include/rssfuncs.php

   1 <?php
   2         define('DAEMON_UPDATE_LOGIN_LIMIT', 30);
   3         define('DAEMON_FEED_LIMIT', 100);
   4         define('DAEMON_SLEEP_INTERVAL', 60);
   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 '5 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 5 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                 $random_qpart = sql_random_function();
 117
 118                 // We search for feed needing update.
 119                 $result = db_query("SELECT DISTINCT ttrss_feeds.feed_url,$random_qpart
 120                         FROM
 121                                 ttrss_feeds, ttrss_users, ttrss_user_prefs
 122                         WHERE
 123                                 ttrss_feeds.owner_uid = ttrss_users.id
 124                                 AND ttrss_users.id = ttrss_user_prefs.owner_uid
 125                                 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
 126                                 $login_thresh_qpart $update_limit_qpart
 127                                 $updstart_thresh_qpart
 128                         ORDER BY $random_qpart $query_limit");
 129
 130                 if($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(',', $feeds_quoted)));
 151                 }
 152
 153                 expire_cached_files($debug);
 154                 expire_lock_files($debug);
 155                 expire_error_log($debug);
 156
 157                 $nf = 0;
 158
 159                 // For each feed, we call the feed update function.
 160                 foreach ($feeds_to_update as $feed) {
 161                         if($debug) _debug("Base feed: $feed");
 162
 163                         //update_rss_feed($line["id"], true);
 164
 165                         // since we have the data cached, we can deal with other feeds with the same url
 166
 167                         $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated
 168                         FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
 169                                 ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
 170                                 ttrss_users.id = ttrss_user_prefs.owner_uid AND
 171                                 ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
 172                                 feed_url = '".db_escape_string($feed)."' AND
 173                                 (ttrss_feeds.update_interval > 0 OR
 174                                         ttrss_user_prefs.value != '-1')
 175                                 $login_thresh_qpart
 176                         ORDER BY ttrss_feeds.id $query_limit");
 177
 178                         if (db_num_rows($tmp_result) > 0) {
 179                                 while ($tline = db_fetch_assoc($tmp_result)) {
 180                                         if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"]);
 181                                         update_rss_feed($tline["id"], true);
 182                                         ++$nf;
 183                                 }
 184                         }
 185                 }
 186
 187                 require_once "digest.php";
 188
 189                 // Send feed digests by email if needed.
 190                 send_headlines_digests($debug);
 191
 192                 return $nf;
 193
 194         } // function update_daemon_common
 195
 196         // ignore_daemon is not used
 197         function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false,
 198                 $override_url = false) {
 199
 200                 require_once "lib/simplepie/simplepie.inc";
 201
 202                 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
 203
 204                 if ($debug_enabled) {
 205                         _debug("update_rss_feed: start");
 206                 }
 207
 208                 $result = db_query("SELECT id,update_interval,auth_login,
 209                         feed_url,auth_pass,cache_images,last_updated,
 210                         mark_unread_on_update, owner_uid,
 211                         pubsub_state, auth_pass_encrypted
 212                         FROM ttrss_feeds WHERE id = '$feed'");
 213
 214                 if (db_num_rows($result) == 0) {
 215                         if ($debug_enabled) {
 216                                 _debug("update_rss_feed: feed $feed NOT FOUND/SKIPPED");
 217                         }
 218                         return false;
 219                 }
 220
 221                 $last_updated = db_fetch_result($result, 0, "last_updated");
 222                 $owner_uid = db_fetch_result($result, 0, "owner_uid");
 223                 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
 224                         0, "mark_unread_on_update"));
 225                 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
 226                 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
 227                         0, "auth_pass_encrypted"));
 228
 229                 db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
 230                         WHERE id = '$feed'");
 231
 232                 $auth_login = db_fetch_result($result, 0, "auth_login");
 233                 $auth_pass = db_fetch_result($result, 0, "auth_pass");
 234
 235                 if ($auth_pass_encrypted) {
 236                         require_once "crypt.php";
 237                         $auth_pass = decrypt_string($auth_pass);
 238                 }
 239
 240                 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
 241                 $fetch_url = db_fetch_result($result, 0, "feed_url");
 242
 243                 $feed = db_escape_string($feed);
 244
 245                 if ($override_url) $fetch_url = $override_url;
 246
 247                 $date_feed_processed = date('Y-m-d H:i');
 248
 249                 $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".feed";
 250
 251                 // Ignore cache if new feed or manual update.
 252                 $cache_age = ($no_cache || is_null($last_updated) || $last_updated == '1970-01-01 00:00:00') ?
