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 exceeded).
  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                                         AND ttrss_user_prefs.value != '-1')
  93                                 OR (last_updated = '1970-01-01 00:00:00'
  94                                         AND ttrss_user_prefs.value != '-1'))";
  95                 } else {
  96                         $update_limit_qpart = "AND ((
  97                                         ttrss_feeds.update_interval = 0
  98                                         AND ttrss_user_prefs.value != '-1'
  99                                         AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
 100                                 ) OR (
 101                                         ttrss_feeds.update_interval > 0
 102                                         AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
 103                                 ) OR (ttrss_feeds.last_updated IS NULL
 104                                         AND ttrss_user_prefs.value != '-1')
 105                                 OR (last_updated = '1970-01-01 00:00:00'
 106                                         AND ttrss_user_prefs.value != '-1'))";
 107                 }
 108
 109                 // Test if feed is currently being updated by another process.
 110                 if (DB_TYPE == "pgsql") {
 111                         $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
 112                 } else {
 113                         $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 10 MINUTE))";
 114                 }
 115
 116                 // Test if there is a limit to number of updated feeds
 117                 $query_limit = "";
 118                 if($limit) $query_limit = sprintf("LIMIT %d", $limit);
 119
 120                 $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
 121                         FROM
 122                                 ttrss_feeds, ttrss_users, ttrss_user_prefs
 123                         WHERE
 124                                 ttrss_feeds.owner_uid = ttrss_users.id
 125                                 AND ttrss_user_prefs.profile IS NULL
 126                                 AND ttrss_users.id = ttrss_user_prefs.owner_uid
 127                                 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
 128                                 $login_thresh_qpart $update_limit_qpart
 129                                 $updstart_thresh_qpart
 130                                 ORDER BY last_updated $query_limit";
 131
 132                 // We search for feed needing update.
 133                 $result = db_query($query);
 134
 135                 if($debug) _debug(sprintf("Scheduled %d feeds to update...", db_num_rows($result)));
 136
 137                 // Here is a little cache magic in order to minimize risk of double feed updates.
 138                 $feeds_to_update = array();
 139                 while ($line = db_fetch_assoc($result)) {
 140                         array_push($feeds_to_update, db_escape_string($line['feed_url']));
 141                 }
 142
 143                 // We update the feed last update started date before anything else.
 144                 // There is no lag due to feed contents downloads
 145                 // It prevent an other process to update the same feed.
 146
 147                 if(count($feeds_to_update) > 0) {
 148                         $feeds_quoted = array();
 149
 150                         foreach ($feeds_to_update as $feed) {
 151                                 array_push($feeds_quoted, "'" . db_escape_string($feed) . "'");
 152                         }
 153
 154                         db_query(sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
 155                                 WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
 156                 }
 157
 158                 $nf = 0;
 159
 160                 // For each feed, we call the feed update function.
 161                 foreach ($feeds_to_update as $feed) {
 162                         if($debug) _debug("Base feed: $feed");
 163
 164                         //update_rss_feed($line["id"], true);
 165
 166                         // since we have the data cached, we can deal with other feeds with the same url
 167
 168                         $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
 169                         FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
 170                                 ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
 171                                 ttrss_users.id = ttrss_user_prefs.owner_uid AND
 172                                 ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
 173                                 ttrss_user_prefs.profile IS NULL AND
 174                                 feed_url = '".db_escape_string($feed)."' AND
 175                                 (ttrss_feeds.update_interval > 0 OR
 176                                         ttrss_user_prefs.value != '-1')
 177                                 $login_thresh_qpart
 178                         ORDER BY ttrss_feeds.id $query_limit");
 179
 180                         if (db_num_rows($tmp_result) > 0) {
 181                                 $rss = false;
 182
 183                                 while ($tline = db_fetch_assoc($tmp_result)) {
 184                                         if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
 185
 186                                         $rss = update_rss_feed($tline["id"], true, false, $rss);
 187                                         _debug_suppress(false);
 188                                         ++$nf;
 189                                 }
 190                         }
 191                 }
 192
 193                 require_once "digest.php";
 194
 195                 // Send feed digests by email if needed.
