]> git.wh0rd.org - tt-rss.git/blame - include/rssfuncs.php
tag_is_valid: check numeric tags properly
[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
fce451a4 137 $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
2c08214a
AD
138 FROM
139 ttrss_feeds, ttrss_users, ttrss_user_prefs
f4ae0f05 140 WHERE
2c08214a 141 ttrss_feeds.owner_uid = ttrss_users.id
f08426e3 142 AND ttrss_user_prefs.profile IS NULL
2c08214a
AD
143 AND ttrss_users.id = ttrss_user_prefs.owner_uid
144 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
145 $login_thresh_qpart $update_limit_qpart
1c4421fc 146 $updstart_thresh_qpart
fce451a4
AD
147 ORDER BY last_updated $query_limit";
148
149 // We search for feed needing update.
150 $result = db_query($query);
2c08214a 151
ca5ff2d9 152 if($debug) _debug(sprintf("Scheduled %d feeds to update...", db_num_rows($result)));
2c08214a
AD
153
154 // Here is a little cache magic in order to minimize risk of double feed updates.
155 $feeds_to_update = array();
156 while ($line = db_fetch_assoc($result)) {
a42c55f0 157 array_push($feeds_to_update, db_escape_string($line['feed_url']));
2c08214a
AD
158 }
159
160 // We update the feed last update started date before anything else.
161 // There is no lag due to feed contents downloads
162 // It prevent an other process to update the same feed.
1c4421fc
AD
163
164 if(count($feeds_to_update) > 0) {
165 $feeds_quoted = array();
166
167 foreach ($feeds_to_update as $feed) {
a42c55f0 168 array_push($feeds_quoted, "'" . db_escape_string($feed) . "'");
1c4421fc
AD
169 }
170
a42c55f0 171 db_query(sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
1c4421fc 172 WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
2c08214a
AD
173 }
174
8292d05b 175 $nf = 0;
2d9c5684 176 $bstarted = microtime(true);
8292d05b 177
2c08214a 178 // For each feed, we call the feed update function.
1c4421fc
AD
179 foreach ($feeds_to_update as $feed) {
180 if($debug) _debug("Base feed: $feed");
2c08214a 181
a42c55f0 182 //update_rss_feed($line["id"], true);
2c08214a 183
1c4421fc
AD
184 // since we have the data cached, we can deal with other feeds with the same url
185
15c762be 186 $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
ee0542ce
AD
187 FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
188 ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
189 ttrss_users.id = ttrss_user_prefs.owner_uid AND
190 ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
f08426e3 191 ttrss_user_prefs.profile IS NULL AND
a42c55f0 192 feed_url = '".db_escape_string($feed)."' AND
ee0542ce
AD
193 (ttrss_feeds.update_interval > 0 OR
194 ttrss_user_prefs.value != '-1')
1c4421fc 195 $login_thresh_qpart
5929a0c1 196 ORDER BY ttrss_feeds.id $query_limit");
1c4421fc
AD
197
198 if (db_num_rows($tmp_result) > 0) {
88edaa93
AD
199 $rss = false;
200
1c4421fc 201 while ($tline = db_fetch_assoc($tmp_result)) {
15c762be 202 if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
f08426e3 203
2d9c5684 204 $fstarted = microtime(true);
f77c7203 205 $rss = update_rss_feed($tline["id"], true, false);
4f71d743 206 _debug_suppress(false);
2d9c5684
AD
207
208 _debug(sprintf(" %.4f (sec)", microtime(true) - $fstarted));
209
8292d05b 210 ++$nf;
1c4421fc
AD
211 }
212 }
2c08214a
AD
213 }
214
2d9c5684
AD
215 if ($nf > 0) {
216 _debug(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
217 microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
218 }
219
f1611683
AD
220 require_once "digest.php";
221
2c08214a 222 // Send feed digests by email if needed.
