]> git.wh0rd.org - tt-rss.git/blame - include/rssfuncs.php
daemon common: use proper update condition checking for secondary-selected feeds
[tt-rss.git] / include / rssfuncs.php
CommitLineData
2c08214a 1<?php
fce451a4
AD
2 define_default('DAEMON_UPDATE_LOGIN_LIMIT', 30);
3 define_default('DAEMON_FEED_LIMIT', 500);
e2cf81e2 4 define_default('DAEMON_SLEEP_INTERVAL', 120);
a3d50184 5 define_default('_MIN_CACHE_IMAGE_SIZE', 1024);
09e8bdfd 6
eb16bd9f 7 function calculate_article_hash($article, $pluginhost) {
af244f92
AD
8 $tmp = "";
9
10 foreach ($article as $k => $v) {
11 if ($k != "feed" && isset($v)) {
24e6ff5d
AD
12 $x = strip_tags(is_array($v) ? implode(",", $v) : $v);
13
14 //_debug("$k:" . sha1($x) . ":" . htmlspecialchars($x), true);
15
16 $tmp .= sha1("$k:" . sha1($x));
af244f92
AD
17 }
18 }
19
eb16bd9f 20 return sha1(implode(",", $pluginhost->get_plugin_names()) . $tmp);
b1840673
AD
21 }
22
6322ac79 23 function update_feedbrowser_cache() {
79178062 24
a42c55f0 25 $result = db_query("SELECT feed_url, site_url, title, COUNT(id) AS subscribers
79178062
AD
26 FROM ttrss_feeds WHERE (SELECT COUNT(id) = 0 FROM ttrss_feeds AS tf
27 WHERE tf.feed_url = ttrss_feeds.feed_url
28 AND (private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%'))
29 GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
30
a42c55f0 31 db_query("BEGIN");
79178062 32
a42c55f0 33 db_query("DELETE FROM ttrss_feedbrowser_cache");
79178062
AD
34
35 $count = 0;
36
37 while ($line = db_fetch_assoc($result)) {
a42c55f0
AD
38 $subscribers = db_escape_string($line["subscribers"]);
39 $feed_url = db_escape_string($line["feed_url"]);
40 $title = db_escape_string($line["title"]);
41 $site_url = db_escape_string($line["site_url"]);
79178062 42
a42c55f0 43 $tmp_result = db_query("SELECT subscribers FROM
79178062
AD
44 ttrss_feedbrowser_cache WHERE feed_url = '$feed_url'");
45
46 if (db_num_rows($tmp_result) == 0) {
47
a42c55f0 48 db_query("INSERT INTO ttrss_feedbrowser_cache
79178062
AD
49 (feed_url, site_url, title, subscribers) VALUES ('$feed_url',
50 '$site_url', '$title', '$subscribers')");
51
52 ++$count;
53
54 }
55
56 }
57
a42c55f0 58 db_query("COMMIT");
79178062
AD
59
60 return $count;
61
62 }
63
64
2c08214a
AD
65 /**
66 * Update a feed batch.
67 * Used by daemons to update n feeds by run.
68 * Only update feed needing a update, and not being processed
69 * by another process.
70 *
71 * @param mixed $link Database link
72 * @param integer $limit Maximum number of feeds in update batch. Default to DAEMON_FEED_LIMIT.
73 * @param boolean $from_http Set to true if you call this function from http to disable cli specific code.
74 * @param boolean $debug Set to false to disable debug output. Default to true.
75 * @return void
76 */
a42c55f0 77 function update_daemon_common($limit = DAEMON_FEED_LIMIT, $from_http = false, $debug = true) {
2c08214a
AD
78 // Process all other feeds using last_updated and interval parameters
79
6322ac79 80 $schema_version = get_schema_version();
857efe49
AD
81
82 if ($schema_version != SCHEMA_VERSION) {
83 die("Schema version is wrong, please upgrade the database.\n");
84 }
85
61c1812f
AD
86 define('PREFS_NO_CACHE', true);
87
2c08214a 88 // Test if the user has loggued in recently. If not, it does not update its feeds.
09e8bdfd 89 if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
2c08214a
AD
90 if (DB_TYPE == "pgsql") {
91 $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
92 } else {
93 $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
94 }
95 } else {
96 $login_thresh_qpart = "";
97 }
98
537eca40 99 // Test if the feed need a update (update interval exceeded).
2c08214a
AD
100 if (DB_TYPE == "pgsql") {
101 $update_limit_qpart = "AND ((
102 ttrss_feeds.update_interval = 0
ee0542ce 103 AND ttrss_user_prefs.value != '-1'
2c08214a
AD
104 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
105 ) OR (
106 ttrss_feeds.update_interval > 0
107 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
f08426e3
AD
108 ) OR (ttrss_feeds.last_updated IS NULL
109 AND ttrss_user_prefs.value != '-1')
110 OR (last_updated = '1970-01-01 00:00:00'
111 AND ttrss_user_prefs.value != '-1'))";
2c08214a
AD
112 } else {
113 $update_limit_qpart = "AND ((
114 ttrss_feeds.update_interval = 0
ee0542ce 115 AND ttrss_user_prefs.value != '-1'
2c08214a
AD
116 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
117 ) OR (
118 ttrss_feeds.update_interval > 0
119 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
f08426e3
AD
120 ) OR (ttrss_feeds.last_updated IS NULL
121 AND ttrss_user_prefs.value != '-1')
122 OR (last_updated = '1970-01-01 00:00:00'
123 AND ttrss_user_prefs.value != '-1'))";
2c08214a
AD
124 }
125
126 // Test if feed is currently being updated by another process.
127 if (DB_TYPE == "pgsql") {
566417c4 128 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
2c08214a 129 } else {
566417c4 130 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 10 MINUTE))";
2c08214a
AD
131 }
132
133 // Test if there is a limit to number of updated feeds
134 $query_limit = "";
135 if($limit) $query_limit = sprintf("LIMIT %d", $limit);
136
98070db0
TK
137 // Update the least recently updated feeds first
138 $query_order = "ORDER BY last_updated";
139 if (DB_TYPE == "pgsql") $query_order .= " NULLS FIRST";
140
fce451a4 141 $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
2c08214a
AD
142 FROM
143 ttrss_feeds, ttrss_users, ttrss_user_prefs
f4ae0f05 144 WHERE
2c08214a 145 ttrss_feeds.owner_uid = ttrss_users.id
f08426e3 146 AND ttrss_user_prefs.profile IS NULL
2c08214a
AD
147 AND ttrss_users.id = ttrss_user_prefs.owner_uid
148 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
149 $login_thresh_qpart $update_limit_qpart
1c4421fc 150 $updstart_thresh_qpart
98070db0 151 $query_order $query_limit";
fce451a4
AD
152
153 // We search for feed needing update.
154 $result = db_query($query);
2c08214a 155
ca5ff2d9 156 if($debug) _debug(sprintf("Scheduled %d feeds to update...", db_num_rows($result)));
2c08214a
AD
157
158 // Here is a little cache magic in order to minimize risk of double feed updates.
159 $feeds_to_update = array();
160 while ($line = db_fetch_assoc($result)) {
a42c55f0 161 array_push($feeds_to_update, db_escape_string($line['feed_url']));
2c08214a
AD
162 }
163
164 // We update the feed last update started date before anything else.
