]> git.wh0rd.org - tt-rss.git/blame - include/rssfuncs.php
git.wh0rd.org / tt-rss.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
disable csrf logging
[tt-rss.git] / include / rssfuncs.php
CommitLineData
2c08214a 1<?php
09e8bdfd
AD		 2 define('DAEMON_UPDATE_LOGIN_LIMIT', 30);
3 define('DAEMON_FEED_LIMIT', 100);
4 define('DAEMON_SLEEP_INTERVAL', 120);
5
79178062
AD		 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
2c08214a
AD		 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 // Test if the user has loggued in recently. If not, it does not update its feeds.
09e8bdfd 64 if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
2c08214a
AD		 65 if (DB_TYPE == "pgsql") {
66 $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
67 } else {
68 $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
69 }
70 } else {
71 $login_thresh_qpart = "";
72 }
73
74 // Test if the feed need a update (update interval exceded).
75 if (DB_TYPE == "pgsql") {
76 $update_limit_qpart = "AND ((
77 ttrss_feeds.update_interval = 0
78 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
79 ) OR (
80 ttrss_feeds.update_interval > 0
81 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
82 ) OR ttrss_feeds.last_updated IS NULL)";
83 } else {
84 $update_limit_qpart = "AND ((
85 ttrss_feeds.update_interval = 0
86 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
87 ) OR (
88 ttrss_feeds.update_interval > 0
89 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
90 ) OR ttrss_feeds.last_updated IS NULL)";
91 }
92
93 // Test if feed is currently being updated by another process.
94 if (DB_TYPE == "pgsql") {
95 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '5 minutes')";
96 } else {
97 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 5 MINUTE))";
98 }
99
100 // Test if there is a limit to number of updated feeds
101 $query_limit = "";
102 if($limit) $query_limit = sprintf("LIMIT %d", $limit);
103
104 $random_qpart = sql_random_function();
105
106 // We search for feed needing update.
107 $result = db_query($link, "SELECT ttrss_feeds.feed_url,ttrss_feeds.id, ttrss_feeds.owner_uid,
108 ".SUBSTRING_FOR_DATE."(ttrss_feeds.last_updated,1,19) AS last_updated,
109 ttrss_feeds.update_interval
110 FROM
111 ttrss_feeds, ttrss_users, ttrss_user_prefs
112 WHERE
113 ttrss_feeds.owner_uid = ttrss_users.id
114 AND ttrss_users.id = ttrss_user_prefs.owner_uid
115 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
116 $login_thresh_qpart $update_limit_qpart
117 $updstart_thresh_qpart
118 ORDER BY $random_qpart $query_limit");
119
120 $user_prefs_cache = array();
121
122 if($debug) _debug(sprintf("Scheduled %d feeds to update...\n", db_num_rows($result)));
123
124 // Here is a little cache magic in order to minimize risk of double feed updates.
125 $feeds_to_update = array();
126 while ($line = db_fetch_assoc($result)) {
127 $feeds_to_update[$line['id']] = $line;
128 }
129
130 // We update the feed last update started date before anything else.
131 // There is no lag due to feed contents downloads
132 // It prevent an other process to update the same feed.
133 $feed_ids = array_keys($feeds_to_update);
134 if($feed_ids) {
135 db_query($link, sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
136 WHERE id IN (%s)", implode(',', $feed_ids)));
137 }
138
139 // For each feed, we call the feed update function.
140 while ($line = array_pop($feeds_to_update)) {
141
142 if($debug) _debug("Feed: " . $line["feed_url"] . ", " . $line["last_updated"]);
143
144 update_rss_feed($link, $line["id"], true);
145
146 sleep(1); // prevent flood (FIXME make this an option?)
147 }
148
149 // Send feed digests by email if needed.
036cd3a4 150 send_headlines_digests($link, 100, $debug);
2c08214a
AD		 151
152 } // function update_daemon_common
153
154 function fetch_twitter_rss($link, $url, $owner_uid) {
155
156 require_once 'lib/tmhoauth/tmhOAuth.php';
157
158 $result = db_query($link, "SELECT twitter_oauth FROM ttrss_users
159 WHERE id = $owner_uid");
160
161 $access_token = json_decode(db_fetch_result($result, 0, 'twitter_oauth'), true);
162 $url_escaped = db_escape_string($url);
163
164 if ($access_token) {
165
166 $tmhOAuth = new tmhOAuth(array(
167 'consumer_key' => CONSUMER_KEY,
168 'consumer_secret' => CONSUMER_SECRET,
169 'user_token' => $access_token['oauth_token'],
170 'user_secret' => $access_token['oauth_token_secret'],
171 ));
172
173 $code = $tmhOAuth->request('GET', $url);
174
175 if ($code == 200) {
176
177 $content = $tmhOAuth->response['response'];
178
179 define('MAGPIE_CACHE_ON', false);
180
181 $rss = new MagpieRSS($content, MAGPIE_OUTPUT_ENCODING,
182 MAGPIE_INPUT_ENCODING, MAGPIE_DETECT_ENCODING );
183
184 return $rss;