a42c55f0 223 send_headlines_digests($debug);
2c08214a 224
8292d05b
AD
225 return $nf;
226
2c08214a
AD
227 } // function update_daemon_common
228
6022776d
AD
229 // this is used when subscribing
230 function set_basic_feed_info($feed) {
231
232 $feed = db_escape_string($feed);
233
9bd422c2 234 $result = db_query("SELECT feed_url,auth_pass,auth_login,auth_pass_encrypted
6022776d
AD
235 FROM ttrss_feeds WHERE id = '$feed'");
236
237 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
238 0, "auth_pass_encrypted"));
239
240 $auth_login = db_fetch_result($result, 0, "auth_login");
241 $auth_pass = db_fetch_result($result, 0, "auth_pass");
242
243 if ($auth_pass_encrypted) {
244 require_once "crypt.php";
245 $auth_pass = decrypt_string($auth_pass);
246 }
247
248 $fetch_url = db_fetch_result($result, 0, "feed_url");
249
250 $feed_data = fetch_file_contents($fetch_url, false,
251 $auth_login, $auth_pass, false,
252 FEED_FETCH_TIMEOUT_TIMEOUT,
253 0);
254
255 global $fetch_curl_used;
256
257 if (!$fetch_curl_used) {
258 $tmp = @gzdecode($feed_data);
259
260 if ($tmp) $feed_data = $tmp;
261 }
262
263 $feed_data = trim($feed_data);
264
265 $rss = new FeedParser($feed_data);
266 $rss->init();
267
268 if (!$rss->error()) {
269
270 $result = db_query("SELECT title, site_url FROM ttrss_feeds WHERE id = '$feed'");
271
272 $registered_title = db_fetch_result($result, 0, "title");
273 $orig_site_url = db_fetch_result($result, 0, "site_url");
274
275 $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
276 $feed_title = db_escape_string(mb_substr($rss->get_title(), 0, 199));
277
278 if ($feed_title && (!$registered_title || $registered_title == "[Unknown]")) {
279 db_query("UPDATE ttrss_feeds SET
280 title = '$feed_title' WHERE id = '$feed'");
281 }
282
283 if ($site_url && $orig_site_url != $site_url) {
284 db_query("UPDATE ttrss_feeds SET
285 site_url = '$site_url' WHERE id = '$feed'");
286 }
287 }
288 }
289
87764a50 290 // ignore_daemon is not used
88edaa93 291 function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
2c08214a 292
2c08214a
AD
293 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
294
4f71d743 295 _debug_suppress(!$debug_enabled);
68cccafc 296 _debug("start", $debug_enabled);
2c08214a 297
6bb96beb
AD
298 $result = db_query("SELECT title FROM ttrss_feeds
299 WHERE id = '$feed'");
300 $title = db_fetch_result($result, 0, "title");
301
302 // feed was batch-subscribed or something, we need to get basic info
303 // this is not optimal currently as it fetches stuff separately TODO: optimize
304 if ($title == "[Unknown]") {
305 _debug("setting basic feed info for $feed...");
306 set_basic_feed_info($feed);
307 }
308
a42c55f0 309 $result = db_query("SELECT id,update_interval,auth_login,
5ba1ddd4 310 feed_url,auth_pass,cache_images,
5321e775 311 mark_unread_on_update, owner_uid,
90e5f4f1 312 pubsub_state, auth_pass_encrypted,
df659891 313 feed_language,
90e5f4f1
AD
314 (SELECT max(date_entered) FROM
315 ttrss_entries, ttrss_user_entries where ref_id = id AND feed_id = '$feed') AS last_article_timestamp
87764a50 316 FROM ttrss_feeds WHERE id = '$feed'");
2c08214a
AD
317
318 if (db_num_rows($result) == 0) {
68cccafc 319 _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
2c08214a
AD
320 return false;
321 }
322
90e5f4f1 323 $last_article_timestamp = @strtotime(db_fetch_result($result, 0, "last_article_timestamp"));
5d0d3887
AD
324
325 if (defined('_DISABLE_HTTP_304'))
326 $last_article_timestamp = 0;
327
2c08214a
AD
328 $owner_uid = db_fetch_result($result, 0, "owner_uid");
329 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
330 0, "mark_unread_on_update"));
2c08214a 331 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
044cff2d
AD
332 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
333 0, "auth_pass_encrypted"));
2c08214a 334
a42c55f0 335 db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
2c08214a
AD
336 WHERE id = '$feed'");
337
338 $auth_login = db_fetch_result($result, 0, "auth_login");
339 $auth_pass = db_fetch_result($result, 0, "auth_pass");
340
044cff2d
AD
341 if ($auth_pass_encrypted) {
342 require_once "crypt.php";
343 $auth_pass = decrypt_string($auth_pass);
344 }
345
2c08214a
AD
346 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
347 $fetch_url = db_fetch_result($result, 0, "feed_url");
df659891
AD
348 $feed_language = db_escape_string(mb_strtolower(db_fetch_result($result, 0, "feed_language")));
349 if (!$feed_language) $feed_language = 'english';
2c08214a 350
a42c55f0 351 $feed = db_escape_string($feed);
2c08214a 352
f074ffe9 353 $date_feed_processed = date('Y-m-d H:i');
2c08214a 354
865a3ed6 355 $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
f074ffe9 356
ee65bef4
AD
357 $pluginhost = new PluginHost();
358 $pluginhost->set_debug($debug_enabled);
359 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
360
361 $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
362 $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
363 $pluginhost->load_data();
364
88edaa93
AD
