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
 145                 // For each feed, we call the feed update function.
 146                 while ($line = array_pop($feeds_to_update)) {
 147
 148                         if($debug) _debug("Feed: " . $line["feed_url"] . ", " . $line["last_updated"]);
 149
 150                         update_rss_feed($link, $line["id"], true);
 151
 152                         sleep(1); // prevent flood (FIXME make this an option?)
 153                 }
 154
 155                 // Send feed digests by email if needed.
 156                 send_headlines_digests($link, $debug);
 157
 158         } // function update_daemon_common
 159
 160         // ignore_daemon is not used
 161         function update_rss_feed($link, $feed, $ignore_daemon = false, $no_cache = false,
 162                 $override_url = false) {
 163
 164                 require_once "lib/simplepie/simplepie.inc";
 165                 require_once "lib/magpierss/rss_fetch.inc";
 166                 require_once 'lib/magpierss/rss_utils.inc';
 167
 168                 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
 169
 170                 if ($debug_enabled) {
 171                         _debug("update_rss_feed: start");
 172                 }
 173
 174                 $result = db_query($link, "SELECT id,update_interval,auth_login,
 175                         feed_url,auth_pass,cache_images,update_method,last_updated,cache_content,
 176                         mark_unread_on_update, owner_uid, update_on_checksum_change,
 177                         pubsub_state
 178                         FROM ttrss_feeds WHERE id = '$feed'");
 179
 180                 if (db_num_rows($result) == 0) {
 181                         if ($debug_enabled) {
 182                                 _debug("update_rss_feed: feed $feed NOT FOUND/SKIPPED");
 183                         }
 184                         return false;
 185                 }
 186
 187                 $update_method = db_fetch_result($result, 0, "update_method");
 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                 $update_on_checksum_change = sql_bool_to_bool(db_fetch_result($result,
 193                         0, "update_on_checksum_change"));
 194                 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
 195
 196                 db_query($link, "UPDATE ttrss_feeds SET last_update_started = NOW()
 197                         WHERE id = '$feed'");
 198
 199                 $auth_login = db_fetch_result($result, 0, "auth_login");
 200                 $auth_pass = db_fetch_result($result, 0, "auth_pass");
 201
 202                 if ($update_method == 0)
 203                         $update_method = DEFAULT_UPDATE_METHOD + 1;
 204
 205                 // 1 - Magpie
 206                 // 2 - SimplePie
 207                 // 3 - Twitter OAuth
 208
 209                 if ($update_method == 2)
 210                         $use_simplepie = true;
 211                 else
 212                         $use_simplepie = false;
 213
 214                 if ($debug_enabled) {
 215                         _debug("update method: $update_method (feed setting: $update_method) (use simplepie: $use_simplepie)\n");
 216                 }
 217
 218                 if ($update_method == 1) {
 219                         $auth_login = urlencode($auth_login);
 220                         $auth_pass = urlencode($auth_pass);
 221                 }
 222
 223                 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
 224                 $cache_content = sql_bool_to_bool(db_fetch_result($result, 0, "cache_content"));
 225                 $fetch_url = db_fetch_result($result, 0, "feed_url");
 226
 227                 $feed = db_escape_string($feed);
 228
 229                 if ($auth_login && $auth_pass ){
 230                         $url_parts = array();
 231                         preg_match("/(^[^:]*):\/\/(.*)/", $fetch_url, $url_parts);
 232
 233                         if ($url_parts[1] && $url_parts[2]) {
 234                                 $fetch_url = $url_parts[1] . "://$auth_login:$auth_pass@" . $url_parts[2];
 235                         }
 236
 237                 }
 238
 239                 if ($override_url)
 240                         $fetch_url = $override_url;
 241
 242                 if ($debug_enabled) {
 243                         _debug("update_rss_feed: fetching [$fetch_url]...");
 244                 }
 245
 246                 // Ignore cache if new feed or manual update.
 247                 $cache_age = (is_null($last_updated) || $last_updated == '1970-01-01 00:00:00') ?
 248                         -1 : get_feed_update_interval($link, $feed) * 60;
 249
 250                 if ($update_method == 1) {
 251
 252                         define('MAGPIE_CACHE_AGE', $cache_age);
 253                         define('MAGPIE_CACHE_ON', !$no_cache);
 254                         define('MAGPIE_FETCH_TIME_OUT', $no_cache ? 15 : 60);
 255                         define('MAGPIE_CACHE_DIR', CACHE_DIR . "/magpie");
 256
 257                         $rss = @fetch_rss($fetch_url);
 258                 } else {
 259                         $simplepie_cache_dir = CACHE_DIR . "/simplepie";
 260
 261                         if (!is_dir($simplepie_cache_dir)) {
 262                                 mkdir($simplepie_cache_dir);
 263                         }
 264
 265                         $rss = new SimplePie();
 266                         $rss->set_useragent(SELF_USER_AGENT);
 267                         $rss->set_timeout($no_cache ? 15 : 60);
 268                         $rss->set_feed_url($fetch_url);
 269                         $rss->set_output_encoding('UTF-8');
 270                         //$rss->force_feed(true);
 271
 272                         if ($debug_enabled) {
 273                                 _debug("feed update interval (sec): " .
