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, $debug);
 157
 158         } // function update_daemon_common
 159
 160         function update_rss_feed($link, $feed, $ignore_daemon = false, $no_cache = false,
 161                 $override_url = false) {
 162
 163                 require_once "lib/simplepie/simplepie.inc";
 164                 require_once "lib/magpierss/rss_fetch.inc";
 165                 require_once 'lib/magpierss/rss_utils.inc';
 166
 167                 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
 168
 169                 if (!$_REQUEST["daemon"] && !$ignore_daemon) {
 170                         return false;
 171                 }
 172
 173                 if ($debug_enabled) {
 174                         _debug("update_rss_feed: start");
 175                 }
 176
 177                 if (!$ignore_daemon) {
 178
 179                         if (DB_TYPE == "pgsql") {
 180                                         $updstart_thresh_qpart = "(ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '120 seconds')";
 181                                 } else {
 182                                         $updstart_thresh_qpart = "(ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 120 SECOND))";
 183                                 }
 184
 185                         $result = db_query($link, "SELECT id,update_interval,auth_login,
 186                                 auth_pass,cache_images,update_method,last_updated
 187                                 FROM ttrss_feeds WHERE id = '$feed' AND $updstart_thresh_qpart");
 188
 189                 } else {
 190
 191                         $result = db_query($link, "SELECT id,update_interval,auth_login,
 192                                 feed_url,auth_pass,cache_images,update_method,last_updated,
 193                                 mark_unread_on_update, owner_uid, update_on_checksum_change,
 194                                 pubsub_state
 195                                 FROM ttrss_feeds WHERE id = '$feed'");
 196
 197                 }
 198
 199                 if (db_num_rows($result) == 0) {
 200                         if ($debug_enabled) {
 201                                 _debug("update_rss_feed: feed $feed NOT FOUND/SKIPPED");
 202                         }
 203                         return false;
 204                 }
 205
 206                 $update_method = db_fetch_result($result, 0, "update_method");
 207                 $last_updated = db_fetch_result($result, 0, "last_updated");
 208                 $owner_uid = db_fetch_result($result, 0, "owner_uid");
 209                 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
 210                         0, "mark_unread_on_update"));
 211                 $update_on_checksum_change = sql_bool_to_bool(db_fetch_result($result,
 212                         0, "update_on_checksum_change"));
 213                 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
 214
 215                 db_query($link, "UPDATE ttrss_feeds SET last_update_started = NOW()
 216                         WHERE id = '$feed'");
 217
 218                 $auth_login = db_fetch_result($result, 0, "auth_login");
 219                 $auth_pass = db_fetch_result($result, 0, "auth_pass");
 220
 221                 if ($update_method == 0)
 222                         $update_method = DEFAULT_UPDATE_METHOD + 1;
 223
 224                 // 1 - Magpie
 225                 // 2 - SimplePie
 226                 // 3 - Twitter OAuth
 227
 228                 if ($update_method == 2)
 229                         $use_simplepie = true;
 230                 else
 231                         $use_simplepie = false;
 232
 233                 if ($debug_enabled) {
 234                         _debug("update method: $update_method (feed setting: $update_method) (use simplepie: $use_simplepie)\n");
 235                 }
 236
 237                 if ($update_method == 1) {
 238                         $auth_login = urlencode($auth_login);
 239                         $auth_pass = urlencode($auth_pass);
 240                 }
 241
 242                 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
 243                 $fetch_url = db_fetch_result($result, 0, "feed_url");
 244
 245                 $feed = db_escape_string($feed);
 246
 247                 if ($auth_login && $auth_pass ){
 248                         $url_parts = array();
 249                         preg_match("/(^[^:]*):\/\/(.*)/", $fetch_url, $url_parts);
 250
 251                         if ($url_parts[1] && $url_parts[2]) {
 252                                 $fetch_url = $url_parts[1] . "://$auth_login:$auth_pass@" . $url_parts[2];
 253                         }
 254
 255                 }
 256
 257                 if ($override_url)
 258                         $fetch_url = $override_url;
 259
 260                 if ($debug_enabled) {
 261                         _debug("update_rss_feed: fetching [$fetch_url]...");
 262                 }
 263
 264                 // Ignore cache if new feed or manual update.
 265                 $cache_age = (is_null($last_updated) || $last_updated == '1970-01-01 00:00:00') ?
 266                         -1 : get_feed_update_interval($link, $feed) * 60;
 267
 268                 if ($update_method == 1) {
 269
 270                         define('MAGPIE_CACHE_AGE', $cache_age);
 271                         define('MAGPIE_CACHE_ON', !$no_cache);
 272                         define('MAGPIE_FETCH_TIME_OUT', $no_cache ? 15 : 60);
 273                         define('MAGPIE_CACHE_DIR', CACHE_DIR . "/magpie");
 274
 275                         $rss = @fetch_rss($fetch_url);
 276                 } else {
 277                         $simplepie_cache_dir = CACHE_DIR . "/simplepie";
 278
 279                         if (!is_dir($simplepie_cache_dir)) {
 280                                 mkdir($simplepie_cache_dir);
 281                         }
 282
 283                         $rss = new SimplePie();
 284                         $rss->set_useragent(SELF_USER_AGENT);
 285                         $rss->set_timeout($no_cache ? 15 : 60);
 286                         $rss->set_feed_url($fetch_url);
 287                         $rss->set_output_encoding('UTF-8');
 288                         //$rss->force_feed(true);
 289
 290                         if ($debug_enabled) {
 291                                 _debug("feed update interval (sec): " .
