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         define_default('_MIN_CACHE_IMAGE_SIZE', 1024);
   6
   7         function calculate_article_hash($article, $pluginhost) {
   8                 $tmp = "";
   9
  10                 foreach ($article as $k => $v) {
  11                         if ($k != "feed" && isset($v)) {
  12                                 $x = strip_tags(is_array($v) ? implode(",", $v) : $v);
  13
  14                                 //_debug("$k:" . sha1($x) . ":" . htmlspecialchars($x), true);
  15
  16                                 $tmp .= sha1("$k:" . sha1($x));
  17                         }
  18                 }
  19
  20                 return sha1(implode(",", $pluginhost->get_plugin_names()) . $tmp);
  21         }
  22
  23         function update_feedbrowser_cache() {
  24
  25                 $result = db_query("SELECT feed_url, site_url, title, COUNT(id) AS subscribers
  26                         FROM ttrss_feeds WHERE (SELECT COUNT(id) = 0 FROM ttrss_feeds AS tf
  27                                 WHERE tf.feed_url = ttrss_feeds.feed_url
  28                                 AND (private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%'))
  29                                 GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
  30
  31                 db_query("BEGIN");
  32
  33                 db_query("DELETE FROM ttrss_feedbrowser_cache");
  34
  35                 $count = 0;
  36
  37                 while ($line = db_fetch_assoc($result)) {
  38                         $subscribers = db_escape_string($line["subscribers"]);
  39                         $feed_url = db_escape_string($line["feed_url"]);
  40                         $title = db_escape_string($line["title"]);
  41                         $site_url = db_escape_string($line["site_url"]);
  42
  43                         $tmp_result = db_query("SELECT subscribers FROM
  44                                 ttrss_feedbrowser_cache WHERE feed_url = '$feed_url'");
  45
  46                         if (db_num_rows($tmp_result) == 0) {
  47
  48                                 db_query("INSERT INTO ttrss_feedbrowser_cache
  49                                         (feed_url, site_url, title, subscribers) VALUES ('$feed_url',
  50                                                 '$site_url', '$title', '$subscribers')");
  51
  52                                 ++$count;
  53
  54                         }
  55
  56                 }
  57
  58                 db_query("COMMIT");
  59
  60                 return $count;
  61
  62         }
  63
  64
  65         /**
  66          * Update a feed batch.
  67          * Used by daemons to update n feeds by run.
  68          * Only update feed needing a update, and not being processed
  69          * by another process.
  70          *
  71          * @param mixed $link Database link
  72          * @param integer $limit Maximum number of feeds in update batch. Default to DAEMON_FEED_LIMIT.
  73          * @param boolean $from_http Set to true if you call this function from http to disable cli specific code.
  74          * @param boolean $debug Set to false to disable debug output. Default to true.
  75          * @return void
  76          */
  77         function update_daemon_common($limit = DAEMON_FEED_LIMIT, $from_http = false, $debug = true) {
  78                 // Process all other feeds using last_updated and interval parameters
  79
  80                 $schema_version = get_schema_version();
  81
  82                 if ($schema_version != SCHEMA_VERSION) {
  83                         die("Schema version is wrong, please upgrade the database.\n");
  84                 }
  85
  86                 define('PREFS_NO_CACHE', true);
  87
  88                 // Test if the user has loggued in recently. If not, it does not update its feeds.
  89                 if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
  90                         if (DB_TYPE == "pgsql") {
  91                                 $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
  92                         } else {
  93                                 $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
  94                         }
  95                 } else {
  96                         $login_thresh_qpart = "";
  97                 }
  98
  99                 // Test if the feed need a update (update interval exceeded).
 100                 if (DB_TYPE == "pgsql") {
 101                         $update_limit_qpart = "AND ((
 102                                         ttrss_feeds.update_interval = 0
 103                                         AND ttrss_user_prefs.value != '-1'
 104                                         AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
 105                                 ) OR (
 106                                         ttrss_feeds.update_interval > 0
 107                                         AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
 108                                 ) OR (ttrss_feeds.last_updated IS NULL
 109                                         AND ttrss_user_prefs.value != '-1')
 110                                 OR (last_updated = '1970-01-01 00:00:00'
 111                                         AND ttrss_user_prefs.value != '-1'))";
 112                 } else {
 113                         $update_limit_qpart = "AND ((
 114                                         ttrss_feeds.update_interval = 0
 115                                         AND ttrss_user_prefs.value != '-1'
 116                                         AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
 117                                 ) OR (
 118                                         ttrss_feeds.update_interval > 0
 119                                         AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
 120                                 ) OR (ttrss_feeds.last_updated IS NULL
 121                                         AND ttrss_user_prefs.value != '-1')
 122                                 OR (last_updated = '1970-01-01 00:00:00'
 123                                         AND ttrss_user_prefs.value != '-1'))";
 124                 }
 125
 126                 // Test if feed is currently being updated by another process.
 127                 if (DB_TYPE == "pgsql") {
 128                         $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
 129                 } else {
 130                         $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 10 MINUTE))";
 131                 }
 132
 133                 // Test if there is a limit to number of updated feeds
 134                 $query_limit = "";
 135                 if($limit) $query_limit = sprintf("LIMIT %d", $limit);
 136
 137                 // Update the least recently updated feeds first
 138                 $query_order = "ORDER BY last_updated";
 139                 if (DB_TYPE == "pgsql") $query_order .= " NULLS FIRST";
 140
 141                 $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
 142                         FROM
 143                                 ttrss_feeds, ttrss_users, ttrss_user_prefs
 144                         WHERE
 145                                 ttrss_feeds.owner_uid = ttrss_users.id
 146                                 AND ttrss_user_prefs.profile IS NULL
 147                                 AND ttrss_users.id = ttrss_user_prefs.owner_uid
 148                                 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
 149                                 $login_thresh_qpart $update_limit_qpart
 150                                 $updstart_thresh_qpart
 151                                 $query_order $query_limit";
 152
 153                 // We search for feed needing update.
 154                 $result = db_query($query);
 155
 156                 if($debug) _debug(sprintf("Scheduled %d feeds to update...", db_num_rows($result)));
 157
 158                 // Here is a little cache magic in order to minimize risk of double feed updates.
 159                 $feeds_to_update = array();
 160                 while ($line = db_fetch_assoc($result)) {
 161                         array_push($feeds_to_update, db_escape_string($line['feed_url']));
 162                 }
 163
 164                 // We update the feed last update started date before anything else.
 165                 // There is no lag due to feed contents downloads
 166                 // It prevent an other process to update the same feed.
