]> git.wh0rd.org - tt-rss.git/blob - include/rssfuncs.php
do not try to load plugins which aren't directories
[tt-rss.git] / 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 $subscribers = db_escape_string($link, $line["subscribers"]);
22 $feed_url = db_escape_string($link, $line["feed_url"]);
23 $title = db_escape_string($link, $line["title"]);
24 $site_url = db_escape_string($link, $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_user_prefs.value != '-1'
81 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
82 ) OR (
83 ttrss_feeds.update_interval > 0
84 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
85 ) OR ttrss_feeds.last_updated IS NULL
86 OR last_updated = '1970-01-01 00:00:00')";
87 } else {
88 $update_limit_qpart = "AND ((
89 ttrss_feeds.update_interval = 0
90 AND ttrss_user_prefs.value != '-1'
91 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
92 ) OR (
93 ttrss_feeds.update_interval > 0
94 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
95 ) OR ttrss_feeds.last_updated IS NULL
96 OR last_updated = '1970-01-01 00:00:00')";
97 }
98
99 // Test if feed is currently being updated by another process.
100 if (DB_TYPE == "pgsql") {
101 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '5 minutes')";
102 } else {
103 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 5 MINUTE))";
104 }
105
106 // Test if there is a limit to number of updated feeds
107 $query_limit = "";
108 if($limit) $query_limit = sprintf("LIMIT %d", $limit);
109
110 $random_qpart = sql_random_function();
111
112 // We search for feed needing update.
113 $result = db_query($link, "SELECT DISTINCT ttrss_feeds.feed_url,$random_qpart
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...", 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($result)) {
131 array_push($feeds_to_update, db_escape_string($link, $line['feed_url']));
132 }
133
134 // We update the feed last update started date before anything else.
135 // There is no lag due to feed contents downloads
136 // It prevent an other process to update the same feed.
137
138 if(count($feeds_to_update) > 0) {
139 $feeds_quoted = array();
140
141 foreach ($feeds_to_update as $feed) {
142 array_push($feeds_quoted, "'" . db_escape_string($link, $feed) . "'");
143 }
144
145 db_query($link, sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
146 WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
147 }
148
149 expire_cached_files($debug);
150 expire_lock_files($debug);
151
152 $nf = 0;
153
154 // For each feed, we call the feed update function.
155 foreach ($feeds_to_update as $feed) {
156 if($debug) _debug("Base feed: $feed");
157
158 //update_rss_feed($link, $line["id"], true);
159
160 // since we have the data cached, we can deal with other feeds with the same url
161
162 $tmp_result = db_query($link, "SELECT DISTINCT ttrss_feeds.id,last_updated
163 FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
164 ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
165 ttrss_users.id = ttrss_user_prefs.owner_uid AND
166 ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
167 feed_url = '".db_escape_string($link, $feed)."' AND
168 (ttrss_feeds.update_interval > 0 OR
169 ttrss_user_prefs.value != '-1')
170 $login_thresh_qpart
171 ORDER BY ttrss_feeds.id $query_limit");
172
173 if (db_num_rows($tmp_result) > 0) {
174 while ($tline = db_fetch_assoc($tmp_result)) {
175 if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"]);
176 update_rss_feed($link, $tline["id"], true);
177 ++$nf;
178 }
179 }
180 }
181
182 require_once "digest.php";
183
184 // Send feed digests by email if needed.
185 send_headlines_digests($link, $debug);
186
187 return $nf;
188
189 } // function update_daemon_common
190
191 // ignore_daemon is not used
192 function update_rss_feed($link, $feed, $ignore_daemon = false, $no_cache = false,
193 $override_url = false) {
194
195 require_once "lib/simplepie/simplepie.inc";
196
197 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
198
199 if ($debug_enabled) {
200 _debug("update_rss_feed: start");
201 }
202
203 $result = db_query($link, "SELECT id,update_interval,auth_login,
204 feed_url,auth_pass,cache_images,last_updated,
205 mark_unread_on_update, owner_uid,
206 pubsub_state, auth_pass_encrypted
207 FROM ttrss_feeds WHERE id = '$feed'");
208
209 if (db_num_rows($result) == 0) {
210 if ($debug_enabled) {
211 _debug("update_rss_feed: feed $feed NOT FOUND/SKIPPED");
212 }
213 return false;
214 }
215
216 $last_updated = db_fetch_result($result, 0, "last_updated");
217 $owner_uid = db_fetch_result($result, 0, "owner_uid");
218 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
219 0, "mark_unread_on_update"));
220 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
221 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
222 0, "auth_pass_encrypted"));
223
224 db_query($link, "UPDATE ttrss_feeds SET last_update_started = NOW()
225 WHERE id = '$feed'");
226
227 $auth_login = db_fetch_result($result, 0, "auth_login");
228 $auth_pass = db_fetch_result($result, 0, "auth_pass");
229
230 if ($auth_pass_encrypted) {
231 require_once "crypt.php";
232 $auth_pass = decrypt_string($auth_pass);
233 }
234
235 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
236 $fetch_url = db_fetch_result($result, 0, "feed_url");
237
238 $feed = db_escape_string($link, $feed);
239
240 if ($override_url) $fetch_url = $override_url;
241
242 $date_feed_processed = date('Y-m-d H:i');
243
244 $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".feed";
245
246 // Ignore cache if new feed or manual update.
