classes/rssutils.php

   1 <?php
   2 class RSSUtils {
   3         static function calculate_article_hash($article, $pluginhost) {
   4                 $tmp = "";
   5
   6                 foreach ($article as $k => $v) {
   7                         if ($k != "feed" && isset($v)) {
   8                                 $x = strip_tags(is_array($v) ? implode(",", $v) : $v);
   9
  10                                 //_debug("$k:" . sha1($x) . ":" . htmlspecialchars($x), true);
  11
  12                                 $tmp .= sha1("$k:" . sha1($x));
  13                         }
  14                 }
  15
  16                 return sha1(implode(",", $pluginhost->get_plugin_names()) . $tmp);
  17         }
  18
  19         static function update_feedbrowser_cache() {
  20
  21                 $result = db_query("SELECT feed_url, site_url, title, COUNT(id) AS subscribers
  22                         FROM ttrss_feeds WHERE feed_url NOT IN (SELECT feed_url FROM ttrss_feeds
  23                                 WHERE private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%')
  24                                 GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
  25
  26                 db_query("BEGIN");
  27
  28                 db_query("DELETE FROM ttrss_feedbrowser_cache");
  29
  30                 $count = 0;
  31
  32                 while ($line = db_fetch_assoc($result)) {
  33                         $subscribers = db_escape_string($line["subscribers"]);
  34                         $feed_url = db_escape_string($line["feed_url"]);
  35                         $title = db_escape_string($line["title"]);
  36                         $site_url = db_escape_string($line["site_url"]);
  37
  38                         $tmp_result = db_query("SELECT subscribers FROM
  39                                 ttrss_feedbrowser_cache WHERE feed_url = '$feed_url'");
  40
  41                         if (db_num_rows($tmp_result) == 0) {
  42
  43                                 db_query("INSERT INTO ttrss_feedbrowser_cache
  44                                         (feed_url, site_url, title, subscribers) VALUES ('$feed_url',
  45                                                 '$site_url', '$title', '$subscribers')");
  46
  47                                 ++$count;
  48
  49                         }
  50
  51                 }
  52
  53                 db_query("COMMIT");
  54
  55                 return $count;
  56
  57         }
  58
  59         static function update_daemon_common($limit = DAEMON_FEED_LIMIT, $debug = true) {
  60                 $schema_version = get_schema_version();
  61
  62                 if ($schema_version != SCHEMA_VERSION) {
  63                         die("Schema version is wrong, please upgrade the database.\n");
  64                 }
  65
  66                 if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
  67                         if (DB_TYPE == "pgsql") {
  68                                 $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
  69                         } else {
  70                                 $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
  71                         }
  72                 } else {
  73                         $login_thresh_qpart = "";
  74                 }
  75
  76                 if (DB_TYPE == "pgsql") {
  77                         $update_limit_qpart = "AND ((
  78                                         ttrss_feeds.update_interval = 0
  79                                         AND ttrss_user_prefs.value != '-1'
  80                                         AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
  81                                 ) OR (
  82                                         ttrss_feeds.update_interval > 0
  83                                         AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
  84                                 ) OR (ttrss_feeds.last_updated IS NULL
  85                                         AND ttrss_user_prefs.value != '-1')
  86                                 OR (last_updated = '1970-01-01 00:00:00'
  87                                         AND ttrss_user_prefs.value != '-1'))";
  88                 } else {
  89                         $update_limit_qpart = "AND ((
  90                                         ttrss_feeds.update_interval = 0
  91                                         AND ttrss_user_prefs.value != '-1'
  92                                         AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
  93                                 ) OR (
  94                                         ttrss_feeds.update_interval > 0
  95                                         AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
  96                                 ) OR (ttrss_feeds.last_updated IS NULL
  97                                         AND ttrss_user_prefs.value != '-1')
  98                                 OR (last_updated = '1970-01-01 00:00:00'
  99                                         AND ttrss_user_prefs.value != '-1'))";
 100                 }
 101
 102                 // Test if feed is currently being updated by another process.
 103                 if (DB_TYPE == "pgsql") {
 104                         $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
 105                 } else {
 106                         $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 10 MINUTE))";
 107                 }
 108
 109                 $query_limit = $limit ? sprintf("LIMIT %d", $limit) : "";
 110
 111                 // Update the least recently updated feeds first
 112                 $query_order = "ORDER BY last_updated";
 113                 if (DB_TYPE == "pgsql") $query_order .= " NULLS FIRST";
 114
 115                 $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
 116                         FROM
 117                                 ttrss_feeds, ttrss_users, ttrss_user_prefs
 118                         WHERE
 119                                 ttrss_feeds.owner_uid = ttrss_users.id
 120                                 AND ttrss_user_prefs.profile IS NULL
 121                                 AND ttrss_users.id = ttrss_user_prefs.owner_uid
 122                                 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
 123                                 $login_thresh_qpart $update_limit_qpart
 124                                 $updstart_thresh_qpart
 125                                 $query_order $query_limit";
 126
 127                 $result = db_query($query);
 128
 129                 if ($debug) _debug(sprintf("Scheduled %d feeds to update...", db_num_rows($result)));
 130
 131                 $feeds_to_update = array();
 132                 while ($line = db_fetch_assoc($result)) {
 133                         array_push($feeds_to_update, $line['feed_url']);
 134                 }
 135
 136                 // Update last_update_started before actually starting the batch
 137                 // in order to minimize collision risk for parallel daemon tasks
 138                 if (count($feeds_to_update) > 0) {
 139                         $feeds_quoted = array_map(function ($s) { return "'" . db_escape_string($s) . "'"; }, $feeds_to_update);
 140
 141                         db_query(sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
 142                                 WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
 143                 }
 144
 145                 $nf = 0;
 146                 $bstarted = microtime(true);
 147
 148                 $batch_owners = array();
 149
 150                 foreach ($feeds_to_update as $feed) {
 151                         if($debug) _debug("Base feed: $feed");
 152
 153                         //update_rss_feed($line["id"], true);
 154
 155                         // since we have the data cached, we can deal with other feeds with the same url
 156                         $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
 157                         FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
 158                                 ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
 159                                 ttrss_users.id = ttrss_user_prefs.owner_uid AND
 160                                 ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
 161                                 ttrss_user_prefs.profile IS NULL AND
 162                                 feed_url = '".db_escape_string($feed)."'