 167
 168                 if(count($feeds_to_update) > 0) {
 169                         $feeds_quoted = array();
 170
 171                         foreach ($feeds_to_update as $feed) {
 172                                 array_push($feeds_quoted, "'" . db_escape_string($feed) . "'");
 173                         }
 174
 175                         db_query(sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
 176                                 WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
 177                 }
 178
 179                 $nf = 0;
 180                 $bstarted = microtime(true);
 181
 182                 $batch_owners = array();
 183
 184                 // For each feed, we call the feed update function.
 185                 foreach ($feeds_to_update as $feed) {
 186                         if($debug) _debug("Base feed: $feed");
 187
 188                         //update_rss_feed($line["id"], true);
 189
 190                         // since we have the data cached, we can deal with other feeds with the same url
 191
 192                         $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
 193                         FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
 194                                 ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
 195                                 ttrss_users.id = ttrss_user_prefs.owner_uid AND
 196                                 ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
 197                                 ttrss_user_prefs.profile IS NULL AND
 198                                 feed_url = '".db_escape_string($feed)."'
 199                                 $update_limit_qpart
 200                                 $login_thresh_qpart
 201                         ORDER BY ttrss_feeds.id $query_limit");
 202
 203                         if (db_num_rows($tmp_result) > 0) {
 204                                 $rss = false;
 205
 206                                 while ($tline = db_fetch_assoc($tmp_result)) {
 207                                         if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
 208
 209                                         if (array_search($tline["owner_uid"], $batch_owners) === FALSE)
 210                                                 array_push($batch_owners, $tline["owner_uid"]);
 211
 212                                         $fstarted = microtime(true);
 213                                         $rss = update_rss_feed($tline["id"], true, false);
 214                                         _debug_suppress(false);
 215
 216                                         _debug(sprintf("    %.4f (sec)", microtime(true) - $fstarted));
 217
 218                                         ++$nf;
 219                                 }
 220                         }
 221                 }
 222
 223                 if ($nf > 0) {
 224                         _debug(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
 225                                 microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
 226                 }
 227
 228                 foreach ($batch_owners as $owner_uid) {
 229                         _debug("Running housekeeping tasks for user $owner_uid...");
 230
 231                         housekeeping_user($owner_uid);
 232                 }
 233
 234                 require_once "digest.php";
 235
 236                 // Send feed digests by email if needed.
 237                 send_headlines_digests($debug);
 238
 239                 return $nf;
 240
 241         } // function update_daemon_common
 242
 243         // this is used when subscribing
 244         function set_basic_feed_info($feed) {
 245
 246                 $feed = db_escape_string($feed);
 247
 248                 $result = db_query("SELECT feed_url,auth_pass,auth_login,auth_pass_encrypted
 249                                         FROM ttrss_feeds WHERE id = '$feed'");
 250
 251                 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
 252                         0, "auth_pass_encrypted"));
 253
 254                 $auth_login = db_fetch_result($result, 0, "auth_login");
 255                 $auth_pass = db_fetch_result($result, 0, "auth_pass");
 256
 257                 if ($auth_pass_encrypted) {
 258                         require_once "crypt.php";
 259                         $auth_pass = decrypt_string($auth_pass);
 260                 }
 261
 262                 $fetch_url = db_fetch_result($result, 0, "feed_url");
 263
 264                 $feed_data = fetch_file_contents($fetch_url, false,
 265                         $auth_login, $auth_pass, false,
 266                         FEED_FETCH_TIMEOUT_TIMEOUT,
 267                         0);
 268
 269                 global $fetch_curl_used;
 270
 271                 if (!$fetch_curl_used) {
 272                         $tmp = @gzdecode($feed_data);
 273
 274                         if ($tmp) $feed_data = $tmp;
 275                 }
 276
 277                 $feed_data = trim($feed_data);
 278
 279                 $rss = new FeedParser($feed_data);
 280                 $rss->init();
 281
 282                 if (!$rss->error()) {
 283
 284                         $result = db_query("SELECT title, site_url FROM ttrss_feeds WHERE id = '$feed'");
 285
 286                         $registered_title = db_fetch_result($result, 0, "title");
 287                         $orig_site_url = db_fetch_result($result, 0, "site_url");
 288
 289                         $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
 290                         $feed_title = db_escape_string(mb_substr($rss->get_title(), 0, 199));
 291
 292                         if ($feed_title && (!$registered_title || $registered_title == "[Unknown]")) {
 293                                 db_query("UPDATE ttrss_feeds SET
 294                                         title = '$feed_title' WHERE id = '$feed'");
 295                         }
 296
 297                         if ($site_url && $orig_site_url != $site_url) {
 298                                 db_query("UPDATE ttrss_feeds SET
 299                                                         site_url = '$site_url' WHERE id = '$feed'");
 300                         }
 301                 }
 302         }
 303
 304         // ignore_daemon is not used
 305         function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
 306
 307                 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
 308
 309                 _debug_suppress(!$debug_enabled);
 310                 _debug("start", $debug_enabled);
 311
 312                 $result = db_query("SELECT title FROM ttrss_feeds
 313                         WHERE id = '$feed'");
 314                 $title = db_fetch_result($result, 0, "title");
 315
 316                 // feed was batch-subscribed or something, we need to get basic info
 317                 // this is not optimal currently as it fetches stuff separately TODO: optimize
 318                 if ($title == "[Unknown]") {
 319                         _debug("setting basic feed info for $feed...");
 320                         set_basic_feed_info($feed);
 321                 }
 322
 323                 $result = db_query("SELECT id,update_interval,auth_login,
 324                         feed_url,auth_pass,cache_images,
 325                         mark_unread_on_update, owner_uid,
 326                         pubsub_state, auth_pass_encrypted,
 327                         feed_language
 328                         FROM ttrss_feeds WHERE id = '$feed'");
 329
 330                 if (db_num_rows($result) == 0) {
 331                         _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
 332                         user_error("Attempt to update unknown/invalid feed $feed", E_USER_WARNING);
 333                         return false;
 334                 }
 335
 336                 $owner_uid = db_fetch_result($result, 0, "owner_uid");
 337                 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
 338                         0, "mark_unread_on_update"));
 339                 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
 340                 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
 341                         0, "auth_pass_encrypted"));
 342
 343                 db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
 344                         WHERE id = '$feed'");
 345
 346                 $auth_login = db_fetch_result($result, 0, "auth_login");
 347                 $auth_pass = db_fetch_result($result, 0, "auth_pass");
 348
 349                 if ($auth_pass_encrypted) {
 350                         require_once "crypt.php";
 351                         $auth_pass = decrypt_string($auth_pass);
 352                 }
 353
 354                 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
 355                 $fetch_url = db_fetch_result($result, 0, "feed_url");
 356                 $feed_language = db_escape_string(mb_strtolower(db_fetch_result($result, 0, "feed_language")));
 357                 if (!$feed_language) $feed_language = 'english';
 358
 359                 $feed = db_escape_string($feed);
 360
 361                 $date_feed_processed = date('Y-m-d H:i');
 362
 363                 $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
 364
 365                 $pluginhost = new PluginHost();
 366                 $pluginhost->set_debug($debug_enabled);
 367                 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
 368
 369                 $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
 370                 $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
 371                 $pluginhost->load_data();
 372
 373                 if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
 374                         _debug("using previously initialized parser object");
 375                 } else {
 376                         $rss_hash = false;
 377
 378                         $force_refetch = isset($_REQUEST["force_refetch"]);
 379                         $feed_data = "";
 380
 381                         foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
 382                                 $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, 0, $auth_login, $auth_pass);
 383                         }
 384
 385                         // try cache
 386                         if (!$feed_data &&
 387                                 file_exists($cache_filename) &&
 388                                 is_readable($cache_filename) &&