247 $cache_age = ($no_cache || is_null($last_updated) || $last_updated == '1970-01-01 00:00:00') ?
248 30 : get_feed_update_interval($link, $feed) * 60;
249
250 if ($debug_enabled) {
251 _debug("update_rss_feed: cache filename: $cache_filename exists: " . file_exists($cache_filename));
252 _debug("update_rss_feed: cache age: $cache_age; no cache: $no_cache");
253 }
254
255 $cached_feed_data_hash = false;
256
257 $rss = false;
258 $rss_hash = false;
259 $cache_timestamp = file_exists($cache_filename) ? filemtime($cache_filename) : 0;
260 $last_updated_timestamp = strtotime($last_updated);
261
262 $force_refetch = isset($_REQUEST["force_refetch"]);
263
264 if (file_exists($cache_filename) &&
265 is_readable($cache_filename) &&
266 !$auth_login && !$auth_pass &&
267 filemtime($cache_filename) > time() - $cache_age) {
268
269 if ($debug_enabled) {
270 _debug("update_rss_feed: using local cache.");
271 }
272
273 if ($cache_timestamp > $last_updated_timestamp) {
274 @$rss_data = file_get_contents($cache_filename);
275
276 if ($rss_data) {
277 $rss_hash = sha1($rss_data);
278 @$rss = unserialize($rss_data);
279 }
280 } else if (!$force_refetch) {
281 if ($debug_enabled) {
282 _debug("update_rss_feed: local cache valid and older than last_updated, nothing to do.");
283 }
284 return;
285 }
286 }
287
288 if (!$rss) {
289
290 if (!$feed_data) {
291 if ($debug_enabled) {
292 _debug("update_rss_feed: fetching [$fetch_url] (ts: $cache_timestamp/$last_updated_timestamp)");
293 }
294
295 $feed_data = fetch_file_contents($fetch_url, false,
296 $auth_login, $auth_pass, false,
297 $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
298 $force_refetch ? 0 : max($last_updated_timestamp, $cache_timestamp));
299
300 if ($debug_enabled) {
301 _debug("update_rss_feed: fetch done.");
302 }
303
304 }
305
306 if (!$feed_data) {
307 global $fetch_last_error;
308 global $fetch_last_error_code;
309
310 if ($debug_enabled) {
311 _debug("update_rss_feed: unable to fetch: $fetch_last_error [$fetch_last_error_code]");
312 }
313
314 $error_escaped = '';
315
316 // If-Modified-Since
317 if ($fetch_last_error_code != 304) {
318 $error_escaped = db_escape_string($link, $fetch_last_error);
319 } else {
320 if ($debug_enabled) {
321 _debug("update_rss_feed: source claims data not modified, nothing to do.");
322 }
323 }
324
325 db_query($link,
326 "UPDATE ttrss_feeds SET last_error = '$error_escaped',
327 last_updated = NOW() WHERE id = '$feed'");
328
329 return;
330 }
331 }
332
333 $pluginhost = new PluginHost($link);
334 $pluginhost->set_debug($debug_enabled);
335 $user_plugins = get_pref($link, "_ENABLED_PLUGINS", $owner_uid);
336
337 $pluginhost->load(PLUGINS, $pluginhost::KIND_ALL);
338 $pluginhost->load($user_plugins, $pluginhost::KIND_USER, $owner_uid);
339 $pluginhost->load_data();
340
341 foreach ($pluginhost->get_hooks($pluginhost::HOOK_FEED_FETCHED) as $plugin) {
342 $feed_data = $plugin->hook_feed_fetched($feed_data);
343 }
344
345 if (!$rss) {
346 $rss = new SimplePie();
347 $rss->set_sanitize_class("SanitizeDummy");
348 // simplepie ignores the above and creates default sanitizer anyway,
349 // so let's override it...
