include/rssfuncs.php

   1 <?php
   2         define('DAEMON_UPDATE_LOGIN_LIMIT', 30);
   3         define('DAEMON_FEED_LIMIT', 100);
   4         define('DAEMON_SLEEP_INTERVAL', 60);
   5
   6         function update_feedbrowser_cache($link) {
   7
   8                 $result = db_query($link, "SELECT feed_url, site_url, title, COUNT(id) AS subscribers
   9                         FROM ttrss_feeds WHERE (SELECT COUNT(id) = 0 FROM ttrss_feeds AS tf
  10                                 WHERE tf.feed_url = ttrss_feeds.feed_url
  11                                 AND (private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%'))
  12                                 GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
  13
  14                 db_query($link, "BEGIN");
  15
  16                 db_query($link, "DELETE FROM ttrss_feedbrowser_cache");
  17
  18                 $count = 0;
  19
  20                 while ($line = db_fetch_assoc($result)) {
  21                         $subscribers = db_escape_string($link, $line["subscribers"]);
  22                         $feed_url = db_escape_string($link, $line["feed_url"]);
  23                         $title = db_escape_string($link, $line["title"]);
  24                         $site_url = db_escape_string($link, $line["site_url"]);
  25
  26                         $tmp_result = db_query($link, "SELECT subscribers FROM
  27                                 ttrss_feedbrowser_cache WHERE feed_url = '$feed_url'");
  28
  29                         if (db_num_rows($tmp_result) == 0) {
  30
  31                                 db_query($link, "INSERT INTO ttrss_feedbrowser_cache
  32                                         (feed_url, site_url, title, subscribers) VALUES ('$feed_url',
  33                                                 '$site_url', '$title', '$subscribers')");
  34
  35                                 ++$count;
  36
  37                         }
  38
  39                 }
  40
  41                 db_query($link, "COMMIT");
  42
  43                 return $count;
  44
  45         }
  46
  47
  48         /**
  49          * Update a feed batch.
  50          * Used by daemons to update n feeds by run.
  51          * Only update feed needing a update, and not being processed
  52          * by another process.
  53          *
  54          * @param mixed $link Database link
  55          * @param integer $limit Maximum number of feeds in update batch. Default to DAEMON_FEED_LIMIT.
  56          * @param boolean $from_http Set to true if you call this function from http to disable cli specific code.
  57          * @param boolean $debug Set to false to disable debug output. Default to true.
  58          * @return void
  59          */
  60         function update_daemon_common($link, $limit = DAEMON_FEED_LIMIT, $from_http = false, $debug = true) {
  61                 // Process all other feeds using last_updated and interval parameters
  62
  63                 define('PREFS_NO_CACHE', true);
  64
  65                 // Test if the user has loggued in recently. If not, it does not update its feeds.
  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                 // Test if the feed need a update (update interval exceded).
  77                 if (DB_TYPE == "pgsql") {
  78                         $update_limit_qpart = "AND ((
  79                                         ttrss_feeds.update_interval = 0
  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                                 OR last_updated = '1970-01-01 00:00:00')";
  86                 } else {
  87                         $update_limit_qpart = "AND ((
  88                                         ttrss_feeds.update_interval = 0
  89                                         AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
  90                                 ) OR (
  91                                         ttrss_feeds.update_interval > 0
  92                                         AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
  93                                 ) OR ttrss_feeds.last_updated IS NULL
  94                                 OR last_updated = '1970-01-01 00:00:00')";
  95                 }
  96
  97                 // Test if feed is currently being updated by another process.
  98                 if (DB_TYPE == "pgsql") {
  99                         $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '5 minutes')";
 100                 } else {
 101                         $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 5 MINUTE))";
 102                 }
 103
 104                 // Test if there is a limit to number of updated feeds
 105                 $query_limit = "";
 106                 if($limit) $query_limit = sprintf("LIMIT %d", $limit);
 107
 108                 $random_qpart = sql_random_function();
 109
 110                 // We search for feed needing update.
 111                 $result = db_query($link, "SELECT DISTINCT ttrss_feeds.feed_url
 112                         FROM
 113                                 ttrss_feeds, ttrss_users, ttrss_user_prefs
 114                         WHERE
 115                                 ttrss_feeds.owner_uid = ttrss_users.id
 116                                 AND ttrss_users.id = ttrss_user_prefs.owner_uid
 117                                 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
 118                                 $login_thresh_qpart $update_limit_qpart
 119                                 $updstart_thresh_qpart
 120                         ORDER BY feed_url $query_limit");
 121
 122                 $user_prefs_cache = array();
 123
 124                 if($debug) _debug(sprintf("Scheduled %d feeds to update...", db_num_rows($result)));
 125
 126                 // Here is a little cache magic in order to minimize risk of double feed updates.
 127                 $feeds_to_update = array();
 128                 while ($line = db_fetch_assoc($result)) {
 129                         array_push($feeds_to_update, db_escape_string($link, $line['feed_url']));
 130                 }
 131
 132                 // We update the feed last update started date before anything else.
 133                 // There is no lag due to feed contents downloads
 134                 // It prevent an other process to update the same feed.
 135
 136                 if(count($feeds_to_update) > 0) {
 137                         $feeds_quoted = array();
 138
 139                         foreach ($feeds_to_update as $feed) {
 140                                 array_push($feeds_quoted, "'" . db_escape_string($link, $feed) . "'");
 141                         }
 142
 143                         db_query($link, sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
 144                                 WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
 145                 }
 146
 147                 expire_cached_files($debug);
 148                 expire_lock_files($debug);
 149
 150                 // For each feed, we call the feed update function.