 292                                         get_feed_update_interval($link, $feed)*60);
 293                         }
 294
 295                         $rss->enable_cache(!$no_cache);
 296
 297                         if (!$no_cache) {
 298                                 $rss->set_cache_location($simplepie_cache_dir);
 299                                 $rss->set_cache_duration($cache_age);
 300                         }
 301
 302                         $rss->init();
 303                 }
 304
 305 //              print_r($rss);
 306
 307                 if ($debug_enabled) {
 308                         _debug("update_rss_feed: fetch done, parsing...");
 309                 }
 310
 311                 $feed = db_escape_string($feed);
 312
 313                 if ($update_method == 2) {
 314                         $fetch_ok = !$rss->error();
 315                 } else {
 316                         $fetch_ok = !!$rss;
 317                 }
 318
 319                 if ($fetch_ok) {
 320
 321                         if ($debug_enabled) {
 322                                 _debug("update_rss_feed: processing feed data...");
 323                         }
 324
 325 //                      db_query($link, "BEGIN");
 326
 327                         if (DB_TYPE == "pgsql") {
 328                                 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
 329                         } else {
 330                                 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
 331                         }
 332
 333                         $result = db_query($link, "SELECT title,icon_url,site_url,owner_uid,
 334                                 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
 335                                                 favicon_needs_check
 336                                 FROM ttrss_feeds WHERE id = '$feed'");
 337
 338                         $registered_title = db_fetch_result($result, 0, "title");
 339                         $orig_icon_url = db_fetch_result($result, 0, "icon_url");
 340                         $orig_site_url = db_fetch_result($result, 0, "site_url");
 341                         $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
 342                                 "favicon_needs_check"));
 343
 344                         $owner_uid = db_fetch_result($result, 0, "owner_uid");
 345
 346                         if ($use_simplepie) {
 347                                 $site_url = db_escape_string(trim($rss->get_link()));
 348                         } else {
 349                                 $site_url = db_escape_string(trim($rss->channel["link"]));
 350                         }
 351
 352                         // weird, weird Magpie
 353                         if (!$use_simplepie) {
 354                                 if (!$site_url) $site_url = db_escape_string($rss->channel["link_"]);
 355                         }
 356
 357                         $site_url = rewrite_relative_url($fetch_url, $site_url);
 358                         $site_url = substr($site_url, 0, 250);
 359
 360                         if ($debug_enabled) {
 361                                 _debug("update_rss_feed: checking favicon...");
 362                         }
 363
 364                         if ($favicon_needs_check) {
 365                                 check_feed_favicon($site_url, $feed, $link);
 366
 367                                 db_query($link, "UPDATE ttrss_feeds SET favicon_last_checked = NOW()
 368                                         WHERE id = '$feed'");
 369                         }
 370
 371                         if (!$registered_title || $registered_title == "[Unknown]") {
 372
 373                                 if ($use_simplepie) {
 374                                         $feed_title = db_escape_string($rss->get_title());
 375                                 } else {
 376                                         $feed_title = db_escape_string($rss->channel["title"]);
 377                                 }
 378
 379                                 if ($debug_enabled) {
 380                                         _debug("update_rss_feed: registering title: $feed_title");
 381                                 }
 382
 383                                 db_query($link, "UPDATE ttrss_feeds SET
 384                                         title = '$feed_title' WHERE id = '$feed'");
 385                         }
 386
 387                         if ($site_url && $orig_site_url != $site_url) {
 388                                 db_query($link, "UPDATE ttrss_feeds SET
 389                                         site_url = '$site_url' WHERE id = '$feed'");
 390                         }
 391
 392 //                      print "I: " . $rss->channel["image"]["url"];
 393
 394                         if (!$use_simplepie) {
 395                                 $icon_url = db_escape_string(trim($rss->image["url"]));
 396                         } else {
 397                                 $icon_url = db_escape_string(trim($rss->get_image_url()));
 398                         }
 399
 400                         $icon_url = rewrite_relative_url($fetch_url, $icon_url);
 401                         $icon_url = substr($icon_url, 0, 250);
 402
 403                         if ($icon_url && $orig_icon_url != $icon_url) {
 404                                 db_query($link, "UPDATE ttrss_feeds SET icon_url = '$icon_url' WHERE id = '$feed'");
 405                         }
 406
 407                         if ($debug_enabled) {
 408                                 _debug("update_rss_feed: loading filters & labels...");
 409                         }
 410
 411                         $filters = load_filters($link, $feed, $owner_uid);
 412                         $labels = get_all_labels($link, $owner_uid);
 413
 414                         if ($debug_enabled) {
 415                                 //print_r($filters);
 416                                 _debug("update_rss_feed: " . count($filters) . " filters loaded.");
 417                         }
 418
 419                         if ($use_simplepie) {
 420                                 $iterator = $rss->get_items();
 421                         } else {
 422                                 $iterator = $rss->items;
 423                                 if (!$iterator || !is_array($iterator)) $iterator = $rss->entries;
 424                                 if (!$iterator || !is_array($iterator)) $iterator = $rss;