165 // There is no lag due to feed contents downloads
166 // It prevent an other process to update the same feed.
1c4421fc
AD
167
168 if(count($feeds_to_update) > 0) {
169 $feeds_quoted = array();
170
171 foreach ($feeds_to_update as $feed) {
a42c55f0 172 array_push($feeds_quoted, "'" . db_escape_string($feed) . "'");
1c4421fc
AD
173 }
174
a42c55f0 175 db_query(sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
1c4421fc 176 WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
2c08214a
AD
177 }
178
8292d05b 179 $nf = 0;
2d9c5684 180 $bstarted = microtime(true);
8292d05b 181
2c08214a 182 // For each feed, we call the feed update function.
1c4421fc
AD
183 foreach ($feeds_to_update as $feed) {
184 if($debug) _debug("Base feed: $feed");
2c08214a 185
a42c55f0 186 //update_rss_feed($line["id"], true);
2c08214a 187
1c4421fc
AD
188 // since we have the data cached, we can deal with other feeds with the same url
189
15c762be 190 $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
ee0542ce
AD
191 FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
192 ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
193 ttrss_users.id = ttrss_user_prefs.owner_uid AND
194 ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
f08426e3 195 ttrss_user_prefs.profile IS NULL AND
9e84bab4
AD
196 feed_url = '".db_escape_string($feed)."'
197 $update_limit_qpart
1c4421fc 198 $login_thresh_qpart
5929a0c1 199 ORDER BY ttrss_feeds.id $query_limit");
1c4421fc
AD
200
201 if (db_num_rows($tmp_result) > 0) {
88edaa93
AD
202 $rss = false;
203
1c4421fc 204 while ($tline = db_fetch_assoc($tmp_result)) {
15c762be 205 if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
f08426e3 206
2d9c5684 207 $fstarted = microtime(true);
f77c7203 208 $rss = update_rss_feed($tline["id"], true, false);
4f71d743 209 _debug_suppress(false);
2d9c5684
AD
210
211 _debug(sprintf(" %.4f (sec)", microtime(true) - $fstarted));
212
8292d05b 213 ++$nf;
1c4421fc
AD
214 }
215 }
2c08214a
AD
216 }
217
2d9c5684
AD
218 if ($nf > 0) {
219 _debug(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
220 microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
221 }
222
f1611683
AD
223 require_once "digest.php";
224
2c08214a 225 // Send feed digests by email if needed.
a42c55f0 226 send_headlines_digests($debug);
2c08214a 227
8292d05b
AD
228 return $nf;
229
2c08214a
AD
230 } // function update_daemon_common
231
6022776d
AD
232 // this is used when subscribing
233 function set_basic_feed_info($feed) {
234
235 $feed = db_escape_string($feed);
236
9bd422c2 237 $result = db_query("SELECT feed_url,auth_pass,auth_login,auth_pass_encrypted
6022776d
AD
238 FROM ttrss_feeds WHERE id = '$feed'");
239
240 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
241 0, "auth_pass_encrypted"));
242
243 $auth_login = db_fetch_result($result, 0, "auth_login");
244 $auth_pass = db_fetch_result($result, 0, "auth_pass");
245
246 if ($auth_pass_encrypted) {
247 require_once "crypt.php";
248 $auth_pass = decrypt_string($auth_pass);
249 }
250
251 $fetch_url = db_fetch_result($result, 0, "feed_url");
252
253 $feed_data = fetch_file_contents($fetch_url, false,
254 $auth_login, $auth_pass, false,
255 FEED_FETCH_TIMEOUT_TIMEOUT,
256 0);
257
258 global $fetch_curl_used;
259
260 if (!$fetch_curl_used) {
261 $tmp = @gzdecode($feed_data);
262
263 if ($tmp) $feed_data = $tmp;
264 }
265
266 $feed_data = trim($feed_data);
267
268 $rss = new FeedParser($feed_data);
269 $rss->init();
270
271 if (!$rss->error()) {
272
273 $result = db_query("SELECT title, site_url FROM ttrss_feeds WHERE id = '$feed'");
274
275 $registered_title = db_fetch_result($result, 0, "title");
276 $orig_site_url = db_fetch_result($result, 0, "site_url");
277
278 $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
279 $feed_title = db_escape_string(mb_substr($rss->get_title(), 0, 199));
280
281 if ($feed_title && (!$registered_title || $registered_title == "[Unknown]")) {
282 db_query("UPDATE ttrss_feeds SET
283 title = '$feed_title' WHERE id = '$feed'");
284 }
285
286 if ($site_url && $orig_site_url != $site_url) {
287 db_query("UPDATE ttrss_feeds SET
288 site_url = '$site_url' WHERE id = '$feed'");
289 }
290 }
291 }
292
87764a50 293 // ignore_daemon is not used
88edaa93 294 function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
2c08214a 295
2c08214a
AD
296 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
297
4f71d743 298 _debug_suppress(!$debug_enabled);
68cccafc 299 _debug("start", $debug_enabled);
2c08214a 300
6bb96beb
AD
301 $result = db_query("SELECT title FROM ttrss_feeds
302 WHERE id = '$feed'");
303 $title = db_fetch_result($result, 0, "title");
304
305 // feed was batch-subscribed or something, we need to get basic info
306 // this is not optimal currently as it fetches stuff separately TODO: optimize
307 if ($title == "[Unknown]") {
308 _debug("setting basic feed info for $feed...");
309 set_basic_feed_info($feed);
310 }
311
a42c55f0 312 $result = db_query("SELECT id,update_interval,auth_login,
5ba1ddd4 313 feed_url,auth_pass,cache_images,
5321e775 314 mark_unread_on_update, owner_uid,
90e5f4f1 315 pubsub_state, auth_pass_encrypted,
0edf1d0d 316 feed_language
87764a50 317 FROM ttrss_feeds WHERE id = '$feed'");
2c08214a
AD
318
319 if (db_num_rows($result) == 0) {
68cccafc 320 _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
2c08214a
AD
321 return false;
322 }
323
2c08214a
AD
324 $owner_uid = db_fetch_result($result, 0, "owner_uid");
325 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
326 0, "mark_unread_on_update"));
2c08214a 327 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
044cff2d
AD
328 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
329 0, "auth_pass_encrypted"));
2c08214a 330
a42c55f0 331 db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
2c08214a
AD
332 WHERE id = '$feed'");
333
334 $auth_login = db_fetch_result($result, 0, "auth_login");
335 $auth_pass = db_fetch_result($result, 0, "auth_pass");
336
044cff2d
AD
337 if ($auth_pass_encrypted) {
338 require_once "crypt.php";
339 $auth_pass = decrypt_string($auth_pass);
340 }
341
2c08214a
AD
342 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
343 $fetch_url = db_fetch_result($result, 0, "feed_url");
df659891
AD
344 $feed_language = db_escape_string(mb_strtolower(db_fetch_result($result, 0, "feed_language")));
345 if (!$feed_language) $feed_language = 'english';
2c08214a 346
a42c55f0 347 $feed = db_escape_string($feed);
2c08214a 348
f074ffe9 349 $date_feed_processed = date('Y-m-d H:i');
2c08214a 350
865a3ed6 351 $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
f074ffe9 352
ee65bef4
AD
353 $pluginhost = new PluginHost();
354 $pluginhost->set_debug($debug_enabled);
355 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
356
357 $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
358 $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
359 $pluginhost->load_data();
360
88edaa93
AD
361 if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
362 _debug("using previously initialized parser object");
363 } else {
364 $rss_hash = false;
4f9cbdff 365
88edaa93 366 $force_refetch = isset($_REQUEST["force_refetch"]);
0edf1d0d 367 $feed_data = "";
687a4f59 368
95beaa14 369 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