 151                 foreach ($feeds_to_update as $feed) {
 152                         if($debug) _debug("Base feed: $feed");
 153
 154                         //update_rss_feed($link, $line["id"], true);
 155
 156                         // since we have the data cached, we can deal with other feeds with the same url
 157
 158                         $tmp_result = db_query($link, "SELECT ttrss_feeds.feed_url,ttrss_feeds.id,last_updated
 159                         FROM ttrss_feeds, ttrss_users WHERE
 160                                 ttrss_users.id = ttrss_feeds.owner_uid AND
 161                                 feed_url = '".db_escape_string($link, $feed)."' AND
 162                                 ttrss_feeds.update_interval != -1
 163                                 $login_thresh_qpart
 164                         ORDER BY feed_url $query_limit");
 165
 166                         if (db_num_rows($tmp_result) > 0) {
 167                                 while ($tline = db_fetch_assoc($tmp_result)) {
 168                                         if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"]);
 169                                         update_rss_feed($link, $tline["id"], true);
 170                                 }
 171                         }
 172                 }
 173
 174                 require_once "digest.php";
 175
 176                 // Send feed digests by email if needed.
 177                 send_headlines_digests($link, $debug);
 178
 179         } // function update_daemon_common
 180
 181         // ignore_daemon is not used
 182         function update_rss_feed($link, $feed, $ignore_daemon = false, $no_cache = false,
 183                 $override_url = false) {
 184
 185                 require_once "lib/simplepie/simplepie.inc";
 186
 187                 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
 188
 189                 if ($debug_enabled) {
 190                         _debug("update_rss_feed: start");
 191                 }
 192
 193                 $result = db_query($link, "SELECT id,update_interval,auth_login,
 194                         feed_url,auth_pass,cache_images,last_updated,
 195                         mark_unread_on_update, owner_uid,
 196                         pubsub_state
 197                         FROM ttrss_feeds WHERE id = '$feed'");
 198
 199                 if (db_num_rows($result) == 0) {
 200                         if ($debug_enabled) {
 201                                 _debug("update_rss_feed: feed $feed NOT FOUND/SKIPPED");
 202                         }
 203                         return false;
 204                 }
 205
 206                 $last_updated = db_fetch_result($result, 0, "last_updated");
 207                 $owner_uid = db_fetch_result($result, 0, "owner_uid");
 208                 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
 209                         0, "mark_unread_on_update"));
 210                 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
 211
 212                 db_query($link, "UPDATE ttrss_feeds SET last_update_started = NOW()
 213                         WHERE id = '$feed'");
 214
 215                 $auth_login = db_fetch_result($result, 0, "auth_login");
 216                 $auth_pass = db_fetch_result($result, 0, "auth_pass");
 217
 218                 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
 219                 $fetch_url = db_fetch_result($result, 0, "feed_url");
 220
 221                 $feed = db_escape_string($link, $feed);
 222
 223                 if ($override_url) $fetch_url = $override_url;
 224
 225                 $date_feed_processed = date('Y-m-d H:i');
 226
 227                 $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".feed";
 228
 229                 // Ignore cache if new feed or manual update.
 230                 $cache_age = ($no_cache || is_null($last_updated) || $last_updated == '1970-01-01 00:00:00') ?
 231                         30 : get_feed_update_interval($link, $feed) * 60;
 232
 233                 if ($debug_enabled) {
 234                         _debug("update_rss_feed: cache filename: $cache_filename exists: " . file_exists($cache_filename));
 235                         _debug("update_rss_feed: cache age: $cache_age; no cache: $no_cache");
 236                 }
 237
 238                 $cached_feed_data_hash = false;
 239
 240                 $rss = false;
 241                 $rss_hash = false;
 242                 $cache_timestamp = file_exists($cache_filename) ? filemtime($cache_filename) : 0;
 243                 $last_updated_timestamp = strtotime($last_updated);
 244
 245                 if (file_exists($cache_filename) &&
 246                         is_readable($cache_filename) &&
 247                         !$auth_login && !$auth_pass &&
 248                         filemtime($cache_filename) > time() - $cache_age) {
 249
 250                                 if ($debug_enabled) {
 251                                         _debug("update_rss_feed: using local cache.");
 252                                 }
 253
 254                                 @$rss_data = file_get_contents($cache_filename);
 255
 256                                 if ($rss_data) {
 257                                         $rss_hash = sha1($rss_data);
 258                                         @$rss = unserialize($rss_data);
 259                                 }
 260                 }
 261
 262                 if (!$rss) {
 263
 264                         if (!$feed_data) {
 265                                 if ($debug_enabled) {
 266                                         _debug("update_rss_feed: fetching [$fetch_url] (ts: $cache_timestamp/$last_updated_timestamp)");
 267                                 }
 268
 269                                 $feed_data = fetch_file_contents($fetch_url, false,
 270                                         $auth_login, $auth_pass, false, $no_cache ? 15 : 45,
 271                                         max($last_updated_timestamp, $cache_timestamp));
 272
 273                                 if ($debug_enabled) {
 274                                         _debug("update_rss_feed: fetch done.");
 275                                 }
 276
 277                         }
 278
 279                         if (!$feed_data) {
 280                                 global $fetch_last_error;
 281                                 global $fetch_last_error_code;
 282
 283                                 if ($debug_enabled) {
 284                                         _debug("update_rss_feed: unable to fetch: $fetch_last_error [$fetch_last_error_code]");
 285                                 }
 286
 287                                 $error_escaped = '';
 288
 289                                 // If-Modified-Since
 290                                 if ($fetch_last_error_code != 304) {
 291                                         $error_escaped = db_escape_string($link, $fetch_last_error);
 292                                 } else {
 293                                         if ($debug_enabled) {
 294                                                 _debug("update_rss_feed: source claims data not modified, nothing to do.");
 295                                         }
 296                                 }
 297
 298                                 db_query($link,
 299                                         "UPDATE ttrss_feeds SET last_error = '$error_escaped',
 300                                                 last_updated = NOW() WHERE id = '$feed'");
 301
 302                                 return;
 303                         }
 304                 }
 305
 306                 $pluginhost = new PluginHost($link);
 307                 $pluginhost->set_debug($debug_enabled);
 308                 $user_plugins = get_pref($link, "_ENABLED_PLUGINS", $owner_uid);
 309
 310                 $pluginhost->load(PLUGINS, $pluginhost::KIND_ALL);
 311                 $pluginhost->load($user_plugins, $pluginhost::KIND_USER, $owner_uid);
 312                 $pluginhost->load_data();
 313
 314                 foreach ($pluginhost->get_hooks($pluginhost::HOOK_FEED_FETCHED) as $plugin) {
 315                         $feed_data = $plugin->hook_feed_fetched($feed_data);
 316                 }
 317
 318                 if (!$rss) {
 319                         $rss = new SimplePie();
 320                         $rss->set_sanitize_class("SanitizeDummy");
 321                         // simplepie ignores the above and creates default sanitizer anyway,
 322                         // so let's override it...