 389                                 !$auth_login && !$auth_pass &&
 390                                 filemtime($cache_filename) > time() - 30) {
 391
 392                                 _debug("using local cache [$cache_filename].", $debug_enabled);
 393
 394                                 @$feed_data = file_get_contents($cache_filename);
 395
 396                                 if ($feed_data) {
 397                                         $rss_hash = sha1($feed_data);
 398                                 }
 399
 400                         } else {
 401                                 _debug("local cache will not be used for this feed", $debug_enabled);
 402                         }
 403
 404                         // fetch feed from source
 405                         if (!$feed_data) {
 406                                 _debug("fetching [$fetch_url]...", $debug_enabled);
 407
 408                                 if (ini_get("open_basedir") && function_exists("curl_init")) {
 409                                         _debug("not using CURL due to open_basedir restrictions");
 410                                 }
 411
 412                                 $feed_data = fetch_file_contents($fetch_url, false,
 413                                         $auth_login, $auth_pass, false,
 414                                         $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
 415                                         0);
 416
 417                                 global $fetch_curl_used;
 418
 419                                 if (!$fetch_curl_used) {
 420                                         $tmp = @gzdecode($feed_data);
 421
 422                                         if ($tmp) $feed_data = $tmp;
 423                                 }
 424
 425                                 $feed_data = trim($feed_data);
 426
 427                                 _debug("fetch done.", $debug_enabled);
 428
 429                                 // cache vanilla feed data for re-use
 430                                 if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
 431                                         $new_rss_hash = sha1($feed_data);
 432
 433                                         if ($new_rss_hash != $rss_hash) {
 434                                                 _debug("saving $cache_filename", $debug_enabled);
 435                                                 @file_put_contents($cache_filename, $feed_data);
 436                                         }
 437                                 }
 438                         }
 439
 440                         if (!$feed_data) {
 441                                 global $fetch_last_error;
 442                                 global $fetch_last_error_code;
 443
 444                                 _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
 445
 446                                 $error_escaped = '';
 447
 448                                 // If-Modified-Since
 449                                 if ($fetch_last_error_code != 304) {
 450                                         $error_escaped = db_escape_string($fetch_last_error);
 451                                 } else {
 452                                         _debug("source claims data not modified, nothing to do.", $debug_enabled);
 453                                 }
 454
 455                                 db_query(
 456                                         "UPDATE ttrss_feeds SET last_error = '$error_escaped',
 457                                                 last_updated = NOW() WHERE id = '$feed'");
 458
 459                                 return;
 460                         }
 461                 }
 462
 463                 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
 464                         $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
 465                 }
 466
 467                 // set last update to now so if anything *simplepie* crashes later we won't be
 468                 // continuously failing on the same feed
 469                 //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
 470
 471                 if (!$rss) {
 472                         $rss = new FeedParser($feed_data);
 473                         $rss->init();
 474                 }
 475
 476 //              print_r($rss);
 477
 478                 $feed = db_escape_string($feed);
 479
 480                 if (!$rss->error()) {
 481
 482                         // We use local pluginhost here because we need to load different per-user feed plugins
 483                         $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
 484
 485                         _debug("language: $feed_language", $debug_enabled);
 486                         _debug("processing feed data...", $debug_enabled);
 487
 488 //                      db_query("BEGIN");
 489
 490                         if (DB_TYPE == "pgsql") {
 491                                 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
 492                         } else {
 493                                 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
 494                         }
 495
 496                         $result = db_query("SELECT owner_uid,favicon_avg_color,
 497                                 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
 498                                                 favicon_needs_check
 499                                 FROM ttrss_feeds WHERE id = '$feed'");
 500
 501                         $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
 502                                 "favicon_needs_check"));
 503                         $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
 504
 505                         $owner_uid = db_fetch_result($result, 0, "owner_uid");
 506
 507                         $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
 508
 509                         _debug("site_url: $site_url", $debug_enabled);
 510                         _debug("feed_title: " . $rss->get_title(), $debug_enabled);
 511
 512                         if ($favicon_needs_check || $force_refetch) {
 513
 514                                 /* terrible hack: if we crash on floicon shit here, we won't check
 515                                  * the icon avgcolor again (unless the icon got updated) */
 516
 517                                 $favicon_file = ICONS_DIR . "/$feed.ico";
 518                                 $favicon_modified = @filemtime($favicon_file);
 519
 520                                 _debug("checking favicon...", $debug_enabled);
 521
 522                                 check_feed_favicon($site_url, $feed);
 523                                 $favicon_modified_new = @filemtime($favicon_file);
 524
 525                                 if ($favicon_modified_new > $favicon_modified)
 526                                         $favicon_avg_color = '';
 527
 528                                 if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
 529                                                 require_once "colors.php";
 530
 531                                                 db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
 532                                                         id = '$feed'");
 533
 534                                                 $favicon_color = db_escape_string(
 535                                                         calculate_avg_color($favicon_file));
 536
 537                                                 $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
 538                                 } else if ($favicon_avg_color == 'fail') {
 539                                         _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
 540                                 }
 541
 542                                 db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
 543                                         $favicon_colorstring
 544                                         WHERE id = '$feed'");
 545                         }
 546
 547                         _debug("loading filters & labels...", $debug_enabled);
 548
 549                         $filters = load_filters($feed, $owner_uid);
 550
 551                         _debug("" . count($filters) . " filters loaded.", $debug_enabled);
 552
 553                         $items = $rss->get_items();
 554
 555                         if (!is_array($items)) {
 556                                 _debug("no articles found.", $debug_enabled);
 557
 558                                 db_query("UPDATE ttrss_feeds
 559                                         SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
 560
 561                                 return; // no articles
 562                         }
 563
 564                         if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
 565
 566                                 _debug("checking for PUSH hub...", $debug_enabled);
 567
 568                                 $feed_hub_url = false;
 569
 570                                 $links = $rss->get_links('hub');
 571
 572                                 if ($links && is_array($links)) {
 573                                         foreach ($links as $l) {
 574                                                 $feed_hub_url = $l;
 575                                                 break;
 576                                         }
 577                                 }
 578
 579                                 _debug("feed hub url: $feed_hub_url", $debug_enabled);
 580
 581                                 $feed_self_url = $fetch_url;
 582
 583                                 $links = $rss->get_links('self');
 584
 585                                 if ($links && is_array($links)) {
 586                                         foreach ($links as $l) {
 587                                                 $feed_self_url = $l;
 588                                                 break;
 589                                         }
 590                                 }
 591
 592                                 _debug("feed self url = $feed_self_url");
 593
 594                                 if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
 595                                         !ini_get("open_basedir")) {
 596
 597                                         require_once 'lib/pubsubhubbub/subscriber.php';
 598
 599                                         $callback_url = get_self_url_prefix() .