 253                         30 : get_feed_update_interval($feed) * 60;
 254
 255                 if ($debug_enabled) {
 256                         _debug("update_rss_feed: cache filename: $cache_filename exists: " . file_exists($cache_filename));
 257                         _debug("update_rss_feed: cache age: $cache_age; no cache: $no_cache");
 258                 }
 259
 260                 $cached_feed_data_hash = false;
 261
 262                 $rss = false;
 263                 $rss_hash = false;
 264                 $cache_timestamp = file_exists($cache_filename) ? filemtime($cache_filename) : 0;
 265                 $last_updated_timestamp = strtotime($last_updated);
 266
 267                 $force_refetch = isset($_REQUEST["force_refetch"]);
 268
 269                 if (file_exists($cache_filename) &&
 270                         is_readable($cache_filename) &&
 271                         !$auth_login && !$auth_pass &&
 272                         filemtime($cache_filename) > time() - $cache_age) {
 273
 274                                 if ($debug_enabled) {
 275                                         _debug("update_rss_feed: using local cache.");
 276                                 }
 277
 278                                 if ($cache_timestamp > $last_updated_timestamp) {
 279                                         @$rss_data = file_get_contents($cache_filename);
 280
 281                                         if ($rss_data) {
 282                                                 $rss_hash = sha1($rss_data);
 283                                                 @$rss = unserialize($rss_data);
 284                                         }
 285                                 } else if (!$force_refetch) {
 286                                         if ($debug_enabled) {
 287                                                 _debug("update_rss_feed: local cache valid and older than last_updated, nothing to do.");
 288                                         }
 289                                         return;
 290                                 }
 291                 }
 292
 293                 if (!$rss) {
 294
 295                         if (!$feed_data) {
 296                                 if ($debug_enabled) {
 297                                         _debug("update_rss_feed: fetching [$fetch_url] (ts: $cache_timestamp/$last_updated_timestamp)");
 298                                 }
 299
 300                                 $feed_data = fetch_file_contents($fetch_url, false,
 301                                         $auth_login, $auth_pass, false,
 302                                         $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
 303                                         $force_refetch ? 0 : max($last_updated_timestamp, $cache_timestamp));
 304
 305                                 global $fetch_curl_used;
 306
 307                                 if (!$fetch_curl_used) {
 308                                         $tmp = @gzdecode($feed_data);
 309
 310                                         if ($tmp) $feed_data = $tmp;
 311                                 }
 312
 313                                 if ($debug_enabled) {
 314                                         _debug("update_rss_feed: fetch done.");
 315                                 }
 316
 317                         }
 318
 319                         if (!$feed_data) {
 320                                 global $fetch_last_error;
 321                                 global $fetch_last_error_code;
 322
 323                                 if ($debug_enabled) {
 324                                         _debug("update_rss_feed: unable to fetch: $fetch_last_error [$fetch_last_error_code]");
 325                                 }
 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                                         if ($debug_enabled) {
 334                                                 _debug("update_rss_feed: source claims data not modified, nothing to do.");
 335                                         }
 336                                 }
 337
 338                                 db_query(
 339                                         "UPDATE ttrss_feeds SET last_error = '$error_escaped',
 340                                                 last_updated = NOW() WHERE id = '$feed'");
 341
 342                                 return;
 343                         }
 344                 }
 345
 346                 $pluginhost = new PluginHost();
 347                 $pluginhost->set_debug($debug_enabled);
 348                 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
 349
 350                 $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
 351                 $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
 352                 $pluginhost->load_data();
 353
 354                 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
 355                         $feed_data = $plugin->hook_feed_fetched($feed_data);
 356                 }
 357
 358                 if (!$rss) {
 359                         $rss = new SimplePie();
 360                         $rss->set_sanitize_class("SanitizeDummy");
 361                         // simplepie ignores the above and creates default sanitizer anyway,
 362                         // so let's override it...
 363                         $rss->sanitize = new SanitizeDummy();
 364                         $rss->set_output_encoding('UTF-8');
 365                         $rss->set_raw_data($feed_data);
 366                         $rss->enable_cache(false);
 367
 368                         @$rss->init();
 369                 }
 370
 371 //              print_r($rss);
 372
 373                 $feed = db_escape_string($feed);
 374
 375                 if (!$rss->error()) {
 376
 377                         // cache data for later
 378                         if (!$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
 379                                 $rss_data = serialize($rss);
 380                                 $new_rss_hash = sha1($rss_data);
 381
 382                                 if ($new_rss_hash != $rss_hash) {
 383                                         if ($debug_enabled) {
 384                                                 _debug("update_rss_feed: saving $cache_filename");
 385                                         }
 386                                         @file_put_contents($cache_filename, serialize($rss));
 387                                 }
 388                         }
 389
 390                         // We use local pluginhost here because we need to load different per-user feed plugins
 391                         $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
 392
 393                         if ($debug_enabled) {
 394                                 _debug("update_rss_feed: processing feed data...");
 395                         }
 396
 397 //                      db_query("BEGIN");
 398
 399                         if (DB_TYPE == "pgsql") {
 400                                 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
 401                         } else {
 402                                 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
 403                         }
 404
 405                         $result = db_query("SELECT title,site_url,owner_uid,
 406                                 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
 407                                                 favicon_needs_check
 408                                 FROM ttrss_feeds WHERE id = '$feed'");
 409
 410                         $registered_title = db_fetch_result($result, 0, "title");
 411                         $orig_site_url = db_fetch_result($result, 0, "site_url");
 412                         $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
 413                                 "favicon_needs_check"));
 414
 415                         $owner_uid = db_fetch_result($result, 0, "owner_uid");
 416
 417                         $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
 418
 419                         if ($favicon_needs_check || $force_refetch) {
 420                                 if ($debug_enabled) {
 421                                         _debug("update_rss_feed: checking favicon...");
 422                                 }
 423
 424                                 check_feed_favicon($site_url, $feed, $link);
 425                                 $favicon_file = ICONS_DIR . "/$feed.ico";
 426
 427                                 if (file_exists($favicon_file) && function_exists("imagecreatefromstring")) {
 428                                                 require_once "colors.php";
 429
 430                                                 $favicon_color = db_escape_string(
 431                                                         calculate_avg_color($favicon_file));
 432
 433                                                 $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
 434                                 }
 435
 436                                 db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
 437                                         $favicon_colorstring
 438                                         WHERE id = '$feed'");
 439                         }
 440
 441                         if (!$registered_title || $registered_title == "[Unknown]") {
 442
 443                                 $feed_title = db_escape_string($rss->get_title());
 444
 445                                 if ($debug_enabled) {
 446                                         _debug("update_rss_feed: registering title: $feed_title");
 447                                 }
 448
 449                                 db_query("UPDATE ttrss_feeds SET
 450                                         title = '$feed_title' WHERE id = '$feed'");
 451                         }
 452
 453                         if ($site_url && $orig_site_url != $site_url) {
 454                                 db_query("UPDATE ttrss_feeds SET