0edf1d0d 370 $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, 0, $auth_login, $auth_pass);
95beaa14
RA
371 }
372
373 // try cache
374 if (!$feed_data &&
375 file_exists($cache_filename) &&
88edaa93
AD
376 is_readable($cache_filename) &&
377 !$auth_login && !$auth_pass &&
378 filemtime($cache_filename) > time() - 30) {
2c08214a 379
15d0528a 380 _debug("using local cache [$cache_filename].", $debug_enabled);
be574731 381
88edaa93 382 @$feed_data = file_get_contents($cache_filename);
52637d3b 383
88edaa93
AD
384 if ($feed_data) {
385 $rss_hash = sha1($feed_data);
386 }
f074ffe9 387
88edaa93
AD
388 } else {
389 _debug("local cache will not be used for this feed", $debug_enabled);
390 }
ee65bef4 391
95beaa14 392 // fetch feed from source
fd687300
AD
393 if (!$feed_data) {
394 _debug("fetching [$fetch_url]...", $debug_enabled);
312742db
AD
395
396 if (ini_get("open_basedir") && function_exists("curl_init")) {
397 _debug("not using CURL due to open_basedir restrictions");
398 }
399
fd687300
AD
400 $feed_data = fetch_file_contents($fetch_url, false,
401 $auth_login, $auth_pass, false,
402 $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
0edf1d0d 403 0);
3f6f0857 404
fd687300 405 global $fetch_curl_used;
3f6f0857 406
fd687300
AD
407 if (!$fetch_curl_used) {
408 $tmp = @gzdecode($feed_data);
3f6f0857 409
fd687300
AD
410 if ($tmp) $feed_data = $tmp;
411 }
1367bc3f 412
fd687300 413 $feed_data = trim($feed_data);
017401dd 414
fd687300
AD
415 _debug("fetch done.", $debug_enabled);
416
95beaa14
RA
417 // cache vanilla feed data for re-use
418 if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
419 $new_rss_hash = sha1($feed_data);
420
421 if ($new_rss_hash != $rss_hash) {
422 _debug("saving $cache_filename", $debug_enabled);
423 @file_put_contents($cache_filename, $feed_data);
424 }
425 }
4f9cbdff 426 }
017401dd 427
fd687300
AD
428 if (!$feed_data) {
429 global $fetch_last_error;
430 global $fetch_last_error_code;
f074ffe9 431
fd687300 432 _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
f074ffe9 433
fd687300 434 $error_escaped = '';
7a01dc77 435
fd687300
AD
436 // If-Modified-Since
437 if ($fetch_last_error_code != 304) {
438 $error_escaped = db_escape_string($fetch_last_error);
439 } else {
440 _debug("source claims data not modified, nothing to do.", $debug_enabled);
441 }
4f9cbdff 442
fd687300
AD
443 db_query(
444 "UPDATE ttrss_feeds SET last_error = '$error_escaped',
445 last_updated = NOW() WHERE id = '$feed'");
4f9cbdff 446
fd687300
AD
447 return;
448 }
f074ffe9
AD
449 }
450
1ffe3391 451 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
6791af0c 452 $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
017401dd
AD
453 }
454
5ef84097
AD
455 // set last update to now so if anything *simplepie* crashes later we won't be
456 // continuously failing on the same feed
5de40104 457 //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
5ef84097 458
4f9cbdff 459 if (!$rss) {
cd07592c
AD
460 $rss = new FeedParser($feed_data);
461 $rss->init();
5de51df7
AD
462 }
463
2c08214a
AD
464// print_r($rss);
465
a42c55f0 466 $feed = db_escape_string($feed);
2c08214a 467
19b3992b 468 if (!$rss->error()) {
2c08214a 469
d2a421e3 470 // We use local pluginhost here because we need to load different per-user feed plugins
1ffe3391 471 $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
4412b877 472
df659891 473 _debug("language: $feed_language", $debug_enabled);
68cccafc 474 _debug("processing feed data...", $debug_enabled);
2c08214a 475
a42c55f0 476// db_query("BEGIN");
2c08214a 477
382268c6
AD
478 if (DB_TYPE == "pgsql") {
479 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
480 } else {
481 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
482 }
483
fe4535e6 484 $result = db_query("SELECT owner_uid,favicon_avg_color,
382268c6
AD
485 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
486 favicon_needs_check
2c08214a
AD
487 FROM ttrss_feeds WHERE id = '$feed'");
488
382268c6
AD
489 $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
490 "favicon_needs_check"));
36490f11 491 $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
2c08214a
AD
492
493 $owner_uid = db_fetch_result($result, 0, "owner_uid");
494
a42c55f0 495 $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
2c08214a 496
cd07592c
AD
497 _debug("site_url: $site_url", $debug_enabled);
498 _debug("feed_title: " . $rss->get_title(), $debug_enabled);
499
687a4f59 500 if ($favicon_needs_check || $force_refetch) {
36490f11
AD
501
502 /* terrible hack: if we crash on floicon shit here, we won't check
560cbd8c 503 * the icon avgcolor again (unless the icon got updated) */
36490f11 504
560cbd8c
AD
505 $favicon_file = ICONS_DIR . "/$feed.ico";
506 $favicon_modified = @filemtime($favicon_file);
507
68cccafc 508 _debug("checking favicon...", $debug_enabled);
687a4f59 509
6f7798b6 510 check_feed_favicon($site_url, $feed);
560cbd8c
AD
511 $favicon_modified_new = @filemtime($favicon_file);
512
513 if ($favicon_modified_new > $favicon_modified)
514 $favicon_avg_color = '';
687a4f59 515
6ee0d4b0 516 if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
bc7a144d 517 require_once "colors.php";
687a4f59 518
aafd55ba
AD
519 db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
520 id = '$feed'");
521
6322ac79 522 $favicon_color = db_escape_string(
bc7a144d 523 calculate_avg_color($favicon_file));
63c323f7 524
bc7a144d 525 $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
36490f11 526 } else if ($favicon_avg_color == 'fail') {
84ceb6bd 527 _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
6ac722d5 528 }
687a4f59 529
a42c55f0 530 db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
bc7a144d 531 $favicon_colorstring
f2798eb6
AD
532 WHERE id = '$feed'");
533 }
2c08214a 534
68cccafc 535 _debug("loading filters & labels...", $debug_enabled);
2c08214a 536
a42c55f0 537 $filters = load_filters($feed, $owner_uid);
2c08214a 538
68cccafc 539 _debug("" . count($filters) . " filters loaded.", $debug_enabled);
2c08214a 540
19b3992b 541 $items = $rss->get_items();
2c08214a 542
19b3992b 543 if (!is_array($items)) {
68cccafc 544 _debug("no articles found.", $debug_enabled);
2c08214a 545
a42c55f0 546 db_query("UPDATE ttrss_feeds
2c08214a
AD
547 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
548
549 return; // no articles
550 }
551
552 if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
553
68cccafc 554 _debug("checking for PUSH hub...", $debug_enabled);
2c08214a
AD
555
556 $feed_hub_url = false;
2c08214a 557
19b3992b 558 $links = $rss->get_links('hub');
2c08214a 559
19b3992b
AD
560 if ($links && is_array($links)) {
561 foreach ($links as $l) {
562 $feed_hub_url = $l;
563 break;
2c08214a
AD
564 }
565 }
566
68cccafc 567 _debug("feed hub url: $feed_hub_url", $debug_enabled);
2c08214a 568
d8179cb9
AD
569 $feed_self_url = $fetch_url;
570
571 $links = $rss->get_links('self');
572
573 if ($links && is_array($links)) {
574 foreach ($links as $l) {
575 $feed_self_url = $l;
576 break;
577 }
578 }
579
580 _debug("feed self url = $feed_self_url");
581
582 if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
2c08214a
AD
583 !ini_get("open_basedir")) {
584
585 require_once 'lib/pubsubhubbub/subscriber.php';
586
587 $callback_url = get_self_url_prefix() .