 600                                                 "/public.php?op=pubsub&id=$feed";
 601
 602                                         $s = new Subscriber($feed_hub_url, $callback_url);
 603
 604                                         $rc = $s->subscribe($feed_self_url);
 605
 606                                         _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
 607
 608                                         db_query("UPDATE ttrss_feeds SET pubsub_state = 1
 609                                                 WHERE id = '$feed'");
 610                                 }
 611                         }
 612
 613                         _debug("processing articles...", $debug_enabled);
 614
 615                         $tstart = time();
 616
 617                         foreach ($items as $item) {
 618                                 if ($_REQUEST['xdebug'] == 3) {
 619                                         print_r($item);
 620                                 }
 621
 622                                 if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
 623                                         _debug("looks like there's too many articles to process at once, breaking out", $debug_enabled);
 624                                         break;
 625                                 }
 626
 627                                 $entry_guid = $item->get_id();
 628                                 if (!$entry_guid) $entry_guid = $item->get_link();
 629                                 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
 630                                 if (!$entry_guid) continue;
 631
 632                                 $entry_guid = "$owner_uid,$entry_guid";
 633
 634                                 $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
 635
 636                                 _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
 637
 638                                 $entry_timestamp = "";
 639
 640                                 $entry_timestamp = $item->get_date();
 641
 642                                 _debug("orig date: " . $item->get_date(), $debug_enabled);
 643
 644                                 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
 645                                         $entry_timestamp = time();
 646                                 }
 647
 648                                 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
 649
 650                                 _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
 651
 652 //                              $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
 653 //                              $entry_title = decode_numeric_entities($entry_title);
 654                                 $entry_title = $item->get_title();
 655
 656                                 $entry_link = rewrite_relative_url($site_url, $item->get_link());
 657
 658                                 _debug("title $entry_title", $debug_enabled);
 659                                 _debug("link $entry_link", $debug_enabled);
 660
 661                                 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
 662
 663                                 $entry_content = $item->get_content();
 664                                 if (!$entry_content) $entry_content = $item->get_description();
 665
 666                                 if ($_REQUEST["xdebug"] == 2) {
 667                                         print "content: ";
 668                                         print htmlspecialchars($entry_content);
 669                                         print "\n";
 670                                 }
 671
 672                                 $entry_comments = $item->get_comments_url();
 673                                 $entry_author = $item->get_author();
 674
 675                                 $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
 676
 677                                 $entry_comments = db_escape_string(mb_substr(trim($entry_comments), 0, 245));
 678                                 $entry_author = db_escape_string(mb_substr(trim($entry_author), 0, 245));
 679
 680                                 $num_comments = (int) $item->get_comments_count();
 681
 682                                 _debug("author $entry_author", $debug_enabled);
 683                                 _debug("num_comments: $num_comments", $debug_enabled);
 684                                 _debug("looking for tags...", $debug_enabled);
 685
 686                                 // parse <category> entries into tags
 687
 688                                 $additional_tags = array();
 689
 690                                 $additional_tags_src = $item->get_categories();
 691
 692                                 if (is_array($additional_tags_src)) {
 693                                         foreach ($additional_tags_src as $tobj) {
 694                                                 array_push($additional_tags, $tobj);
 695                                         }
 696                                 }
 697
 698                                 $entry_tags = array_unique($additional_tags);
 699
 700                                 for ($i = 0; $i < count($entry_tags); $i++)
 701                                         $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
 702
 703                                 _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
 704
 705                                 _debug("done collecting data.", $debug_enabled);
 706
 707                                 $result = db_query("SELECT id, content_hash, lang FROM ttrss_entries
 708                                         WHERE guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed'");
 709
 710                                 if (db_num_rows($result) != 0) {
 711                                         $base_entry_id = db_fetch_result($result, 0, "id");
 712                                         $entry_stored_hash = db_fetch_result($result, 0, "content_hash");
 713                                         $article_labels = get_article_labels($base_entry_id, $owner_uid);
 714                                         $entry_language = db_fetch_result($result, 0, "lang");
 715
 716                                         $existing_tags = get_article_tags($base_entry_id, $owner_uid);
 717                                         $entry_tags = array_unique(array_merge($entry_tags, $existing_tags));
 718
 719                                 } else {
 720                                         $base_entry_id = false;
 721                                         $entry_stored_hash = "";
 722                                         $article_labels = array();
 723                                         $entry_language = "";
 724                                 }
 725
 726                                 $article = array("owner_uid" => $owner_uid, // read only
 727                                         "guid" => $entry_guid, // read only
 728                                         "guid_hashed" => $entry_guid_hashed, // read only
 729                                         "title" => $entry_title,
 730                                         "content" => $entry_content,
 731                                         "link" => $entry_link,
 732                                         "labels" => $article_labels, // current limitation: can add labels to article, can't remove them
 733                                         "tags" => $entry_tags,
 734                                         "author" => $entry_author,
 735                                         "force_catchup" => false, // ugly hack for the time being
 736                                         "score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
 737                                         "language" => $entry_language,
 738                                         "feed" => array("id" => $feed,
 739                                                 "fetch_url" => $fetch_url,
 740                                                 "site_url" => $site_url,
 741                                                 "cache_images" => $cache_images)
 742                                         );
 743
 744                                 $entry_plugin_data = "";
 745                                 $entry_current_hash = calculate_article_hash($article, $pluginhost);
 746
 747                                 _debug("article hash: $entry_current_hash [stored=$entry_stored_hash]", $debug_enabled);
 748
 749                                 if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
 750                                         _debug("stored article seems up to date [IID: $base_entry_id], updating timestamp only", $debug_enabled);
 751
 752                                         // we keep encountering the entry in feeds, so we need to
 753                                         // update date_updated column so that we don't get horrible
 754                                         // dupes when the entry gets purged and reinserted again e.g.
