]> git.wh0rd.org - tt-rss.git/blame - include/rssfuncs.php
rename cache images option label (rip translations)
[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);
41bead9b 5 define_default('_MIN_CACHE_FILE_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
45378752
LD
26 FROM ttrss_feeds WHERE feed_url NOT IN (SELECT feed_url FROM ttrss_feeds
27 WHERE private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%')
79178062
AD
28 GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
29
a42c55f0 30 db_query("BEGIN");
79178062 31
a42c55f0 32 db_query("DELETE FROM ttrss_feedbrowser_cache");
79178062
AD
33
34 $count = 0;
35
36 while ($line = db_fetch_assoc($result)) {
a42c55f0
AD
37 $subscribers = db_escape_string($line["subscribers"]);
38 $feed_url = db_escape_string($line["feed_url"]);
39 $title = db_escape_string($line["title"]);
40 $site_url = db_escape_string($line["site_url"]);
79178062 41
a42c55f0 42 $tmp_result = db_query("SELECT subscribers FROM
79178062
AD
43 ttrss_feedbrowser_cache WHERE feed_url = '$feed_url'");
44
45 if (db_num_rows($tmp_result) == 0) {
46
a42c55f0 47 db_query("INSERT INTO ttrss_feedbrowser_cache
79178062
AD
48 (feed_url, site_url, title, subscribers) VALUES ('$feed_url',
49 '$site_url', '$title', '$subscribers')");
50
51 ++$count;
52
53 }
54
55 }
56
a42c55f0 57 db_query("COMMIT");
79178062
AD
58
59 return $count;
60
61 }
62
63
2c08214a
AD
64 /**
65 * Update a feed batch.
66 * Used by daemons to update n feeds by run.
67 * Only update feed needing a update, and not being processed
68 * by another process.
69 *
70 * @param mixed $link Database link
71 * @param integer $limit Maximum number of feeds in update batch. Default to DAEMON_FEED_LIMIT.
72 * @param boolean $from_http Set to true if you call this function from http to disable cli specific code.
73 * @param boolean $debug Set to false to disable debug output. Default to true.
74 * @return void
75 */
a42c55f0 76 function update_daemon_common($limit = DAEMON_FEED_LIMIT, $from_http = false, $debug = true) {
2c08214a
AD
77 // Process all other feeds using last_updated and interval parameters
78
6322ac79 79 $schema_version = get_schema_version();
857efe49
AD
80
81 if ($schema_version != SCHEMA_VERSION) {
82 die("Schema version is wrong, please upgrade the database.\n");
83 }
84
61c1812f
AD
85 define('PREFS_NO_CACHE', true);
86
2c08214a 87 // Test if the user has loggued in recently. If not, it does not update its feeds.
09e8bdfd 88 if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
2c08214a
AD
89 if (DB_TYPE == "pgsql") {
90 $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
91 } else {
92 $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
93 }
94 } else {
95 $login_thresh_qpart = "";
96 }
97
537eca40 98 // Test if the feed need a update (update interval exceeded).
2c08214a
AD
99 if (DB_TYPE == "pgsql") {
100 $update_limit_qpart = "AND ((
101 ttrss_feeds.update_interval = 0
ee0542ce 102 AND ttrss_user_prefs.value != '-1'
2c08214a
AD
103 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
104 ) OR (
105 ttrss_feeds.update_interval > 0
106 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
f08426e3
AD
107 ) OR (ttrss_feeds.last_updated IS NULL
108 AND ttrss_user_prefs.value != '-1')
109 OR (last_updated = '1970-01-01 00:00:00'
110 AND ttrss_user_prefs.value != '-1'))";
2c08214a
AD
111 } else {
112 $update_limit_qpart = "AND ((
113 ttrss_feeds.update_interval = 0
ee0542ce 114 AND ttrss_user_prefs.value != '-1'
2c08214a
AD
115 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
116 ) OR (
117 ttrss_feeds.update_interval > 0
118 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
f08426e3
AD
119 ) OR (ttrss_feeds.last_updated IS NULL
120 AND ttrss_user_prefs.value != '-1')
121 OR (last_updated = '1970-01-01 00:00:00'
122 AND ttrss_user_prefs.value != '-1'))";
2c08214a
AD
123 }
124
125 // Test if feed is currently being updated by another process.
126 if (DB_TYPE == "pgsql") {
566417c4 127 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
2c08214a 128 } else {
566417c4 129 $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
130 }
131
132 // Test if there is a limit to number of updated feeds
133 $query_limit = "";
134 if($limit) $query_limit = sprintf("LIMIT %d", $limit);
135
98070db0
TK
136 // Update the least recently updated feeds first
137 $query_order = "ORDER BY last_updated";
138 if (DB_TYPE == "pgsql") $query_order .= " NULLS FIRST";
139
fce451a4 140 $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
2c08214a
AD
141 FROM
142 ttrss_feeds, ttrss_users, ttrss_user_prefs
f4ae0f05 143 WHERE
2c08214a 144 ttrss_feeds.owner_uid = ttrss_users.id
f08426e3 145 AND ttrss_user_prefs.profile IS NULL
2c08214a
AD
146 AND ttrss_users.id = ttrss_user_prefs.owner_uid
147 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
148 $login_thresh_qpart $update_limit_qpart
1c4421fc 149 $updstart_thresh_qpart
98070db0 150 $query_order $query_limit";
fce451a4
AD
151
152 // We search for feed needing update.
153 $result = db_query($query);
2c08214a 154
ca5ff2d9 155 if($debug) _debug(sprintf("Scheduled %d feeds to update...", db_num_rows($result)));
2c08214a
AD
156
157 // Here is a little cache magic in order to minimize risk of double feed updates.
158 $feeds_to_update = array();
159 while ($line = db_fetch_assoc($result)) {
a42c55f0 160 array_push($feeds_to_update, db_escape_string($line['feed_url']));
2c08214a
AD
161 }
162
163 // We update the feed last update started date before anything else.
164 // There is no lag due to feed contents downloads
165 // It prevent an other process to update the same feed.
1c4421fc
AD
166
167 if(count($feeds_to_update) > 0) {
168 $feeds_quoted = array();
169
170 foreach ($feeds_to_update as $feed) {
a42c55f0 171 array_push($feeds_quoted, "'" . db_escape_string($feed) . "'");
1c4421fc
AD
172 }
173
a42c55f0 174 db_query(sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
1c4421fc 175 WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
2c08214a
AD
176 }
177
8292d05b 178 $nf = 0;
2d9c5684 179 $bstarted = microtime(true);
8292d05b 180
5cbd1fe8
AD
181 $batch_owners = array();
182
2c08214a 183 // For each feed, we call the feed update function.
1c4421fc
AD
184 foreach ($feeds_to_update as $feed) {
185 if($debug) _debug("Base feed: $feed");
2c08214a 186
a42c55f0 187 //update_rss_feed($line["id"], true);
2c08214a 188
1c4421fc
AD
189 // since we have the data cached, we can deal with other feeds with the same url
190
15c762be 191 $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
ee0542ce
AD
192 FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
193 ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
194 ttrss_users.id = ttrss_user_prefs.owner_uid AND
195 ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
f08426e3 196 ttrss_user_prefs.profile IS NULL AND
9e84bab4
AD
197 feed_url = '".db_escape_string($feed)."'