365 if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
366 _debug("using previously initialized parser object");
367 } else {
368 $rss_hash = false;
4f9cbdff 369
88edaa93 370 $force_refetch = isset($_REQUEST["force_refetch"]);
687a4f59 371
95beaa14 372 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
01465325 373 $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, $last_article_timestamp, $auth_login, $auth_pass);
95beaa14
RA
374 }
375
376 // try cache
377 if (!$feed_data &&
378 file_exists($cache_filename) &&
88edaa93
AD
379 is_readable($cache_filename) &&
380 !$auth_login && !$auth_pass &&
381 filemtime($cache_filename) > time() - 30) {
2c08214a 382
15d0528a 383 _debug("using local cache [$cache_filename].", $debug_enabled);
be574731 384
88edaa93 385 @$feed_data = file_get_contents($cache_filename);
52637d3b 386
88edaa93
AD
387 if ($feed_data) {
388 $rss_hash = sha1($feed_data);
389 }
f074ffe9 390
88edaa93
AD
391 } else {
392 _debug("local cache will not be used for this feed", $debug_enabled);
393 }
ee65bef4 394
95beaa14 395 // fetch feed from source
fd687300
AD
396 if (!$feed_data) {
397 _debug("fetching [$fetch_url]...", $debug_enabled);
398 _debug("If-Modified-Since: ".gmdate('D, d M Y H:i:s \G\M\T', $last_article_timestamp), $debug_enabled);
96f98cb0 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,
403 $force_refetch ? 0 : $last_article_timestamp);
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: ";
487f0750 656 print $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
e44280a0 749 // if we allow duplicate posts, we have to continue to
da7e3e91
GP
750 // create the user entries for this feed
751 if (!get_pref("ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
752 continue;
753 }
b1840673
AD
754 }
755
756 _debug("hash differs, applying plugin filters:", $debug_enabled);
757
1ffe3391 758 foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
b1840673
AD
759 _debug("... " . get_class($plugin), $debug_enabled);
760
761 $start = microtime(true);
19b3992b 762 $article = $plugin->hook_article_filter($article);
0084f0d1 763
b1840673
AD
764 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
765
766 $entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
cc85704f
AD
767 }
768
b1840673
AD
769 $entry_plugin_data = db_escape_string($entry_plugin_data);
770
771 _debug("plugin data: $entry_plugin_data", $debug_enabled);
772
35c12dc4
AD
773 // Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
774 if (DB_TYPE == "mysql") {
775 foreach ($article as $k => $v) {
35c37354
AD
776
777 // i guess we'll have to take the risk of 4byte unicode labels & tags here
778 if (!is_array($article[$k])) {
779 $article[$k] = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $v);
780 }
35c12dc4
AD
781 }
782 }
783
b8774453
AD
784 /* Collect article tags here so we could filter by them: */
785
786 $article_filters = get_article_filters($filters, $article["title"],
787 $article["content"], $article["link"], 0, $article["author"],
788 $article["tags"]);
789
790 if ($debug_enabled) {
791 _debug("article filters: ", $debug_enabled);
792 if (count($article_filters) != 0) {
793 print_r($article_filters);
794 }
795 }
796
797 $plugin_filter_names = find_article_filters($article_filters, "plugin");
798 $plugin_filter_actions = $pluginhost->get_filter_actions();
799
800 if (count($plugin_filter_names) > 0) {
801 _debug("applying plugin filter actions...", $debug_enabled);
802
803 foreach ($plugin_filter_names as $pfn) {
804 list($pfclass,$pfaction) = explode(":", $pfn["param"]);
805
806 if (isset($plugin_filter_actions[$pfclass])) {
807 $plugin = $pluginhost->get_plugin($pfclass);
808
809 _debug("... $pfclass: $pfaction", $debug_enabled);
810
811 if ($plugin) {
812 $start = microtime(true);
813 $article = $plugin->hook_article_filter_action($article, $pfaction);
814
815 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
816 } else {
817 _debug("??? $pfclass: plugin object not found.");
818 }
819 } else {
820 _debug("??? $pfclass: filter plugin not registered.");
821 }
822 }
823 }
824
19b3992b 825 $entry_tags = $article["tags"];
a42c55f0
AD
826 $entry_guid = db_escape_string($entry_guid);
827 $entry_title = db_escape_string($article["title"]);
828 $entry_author = db_escape_string($article["author"]);
829 $entry_link = db_escape_string($article["link"]);
f935d98e 830 $entry_content = $article["content"]; // escaped below
c9299c28 831 $entry_force_catchup = $article["force_catchup"];
a29fe121 832 $article_labels = $article["labels"];
6de3a1be 833 $entry_score_modifier = (int) $article["score_modifier"];
3318d324 834 $entry_language = db_escape_string($article["language"]);
a29fe121
AD
835
836 if ($debug_enabled) {
837 _debug("article labels:", $debug_enabled);
838 print_r($article_labels);
839 }
c9299c28
AD
840
841 _debug("force catchup: $entry_force_catchup");
f935d98e 842
0a3fd79b 843 if ($cache_images && is_writable(CACHE_DIR . '/images'))
f0bd8e65 844 cache_images($entry_content, $site_url, $debug_enabled);
0a3fd79b 845