 755                                         // in the case of SLOW SLOW OMG SLOW updating feeds
 756
 757                                         $base_entry_id = db_fetch_result($result, 0, "id");
 758
 759                                         db_query("UPDATE ttrss_entries SET date_updated = NOW()
 760                                                 WHERE id = '$base_entry_id'");
 761
 762                                         continue;
 763                                 }
 764
 765                                 _debug("hash differs, applying plugin filters:", $debug_enabled);
 766
 767                                 foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
 768                                         _debug("... " . get_class($plugin), $debug_enabled);
 769
 770                                         $start = microtime(true);
 771                                         $article = $plugin->hook_article_filter($article);
 772
 773                                         _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
 774
 775                                         $entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
 776                                 }
 777
 778                                 if ($_REQUEST["xdebug"] == 2) {
 779                                         print "processed content: ";
 780                                         print htmlspecialchars($article["content"]);
 781                                         print "\n";
 782                                 }
 783
 784                                 $entry_plugin_data = db_escape_string($entry_plugin_data);
 785
 786                                 _debug("plugin data: $entry_plugin_data", $debug_enabled);
 787
 788                                 // Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
 789                                 if (DB_TYPE == "mysql") {
 790                                         foreach ($article as $k => $v) {
 791
 792                                                 // i guess we'll have to take the risk of 4byte unicode labels & tags here
 793                                                 if (is_string($article[$k])) {
 794                                                         $article[$k] = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $v);
 795                                                 }
 796                                         }
 797                                 }
 798
 799                                 /* Collect article tags here so we could filter by them: */
 800
 801                                 $article_filters = get_article_filters($filters, $article["title"],
 802                                         $article["content"], $article["link"], 0, $article["author"],
 803                                         $article["tags"]);
 804
 805                                 if ($debug_enabled) {
 806                                         _debug("article filters: ", $debug_enabled);
 807                                         if (count($article_filters) != 0) {
 808                                                 print_r($article_filters);
 809                                         }
 810                                 }
 811
 812                                 $plugin_filter_names = find_article_filters($article_filters, "plugin");
 813                                 $plugin_filter_actions = $pluginhost->get_filter_actions();
 814
 815                                 if (count($plugin_filter_names) > 0) {
 816                                         _debug("applying plugin filter actions...", $debug_enabled);
 817
 818                                         foreach ($plugin_filter_names as $pfn) {
 819                                                 list($pfclass,$pfaction) = explode(":", $pfn["param"]);
 820
 821                                                 if (isset($plugin_filter_actions[$pfclass])) {
 822                                                         $plugin = $pluginhost->get_plugin($pfclass);
 823
 824                                                         _debug("... $pfclass: $pfaction", $debug_enabled);
 825
 826                                                         if ($plugin) {
 827                                                                 $start = microtime(true);
 828                                                                 $article = $plugin->hook_article_filter_action($article, $pfaction);
 829
 830                                                                 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
 831                                                         } else {
 832                                                                 _debug("??? $pfclass: plugin object not found.");
 833                                                         }
 834                                                 } else {
 835                                                         _debug("??? $pfclass: filter plugin not registered.");
 836                                                 }
 837                                         }
 838                                 }
 839
 840                                 $entry_tags = $article["tags"];
 841                                 $entry_guid = db_escape_string($entry_guid);
 842                                 $entry_title = db_escape_string($article["title"]);
 843                                 $entry_author = db_escape_string($article["author"]);
 844                                 $entry_link = db_escape_string($article["link"]);
 845                                 $entry_content = $article["content"]; // escaped below
 846                                 $entry_force_catchup = $article["force_catchup"];
 847                                 $article_labels = $article["labels"];
 848                                 $entry_score_modifier = (int) $article["score_modifier"];
 849                                 $entry_language = db_escape_string($article["language"]);
 850
 851                                 if ($debug_enabled) {
 852                                         _debug("article labels:", $debug_enabled);
 853                                         print_r($article_labels);
 854                                 }
 855
 856                                 _debug("force catchup: $entry_force_catchup");
 857
 858                                 if ($cache_images && is_writable(CACHE_DIR . '/images'))
 859                                         cache_images($entry_content, $site_url, $debug_enabled);
 860
 861                                 $entry_content = db_escape_string($entry_content, false);
 862
 863                                 //db_query("BEGIN");
 864
 865                                 $result = db_query("SELECT id FROM      ttrss_entries
 866                                         WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
 867
 868                                 if (db_num_rows($result) == 0) {
 869
 870                                         _debug("base guid [$entry_guid] not found", $debug_enabled);
 871
 872                                         // base post entry does not exist, create it
 873
 874                                         $result = db_query(
 875                                                 "INSERT INTO ttrss_entries
 876                                                         (title,
 877                                                         guid,
 878                                                         link,
 879                                                         updated,
 880                                                         content,
 881                                                         content_hash,
 882                                                         no_orig_date,
 883                                                         date_updated,
 884                                                         date_entered,
 885                                                         comments,
 886                                                         num_comments,
 887                                                         plugin_data,
 888                                                         lang,
 889                                                         author)
 890                                                 VALUES
 891                                                         ('$entry_title',
 892                                                         '$entry_guid_hashed',
 893                                                         '$entry_link',
 894                                                         '$entry_timestamp_fmt',
 895                                                         '$entry_content',
 896                                                         '$entry_current_hash',
 897                                                         false,
 898                                                         NOW(),
 899                                                         '$date_feed_processed',
 900                                                         '$entry_comments',
 901                                                         '$num_comments',
 902                                                         '$entry_plugin_data',
 903                                                         '$entry_language',
 904                                                         '$entry_author')");
 905
 906                                 } else {
 907                                         $base_entry_id = db_fetch_result($result, 0, "id");
 908                                 }
 909
 910                                 // now it should exist, if not - bad luck then
 911
 912                                 $result = db_query("SELECT id FROM ttrss_entries
 913                                         WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
 914
 915                                 $entry_ref_id = 0;
 916                                 $entry_int_id = 0;
 917
 918                                 if (db_num_rows($result) == 1) {
 919
 920                                         _debug("base guid found, checking for user record", $debug_enabled);
 921
 922                                         $ref_id = db_fetch_result($result, 0, "id");
 923                                         $entry_ref_id = $ref_id;
 924
 925                                         /* $stored_guid = db_fetch_result($result, 0, "guid");
 926                                         if ($stored_guid != $entry_guid_hashed) {