 196                 send_headlines_digests($debug);
 197
 198                 return $nf;
 199
 200         } // function update_daemon_common
 201
 202         // ignore_daemon is not used
 203         function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
 204
 205                 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
 206
 207                 _debug_suppress(!$debug_enabled);
 208                 _debug("start", $debug_enabled);
 209
 210                 $result = db_query("SELECT id,update_interval,auth_login,
 211                         feed_url,auth_pass,cache_images,
 212                         mark_unread_on_update, owner_uid,
 213                         pubsub_state, auth_pass_encrypted,
 214                         (SELECT max(date_entered) FROM
 215                                 ttrss_entries, ttrss_user_entries where ref_id = id AND feed_id = '$feed') AS last_article_timestamp
 216                         FROM ttrss_feeds WHERE id = '$feed'");
 217
 218                 if (db_num_rows($result) == 0) {
 219                         _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
 220                         return false;
 221                 }
 222
 223                 $last_article_timestamp = @strtotime(db_fetch_result($result, 0, "last_article_timestamp"));
 224
 225                 if (defined('_DISABLE_HTTP_304'))
 226                         $last_article_timestamp = 0;
 227
 228                 $owner_uid = db_fetch_result($result, 0, "owner_uid");
 229                 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
 230                         0, "mark_unread_on_update"));
 231                 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
 232                 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
 233                         0, "auth_pass_encrypted"));
 234
 235                 db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
 236                         WHERE id = '$feed'");
 237
 238                 $auth_login = db_fetch_result($result, 0, "auth_login");
 239                 $auth_pass = db_fetch_result($result, 0, "auth_pass");
 240
 241                 if ($auth_pass_encrypted) {
 242                         require_once "crypt.php";
 243                         $auth_pass = decrypt_string($auth_pass);
 244                 }
 245
 246                 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
 247                 $fetch_url = db_fetch_result($result, 0, "feed_url");
 248
 249                 $feed = db_escape_string($feed);
 250
 251                 $date_feed_processed = date('Y-m-d H:i');
 252
 253                 $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
 254
 255                 $pluginhost = new PluginHost();
 256                 $pluginhost->set_debug($debug_enabled);
 257                 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
 258
 259                 $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
 260                 $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
 261                 $pluginhost->load_data();
 262
 263                 if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
 264                         _debug("using previously initialized parser object");
 265                 } else {
 266                         $rss_hash = false;
 267
 268                         $force_refetch = isset($_REQUEST["force_refetch"]);
 269
 270                         foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
 271                                 $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed);
 272                         }
 273
 274                         // try cache
 275                         if (!$feed_data &&
 276                                 file_exists($cache_filename) &&
 277                                 is_readable($cache_filename) &&
 278                                 !$auth_login && !$auth_pass &&
 279                                 filemtime($cache_filename) > time() - 30) {
 280
 281                                 _debug("using local cache.", $debug_enabled);
 282
 283                                 @$feed_data = file_get_contents($cache_filename);
 284
 285                                 if ($feed_data) {
 286                                         $rss_hash = sha1($feed_data);
 287                                 }
 288
 289                         } else {
 290                                 _debug("local cache will not be used for this feed", $debug_enabled);
 291                         }
 292
 293                         // fetch feed from source
 294                         if (!$feed_data) {
 295                                 _debug("fetching [$fetch_url]...", $debug_enabled);
 296                                 _debug("If-Modified-Since: ".gmdate('D, d M Y H:i:s \G\M\T', $last_article_timestamp), $debug_enabled);
 297
 298                                 $feed_data = fetch_file_contents($fetch_url, false,
 299                                         $auth_login, $auth_pass, false,
 300                                         $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
 301                                         $force_refetch ? 0 : $last_article_timestamp);
 302
 303                                 global $fetch_curl_used;
 304
 305                                 if (!$fetch_curl_used) {
 306                                         $tmp = @gzdecode($feed_data);
 307
 308                                         if ($tmp) $feed_data = $tmp;
 309                                 }
 310
 311                                 $feed_data = trim($feed_data);
 312
 313                                 _debug("fetch done.", $debug_enabled);
 314
 315                                 /* if ($feed_data) {
 316                                         $error = verify_feed_xml($feed_data);
 317
 318                                         if ($error) {
 319                                                 _debug("error verifying XML, code: " . $error->code, $debug_enabled);
 320
 321                                                 if ($error->code == 26) {
 322                                                         _debug("got error 26, trying to decode entities...", $debug_enabled);
 323
 324                                                         $feed_data = html_entity_decode($feed_data, ENT_COMPAT, 'UTF-8');
 325
 326                                                         $error = verify_feed_xml($feed_data);
 327
 328                                                         if ($error) $feed_data = '';
 329                                                 }
 330                                         }
 331                                 } */
 332
 333                                 // cache vanilla feed data for re-use
 334                                 if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
 335                                         $new_rss_hash = sha1($feed_data);
 336
 337                                         if ($new_rss_hash != $rss_hash) {
 338                                                 _debug("saving $cache_filename", $debug_enabled);
 339                                                 @file_put_contents($cache_filename, $feed_data);
 340                                         }
 341                                 }
 342                         }
 343
 344                         if (!$feed_data) {
 345                                 global $fetch_last_error;
 346                                 global $fetch_last_error_code;
 347
 348                                 _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
 349
 350                                 $error_escaped = '';
 351
 352                                 // If-Modified-Since
 353                                 if ($fetch_last_error_code != 304) {
 354                                         $error_escaped = db_escape_string($fetch_last_error);
 355                                 } else {
 356                                         _debug("source claims data not modified, nothing to do.", $debug_enabled);
 357                                 }
 358
 359                                 db_query(
 360                                         "UPDATE ttrss_feeds SET last_error = '$error_escaped',
 361                                                 last_updated = NOW() WHERE id = '$feed'");
 362
 363                                 return;
 364                         }
 365                 }
 366
 367                 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
 368                         $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
 369                 }
 370
 371                 // set last update to now so if anything *simplepie* crashes later we won't be
 372                 // continuously failing on the same feed
 373                 //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
 374
 375                 if (!$rss) {
 376                         $rss = new FeedParser($feed_data);
 377                         $rss->init();
 378                 }
 379
 380                 if (DETECT_ARTICLE_LANGUAGE) {
 381                         require_once "lib/languagedetect/LanguageDetect.php";
 382
 383                         $lang = new Text_LanguageDetect();
 384                         $lang->setNameMode(2);
 385                 }
 386
 387 //              print_r($rss);
 388
 389                 $feed = db_escape_string($feed);
 390
 391                 if (!$rss->error()) {
 392
 393                         // We use local pluginhost here because we need to load different per-user feed plugins
 394                         $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
 395
 396                         _debug("processing feed data...", $debug_enabled);
 397
 398 //                      db_query("BEGIN");
 399
 400                         if (DB_TYPE == "pgsql") {
 401                                 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
 402                         } else {
 403                                 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
 404                         }
 405
 406                         $result = db_query("SELECT title,site_url,owner_uid,favicon_avg_color,