350 $rss->sanitize = new SanitizeDummy();
351 $rss->set_output_encoding('UTF-8');
352 $rss->set_raw_data($feed_data);
353 $rss->enable_cache(false);
354
355 @$rss->init();
356 }
357
358 // print_r($rss);
359
360 $feed = db_escape_string($link, $feed);
361
362 if (!$rss->error()) {
363
364 // cache data for later
365 if (!$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
366 $rss_data = serialize($rss);
367 $new_rss_hash = sha1($rss_data);
368
369 if ($new_rss_hash != $rss_hash) {
370 if ($debug_enabled) {
371 _debug("update_rss_feed: saving $cache_filename");
372 }
373 @file_put_contents($cache_filename, serialize($rss));
374 }
375 }
376
377 // We use local pluginhost here because we need to load different per-user feed plugins
378 $pluginhost->run_hooks($pluginhost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
379
380 if ($debug_enabled) {
381 _debug("update_rss_feed: processing feed data...");
382 }
383
384 // db_query($link, "BEGIN");
385
386 if (DB_TYPE == "pgsql") {
387 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
388 } else {
389 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
390 }
391
392 $result = db_query($link, "SELECT title,site_url,owner_uid,
393 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
394 favicon_needs_check
395 FROM ttrss_feeds WHERE id = '$feed'");
396
397 $registered_title = db_fetch_result($result, 0, "title");
398 $orig_site_url = db_fetch_result($result, 0, "site_url");
399 $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
400 "favicon_needs_check"));
401
402 $owner_uid = db_fetch_result($result, 0, "owner_uid");
403
404 $site_url = db_escape_string($link, mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
405
406 if ($favicon_needs_check || $force_refetch) {
407 if ($debug_enabled) {
408 _debug("update_rss_feed: checking favicon...");
409 }
410
411 check_feed_favicon($site_url, $feed, $link);
412 $favicon_file = ICONS_DIR . "/$feed.ico";
413
414 if (file_exists($favicon_file)) {
415 require_once "colors.php";
416
417 $favicon_color = db_escape_string($link,
418 calculate_avg_color($favicon_file));
419
420 $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
421 }
422
423 db_query($link, "UPDATE ttrss_feeds SET favicon_last_checked = NOW()
424 $favicon_colorstring
425 WHERE id = '$feed'");
426 }
427
428 if (!$registered_title || $registered_title == "[Unknown]") {
429
430 $feed_title = db_escape_string($link, $rss->get_title());
431
432 if ($debug_enabled) {
433 _debug("update_rss_feed: registering title: $feed_title");
434 }
435
436 db_query($link, "UPDATE ttrss_feeds SET
437 title = '$feed_title' WHERE id = '$feed'");
438 }
439
440 if ($site_url && $orig_site_url != $site_url) {
441 db_query($link, "UPDATE ttrss_feeds SET
442 site_url = '$site_url' WHERE id = '$feed'");
443 }
444
445 if ($debug_enabled) {
446 _debug("update_rss_feed: loading filters & labels...");
447 }
448
449 $filters = load_filters($link, $feed, $owner_uid);
450 $labels = get_all_labels($link, $owner_uid);
451
452 if ($debug_enabled) {
453 //print_r($filters);
454 _debug("update_rss_feed: " . count($filters) . " filters loaded.");
455 }
456
457 $items = $rss->get_items();
458
459 if (!is_array($items)) {
460 if ($debug_enabled) {
461 _debug("update_rss_feed: no articles found.");
462 }
463
464 db_query($link, "UPDATE ttrss_feeds
465 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
466
467 return; // no articles
468 }
469
470 if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
471
472 if ($debug_enabled) _debug("update_rss_feed: checking for PUSH hub...");
473
474 $feed_hub_url = false;
475
476 $links = $rss->get_links('hub');
477
478 if ($links && is_array($links)) {
479 foreach ($links as $l) {
480 $feed_hub_url = $l;
481 break;
482 }
483 }
484
485 if ($debug_enabled) _debug("update_rss_feed: feed hub url: $feed_hub_url");
486
487 if ($feed_hub_url && function_exists('curl_init') &&
488 !ini_get("open_basedir")) {
489
490 require_once 'lib/pubsubhubbub/subscriber.php';
491
492 $callback_url = get_self_url_prefix() .
493 "/public.php?op=pubsub&id=$feed";
494
495 $s = new Subscriber($feed_hub_url, $callback_url);
496
497 $rc = $s->subscribe($fetch_url);
498
499 if ($debug_enabled)
500 _debug("update_rss_feed: feed hub url found, subscribe request sent.");
501
502 db_query($link, "UPDATE ttrss_feeds SET pubsub_state = 1
503 WHERE id = '$feed'");
504 }
505 }
506
507 if ($debug_enabled) {
508 _debug("update_rss_feed: processing articles...");
509 }
510
511 foreach ($items as $item) {
512 if ($_REQUEST['xdebug'] == 3) {
513 print_r($item);
514 }
515
516 $entry_guid = $item->get_id();
517 if (!$entry_guid) $entry_guid = $item->get_link();
518 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
519
520 if (!$entry_guid) continue;
521
522 $entry_guid = "$owner_uid,$entry_guid";
523
524 $entry_guid_hashed = db_escape_string($link, 'SHA1:' . sha1($entry_guid));
525
526 if ($debug_enabled) {
527 _debug("update_rss_feed: guid $entry_guid / $entry_guid_hashed");
528 }
529
530 $entry_timestamp = "";
531
532 $entry_timestamp = strtotime($item->get_date());
533
534 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
535 $entry_timestamp = time();
536 $no_orig_date = 'true';
537 } else {
538 $no_orig_date = 'false';
539 }
540
541 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
542
543 if ($debug_enabled) {
544 _debug("update_rss_feed: date $entry_timestamp [$entry_timestamp_fmt]");
545 }
546
547 $entry_title = html_entity_decode($item->get_title());
548
549 $entry_link = rewrite_relative_url($site_url, $item->get_link());
550
551 if ($debug_enabled) {
552 _debug("update_rss_feed: title $entry_title");