 455                                         site_url = '$site_url' WHERE id = '$feed'");
 456                         }
 457
 458                         if ($debug_enabled) {
 459                                 _debug("update_rss_feed: loading filters & labels...");
 460                         }
 461
 462                         $filters = load_filters($feed, $owner_uid);
 463                         $labels = get_all_labels($owner_uid);
 464
 465                         if ($debug_enabled) {
 466                                 //print_r($filters);
 467                                 _debug("update_rss_feed: " . count($filters) . " filters loaded.");
 468                         }
 469
 470                         $items = $rss->get_items();
 471
 472                         if (!is_array($items)) {
 473                                 if ($debug_enabled) {
 474                                         _debug("update_rss_feed: no articles found.");
 475                                 }
 476
 477                                 db_query("UPDATE ttrss_feeds
 478                                         SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
 479
 480                                 return; // no articles
 481                         }
 482
 483                         if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
 484
 485                                 if ($debug_enabled) _debug("update_rss_feed: checking for PUSH hub...");
 486
 487                                 $feed_hub_url = false;
 488
 489                                 $links = $rss->get_links('hub');
 490
 491                                 if ($links && is_array($links)) {
 492                                         foreach ($links as $l) {
 493                                                 $feed_hub_url = $l;
 494                                                 break;
 495                                         }
 496                                 }
 497
 498                                 if ($debug_enabled) _debug("update_rss_feed: feed hub url: $feed_hub_url");
 499
 500                                 if ($feed_hub_url && function_exists('curl_init') &&
 501                                         !ini_get("open_basedir")) {
 502
 503                                         require_once 'lib/pubsubhubbub/subscriber.php';
 504
 505                                         $callback_url = get_self_url_prefix() .
 506                                                 "/public.php?op=pubsub&id=$feed";
 507
 508                                         $s = new Subscriber($feed_hub_url, $callback_url);
 509
 510                                         $rc = $s->subscribe($fetch_url);
 511
 512                                         if ($debug_enabled)
 513                                                 _debug("update_rss_feed: feed hub url found, subscribe request sent.");
 514
 515                                         db_query("UPDATE ttrss_feeds SET pubsub_state = 1
 516                                                 WHERE id = '$feed'");
 517                                 }
 518                         }
 519
 520                         if ($debug_enabled) {
 521                                 _debug("update_rss_feed: processing articles...");
 522                         }
 523
 524                         foreach ($items as $item) {
 525                                 if ($_REQUEST['xdebug'] == 3) {
 526                                         print_r($item);
 527                                 }
 528
 529                                 $entry_guid = $item->get_id();
 530                                 if (!$entry_guid) $entry_guid = $item->get_link();
 531                                 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
 532
 533                                 if (!$entry_guid) continue;
 534
 535                                 $entry_guid = "$owner_uid,$entry_guid";
 536
 537                                 $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
 538
 539                                 if ($debug_enabled) {
 540                                         _debug("update_rss_feed: guid $entry_guid / $entry_guid_hashed");
 541                                 }
 542
 543                                 $entry_timestamp = "";
 544
 545                                 $entry_timestamp = strtotime($item->get_date());
 546
 547                                 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
 548                                         $entry_timestamp = time();
 549                                         $no_orig_date = 'true';
 550                                 } else {
 551                                         $no_orig_date = 'false';
 552                                 }
 553
 554                                 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
 555
 556                                 if ($debug_enabled) {
 557                                         _debug("update_rss_feed: date $entry_timestamp [$entry_timestamp_fmt]");
 558                                 }
 559
 560                                 $entry_title = html_entity_decode($item->get_title());
 561
 562                                 $entry_link = rewrite_relative_url($site_url, $item->get_link());
 563
 564                                 if ($debug_enabled) {
 565                                         _debug("update_rss_feed: title $entry_title");
 566                                         _debug("update_rss_feed: link $entry_link");
 567                                 }
 568
 569                                 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
 570
 571                                 $entry_content = $item->get_content();
 572                                 if (!$entry_content) $entry_content = $item->get_description();
 573
 574                                 if ($_REQUEST["xdebug"] == 2) {
 575                                         print "update_rss_feed: content: ";
 576                                         print $entry_content;
 577                                         print "\n";
 578                                 }
 579
 580                                 $entry_comments = $item->data["comments"];
 581
 582                                 if ($item->get_author()) {
 583                                         $entry_author_item = $item->get_author();
 584                                         $entry_author = $entry_author_item->get_name();
 585                                         if (!$entry_author) $entry_author = $entry_author_item->get_email();
 586                                 }
 587
 588                                 $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
 589
 590                                 $entry_comments = db_escape_string(mb_substr(trim($entry_comments), 0, 245));
 591                                 $entry_author = db_escape_string(mb_substr(trim($entry_author), 0, 245));
 592
 593                                 $num_comments = $item->get_item_tags('http://purl.org/rss/1.0/modules/slash/', 'comments');
 594
 595                                 if (is_array($num_comments) && is_array($num_comments[0])) {
 596                                         $num_comments = (int) $num_comments[0]["data"];
 597                                 } else {
 598                                         $num_comments = 0;
 599                                 }
 600
 601                                 if ($debug_enabled) {
 602                                         _debug("update_rss_feed: author $entry_author");
 603                                         _debug("update_rss_feed: num_comments: $num_comments");
 604                                         _debug("update_rss_feed: looking for tags [1]...");
 605                                 }
 606
 607                                 // parse <category> entries into tags
 608
 609                                 $additional_tags = array();
 610
 611                                 $additional_tags_src = $item->get_categories();
 612
 613                                 if (is_array($additional_tags_src)) {
 614                                         foreach ($additional_tags_src as $tobj) {
 615                                                 array_push($additional_tags, $tobj->get_term());
 616                                         }
 617                                 }
 618
 619                                 if ($debug_enabled) {
 620                                         _debug("update_rss_feed: category tags:");
 621                                         print_r($additional_tags);
 622                                 }
 623
 624                                 if ($debug_enabled) {
 625                                         _debug("update_rss_feed: looking for tags [2]...");
 626                                 }
 627
 628                                 $entry_tags = array_unique($additional_tags);
 629
 630                                 for ($i = 0; $i < count($entry_tags); $i++)
 631                                         $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
 632
 633                                 if ($debug_enabled) {
 634                                         //_debug("update_rss_feed: unfiltered tags found:");
 635                                         //print_r($entry_tags);
 636                                 }
 637
 638                                 if ($debug_enabled) {
 639                                         _debug("update_rss_feed: done collecting data.");
 640                                 }
 641
 642                                 // TODO: less memory-hungry implementation
 643
 644                                 if ($debug_enabled) {
 645                                         _debug("update_rss_feed: applying plugin filters..");
 646                                 }
 647
 648                                 // FIXME not sure if owner_uid is a good idea here, we may have a base entry without user entry (?)
