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) {
   8                 $ph = PluginHost::getInstance();
   9                 return sha1(implode(":", $ph->get_plugin_names()) . serialize($article));
  10         }
  11
  12         function update_feedbrowser_cache() {
  13
  14                 $result = db_query("SELECT feed_url, site_url, title, COUNT(id) AS subscribers
  15                         FROM ttrss_feeds WHERE (SELECT COUNT(id) = 0 FROM ttrss_feeds AS tf
  16                                 WHERE tf.feed_url = ttrss_feeds.feed_url
  17                                 AND (private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%'))
  18                                 GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
  19
  20                 db_query("BEGIN");
  21
  22                 db_query("DELETE FROM ttrss_feedbrowser_cache");
  23
  24                 $count = 0;
  25
  26                 while ($line = db_fetch_assoc($result)) {
  27                         $subscribers = db_escape_string($line["subscribers"]);
  28                         $feed_url = db_escape_string($line["feed_url"]);
  29                         $title = db_escape_string($line["title"]);
  30                         $site_url = db_escape_string($line["site_url"]);
  31
  32                         $tmp_result = db_query("SELECT subscribers FROM
  33                                 ttrss_feedbrowser_cache WHERE feed_url = '$feed_url'");
  34
  35                         if (db_num_rows($tmp_result) == 0) {
  36
  37                                 db_query("INSERT INTO ttrss_feedbrowser_cache
  38                                         (feed_url, site_url, title, subscribers) VALUES ('$feed_url',
  39                                                 '$site_url', '$title', '$subscribers')");
  40
  41                                 ++$count;
  42
  43                         }
  44
  45                 }
  46
  47                 db_query("COMMIT");
  48
  49                 return $count;
  50
  51         }
  52
  53
  54         /**
  55          * Update a feed batch.
  56          * Used by daemons to update n feeds by run.
  57          * Only update feed needing a update, and not being processed
  58          * by another process.
  59          *
  60          * @param mixed $link Database link
  61          * @param integer $limit Maximum number of feeds in update batch. Default to DAEMON_FEED_LIMIT.
  62          * @param boolean $from_http Set to true if you call this function from http to disable cli specific code.
  63          * @param boolean $debug Set to false to disable debug output. Default to true.
  64          * @return void
  65          */
  66         function update_daemon_common($limit = DAEMON_FEED_LIMIT, $from_http = false, $debug = true) {
  67                 // Process all other feeds using last_updated and interval parameters
  68
  69                 $schema_version = get_schema_version();
  70
  71                 if ($schema_version != SCHEMA_VERSION) {
  72                         die("Schema version is wrong, please upgrade the database.\n");
  73                 }
  74
  75                 define('PREFS_NO_CACHE', true);
  76
  77                 // Test if the user has loggued in recently. If not, it does not update its feeds.
  78                 if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
  79                         if (DB_TYPE == "pgsql") {
  80                                 $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
  81                         } else {
  82                                 $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
  83                         }
  84                 } else {
  85                         $login_thresh_qpart = "";
  86                 }
  87
  88                 // Test if the feed need a update (update interval exceeded).
  89                 if (DB_TYPE == "pgsql") {
  90                         $update_limit_qpart = "AND ((
  91                                         ttrss_feeds.update_interval = 0
  92                                         AND ttrss_user_prefs.value != '-1'
  93                                         AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
  94                                 ) OR (
  95                                         ttrss_feeds.update_interval > 0
  96                                         AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
  97                                 ) OR (ttrss_feeds.last_updated IS NULL
  98                                         AND ttrss_user_prefs.value != '-1')
  99                                 OR (last_updated = '1970-01-01 00:00:00'
 100                                         AND ttrss_user_prefs.value != '-1'))";
 101                 } else {
 102                         $update_limit_qpart = "AND ((
 103                                         ttrss_feeds.update_interval = 0
 104                                         AND ttrss_user_prefs.value != '-1'
 105                                         AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
 106                                 ) OR (
 107                                         ttrss_feeds.update_interval > 0
 108                                         AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
 109                                 ) OR (ttrss_feeds.last_updated IS NULL
 110                                         AND ttrss_user_prefs.value != '-1')
 111                                 OR (last_updated = '1970-01-01 00:00:00'
 112                                         AND ttrss_user_prefs.value != '-1'))";
 113                 }
 114
 115                 // Test if feed is currently being updated by another process.
 116                 if (DB_TYPE == "pgsql") {
 117                         $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
 118                 } else {
 119                         $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 10 MINUTE))";
 120                 }
 121
 122                 // Test if there is a limit to number of updated feeds
 123                 $query_limit = "";
 124                 if($limit) $query_limit = sprintf("LIMIT %d", $limit);
 125
 126                 $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
 127                         FROM
 128                                 ttrss_feeds, ttrss_users, ttrss_user_prefs
 129                         WHERE
 130                                 ttrss_feeds.owner_uid = ttrss_users.id
 131                                 AND ttrss_user_prefs.profile IS NULL
 132                                 AND ttrss_users.id = ttrss_user_prefs.owner_uid
 133                                 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
 134                                 $login_thresh_qpart $update_limit_qpart
 135                                 $updstart_thresh_qpart
 136                                 ORDER BY last_updated $query_limit";
 137
 138                 // We search for feed needing update.
 139                 $result = db_query($query);
 140
 141                 if($debug) _debug(sprintf("Scheduled %d feeds to update...", db_num_rows($result)));
 142
 143                 // Here is a little cache magic in order to minimize risk of double feed updates.
 144                 $feeds_to_update = array();
 145                 while ($line = db_fetch_assoc($result)) {
 146                         array_push($feeds_to_update, db_escape_string($line['feed_url']));
 147                 }
 148
 149                 // We update the feed last update started date before anything else.
 150                 // There is no lag due to feed contents downloads
 151                 // It prevent an other process to update the same feed.