198 $update_limit_qpart
1c4421fc 199 $login_thresh_qpart
5929a0c1 200 ORDER BY ttrss_feeds.id $query_limit");
1c4421fc
AD
201
202 if (db_num_rows($tmp_result) > 0) {
88edaa93
AD
203 $rss = false;
204
1c4421fc 205 while ($tline = db_fetch_assoc($tmp_result)) {
15c762be 206 if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
f08426e3 207
5cbd1fe8
AD
208 if (array_search($tline["owner_uid"], $batch_owners) === FALSE)
209 array_push($batch_owners, $tline["owner_uid"]);
210
2d9c5684 211 $fstarted = microtime(true);
f77c7203 212 $rss = update_rss_feed($tline["id"], true, false);
4f71d743 213 _debug_suppress(false);
2d9c5684
AD
214
215 _debug(sprintf(" %.4f (sec)", microtime(true) - $fstarted));
216
8292d05b 217 ++$nf;
1c4421fc
AD
218 }
219 }
2c08214a
AD
220 }
221
2d9c5684
AD
222 if ($nf > 0) {
223 _debug(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
224 microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
225 }
226
5cbd1fe8
AD
227 foreach ($batch_owners as $owner_uid) {
228 _debug("Running housekeeping tasks for user $owner_uid...");
229
230 housekeeping_user($owner_uid);
231 }
232
f1611683
AD
233 require_once "digest.php";
234
2c08214a 235 // Send feed digests by email if needed.
a42c55f0 236 send_headlines_digests($debug);
2c08214a 237
8292d05b
AD
238 return $nf;
239
2c08214a
AD
240 } // function update_daemon_common
241
6022776d
AD
242 // this is used when subscribing
243 function set_basic_feed_info($feed) {
244
245 $feed = db_escape_string($feed);
246
9bd422c2 247 $result = db_query("SELECT feed_url,auth_pass,auth_login,auth_pass_encrypted
6022776d
AD
248 FROM ttrss_feeds WHERE id = '$feed'");
249
250 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
251 0, "auth_pass_encrypted"));
252
253 $auth_login = db_fetch_result($result, 0, "auth_login");
254 $auth_pass = db_fetch_result($result, 0, "auth_pass");
255
17a8e61d 256 if ($auth_pass_encrypted && function_exists("mcrypt_decrypt")) {
6022776d
AD
257 require_once "crypt.php";
258 $auth_pass = decrypt_string($auth_pass);
259 }
260
261 $fetch_url = db_fetch_result($result, 0, "feed_url");
262
263 $feed_data = fetch_file_contents($fetch_url, false,
264 $auth_login, $auth_pass, false,
3d5d2890 265 FEED_FETCH_TIMEOUT,
6022776d
AD
266 0);
267
268 global $fetch_curl_used;
269
270 if (!$fetch_curl_used) {
271 $tmp = @gzdecode($feed_data);
272
273 if ($tmp) $feed_data = $tmp;
274 }
275
276 $feed_data = trim($feed_data);
277
278 $rss = new FeedParser($feed_data);
279 $rss->init();
280
281 if (!$rss->error()) {
282
283 $result = db_query("SELECT title, site_url FROM ttrss_feeds WHERE id = '$feed'");
284
285 $registered_title = db_fetch_result($result, 0, "title");
286 $orig_site_url = db_fetch_result($result, 0, "site_url");
287
288 $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
289 $feed_title = db_escape_string(mb_substr($rss->get_title(), 0, 199));
290
291 if ($feed_title && (!$registered_title || $registered_title == "[Unknown]")) {
292 db_query("UPDATE ttrss_feeds SET
293 title = '$feed_title' WHERE id = '$feed'");
294 }
295
296 if ($site_url && $orig_site_url != $site_url) {
297 db_query("UPDATE ttrss_feeds SET
298 site_url = '$site_url' WHERE id = '$feed'");
299 }
300 }
301 }
302
87764a50 303 // ignore_daemon is not used
88edaa93 304 function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
2c08214a 305
2c08214a
AD
306 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
307
4f71d743 308 _debug_suppress(!$debug_enabled);
68cccafc 309 _debug("start", $debug_enabled);
2c08214a 310
6bb96beb
AD
311 $result = db_query("SELECT title FROM ttrss_feeds
312 WHERE id = '$feed'");
bfe1eb4e
AD
313
314 if (db_num_rows($result) == 0) {
315 _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
316 user_error("Attempt to update unknown/invalid feed $feed", E_USER_WARNING);
317 return false;
318 }
319
6bb96beb
AD
320 $title = db_fetch_result($result, 0, "title");
321
322 // feed was batch-subscribed or something, we need to get basic info
323 // this is not optimal currently as it fetches stuff separately TODO: optimize
324 if ($title == "[Unknown]") {
325 _debug("setting basic feed info for $feed...");
326 set_basic_feed_info($feed);
327 }
328
a42c55f0 329 $result = db_query("SELECT id,update_interval,auth_login,
5ba1ddd4 330 feed_url,auth_pass,cache_images,
5321e775 331 mark_unread_on_update, owner_uid,
90e5f4f1 332 pubsub_state, auth_pass_encrypted,
0edf1d0d 333 feed_language
87764a50 334 FROM ttrss_feeds WHERE id = '$feed'");
2c08214a 335
2c08214a
AD
336 $owner_uid = db_fetch_result($result, 0, "owner_uid");
337 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
338 0, "mark_unread_on_update"));
2c08214a 339 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
044cff2d
AD
340 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
341 0, "auth_pass_encrypted"));
2c08214a 342
a42c55f0 343 db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
2c08214a
AD
344 WHERE id = '$feed'");
345
346 $auth_login = db_fetch_result($result, 0, "auth_login");
347 $auth_pass = db_fetch_result($result, 0, "auth_pass");
348
17a8e61d 349 if ($auth_pass_encrypted && function_exists("mcrypt_decrypt")) {
044cff2d
AD
350 require_once "crypt.php";
351 $auth_pass = decrypt_string($auth_pass);
352 }
353
2c08214a
AD
354 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
355 $fetch_url = db_fetch_result($result, 0, "feed_url");
df659891
AD
356 $feed_language = db_escape_string(mb_strtolower(db_fetch_result($result, 0, "feed_language")));
357 if (!$feed_language) $feed_language = 'english';
2c08214a 358
a42c55f0 359 $feed = db_escape_string($feed);
2c08214a 360
f074ffe9 361 $date_feed_processed = date('Y-m-d H:i');
2c08214a 362
865a3ed6 363 $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
f074ffe9 364
ee65bef4
AD
365 $pluginhost = new PluginHost();
366 $pluginhost->set_debug($debug_enabled);
367 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
368
369 $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
370 $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
371 $pluginhost->load_data();
372
88edaa93
AD
373 if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
374 _debug("using previously initialized parser object");
375 } else {
376 $rss_hash = false;
4f9cbdff 377
88edaa93 378 $force_refetch = isset($_REQUEST["force_refetch"]);
0edf1d0d 379 $feed_data = "";
687a4f59 380
95beaa14 381 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
0edf1d0d 382 $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, 0, $auth_login, $auth_pass);
95beaa14
RA
383 }
384
385 // try cache
386 if (!$feed_data &&
387 file_exists($cache_filename) &&
88edaa93
AD
388 is_readable($cache_filename) &&
389 !$auth_login && !$auth_pass &&
390 filemtime($cache_filename) > time() - 30) {
2c08214a 391
15d0528a 392 _debug("using local cache [$cache_filename].", $debug_enabled);
be574731 393