 163                                 $update_limit_qpart
 164                                 $login_thresh_qpart
 165                         ORDER BY ttrss_feeds.id $query_limit");
 166
 167                         if (db_num_rows($tmp_result) > 0) {
 168                                 while ($tline = db_fetch_assoc($tmp_result)) {
 169                                         if ($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
 170
 171                                         if (array_search($tline["owner_uid"], $batch_owners) === FALSE)
 172                                                 array_push($batch_owners, $tline["owner_uid"]);
 173
 174                                         $fstarted = microtime(true);
 175                                         RSSUtils::update_rss_feed($tline["id"], true, false);
 176                                         _debug_suppress(false);
 177
 178                                         _debug(sprintf("    %.4f (sec)", microtime(true) - $fstarted));
 179
 180                                         ++$nf;
 181                                 }
 182                         }
 183                 }
 184
 185                 if ($nf > 0) {
 186                         _debug(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
 187                                 microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
 188                 }
 189
 190                 foreach ($batch_owners as $owner_uid) {
 191                         _debug("Running housekeeping tasks for user $owner_uid...");
 192
 193                         RSSUtils::housekeeping_user($owner_uid);
 194                 }
 195
 196                 // Send feed digests by email if needed.
 197                 Digest::send_headlines_digests($debug);
 198
 199                 return $nf;
 200
 201         }
 202
 203         // this is used when subscribing
 204         static function set_basic_feed_info($feed) {
 205
 206                 $feed = db_escape_string($feed);
 207
 208                 $result = db_query("SELECT feed_url,auth_pass,auth_login,auth_pass_encrypted
 209                                         FROM ttrss_feeds WHERE id = '$feed'");
 210
 211                 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
 212                         0, "auth_pass_encrypted"));
 213
 214                 $auth_login = db_fetch_result($result, 0, "auth_login");
 215                 $auth_pass = db_fetch_result($result, 0, "auth_pass");
 216
 217                 if ($auth_pass_encrypted && function_exists("mcrypt_decrypt")) {
 218                         require_once "crypt.php";
 219                         $auth_pass = decrypt_string($auth_pass);
 220                 }
 221
 222                 $fetch_url = db_fetch_result($result, 0, "feed_url");
 223
 224                 $feed_data = fetch_file_contents($fetch_url, false,
 225                         $auth_login, $auth_pass, false,
 226                         FEED_FETCH_TIMEOUT,
 227                         0);
 228
 229                 global $fetch_curl_used;
 230
 231                 if (!$fetch_curl_used) {
 232                         $tmp = @gzdecode($feed_data);
 233
 234                         if ($tmp) $feed_data = $tmp;
 235                 }
 236
 237                 $feed_data = trim($feed_data);
 238
 239                 $rss = new FeedParser($feed_data);
 240                 $rss->init();
 241
 242                 if (!$rss->error()) {
 243
 244                         $result = db_query("SELECT title, site_url FROM ttrss_feeds WHERE id = '$feed'");
 245
 246                         $registered_title = db_fetch_result($result, 0, "title");
 247                         $orig_site_url = db_fetch_result($result, 0, "site_url");
 248
 249                         $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
 250                         $feed_title = db_escape_string(mb_substr($rss->get_title(), 0, 199));
 251
 252                         if ($feed_title && (!$registered_title || $registered_title == "[Unknown]")) {
 253                                 db_query("UPDATE ttrss_feeds SET
 254                                         title = '$feed_title' WHERE id = '$feed'");
 255                         }
 256
 257                         if ($site_url && $orig_site_url != $site_url) {
 258                                 db_query("UPDATE ttrss_feeds SET
 259                                                         site_url = '$site_url' WHERE id = '$feed'");
 260                         }
 261                 }
 262         }
 263
 264         /**
 265          * @SuppressWarnings(PHPMD.UnusedFormalParameter)
 266          */
 267         static function update_rss_feed($feed, $no_cache = false) {
 268
 269                 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
 270
 271                 _debug_suppress(!$debug_enabled);
 272                 _debug("start", $debug_enabled);
 273
 274                 $result = db_query("SELECT title FROM ttrss_feeds
 275                         WHERE id = '$feed'");
 276
 277                 if (db_num_rows($result) == 0) {
 278                         _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
 279                         user_error("Attempt to update unknown/invalid feed $feed", E_USER_WARNING);
 280                         return false;
 281                 }
 282
 283                 $title = db_fetch_result($result, 0, "title");
 284
 285                 // feed was batch-subscribed or something, we need to get basic info
 286                 // this is not optimal currently as it fetches stuff separately TODO: optimize
 287                 if ($title == "[Unknown]") {
 288                         _debug("setting basic feed info for $feed...");
 289                         RSSUtils::set_basic_feed_info($feed);
 290                 }
 291
 292                 $result = db_query("SELECT id,update_interval,auth_login,
 293                         feed_url,auth_pass,cache_images,
 294                         mark_unread_on_update, owner_uid,
 295                         auth_pass_encrypted, feed_language
 296                         FROM ttrss_feeds WHERE id = '$feed'");
 297
 298                 $owner_uid = db_fetch_result($result, 0, "owner_uid");
 299                 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
 300                         0, "mark_unread_on_update"));
 301                 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
 302                         0, "auth_pass_encrypted"));
 303
 304                 db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
 305                         WHERE id = '$feed'");
 306
 307                 $auth_login = db_fetch_result($result, 0, "auth_login");
 308                 $auth_pass = db_fetch_result($result, 0, "auth_pass");
 309
 310                 if ($auth_pass_encrypted && function_exists("mcrypt_decrypt")) {
 311                         require_once "crypt.php";
 312                         $auth_pass = decrypt_string($auth_pass);
 313                 }
 314
 315                 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
 316                 $fetch_url = db_fetch_result($result, 0, "feed_url");
 317                 $feed_language = db_escape_string(mb_strtolower(db_fetch_result($result, 0, "feed_language")));
 318                 if (!$feed_language) $feed_language = 'english';
 319
 320                 $feed = db_escape_string($feed);
 321
 322                 $date_feed_processed = date('Y-m-d H:i');
 323
 324                 $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
 325
 326                 $pluginhost = new PluginHost();
 327                 $pluginhost->set_debug($debug_enabled);
 328                 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
 329
 330                 $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
 331                 $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
 332                 $pluginhost->load_data();
 333
 334                 $rss_hash = false;
 335
 336                 $force_refetch = isset($_REQUEST["force_refetch"]);
 337                 $feed_data = "";
 338
 339                 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
 340                         $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, 0, $auth_login, $auth_pass);
 341                 }
 342
 343                 // try cache
 344                 if (!$feed_data &&
 345                         file_exists($cache_filename) &&
 346                         is_readable($cache_filename) &&
 347                         !$auth_login && !$auth_pass &&
 348                         filemtime($cache_filename) > time() - 30) {
 349
 350                         _debug("using local cache [$cache_filename].", $debug_enabled);
 351
 352                         @$feed_data = file_get_contents($cache_filename);
 353
 354                         if ($feed_data) {
 355                                 $rss_hash = sha1($feed_data);
 356                         }
 357
 358                 } else {
 359                         _debug("local cache will not be used for this feed", $debug_enabled);
 360                 }
 361
 362                 // fetch feed from source
 363                 if (!$feed_data) {
 364                         _debug("fetching [$fetch_url]...", $debug_enabled);
 365
 366                         if (ini_get("open_basedir") && function_exists("curl_init")) {
 367                                 _debug("not using CURL due to open_basedir restrictions");
 368                         }
 369
 370                         $feed_data = fetch_file_contents($fetch_url, false,
 371                                 $auth_login, $auth_pass, false,
 372                                 $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
 373                                 0);
 374
 375                         global $fetch_curl_used;
 376
 377                         if (!$fetch_curl_used) {
 378                                 $tmp = @gzdecode($feed_data);
 379
 380                                 if ($tmp) $feed_data = $tmp;
 381                         }
 382