 152
 153                 if(count($feeds_to_update) > 0) {
 154                         $feeds_quoted = array();
 155
 156                         foreach ($feeds_to_update as $feed) {
 157                                 array_push($feeds_quoted, "'" . db_escape_string($feed) . "'");
 158                         }
 159
 160                         db_query(sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
 161                                 WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
 162                 }
 163
 164                 $nf = 0;
 165
 166                 // For each feed, we call the feed update function.
 167                 foreach ($feeds_to_update as $feed) {
 168                         if($debug) _debug("Base feed: $feed");
 169
 170                         //update_rss_feed($line["id"], true);
 171
 172                         // since we have the data cached, we can deal with other feeds with the same url
 173
 174                         $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
 175                         FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
 176                                 ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
 177                                 ttrss_users.id = ttrss_user_prefs.owner_uid AND
 178                                 ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
 179                                 ttrss_user_prefs.profile IS NULL AND
 180                                 feed_url = '".db_escape_string($feed)."' AND
 181                                 (ttrss_feeds.update_interval > 0 OR
 182                                         ttrss_user_prefs.value != '-1')
 183                                 $login_thresh_qpart
 184                         ORDER BY ttrss_feeds.id $query_limit");
 185
 186                         if (db_num_rows($tmp_result) > 0) {
 187                                 $rss = false;
 188
 189                                 while ($tline = db_fetch_assoc($tmp_result)) {
 190                                         if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
 191
 192                                         $rss = update_rss_feed($tline["id"], true, false);
 193                                         _debug_suppress(false);
 194                                         ++$nf;
 195                                 }
 196                         }
 197                 }
 198
 199                 require_once "digest.php";
 200
 201                 // Send feed digests by email if needed.
 202                 send_headlines_digests($debug);
 203
 204                 return $nf;
 205
 206         } // function update_daemon_common
 207
 208         // ignore_daemon is not used
 209         function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
 210
 211                 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
 212
 213                 _debug_suppress(!$debug_enabled);
 214                 _debug("start", $debug_enabled);
 215
 216                 $result = db_query("SELECT id,update_interval,auth_login,
 217                         feed_url,auth_pass,cache_images,
 218                         mark_unread_on_update, owner_uid,
 219                         pubsub_state, auth_pass_encrypted,
 220                         (SELECT max(date_entered) FROM
 221                                 ttrss_entries, ttrss_user_entries where ref_id = id AND feed_id = '$feed') AS last_article_timestamp
 222                         FROM ttrss_feeds WHERE id = '$feed'");
 223
 224                 if (db_num_rows($result) == 0) {
 225                         _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
 226                         return false;
 227                 }
 228
 229                 $last_article_timestamp = @strtotime(db_fetch_result($result, 0, "last_article_timestamp"));
 230
 231                 if (defined('_DISABLE_HTTP_304'))
 232                         $last_article_timestamp = 0;
 233
 234                 $owner_uid = db_fetch_result($result, 0, "owner_uid");
 235                 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
 236                         0, "mark_unread_on_update"));
 237                 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
 238                 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
 239                         0, "auth_pass_encrypted"));
 240
 241                 db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
 242                         WHERE id = '$feed'");
 243
 244                 $auth_login = db_fetch_result($result, 0, "auth_login");
 245                 $auth_pass = db_fetch_result($result, 0, "auth_pass");
 246
 247                 if ($auth_pass_encrypted) {
 248                         require_once "crypt.php";
 249                         $auth_pass = decrypt_string($auth_pass);
 250                 }
 251
 252                 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
 253                 $fetch_url = db_fetch_result($result, 0, "feed_url");
 254
 255                 $feed = db_escape_string($feed);
 256
 257                 $date_feed_processed = date('Y-m-d H:i');
 258
 259                 $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
 260
 261                 $pluginhost = new PluginHost();
 262                 $pluginhost->set_debug($debug_enabled);
 263                 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
 264
 265                 $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
 266                 $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
 267                 $pluginhost->load_data();
 268
 269                 if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
 270                         _debug("using previously initialized parser object");
 271                 } else {
 272                         $rss_hash = false;
 273
 274                         $force_refetch = isset($_REQUEST["force_refetch"]);
 275
 276                         foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
 277                                 $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, $last_article_timestamp, $auth_login, $auth_pass);
 278                         }
 279
 280                         // try cache
 281                         if (!$feed_data &&
 282                                 file_exists($cache_filename) &&
 283                                 is_readable($cache_filename) &&
 284                                 !$auth_login && !$auth_pass &&
 285                                 filemtime($cache_filename) > time() - 30) {
 286
 287                                 _debug("using local cache [$cache_filename].", $debug_enabled);
 288
 289                                 @$feed_data = file_get_contents($cache_filename);
 290
 291                                 if ($feed_data) {
 292                                         $rss_hash = sha1($feed_data);
 293                                 }
 294
 295                         } else {
 296                                 _debug("local cache will not be used for this feed", $debug_enabled);
 297                         }
 298
 299                         // fetch feed from source
 300                         if (!$feed_data) {
 301                                 _debug("fetching [$fetch_url]...", $debug_enabled);
 302                                 _debug("If-Modified-Since: ".gmdate('D, d M Y H:i:s \G\M\T', $last_article_timestamp), $debug_enabled);
 303
 304                                 $feed_data = fetch_file_contents($fetch_url, false,
 305                                         $auth_login, $auth_pass, false,
 306                                         $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
 307                                         $force_refetch ? 0 : $last_article_timestamp);
 308
 309                                 global $fetch_curl_used;
 310
 311                                 if (!$fetch_curl_used) {
 312                                         $tmp = @gzdecode($feed_data);
 313
 314                                         if ($tmp) $feed_data = $tmp;
 315                                 }
 316
 317                                 $feed_data = trim($feed_data);
 318
 319                                 _debug("fetch done.", $debug_enabled);
 320
 321                                 /* if ($feed_data) {
 322                                         $error = verify_feed_xml($feed_data);
 323
 324                                         if ($error) {
 325                                                 _debug("error verifying XML, code: " . $error->code, $debug_enabled);
 326
 327                                                 if ($error->code == 26) {
 328                                                         _debug("got error 26, trying to decode entities...", $debug_enabled);
 329
 330                                                         $feed_data = html_entity_decode($feed_data, ENT_COMPAT, 'UTF-8');
 331
 332                                                         $error = verify_feed_xml($feed_data);
 333
 334                                                         if ($error) $feed_data = '';
 335                                                 }
 336                                         }
 337                                 } */
 338
 339                                 // cache vanilla feed data for re-use
 340                                 if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
 341                                         $new_rss_hash = sha1($feed_data);
 342
 343                                         if ($new_rss_hash != $rss_hash) {