88edaa93 394 @$feed_data = file_get_contents($cache_filename);
52637d3b 395
88edaa93
AD
396 if ($feed_data) {
397 $rss_hash = sha1($feed_data);
398 }
f074ffe9 399
88edaa93
AD
400 } else {
401 _debug("local cache will not be used for this feed", $debug_enabled);
402 }
ee65bef4 403
95beaa14 404 // fetch feed from source
fd687300
AD
405 if (!$feed_data) {
406 _debug("fetching [$fetch_url]...", $debug_enabled);
312742db
AD
407
408 if (ini_get("open_basedir") && function_exists("curl_init")) {
409 _debug("not using CURL due to open_basedir restrictions");
410 }
411
fd687300
AD
412 $feed_data = fetch_file_contents($fetch_url, false,
413 $auth_login, $auth_pass, false,
414 $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
0edf1d0d 415 0);
3f6f0857 416
fd687300 417 global $fetch_curl_used;
3f6f0857 418
fd687300
AD
419 if (!$fetch_curl_used) {
420 $tmp = @gzdecode($feed_data);
3f6f0857 421
fd687300
AD
422 if ($tmp) $feed_data = $tmp;
423 }
1367bc3f 424
fd687300 425 $feed_data = trim($feed_data);
017401dd 426
fd687300
AD
427 _debug("fetch done.", $debug_enabled);
428
95beaa14
RA
429 // cache vanilla feed data for re-use
430 if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
431 $new_rss_hash = sha1($feed_data);
432
433 if ($new_rss_hash != $rss_hash) {
434 _debug("saving $cache_filename", $debug_enabled);
435 @file_put_contents($cache_filename, $feed_data);
436 }
437 }
4f9cbdff 438 }
017401dd 439
fd687300
AD
440 if (!$feed_data) {
441 global $fetch_last_error;
442 global $fetch_last_error_code;
f074ffe9 443
fd687300 444 _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
f074ffe9 445
fd687300 446 $error_escaped = '';
7a01dc77 447
fd687300
AD
448 // If-Modified-Since
449 if ($fetch_last_error_code != 304) {
450 $error_escaped = db_escape_string($fetch_last_error);
451 } else {
452 _debug("source claims data not modified, nothing to do.", $debug_enabled);
453 }
4f9cbdff 454
fd687300
AD
455 db_query(
456 "UPDATE ttrss_feeds SET last_error = '$error_escaped',
457 last_updated = NOW() WHERE id = '$feed'");
4f9cbdff 458
fd687300
AD
459 return;
460 }
f074ffe9
AD
461 }
462
1ffe3391 463 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
6791af0c 464 $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
017401dd
AD
465 }
466
5ef84097
AD
467 // set last update to now so if anything *simplepie* crashes later we won't be
468 // continuously failing on the same feed
5de40104 469 //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
5ef84097 470
4f9cbdff 471 if (!$rss) {
cd07592c
AD
472 $rss = new FeedParser($feed_data);
473 $rss->init();
5de51df7
AD
474 }
475
2c08214a
AD
476// print_r($rss);
477
a42c55f0 478 $feed = db_escape_string($feed);
2c08214a 479
19b3992b 480 if (!$rss->error()) {
2c08214a 481
d2a421e3 482 // We use local pluginhost here because we need to load different per-user feed plugins
1ffe3391 483 $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
4412b877 484
df659891 485 _debug("language: $feed_language", $debug_enabled);
68cccafc 486 _debug("processing feed data...", $debug_enabled);
2c08214a 487
a42c55f0 488// db_query("BEGIN");
2c08214a 489
382268c6
AD
490 if (DB_TYPE == "pgsql") {
491 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
492 } else {
493 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
494 }
495
fe4535e6 496 $result = db_query("SELECT owner_uid,favicon_avg_color,
382268c6
AD
497 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
498 favicon_needs_check
2c08214a
AD
499 FROM ttrss_feeds WHERE id = '$feed'");
500
382268c6
AD
501 $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
502 "favicon_needs_check"));
36490f11 503 $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
2c08214a
AD
504
505 $owner_uid = db_fetch_result($result, 0, "owner_uid");
506
a42c55f0 507 $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
2c08214a 508
cd07592c
AD
509 _debug("site_url: $site_url", $debug_enabled);
510 _debug("feed_title: " . $rss->get_title(), $debug_enabled);
511
687a4f59 512 if ($favicon_needs_check || $force_refetch) {
36490f11
AD
513
514 /* terrible hack: if we crash on floicon shit here, we won't check
560cbd8c 515 * the icon avgcolor again (unless the icon got updated) */
36490f11 516
560cbd8c
AD
517 $favicon_file = ICONS_DIR . "/$feed.ico";
518 $favicon_modified = @filemtime($favicon_file);
519
68cccafc 520 _debug("checking favicon...", $debug_enabled);
687a4f59 521
6f7798b6 522 check_feed_favicon($site_url, $feed);
560cbd8c
AD
523 $favicon_modified_new = @filemtime($favicon_file);
524
525 if ($favicon_modified_new > $favicon_modified)
526 $favicon_avg_color = '';
687a4f59 527
6ee0d4b0 528 if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
bc7a144d 529 require_once "colors.php";
687a4f59 530
aafd55ba
AD
531 db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
532 id = '$feed'");
533
6322ac79 534 $favicon_color = db_escape_string(
bc7a144d 535 calculate_avg_color($favicon_file));
63c323f7 536
bc7a144d 537 $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
36490f11 538 } else if ($favicon_avg_color == 'fail') {
84ceb6bd 539 _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
6ac722d5 540 }
687a4f59 541
a42c55f0 542 db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
bc7a144d 543 $favicon_colorstring
f2798eb6
AD
544 WHERE id = '$feed'");
545 }
2c08214a 546
68cccafc 547 _debug("loading filters & labels...", $debug_enabled);
2c08214a 548
a42c55f0 549 $filters = load_filters($feed, $owner_uid);
2c08214a 550
68cccafc 551 _debug("" . count($filters) . " filters loaded.", $debug_enabled);
2c08214a 552
19b3992b 553 $items = $rss->get_items();
2c08214a 554
19b3992b 555 if (!is_array($items)) {
68cccafc 556 _debug("no articles found.", $debug_enabled);
2c08214a 557
a42c55f0 558 db_query("UPDATE ttrss_feeds
2c08214a
AD
559 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
560
561 return; // no articles
562 }
563
564 if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
565
68cccafc 566 _debug("checking for PUSH hub...", $debug_enabled);
2c08214a
AD
567
568 $feed_hub_url = false;
2c08214a 569
19b3992b 570 $links = $rss->get_links('hub');
2c08214a 571
19b3992b
AD
572 if ($links && is_array($links)) {
573 foreach ($links as $l) {
574 $feed_hub_url = $l;
575 break;
2c08214a
AD
576 }
577 }
578
68cccafc 579 _debug("feed hub url: $feed_hub_url", $debug_enabled);
2c08214a 580
d8179cb9
AD
581 $feed_self_url = $fetch_url;
582
583 $links = $rss->get_links('self');
584
585 if ($links && is_array($links)) {
586 foreach ($links as $l) {
587 $feed_self_url = $l;
588 break;
589 }
590 }
591
592 _debug("feed self url = $feed_self_url");
593
594 if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
2c08214a
AD
595 !ini_get("open_basedir")) {
596
becd215a 597 require_once 'lib/pubsubhubbub/Subscriber.php';
2c08214a
AD
598
599 $callback_url = get_self_url_prefix() .