185
186 } else {
187
188 db_query($link, "UPDATE ttrss_feeds
189 SET last_error = 'OAuth authorization failed ($code).'
190 WHERE feed_url = '$url_escaped' AND owner_uid = $owner_uid");
191 }
192
193 } else {
194
195 db_query($link, "UPDATE ttrss_feeds
196 SET last_error = 'OAuth information not found.'
197 WHERE feed_url = '$url_escaped' AND owner_uid = $owner_uid");
198
199 return false;
200 }
201 }
202
203 function update_rss_feed($link, $feed, $ignore_daemon = false, $no_cache = false) {
204
205 global $memcache;
206
207 /* Update all feeds with the same URL to utilize memcache */
208
209 if ($memcache) {
210 $result = db_query($link, "SELECT f1.id
211 FROM ttrss_feeds AS f1, ttrss_feeds AS f2
212 WHERE f2.feed_url = f1.feed_url AND f2.id = '$feed'");
213
214 while ($line = db_fetch_assoc($result)) {
215 update_rss_feed_real($link, $line["id"], $ignore_daemon, $no_cache);
216 }
217 } else {
218 update_rss_feed_real($link, $feed, $ignore_daemon, $no_cache);
219 }
220 }
221
222 function update_rss_feed_real($link, $feed, $ignore_daemon = false, $no_cache = false,
223 $override_url = false) {
224
225 require_once "lib/simplepie/simplepie.inc";
226 require_once "lib/magpierss/rss_fetch.inc";
227 require_once 'lib/magpierss/rss_utils.inc';
228
229 global $memcache;
230
231 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
232
233 if (!$_REQUEST["daemon"] && !$ignore_daemon) {
234 return false;
235 }
236
237 if ($debug_enabled) {
238 _debug("update_rss_feed: start");
239 }
240
241 if (!$ignore_daemon) {
242
243 if (DB_TYPE == "pgsql") {
244 $updstart_thresh_qpart = "(ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '120 seconds')";
245 } else {
246 $updstart_thresh_qpart = "(ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 120 SECOND))";
247 }
248
249 $result = db_query($link, "SELECT id,update_interval,auth_login,
250 auth_pass,cache_images,update_method
251 FROM ttrss_feeds WHERE id = '$feed' AND $updstart_thresh_qpart");
252
253 } else {
254
255 $result = db_query($link, "SELECT id,update_interval,auth_login,
256 feed_url,auth_pass,cache_images,update_method,last_updated,
257 mark_unread_on_update, owner_uid, update_on_checksum_change,
258 pubsub_state
259 FROM ttrss_feeds WHERE id = '$feed'");
260
261 }
262
263 if (db_num_rows($result) == 0) {
264 if ($debug_enabled) {
265 _debug("update_rss_feed: feed $feed NOT FOUND/SKIPPED");
266 }
267 return false;
268 }
269
270 $update_method = db_fetch_result($result, 0, "update_method");
271 $last_updated = db_fetch_result($result, 0, "last_updated");
272 $owner_uid = db_fetch_result($result, 0, "owner_uid");
273 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
274 0, "mark_unread_on_update"));
275 $update_on_checksum_change = sql_bool_to_bool(db_fetch_result($result,
276 0, "update_on_checksum_change"));
277 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
278
279 db_query($link, "UPDATE ttrss_feeds SET last_update_started = NOW()
280 WHERE id = '$feed'");
281
282 $auth_login = db_fetch_result($result, 0, "auth_login");
283 $auth_pass = db_fetch_result($result, 0, "auth_pass");
284
285 if ($update_method == 0)
286 $update_method = DEFAULT_UPDATE_METHOD + 1;
287
288 // 1 - Magpie
289 // 2 - SimplePie
290 // 3 - Twitter OAuth
291
292 if ($update_method == 2)
293 $use_simplepie = true;
294 else
295 $use_simplepie = false;
296
297 if ($debug_enabled) {
298 _debug("update method: $update_method (feed setting: $update_method) (use simplepie: $use_simplepie)\n");
299 }
300
301 if ($update_method == 1) {
302 $auth_login = urlencode($auth_login);
303 $auth_pass = urlencode($auth_pass);
304 }
305
306 $update_interval = db_fetch_result($result, 0, "update_interval");
307 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
308 $fetch_url = db_fetch_result($result, 0, "feed_url");
309
310 if ($update_interval < 0) { return false; }
311
312 $feed = db_escape_string($feed);
313
314 if ($auth_login && $auth_pass ){
315 $url_parts = array();
316 preg_match("/(^[^:]*):\/\/(.*)/", $fetch_url, $url_parts);
317
318 if ($url_parts[1] && $url_parts[2]) {
319 $fetch_url = $url_parts[1] . "://$auth_login:$auth_pass@" . $url_parts[2];
320 }
321
322 }
323
324 if ($override_url)
325 $fetch_url = $override_url;
326
327 if ($debug_enabled) {
328 _debug("update_rss_feed: fetching [$fetch_url]...");
329 }
330
331 $obj_id = md5("FDATA:$use_simplepie:$fetch_url");
332
333 if ($memcache && $obj = $memcache->get($obj_id)) {
334
335 if ($debug_enabled) {
336 _debug("update_rss_feed: data found in memcache.");
337 }
338
339 $rss = $obj;
340
341 } else {
342
343 if ($update_method == 3) {
344 $rss = fetch_twitter_rss($link, $fetch_url, $owner_uid);
345 } else if ($update_method == 1) {
346
347 define('MAGPIE_CACHE_AGE', get_feed_update_interval($link, $feed) * 60);
348 define('MAGPIE_CACHE_ON', !$no_cache);
349 define('MAGPIE_FETCH_TIME_OUT', 60);
350 define('MAGPIE_CACHE_DIR', CACHE_DIR . "/magpie");
351
352 $rss = @fetch_rss($fetch_url);
353 } else {
354 $simplepie_cache_dir = CACHE_DIR . "/simplepie";
355
356 if (!is_dir($simplepie_cache_dir)) {
357 mkdir($simplepie_cache_dir);
358 }
359
360 $rss = new SimplePie();
361 $rss->set_useragent(SELF_USER_AGENT);
362 # $rss->set_timeout(10);
363 $rss->set_feed_url($fetch_url);
364 $rss->set_output_encoding('UTF-8');
365 $rss->force_feed(true);
366
367 if (SIMPLEPIE_CACHE_IMAGES && $cache_images) {
368
369 if ($debug_enabled) {
370 _debug("enabling image cache");
371 }
372
373 $rss->set_image_handler("image.php", 'i');
374 }
375
376 if ($debug_enabled) {
377 _debug("feed update interval (sec): " .