 927                                                 if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
 928
 929                                                 db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
 930                                                         id = '$ref_id'");
 931                                         } */
 932
 933                                         if (find_article_filter($article_filters, "filter")) {
 934                                                 //db_query("COMMIT"); // close transaction in progress
 935                                                 continue;
 936                                         }
 937
 938                                         $score = calculate_article_score($article_filters) + $entry_score_modifier;
 939
 940                                         _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
 941
 942                                         // check for user post link to main table
 943
 944                                         $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
 945                                                         ref_id = '$ref_id' AND owner_uid = '$owner_uid'";
 946
 947 //                                      if ($_REQUEST["xdebug"]) print "$query\n";
 948
 949                                         $result = db_query($query);
 950
 951                                         // okay it doesn't exist - create user entry
 952                                         if (db_num_rows($result) == 0) {
 953
 954                                                 _debug("user record not found, creating...", $debug_enabled);
 955
 956                                                 if ($score >= -500 && !find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
 957                                                         $unread = 'true';
 958                                                         $last_read_qpart = 'NULL';
 959                                                 } else {
 960                                                         $unread = 'false';
 961                                                         $last_read_qpart = 'NOW()';
 962                                                 }
 963
 964                                                 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
 965                                                         $marked = 'true';
 966                                                 } else {
 967                                                         $marked = 'false';
 968                                                 }
 969
 970                                                 if (find_article_filter($article_filters, 'publish')) {
 971                                                         $published = 'true';
 972                                                 } else {
 973                                                         $published = 'false';
 974                                                 }
 975
 976                                                 // N-grams
 977
 978                                                 /* if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
 979
 980                                                         $result = db_query("SELECT COUNT(*) AS similar FROM
 981                                                                         ttrss_entries,ttrss_user_entries
 982                                                                 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
 983                                                                         AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
 984                                                                         AND owner_uid = $owner_uid");
 985
 986                                                         $ngram_similar = db_fetch_result($result, 0, "similar");
 987
 988                                                         _debug("N-gram similar results: $ngram_similar", $debug_enabled);
 989
 990                                                         if ($ngram_similar > 0) {
 991                                                                 $unread = 'false';
 992                                                         }
 993                                                 } */
 994
 995                                                 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
 996                                                 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
 997
 998                                                 $result = db_query(
 999                                                         "INSERT INTO ttrss_user_entries
1000                                                                 (ref_id, owner_uid, feed_id, unread, last_read, marked,
1001                                                                 published, score, tag_cache, label_cache, uuid,
1002                                                                 last_marked, last_published)
1003                                                         VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
1004                                                                 $last_read_qpart, $marked, $published, '$score', '', '',
1005                                                                 '', $last_marked, $last_published)");
1006
1007                                                 if (PUBSUBHUBBUB_HUB && $published == 'true') {
1008                                                         $rss_link = get_self_url_prefix() .
1009                                                                 "/public.php?op=rss&id=-2&key=" .
1010                                                                 get_feed_access_key(-2, false, $owner_uid);
1011
1012                                                         $p = new Publisher(PUBSUBHUBBUB_HUB);
1013
1014                                                         /* $pubsub_result = */ $p->publish_update($rss_link);
1015                                                 }
1016
1017                                                 $result = db_query(
1018                                                         "SELECT int_id FROM ttrss_user_entries WHERE
1019                                                                 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
1020                                                                 feed_id = '$feed' LIMIT 1");
1021
1022                                                 if (db_num_rows($result) == 1) {
1023                                                         $entry_int_id = db_fetch_result($result, 0, "int_id");
1024                                                 }
1025                                         } else {
1026                                                 _debug("user record FOUND", $debug_enabled);
1027
1028                                                 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
1029                                                 $entry_int_id = db_fetch_result($result, 0, "int_id");
1030                                         }
1031
1032                                         _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
1033
1034                                         if (DB_TYPE == "pgsql") {
1035                                            $tsvector_combined = db_escape_string(mb_substr($entry_title . ' ' . strip_tags(str_replace('<', ' <', $entry_content)),
1036                                                         0, 1000000));
1037
1038                                                 $tsvector_qpart = "tsvector_combined = to_tsvector('$feed_language', '$tsvector_combined'),";
1039
1040                                         } else {
1041                                                 $tsvector_qpart = "";
1042                                         }
1043
1044                                         db_query("UPDATE ttrss_entries
1045                                                 SET title = '$entry_title',
1046                                                         content = '$entry_content',
1047                                                         content_hash = '$entry_current_hash',
1048                                                         updated = '$entry_timestamp_fmt',
1049                                                         $tsvector_qpart
1050                                                         num_comments = '$num_comments',
1051                                                         plugin_data = '$entry_plugin_data',
1052                                                         author = '$entry_author',
1053                                                         lang = '$entry_language'
1054                                                 WHERE id = '$ref_id'");
1055
1056                                         // update aux data
1057                                         db_query("UPDATE ttrss_user_entries
1058                                                         SET score = '$score' WHERE ref_id = '$ref_id'");
1059
1060                                         if ($mark_unread_on_update) {
1061                                                 _debug("article updated, marking unread as requested.", $debug_enabled);
1062
1063                                                 db_query("UPDATE ttrss_user_entries
1064                                                         SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
1065                                         }
1066                                 }
1067
1068                                 //db_query("COMMIT");
1069
1070                                 _debug("assigning labels [other]...", $debug_enabled);
1071
1072                                 foreach ($article_labels as $label) {
1073                                         label_add_article($entry_ref_id, $label[1], $owner_uid);
1074                                 }
1075
1076                                 _debug("assigning labels [filters]...", $debug_enabled);
1077
1078                                 assign_article_to_label_filters($entry_ref_id, $article_filters,
1079                                         $owner_uid, $article_labels);
1080
1081                                 _debug("looking for enclosures...", $debug_enabled);
1082
1083                                 // enclosures
1084
1085                                 $enclosures = array();
1086
1087                                 $encs = $item->get_enclosures();