553 _debug("update_rss_feed: link $entry_link");
554 }
555
556 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
557
558 $entry_content = $item->get_content();
559 if (!$entry_content) $entry_content = $item->get_description();
560
561 if ($_REQUEST["xdebug"] == 2) {
562 print "update_rss_feed: content: ";
563 print $entry_content;
564 print "\n";
565 }
566
567 $entry_comments = $item->data["comments"];
568
569 if ($item->get_author()) {
570 $entry_author_item = $item->get_author();
571 $entry_author = $entry_author_item->get_name();
572 if (!$entry_author) $entry_author = $entry_author_item->get_email();
573
574 $entry_author = db_escape_string($link, $entry_author);
575 }
576
577 $entry_guid = db_escape_string($link, mb_substr($entry_guid, 0, 245));
578
579 $entry_comments = db_escape_string($link, mb_substr($entry_comments, 0, 245));
580 $entry_author = db_escape_string($link, mb_substr($entry_author, 0, 245));
581
582 $num_comments = $item->get_item_tags('http://purl.org/rss/1.0/modules/slash/', 'comments');
583
584 if (is_array($num_comments) && is_array($num_comments[0])) {
585 $num_comments = (int) $num_comments[0]["data"];
586 } else {
587 $num_comments = 0;
588 }
589
590 if ($debug_enabled) {
591 _debug("update_rss_feed: num_comments: $num_comments");
592 _debug("update_rss_feed: looking for tags [1]...");
593 }
594
595 // parse <category> entries into tags
596
597 $additional_tags = array();
598
599 $additional_tags_src = $item->get_categories();
600
601 if (is_array($additional_tags_src)) {
602 foreach ($additional_tags_src as $tobj) {
603 array_push($additional_tags, $tobj->get_term());
604 }
605 }
606
607 if ($debug_enabled) {
608 _debug("update_rss_feed: category tags:");
609 print_r($additional_tags);
610 }
611
612 if ($debug_enabled) {
613 _debug("update_rss_feed: looking for tags [2]...");
614 }
615
616 $entry_tags = array_unique($additional_tags);
617
618 for ($i = 0; $i < count($entry_tags); $i++)
619 $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
620
621 if ($debug_enabled) {
622 //_debug("update_rss_feed: unfiltered tags found:");
623 //print_r($entry_tags);
624 }
625
626 if ($debug_enabled) {
627 _debug("update_rss_feed: done collecting data.");
628 }
629
630 // TODO: less memory-hungry implementation
631
632 if ($debug_enabled) {
633 _debug("update_rss_feed: applying plugin filters..");
634 }
635
636 // FIXME not sure if owner_uid is a good idea here, we may have a base entry without user entry (?)
637 $result = db_query($link, "SELECT plugin_data,title,content,link,tag_cache,author FROM ttrss_entries, ttrss_user_entries
638 WHERE ref_id = id AND (guid = '".db_escape_string($link, $entry_guid)."' OR guid = '$entry_guid_hashed') AND owner_uid = $owner_uid");
639
640 if (db_num_rows($result) != 0) {
641 $entry_plugin_data = db_fetch_result($result, 0, "plugin_data");
642 $stored_article = array("title" => db_fetch_result($result, 0, "title"),
643 "content" => db_fetch_result($result, 0, "content"),
644 "link" => db_fetch_result($result, 0, "link"),
645 "tags" => explode(",", db_fetch_result($result, 0, "tag_cache")),
646 "author" => db_fetch_result($result, 0, "author"));
647 } else {
648 $entry_plugin_data = "";
649 $stored_article = array();
650 }
651
652 $article = array("owner_uid" => $owner_uid, // read only
653 "guid" => $entry_guid, // read only
654 "title" => $entry_title,
655 "content" => $entry_content,
656 "link" => $entry_link,
657 "tags" => $entry_tags,
658 "plugin_data" => $entry_plugin_data,
659 "author" => $entry_author,
660 "stored" => $stored_article);
661
662 foreach ($pluginhost->get_hooks($pluginhost::HOOK_ARTICLE_FILTER) as $plugin) {
663 $article = $plugin->hook_article_filter($article);
664 }
665
666 $entry_tags = $article["tags"];
667 $entry_guid = db_escape_string($link, $entry_guid);
668 $entry_title = db_escape_string($link, $article["title"]);
669 $entry_author = db_escape_string($link, $article["author"]);
670 $entry_link = db_escape_string($link, $article["link"]);
671 $entry_plugin_data = db_escape_string($link, $article["plugin_data"]);
672 $entry_content = $article["content"]; // escaped below
673
674
675 if ($debug_enabled) {
676 _debug("update_rss_feed: plugin data: $entry_plugin_data");
677 }
678
679 if ($cache_images && is_writable(CACHE_DIR . '/images'))
680 cache_images($entry_content, $site_url, $debug_enabled);
681
682 $entry_content = db_escape_string($link, $entry_content, false);
683
684 $content_hash = "SHA1:" . sha1($entry_content);
685
686 db_query($link, "BEGIN");
687
688 $result = db_query($link, "SELECT id FROM ttrss_entries
689 WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
690
691 if (db_num_rows($result) == 0) {
692
693 if ($debug_enabled) {
694 _debug("update_rss_feed: base guid [$entry_guid] not found");
695 }
696
697 // base post entry does not exist, create it
698
699 $result = db_query($link,
700 "INSERT INTO ttrss_entries
701 (title,
702 guid,
703 link,
704 updated,
705 content,
706 content_hash,
707 cached_content,
708 no_orig_date,
709 date_updated,
710 date_entered,
711 comments,
712 num_comments,
713 plugin_data,
714 author)
715 VALUES
716 ('$entry_title',
717 '$entry_guid_hashed',
718 '$entry_link',
719 '$entry_timestamp_fmt',
720 '$entry_content',
721 '$content_hash',
722 '',
723 $no_orig_date,
724 NOW(),
725 '$date_feed_processed',
726 '$entry_comments',
727 '$num_comments',
728 '$entry_plugin_data',
729 '$entry_author')");
730
731 $article_labels = array();
732
733 } else {
734 // we keep encountering the entry in feeds, so we need to
735 // update date_updated column so that we don't get horrible
736 // dupes when the entry gets purged and reinserted again e.g.