 649                                 $result = db_query("SELECT plugin_data,title,content,link,tag_cache,author FROM ttrss_entries, ttrss_user_entries
 650                                         WHERE ref_id = id AND (guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed') AND owner_uid = $owner_uid");
 651
 652                                 if (db_num_rows($result) != 0) {
 653                                         $entry_plugin_data = db_fetch_result($result, 0, "plugin_data");
 654                                         $stored_article = array("title" => db_fetch_result($result, 0, "title"),
 655                                                 "content" => db_fetch_result($result, 0, "content"),
 656                                                 "link" => db_fetch_result($result, 0, "link"),
 657                                                 "tags" => explode(",", db_fetch_result($result, 0, "tag_cache")),
 658                                                 "author" => db_fetch_result($result, 0, "author"));
 659                                 } else {
 660                                         $entry_plugin_data = "";
 661                                         $stored_article = array();
 662                                 }
 663
 664                                 $article = array("owner_uid" => $owner_uid, // read only
 665                                         "guid" => $entry_guid, // read only
 666                                         "title" => $entry_title,
 667                                         "content" => $entry_content,
 668                                         "link" => $entry_link,
 669                                         "tags" => $entry_tags,
 670                                         "plugin_data" => $entry_plugin_data,
 671                                         "author" => $entry_author,
 672                                         "stored" => $stored_article);
 673
 674                                 foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
 675                                         $article = $plugin->hook_article_filter($article);
 676                                 }
 677
 678                                 $entry_tags = $article["tags"];
 679                                 $entry_guid = db_escape_string($entry_guid);
 680                                 $entry_title = db_escape_string($article["title"]);
 681                                 $entry_author = db_escape_string($article["author"]);
 682                                 $entry_link = db_escape_string($article["link"]);
 683                                 $entry_plugin_data = db_escape_string($article["plugin_data"]);
 684                                 $entry_content = $article["content"]; // escaped below
 685
 686
 687                                 if ($debug_enabled) {
 688                                         _debug("update_rss_feed: plugin data: $entry_plugin_data");
 689                                 }
 690
 691                                 if ($cache_images && is_writable(CACHE_DIR . '/images'))
 692                                         cache_images($entry_content, $site_url, $debug_enabled);
 693
 694                                 $entry_content = db_escape_string($entry_content, false);
 695
 696                                 $content_hash = "SHA1:" . sha1($entry_content);
 697
 698                                 db_query("BEGIN");
 699
 700                                 $result = db_query("SELECT id FROM      ttrss_entries
 701                                         WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
 702
 703                                 if (db_num_rows($result) == 0) {
 704
 705                                         if ($debug_enabled) {
 706                                                 _debug("update_rss_feed: base guid [$entry_guid] not found");
 707                                         }
 708
 709                                         // base post entry does not exist, create it
 710
 711                                         $result = db_query(
 712                                                 "INSERT INTO ttrss_entries
 713                                                         (title,
 714                                                         guid,
 715                                                         link,
 716                                                         updated,
 717                                                         content,
 718                                                         content_hash,
 719                                                         cached_content,
 720                                                         no_orig_date,
 721                                                         date_updated,
 722                                                         date_entered,
 723                                                         comments,
 724                                                         num_comments,
 725                                                         plugin_data,
 726                                                         author)
 727                                                 VALUES
 728                                                         ('$entry_title',
 729                                                         '$entry_guid_hashed',
 730                                                         '$entry_link',
 731                                                         '$entry_timestamp_fmt',
 732                                                         '$entry_content',
 733                                                         '$content_hash',
 734                                                         '',
 735                                                         $no_orig_date,
 736                                                         NOW(),
 737                                                         '$date_feed_processed',
 738                                                         '$entry_comments',
 739                                                         '$num_comments',
 740                                                         '$entry_plugin_data',
 741                                                         '$entry_author')");
 742
 743                                         $article_labels = array();
 744
 745                                 } else {
 746                                         // we keep encountering the entry in feeds, so we need to
 747                                         // update date_updated column so that we don't get horrible
 748                                         // dupes when the entry gets purged and reinserted again e.g.