378 get_feed_update_interval($link, $feed)*60);
379 }
380
381 $rss->enable_cache(!$no_cache);
382
383 if (!$no_cache) {
384 $rss->set_cache_location($simplepie_cache_dir);
385 $rss->set_cache_duration(get_feed_update_interval($link, $feed) * 60);
386 }
387
388 $rss->init();
389 }
390
391 if ($memcache && $rss) $memcache->add($obj_id, $rss, 0, 300);
392 }
393
394// print_r($rss);
395
396 if ($debug_enabled) {
397 _debug("update_rss_feed: fetch done, parsing...");
398 }
399
400 $feed = db_escape_string($feed);
401
402 if ($update_method == 2) {
403 $fetch_ok = !$rss->error();
404 } else {
405 $fetch_ok = !!$rss;
406 }
407
408 if ($fetch_ok) {
409
410 if ($debug_enabled) {
411 _debug("update_rss_feed: processing feed data...");
412 }
413
414// db_query($link, "BEGIN");
415
416 $result = db_query($link, "SELECT title,icon_url,site_url,owner_uid
417 FROM ttrss_feeds WHERE id = '$feed'");
418
419 $registered_title = db_fetch_result($result, 0, "title");
420 $orig_icon_url = db_fetch_result($result, 0, "icon_url");
421 $orig_site_url = db_fetch_result($result, 0, "site_url");
422
423 $owner_uid = db_fetch_result($result, 0, "owner_uid");
424
425 if ($use_simplepie) {
426 $site_url = $rss->get_link();
427 } else {
428 $site_url = $rss->channel["link"];
429 }
430
431 $site_url = rewrite_relative_url($fetch_url, $site_url);
432
433 if ($debug_enabled) {
434 _debug("update_rss_feed: checking favicon...");
435 }
436
437 check_feed_favicon($site_url, $feed, $link);
438
439 if (!$registered_title || $registered_title == "[Unknown]") {
440
441 if ($use_simplepie) {
442 $feed_title = db_escape_string($rss->get_title());
443 } else {
444 $feed_title = db_escape_string($rss->channel["title"]);
445 }
446
447 if ($debug_enabled) {
448 _debug("update_rss_feed: registering title: $feed_title");
449 }
450
451 db_query($link, "UPDATE ttrss_feeds SET
452 title = '$feed_title' WHERE id = '$feed'");
453 }
454
455 // weird, weird Magpie
456 if (!$use_simplepie) {
457 if (!$site_url) $site_url = db_escape_string($rss->channel["link_"]);
458 }
459
460 if ($site_url && $orig_site_url != db_escape_string($site_url)) {
461 db_query($link, "UPDATE ttrss_feeds SET
462 site_url = '$site_url' WHERE id = '$feed'");
463 }
464
465// print "I: " . $rss->channel["image"]["url"];
466
467 if (!$use_simplepie) {
468 $icon_url = db_escape_string($rss->image["url"]);
469 } else {
470 $icon_url = db_escape_string($rss->get_image_url());
471 }
472
473 $icon_url = substr($icon_url, 0, 250);
474
475 if ($icon_url && $orig_icon_url != $icon_url) {
476 db_query($link, "UPDATE ttrss_feeds SET icon_url = '$icon_url' WHERE id = '$feed'");
477 }
478
479 if ($debug_enabled) {
480 _debug("update_rss_feed: loading filters...");
481 }
482
483 $filters = load_filters($link, $feed, $owner_uid);
484
485// if ($debug_enabled) {
486// print_r($filters);
487// }
488
489 if ($use_simplepie) {
490 $iterator = $rss->get_items();
491 } else {
492 $iterator = $rss->items;
493 if (!$iterator || !is_array($iterator)) $iterator = $rss->entries;
494 if (!$iterator || !is_array($iterator)) $iterator = $rss;