 383                         $feed_data = trim($feed_data);
 384
 385                         _debug("fetch done.", $debug_enabled);
 386
 387                         // cache vanilla feed data for re-use
 388                         if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
 389                                 $new_rss_hash = sha1($feed_data);
 390
 391                                 if ($new_rss_hash != $rss_hash) {
 392                                         _debug("saving $cache_filename", $debug_enabled);
 393                                         @file_put_contents($cache_filename, $feed_data);
 394                                 }
 395                         }
 396                 }
 397
 398                 if (!$feed_data) {
 399                         global $fetch_last_error;
 400                         global $fetch_last_error_code;
 401
 402                         _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
 403
 404                         $error_escaped = '';
 405
 406                         // If-Modified-Since
 407                         if ($fetch_last_error_code != 304) {
 408                                 $error_escaped = db_escape_string($fetch_last_error);
 409                         } else {
 410                                 _debug("source claims data not modified, nothing to do.", $debug_enabled);
 411                         }
 412
 413                         db_query(
 414                                 "UPDATE ttrss_feeds SET last_error = '$error_escaped',
 415                                         last_updated = NOW() WHERE id = '$feed'");
 416
 417                         return;
 418                 }
 419
 420                 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
 421                         $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
 422                 }
 423
 424                 $rss = new FeedParser($feed_data);
 425                 $rss->init();
 426
 427                 $feed = db_escape_string($feed);
 428
 429                 if (!$rss->error()) {
 430
 431                         // We use local pluginhost here because we need to load different per-user feed plugins
 432                         $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
 433
 434                         _debug("language: $feed_language", $debug_enabled);
 435                         _debug("processing feed data...", $debug_enabled);
 436
 437 //                      db_query("BEGIN");
 438
 439                         if (DB_TYPE == "pgsql") {
 440                                 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
 441                         } else {
 442                                 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
 443                         }
 444
 445                         $result = db_query("SELECT owner_uid,favicon_avg_color,
 446                                 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
 447                                                 favicon_needs_check
 448                                 FROM ttrss_feeds WHERE id = '$feed'");
 449
 450                         $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
 451                                 "favicon_needs_check"));
 452                         $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
 453
 454                         $owner_uid = db_fetch_result($result, 0, "owner_uid");
 455
 456                         $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
 457
 458                         _debug("site_url: $site_url", $debug_enabled);
 459                         _debug("feed_title: " . $rss->get_title(), $debug_enabled);
 460
 461                         if ($favicon_needs_check || $force_refetch) {
 462
 463                                 /* terrible hack: if we crash on floicon shit here, we won't check
 464                                  * the icon avgcolor again (unless the icon got updated) */
 465
 466                                 $favicon_file = ICONS_DIR . "/$feed.ico";
 467                                 $favicon_modified = @filemtime($favicon_file);
 468
 469                                 _debug("checking favicon...", $debug_enabled);
 470
 471                                 RSSUtils::check_feed_favicon($site_url, $feed);
 472                                 $favicon_modified_new = @filemtime($favicon_file);
 473
 474                                 if ($favicon_modified_new > $favicon_modified)
 475                                         $favicon_avg_color = '';
 476
 477                                 if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
 478                                         require_once "colors.php";
 479
 480                                         db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
 481                                                         id = '$feed'");
 482
 483                                         $favicon_color = db_escape_string(
 484                                                 calculate_avg_color($favicon_file));
 485
 486                                         $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
 487                                 } else if ($favicon_avg_color == 'fail') {
 488                                         _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
 489                                 }
 490
 491                                 db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
 492                                         $favicon_colorstring
 493                                         WHERE id = '$feed'");
 494                         }
 495
 496                         _debug("loading filters & labels...", $debug_enabled);
 497
 498                         $filters = load_filters($feed, $owner_uid);
 499
 500                         _debug("" . count($filters) . " filters loaded.", $debug_enabled);
 501
 502                         $items = $rss->get_items();
 503
 504                         if (!is_array($items)) {
 505                                 _debug("no articles found.", $debug_enabled);
 506
 507                                 db_query("UPDATE ttrss_feeds
 508                                         SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
 509
 510                                 return; // no articles
 511                         }
 512
 513                         _debug("processing articles...", $debug_enabled);
 514
 515                         $tstart = time();
 516
 517                         foreach ($items as $item) {
 518                                 if ($_REQUEST['xdebug'] == 3) {
 519                                         print_r($item);
 520                                 }
 521
 522                                 if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
 523                                         _debug("looks like there's too many articles to process at once, breaking out", $debug_enabled);
 524                                         break;
 525                                 }
 526
 527                                 $entry_guid = $item->get_id();
 528                                 if (!$entry_guid) $entry_guid = $item->get_link();
 529                                 if (!$entry_guid) $entry_guid = RSSUtils::make_guid_from_title($item->get_title());
 530                                 if (!$entry_guid) continue;
 531
 532                                 $entry_guid = "$owner_uid,$entry_guid";
 533
 534                                 $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
 535
 536                                 _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
 537
 538                                 $entry_timestamp = "";
 539
 540                                 $entry_timestamp = $item->get_date();
 541
 542                                 _debug("orig date: " . $item->get_date(), $debug_enabled);
 543
 544                                 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
 545                                         $entry_timestamp = time();
 546                                 }
 547
 548                                 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
 549
 550                                 _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
 551
 552 //                              $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
 553 //                              $entry_title = decode_numeric_entities($entry_title);
 554                                 $entry_title = $item->get_title();
 555
 556                                 $entry_link = rewrite_relative_url($site_url, $item->get_link());
 557
 558                                 _debug("title $entry_title", $debug_enabled);
 559                                 _debug("link $entry_link", $debug_enabled);
 560
 561                                 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
 562
 563                                 $entry_content = $item->get_content();
 564                                 if (!$entry_content) $entry_content = $item->get_description();
 565
 566                                 if ($_REQUEST["xdebug"] == 2) {
 567                                         print "content: ";
 568                                         print htmlspecialchars($entry_content);
 569                                         print "\n";
 570                                 }
 571
 572                                 $entry_comments = db_escape_string(mb_substr($item->get_comments_url(), 0, 245));
 573                                 $num_comments = (int) $item->get_comments_count();
 574
 575                                 $entry_author = $item->get_author(); // escaped later
 576                                 $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
 577
 578                                 _debug("author $entry_author", $debug_enabled);
 579                                 _debug("num_comments: $num_comments", $debug_enabled);
 580                                 _debug("looking for tags...", $debug_enabled);
 581
