                                         $enc_dur = db_escape_string($enc[2]);
 955
 956                                         $result = db_query("SELECT id FROM ttrss_enclosures
 957                                                 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
 958
 959                                         if (db_num_rows($result) == 0) {
 960                                                 db_query("INSERT INTO ttrss_enclosures
 961                                                         (content_url, content_type, title, duration, post_id) VALUES
 962                                                         ('$enc_url', '$enc_type', '', '$enc_dur', '$entry_ref_id')");
 963                                         }
 964                                 }
 965
 966                                 db_query("COMMIT");
 967
 968                                 // check for manual tags (we have to do it here since they're loaded from filters)
 969
 970                                 foreach ($article_filters as $f) {
 971                                         if ($f["type"] == "tag") {
 972
 973                                                 $manual_tags = trim_array(explode(",", $f["param"]));
 974
 975                                                 foreach ($manual_tags as $tag) {
 976                                                         if (tag_is_valid($tag)) {
 977                                                                 array_push($entry_tags, $tag);
 978                                                         }
 979                                                 }
 980                                         }
 981                                 }
 982
 983                                 // Skip boring tags
 984
 985                                 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
 986                                         'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
 987
 988                                 $filtered_tags = array();
 989                                 $tags_to_cache = array();
 990
 991                                 if ($entry_tags && is_array($entry_tags)) {
 992                                         foreach ($entry_tags as $tag) {
 993                                                 if (array_search($tag, $boring_tags) === false) {
 994                                                         array_push($filtered_tags, $tag);
 995                                                 }
 996                                         }
 997                                 }
 998
 999                                 $filtered_tags = array_unique($filtered_tags);
1000
1001                                 if ($debug_enabled) {
1002                                         _debug("filtered article tags:", $debug_enabled);
1003                                         print_r($filtered_tags);
1004                                 }
1005
1006                                 // Save article tags in the database
1007
1008                                 if (count($filtered_tags) > 0) {
1009
1010                                         db_query("BEGIN");
1011
1012                                         foreach ($filtered_tags as $tag) {
1013
1014                                                 $tag = sanitize_tag($tag);
1015                                                 $tag = db_escape_string($tag);
1016
1017                                                 if (!tag_is_valid($tag)) continue;
1018
1019                                                 $result = db_query("SELECT id FROM ttrss_tags
1020                                                         WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1021                                                         owner_uid = '$owner_uid' LIMIT 1");
1022
1023                                                         if ($result && db_num_rows($result) == 0) {
1024
1025                                                                 db_query("INSERT INTO ttrss_tags
1026                                                                         (owner_uid,tag_name,post_int_id)
1027                                                                         VALUES ('$owner_uid','$tag', '$entry_int_id')");
1028                                                         }
1029
1030                                                 array_push($tags_to_cache, $tag);
1031                                         }
1032
1033                                         /* update the cache */
1034
1035                                         $tags_to_cache = array_unique($tags_to_cache);
1036
1037                                         $tags_str = db_escape_string(join(",", $tags_to_cache));
1038
1039                                         db_query("UPDATE ttrss_user_entries
1040                                                 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1041                                                 AND owner_uid = $owner_uid");
1042
1043                                         db_query("COMMIT");
1044                                 }
1045
1046                                 if (get_pref("AUTO_ASSIGN_LABELS", $owner_uid, false)) {
1047                                         _debug("auto-assigning labels...", $debug_enabled);
1048
1049                                         foreach ($labels as $label) {
1050                                                 $caption = preg_quote($label["caption"]);
1051
1052                                                 if ($caption && preg_match("/\b$caption\b/i", "$tags_str " . strip_tags($entry_content) . " $entry_title")) {
1053                                                         if (!labels_contains_caption($article_labels, $caption)) {
1054                                                                 label_add_article($entry_ref_id, $caption, $owner_uid);
1055                                                         }
1056                                                 }
1057                                         }
1058                                 }
1059
1060                                 _debug("article processed", $debug_enabled);
1061                         }
1062
1063                         _debug("purging feed...", $debug_enabled);
1064
1065                         purge_feed($feed, 0, $debug_enabled);
1066
1067                         db_query("UPDATE ttrss_feeds