 407                                 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
 408                                                 favicon_needs_check
 409                                 FROM ttrss_feeds WHERE id = '$feed'");
 410
 411                         $registered_title = db_fetch_result($result, 0, "title");
 412                         $orig_site_url = db_fetch_result($result, 0, "site_url");
 413                         $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
 414                                 "favicon_needs_check"));
 415                         $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
 416
 417                         $owner_uid = db_fetch_result($result, 0, "owner_uid");
 418
 419                         $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
 420
 421                         _debug("site_url: $site_url", $debug_enabled);
 422                         _debug("feed_title: " . $rss->get_title(), $debug_enabled);
 423
 424                         if ($favicon_needs_check || $force_refetch) {
 425
 426                                 /* terrible hack: if we crash on floicon shit here, we won't check
 427                                  * the icon avgcolor again (unless the icon got updated) */
 428
 429                                 $favicon_file = ICONS_DIR . "/$feed.ico";
 430                                 $favicon_modified = @filemtime($favicon_file);
 431
 432                                 _debug("checking favicon...", $debug_enabled);
 433
 434                                 check_feed_favicon($site_url, $feed);
 435                                 $favicon_modified_new = @filemtime($favicon_file);
 436
 437                                 if ($favicon_modified_new > $favicon_modified)
 438                                         $favicon_avg_color = '';
 439
 440                                 if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
 441                                                 require_once "colors.php";
 442
 443                                                 db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
 444                                                         id = '$feed'");
 445
 446                                                 $favicon_color = db_escape_string(
 447                                                         calculate_avg_color($favicon_file));
 448
 449                                                 $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
 450                                 } else if ($favicon_avg_color == 'fail') {
 451                                         _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
 452                                 }
 453
 454                                 db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
 455                                         $favicon_colorstring
 456                                         WHERE id = '$feed'");
 457                         }
 458
 459                         if (!$registered_title || $registered_title == "[Unknown]") {
 460
 461                                 $feed_title = db_escape_string($rss->get_title());
 462
 463                                 if ($feed_title) {
 464                                         _debug("registering title: $feed_title", $debug_enabled);
 465
 466                                         db_query("UPDATE ttrss_feeds SET
 467                                                 title = '$feed_title' WHERE id = '$feed'");
 468                                 }
 469                         }
 470
 471                         if ($site_url && $orig_site_url != $site_url) {
 472                                 db_query("UPDATE ttrss_feeds SET
 473                                         site_url = '$site_url' WHERE id = '$feed'");
 474                         }
 475
 476                         _debug("loading filters & labels...", $debug_enabled);
 477
 478                         $filters = load_filters($feed, $owner_uid);
 479                         $labels = get_all_labels($owner_uid);
 480
 481                         _debug("" . count($filters) . " filters loaded.", $debug_enabled);
 482
 483                         $items = $rss->get_items();
 484
 485                         if (!is_array($items)) {
 486                                 _debug("no articles found.", $debug_enabled);
 487
 488                                 db_query("UPDATE ttrss_feeds
 489                                         SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
 490
 491                                 return; // no articles
 492                         }
 493
 494                         if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
 495
 496                                 _debug("checking for PUSH hub...", $debug_enabled);
 497
 498                                 $feed_hub_url = false;
 499
 500                                 $links = $rss->get_links('hub');
 501
 502                                 if ($links && is_array($links)) {
 503                                         foreach ($links as $l) {
 504                                                 $feed_hub_url = $l;
 505                                                 break;
 506                                         }
 507                                 }
 508
 509                                 _debug("feed hub url: $feed_hub_url", $debug_enabled);
 510
 511                                 $feed_self_url = $fetch_url;
 512
 513                                 $links = $rss->get_links('self');
 514
 515                                 if ($links && is_array($links)) {
 516                                         foreach ($links as $l) {
 517                                                 $feed_self_url = $l;
 518                                                 break;
 519                                         }
 520                                 }
 521
 522                                 _debug("feed self url = $feed_self_url");
 523
 524                                 if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
 525                                         !ini_get("open_basedir")) {
 526
 527                                         require_once 'lib/pubsubhubbub/subscriber.php';
 528
 529                                         $callback_url = get_self_url_prefix() .
 530                                                 "/public.php?op=pubsub&id=$feed";
 531
 532                                         $s = new Subscriber($feed_hub_url, $callback_url);
 533
 534                                         $rc = $s->subscribe($feed_self_url);
 535
 536                                         _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
 537
 538                                         db_query("UPDATE ttrss_feeds SET pubsub_state = 1
 539                                                 WHERE id = '$feed'");
 540                                 }
 541                         }
 542
 543                         _debug("processing articles...", $debug_enabled);
 544
 545                         foreach ($items as $item) {
 546                                 if ($_REQUEST['xdebug'] == 3) {
 547                                         print_r($item);
 548                                 }
 549
 550                                 $entry_guid = $item->get_id();
 551                                 if (!$entry_guid) $entry_guid = $item->get_link();
 552                                 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
 553
 554                                 _debug("f_guid $entry_guid", $debug_enabled);
 555
 556                                 if (!$entry_guid) continue;
 557
 558                                 $entry_guid = "$owner_uid,$entry_guid";
 559
 560                                 $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
 561
 562                                 _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
 563
 564                                 $entry_timestamp = "";
 565
 566                                 $entry_timestamp = $item->get_date();
 567
 568                                 _debug("orig date: " . $item->get_date(), $debug_enabled);
 569
 570                                 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
 571                                         $entry_timestamp = time();
 572                                 }
 573
 574                                 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
 575
 576                                 _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
 577
 578 //                              $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
 579 //                              $entry_title = decode_numeric_entities($entry_title);
 580                                 $entry_title = $item->get_title();
 581
 582                                 $entry_link = rewrite_relative_url($site_url, $item->get_link());
 583
 584                                 _debug("title $entry_title", $debug_enabled);
 585                                 _debug("link $entry_link", $debug_enabled);
 586
 587                                 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
 588
 589                                 $entry_content = $item->get_content();
 590                                 if (!$entry_content) $entry_content = $item->get_description();
 591
 592                                 if ($_REQUEST["xdebug"] == 2) {
 593                                         print "content: ";
 594                                         print $entry_content;
 595                                         print "\n";
 596                                 }
 597
 598                                 $entry_language = "";
 599
 600                                 if (DETECT_ARTICLE_LANGUAGE) {
 601                                         $entry_language = $lang->detect($entry_title . " " . $entry_content, 1);
 602
 603                                         if (count($entry_language) > 0) {
 604                                                 $entry_language = array_keys($entry_language);
 605
 606                                                 // the fuck?