600 "/public.php?op=pubsub&id=$feed";
601
cb3f8773 602 $s = new Pubsubhubbub\Subscriber\Subscriber($feed_hub_url, $callback_url);
2c08214a 603
d8179cb9 604 $rc = $s->subscribe($feed_self_url);
2c08214a 605
5ba1ddd4 606 _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
2c08214a 607
a42c55f0 608 db_query("UPDATE ttrss_feeds SET pubsub_state = 1
2c08214a
AD
609 WHERE id = '$feed'");
610 }
611 }
612
68cccafc 613 _debug("processing articles...", $debug_enabled);
2c08214a 614
6c9f3d4a
AD
615 $tstart = time();
616
19b3992b 617 foreach ($items as $item) {
5d56d100 618 if ($_REQUEST['xdebug'] == 3) {
2c08214a
AD
619 print_r($item);
620 }
621
6c9f3d4a
AD
622 if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
623 _debug("looks like there's too many articles to process at once, breaking out", $debug_enabled);
624 break;
625 }
626
19b3992b
AD
627 $entry_guid = $item->get_id();
628 if (!$entry_guid) $entry_guid = $item->get_link();
629 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
2c08214a
AD
630 if (!$entry_guid) continue;
631
3a4c8973
AD
632 $entry_guid = "$owner_uid,$entry_guid";
633
a42c55f0 634 $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
5e3d5480 635
68cccafc 636 _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
5e3d5480 637
2c08214a
AD
638 $entry_timestamp = "";
639
04d2f9c8
AD
640 $entry_timestamp = $item->get_date();
641
642 _debug("orig date: " . $item->get_date(), $debug_enabled);
2c08214a 643
30123fe6 644 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
2c08214a 645 $entry_timestamp = time();
2c08214a
AD
646 }
647
648 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
649
68cccafc 650 _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
2c08214a 651
99429e57
AD
652// $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
653// $entry_title = decode_numeric_entities($entry_title);
654 $entry_title = $item->get_title();
1b35d30c 655
5d56d100 656 $entry_link = rewrite_relative_url($site_url, $item->get_link());
2c08214a 657
68cccafc
AD
658 _debug("title $entry_title", $debug_enabled);
659 _debug("link $entry_link", $debug_enabled);
2c08214a
AD
660
661 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
662
19b3992b
AD
663 $entry_content = $item->get_content();
664 if (!$entry_content) $entry_content = $item->get_description();
2c08214a
AD
665
666 if ($_REQUEST["xdebug"] == 2) {
9ec10352 667 print "content: ";
0bc503ff 668 print htmlspecialchars($entry_content);
3c696512 669 print "\n";
2c08214a
AD
670 }
671
12ff230b
AD
672 $entry_comments = db_escape_string(mb_substr($item->get_comments_url(), 0, 245));
673 $num_comments = (int) $item->get_comments_count();
2c08214a 674
12ff230b 675 $entry_author = $item->get_author(); // escaped later
a42c55f0 676 $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
2c08214a 677
68cccafc
AD
678 _debug("author $entry_author", $debug_enabled);
679 _debug("num_comments: $num_comments", $debug_enabled);
ee78f81c 680 _debug("looking for tags...", $debug_enabled);
2c08214a
AD
681
682 // parse <category> entries into tags
683
684 $additional_tags = array();
685
19b3992b 686 $additional_tags_src = $item->get_categories();
2c08214a 687
19b3992b
AD
688 if (is_array($additional_tags_src)) {
689 foreach ($additional_tags_src as $tobj) {
cd07592c 690 array_push($additional_tags, $tobj);
2c08214a 691 }
19b3992b 692 }
2c08214a 693
fa6fbd36 694 $entry_tags = array_unique($additional_tags);
2c08214a
AD
695
696 for ($i = 0; $i < count($entry_tags); $i++)
697 $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
698
ee78f81c
AD
699 _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
700
68cccafc 701 _debug("done collecting data.", $debug_enabled);
2c08214a 702
3318d324 703 $result = db_query("SELECT id, content_hash, lang FROM ttrss_entries
b1840673 704 WHERE guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed'");
b30abdad
AD
705
706 if (db_num_rows($result) != 0) {
b1840673
AD
707 $base_entry_id = db_fetch_result($result, 0, "id");
708 $entry_stored_hash = db_fetch_result($result, 0, "content_hash");
a29fe121 709 $article_labels = get_article_labels($base_entry_id, $owner_uid);
3318d324 710 $entry_language = db_fetch_result($result, 0, "lang");
a8ac7661 711
040aa540 712 $existing_tags = get_article_tags($base_entry_id, $owner_uid);
a8ac7661
AD
713 $entry_tags = array_unique(array_merge($entry_tags, $existing_tags));
714
b30abdad 715 } else {
b1840673
AD
716 $base_entry_id = false;
717 $entry_stored_hash = "";
a29fe121 718 $article_labels = array();
3318d324 719 $entry_language = "";
b30abdad
AD
720 }
721
455b1401 722 $article = array("owner_uid" => $owner_uid, // read only
b30abdad 723 "guid" => $entry_guid, // read only
59e83455 724 "guid_hashed" => $entry_guid_hashed, // read only
19b3992b
AD
725 "title" => $entry_title,
726 "content" => $entry_content,
727 "link" => $entry_link,
a29fe121 728 "labels" => $article_labels, // current limitation: can add labels to article, can't remove them
19b3992b 729 "tags" => $entry_tags,
e02555c1 730 "author" => $entry_author,
c9299c28 731 "force_catchup" => false, // ugly hack for the time being
6de3a1be 732 "score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
3318d324 733 "language" => $entry_language,
f73e03e0
AD
734 "feed" => array("id" => $feed,
735 "fetch_url" => $fetch_url,
babfadbf
J
736 "site_url" => $site_url,
737 "cache_images" => $cache_images)
f73e03e0 738 );
cc85704f 739
b1840673 740 $entry_plugin_data = "";
eb16bd9f 741 $entry_current_hash = calculate_article_hash($article, $pluginhost);
b1840673
AD
742
743 _debug("article hash: $entry_current_hash [stored=$entry_stored_hash]", $debug_enabled);
744
522e8b35 745 if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
b1840673
AD
746 _debug("stored article seems up to date [IID: $base_entry_id], updating timestamp only", $debug_enabled);
747
748 // we keep encountering the entry in feeds, so we need to
749 // update date_updated column so that we don't get horrible
750 // dupes when the entry gets purged and reinserted again e.g.
