]> 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
support comments on simplepie properly
[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);
0f556e3e 4 define('DAEMON_SLEEP_INTERVAL', 60);
09e8bdfd 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
61c1812f
AD		 63 define('PREFS_NO_CACHE', true);
64
2c08214a 65 // Test if the user has loggued in recently. If not, it does not update its feeds.
09e8bdfd 66 if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
2c08214a
AD		 67 if (DB_TYPE == "pgsql") {
68 $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
69 } else {
70 $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
71 }
72 } else {
73 $login_thresh_qpart = "";
74 }
75
76 // Test if the feed need a update (update interval exceded).
77 if (DB_TYPE == "pgsql") {
78 $update_limit_qpart = "AND ((
79 ttrss_feeds.update_interval = 0
80 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
81 ) OR (
82 ttrss_feeds.update_interval > 0
83 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
941e48a4
AD		 84 ) OR ttrss_feeds.last_updated IS NULL
85 OR last_updated = '1970-01-01 00:00:00')";
2c08214a
AD		 86 } else {
87 $update_limit_qpart = "AND ((
88 ttrss_feeds.update_interval = 0
89 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
90 ) OR (
91 ttrss_feeds.update_interval > 0
92 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
941e48a4
AD		 93 ) OR ttrss_feeds.last_updated IS NULL
94 OR last_updated = '1970-01-01 00:00:00')";
2c08214a
AD		 95 }
96
97 // Test if feed is currently being updated by another process.
98 if (DB_TYPE == "pgsql") {
99 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '5 minutes')";
100 } else {
101 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 5 MINUTE))";
102 }
103
104 // Test if there is a limit to number of updated feeds
105 $query_limit = "";
106 if($limit) $query_limit = sprintf("LIMIT %d", $limit);
107
108 $random_qpart = sql_random_function();
109
110 // We search for feed needing update.
111 $result = db_query($link, "SELECT ttrss_feeds.feed_url,ttrss_feeds.id, ttrss_feeds.owner_uid,
112 ".SUBSTRING_FOR_DATE."(ttrss_feeds.last_updated,1,19) AS last_updated,
113 ttrss_feeds.update_interval
114 FROM
115 ttrss_feeds, ttrss_users, ttrss_user_prefs
116 WHERE
117 ttrss_feeds.owner_uid = ttrss_users.id
118 AND ttrss_users.id = ttrss_user_prefs.owner_uid
119 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
120 $login_thresh_qpart $update_limit_qpart
121 $updstart_thresh_qpart
122 ORDER BY $random_qpart $query_limit");
123
124 $user_prefs_cache = array();
125
126 if($debug) _debug(sprintf("Scheduled %d feeds to update...\n", db_num_rows($result)));
127
128 // Here is a little cache magic in order to minimize risk of double feed updates.
129 $feeds_to_update = array();
130 while ($line = db_fetch_assoc($result)) {
131 $feeds_to_update[$line['id']] = $line;
132 }
133
134 // We update the feed last update started date before anything else.
135 // There is no lag due to feed contents downloads
136 // It prevent an other process to update the same feed.
137 $feed_ids = array_keys($feeds_to_update);
138 if($feed_ids) {
139 db_query($link, sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
140 WHERE id IN (%s)", implode(',', $feed_ids)));
141 }
142
3c696512 143 expire_cached_files($debug);
2a91b6ff 144 expire_lock_files($debug);
3c696512 145
2c08214a
AD		 146 // For each feed, we call the feed update function.
147 while ($line = array_pop($feeds_to_update)) {
148
149 if($debug) _debug("Feed: " . $line["feed_url"] . ", " . $line["last_updated"]);
150
151 update_rss_feed($link, $line["id"], true);
152
153 sleep(1); // prevent flood (FIXME make this an option?)
154 }
155
156 // Send feed digests by email if needed.
8437c066 157 send_headlines_digests($link, $debug);
2c08214a
AD		 158
159 } // function update_daemon_common
160
87764a50 161 // ignore_daemon is not used
0e4a7d7a 162 function update_rss_feed($link, $feed, $ignore_daemon = false, $no_cache = false,
2c08214a
AD		 163 $override_url = false) {
164
165 require_once "lib/simplepie/simplepie.inc";
166 require_once "lib/magpierss/rss_fetch.inc";
167 require_once 'lib/magpierss/rss_utils.inc';
168
2c08214a
AD		 169 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
170
2c08214a
AD		 171 if ($debug_enabled) {
172 _debug("update_rss_feed: start");
173 }
174
87764a50
AD		 175 $result = db_query($link, "SELECT id,update_interval,auth_login,
176 feed_url,auth_pass,cache_images,update_method,last_updated,cache_content,
177 mark_unread_on_update, owner_uid, update_on_checksum_change,
178 pubsub_state
179 FROM ttrss_feeds WHERE id = '$feed'");
2c08214a
AD		 180
181 if (db_num_rows($result) == 0) {
182 if ($debug_enabled) {
183 _debug("update_rss_feed: feed $feed NOT FOUND/SKIPPED");
184 }
185 return false;
186 }
187
188 $update_method = db_fetch_result($result, 0, "update_method");
189 $last_updated = db_fetch_result($result, 0, "last_updated");
190 $owner_uid = db_fetch_result($result, 0, "owner_uid");
191 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
192 0, "mark_unread_on_update"));
193 $update_on_checksum_change = sql_bool_to_bool(db_fetch_result($result,
194 0, "update_on_checksum_change"));
195 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
196
197 db_query($link, "UPDATE ttrss_feeds SET last_update_started = NOW()
198 WHERE id = '$feed'");
199
200 $auth_login = db_fetch_result($result, 0, "auth_login");
201 $auth_pass = db_fetch_result($result, 0, "auth_pass");
202
203 if ($update_method == 0)
204 $update_method = DEFAULT_UPDATE_METHOD + 1;
205
206 // 1 - Magpie
207 // 2 - SimplePie
208 // 3 - Twitter OAuth
209
210 if ($update_method == 2)
211 $use_simplepie = true;
212 else
213 $use_simplepie = false;
214
215 if ($debug_enabled) {
216 _debug("update method: $update_method (feed setting: $update_method) (use simplepie: $use_simplepie)\n");
217 }
218
219 if ($update_method == 1) {
220 $auth_login = urlencode($auth_login);
221 $auth_pass = urlencode($auth_pass);
222 }
223
2c08214a 224 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
87764a50 225 $cache_content = sql_bool_to_bool(db_fetch_result($result, 0, "cache_content"));
2c08214a
AD		 226 $fetch_url = db_fetch_result($result, 0, "feed_url");
227
2c08214a
AD		 228 $feed = db_escape_string($feed);
229
230 if ($auth_login && $auth_pass ){
231 $url_parts = array();
232 preg_match("/(^[^:]*):\/\/(.*)/", $fetch_url, $url_parts);
233
234 if ($url_parts[1] && $url_parts[2]) {
235 $fetch_url = $url_parts[1] . "://$auth_login:$auth_pass@" . $url_parts[2];
236 }
237
238 }
239
240 if ($override_url)
241 $fetch_url = $override_url;
242
243 if ($debug_enabled) {
244 _debug("update_rss_feed: fetching [$fetch_url]...");
245 }
246
0e4a7d7a
AD		 247 // Ignore cache if new feed or manual update.