588 "/public.php?op=pubsub&id=$feed";
589
590 $s = new Subscriber($feed_hub_url, $callback_url);
591
d8179cb9 592 $rc = $s->subscribe($feed_self_url);
2c08214a 593
5ba1ddd4 594 _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
2c08214a 595
a42c55f0 596 db_query("UPDATE ttrss_feeds SET pubsub_state = 1
2c08214a
AD
597 WHERE id = '$feed'");
598 }
599 }
600
68cccafc 601 _debug("processing articles...", $debug_enabled);
2c08214a 602
6c9f3d4a
AD
603 $tstart = time();
604
19b3992b 605 foreach ($items as $item) {
5d56d100 606 if ($_REQUEST['xdebug'] == 3) {
2c08214a
AD
607 print_r($item);
608 }
609
6c9f3d4a
AD
610 if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
611 _debug("looks like there's too many articles to process at once, breaking out", $debug_enabled);
612 break;
613 }
614
19b3992b
AD
615 $entry_guid = $item->get_id();
616 if (!$entry_guid) $entry_guid = $item->get_link();
617 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
2c08214a
AD
618 if (!$entry_guid) continue;
619
3a4c8973
AD
620 $entry_guid = "$owner_uid,$entry_guid";
621
a42c55f0 622 $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
5e3d5480 623
68cccafc 624 _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
5e3d5480 625
2c08214a
AD
626 $entry_timestamp = "";
627
04d2f9c8
AD
628 $entry_timestamp = $item->get_date();
629
630 _debug("orig date: " . $item->get_date(), $debug_enabled);
2c08214a 631
30123fe6 632 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
2c08214a 633 $entry_timestamp = time();
2c08214a
AD
634 }
635
636 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
637
68cccafc 638 _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
2c08214a 639
99429e57
AD
640// $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
641// $entry_title = decode_numeric_entities($entry_title);
642 $entry_title = $item->get_title();
1b35d30c 643
5d56d100 644 $entry_link = rewrite_relative_url($site_url, $item->get_link());
2c08214a 645
68cccafc
AD
646 _debug("title $entry_title", $debug_enabled);
647 _debug("link $entry_link", $debug_enabled);
2c08214a
AD
648
649 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
650
19b3992b
AD
651 $entry_content = $item->get_content();
652 if (!$entry_content) $entry_content = $item->get_description();
2c08214a
AD
653
654 if ($_REQUEST["xdebug"] == 2) {
9ec10352 655 print "content: ";
0bc503ff 656 print htmlspecialchars($entry_content);
3c696512 657 print "\n";
2c08214a
AD
658 }
659
cd07592c
AD
660 $entry_comments = $item->get_comments_url();
661 $entry_author = $item->get_author();
2c08214a 662
a42c55f0 663 $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
2c08214a 664
ce2b219c
AD
665 $entry_comments = db_escape_string(mb_substr(trim($entry_comments), 0, 245));
666 $entry_author = db_escape_string(mb_substr(trim($entry_author), 0, 245));
2c08214a 667
d4992d6b 668 $num_comments = (int) $item->get_comments_count();
2c08214a 669
68cccafc
AD
670 _debug("author $entry_author", $debug_enabled);
671 _debug("num_comments: $num_comments", $debug_enabled);
ee78f81c 672 _debug("looking for tags...", $debug_enabled);
2c08214a
AD
673
674 // parse <category> entries into tags
675
676 $additional_tags = array();
677
19b3992b 678 $additional_tags_src = $item->get_categories();
2c08214a 679
19b3992b
AD
680 if (is_array($additional_tags_src)) {
681 foreach ($additional_tags_src as $tobj) {
cd07592c 682 array_push($additional_tags, $tobj);
2c08214a 683 }
19b3992b 684 }
2c08214a 685
fa6fbd36 686 $entry_tags = array_unique($additional_tags);
2c08214a
AD
687
688 for ($i = 0; $i < count($entry_tags); $i++)
689 $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
690
ee78f81c
AD
691 _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
692
68cccafc 693 _debug("done collecting data.", $debug_enabled);
2c08214a 694
3318d324 695 $result = db_query("SELECT id, content_hash, lang FROM ttrss_entries
b1840673 696 WHERE guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed'");
b30abdad
AD
697
698 if (db_num_rows($result) != 0) {
b1840673
AD
699 $base_entry_id = db_fetch_result($result, 0, "id");
700 $entry_stored_hash = db_fetch_result($result, 0, "content_hash");
a29fe121 701 $article_labels = get_article_labels($base_entry_id, $owner_uid);
3318d324 702 $entry_language = db_fetch_result($result, 0, "lang");
a8ac7661 703
040aa540 704 $existing_tags = get_article_tags($base_entry_id, $owner_uid);
a8ac7661
AD
705 $entry_tags = array_unique(array_merge($entry_tags, $existing_tags));
706
b30abdad 707 } else {
b1840673
AD
708 $base_entry_id = false;
709 $entry_stored_hash = "";
a29fe121 710 $article_labels = array();
3318d324 711 $entry_language = "";
b30abdad
AD
712 }
713
455b1401 714 $article = array("owner_uid" => $owner_uid, // read only
b30abdad 715 "guid" => $entry_guid, // read only
59e83455 716 "guid_hashed" => $entry_guid_hashed, // read only
19b3992b
AD
717 "title" => $entry_title,
718 "content" => $entry_content,
719 "link" => $entry_link,
a29fe121 720 "labels" => $article_labels, // current limitation: can add labels to article, can't remove them
19b3992b 721 "tags" => $entry_tags,
e02555c1 722 "author" => $entry_author,
c9299c28 723 "force_catchup" => false, // ugly hack for the time being
6de3a1be 724 "score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
3318d324 725 "language" => $entry_language,
f73e03e0
AD
726 "feed" => array("id" => $feed,
727 "fetch_url" => $fetch_url,
728 "site_url" => $site_url)
729 );
cc85704f 730
b1840673 731 $entry_plugin_data = "";
eb16bd9f 732 $entry_current_hash = calculate_article_hash($article, $pluginhost);
b1840673
AD
733
734 _debug("article hash: $entry_current_hash [stored=$entry_stored_hash]", $debug_enabled);
735
522e8b35 736 if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
b1840673
AD
737 _debug("stored article seems up to date [IID: $base_entry_id], updating timestamp only", $debug_enabled);
738
739 // we keep encountering the entry in feeds, so we need to
740 // update date_updated column so that we don't get horrible
741 // dupes when the entry gets purged and reinserted again e.g.