 425                         }
 426
 427                         if (!is_array($iterator)) {
 428                                 /* db_query($link, "UPDATE ttrss_feeds
 429                                         SET last_error = 'Parse error: can\'t find any articles.'
 430                                         WHERE id = '$feed'"); */
 431
 432                                 // clear any errors and mark feed as updated if fetched okay
 433                                 // even if it's blank
 434
 435                                 if ($debug_enabled) {
 436                                         _debug("update_rss_feed: entry iterator is not an array, no articles?");
 437                                 }
 438
 439                                 db_query($link, "UPDATE ttrss_feeds
 440                                         SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
 441
 442                                 return; // no articles
 443                         }
 444
 445                         if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
 446
 447                                 if ($debug_enabled) _debug("update_rss_feed: checking for PUSH hub...");
 448
 449                                 $feed_hub_url = false;
 450                                 if ($use_simplepie) {
 451                                         $links = $rss->get_links('hub');
 452
 453                                         if ($links && is_array($links)) {
 454                                                 foreach ($links as $l) {
 455                                                         $feed_hub_url = $l;
 456                                                         break;
 457                                                 }
 458                                         }
 459
 460                                 } else {
 461                                         $atom = $rss->channel['atom'];
 462
 463                                         if ($atom) {
 464                                                 if ($atom['link@rel'] == 'hub') {
 465                                                         $feed_hub_url = $atom['link@href'];
 466                                                 }
 467
 468                                                 if (!$feed_hub_url && $atom['link#'] > 1) {
 469                                                         for ($i = 2; $i <= $atom['link#']; $i++) {
 470                                                                 if ($atom["link#$i@rel"] == 'hub') {
 471                                                                         $feed_hub_url = $atom["link#$i@href"];
 472                                                                         break;
 473                                                                 }
 474                                                         }
 475                                                 }
 476                                         } else {
 477                                                 $feed_hub_url = $rss->channel['link_hub'];
 478                                         }
 479                                 }
 480
 481                                 if ($debug_enabled) _debug("update_rss_feed: feed hub url: $feed_hub_url");
 482
 483                                 if ($feed_hub_url && function_exists('curl_init') &&
 484                                         !ini_get("open_basedir")) {
 485
 486                                         require_once 'lib/pubsubhubbub/subscriber.php';
 487
 488                                         $callback_url = get_self_url_prefix() .
 489                                                 "/public.php?op=pubsub&id=$feed";
 490
 491                                         $s = new Subscriber($feed_hub_url, $callback_url);
 492
 493                                         $rc = $s->subscribe($fetch_url);
 494
 495                                         if ($debug_enabled)
 496                                                 _debug("update_rss_feed: feed hub url found, subscribe request sent.");
 497
 498                                         db_query($link, "UPDATE ttrss_feeds SET pubsub_state = 1
 499                                                 WHERE id = '$feed'");
 500                                 }
 501                         }
 502
 503                         if ($debug_enabled) {
 504                                 _debug("update_rss_feed: processing articles...");
 505                         }
 506
 507                         foreach ($iterator as $item) {
 508                                 if ($_REQUEST['xdebug'] == 2) {
 509                                         print_r($item);
 510                                 }
 511
 512                                 if ($use_simplepie) {
 513                                         $entry_guid = $item->get_id();
 514                                         if (!$entry_guid) $entry_guid = $item->get_link();
 515                                         if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
 516
 517                                 } else {
 518
 519                                         $entry_guid = $item["id"];
 520
 521                                         if (!$entry_guid) $entry_guid = $item["guid"];
 522                                         if (!$entry_guid) $entry_guid = $item["about"];
 523                                         if (!$entry_guid) $entry_guid = $item["link"];
 524                                         if (!$entry_guid) $entry_guid = make_guid_from_title($item["title"]);
 525                                 }
 526
 527                                 if ($debug_enabled) {
 528                                         _debug("update_rss_feed: guid $entry_guid");
 529                                 }
 530
 531                                 if (!$entry_guid) continue;
 532
 533                                 $entry_timestamp = "";
 534
 535                                 if ($use_simplepie) {
 536                                         $entry_timestamp = strtotime($item->get_date());
 537                                 } else {
 538                                         $rss_2_date = $item['pubdate'];
 539                                         $rss_1_date = $item['dc']['date'];
 540                                         $atom_date = $item['issued'];
 541                                         if (!$atom_date) $atom_date = $item['updated'];
 542
 543                                         if ($atom_date != "") $entry_timestamp = parse_w3cdtf($atom_date);
 544                                         if ($rss_1_date != "") $entry_timestamp = parse_w3cdtf($rss_1_date);
 545                                         if ($rss_2_date != "") $entry_timestamp = strtotime($rss_2_date);
 546
 547                                 }
 548
 549                                 if ($entry_timestamp == "" || $entry_timestamp == -1 || !$entry_timestamp) {
 550                                         $entry_timestamp = time();
 551                                         $no_orig_date = 'true';
 552                                 } else {
 553                                         $no_orig_date = 'false';
 554                                 }
 555
 556                                 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
 557
 558                                 if ($debug_enabled) {
 559                                         _debug("update_rss_feed: date $entry_timestamp [$entry_timestamp_fmt]");
 560                                 }
 561
 562                                 if ($use_simplepie) {
 563                                         $entry_title = $item->get_title();
 564                                 } else {
 565                                         $entry_title = trim(strip_tags($item["title"]));
 566                                 }
 567
 568                                 if ($use_simplepie) {
 569                                         $entry_link = $item->get_link();
 570                                 } else {
 571                                         // strange Magpie workaround
 572                                         $entry_link = $item["link_"];
 573                                         if (!$entry_link) $entry_link = $item["link"];
 574                                 }
 575
 576                                 $entry_link = rewrite_relative_url($site_url, $entry_link);
 577
 578                                 if ($debug_enabled) {
 579                                         _debug("update_rss_feed: title $entry_title");
 580                                         _debug("update_rss_feed: link $entry_link");
 581                                 }
 582
 583                                 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
 584
 585                                 $entry_link = strip_tags($entry_link);
 586
 587                                 if ($use_simplepie) {
 588                                         $entry_content = $item->get_content();
 589                                         if (!$entry_content) $entry_content = $item->get_description();
 590                                 } else {
 591                                         $entry_content = $item["content:escaped"];
 592
 593                                         if (!$entry_content) $entry_content = $item["content:encoded"];
 594                                         if (!$entry_content && is_array($entry_content)) $entry_content = $item["content"]["encoded"];
 595                                         if (!$entry_content) $entry_content = $item["content"];
 596
 597                                         if (is_array($entry_content)) $entry_content = $entry_content[0];
 598
 599                                         // Magpie bugs are getting ridiculous
 600                                         if (trim($entry_content) == "Array") $entry_content = false;
 601
 602                                         if (!$entry_content) $entry_content = $item["atom_content"];
 603                                         if (!$entry_content) $entry_content = $item["summary"];
