]> git.wh0rd.org - tt-rss.git/blob - classes/rssutils.php
git.wh0rd.org / tt-rss.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge branch 'pullreq-enclosure-content-type' of tkappe/tt-rss into master
[tt-rss.git] / classes / rssutils.php
1 <?php
2 class RSSUtils {
3 static function calculate_article_hash($article, $pluginhost) {
4 $tmp = "";
5
6 foreach ($article as $k => $v) {
7 if ($k != "feed" && isset($v)) {
8 $x = strip_tags(is_array($v) ? implode(",", $v) : $v);
9
10 //_debug("$k:" . sha1($x) . ":" . htmlspecialchars($x), true);
11
12 $tmp .= sha1("$k:" . sha1($x));
13 }
14 }
15
16 return sha1(implode(",", $pluginhost->get_plugin_names()) . $tmp);
17 }
18
19 // Strips utf8mb4 characters (i.e. emoji) for mysql
20 static function strip_utf8mb4($str) {
21 return preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $str);
22 }
23
24 static function update_feedbrowser_cache() {
25
26 $pdo = Db::pdo();
27
28 $sth = $pdo->query("SELECT feed_url, site_url, title, COUNT(id) AS subscribers
29 FROM ttrss_feeds WHERE feed_url NOT IN (SELECT feed_url FROM ttrss_feeds
30 WHERE private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%')
31 GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
32
33 $pdo->beginTransaction();
34
35 $pdo->query("DELETE FROM ttrss_feedbrowser_cache");
36
37 $count = 0;
38
39 while ($line = $sth->fetch()) {
40
41 $subscribers = $line["subscribers"];
42 $feed_url = $line["feed_url"];
43 $title = $line["title"];
44 $site_url = $line["site_url"];
45
46 $tmph = $pdo->prepare("SELECT subscribers FROM
47 ttrss_feedbrowser_cache WHERE feed_url = ?");
48 $tmph->execute([$feed_url]);
49
50 if (!$tmph->fetch()) {
51
52 $tmph = $pdo->prepare("INSERT INTO ttrss_feedbrowser_cache
53 (feed_url, site_url, title, subscribers)
54 VALUES
55 (?, ?, ?, ?)");
56
57 $tmph->execute([$feed_url, $site_url, $title, $subscribers]);
58
59 ++$count;
60
61 }
62
63 }
64
65 $pdo->commit();
66
67 return $count;
68
69 }
70
71 static function update_daemon_common($limit = DAEMON_FEED_LIMIT, $debug = true) {
72 $schema_version = get_schema_version();
73
74 if ($schema_version != SCHEMA_VERSION) {
75 die("Schema version is wrong, please upgrade the database.\n");
76 }
77
78 $pdo = Db::pdo();
79
80 if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
81 if (DB_TYPE == "pgsql") {
82 $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
83 } else {
84 $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
85 }
86 } else {
87 $login_thresh_qpart = "";
88 }
89
90 if (DB_TYPE == "pgsql") {
91 $update_limit_qpart = "AND ((
92 ttrss_feeds.update_interval = 0
93 AND ttrss_user_prefs.value != '-1'
94 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
95 ) OR (
96 ttrss_feeds.update_interval > 0
97 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
98 ) OR (ttrss_feeds.last_updated IS NULL
99 AND ttrss_user_prefs.value != '-1')
100 OR (last_updated = '1970-01-01 00:00:00'
101 AND ttrss_user_prefs.value != '-1'))";
102 } else {
103 $update_limit_qpart = "AND ((
104 ttrss_feeds.update_interval = 0
105 AND ttrss_user_prefs.value != '-1'
106 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
107 ) OR (
108 ttrss_feeds.update_interval > 0
109 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
110 ) OR (ttrss_feeds.last_updated IS NULL
111 AND ttrss_user_prefs.value != '-1')
112 OR (last_updated = '1970-01-01 00:00:00'
113 AND ttrss_user_prefs.value != '-1'))";
114 }
115
116 // Test if feed is currently being updated by another process.
117 if (DB_TYPE == "pgsql") {
118 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
119 } else {
120 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 10 MINUTE))";
121 }
122
123 $query_limit = $limit ? sprintf("LIMIT %d", $limit) : "";
124
125 // Update the least recently updated feeds first
126 $query_order = "ORDER BY last_updated";
127 if (DB_TYPE == "pgsql") $query_order .= " NULLS FIRST";
128
129 $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
130 FROM
131 ttrss_feeds, ttrss_users, ttrss_user_prefs
132 WHERE
133 ttrss_feeds.owner_uid = ttrss_users.id
134 AND ttrss_user_prefs.profile IS NULL
135 AND ttrss_users.id = ttrss_user_prefs.owner_uid
136 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
137 $login_thresh_qpart $update_limit_qpart
138 $updstart_thresh_qpart
139 $query_order $query_limit";
140
141 $res = $pdo->query($query);
142
143 $feeds_to_update = array();
144 while ($line = $res->fetch()) {
145 array_push($feeds_to_update, $line['feed_url']);
146 }
147
148 if ($debug) _debug(sprintf("Scheduled %d feeds to update...", count($feeds_to_update)));
149
150 // Update last_update_started before actually starting the batch
151 // in order to minimize collision risk for parallel daemon tasks
152 if (count($feeds_to_update) > 0) {
153 $feeds_qmarks = arr_qmarks($feeds_to_update);
154
155 $tmph = $pdo->prepare("UPDATE ttrss_feeds SET last_update_started = NOW()
156 WHERE feed_url IN ($feeds_qmarks)");
157 $tmph->execute($feeds_to_update);
158 }
159
160 $nf = 0;
161 $bstarted = microtime(true);
162
163 $batch_owners = array();
164
165 // since we have the data cached, we can deal with other feeds with the same url
166 $usth = $pdo->prepare("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
167 FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
168 ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
169 ttrss_users.id = ttrss_user_prefs.owner_uid AND
170 ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
171 ttrss_user_prefs.profile IS NULL AND
172 feed_url = ?
173 $update_limit_qpart
174 $login_thresh_qpart
175 ORDER BY ttrss_feeds.id $query_limit");
176
177 foreach ($feeds_to_update as $feed) {
178 if($debug) _debug("Base feed: $feed");
179
180 $usth->execute([$feed]);
181 //update_rss_feed($line["id"], true);
182
183 if ($tline = $usth->fetch()) {
184 if ($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
185
186 if (array_search($tline["owner_uid"], $batch_owners) === FALSE)
187 array_push($batch_owners, $tline["owner_uid"]);
188
189 $fstarted = microtime(true);
190 RSSUtils::update_rss_feed($tline["id"], true, false);
191 _debug_suppress(false);
192
193 _debug(sprintf(" %.4f (sec)", microtime(true) - $fstarted));
194
195 ++$nf;
196 }
197 }
198
199 if ($nf > 0) {
200 _debug(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
201 microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
202 }
203
204 foreach ($batch_owners as $owner_uid) {
205 _debug("Running housekeeping tasks for user $owner_uid...");
206
207 RSSUtils::housekeeping_user($owner_uid);
208 }
209
210 // Send feed digests by email if needed.