248 $cache_age = (is_null($last_updated) || $last_updated == '1970-01-01 00:00:00') ?
249 -1 : get_feed_update_interval($link, $feed) * 60;
2c08214a 250
304aadb9 251 if ($update_method == 1) {
2c08214a 252
0e4a7d7a
AD		 253 define('MAGPIE_CACHE_AGE', $cache_age);
254 define('MAGPIE_CACHE_ON', !$no_cache);
c0c2abba 255 define('MAGPIE_FETCH_TIME_OUT', $no_cache ? 15 : 60);
0e4a7d7a 256 define('MAGPIE_CACHE_DIR', CACHE_DIR . "/magpie");
2c08214a 257
0e4a7d7a 258 $rss = @fetch_rss($fetch_url);
2c08214a 259 } else {
0e4a7d7a 260 $simplepie_cache_dir = CACHE_DIR . "/simplepie";
2c08214a 261
0e4a7d7a
AD		 262 if (!is_dir($simplepie_cache_dir)) {
263 mkdir($simplepie_cache_dir);
264 }
2c08214a 265
0e4a7d7a
AD		 266 $rss = new SimplePie();
267 $rss->set_useragent(SELF_USER_AGENT);
c0c2abba 268 $rss->set_timeout($no_cache ? 15 : 60);
0e4a7d7a
AD		 269 $rss->set_feed_url($fetch_url);
270 $rss->set_output_encoding('UTF-8');
271 //$rss->force_feed(true);
2c08214a 272
0e4a7d7a
AD		 273 if ($debug_enabled) {
274 _debug("feed update interval (sec): " .
275 get_feed_update_interval($link, $feed)*60);
276 }
2c08214a 277
0e4a7d7a 278 $rss->enable_cache(!$no_cache);
2c08214a 279
0e4a7d7a
AD		 280 if (!$no_cache) {
281 $rss->set_cache_location($simplepie_cache_dir);
282 $rss->set_cache_duration($cache_age);
2c08214a
AD		 283 }
284
0e4a7d7a 285 $rss->init();
2c08214a
AD		 286 }
287
288// print_r($rss);
289
290 if ($debug_enabled) {
291 _debug("update_rss_feed: fetch done, parsing...");
292 }
293
294 $feed = db_escape_string($feed);
295
296 if ($update_method == 2) {
297 $fetch_ok = !$rss->error();
298 } else {
299 $fetch_ok = !!$rss;
300 }
301
302 if ($fetch_ok) {
303
304 if ($debug_enabled) {
305 _debug("update_rss_feed: processing feed data...");
306 }
307
308// db_query($link, "BEGIN");
309
382268c6
AD		 310 if (DB_TYPE == "pgsql") {
311 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
312 } else {
313 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
314 }
315
316 $result = db_query($link, "SELECT title,icon_url,site_url,owner_uid,
317 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
318 favicon_needs_check
2c08214a
AD		 319 FROM ttrss_feeds WHERE id = '$feed'");
320
321 $registered_title = db_fetch_result($result, 0, "title");
322 $orig_icon_url = db_fetch_result($result, 0, "icon_url");
323 $orig_site_url = db_fetch_result($result, 0, "site_url");
382268c6
AD		 324 $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
325 "favicon_needs_check"));
2c08214a
AD		 326
327 $owner_uid = db_fetch_result($result, 0, "owner_uid");
328
329 if ($use_simplepie) {
0cf81637 330 $site_url = db_escape_string(trim($rss->get_link()));
2c08214a 331 } else {
0cf81637
AD		 332 $site_url = db_escape_string(trim($rss->channel["link"]));
333 }
334
335 // weird, weird Magpie
336 if (!$use_simplepie) {
337 if (!$site_url) $site_url = db_escape_string($rss->channel["link_"]);
2c08214a
AD		 338 }
339
340 $site_url = rewrite_relative_url($fetch_url, $site_url);
0cf81637 341 $site_url = substr($site_url, 0, 250);
2c08214a
AD		 342
343 if ($debug_enabled) {
344 _debug("update_rss_feed: checking favicon...");
345 }
346
f2798eb6
AD		 347 if ($favicon_needs_check) {
348 check_feed_favicon($site_url, $feed, $link);
349
350 db_query($link, "UPDATE ttrss_feeds SET favicon_last_checked = NOW()
351 WHERE id = '$feed'");
352 }
2c08214a
AD		 353
354 if (!$registered_title || $registered_title == "[Unknown]") {
355
356 if ($use_simplepie) {
357 $feed_title = db_escape_string($rss->get_title());
358 } else {
359 $feed_title = db_escape_string($rss->channel["title"]);
360 }
361
362 if ($debug_enabled) {
363 _debug("update_rss_feed: registering title: $feed_title");
364 }
365
366 db_query($link, "UPDATE ttrss_feeds SET
367 title = '$feed_title' WHERE id = '$feed'");
368 }
369
0cf81637 370 if ($site_url && $orig_site_url != $site_url) {
2c08214a
AD		 371 db_query($link, "UPDATE ttrss_feeds SET
372 site_url = '$site_url' WHERE id = '$feed'");
373 }
374
375// print "I: " . $rss->channel["image"]["url"];
376
377 if (!$use_simplepie) {
0cf81637 378 $icon_url = db_escape_string(trim($rss->image["url"]));
2c08214a 379 } else {
0cf81637 380 $icon_url = db_escape_string(trim($rss->get_image_url()));
2c08214a
AD		 381 }
382
0cf81637 383 $icon_url = rewrite_relative_url($fetch_url, $icon_url);
2c08214a
AD		 384 $icon_url = substr($icon_url, 0, 250);
385
386 if ($icon_url && $orig_icon_url != $icon_url) {
387 db_query($link, "UPDATE ttrss_feeds SET icon_url = '$icon_url' WHERE id = '$feed'");
388 }
389
390 if ($debug_enabled) {
b24504b1 391 _debug("update_rss_feed: loading filters & labels...");
2c08214a
AD		 392 }
393
394 $filters = load_filters($link, $feed, $owner_uid);
b24504b1 395 $labels = get_all_labels($link, $owner_uid);
2c08214a 396
67bd0b1f
AD		 397 if ($debug_enabled) {
398 //print_r($filters);
399 _debug("update_rss_feed: " . count($filters) . " filters loaded.");
400 }
2c08214a 401
cc85704f
AD		 402 $filter_plugins = array();
403
404 if (defined('_ARTICLE_FILTER_PLUGINS')) {
405 foreach (explode(",", _ARTICLE_FILTER_PLUGINS) as $p) {
406 $pclass = "filter_" . trim($p);
407
408 if (class_exists($pclass)) {
409 $plugin = new $pclass($link);
410 array_push($filter_plugins, $plugin);
411 }
412 }
413 }
414
415 if ($debug_enabled) {
416 _debug("update_rss_feed: " . count($filter_plugins) . " filter plugins loaded.");
417 }
418
2c08214a
AD		 419 if ($use_simplepie) {
420 $iterator = $rss->get_items();
421 } else {
422 $iterator = $rss->items;
423 if (!$iterator || !is_array($iterator)) $iterator = $rss->entries;
424 if (!$iterator || !is_array($iterator)) $iterator = $rss;