 604
 605                                         if (!$entry_content ||
 606                                                 strlen($entry_content) < strlen($item["description"])) {
 607                                                         $entry_content = $item["description"];
 608                                         };
 609
 610                                         // WTF
 611                                         if (is_array($entry_content)) {
 612                                                 $entry_content = $entry_content["encoded"];
 613                                                 if (!$entry_content) $entry_content = $entry_content["escaped"];
 614                                         }
 615                                 }
 616
 617                                 if ($cache_images && is_writable(CACHE_DIR . '/images'))
 618                                         $entry_content = cache_images($entry_content, $site_url, $debug_enabled);
 619
 620                                 if ($_REQUEST["xdebug"] == 2) {
 621                                         print "update_rss_feed: content: ";
 622                                         print $entry_content;
 623                                         print "\n";
 624                                 }
 625
 626                                 $entry_content_unescaped = $entry_content;
 627
 628                                 if ($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                                 $content_hash = "SHA1:" . sha1(strip_tags($entry_content));
 672
 673                                 $entry_title = db_escape_string($entry_title);
 674                                 $entry_link = db_escape_string($entry_link);
 675                                 $entry_comments = mb_substr(db_escape_string($entry_comments), 0, 250);
 676                                 $entry_author = mb_substr($entry_author, 0, 250);
 677
 678                                 if ($use_simplepie) {
 679                                         $num_comments = 0; #FIXME#
 680                                 } else {
 681                                         $num_comments = db_escape_string($item["slash"]["comments"]);
 682                                 }
 683
 684                                 if (!$num_comments) $num_comments = 0;
 685
 686                                 if ($debug_enabled) {
 687                                         _debug("update_rss_feed: looking for tags [1]...");
 688                                 }
 689
 690                                 // parse <category> entries into tags
 691
 692                                 $additional_tags = array();
 693
 694                                 if ($use_simplepie) {
 695
 696                                         $additional_tags_src = $item->get_categories();
 697
 698                                         if (is_array($additional_tags_src)) {
 699                                                 foreach ($additional_tags_src as $tobj) {
 700                                                         array_push($additional_tags, $tobj->get_term());
 701                                                 }
 702                                         }
 703
 704                                         if ($debug_enabled) {
 705                                                 _debug("update_rss_feed: category tags:");
 706                                                 print_r($additional_tags);
 707                                         }
 708
 709                                 } else {
 710
 711                                         $t_ctr = $item['category#'];
 712
 713                                         if ($t_ctr == 0) {
 714                                                 $additional_tags = array();
 715                                         } else if ($t_ctr > 0) {
 716                                                 $additional_tags = array($item['category']);
 717
 718                                                 if ($item['category@term']) {
 719                                                         array_push($additional_tags, $item['category@term']);
 720                                                 }
 721
 722                                                 for ($i = 0; $i <= $t_ctr; $i++ ) {
 723                                                         if ($item["category#$i"]) {
 724                                                                 array_push($additional_tags, $item["category#$i"]);
 725                                                         }
 726
 727                                                         if ($item["category#$i@term"]) {
 728                                                                 array_push($additional_tags, $item["category#$i@term"]);
 729                                                         }
 730                                                 }
 731                                         }
 732
 733                                         // parse <dc:subject> elements
 734
 735                                         $t_ctr = $item['dc']['subject#'];
 736
 737                                         if ($t_ctr > 0) {
 738                                                 array_push($additional_tags, $item['dc']['subject']);
 739
 740                                                 for ($i = 0; $i <= $t_ctr; $i++ ) {
 741                                                         if ($item['dc']["subject#$i"]) {
 742                                                                 array_push($additional_tags, $item['dc']["subject#$i"]);
 743                                                         }
 744                                                 }
 745                                         }
 746                                 }
 747
 748                                 if ($debug_enabled) {
 749                                         _debug("update_rss_feed: looking for tags [2]...");
 750                                 }
 751
 752                                 /* taaaags */
 753                                 // <a href="..." rel="tag">Xorg</a>, //
 754
 755                                 $entry_tags = null;
 756
 757                                 preg_match_all("/<a.*?rel=['\"]tag['\"].*?\>([^<]+)<\/a>/i",
 758                                         $entry_content_unescaped, $entry_tags);
 759
 760                                 $entry_tags = $entry_tags[1];
 761
 762                                 $entry_tags = array_merge($entry_tags, $additional_tags);
 763                                 $entry_tags = array_unique($entry_tags);
 764
 765                                 for ($i = 0; $i < count($entry_tags); $i++)
 766                                         $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
 767
 768                                 if ($debug_enabled) {
 769                                         //_debug("update_rss_feed: unfiltered tags found:");
 770                                         //print_r($entry_tags);
 771                                 }
 772
 773                                 if ($debug_enabled) {
 774                                         _debug("update_rss_feed: done collecting data [TITLE:$entry_title]");
 775                                 }
 776
 777                                 db_query($link, "BEGIN");
 778
 779                                 if (db_num_rows($result) == 0) {
 780
 781                                         if ($debug_enabled) {
 782                                                 _debug("update_rss_feed: base guid not found");
 783                                         }
 784
 785                                         // base post entry does not exist, create it
 786
 787                                         $result = db_query($link,
 788                                                 "INSERT INTO ttrss_entries
 789                                                         (title,
 790                                                         guid,
 791                                                         link,
 792                                                         updated,
 793                                                         content,
 794                                                         content_hash,
 795                                                         no_orig_date,
 796                                                         date_updated,
 797                                                         date_entered,
 798                                                         comments,
 799                                                         num_comments,
 800                                                         author)
 801                                                 VALUES
 802                                                         ('$entry_title',
 803                                                         '$entry_guid',
 804                                                         '$entry_link',
 805                                                         '$entry_timestamp_fmt',
 806                                                         '$entry_content',
 807                                                         '$content_hash',
 808                                                         $no_orig_date,
 809                                                         NOW(),
 810                                                         NOW(),
 811                                                         '$entry_comments',
 812                                                         '$num_comments',
 813                                                         '$entry_author')");
 814
 815                                         $article_labels = array();
 816
 817                                 } else {
 818                                         // we keep encountering the entry in feeds, so we need to
 819                                         // update date_updated column so that we don't get horrible
 820                                         // dupes when the entry gets purged and reinserted again e.g.