751 // in the case of SLOW SLOW OMG SLOW updating feeds
752
753 $base_entry_id = db_fetch_result($result, 0, "id");
754
755 db_query("UPDATE ttrss_entries SET date_updated = NOW()
756 WHERE id = '$base_entry_id'");
757
5bdcb8fd 758 continue;
b1840673
AD
759 }
760
761 _debug("hash differs, applying plugin filters:", $debug_enabled);
762
1ffe3391 763 foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
b1840673
AD
764 _debug("... " . get_class($plugin), $debug_enabled);
765
766 $start = microtime(true);
19b3992b 767 $article = $plugin->hook_article_filter($article);
0084f0d1 768
b1840673
AD
769 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
770
771 $entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
cc85704f
AD
772 }
773
0bc503ff
AD
774 if ($_REQUEST["xdebug"] == 2) {
775 print "processed content: ";
776 print htmlspecialchars($article["content"]);
777 print "\n";
778 }
779
b1840673
AD
780 $entry_plugin_data = db_escape_string($entry_plugin_data);
781
782 _debug("plugin data: $entry_plugin_data", $debug_enabled);
783
35c12dc4
AD
784 // Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
785 if (DB_TYPE == "mysql") {
786 foreach ($article as $k => $v) {
35c37354
AD
787
788 // i guess we'll have to take the risk of 4byte unicode labels & tags here
dae16f72 789 if (is_string($article[$k])) {
35c37354
AD
790 $article[$k] = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $v);
791 }
35c12dc4
AD
792 }
793 }
794
b8774453
AD
795 /* Collect article tags here so we could filter by them: */
796
557d86fe
AD
797 $matched_rules = array();
798
b8774453
AD
799 $article_filters = get_article_filters($filters, $article["title"],
800 $article["content"], $article["link"], 0, $article["author"],
557d86fe 801 $article["tags"], $matched_rules);
b8774453
AD
802
803 if ($debug_enabled) {
557d86fe
AD
804 _debug("matched filter rules: ", $debug_enabled);
805
806 if (count($matched_rules) != 0) {
807 print_r($matched_rules);
808 }
809
810 _debug("filter actions: ", $debug_enabled);
811
b8774453
AD
812 if (count($article_filters) != 0) {
813 print_r($article_filters);
814 }
815 }
816
817 $plugin_filter_names = find_article_filters($article_filters, "plugin");
818 $plugin_filter_actions = $pluginhost->get_filter_actions();
819
820 if (count($plugin_filter_names) > 0) {
821 _debug("applying plugin filter actions...", $debug_enabled);
822
823 foreach ($plugin_filter_names as $pfn) {
824 list($pfclass,$pfaction) = explode(":", $pfn["param"]);
825
826 if (isset($plugin_filter_actions[$pfclass])) {
827 $plugin = $pluginhost->get_plugin($pfclass);
828
829 _debug("... $pfclass: $pfaction", $debug_enabled);
830
831 if ($plugin) {
832 $start = microtime(true);
833 $article = $plugin->hook_article_filter_action($article, $pfaction);
834
835 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
836 } else {
837 _debug("??? $pfclass: plugin object not found.");
838 }
839 } else {
840 _debug("??? $pfclass: filter plugin not registered.");
841 }
842 }
843 }
844
19b3992b 845 $entry_tags = $article["tags"];
a42c55f0
AD
846 $entry_guid = db_escape_string($entry_guid);
847 $entry_title = db_escape_string($article["title"]);
12ff230b 848 $entry_author = db_escape_string(mb_substr($article["author"], 0, 245));
a42c55f0 849 $entry_link = db_escape_string($article["link"]);
f935d98e 850 $entry_content = $article["content"]; // escaped below
c9299c28 851 $entry_force_catchup = $article["force_catchup"];
a29fe121 852 $article_labels = $article["labels"];
6de3a1be 853 $entry_score_modifier = (int) $article["score_modifier"];
3318d324 854 $entry_language = db_escape_string($article["language"]);
a29fe121
AD
855
856 if ($debug_enabled) {
857 _debug("article labels:", $debug_enabled);
557d86fe
AD
858
859 if (count($article_labels) != 0) {
860 print_r($article_labels);
861 }
a29fe121 862 }
c9299c28
AD
863
864 _debug("force catchup: $entry_force_catchup");
f935d98e 865
0a3fd79b 866 if ($cache_images && is_writable(CACHE_DIR . '/images'))
41bead9b 867 cache_media($entry_content, $site_url, $debug_enabled);
0a3fd79b 868
a42c55f0 869 $entry_content = db_escape_string($entry_content, false);
1f45c857 870
07877caf 871 //db_query("BEGIN");
2c08214a 872
a42c55f0 873 $result = db_query("SELECT id FROM ttrss_entries
5e3d5480 874 WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
9e222305 875
2c08214a
AD
876 if (db_num_rows($result) == 0) {
877
68cccafc 878 _debug("base guid [$entry_guid] not found", $debug_enabled);
2c08214a
AD
879
880 // base post entry does not exist, create it
881
6322ac79 882 $result = db_query(
2c08214a
AD
883 "INSERT INTO ttrss_entries
884 (title,
885 guid,
886 link,
887 updated,
888 content,
889 content_hash,
890 no_orig_date,
891 date_updated,
892 date_entered,
893 comments,
894 num_comments,
b30abdad 895 plugin_data,
6b461797 896 lang,
2c08214a
AD
897 author)
898 VALUES
899 ('$entry_title',
5e3d5480 900 '$entry_guid_hashed',
2c08214a
AD
901 '$entry_link',
902 '$entry_timestamp_fmt',
903 '$entry_content',
b1840673 904 '$entry_current_hash',
5ba1ddd4 905 false,
2c08214a 906 NOW(),
be574731 907 '$date_feed_processed',
2c08214a
AD
908 '$entry_comments',
909 '$num_comments',
b30abdad 910 '$entry_plugin_data',
6b461797 911 '$entry_language',
2c08214a 912 '$entry_author')");
e8291805 913
2c08214a 914 } else {
2c08214a 915 $base_entry_id = db_fetch_result($result, 0, "id");
2c08214a
AD
916 }
917
918 // now it should exist, if not - bad luck then
919
b1840673 920 $result = db_query("SELECT id FROM ttrss_entries
5e3d5480 921 WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
2c08214a
AD
922
923 $entry_ref_id = 0;
924 $entry_int_id = 0;
925
926 if (db_num_rows($result) == 1) {
927
68cccafc 928 _debug("base guid found, checking for user record", $debug_enabled);
2c08214a 929
2c08214a
AD
930 $ref_id = db_fetch_result($result, 0, "id");
931 $entry_ref_id = $ref_id;
932
5e3d5480
AD
933 /* $stored_guid = db_fetch_result($result, 0, "guid");
934 if ($stored_guid != $entry_guid_hashed) {
68cccafc 935 if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
5e3d5480 936
a42c55f0 937 db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
5e3d5480
AD
938 id = '$ref_id'");
939 } */
940
2c08214a 941 if (find_article_filter($article_filters, "filter")) {
07877caf 942 //db_query("COMMIT"); // close transaction in progress
2c08214a
AD
943 continue;
944 }
945
6de3a1be 946 $score = calculate_article_score($article_filters) + $entry_score_modifier;
2c08214a 947
6de3a1be 948 _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
2c08214a 949
4f186b1f
AD
950 // check for user post link to main table
951
2c08214a 952 $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
5bdcb8fd 953 ref_id = '$ref_id' AND owner_uid = '$owner_uid'";
2c08214a
AD
954
955// if ($_REQUEST["xdebug"]) print "$query\n";
956
a42c55f0 957 $result = db_query($query);
2c08214a
AD
958
959 // okay it doesn't exist - create user entry
960 if (db_num_rows($result) == 0) {
961
68cccafc 962 _debug("user record not found, creating...", $debug_enabled);
2c08214a 963
c9299c28 964 if ($score >= -500 && !find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
2c08214a
AD
965 $unread = 'true';
966 $last_read_qpart = 'NULL';
967 } else {
968 $unread = 'false';
969 $last_read_qpart = 'NOW()';
970 }
971
972 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
973 $marked = 'true';
974 } else {
975 $marked = 'false';
976 }
977
978 if (find_article_filter($article_filters, 'publish')) {
979 $published = 'true';
980 } else {
981 $published = 'false';
982 }
983
7873d588
AD
984 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
985 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
986
6322ac79 987 $result = db_query(
2c08214a
AD
988 "INSERT INTO ttrss_user_entries
989 (ref_id, owner_uid, feed_id, unread, last_read, marked,
7873d588
AD
990 published, score, tag_cache, label_cache, uuid,
991 last_marked, last_published)
2c08214a 992 VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
7873d588
AD
993 $last_read_qpart, $marked, $published, '$score', '', '',
994 '', $last_marked, $last_published)");
2c08214a
AD
995
996 if (PUBSUBHUBBUB_HUB && $published == 'true') {
997 $rss_link = get_self_url_prefix() .