742 // in the case of SLOW SLOW OMG SLOW updating feeds
743
744 $base_entry_id = db_fetch_result($result, 0, "id");
745
746 db_query("UPDATE ttrss_entries SET date_updated = NOW()
747 WHERE id = '$base_entry_id'");
748
5bdcb8fd 749 continue;
b1840673
AD
750 }
751
752 _debug("hash differs, applying plugin filters:", $debug_enabled);
753
1ffe3391 754 foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
b1840673
AD
755 _debug("... " . get_class($plugin), $debug_enabled);
756
757 $start = microtime(true);
19b3992b 758 $article = $plugin->hook_article_filter($article);
0084f0d1 759
b1840673
AD
760 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
761
762 $entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
cc85704f
AD
763 }
764
0bc503ff
AD
765 if ($_REQUEST["xdebug"] == 2) {
766 print "processed content: ";
767 print htmlspecialchars($article["content"]);
768 print "\n";
769 }
770
b1840673
AD
771 $entry_plugin_data = db_escape_string($entry_plugin_data);
772
773 _debug("plugin data: $entry_plugin_data", $debug_enabled);
774
35c12dc4
AD
775 // Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
776 if (DB_TYPE == "mysql") {
777 foreach ($article as $k => $v) {
35c37354
AD
778
779 // i guess we'll have to take the risk of 4byte unicode labels & tags here
780 if (!is_array($article[$k])) {
781 $article[$k] = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $v);
782 }
35c12dc4
AD
783 }
784 }
785
b8774453
AD
786 /* Collect article tags here so we could filter by them: */
787
788 $article_filters = get_article_filters($filters, $article["title"],
789 $article["content"], $article["link"], 0, $article["author"],
790 $article["tags"]);
791
792 if ($debug_enabled) {
793 _debug("article filters: ", $debug_enabled);
794 if (count($article_filters) != 0) {
795 print_r($article_filters);
796 }
797 }
798
799 $plugin_filter_names = find_article_filters($article_filters, "plugin");
800 $plugin_filter_actions = $pluginhost->get_filter_actions();
801
802 if (count($plugin_filter_names) > 0) {
803 _debug("applying plugin filter actions...", $debug_enabled);
804
805 foreach ($plugin_filter_names as $pfn) {
806 list($pfclass,$pfaction) = explode(":", $pfn["param"]);
807
808 if (isset($plugin_filter_actions[$pfclass])) {
809 $plugin = $pluginhost->get_plugin($pfclass);
810
811 _debug("... $pfclass: $pfaction", $debug_enabled);
812
813 if ($plugin) {
814 $start = microtime(true);
815 $article = $plugin->hook_article_filter_action($article, $pfaction);
816
817 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
818 } else {
819 _debug("??? $pfclass: plugin object not found.");
820 }
821 } else {
822 _debug("??? $pfclass: filter plugin not registered.");
823 }
824 }
825 }
826
19b3992b 827 $entry_tags = $article["tags"];
a42c55f0
AD
828 $entry_guid = db_escape_string($entry_guid);
829 $entry_title = db_escape_string($article["title"]);
830 $entry_author = db_escape_string($article["author"]);
831 $entry_link = db_escape_string($article["link"]);
f935d98e 832 $entry_content = $article["content"]; // escaped below
c9299c28 833 $entry_force_catchup = $article["force_catchup"];
a29fe121 834 $article_labels = $article["labels"];
6de3a1be 835 $entry_score_modifier = (int) $article["score_modifier"];
3318d324 836 $entry_language = db_escape_string($article["language"]);
a29fe121
AD
837
838 if ($debug_enabled) {
839 _debug("article labels:", $debug_enabled);
840 print_r($article_labels);
841 }
c9299c28
AD
842
843 _debug("force catchup: $entry_force_catchup");
f935d98e 844
0a3fd79b 845 if ($cache_images && is_writable(CACHE_DIR . '/images'))
f0bd8e65 846 cache_images($entry_content, $site_url, $debug_enabled);
0a3fd79b 847
a42c55f0 848 $entry_content = db_escape_string($entry_content, false);
1f45c857 849
07877caf 850 //db_query("BEGIN");
2c08214a 851
a42c55f0 852 $result = db_query("SELECT id FROM ttrss_entries
5e3d5480 853 WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
9e222305 854
2c08214a
AD
855 if (db_num_rows($result) == 0) {
856
68cccafc 857 _debug("base guid [$entry_guid] not found", $debug_enabled);
2c08214a
AD
858
859 // base post entry does not exist, create it
860
6322ac79 861 $result = db_query(
2c08214a
AD
862 "INSERT INTO ttrss_entries
863 (title,
864 guid,
865 link,
866 updated,
867 content,
868 content_hash,
869 no_orig_date,
870 date_updated,
871 date_entered,
872 comments,
873 num_comments,
b30abdad 874 plugin_data,
6b461797 875 lang,
2c08214a
AD
876 author)
877 VALUES
878 ('$entry_title',
5e3d5480 879 '$entry_guid_hashed',
2c08214a
AD
880 '$entry_link',
881 '$entry_timestamp_fmt',
882 '$entry_content',
b1840673 883 '$entry_current_hash',
5ba1ddd4 884 false,
2c08214a 885 NOW(),
be574731 886 '$date_feed_processed',
2c08214a
AD
887 '$entry_comments',
888 '$num_comments',
b30abdad 889 '$entry_plugin_data',
6b461797 890 '$entry_language',
2c08214a 891 '$entry_author')");
e8291805 892
2c08214a 893 } else {
2c08214a 894 $base_entry_id = db_fetch_result($result, 0, "id");
2c08214a
AD
895 }
896
897 // now it should exist, if not - bad luck then
898
b1840673 899 $result = db_query("SELECT id FROM ttrss_entries
5e3d5480 900 WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
2c08214a
AD
901
902 $entry_ref_id = 0;
903 $entry_int_id = 0;
904
905 if (db_num_rows($result) == 1) {
906
68cccafc 907 _debug("base guid found, checking for user record", $debug_enabled);
2c08214a 908
2c08214a
AD
909 $ref_id = db_fetch_result($result, 0, "id");
910 $entry_ref_id = $ref_id;
911
5e3d5480
AD
912 /* $stored_guid = db_fetch_result($result, 0, "guid");
913 if ($stored_guid != $entry_guid_hashed) {
68cccafc 914 if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
5e3d5480 915
a42c55f0 916 db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
5e3d5480
AD
917 id = '$ref_id'");
918 } */
919
2c08214a 920 if (find_article_filter($article_filters, "filter")) {
07877caf 921 //db_query("COMMIT"); // close transaction in progress
2c08214a
AD
922 continue;
923 }
924
6de3a1be 925 $score = calculate_article_score($article_filters) + $entry_score_modifier;