 821                                         // in the case of SLOW SLOW OMG SLOW updating feeds
 822
 823                                         $base_entry_id = db_fetch_result($result, 0, "id");
 824
 825                                         db_query($link, "UPDATE ttrss_entries SET date_updated = NOW()
 826                                                 WHERE id = '$base_entry_id'");
 827
 828                                         $article_labels = get_article_labels($link, $base_entry_id, $owner_uid);
 829                                 }
 830
 831                                 // now it should exist, if not - bad luck then
 832
 833                                 $result = db_query($link, "SELECT
 834                                                 id,content_hash,no_orig_date,title,
 835                                                 ".SUBSTRING_FOR_DATE."(date_updated,1,19) as date_updated,
 836                                                 ".SUBSTRING_FOR_DATE."(updated,1,19) as updated,
 837                                                 num_comments
 838                                         FROM
 839                                                 ttrss_entries
 840                                         WHERE guid = '$entry_guid'");
 841
 842                                 $entry_ref_id = 0;
 843                                 $entry_int_id = 0;
 844
 845                                 if (db_num_rows($result) == 1) {
 846
 847                                         if ($debug_enabled) {
 848                                                 _debug("update_rss_feed: base guid found, checking for user record");
 849                                         }
 850
 851                                         // this will be used below in update handler
 852                                         $orig_content_hash = db_fetch_result($result, 0, "content_hash");
 853                                         $orig_title = db_fetch_result($result, 0, "title");
 854                                         $orig_num_comments = db_fetch_result($result, 0, "num_comments");
 855                                         $orig_date_updated = strtotime(db_fetch_result($result,
 856                                                 0, "date_updated"));
 857
 858                                         $ref_id = db_fetch_result($result, 0, "id");
 859                                         $entry_ref_id = $ref_id;
 860
 861                                         // check for user post link to main table
 862
 863                                         // do we allow duplicate posts with same GUID in different feeds?
 864                                         if (get_pref($link, "ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
 865                                                 $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
 866                                         } else {
 867                                                 $dupcheck_qpart = "";
 868                                         }
 869
 870                                         /* Collect article tags here so we could filter by them: */
 871
 872                                         $article_filters = get_article_filters($filters, $entry_title,
 873                                                 strip_tags($entry_content), $entry_link, $entry_timestamp, $entry_author,
 874                                                 $entry_tags);
 875
 876                                         if ($debug_enabled) {
 877                                                 _debug("update_rss_feed: article filters: ");
 878                                                 if (count($article_filters) != 0) {
 879                                                         print_r($article_filters);
 880                                                 }
 881                                         }
 882
 883                                         if (find_article_filter($article_filters, "filter")) {
 884                                                 db_query($link, "COMMIT"); // close transaction in progress
 885                                                 continue;
 886                                         }
 887
 888                                         $score = calculate_article_score($article_filters);
 889
 890                                         if ($debug_enabled) {
 891                                                 _debug("update_rss_feed: initial score: $score");
 892                                         }
 893
 894                                         $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
 895                                                         ref_id = '$ref_id' AND owner_uid = '$owner_uid'
 896                                                         $dupcheck_qpart";
 897
 898 //                                      if ($_REQUEST["xdebug"]) print "$query\n";
 899
 900                                         $result = db_query($link, $query);
 901
 902                                         // okay it doesn't exist - create user entry
 903                                         if (db_num_rows($result) == 0) {
 904
 905                                                 if ($debug_enabled) {
 906                                                         _debug("update_rss_feed: user record not found, creating...");
 907                                                 }
 908
 909                                                 if ($score >= -500 && !find_article_filter($article_filters, 'catchup')) {
 910                                                         $unread = 'true';
 911                                                         $last_read_qpart = 'NULL';
 912                                                 } else {
 913                                                         $unread = 'false';
 914                                                         $last_read_qpart = 'NOW()';
 915                                                 }
 916
 917                                                 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
 918                                                         $marked = 'true';
 919                                                 } else {
 920                                                         $marked = 'false';
 921                                                 }
 922
 923                                                 if (find_article_filter($article_filters, 'publish')) {
 924                                                         $published = 'true';
 925                                                 } else {
 926                                                         $published = 'false';
 927                                                 }
 928
 929                                                 // N-grams
 930
 931                                                 if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
 932
 933                                                         $result = db_query($link, "SELECT COUNT(*) AS similar FROM
 934                                                                         ttrss_entries,ttrss_user_entries
 935                                                                 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
 936                                                                         AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
 937                                                                         AND owner_uid = $owner_uid");
 938
 939                                                         $ngram_similar = db_fetch_result($result, 0, "similar");
 940
 941                                                         if ($debug_enabled) {
 942                                                                 _debug("update_rss_feed: N-gram similar results: $ngram_similar");
 943                                                         }
 944
 945                                                         if ($ngram_similar > 0) {
 946                                                                 $unread = 'false';
 947                                                         }
 948                                                 }
 949
 950                                                 $result = db_query($link,
 951                                                         "INSERT INTO ttrss_user_entries
 952                                                                 (ref_id, owner_uid, feed_id, unread, last_read, marked,
 953                                                                         published, score, tag_cache, label_cache, uuid)
 954                                                         VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
 955                                                                 $last_read_qpart, $marked, $published, '$score', '', '', '')");
 956
 957                                                 if (PUBSUBHUBBUB_HUB && $published == 'true') {
 958                                                         $rss_link = get_self_url_prefix() .
 959                                                                 "/public.php?op=rss&id=-2&key=" .