 274                                         get_feed_update_interval($link, $feed)*60);
 275                         }
 276
 277                         $rss->enable_cache(!$no_cache);
 278
 279                         if (!$no_cache) {
 280                                 $rss->set_cache_location($simplepie_cache_dir);
 281                                 $rss->set_cache_duration($cache_age);
 282                         }
 283
 284                         $rss->init();
 285                 }
 286
 287 //              print_r($rss);
 288
 289                 if ($debug_enabled) {
 290                         _debug("update_rss_feed: fetch done, parsing...");
 291                 }
 292
 293                 $feed = db_escape_string($feed);
 294
 295                 if ($update_method == 2) {
 296                         $fetch_ok = !$rss->error();
 297                 } else {
 298                         $fetch_ok = !!$rss;
 299                 }
 300
 301                 if ($fetch_ok) {
 302
 303                         if ($debug_enabled) {
 304                                 _debug("update_rss_feed: processing feed data...");
 305                         }
 306
 307 //                      db_query($link, "BEGIN");
 308
 309                         if (DB_TYPE == "pgsql") {
 310                                 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
 311                         } else {
 312                                 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
 313                         }
 314
 315                         $result = db_query($link, "SELECT title,icon_url,site_url,owner_uid,
 316                                 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
 317                                                 favicon_needs_check
 318                                 FROM ttrss_feeds WHERE id = '$feed'");
 319
 320                         $registered_title = db_fetch_result($result, 0, "title");
 321                         $orig_icon_url = db_fetch_result($result, 0, "icon_url");
 322                         $orig_site_url = db_fetch_result($result, 0, "site_url");
 323                         $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
 324                                 "favicon_needs_check"));
 325
 326                         $owner_uid = db_fetch_result($result, 0, "owner_uid");
 327
 328                         if ($use_simplepie) {
 329                                 $site_url = db_escape_string(trim($rss->get_link()));
 330                         } else {
 331                                 $site_url = db_escape_string(trim($rss->channel["link"]));
 332                         }
 333
 334                         // weird, weird Magpie
 335                         if (!$use_simplepie) {
 336                                 if (!$site_url) $site_url = db_escape_string($rss->channel["link_"]);
 337                         }
 338
 339                         $site_url = rewrite_relative_url($fetch_url, $site_url);
 340                         $site_url = substr($site_url, 0, 250);
 341
 342                         if ($debug_enabled) {
 343                                 _debug("update_rss_feed: checking favicon...");
 344                         }
 345
 346                         if ($favicon_needs_check) {
 347                                 check_feed_favicon($site_url, $feed, $link);
 348
 349                                 db_query($link, "UPDATE ttrss_feeds SET favicon_last_checked = NOW()
 350                                         WHERE id = '$feed'");
 351                         }
 352
 353                         if (!$registered_title || $registered_title == "[Unknown]") {
 354
 355                                 if ($use_simplepie) {
 356                                         $feed_title = db_escape_string($rss->get_title());
 357                                 } else {
 358                                         $feed_title = db_escape_string($rss->channel["title"]);
 359                                 }
 360
 361                                 if ($debug_enabled) {
 362                                         _debug("update_rss_feed: registering title: $feed_title");
 363                                 }
 364
 365                                 db_query($link, "UPDATE ttrss_feeds SET
 366                                         title = '$feed_title' WHERE id = '$feed'");
 367                         }
 368
 369                         if ($site_url && $orig_site_url != $site_url) {
 370                                 db_query($link, "UPDATE ttrss_feeds SET
 371                                         site_url = '$site_url' WHERE id = '$feed'");
 372                         }
 373
 374 //                      print "I: " . $rss->channel["image"]["url"];
 375
 376                         if (!$use_simplepie) {
 377                                 $icon_url = db_escape_string(trim($rss->image["url"]));
 378                         } else {
 379                                 $icon_url = db_escape_string(trim($rss->get_image_url()));
 380                         }
 381
 382                         $icon_url = rewrite_relative_url($fetch_url, $icon_url);
 383                         $icon_url = substr($icon_url, 0, 250);
 384
 385                         if ($icon_url && $orig_icon_url != $icon_url) {
 386                                 db_query($link, "UPDATE ttrss_feeds SET icon_url = '$icon_url' WHERE id = '$feed'");
 387                         }
 388
 389                         if ($debug_enabled) {
 390                                 _debug("update_rss_feed: loading filters & labels...");
 391                         }
 392
 393                         $filters = load_filters($link, $feed, $owner_uid);
 394                         $labels = get_all_labels($link, $owner_uid);
 395
 396                         if ($debug_enabled) {
 397                                 //print_r($filters);
 398                                 _debug("update_rss_feed: " . count($filters) . " filters loaded.");
 399                         }
 400
 401                         $filter_plugins = array();
 402
 403                         if (defined('_ARTICLE_FILTER_PLUGINS')) {
 404                                 foreach (explode(",", _ARTICLE_FILTER_PLUGINS) as $p) {
 405                                         $pclass = "filter_" . trim($p);
 406
 407                                         if (class_exists($pclass)) {
 408                                                 $plugin = new $pclass($link);
 409                                                 array_push($filter_plugins, $plugin);
 410                                         }
 411                                 }
 412                         }
 413
 414                         if ($debug_enabled) {
 415                                 _debug("update_rss_feed: " . count($filter_plugins) . " filter plugins loaded.");
 416                         }
 417
 418                         if ($use_simplepie) {
 419                                 $iterator = $rss->get_items();
 420                         } else {
 421                                 $iterator = $rss->items;
 422                                 if (!$iterator || !is_array($iterator)) $iterator = $rss->entries;
 423                                 if (!$iterator || !is_array($iterator)) $iterator = $rss;
 424                         }
 425
 426                         if (!is_array($iterator)) {
 427                                 /* db_query($link, "UPDATE ttrss_feeds
 428                                         SET last_error = 'Parse error: can\'t find any articles.'
 429                                         WHERE id = '$feed'"); */
 430
 431                                 // clear any errors and mark feed as updated if fetched okay
 432                                 // even if it's blank
 433
 434                                 if ($debug_enabled) {
 435                                         _debug("update_rss_feed: entry iterator is not an array, no articles?");
 436                                 }
 437
 438                                 db_query($link, "UPDATE ttrss_feeds
 439                                         SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
 440
 441                                 return; // no articles
 442                         }
 443
 444                         if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
 445
 446                                 if ($debug_enabled) _debug("update_rss_feed: checking for PUSH hub...");
 447
 448                                 $feed_hub_url = false;
 449                                 if ($use_simplepie) {
 450                                         $links = $rss->get_links('hub');
 451
 452                                         if ($links && is_array($links)) {
 453                                                 foreach ($links as $l) {
 454                                                         $feed_hub_url = $l;
 455                                                         break;
 456                                                 }
 457                                         }
 458
 459                                 } else {
 460                                         $atom = $rss->channel['atom'];
 461
 462                                         if ($atom) {
 463                                                 if ($atom['link@rel'] == 'hub') {
 464                                                         $feed_hub_url = $atom['link@href'];
 465                                                 }
 466
 467                                                 if (!$feed_hub_url && $atom['link#'] > 1) {
 468                                                         for ($i = 2; $i <= $atom['link#']; $i++) {
 469                                                                 if ($atom["link#$i@rel"] == 'hub') {
 470                                                                         $feed_hub_url = $atom["link#$i@href"];
 471                                                                         break;
 472                                                                 }
 473                                                         }
 474                                                 }
 475                                         } else {
 476                                                 $feed_hub_url = $rss->channel['link_hub'];
 477                                         }
 478                                 }
 479
 480                                 if ($debug_enabled) _debug("update_rss_feed: feed hub url: $feed_hub_url");
 481
 482                                 if ($feed_hub_url && function_exists('curl_init') &&
 483                                         !ini_get("open_basedir")) {
 484
 485                                         require_once 'lib/pubsubhubbub/subscriber.php';
 486
 487                                         $callback_url = get_self_url_prefix() .