a42c55f0 846 $entry_content = db_escape_string($entry_content, false);
1f45c857 847
07877caf 848 //db_query("BEGIN");
2c08214a 849
a42c55f0 850 $result = db_query("SELECT id FROM ttrss_entries
5e3d5480 851 WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
9e222305 852
2c08214a
AD
853 if (db_num_rows($result) == 0) {
854
68cccafc 855 _debug("base guid [$entry_guid] not found", $debug_enabled);
2c08214a
AD
856
857 // base post entry does not exist, create it
858
6322ac79 859 $result = db_query(
2c08214a
AD
860 "INSERT INTO ttrss_entries
861 (title,
862 guid,
863 link,
864 updated,
865 content,
866 content_hash,
867 no_orig_date,
868 date_updated,
869 date_entered,
870 comments,
871 num_comments,
b30abdad 872 plugin_data,
6b461797 873 lang,
2c08214a
AD
874 author)
875 VALUES
876 ('$entry_title',
5e3d5480 877 '$entry_guid_hashed',
2c08214a
AD
878 '$entry_link',
879 '$entry_timestamp_fmt',
880 '$entry_content',
b1840673 881 '$entry_current_hash',
5ba1ddd4 882 false,
2c08214a 883 NOW(),
be574731 884 '$date_feed_processed',
2c08214a
AD
885 '$entry_comments',
886 '$num_comments',
b30abdad 887 '$entry_plugin_data',
6b461797 888 '$entry_language',
2c08214a 889 '$entry_author')");
e8291805 890
2c08214a 891 } else {
2c08214a 892 $base_entry_id = db_fetch_result($result, 0, "id");
2c08214a
AD
893 }
894
895 // now it should exist, if not - bad luck then
896
b1840673 897 $result = db_query("SELECT id FROM ttrss_entries
5e3d5480 898 WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
2c08214a
AD
899
900 $entry_ref_id = 0;
901 $entry_int_id = 0;
902
903 if (db_num_rows($result) == 1) {
904
68cccafc 905 _debug("base guid found, checking for user record", $debug_enabled);
2c08214a 906
2c08214a
AD
907 $ref_id = db_fetch_result($result, 0, "id");
908 $entry_ref_id = $ref_id;
909
5e3d5480
AD
910 /* $stored_guid = db_fetch_result($result, 0, "guid");
911 if ($stored_guid != $entry_guid_hashed) {
68cccafc 912 if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
5e3d5480 913
a42c55f0 914 db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
5e3d5480
AD
915 id = '$ref_id'");
916 } */
917
2c08214a
AD
918 // check for user post link to main table
919
920 // do we allow duplicate posts with same GUID in different feeds?
a42c55f0 921 if (get_pref("ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
2c08214a
AD
922 $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
923 } else {
924 $dupcheck_qpart = "";
925 }
926
2c08214a 927 if (find_article_filter($article_filters, "filter")) {
07877caf 928 //db_query("COMMIT"); // close transaction in progress
2c08214a
AD
929 continue;
930 }
931
6de3a1be 932 $score = calculate_article_score($article_filters) + $entry_score_modifier;
2c08214a 933
6de3a1be 934 _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
2c08214a
AD
935
936 $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
937 ref_id = '$ref_id' AND owner_uid = '$owner_uid'
938 $dupcheck_qpart";
939
940// if ($_REQUEST["xdebug"]) print "$query\n";
941
a42c55f0 942 $result = db_query($query);
2c08214a
AD
943
944 // okay it doesn't exist - create user entry
945 if (db_num_rows($result) == 0) {
946
68cccafc 947 _debug("user record not found, creating...", $debug_enabled);
2c08214a 948
c9299c28 949 if ($score >= -500 && !find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
2c08214a
AD
950 $unread = 'true';
951 $last_read_qpart = 'NULL';
952 } else {
953 $unread = 'false';
954 $last_read_qpart = 'NOW()';
955 }
956
957 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
958 $marked = 'true';
959 } else {
960 $marked = 'false';
961 }
962
963 if (find_article_filter($article_filters, 'publish')) {
964 $published = 'true';
965 } else {
966 $published = 'false';
967 }
968
2ea9bbfd
AD
969 // N-grams
970
c9299c28 971 /* if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
2ea9bbfd 972
a42c55f0 973 $result = db_query("SELECT COUNT(*) AS similar FROM
2ea9bbfd
AD
974 ttrss_entries,ttrss_user_entries
975 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
976 AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
977 AND owner_uid = $owner_uid");
978
979 $ngram_similar = db_fetch_result($result, 0, "similar");
980
68cccafc 981 _debug("N-gram similar results: $ngram_similar", $debug_enabled);
2ea9bbfd
AD
982
983 if ($ngram_similar > 0) {
984 $unread = 'false';
985 }
c9299c28 986 } */
2ea9bbfd 987
7873d588
AD
988 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
989 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
990
6322ac79 991 $result = db_query(
2c08214a
AD
992 "INSERT INTO ttrss_user_entries
993 (ref_id, owner_uid, feed_id, unread, last_read, marked,
7873d588
AD
994 published, score, tag_cache, label_cache, uuid,
995 last_marked, last_published)
2c08214a 996 VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
7873d588
AD
997 $last_read_qpart, $marked, $published, '$score', '', '',
998 '', $last_marked, $last_published)");
2c08214a
AD
999
1000 if (PUBSUBHUBBUB_HUB && $published == 'true') {
1001 $rss_link = get_self_url_prefix() .