495 }
496
497 if (!is_array($iterator)) {
498 /* db_query($link, "UPDATE ttrss_feeds
499 SET last_error = 'Parse error: can\'t find any articles.'
500 WHERE id = '$feed'"); */
501
502 // clear any errors and mark feed as updated if fetched okay
503 // even if it's blank
504
505 if ($debug_enabled) {
506 _debug("update_rss_feed: entry iterator is not an array, no articles?");
507 }
508
509 db_query($link, "UPDATE ttrss_feeds
510 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
511
512 return; // no articles
513 }
514
515 if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
516
517 if ($debug_enabled) _debug("update_rss_feed: checking for PUSH hub...");
518
519 $feed_hub_url = false;
520 if ($use_simplepie) {
521 $links = $rss->get_links('hub');
522
523 if ($links && is_array($links)) {
524 foreach ($links as $l) {
525 $feed_hub_url = $l;
526 break;
527 }
528 }
529
530 } else {
531 $atom = $rss->channel['atom'];
532
533 if ($atom) {
534 if ($atom['link@rel'] == 'hub') {
535 $feed_hub_url = $atom['link@href'];
536 }
537
538 if (!$feed_hub_url && $atom['link#'] > 1) {
539 for ($i = 2; $i <= $atom['link#']; $i++) {
540 if ($atom["link#$i@rel"] == 'hub') {
541 $feed_hub_url = $atom["link#$i@href"];
542 break;
543 }
544 }
545 }
546 } else {
547 $feed_hub_url = $rss->channel['link_hub'];
548 }
549 }
550
551 if ($debug_enabled) _debug("update_rss_feed: feed hub url: $feed_hub_url");
552
553 if ($feed_hub_url && function_exists('curl_init') &&
554 !ini_get("open_basedir")) {
555
556 require_once 'lib/pubsubhubbub/subscriber.php';
557
558 $callback_url = get_self_url_prefix() .
559 "/public.php?op=pubsub&id=$feed";
560
561 $s = new Subscriber($feed_hub_url, $callback_url);
562
563 $rc = $s->subscribe($fetch_url);
564
565 if ($debug_enabled)
566 _debug("update_rss_feed: feed hub url found, subscribe request sent.");
567
568 db_query($link, "UPDATE ttrss_feeds SET pubsub_state = 1
569 WHERE id = '$feed'");
570 }
571 }
572
573 if ($debug_enabled) {
574 _debug("update_rss_feed: processing articles...");
575 }
576
577 foreach ($iterator as $item) {
578
579 if ($_REQUEST['xdebug'] == 2) {
580 print_r($item);
581 }
582
583 if ($use_simplepie) {
584 $entry_guid = $item->get_id();
585 if (!$entry_guid) $entry_guid = $item->get_link();
586 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
587
588 } else {
589
590 $entry_guid = $item["id"];
591
592 if (!$entry_guid) $entry_guid = $item["guid"];
593 if (!$entry_guid) $entry_guid = $item["about"];
594 if (!$entry_guid) $entry_guid = $item["link"];
595 if (!$entry_guid) $entry_guid = make_guid_from_title($item["title"]);
596 }
597
598 if ($debug_enabled) {
599 _debug("update_rss_feed: guid $entry_guid");
600 }
601
602 if (!$entry_guid) continue;
603
604 $entry_timestamp = "";
605
606 if ($use_simplepie) {
607 $entry_timestamp = strtotime($item->get_date());
608 } else {
609 $rss_2_date = $item['pubdate'];
610 $rss_1_date = $item['dc']['date'];
611 $atom_date = $item['issued'];
612 if (!$atom_date) $atom_date = $item['updated'];
613
614 if ($atom_date != "") $entry_timestamp = parse_w3cdtf($atom_date);
615 if ($rss_1_date != "") $entry_timestamp = parse_w3cdtf($rss_1_date);
616 if ($rss_2_date != "") $entry_timestamp = strtotime($rss_2_date);
617
618 }
619
620 if ($entry_timestamp == "" || $entry_timestamp == -1 || !$entry_timestamp) {
621 $entry_timestamp = time();
622 $no_orig_date = 'true';
623 } else {
624 $no_orig_date = 'false';
625 }
626
627 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
628
629 if ($debug_enabled) {
630 _debug("update_rss_feed: date $entry_timestamp [$entry_timestamp_fmt]");
631 }
632
633 if ($use_simplepie) {
634 $entry_title = $item->get_title();
635 } else {
636 $entry_title = trim(strip_tags($item["title"]));
637 }
638
639 if ($use_simplepie) {
640 $entry_link = $item->get_link();
641 } else {
642 // strange Magpie workaround
643 $entry_link = $item["link_"];
644 if (!$entry_link) $entry_link = $item["link"];
645 }
646
647 $entry_link = rewrite_relative_url($site_url, $entry_link);
648
649 if ($debug_enabled) {
650 _debug("update_rss_feed: title $entry_title");
651 _debug("update_rss_feed: link $entry_link");
652 }
653
654 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
655
656 $entry_link = strip_tags($entry_link);
657
658 if ($use_simplepie) {
659 $entry_content = $item->get_content();
660 if (!$entry_content) $entry_content = $item->get_description();
661 } else {
662 $entry_content = $item["content:escaped"];
663
664 if (!$entry_content) $entry_content = $item["content:encoded"];
665 if (!$entry_content) $entry_content = $item["content"]["encoded"];
666 if (!$entry_content) $entry_content = $item["content"];