 344                                                 _debug("saving $cache_filename", $debug_enabled);
 345                                                 @file_put_contents($cache_filename, $feed_data);
 346                                         }
 347                                 }
 348                         }
 349
 350                         if (!$feed_data) {
 351                                 global $fetch_last_error;
 352                                 global $fetch_last_error_code;
 353
 354                                 _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
 355
 356                                 $error_escaped = '';
 357
 358                                 // If-Modified-Since
 359                                 if ($fetch_last_error_code != 304) {
 360                                         $error_escaped = db_escape_string($fetch_last_error);
 361                                 } else {
 362                                         _debug("source claims data not modified, nothing to do.", $debug_enabled);
 363                                 }
 364
 365                                 db_query(
 366                                         "UPDATE ttrss_feeds SET last_error = '$error_escaped',
 367                                                 last_updated = NOW() WHERE id = '$feed'");
 368
 369                                 return;
 370                         }
 371                 }
 372
 373                 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
 374                         $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
 375                 }
 376
 377                 // set last update to now so if anything *simplepie* crashes later we won't be
 378                 // continuously failing on the same feed
 379                 //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
 380
 381                 if (!$rss) {
 382                         $rss = new FeedParser($feed_data);
 383                         $rss->init();
 384                 }
 385
 386                 if (DETECT_ARTICLE_LANGUAGE) {
 387                         require_once "lib/languagedetect/LanguageDetect.php";
 388
 389                         $lang = new Text_LanguageDetect();
 390                         $lang->setNameMode(2);
 391                 }
 392
 393 //              print_r($rss);
 394
 395                 $feed = db_escape_string($feed);
 396
 397                 if (!$rss->error()) {
 398
 399                         // We use local pluginhost here because we need to load different per-user feed plugins
 400                         $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
 401
 402                         _debug("processing feed data...", $debug_enabled);
 403
 404 //                      db_query("BEGIN");
 405
 406                         if (DB_TYPE == "pgsql") {
 407                                 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
 408                         } else {
 409                                 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
 410                         }
 411
 412                         $result = db_query("SELECT title,site_url,owner_uid,favicon_avg_color,
 413                                 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
 414                                                 favicon_needs_check
 415                                 FROM ttrss_feeds WHERE id = '$feed'");
 416
 417                         $registered_title = db_fetch_result($result, 0, "title");
 418                         $orig_site_url = db_fetch_result($result, 0, "site_url");
 419                         $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
 420                                 "favicon_needs_check"));
 421                         $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
 422
 423                         $owner_uid = db_fetch_result($result, 0, "owner_uid");
 424
 425                         $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
 426
 427                         _debug("site_url: $site_url", $debug_enabled);
 428                         _debug("feed_title: " . $rss->get_title(), $debug_enabled);
 429
 430                         if ($favicon_needs_check || $force_refetch) {
 431
 432                                 /* terrible hack: if we crash on floicon shit here, we won't check
 433                                  * the icon avgcolor again (unless the icon got updated) */
 434
 435                                 $favicon_file = ICONS_DIR . "/$feed.ico";
 436                                 $favicon_modified = @filemtime($favicon_file);
 437
 438                                 _debug("checking favicon...", $debug_enabled);
 439
 440                                 check_feed_favicon($site_url, $feed);
 441                                 $favicon_modified_new = @filemtime($favicon_file);
 442
 443                                 if ($favicon_modified_new > $favicon_modified)
 444                                         $favicon_avg_color = '';
 445
 446                                 if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
 447                                                 require_once "colors.php";
 448
 449                                                 db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
 450                                                         id = '$feed'");
 451
 452                                                 $favicon_color = db_escape_string(
 453                                                         calculate_avg_color($favicon_file));
 454
 455                                                 $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
 456                                 } else if ($favicon_avg_color == 'fail') {
 457                                         _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
 458                                 }
 459
 460                                 db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
 461                                         $favicon_colorstring
 462                                         WHERE id = '$feed'");
 463                         }
 464
 465                         if (!$registered_title || $registered_title == "[Unknown]") {
 466
 467                                 $feed_title = db_escape_string($rss->get_title());
 468
 469                                 if ($feed_title) {
 470                                         _debug("registering title: $feed_title", $debug_enabled);
 471
 472                                         db_query("UPDATE ttrss_feeds SET
 473                                                 title = '$feed_title' WHERE id = '$feed'");
 474                                 }
 475                         }
 476
 477                         if ($site_url && $orig_site_url != $site_url) {
 478                                 db_query("UPDATE ttrss_feeds SET
 479                                         site_url = '$site_url' WHERE id = '$feed'");
 480                         }
 481
 482                         _debug("loading filters & labels...", $debug_enabled);
 483
 484                         $filters = load_filters($feed, $owner_uid);
 485                         $labels = get_all_labels($owner_uid);
 486
 487                         _debug("" . count($filters) . " filters loaded.", $debug_enabled);
 488
 489                         $items = $rss->get_items();
 490
 491                         if (!is_array($items)) {
 492                                 _debug("no articles found.", $debug_enabled);
 493
 494                                 db_query("UPDATE ttrss_feeds
 495                                         SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
 496
 497                                 return; // no articles
 498                         }
 499
 500                         if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
 501
 502                                 _debug("checking for PUSH hub...", $debug_enabled);
 503
 504                                 $feed_hub_url = false;
 505
 506                                 $links = $rss->get_links('hub');
 507
 508                                 if ($links && is_array($links)) {
 509                                         foreach ($links as $l) {
 510                                                 $feed_hub_url = $l;
 511                                                 break;
 512                                         }
 513                                 }
 514
 515                                 _debug("feed hub url: $feed_hub_url", $debug_enabled);
 516
 517                                 $feed_self_url = $fetch_url;
 518
 519                                 $links = $rss->get_links('self');
 520
 521                                 if ($links && is_array($links)) {
 522                                         foreach ($links as $l) {
 523                                                 $feed_self_url = $l;
 524                                                 break;
 525                                         }
 526                                 }
 527
 528                                 _debug("feed self url = $feed_self_url");
 529
 530                                 if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
 531                                         !ini_get("open_basedir")) {
 532
 533                                         require_once 'lib/pubsubhubbub/subscriber.php';
 534
 535                                         $callback_url = get_self_url_prefix() .