 749                                         // in the case of SLOW SLOW OMG SLOW updating feeds
 750
 751                                         $base_entry_id = db_fetch_result($result, 0, "id");
 752
 753                                         db_query("UPDATE ttrss_entries SET date_updated = NOW()
 754                                                 WHERE id = '$base_entry_id'");
 755
 756                                         $article_labels = get_article_labels($base_entry_id, $owner_uid);
 757                                 }
 758
 759                                 // now it should exist, if not - bad luck then
 760
 761                                 $result = db_query("SELECT
 762                                                 id,content_hash,no_orig_date,title,plugin_data,guid,
 763                                                 ".SUBSTRING_FOR_DATE."(date_updated,1,19) as date_updated,
 764                                                 ".SUBSTRING_FOR_DATE."(updated,1,19) as updated,
 765                                                 num_comments
 766                                         FROM
 767                                                 ttrss_entries
 768                                         WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
 769
 770                                 $entry_ref_id = 0;
 771                                 $entry_int_id = 0;
 772
 773                                 if (db_num_rows($result) == 1) {
 774
 775                                         if ($debug_enabled) {
 776                                                 _debug("update_rss_feed: base guid found, checking for user record");
 777                                         }
 778
 779                                         // this will be used below in update handler
 780                                         $orig_content_hash = db_fetch_result($result, 0, "content_hash");
 781                                         $orig_title = db_fetch_result($result, 0, "title");
 782                                         $orig_num_comments = db_fetch_result($result, 0, "num_comments");
 783                                         $orig_date_updated = strtotime(db_fetch_result($result,
 784                                                 0, "date_updated"));
 785                                         $orig_plugin_data = db_fetch_result($result, 0, "plugin_data");
 786
 787                                         $ref_id = db_fetch_result($result, 0, "id");
 788                                         $entry_ref_id = $ref_id;
 789
 790                                         /* $stored_guid = db_fetch_result($result, 0, "guid");
 791                                         if ($stored_guid != $entry_guid_hashed) {
 792                                                 if ($debug_enabled) _debug("upgrading compat guid to hashed one");
 793
 794                                                 db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
 795                                                         id = '$ref_id'");
 796                                         } */
 797
 798                                         // check for user post link to main table
 799
 800                                         // do we allow duplicate posts with same GUID in different feeds?
 801                                         if (get_pref("ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
 802                                                 $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
 803                                         } else {
 804                                                 $dupcheck_qpart = "";
 805                                         }
 806
 807                                         /* Collect article tags here so we could filter by them: */
 808
 809                                         $article_filters = get_article_filters($filters, $entry_title,
 810                                                 $entry_content, $entry_link, $entry_timestamp, $entry_author,
 811                                                 $entry_tags);
 812
 813                                         if ($debug_enabled) {
 814                                                 _debug("update_rss_feed: article filters: ");
 815                                                 if (count($article_filters) != 0) {
 816                                                         print_r($article_filters);
 817                                                 }
 818                                         }
 819
 820                                         if (find_article_filter($article_filters, "filter")) {
 821                                                 db_query("COMMIT"); // close transaction in progress
 822                                                 continue;
 823                                         }
 824
 825                                         $score = calculate_article_score($article_filters);
 826
 827                                         if ($debug_enabled) {
 828                                                 _debug("update_rss_feed: initial score: $score");
 829                                         }
 830
 831                                         $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
 832                                                         ref_id = '$ref_id' AND owner_uid = '$owner_uid'
 833                                                         $dupcheck_qpart";
 834
 835 //                                      if ($_REQUEST["xdebug"]) print "$query\n";
 836
 837                                         $result = db_query($query);
 838
 839                                         // okay it doesn't exist - create user entry
 840                                         if (db_num_rows($result) == 0) {
 841
 842                                                 if ($debug_enabled) {
 843                                                         _debug("update_rss_feed: user record not found, creating...");
 844                                                 }
 845
 846                                                 if ($score >= -500 && !find_article_filter($article_filters, 'catchup')) {
 847                                                         $unread = 'true';
 848                                                         $last_read_qpart = 'NULL';
 849                                                 } else {
 850                                                         $unread = 'false';
 851                                                         $last_read_qpart = 'NOW()';
 852                                                 }
 853
 854                                                 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
 855                                                         $marked = 'true';
 856                                                 } else {
 857                                                         $marked = 'false';
 858                                                 }
 859
 860                                                 if (find_article_filter($article_filters, 'publish')) {
 861                                                         $published = 'true';
 862                                                 } else {
 863                                                         $published = 'false';
 864                                                 }
 865
 866                                                 // N-grams
 867
 868                                                 if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
 869
 870                                                         $result = db_query("SELECT COUNT(*) AS similar FROM
 871                                                                         ttrss_entries,ttrss_user_entries
 872                                                                 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
 873                                                                         AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
 874                                                                         AND owner_uid = $owner_uid");
 875
 876                                                         $ngram_similar = db_fetch_result($result, 0, "similar");
 877
 878                                                         if ($debug_enabled) {
 879                                                                 _debug("update_rss_feed: N-gram similar results: $ngram_similar");
 880                                                         }
 881
 882                                                         if ($ngram_similar > 0) {
 883                                                                 $unread = 'false';
 884                                                         }
 885                                                 }
 886
 887                                                 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
 888                                                 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
 889
 890                                                 $result = db_query(
 891                                                         "INSERT INTO ttrss_user_entries
 892                                                                 (ref_id, owner_uid, feed_id, unread, last_read, marked,
 893                                                                 published, score, tag_cache, label_cache, uuid,
 894                                                                 last_marked, last_published)
 895                                                         VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
 896                                                                 $last_read_qpart, $marked, $published, '$score', '', '',
 897                                                                 '', $last_marked, $last_published)");
 898
 899                                                 if (PUBSUBHUBBUB_HUB && $published == 'true') {
 900                                                         $rss_link = get_self_url_prefix() .
 901                                                                 "/public.php?op=rss&id=-2&key=" .