1002 "/public.php?op=rss&id=-2&key=" .
a42c55f0 1003 get_feed_access_key(-2, false, $owner_uid);
2c08214a
AD
1004
1005 $p = new Publisher(PUBSUBHUBBUB_HUB);
1006
5ba1ddd4 1007 /* $pubsub_result = */ $p->publish_update($rss_link);
2c08214a
AD
1008 }
1009
6322ac79 1010 $result = db_query(
2c08214a
AD
1011 "SELECT int_id FROM ttrss_user_entries WHERE
1012 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
1013 feed_id = '$feed' LIMIT 1");
1014
1015 if (db_num_rows($result) == 1) {
1016 $entry_int_id = db_fetch_result($result, 0, "int_id");
1017 }
1018 } else {
68cccafc 1019 _debug("user record FOUND", $debug_enabled);
2c08214a
AD
1020
1021 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
1022 $entry_int_id = db_fetch_result($result, 0, "int_id");
1023 }
1024
68cccafc 1025 _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
2c08214a 1026
e854442e 1027 if (DB_TYPE == "pgsql") {
978989bb 1028 $tsvector_combined = db_escape_string(mb_substr($entry_title . ' ' . strip_tags(str_replace('<', ' <', $entry_content)),
e854442e
AD
1029 0, 1000000));
1030
df659891 1031 $tsvector_qpart = "tsvector_combined = to_tsvector('$feed_language', '$tsvector_combined'),";
e854442e
AD
1032
1033 } else {
1034 $tsvector_qpart = "";
1035 }
1036
b1840673
AD
1037 db_query("UPDATE ttrss_entries
1038 SET title = '$entry_title',
1039 content = '$entry_content',
1040 content_hash = '$entry_current_hash',
1041 updated = '$entry_timestamp_fmt',
e854442e 1042 $tsvector_qpart
b1840673
AD
1043 num_comments = '$num_comments',
1044 plugin_data = '$entry_plugin_data',
1045 author = '$entry_author',
1046 lang = '$entry_language'
1047 WHERE id = '$ref_id'");
1048
59e83455
AD
1049 // update aux data
1050 db_query("UPDATE ttrss_user_entries
1051 SET score = '$score' WHERE ref_id = '$ref_id'");
1052
b1840673 1053 if ($mark_unread_on_update) {
24e6ff5d
AD
1054 _debug("article updated, marking unread as requested.", $debug_enabled);
1055
b1840673
AD
1056 db_query("UPDATE ttrss_user_entries
1057 SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
2c08214a
AD
1058 }
1059 }
1060
07877caf 1061 //db_query("COMMIT");
2c08214a 1062
a29fe121
AD
1063 _debug("assigning labels [other]...", $debug_enabled);
1064
1065 foreach ($article_labels as $label) {
1066 label_add_article($entry_ref_id, $label[1], $owner_uid);
1067 }
1068
1069 _debug("assigning labels [filters]...", $debug_enabled);
2c08214a 1070
a42c55f0 1071 assign_article_to_label_filters($entry_ref_id, $article_filters,
b24504b1 1072 $owner_uid, $article_labels);
2c08214a 1073
68cccafc 1074 _debug("looking for enclosures...", $debug_enabled);
2c08214a
AD
1075
1076 // enclosures
1077
1078 $enclosures = array();
1079
19b3992b 1080 $encs = $item->get_enclosures();
2c08214a 1081
19b3992b
AD
1082 if (is_array($encs)) {
1083 foreach ($encs as $e) {
1084 $e_item = array(
523bd90b 1085 $e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
2c08214a 1086 array_push($enclosures, $e_item);
2c08214a
AD
1087 }
1088 }
1089
2c08214a 1090 if ($debug_enabled) {
68cccafc 1091 _debug("article enclosures:", $debug_enabled);
2c08214a
AD
1092 print_r($enclosures);
1093 }
1094
dab229cd 1095 //db_query("BEGIN");
2c08214a 1096
5c54e683
AD
1097// debugging
1098// db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
1099
2c08214a 1100 foreach ($enclosures as $enc) {
a42c55f0
AD
1101 $enc_url = db_escape_string($enc[0]);
1102 $enc_type = db_escape_string($enc[1]);
1103 $enc_dur = db_escape_string($enc[2]);
5c54e683 1104 $enc_title = db_escape_string($enc[3]);
523bd90b
FE
1105 $enc_width = intval($enc[4]);
1106 $enc_height = intval($enc[5]);
2c08214a 1107
a42c55f0 1108 $result = db_query("SELECT id FROM ttrss_enclosures
2c08214a
AD
1109 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
1110
1111 if (db_num_rows($result) == 0) {
a42c55f0 1112 db_query("INSERT INTO ttrss_enclosures
523bd90b
FE
1113 (content_url, content_type, title, duration, post_id, width, height) VALUES