 536                                                 "/public.php?op=pubsub&id=$feed";
 537
 538                                         $s = new Subscriber($feed_hub_url, $callback_url);
 539
 540                                         $rc = $s->subscribe($feed_self_url);
 541
 542                                         _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
 543
 544                                         db_query("UPDATE ttrss_feeds SET pubsub_state = 1
 545                                                 WHERE id = '$feed'");
 546                                 }
 547                         }
 548
 549                         _debug("processing articles...", $debug_enabled);
 550
 551                         foreach ($items as $item) {
 552                                 if ($_REQUEST['xdebug'] == 3) {
 553                                         print_r($item);
 554                                 }
 555
 556                                 $entry_guid = $item->get_id();
 557                                 if (!$entry_guid) $entry_guid = $item->get_link();
 558                                 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
 559
 560                                 _debug("f_guid $entry_guid", $debug_enabled);
 561
 562                                 if (!$entry_guid) continue;
 563
 564                                 $entry_guid = "$owner_uid,$entry_guid";
 565
 566                                 $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
 567
 568                                 _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
 569
 570                                 $entry_timestamp = "";
 571
 572                                 $entry_timestamp = $item->get_date();
 573
 574                                 _debug("orig date: " . $item->get_date(), $debug_enabled);
 575
 576                                 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
 577                                         $entry_timestamp = time();
 578                                 }
 579
 580                                 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
 581
 582                                 _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
 583
 584 //                              $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
 585 //                              $entry_title = decode_numeric_entities($entry_title);
 586                                 $entry_title = $item->get_title();
 587
 588                                 $entry_link = rewrite_relative_url($site_url, $item->get_link());
 589
 590                                 _debug("title $entry_title", $debug_enabled);
 591                                 _debug("link $entry_link", $debug_enabled);
 592
 593                                 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
 594
 595                                 $entry_content = $item->get_content();
 596                                 if (!$entry_content) $entry_content = $item->get_description();
 597
 598                                 if ($_REQUEST["xdebug"] == 2) {
 599                                         print "content: ";
 600                                         print $entry_content;
 601                                         print "\n";
 602                                 }
 603
 604                                 $entry_language = "";
 605
 606                                 if (DETECT_ARTICLE_LANGUAGE) {
 607                                         $entry_language = $lang->detect($entry_title . " " . $entry_content, 1);
 608
 609                                         if (count($entry_language) > 0) {
 610                                                 $entry_language = array_keys($entry_language);
 611
 612                                                 // the fuck?
 613                                                 if (is_array($entry_language))
 614                                                         $entry_language = "";
 615                                                 else
 616                                                         $entry_language = db_escape_string(substr($entry_language[0], 0, 2));
 617
 618                                                 _debug("detected language: $entry_language", $debug_enabled);
 619                                         } else {
 620                                                 $entry_language = "";
 621                                         }
 622                                 }
 623
 624                                 $entry_comments = $item->get_comments_url();
 625                                 $entry_author = $item->get_author();
 626
 627                                 $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
 628
 629                                 $entry_comments = db_escape_string(mb_substr(trim($entry_comments), 0, 245));
 630                                 $entry_author = db_escape_string(mb_substr(trim($entry_author), 0, 245));
 631
 632                                 $num_comments = (int) $item->get_comments_count();
 633
 634                                 _debug("author $entry_author", $debug_enabled);
 635                                 _debug("num_comments: $num_comments", $debug_enabled);
 636                                 _debug("looking for tags...", $debug_enabled);
 637
 638                                 // parse <category> entries into tags
 639
 640                                 $additional_tags = array();
 641
 642                                 $additional_tags_src = $item->get_categories();
 643
 644                                 if (is_array($additional_tags_src)) {
 645                                         foreach ($additional_tags_src as $tobj) {
 646                                                 array_push($additional_tags, $tobj);
 647                                         }
 648                                 }
 649
 650                                 $entry_tags = array_unique($additional_tags);
 651
 652                                 for ($i = 0; $i < count($entry_tags); $i++)
 653                                         $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
 654
 655                                 _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
 656
 657                                 _debug("done collecting data.", $debug_enabled);
 658
 659                                 $result = db_query("SELECT id, content_hash FROM ttrss_entries
 660                                         WHERE guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed'");
 661
 662                                 if (db_num_rows($result) != 0) {
 663                                         $base_entry_id = db_fetch_result($result, 0, "id");
 664                                         $entry_stored_hash = db_fetch_result($result, 0, "content_hash");
 665                                 } else {
 666                                         $base_entry_id = false;
 667                                         $entry_stored_hash = "";
 668                                 }
 669
 670                                 $article = array("owner_uid" => $owner_uid, // read only
 671                                         "guid" => $entry_guid, // read only
 672                                         "title" => $entry_title,
 673                                         "content" => $entry_content,
 674                                         "link" => $entry_link,
 675                                         "tags" => $entry_tags,
 676                                         "author" => $entry_author,
 677                                         "feed" => array("id" => $feed,
 678                                                 "fetch_url" => $fetch_url,
 679                                                 "site_url" => $site_url)
 680                                         );
 681
 682                                 $entry_plugin_data = "";
 683                                 $entry_current_hash = calculate_article_hash($article);
 684
 685                                 _debug("article hash: $entry_current_hash [stored=$entry_stored_hash]", $debug_enabled);
 686
 687                                 if ($entry_current_hash == $entry_stored_hash) {
 688                                         _debug("stored article seems up to date [IID: $base_entry_id], updating timestamp only", $debug_enabled);
 689
 690                                         // we keep encountering the entry in feeds, so we need to
 691                                         // update date_updated column so that we don't get horrible
 692                                         // dupes when the entry gets purged and reinserted again e.g.