 582                                 // parse <category> entries into tags
 583
 584                                 $additional_tags = array();
 585
 586                                 $additional_tags_src = $item->get_categories();
 587
 588                                 if (is_array($additional_tags_src)) {
 589                                         foreach ($additional_tags_src as $tobj) {
 590                                                 array_push($additional_tags, $tobj);
 591                                         }
 592                                 }
 593
 594                                 $entry_tags = array_unique($additional_tags);
 595
 596                                 for ($i = 0; $i < count($entry_tags); $i++)
 597                                         $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
 598
 599                                 _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
 600
 601                                 _debug("done collecting data.", $debug_enabled);
 602
 603                                 $result = db_query("SELECT id, content_hash, lang FROM ttrss_entries
 604                                         WHERE guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed'");
 605
 606                                 if (db_num_rows($result) != 0) {
 607                                         $base_entry_id = db_fetch_result($result, 0, "id");
 608                                         $entry_stored_hash = db_fetch_result($result, 0, "content_hash");
 609                                         $article_labels = Article::get_article_labels($base_entry_id, $owner_uid);
 610                                         $entry_language = db_fetch_result($result, 0, "lang");
 611
 612                                         $existing_tags = Article::get_article_tags($base_entry_id, $owner_uid);
 613                                         $entry_tags = array_unique(array_merge($entry_tags, $existing_tags));
 614
 615                                 } else {
 616                                         $base_entry_id = false;
 617                                         $entry_stored_hash = "";
 618                                         $article_labels = array();
 619                                         $entry_language = "";
 620                                 }
 621
 622                                 $article = array("owner_uid" => $owner_uid, // read only
 623                                         "guid" => $entry_guid, // read only
 624                                         "guid_hashed" => $entry_guid_hashed, // read only
 625                                         "title" => $entry_title,
 626                                         "content" => $entry_content,
 627                                         "link" => $entry_link,
 628                                         "labels" => $article_labels, // current limitation: can add labels to article, can't remove them
 629                                         "tags" => $entry_tags,
 630                                         "author" => $entry_author,
 631                                         "force_catchup" => false, // ugly hack for the time being
 632                                         "score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
 633                                         "language" => $entry_language,
 634                                         "feed" => array("id" => $feed,
 635                                                 "fetch_url" => $fetch_url,
 636                                                 "site_url" => $site_url,
 637                                                 "cache_images" => $cache_images)
 638                                 );
 639
 640                                 $entry_plugin_data = "";
 641                                 $entry_current_hash = RSSUtils::calculate_article_hash($article, $pluginhost);
 642
 643                                 _debug("article hash: $entry_current_hash [stored=$entry_stored_hash]", $debug_enabled);
 644
 645                                 if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
 646                                         _debug("stored article seems up to date [IID: $base_entry_id], updating timestamp only", $debug_enabled);
 647
 648                                         // we keep encountering the entry in feeds, so we need to
 649                                         // update date_updated column so that we don't get horrible
 650                                         // dupes when the entry gets purged and reinserted again e.g.
 651                                         // in the case of SLOW SLOW OMG SLOW updating feeds
 652
 653                                         $base_entry_id = db_fetch_result($result, 0, "id");
 654
 655                                         db_query("UPDATE ttrss_entries SET date_updated = NOW()
 656                                                 WHERE id = '$base_entry_id'");
 657
 658                                         continue;
 659                                 }
 660
 661                                 _debug("hash differs, applying plugin filters:", $debug_enabled);
 662
 663                                 foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
 664                                         _debug("... " . get_class($plugin), $debug_enabled);
 665
 666                                         $start = microtime(true);
 667                                         $article = $plugin->hook_article_filter($article);
 668
 669                                         _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
 670
 671                                         $entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
 672                                 }
 673
 674                                 if ($_REQUEST["xdebug"] == 2) {
 675                                         print "processed content: ";
 676                                         print htmlspecialchars($article["content"]);
 677                                         print "\n";
 678                                 }
 679
 680                                 $entry_plugin_data = db_escape_string($entry_plugin_data);
 681
 682                                 _debug("plugin data: $entry_plugin_data", $debug_enabled);
 683
 684                                 // Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
 685                                 if (DB_TYPE == "mysql") {
 686                                         foreach ($article as $k => $v) {
 687
 688                                                 // i guess we'll have to take the risk of 4byte unicode labels & tags here
 689                                                 if (is_string($article[$k])) {
 690                                                         $article[$k] = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $v);
 691                                                 }
 692                                         }
 693                                 }
 694
 695                                 /* Collect article tags here so we could filter by them: */
 696
 697                                 $matched_rules = array();
 698
 699                                 $article_filters = RSSUtils::get_article_filters($filters, $article["title"],
 700                                         $article["content"], $article["link"], $article["author"],
 701                                         $article["tags"], $matched_rules);
 702
 703                                 if ($debug_enabled) {
 704                                         _debug("matched filter rules: ", $debug_enabled);
 705
 706                                         if (count($matched_rules) != 0) {
 707                                                 print_r($matched_rules);
 708                                         }
 709
 710                                         _debug("filter actions: ", $debug_enabled);
 711
 712                                         if (count($article_filters) != 0) {
 713                                                 print_r($article_filters);
 714                                         }
 715                                 }
 716
 717                                 $plugin_filter_names = RSSUtils::find_article_filters($article_filters, "plugin");
 718                                 $plugin_filter_actions = $pluginhost->get_filter_actions();
 719
 720                                 if (count($plugin_filter_names) > 0) {
 721                                         _debug("applying plugin filter actions...", $debug_enabled);
 722
 723                                         foreach ($plugin_filter_names as $pfn) {
 724                                                 list($pfclass,$pfaction) = explode(":", $pfn["param"]);
 725
 726                                                 if (isset($plugin_filter_actions[$pfclass])) {
 727                                                         $plugin = $pluginhost->get_plugin($pfclass);
 728
 729                                                         _debug("... $pfclass: $pfaction", $debug_enabled);
 730
 731                                                         if ($plugin) {
 732                                                                 $start = microtime(true);
 733                                                                 $article = $plugin->hook_article_filter_action($article, $pfaction);
 734
 735                                                                 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
 736                                                         } else {
 737                                                                 _debug("??? $pfclass: plugin object not found.");
 738                                                         }
 739                                                 } else {
 740                                                         _debug("??? $pfclass: filter plugin not registered.");
 741                                                 }
 742                                         }
 743                                 }