 488                                                 "/public.php?op=pubsub&id=$feed";
 489
 490                                         $s = new Subscriber($feed_hub_url, $callback_url);
 491
 492                                         $rc = $s->subscribe($fetch_url);
 493
 494                                         if ($debug_enabled)
 495                                                 _debug("update_rss_feed: feed hub url found, subscribe request sent.");
 496
 497                                         db_query($link, "UPDATE ttrss_feeds SET pubsub_state = 1
 498                                                 WHERE id = '$feed'");
 499                                 }
 500                         }
 501
 502                         if ($debug_enabled) {
 503                                 _debug("update_rss_feed: processing articles...");
 504                         }
 505
 506                         foreach ($iterator as $item) {
 507                                 if ($_REQUEST['xdebug'] == 2) {
 508                                         print_r($item);
 509                                 }
 510
 511                                 if ($use_simplepie) {
 512                                         $entry_guid = $item->get_id();
 513                                         if (!$entry_guid) $entry_guid = $item->get_link();
 514                                         if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
 515
 516                                 } else {
 517
 518                                         $entry_guid = $item["id"];
 519
 520                                         if (!$entry_guid) $entry_guid = $item["guid"];
 521                                         if (!$entry_guid) $entry_guid = $item["about"];
 522                                         if (!$entry_guid) $entry_guid = $item["link"];
 523                                         if (!$entry_guid) $entry_guid = make_guid_from_title($item["title"]);
 524                                 }
 525
 526                                 if ($debug_enabled) {
 527                                         _debug("update_rss_feed: guid $entry_guid");
 528                                 }
 529
 530                                 if (!$entry_guid) continue;
 531
 532                                 $entry_timestamp = "";
 533
 534                                 if ($use_simplepie) {
 535                                         $entry_timestamp = strtotime($item->get_date());
 536                                 } else {
 537                                         $rss_2_date = $item['pubdate'];
 538                                         $rss_1_date = $item['dc']['date'];
 539                                         $atom_date = $item['issued'];
 540                                         if (!$atom_date) $atom_date = $item['updated'];
 541
 542                                         if ($atom_date != "") $entry_timestamp = parse_w3cdtf($atom_date);
 543                                         if ($rss_1_date != "") $entry_timestamp = parse_w3cdtf($rss_1_date);
 544                                         if ($rss_2_date != "") $entry_timestamp = strtotime($rss_2_date);
 545
 546                                 }
 547
 548                                 if ($entry_timestamp == "" || $entry_timestamp == -1 || !$entry_timestamp) {
 549                                         $entry_timestamp = time();
 550                                         $no_orig_date = 'true';
 551                                 } else {
 552                                         $no_orig_date = 'false';
 553                                 }
 554
 555                                 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
 556
 557                                 if ($debug_enabled) {
 558                                         _debug("update_rss_feed: date $entry_timestamp [$entry_timestamp_fmt]");
 559                                 }
 560
 561                                 if ($use_simplepie) {
 562                                         $entry_title = $item->get_title();
 563                                 } else {
 564                                         $entry_title = trim(strip_tags($item["title"]));
 565                                 }
 566
 567                                 if ($use_simplepie) {
 568                                         $entry_link = $item->get_link();
 569                                 } else {
 570                                         // strange Magpie workaround
 571                                         $entry_link = $item["link_"];
 572                                         if (!$entry_link) $entry_link = $item["link"];
 573                                 }
 574
 575                                 $entry_link = rewrite_relative_url($site_url, $entry_link);
 576
 577                                 if ($debug_enabled) {
 578                                         _debug("update_rss_feed: title $entry_title");
 579                                         _debug("update_rss_feed: link $entry_link");
 580                                 }
 581
 582                                 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
 583
 584                                 $entry_link = strip_tags($entry_link);
 585
 586                                 if ($use_simplepie) {
 587                                         $entry_content = $item->get_content();
 588                                         if (!$entry_content) $entry_content = $item->get_description();
 589                                 } else {
 590                                         $entry_content = $item["content:escaped"];
 591
 592                                         if (!$entry_content) $entry_content = $item["content:encoded"];
 593                                         if (!$entry_content && is_array($entry_content)) $entry_content = $item["content"]["encoded"];
 594                                         if (!$entry_content) $entry_content = $item["content"];
 595
 596                                         if (is_array($entry_content)) $entry_content = $entry_content[0];
 597
 598                                         // Magpie bugs are getting ridiculous
 599                                         if (trim($entry_content) == "Array") $entry_content = false;
 600
 601                                         if (!$entry_content) $entry_content = $item["atom_content"];
 602                                         if (!$entry_content) $entry_content = $item["summary"];
 603
 604                                         if (!$entry_content ||
 605                                                 strlen($entry_content) < strlen($item["description"])) {
 606                                                         $entry_content = $item["description"];
 607                                         };
 608
 609                                         // WTF
 610                                         if (is_array($entry_content)) {
 611                                                 $entry_content = $entry_content["encoded"];
 612                                                 if (!$entry_content) $entry_content = $entry_content["escaped"];
 613                                         }
 614                                 }
 615
 616                                 if ($cache_images && is_writable(CACHE_DIR . '/images'))
 617                                         $entry_content = cache_images($entry_content, $site_url, $debug_enabled);
 618
 619                                 if ($_REQUEST["xdebug"] == 2) {
 620                                         print "update_rss_feed: content: ";
 621                                         print $entry_content;
 622                                         print "\n";
 623                                 }
 624
 625                                 $entry_content_unescaped = $entry_content;
 626                                 $entry_cached_content = "";
 627
 628                                 if ($use_simplepie) {
 629                                         $entry_comments = strip_tags($item->data["comments"]);
 630                                         if ($item->get_author()) {
 631                                                 $entry_author_item = $item->get_author();
 632                                                 $entry_author = $entry_author_item->get_name();
 633                                                 if (!$entry_author) $entry_author = $entry_author_item->get_email();
 634
 635                                                 $entry_author = db_escape_string($entry_author);
 636                                         }
 637                                 } else {
 638                                         $entry_comments = strip_tags($item["comments"]);
 639
 640                                         $entry_author = db_escape_string(strip_tags($item['dc']['creator']));
 641
 642                                         if ($item['author']) {
 643
 644                                                 if (is_array($item['author'])) {
 645
 646                                                         if (!$entry_author) {
 647                                                                 $entry_author = db_escape_string(strip_tags($item['author']['name']));
 648                                                         }
 649
 650                                                         if (!$entry_author) {
 651                                                                 $entry_author = db_escape_string(strip_tags($item['author']['email']));
 652                                                         }
 653                                                 }
 654
 655                                                 if (!$entry_author) {
 656                                                         $entry_author = db_escape_string(strip_tags($item['author']));
 657                                                 }
 658                                         }
 659                                 }
 660
 661                                 if (preg_match('/^[\t\n\r ]*$/', $entry_author)) $entry_author = '';
 662
 663                                 $entry_guid = db_escape_string(strip_tags($entry_guid));
 664                                 $entry_guid = mb_substr($entry_guid, 0, 250);
 665
 666                                 $result = db_query($link, "SELECT id FROM       ttrss_entries
 667                                         WHERE guid = '$entry_guid'");
 668
 669                                 $entry_content = db_escape_string($entry_content, false);
 670
 671                                 $entry_title = db_escape_string($entry_title);
 672                                 $entry_link = db_escape_string($entry_link);
 673                                 $entry_comments = mb_substr(db_escape_string($entry_comments), 0, 250);