211 Digest::send_headlines_digests($debug);
212
213 return $nf;
214 }
215
216 // this is used when subscribing
217 static function set_basic_feed_info($feed) {
218
219 $pdo = Db::pdo();
220
221 $sth = $pdo->prepare("SELECT owner_uid,feed_url,auth_pass,auth_login
222 FROM ttrss_feeds WHERE id = ?");
223 $sth->execute([$feed]);
224
225 if ($row = $sth->fetch()) {
226
227 $owner_uid = $row["owner_uid"];
228 $auth_login = $row["auth_login"];
229 $auth_pass = $row["auth_pass"];
230 $fetch_url = $row["feed_url"];
231
232 $pluginhost = new PluginHost();
233 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
234
235 $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
236 $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
237 $pluginhost->load_data();
238
239 $basic_info = array();
240 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_BASIC_INFO) as $plugin) {
241 $basic_info = $plugin->hook_feed_basic_info($basic_info, $fetch_url, $owner_uid, $feed, $auth_login, $auth_pass);
242 }
243
244 if (!$basic_info) {
245 $feed_data = fetch_file_contents($fetch_url, false,
246 $auth_login, $auth_pass, false,
247 FEED_FETCH_TIMEOUT,
248 0);
249
250 global $fetch_curl_used;
251
252 if (!$fetch_curl_used) {
253 $tmp = @gzdecode($feed_data);
254
255 if ($tmp) $feed_data = $tmp;
256 }
257
258 $feed_data = trim($feed_data);
259
260 $rss = new FeedParser($feed_data);
261 $rss->init();
262
263 if (!$rss->error()) {
264 $basic_info = array(
265 'title' => mb_substr($rss->get_title(), 0, 199),
266 'site_url' => mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245)
267 );
268 }
269 }
270
271 if ($basic_info && is_array($basic_info)) {
272 $sth = $pdo->prepare("SELECT title, site_url FROM ttrss_feeds WHERE id = ?");
273 $sth->execute([$feed]);
274
275 if ($row = $sth->fetch()) {
276
277 $registered_title = $row["title"];
278 $orig_site_url = $row["site_url"];
279
280 if ($basic_info['title'] && (!$registered_title || $registered_title == "[Unknown]")) {
281
282 $sth = $pdo->prepare("UPDATE ttrss_feeds SET
283 title = ? WHERE id = ?");
284 $sth->execute([$basic_info['title'], $feed]);
285 }
286
287 if ($basic_info['site_url'] && $orig_site_url != $basic_info['site_url']) {
288 $sth = $pdo->prepare("UPDATE ttrss_feeds SET
289 site_url = ? WHERE id = ?");
290 $sth->execute([$basic_info['site_url'], $feed]);
291 }
292
293 }
294 }
295 }
296 }
297
298 /**
299 * @SuppressWarnings(PHPMD.UnusedFormalParameter)
300 */
301 static function update_rss_feed($feed, $no_cache = false) {
302
303 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || clean($_REQUEST['xdebug']);
304
305 _debug_suppress(!$debug_enabled);
306 _debug("start", $debug_enabled);
307
308 $pdo = Db::pdo();
309
310 $sth = $pdo->prepare("SELECT title FROM ttrss_feeds WHERE id = ?");
311 $sth->execute([$feed]);
312
313 if (!$row = $sth->fetch()) {
314 _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
315 user_error("Attempt to update unknown/invalid feed $feed", E_USER_WARNING);
316 return false;
317 }
318
319 $title = $row["title"];
320
321 // feed was batch-subscribed or something, we need to get basic info
322 // this is not optimal currently as it fetches stuff separately TODO: optimize
323 if ($title == "[Unknown]") {
324 _debug("setting basic feed info for $feed...");
325 RSSUtils::set_basic_feed_info($feed);
326 }
327
328 $sth = $pdo->prepare("SELECT id,update_interval,auth_login,
329 feed_url,auth_pass,cache_images,
330 mark_unread_on_update, owner_uid,
331 auth_pass_encrypted, feed_language,
332 last_modified,
333 ".SUBSTRING_FOR_DATE."(last_unconditional, 1, 19) AS last_unconditional
334 FROM ttrss_feeds WHERE id = ?");
335 $sth->execute([$feed]);
336
337 if ($row = $sth->fetch()) {
338
339 $owner_uid = $row["owner_uid"];
340 $mark_unread_on_update = $row["mark_unread_on_update"];
341
342 $sth = $pdo->prepare("UPDATE ttrss_feeds SET last_update_started = NOW()
343 WHERE id = ?");
344 $sth->execute([$feed]);
345
346 $auth_login = $row["auth_login"];
347 $auth_pass = $row["auth_pass"];
348 $stored_last_modified = $row["last_modified"];
349 $last_unconditional = $row["last_unconditional"];
350 $cache_images = $row["cache_images"];
351 $fetch_url = $row["feed_url"];
352
353 $feed_language = mb_strtolower($row["feed_language"]);
354 if (!$feed_language) $feed_language = 'english';
355
356 } else {
357 return false;
358 }
359
360 $date_feed_processed = date('Y-m-d H:i');
361
362 $cache_filename = CACHE_DIR . "/feeds/" . sha1($fetch_url) . ".xml";
363
364 $pluginhost = new PluginHost();
365 $pluginhost->set_debug($debug_enabled);
366 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
367
368 $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
369 $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
370 $pluginhost->load_data();
371
372 $rss_hash = false;
373
374 $force_refetch = isset($_REQUEST["force_refetch"]);
375 $feed_data = "";
376
377 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
378 $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, 0, $auth_login, $auth_pass);
379 }
380
381 // try cache
382 if (!$feed_data &&
383 file_exists($cache_filename) &&
384 is_readable($cache_filename) &&
385 !$auth_login && !$auth_pass &&
386 filemtime($cache_filename) > time() - 30) {
387
388 _debug("using local cache [$cache_filename].", $debug_enabled);
389
390 @$feed_data = file_get_contents($cache_filename);
391
392 if ($feed_data) {
393 $rss_hash = sha1($feed_data);
394 }
395
396 } else {
397 _debug("local cache will not be used for this feed", $debug_enabled);
398 }
399
400 global $fetch_last_modified;
401
402 // fetch feed from source
403 if (!$feed_data) {
404 _debug("last unconditional update request: $last_unconditional");