1068                                 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
1069
1070 //                      db_query("COMMIT");
1071
1072                 } else {
1073
1074                         $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
1075
1076                         _debug("error fetching feed: $error_msg", $debug_enabled);
1077
1078                         db_query(
1079                                 "UPDATE ttrss_feeds SET last_error = '$error_msg',
1080                                         last_updated = NOW() WHERE id = '$feed'");
1081                 }
1082
1083                 unset($rss);
1084
1085                 _debug("done", $debug_enabled);
1086         }
1087
1088         function cache_images($html, $site_url, $debug) {
1089                 $cache_dir = CACHE_DIR . "/images";
1090
1091                 libxml_use_internal_errors(true);
1092
1093                 $charset_hack = '<head>
1094                         <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1095                 </head>';
1096
1097                 $doc = new DOMDocument();
1098                 $doc->loadHTML($charset_hack . $html);
1099                 $xpath = new DOMXPath($doc);
1100
1101                 $entries = $xpath->query('(//img[@src])');
1102
1103                 foreach ($entries as $entry) {
1104                         if ($entry->hasAttribute('src')) {
1105                                 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1106
1107                                 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1108
1109                                 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1110
1111                                 if (!file_exists($local_filename)) {
1112                                         $file_content = fetch_file_contents($src);
1113
1114                                         if ($file_content && strlen($file_content) > 1024) {
1115                                                 file_put_contents($local_filename, $file_content);
1116                                         }
1117                                 }
1118
1119                                 if (file_exists($local_filename)) {
1120                                         $entry->setAttribute('src', SELF_URL_PATH . '/image.php?url=' .
1121                                                 base64_encode($src));
1122                                 }
1123                         }
1124                 }
1125
1126                 $node = $doc->getElementsByTagName('body')->item(0);
1127
1128                 return $doc->saveXML($node);
1129         }
1130
1131         function expire_error_log($debug) {
1132                 if ($debug) _debug("Removing old error log entries...");
1133
1134                 if (DB_TYPE == "pgsql") {
1135                         db_query("DELETE FROM ttrss_error_log
1136                                 WHERE created_at < NOW() - INTERVAL '7 days'");
1137                 } else {
1138                         db_query("DELETE FROM ttrss_error_log
1139                                 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1140                 }
1141
1142         }
1143
1144         function expire_lock_files($debug) {
1145                 //if ($debug) _debug("Removing old lock files...");
1146
1147                 $num_deleted = 0;
1148
1149                 if (is_writable(LOCK_DIRECTORY)) {
1150                         $files = glob(LOCK_DIRECTORY . "/*.lock");
1151
1152                         if ($files) {
1153                                 foreach ($files as $file) {
1154                                         if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
1155                                                 unlink($file);
1156                                                 ++$num_deleted;
1157                                         }
1158                                 }
1159                         }
1160                 }
1161
1162                 if ($debug) _debug("Removed $num_deleted old lock files.");
1163         }
1164
1165         function expire_cached_files($debug) {
1166                 foreach (array("simplepie", "images", "export", "upload") as $dir) {
1167                         $cache_dir = CACHE_DIR . "/$dir";
1168
1169 //                      if ($debug) _debug("Expiring $cache_dir");
1170
1171                         $num_deleted = 0;
1172
1173                         if (is_writable($cache_dir)) {
1174                                 $files = glob("$cache_dir/*");
1175
1176                                 if ($files) {
1177                                         foreach ($files as $file) {
1178                                                 if (time() - filemtime($file) > 86400*7) {
1179                                                         unlink($file);
1180
1181                                                         ++$num_deleted;
1182                                                 }
1183                                         }
1184                                 }
1185                         }
1186
1187                         if ($debug) _debug("$cache_dir: removed $num_deleted files.");
1188                 }
1189         }
1190
1191         /**
1192         * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1193         * Returns the url query as associative array
1194         *
1195         * @param    string    query
1196         * @return    array    params
1197         */
1198         function convertUrlQuery($query) {
1199                 $queryParts = explode('&', $query);
1200
1201                 $params = array();
1202
1203                 foreach ($queryParts as $param) {
1204                         $item = explode('=', $param);
1205                         $params[$item[0]] = $item[1];
1206                 }
1207
1208                 return $params;
1209         }
1210
1211         function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
1212                 $matches = array();
1213
1214                 foreach ($filters as $filter) {
1215                         $match_any_rule = $filter["match_any_rule"];
1216                         $inverse = $filter["inverse"];
1217                         $filter_match = false;
1218
1219                         foreach ($filter["rules"] as $rule) {
1220                                 $match = false;
1221                                 $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