 674                                 $entry_author = mb_substr($entry_author, 0, 250);
 675
 676                                 if ($use_simplepie) {
 677                                         $num_comments = 0; #FIXME#
 678                                 } else {
 679                                         $num_comments = db_escape_string($item["slash"]["comments"]);
 680                                 }
 681
 682                                 if (!$num_comments) $num_comments = 0;
 683
 684                                 if ($debug_enabled) {
 685                                         _debug("update_rss_feed: looking for tags [1]...");
 686                                 }
 687
 688                                 // parse <category> entries into tags
 689
 690                                 $additional_tags = array();
 691
 692                                 if ($use_simplepie) {
 693
 694                                         $additional_tags_src = $item->get_categories();
 695
 696                                         if (is_array($additional_tags_src)) {
 697                                                 foreach ($additional_tags_src as $tobj) {
 698                                                         array_push($additional_tags, $tobj->get_term());
 699                                                 }
 700                                         }
 701
 702                                         if ($debug_enabled) {
 703                                                 _debug("update_rss_feed: category tags:");
 704                                                 print_r($additional_tags);
 705                                         }
 706
 707                                 } else {
 708
 709                                         $t_ctr = $item['category#'];
 710
 711                                         if ($t_ctr == 0) {
 712                                                 $additional_tags = array();
 713                                         } else if ($t_ctr > 0) {
 714                                                 $additional_tags = array($item['category']);
 715
 716                                                 if ($item['category@term']) {
 717                                                         array_push($additional_tags, $item['category@term']);
 718                                                 }
 719
 720                                                 for ($i = 0; $i <= $t_ctr; $i++ ) {
 721                                                         if ($item["category#$i"]) {
 722                                                                 array_push($additional_tags, $item["category#$i"]);
 723                                                         }
 724
 725                                                         if ($item["category#$i@term"]) {
 726                                                                 array_push($additional_tags, $item["category#$i@term"]);
 727                                                         }
 728                                                 }
 729                                         }
 730
 731                                         // parse <dc:subject> elements
 732
 733                                         $t_ctr = $item['dc']['subject#'];
 734
 735                                         if ($t_ctr > 0) {
 736                                                 array_push($additional_tags, $item['dc']['subject']);
 737
 738                                                 for ($i = 0; $i <= $t_ctr; $i++ ) {
 739                                                         if ($item['dc']["subject#$i"]) {
 740                                                                 array_push($additional_tags, $item['dc']["subject#$i"]);
 741                                                         }
 742                                                 }
 743                                         }
 744                                 }
 745
 746                                 if ($debug_enabled) {
 747                                         _debug("update_rss_feed: looking for tags [2]...");
 748                                 }
 749
 750                                 /* taaaags */
 751                                 // <a href="..." rel="tag">Xorg</a>, //
 752
 753                                 $entry_tags = null;
 754
 755                                 preg_match_all("/<a.*?rel=['\"]tag['\"].*?\>([^<]+)<\/a>/i",
 756                                         $entry_content_unescaped, $entry_tags);
 757
 758                                 $entry_tags = $entry_tags[1];
 759
 760                                 $entry_tags = array_merge($entry_tags, $additional_tags);
 761                                 $entry_tags = array_unique($entry_tags);
 762
 763                                 for ($i = 0; $i < count($entry_tags); $i++)
 764                                         $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
 765
 766                                 if ($debug_enabled) {
 767                                         //_debug("update_rss_feed: unfiltered tags found:");
 768                                         //print_r($entry_tags);
 769                                 }
 770
 771                                 if ($debug_enabled) {
 772                                         _debug("update_rss_feed: done collecting data [TITLE:$entry_title]");
 773                                 }
 774
 775                                 // TODO: less memory-hungry implementation
 776                                 if (count($filter_plugins) > 0) {
 777                                         if ($debug_enabled) {
 778                                                 _debug("update_rss_feed: applying plugin filters...");
 779                                         }
 780
 781                                         $article = array("owner_uid" => $owner_uid,
 782                                                 "title" => $entry_title,
 783                                                 "content" => $entry_content,
 784                                                 "link" => $entry_link,
 785                                                 "tags" => $entry_tags,
 786                                                 "author" => $entry_author);
 787
 788                                         foreach ($filter_plugins as $plugin) {
 789                                                 $article = $plugin->filter_article($article);
 790                                         }
 791
 792                                         $entry_title = $article["title"];
 793                                         $entry_content = $article["content"];
 794                                         $entry_tags = $article["tags"];
 795                                         $entry_author = $article["author"];
 796                                 }
 797
 798                                 $content_hash = "SHA1:" . sha1(strip_tags($entry_content));
 799
 800                                 db_query($link, "BEGIN");
 801
 802                                 if (db_num_rows($result) == 0) {
 803
 804                                         if ($debug_enabled) {
 805                                                 _debug("update_rss_feed: base guid not found");
 806                                         }
 807
 808                                         if ($cache_content) {
 809                                                 if ($debug_enabled) {
 810                                                         _debug("update_rss_feed: caching content...");
 811                                                 }
 812
 813                                                 $entry_cached_content = cache_content($link, $entry_link, $auth_login, $auth_pass);
 814
 815                                                 if ($cache_images && is_writable(CACHE_DIR . '/images'))
 816                                                         $entry_cached_content = cache_images($entry_cached_content, $site_url, $debug_enabled);
 817
 818                                                 $entry_cached_content = db_escape_string($entry_cached_content, false);
 819
 820                                         }
 821
 822                                         // base post entry does not exist, create it
 823
 824                                         $result = db_query($link,
 825                                                 "INSERT INTO ttrss_entries
 826                                                         (title,
 827                                                         guid,
 828                                                         link,
 829                                                         updated,
 830                                                         content,
 831                                                         content_hash,
 832                                                         cached_content,
 833                                                         no_orig_date,
 834                                                         date_updated,
 835                                                         date_entered,
 836                                                         comments,
 837                                                         num_comments,
 838                                                         author)
 839                                                 VALUES
 840                                                         ('$entry_title',
 841                                                         '$entry_guid',
 842                                                         '$entry_link',
 843                                                         '$entry_timestamp_fmt',
 844                                                         '$entry_content',
 845                                                         '$content_hash',
 846                                                         '$entry_cached_content',
 847                                                         $no_orig_date,
 848                                                         NOW(),
 849                                                         NOW(),
 850                                                         '$entry_comments',
 851                                                         '$num_comments',
 852                                                         '$entry_author')");
 853
 854                                         $article_labels = array();
 855
 856                                 } else {
 857                                         // we keep encountering the entry in feeds, so we need to
 858                                         // update date_updated column so that we don't get horrible
 859                                         // dupes when the entry gets purged and reinserted again e.g.