998 "/public.php?op=rss&id=-2&key=" .
a42c55f0 999 get_feed_access_key(-2, false, $owner_uid);
2c08214a 1000
cb3f8773 1001 $p = new pubsubhubbub\publisher\Publisher(PUBSUBHUBBUB_HUB);
2c08214a 1002
5ba1ddd4 1003 /* $pubsub_result = */ $p->publish_update($rss_link);
2c08214a
AD
1004 }
1005
6322ac79 1006 $result = db_query(
2c08214a
AD
1007 "SELECT int_id FROM ttrss_user_entries WHERE
1008 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
1009 feed_id = '$feed' LIMIT 1");
1010
1011 if (db_num_rows($result) == 1) {
1012 $entry_int_id = db_fetch_result($result, 0, "int_id");
1013 }
1014 } else {
68cccafc 1015 _debug("user record FOUND", $debug_enabled);
2c08214a
AD
1016
1017 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
1018 $entry_int_id = db_fetch_result($result, 0, "int_id");
1019 }
1020
68cccafc 1021 _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
2c08214a 1022
e854442e 1023 if (DB_TYPE == "pgsql") {
978989bb 1024 $tsvector_combined = db_escape_string(mb_substr($entry_title . ' ' . strip_tags(str_replace('<', ' <', $entry_content)),
e854442e
AD
1025 0, 1000000));
1026
df659891 1027 $tsvector_qpart = "tsvector_combined = to_tsvector('$feed_language', '$tsvector_combined'),";
e854442e
AD
1028
1029 } else {
1030 $tsvector_qpart = "";
1031 }
1032
b1840673
AD
1033 db_query("UPDATE ttrss_entries
1034 SET title = '$entry_title',
1035 content = '$entry_content',
1036 content_hash = '$entry_current_hash',
1037 updated = '$entry_timestamp_fmt',
e854442e 1038 $tsvector_qpart
b1840673
AD
1039 num_comments = '$num_comments',
1040 plugin_data = '$entry_plugin_data',
1041 author = '$entry_author',
1042 lang = '$entry_language'
1043 WHERE id = '$ref_id'");
1044
59e83455
AD
1045 // update aux data
1046 db_query("UPDATE ttrss_user_entries
1047 SET score = '$score' WHERE ref_id = '$ref_id'");
1048
b1840673 1049 if ($mark_unread_on_update) {
24e6ff5d
AD
1050 _debug("article updated, marking unread as requested.", $debug_enabled);
1051
b1840673
AD
1052 db_query("UPDATE ttrss_user_entries
1053 SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
2c08214a
AD
1054 }
1055 }
1056
07877caf 1057 //db_query("COMMIT");
2c08214a 1058
a29fe121
AD
1059 _debug("assigning labels [other]...", $debug_enabled);
1060
1061 foreach ($article_labels as $label) {
1062 label_add_article($entry_ref_id, $label[1], $owner_uid);
1063 }
1064
1065 _debug("assigning labels [filters]...", $debug_enabled);
2c08214a 1066
a42c55f0 1067 assign_article_to_label_filters($entry_ref_id, $article_filters,
b24504b1 1068 $owner_uid, $article_labels);
2c08214a 1069
68cccafc 1070 _debug("looking for enclosures...", $debug_enabled);
2c08214a
AD
1071
1072 // enclosures
1073
1074 $enclosures = array();
1075
19b3992b 1076 $encs = $item->get_enclosures();
2c08214a 1077
19b3992b
AD
1078 if (is_array($encs)) {
1079 foreach ($encs as $e) {
1080 $e_item = array(
523bd90b 1081 $e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
2c08214a 1082 array_push($enclosures, $e_item);
2c08214a
AD
1083 }
1084 }
1085
2c08214a 1086 if ($debug_enabled) {
68cccafc 1087 _debug("article enclosures:", $debug_enabled);
2c08214a
AD
1088 print_r($enclosures);
1089 }
1090
dab229cd 1091 //db_query("BEGIN");
2c08214a 1092
5c54e683
AD
1093// debugging
1094// db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
1095
2c08214a 1096 foreach ($enclosures as $enc) {
a42c55f0
AD
1097 $enc_url = db_escape_string($enc[0]);
1098 $enc_type = db_escape_string($enc[1]);
1099 $enc_dur = db_escape_string($enc[2]);
5c54e683 1100 $enc_title = db_escape_string($enc[3]);
523bd90b
FE
1101 $enc_width = intval($enc[4]);
1102 $enc_height = intval($enc[5]);
2c08214a 1103
a42c55f0 1104 $result = db_query("SELECT id FROM ttrss_enclosures
2c08214a
AD
1105 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
1106
1107 if (db_num_rows($result) == 0) {
a42c55f0 1108 db_query("INSERT INTO ttrss_enclosures
523bd90b
FE
1109 (content_url, content_type, title, duration, post_id, width, height) VALUES
1110 ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
2c08214a
AD
1111 }
1112 }
1113
dab229cd 1114 //db_query("COMMIT");
2c08214a
AD
1115
1116 // check for manual tags (we have to do it here since they're loaded from filters)
1117
1118 foreach ($article_filters as $f) {
6aff7845 1119 if ($f["type"] == "tag") {
2c08214a 1120
6aff7845 1121 $manual_tags = trim_array(explode(",", $f["param"]));
2c08214a
AD
1122
1123 foreach ($manual_tags as $tag) {
1124 if (tag_is_valid($tag)) {
1125 array_push($entry_tags, $tag);
1126 }
1127 }
1128 }
1129 }
1130
1131 // Skip boring tags
1132
6322ac79 1133 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
2c08214a
AD
1134 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1135
1136 $filtered_tags = array();
1137 $tags_to_cache = array();
1138
1139 if ($entry_tags && is_array($entry_tags)) {
1140 foreach ($entry_tags as $tag) {
1141 if (array_search($tag, $boring_tags) === false) {
1142 array_push($filtered_tags, $tag);
1143 }
1144 }
1145 }
1146
1147 $filtered_tags = array_unique($filtered_tags);
1148
1149 if ($debug_enabled) {
68cccafc 1150 _debug("filtered article tags:", $debug_enabled);
2c08214a
AD
1151 print_r($filtered_tags);
1152 }
1153
1154 // Save article tags in the database
1155
1156 if (count($filtered_tags) > 0) {
1157
dab229cd 1158 //db_query("BEGIN");
2c08214a
AD
1159
1160 foreach ($filtered_tags as $tag) {
1161
1162 $tag = sanitize_tag($tag);
a42c55f0 1163 $tag = db_escape_string($tag);
2c08214a