737 // in the case of SLOW SLOW OMG SLOW updating feeds
738
739 $base_entry_id = db_fetch_result($result, 0, "id");
740
741 db_query($link, "UPDATE ttrss_entries SET date_updated = NOW()
742 WHERE id = '$base_entry_id'");
743
744 $article_labels = get_article_labels($link, $base_entry_id, $owner_uid);
745 }
746
747 // now it should exist, if not - bad luck then
748
749 $result = db_query($link, "SELECT
750 id,content_hash,no_orig_date,title,plugin_data,guid,
751 ".SUBSTRING_FOR_DATE."(date_updated,1,19) as date_updated,
752 ".SUBSTRING_FOR_DATE."(updated,1,19) as updated,
753 num_comments
754 FROM
755 ttrss_entries
756 WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
757
758 $entry_ref_id = 0;
759 $entry_int_id = 0;
760
761 if (db_num_rows($result) == 1) {
762
763 if ($debug_enabled) {
764 _debug("update_rss_feed: base guid found, checking for user record");
765 }
766
767 // this will be used below in update handler
768 $orig_content_hash = db_fetch_result($result, 0, "content_hash");
769 $orig_title = db_fetch_result($result, 0, "title");
770 $orig_num_comments = db_fetch_result($result, 0, "num_comments");
771 $orig_date_updated = strtotime(db_fetch_result($result,
772 0, "date_updated"));
773 $orig_plugin_data = db_fetch_result($result, 0, "plugin_data");
774
775 $ref_id = db_fetch_result($result, 0, "id");
776 $entry_ref_id = $ref_id;
777
778 /* $stored_guid = db_fetch_result($result, 0, "guid");
779 if ($stored_guid != $entry_guid_hashed) {
780 if ($debug_enabled) _debug("upgrading compat guid to hashed one");
781
782 db_query($link, "UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
783 id = '$ref_id'");
784 } */
785
786 // check for user post link to main table
787
788 // do we allow duplicate posts with same GUID in different feeds?
789 if (get_pref($link, "ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
790 $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
791 } else {
792 $dupcheck_qpart = "";
793 }
794
795 /* Collect article tags here so we could filter by them: */
796
797 $article_filters = get_article_filters($filters, $entry_title,
798 $entry_content, $entry_link, $entry_timestamp, $entry_author,
799 $entry_tags);
800
801 if ($debug_enabled) {
802 _debug("update_rss_feed: article filters: ");
803 if (count($article_filters) != 0) {
804 print_r($article_filters);
805 }
806 }
807
808 if (find_article_filter($article_filters, "filter")) {
809 db_query($link, "COMMIT"); // close transaction in progress
810 continue;
811 }
812
813 $score = calculate_article_score($article_filters);
814
815 if ($debug_enabled) {
816 _debug("update_rss_feed: initial score: $score");
817 }
818
819 $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
820 ref_id = '$ref_id' AND owner_uid = '$owner_uid'
821 $dupcheck_qpart";
822
823 // if ($_REQUEST["xdebug"]) print "$query\n";
824
825 $result = db_query($link, $query);
826
827 // okay it doesn't exist - create user entry
828 if (db_num_rows($result) == 0) {
829
830 if ($debug_enabled) {
831 _debug("update_rss_feed: user record not found, creating...");
832 }
833
834 if ($score >= -500 && !find_article_filter($article_filters, 'catchup')) {
835 $unread = 'true';
836 $last_read_qpart = 'NULL';
837 } else {
838 $unread = 'false';
839 $last_read_qpart = 'NOW()';
840 }
841
842 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
843 $marked = 'true';
844 } else {
845 $marked = 'false';
846 }
847
848 if (find_article_filter($article_filters, 'publish')) {
849 $published = 'true';
850 } else {
851 $published = 'false';
852 }
853
854 // N-grams
855
856 if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
857
858 $result = db_query($link, "SELECT COUNT(*) AS similar FROM
859 ttrss_entries,ttrss_user_entries
860 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
861 AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
862 AND owner_uid = $owner_uid");
863
864 $ngram_similar = db_fetch_result($result, 0, "similar");
865
866 if ($debug_enabled) {
867 _debug("update_rss_feed: N-gram similar results: $ngram_similar");
868 }
869
870 if ($ngram_similar > 0) {
871 $unread = 'false';
872 }
873 }
874
875 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
876 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
877
878 $result = db_query($link,
879 "INSERT INTO ttrss_user_entries
880 (ref_id, owner_uid, feed_id, unread, last_read, marked,
881 published, score, tag_cache, label_cache, uuid,
882 last_marked, last_published)
883 VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
884 $last_read_qpart, $marked, $published, '$score', '', '',
885 '', $last_marked, $last_published)");
886
887 if (PUBSUBHUBBUB_HUB && $published == 'true') {
888 $rss_link = get_self_url_prefix() .