405
406 if (ini_get("open_basedir") && function_exists("curl_init")) {
407 _debug("not using CURL due to open_basedir restrictions");
408 }
409
410 if (time() - strtotime($last_unconditional) > MAX_CONDITIONAL_INTERVAL) {
411 _debug("maximum allowed interval for conditional requests exceeded, forcing refetch");
412
413 $force_refetch = true;
414 } else {
415 _debug("stored last modified for conditional request: $stored_last_modified", $debug_enabled);
416 }
417
418 _debug("fetching [$fetch_url] (force_refetch: $force_refetch)...", $debug_enabled);
419
420 $feed_data = fetch_file_contents([
421 "url" => $fetch_url,
422 "login" => $auth_login,
423 "pass" => $auth_pass,
424 "timeout" => $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
425 "last_modified" => $force_refetch ? "" : $stored_last_modified
426 ]);
427
428 global $fetch_curl_used;
429
430 if (!$fetch_curl_used) {
431 $tmp = @gzdecode($feed_data);
432
433 if ($tmp) $feed_data = $tmp;
434 }
435
436 $feed_data = trim($feed_data);
437
438 _debug("fetch done.", $debug_enabled);
439 _debug("source last modified: " . $fetch_last_modified, $debug_enabled);
440
441 if ($feed_data && $fetch_last_modified != $stored_last_modified) {
442 $sth = $pdo->prepare("UPDATE ttrss_feeds SET last_modified = ? WHERE id = ?");
443 $sth->execute([substr($fetch_last_modified, 0, 245), $feed]);
444 }
445
446 // cache vanilla feed data for re-use
447 if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/feeds")) {
448 $new_rss_hash = sha1($feed_data);
449
450 if ($new_rss_hash != $rss_hash) {
451 _debug("saving $cache_filename", $debug_enabled);
452 @file_put_contents($cache_filename, $feed_data);
453 }
454 }
455 }
456
457 if (!$feed_data) {
458 global $fetch_last_error;
459 global $fetch_last_error_code;
460
461 _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
462
463 // If-Modified-Since
464 if ($fetch_last_error_code != 304) {
465 $error_message = $fetch_last_error;
466 } else {
467 _debug("source claims data not modified, nothing to do.", $debug_enabled);
468 $error_message = "";
469 }
470
471 $sth = $pdo->prepare("UPDATE ttrss_feeds SET last_error = ?,
472 last_updated = NOW() WHERE id = ?");
473 $sth->execute([$error_message, $feed]);
474
475 return;
476 }
477
478 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
479 $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
480 }
481
482 $rss = new FeedParser($feed_data);
483 $rss->init();
484
485 if (!$rss->error()) {
486
487 // We use local pluginhost here because we need to load different per-user feed plugins
488 $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
489
490 _debug("language: $feed_language", $debug_enabled);
491 _debug("processing feed data...", $debug_enabled);
492
493 if (DB_TYPE == "pgsql") {
494 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
495 } else {
496 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
497 }
498
499 $sth = $pdo->prepare("SELECT owner_uid,favicon_avg_color,
500 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
501 favicon_needs_check
502 FROM ttrss_feeds WHERE id = ?");
503 $sth->execute([$feed]);
504
505 if ($row = $sth->fetch()) {
506 $favicon_needs_check = $row["favicon_needs_check"];
507 $favicon_avg_color = $row["favicon_avg_color"];
508 $owner_uid = $row["owner_uid"];
509 } else {
510 return false;
511 }
512
513 $site_url = mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245);
514
515 _debug("site_url: $site_url", $debug_enabled);
516 _debug("feed_title: " . $rss->get_title(), $debug_enabled);
517
518 if ($favicon_needs_check || $force_refetch) {
519
520 /* terrible hack: if we crash on floicon shit here, we won't check
521 * the icon avgcolor again (unless the icon got updated) */
522
523 $favicon_file = ICONS_DIR . "/$feed.ico";
524 $favicon_modified = @filemtime($favicon_file);
525
526 _debug("checking favicon...", $debug_enabled);
527
528 RSSUtils::check_feed_favicon($site_url, $feed);
529 $favicon_modified_new = @filemtime($favicon_file);
530
531 if ($favicon_modified_new > $favicon_modified)
532 $favicon_avg_color = '';
533
534 $favicon_colorstring = "";
535 if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
536 require_once "colors.php";
537
538 $sth = $pdo->prepare("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
539 id = ?");
540 $sth->execute([$feed]);
541
542 $favicon_color = calculate_avg_color($favicon_file);
543
544 $favicon_colorstring = ",favicon_avg_color = " . $pdo->quote($favicon_color);
545
546 } else if ($favicon_avg_color == 'fail') {
547 _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
548 }
549
550 $sth = $pdo->prepare("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
551 $favicon_colorstring WHERE id = ?");
552 $sth->execute([$feed]);
553 }
554
555 _debug("loading filters & labels...", $debug_enabled);
556
557 $filters = load_filters($feed, $owner_uid);
558
559 if ($debug_enabled) {
560 print_r($filters);
561 }
562
563 _debug("" . count($filters) . " filters loaded.", $debug_enabled);
564
565 $items = $rss->get_items();
566
567 if (!is_array($items)) {
568 _debug("no articles found.", $debug_enabled);
569
570 $sth = $pdo->prepare("UPDATE ttrss_feeds
571 SET last_updated = NOW(), last_unconditional = NOW(), last_error = '' WHERE id = ?");
572 $sth->execute([$feed]);
573
574 return true; // no articles
575 }
576
577 _debug("processing articles...", $debug_enabled);
578
579 $tstart = time();
580
581 foreach ($items as $item) {
582 $pdo->beginTransaction();
583
584 if (clean($_REQUEST['xdebug']) == 3) {
585 print_r($item);
586 }
587
588 if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