425 }
426
427 if (!is_array($iterator)) {
428 /* db_query($link, "UPDATE ttrss_feeds
429 SET last_error = 'Parse error: can\'t find any articles.'
430 WHERE id = '$feed'"); */
431
432 // clear any errors and mark feed as updated if fetched okay
433 // even if it's blank
434
435 if ($debug_enabled) {
436 _debug("update_rss_feed: entry iterator is not an array, no articles?");
437 }
438
439 db_query($link, "UPDATE ttrss_feeds
440 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
441
442 return; // no articles
443 }
444
445 if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
446
447 if ($debug_enabled) _debug("update_rss_feed: checking for PUSH hub...");
448
449 $feed_hub_url = false;
450 if ($use_simplepie) {
451 $links = $rss->get_links('hub');
452
453 if ($links && is_array($links)) {
454 foreach ($links as $l) {
455 $feed_hub_url = $l;
456 break;
457 }
458 }
459
460 } else {
461 $atom = $rss->channel['atom'];
462
463 if ($atom) {
464 if ($atom['link@rel'] == 'hub') {
465 $feed_hub_url = $atom['link@href'];
466 }
467
468 if (!$feed_hub_url && $atom['link#'] > 1) {
469 for ($i = 2; $i <= $atom['link#']; $i++) {
470 if ($atom["link#$i@rel"] == 'hub') {
471 $feed_hub_url = $atom["link#$i@href"];
472 break;
473 }
474 }
475 }
476 } else {
477 $feed_hub_url = $rss->channel['link_hub'];
478 }
479 }
480
481 if ($debug_enabled) _debug("update_rss_feed: feed hub url: $feed_hub_url");
482
483 if ($feed_hub_url && function_exists('curl_init') &&
484 !ini_get("open_basedir")) {
485
486 require_once 'lib/pubsubhubbub/subscriber.php';
487
488 $callback_url = get_self_url_prefix() .
489 "/public.php?op=pubsub&id=$feed";
490
491 $s = new Subscriber($feed_hub_url, $callback_url);
492
493 $rc = $s->subscribe($fetch_url);
494
495 if ($debug_enabled)
496 _debug("update_rss_feed: feed hub url found, subscribe request sent.");
497
498 db_query($link, "UPDATE ttrss_feeds SET pubsub_state = 1
499 WHERE id = '$feed'");
500 }
501 }
502
503 if ($debug_enabled) {
504 _debug("update_rss_feed: processing articles...");
505 }
506
507 foreach ($iterator as $item) {
2c08214a
AD		 508 if ($_REQUEST['xdebug'] == 2) {
509 print_r($item);
510 }
511
512 if ($use_simplepie) {
513 $entry_guid = $item->get_id();
514 if (!$entry_guid) $entry_guid = $item->get_link();
515 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
516
517 } else {
518
519 $entry_guid = $item["id"];
520
521 if (!$entry_guid) $entry_guid = $item["guid"];
522 if (!$entry_guid) $entry_guid = $item["about"];
523 if (!$entry_guid) $entry_guid = $item["link"];
524 if (!$entry_guid) $entry_guid = make_guid_from_title($item["title"]);
525 }
526
765509c5 527 if ($cache_content) {
d5e9cf28 528 $entry_guid = "ccache:$entry_guid";
765509c5
AD		 529 }
530
531 if ($auth_login || $auth_pass) {
532 $entry_guid = "auth,$owner_uid:$entry_guid";
533 }
534
2c08214a
AD		 535 if ($debug_enabled) {
536 _debug("update_rss_feed: guid $entry_guid");
537 }
538
539 if (!$entry_guid) continue;
540
541 $entry_timestamp = "";
542
543 if ($use_simplepie) {
544 $entry_timestamp = strtotime($item->get_date());
545 } else {
546 $rss_2_date = $item['pubdate'];
547 $rss_1_date = $item['dc']['date'];
548 $atom_date = $item['issued'];
549 if (!$atom_date) $atom_date = $item['updated'];
550
551 if ($atom_date != "") $entry_timestamp = parse_w3cdtf($atom_date);
552 if ($rss_1_date != "") $entry_timestamp = parse_w3cdtf($rss_1_date);
553 if ($rss_2_date != "") $entry_timestamp = strtotime($rss_2_date);
554
555 }
556
557 if ($entry_timestamp == "" || $entry_timestamp == -1 || !$entry_timestamp) {
558 $entry_timestamp = time();
559 $no_orig_date = 'true';
560 } else {
561 $no_orig_date = 'false';
562 }
563
564 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
565
566 if ($debug_enabled) {
567 _debug("update_rss_feed: date $entry_timestamp [$entry_timestamp_fmt]");
568 }
569
570 if ($use_simplepie) {
571 $entry_title = $item->get_title();
572 } else {
573 $entry_title = trim(strip_tags($item["title"]));
574 }
575
576 if ($use_simplepie) {
577 $entry_link = $item->get_link();
578 } else {
579 // strange Magpie workaround
580 $entry_link = $item["link_"];
581 if (!$entry_link) $entry_link = $item["link"];
582 }
583
584 $entry_link = rewrite_relative_url($site_url, $entry_link);
585
586 if ($debug_enabled) {
587 _debug("update_rss_feed: title $entry_title");
588 _debug("update_rss_feed: link $entry_link");
589 }
590
591 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
592
593 $entry_link = strip_tags($entry_link);
594
595 if ($use_simplepie) {
596 $entry_content = $item->get_content();
597 if (!$entry_content) $entry_content = $item->get_description();
598 } else {
599 $entry_content = $item["content:escaped"];
600
601 if (!$entry_content) $entry_content = $item["content:encoded"];
eb38af4e 602 if (!$entry_content && is_array($entry_content)) $entry_content = $item["content"]["encoded"];
2c08214a
AD		 603 if (!$entry_content) $entry_content = $item["content"];
604
605 if (is_array($entry_content)) $entry_content = $entry_content[0];
606
607 // Magpie bugs are getting ridiculous
608 if (trim($entry_content) == "Array") $entry_content = false;
609
610 if (!$entry_content) $entry_content = $item["atom_content"];
611 if (!$entry_content) $entry_content = $item["summary"];
612
613 if (!$entry_content ||
614 strlen($entry_content) < strlen($item["description"])) {
615 $entry_content = $item["description"];
616 };
617
618 // WTF
619 if (is_array($entry_content)) {
620 $entry_content = $entry_content["encoded"];
621 if (!$entry_content) $entry_content = $entry_content["escaped"];
622 }
623 }
624
c5867798 625 if ($cache_images && is_writable(CACHE_DIR . '/images'))
3c696512