1114 ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
2c08214a
AD
1115 }
1116 }
1117
dab229cd 1118 //db_query("COMMIT");
2c08214a
AD
1119
1120 // check for manual tags (we have to do it here since they're loaded from filters)
1121
1122 foreach ($article_filters as $f) {
6aff7845 1123 if ($f["type"] == "tag") {
2c08214a 1124
6aff7845 1125 $manual_tags = trim_array(explode(",", $f["param"]));
2c08214a
AD
1126
1127 foreach ($manual_tags as $tag) {
1128 if (tag_is_valid($tag)) {
1129 array_push($entry_tags, $tag);
1130 }
1131 }
1132 }
1133 }
1134
1135 // Skip boring tags
1136
6322ac79 1137 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
2c08214a
AD
1138 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1139
1140 $filtered_tags = array();
1141 $tags_to_cache = array();
1142
1143 if ($entry_tags && is_array($entry_tags)) {
1144 foreach ($entry_tags as $tag) {
1145 if (array_search($tag, $boring_tags) === false) {
1146 array_push($filtered_tags, $tag);
1147 }
1148 }
1149 }
1150
1151 $filtered_tags = array_unique($filtered_tags);
1152
1153 if ($debug_enabled) {
68cccafc 1154 _debug("filtered article tags:", $debug_enabled);
2c08214a
AD
1155 print_r($filtered_tags);
1156 }
1157
1158 // Save article tags in the database
1159
1160 if (count($filtered_tags) > 0) {
1161
dab229cd 1162 //db_query("BEGIN");
2c08214a
AD
1163
1164 foreach ($filtered_tags as $tag) {
1165
1166 $tag = sanitize_tag($tag);
a42c55f0 1167 $tag = db_escape_string($tag);
2c08214a
AD
1168
1169 if (!tag_is_valid($tag)) continue;
1170
a42c55f0 1171 $result = db_query("SELECT id FROM ttrss_tags
2c08214a
AD
1172 WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1173 owner_uid = '$owner_uid' LIMIT 1");
1174
1175 if ($result && db_num_rows($result) == 0) {
1176
a42c55f0 1177 db_query("INSERT INTO ttrss_tags
2c08214a
AD
1178 (owner_uid,tag_name,post_int_id)
1179 VALUES ('$owner_uid','$tag', '$entry_int_id')");
1180 }
1181
1182 array_push($tags_to_cache, $tag);
1183 }
1184
1185 /* update the cache */
1186
1187 $tags_to_cache = array_unique($tags_to_cache);
1188
a42c55f0 1189 $tags_str = db_escape_string(join(",", $tags_to_cache));
2c08214a 1190
a42c55f0 1191 db_query("UPDATE ttrss_user_entries
2c08214a
AD
1192 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1193 AND owner_uid = $owner_uid");
1194
dab229cd 1195 //db_query("COMMIT");
2c08214a
AD
1196 }
1197
68cccafc 1198 _debug("article processed", $debug_enabled);
2c08214a
AD
1199 }
1200
68cccafc 1201 _debug("purging feed...", $debug_enabled);
2c08214a 1202
a42c55f0 1203 purge_feed($feed, 0, $debug_enabled);
2c08214a 1204
a42c55f0 1205 db_query("UPDATE ttrss_feeds
f2798eb6 1206 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
2c08214a 1207
a42c55f0 1208// db_query("COMMIT");
2c08214a
AD
1209
1210 } else {
1211
a42c55f0 1212 $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
2c08214a 1213
4ad04ee2
AD
1214 _debug("fetch error: $error_msg", $debug_enabled);
1215
1216 if (count($rss->errors()) > 1) {
1217 foreach ($rss->errors() as $error) {
1218 _debug("+ $error");
1219 }
1220 }
2c08214a 1221
6322ac79 1222 db_query(
2c08214a 1223 "UPDATE ttrss_feeds SET last_error = '$error_msg',
88edaa93 1224 last_updated = NOW() WHERE id = '$feed'");
2c08214a 1225
88edaa93
AD
1226 unset($rss);
1227 }
2c08214a 1228
68cccafc 1229 _debug("done", $debug_enabled);
88edaa93
AD
1230
1231 return $rss;
2c08214a
AD
1232 }
1233
3c696512 1234 function cache_images($html, $site_url, $debug) {
3c696512
AD
1235 libxml_use_internal_errors(true);
1236
1237 $charset_hack = '<head>
1238 <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1239 </head>';
1240
1241 $doc = new DOMDocument();
1242 $doc->loadHTML($charset_hack . $html);
1243 $xpath = new DOMXPath($doc);
1244
1245 $entries = $xpath->query('(//img[@src])');
1246
1247 foreach ($entries as $entry) {
1248 if ($entry->hasAttribute('src')) {