889 "/public.php?op=rss&id=-2&key=" .
890 get_feed_access_key($link, -2, false, $owner_uid);
891
892 $p = new Publisher(PUBSUBHUBBUB_HUB);
893
894 $pubsub_result = $p->publish_update($rss_link);
895 }
896
897 $result = db_query($link,
898 "SELECT int_id FROM ttrss_user_entries WHERE
899 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
900 feed_id = '$feed' LIMIT 1");
901
902 if (db_num_rows($result) == 1) {
903 $entry_int_id = db_fetch_result($result, 0, "int_id");
904 }
905 } else {
906 if ($debug_enabled) {
907 _debug("update_rss_feed: user record FOUND");
908 }
909
910 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
911 $entry_int_id = db_fetch_result($result, 0, "int_id");
912 }
913
914 if ($debug_enabled) {
915 _debug("update_rss_feed: RID: $entry_ref_id, IID: $entry_int_id");
916 }
917
918 $post_needs_update = false;
919 $update_insignificant = false;
920
921 if ($orig_num_comments != $num_comments) {
922 $post_needs_update = true;
923 $update_insignificant = true;
924 }
925
926 if ($entry_plugin_data != $orig_plugin_data) {
927 $post_needs_update = true;
928 $update_insignificant = true;
929 }
930
931 if ($content_hash != $orig_content_hash) {
932 $post_needs_update = true;
933 $update_insignificant = false;
934 }
935
936 if (db_escape_string($link, $orig_title) != $entry_title) {
937 $post_needs_update = true;
938 $update_insignificant = false;
939 }
940
941 // if post needs update, update it and mark all user entries
942 // linking to this post as updated
943 if ($post_needs_update) {
944
945 if (defined('DAEMON_EXTENDED_DEBUG')) {
946 _debug("update_rss_feed: post $entry_guid_hashed needs update...");
947 }
948
949 // print "<!-- post $orig_title needs update : $post_needs_update -->";
950
951 db_query($link, "UPDATE ttrss_entries
952 SET title = '$entry_title', content = '$entry_content',
953 content_hash = '$content_hash',
954 updated = '$entry_timestamp_fmt',
955 num_comments = '$num_comments',
956 plugin_data = '$entry_plugin_data'
957 WHERE id = '$ref_id'");
958
959 if (!$update_insignificant) {
960 if ($mark_unread_on_update) {
961 db_query($link, "UPDATE ttrss_user_entries
962 SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
963 }
964 }
965 }
966 }
967
968 db_query($link, "COMMIT");
969
970 if ($debug_enabled) {
971 _debug("update_rss_feed: assigning labels...");
972 }
973
974 assign_article_to_label_filters($link, $entry_ref_id, $article_filters,
975 $owner_uid, $article_labels);
976
977 if ($debug_enabled) {
978 _debug("update_rss_feed: looking for enclosures...");
979 }
980
981 // enclosures
982
983 $enclosures = array();
984
985 $encs = $item->get_enclosures();
986
987 if (is_array($encs)) {
988 foreach ($encs as $e) {
989 $e_item = array(
990 $e->link, $e->type, $e->length);
991 array_push($enclosures, $e_item);
992 }
993 }
994
995 if ($debug_enabled) {
996 _debug("update_rss_feed: article enclosures:");
997 print_r($enclosures);
998 }
999
1000 db_query($link, "BEGIN");
1001
1002 foreach ($enclosures as $enc) {
1003 $enc_url = db_escape_string($link, $enc[0]);
1004 $enc_type = db_escape_string($link, $enc[1]);
1005 $enc_dur = db_escape_string($link, $enc[2]);
1006
1007 $result = db_query($link, "SELECT id FROM ttrss_enclosures
1008 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
1009
1010 if (db_num_rows($result) == 0) {
1011 db_query($link, "INSERT INTO ttrss_enclosures
1012 (content_url, content_type, title, duration, post_id) VALUES
1013 ('$enc_url', '$enc_type', '', '$enc_dur', '$entry_ref_id')");
1014 }
1015 }
1016
1017 db_query($link, "COMMIT");
1018
1019 // check for manual tags (we have to do it here since they're loaded from filters)
1020
1021 foreach ($article_filters as $f) {
1022 if ($f["type"] == "tag") {
1023
1024 $manual_tags = trim_array(explode(",", $f["param"]));
1025
1026 foreach ($manual_tags as $tag) {
1027 if (tag_is_valid($tag)) {
1028 array_push($entry_tags, $tag);
1029 }
1030 }
1031 }
1032 }
1033
1034 // Skip boring tags
1035
1036 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref($link,
1037 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1038