2c08214a 926
6de3a1be 927 _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
2c08214a 928
4f186b1f
AD
929 // check for user post link to main table
930
2c08214a 931 $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
5bdcb8fd 932 ref_id = '$ref_id' AND owner_uid = '$owner_uid'";
2c08214a
AD
933
934// if ($_REQUEST["xdebug"]) print "$query\n";
935
a42c55f0 936 $result = db_query($query);
2c08214a
AD
937
938 // okay it doesn't exist - create user entry
939 if (db_num_rows($result) == 0) {
940
68cccafc 941 _debug("user record not found, creating...", $debug_enabled);
2c08214a 942
c9299c28 943 if ($score >= -500 && !find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
2c08214a
AD
944 $unread = 'true';
945 $last_read_qpart = 'NULL';
946 } else {
947 $unread = 'false';
948 $last_read_qpart = 'NOW()';
949 }
950
951 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
952 $marked = 'true';
953 } else {
954 $marked = 'false';
955 }
956
957 if (find_article_filter($article_filters, 'publish')) {
958 $published = 'true';
959 } else {
960 $published = 'false';
961 }
962
2ea9bbfd
AD
963 // N-grams
964
c9299c28 965 /* if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
2ea9bbfd 966
a42c55f0 967 $result = db_query("SELECT COUNT(*) AS similar FROM
2ea9bbfd
AD
968 ttrss_entries,ttrss_user_entries
969 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
970 AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
971 AND owner_uid = $owner_uid");
972
973 $ngram_similar = db_fetch_result($result, 0, "similar");
974
68cccafc 975 _debug("N-gram similar results: $ngram_similar", $debug_enabled);
2ea9bbfd
AD
976
977 if ($ngram_similar > 0) {
978 $unread = 'false';
979 }
c9299c28 980 } */
2ea9bbfd 981
7873d588
AD
982 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
983 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
984
6322ac79 985 $result = db_query(
2c08214a
AD
986 "INSERT INTO ttrss_user_entries
987 (ref_id, owner_uid, feed_id, unread, last_read, marked,
7873d588
AD
988 published, score, tag_cache, label_cache, uuid,
989 last_marked, last_published)
2c08214a 990 VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
7873d588
AD
991 $last_read_qpart, $marked, $published, '$score', '', '',
992 '', $last_marked, $last_published)");
2c08214a
AD
993
994 if (PUBSUBHUBBUB_HUB && $published == 'true') {
995 $rss_link = get_self_url_prefix() .
996 "/public.php?op=rss&id=-2&key=" .
a42c55f0 997 get_feed_access_key(-2, false, $owner_uid);
2c08214a
AD
998
999 $p = new Publisher(PUBSUBHUBBUB_HUB);
1000
5ba1ddd4 1001 /* $pubsub_result = */ $p->publish_update($rss_link);
2c08214a
AD
1002 }
1003
6322ac79 1004 $result = db_query(
2c08214a
AD
1005 "SELECT int_id FROM ttrss_user_entries WHERE
1006 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
1007 feed_id = '$feed' LIMIT 1");
1008
1009 if (db_num_rows($result) == 1) {
1010 $entry_int_id = db_fetch_result($result, 0, "int_id");
1011 }
1012 } else {
68cccafc 1013 _debug("user record FOUND", $debug_enabled);
2c08214a
AD
1014
1015 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
1016 $entry_int_id = db_fetch_result($result, 0, "int_id");
1017 }
1018
68cccafc 1019 _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
2c08214a 1020
e854442e 1021 if (DB_TYPE == "pgsql") {
978989bb 1022 $tsvector_combined = db_escape_string(mb_substr($entry_title . ' ' . strip_tags(str_replace('<', ' <', $entry_content)),
e854442e
AD
1023 0, 1000000));
1024
df659891 1025 $tsvector_qpart = "tsvector_combined = to_tsvector('$feed_language', '$tsvector_combined'),";
e854442e
AD
1026
1027 } else {
1028 $tsvector_qpart = "";
1029 }
1030
b1840673
AD
1031 db_query("UPDATE ttrss_entries
1032 SET title = '$entry_title',
1033 content = '$entry_content',
1034 content_hash = '$entry_current_hash',
1035 updated = '$entry_timestamp_fmt',
e854442e 1036 $tsvector_qpart
b1840673
AD
1037 num_comments = '$num_comments',
1038 plugin_data = '$entry_plugin_data',
1039 author = '$entry_author',
1040 lang = '$entry_language'
1041 WHERE id = '$ref_id'");
1042
59e83455
AD
1043 // update aux data
1044 db_query("UPDATE ttrss_user_entries
1045 SET score = '$score' WHERE ref_id = '$ref_id'");
1046
b1840673 1047 if ($mark_unread_on_update) {
24e6ff5d
AD
1048 _debug("article updated, marking unread as requested.", $debug_enabled);
1049
b1840673
AD
1050 db_query("UPDATE ttrss_user_entries
1051 SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
2c08214a
AD
1052 }
1053 }
1054
07877caf 1055 //db_query("COMMIT");
2c08214a 1056
a29fe121
AD
1057 _debug("assigning labels [other]...", $debug_enabled);
1058
1059 foreach ($article_labels as $label) {
1060 label_add_article($entry_ref_id, $label[1], $owner_uid);
1061 }
1062
1063 _debug("assigning labels [filters]...", $debug_enabled);
2c08214a 1064
a42c55f0 1065 assign_article_to_label_filters($entry_ref_id, $article_filters,
b24504b1 1066 $owner_uid, $article_labels);
2c08214a 1067
68cccafc 1068 _debug("looking for enclosures...", $debug_enabled);
2c08214a
AD
1069
1070 // enclosures
1071
1072 $enclosures = array();
1073
19b3992b 1074 $encs = $item->get_enclosures();
2c08214a 1075
19b3992b
AD
1076 if (is_array($encs)) {
1077 foreach ($encs as $e) {
1078 $e_item = array(
523bd90b 1079 $e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
2c08214a 1080 array_push($enclosures, $e_item);
2c08214a
AD
1081 }
1082 }
1083
2c08214a 1084 if ($debug_enabled) {
68cccafc 1085 _debug("article enclosures:", $debug_enabled);
2c08214a
AD
1086 print_r($enclosures);
1087 }
1088
dab229cd 1089 //db_query("BEGIN");
2c08214a 1090
5c54e683
AD
1091// debugging
1092// db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
1093
2c08214a 1094 foreach ($enclosures as $enc) {
a42c55f0
AD
1095 $enc_url = db_escape_string($enc[0]);
1096 $enc_type = db_escape_string($enc[1]);
1097 $enc_dur = db_escape_string($enc[2]);
5c54e683 1098 $enc_title = db_escape_string($enc[3]);
523bd90b
FE
1099 $enc_width = intval($enc[4]);
1100 $enc_height = intval($enc[5]);
2c08214a 1101