 744
 745                                 $entry_tags = $article["tags"];
 746                                 $entry_guid = db_escape_string($entry_guid);
 747                                 $entry_title = db_escape_string($article["title"]);
 748                                 $entry_author = db_escape_string(mb_substr($article["author"], 0, 245));
 749                                 $entry_link = db_escape_string($article["link"]);
 750                                 $entry_content = $article["content"]; // escaped below
 751                                 $entry_force_catchup = $article["force_catchup"];
 752                                 $article_labels = $article["labels"];
 753                                 $entry_score_modifier = (int) $article["score_modifier"];
 754                                 $entry_language = db_escape_string($article["language"]);
 755
 756                                 if ($debug_enabled) {
 757                                         _debug("article labels:", $debug_enabled);
 758
 759                                         if (count($article_labels) != 0) {
 760                                                 print_r($article_labels);
 761                                         }
 762                                 }
 763
 764                                 _debug("force catchup: $entry_force_catchup");
 765
 766                                 if ($cache_images && is_writable(CACHE_DIR . '/images'))
 767                                         RSSUtils::cache_media($entry_content, $site_url, $debug_enabled);
 768
 769                                 $entry_content = db_escape_string($entry_content, false);
 770
 771                                 //db_query("BEGIN");
 772
 773                                 $result = db_query("SELECT id FROM      ttrss_entries
 774                                         WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
 775
 776                                 if (db_num_rows($result) == 0) {
 777
 778                                         _debug("base guid [$entry_guid or $entry_guid_hashed] not found, creating...", $debug_enabled);
 779
 780                                         // base post entry does not exist, create it
 781
 782                                         db_query(
 783                                                 "INSERT INTO ttrss_entries
 784                                                         (title,
 785                                                         guid,
 786                                                         link,
 787                                                         updated,
 788                                                         content,
 789                                                         content_hash,
 790                                                         no_orig_date,
 791                                                         date_updated,
 792                                                         date_entered,
 793                                                         comments,
 794                                                         num_comments,
 795                                                         plugin_data,
 796                                                         lang,
 797                                                         author)
 798                                                 VALUES
 799                                                         ('$entry_title',
 800                                                         '$entry_guid_hashed',
 801                                                         '$entry_link',
 802                                                         '$entry_timestamp_fmt',
 803                                                         '$entry_content',
 804                                                         '$entry_current_hash',
 805                                                         false,
 806                                                         NOW(),
 807                                                         '$date_feed_processed',
 808                                                         '$entry_comments',
 809                                                         '$num_comments',
 810                                                         '$entry_plugin_data',
 811                                                         '$entry_language',
 812                                                         '$entry_author')");
 813
 814                                 }
 815
 816                                 // now it should exist, if not - bad luck then
 817
 818                                 $result = db_query("SELECT id FROM ttrss_entries
 819                                         WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
 820
 821                                 $entry_ref_id = 0;
 822                                 $entry_int_id = 0;
 823
 824                                 if (db_num_rows($result) == 1) {
 825
 826                                         _debug("base guid found, checking for user record", $debug_enabled);
 827
 828                                         $ref_id = db_fetch_result($result, 0, "id");
 829                                         $entry_ref_id = $ref_id;
 830
 831                                         /* $stored_guid = db_fetch_result($result, 0, "guid");
 832                                         if ($stored_guid != $entry_guid_hashed) {
 833                                                 if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
 834
 835                                                 db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
 836                                                         id = '$ref_id'");
 837                                         } */
 838
 839                                         if (RSSUtils::find_article_filter($article_filters, "filter")) {
 840                                                 //db_query("COMMIT"); // close transaction in progress
 841                                                 continue;
 842                                         }
 843
 844                                         $score = RSSUtils::calculate_article_score($article_filters) + $entry_score_modifier;
 845
 846                                         _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
 847
 848                                         // check for user post link to main table
 849
 850                                         $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
 851                                                         ref_id = '$ref_id' AND owner_uid = '$owner_uid'";
 852
 853 //                                      if ($_REQUEST["xdebug"]) print "$query\n";
 854
 855                                         $result = db_query($query);
 856
 857                                         // okay it doesn't exist - create user entry
 858                                         if (db_num_rows($result) == 0) {
 859
 860                                                 _debug("user record not found, creating...", $debug_enabled);
 861
 862                                                 if ($score >= -500 && !RSSUtils::find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
 863                                                         $unread = 'true';
 864                                                         $last_read_qpart = 'NULL';
 865                                                 } else {
 866                                                         $unread = 'false';
 867                                                         $last_read_qpart = 'NOW()';
 868                                                 }
 869
 870                                                 if (RSSUtils::find_article_filter($article_filters, 'mark') || $score > 1000) {
 871                                                         $marked = 'true';
 872                                                 } else {
 873                                                         $marked = 'false';
 874                                                 }
 875
 876                                                 if (RSSUtils::find_article_filter($article_filters, 'publish')) {
 877                                                         $published = 'true';
 878                                                 } else {
 879                                                         $published = 'false';
 880                                                 }
 881
 882                                                 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
 883                                                 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
 884
 885                                                 $result = db_query(
 886                                                         "INSERT INTO ttrss_user_entries
 887                                                                 (ref_id, owner_uid, feed_id, unread, last_read, marked,
 888                                                                 published, score, tag_cache, label_cache, uuid,
 889                                                                 last_marked, last_published)
 890                                                         VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
 891                                                                 $last_read_qpart, $marked, $published, '$score', '', '',
 892                                                                 '', $last_marked, $last_published)");
 893
 894                                                 $result = db_query(
 895                                                         "SELECT int_id FROM ttrss_user_entries WHERE
 896                                                                 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
 897                                                                 feed_id = '$feed' LIMIT 1");
 898
 899                                                 if (db_num_rows($result) == 1) {