 860                                         // in the case of SLOW SLOW OMG SLOW updating feeds
 861
 862                                         $base_entry_id = db_fetch_result($result, 0, "id");
 863
 864                                         db_query($link, "UPDATE ttrss_entries SET date_updated = NOW()
 865                                                 WHERE id = '$base_entry_id'");
 866
 867                                         $article_labels = get_article_labels($link, $base_entry_id, $owner_uid);
 868                                 }
 869
 870                                 // now it should exist, if not - bad luck then
 871
 872                                 $result = db_query($link, "SELECT
 873                                                 id,content_hash,no_orig_date,title,
 874                                                 ".SUBSTRING_FOR_DATE."(date_updated,1,19) as date_updated,
 875                                                 ".SUBSTRING_FOR_DATE."(updated,1,19) as updated,
 876                                                 num_comments
 877                                         FROM
 878                                                 ttrss_entries
 879                                         WHERE guid = '$entry_guid'");
 880
 881                                 $entry_ref_id = 0;
 882                                 $entry_int_id = 0;
 883
 884                                 if (db_num_rows($result) == 1) {
 885
 886                                         if ($debug_enabled) {
 887                                                 _debug("update_rss_feed: base guid found, checking for user record");
 888                                         }
 889
 890                                         // this will be used below in update handler
 891                                         $orig_content_hash = db_fetch_result($result, 0, "content_hash");
 892                                         $orig_title = db_fetch_result($result, 0, "title");
 893                                         $orig_num_comments = db_fetch_result($result, 0, "num_comments");
 894                                         $orig_date_updated = strtotime(db_fetch_result($result,
 895                                                 0, "date_updated"));
 896
 897                                         $ref_id = db_fetch_result($result, 0, "id");
 898                                         $entry_ref_id = $ref_id;
 899
 900                                         // check for user post link to main table
 901
 902                                         // do we allow duplicate posts with same GUID in different feeds?
 903                                         if (get_pref($link, "ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
 904                                                 $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
 905                                         } else {
 906                                                 $dupcheck_qpart = "";
 907                                         }
 908
 909                                         /* Collect article tags here so we could filter by them: */
 910
 911                                         $article_filters = get_article_filters($filters, $entry_title,
 912                                                 $entry_content, $entry_link, $entry_timestamp, $entry_author,
 913                                                 $entry_tags);
 914
 915                                         if ($debug_enabled) {
 916                                                 _debug("update_rss_feed: article filters: ");
 917                                                 if (count($article_filters) != 0) {
 918                                                         print_r($article_filters);
 919                                                 }
 920                                         }
 921
 922                                         if (find_article_filter($article_filters, "filter")) {
 923                                                 db_query($link, "COMMIT"); // close transaction in progress
 924                                                 continue;
 925                                         }
 926
 927                                         $score = calculate_article_score($article_filters);
 928
 929                                         if ($debug_enabled) {
 930                                                 _debug("update_rss_feed: initial score: $score");
 931                                         }
 932
 933                                         $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
 934                                                         ref_id = '$ref_id' AND owner_uid = '$owner_uid'
 935                                                         $dupcheck_qpart";
 936
 937 //                                      if ($_REQUEST["xdebug"]) print "$query\n";
 938
 939                                         $result = db_query($link, $query);
 940
 941                                         // okay it doesn't exist - create user entry
 942                                         if (db_num_rows($result) == 0) {
 943
 944                                                 if ($debug_enabled) {
 945                                                         _debug("update_rss_feed: user record not found, creating...");
 946                                                 }
 947
 948                                                 if ($score >= -500 && !find_article_filter($article_filters, 'catchup')) {
 949                                                         $unread = 'true';
 950                                                         $last_read_qpart = 'NULL';
 951                                                 } else {
 952                                                         $unread = 'false';
 953                                                         $last_read_qpart = 'NOW()';
 954                                                 }
 955
 956                                                 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
 957                                                         $marked = 'true';
 958                                                 } else {
 959                                                         $marked = 'false';
 960                                                 }
 961
 962                                                 if (find_article_filter($article_filters, 'publish')) {
 963                                                         $published = 'true';
 964                                                 } else {
 965                                                         $published = 'false';
 966                                                 }
 967
 968                                                 // N-grams
 969
 970                                                 if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
 971
 972                                                         $result = db_query($link, "SELECT COUNT(*) AS similar FROM
 973                                                                         ttrss_entries,ttrss_user_entries
 974                                                                 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
 975                                                                         AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
 976                                                                         AND owner_uid = $owner_uid");
 977
 978                                                         $ngram_similar = db_fetch_result($result, 0, "similar");
 979
 980                                                         if ($debug_enabled) {
 981                                                                 _debug("update_rss_feed: N-gram similar results: $ngram_similar");
 982                                                         }
 983
 984                                                         if ($ngram_similar > 0) {
 985                                                                 $unread = 'false';
 986                                                         }
 987                                                 }
 988
 989                                                 $result = db_query($link,
 990                                                         "INSERT INTO ttrss_user_entries
 991                                                                 (ref_id, owner_uid, feed_id, unread, last_read, marked,
 992                                                                         published, score, tag_cache, label_cache, uuid)
 993                                                         VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
 994                                                                 $last_read_qpart, $marked, $published, '$score', '', '', '')");
 995
 996                                                 if (PUBSUBHUBBUB_HUB && $published == 'true') {
 997                                                         $rss_link = get_self_url_prefix() .