 693                                         // in the case of SLOW SLOW OMG SLOW updating feeds
 694
 695                                         $base_entry_id = db_fetch_result($result, 0, "id");
 696
 697                                         db_query("UPDATE ttrss_entries SET date_updated = NOW()
 698                                                 WHERE id = '$base_entry_id'");
 699
 700                                         continue;
 701                                 }
 702
 703                                 _debug("hash differs, applying plugin filters:", $debug_enabled);
 704
 705                                 foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
 706                                         _debug("... " . get_class($plugin), $debug_enabled);
 707
 708                                         $start = microtime(true);
 709                                         $article = $plugin->hook_article_filter($article);
 710
 711                                         _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
 712
 713                                         $entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
 714                                 }
 715
 716                                 $entry_plugin_data = db_escape_string($entry_plugin_data);
 717
 718                                 _debug("plugin data: $entry_plugin_data", $debug_enabled);
 719
 720                                 $entry_tags = $article["tags"];
 721                                 $entry_guid = db_escape_string($entry_guid);
 722                                 $entry_title = db_escape_string($article["title"]);
 723                                 $entry_author = db_escape_string($article["author"]);
 724                                 $entry_link = db_escape_string($article["link"]);
 725                                 $entry_content = $article["content"]; // escaped below
 726
 727                                 if ($cache_images && is_writable(CACHE_DIR . '/images'))
 728                                         cache_images($entry_content, $site_url, $debug_enabled);
 729
 730                                 $entry_content = db_escape_string($entry_content, false);
 731
 732                                 db_query("BEGIN");
 733
 734                                 $result = db_query("SELECT id FROM      ttrss_entries
 735                                         WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
 736
 737                                 if (db_num_rows($result) == 0) {
 738
 739                                         _debug("base guid [$entry_guid] not found", $debug_enabled);
 740
 741                                         // base post entry does not exist, create it
 742
 743                                         $result = db_query(
 744                                                 "INSERT INTO ttrss_entries
 745                                                         (title,
 746                                                         guid,
 747                                                         link,
 748                                                         updated,
 749                                                         content,
 750                                                         content_hash,
 751                                                         no_orig_date,
 752                                                         date_updated,
 753                                                         date_entered,
 754                                                         comments,
 755                                                         num_comments,
 756                                                         plugin_data,
 757                                                         lang,
 758                                                         author)
 759                                                 VALUES
 760                                                         ('$entry_title',
 761                                                         '$entry_guid_hashed',
 762                                                         '$entry_link',
 763                                                         '$entry_timestamp_fmt',
 764                                                         '$entry_content',
 765                                                         '$entry_current_hash',
 766                                                         false,
 767                                                         NOW(),
 768                                                         '$date_feed_processed',
 769                                                         '$entry_comments',
 770                                                         '$num_comments',
 771                                                         '$entry_plugin_data',
 772                                                         '$entry_language',
 773                                                         '$entry_author')");
 774
 775                                         $article_labels = array();
 776
 777                                 } else {
 778                                         $base_entry_id = db_fetch_result($result, 0, "id");
 779
 780                                         $article_labels = get_article_labels($base_entry_id, $owner_uid);
 781                                 }
 782
 783                                 // now it should exist, if not - bad luck then
 784
 785                                 $result = db_query("SELECT id FROM ttrss_entries
 786                                         WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
 787
 788                                 $entry_ref_id = 0;
 789                                 $entry_int_id = 0;
 790
 791                                 if (db_num_rows($result) == 1) {
 792
 793                                         _debug("base guid found, checking for user record", $debug_enabled);
 794
 795                                         $ref_id = db_fetch_result($result, 0, "id");
 796                                         $entry_ref_id = $ref_id;
 797
 798                                         /* $stored_guid = db_fetch_result($result, 0, "guid");
 799                                         if ($stored_guid != $entry_guid_hashed) {
 800                                                 if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
 801
 802                                                 db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
 803                                                         id = '$ref_id'");
 804                                         } */
 805
 806                                         // check for user post link to main table
 807
 808                                         // do we allow duplicate posts with same GUID in different feeds?