589 _debug("looks like there's too many articles to process at once, breaking out", $debug_enabled);
590 $pdo->commit();
591 break;
592 }
593
594 $entry_guid = strip_tags($item->get_id());
595 if (!$entry_guid) $entry_guid = strip_tags($item->get_link());
596 if (!$entry_guid) $entry_guid = RSSUtils::make_guid_from_title($item->get_title());
597
598 if (!$entry_guid) {
599 $pdo->commit();
600 continue;
601 }
602
603 $entry_guid = "$owner_uid,$entry_guid";
604
605 $entry_guid_hashed = 'SHA1:' . sha1($entry_guid);
606
607 _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
608
609 $entry_timestamp = strip_tags($item->get_date());
610
611 _debug("orig date: " . $item->get_date(), $debug_enabled);
612
613 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
614 $entry_timestamp = time();
615 }
616
617 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
618
619 _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
620
621 $entry_title = strip_tags($item->get_title());
622
623 $entry_link = rewrite_relative_url($site_url, $item->get_link());
624
625 $entry_language = mb_substr(trim($item->get_language()), 0, 2);
626
627 _debug("title $entry_title", $debug_enabled);
628 _debug("link $entry_link", $debug_enabled);
629 _debug("language $entry_language", $debug_enabled);
630
631 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
632
633 $entry_content = $item->get_content();
634 if (!$entry_content) $entry_content = $item->get_description();
635
636 if (clean($_REQUEST["xdebug"]) == 2) {
637 print "content: ";
638 print htmlspecialchars($entry_content);
639 print "\n";
640 }
641
642 $entry_comments = mb_substr(strip_tags($item->get_comments_url()), 0, 245);
643 $num_comments = (int) $item->get_comments_count();
644
645 $entry_author = strip_tags($item->get_author());
646 $entry_guid = mb_substr($entry_guid, 0, 245);
647
648 _debug("author $entry_author", $debug_enabled);
649 _debug("num_comments: $num_comments", $debug_enabled);
650 _debug("looking for tags...", $debug_enabled);
651
652 // parse <category> entries into tags
653
654 $additional_tags = array();
655
656 $additional_tags_src = $item->get_categories();
657
658 if (is_array($additional_tags_src)) {
659 foreach ($additional_tags_src as $tobj) {
660 array_push($additional_tags, $tobj);
661 }
662 }
663
664 $entry_tags = array_unique($additional_tags);
665
666 for ($i = 0; $i < count($entry_tags); $i++) {
667 $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
668
669 // we don't support numeric tags, let's prefix them
670 if (is_numeric($entry_tags[$i])) $entry_tags[$i] = 't:' . $entry_tags[$i];
671 }
672
673 _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
674
675 _debug("done collecting data.", $debug_enabled);
676
677 $sth = $pdo->prepare("SELECT id, content_hash, lang FROM ttrss_entries
678 WHERE guid = ? OR guid = ?");
679 $sth->execute([$entry_guid, $entry_guid_hashed]);
680
681 if ($row = $sth->fetch()) {
682 $base_entry_id = $row["id"];
683 $entry_stored_hash = $row["content_hash"];
684 $article_labels = Article::get_article_labels($base_entry_id, $owner_uid);
685
686 $existing_tags = Article::get_article_tags($base_entry_id, $owner_uid);
687 $entry_tags = array_unique(array_merge($entry_tags, $existing_tags));
688 } else {
689 $base_entry_id = false;
690 $entry_stored_hash = "";
691 $article_labels = array();
692 }
693
694 $article = array("owner_uid" => $owner_uid, // read only
695 "guid" => $entry_guid, // read only
696 "guid_hashed" => $entry_guid_hashed, // read only
697 "title" => $entry_title,
698 "content" => $entry_content,
699 "link" => $entry_link,
700 "labels" => $article_labels, // current limitation: can add labels to article, can't remove them
701 "tags" => $entry_tags,
702 "author" => $entry_author,
703 "force_catchup" => false, // ugly hack for the time being
704 "score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
705 "language" => $entry_language,
706 "num_comments" => $num_comments, // read only
707 "feed" => array("id" => $feed,
708 "fetch_url" => $fetch_url,
709 "site_url" => $site_url,
710 "cache_images" => $cache_images)
711 );
712
713 $entry_plugin_data = "";
714 $entry_current_hash = RSSUtils::calculate_article_hash($article, $pluginhost);
715
716 _debug("article hash: $entry_current_hash [stored=$entry_stored_hash]", $debug_enabled);
717
718 if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
719 _debug("stored article seems up to date [IID: $base_entry_id], updating timestamp only", $debug_enabled);
720
721 // we keep encountering the entry in feeds, so we need to
722 // update date_updated column so that we don't get horrible
723 // dupes when the entry gets purged and reinserted again e.g.
724 // in the case of SLOW SLOW OMG SLOW updating feeds
725
726 $sth = $pdo->prepare("UPDATE ttrss_entries SET date_updated = NOW()
727 WHERE id = ?");
728 $sth->execute([$base_entry_id]);
729
730 $pdo->commit();
731 continue;
732 }
733
734 _debug("hash differs, applying plugin filters:", $debug_enabled);
735
736 foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
737 _debug("... " . get_class($plugin), $debug_enabled);
738
739 $start = microtime(true);
740 $article = $plugin->hook_article_filter($article);
741
742 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
743
744 $entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
745 }
746
747 if (clean($_REQUEST["xdebug"]) == 2) {
748 print "processed content: ";
749 print htmlspecialchars($article["content"]);
750 print "\n";
751 }
752
753 _debug("plugin data: $entry_plugin_data", $debug_enabled);
754
755 // Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
756 if (DB_TYPE == "mysql" && MYSQL_CHARSET != "UTF8MB4") {
757 foreach ($article as $k => $v) {