AD
1164
1165 if (!tag_is_valid($tag)) continue;
1166
a42c55f0 1167 $result = db_query("SELECT id FROM ttrss_tags
2c08214a
AD
1168 WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1169 owner_uid = '$owner_uid' LIMIT 1");
1170
1171 if ($result && db_num_rows($result) == 0) {
1172
a42c55f0 1173 db_query("INSERT INTO ttrss_tags
2c08214a
AD
1174 (owner_uid,tag_name,post_int_id)
1175 VALUES ('$owner_uid','$tag', '$entry_int_id')");
1176 }
1177
1178 array_push($tags_to_cache, $tag);
1179 }
1180
1181 /* update the cache */
1182
1183 $tags_to_cache = array_unique($tags_to_cache);
1184
a42c55f0 1185 $tags_str = db_escape_string(join(",", $tags_to_cache));
2c08214a 1186
a42c55f0 1187 db_query("UPDATE ttrss_user_entries
2c08214a
AD
1188 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1189 AND owner_uid = $owner_uid");
1190
dab229cd 1191 //db_query("COMMIT");
2c08214a
AD
1192 }
1193
68cccafc 1194 _debug("article processed", $debug_enabled);
2c08214a
AD
1195 }
1196
68cccafc 1197 _debug("purging feed...", $debug_enabled);
2c08214a 1198
a42c55f0 1199 purge_feed($feed, 0, $debug_enabled);
2c08214a 1200
a42c55f0 1201 db_query("UPDATE ttrss_feeds
f2798eb6 1202 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
2c08214a 1203
a42c55f0 1204// db_query("COMMIT");
2c08214a
AD
1205
1206 } else {
1207
a42c55f0 1208 $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
2c08214a 1209
4ad04ee2
AD
1210 _debug("fetch error: $error_msg", $debug_enabled);
1211
1212 if (count($rss->errors()) > 1) {
1213 foreach ($rss->errors() as $error) {
1214 _debug("+ $error");
1215 }
1216 }
2c08214a 1217
6322ac79 1218 db_query(
2c08214a 1219 "UPDATE ttrss_feeds SET last_error = '$error_msg',
88edaa93 1220 last_updated = NOW() WHERE id = '$feed'");
2c08214a 1221
88edaa93
AD
1222 unset($rss);
1223 }
2c08214a 1224
68cccafc 1225 _debug("done", $debug_enabled);
88edaa93
AD
1226
1227 return $rss;
2c08214a
AD
1228 }
1229
41bead9b 1230 function cache_media($html, $site_url, $debug) {
3c696512
AD
1231 libxml_use_internal_errors(true);
1232
1233 $charset_hack = '<head>
1234 <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1235 </head>';
1236
1237 $doc = new DOMDocument();
1238 $doc->loadHTML($charset_hack . $html);
1239 $xpath = new DOMXPath($doc);
1240
5edd605a 1241 $entries = $xpath->query('(//img[@src])|(//video/source[@src])');
3c696512
AD
1242
1243 foreach ($entries as $entry) {
5edd605a 1244 if ($entry->hasAttribute('src') && strpos($entry->getAttribute('src'), "data:") !== 0) {
3c696512
AD
1245 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1246
41bead9b 1247 $local_filename = CACHE_DIR . "/images/" . sha1($src);
3c696512 1248
41bead9b 1249 if ($debug) _debug("cache_media: downloading: $src to $local_filename");
3c696512
AD
1250
1251 if (!file_exists($local_filename)) {
1252 $file_content = fetch_file_contents($src);
1253
41bead9b 1254 if ($file_content && strlen($file_content) > _MIN_CACHE_FILE_SIZE) {
3c696512
AD
1255 file_put_contents($local_filename, $file_content);
1256 }
4a27966e
J
1257 } else {
1258 touch($local_filename);
3c696512 1259 }
3c696512
AD
1260 }
1261 }
3c696512
AD
1262 }
1263
a42c55f0 1264 function expire_error_log($debug) {
e2261e17
AD
1265 if ($debug) _debug("Removing old error log entries...");
1266
1267 if (DB_TYPE == "pgsql") {
a42c55f0 1268 db_query("DELETE FROM ttrss_error_log
e2261e17
AD
1269 WHERE created_at < NOW() - INTERVAL '7 days'");
1270 } else {
a42c55f0 1271 db_query("DELETE FROM ttrss_error_log
e2261e17
AD
1272 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1273 }
1274
1275 }
1276
2a91b6ff 1277 function expire_lock_files($debug) {
65465085 1278 //if ($debug) _debug("Removing old lock files...");
2a91b6ff
AD
1279
1280 $num_deleted = 0;
1281
1282 if (is_writable(LOCK_DIRECTORY)) {
1283 $files = glob(LOCK_DIRECTORY . "/*.lock");
1284
1285 if ($files) {
1286 foreach ($files as $file) {
11344971 1287 if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
2a91b6ff
AD
1288 unlink($file);
1289 ++$num_deleted;
1290 }
1291 }
1292 }
1293 }
1294
65465085 1295 if ($debug) _debug("Removed $num_deleted old lock files.");
2a91b6ff
AD
1296 }
1297
3c696512 1298 function expire_cached_files($debug) {
3306daec 1299 foreach (array("simplepie", "images", "export", "upload") as $dir) {
3c696512 1300 $cache_dir = CACHE_DIR . "/$dir";
2c08214a 1301
65465085 1302// if ($debug) _debug("Expiring $cache_dir");
2c08214a 1303
3c696512
AD
1304 $num_deleted = 0;
1305
1306 if (is_writable($cache_dir)) {
1307 $files = glob("$cache_dir/*");
1308
2a91b6ff 1309 if ($files) {
2ab20c31
AD
1310 foreach ($files as $file) {
1311 if (time() - filemtime($file) > 86400*7) {
1312 unlink($file);
3c696512 1313
2ab20c31
AD
1314 ++$num_deleted;
1315 }
3c696512
AD
1316 }
1317 }
2a91b6ff 1318 }
3c696512 1319
65465085 1320 if ($debug) _debug("$cache_dir: removed $num_deleted files.");
3c696512
AD
1321 }
1322 }
2c08214a 1323
a3e0bdcf
AD
1324 /**
1325 * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1326 * Returns the url query as associative array
1327 *
1328 * @param string query
1329 * @return array params
1330 */
1331 function convertUrlQuery($query) {
1332 $queryParts = explode('&', $query);
1333
1334 $params = array();
1335
1336 foreach ($queryParts as $param) {
1337 $item = explode('=', $param);
1338 $params[$item[0]] = $item[1];
1339 }
1340
1341 return $params;
1342 }
92c14e9d 1343
557d86fe 1344 function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags, &$matched_rules = false) {