 809                                         if (get_pref("ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
 810                                                 $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
 811                                         } else {
 812                                                 $dupcheck_qpart = "";
 813                                         }
 814
 815                                         /* Collect article tags here so we could filter by them: */
 816
 817                                         $article_filters = get_article_filters($filters, $entry_title,
 818                                                 $entry_content, $entry_link, $entry_timestamp, $entry_author,
 819                                                 $entry_tags);
 820
 821                                         if ($debug_enabled) {
 822                                                 _debug("article filters: ", $debug_enabled);
 823                                                 if (count($article_filters) != 0) {
 824                                                         print_r($article_filters);
 825                                                 }
 826                                         }
 827
 828                                         if (find_article_filter($article_filters, "filter")) {
 829                                                 db_query("COMMIT"); // close transaction in progress
 830                                                 continue;
 831                                         }
 832
 833                                         $score = calculate_article_score($article_filters);
 834
 835                                         _debug("initial score: $score", $debug_enabled);
 836
 837                                         $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
 838                                                         ref_id = '$ref_id' AND owner_uid = '$owner_uid'
 839                                                         $dupcheck_qpart";
 840
 841 //                                      if ($_REQUEST["xdebug"]) print "$query\n";
 842
 843                                         $result = db_query($query);
 844
 845                                         // okay it doesn't exist - create user entry
 846                                         if (db_num_rows($result) == 0) {
 847
 848                                                 _debug("user record not found, creating...", $debug_enabled);
 849
 850                                                 if ($score >= -500 && !find_article_filter($article_filters, 'catchup')) {
 851                                                         $unread = 'true';
 852                                                         $last_read_qpart = 'NULL';
 853                                                 } else {
 854                                                         $unread = 'false';
 855                                                         $last_read_qpart = 'NOW()';
 856                                                 }
 857
 858                                                 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
 859                                                         $marked = 'true';
 860                                                 } else {
 861                                                         $marked = 'false';
 862                                                 }
 863
 864                                                 if (find_article_filter($article_filters, 'publish')) {
 865                                                         $published = 'true';
 866                                                 } else {
 867                                                         $published = 'false';
 868                                                 }
 869
 870                                                 // N-grams
 871
 872                                                 if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
 873
 874                                                         $result = db_query("SELECT COUNT(*) AS similar FROM
 875                                                                         ttrss_entries,ttrss_user_entries
 876                                                                 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
 877                                                                         AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
 878                                                                         AND owner_uid = $owner_uid");
 879
 880                                                         $ngram_similar = db_fetch_result($result, 0, "similar");
 881
 882                                                         _debug("N-gram similar results: $ngram_similar", $debug_enabled);
 883
 884                                                         if ($ngram_similar > 0) {
 885                                                                 $unread = 'false';
 886                                                         }
 887                                                 }
 888
 889                                                 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
 890                                                 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
 891
 892                                                 $result = db_query(
 893                                                         "INSERT INTO ttrss_user_entries
 894                                                                 (ref_id, owner_uid, feed_id, unread, last_read, marked,
 895                                                                 published, score, tag_cache, label_cache, uuid,
 896                                                                 last_marked, last_published)
 897                                                         VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
 898                                                                 $last_read_qpart, $marked, $published, '$score', '', '',
 899                                                                 '', $last_marked, $last_published)");
 900
 901                                                 if (PUBSUBHUBBUB_HUB && $published == 'true') {
 902                                                         $rss_link = get_self_url_prefix() .
 903                                                                 "/public.php?op=rss&id=-2&key=" .
 904                                                                 get_feed_access_key(-2, false, $owner_uid);
 905
 906                                                         $p = new Publisher(PUBSUBHUBBUB_HUB);
 907
 908                                                         /* $pubsub_result = */ $p->publish_update($rss_link);
 909                                                 }
 910
 911                                                 $result = db_query(
 912                                                         "SELECT int_id FROM ttrss_user_entries WHERE
 913                                                                 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
 914                                                                 feed_id = '$feed' LIMIT 1");
 915
 916                                                 if (db_num_rows($result) == 1) {
 917                                                         $entry_int_id = db_fetch_result($result, 0, "int_id");
 918                                                 }
 919                                         } else {
 920                                                 _debug("user record FOUND", $debug_enabled);
 921
 922                                                 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
 923                                                 $entry_int_id = db_fetch_result($result, 0, "int_id");
 924                                         }
 925
 926                                         _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
 927
 928                                         db_query("UPDATE ttrss_entries
 929                                                 SET title = '$entry_title',
 930                                                         content = '$entry_content',
 931                                                         content_hash = '$entry_current_hash',
 932                                                         updated = '$entry_timestamp_fmt',
 933                                                         num_comments = '$num_comments',
 934                                                         plugin_data = '$entry_plugin_data',
 935                                                         author = '$entry_author',
 936                                                         lang = '$entry_language'
 937                                                 WHERE id = '$ref_id'");
 938
 939                                         if ($mark_unread_on_update) {
 940                                                 db_query("UPDATE ttrss_user_entries
 941                                                         SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
 942                                         }
 943                                 }
 944
 945                                 db_query("COMMIT");
 946
 947                                 _debug("assigning labels...", $debug_enabled);
 948
 949                                 assign_article_to_label_filters($entry_ref_id, $article_filters,
 950                                         $owner_uid, $article_labels);
 951
 952                                 _debug("looking for enclosures...", $debug_enabled);
 953
 954                                 // enclosures
 955
 956                                 $enclosures = array();
 957
 958                                 $encs = $item->get_enclosures();
 959
 960                                 if (is_array($encs)) {
 961                                         foreach ($encs as $e) {
 962                                                 $e_item = array(