 323                         $rss->sanitize = new SanitizeDummy();
 324                         $rss->set_output_encoding('UTF-8');
 325                         $rss->set_raw_data($feed_data);
 326                         $rss->enable_cache(false);
 327
 328                         @$rss->init();
 329                 }
 330
 331 //              print_r($rss);
 332
 333                 $feed = db_escape_string($link, $feed);
 334
 335                 if (!$rss->error()) {
 336
 337                         // cache data for later
 338                         if (!$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
 339                                 $rss_data = serialize($rss);
 340                                 $new_rss_hash = sha1($rss_data);
 341
 342                                 if ($new_rss_hash != $rss_hash) {
 343                                         if ($debug_enabled) {
 344                                                 _debug("update_rss_feed: saving $cache_filename");
 345                                         }
 346                                         @file_put_contents($cache_filename, serialize($rss));
 347                                 }
 348                         }
 349
 350                         // We use local pluginhost here because we need to load different per-user feed plugins
 351                         $pluginhost->run_hooks($pluginhost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
 352
 353                         if ($debug_enabled) {
 354                                 _debug("update_rss_feed: processing feed data...");
 355                         }
 356
 357 //                      db_query($link, "BEGIN");
 358
 359                         if (DB_TYPE == "pgsql") {
 360                                 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
 361                         } else {
 362                                 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
 363                         }
 364
 365                         $result = db_query($link, "SELECT title,site_url,owner_uid,
 366                                 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
 367                                                 favicon_needs_check
 368                                 FROM ttrss_feeds WHERE id = '$feed'");
 369
 370                         $registered_title = db_fetch_result($result, 0, "title");
 371                         $orig_site_url = db_fetch_result($result, 0, "site_url");
 372                         $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
 373                                 "favicon_needs_check"));
 374
 375                         $owner_uid = db_fetch_result($result, 0, "owner_uid");
 376
 377                         $site_url = db_escape_string($link, mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
 378
 379                         if ($debug_enabled) {
 380                                 _debug("update_rss_feed: checking favicon...");
 381                         }
 382
 383                         if ($favicon_needs_check) {
 384                                 check_feed_favicon($site_url, $feed, $link);
 385
 386                                 db_query($link, "UPDATE ttrss_feeds SET favicon_last_checked = NOW()
 387                                         WHERE id = '$feed'");
 388                         }
 389
 390                         if (!$registered_title || $registered_title == "[Unknown]") {
 391
 392                                 $feed_title = db_escape_string($link, $rss->get_title());
 393
 394                                 if ($debug_enabled) {
 395                                         _debug("update_rss_feed: registering title: $feed_title");
 396                                 }
 397
 398                                 db_query($link, "UPDATE ttrss_feeds SET
 399                                         title = '$feed_title' WHERE id = '$feed'");
 400                         }
 401
 402                         if ($site_url && $orig_site_url != $site_url) {
 403                                 db_query($link, "UPDATE ttrss_feeds SET
 404                                         site_url = '$site_url' WHERE id = '$feed'");
 405                         }
 406
 407                         if ($debug_enabled) {
 408                                 _debug("update_rss_feed: loading filters & labels...");
 409                         }
 410
 411                         $filters = load_filters($link, $feed, $owner_uid);
 412                         $labels = get_all_labels($link, $owner_uid);
 413
 414                         if ($debug_enabled) {
 415                                 //print_r($filters);
 416                                 _debug("update_rss_feed: " . count($filters) . " filters loaded.");
 417                         }
 418
 419                         $items = $rss->get_items();
 420
 421                         if (!is_array($items)) {
 422                                 if ($debug_enabled) {
 423                                         _debug("update_rss_feed: no articles found.");
 424                                 }
 425
 426                                 db_query($link, "UPDATE ttrss_feeds
 427                                         SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
 428
 429                                 return; // no articles
 430                         }
 431
 432                         if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
 433
 434                                 if ($debug_enabled) _debug("update_rss_feed: checking for PUSH hub...");
 435
 436                                 $feed_hub_url = false;
 437
 438                                 $links = $rss->get_links('hub');
 439
 440                                 if ($links && is_array($links)) {
 441                                         foreach ($links as $l) {
 442                                                 $feed_hub_url = $l;
 443                                                 break;
 444                                         }
 445                                 }
 446
 447                                 if ($debug_enabled) _debug("update_rss_feed: feed hub url: $feed_hub_url");
 448
 449                                 if ($feed_hub_url && function_exists('curl_init') &&
 450                                         !ini_get("open_basedir")) {
 451
 452                                         require_once 'lib/pubsubhubbub/subscriber.php';
 453
 454                                         $callback_url = get_self_url_prefix() .