AD		 626 $entry_content = cache_images($entry_content, $site_url, $debug_enabled);
627
2c08214a
AD		 628 if ($_REQUEST["xdebug"] == 2) {
629 print "update_rss_feed: content: ";
487f0750 630 print $entry_content;
3c696512 631 print "\n";
2c08214a
AD		 632 }
633
87764a50 634 $entry_cached_content = "";
2c08214a
AD		 635
636 if ($use_simplepie) {
35cb2b8f 637 $entry_comments = strip_tags($item->data["comments"]);
2c08214a
AD		 638 if ($item->get_author()) {
639 $entry_author_item = $item->get_author();
640 $entry_author = $entry_author_item->get_name();
641 if (!$entry_author) $entry_author = $entry_author_item->get_email();
642
643 $entry_author = db_escape_string($entry_author);
644 }
645 } else {
35cb2b8f 646 $entry_comments = strip_tags($item["comments"]);
2c08214a
AD		 647
648 $entry_author = db_escape_string(strip_tags($item['dc']['creator']));
649
650 if ($item['author']) {
651
652 if (is_array($item['author'])) {
653
654 if (!$entry_author) {
655 $entry_author = db_escape_string(strip_tags($item['author']['name']));
656 }
657
658 if (!$entry_author) {
659 $entry_author = db_escape_string(strip_tags($item['author']['email']));
660 }
661 }
662
663 if (!$entry_author) {
664 $entry_author = db_escape_string(strip_tags($item['author']));
665 }
666 }
667 }
668
669 if (preg_match('/^[\t\n\r ]*$/', $entry_author)) $entry_author = '';
670
671 $entry_guid = db_escape_string(strip_tags($entry_guid));
672 $entry_guid = mb_substr($entry_guid, 0, 250);
673
674 $result = db_query($link, "SELECT id FROM ttrss_entries
675 WHERE guid = '$entry_guid'");
676
2c08214a
AD		 677 $entry_comments = mb_substr(db_escape_string($entry_comments), 0, 250);
678 $entry_author = mb_substr($entry_author, 0, 250);
679
680 if ($use_simplepie) {
83e6e313
AD		 681 $num_comments = $item->get_item_tags('http://purl.org/rss/1.0/modules/slash/', 'comments');
682
683 if (is_array($num_comments) && is_array($num_comments[0])) {
684 $num_comments = (int) $num_comments[0]["data"];
685 } else {
686 $num_comments = 0;
687 }
2c08214a 688 } else {
83e6e313 689 $num_comments = (int) $item["slash"]["comments"];
2c08214a
AD		 690 }
691
2c08214a 692 if ($debug_enabled) {
83e6e313 693 _debug("update_rss_feed: num_comments: $num_comments");
2c08214a
AD		 694 _debug("update_rss_feed: looking for tags [1]...");
695 }
696
697 // parse <category> entries into tags
698
699 $additional_tags = array();
700
701 if ($use_simplepie) {
702
703 $additional_tags_src = $item->get_categories();
704
705 if (is_array($additional_tags_src)) {
706 foreach ($additional_tags_src as $tobj) {
707 array_push($additional_tags, $tobj->get_term());
708 }
709 }
710
711 if ($debug_enabled) {
712 _debug("update_rss_feed: category tags:");
713 print_r($additional_tags);
714 }
715
716 } else {
717
718 $t_ctr = $item['category#'];
719
720 if ($t_ctr == 0) {
721 $additional_tags = array();
722 } else if ($t_ctr > 0) {
723 $additional_tags = array($item['category']);
724
725 if ($item['category@term']) {
726 array_push($additional_tags, $item['category@term']);
727 }
728
729 for ($i = 0; $i <= $t_ctr; $i++ ) {
730 if ($item["category#$i"]) {
731 array_push($additional_tags, $item["category#$i"]);
732 }
733
734 if ($item["category#$i@term"]) {
735 array_push($additional_tags, $item["category#$i@term"]);
736 }
737 }
738 }
739
740 // parse <dc:subject> elements
741
742 $t_ctr = $item['dc']['subject#'];
743
744 if ($t_ctr > 0) {
745 array_push($additional_tags, $item['dc']['subject']);
746
747 for ($i = 0; $i <= $t_ctr; $i++ ) {
748 if ($item['dc']["subject#$i"]) {
749 array_push($additional_tags, $item['dc']["subject#$i"]);
750 }
751 }
752 }
753 }
754
755 if ($debug_enabled) {
756 _debug("update_rss_feed: looking for tags [2]...");
757 }
758
759 /* taaaags */
760 // <a href="..." rel="tag">Xorg</a>, //
761
762 $entry_tags = null;
763
764 preg_match_all("/<a.*?rel=['\"]tag['\"].*?\>([^<]+)<\/a>/i",
2bbd6994 765 $entry_content, $entry_tags);
2c08214a
AD		 766
767 $entry_tags = $entry_tags[1];
768
769 $entry_tags = array_merge($entry_tags, $additional_tags);
770 $entry_tags = array_unique($entry_tags);
771
772 for ($i = 0; $i < count($entry_tags); $i++)
773 $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
774
775 if ($debug_enabled) {
6aff7845
AD		 776 //_debug("update_rss_feed: unfiltered tags found:");
777 //print_r($entry_tags);
2c08214a
AD		 778 }
779
2c08214a
AD		 780 if ($debug_enabled) {
781 _debug("update_rss_feed: done collecting data [TITLE:$entry_title]");
782 }
783
cc85704f
AD		 784 // TODO: less memory-hungry implementation
785 if (count($filter_plugins) > 0) {
786 if ($debug_enabled) {
787 _debug("update_rss_feed: applying plugin filters...");
788 }
789
790 $article = array("owner_uid" => $owner_uid,
791 "title" => $entry_title,
792 "content" => $entry_content,
793 "link" => $entry_link,
794 "tags" => $entry_tags,
795 "author" => $entry_author);
796
797 foreach ($filter_plugins as $plugin) {
798 $article = $plugin->filter_article($article);
799 }
800
801 $entry_title = $article["title"];
802 $entry_content = $article["content"];
803 $entry_tags = $article["tags"];
804 $entry_author = $article["author"];
805 }
806
2bbd6994
AD		 807 $entry_content = db_escape_string($entry_content, false);
808 $entry_title = db_escape_string($entry_title);
809 $entry_author = db_escape_string($entry_author);
810 $entry_link = db_escape_string($entry_link);
811
cc85704f
AD		 812 $content_hash = "SHA1:" . sha1(strip_tags($entry_content));
813
2c08214a
AD		 814 db_query($link, "BEGIN");
815
816 if (db_num_rows($result) == 0) {
817
818 if ($debug_enabled) {
819 _debug("update_rss_feed: base guid not found");
820 }
821
87764a50
AD		 822 if ($cache_content) {
823 if ($debug_enabled) {