92c14e9d
AD
1345 $matches = array();
1346
1347 foreach ($filters as $filter) {
1348 $match_any_rule = $filter["match_any_rule"];
a3a896a1 1349 $inverse = $filter["inverse"];
92c14e9d
AD
1350 $filter_match = false;
1351
1352 foreach ($filter["rules"] as $rule) {
1353 $match = false;
ffa1bd7b 1354 $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
a3a896a1 1355 $rule_inverse = $rule["inverse"];
92c14e9d
AD
1356
1357 if (!$reg_exp)
1358 continue;
1359
1360 switch ($rule["type"]) {
1361 case "title":
e3dceca6 1362 $match = @preg_match("/$reg_exp/iu", $title);
92c14e9d
AD
1363 break;
1364 case "content":
d03ae73e
AD
1365 // we don't need to deal with multiline regexps
1366 $content = preg_replace("/[\r\n\t]/", "", $content);
1367
e3dceca6 1368 $match = @preg_match("/$reg_exp/iu", $content);
92c14e9d
AD
1369 break;
1370 case "both":
d03ae73e
AD
1371 // we don't need to deal with multiline regexps
1372 $content = preg_replace("/[\r\n\t]/", "", $content);
1373
e3dceca6 1374 $match = (@preg_match("/$reg_exp/iu", $title) || @preg_match("/$reg_exp/iu", $content));
92c14e9d
AD
1375 break;
1376 case "link":
e3dceca6 1377 $match = @preg_match("/$reg_exp/iu", $link);
92c14e9d
AD
1378 break;
1379 case "author":
e3dceca6 1380 $match = @preg_match("/$reg_exp/iu", $author);
92c14e9d
AD
1381 break;
1382 case "tag":
7b80b5e1 1383 foreach ($tags as $tag) {
e3dceca6 1384 if (@preg_match("/$reg_exp/iu", $tag)) {
7b80b5e1
DK
1385 $match = true;
1386 break;
1387 }
1388 }
92c14e9d
AD
1389 break;
1390 }
1391
a3a896a1
AD
1392 if ($rule_inverse) $match = !$match;
1393
92c14e9d
AD
1394 if ($match_any_rule) {
1395 if ($match) {
1396 $filter_match = true;
1397 break;
1398 }
1399 } else {
1400 $filter_match = $match;
1401 if (!$match) {
1402 break;
1403 }
1404 }
1405 }
1406
a3a896a1
AD
1407 if ($inverse) $filter_match = !$filter_match;
1408
92c14e9d 1409 if ($filter_match) {
557d86fe
AD
1410 if (is_array($matched_rules)) array_push($matched_rules, $rule);
1411
92c14e9d
AD
1412 foreach ($filter["actions"] AS $action) {
1413 array_push($matches, $action);
5e736e45
AD
1414
1415 // if Stop action encountered, perform no further processing
fd3e5e8d 1416 if (isset($action["type"]) && $action["type"] == "stop") return $matches;
92c14e9d
AD
1417 }
1418 }
1419 }
1420
1421 return $matches;
1422 }
1423
1424 function find_article_filter($filters, $filter_name) {
1425 foreach ($filters as $f) {
1426 if ($f["type"] == $filter_name) {
1427 return $f;
1428 };
1429 }
1430 return false;
1431 }
1432
1433 function find_article_filters($filters, $filter_name) {
1434 $results = array();
1435
1436 foreach ($filters as $f) {
1437 if ($f["type"] == $filter_name) {
1438 array_push($results, $f);
1439 };
1440 }
1441 return $results;
1442 }
1443
1444 function calculate_article_score($filters) {
1445 $score = 0;
1446
1447 foreach ($filters as $f) {
1448 if ($f["type"] == "score") {
1449 $score += $f["param"];
1450 };
1451 }
1452 return $score;
1453 }
1454
b24504b1
AD
1455 function labels_contains_caption($labels, $caption) {
1456 foreach ($labels as $label) {
1457 if ($label[1] == $caption) {
1458 return true;
1459 }
1460 }
1461
1462 return false;
1463 }
1464
a42c55f0 1465 function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
92c14e9d
AD
1466 foreach ($filters as $f) {
1467 if ($f["type"] == "label") {
b24504b1 1468 if (!labels_contains_caption($article_labels, $f["param"])) {
a42c55f0 1469 label_add_article($id, $f["param"], $owner_uid);
b24504b1
AD
1470 }
1471 }
92c14e9d
AD
1472 }
1473 }
87764a50 1474
87d7e850
AD
1475 function make_guid_from_title($title) {
1476 return preg_replace("/[ \"\',.:;]/", "-",
1477 mb_strtolower(strip_tags($title), 'utf-8'));
1478 }
1479
d1f3fa97 1480 /* function verify_feed_xml($feed_data) {
ebec81a6
AD
1481 libxml_use_internal_errors(true);
1482 $doc = new DOMDocument();
1483 $doc->loadXML($feed_data);
1484 $error = libxml_get_last_error();
1485 libxml_clear_errors();
1486 return $error;
d1f3fa97 1487 } */
87d7e850 1488
168cf351
AD
1489 function cleanup_counters_cache($debug) {
1490 $result = db_query("DELETE FROM ttrss_counters_cache
1491 WHERE feed_id > 0 AND
1492 (SELECT COUNT(id) FROM ttrss_feeds WHERE
1493 id = feed_id AND
1494 ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
1495 $frows = db_affected_rows($result);
1496
1497 $result = db_query("DELETE FROM ttrss_cat_counters_cache
1498 WHERE feed_id > 0 AND
1499 (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
1500 id = feed_id AND
1501 ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
1502 $crows = db_affected_rows($result);
1503
1504 _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
1505 }
1506
5cbd1fe8
AD
1507 function housekeeping_user($owner_uid) {
1508 $tmph = new PluginHost();
1509
1510 load_user_plugins($owner_uid, $tmph);
1511
1512 $tmph->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
1513 }
1514
e2cf81e2
AD
1515 function housekeeping_common($debug) {
1516 expire_cached_files($debug);
1517 expire_lock_files($debug);
1518 expire_error_log($debug);
1519
1520 $count = update_feedbrowser_cache();
1521 _debug("Feedbrowser updated, $count feeds processed.");
1522
1523 purge_orphans( true);
168cf351 1524 cleanup_counters_cache($debug);
e2cf81e2 1525
9b736a20
AD
1526 //$rc = cleanup_tags( 14, 50000);
1527 //_debug("Cleaned $rc cached tags.");
8e470220 1528
00f22824 1529 PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
e2cf81e2 1530 }
2c08214a 1531?>