667
668 if (is_array($entry_content)) $entry_content = $entry_content[0];
669
670 // Magpie bugs are getting ridiculous
671 if (trim($entry_content) == "Array") $entry_content = false;
672
673 if (!$entry_content) $entry_content = $item["atom_content"];
674 if (!$entry_content) $entry_content = $item["summary"];
675
676 if (!$entry_content ||
677 strlen($entry_content) < strlen($item["description"])) {
678 $entry_content = $item["description"];
679 };
680
681 // WTF
682 if (is_array($entry_content)) {
683 $entry_content = $entry_content["encoded"];
684 if (!$entry_content) $entry_content = $entry_content["escaped"];
685 }
686 }
687
688 if ($_REQUEST["xdebug"] == 2) {
689 print "update_rss_feed: content: ";
690 print_r(htmlspecialchars($entry_content));
691 }
692
693 $entry_content_unescaped = $entry_content;
694
695 if ($use_simplepie) {
696 $entry_comments = strip_tags($item->data["comments"]);
697 if ($item->get_author()) {
698 $entry_author_item = $item->get_author();
699 $entry_author = $entry_author_item->get_name();
700 if (!$entry_author) $entry_author = $entry_author_item->get_email();
701
702 $entry_author = db_escape_string($entry_author);
703 }
704 } else {
705 $entry_comments = strip_tags($item["comments"]);
706
707 $entry_author = db_escape_string(strip_tags($item['dc']['creator']));
708
709 if ($item['author']) {
710
711 if (is_array($item['author'])) {
712
713 if (!$entry_author) {
714 $entry_author = db_escape_string(strip_tags($item['author']['name']));
715 }
716
717 if (!$entry_author) {
718 $entry_author = db_escape_string(strip_tags($item['author']['email']));
719 }
720 }
721
722 if (!$entry_author) {
723 $entry_author = db_escape_string(strip_tags($item['author']));
724 }
725 }
726 }
727
728 if (preg_match('/^[\t\n\r ]*$/', $entry_author)) $entry_author = '';
729
730 $entry_guid = db_escape_string(strip_tags($entry_guid));
731 $entry_guid = mb_substr($entry_guid, 0, 250);
732
733 $result = db_query($link, "SELECT id FROM ttrss_entries
734 WHERE guid = '$entry_guid'");
735
736 $entry_content = db_escape_string($entry_content, false);
737
738 $content_hash = "SHA1:" . sha1(strip_tags($entry_content));
739
740 $entry_title = db_escape_string($entry_title);
741 $entry_link = db_escape_string($entry_link);
742 $entry_comments = mb_substr(db_escape_string($entry_comments), 0, 250);
743 $entry_author = mb_substr($entry_author, 0, 250);
744
745 if ($use_simplepie) {
746 $num_comments = 0; #FIXME#
747 } else {
748 $num_comments = db_escape_string($item["slash"]["comments"]);
749 }
750
751 if (!$num_comments) $num_comments = 0;
752
753 if ($debug_enabled) {
754 _debug("update_rss_feed: looking for tags [1]...");
755 }
756
757 // parse <category> entries into tags
758
759 $additional_tags = array();
760
761 if ($use_simplepie) {
762
763 $additional_tags_src = $item->get_categories();
764
765 if (is_array($additional_tags_src)) {
766 foreach ($additional_tags_src as $tobj) {
767 array_push($additional_tags, $tobj->get_term());
768 }
769 }
770
771 if ($debug_enabled) {
772 _debug("update_rss_feed: category tags:");
773 print_r($additional_tags);
774 }
775
776 } else {
777
778 $t_ctr = $item['category#'];
779
780 if ($t_ctr == 0) {
781 $additional_tags = array();
782 } else if ($t_ctr > 0) {
783 $additional_tags = array($item['category']);
784
785 if ($item['category@term']) {
786 array_push($additional_tags, $item['category@term']);
787 }
788
789 for ($i = 0; $i <= $t_ctr; $i++ ) {
790 if ($item["category#$i"]) {
791 array_push($additional_tags, $item["category#$i"]);
792 }
793
794 if ($item["category#$i@term"]) {
795 array_push($additional_tags, $item["category#$i@term"]);
796 }
797 }
798 }
799
800 // parse <dc:subject> elements
801
802 $t_ctr = $item['dc']['subject#'];
803
804 if ($t_ctr > 0) {
805 array_push($additional_tags, $item['dc']['subject']);
806
807 for ($i = 0; $i <= $t_ctr; $i++ ) {
808 if ($item['dc']["subject#$i"]) {
809 array_push($additional_tags, $item['dc']["subject#$i"]);
810 }
811 }
812 }
813 }
814
815 if ($debug_enabled) {
816 _debug("update_rss_feed: looking for tags [2]...");
817 }
818
819 /* taaaags */
820 // <a href="..." rel="tag">Xorg</a>, //
821
822 $entry_tags = null;
823
824 preg_match_all("/<a.*?rel=['\"]tag['\"].*?\>([^<]+)<\/a>/i",
825 $entry_content_unescaped, $entry_tags);
826
827 $entry_tags = $entry_tags[1];
828
829 $entry_tags = array_merge($entry_tags, $additional_tags);
830 $entry_tags = array_unique($entry_tags);
831
832 for ($i = 0; $i < count($entry_tags); $i++)
833 $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
834
835 if ($debug_enabled) {
836 _debug("update_rss_feed: unfiltered tags found:");