758 // i guess we'll have to take the risk of 4byte unicode labels & tags here
759 if (is_string($article[$k])) {
760 $article[$k] = RSSUtils::strip_utf8mb4($v);
761 }
762 }
763 }
764
765 /* Collect article tags here so we could filter by them: */
766
767 $matched_rules = array();
768
769 $article_filters = RSSUtils::get_article_filters($filters, $article["title"],
770 $article["content"], $article["link"], $article["author"],
771 $article["tags"], $matched_rules);
772
773 if ($debug_enabled) {
774 _debug("matched filter rules: ", $debug_enabled);
775
776 if (count($matched_rules) != 0) {
777 print_r($matched_rules);
778 }
779
780 _debug("filter actions: ", $debug_enabled);
781
782 if (count($article_filters) != 0) {
783 print_r($article_filters);
784 }
785 }
786
787 $plugin_filter_names = RSSUtils::find_article_filters($article_filters, "plugin");
788 $plugin_filter_actions = $pluginhost->get_filter_actions();
789
790 if (count($plugin_filter_names) > 0) {
791 _debug("applying plugin filter actions...", $debug_enabled);
792
793 foreach ($plugin_filter_names as $pfn) {
794 list($pfclass,$pfaction) = explode(":", $pfn["param"]);
795
796 if (isset($plugin_filter_actions[$pfclass])) {
797 $plugin = $pluginhost->get_plugin($pfclass);
798
799 _debug("... $pfclass: $pfaction", $debug_enabled);
800
801 if ($plugin) {
802 $start = microtime(true);
803 $article = $plugin->hook_article_filter_action($article, $pfaction);
804
805 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
806 } else {
807 _debug("??? $pfclass: plugin object not found.");
808 }
809 } else {
810 _debug("??? $pfclass: filter plugin not registered.");
811 }
812 }
813 }
814
815 $entry_tags = $article["tags"];
816 $entry_title = strip_tags($article["title"]);
817 $entry_author = mb_substr(strip_tags($article["author"]), 0, 245);
818 $entry_link = strip_tags($article["link"]);
819 $entry_content = $article["content"]; // escaped below
820 $entry_force_catchup = $article["force_catchup"];
821 $article_labels = $article["labels"];
822 $entry_score_modifier = (int) $article["score_modifier"];
823 $entry_language = $article["language"];
824
825 if ($debug_enabled) {
826 _debug("article labels:", $debug_enabled);
827
828 if (count($article_labels) != 0) {
829 print_r($article_labels);
830 }
831 }
832
833 _debug("force catchup: $entry_force_catchup");
834
835 if ($cache_images && is_writable(CACHE_DIR . '/images'))
836 RSSUtils::cache_media($entry_content, $site_url, $debug_enabled);
837
838 $csth = $pdo->prepare("SELECT id FROM ttrss_entries
839 WHERE guid = ? OR guid = ?");
840 $csth->execute([$entry_guid, $entry_guid_hashed]);
841
842 if (!$row = $csth->fetch()) {
843
844 _debug("base guid [$entry_guid or $entry_guid_hashed] not found, creating...", $debug_enabled);
845
846 // base post entry does not exist, create it
847
848 $usth = $pdo->prepare(
849 "INSERT INTO ttrss_entries
850 (title,
851 guid,
852 link,
853 updated,
854 content,
855 content_hash,
856 no_orig_date,
857 date_updated,
858 date_entered,
859 comments,
860 num_comments,
861 plugin_data,
862 lang,
863 author)
864 VALUES
865 (?, ?, ?, ?, ?, ?,
866 false,
867 NOW(),
868 ?, ?, ?, ?, ?, ?)");
869
870 $usth->execute([$entry_title,
871 $entry_guid_hashed,
872 $entry_link,
873 $entry_timestamp_fmt,
874 "$entry_content",
875 $entry_current_hash,
876 $date_feed_processed,
877 $entry_comments,
878 (int)$num_comments,
879 $entry_plugin_data,
880 "$entry_language",
881 "$entry_author"]);
882
883 }
884
885 $csth->execute([$entry_guid, $entry_guid_hashed]);
886
887 $entry_ref_id = 0;
888 $entry_int_id = 0;
889
890 if ($row = $csth->fetch()) {
891
892 _debug("base guid found, checking for user record", $debug_enabled);
893
894 $ref_id = $row['id'];
895 $entry_ref_id = $ref_id;
896
897 if (RSSUtils::find_article_filter($article_filters, "filter")) {
898 $pdo->commit();
899 continue;
900 }
901
902 $score = RSSUtils::calculate_article_score($article_filters) + $entry_score_modifier;
903
904 _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
905
906 // check for user post link to main table
907
908 $sth = $pdo->prepare("SELECT ref_id, int_id FROM ttrss_user_entries WHERE
909 ref_id = ? AND owner_uid = ?");
910 $sth->execute([$ref_id, $owner_uid]);
911
912 // okay it doesn't exist - create user entry
913 if ($row = $sth->fetch()) {
914 $entry_ref_id = $row["ref_id"];
915 $entry_int_id = $row["int_id"];
916
917 _debug("user record FOUND: RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
918 } else {
919
920 _debug("user record not found, creating...", $debug_enabled);
921
922 if ($score >= -500 && !RSSUtils::find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
923 $unread = 1;
924 $last_read_qpart = null;
925 } else {
926 $unread = 0;
927 $last_read_qpart = date("Y-m-d H:i"); // we can't use NOW() here because it gets quoted
928 }
929
930 if (RSSUtils::find_article_filter($article_filters, 'mark') || $score > 1000) {
931 $marked = 1;
932 } else {
933 $marked = 0;
934 }
935
936 if (RSSUtils::find_article_filter($article_filters, 'publish')) {
937 $published = 1;
938 } else {
939 $published = 0;
940 }
941
942 $last_marked = ($marked == 1) ? 'NOW()' : 'NULL';
943 $last_published = ($published == 1) ? 'NOW()' : 'NULL';
944
945 $sth = $pdo->prepare(
946 "INSERT INTO ttrss_user_entries
947 (ref_id, owner_uid, feed_id, unread, last_read, marked,
948 published, score, tag_cache, label_cache, uuid,
949 last_marked, last_published)
950 VALUES (?, ?, ?, ?, ?, ?, ?, ?, '', '', '', ".$last_marked.", ".$last_published.")");
951
952 $sth->execute([$ref_id, $owner_uid, $feed, $unread, $last_read_qpart, $marked,
953 $published, $score]);
954