 960                                                                 get_feed_access_key($link, -2, false, $owner_uid);
 961
 962                                                         $p = new Publisher(PUBSUBHUBBUB_HUB);
 963
 964                                                         $pubsub_result = $p->publish_update($rss_link);
 965                                                 }
 966
 967                                                 $result = db_query($link,
 968                                                         "SELECT int_id FROM ttrss_user_entries WHERE
 969                                                                 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
 970                                                                 feed_id = '$feed' LIMIT 1");
 971
 972                                                 if (db_num_rows($result) == 1) {
 973                                                         $entry_int_id = db_fetch_result($result, 0, "int_id");
 974                                                 }
 975                                         } else {
 976                                                 if ($debug_enabled) {
 977                                                         _debug("update_rss_feed: user record FOUND");
 978                                                 }
 979
 980                                                 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
 981                                                 $entry_int_id = db_fetch_result($result, 0, "int_id");
 982                                         }
 983
 984                                         if ($debug_enabled) {
 985                                                 _debug("update_rss_feed: RID: $entry_ref_id, IID: $entry_int_id");
 986                                         }
 987
 988                                         $post_needs_update = false;
 989                                         $update_insignificant = false;
 990
 991                                         if ($orig_num_comments != $num_comments) {
 992                                                 $post_needs_update = true;
 993                                                 $update_insignificant = true;
 994                                         }
 995
 996                                         if ($content_hash != $orig_content_hash) {
 997                                                 $post_needs_update = true;
 998                                                 $update_insignificant = false;
 999                                         }
1000
1001                                         if (db_escape_string($orig_title) != $entry_title) {
1002                                                 $post_needs_update = true;
1003                                                 $update_insignificant = false;
1004                                         }
1005
1006                                         // if post needs update, update it and mark all user entries
1007                                         // linking to this post as updated
1008                                         if ($post_needs_update) {
1009
1010                                                 if (defined('DAEMON_EXTENDED_DEBUG')) {
1011                                                         _debug("update_rss_feed: post $entry_guid needs update...");
1012                                                 }
1013
1014 //                                              print "<!-- post $orig_title needs update : $post_needs_update -->";
1015
1016                                                 db_query($link, "UPDATE ttrss_entries
1017                                                         SET title = '$entry_title', content = '$entry_content',
1018                                                                 content_hash = '$content_hash',
1019                                                                 updated = '$entry_timestamp_fmt',
1020                                                                 num_comments = '$num_comments'
1021                                                         WHERE id = '$ref_id'");
1022
1023                                                 if (!$update_insignificant) {
1024                                                         if ($mark_unread_on_update) {
1025                                                                 db_query($link, "UPDATE ttrss_user_entries
1026                                                                         SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
1027                                                         } else if ($update_on_checksum_change) {
1028                                                                 db_query($link, "UPDATE ttrss_user_entries
1029                                                                         SET last_read = null WHERE ref_id = '$ref_id'
1030                                                                                 AND unread = false");
1031                                                         }
1032                                                 }
1033                                         }
1034                                 }
1035
1036                                 db_query($link, "COMMIT");
1037
1038                                 if ($debug_enabled) {
1039                                         _debug("update_rss_feed: assigning labels...");
1040                                 }
1041
1042                                 assign_article_to_label_filters($link, $entry_ref_id, $article_filters,
1043                                         $owner_uid, $article_labels);
1044
1045                                 if ($debug_enabled) {
1046                                         _debug("update_rss_feed: looking for enclosures...");
1047                                 }
1048
1049                                 // enclosures
1050
1051                                 $enclosures = array();
1052
1053                                 if ($use_simplepie) {
1054                                         $encs = $item->get_enclosures();
1055
1056                                         if (is_array($encs)) {
1057                                                 foreach ($encs as $e) {
1058                                                         $e_item = array(
1059                                                                 $e->link, $e->type, $e->length);
1060
1061                                                         array_push($enclosures, $e_item);
1062                                                 }
1063                                         }
1064
1065                                 } else {
1066                                         // <enclosure>
1067
1068                                         $e_ctr = $item['enclosure#'];
1069
1070                                         if ($e_ctr > 0) {
1071                                                 $e_item = array($item['enclosure@url'],
1072                                                         $item['enclosure@type'],
1073                                                         $item['enclosure@length']);
1074
1075                                                 array_push($enclosures, $e_item);
1076
1077                                                 for ($i = 0; $i <= $e_ctr; $i++ ) {
1078
1079                                                         if ($item["enclosure#$i@url"]) {
1080                                                                 $e_item = array($item["enclosure#$i@url"],
1081                                                                         $item["enclosure#$i@type"],
1082                                                                         $item["enclosure#$i@length"]);
1083                                                                 array_push($enclosures, $e_item);
1084                                                         }
1085                                                 }
1086                                         }
1087
1088                                         // <media:content>
1089                                         // can there be many of those? yes -fox
1090
1091                                         $m_ctr = $item['media']['content#'];
1092
1093                                         if ($m_ctr > 0) {
1094                                                 $e_item = array($item['media']['content@url'],
1095                                                         $item['media']['content@medium'],
1096                                                         $item['media']['content@length']);
1097
1098                                                 array_push($enclosures, $e_item);
1099
1100                                                 for ($i = 0; $i <= $m_ctr; $i++ ) {
1101
1102                                                         if ($item["media"]["content#$i@url"]) {
1103                                                                 $e_item = array($item["media"]["content#$i@url"],
1104                                                                         $item["media"]["content#$i@medium"],
1105                                                                         $item["media"]["content#$i@length"]);
1106                                                                 array_push($enclosures, $e_item);
1107                                                         }
1108                                                 }
1109
1110                                         }
1111                                 }
1112
1113
1114                                 if ($debug_enabled) {
1115                                         _debug("update_rss_feed: article enclosures:");
1116                                         print_r($enclosures);
1117                                 }
1118
1119                                 db_query($link, "BEGIN");
1120
1121                                 foreach ($enclosures as $enc) {
1122                                         $enc_url = db_escape_string($enc[0]);