1249 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1250
1251 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1252
1253 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1254
1255 if (!file_exists($local_filename)) {
1256 $file_content = fetch_file_contents($src);
1257
a3d50184 1258 if ($file_content && strlen($file_content) > _MIN_CACHE_IMAGE_SIZE) {
3c696512
AD
1259 file_put_contents($local_filename, $file_content);
1260 }
4a27966e
J
1261 } else {
1262 touch($local_filename);
3c696512 1263 }
3c696512
AD
1264 }
1265 }
3c696512
AD
1266 }
1267
a42c55f0 1268 function expire_error_log($debug) {
e2261e17
AD
1269 if ($debug) _debug("Removing old error log entries...");
1270
1271 if (DB_TYPE == "pgsql") {
a42c55f0 1272 db_query("DELETE FROM ttrss_error_log
e2261e17
AD
1273 WHERE created_at < NOW() - INTERVAL '7 days'");
1274 } else {
a42c55f0 1275 db_query("DELETE FROM ttrss_error_log
e2261e17
AD
1276 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1277 }
1278
1279 }
1280
2a91b6ff 1281 function expire_lock_files($debug) {
65465085 1282 //if ($debug) _debug("Removing old lock files...");
2a91b6ff
AD
1283
1284 $num_deleted = 0;
1285
1286 if (is_writable(LOCK_DIRECTORY)) {
1287 $files = glob(LOCK_DIRECTORY . "/*.lock");
1288
1289 if ($files) {
1290 foreach ($files as $file) {
11344971 1291 if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
2a91b6ff
AD
1292 unlink($file);
1293 ++$num_deleted;
1294 }
1295 }
1296 }
1297 }
1298
65465085 1299 if ($debug) _debug("Removed $num_deleted old lock files.");
2a91b6ff
AD
1300 }
1301
3c696512 1302 function expire_cached_files($debug) {
3306daec 1303 foreach (array("simplepie", "images", "export", "upload") as $dir) {
3c696512 1304 $cache_dir = CACHE_DIR . "/$dir";
2c08214a 1305
65465085 1306// if ($debug) _debug("Expiring $cache_dir");
2c08214a 1307
3c696512
AD
1308 $num_deleted = 0;
1309
1310 if (is_writable($cache_dir)) {
1311 $files = glob("$cache_dir/*");
1312
2a91b6ff 1313 if ($files) {
2ab20c31
AD
1314 foreach ($files as $file) {
1315 if (time() - filemtime($file) > 86400*7) {
1316 unlink($file);
3c696512 1317
2ab20c31
AD
1318 ++$num_deleted;
1319 }
3c696512
AD
1320 }
1321 }
2a91b6ff 1322 }
3c696512 1323
65465085 1324 if ($debug) _debug("$cache_dir: removed $num_deleted files.");
3c696512
AD
1325 }
1326 }
2c08214a 1327
a3e0bdcf
AD
1328 /**
1329 * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1330 * Returns the url query as associative array
1331 *
1332 * @param string query
1333 * @return array params
1334 */
1335 function convertUrlQuery($query) {
1336 $queryParts = explode('&', $query);
1337
1338 $params = array();
1339
1340 foreach ($queryParts as $param) {
1341 $item = explode('=', $param);
1342 $params[$item[0]] = $item[1];
1343 }
1344
1345 return $params;
1346 }
92c14e9d 1347
48cb2536 1348 function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
92c14e9d
AD
1349 $matches = array();
1350
1351 foreach ($filters as $filter) {
1352 $match_any_rule = $filter["match_any_rule"];
a3a896a1 1353 $inverse = $filter["inverse"];
92c14e9d
AD
1354 $filter_match = false;
1355
1356 foreach ($filter["rules"] as $rule) {
1357 $match = false;
ffa1bd7b 1358 $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
a3a896a1 1359 $rule_inverse = $rule["inverse"];
92c14e9d
AD
1360
1361 if (!$reg_exp)
1362 continue;
1363
1364 switch ($rule["type"]) {
1365 case "title":
1366 $match = @preg_match("/$reg_exp/i", $title);
1367 break;
1368 case "content":
d03ae73e
AD
1369 // we don't need to deal with multiline regexps
1370 $content = preg_replace("/[\r\n\t]/", "", $content);
1371
92c14e9d
AD
1372 $match = @preg_match("/$reg_exp/i", $content);
1373 break;
1374 case "both":
d03ae73e
AD
1375 // we don't need to deal with multiline regexps
1376 $content = preg_replace("/[\r\n\t]/", "", $content);
1377
72d1d067 1378 $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
92c14e9d
AD
1379 break;
1380 case "link":