955 $sth = $pdo->prepare("SELECT int_id FROM ttrss_user_entries WHERE
956 ref_id = ? AND owner_uid = ? AND
957 feed_id = ? LIMIT 1");
958
959 $sth->execute([$ref_id, $owner_uid, $feed]);
960
961 if ($row = $sth->fetch())
962 $entry_int_id = $row['int_id'];
963 }
964
965 _debug("resulting RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
966
967 if (DB_TYPE == "pgsql")
968 $tsvector_qpart = "tsvector_combined = to_tsvector(:ts_lang, :ts_content),";
969 else
970 $tsvector_qpart = "";
971
972 $sth = $pdo->prepare("UPDATE ttrss_entries
973 SET title = :title,
974 $tsvector_qpart
975 content = :content,
976 content_hash = :content_hash,
977 updated = :updated,
978 date_updated = NOW(),
979 num_comments = :num_comments,
980 plugin_data = :plugin_data,
981 author = :author,
982 lang = :lang
983 WHERE id = :id");
984
985 $params = [":title" => $entry_title,
986 ":content" => "$entry_content",
987 ":content_hash" => $entry_current_hash,
988 ":updated" => $entry_timestamp_fmt,
989 ":num_comments" => (int)$num_comments,
990 ":plugin_data" => $entry_plugin_data,
991 ":author" => "$entry_author",
992 ":lang" => $entry_language,
993 ":id" => $ref_id];
994
995 if (DB_TYPE == "pgsql") {
996 $params[":ts_lang"] = $feed_language;
997 $params[":ts_content"] = mb_substr(strip_tags($entry_title . " " . $entry_content), 0, 900000);
998 }
999
1000 $sth->execute($params);
1001
1002 // update aux data
1003 $sth = $pdo->prepare("UPDATE ttrss_user_entries
1004 SET score = ? WHERE ref_id = ?");
1005 $sth->execute([$score, $ref_id]);
1006
1007 if ($mark_unread_on_update) {
1008 _debug("article updated, marking unread as requested.", $debug_enabled);
1009
1010 $sth = $pdo->prepare("UPDATE ttrss_user_entries
1011 SET last_read = null, unread = true WHERE ref_id = ?");
1012 $sth->execute([$ref_id]);
1013 }
1014 }
1015
1016 _debug("assigning labels [other]...", $debug_enabled);
1017
1018 foreach ($article_labels as $label) {
1019 Labels::add_article($entry_ref_id, $label[1], $owner_uid);
1020 }
1021
1022 _debug("assigning labels [filters]...", $debug_enabled);
1023
1024 RSSUtils::assign_article_to_label_filters($entry_ref_id, $article_filters,
1025 $owner_uid, $article_labels);
1026
1027 _debug("looking for enclosures...", $debug_enabled);
1028
1029 // enclosures
1030
1031 $enclosures = array();
1032
1033 $encs = $item->get_enclosures();
1034
1035 if (is_array($encs)) {
1036 foreach ($encs as $e) {
1037 $e_item = array(
1038 rewrite_relative_url($site_url, $e->link),
1039 $e->type, $e->length, $e->title, $e->width, $e->height);
1040
1041 // Yet another episode of "mysql utf8_general_ci is gimped"
1042 if (DB_TYPE == "mysql" && MYSQL_CHARSET != "UTF8MB4") {
1043 for ($i = 0; $i < count($e_item); $i++) {
1044 if (is_string($e_item[$i])) {
1045 $e_item[$i] = RSSUtils::strip_utf8mb4($e_item[$i]);
1046 }
1047 }
1048 }
1049
1050 array_push($enclosures, $e_item);
1051 }
1052 }
1053
1054 if ($cache_images && is_writable(CACHE_DIR . '/images'))
1055 RSSUtils::cache_enclosures($enclosures, $site_url, $debug_enabled);
1056
1057 if ($debug_enabled) {
1058 _debug("article enclosures:", $debug_enabled);
1059 print_r($enclosures);
1060 }
1061
1062 $esth = $pdo->prepare("SELECT id FROM ttrss_enclosures
1063 WHERE content_url = ? AND content_type = ? AND post_id = ?");
1064
1065 $usth = $pdo->prepare("INSERT INTO ttrss_enclosures
1066 (content_url, content_type, title, duration, post_id, width, height) VALUES
1067 (?, ?, ?, ?, ?, ?, ?)");
1068
1069 foreach ($enclosures as $enc) {
1070 $enc_url = $enc[0];
1071 $enc_type = $enc[1];
1072 $enc_dur = (int)$enc[2];
1073 $enc_title = $enc[3];
1074 $enc_width = intval($enc[4]);
1075 $enc_height = intval($enc[5]);
1076
1077 $esth->execute([$enc_url, $enc_type, $entry_ref_id]);
1078
1079 if (!$esth->fetch()) {
1080 $usth->execute([$enc_url, $enc_type, (string)$enc_title, $enc_dur, $entry_ref_id, $enc_width, $enc_height]);
1081 }
1082 }
1083
1084 // check for manual tags (we have to do it here since they're loaded from filters)
1085
1086 foreach ($article_filters as $f) {
1087 if ($f["type"] == "tag") {
1088
1089 $manual_tags = trim_array(explode(",", $f["param"]));
1090
1091 foreach ($manual_tags as $tag) {
1092 if (tag_is_valid($tag)) {
1093 array_push($entry_tags, $tag);
1094 }
1095 }
1096 }
1097 }
1098
1099 // Skip boring tags
1100
1101 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
1102 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1103
1104 $filtered_tags = array();
1105 $tags_to_cache = array();
1106
1107 if ($entry_tags && is_array($entry_tags)) {
1108 foreach ($entry_tags as $tag) {
1109 if (array_search($tag, $boring_tags) === false) {
1110 array_push($filtered_tags, $tag);
1111 }
1112 }
1113 }
1114
1115 $filtered_tags = array_unique($filtered_tags);
1116
1117 if ($debug_enabled) {
1118 _debug("filtered article tags:", $debug_enabled);
1119 print_r($filtered_tags);
1120 }
1121
1122 // Save article tags in the database
1123
1124 if (count($filtered_tags) > 0) {
1125
1126 $tsth = $pdo->prepare("SELECT id FROM ttrss_tags
1127 WHERE tag_name = ? AND post_int_id = ? AND
1128 owner_uid = ? LIMIT 1");
1129
1130 $usth = $pdo->prepare("INSERT INTO ttrss_tags
1131 (owner_uid,tag_name,post_int_id)
1132 VALUES (?, ?, ?)");
1133
1134 foreach ($filtered_tags as $tag) {
1135
1136 $tag = sanitize_tag($tag);
1137
1138 if (!tag_is_valid($tag)) continue;
1139
1140 $tsth->execute([$tag, $entry_int_id, $owner_uid]);
1141
1142 if (!$tsth->fetch()) {
1143 $usth->execute([$owner_uid, $tag, $entry_int_id]);
1144 }
1145
1146 array_push($tags_to_cache, $tag);
1147 }
1148
1149 /* update the cache */
1150
1151 $tags_to_cache = array_unique($tags_to_cache);
1152
1153 $tags_str = join(",", $tags_to_cache);