 963                                                         $e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
 964                                                 array_push($enclosures, $e_item);
 965                                         }
 966                                 }
 967
 968                                 if ($debug_enabled) {
 969                                         _debug("article enclosures:", $debug_enabled);
 970                                         print_r($enclosures);
 971                                 }
 972
 973                                 db_query("BEGIN");
 974
 975 //                              debugging
 976 //                              db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
 977
 978                                 foreach ($enclosures as $enc) {
 979                                         $enc_url = db_escape_string($enc[0]);
 980                                         $enc_type = db_escape_string($enc[1]);
 981                                         $enc_dur = db_escape_string($enc[2]);
 982                                         $enc_title = db_escape_string($enc[3]);
 983                                         $enc_width = intval($enc[4]);
 984                                         $enc_height = intval($enc[5]);
 985
 986                                         $result = db_query("SELECT id FROM ttrss_enclosures
 987                                                 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
 988
 989                                         if (db_num_rows($result) == 0) {
 990                                                 db_query("INSERT INTO ttrss_enclosures
 991                                                         (content_url, content_type, title, duration, post_id, width, height) VALUES
 992                                                         ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
 993                                         }
 994                                 }
 995
 996                                 db_query("COMMIT");
 997
 998                                 // check for manual tags (we have to do it here since they're loaded from filters)
 999
1000                                 foreach ($article_filters as $f) {
1001                                         if ($f["type"] == "tag") {
1002
1003                                                 $manual_tags = trim_array(explode(",", $f["param"]));
1004
1005                                                 foreach ($manual_tags as $tag) {
1006                                                         if (tag_is_valid($tag)) {
1007                                                                 array_push($entry_tags, $tag);
1008                                                         }
1009                                                 }
1010                                         }
1011                                 }
1012
1013                                 // Skip boring tags
1014
1015                                 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
1016                                         'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1017
1018                                 $filtered_tags = array();
1019                                 $tags_to_cache = array();
1020
1021                                 if ($entry_tags && is_array($entry_tags)) {
1022                                         foreach ($entry_tags as $tag) {
1023                                                 if (array_search($tag, $boring_tags) === false) {
1024                                                         array_push($filtered_tags, $tag);
1025                                                 }
1026                                         }
1027                                 }
1028
1029                                 $filtered_tags = array_unique($filtered_tags);
1030
1031                                 if ($debug_enabled) {
1032                                         _debug("filtered article tags:", $debug_enabled);
1033                                         print_r($filtered_tags);
1034                                 }
1035
1036                                 // Save article tags in the database
1037
1038                                 if (count($filtered_tags) > 0) {
1039
1040                                         db_query("BEGIN");
1041
1042                                         foreach ($filtered_tags as $tag) {
1043
1044                                                 $tag = sanitize_tag($tag);
1045                                                 $tag = db_escape_string($tag);
1046
1047                                                 if (!tag_is_valid($tag)) continue;
1048
1049                                                 $result = db_query("SELECT id FROM ttrss_tags
1050                                                         WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1051                                                         owner_uid = '$owner_uid' LIMIT 1");
1052
1053                                                         if ($result && db_num_rows($result) == 0) {
1054
1055                                                                 db_query("INSERT INTO ttrss_tags
1056                                                                         (owner_uid,tag_name,post_int_id)
1057                                                                         VALUES ('$owner_uid','$tag', '$entry_int_id')");
1058                                                         }
1059
1060                                                 array_push($tags_to_cache, $tag);
1061                                         }
1062
1063                                         /* update the cache */
1064
1065                                         $tags_to_cache = array_unique($tags_to_cache);
1066
1067                                         $tags_str = db_escape_string(join(",", $tags_to_cache));
1068
1069                                         db_query("UPDATE ttrss_user_entries
1070                                                 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1071                                                 AND owner_uid = $owner_uid");
1072
1073                                         db_query("COMMIT");
1074                                 }
1075
1076                                 if (get_pref("AUTO_ASSIGN_LABELS", $owner_uid, false)) {
1077                                         _debug("auto-assigning labels...", $debug_enabled);
1078
1079                                         foreach ($labels as $label) {
1080                                                 $caption = preg_quote($label["caption"]);
1081
1082                                                 if ($caption && preg_match("/\b$caption\b/i", "$tags_str " . strip_tags($entry_content) . " $entry_title")) {
1083                                                         if (!labels_contains_caption($article_labels, $caption)) {
1084                                                                 label_add_article($entry_ref_id, $caption, $owner_uid);
1085                                                         }
1086                                                 }
1087                                         }
1088                                 }
1089
1090                                 _debug("article processed", $debug_enabled);
1091                         }
1092
1093                         _debug("purging feed...", $debug_enabled);
1094
1095                         purge_feed($feed, 0, $debug_enabled);
1096
1097                         db_query("UPDATE ttrss_feeds
1098                                 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
1099
1100 //                      db_query("COMMIT");
1101
1102                 } else {
1103
1104                         $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
1105
1106                         _debug("fetch error: $error_msg", $debug_enabled);
1107
1108                         if (count($rss->errors()) > 1) {
1109                                 foreach ($rss->errors() as $error) {
1110                                         _debug("+ $error");
1111                                 }
1112                         }
1113
1114                         db_query(
1115                                 "UPDATE ttrss_feeds SET last_error = '$error_msg',
1116                                 last_updated = NOW() WHERE id = '$feed'");
1117
1118                         unset($rss);
1119                 }
1120
1121                 _debug("done", $debug_enabled);
1122
1123                 return $rss;
1124         }
1125
1126         function cache_images($html, $site_url, $debug) {
1127                 libxml_use_internal_errors(true);
1128
1129                 $charset_hack = '<head>
1130                         <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1131                 </head>';
1132
1133                 $doc = new DOMDocument();
1134                 $doc->loadHTML($charset_hack . $html);
1135                 $xpath = new DOMXPath($doc);
1136
1137                 $entries = $xpath->query('(//img[@src])');
1138
1139                 foreach ($entries as $entry) {
1140                         if ($entry->hasAttribute('src')) {
1141                                 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1142
1143                                 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1144
1145                                 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1146
1147                                 if (!file_exists($local_filename)) {
1148                                         $file_content = fetch_file_contents($src);
1149
1150                                         if ($file_content && strlen($file_content) > _MIN_CACHE_IMAGE_SIZE) {
1151                                                 file_put_contents($local_filename, $file_content);
1152                                         }
1153                                 }
1154
1155                                 /* if (file_exists($local_filename)) {
1156                                         $entry->setAttribute('src', SELF_URL_PATH . '/image.php?url=' .