 902                                                                 get_feed_access_key(-2, false, $owner_uid);
 903
 904                                                         $p = new Publisher(PUBSUBHUBBUB_HUB);
 905
 906                                                         $pubsub_result = $p->publish_update($rss_link);
 907                                                 }
 908
 909                                                 $result = db_query(
 910                                                         "SELECT int_id FROM ttrss_user_entries WHERE
 911                                                                 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
 912                                                                 feed_id = '$feed' LIMIT 1");
 913
 914                                                 if (db_num_rows($result) == 1) {
 915                                                         $entry_int_id = db_fetch_result($result, 0, "int_id");
 916                                                 }
 917                                         } else {
 918                                                 if ($debug_enabled) {
 919                                                         _debug("update_rss_feed: user record FOUND");
 920                                                 }
 921
 922                                                 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
 923                                                 $entry_int_id = db_fetch_result($result, 0, "int_id");
 924                                         }
 925
 926                                         if ($debug_enabled) {
 927                                                 _debug("update_rss_feed: RID: $entry_ref_id, IID: $entry_int_id");
 928                                         }
 929
 930                                         $post_needs_update = false;
 931                                         $update_insignificant = false;
 932
 933                                         if ($orig_num_comments != $num_comments) {
 934                                                 $post_needs_update = true;
 935                                                 $update_insignificant = true;
 936                                         }
 937
 938                                         if ($entry_plugin_data != $orig_plugin_data) {
 939                                                 $post_needs_update = true;
 940                                                 $update_insignificant = true;
 941                                         }
 942
 943                                         if ($content_hash != $orig_content_hash) {
 944                                                 $post_needs_update = true;
 945                                                 $update_insignificant = false;
 946                                         }
 947
 948                                         if (db_escape_string($orig_title) != $entry_title) {
 949                                                 $post_needs_update = true;
 950                                                 $update_insignificant = false;
 951                                         }
 952
 953                                         // if post needs update, update it and mark all user entries
 954                                         // linking to this post as updated
 955                                         if ($post_needs_update) {
 956
 957                                                 if (defined('DAEMON_EXTENDED_DEBUG')) {
 958                                                         _debug("update_rss_feed: post $entry_guid_hashed needs update...");
 959                                                 }
 960
 961 //                                              print "<!-- post $orig_title needs update : $post_needs_update -->";
 962
 963                                                 db_query("UPDATE ttrss_entries
 964                                                         SET title = '$entry_title', content = '$entry_content',
 965                                                                 content_hash = '$content_hash',
 966                                                                 updated = '$entry_timestamp_fmt',
 967                                                                 num_comments = '$num_comments',
 968                                                                 plugin_data = '$entry_plugin_data'
 969                                                         WHERE id = '$ref_id'");
 970
 971                                                 if (!$update_insignificant) {
 972                                                         if ($mark_unread_on_update) {
 973                                                                 db_query("UPDATE ttrss_user_entries
 974                                                                         SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
 975                                                         }
 976                                                 }
 977                                         }
 978                                 }
 979
 980                                 db_query("COMMIT");
 981
 982                                 if ($debug_enabled) {
 983                                         _debug("update_rss_feed: assigning labels...");
 984                                 }
 985
 986                                 assign_article_to_label_filters($entry_ref_id, $article_filters,
 987                                         $owner_uid, $article_labels);
 988
 989                                 if ($debug_enabled) {
 990                                         _debug("update_rss_feed: looking for enclosures...");
 991                                 }
 992
 993                                 // enclosures
 994
 995                                 $enclosures = array();
 996
 997                                 $encs = $item->get_enclosures();
 998
 999                                 if (is_array($encs)) {
1000                                         foreach ($encs as $e) {
1001                                                 $e_item = array(
1002                                                         $e->link, $e->type, $e->length);
1003                                                 array_push($enclosures, $e_item);
1004                                         }
1005                                 }
1006
1007                                 if ($debug_enabled) {
1008                                         _debug("update_rss_feed: article enclosures:");
1009                                         print_r($enclosures);
1010                                 }
1011
1012                                 db_query("BEGIN");
1013
1014                                 foreach ($enclosures as $enc) {
1015                                         $enc_url = db_escape_string($enc[0]);
1016                                         $enc_type = db_escape_string($enc[1]);
1017                                         $enc_dur = db_escape_string($enc[2]);
1018
1019                                         $result = db_query("SELECT id FROM ttrss_enclosures
1020                                                 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
1021
1022                                         if (db_num_rows($result) == 0) {
1023                                                 db_query("INSERT INTO ttrss_enclosures
1024                                                         (content_url, content_type, title, duration, post_id) VALUES
1025                                                         ('$enc_url', '$enc_type', '', '$enc_dur', '$entry_ref_id')");
1026                                         }
1027                                 }
1028
1029                                 db_query("COMMIT");
1030
1031                                 // check for manual tags (we have to do it here since they're loaded from filters)
1032
1033                                 foreach ($article_filters as $f) {
1034                                         if ($f["type"] == "tag") {
1035
1036                                                 $manual_tags = trim_array(explode(",", $f["param"]));
1037