1154
1155 $tsth = $pdo->prepare("UPDATE ttrss_user_entries
1156 SET tag_cache = ? WHERE ref_id = ?
1157 AND owner_uid = ?");
1158 $tsth->execute([$tags_str, $entry_ref_id, $owner_uid]);
1159 }
1160
1161 _debug("article processed", $debug_enabled);
1162
1163 $pdo->commit();
1164 }
1165
1166 _debug("purging feed...", $debug_enabled);
1167
1168 purge_feed($feed, 0, $debug_enabled);
1169
1170 $sth = $pdo->prepare("UPDATE ttrss_feeds
1171 SET last_updated = NOW(), last_unconditional = NOW(), last_error = '' WHERE id = ?");
1172 $sth->execute([$feed]);
1173
1174 } else {
1175
1176 $error_msg = mb_substr($rss->error(), 0, 245);
1177
1178 _debug("fetch error: $error_msg", $debug_enabled);
1179
1180 if (count($rss->errors()) > 1) {
1181 foreach ($rss->errors() as $error) {
1182 _debug("+ $error");
1183 }
1184 }
1185
1186 $sth = $pdo->prepare("UPDATE ttrss_feeds SET last_error = ?,
1187 last_updated = NOW(), last_unconditional = NOW() WHERE id = ?");
1188 $sth->execute([$error_msg, $feed]);
1189
1190 unset($rss);
1191 return false;
1192 }
1193
1194 _debug("done", $debug_enabled);
1195
1196 return true;
1197 }
1198
1199 static function cache_enclosures($enclosures, $site_url, $debug) {
1200 foreach ($enclosures as $enc) {
1201
1202 if (preg_match("/(image|audio|video)/", $enc[1])) {
1203
1204 $src = rewrite_relative_url($site_url, $enc[0]);
1205
1206 $local_filename = CACHE_DIR . "/images/" . sha1($src);
1207
1208 if ($debug) _debug("cache_enclosures: downloading: $src to $local_filename");
1209
1210 if (!file_exists($local_filename)) {
1211 $file_content = fetch_file_contents($src);
1212
1213 if ($file_content && strlen($file_content) > MIN_CACHE_FILE_SIZE) {
1214 file_put_contents($local_filename, $file_content);
1215 }
1216 } else {
1217 touch($local_filename);
1218 }
1219 }
1220 }
1221 }
1222
1223 static function cache_media($html, $site_url, $debug) {
1224 libxml_use_internal_errors(true);
1225
1226 $charset_hack = '<head>
1227 <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1228 </head>';
1229
1230 $doc = new DOMDocument();
1231 $doc->loadHTML($charset_hack . $html);
1232 $xpath = new DOMXPath($doc);
1233
1234 $entries = $xpath->query('(//img[@src])|(//video/source[@src])|(//audio/source[@src])');
1235
1236 foreach ($entries as $entry) {
1237 if ($entry->hasAttribute('src') && strpos($entry->getAttribute('src'), "data:") !== 0) {
1238 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1239
1240 $local_filename = CACHE_DIR . "/images/" . sha1($src);
1241
1242 if ($debug) _debug("cache_media: checking $src");
1243
1244 if (!file_exists($local_filename)) {
1245 if ($debug) _debug("cache_media: downloading: $src to $local_filename");
1246
1247 $file_content = fetch_file_contents($src);
1248
1249 if ($file_content && strlen($file_content) > MIN_CACHE_FILE_SIZE) {
1250 file_put_contents($local_filename, $file_content);
1251 }
1252 } else {
1253 touch($local_filename);
1254 }
1255 }
1256 }
1257 }
1258
1259 static function expire_error_log($debug) {
1260 if ($debug) _debug("Removing old error log entries...");
1261
1262 $pdo = Db::pdo();
1263
1264 if (DB_TYPE == "pgsql") {
1265 $pdo->query("DELETE FROM ttrss_error_log
1266 WHERE created_at < NOW() - INTERVAL '7 days'");
1267 } else {
1268 $pdo->query("DELETE FROM ttrss_error_log
1269 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1270 }
1271 }
1272
1273 static function expire_lock_files($debug) {
1274 //if ($debug) _debug("Removing old lock files...");
1275
1276 $num_deleted = 0;
1277
1278 if (is_writable(LOCK_DIRECTORY)) {
1279 $files = glob(LOCK_DIRECTORY . "/*.lock");
1280
1281 if ($files) {
1282 foreach ($files as $file) {
1283 if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
1284 unlink($file);
1285 ++$num_deleted;
1286 }
1287 }
1288 }
1289 }
1290
1291 if ($debug) _debug("Removed $num_deleted old lock files.");
1292 }
1293
1294 static function expire_cached_files($debug) {
1295 foreach (array("simplepie", "feeds", "images", "export", "upload") as $dir) {
1296 $cache_dir = CACHE_DIR . "/$dir";
1297
1298 // if ($debug) _debug("Expiring $cache_dir");
1299
1300 $num_deleted = 0;
1301
1302 if (is_writable($cache_dir)) {
1303 $files = glob("$cache_dir/*");
1304
1305 if ($files) {
1306 foreach ($files as $file) {
1307 if (time() - filemtime($file) > 86400*CACHE_MAX_DAYS) {
1308 unlink($file);
1309
1310 ++$num_deleted;
1311 }
1312 }
1313 }
1314 }
1315
1316 if ($debug) _debug("$cache_dir: removed $num_deleted files.");
1317 }
1318 }
1319
1320 /**
1321 * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1322 * Returns the url query as associative array
1323 *
1324 * @param string query
1325 * @return array params
1326 */
1327 static function convertUrlQuery($query) {
1328 $queryParts = explode('&', $query);
1329
1330 $params = array();
1331
1332 foreach ($queryParts as $param) {
1333 $item = explode('=', $param);
1334 $params[$item[0]] = $item[1];
1335 }
1336
1337 return $params;
1338 }
1339
1340 static function get_article_filters($filters, $title, $content, $link, $author, $tags, &$matched_rules = false) {
1341 $matches = array();
1342
1343 foreach ($filters as $filter) {
1344 $match_any_rule = $filter["match_any_rule"];
1345 $inverse = $filter["inverse"];
1346 $filter_match = false;
1347
1348 foreach ($filter["rules"] as $rule) {
1349 $match = false;
1350 $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
1351 $rule_inverse = $rule["inverse"];
1352
1353 if (!$reg_exp)
1354 continue;
1355
1356 switch ($rule["type"]) {
1357 case "title":
1358 $match = @preg_match("/$reg_exp/iu", $title);
1359 break;
1360 case "content":
1361 // we don't need to deal with multiline regexps
1362 $content = preg_replace("/[\r\n\t]/", "", $content);
1363
1364 $match = @preg_match("/$reg_exp/iu", $content);
1365 break;
1366 case "both":
1367 // we don't need to deal with multiline regexps
1368 $content = preg_replace("/[\r\n\t]/", "", $content);