1157                                                 base64_encode($src));
1158                                 } */
1159                         }
1160                 }
1161
1162                 //$node = $doc->getElementsByTagName('body')->item(0);
1163                 //return $doc->saveXML($node);
1164         }
1165
1166         function expire_error_log($debug) {
1167                 if ($debug) _debug("Removing old error log entries...");
1168
1169                 if (DB_TYPE == "pgsql") {
1170                         db_query("DELETE FROM ttrss_error_log
1171                                 WHERE created_at < NOW() - INTERVAL '7 days'");
1172                 } else {
1173                         db_query("DELETE FROM ttrss_error_log
1174                                 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1175                 }
1176
1177         }
1178
1179         function expire_lock_files($debug) {
1180                 //if ($debug) _debug("Removing old lock files...");
1181
1182                 $num_deleted = 0;
1183
1184                 if (is_writable(LOCK_DIRECTORY)) {
1185                         $files = glob(LOCK_DIRECTORY . "/*.lock");
1186
1187                         if ($files) {
1188                                 foreach ($files as $file) {
1189                                         if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
1190                                                 unlink($file);
1191                                                 ++$num_deleted;
1192                                         }
1193                                 }
1194                         }
1195                 }
1196
1197                 if ($debug) _debug("Removed $num_deleted old lock files.");
1198         }
1199
1200         function expire_cached_files($debug) {
1201                 foreach (array("simplepie", "images", "export", "upload") as $dir) {
1202                         $cache_dir = CACHE_DIR . "/$dir";
1203
1204 //                      if ($debug) _debug("Expiring $cache_dir");
1205
1206                         $num_deleted = 0;
1207
1208                         if (is_writable($cache_dir)) {
1209                                 $files = glob("$cache_dir/*");
1210
1211                                 if ($files) {
1212                                         foreach ($files as $file) {
1213                                                 if (time() - filemtime($file) > 86400*7) {
1214                                                         unlink($file);
1215
1216                                                         ++$num_deleted;
1217                                                 }
1218                                         }
1219                                 }
1220                         }
1221
1222                         if ($debug) _debug("$cache_dir: removed $num_deleted files.");
1223                 }
1224         }
1225
1226         /**
1227         * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1228         * Returns the url query as associative array
1229         *
1230         * @param    string    query
1231         * @return    array    params
1232         */
1233         function convertUrlQuery($query) {
1234                 $queryParts = explode('&', $query);
1235
1236                 $params = array();
1237
1238                 foreach ($queryParts as $param) {
1239                         $item = explode('=', $param);
1240                         $params[$item[0]] = $item[1];
1241                 }
1242
1243                 return $params;
1244         }
1245
1246         function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
1247                 $matches = array();
1248
1249                 foreach ($filters as $filter) {
1250                         $match_any_rule = $filter["match_any_rule"];
1251                         $inverse = $filter["inverse"];
1252                         $filter_match = false;
1253
1254                         foreach ($filter["rules"] as $rule) {
1255                                 $match = false;
1256                                 $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
1257                                 $rule_inverse = $rule["inverse"];
1258
1259                                 if (!$reg_exp)
1260                                         continue;
1261
1262                                 switch ($rule["type"]) {
1263                                 case "title":
1264                                         $match = @preg_match("/$reg_exp/i", $title);
1265                                         break;
1266                                 case "content":
1267                                         // we don't need to deal with multiline regexps
1268                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1269
1270                                         $match = @preg_match("/$reg_exp/i", $content);
1271                                         break;
1272                                 case "both":
1273                                         // we don't need to deal with multiline regexps
1274                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1275
1276                                         $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
1277                                         break;
1278                                 case "link":
1279                                         $match = @preg_match("/$reg_exp/i", $link);
1280                                         break;
1281                                 case "author":
1282                                         $match = @preg_match("/$reg_exp/i", $author);
1283                                         break;
1284                                 case "tag":
1285                                         foreach ($tags as $tag) {
1286                                                 if (@preg_match("/$reg_exp/i", $tag)) {
1287                                                         $match = true;
1288                                                         break;
1289                                                 }
1290                                         }
1291                                         break;
1292                                 }
1293
1294                                 if ($rule_inverse) $match = !$match;
1295
1296                                 if ($match_any_rule) {
1297                                         if ($match) {
1298                                                 $filter_match = true;
1299                                                 break;
1300                                         }
1301                                 } else {
1302                                         $filter_match = $match;
1303                                         if (!$match) {
1304                                                 break;
1305                                         }
1306                                 }
1307                         }
1308
1309                         if ($inverse) $filter_match = !$filter_match;
1310
1311                         if ($filter_match) {
1312                                 foreach ($filter["actions"] AS $action) {
1313                                         array_push($matches, $action);
1314
1315                                         // if Stop action encountered, perform no further processing
1316                                         if ($action["type"] == "stop") return $matches;
1317                                 }
1318                         }
1319                 }
1320
1321                 return $matches;
1322         }
1323
1324         function find_article_filter($filters, $filter_name) {
1325                 foreach ($filters as $f) {
1326                         if ($f["type"] == $filter_name) {
1327                                 return $f;
1328                         };
1329                 }
1330                 return false;
1331         }
1332
1333         function find_article_filters($filters, $filter_name) {
1334                 $results = array();
1335
1336                 foreach ($filters as $f) {
1337                         if ($f["type"] == $filter_name) {
1338                                 array_push($results, $f);
1339                         };
1340                 }
1341                 return $results;
1342         }
1343
1344         function calculate_article_score($filters) {
1345                 $score = 0;
1346
1347                 foreach ($filters as $f) {
1348                         if ($f["type"] == "score") {
1349                                 $score += $f["param"];
1350                         };
1351                 }
1352                 return $score;
1353         }
1354
1355         function labels_contains_caption($labels, $caption) {
1356                 foreach ($labels as $label) {
1357                         if ($label[1] == $caption) {
1358                                 return true;
1359                         }
1360                 }
1361
1362                 return false;
1363         }
1364
1365         function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
1366                 foreach ($filters as $f) {
1367                         if ($f["type"] == "label") {
1368                                 if (!labels_contains_caption($article_labels, $f["param"])) {
1369                                         label_add_article($id, $f["param"], $owner_uid);
1370                                 }
1371                         }
1372                 }
1373         }
1374
1375         function make_guid_from_title($title) {
1376                 return preg_replace("/[ \"\',.:;]/", "-",
1377                         mb_strtolower(strip_tags($title), 'utf-8'));
1378         }
1379
1380         /* function verify_feed_xml($feed_data) {
1381                 libxml_use_internal_errors(true);
1382                 $doc = new DOMDocument();
1383                 $doc->loadXML($feed_data);
1384                 $error = libxml_get_last_error();
1385                 libxml_clear_errors();
1386                 return $error;
1387         } */
1388
1389         function housekeeping_common($debug) {
1390                 expire_cached_files($debug);
1391                 expire_lock_files($debug);
1392                 expire_error_log($debug);
1393
1394                 $count = update_feedbrowser_cache();
1395                 _debug("Feedbrowser updated, $count feeds processed.");
1396
1397                 purge_orphans( true);
1398                 $rc = cleanup_tags( 14, 50000);
1399
1400                 _debug("Cleaned $rc cached tags.");
1401
1402                 PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
1403
1404         }
1405 ?>