837 print_r($entry_tags);
838 }
839
840 # sanitize content
841
842 $entry_content = sanitize_article_content($entry_content);
843 $entry_title = sanitize_article_content($entry_title);
844
845 if ($debug_enabled) {
846 _debug("update_rss_feed: done collecting data [TITLE:$entry_title]");
847 }
848
849 db_query($link, "BEGIN");
850
851 if (db_num_rows($result) == 0) {
852
853 if ($debug_enabled) {
854 _debug("update_rss_feed: base guid not found");
855 }
856
857 // base post entry does not exist, create it
858
859 $result = db_query($link,
860 "INSERT INTO ttrss_entries
861 (title,
862 guid,
863 link,
864 updated,
865 content,
866 content_hash,
867 no_orig_date,
868 date_updated,
869 date_entered,
870 comments,
871 num_comments,
872 author)
873 VALUES
874 ('$entry_title',
875 '$entry_guid',
876 '$entry_link',
877 '$entry_timestamp_fmt',
878 '$entry_content',
879 '$content_hash',
880 $no_orig_date,
881 NOW(),
882 NOW(),
883 '$entry_comments',
884 '$num_comments',
885 '$entry_author')");
886 } else {
887 // we keep encountering the entry in feeds, so we need to
888 // update date_updated column so that we don't get horrible
889 // dupes when the entry gets purged and reinserted again e.g.
890 // in the case of SLOW SLOW OMG SLOW updating feeds
891
892 $base_entry_id = db_fetch_result($result, 0, "id");
893
894 db_query($link, "UPDATE ttrss_entries SET date_updated = NOW()
895 WHERE id = '$base_entry_id'");
896 }
897
898 // now it should exist, if not - bad luck then
899
900 $result = db_query($link, "SELECT
901 id,content_hash,no_orig_date,title,
902 ".SUBSTRING_FOR_DATE."(date_updated,1,19) as date_updated,
903 ".SUBSTRING_FOR_DATE."(updated,1,19) as updated,
904 num_comments
905 FROM
906 ttrss_entries
907 WHERE guid = '$entry_guid'");
908
909 $entry_ref_id = 0;
910 $entry_int_id = 0;
911
912 if (db_num_rows($result) == 1) {
913
914 if ($debug_enabled) {
915 _debug("update_rss_feed: base guid found, checking for user record");
916 }
917
918 // this will be used below in update handler
919 $orig_content_hash = db_fetch_result($result, 0, "content_hash");
920 $orig_title = db_fetch_result($result, 0, "title");
921 $orig_num_comments = db_fetch_result($result, 0, "num_comments");
922 $orig_date_updated = strtotime(db_fetch_result($result,
923 0, "date_updated"));
924
925 $ref_id = db_fetch_result($result, 0, "id");
926 $entry_ref_id = $ref_id;
927
928 // check for user post link to main table
929
930 // do we allow duplicate posts with same GUID in different feeds?
931 if (get_pref($link, "ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
932 $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
933 } else {
934 $dupcheck_qpart = "";
935 }
936
937 /* Collect article tags here so we could filter by them: */
938
939 $article_filters = get_article_filters($filters, $entry_title,
940 $entry_content, $entry_link, $entry_timestamp, $entry_author,
941 $entry_tags);
942
943 if ($debug_enabled) {
944 _debug("update_rss_feed: article filters: ");
945 if (count($article_filters) != 0) {
946 print_r($article_filters);
947 }
948 }
949
950 if (find_article_filter($article_filters, "filter")) {
951 db_query($link, "COMMIT"); // close transaction in progress
952 continue;
953 }
954
955 $score = calculate_article_score($article_filters);
956
957 if ($debug_enabled) {
958 _debug("update_rss_feed: initial score: $score");
959 }
960
961 $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
962 ref_id = '$ref_id' AND owner_uid = '$owner_uid'
963 $dupcheck_qpart";
964
965// if ($_REQUEST["xdebug"]) print "$query\n";
966
967 $result = db_query($link, $query);
968
969 // okay it doesn't exist - create user entry
970 if (db_num_rows($result) == 0) {
971
972 if ($debug_enabled) {
973 _debug("update_rss_feed: user record not found, creating...");
974 }
975
976 if ($score >= -500 && !find_article_filter($article_filters, 'catchup')) {
977 $unread = 'true';
978 $last_read_qpart = 'NULL';
979 } else {
980 $unread = 'false';
981 $last_read_qpart = 'NOW()';
982 }
983
984 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
985 $marked = 'true';
986 } else {
987 $marked = 'false';
988 }
989
990 if (find_article_filter($article_filters, 'publish')) {
991 $published = 'true';
992 } else {
993 $published = 'false';
994 }
995
996 $result = db_query($link,
997 "INSERT INTO ttrss_user_entries
998 (ref_id, owner_uid, feed_id, unread, last_read, marked,
999 published, score, tag_cache, label_cache, uuid)
1000 VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
1001 $last_read_qpart, $marked, $published, '$score', '', '', '')");
1002
1003 if (PUBSUBHUBBUB_HUB && $published == 'true') {
1004 $rss_link = get_self_url_prefix() .