1038                                                 foreach ($manual_tags as $tag) {
1039                                                         if (tag_is_valid($tag)) {
1040                                                                 array_push($entry_tags, $tag);
1041                                                         }
1042                                                 }
1043                                         }
1044                                 }
1045
1046                                 // Skip boring tags
1047
1048                                 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
1049                                         'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1050
1051                                 $filtered_tags = array();
1052                                 $tags_to_cache = array();
1053
1054                                 if ($entry_tags && is_array($entry_tags)) {
1055                                         foreach ($entry_tags as $tag) {
1056                                                 if (array_search($tag, $boring_tags) === false) {
1057                                                         array_push($filtered_tags, $tag);
1058                                                 }
1059                                         }
1060                                 }
1061
1062                                 $filtered_tags = array_unique($filtered_tags);
1063
1064                                 if ($debug_enabled) {
1065                                         _debug("update_rss_feed: filtered article tags:");
1066                                         print_r($filtered_tags);
1067                                 }
1068
1069                                 // Save article tags in the database
1070
1071                                 if (count($filtered_tags) > 0) {
1072
1073                                         db_query("BEGIN");
1074
1075                                         foreach ($filtered_tags as $tag) {
1076
1077                                                 $tag = sanitize_tag($tag);
1078                                                 $tag = db_escape_string($tag);
1079
1080                                                 if (!tag_is_valid($tag)) continue;
1081
1082                                                 $result = db_query("SELECT id FROM ttrss_tags
1083                                                         WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1084                                                         owner_uid = '$owner_uid' LIMIT 1");
1085
1086                                                         if ($result && db_num_rows($result) == 0) {
1087
1088                                                                 db_query("INSERT INTO ttrss_tags
1089                                                                         (owner_uid,tag_name,post_int_id)
1090                                                                         VALUES ('$owner_uid','$tag', '$entry_int_id')");
1091                                                         }
1092
1093                                                 array_push($tags_to_cache, $tag);
1094                                         }
1095
1096                                         /* update the cache */
1097
1098                                         $tags_to_cache = array_unique($tags_to_cache);
1099
1100                                         $tags_str = db_escape_string(join(",", $tags_to_cache));
1101
1102                                         db_query("UPDATE ttrss_user_entries
1103                                                 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1104                                                 AND owner_uid = $owner_uid");
1105
1106                                         db_query("COMMIT");
1107                                 }
1108
1109                                 if (get_pref("AUTO_ASSIGN_LABELS", $owner_uid, false)) {
1110                                         if ($debug_enabled) {
1111                                                 _debug("update_rss_feed: auto-assigning labels...");
1112                                         }
1113
1114                                         foreach ($labels as $label) {
1115                                                 $caption = preg_quote($label["caption"]);
1116
1117                                                 if ($caption && preg_match("/\b$caption\b/i", "$tags_str " . strip_tags($entry_content) . " $entry_title")) {
1118                                                         if (!labels_contains_caption($article_labels, $caption)) {
1119                                                                 label_add_article($entry_ref_id, $caption, $owner_uid);
1120                                                         }
1121                                                 }
1122                                         }
1123                                 }
1124
1125                                 if ($debug_enabled) {
1126                                         _debug("update_rss_feed: article processed");
1127                                 }
1128                         }
1129
1130                         if (!$last_updated) {
1131                                 if ($debug_enabled) {
1132                                         _debug("update_rss_feed: new feed, catching it up...");
1133                                 }
1134                                 catchup_feed($feed, false, $owner_uid);
1135                         }
1136
1137                         if ($debug_enabled) {
1138                                 _debug("purging feed...");
1139                         }
1140
1141                         purge_feed($feed, 0, $debug_enabled);
1142
1143                         db_query("UPDATE ttrss_feeds
1144                                 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
1145
1146 //                      db_query("COMMIT");
1147
1148                 } else {
1149
1150                         $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
1151
1152                         if ($debug_enabled) {
1153                                 _debug("update_rss_feed: error fetching feed: $error_msg");
1154                         }
1155
1156                         db_query(
1157                                 "UPDATE ttrss_feeds SET last_error = '$error_msg',
1158                                         last_updated = NOW() WHERE id = '$feed'");
1159                 }
1160
1161                 unset($rss);
1162
1163                 if ($debug_enabled) {
1164                         _debug("update_rss_feed: done");
1165                 }
1166
1167         }
1168
1169         function cache_images($html, $site_url, $debug) {
1170                 $cache_dir = CACHE_DIR . "/images";
1171
1172                 libxml_use_internal_errors(true);
1173
1174                 $charset_hack = '<head>
1175                         <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1176                 </head>';
1177
1178                 $doc = new DOMDocument();
1179                 $doc->loadHTML($charset_hack . $html);
1180                 $xpath = new DOMXPath($doc);
1181
1182                 $entries = $xpath->query('(//img[@src])');
1183
1184                 foreach ($entries as $entry) {
1185                         if ($entry->hasAttribute('src')) {
1186                                 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1187
1188                                 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1189
1190                                 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1191
1192                                 if (!file_exists($local_filename)) {
1193                                         $file_content = fetch_file_contents($src);
1194
1195                                         if ($file_content && strlen($file_content) > 1024) {
1196                                                 file_put_contents($local_filename, $file_content);
1197                                         }
1198                                 }
1199
1200                                 if (file_exists($local_filename)) {
1201                                         $entry->setAttribute('src', SELF_URL_PATH . '/image.php?url=' .