 607                                                 if (is_array($entry_language))
 608                                                         $entry_language = "";
 609                                                 else
 610                                                         $entry_language = db_escape_string(substr($entry_language[0], 0, 2));
 611
 612                                                 _debug("detected language: $entry_language", $debug_enabled);
 613                                         } else {
 614                                                 $entry_language = "";
 615                                         }
 616                                 }
 617
 618                                 $entry_comments = $item->get_comments_url();
 619                                 $entry_author = $item->get_author();
 620
 621                                 $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
 622
 623                                 $entry_comments = db_escape_string(mb_substr(trim($entry_comments), 0, 245));
 624                                 $entry_author = db_escape_string(mb_substr(trim($entry_author), 0, 245));
 625
 626                                 $num_comments = (int) $item->get_comments_count();
 627
 628                                 _debug("author $entry_author", $debug_enabled);
 629                                 _debug("num_comments: $num_comments", $debug_enabled);
 630                                 _debug("looking for tags...", $debug_enabled);
 631
 632                                 // parse <category> entries into tags
 633
 634                                 $additional_tags = array();
 635
 636                                 $additional_tags_src = $item->get_categories();
 637
 638                                 if (is_array($additional_tags_src)) {
 639                                         foreach ($additional_tags_src as $tobj) {
 640                                                 array_push($additional_tags, $tobj);
 641                                         }
 642                                 }
 643
 644                                 $entry_tags = array_unique($additional_tags);
 645
 646                                 for ($i = 0; $i < count($entry_tags); $i++)
 647                                         $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
 648
 649                                 _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
 650
 651                                 _debug("done collecting data.", $debug_enabled);
 652
 653                                 // TODO: less memory-hungry implementation
 654
 655                                 _debug("applying plugin filters..", $debug_enabled);
 656
 657                                 // FIXME not sure if owner_uid is a good idea here, we may have a base entry without user entry (?)
 658                                 $result = db_query("SELECT plugin_data,title,content,link,tag_cache,author FROM ttrss_entries, ttrss_user_entries
 659                                         WHERE ref_id = id AND (guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed') AND owner_uid = $owner_uid");
 660
 661                                 if (db_num_rows($result) != 0) {
 662                                         $entry_plugin_data = db_fetch_result($result, 0, "plugin_data");
 663                                         $stored_article = array("title" => db_fetch_result($result, 0, "title"),
 664                                                 "content" => db_fetch_result($result, 0, "content"),
 665                                                 "link" => db_fetch_result($result, 0, "link"),
 666                                                 "tags" => explode(",", db_fetch_result($result, 0, "tag_cache")),
 667                                                 "author" => db_fetch_result($result, 0, "author"));
 668                                 } else {
 669                                         $entry_plugin_data = "";
 670                                         $stored_article = array();
 671                                 }
 672
 673                                 $article = array("owner_uid" => $owner_uid, // read only
 674                                         "guid" => $entry_guid, // read only
 675                                         "title" => $entry_title,
 676                                         "content" => $entry_content,
 677                                         "link" => $entry_link,
 678                                         "tags" => $entry_tags,
 679                                         "plugin_data" => $entry_plugin_data,
 680                                         "author" => $entry_author,
 681                                         "stored" => $stored_article,
 682                                         "feed" => array("id" => $feed,
 683                                                 "fetch_url" => $fetch_url,
 684                                                 "site_url" => $site_url)
 685                                         );
 686
 687                                 foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
 688                                         $article = $plugin->hook_article_filter($article);
 689                                 }
 690
 691                                 $entry_tags = $article["tags"];
 692                                 $entry_guid = db_escape_string($entry_guid);
 693                                 $entry_title = db_escape_string($article["title"]);
 694                                 $entry_author = db_escape_string($article["author"]);
 695                                 $entry_link = db_escape_string($article["link"]);
 696                                 $entry_plugin_data = db_escape_string($article["plugin_data"]);
 697                                 $entry_content = $article["content"]; // escaped below
 698
 699
 700                                 _debug("plugin data: $entry_plugin_data", $debug_enabled);
 701
 702                                 if ($cache_images && is_writable(CACHE_DIR . '/images'))
 703                                         cache_images($entry_content, $site_url, $debug_enabled);
 704
 705                                 $entry_content = db_escape_string($entry_content, false);
 706
 707                                 $content_hash = "SHA1:" . sha1($entry_content);
 708
 709                                 db_query("BEGIN");
 710
 711                                 $result = db_query("SELECT id FROM      ttrss_entries
 712                                         WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
 713
 714                                 if (db_num_rows($result) == 0) {
 715
 716                                         _debug("base guid [$entry_guid] not found", $debug_enabled);
 717
 718                                         // base post entry does not exist, create it
 719
 720                                         $result = db_query(
 721                                                 "INSERT INTO ttrss_entries
 722                                                         (title,
 723                                                         guid,
 724                                                         link,
 725                                                         updated,
 726                                                         content,
 727                                                         content_hash,
 728                                                         no_orig_date,
 729                                                         date_updated,
 730                                                         date_entered,
 731                                                         comments,
 732                                                         num_comments,
 733                                                         plugin_data,
 734                                                         lang,
 735                                                         author)
 736                                                 VALUES
 737                                                         ('$entry_title',
 738                                                         '$entry_guid_hashed',
 739                                                         '$entry_link',
 740                                                         '$entry_timestamp_fmt',
 741                                                         '$entry_content',
 742                                                         '$content_hash',
 743                                                         false,
 744                                                         NOW(),
 745                                                         '$date_feed_processed',
 746                                                         '$entry_comments',
 747                                                         '$num_comments',
 748                                                         '$entry_plugin_data',
 749                                                         '$entry_language',
 750                                                         '$entry_author')");
 751
 752                                         $article_labels = array();
 753
 754                                 } else {
 755                                         // we keep encountering the entry in feeds, so we need to
 756                                         // update date_updated column so that we don't get horrible
 757                                         // dupes when the entry gets purged and reinserted again e.g.