 998                                                                 "/public.php?op=rss&id=-2&key=" .
 999                                                                 get_feed_access_key($link, -2, false, $owner_uid);
1000
1001                                                         $p = new Publisher(PUBSUBHUBBUB_HUB);
1002
1003                                                         $pubsub_result = $p->publish_update($rss_link);
1004                                                 }
1005
1006                                                 $result = db_query($link,
1007                                                         "SELECT int_id FROM ttrss_user_entries WHERE
1008                                                                 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
1009                                                                 feed_id = '$feed' LIMIT 1");
1010
1011                                                 if (db_num_rows($result) == 1) {
1012                                                         $entry_int_id = db_fetch_result($result, 0, "int_id");
1013                                                 }
1014                                         } else {
1015                                                 if ($debug_enabled) {
1016                                                         _debug("update_rss_feed: user record FOUND");
1017                                                 }
1018
1019                                                 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
1020                                                 $entry_int_id = db_fetch_result($result, 0, "int_id");
1021                                         }
1022
1023                                         if ($debug_enabled) {
1024                                                 _debug("update_rss_feed: RID: $entry_ref_id, IID: $entry_int_id");
1025                                         }
1026
1027                                         $post_needs_update = false;
1028                                         $update_insignificant = false;
1029
1030                                         if ($orig_num_comments != $num_comments) {
1031                                                 $post_needs_update = true;
1032                                                 $update_insignificant = true;
1033                                         }
1034
1035                                         if ($content_hash != $orig_content_hash) {
1036                                                 $post_needs_update = true;
1037                                                 $update_insignificant = false;
1038
1039                                                 if ($cache_content) {
1040                                                         if ($debug_enabled) {
1041                                                                 _debug("update_rss_feed: caching content because original checksum changed...");
1042                                                         }
1043
1044                                                         $entry_cached_content = cache_content($link, $entry_link, $auth_login, $auth_pass);
1045
1046                                                         if ($cache_images && is_writable(CACHE_DIR . '/images'))
1047                                                                 $entry_cached_content = cache_images($entry_cached_content, $site_url, $debug_enabled);
1048
1049                                                         $entry_cached_content = db_escape_string($entry_cached_content, false);
1050                                                 }
1051                                         }
1052
1053                                         if (db_escape_string($orig_title) != $entry_title) {
1054                                                 $post_needs_update = true;
1055                                                 $update_insignificant = false;
1056                                         }
1057
1058                                         // if post needs update, update it and mark all user entries
1059                                         // linking to this post as updated
1060                                         if ($post_needs_update) {
1061
1062                                                 if (defined('DAEMON_EXTENDED_DEBUG')) {
1063                                                         _debug("update_rss_feed: post $entry_guid needs update...");
1064                                                 }
1065
1066 //                                              print "<!-- post $orig_title needs update : $post_needs_update -->";
1067
1068                                                 db_query($link, "UPDATE ttrss_entries
1069                                                         SET title = '$entry_title', content = '$entry_content',
1070                                                                 content_hash = '$content_hash',
1071                                                                 cached_content = '$entry_cached_content',
1072                                                                 updated = '$entry_timestamp_fmt',
1073                                                                 num_comments = '$num_comments'
1074                                                         WHERE id = '$ref_id'");
1075
1076                                                 if (!$update_insignificant) {
1077                                                         if ($mark_unread_on_update) {
1078                                                                 db_query($link, "UPDATE ttrss_user_entries
1079                                                                         SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
1080                                                         } else if ($update_on_checksum_change) {
1081                                                                 db_query($link, "UPDATE ttrss_user_entries
1082                                                                         SET last_read = null WHERE ref_id = '$ref_id'
1083                                                                                 AND unread = false");
1084                                                         }
1085                                                 }
1086                                         }
1087                                 }
1088
1089                                 db_query($link, "COMMIT");
1090
1091                                 if ($debug_enabled) {
1092                                         _debug("update_rss_feed: assigning labels...");
1093                                 }
1094
1095                                 assign_article_to_label_filters($link, $entry_ref_id, $article_filters,
1096                                         $owner_uid, $article_labels);
1097
1098                                 if ($debug_enabled) {
1099                                         _debug("update_rss_feed: looking for enclosures...");
1100                                 }
1101
1102                                 // enclosures
1103
1104                                 $enclosures = array();
1105
1106                                 if ($use_simplepie) {
1107                                         $encs = $item->get_enclosures();
1108
1109                                         if (is_array($encs)) {
1110                                                 foreach ($encs as $e) {
1111                                                         $e_item = array(
1112                                                                 $e->link, $e->type, $e->length);
1113
1114                                                         array_push($enclosures, $e_item);
1115                                                 }
1116                                         }
1117
1118                                 } else {
1119                                         // <enclosure>
1120
1121                                         $e_ctr = $item['enclosure#'];
1122
1123                                         if ($e_ctr > 0) {
1124                                                 $e_item = array($item['enclosure@url'],
1125                                                         $item['enclosure@type'],
1126                                                         $item['enclosure@length']);
1127
1128                                                 array_push($enclosures, $e_item);
1129
1130                                                 for ($i = 0; $i <= $e_ctr; $i++ ) {
1131
1132                                                         if ($item["enclosure#$i@url"]) {
1133                                                                 $e_item = array($item["enclosure#$i@url"],
1134                                                                         $item["enclosure#$i@type"],
1135                                                                         $item["enclosure#$i@length"]);
1136                                                                 array_push($enclosures, $e_item);
1137                                                         }
1138                                                 }
1139                                         }
1140
1141                                         // <media:content>
1142                                         // can there be many of those? yes -fox
1143
1144                                         $m_ctr = $item['media']['content#'];
1145
1146                                         if ($m_ctr > 0) {
1147                                                 $e_item = array($item['media']['content@url'],
1148                                                         $item['media']['content@medium'],
1149                                                         $item['media']['content@length']);
1150
1151                                                 array_push($enclosures, $e_item);
1152
1153                                                 for ($i = 0; $i <= $m_ctr; $i++ ) {
1154
1155                                                         if ($item["media"]["content#$i@url"]) {
1156                                                                 $e_item = array($item["media"]["content#$i@url"],
1157                                                                         $item["media"]["content#$i@medium"],
1158                                                                         $item["media"]["content#$i@length"]);
1159                                                                 array_push($enclosures, $e_item);
1160                                                         }
1161                                                 }
