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