 455                                                 "/public.php?op=pubsub&id=$feed";
 456
 457                                         $s = new Subscriber($feed_hub_url, $callback_url);
 458
 459                                         $rc = $s->subscribe($fetch_url);
 460
 461                                         if ($debug_enabled)
 462                                                 _debug("update_rss_feed: feed hub url found, subscribe request sent.");
 463
 464                                         db_query($link, "UPDATE ttrss_feeds SET pubsub_state = 1
 465                                                 WHERE id = '$feed'");
 466                                 }
 467                         }
 468
 469                         if ($debug_enabled) {
 470                                 _debug("update_rss_feed: processing articles...");
 471                         }
 472
 473                         foreach ($items as $item) {
 474                                 if ($_REQUEST['xdebug'] == 3) {
 475                                         print_r($item);
 476                                 }
 477
 478                                 $entry_guid = $item->get_id();
 479                                 if (!$entry_guid) $entry_guid = $item->get_link();
 480                                 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
 481
 482                                 if ($debug_enabled) {
 483                                         _debug("update_rss_feed: guid $entry_guid");
 484                                 }
 485
 486                                 if (!$entry_guid) continue;
 487
 488                                 $entry_guid = "$owner_uid,$entry_guid";
 489
 490                                 $entry_timestamp = "";
 491
 492                                 $entry_timestamp = strtotime($item->get_date());
 493
 494                                 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
 495                                         $entry_timestamp = time();
 496                                         $no_orig_date = 'true';
 497                                 } else {
 498                                         $no_orig_date = 'false';
 499                                 }
 500
 501                                 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
 502
 503                                 if ($debug_enabled) {
 504                                         _debug("update_rss_feed: date $entry_timestamp [$entry_timestamp_fmt]");
 505                                 }
 506
 507                                 $entry_title = $item->get_title();
 508
 509                                 $entry_link = rewrite_relative_url($site_url, $item->get_link());
 510
 511                                 if ($debug_enabled) {
 512                                         _debug("update_rss_feed: title $entry_title");
 513                                         _debug("update_rss_feed: link $entry_link");
 514                                 }
 515
 516                                 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
 517
 518                                 $entry_content = $item->get_content();
 519                                 if (!$entry_content) $entry_content = $item->get_description();
 520
 521                                 if ($_REQUEST["xdebug"] == 2) {
 522                                         print "update_rss_feed: content: ";
 523                                         print $entry_content;
 524                                         print "\n";
 525                                 }
 526
 527                                 $entry_comments = $item->data["comments"];
 528
 529                                 if ($item->get_author()) {
 530                                         $entry_author_item = $item->get_author();
 531                                         $entry_author = $entry_author_item->get_name();
 532                                         if (!$entry_author) $entry_author = $entry_author_item->get_email();
 533
 534                                         $entry_author = db_escape_string($link, $entry_author);
 535                                 }
 536
 537                                 $entry_guid = db_escape_string($link, mb_substr($entry_guid, 0, 245));
 538
 539                                 $entry_comments = db_escape_string($link, mb_substr($entry_comments, 0, 245));
 540                                 $entry_author = db_escape_string($link, mb_substr($entry_author, 0, 245));
 541
 542                                 $num_comments = $item->get_item_tags('http://purl.org/rss/1.0/modules/slash/', 'comments');
 543
 544                                 if (is_array($num_comments) && is_array($num_comments[0])) {
 545                                         $num_comments = (int) $num_comments[0]["data"];
 546                                 } else {
 547                                         $num_comments = 0;
 548                                 }
 549
 550                                 if ($debug_enabled) {
 551                                         _debug("update_rss_feed: num_comments: $num_comments");
 552                                         _debug("update_rss_feed: looking for tags [1]...");
 553                                 }
 554
 555                                 // parse <category> entries into tags
 556
 557                                 $additional_tags = array();
 558
 559                                 $additional_tags_src = $item->get_categories();
 560
 561                                 if (is_array($additional_tags_src)) {
 562                                         foreach ($additional_tags_src as $tobj) {
 563                                                 array_push($additional_tags, $tobj->get_term());
 564                                         }
 565                                 }
 566
 567                                 if ($debug_enabled) {
 568                                         _debug("update_rss_feed: category tags:");
 569                                         print_r($additional_tags);
 570                                 }
 571
 572                                 if ($debug_enabled) {
 573                                         _debug("update_rss_feed: looking for tags [2]...");
 574                                 }
 575
 576                                 $entry_tags = array_unique($additional_tags);
 577
 578                                 for ($i = 0; $i < count($entry_tags); $i++)
 579                                         $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
 580
 581                                 if ($debug_enabled) {
 582                                         //_debug("update_rss_feed: unfiltered tags found:");
 583                                         //print_r($entry_tags);
 584                                 }
 585
 586                                 if ($debug_enabled) {
 587                                         _debug("update_rss_feed: done collecting data.");
 588                                 }
 589
 590                                 // TODO: less memory-hungry implementation
 591
 592                                 if ($debug_enabled) {
 593                                         _debug("update_rss_feed: applying plugin filters..");
 594                                 }
 595
 596                                 // FIXME not sure if owner_uid is a good idea here, we may have a base entry without user entry (?)
 597                                 $result = db_query($link, "SELECT plugin_data,title,content,link,tag_cache,author FROM ttrss_entries, ttrss_user_entries
 598                                         WHERE ref_id = id AND guid = '".db_escape_string($link, $entry_guid)."' AND owner_uid = $owner_uid");
 599
 600                                 if (db_num_rows($result) != 0) {
 601                                         $entry_plugin_data = db_fetch_result($result, 0, "plugin_data");
 602                                         $stored_article = array("title" => db_fetch_result($result, 0, "title"),
 603                                                 "content" => db_fetch_result($result, 0, "content"),
 604                                                 "link" => db_fetch_result($result, 0, "link"),
 605                                                 "tags" => explode(",", db_fetch_result($result, 0, "tag_cache")),
 606                                                 "author" => db_fetch_result($result, 0, "author"));
 607                                 } else {
 608                                         $entry_plugin_data = "";
 609                                         $stored_article = array();
 610                                 }
 611
 612                                 $article = array("owner_uid" => $owner_uid, // read only
 613                                         "guid" => $entry_guid, // read only
 614                                         "title" => $entry_title,
 615                                         "content" => $entry_content,
 616                                         "link" => $entry_link,
 617                                         "tags" => $entry_tags,
 618                                         "plugin_data" => $entry_plugin_data,
 619                                         "author" => $entry_author,
 620                                         "stored" => $stored_article);
 621
 622                                 foreach ($pluginhost->get_hooks($pluginhost::HOOK_ARTICLE_FILTER) as $plugin) {
 623                                         $article = $plugin->hook_article_filter($article);
 624                                 }
 625
 626                                 $entry_tags = $article["tags"];
 627                                 $entry_guid = db_escape_string($link, $entry_guid);
 628                                 $entry_title = db_escape_string($link, $article["title"]);
 629                                 $entry_author = db_escape_string($link, $article["author"]);
 630                                 $entry_link = db_escape_string($link, $article["link"]);
 631                                 $entry_plugin_data = db_escape_string($link, $article["plugin_data"]);
 632                                 $entry_content = $article["content"]; // escaped below
 633
 634
 635                                 if ($debug_enabled) {
 636                                         _debug("update_rss_feed: plugin data: $entry_plugin_data");
 637                                 }
 638
 639                                 if ($cache_images && is_writable(CACHE_DIR . '/images'))
 640                                         cache_images($entry_content, $site_url, $debug_enabled);
 641
 642                                 $entry_content = db_escape_string($link, $entry_content, false);
 643
 644                                 $content_hash = "SHA1:" . sha1($entry_content);
 645
 646                                 db_query($link, "BEGIN");
 647
 648                                 $result = db_query($link, "SELECT id FROM       ttrss_entries
 649                                         WHERE guid = '$entry_guid'");
 650
 651                                 if (db_num_rows($result) == 0) {
 652
 653                                         if ($debug_enabled) {
 654                                                 _debug("update_rss_feed: base guid [$entry_guid] not found");
 655                                         }
 656
 657                                         // base post entry does not exist, create it
 658
 659                                         $result = db_query($link,
 660                                                 "INSERT INTO ttrss_entries
 661                                                         (title,
 662                                                         guid,
 663                                                         link,
 664                                                         updated,