1088
1089                                 if (is_array($encs)) {
1090                                         foreach ($encs as $e) {
1091                                                 $e_item = array(
1092                                                         $e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
1093                                                 array_push($enclosures, $e_item);
1094                                         }
1095                                 }
1096
1097                                 if ($debug_enabled) {
1098                                         _debug("article enclosures:", $debug_enabled);
1099                                         print_r($enclosures);
1100                                 }
1101
1102                                 //db_query("BEGIN");
1103
1104 //                              debugging
1105 //                              db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
1106
1107                                 foreach ($enclosures as $enc) {
1108                                         $enc_url = db_escape_string($enc[0]);
1109                                         $enc_type = db_escape_string($enc[1]);
1110                                         $enc_dur = db_escape_string($enc[2]);
1111                                         $enc_title = db_escape_string($enc[3]);
1112                                         $enc_width = intval($enc[4]);
1113                                         $enc_height = intval($enc[5]);
1114
1115                                         $result = db_query("SELECT id FROM ttrss_enclosures
1116                                                 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
1117
1118                                         if (db_num_rows($result) == 0) {
1119                                                 db_query("INSERT INTO ttrss_enclosures
1120                                                         (content_url, content_type, title, duration, post_id, width, height) VALUES
1121                                                         ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
1122                                         }
1123                                 }
1124
1125                                 //db_query("COMMIT");
1126
1127                                 // check for manual tags (we have to do it here since they're loaded from filters)
1128
1129                                 foreach ($article_filters as $f) {
1130                                         if ($f["type"] == "tag") {
1131
1132                                                 $manual_tags = trim_array(explode(",", $f["param"]));
1133
1134                                                 foreach ($manual_tags as $tag) {
1135                                                         if (tag_is_valid($tag)) {
1136                                                                 array_push($entry_tags, $tag);
1137                                                         }
1138                                                 }
1139                                         }
1140                                 }
1141
1142                                 // Skip boring tags
1143
1144                                 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
1145                                         'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1146
1147                                 $filtered_tags = array();
1148                                 $tags_to_cache = array();
1149
1150                                 if ($entry_tags && is_array($entry_tags)) {
1151                                         foreach ($entry_tags as $tag) {
1152                                                 if (array_search($tag, $boring_tags) === false) {
1153                                                         array_push($filtered_tags, $tag);
1154                                                 }
1155                                         }
1156                                 }
1157
1158                                 $filtered_tags = array_unique($filtered_tags);
1159
1160                                 if ($debug_enabled) {
1161                                         _debug("filtered article tags:", $debug_enabled);
1162                                         print_r($filtered_tags);
1163                                 }
1164
1165                                 // Save article tags in the database
1166
1167                                 if (count($filtered_tags) > 0) {
1168
1169                                         //db_query("BEGIN");
1170
1171                                         foreach ($filtered_tags as $tag) {
1172
1173                                                 $tag = sanitize_tag($tag);
1174                                                 $tag = db_escape_string($tag);
1175
1176                                                 if (!tag_is_valid($tag)) continue;
1177
1178                                                 $result = db_query("SELECT id FROM ttrss_tags
1179                                                         WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1180                                                         owner_uid = '$owner_uid' LIMIT 1");
1181
1182                                                         if ($result && db_num_rows($result) == 0) {
1183
1184                                                                 db_query("INSERT INTO ttrss_tags
1185                                                                         (owner_uid,tag_name,post_int_id)
1186                                                                         VALUES ('$owner_uid','$tag', '$entry_int_id')");
1187                                                         }
1188
1189                                                 array_push($tags_to_cache, $tag);
1190                                         }
1191
1192                                         /* update the cache */
1193
1194                                         $tags_to_cache = array_unique($tags_to_cache);
1195
1196                                         $tags_str = db_escape_string(join(",", $tags_to_cache));
1197
1198                                         db_query("UPDATE ttrss_user_entries
1199                                                 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1200                                                 AND owner_uid = $owner_uid");
1201
1202                                         //db_query("COMMIT");
1203                                 }
1204
1205                                 _debug("article processed", $debug_enabled);
1206                         }
1207
1208                         _debug("purging feed...", $debug_enabled);
1209
1210                         purge_feed($feed, 0, $debug_enabled);
1211
1212                         db_query("UPDATE ttrss_feeds
1213                                 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
1214
1215 //                      db_query("COMMIT");
1216
1217                 } else {
1218
1219                         $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
1220
1221                         _debug("fetch error: $error_msg", $debug_enabled);
1222
1223                         if (count($rss->errors()) > 1) {
1224                                 foreach ($rss->errors() as $error) {
1225                                         _debug("+ $error");
1226                                 }
1227                         }
1228
1229                         db_query(
1230                                 "UPDATE ttrss_feeds SET last_error = '$error_msg',
1231                                 last_updated = NOW() WHERE id = '$feed'");
1232
1233                         unset($rss);
1234                 }
1235
1236                 _debug("done", $debug_enabled);
1237
1238                 return $rss;
1239         }
1240
1241         function cache_images($html, $site_url, $debug) {
1242                 libxml_use_internal_errors(true);
1243
1244                 $charset_hack = '<head>
1245                         <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1246                 </head>';
1247
1248                 $doc = new DOMDocument();
1249                 $doc->loadHTML($charset_hack . $html);
1250                 $xpath = new DOMXPath($doc);
1251
1252                 $entries = $xpath->query('(//img[@src])');
1253
1254                 foreach ($entries as $entry) {
1255                         if ($entry->hasAttribute('src')) {
1256                                 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1257
1258                                 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1259
1260                                 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1261
1262                                 if (!file_exists($local_filename)) {
1263                                         $file_content = fetch_file_contents($src);
1264
1265                                         if ($file_content && strlen($file_content) > _MIN_CACHE_IMAGE_SIZE) {
1266                                                 file_put_contents($local_filename, $file_content);
1267                                         }
1268                                 } else {
1269                                         touch($local_filename);
1270                                 }
1271                         }
1272                 }
1273         }
1274
1275         function expire_error_log($debug) {
1276                 if ($debug) _debug("Removing old error log entries...");
1277
1278                 if (DB_TYPE == "pgsql") {
1279                         db_query("DELETE FROM ttrss_error_log
1280                                 WHERE created_at < NOW() - INTERVAL '7 days'");
1281                 } else {
1282                         db_query("DELETE FROM ttrss_error_log
1283                                 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1284                 }
1285
1286         }
1287
1288         function expire_lock_files($debug) {
1289                 //if ($debug) _debug("Removing old lock files...");
1290
1291                 $num_deleted = 0;
1292
1293                 if (is_writable(LOCK_DIRECTORY)) {
1294                         $files = glob(LOCK_DIRECTORY . "/*.lock");
1295
1296                         if ($files) {
1297                                 foreach ($files as $file) {
1298                                         if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
1299                                                 unlink($file);