1369
1370 $match = (@preg_match("/$reg_exp/iu", $title) || @preg_match("/$reg_exp/iu", $content));
1371 break;
1372 case "link":
1373 $match = @preg_match("/$reg_exp/iu", $link);
1374 break;
1375 case "author":
1376 $match = @preg_match("/$reg_exp/iu", $author);
1377 break;
1378 case "tag":
1379 foreach ($tags as $tag) {
1380 if (@preg_match("/$reg_exp/iu", $tag)) {
1381 $match = true;
1382 break;
1383 }
1384 }
1385 break;
1386 }
1387
1388 if ($rule_inverse) $match = !$match;
1389
1390 if ($match_any_rule) {
1391 if ($match) {
1392 $filter_match = true;
1393 break;
1394 }
1395 } else {
1396 $filter_match = $match;
1397 if (!$match) {
1398 break;
1399 }
1400 }
1401 }
1402
1403 if ($inverse) $filter_match = !$filter_match;
1404
1405 if ($filter_match) {
1406 if (is_array($matched_rules)) array_push($matched_rules, $rule);
1407
1408 foreach ($filter["actions"] AS $action) {
1409 array_push($matches, $action);
1410
1411 // if Stop action encountered, perform no further processing
1412 if (isset($action["type"]) && $action["type"] == "stop") return $matches;
1413 }
1414 }
1415 }
1416
1417 return $matches;
1418 }
1419
1420 static function find_article_filter($filters, $filter_name) {
1421 foreach ($filters as $f) {
1422 if ($f["type"] == $filter_name) {
1423 return $f;
1424 };
1425 }
1426 return false;
1427 }
1428
1429 static function find_article_filters($filters, $filter_name) {
1430 $results = array();
1431
1432 foreach ($filters as $f) {
1433 if ($f["type"] == $filter_name) {
1434 array_push($results, $f);
1435 };
1436 }
1437 return $results;
1438 }
1439
1440 static function calculate_article_score($filters) {
1441 $score = 0;
1442
1443 foreach ($filters as $f) {
1444 if ($f["type"] == "score") {
1445 $score += $f["param"];
1446 };
1447 }
1448 return $score;
1449 }
1450
1451 static function labels_contains_caption($labels, $caption) {
1452 foreach ($labels as $label) {
1453 if ($label[1] == $caption) {
1454 return true;
1455 }
1456 }
1457
1458 return false;
1459 }
1460
1461 static function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
1462 foreach ($filters as $f) {
1463 if ($f["type"] == "label") {
1464 if (!RSSUtils::labels_contains_caption($article_labels, $f["param"])) {
1465 Labels::add_article($id, $f["param"], $owner_uid);
1466 }
1467 }
1468 }
1469 }
1470
1471 static function make_guid_from_title($title) {
1472 return preg_replace("/[ \"\',.:;]/", "-",
1473 mb_strtolower(strip_tags($title), 'utf-8'));
1474 }
1475
1476 static function cleanup_counters_cache($debug) {
1477 $pdo = Db::pdo();
1478
1479 $res = $pdo->query("DELETE FROM ttrss_counters_cache
1480 WHERE feed_id > 0 AND
1481 (SELECT COUNT(id) FROM ttrss_feeds WHERE
1482 id = feed_id AND
1483 ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
1484
1485 $frows = $res->rowCount();
1486
1487 $res = $pdo->query("DELETE FROM ttrss_cat_counters_cache
1488 WHERE feed_id > 0 AND
1489 (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
1490 id = feed_id AND
1491 ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
1492
1493 $crows = $res->rowCount();
1494
1495 if ($debug) _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
1496 }
1497
1498 static function housekeeping_user($owner_uid) {
1499 $tmph = new PluginHost();
1500
1501 load_user_plugins($owner_uid, $tmph);
1502
1503 $tmph->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
1504 }
1505
1506 static function housekeeping_common($debug) {
1507 RSSUtils::expire_cached_files($debug);
1508 RSSUtils::expire_lock_files($debug);
1509 RSSUtils::expire_error_log($debug);
1510
1511 $count = RSSUtils::update_feedbrowser_cache();
1512 _debug("Feedbrowser updated, $count feeds processed.");
1513
1514 Article::purge_orphans( true);
1515 RSSUtils::cleanup_counters_cache($debug);
1516
1517 //$rc = cleanup_tags( 14, 50000);
1518 //_debug("Cleaned $rc cached tags.");
1519
1520 PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
1521 }
1522
1523 static function check_feed_favicon($site_url, $feed) {
1524 # print "FAVICON [$site_url]: $favicon_url\n";
1525
1526 $icon_file = ICONS_DIR . "/$feed.ico";
1527
1528 if (!file_exists($icon_file)) {
1529 $favicon_url = get_favicon_url($site_url);
1530
1531 if ($favicon_url) {
1532 // Limiting to "image" type misses those served with text/plain
1533 $contents = fetch_file_contents($favicon_url); // , "image");
1534
1535 if ($contents) {
1536 // Crude image type matching.
1537 // Patterns gleaned from the file(1) source code.
1538 if (preg_match('/^\x00\x00\x01\x00/', $contents)) {
1539 // 0 string \000\000\001\000 MS Windows icon resource
1540 //error_log("check_feed_favicon: favicon_url=$favicon_url isa MS Windows icon resource");
1541 }
1542 elseif (preg_match('/^GIF8/', $contents)) {
1543 // 0 string GIF8 GIF image data
1544 //error_log("check_feed_favicon: favicon_url=$favicon_url isa GIF image");
1545 }
1546 elseif (preg_match('/^\x89PNG\x0d\x0a\x1a\x0a/', $contents)) {
1547 // 0 string \x89PNG\x0d\x0a\x1a\x0a PNG image data
1548 //error_log("check_feed_favicon: favicon_url=$favicon_url isa PNG image");
1549 }
1550 elseif (preg_match('/^\xff\xd8/', $contents)) {
1551 // 0 beshort 0xffd8 JPEG image data
1552 //error_log("check_feed_favicon: favicon_url=$favicon_url isa JPG image");
1553 }
1554 elseif (preg_match('/^BM/', $contents)) {
1555 // 0 string BM PC bitmap (OS2, Windows BMP files)
1556 //error_log("check_feed_favicon, favicon_url=$favicon_url isa BMP image");
1557 }
1558 else {
1559 //error_log("check_feed_favicon: favicon_url=$favicon_url isa UNKNOWN type");
1560 $contents = "";
1561 }
1562 }
1563
1564 if ($contents) {
1565 $fp = @fopen($icon_file, "w");
1566
1567 if ($fp) {
1568 fwrite($fp, $contents);
1569 fclose($fp);
1570 chmod($icon_file, 0644);
1571 }
1572 }
1573 }
1574 return $icon_file;
1575 }
1576 }
1577
1578
1579
1580 }
web-based news feed (RSS/Atom) reader and aggregator; see https://tt-rss.org/