 665                                                         content,
 666                                                         content_hash,
 667                                                         cached_content,
 668                                                         no_orig_date,
 669                                                         date_updated,
 670                                                         date_entered,
 671                                                         comments,
 672                                                         num_comments,
 673                                                         plugin_data,
 674                                                         author)
 675                                                 VALUES
 676                                                         ('$entry_title',
 677                                                         '$entry_guid',
 678                                                         '$entry_link',
 679                                                         '$entry_timestamp_fmt',
 680                                                         '$entry_content',
 681                                                         '$content_hash',
 682                                                         '',
 683                                                         $no_orig_date,
 684                                                         NOW(),
 685                                                         '$date_feed_processed',
 686                                                         '$entry_comments',
 687                                                         '$num_comments',
 688                                                         '$entry_plugin_data',
 689                                                         '$entry_author')");
 690
 691                                         $article_labels = array();
 692
 693                                 } else {
 694                                         // we keep encountering the entry in feeds, so we need to
 695                                         // update date_updated column so that we don't get horrible
 696                                         // dupes when the entry gets purged and reinserted again e.g.
 697                                         // in the case of SLOW SLOW OMG SLOW updating feeds
 698
 699                                         $base_entry_id = db_fetch_result($result, 0, "id");
 700
 701                                         db_query($link, "UPDATE ttrss_entries SET date_updated = NOW()
 702                                                 WHERE id = '$base_entry_id'");
 703
 704                                         $article_labels = get_article_labels($link, $base_entry_id, $owner_uid);
 705                                 }
 706
 707                                 // now it should exist, if not - bad luck then
 708
 709                                 $result = db_query($link, "SELECT
 710                                                 id,content_hash,no_orig_date,title,plugin_data,
 711                                                 ".SUBSTRING_FOR_DATE."(date_updated,1,19) as date_updated,
 712                                                 ".SUBSTRING_FOR_DATE."(updated,1,19) as updated,
 713                                                 num_comments
 714                                         FROM
 715                                                 ttrss_entries
 716                                         WHERE guid = '$entry_guid'");
 717
 718                                 $entry_ref_id = 0;
 719                                 $entry_int_id = 0;
 720
 721                                 if (db_num_rows($result) == 1) {
 722
 723                                         if ($debug_enabled) {
 724                                                 _debug("update_rss_feed: base guid [$entry_guid] found, checking for user record");
 725                                         }
 726
 727                                         // this will be used below in update handler
 728                                         $orig_content_hash = db_fetch_result($result, 0, "content_hash");
 729                                         $orig_title = db_fetch_result($result, 0, "title");
 730                                         $orig_num_comments = db_fetch_result($result, 0, "num_comments");
 731                                         $orig_date_updated = strtotime(db_fetch_result($result,
 732                                                 0, "date_updated"));
 733                                         $orig_plugin_data = db_fetch_result($result, 0, "plugin_data");
 734
 735                                         $ref_id = db_fetch_result($result, 0, "id");
 736                                         $entry_ref_id = $ref_id;
 737
 738                                         // check for user post link to main table
 739
 740                                         // do we allow duplicate posts with same GUID in different feeds?
 741                                         if (get_pref($link, "ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
 742                                                 $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
 743                                         } else {
 744                                                 $dupcheck_qpart = "";
 745                                         }
 746
 747                                         /* Collect article tags here so we could filter by them: */
 748
 749                                         $article_filters = get_article_filters($filters, $entry_title,
 750                                                 $entry_content, $entry_link, $entry_timestamp, $entry_author,
 751                                                 $entry_tags);
 752
 753                                         if ($debug_enabled) {
 754                                                 _debug("update_rss_feed: article filters: ");
 755                                                 if (count($article_filters) != 0) {
 756                                                         print_r($article_filters);
 757                                                 }
 758                                         }
 759
 760                                         if (find_article_filter($article_filters, "filter")) {
 761                                                 db_query($link, "COMMIT"); // close transaction in progress
 762                                                 continue;
 763                                         }
 764
 765                                         $score = calculate_article_score($article_filters);
 766
 767                                         if ($debug_enabled) {
 768                                                 _debug("update_rss_feed: initial score: $score");
 769                                         }
 770
 771                                         $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
 772                                                         ref_id = '$ref_id' AND owner_uid = '$owner_uid'
 773                                                         $dupcheck_qpart";
 774
 775 //                                      if ($_REQUEST["xdebug"]) print "$query\n";
 776
 777                                         $result = db_query($link, $query);
 778
 779                                         // okay it doesn't exist - create user entry
 780                                         if (db_num_rows($result) == 0) {
 781
 782                                                 if ($debug_enabled) {
 783                                                         _debug("update_rss_feed: user record not found, creating...");
 784                                                 }
 785
 786                                                 if ($score >= -500 && !find_article_filter($article_filters, 'catchup')) {
 787                                                         $unread = 'true';
 788                                                         $last_read_qpart = 'NULL';
 789                                                 } else {
 790                                                         $unread = 'false';
 791                                                         $last_read_qpart = 'NOW()';
 792                                                 }
 793
 794                                                 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
 795                                                         $marked = 'true';
 796                                                 } else {
 797                                                         $marked = 'false';
 798                                                 }
 799
 800                                                 if (find_article_filter($article_filters, 'publish')) {
 801                                                         $published = 'true';
 802                                                 } else {
 803                                                         $published = 'false';
 804                                                 }
 805
 806                                                 // N-grams
 807
 808                                                 if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
 809
 810                                                         $result = db_query($link, "SELECT COUNT(*) AS similar FROM
 811                                                                         ttrss_entries,ttrss_user_entries
 812                                                                 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
 813                                                                         AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
 814                                                                         AND owner_uid = $owner_uid");
 815
 816                                                         $ngram_similar = db_fetch_result($result, 0, "similar");
 817
 818                                                         if ($debug_enabled) {
 819                                                                 _debug("update_rss_feed: N-gram similar results: $ngram_similar");
 820                                                         }
 821
 822                                                         if ($ngram_similar > 0) {
 823                                                                 $unread = 'false';
 824                                                         }
 825                                                 }
 826
 827                                                 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
 828                                                 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
 829
 830                                                 $result = db_query($link,
 831                                                         "INSERT INTO ttrss_user_entries
 832                                                                 (ref_id, owner_uid, feed_id, unread, last_read, marked,
 833                                                                 published, score, tag_cache, label_cache, uuid,
 834                                                                 last_marked, last_published)
 835                                                         VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
 836                                                                 $last_read_qpart, $marked, $published, '$score', '', '',
 837                                                                 '', $last_marked, $last_published)");
 838
 839                                                 if (PUBSUBHUBBUB_HUB && $published == 'true') {
 840                                                         $rss_link = get_self_url_prefix() .