a42c55f0 1102 $result = db_query("SELECT id FROM ttrss_enclosures
2c08214a
AD
1103 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
1104
1105 if (db_num_rows($result) == 0) {
a42c55f0 1106 db_query("INSERT INTO ttrss_enclosures
523bd90b
FE
1107 (content_url, content_type, title, duration, post_id, width, height) VALUES
1108 ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
2c08214a
AD
1109 }
1110 }
1111
dab229cd 1112 //db_query("COMMIT");
2c08214a
AD
1113
1114 // check for manual tags (we have to do it here since they're loaded from filters)
1115
1116 foreach ($article_filters as $f) {
6aff7845 1117 if ($f["type"] == "tag") {
2c08214a 1118
6aff7845 1119 $manual_tags = trim_array(explode(",", $f["param"]));
2c08214a
AD
1120
1121 foreach ($manual_tags as $tag) {
1122 if (tag_is_valid($tag)) {
1123 array_push($entry_tags, $tag);
1124 }
1125 }
1126 }
1127 }
1128
1129 // Skip boring tags
1130
6322ac79 1131 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
2c08214a
AD
1132 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1133
1134 $filtered_tags = array();
1135 $tags_to_cache = array();
1136
1137 if ($entry_tags && is_array($entry_tags)) {
1138 foreach ($entry_tags as $tag) {
1139 if (array_search($tag, $boring_tags) === false) {
1140 array_push($filtered_tags, $tag);
1141 }
1142 }
1143 }
1144
1145 $filtered_tags = array_unique($filtered_tags);
1146
1147 if ($debug_enabled) {
68cccafc 1148 _debug("filtered article tags:", $debug_enabled);
2c08214a
AD
1149 print_r($filtered_tags);
1150 }
1151
1152 // Save article tags in the database
1153
1154 if (count($filtered_tags) > 0) {
1155
dab229cd 1156 //db_query("BEGIN");
2c08214a
AD
1157
1158 foreach ($filtered_tags as $tag) {
1159
1160 $tag = sanitize_tag($tag);
a42c55f0 1161 $tag = db_escape_string($tag);
2c08214a
AD
1162
1163 if (!tag_is_valid($tag)) continue;
1164
a42c55f0 1165 $result = db_query("SELECT id FROM ttrss_tags
2c08214a
AD
1166 WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1167 owner_uid = '$owner_uid' LIMIT 1");
1168
1169 if ($result && db_num_rows($result) == 0) {
1170
a42c55f0 1171 db_query("INSERT INTO ttrss_tags
2c08214a
AD
1172 (owner_uid,tag_name,post_int_id)
1173 VALUES ('$owner_uid','$tag', '$entry_int_id')");
1174 }
1175
1176 array_push($tags_to_cache, $tag);
1177 }
1178
1179 /* update the cache */
1180
1181 $tags_to_cache = array_unique($tags_to_cache);
1182
a42c55f0 1183 $tags_str = db_escape_string(join(",", $tags_to_cache));
2c08214a 1184
a42c55f0 1185 db_query("UPDATE ttrss_user_entries
2c08214a
AD
1186 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1187 AND owner_uid = $owner_uid");
1188
dab229cd 1189 //db_query("COMMIT");
2c08214a
AD
1190 }
1191
68cccafc 1192 _debug("article processed", $debug_enabled);
2c08214a
AD
1193 }
1194
68cccafc 1195 _debug("purging feed...", $debug_enabled);
2c08214a 1196
a42c55f0 1197 purge_feed($feed, 0, $debug_enabled);
2c08214a 1198
a42c55f0 1199 db_query("UPDATE ttrss_feeds
f2798eb6 1200 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
2c08214a 1201
a42c55f0 1202// db_query("COMMIT");
2c08214a
AD
1203
1204 } else {
1205
a42c55f0 1206 $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
2c08214a 1207
4ad04ee2
AD
1208 _debug("fetch error: $error_msg", $debug_enabled);
1209
1210 if (count($rss->errors()) > 1) {
1211 foreach ($rss->errors() as $error) {
1212 _debug("+ $error");
1213 }
1214 }
2c08214a 1215
6322ac79 1216 db_query(
2c08214a 1217 "UPDATE ttrss_feeds SET last_error = '$error_msg',
88edaa93 1218 last_updated = NOW() WHERE id = '$feed'");
2c08214a 1219
88edaa93
AD
1220 unset($rss);
1221 }
2c08214a 1222
68cccafc 1223 _debug("done", $debug_enabled);
88edaa93
AD
1224
1225 return $rss;
2c08214a
AD
1226 }
1227
3c696512 1228 function cache_images($html, $site_url, $debug) {
3c696512
AD
1229 libxml_use_internal_errors(true);
1230
1231 $charset_hack = '<head>
1232 <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1233 </head>';
1234
1235 $doc = new DOMDocument();
1236 $doc->loadHTML($charset_hack . $html);
1237 $xpath = new DOMXPath($doc);
1238
1239 $entries = $xpath->query('(//img[@src])');
1240
1241 foreach ($entries as $entry) {
1242 if ($entry->hasAttribute('src')) {
1243 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1244
1245 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1246
1247 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1248
1249 if (!file_exists($local_filename)) {
1250 $file_content = fetch_file_contents($src);
1251
a3d50184 1252 if ($file_content && strlen($file_content) > _MIN_CACHE_IMAGE_SIZE) {
3c696512
AD
1253 file_put_contents($local_filename, $file_content);
1254 }
4a27966e
J
1255 } else {
1256 touch($local_filename);
3c696512 1257 }
3c696512
AD
1258 }
1259 }
3c696512
AD
1260 }
1261
a42c55f0 1262 function expire_error_log($debug) {
e2261e17
AD
1263 if ($debug) _debug("Removing old error log entries...");
1264
1265 if (DB_TYPE == "pgsql") {
a42c55f0 1266 db_query("DELETE FROM ttrss_error_log
e2261e17
AD
1267 WHERE created_at < NOW() - INTERVAL '7 days'");
1268 } else {
a42c55f0 1269 db_query("DELETE FROM ttrss_error_log
e2261e17
AD
1270 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1271 }
1272
1273 }
1274
2a91b6ff 1275 function expire_lock_files($debug) {
65465085 1276 //if ($debug) _debug("Removing old lock files...");
2a91b6ff
AD
1277
1278 $num_deleted = 0;
1279
1280 if (is_writable(LOCK_DIRECTORY)) {
1281 $files = glob(LOCK_DIRECTORY . "/*.lock");
1282
1283 if ($files) {
1284 foreach ($files as $file) {
11344971 1285 if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
2a91b6ff
AD
1286 unlink($file);
1287 ++$num_deleted;
1288 }
1289 }
1290 }
1291 }
1292
65465085 1293 if ($debug) _debug("Removed $num_deleted old lock files.");
2a91b6ff
AD
1294 }
1295
3c696512 1296 function expire_cached_files($debug) {
3306daec 1297 foreach (array("simplepie", "images", "export", "upload") as $dir) {
3c696512 1298 $cache_dir = CACHE_DIR . "/$dir";
2c08214a 1299
65465085 1300// if ($debug) _debug("Expiring $cache_dir");
2c08214a 1301
3c696512