1162
1163                                         }
1164                                 }
1165
1166
1167                                 if ($debug_enabled) {
1168                                         _debug("update_rss_feed: article enclosures:");
1169                                         print_r($enclosures);
1170                                 }
1171
1172                                 db_query($link, "BEGIN");
1173
1174                                 foreach ($enclosures as $enc) {
1175                                         $enc_url = db_escape_string($enc[0]);
1176                                         $enc_type = db_escape_string($enc[1]);
1177                                         $enc_dur = db_escape_string($enc[2]);
1178
1179                                         $result = db_query($link, "SELECT id FROM ttrss_enclosures
1180                                                 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
1181
1182                                         if (db_num_rows($result) == 0) {
1183                                                 db_query($link, "INSERT INTO ttrss_enclosures
1184                                                         (content_url, content_type, title, duration, post_id) VALUES
1185                                                         ('$enc_url', '$enc_type', '', '$enc_dur', '$entry_ref_id')");
1186                                         }
1187                                 }
1188
1189                                 db_query($link, "COMMIT");
1190
1191                                 // check for manual tags (we have to do it here since they're loaded from filters)
1192
1193                                 foreach ($article_filters as $f) {
1194                                         if ($f["type"] == "tag") {
1195
1196                                                 $manual_tags = trim_array(explode(",", $f["param"]));
1197
1198                                                 foreach ($manual_tags as $tag) {
1199                                                         if (tag_is_valid($tag)) {
1200                                                                 array_push($entry_tags, $tag);
1201                                                         }
1202                                                 }
1203                                         }
1204                                 }
1205
1206                                 // Skip boring tags
1207
1208                                 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref($link,
1209                                         'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1210
1211                                 $filtered_tags = array();
1212                                 $tags_to_cache = array();
1213
1214                                 if ($entry_tags && is_array($entry_tags)) {
1215                                         foreach ($entry_tags as $tag) {
1216                                                 if (array_search($tag, $boring_tags) === false) {
1217                                                         array_push($filtered_tags, $tag);
1218                                                 }
1219                                         }
1220                                 }
1221
1222                                 $filtered_tags = array_unique($filtered_tags);
1223
1224                                 if ($debug_enabled) {
1225                                         _debug("update_rss_feed: filtered article tags:");
1226                                         print_r($filtered_tags);
1227                                 }
1228
1229                                 // Save article tags in the database
1230
1231                                 if (count($filtered_tags) > 0) {
1232
1233                                         db_query($link, "BEGIN");
1234
1235                                         foreach ($filtered_tags as $tag) {
1236
1237                                                 $tag = sanitize_tag($tag);
1238                                                 $tag = db_escape_string($tag);
1239
1240                                                 if (!tag_is_valid($tag)) continue;
1241
1242                                                 $result = db_query($link, "SELECT id FROM ttrss_tags
1243                                                         WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1244                                                         owner_uid = '$owner_uid' LIMIT 1");
1245
1246                                                         if ($result && db_num_rows($result) == 0) {
1247
1248                                                                 db_query($link, "INSERT INTO ttrss_tags
1249                                                                         (owner_uid,tag_name,post_int_id)
1250                                                                         VALUES ('$owner_uid','$tag', '$entry_int_id')");
1251                                                         }
1252
1253                                                 array_push($tags_to_cache, $tag);
1254                                         }
1255
1256                                         /* update the cache */
1257
1258                                         $tags_to_cache = array_unique($tags_to_cache);
1259
1260                                         $tags_str = db_escape_string(join(",", $tags_to_cache));
1261
1262                                         db_query($link, "UPDATE ttrss_user_entries
1263                                                 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1264                                                 AND owner_uid = $owner_uid");
1265
1266                                         db_query($link, "COMMIT");
1267                                 }
1268
1269                                 if (get_pref($link, "AUTO_ASSIGN_LABELS", $owner_uid, false)) {
1270                                         if ($debug_enabled) {
1271                                                 _debug("update_rss_feed: auto-assigning labels...");
1272                                         }
1273
1274                                         foreach ($labels as $label) {
1275                                                 $caption = $label["caption"];
1276
1277                                                 if (preg_match("/\b$caption\b/i", "$tags_str " . strip_tags($entry_content) . " $entry_title")) {
1278                                                         if (!labels_contains_caption($article_labels, $caption)) {
1279                                                                 label_add_article($link, $entry_ref_id, $caption, $owner_uid);
1280                                                         }
1281                                                 }
1282                                         }
1283                                 }
1284
1285                                 if ($debug_enabled) {
1286                                         _debug("update_rss_feed: article processed");
1287                                 }
1288                         }
1289
1290                         if (!$last_updated) {
1291                                 if ($debug_enabled) {
1292                                         _debug("update_rss_feed: new feed, catching it up...");
1293                                 }
1294                                 catchup_feed($link, $feed, false, $owner_uid);
1295                         }
1296
1297                         if ($debug_enabled) {
1298                                 _debug("purging feed...");
1299                         }
1300
1301                         purge_feed($link, $feed, 0, $debug_enabled);
1302
1303                         db_query($link, "UPDATE ttrss_feeds
1304                                 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
1305
1306 //                      db_query($link, "COMMIT");
1307
1308                 } else {
1309
1310                         if ($use_simplepie) {
1311                                 $error_msg = mb_substr($rss->error(), 0, 250);
1312                         } else {
1313                                 $error_msg = mb_substr(magpie_error(), 0, 250);
1314                         }
1315
1316                         if ($debug_enabled) {
1317                                 _debug("update_rss_feed: error fetching feed: $error_msg");
1318                         }
1319
1320                         $error_msg = db_escape_string($error_msg);
1321
1322                         db_query($link,
1323                                 "UPDATE ttrss_feeds SET last_error = '$error_msg',
1324                                         last_updated = NOW() WHERE id = '$feed'");
1325                 }
1326
1327                 if ($use_simplepie) {
1328                         unset($rss);
1329                 }
1330
1331                 if ($debug_enabled) {
1332                         _debug("update_rss_feed: done");
1333                 }
1334
1335         }
1336
1337         function cache_images($html, $site_url, $debug) {
1338                 $cache_dir = CACHE_DIR . "/images";
1339
1340                 libxml_use_internal_errors(true);
1341
1342                 $charset_hack = '<head>
1343                         <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1344                 </head>';
1345
1346                 $doc = new DOMDocument();
1347                 $doc->loadHTML($charset_hack . $html);
1348                 $xpath = new DOMXPath($doc);
1349
1350                 $entries = $xpath->query('(//img[@src])');
1351
1352                 foreach ($entries as $entry) {
1353                         if ($entry->hasAttribute('src')) {
1354                                 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1355
1356                                 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1357
1358                                 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1359
1360                                 if (!file_exists($local_filename)) {
1361                                         $file_content = fetch_file_contents($src);
1362
1363                                         if ($file_content && strlen($file_content) > 1024) {
1364                                                 file_put_contents($local_filename, $file_content);
1365                                         }
1366                                 }
1367
1368                                 if (file_exists($local_filename)) {
1369                                         $entry->setAttribute('src', SELF_URL_PATH . '/image.php?url=' .