1300                                                 ++$num_deleted;
1301                                         }
1302                                 }
1303                         }
1304                 }
1305
1306                 if ($debug) _debug("Removed $num_deleted old lock files.");
1307         }
1308
1309         function expire_cached_files($debug) {
1310                 foreach (array("simplepie", "images", "export", "upload") as $dir) {
1311                         $cache_dir = CACHE_DIR . "/$dir";
1312
1313 //                      if ($debug) _debug("Expiring $cache_dir");
1314
1315                         $num_deleted = 0;
1316
1317                         if (is_writable($cache_dir)) {
1318                                 $files = glob("$cache_dir/*");
1319
1320                                 if ($files) {
1321                                         foreach ($files as $file) {
1322                                                 if (time() - filemtime($file) > 86400*7) {
1323                                                         unlink($file);
1324
1325                                                         ++$num_deleted;
1326                                                 }
1327                                         }
1328                                 }
1329                         }
1330
1331                         if ($debug) _debug("$cache_dir: removed $num_deleted files.");
1332                 }
1333         }
1334
1335         /**
1336         * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1337         * Returns the url query as associative array
1338         *
1339         * @param    string    query
1340         * @return    array    params
1341         */
1342         function convertUrlQuery($query) {
1343                 $queryParts = explode('&', $query);
1344
1345                 $params = array();
1346
1347                 foreach ($queryParts as $param) {
1348                         $item = explode('=', $param);
1349                         $params[$item[0]] = $item[1];
1350                 }
1351
1352                 return $params;
1353         }
1354
1355         function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
1356                 $matches = array();
1357
1358                 foreach ($filters as $filter) {
1359                         $match_any_rule = $filter["match_any_rule"];
1360                         $inverse = $filter["inverse"];
1361                         $filter_match = false;
1362
1363                         foreach ($filter["rules"] as $rule) {
1364                                 $match = false;
1365                                 $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
1366                                 $rule_inverse = $rule["inverse"];
1367
1368                                 if (!$reg_exp)
1369                                         continue;
1370
1371                                 switch ($rule["type"]) {
1372                                 case "title":
1373                                         $match = @preg_match("/$reg_exp/i", $title);
1374                                         break;
1375                                 case "content":
1376                                         // we don't need to deal with multiline regexps
1377                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1378
1379                                         $match = @preg_match("/$reg_exp/i", $content);
1380                                         break;
1381                                 case "both":
1382                                         // we don't need to deal with multiline regexps
1383                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1384
1385                                         $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
1386                                         break;
1387                                 case "link":
1388                                         $match = @preg_match("/$reg_exp/i", $link);
1389                                         break;
1390                                 case "author":
1391                                         $match = @preg_match("/$reg_exp/i", $author);
1392                                         break;
1393                                 case "tag":
1394                                         foreach ($tags as $tag) {
1395                                                 if (@preg_match("/$reg_exp/i", $tag)) {
1396                                                         $match = true;
1397                                                         break;
1398                                                 }
1399                                         }
1400                                         break;
1401                                 }
1402
1403                                 if ($rule_inverse) $match = !$match;
1404
1405                                 if ($match_any_rule) {
1406                                         if ($match) {
1407                                                 $filter_match = true;
1408                                                 break;
1409                                         }
1410                                 } else {
1411                                         $filter_match = $match;
1412                                         if (!$match) {
1413                                                 break;
1414                                         }
1415                                 }
1416                         }
1417
1418                         if ($inverse) $filter_match = !$filter_match;
1419
1420                         if ($filter_match) {
1421                                 foreach ($filter["actions"] AS $action) {
1422                                         array_push($matches, $action);
1423
1424                                         // if Stop action encountered, perform no further processing
1425                                         if (isset($action["type"]) && $action["type"] == "stop") return $matches;
1426                                 }
1427                         }
1428                 }
1429
1430                 return $matches;
1431         }
1432
1433         function find_article_filter($filters, $filter_name) {
1434                 foreach ($filters as $f) {
1435                         if ($f["type"] == $filter_name) {
1436                                 return $f;
1437                         };
1438                 }
1439                 return false;
1440         }
1441
1442         function find_article_filters($filters, $filter_name) {
1443                 $results = array();
1444
1445                 foreach ($filters as $f) {
1446                         if ($f["type"] == $filter_name) {
1447                                 array_push($results, $f);
1448                         };
1449                 }
1450                 return $results;
1451         }
1452
1453         function calculate_article_score($filters) {
1454                 $score = 0;
1455
1456                 foreach ($filters as $f) {
1457                         if ($f["type"] == "score") {
1458                                 $score += $f["param"];
1459                         };
1460                 }
1461                 return $score;
1462         }
1463
1464         function labels_contains_caption($labels, $caption) {
1465                 foreach ($labels as $label) {
1466                         if ($label[1] == $caption) {
1467                                 return true;
1468                         }
1469                 }
1470
1471                 return false;
1472         }
1473
1474         function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
1475                 foreach ($filters as $f) {
1476                         if ($f["type"] == "label") {
1477                                 if (!labels_contains_caption($article_labels, $f["param"])) {
1478                                         label_add_article($id, $f["param"], $owner_uid);
1479                                 }
1480                         }
1481                 }
1482         }
1483
1484         function make_guid_from_title($title) {
1485                 return preg_replace("/[ \"\',.:;]/", "-",
1486                         mb_strtolower(strip_tags($title), 'utf-8'));
1487         }
1488
1489         /* function verify_feed_xml($feed_data) {
1490                 libxml_use_internal_errors(true);
1491                 $doc = new DOMDocument();
1492                 $doc->loadXML($feed_data);
1493                 $error = libxml_get_last_error();
1494                 libxml_clear_errors();
1495                 return $error;
1496         } */
1497
1498         function cleanup_counters_cache($debug) {
1499                 $result = db_query("DELETE FROM ttrss_counters_cache
1500                         WHERE feed_id > 0 AND
1501                         (SELECT COUNT(id) FROM ttrss_feeds WHERE
1502                                 id = feed_id AND
1503                                 ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
1504                 $frows = db_affected_rows($result);
1505
1506                 $result = db_query("DELETE FROM ttrss_cat_counters_cache
1507                         WHERE feed_id > 0 AND
1508                         (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
1509                                 id = feed_id AND
1510                                 ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
1511                 $crows = db_affected_rows($result);
1512
1513                 _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
1514         }
1515
1516         function housekeeping_user($owner_uid) {
1517                 $tmph = new PluginHost();
1518
1519                 load_user_plugins($owner_uid, $tmph);
1520
1521                 $tmph->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
1522         }
1523
1524         function housekeeping_common($debug) {
1525                 expire_cached_files($debug);
1526                 expire_lock_files($debug);
1527                 expire_error_log($debug);
1528
1529                 $count = update_feedbrowser_cache();
1530                 _debug("Feedbrowser updated, $count feeds processed.");
1531
1532                 purge_orphans( true);
1533                 cleanup_counters_cache($debug);
1534
1535                 //$rc = cleanup_tags( 14, 50000);
1536                 //_debug("Cleaned $rc cached tags.");
1537
1538                 PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
1539         }
1540 ?>