 758                                         // in the case of SLOW SLOW OMG SLOW updating feeds
 759
 760                                         $base_entry_id = db_fetch_result($result, 0, "id");
 761
 762                                         db_query("UPDATE ttrss_entries SET date_updated = NOW()
 763                                                 WHERE id = '$base_entry_id'");
 764
 765                                         $article_labels = get_article_labels($base_entry_id, $owner_uid);
 766                                 }
 767
 768                                 // now it should exist, if not - bad luck then
 769
 770                                 $result = db_query("SELECT
 771                                                 id,content_hash,title,plugin_data,guid,
 772                                                 num_comments
 773                                         FROM
 774                                                 ttrss_entries
 775                                         WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
 776
 777                                 $entry_ref_id = 0;
 778                                 $entry_int_id = 0;
 779
 780                                 if (db_num_rows($result) == 1) {
 781
 782                                         _debug("base guid found, checking for user record", $debug_enabled);
 783
 784                                         // this will be used below in update handler
 785                                         $orig_content_hash = db_fetch_result($result, 0, "content_hash");
 786                                         $orig_title = db_fetch_result($result, 0, "title");
 787                                         $orig_num_comments = db_fetch_result($result, 0, "num_comments");
 788                                         $orig_plugin_data = db_fetch_result($result, 0, "plugin_data");
 789
 790                                         $ref_id = db_fetch_result($result, 0, "id");
 791                                         $entry_ref_id = $ref_id;
 792
 793                                         /* $stored_guid = db_fetch_result($result, 0, "guid");
 794                                         if ($stored_guid != $entry_guid_hashed) {
 795                                                 if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
 796
 797                                                 db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
 798                                                         id = '$ref_id'");
 799                                         } */
 800
 801                                         // check for user post link to main table
 802
 803                                         // do we allow duplicate posts with same GUID in different feeds?
 804                                         if (get_pref("ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
 805                                                 $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
 806                                         } else {
 807                                                 $dupcheck_qpart = "";
 808                                         }
 809
 810                                         /* Collect article tags here so we could filter by them: */
 811
 812                                         $article_filters = get_article_filters($filters, $entry_title,
 813                                                 $entry_content, $entry_link, $entry_timestamp, $entry_author,
 814                                                 $entry_tags);
 815
 816                                         if ($debug_enabled) {
 817                                                 _debug("article filters: ", $debug_enabled);
 818                                                 if (count($article_filters) != 0) {
 819                                                         print_r($article_filters);
 820                                                 }
 821                                         }
 822
 823                                         if (find_article_filter($article_filters, "filter")) {
 824                                                 db_query("COMMIT"); // close transaction in progress
 825                                                 continue;
 826                                         }
 827
 828                                         $score = calculate_article_score($article_filters);
 829
 830                                         _debug("initial score: $score", $debug_enabled);
 831
 832                                         $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
 833                                                         ref_id = '$ref_id' AND owner_uid = '$owner_uid'
 834                                                         $dupcheck_qpart";
 835
 836 //                                      if ($_REQUEST["xdebug"]) print "$query\n";
 837
 838                                         $result = db_query($query);
 839
 840                                         // okay it doesn't exist - create user entry
 841                                         if (db_num_rows($result) == 0) {
 842
 843                                                 _debug("user record not found, creating...", $debug_enabled);
 844
 845                                                 if ($score >= -500 && !find_article_filter($article_filters, 'catchup')) {
 846                                                         $unread = 'true';
 847                                                         $last_read_qpart = 'NULL';
 848                                                 } else {
 849                                                         $unread = 'false';
 850                                                         $last_read_qpart = 'NOW()';
 851                                                 }
 852
 853                                                 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
 854                                                         $marked = 'true';
 855                                                 } else {
 856                                                         $marked = 'false';
 857                                                 }
 858
 859                                                 if (find_article_filter($article_filters, 'publish')) {
 860                                                         $published = 'true';
 861                                                 } else {
 862                                                         $published = 'false';
 863                                                 }
 864
 865                                                 // N-grams
 866
 867                                                 if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
 868
 869                                                         $result = db_query("SELECT COUNT(*) AS similar FROM
 870                                                                         ttrss_entries,ttrss_user_entries
 871                                                                 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
 872                                                                         AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
 873                                                                         AND owner_uid = $owner_uid");
 874
 875                                                         $ngram_similar = db_fetch_result($result, 0, "similar");
 876
 877                                                         _debug("N-gram similar results: $ngram_similar", $debug_enabled);
 878
 879                                                         if ($ngram_similar > 0) {
 880                                                                 $unread = 'false';
 881                                                         }
 882                                                 }
 883
 884                                                 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
 885                                                 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
 886
 887                                                 $result = db_query(
 888                                                         "INSERT INTO ttrss_user_entries
 889                                                                 (ref_id, owner_uid, feed_id, unread, last_read, marked,
 890                                                                 published, score, tag_cache, label_cache, uuid,
 891                                                                 last_marked, last_published)
 892                                                         VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
 893                                                                 $last_read_qpart, $marked, $published, '$score', '', '',
 894                                                                 '', $last_marked, $last_published)");
 895
 896                                                 if (PUBSUBHUBBUB_HUB && $published == 'true') {
 897                                                         $rss_link = get_self_url_prefix() .