8054439f 824 _debug("update_rss_feed: caching content (initial)...");
87764a50
AD		 825 }
826
827 $entry_cached_content = cache_content($link, $entry_link, $auth_login, $auth_pass);
828
829 if ($cache_images && is_writable(CACHE_DIR . '/images'))
830 $entry_cached_content = cache_images($entry_cached_content, $site_url, $debug_enabled);
831
832 $entry_cached_content = db_escape_string($entry_cached_content, false);
87764a50
AD		 833 }
834
2c08214a
AD		 835 // base post entry does not exist, create it
836
837 $result = db_query($link,
838 "INSERT INTO ttrss_entries
839 (title,
840 guid,
841 link,
842 updated,
843 content,
844 content_hash,
87764a50 845 cached_content,
2c08214a
AD		 846 no_orig_date,
847 date_updated,
848 date_entered,
849 comments,
850 num_comments,
851 author)
852 VALUES
853 ('$entry_title',
854 '$entry_guid',
855 '$entry_link',
856 '$entry_timestamp_fmt',
857 '$entry_content',
858 '$content_hash',
cb93a5de 859 '$entry_cached_content',
2c08214a
AD		 860 $no_orig_date,
861 NOW(),
862 NOW(),
863 '$entry_comments',
864 '$num_comments',
865 '$entry_author')");
e8291805
AD		 866
867 $article_labels = array();
868
2c08214a
AD		 869 } else {
870 // we keep encountering the entry in feeds, so we need to
871 // update date_updated column so that we don't get horrible
872 // dupes when the entry gets purged and reinserted again e.g.
873 // in the case of SLOW SLOW OMG SLOW updating feeds
874
875 $base_entry_id = db_fetch_result($result, 0, "id");
876
877 db_query($link, "UPDATE ttrss_entries SET date_updated = NOW()
878 WHERE id = '$base_entry_id'");
e8291805
AD		 879
880 $article_labels = get_article_labels($link, $base_entry_id, $owner_uid);
2c08214a
AD		 881 }
882
883 // now it should exist, if not - bad luck then
884
885 $result = db_query($link, "SELECT
886 id,content_hash,no_orig_date,title,
887 ".SUBSTRING_FOR_DATE."(date_updated,1,19) as date_updated,
888 ".SUBSTRING_FOR_DATE."(updated,1,19) as updated,
130b0781 889 num_comments, cached_content
2c08214a
AD		 890 FROM
891 ttrss_entries
892 WHERE guid = '$entry_guid'");
893
894 $entry_ref_id = 0;
895 $entry_int_id = 0;
896
897 if (db_num_rows($result) == 1) {
898
899 if ($debug_enabled) {
900 _debug("update_rss_feed: base guid found, checking for user record");
901 }
902
903 // this will be used below in update handler
904 $orig_content_hash = db_fetch_result($result, 0, "content_hash");
905 $orig_title = db_fetch_result($result, 0, "title");
906 $orig_num_comments = db_fetch_result($result, 0, "num_comments");
130b0781 907 $orig_cached_content = trim(db_fetch_result($result, 0, "cached_content"));
2c08214a
AD		 908 $orig_date_updated = strtotime(db_fetch_result($result,
909 0, "date_updated"));
910
911 $ref_id = db_fetch_result($result, 0, "id");
912 $entry_ref_id = $ref_id;
913
914 // check for user post link to main table
915
916 // do we allow duplicate posts with same GUID in different feeds?
917 if (get_pref($link, "ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
918 $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
919 } else {
920 $dupcheck_qpart = "";
921 }
922
923 /* Collect article tags here so we could filter by them: */
924
925 $article_filters = get_article_filters($filters, $entry_title,
4021d61a 926 $entry_content, $entry_link, $entry_timestamp, $entry_author,
2c08214a
AD		 927 $entry_tags);
928
929 if ($debug_enabled) {
930 _debug("update_rss_feed: article filters: ");
931 if (count($article_filters) != 0) {
932 print_r($article_filters);
933 }
934 }
935
936 if (find_article_filter($article_filters, "filter")) {
937 db_query($link, "COMMIT"); // close transaction in progress
938 continue;
939 }
940
941 $score = calculate_article_score($article_filters);
942
943 if ($debug_enabled) {
944 _debug("update_rss_feed: initial score: $score");
945 }
946
947 $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
948 ref_id = '$ref_id' AND owner_uid = '$owner_uid'
949 $dupcheck_qpart";
950
951// if ($_REQUEST["xdebug"]) print "$query\n";
952
953 $result = db_query($link, $query);
954
955 // okay it doesn't exist - create user entry
956 if (db_num_rows($result) == 0) {
957
958 if ($debug_enabled) {
959 _debug("update_rss_feed: user record not found, creating...");
960 }
961
962 if ($score >= -500 && !find_article_filter($article_filters, 'catchup')) {
963 $unread = 'true';
964 $last_read_qpart = 'NULL';
965 } else {
966 $unread = 'false';
967 $last_read_qpart = 'NOW()';
968 }
969
970 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
971 $marked = 'true';
972 } else {
973 $marked = 'false';
974 }
975
976 if (find_article_filter($article_filters, 'publish')) {
977 $published = 'true';
978 } else {
979 $published = 'false';
980 }
981
2ea9bbfd
AD		 982 // N-grams
983
984 if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
985
986 $result = db_query($link, "SELECT COUNT(*) AS similar FROM
987 ttrss_entries,ttrss_user_entries
988 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
989 AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
990 AND owner_uid = $owner_uid");
991
992 $ngram_similar = db_fetch_result($result, 0, "similar");
993
994 if ($debug_enabled) {
995 _debug("update_rss_feed: N-gram similar results: $ngram_similar");
996 }
997
998 if ($ngram_similar > 0) {
999 $unread = 'false';
1000 }
1001 }
1002
2c08214a
AD		 1003 $result = db_query($link,
1004 "INSERT INTO ttrss_user_entries
1005 (ref_id, owner_uid, feed_id, unread, last_read, marked,
1006 published, score, tag_cache, label_cache, uuid)
1007 VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
1008 $last_read_qpart, $marked, $published, '$score', '', '', '')");
1009
1010 if (PUBSUBHUBBUB_HUB && $published == 'true') {
1011 $rss_link = get_self_url_prefix() .