 841                                                                 "/public.php?op=rss&id=-2&key=" .
 842                                                                 get_feed_access_key($link, -2, false, $owner_uid);
 843
 844                                                         $p = new Publisher(PUBSUBHUBBUB_HUB);
 845
 846                                                         $pubsub_result = $p->publish_update($rss_link);
 847                                                 }
 848
 849                                                 $result = db_query($link,
 850                                                         "SELECT int_id FROM ttrss_user_entries WHERE
 851                                                                 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
 852                                                                 feed_id = '$feed' LIMIT 1");
 853
 854                                                 if (db_num_rows($result) == 1) {
 855                                                         $entry_int_id = db_fetch_result($result, 0, "int_id");
 856                                                 }
 857                                         } else {
 858                                                 if ($debug_enabled) {
 859                                                         _debug("update_rss_feed: user record FOUND");
 860                                                 }
 861
 862                                                 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
 863                                                 $entry_int_id = db_fetch_result($result, 0, "int_id");
 864                                         }
 865
 866                                         if ($debug_enabled) {
 867                                                 _debug("update_rss_feed: RID: $entry_ref_id, IID: $entry_int_id");
 868                                         }
 869
 870                                         $post_needs_update = false;
 871                                         $update_insignificant = false;
 872
 873                                         if ($orig_num_comments != $num_comments) {
 874                                                 $post_needs_update = true;
 875                                                 $update_insignificant = true;
 876                                         }
 877
 878                                         if ($entry_plugin_data != $orig_plugin_data) {
 879                                                 $post_needs_update = true;
 880                                                 $update_insignificant = true;
 881                                         }
 882
 883                                         if ($content_hash != $orig_content_hash) {
 884                                                 $post_needs_update = true;
 885                                                 $update_insignificant = false;
 886                                         }
 887
 888                                         if (db_escape_string($link, $orig_title) != $entry_title) {
 889                                                 $post_needs_update = true;
 890                                                 $update_insignificant = false;
 891                                         }
 892
 893                                         // if post needs update, update it and mark all user entries
 894                                         // linking to this post as updated
 895                                         if ($post_needs_update) {
 896
 897                                                 if (defined('DAEMON_EXTENDED_DEBUG')) {
 898                                                         _debug("update_rss_feed: post $entry_guid needs update...");
 899                                                 }
 900
 901 //                                              print "<!-- post $orig_title needs update : $post_needs_update -->";
 902
 903                                                 db_query($link, "UPDATE ttrss_entries
 904                                                         SET title = '$entry_title', content = '$entry_content',
 905                                                                 content_hash = '$content_hash',
 906                                                                 updated = '$entry_timestamp_fmt',
 907                                                                 num_comments = '$num_comments',
 908                                                                 plugin_data = '$entry_plugin_data'
 909                                                         WHERE id = '$ref_id'");
 910
 911                                                 if (!$update_insignificant) {
 912                                                         if ($mark_unread_on_update) {
 913                                                                 db_query($link, "UPDATE ttrss_user_entries
 914                                                                         SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
 915                                                         }
 916                                                 }
 917                                         }
 918                                 }
 919
 920                                 db_query($link, "COMMIT");
 921
 922                                 if ($debug_enabled) {
 923                                         _debug("update_rss_feed: assigning labels...");
 924                                 }
 925
 926                                 assign_article_to_label_filters($link, $entry_ref_id, $article_filters,
 927                                         $owner_uid, $article_labels);
 928
 929                                 if ($debug_enabled) {
 930                                         _debug("update_rss_feed: looking for enclosures...");
 931                                 }
 932
 933                                 // enclosures
 934
 935                                 $enclosures = array();
 936
 937                                 $encs = $item->get_enclosures();
 938
 939                                 if (is_array($encs)) {
 940                                         foreach ($encs as $e) {
 941                                                 $e_item = array(
 942                                                         $e->link, $e->type, $e->length);
 943                                                 array_push($enclosures, $e_item);
 944                                         }
 945                                 }
 946
 947                                 if ($debug_enabled) {
 948                                         _debug("update_rss_feed: article enclosures:");
 949                                         print_r($enclosures);
 950                                 }
 951
 952                                 db_query($link, "BEGIN");
 953
 954                                 foreach ($enclosures as $enc) {
 955                                         $enc_url = db_escape_string($link, $enc[0]);
 956                                         $enc_type = db_escape_string($link, $enc[1]);
 957                                         $enc_dur = db_escape_string($link, $enc[2]);
 958
 959                                         $result = db_query($link, "SELECT id FROM ttrss_enclosures
 960                                                 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
 961
 962                                         if (db_num_rows($result) == 0) {
 963                                                 db_query($link, "INSERT INTO ttrss_enclosures
 964                                                         (content_url, content_type, title, duration, post_id) VALUES
 965                                                         ('$enc_url', '$enc_type', '', '$enc_dur', '$entry_ref_id')");
 966                                         }
 967                                 }
 968
 969                                 db_query($link, "COMMIT");
 970
 971                                 // check for manual tags (we have to do it here since they're loaded from filters)
 972
 973                                 foreach ($article_filters as $f) {
 974                                         if ($f["type"] == "tag") {
 975
 976                                                 $manual_tags = trim_array(explode(",", $f["param"]));
 977