 900                                                         $entry_int_id = db_fetch_result($result, 0, "int_id");
 901                                                 }
 902                                         } else {
 903                                                 _debug("user record FOUND", $debug_enabled);
 904
 905                                                 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
 906                                                 $entry_int_id = db_fetch_result($result, 0, "int_id");
 907                                         }
 908
 909                                         _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
 910
 911                                         if (DB_TYPE == "pgsql") {
 912                                                 $tsvector_combined = db_escape_string(mb_substr($entry_title . ' ' . strip_tags(str_replace('<', ' <', $entry_content)),
 913                                                         0, 1000000));
 914
 915                                                 $tsvector_qpart = "tsvector_combined = to_tsvector('$feed_language', '$tsvector_combined'),";
 916
 917                                         } else {
 918                                                 $tsvector_qpart = "";
 919                                         }
 920
 921                                         db_query("UPDATE ttrss_entries
 922                                                 SET title = '$entry_title',
 923                                                         content = '$entry_content',
 924                                                         content_hash = '$entry_current_hash',
 925                                                         updated = '$entry_timestamp_fmt',
 926                                                         $tsvector_qpart
 927                                                         num_comments = '$num_comments',
 928                                                         plugin_data = '$entry_plugin_data',
 929                                                         author = '$entry_author',
 930                                                         lang = '$entry_language'
 931                                                 WHERE id = '$ref_id'");
 932
 933                                         // update aux data
 934                                         db_query("UPDATE ttrss_user_entries
 935                                                         SET score = '$score' WHERE ref_id = '$ref_id'");
 936
 937                                         if ($mark_unread_on_update) {
 938                                                 _debug("article updated, marking unread as requested.", $debug_enabled);
 939
 940                                                 db_query("UPDATE ttrss_user_entries
 941                                                         SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
 942                                         }
 943                                 }
 944
 945                                 //db_query("COMMIT");
 946
 947                                 _debug("assigning labels [other]...", $debug_enabled);
 948
 949                                 foreach ($article_labels as $label) {
 950                                         Labels::add_article($entry_ref_id, $label[1], $owner_uid);
 951                                 }
 952
 953                                 _debug("assigning labels [filters]...", $debug_enabled);
 954
 955                                 RSSUtils::assign_article_to_label_filters($entry_ref_id, $article_filters,
 956                                         $owner_uid, $article_labels);
 957
 958                                 _debug("looking for enclosures...", $debug_enabled);
 959
 960                                 // enclosures
 961
 962                                 $enclosures = array();
 963
 964                                 $encs = $item->get_enclosures();
 965
 966                                 if (is_array($encs)) {
 967                                         foreach ($encs as $e) {
 968                                                 $e_item = array(
 969                                                         rewrite_relative_url($site_url, $e->link),
 970                                                         $e->type, $e->length, $e->title, $e->width, $e->height);
 971                                                 array_push($enclosures, $e_item);
 972                                         }
 973                                 }
 974
 975                                 if ($cache_images && is_writable(CACHE_DIR . '/images'))
 976                                         RSSUtils::cache_enclosures($enclosures, $site_url, $debug_enabled);
 977
 978                                 if ($debug_enabled) {
 979                                         _debug("article enclosures:", $debug_enabled);
 980                                         print_r($enclosures);
 981                                 }
 982
 983                                 //db_query("BEGIN");
 984
 985 //                              debugging
 986 //                              db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
 987
 988                                 foreach ($enclosures as $enc) {
 989                                         $enc_url = db_escape_string($enc[0]);
 990                                         $enc_type = db_escape_string($enc[1]);
 991                                         $enc_dur = db_escape_string($enc[2]);
 992                                         $enc_title = db_escape_string($enc[3]);
 993                                         $enc_width = intval($enc[4]);
 994                                         $enc_height = intval($enc[5]);
 995
 996                                         $result = db_query("SELECT id FROM ttrss_enclosures
 997                                                 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
 998
 999                                         if (db_num_rows($result) == 0) {
1000                                                 db_query("INSERT INTO ttrss_enclosures
1001                                                         (content_url, content_type, title, duration, post_id, width, height) VALUES
1002                                                         ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
1003                                         }
1004                                 }
1005
1006                                 //db_query("COMMIT");
1007
1008                                 // check for manual tags (we have to do it here since they're loaded from filters)
1009
1010                                 foreach ($article_filters as $f) {
1011                                         if ($f["type"] == "tag") {
1012
1013                                                 $manual_tags = trim_array(explode(",", $f["param"]));
1014
1015                                                 foreach ($manual_tags as $tag) {
1016                                                         if (tag_is_valid($tag)) {
1017                                                                 array_push($entry_tags, $tag);
1018                                                         }
1019                                                 }
1020                                         }
1021                                 }
1022
1023                                 // Skip boring tags
1024
1025                                 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
1026                                         'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1027
1028                                 $filtered_tags = array();
1029                                 $tags_to_cache = array();
1030
1031                                 if ($entry_tags && is_array($entry_tags)) {
1032                                         foreach ($entry_tags as $tag) {
1033                                                 if (array_search($tag, $boring_tags) === false) {
1034                                                         array_push($filtered_tags, $tag);
1035                                                 }
1036                                         }
1037                                 }
1038
1039                                 $filtered_tags = array_unique($filtered_tags);
1040
1041                                 if ($debug_enabled) {
1042                                         _debug("filtered article tags:", $debug_enabled);
1043                                         print_r($filtered_tags);
1044                                 }
1045
1046                                 // Save article tags in the database
1047
1048                                 if (count($filtered_tags) > 0) {
1049
1050                                         //db_query("BEGIN");
1051
1052                                         foreach ($filtered_tags as $tag) {
1053
1054                                                 $tag = sanitize_tag($tag);
1055                                                 $tag = db_escape_string($tag);
1056
1057                                                 if (!tag_is_valid($tag)) continue;
1058
1059                                                 $result = db_query("SELECT id FROM ttrss_tags
1060                                                         WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1061                                                         owner_uid = '$owner_uid' LIMIT 1");
1062
1063                                                 if ($result && db_num_rows($result) == 0) {
1064
1065                                                         db_query("INSERT INTO ttrss_tags
1066                                                                         (owner_uid,tag_name,post_int_id)
1067                                                                         VALUES ('$owner_uid','$tag', '$entry_int_id')");
1068                                                 }
1069
1070                                                 array_push($tags_to_cache, $tag);