1222                                 $rule_inverse = $rule["inverse"];
1223
1224                                 if (!$reg_exp)
1225                                         continue;
1226
1227                                 switch ($rule["type"]) {
1228                                 case "title":
1229                                         $match = @preg_match("/$reg_exp/i", $title);
1230                                         break;
1231                                 case "content":
1232                                         // we don't need to deal with multiline regexps
1233                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1234
1235                                         $match = @preg_match("/$reg_exp/i", $content);
1236                                         break;
1237                                 case "both":
1238                                         // we don't need to deal with multiline regexps
1239                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1240
1241                                         $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
1242                                         break;
1243                                 case "link":
1244                                         $match = @preg_match("/$reg_exp/i", $link);
1245                                         break;
1246                                 case "author":
1247                                         $match = @preg_match("/$reg_exp/i", $author);
1248                                         break;
1249                                 case "tag":
1250                                         foreach ($tags as $tag) {
1251                                                 if (@preg_match("/$reg_exp/i", $tag)) {
1252                                                         $match = true;
1253                                                         break;
1254                                                 }
1255                                         }
1256                                         break;
1257                                 }
1258
1259                                 if ($rule_inverse) $match = !$match;
1260
1261                                 if ($match_any_rule) {
1262                                         if ($match) {
1263                                                 $filter_match = true;
1264                                                 break;
1265                                         }
1266                                 } else {
1267                                         $filter_match = $match;
1268                                         if (!$match) {
1269                                                 break;
1270                                         }
1271                                 }
1272                         }
1273
1274                         if ($inverse) $filter_match = !$filter_match;
1275
1276                         if ($filter_match) {
1277                                 foreach ($filter["actions"] AS $action) {
1278                                         array_push($matches, $action);
1279
1280                                         // if Stop action encountered, perform no further processing
1281                                         if ($action["type"] == "stop") return $matches;
1282                                 }
1283                         }
1284                 }
1285
1286                 return $matches;
1287         }
1288
1289         function find_article_filter($filters, $filter_name) {
1290                 foreach ($filters as $f) {
1291                         if ($f["type"] == $filter_name) {
1292                                 return $f;
1293                         };
1294                 }
1295                 return false;
1296         }
1297
1298         function find_article_filters($filters, $filter_name) {
1299                 $results = array();
1300
1301                 foreach ($filters as $f) {
1302                         if ($f["type"] == $filter_name) {
1303                                 array_push($results, $f);
1304                         };
1305                 }
1306                 return $results;
1307         }
1308
1309         function calculate_article_score($filters) {
1310                 $score = 0;
1311
1312                 foreach ($filters as $f) {
1313                         if ($f["type"] == "score") {
1314                                 $score += $f["param"];
1315                         };
1316                 }
1317                 return $score;
1318         }
1319
1320         function labels_contains_caption($labels, $caption) {
1321                 foreach ($labels as $label) {
1322                         if ($label[1] == $caption) {
1323                                 return true;
1324                         }
1325                 }
1326
1327                 return false;
1328         }
1329
1330         function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
1331                 foreach ($filters as $f) {
1332                         if ($f["type"] == "label") {
1333                                 if (!labels_contains_caption($article_labels, $f["param"])) {
1334                                         label_add_article($id, $f["param"], $owner_uid);
1335                                 }
1336                         }
1337                 }
1338         }
1339
1340         function make_guid_from_title($title) {
1341                 return preg_replace("/[ \"\',.:;]/", "-",
1342                         mb_strtolower(strip_tags($title), 'utf-8'));
1343         }
1344
1345         /* function verify_feed_xml($feed_data) {
1346                 libxml_use_internal_errors(true);
1347                 $doc = new DOMDocument();
1348                 $doc->loadXML($feed_data);
1349                 $error = libxml_get_last_error();
1350                 libxml_clear_errors();
1351                 return $error;
1352         } */
1353
1354         function housekeeping_common($debug) {
1355                 expire_cached_files($debug);
1356                 expire_lock_files($debug);
1357                 expire_error_log($debug);
1358
1359                 $count = update_feedbrowser_cache();
1360                 _debug("Feedbrowser updated, $count feeds processed.");
1361
1362                 purge_orphans( true);
1363                 $rc = cleanup_tags( 14, 50000);
1364
1365                 _debug("Cleaned $rc cached tags.");
1366         }
1367 ?>