 978                                                 foreach ($manual_tags as $tag) {
 979                                                         if (tag_is_valid($tag)) {
 980                                                                 array_push($entry_tags, $tag);
 981                                                         }
 982                                                 }
 983                                         }
 984                                 }
 985
 986                                 // Skip boring tags
 987
 988                                 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref($link,
 989                                         'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
 990
 991                                 $filtered_tags = array();
 992                                 $tags_to_cache = array();
 993
 994                                 if ($entry_tags && is_array($entry_tags)) {
 995                                         foreach ($entry_tags as $tag) {
 996                                                 if (array_search($tag, $boring_tags) === false) {
 997                                                         array_push($filtered_tags, $tag);
 998                                                 }
 999                                         }
1000                                 }
1001
1002                                 $filtered_tags = array_unique($filtered_tags);
1003
1004                                 if ($debug_enabled) {
1005                                         _debug("update_rss_feed: filtered article tags:");
1006                                         print_r($filtered_tags);
1007                                 }
1008
1009                                 // Save article tags in the database
1010
1011                                 if (count($filtered_tags) > 0) {
1012
1013                                         db_query($link, "BEGIN");
1014
1015                                         foreach ($filtered_tags as $tag) {
1016
1017                                                 $tag = sanitize_tag($tag);
1018                                                 $tag = db_escape_string($link, $tag);
1019
1020                                                 if (!tag_is_valid($tag)) continue;
1021
1022                                                 $result = db_query($link, "SELECT id FROM ttrss_tags
1023                                                         WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1024                                                         owner_uid = '$owner_uid' LIMIT 1");
1025
1026                                                         if ($result && db_num_rows($result) == 0) {
1027
1028                                                                 db_query($link, "INSERT INTO ttrss_tags
1029                                                                         (owner_uid,tag_name,post_int_id)
1030                                                                         VALUES ('$owner_uid','$tag', '$entry_int_id')");
1031                                                         }
1032
1033                                                 array_push($tags_to_cache, $tag);
1034                                         }
1035
1036                                         /* update the cache */
1037
1038                                         $tags_to_cache = array_unique($tags_to_cache);
1039
1040                                         $tags_str = db_escape_string($link, join(",", $tags_to_cache));
1041
1042                                         db_query($link, "UPDATE ttrss_user_entries
1043                                                 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1044                                                 AND owner_uid = $owner_uid");
1045
1046                                         db_query($link, "COMMIT");
1047                                 }
1048
1049                                 if (get_pref($link, "AUTO_ASSIGN_LABELS", $owner_uid, false)) {
1050                                         if ($debug_enabled) {
1051                                                 _debug("update_rss_feed: auto-assigning labels...");
1052                                         }
1053
1054                                         foreach ($labels as $label) {
1055                                                 $caption = preg_quote($label["caption"]);
1056
1057                                                 if ($caption && preg_match("/\b$caption\b/i", "$tags_str " . strip_tags($entry_content) . " $entry_title")) {
1058                                                         if (!labels_contains_caption($article_labels, $caption)) {
1059                                                                 label_add_article($link, $entry_ref_id, $caption, $owner_uid);
1060                                                         }
1061                                                 }
1062                                         }
1063                                 }
1064
1065                                 if ($debug_enabled) {
1066                                         _debug("update_rss_feed: article processed");
1067                                 }
1068                         }
1069
1070                         if (!$last_updated) {
1071                                 if ($debug_enabled) {
1072                                         _debug("update_rss_feed: new feed, catching it up...");
1073                                 }
1074                                 catchup_feed($link, $feed, false, $owner_uid);
1075                         }
1076
1077                         if ($debug_enabled) {
1078                                 _debug("purging feed...");
1079                         }
1080
1081                         purge_feed($link, $feed, 0, $debug_enabled);
1082
1083                         db_query($link, "UPDATE ttrss_feeds
1084                                 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
1085
1086 //                      db_query($link, "COMMIT");
1087
1088                 } else {
1089
1090                         $error_msg = db_escape_string($link, mb_substr($rss->error(), 0, 245));
1091
1092                         if ($debug_enabled) {
1093                                 _debug("update_rss_feed: error fetching feed: $error_msg");
1094                         }
1095
1096                         db_query($link,
1097                                 "UPDATE ttrss_feeds SET last_error = '$error_msg',
1098                                         last_updated = NOW() WHERE id = '$feed'");
1099                 }
1100
1101                 unset($rss);
1102
1103                 if ($debug_enabled) {
1104                         _debug("update_rss_feed: done");
1105                 }
1106
1107         }
1108
1109         function cache_images($html, $site_url, $debug) {
1110                 $cache_dir = CACHE_DIR . "/images";
1111
1112                 libxml_use_internal_errors(true);
1113
1114                 $charset_hack = '<head>
1115                         <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1116                 </head>';
1117
1118                 $doc = new DOMDocument();
1119                 $doc->loadHTML($charset_hack . $html);
1120                 $xpath = new DOMXPath($doc);
1121
1122                 $entries = $xpath->query('(//img[@src])');
1123
1124                 foreach ($entries as $entry) {
1125                         if ($entry->hasAttribute('src')) {
1126                                 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1127
1128                                 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1129
1130                                 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1131
1132                                 if (!file_exists($local_filename)) {
1133                                         $file_content = fetch_file_contents($src);
1134
1135                                         if ($file_content && strlen($file_content) > 1024) {
1136                                                 file_put_contents($local_filename, $file_content);
1137                                         }
1138                                 }
1139
1140                                 if (file_exists($local_filename)) {
1141                                         $entry->setAttribute('src', SELF_URL_PATH . '/image.php?url=' .