1123                                         $enc_type = db_escape_string($enc[1]);
1124                                         $enc_dur = db_escape_string($enc[2]);
1125
1126                                         $result = db_query($link, "SELECT id FROM ttrss_enclosures
1127                                                 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
1128
1129                                         if (db_num_rows($result) == 0) {
1130                                                 db_query($link, "INSERT INTO ttrss_enclosures
1131                                                         (content_url, content_type, title, duration, post_id) VALUES
1132                                                         ('$enc_url', '$enc_type', '', '$enc_dur', '$entry_ref_id')");
1133                                         }
1134                                 }
1135
1136                                 db_query($link, "COMMIT");
1137
1138                                 // check for manual tags (we have to do it here since they're loaded from filters)
1139
1140                                 foreach ($article_filters as $f) {
1141                                         if ($f["type"] == "tag") {
1142
1143                                                 $manual_tags = trim_array(explode(",", $f["param"]));
1144
1145                                                 foreach ($manual_tags as $tag) {
1146                                                         if (tag_is_valid($tag)) {
1147                                                                 array_push($entry_tags, $tag);
1148                                                         }
1149                                                 }
1150                                         }
1151                                 }
1152
1153                                 // Skip boring tags
1154
1155                                 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref($link,
1156                                         'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1157
1158                                 $filtered_tags = array();
1159                                 $tags_to_cache = array();
1160
1161                                 if ($entry_tags && is_array($entry_tags)) {
1162                                         foreach ($entry_tags as $tag) {
1163                                                 if (array_search($tag, $boring_tags) === false) {
1164                                                         array_push($filtered_tags, $tag);
1165                                                 }
1166                                         }
1167                                 }
1168
1169                                 $filtered_tags = array_unique($filtered_tags);
1170
1171                                 if ($debug_enabled) {
1172                                         _debug("update_rss_feed: filtered article tags:");
1173                                         print_r($filtered_tags);
1174                                 }
1175
1176                                 // Save article tags in the database
1177
1178                                 if (count($filtered_tags) > 0) {
1179
1180                                         db_query($link, "BEGIN");
1181
1182                                         foreach ($filtered_tags as $tag) {
1183
1184                                                 $tag = sanitize_tag($tag);
1185                                                 $tag = db_escape_string($tag);
1186
1187                                                 if (!tag_is_valid($tag)) continue;
1188
1189                                                 $result = db_query($link, "SELECT id FROM ttrss_tags
1190                                                         WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1191                                                         owner_uid = '$owner_uid' LIMIT 1");
1192
1193                                                         if ($result && db_num_rows($result) == 0) {
1194
1195                                                                 db_query($link, "INSERT INTO ttrss_tags
1196                                                                         (owner_uid,tag_name,post_int_id)
1197                                                                         VALUES ('$owner_uid','$tag', '$entry_int_id')");
1198                                                         }
1199
1200                                                 array_push($tags_to_cache, $tag);
1201                                         }
1202
1203                                         /* update the cache */
1204
1205                                         $tags_to_cache = array_unique($tags_to_cache);
1206
1207                                         $tags_str = db_escape_string(join(",", $tags_to_cache));
1208
1209                                         db_query($link, "UPDATE ttrss_user_entries
1210                                                 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1211                                                 AND owner_uid = $owner_uid");
1212
1213                                         db_query($link, "COMMIT");
1214                                 }
1215
1216                                 if (get_pref($link, "AUTO_ASSIGN_LABELS", $owner_uid, false)) {
1217                                         if ($debug_enabled) {
1218                                                 _debug("update_rss_feed: auto-assigning labels...");
1219                                         }
1220
1221                                         foreach ($labels as $label) {
1222                                                 $caption = $label["caption"];
1223
1224                                                 if (preg_match("/\b$caption\b/i", "$tags_str " . strip_tags($entry_content) . " $entry_title")) {
1225                                                         if (!labels_contains_caption($article_labels, $caption)) {
1226                                                                 label_add_article($link, $entry_ref_id, $caption, $owner_uid);
1227                                                         }
1228                                                 }
1229                                         }
1230                                 }
1231
1232                                 if ($debug_enabled) {
1233                                         _debug("update_rss_feed: article processed");
1234                                 }
1235                         }
1236
1237                         if (!$last_updated) {
1238                                 if ($debug_enabled) {
1239                                         _debug("update_rss_feed: new feed, catching it up...");
1240                                 }
1241                                 catchup_feed($link, $feed, false, $owner_uid);
1242                         }
1243
1244                         if ($debug_enabled) {
1245                                 _debug("purging feed...");
1246                         }
1247
1248                         purge_feed($link, $feed, 0, $debug_enabled);
1249
1250                         db_query($link, "UPDATE ttrss_feeds
1251                                 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
1252
1253 //                      db_query($link, "COMMIT");
1254
1255                 } else {
1256
1257                         if ($use_simplepie) {
1258                                 $error_msg = mb_substr($rss->error(), 0, 250);
1259                         } else {
1260                                 $error_msg = mb_substr(magpie_error(), 0, 250);
1261                         }
1262
1263                         if ($debug_enabled) {
1264                                 _debug("update_rss_feed: error fetching feed: $error_msg");
1265                         }
1266
1267                         $error_msg = db_escape_string($error_msg);
1268
1269                         db_query($link,
1270                                 "UPDATE ttrss_feeds SET last_error = '$error_msg',
1271                                         last_updated = NOW() WHERE id = '$feed'");
1272                 }
1273
1274                 if ($use_simplepie) {
1275                         unset($rss);
1276                 }
1277
1278                 if ($debug_enabled) {
1279                         _debug("update_rss_feed: done");
1280                 }
1281
1282         }
1283
1284         function cache_images($html, $site_url, $debug) {
1285                 $cache_dir = CACHE_DIR . "/images";
1286
1287                 libxml_use_internal_errors(true);
1288
1289                 $charset_hack = '<head>
1290                         <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1291                 </head>';
1292
1293                 $doc = new DOMDocument();
1294                 $doc->loadHTML($charset_hack . $html);
1295                 $xpath = new DOMXPath($doc);
1296
1297                 $entries = $xpath->query('(//img[@src])');
1298
1299                 foreach ($entries as $entry) {
1300                         if ($entry->hasAttribute('src')) {
1301                                 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1302
1303                                 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1304
1305                                 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1306
1307                                 if (!file_exists($local_filename)) {
1308                                         $file_content = fetch_file_contents($src);
1309
1310                                         if ($file_content && strlen($file_content) > 1024) {
1311                                                 file_put_contents($local_filename, $file_content);
1312                                         }
1313                                 }
1314
1315                                 if (file_exists($local_filename)) {
1316                                         $entry->setAttribute('src', SELF_URL_PATH . '/image.php?url=' .