1370                                                 base64_encode($src));
1371                                 }
1372                         }
1373                 }
1374
1375                 $node = $doc->getElementsByTagName('body')->item(0);
1376
1377                 return $doc->saveXML($node, LIBXML_NOEMPTYTAG);
1378         }
1379
1380         function expire_cached_files($debug) {
1381                 foreach (array("magpie", "simplepie", "images", "export") as $dir) {
1382                         $cache_dir = CACHE_DIR . "/$dir";
1383
1384                         if ($debug) _debug("Expiring $cache_dir");
1385
1386                         $num_deleted = 0;
1387
1388                         if (is_writable($cache_dir)) {
1389                                 $files = glob("$cache_dir/*");
1390
1391                                 if ($files)
1392                                         foreach ($files as $file) {
1393                                                 if (time() - filemtime($file) > 86400*7) {
1394                                                         unlink($file);
1395
1396                                                         ++$num_deleted;
1397                                                 }
1398                                         }
1399                                 }
1400
1401                         if ($debug) _debug("Removed $num_deleted files.");
1402                 }
1403         }
1404
1405         /**
1406         * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1407         * Returns the url query as associative array
1408         *
1409         * @param    string    query
1410         * @return    array    params
1411         */
1412         function convertUrlQuery($query) {
1413                 $queryParts = explode('&', $query);
1414
1415                 $params = array();
1416
1417                 foreach ($queryParts as $param) {
1418                         $item = explode('=', $param);
1419                         $params[$item[0]] = $item[1];
1420                 }
1421
1422                 return $params;
1423         }
1424
1425         function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
1426                 $matches = array();
1427
1428                 foreach ($filters as $filter) {
1429                         $match_any_rule = $filter["match_any_rule"];
1430                         $filter_match = false;
1431
1432                         foreach ($filter["rules"] as $rule) {
1433                                 $match = false;
1434                                 $reg_exp = $rule["reg_exp"];
1435
1436                                 if (!$reg_exp)
1437                                         continue;
1438
1439                                 switch ($rule["type"]) {
1440                                 case "title":
1441                                         $match = @preg_match("/$reg_exp/i", $title);
1442                                         break;
1443                                 case "content":
1444                                         // we don't need to deal with multiline regexps
1445                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1446
1447                                         $match = @preg_match("/$reg_exp/i", $content);
1448                                         break;
1449                                 case "both":
1450                                         // we don't need to deal with multiline regexps
1451                                         $content = preg_replace("/[\r\n\t]/", "", $content);
1452
1453                                         $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
1454                                         break;
1455                                 case "link":
1456                                         $match = @preg_match("/$reg_exp/i", $link);
1457                                         break;
1458                                 case "author":
1459                                         $match = @preg_match("/$reg_exp/i", $author);
1460                                         break;
1461                                 case "tag":
1462                                         $tag_string = join(",", $tags);
1463                                         $match = @preg_match("/$reg_exp/i", $tag_string);
1464                                         break;
1465                                 }
1466
1467                                 if ($match_any_rule) {
1468                                         if ($match) {
1469                                                 $filter_match = true;
1470                                                 break;
1471                                         }
1472                                 } else {
1473                                         $filter_match = $match;
1474                                         if (!$match) {
1475                                                 break;
1476                                         }
1477                                 }
1478                         }
1479
1480                         if ($filter_match) {
1481                                 foreach ($filter["actions"] AS $action) {
1482                                         array_push($matches, $action);
1483                                 }
1484                         }
1485                 }
1486
1487                 return $matches;
1488         }
1489
1490         function find_article_filter($filters, $filter_name) {
1491                 foreach ($filters as $f) {
1492                         if ($f["type"] == $filter_name) {
1493                                 return $f;
1494                         };
1495                 }
1496                 return false;
1497         }
1498
1499         function find_article_filters($filters, $filter_name) {
1500                 $results = array();
1501
1502                 foreach ($filters as $f) {
1503                         if ($f["type"] == $filter_name) {
1504                                 array_push($results, $f);
1505                         };
1506                 }
1507                 return $results;
1508         }
1509
1510         function calculate_article_score($filters) {
1511                 $score = 0;
1512
1513                 foreach ($filters as $f) {
1514                         if ($f["type"] == "score") {
1515                                 $score += $f["param"];
1516                         };
1517                 }
1518                 return $score;
1519         }
1520
1521         function labels_contains_caption($labels, $caption) {
1522                 foreach ($labels as $label) {
1523                         if ($label[1] == $caption) {
1524                                 return true;
1525                         }
1526                 }
1527
1528                 return false;
1529         }
1530
1531         function assign_article_to_label_filters($link, $id, $filters, $owner_uid, $article_labels) {
1532                 foreach ($filters as $f) {
1533                         if ($f["type"] == "label") {
1534                                 if (!labels_contains_caption($article_labels, $f["param"])) {
1535                                         label_add_article($link, $id, $f["param"], $owner_uid);
1536                                 }
1537                         }
1538                 }
1539         }
1540
1541         function cache_content($link, $url, $login, $pass) {
1542
1543                 $content = fetch_file_contents($url, $login, $pass);
1544
1545                 if ($content) {
1546                         $doc = new DOMDocument();
1547                         @$doc->loadHTML($content);
1548                         $xpath = new DOMXPath($doc);
1549
1550                         $node = $doc->getElementsByTagName('body')->item(0);
1551
1552                         if ($node) {
1553                                 $content = $doc->saveXML($node, LIBXML_NOEMPTYTAG);
1554
1555                                 return $content;
1556                         }
1557                 }
1558
1559                 return "";
1560         }
1561 ?>