1381 $match = @preg_match("/$reg_exp/i", $link);
1382 break;
1383 case "author":
1384 $match = @preg_match("/$reg_exp/i", $author);
1385 break;
1386 case "tag":
7b80b5e1
DK
1387 foreach ($tags as $tag) {
1388 if (@preg_match("/$reg_exp/i", $tag)) {
1389 $match = true;
1390 break;
1391 }
1392 }
92c14e9d
AD
1393 break;
1394 }
1395
a3a896a1
AD
1396 if ($rule_inverse) $match = !$match;
1397
92c14e9d
AD
1398 if ($match_any_rule) {
1399 if ($match) {
1400 $filter_match = true;
1401 break;
1402 }
1403 } else {
1404 $filter_match = $match;
1405 if (!$match) {
1406 break;
1407 }
1408 }
1409 }
1410
a3a896a1
AD
1411 if ($inverse) $filter_match = !$filter_match;
1412
92c14e9d
AD
1413 if ($filter_match) {
1414 foreach ($filter["actions"] AS $action) {
1415 array_push($matches, $action);
5e736e45
AD
1416
1417 // if Stop action encountered, perform no further processing
fd3e5e8d 1418 if (isset($action["type"]) && $action["type"] == "stop") return $matches;
92c14e9d
AD
1419 }
1420 }
1421 }
1422
1423 return $matches;
1424 }
1425
1426 function find_article_filter($filters, $filter_name) {
1427 foreach ($filters as $f) {
1428 if ($f["type"] == $filter_name) {
1429 return $f;
1430 };
1431 }
1432 return false;
1433 }
1434
1435 function find_article_filters($filters, $filter_name) {
1436 $results = array();
1437
1438 foreach ($filters as $f) {
1439 if ($f["type"] == $filter_name) {
1440 array_push($results, $f);
1441 };
1442 }
1443 return $results;
1444 }
1445
1446 function calculate_article_score($filters) {
1447 $score = 0;
1448
1449 foreach ($filters as $f) {
1450 if ($f["type"] == "score") {
1451 $score += $f["param"];
1452 };
1453 }
1454 return $score;
1455 }
1456
b24504b1
AD
1457 function labels_contains_caption($labels, $caption) {
1458 foreach ($labels as $label) {
1459 if ($label[1] == $caption) {
1460 return true;
1461 }
1462 }
1463
1464 return false;
1465 }
1466
a42c55f0 1467 function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
92c14e9d
AD
1468 foreach ($filters as $f) {
1469 if ($f["type"] == "label") {
b24504b1 1470 if (!labels_contains_caption($article_labels, $f["param"])) {
a42c55f0 1471 label_add_article($id, $f["param"], $owner_uid);
b24504b1
AD
1472 }
1473 }
92c14e9d
AD
1474 }
1475 }
87764a50 1476
87d7e850
AD
1477 function make_guid_from_title($title) {
1478 return preg_replace("/[ \"\',.:;]/", "-",
1479 mb_strtolower(strip_tags($title), 'utf-8'));
1480 }
1481
d1f3fa97 1482 /* function verify_feed_xml($feed_data) {
ebec81a6
AD
1483 libxml_use_internal_errors(true);
1484 $doc = new DOMDocument();
1485 $doc->loadXML($feed_data);
1486 $error = libxml_get_last_error();
1487 libxml_clear_errors();
1488 return $error;
d1f3fa97 1489 } */
87d7e850 1490
168cf351
AD
1491 function cleanup_counters_cache($debug) {
1492 $result = db_query("DELETE FROM ttrss_counters_cache
1493 WHERE feed_id > 0 AND
1494 (SELECT COUNT(id) FROM ttrss_feeds WHERE
1495 id = feed_id AND
1496 ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
1497 $frows = db_affected_rows($result);
1498
1499 $result = db_query("DELETE FROM ttrss_cat_counters_cache
1500 WHERE feed_id > 0 AND
1501 (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
1502 id = feed_id AND
1503 ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
1504 $crows = db_affected_rows($result);
1505
1506 _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
1507 }
1508
e2cf81e2
AD
1509 function housekeeping_common($debug) {
1510 expire_cached_files($debug);
1511 expire_lock_files($debug);
1512 expire_error_log($debug);
1513
1514 $count = update_feedbrowser_cache();
1515 _debug("Feedbrowser updated, $count feeds processed.");
1516
1517 purge_orphans( true);
168cf351 1518 cleanup_counters_cache($debug);
e2cf81e2
AD
1519 $rc = cleanup_tags( 14, 50000);
1520
1521 _debug("Cleaned $rc cached tags.");
8e470220 1522
00f22824 1523 PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
8e470220 1524
e2cf81e2 1525 }
2c08214a 1526?>