1142                                                 base64_encode($src));
1143                                 }
1144                         }
1145                 }
1146
1147                 $node = $doc->getElementsByTagName('body')->item(0);
1148
1149                 return $doc->saveXML($node);
1150         }
1151
1152         function expire_lock_files($debug) {
1153                 if ($debug) _debug("Removing old lock files...");
1154
1155                 $num_deleted = 0;
1156
1157                 if (is_writable(LOCK_DIRECTORY)) {
1158                         $files = glob(LOCK_DIRECTORY . "/*.lock");
1159
1160                         if ($files) {
1161                                 foreach ($files as $file) {
1162                                         if (!file_is_locked($file) && time() - filemtime($file) > 86400*2) {
1163                                                 unlink($file);
1164                                                 ++$num_deleted;
1165                                         }
1166                                 }
1167                         }
1168                 }
1169
1170                 if ($debug) _debug("Removed $num_deleted files.");
1171         }
1172
1173         function expire_cached_files($debug) {
1174                 foreach (array("simplepie", "images", "export") as $dir) {
1175                         $cache_dir = CACHE_DIR . "/$dir";
1176
1177                         if ($debug) _debug("Expiring $cache_dir");
1178
1179                         $num_deleted = 0;
1180
1181                         if (is_writable($cache_dir)) {
1182                                 $files = glob("$cache_dir/*");
1183
1184                                 if ($files) {
1185                                         foreach ($files as $file) {
1186                                                 if (time() - filemtime($file) > 86400*7) {
1187                                                         unlink($file);
1188
1189                                                         ++$num_deleted;
1190                                                 }
1191                                         }
1192                                 }
1193                         }
1194
1195                         if ($debug) _debug("Removed $num_deleted files.");
1196                 }
1197         }
1198
1199         /**
1200         * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1201         * Returns the url query as associative array
1202         *
1203         * @param    string    query
1204         * @return    array    params
1205         */
1206         function convertUrlQuery($query) {
1207                 $queryParts = explode('&', $query);
1208
1209                 $params = array();
1210
1211                 foreach ($queryParts as $param) {
1212                         $item = explode('=', $param);
1213                         $params[$item[0]] = $item[1];
1214                 }
1215
1216                 return $params;
1217         }
1218
1219         function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
1220                 $matches = array();
1221
1222                 foreach ($filters as $filter) {
1223                         $match_any_rule = $filter["match_any_rule"];
1224                         $inverse = $filter["inverse"];
1225                         $filter_match = false;
1226
1227                         foreach ($filter["rules"] as $rule) {
1228                                 $match = false;
1229                                 $reg_exp = $rule["reg_exp"];
1230                                 $rule_inverse = $rule["inverse"];
1231
1232                                 if (!$reg_exp)
1233                                         continue;
1234
1235                                 switch ($rule["type"]) {
1236                                 case "title":
1237                                         $match = @preg_match("/$reg_exp/i", $title);
1238                                         break;
1239                                 case "content":
1240                                         // we don't need to deal with multiline regexps
1241                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1242
1243                                         $match = @preg_match("/$reg_exp/i", $content);
1244                                         break;
1245                                 case "both":
1246                                         // we don't need to deal with multiline regexps
1247                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1248
1249                                         $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
1250                                         break;
1251                                 case "link":
1252                                         $match = @preg_match("/$reg_exp/i", $link);
1253                                         break;
1254                                 case "author":
1255                                         $match = @preg_match("/$reg_exp/i", $author);
1256                                         break;
1257                                 case "tag":
1258                                         $tag_string = join(",", $tags);
1259                                         $match = @preg_match("/$reg_exp/i", $tag_string);
1260                                         break;
1261                                 }
1262
1263                                 if ($rule_inverse) $match = !$match;
1264
1265                                 if ($match_any_rule) {
1266                                         if ($match) {
1267                                                 $filter_match = true;
1268                                                 break;
1269                                         }
1270                                 } else {
1271                                         $filter_match = $match;
1272                                         if (!$match) {
1273                                                 break;
1274                                         }
1275                                 }
1276                         }
1277
1278                         if ($inverse) $filter_match = !$filter_match;
1279
1280                         if ($filter_match) {
1281                                 foreach ($filter["actions"] AS $action) {
1282                                         array_push($matches, $action);
1283                                 }
1284                         }
1285                 }
1286
1287                 return $matches;
1288         }
1289
1290         function find_article_filter($filters, $filter_name) {
1291                 foreach ($filters as $f) {
1292                         if ($f["type"] == $filter_name) {
1293                                 return $f;
1294                         };
1295                 }
1296                 return false;
1297         }
1298
1299         function find_article_filters($filters, $filter_name) {
1300                 $results = array();
1301
1302                 foreach ($filters as $f) {
1303                         if ($f["type"] == $filter_name) {
1304                                 array_push($results, $f);
1305                         };
1306                 }
1307                 return $results;
1308         }
1309
1310         function calculate_article_score($filters) {
1311                 $score = 0;
1312
1313                 foreach ($filters as $f) {
1314                         if ($f["type"] == "score") {
1315                                 $score += $f["param"];
1316                         };
1317                 }
1318                 return $score;
1319         }
1320
1321         function labels_contains_caption($labels, $caption) {
1322                 foreach ($labels as $label) {
1323                         if ($label[1] == $caption) {
1324                                 return true;
1325                         }
1326                 }
1327
1328                 return false;
1329         }
1330
1331         function assign_article_to_label_filters($link, $id, $filters, $owner_uid, $article_labels) {
1332                 foreach ($filters as $f) {
1333                         if ($f["type"] == "label") {
1334                                 if (!labels_contains_caption($article_labels, $f["param"])) {
1335                                         label_add_article($link, $id, $f["param"], $owner_uid);
1336                                 }
1337                         }
1338                 }
1339         }
1340
1341         function make_guid_from_title($title) {
1342                 return preg_replace("/[ \"\',.:;]/", "-",
1343                         mb_strtolower(strip_tags($title), 'utf-8'));
1344         }
1345
1346
1347 ?>