 898                                                                 "/public.php?op=rss&id=-2&key=" .
 899                                                                 get_feed_access_key(-2, false, $owner_uid);
 900
 901                                                         $p = new Publisher(PUBSUBHUBBUB_HUB);
 902
 903                                                         /* $pubsub_result = */ $p->publish_update($rss_link);
 904                                                 }
 905
 906                                                 $result = db_query(
 907                                                         "SELECT int_id FROM ttrss_user_entries WHERE
 908                                                                 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
 909                                                                 feed_id = '$feed' LIMIT 1");
 910
 911                                                 if (db_num_rows($result) == 1) {
 912                                                         $entry_int_id = db_fetch_result($result, 0, "int_id");
 913                                                 }
 914                                         } else {
 915                                                 _debug("user record FOUND", $debug_enabled);
 916
 917                                                 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
 918                                                 $entry_int_id = db_fetch_result($result, 0, "int_id");
 919                                         }
 920
 921                                         _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
 922
 923                                         $post_needs_update = false;
 924                                         $update_insignificant = false;
 925
 926                                         if ($orig_num_comments != $num_comments) {
 927                                                 $post_needs_update = true;
 928                                                 $update_insignificant = true;
 929                                         }
 930
 931                                         if ($entry_plugin_data != $orig_plugin_data) {
 932                                                 $post_needs_update = true;
 933                                                 $update_insignificant = true;
 934                                         }
 935
 936                                         if ($content_hash != $orig_content_hash) {
 937                                                 $post_needs_update = true;
 938                                                 $update_insignificant = false;
 939                                         }
 940
 941                                         if (db_escape_string($orig_title) != $entry_title) {
 942                                                 $post_needs_update = true;
 943                                                 $update_insignificant = false;
 944                                         }
 945
 946                                         // if post needs update, update it and mark all user entries
 947                                         // linking to this post as updated
 948                                         if ($post_needs_update) {
 949
 950                                                 if (defined('DAEMON_EXTENDED_DEBUG')) {
 951                                                         _debug("post $entry_guid_hashed needs update...", $debug_enabled);
 952                                                 }
 953
 954 //                                              print "<!-- post $orig_title needs update : $post_needs_update -->";
 955
 956                                                 db_query("UPDATE ttrss_entries
 957                                                         SET title = '$entry_title', content = '$entry_content',
 958                                                                 content_hash = '$content_hash',
 959                                                                 updated = '$entry_timestamp_fmt',
 960                                                                 num_comments = '$num_comments',
 961                                                                 plugin_data = '$entry_plugin_data'
 962                                                         WHERE id = '$ref_id'");
 963
 964                                                 if (!$update_insignificant) {
 965                                                         if ($mark_unread_on_update) {
 966                                                                 db_query("UPDATE ttrss_user_entries
 967                                                                         SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
 968                                                         }
 969                                                 }
 970                                         }
 971                                 }
 972
 973                                 db_query("COMMIT");
 974
 975                                 _debug("assigning labels...", $debug_enabled);
 976
 977                                 assign_article_to_label_filters($entry_ref_id, $article_filters,
 978                                         $owner_uid, $article_labels);
 979
 980                                 _debug("looking for enclosures...", $debug_enabled);
 981
 982                                 // enclosures
 983
 984                                 $enclosures = array();
 985
 986                                 $encs = $item->get_enclosures();
 987
 988                                 if (is_array($encs)) {
 989                                         foreach ($encs as $e) {
 990                                                 $e_item = array(
 991                                                         $e->link, $e->type, $e->length, $e->title);
 992                                                 array_push($enclosures, $e_item);
 993                                         }
 994                                 }
 995
 996                                 if ($debug_enabled) {
 997                                         _debug("article enclosures:", $debug_enabled);
 998                                         print_r($enclosures);
 999                                 }
1000
1001                                 db_query("BEGIN");
1002
1003 //                              debugging
1004 //                              db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
1005
1006                                 foreach ($enclosures as $enc) {
1007                                         $enc_url = db_escape_string($enc[0]);
1008                                         $enc_type = db_escape_string($enc[1]);
1009                                         $enc_dur = db_escape_string($enc[2]);
1010                                         $enc_title = db_escape_string($enc[3]);
1011
1012                                         $result = db_query("SELECT id FROM ttrss_enclosures
1013                                                 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
1014
1015                                         if (db_num_rows($result) == 0) {
1016                                                 db_query("INSERT INTO ttrss_enclosures
1017                                                         (content_url, content_type, title, duration, post_id) VALUES
1018                                                         ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id')");
1019                                         }
1020                                 }
1021
1022                                 db_query("COMMIT");
1023
1024                                 // check for manual tags (we have to do it here since they're loaded from filters)
1025
1026                                 foreach ($article_filters as $f) {