1005 "/public.php?op=rss&id=-2&key=" .
1006 get_feed_access_key($link, -2, false, $owner_uid);
1007
1008 $p = new Publisher(PUBSUBHUBBUB_HUB);
1009
1010 $pubsub_result = $p->publish_update($rss_link);
1011 }
1012
1013 $result = db_query($link,
1014 "SELECT int_id FROM ttrss_user_entries WHERE
1015 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
1016 feed_id = '$feed' LIMIT 1");
1017
1018 if (db_num_rows($result) == 1) {
1019 $entry_int_id = db_fetch_result($result, 0, "int_id");
1020 }
1021 } else {
1022 if ($debug_enabled) {
1023 _debug("update_rss_feed: user record FOUND");
1024 }
1025
1026 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
1027 $entry_int_id = db_fetch_result($result, 0, "int_id");
1028 }
1029
1030 if ($debug_enabled) {
1031 _debug("update_rss_feed: RID: $entry_ref_id, IID: $entry_int_id");
1032 }
1033
1034 $post_needs_update = false;
1035 $update_insignificant = false;
1036
1037 if ($orig_num_comments != $num_comments) {
1038 $post_needs_update = true;
1039 $update_insignificant = true;
1040 }
1041
1042 if ($content_hash != $orig_content_hash) {
1043 $post_needs_update = true;
1044 $update_insignificant = false;
1045 }
1046
1047 if (db_escape_string($orig_title) != $entry_title) {
1048 $post_needs_update = true;
1049 $update_insignificant = false;
1050 }
1051
1052 // if post needs update, update it and mark all user entries
1053 // linking to this post as updated
1054 if ($post_needs_update) {
1055
1056 if (defined('DAEMON_EXTENDED_DEBUG')) {
1057 _debug("update_rss_feed: post $entry_guid needs update...");
1058 }
1059
1060// print "<!-- post $orig_title needs update : $post_needs_update -->";
1061
1062 db_query($link, "UPDATE ttrss_entries
1063 SET title = '$entry_title', content = '$entry_content',
1064 content_hash = '$content_hash',
1065 updated = '$entry_timestamp_fmt',
1066 num_comments = '$num_comments'
1067 WHERE id = '$ref_id'");
1068
1069 if (!$update_insignificant) {
1070 if ($mark_unread_on_update) {
1071 db_query($link, "UPDATE ttrss_user_entries
1072 SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
1073 } else if ($update_on_checksum_change) {
1074 db_query($link, "UPDATE ttrss_user_entries
1075 SET last_read = null WHERE ref_id = '$ref_id'
1076 AND unread = false");
1077 }
1078 }
1079 }
1080 }
1081
1082 db_query($link, "COMMIT");
1083
1084 if ($debug_enabled) {
1085 _debug("update_rss_feed: assigning labels...");
1086 }
1087
1088 assign_article_to_labels($link, $entry_ref_id, $article_filters,
1089 $owner_uid);
1090
1091 if ($debug_enabled) {
1092 _debug("update_rss_feed: looking for enclosures...");
1093 }
1094
1095 // enclosures
1096
1097 $enclosures = array();
1098
1099 if ($use_simplepie) {
1100 $encs = $item->get_enclosures();
1101
1102 if (is_array($encs)) {
1103 foreach ($encs as $e) {
1104 $e_item = array(
1105 $e->link, $e->type, $e->length);
1106
1107 array_push($enclosures, $e_item);
1108 }
1109 }
1110
1111 } else {
1112 // <enclosure>
1113
1114 $e_ctr = $item['enclosure#'];
1115
1116 if ($e_ctr > 0) {
1117 $e_item = array($item['enclosure@url'],
1118 $item['enclosure@type'],
1119 $item['enclosure@length']);
1120
1121 array_push($enclosures, $e_item);
1122
1123 for ($i = 0; $i <= $e_ctr; $i++ ) {
1124
1125 if ($item["enclosure#$i@url"]) {
1126 $e_item = array($item["enclosure#$i@url"],
1127 $item["enclosure#$i@type"],
1128 $item["enclosure#$i@length"]);
1129 array_push($enclosures, $e_item);
1130 }
1131 }
1132 }
1133
1134 // <media:content>
1135 // can there be many of those? yes -fox
1136
1137 $m_ctr = $item['media']['content#'];
1138
1139 if ($m_ctr > 0) {
1140 $e_item = array($item['media']['content@url'],
1141 $item['media']['content@medium'],
1142 $item['media']['content@length']);
1143
1144 array_push($enclosures, $e_item);
1145
1146 for ($i = 0; $i <= $m_ctr; $i++ ) {
1147
1148 if ($item["media"]["content#$i@url"]) {
1149 $e_item = array($item["media"]["content#$i@url"],
1150 $item["media"]["content#$i@medium"],
1151 $item["media"]["content#$i@length"]);
1152 array_push($enclosures, $e_item);
1153 }
1154 }
1155