1012 "/public.php?op=rss&id=-2&key=" .
1013 get_feed_access_key($link, -2, false, $owner_uid);
1014
1015 $p = new Publisher(PUBSUBHUBBUB_HUB);
1016
1017 $pubsub_result = $p->publish_update($rss_link);
1018 }
1019
1020 $result = db_query($link,
1021 "SELECT int_id FROM ttrss_user_entries WHERE
1022 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
1023 feed_id = '$feed' LIMIT 1");
1024
1025 if (db_num_rows($result) == 1) {
1026 $entry_int_id = db_fetch_result($result, 0, "int_id");
1027 }
1028 } else {
1029 if ($debug_enabled) {
1030 _debug("update_rss_feed: user record FOUND");
1031 }
1032
1033 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
1034 $entry_int_id = db_fetch_result($result, 0, "int_id");
1035 }
1036
1037 if ($debug_enabled) {
1038 _debug("update_rss_feed: RID: $entry_ref_id, IID: $entry_int_id");
1039 }
1040
1041 $post_needs_update = false;
1042 $update_insignificant = false;
130b0781 1043 $cached_content_needs_update = false;
2c08214a
AD		 1044
1045 if ($orig_num_comments != $num_comments) {
1046 $post_needs_update = true;
1047 $update_insignificant = true;
1048 }
1049
1050 if ($content_hash != $orig_content_hash) {
1051 $post_needs_update = true;
1052 $update_insignificant = false;
130b0781
AD		 1053 $cached_content_needs_update = true;
1054 }
87764a50 1055
130b0781
AD		 1056 if ($cache_content) {
1057 if ($debug_enabled) {
1058 _debug("update_rss_feed: caching content because original checksum changed...");
1059 }
87764a50 1060
130b0781 1061 $entry_cached_content = cache_content($link, $entry_link, $auth_login, $auth_pass);
87764a50 1062
130b0781 1063 if ($entry_cached_content) {
87764a50
AD		 1064 if ($cache_images && is_writable(CACHE_DIR . '/images'))
1065 $entry_cached_content = cache_images($entry_cached_content, $site_url, $debug_enabled);
1066
1067 $entry_cached_content = db_escape_string($entry_cached_content, false);
130b0781
AD		 1068 $post_needs_update = true;
1069 } else {
1070 $entry_cached_content = db_escape_string($orig_cached_content);
87764a50 1071 }
130b0781
AD		 1072 } else {
1073 $entry_cached_content = db_escape_string($orig_cached_content);
2c08214a
AD		 1074 }
1075
1076 if (db_escape_string($orig_title) != $entry_title) {
1077 $post_needs_update = true;
1078 $update_insignificant = false;
1079 }
1080
1081 // if post needs update, update it and mark all user entries
1082 // linking to this post as updated
1083 if ($post_needs_update) {
1084
1085 if (defined('DAEMON_EXTENDED_DEBUG')) {
1086 _debug("update_rss_feed: post $entry_guid needs update...");
1087 }
1088
1089// print "<!-- post $orig_title needs update : $post_needs_update -->";
1090
1091 db_query($link, "UPDATE ttrss_entries
1092 SET title = '$entry_title', content = '$entry_content',
1093 content_hash = '$content_hash',
87764a50 1094 cached_content = '$entry_cached_content',
2c08214a
AD		 1095 updated = '$entry_timestamp_fmt',
1096 num_comments = '$num_comments'
1097 WHERE id = '$ref_id'");
1098
1099 if (!$update_insignificant) {
1100 if ($mark_unread_on_update) {
1101 db_query($link, "UPDATE ttrss_user_entries
1102 SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
1103 } else if ($update_on_checksum_change) {
1104 db_query($link, "UPDATE ttrss_user_entries
1105 SET last_read = null WHERE ref_id = '$ref_id'
1106 AND unread = false");
1107 }
1108 }
1109 }
1110 }
1111
1112 db_query($link, "COMMIT");
1113
1114 if ($debug_enabled) {
1115 _debug("update_rss_feed: assigning labels...");
1116 }
1117
92c14e9d 1118 assign_article_to_label_filters($link, $entry_ref_id, $article_filters,
b24504b1 1119 $owner_uid, $article_labels);
2c08214a
AD		 1120
1121 if ($debug_enabled) {
1122 _debug("update_rss_feed: looking for enclosures...");
1123 }
1124
1125 // enclosures
1126
1127 $enclosures = array();
1128
1129 if ($use_simplepie) {
1130 $encs = $item->get_enclosures();
1131
1132 if (is_array($encs)) {
1133 foreach ($encs as $e) {
1134 $e_item = array(
1135 $e->link, $e->type, $e->length);
1136
1137 array_push($enclosures, $e_item);
1138 }
1139 }
1140
1141 } else {
1142 // <enclosure>
1143
1144 $e_ctr = $item['enclosure#'];
1145
1146 if ($e_ctr > 0) {
1147 $e_item = array($item['enclosure@url'],
1148 $item['enclosure@type'],
1149 $item['enclosure@length']);
1150
1151 array_push($enclosures, $e_item);
1152
1153 for ($i = 0; $i <= $e_ctr; $i++ ) {
1154
1155 if ($item["enclosure#$i@url"]) {
1156 $e_item = array($item["enclosure#$i@url"],
1157 $item["enclosure#$i@type"],
1158 $item["enclosure#$i@length"]);
1159 array_push($enclosures, $e_item);
1160 }
1161 }
1162 }
1163
1164 // <media:content>
1165 // can there be many of those? yes -fox
1166
1167 $m_ctr = $item['media']['content#'];
1168
1169 if ($m_ctr > 0) {
1170 $e_item = array($item['media']['content@url'],
1171 $item['media']['content@medium'],
1172 $item['media']['content@length']);
1173
1174 array_push($enclosures, $e_item);
1175
1176 for ($i = 0; $i <= $m_ctr; $i++ ) {
1177
1178 if ($item["media"]["content#$i@url"]) {
1179 $e_item = array($item["media"]["content#$i@url"],
1180 $item["media"]["content#$i@medium"],
1181 $item["media"]["content#$i@length"]);
1182 array_push($enclosures, $e_item);
1183 }
1184 }
1185
1186 }
1187 }
1188
1189
1190 if ($debug_enabled) {
1191 _debug("update_rss_feed: article enclosures:");
1192 print_r($enclosures);
1193 }
1194
1195 db_query($link, "BEGIN");
1196
1197 foreach ($enclosures as $enc) {
1198 $enc_url = db_escape_string($enc[0]);