1027                                         if ($f["type"] == "tag") {
1028
1029                                                 $manual_tags = trim_array(explode(",", $f["param"]));
1030
1031                                                 foreach ($manual_tags as $tag) {
1032                                                         if (tag_is_valid($tag)) {
1033                                                                 array_push($entry_tags, $tag);
1034                                                         }
1035                                                 }
1036                                         }
1037                                 }
1038
1039                                 // Skip boring tags
1040
1041                                 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
1042                                         'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1043
1044                                 $filtered_tags = array();
1045                                 $tags_to_cache = array();
1046
1047                                 if ($entry_tags && is_array($entry_tags)) {
1048                                         foreach ($entry_tags as $tag) {
1049                                                 if (array_search($tag, $boring_tags) === false) {
1050                                                         array_push($filtered_tags, $tag);
1051                                                 }
1052                                         }
1053                                 }
1054
1055                                 $filtered_tags = array_unique($filtered_tags);
1056
1057                                 if ($debug_enabled) {
1058                                         _debug("filtered article tags:", $debug_enabled);
1059                                         print_r($filtered_tags);
1060                                 }
1061
1062                                 // Save article tags in the database
1063
1064                                 if (count($filtered_tags) > 0) {
1065
1066                                         db_query("BEGIN");
1067
1068                                         foreach ($filtered_tags as $tag) {
1069
1070                                                 $tag = sanitize_tag($tag);
1071                                                 $tag = db_escape_string($tag);
1072
1073                                                 if (!tag_is_valid($tag)) continue;
1074
1075                                                 $result = db_query("SELECT id FROM ttrss_tags
1076                                                         WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1077                                                         owner_uid = '$owner_uid' LIMIT 1");
1078
1079                                                         if ($result && db_num_rows($result) == 0) {
1080
1081                                                                 db_query("INSERT INTO ttrss_tags
1082                                                                         (owner_uid,tag_name,post_int_id)
1083                                                                         VALUES ('$owner_uid','$tag', '$entry_int_id')");
1084                                                         }
1085
1086                                                 array_push($tags_to_cache, $tag);
1087                                         }
1088
1089                                         /* update the cache */
1090
1091                                         $tags_to_cache = array_unique($tags_to_cache);
1092
1093                                         $tags_str = db_escape_string(join(",", $tags_to_cache));
1094
1095                                         db_query("UPDATE ttrss_user_entries
1096                                                 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1097                                                 AND owner_uid = $owner_uid");
1098
1099                                         db_query("COMMIT");
1100                                 }
1101
1102                                 if (get_pref("AUTO_ASSIGN_LABELS", $owner_uid, false)) {
1103                                         _debug("auto-assigning labels...", $debug_enabled);
1104
1105                                         foreach ($labels as $label) {
1106                                                 $caption = preg_quote($label["caption"]);
1107
1108                                                 if ($caption && preg_match("/\b$caption\b/i", "$tags_str " . strip_tags($entry_content) . " $entry_title")) {
1109                                                         if (!labels_contains_caption($article_labels, $caption)) {
1110                                                                 label_add_article($entry_ref_id, $caption, $owner_uid);
1111                                                         }
1112                                                 }
1113                                         }
1114                                 }
1115
1116                                 _debug("article processed", $debug_enabled);
1117                         }
1118
1119                         _debug("purging feed...", $debug_enabled);
1120
1121                         purge_feed($feed, 0, $debug_enabled);
1122
1123                         db_query("UPDATE ttrss_feeds
1124                                 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
1125
1126 //                      db_query("COMMIT");
1127
1128                 } else {
1129
1130                         $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
1131
1132                         _debug("fetch error: $error_msg", $debug_enabled);
1133
1134                         if (count($rss->errors()) > 1) {
1135                                 foreach ($rss->errors() as $error) {
1136                                         _debug("+ $error");
1137                                 }
1138                         }
1139
1140                         db_query(
1141                                 "UPDATE ttrss_feeds SET last_error = '$error_msg',
1142                                 last_updated = NOW() WHERE id = '$feed'");
1143
1144                         unset($rss);
1145                 }
1146
1147                 _debug("done", $debug_enabled);
1148
1149                 return $rss;
1150         }
1151
1152         function cache_images($html, $site_url, $debug) {
1153                 libxml_use_internal_errors(true);
1154
1155                 $charset_hack = '<head>
1156                         <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1157                 </head>';
1158
1159                 $doc = new DOMDocument();
1160                 $doc->loadHTML($charset_hack . $html);
1161                 $xpath = new DOMXPath($doc);
1162
1163                 $entries = $xpath->query('(//img[@src])');
1164
1165                 foreach ($entries as $entry) {
1166                         if ($entry->hasAttribute('src')) {
1167                                 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1168
1169                                 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1170
1171                                 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1172
1173                                 if (!file_exists($local_filename)) {
1174                                         $file_content = fetch_file_contents($src);
1175
1176                                         if ($file_content && strlen($file_content) > 1024) {
1177                                                 file_put_contents($local_filename, $file_content);
1178                                         }
1179                                 }
1180
1181                                 /* if (file_exists($local_filename)) {
1182                                         $entry->setAttribute('src', SELF_URL_PATH . '/image.php?url=' .