1039 $filtered_tags = array();
1040 $tags_to_cache = array();
1041
1042 if ($entry_tags && is_array($entry_tags)) {
1043 foreach ($entry_tags as $tag) {
1044 if (array_search($tag, $boring_tags) === false) {
1045 array_push($filtered_tags, $tag);
1046 }
1047 }
1048 }
1049
1050 $filtered_tags = array_unique($filtered_tags);
1051
1052 if ($debug_enabled) {
1053 _debug("update_rss_feed: filtered article tags:");
1054 print_r($filtered_tags);
1055 }
1056
1057 // Save article tags in the database
1058
1059 if (count($filtered_tags) > 0) {
1060
1061 db_query($link, "BEGIN");
1062
1063 foreach ($filtered_tags as $tag) {
1064
1065 $tag = sanitize_tag($tag);
1066 $tag = db_escape_string($link, $tag);
1067
1068 if (!tag_is_valid($tag)) continue;
1069
1070 $result = db_query($link, "SELECT id FROM ttrss_tags
1071 WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1072 owner_uid = '$owner_uid' LIMIT 1");
1073
1074 if ($result && db_num_rows($result) == 0) {
1075
1076 db_query($link, "INSERT INTO ttrss_tags
1077 (owner_uid,tag_name,post_int_id)
1078 VALUES ('$owner_uid','$tag', '$entry_int_id')");
1079 }
1080
1081 array_push($tags_to_cache, $tag);
1082 }
1083
1084 /* update the cache */
1085
1086 $tags_to_cache = array_unique($tags_to_cache);
1087
1088 $tags_str = db_escape_string($link, join(",", $tags_to_cache));
1089
1090 db_query($link, "UPDATE ttrss_user_entries
1091 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1092 AND owner_uid = $owner_uid");
1093
1094 db_query($link, "COMMIT");
1095 }
1096
1097 if (get_pref($link, "AUTO_ASSIGN_LABELS", $owner_uid, false)) {
1098 if ($debug_enabled) {
1099 _debug("update_rss_feed: auto-assigning labels...");
1100 }
1101
1102 foreach ($labels as $label) {
1103 $caption = preg_quote($label["caption"]);
1104
1105 if ($caption && preg_match("/\b$caption\b/i", "$tags_str " . strip_tags($entry_content) . " $entry_title")) {
1106 if (!labels_contains_caption($article_labels, $caption)) {
1107 label_add_article($link, $entry_ref_id, $caption, $owner_uid);
1108 }
1109 }
1110 }
1111 }
1112
1113 if ($debug_enabled) {
1114 _debug("update_rss_feed: article processed");
1115 }
1116 }
1117
1118 if (!$last_updated) {
1119 if ($debug_enabled) {
1120 _debug("update_rss_feed: new feed, catching it up...");
1121 }
1122 catchup_feed($link, $feed, false, $owner_uid);
1123 }
1124
1125 if ($debug_enabled) {
1126 _debug("purging feed...");
1127 }
1128
1129 purge_feed($link, $feed, 0, $debug_enabled);
1130
1131 db_query($link, "UPDATE ttrss_feeds
1132 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
1133
1134 // db_query($link, "COMMIT");
1135
1136 } else {
1137
1138 $error_msg = db_escape_string($link, mb_substr($rss->error(), 0, 245));
1139
1140 if ($debug_enabled) {
1141 _debug("update_rss_feed: error fetching feed: $error_msg");
1142 }
1143
1144 db_query($link,
1145 "UPDATE ttrss_feeds SET last_error = '$error_msg',
1146 last_updated = NOW() WHERE id = '$feed'");
1147 }
1148
1149 unset($rss);
1150
1151 if ($debug_enabled) {
1152 _debug("update_rss_feed: done");
1153 }
1154
1155 }
1156
1157 function cache_images($html, $site_url, $debug) {
1158 $cache_dir = CACHE_DIR . "/images";
1159
1160 libxml_use_internal_errors(true);
1161
1162 $charset_hack = '<head>
1163 <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1164 </head>';
1165
1166 $doc = new DOMDocument();
1167 $doc->loadHTML($charset_hack . $html);
1168 $xpath = new DOMXPath($doc);
1169
1170 $entries = $xpath->query('(//img[@src])');
1171
1172 foreach ($entries as $entry) {
1173 if ($entry->hasAttribute('src')) {
1174 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1175
1176 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1177
1178 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1179
1180 if (!file_exists($local_filename)) {
1181 $file_content = fetch_file_contents($src);
1182
1183 if ($file_content && strlen($file_content) > 1024) {
1184 file_put_contents($local_filename, $file_content);
1185 }
1186 }
1187
1188 if (file_exists($local_filename)) {
1189 $entry->setAttribute('src', SELF_URL_PATH . '/image.php?url=' .