1317                                                 base64_encode($src));
1318                                 }
1319                         }
1320                 }
1321
1322                 $node = $doc->getElementsByTagName('body')->item(0);
1323
1324                 return $doc->saveXML($node, LIBXML_NOEMPTYTAG);
1325         }
1326
1327         function expire_cached_files($debug) {
1328                 foreach (array("magpie", "simplepie", "images", "export") as $dir) {
1329                         $cache_dir = CACHE_DIR . "/$dir";
1330
1331                         if ($debug) _debug("Expiring $cache_dir");
1332
1333                         $num_deleted = 0;
1334
1335                         if (is_writable($cache_dir)) {
1336                                 $files = glob("$cache_dir/*");
1337
1338                                 if ($files)
1339                                         foreach ($files as $file) {
1340                                                 if (time() - filemtime($file) > 86400*7) {
1341                                                         unlink($file);
1342
1343                                                         ++$num_deleted;
1344                                                 }
1345                                         }
1346                                 }
1347
1348                         if ($debug) _debug("Removed $num_deleted files.");
1349                 }
1350         }
1351
1352         /**
1353         * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1354         * Returns the url query as associative array
1355         *
1356         * @param    string    query
1357         * @return    array    params
1358         */
1359         function convertUrlQuery($query) {
1360                 $queryParts = explode('&', $query);
1361
1362                 $params = array();
1363
1364                 foreach ($queryParts as $param) {
1365                         $item = explode('=', $param);
1366                         $params[$item[0]] = $item[1];
1367                 }
1368
1369                 return $params;
1370         }
1371
1372         function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
1373                 $matches = array();
1374
1375                 foreach ($filters as $filter) {
1376                         $match_any_rule = $filter["match_any_rule"];
1377                         $filter_match = false;
1378
1379                         foreach ($filter["rules"] as $rule) {
1380                                 $match = false;
1381                                 $reg_exp = $rule["reg_exp"];
1382
1383                                 if (!$reg_exp)
1384                                         continue;
1385
1386                                 switch ($rule["type"]) {
1387                                 case "title":
1388                                         $match = @preg_match("/$reg_exp/i", $title);
1389                                         break;
1390                                 case "content":
1391                                         $match = @preg_match("/$reg_exp/i", $content);
1392                                         break;
1393                                 case "both":
1394                                         $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $title));
1395                                         break;
1396                                 case "link":
1397                                         $match = @preg_match("/$reg_exp/i", $link);
1398                                         break;
1399                                 case "author":
1400                                         $match = @preg_match("/$reg_exp/i", $author);
1401                                         break;
1402                                 case "tag":
1403                                         $tag_string = join(",", $tags);
1404                                         $match = @preg_match("/$reg_exp/i", $tag_string);
1405                                         break;
1406                                 }
1407
1408                                 if ($match_any_rule) {
1409                                         if ($match) {
1410                                                 $filter_match = true;
1411                                                 break;
1412                                         }
1413                                 } else {
1414                                         $filter_match = $match;
1415                                         if (!$match) {
1416                                                 break;
1417                                         }
1418                                 }
1419                         }
1420
1421                         if ($filter_match) {
1422                                 foreach ($filter["actions"] AS $action) {
1423                                         array_push($matches, $action);
1424                                 }
1425                         }
1426                 }
1427
1428                 return $matches;
1429         }
1430
1431         function find_article_filter($filters, $filter_name) {
1432                 foreach ($filters as $f) {
1433                         if ($f["type"] == $filter_name) {
1434                                 return $f;
1435                         };
1436                 }
1437                 return false;
1438         }
1439
1440         function find_article_filters($filters, $filter_name) {
1441                 $results = array();
1442
1443                 foreach ($filters as $f) {
1444                         if ($f["type"] == $filter_name) {
1445                                 array_push($results, $f);
1446                         };
1447                 }
1448                 return $results;
1449         }
1450
1451         function calculate_article_score($filters) {
1452                 $score = 0;
1453
1454                 foreach ($filters as $f) {
1455                         if ($f["type"] == "score") {
1456                                 $score += $f["param"];
1457                         };
1458                 }
1459                 return $score;
1460         }
1461
1462         function labels_contains_caption($labels, $caption) {
1463                 foreach ($labels as $label) {
1464                         if ($label[1] == $caption) {
1465                                 return true;
1466                         }
1467                 }
1468
1469                 return false;
1470         }
1471
1472         function assign_article_to_label_filters($link, $id, $filters, $owner_uid, $article_labels) {
1473                 foreach ($filters as $f) {
1474                         if ($f["type"] == "label") {
1475                                 if (!labels_contains_caption($article_labels, $f["param"])) {
1476                                         label_add_article($link, $id, $f["param"], $owner_uid);
1477                                 }
1478                         }
1479                 }
1480         }
1481 ?>