1202                                                 base64_encode($src));
1203                                 }
1204                         }
1205                 }
1206
1207                 $node = $doc->getElementsByTagName('body')->item(0);
1208
1209                 return $doc->saveXML($node);
1210         }
1211
1212         function expire_error_log($debug) {
1213                 if ($debug) _debug("Removing old error log entries...");
1214
1215                 if (DB_TYPE == "pgsql") {
1216                         db_query("DELETE FROM ttrss_error_log
1217                                 WHERE created_at < NOW() - INTERVAL '7 days'");
1218                 } else {
1219                         db_query("DELETE FROM ttrss_error_log
1220                                 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1221                 }
1222
1223         }
1224
1225         function expire_lock_files($debug) {
1226                 if ($debug) _debug("Removing old lock files...");
1227
1228                 $num_deleted = 0;
1229
1230                 if (is_writable(LOCK_DIRECTORY)) {
1231                         $files = glob(LOCK_DIRECTORY . "/*.lock");
1232
1233                         if ($files) {
1234                                 foreach ($files as $file) {
1235                                         if (!file_is_locked($file) && time() - filemtime($file) > 86400*2) {
1236                                                 unlink($file);
1237                                                 ++$num_deleted;
1238                                         }
1239                                 }
1240                         }
1241                 }
1242
1243                 if ($debug) _debug("Removed $num_deleted files.");
1244         }
1245
1246         function expire_cached_files($debug) {
1247                 foreach (array("simplepie", "images", "export", "upload") as $dir) {
1248                         $cache_dir = CACHE_DIR . "/$dir";
1249
1250                         if ($debug) _debug("Expiring $cache_dir");
1251
1252                         $num_deleted = 0;
1253
1254                         if (is_writable($cache_dir)) {
1255                                 $files = glob("$cache_dir/*");
1256
1257                                 if ($files) {
1258                                         foreach ($files as $file) {
1259                                                 if (time() - filemtime($file) > 86400*7) {
1260                                                         unlink($file);
1261
1262                                                         ++$num_deleted;
1263                                                 }
1264                                         }
1265                                 }
1266                         }
1267
1268                         if ($debug) _debug("Removed $num_deleted files.");
1269                 }
1270         }
1271
1272         /**
1273         * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1274         * Returns the url query as associative array
1275         *
1276         * @param    string    query
1277         * @return    array    params
1278         */
1279         function convertUrlQuery($query) {
1280                 $queryParts = explode('&', $query);
1281
1282                 $params = array();
1283
1284                 foreach ($queryParts as $param) {
1285                         $item = explode('=', $param);
1286                         $params[$item[0]] = $item[1];
1287                 }
1288
1289                 return $params;
1290         }
1291
1292         function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
1293                 $matches = array();
1294
1295                 foreach ($filters as $filter) {
1296                         $match_any_rule = $filter["match_any_rule"];
1297                         $inverse = $filter["inverse"];
1298                         $filter_match = false;
1299
1300                         foreach ($filter["rules"] as $rule) {
1301                                 $match = false;
1302                                 $reg_exp = $rule["reg_exp"];
1303                                 $rule_inverse = $rule["inverse"];
1304
1305                                 if (!$reg_exp)
1306                                         continue;
1307
1308                                 switch ($rule["type"]) {
1309                                 case "title":
1310                                         $match = @preg_match("/$reg_exp/i", $title);
1311                                         break;
1312                                 case "content":
1313                                         // we don't need to deal with multiline regexps
1314                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1315
1316                                         $match = @preg_match("/$reg_exp/i", $content);
1317                                         break;
1318                                 case "both":
1319                                         // we don't need to deal with multiline regexps
1320                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1321
1322                                         $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
1323                                         break;
1324                                 case "link":
1325                                         $match = @preg_match("/$reg_exp/i", $link);
1326                                         break;
1327                                 case "author":
1328                                         $match = @preg_match("/$reg_exp/i", $author);
1329                                         break;
1330                                 case "tag":
1331                                         $tag_string = join(",", $tags);
1332                                         $match = @preg_match("/$reg_exp/i", $tag_string);
1333                                         break;
1334                                 }
1335
1336                                 if ($rule_inverse) $match = !$match;
1337
1338                                 if ($match_any_rule) {
1339                                         if ($match) {
1340                                                 $filter_match = true;
1341                                                 break;
1342                                         }
1343                                 } else {
1344                                         $filter_match = $match;
1345                                         if (!$match) {
1346                                                 break;
1347                                         }
1348                                 }
1349                         }
1350
1351                         if ($inverse) $filter_match = !$filter_match;
1352
1353                         if ($filter_match) {
1354                                 foreach ($filter["actions"] AS $action) {
1355                                         array_push($matches, $action);
1356
1357                                         // if Stop action encountered, perform no further processing
1358                                         if ($action["type"] == "stop") return $matches;
1359                                 }
1360                         }
1361                 }
1362
1363                 return $matches;
1364         }
1365
1366         function find_article_filter($filters, $filter_name) {
1367                 foreach ($filters as $f) {
1368                         if ($f["type"] == $filter_name) {
1369                                 return $f;
1370                         };
1371                 }
1372                 return false;
1373         }
1374
1375         function find_article_filters($filters, $filter_name) {
1376                 $results = array();
1377
1378                 foreach ($filters as $f) {
1379                         if ($f["type"] == $filter_name) {
1380                                 array_push($results, $f);
1381                         };
1382                 }
1383                 return $results;
1384         }
1385
1386         function calculate_article_score($filters) {
1387                 $score = 0;
1388
1389                 foreach ($filters as $f) {
1390                         if ($f["type"] == "score") {
1391                                 $score += $f["param"];
1392                         };
1393                 }
1394                 return $score;
1395         }
1396
1397         function labels_contains_caption($labels, $caption) {
1398                 foreach ($labels as $label) {
1399                         if ($label[1] == $caption) {
1400                                 return true;
1401                         }
1402                 }
1403
1404                 return false;
1405         }
1406
1407         function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
1408                 foreach ($filters as $f) {
1409                         if ($f["type"] == "label") {
1410                                 if (!labels_contains_caption($article_labels, $f["param"])) {
1411                                         label_add_article($id, $f["param"], $owner_uid);
1412                                 }
1413                         }
1414                 }
1415         }
1416
1417         function make_guid_from_title($title) {
1418                 return preg_replace("/[ \"\',.:;]/", "-",
1419                         mb_strtolower(strip_tags($title), 'utf-8'));
1420         }
1421
1422
1423 ?>