AD
1302 $num_deleted = 0;
1303
1304 if (is_writable($cache_dir)) {
1305 $files = glob("$cache_dir/*");
1306
2a91b6ff 1307 if ($files) {
2ab20c31
AD
1308 foreach ($files as $file) {
1309 if (time() - filemtime($file) > 86400*7) {
1310 unlink($file);
3c696512 1311
2ab20c31
AD
1312 ++$num_deleted;
1313 }
3c696512
AD
1314 }
1315 }
2a91b6ff 1316 }
3c696512 1317
65465085 1318 if ($debug) _debug("$cache_dir: removed $num_deleted files.");
3c696512
AD
1319 }
1320 }
2c08214a 1321
a3e0bdcf
AD
1322 /**
1323 * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1324 * Returns the url query as associative array
1325 *
1326 * @param string query
1327 * @return array params
1328 */
1329 function convertUrlQuery($query) {
1330 $queryParts = explode('&', $query);
1331
1332 $params = array();
1333
1334 foreach ($queryParts as $param) {
1335 $item = explode('=', $param);
1336 $params[$item[0]] = $item[1];
1337 }
1338
1339 return $params;
1340 }
92c14e9d 1341
48cb2536 1342 function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
92c14e9d
AD
1343 $matches = array();
1344
1345 foreach ($filters as $filter) {
1346 $match_any_rule = $filter["match_any_rule"];
a3a896a1 1347 $inverse = $filter["inverse"];
92c14e9d
AD
1348 $filter_match = false;
1349
1350 foreach ($filter["rules"] as $rule) {
1351 $match = false;
ffa1bd7b 1352 $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
a3a896a1 1353 $rule_inverse = $rule["inverse"];
92c14e9d
AD
1354
1355 if (!$reg_exp)
1356 continue;
1357
1358 switch ($rule["type"]) {
1359 case "title":
1360 $match = @preg_match("/$reg_exp/i", $title);
1361 break;
1362 case "content":
d03ae73e
AD
1363 // we don't need to deal with multiline regexps
1364 $content = preg_replace("/[\r\n\t]/", "", $content);
1365
92c14e9d
AD
1366 $match = @preg_match("/$reg_exp/i", $content);
1367 break;
1368 case "both":
d03ae73e
AD
1369 // we don't need to deal with multiline regexps
1370 $content = preg_replace("/[\r\n\t]/", "", $content);
1371
72d1d067 1372 $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
92c14e9d
AD
1373 break;
1374 case "link":
1375 $match = @preg_match("/$reg_exp/i", $link);
1376 break;
1377 case "author":
1378 $match = @preg_match("/$reg_exp/i", $author);
1379 break;
1380 case "tag":
7b80b5e1
DK
1381 foreach ($tags as $tag) {
1382 if (@preg_match("/$reg_exp/i", $tag)) {
1383 $match = true;
1384 break;
1385 }
1386 }
92c14e9d
AD
1387 break;
1388 }
1389
a3a896a1
AD
1390 if ($rule_inverse) $match = !$match;
1391
92c14e9d
AD
1392 if ($match_any_rule) {
1393 if ($match) {
1394 $filter_match = true;
1395 break;
1396 }
1397 } else {
1398 $filter_match = $match;
1399 if (!$match) {
1400 break;
1401 }
1402 }
1403 }
1404
a3a896a1
AD
1405 if ($inverse) $filter_match = !$filter_match;
1406
92c14e9d
AD
1407 if ($filter_match) {
1408 foreach ($filter["actions"] AS $action) {
1409 array_push($matches, $action);
5e736e45
AD
1410
1411 // if Stop action encountered, perform no further processing
fd3e5e8d 1412 if (isset($action["type"]) && $action["type"] == "stop") return $matches;
92c14e9d
AD
1413 }
1414 }
1415 }
1416
1417 return $matches;
1418 }
1419
1420 function find_article_filter($filters, $filter_name) {
1421 foreach ($filters as $f) {
1422 if ($f["type"] == $filter_name) {
1423 return $f;
1424 };
1425 }
1426 return false;
1427 }
1428
1429 function find_article_filters($filters, $filter_name) {
1430 $results = array();
1431
1432 foreach ($filters as $f) {
1433 if ($f["type"] == $filter_name) {
1434 array_push($results, $f);
1435 };
1436 }
1437 return $results;
1438 }
1439
1440 function calculate_article_score($filters) {
1441 $score = 0;
1442
1443 foreach ($filters as $f) {
1444 if ($f["type"] == "score") {
1445 $score += $f["param"];
1446 };
1447 }
1448 return $score;
1449 }
1450
b24504b1
AD
1451 function labels_contains_caption($labels, $caption) {
1452 foreach ($labels as $label) {
1453 if ($label[1] == $caption) {
1454 return true;
1455 }
1456 }
1457
1458 return false;
1459 }
1460
a42c55f0 1461 function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
92c14e9d
AD
1462 foreach ($filters as $f) {
1463 if ($f["type"] == "label") {
b24504b1 1464 if (!labels_contains_caption($article_labels, $f["param"])) {
a42c55f0 1465 label_add_article($id, $f["param"], $owner_uid);
b24504b1
AD
1466 }
1467 }
92c14e9d
AD
1468 }
1469 }
87764a50 1470
87d7e850
AD
1471 function make_guid_from_title($title) {
1472 return preg_replace("/[ \"\',.:;]/", "-",
1473 mb_strtolower(strip_tags($title), 'utf-8'));
1474 }
1475
d1f3fa97 1476 /* function verify_feed_xml($feed_data) {
ebec81a6
AD
1477 libxml_use_internal_errors(true);
1478 $doc = new DOMDocument();
1479 $doc->loadXML($feed_data);
1480 $error = libxml_get_last_error();
1481 libxml_clear_errors();
1482 return $error;
d1f3fa97 1483 } */
87d7e850 1484
168cf351
AD
1485 function cleanup_counters_cache($debug) {
1486 $result = db_query("DELETE FROM ttrss_counters_cache
1487 WHERE feed_id > 0 AND
1488 (SELECT COUNT(id) FROM ttrss_feeds WHERE
1489 id = feed_id AND
1490 ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
1491 $frows = db_affected_rows($result);
1492
1493 $result = db_query("DELETE FROM ttrss_cat_counters_cache
1494 WHERE feed_id > 0 AND
1495 (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
1496 id = feed_id AND
1497 ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
1498 $crows = db_affected_rows($result);
1499
1500 _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
1501 }
1502
e2cf81e2
AD
1503 function housekeeping_common($debug) {
1504 expire_cached_files($debug);
1505 expire_lock_files($debug);
1506 expire_error_log($debug);
1507
1508 $count = update_feedbrowser_cache();
1509 _debug("Feedbrowser updated, $count feeds processed.");
1510
1511 purge_orphans( true);
168cf351 1512 cleanup_counters_cache($debug);
e2cf81e2 1513
9b736a20
AD
1514 //$rc = cleanup_tags( 14, 50000);
1515 //_debug("Cleaned $rc cached tags.");
8e470220 1516
00f22824 1517 PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
8e470220 1518
e2cf81e2 1519 }
2c08214a 1520?>