1071                                         }
1072
1073                                         /* update the cache */
1074
1075                                         $tags_to_cache = array_unique($tags_to_cache);
1076
1077                                         $tags_str = db_escape_string(join(",", $tags_to_cache));
1078
1079                                         db_query("UPDATE ttrss_user_entries
1080                                                 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1081                                                 AND owner_uid = $owner_uid");
1082
1083                                         //db_query("COMMIT");
1084                                 }
1085
1086                                 _debug("article processed", $debug_enabled);
1087                         }
1088
1089                         _debug("purging feed...", $debug_enabled);
1090
1091                         purge_feed($feed, 0, $debug_enabled);
1092
1093                         db_query("UPDATE ttrss_feeds
1094                                 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
1095
1096 //                      db_query("COMMIT");
1097
1098                 } else {
1099
1100                         $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
1101
1102                         _debug("fetch error: $error_msg", $debug_enabled);
1103
1104                         if (count($rss->errors()) > 1) {
1105                                 foreach ($rss->errors() as $error) {
1106                                         _debug("+ $error");
1107                                 }
1108                         }
1109
1110                         db_query(
1111                                 "UPDATE ttrss_feeds SET last_error = '$error_msg',
1112                                 last_updated = NOW() WHERE id = '$feed'");
1113
1114                         unset($rss);
1115                         return;
1116                 }
1117
1118                 _debug("done", $debug_enabled);
1119
1120                 return true;
1121         }
1122
1123         static function cache_enclosures($enclosures, $site_url, $debug) {
1124                 foreach ($enclosures as $enc) {
1125
1126                         if (preg_match("/(image|audio|video)/", $enc[1])) {
1127
1128                                 $src = rewrite_relative_url($site_url, $enc[0]);
1129
1130                                 $local_filename = CACHE_DIR . "/images/" . sha1($src);
1131
1132                                 if ($debug) _debug("cache_enclosures: downloading: $src to $local_filename");
1133
1134                                 if (!file_exists($local_filename)) {
1135                                         $file_content = fetch_file_contents($src);
1136
1137                                         if ($file_content && strlen($file_content) > MIN_CACHE_FILE_SIZE) {
1138                                                 file_put_contents($local_filename, $file_content);
1139                                         }
1140                                 } else {
1141                                         touch($local_filename);
1142                                 }
1143                         }
1144                 }
1145         }
1146
1147         static function cache_media($html, $site_url, $debug) {
1148                 libxml_use_internal_errors(true);
1149
1150                 $charset_hack = '<head>
1151                         <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1152                 </head>';
1153
1154                 $doc = new DOMDocument();
1155                 $doc->loadHTML($charset_hack . $html);
1156                 $xpath = new DOMXPath($doc);
1157
1158                 $entries = $xpath->query('(//img[@src])|(//video/source[@src])|(//audio/source[@src])');
1159
1160                 foreach ($entries as $entry) {
1161                         if ($entry->hasAttribute('src') && strpos($entry->getAttribute('src'), "data:") !== 0) {
1162                                 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1163
1164                                 $local_filename = CACHE_DIR . "/images/" . sha1($src);
1165
1166                                 if ($debug) _debug("cache_media: downloading: $src to $local_filename");
1167
1168                                 if (!file_exists($local_filename)) {
1169                                         $file_content = fetch_file_contents($src);
1170
1171                                         if ($file_content && strlen($file_content) > MIN_CACHE_FILE_SIZE) {
1172                                                 file_put_contents($local_filename, $file_content);
1173                                         }
1174                                 } else {
1175                                         touch($local_filename);
1176                                 }
1177                         }
1178                 }
1179         }
1180
1181         static function expire_error_log($debug) {
1182                 if ($debug) _debug("Removing old error log entries...");
1183
1184                 if (DB_TYPE == "pgsql") {
1185                         db_query("DELETE FROM ttrss_error_log
1186                                 WHERE created_at < NOW() - INTERVAL '7 days'");
1187                 } else {
1188                         db_query("DELETE FROM ttrss_error_log
1189                                 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1190                 }
1191
1192         }
1193
1194         static function expire_lock_files($debug) {
1195                 //if ($debug) _debug("Removing old lock files...");
1196
1197                 $num_deleted = 0;
1198
1199                 if (is_writable(LOCK_DIRECTORY)) {
1200                         $files = glob(LOCK_DIRECTORY . "/*.lock");
1201
1202                         if ($files) {
1203                                 foreach ($files as $file) {
1204                                         if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
1205                                                 unlink($file);
1206                                                 ++$num_deleted;
1207                                         }
1208                                 }
1209                         }
1210                 }
1211
1212                 if ($debug) _debug("Removed $num_deleted old lock files.");
1213         }
1214
1215         static function expire_cached_files($debug) {
1216                 foreach (array("simplepie", "images", "export", "upload") as $dir) {
1217                         $cache_dir = CACHE_DIR . "/$dir";
1218
1219 //                      if ($debug) _debug("Expiring $cache_dir");
1220
1221                         $num_deleted = 0;
1222
1223                         if (is_writable($cache_dir)) {
1224                                 $files = glob("$cache_dir/*");
1225
1226                                 if ($files) {
1227                                         foreach ($files as $file) {
1228                                                 if (time() - filemtime($file) > 86400*CACHE_MAX_DAYS) {
1229                                                         unlink($file);
1230
1231                                                         ++$num_deleted;
1232                                                 }
1233                                         }
1234                                 }
1235                         }
1236
1237                         if ($debug) _debug("$cache_dir: removed $num_deleted files.");
1238                 }
1239         }
1240
1241         /**
1242          * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1243          * Returns the url query as associative array
1244          *
1245          * @param    string    query
1246          * @return    array    params
1247          */
1248         static function convertUrlQuery($query) {
1249                 $queryParts = explode('&', $query);
1250
1251                 $params = array();
1252
1253                 foreach ($queryParts as $param) {
1254                         $item = explode('=', $param);
1255                         $params[$item[0]] = $item[1];
1256                 }
1257
1258                 return $params;
1259         }
1260
1261         static function get_article_filters($filters, $title, $content, $link, $author, $tags, &$matched_rules = false) {
1262                 $matches = array();
1263
1264                 foreach ($filters as $filter) {
1265                         $match_any_rule = $filter["match_any_rule"];
1266                         $inverse = $filter["inverse"];
1267                         $filter_match = false;
1268
1269                         foreach ($filter["rules"] as $rule) {
1270                                 $match = false;
1271                                 $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
1272                                 $rule_inverse = $rule["inverse"];
1273
1274                                 if (!$reg_exp)
1275                                         continue;
1276
1277                                 switch ($rule["type"]) {
1278                                         case "title":
1279                                                 $match = @preg_match("/$reg_exp/iu", $title);
1280                                                 break;
1281                                         case "content":
1282                                                 // we don't need to deal with multiline regexps
1283                                                 $content = preg_replace("/[\r\n\t]/", "", $content);
1284
1285                                                 $match = @preg_match("/$reg_exp/iu", $content);
1286                                                 break;
1287                                         case "both":
1288                                                 // we don't need to deal with multiline regexps
1289                                                 $content = preg_replace("/[\r\n\t]/", "", $content);
1290
1291                                                 $match = (@preg_match("/$reg_exp/iu", $title) || @preg_match("/$reg_exp/iu", $content));