1199 $enc_type = db_escape_string($enc[1]);
1200 $enc_dur = db_escape_string($enc[2]);
1201
1202 $result = db_query($link, "SELECT id FROM ttrss_enclosures
1203 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
1204
1205 if (db_num_rows($result) == 0) {
1206 db_query($link, "INSERT INTO ttrss_enclosures
1207 (content_url, content_type, title, duration, post_id) VALUES
1208 ('$enc_url', '$enc_type', '', '$enc_dur', '$entry_ref_id')");
1209 }
1210 }
1211
1212 db_query($link, "COMMIT");
1213
1214 // check for manual tags (we have to do it here since they're loaded from filters)
1215
1216 foreach ($article_filters as $f) {
6aff7845 1217 if ($f["type"] == "tag") {
2c08214a 1218
6aff7845 1219 $manual_tags = trim_array(explode(",", $f["param"]));
2c08214a
AD		 1220
1221 foreach ($manual_tags as $tag) {
1222 if (tag_is_valid($tag)) {
1223 array_push($entry_tags, $tag);
1224 }
1225 }
1226 }
1227 }
1228
1229 // Skip boring tags
1230
1231 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref($link,
1232 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1233
1234 $filtered_tags = array();
1235 $tags_to_cache = array();
1236
1237 if ($entry_tags && is_array($entry_tags)) {
1238 foreach ($entry_tags as $tag) {
1239 if (array_search($tag, $boring_tags) === false) {
1240 array_push($filtered_tags, $tag);
1241 }
1242 }
1243 }
1244
1245 $filtered_tags = array_unique($filtered_tags);
1246
1247 if ($debug_enabled) {
1248 _debug("update_rss_feed: filtered article tags:");
1249 print_r($filtered_tags);
1250 }
1251
1252 // Save article tags in the database
1253
1254 if (count($filtered_tags) > 0) {
1255
1256 db_query($link, "BEGIN");
1257
1258 foreach ($filtered_tags as $tag) {
1259
1260 $tag = sanitize_tag($tag);
1261 $tag = db_escape_string($tag);
1262
1263 if (!tag_is_valid($tag)) continue;
1264
1265 $result = db_query($link, "SELECT id FROM ttrss_tags
1266 WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1267 owner_uid = '$owner_uid' LIMIT 1");
1268
1269 if ($result && db_num_rows($result) == 0) {
1270
1271 db_query($link, "INSERT INTO ttrss_tags
1272 (owner_uid,tag_name,post_int_id)
1273 VALUES ('$owner_uid','$tag', '$entry_int_id')");
1274 }
1275
1276 array_push($tags_to_cache, $tag);
1277 }
1278
1279 /* update the cache */
1280
1281 $tags_to_cache = array_unique($tags_to_cache);
1282
1283 $tags_str = db_escape_string(join(",", $tags_to_cache));
1284
1285 db_query($link, "UPDATE ttrss_user_entries
1286 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1287 AND owner_uid = $owner_uid");
1288
1289 db_query($link, "COMMIT");
1290 }
1291
b24504b1
AD		 1292 if (get_pref($link, "AUTO_ASSIGN_LABELS", $owner_uid, false)) {
1293 if ($debug_enabled) {
1294 _debug("update_rss_feed: auto-assigning labels...");
1295 }
1296
1297 foreach ($labels as $label) {
1298 $caption = $label["caption"];
1299
3fc6e71a 1300 if (preg_match("/\b$caption\b/i", "$tags_str " . strip_tags($entry_content) . " $entry_title")) {
b24504b1
AD		 1301 if (!labels_contains_caption($article_labels, $caption)) {
1302 label_add_article($link, $entry_ref_id, $caption, $owner_uid);
1303 }
1304 }
1305 }
1306 }
1307
2c08214a
AD		 1308 if ($debug_enabled) {
1309 _debug("update_rss_feed: article processed");
1310 }
1311 }
1312
1313 if (!$last_updated) {
1314 if ($debug_enabled) {
1315 _debug("update_rss_feed: new feed, catching it up...");
1316 }
1317 catchup_feed($link, $feed, false, $owner_uid);
1318 }
1319
1320 if ($debug_enabled) {
1321 _debug("purging feed...");
1322 }
1323
1324 purge_feed($link, $feed, 0, $debug_enabled);
1325
1326 db_query($link, "UPDATE ttrss_feeds
f2798eb6 1327 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
2c08214a
AD		 1328
1329// db_query($link, "COMMIT");
1330
1331 } else {
1332
1333 if ($use_simplepie) {
1334 $error_msg = mb_substr($rss->error(), 0, 250);
1335 } else {
1336 $error_msg = mb_substr(magpie_error(), 0, 250);
1337 }
1338
1339 if ($debug_enabled) {
1340 _debug("update_rss_feed: error fetching feed: $error_msg");
1341 }
1342
1343 $error_msg = db_escape_string($error_msg);
1344
1345 db_query($link,
1346 "UPDATE ttrss_feeds SET last_error = '$error_msg',
1347 last_updated = NOW() WHERE id = '$feed'");
1348 }
1349
1350 if ($use_simplepie) {
1351 unset($rss);
1352 }
1353
1354 if ($debug_enabled) {
1355 _debug("update_rss_feed: done");
1356 }
1357
1358 }
1359
3c696512
AD		 1360 function cache_images($html, $site_url, $debug) {
1361 $cache_dir = CACHE_DIR . "/images";
1362
1363 libxml_use_internal_errors(true);
1364
1365 $charset_hack = '<head>
1366 <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1367 </head>';
1368
1369 $doc = new DOMDocument();
1370 $doc->loadHTML($charset_hack . $html);
1371 $xpath = new DOMXPath($doc);
1372
1373 $entries = $xpath->query('(//img[@src])');
1374
1375 foreach ($entries as $entry) {
1376 if ($entry->hasAttribute('src')) {
1377 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1378
1379 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1380
1381 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1382
1383 if (!file_exists($local_filename)) {
1384 $file_content = fetch_file_contents($src);
1385
b8379e69 1386 if ($file_content && strlen($file_content) > 1024) {
3c696512
AD		 1387 file_put_contents($local_filename, $file_content);
1388 }
1389 }
1390
1391 if (file_exists($local_filename)) {
1392 $entry->setAttribute('src', SELF_URL_PATH . '/image.php?url=' .