1156 }
1157 }
1158
1159
1160 if ($debug_enabled) {
1161 _debug("update_rss_feed: article enclosures:");
1162 print_r($enclosures);
1163 }
1164
1165 db_query($link, "BEGIN");
1166
1167 foreach ($enclosures as $enc) {
1168 $enc_url = db_escape_string($enc[0]);
1169 $enc_type = db_escape_string($enc[1]);
1170 $enc_dur = db_escape_string($enc[2]);
1171
1172 $result = db_query($link, "SELECT id FROM ttrss_enclosures
1173 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
1174
1175 if (db_num_rows($result) == 0) {
1176 db_query($link, "INSERT INTO ttrss_enclosures
1177 (content_url, content_type, title, duration, post_id) VALUES
1178 ('$enc_url', '$enc_type', '', '$enc_dur', '$entry_ref_id')");
1179 }
1180 }
1181
1182 db_query($link, "COMMIT");
1183
1184 // check for manual tags (we have to do it here since they're loaded from filters)
1185
1186 foreach ($article_filters as $f) {
1187 if ($f[0] == "tag") {
1188
1189 $manual_tags = trim_array(explode(",", $f[1]));
1190
1191 foreach ($manual_tags as $tag) {
1192 if (tag_is_valid($tag)) {
1193 array_push($entry_tags, $tag);
1194 }
1195 }
1196 }
1197 }
1198
1199 // Skip boring tags
1200
1201 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref($link,
1202 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1203
1204 $filtered_tags = array();
1205 $tags_to_cache = array();
1206
1207 if ($entry_tags && is_array($entry_tags)) {
1208 foreach ($entry_tags as $tag) {
1209 if (array_search($tag, $boring_tags) === false) {
1210 array_push($filtered_tags, $tag);
1211 }
1212 }
1213 }
1214
1215 $filtered_tags = array_unique($filtered_tags);
1216
1217 if ($debug_enabled) {
1218 _debug("update_rss_feed: filtered article tags:");
1219 print_r($filtered_tags);
1220 }
1221
1222 // Save article tags in the database
1223
1224 if (count($filtered_tags) > 0) {
1225
1226 db_query($link, "BEGIN");
1227
1228 foreach ($filtered_tags as $tag) {
1229
1230 $tag = sanitize_tag($tag);
1231 $tag = db_escape_string($tag);
1232
1233 if (!tag_is_valid($tag)) continue;
1234
1235 $result = db_query($link, "SELECT id FROM ttrss_tags
1236 WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1237 owner_uid = '$owner_uid' LIMIT 1");
1238
1239 if ($result && db_num_rows($result) == 0) {
1240
1241 db_query($link, "INSERT INTO ttrss_tags
1242 (owner_uid,tag_name,post_int_id)
1243 VALUES ('$owner_uid','$tag', '$entry_int_id')");
1244 }
1245
1246 array_push($tags_to_cache, $tag);
1247 }
1248
1249 /* update the cache */
1250
1251 $tags_to_cache = array_unique($tags_to_cache);
1252
1253 $tags_str = db_escape_string(join(",", $tags_to_cache));
1254
1255 db_query($link, "UPDATE ttrss_user_entries
1256 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1257 AND owner_uid = $owner_uid");
1258
1259 db_query($link, "COMMIT");
1260 }
1261
1262 if ($debug_enabled) {
1263 _debug("update_rss_feed: article processed");
1264 }
1265 }
1266
1267 if (!$last_updated) {
1268 if ($debug_enabled) {
1269 _debug("update_rss_feed: new feed, catching it up...");
1270 }
1271 catchup_feed($link, $feed, false, $owner_uid);
1272 }
1273
1274 if ($debug_enabled) {
1275 _debug("purging feed...");
1276 }
1277
1278 purge_feed($link, $feed, 0, $debug_enabled);
1279
1280 db_query($link, "UPDATE ttrss_feeds
1281 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
1282
1283// db_query($link, "COMMIT");
1284
1285 } else {
1286
1287 if ($use_simplepie) {
1288 $error_msg = mb_substr($rss->error(), 0, 250);
1289 } else {
1290 $error_msg = mb_substr(magpie_error(), 0, 250);
1291 }
1292
1293 if ($debug_enabled) {
1294 _debug("update_rss_feed: error fetching feed: $error_msg");
1295 }
1296
1297 $error_msg = db_escape_string($error_msg);
1298
1299 db_query($link,
1300 "UPDATE ttrss_feeds SET last_error = '$error_msg',
1301 last_updated = NOW() WHERE id = '$feed'");
1302 }
1303
1304 if ($use_simplepie) {
1305 unset($rss);
1306 }
1307
1308 if ($debug_enabled) {
1309 _debug("update_rss_feed: done");
1310 }
1311
1312 }
1313
1314
1315
1316
1317?>
web-based news feed (RSS/Atom) reader and aggregator; see https://tt-rss.org/