1292                                                 break;
1293                                         case "link":
1294                                                 $match = @preg_match("/$reg_exp/iu", $link);
1295                                                 break;
1296                                         case "author":
1297                                                 $match = @preg_match("/$reg_exp/iu", $author);
1298                                                 break;
1299                                         case "tag":
1300                                                 foreach ($tags as $tag) {
1301                                                         if (@preg_match("/$reg_exp/iu", $tag)) {
1302                                                                 $match = true;
1303                                                                 break;
1304                                                         }
1305                                                 }
1306                                                 break;
1307                                 }
1308
1309                                 if ($rule_inverse) $match = !$match;
1310
1311                                 if ($match_any_rule) {
1312                                         if ($match) {
1313                                                 $filter_match = true;
1314                                                 break;
1315                                         }
1316                                 } else {
1317                                         $filter_match = $match;
1318                                         if (!$match) {
1319                                                 break;
1320                                         }
1321                                 }
1322                         }
1323
1324                         if ($inverse) $filter_match = !$filter_match;
1325
1326                         if ($filter_match) {
1327                                 if (is_array($matched_rules)) array_push($matched_rules, $rule);
1328
1329                                 foreach ($filter["actions"] AS $action) {
1330                                         array_push($matches, $action);
1331
1332                                         // if Stop action encountered, perform no further processing
1333                                         if (isset($action["type"]) && $action["type"] == "stop") return $matches;
1334                                 }
1335                         }
1336                 }
1337
1338                 return $matches;
1339         }
1340
1341         static function find_article_filter($filters, $filter_name) {
1342                 foreach ($filters as $f) {
1343                         if ($f["type"] == $filter_name) {
1344                                 return $f;
1345                         };
1346                 }
1347                 return false;
1348         }
1349
1350         static function find_article_filters($filters, $filter_name) {
1351                 $results = array();
1352
1353                 foreach ($filters as $f) {
1354                         if ($f["type"] == $filter_name) {
1355                                 array_push($results, $f);
1356                         };
1357                 }
1358                 return $results;
1359         }
1360
1361         static function calculate_article_score($filters) {
1362                 $score = 0;
1363
1364                 foreach ($filters as $f) {
1365                         if ($f["type"] == "score") {
1366                                 $score += $f["param"];
1367                         };
1368                 }
1369                 return $score;
1370         }
1371
1372         static function labels_contains_caption($labels, $caption) {
1373                 foreach ($labels as $label) {
1374                         if ($label[1] == $caption) {
1375                                 return true;
1376                         }
1377                 }
1378
1379                 return false;
1380         }
1381
1382         static function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
1383                 foreach ($filters as $f) {
1384                         if ($f["type"] == "label") {
1385                                 if (!RSSUtils::labels_contains_caption($article_labels, $f["param"])) {
1386                                         Labels::add_article($id, $f["param"], $owner_uid);
1387                                 }
1388                         }
1389                 }
1390         }
1391
1392         static function make_guid_from_title($title) {
1393                 return preg_replace("/[ \"\',.:;]/", "-",
1394                         mb_strtolower(strip_tags($title), 'utf-8'));
1395         }
1396
1397         static function cleanup_counters_cache($debug) {
1398                 $result = db_query("DELETE FROM ttrss_counters_cache
1399                         WHERE feed_id > 0 AND
1400                         (SELECT COUNT(id) FROM ttrss_feeds WHERE
1401                                 id = feed_id AND
1402                                 ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
1403                 $frows = db_affected_rows($result);
1404
1405                 $result = db_query("DELETE FROM ttrss_cat_counters_cache
1406                         WHERE feed_id > 0 AND
1407                         (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
1408                                 id = feed_id AND
1409                                 ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
1410                 $crows = db_affected_rows($result);
1411
1412                 if ($debug) _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
1413         }
1414
1415         static function housekeeping_user($owner_uid) {
1416                 $tmph = new PluginHost();
1417
1418                 load_user_plugins($owner_uid, $tmph);
1419
1420                 $tmph->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
1421         }
1422
1423         static function housekeeping_common($debug) {
1424                 RSSUtils::expire_cached_files($debug);
1425                 RSSUtils::expire_lock_files($debug);
1426                 RSSUtils::expire_error_log($debug);
1427
1428                 $count = RSSUtils::update_feedbrowser_cache();
1429                 _debug("Feedbrowser updated, $count feeds processed.");
1430
1431                 Article::purge_orphans( true);
1432                 RSSUtils::cleanup_counters_cache($debug);
1433
1434                 //$rc = cleanup_tags( 14, 50000);
1435                 //_debug("Cleaned $rc cached tags.");
1436
1437                 PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
1438         }
1439
1440         static function check_feed_favicon($site_url, $feed) {
1441                 #               print "FAVICON [$site_url]: $favicon_url\n";
1442
1443                 $icon_file = ICONS_DIR . "/$feed.ico";
1444
1445                 if (!file_exists($icon_file)) {
1446                         $favicon_url = get_favicon_url($site_url);
1447
1448                         if ($favicon_url) {
1449                                 // Limiting to "image" type misses those served with text/plain
1450                                 $contents = fetch_file_contents($favicon_url); // , "image");
1451
1452                                 if ($contents) {
1453                                         // Crude image type matching.
1454                                         // Patterns gleaned from the file(1) source code.
1455                                         if (preg_match('/^\x00\x00\x01\x00/', $contents)) {
1456                                                 // 0       string  \000\000\001\000        MS Windows icon resource
1457                                                 //error_log("check_feed_favicon: favicon_url=$favicon_url isa MS Windows icon resource");
1458                                         }
1459                                         elseif (preg_match('/^GIF8/', $contents)) {
1460                                                 // 0       string          GIF8            GIF image data
1461                                                 //error_log("check_feed_favicon: favicon_url=$favicon_url isa GIF image");
1462                                         }
1463                                         elseif (preg_match('/^\x89PNG\x0d\x0a\x1a\x0a/', $contents)) {
1464                                                 // 0       string          \x89PNG\x0d\x0a\x1a\x0a         PNG image data
1465                                                 //error_log("check_feed_favicon: favicon_url=$favicon_url isa PNG image");
1466                                         }
1467                                         elseif (preg_match('/^\xff\xd8/', $contents)) {
1468                                                 // 0       beshort         0xffd8          JPEG image data
1469                                                 //error_log("check_feed_favicon: favicon_url=$favicon_url isa JPG image");
1470                                         }
1471                                         else {
1472                                                 //error_log("check_feed_favicon: favicon_url=$favicon_url isa UNKNOWN type");
1473                                                 $contents = "";
1474                                         }
1475                                 }
1476
1477                                 if ($contents) {
1478                                         $fp = @fopen($icon_file, "w");
1479
1480                                         if ($fp) {
1481                                                 fwrite($fp, $contents);
1482                                                 fclose($fp);
1483                                                 chmod($icon_file, 0644);
1484                                         }
1485                                 }
1486                         }
1487                         return $icon_file;
1488                 }
1489         }
1490
1491
1492
1493 }