487f0750 1393 base64_encode($src));
3c696512
AD		 1394 }
1395 }
1396 }
1397
1398 $node = $doc->getElementsByTagName('body')->item(0);
1399
7b8ff151 1400 return $doc->saveXML($node, LIBXML_NOEMPTYTAG);
3c696512
AD		 1401 }
1402
2a91b6ff
AD		 1403 function expire_lock_files($debug) {
1404 if ($debug) _debug("Removing old lock files...");
1405
1406 $num_deleted = 0;
1407
1408 if (is_writable(LOCK_DIRECTORY)) {
1409 $files = glob(LOCK_DIRECTORY . "/*.lock");
1410
1411 if ($files) {
1412 foreach ($files as $file) {
1413 if (!file_is_locked($file) && time() - filemtime($file) > 86400*2) {
1414 unlink($file);
1415 ++$num_deleted;
1416 }
1417 }
1418 }
1419 }
1420
1421 if ($debug) _debug("Removed $num_deleted files.");
1422 }
1423
3c696512 1424 function expire_cached_files($debug) {
d3158e22 1425 foreach (array("magpie", "simplepie", "images", "export") as $dir) {
3c696512 1426 $cache_dir = CACHE_DIR . "/$dir";
2c08214a 1427
3c696512 1428 if ($debug) _debug("Expiring $cache_dir");
2c08214a 1429
3c696512
AD		 1430 $num_deleted = 0;
1431
1432 if (is_writable($cache_dir)) {
1433 $files = glob("$cache_dir/*");
1434
2a91b6ff 1435 if ($files) {
2ab20c31
AD		 1436 foreach ($files as $file) {
1437 if (time() - filemtime($file) > 86400*7) {
1438 unlink($file);
3c696512 1439
2ab20c31
AD		 1440 ++$num_deleted;
1441 }
3c696512
AD		 1442 }
1443 }
2a91b6ff 1444 }
3c696512
AD		 1445
1446 if ($debug) _debug("Removed $num_deleted files.");
1447 }
1448 }
2c08214a 1449
a3e0bdcf
AD		 1450 /**
1451 * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1452 * Returns the url query as associative array
1453 *
1454 * @param string query
1455 * @return array params
1456 */
1457 function convertUrlQuery($query) {
1458 $queryParts = explode('&', $query);
1459
1460 $params = array();
1461
1462 foreach ($queryParts as $param) {
1463 $item = explode('=', $param);
1464 $params[$item[0]] = $item[1];
1465 }
1466
1467 return $params;
1468 }
92c14e9d
AD		 1469
1470 function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
1471 $matches = array();
1472
1473 foreach ($filters as $filter) {
1474 $match_any_rule = $filter["match_any_rule"];
1475 $filter_match = false;
1476
1477 foreach ($filter["rules"] as $rule) {
1478 $match = false;
1479 $reg_exp = $rule["reg_exp"];
1480
1481 if (!$reg_exp)
1482 continue;
1483
1484 switch ($rule["type"]) {
1485 case "title":
1486 $match = @preg_match("/$reg_exp/i", $title);
1487 break;
1488 case "content":
d03ae73e
AD		 1489 // we don't need to deal with multiline regexps
1490 $content = preg_replace("/[\r\n\t]/", "", $content);
1491
92c14e9d
AD		 1492 $match = @preg_match("/$reg_exp/i", $content);
1493 break;
1494 case "both":
d03ae73e
AD		 1495 // we don't need to deal with multiline regexps
1496 $content = preg_replace("/[\r\n\t]/", "", $content);
1497
72d1d067 1498 $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
92c14e9d
AD		 1499 break;
1500 case "link":
1501 $match = @preg_match("/$reg_exp/i", $link);
1502 break;
1503 case "author":
1504 $match = @preg_match("/$reg_exp/i", $author);
1505 break;
1506 case "tag":
1507 $tag_string = join(",", $tags);
1508 $match = @preg_match("/$reg_exp/i", $tag_string);
1509 break;
1510 }
1511
1512 if ($match_any_rule) {
1513 if ($match) {
1514 $filter_match = true;
1515 break;
1516 }
1517 } else {
1518 $filter_match = $match;
1519 if (!$match) {
1520 break;
1521 }
1522 }
1523 }
1524
1525 if ($filter_match) {
1526 foreach ($filter["actions"] AS $action) {
1527 array_push($matches, $action);
1528 }
1529 }
1530 }
1531
1532 return $matches;
1533 }
1534
1535 function find_article_filter($filters, $filter_name) {
1536 foreach ($filters as $f) {
1537 if ($f["type"] == $filter_name) {
1538 return $f;
1539 };
1540 }
1541 return false;
1542 }
1543
1544 function find_article_filters($filters, $filter_name) {
1545 $results = array();
1546
1547 foreach ($filters as $f) {
1548 if ($f["type"] == $filter_name) {
1549 array_push($results, $f);
1550 };
1551 }
1552 return $results;
1553 }
1554
1555 function calculate_article_score($filters) {
1556 $score = 0;
1557
1558 foreach ($filters as $f) {
1559 if ($f["type"] == "score") {
1560 $score += $f["param"];
1561 };
1562 }
1563 return $score;
1564 }
1565
b24504b1
AD		 1566 function labels_contains_caption($labels, $caption) {
1567 foreach ($labels as $label) {
1568 if ($label[1] == $caption) {
1569 return true;
1570 }
1571 }
1572
1573 return false;
1574 }
1575
1576 function assign_article_to_label_filters($link, $id, $filters, $owner_uid, $article_labels) {
92c14e9d
AD		 1577 foreach ($filters as $f) {
1578 if ($f["type"] == "label") {
b24504b1
AD		 1579 if (!labels_contains_caption($article_labels, $f["param"])) {
1580 label_add_article($link, $id, $f["param"], $owner_uid);
1581 }
1582 }
92c14e9d
AD		 1583 }
1584 }
87764a50
AD		 1585
1586 function cache_content($link, $url, $login, $pass) {
1587
1588 $content = fetch_file_contents($url, $login, $pass);
1589
1590 if ($content) {
1591 $doc = new DOMDocument();
1592 @$doc->loadHTML($content);
1593 $xpath = new DOMXPath($doc);
1594
1595 $node = $doc->getElementsByTagName('body')->item(0);
1596
1597 if ($node) {
1598 $content = $doc->saveXML($node, LIBXML_NOEMPTYTAG);
1599
1600 return $content;
1601 }
1602 }
1603
1604 return "";
1605 }
2c08214a 1606?>
web-based news feed (RSS/Atom) reader and aggregator; see https://tt-rss.org/