1183                                                 base64_encode($src));
1184                                 } */
1185                         }
1186                 }
1187
1188                 //$node = $doc->getElementsByTagName('body')->item(0);
1189                 //return $doc->saveXML($node);
1190         }
1191
1192         function expire_error_log($debug) {
1193                 if ($debug) _debug("Removing old error log entries...");
1194
1195                 if (DB_TYPE == "pgsql") {
1196                         db_query("DELETE FROM ttrss_error_log
1197                                 WHERE created_at < NOW() - INTERVAL '7 days'");
1198                 } else {
1199                         db_query("DELETE FROM ttrss_error_log
1200                                 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1201                 }
1202
1203         }
1204
1205         function expire_lock_files($debug) {
1206                 //if ($debug) _debug("Removing old lock files...");
1207
1208                 $num_deleted = 0;
1209
1210                 if (is_writable(LOCK_DIRECTORY)) {
1211                         $files = glob(LOCK_DIRECTORY . "/*.lock");
1212
1213                         if ($files) {
1214                                 foreach ($files as $file) {
1215                                         if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
1216                                                 unlink($file);
1217                                                 ++$num_deleted;
1218                                         }
1219                                 }
1220                         }
1221                 }
1222
1223                 if ($debug) _debug("Removed $num_deleted old lock files.");
1224         }
1225
1226         function expire_cached_files($debug) {
1227                 foreach (array("simplepie", "images", "export", "upload") as $dir) {
1228                         $cache_dir = CACHE_DIR . "/$dir";
1229
1230 //                      if ($debug) _debug("Expiring $cache_dir");
1231
1232                         $num_deleted = 0;
1233
1234                         if (is_writable($cache_dir)) {
1235                                 $files = glob("$cache_dir/*");
1236
1237                                 if ($files) {
1238                                         foreach ($files as $file) {
1239                                                 if (time() - filemtime($file) > 86400*7) {
1240                                                         unlink($file);
1241
1242                                                         ++$num_deleted;
1243                                                 }
1244                                         }
1245                                 }
1246                         }
1247
1248                         if ($debug) _debug("$cache_dir: removed $num_deleted files.");
1249                 }
1250         }
1251
1252         /**
1253         * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1254         * Returns the url query as associative array
1255         *
1256         * @param    string    query
1257         * @return    array    params
1258         */
1259         function convertUrlQuery($query) {
1260                 $queryParts = explode('&', $query);
1261
1262                 $params = array();
1263
1264                 foreach ($queryParts as $param) {
1265                         $item = explode('=', $param);
1266                         $params[$item[0]] = $item[1];
1267                 }
1268
1269                 return $params;
1270         }
1271
1272         function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
1273                 $matches = array();
1274
1275                 foreach ($filters as $filter) {
1276                         $match_any_rule = $filter["match_any_rule"];
1277                         $inverse = $filter["inverse"];
1278                         $filter_match = false;
1279
1280                         foreach ($filter["rules"] as $rule) {
1281                                 $match = false;
1282                                 $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
1283                                 $rule_inverse = $rule["inverse"];
1284
1285                                 if (!$reg_exp)
1286                                         continue;
1287
1288                                 switch ($rule["type"]) {
1289                                 case "title":
1290                                         $match = @preg_match("/$reg_exp/i", $title);
1291                                         break;
1292                                 case "content":
1293                                         // we don't need to deal with multiline regexps
1294                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1295
1296                                         $match = @preg_match("/$reg_exp/i", $content);
1297                                         break;
1298                                 case "both":
1299                                         // we don't need to deal with multiline regexps
1300                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1301
1302                                         $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
1303                                         break;
1304                                 case "link":
1305                                         $match = @preg_match("/$reg_exp/i", $link);
1306                                         break;
1307                                 case "author":
1308                                         $match = @preg_match("/$reg_exp/i", $author);
1309                                         break;
1310                                 case "tag":
1311                                         foreach ($tags as $tag) {
1312                                                 if (@preg_match("/$reg_exp/i", $tag)) {
1313                                                         $match = true;
1314                                                         break;
1315                                                 }
1316                                         }
1317                                         break;
1318                                 }
1319
1320                                 if ($rule_inverse) $match = !$match;
1321
1322                                 if ($match_any_rule) {
1323                                         if ($match) {
1324                                                 $filter_match = true;
1325                                                 break;
1326                                         }
1327                                 } else {
1328                                         $filter_match = $match;
1329                                         if (!$match) {
1330                                                 break;
1331                                         }
1332                                 }
1333                         }
1334
1335                         if ($inverse) $filter_match = !$filter_match;
1336
1337                         if ($filter_match) {
1338                                 foreach ($filter["actions"] AS $action) {
1339                                         array_push($matches, $action);
1340
1341                                         // if Stop action encountered, perform no further processing
1342                                         if ($action["type"] == "stop") return $matches;
1343                                 }
1344                         }
1345                 }
1346
1347                 return $matches;
1348         }
1349
1350         function find_article_filter($filters, $filter_name) {
1351                 foreach ($filters as $f) {
1352                         if ($f["type"] == $filter_name) {
1353                                 return $f;
1354                         };
1355                 }
1356                 return false;
1357         }
1358
1359         function find_article_filters($filters, $filter_name) {
1360                 $results = array();
1361
1362                 foreach ($filters as $f) {
1363                         if ($f["type"] == $filter_name) {
1364                                 array_push($results, $f);
1365                         };
1366                 }
1367                 return $results;
1368         }
1369
1370         function calculate_article_score($filters) {
1371                 $score = 0;
1372
1373                 foreach ($filters as $f) {
1374                         if ($f["type"] == "score") {
1375                                 $score += $f["param"];
1376                         };
1377                 }
1378                 return $score;
1379         }
1380
1381         function labels_contains_caption($labels, $caption) {
1382                 foreach ($labels as $label) {
1383                         if ($label[1] == $caption) {
1384                                 return true;
1385                         }
1386                 }
1387
1388                 return false;
1389         }
1390
1391         function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
1392                 foreach ($filters as $f) {
1393                         if ($f["type"] == "label") {
1394                                 if (!labels_contains_caption($article_labels, $f["param"])) {
1395                                         label_add_article($id, $f["param"], $owner_uid);
1396                                 }
1397                         }
1398                 }
1399         }
1400
1401         function make_guid_from_title($title) {
1402                 return preg_replace("/[ \"\',.:;]/", "-",
1403                         mb_strtolower(strip_tags($title), 'utf-8'));
1404         }
1405
1406         /* function verify_feed_xml($feed_data) {
1407                 libxml_use_internal_errors(true);
1408                 $doc = new DOMDocument();
1409                 $doc->loadXML($feed_data);
1410                 $error = libxml_get_last_error();
1411                 libxml_clear_errors();
1412                 return $error;
1413         } */
1414
1415         function housekeeping_common($debug) {
1416                 expire_cached_files($debug);
1417                 expire_lock_files($debug);
1418                 expire_error_log($debug);
1419
1420                 $count = update_feedbrowser_cache();
1421                 _debug("Feedbrowser updated, $count feeds processed.");
1422
1423                 purge_orphans( true);
1424                 $rc = cleanup_tags( 14, 50000);
1425
1426                 _debug("Cleaned $rc cached tags.");
1427
1428                 PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
1429
1430         }
1431 ?>