1190 base64_encode($src));
1191 }
1192 }
1193 }
1194
1195 $node = $doc->getElementsByTagName('body')->item(0);
1196
1197 return $doc->saveXML($node);
1198 }
1199
1200 function expire_lock_files($debug) {
1201 if ($debug) _debug("Removing old lock files...");
1202
1203 $num_deleted = 0;
1204
1205 if (is_writable(LOCK_DIRECTORY)) {
1206 $files = glob(LOCK_DIRECTORY . "/*.lock");
1207
1208 if ($files) {
1209 foreach ($files as $file) {
1210 if (!file_is_locked($file) && time() - filemtime($file) > 86400*2) {
1211 unlink($file);
1212 ++$num_deleted;
1213 }
1214 }
1215 }
1216 }
1217
1218 if ($debug) _debug("Removed $num_deleted files.");
1219 }
1220
1221 function expire_cached_files($debug) {
1222 foreach (array("simplepie", "images", "export", "upload") as $dir) {
1223 $cache_dir = CACHE_DIR . "/$dir";
1224
1225 if ($debug) _debug("Expiring $cache_dir");
1226
1227 $num_deleted = 0;
1228
1229 if (is_writable($cache_dir)) {
1230 $files = glob("$cache_dir/*");
1231
1232 if ($files) {
1233 foreach ($files as $file) {
1234 if (time() - filemtime($file) > 86400*7) {
1235 unlink($file);
1236
1237 ++$num_deleted;
1238 }
1239 }
1240 }
1241 }
1242
1243 if ($debug) _debug("Removed $num_deleted files.");
1244 }
1245 }
1246
1247 /**
1248 * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1249 * Returns the url query as associative array
1250 *
1251 * @param string query
1252 * @return array params
1253 */
1254 function convertUrlQuery($query) {
1255 $queryParts = explode('&', $query);
1256
1257 $params = array();
1258
1259 foreach ($queryParts as $param) {
1260 $item = explode('=', $param);
1261 $params[$item[0]] = $item[1];
1262 }
1263
1264 return $params;
1265 }
1266
1267 function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
1268 $matches = array();
1269
1270 foreach ($filters as $filter) {
1271 $match_any_rule = $filter["match_any_rule"];
1272 $inverse = $filter["inverse"];
1273 $filter_match = false;
1274
1275 foreach ($filter["rules"] as $rule) {
1276 $match = false;
1277 $reg_exp = $rule["reg_exp"];
1278 $rule_inverse = $rule["inverse"];
1279
1280 if (!$reg_exp)
1281 continue;
1282
1283 switch ($rule["type"]) {
1284 case "title":
1285 $match = @preg_match("/$reg_exp/i", $title);
1286 break;
1287 case "content":
1288 // we don't need to deal with multiline regexps
1289 $content = preg_replace("/[\r\n\t]/", "", $content);
1290
1291 $match = @preg_match("/$reg_exp/i", $content);
1292 break;
1293 case "both":
1294 // we don't need to deal with multiline regexps
1295 $content = preg_replace("/[\r\n\t]/", "", $content);
1296
1297 $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
1298 break;
1299 case "link":
1300 $match = @preg_match("/$reg_exp/i", $link);
1301 break;
1302 case "author":
1303 $match = @preg_match("/$reg_exp/i", $author);
1304 break;
1305 case "tag":
1306 $tag_string = join(",", $tags);
1307 $match = @preg_match("/$reg_exp/i", $tag_string);
1308 break;
1309 }
1310
1311 if ($rule_inverse) $match = !$match;
1312
1313 if ($match_any_rule) {
1314 if ($match) {
1315 $filter_match = true;
1316 break;
1317 }
1318 } else {
1319 $filter_match = $match;
1320 if (!$match) {
1321 break;
1322 }
1323 }
1324 }
1325
1326 if ($inverse) $filter_match = !$filter_match;
1327
1328 if ($filter_match) {
1329 foreach ($filter["actions"] AS $action) {
1330 array_push($matches, $action);
1331
1332 // if Stop action encountered, perform no further processing
1333 if ($action["type"] == "stop") return $matches;
1334 }
1335 }
1336 }
1337
1338 return $matches;
1339 }
1340
1341 function find_article_filter($filters, $filter_name) {
1342 foreach ($filters as $f) {
1343 if ($f["type"] == $filter_name) {
1344 return $f;
1345 };
1346 }
1347 return false;
1348 }
1349
1350 function find_article_filters($filters, $filter_name) {
1351 $results = array();
1352
1353 foreach ($filters as $f) {
1354 if ($f["type"] == $filter_name) {
1355 array_push($results, $f);
1356 };
1357 }
1358 return $results;
1359 }
1360
1361 function calculate_article_score($filters) {
1362 $score = 0;
1363
1364 foreach ($filters as $f) {
1365 if ($f["type"] == "score") {
1366 $score += $f["param"];
1367 };
1368 }
1369 return $score;
1370 }
1371
1372 function labels_contains_caption($labels, $caption) {
1373 foreach ($labels as $label) {
1374 if ($label[1] == $caption) {
1375 return true;
1376 }
1377 }
1378
1379 return false;
1380 }
1381
1382 function assign_article_to_label_filters($link, $id, $filters, $owner_uid, $article_labels) {
1383 foreach ($filters as $f) {
1384 if ($f["type"] == "label") {
1385 if (!labels_contains_caption($article_labels, $f["param"])) {
1386 label_add_article($link, $id, $f["param"], $owner_uid);
1387 }
1388 }
1389 }
1390 }
1391
1392 function make_guid_from_title($title) {
1393 return preg_replace("/[ \"\',.:;]/", "-",
1394 mb_strtolower(strip_tags($title), 'utf-8'));
1395 }
1396
1397
1398 ?>