]> git.wh0rd.org - tt-rss.git/blame - include/rssfuncs.php
catchup maybe add workaround for mysql 5.7 derived_merge
[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)) {
12 $tmp .= sha1("$k:" . (is_array($v) ? implode(",", $v) : $v));
13 }
14 }
15
eb16bd9f 16 return sha1(implode(",", $pluginhost->get_plugin_names()) . $tmp);
b1840673
AD
17 }
18
6322ac79 19 function update_feedbrowser_cache() {
79178062 20
a42c55f0 21 $result = db_query("SELECT feed_url, site_url, title, COUNT(id) AS subscribers
79178062
AD
22 FROM ttrss_feeds WHERE (SELECT COUNT(id) = 0 FROM ttrss_feeds AS tf
23 WHERE tf.feed_url = ttrss_feeds.feed_url
24 AND (private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%'))
25 GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
26
a42c55f0 27 db_query("BEGIN");
79178062 28
a42c55f0 29 db_query("DELETE FROM ttrss_feedbrowser_cache");
79178062
AD
30
31 $count = 0;
32
33 while ($line = db_fetch_assoc($result)) {
a42c55f0
AD
34 $subscribers = db_escape_string($line["subscribers"]);
35 $feed_url = db_escape_string($line["feed_url"]);
36 $title = db_escape_string($line["title"]);
37 $site_url = db_escape_string($line["site_url"]);
79178062 38
a42c55f0 39 $tmp_result = db_query("SELECT subscribers FROM
79178062
AD
40 ttrss_feedbrowser_cache WHERE feed_url = '$feed_url'");
41
42 if (db_num_rows($tmp_result) == 0) {
43
a42c55f0 44 db_query("INSERT INTO ttrss_feedbrowser_cache
79178062
AD
45 (feed_url, site_url, title, subscribers) VALUES ('$feed_url',
46 '$site_url', '$title', '$subscribers')");
47
48 ++$count;
49
50 }
51
52 }
53
a42c55f0 54 db_query("COMMIT");
79178062
AD
55
56 return $count;
57
58 }
59
60
2c08214a
AD
61 /**
62 * Update a feed batch.
63 * Used by daemons to update n feeds by run.
64 * Only update feed needing a update, and not being processed
65 * by another process.
66 *
67 * @param mixed $link Database link
68 * @param integer $limit Maximum number of feeds in update batch. Default to DAEMON_FEED_LIMIT.
69 * @param boolean $from_http Set to true if you call this function from http to disable cli specific code.
70 * @param boolean $debug Set to false to disable debug output. Default to true.
71 * @return void
72 */
a42c55f0 73 function update_daemon_common($limit = DAEMON_FEED_LIMIT, $from_http = false, $debug = true) {
2c08214a
AD
74 // Process all other feeds using last_updated and interval parameters
75
6322ac79 76 $schema_version = get_schema_version();
857efe49
AD
77
78 if ($schema_version != SCHEMA_VERSION) {
79 die("Schema version is wrong, please upgrade the database.\n");
80 }
81
61c1812f
AD
82 define('PREFS_NO_CACHE', true);
83
2c08214a 84 // Test if the user has loggued in recently. If not, it does not update its feeds.
09e8bdfd 85 if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
2c08214a
AD
86 if (DB_TYPE == "pgsql") {
87 $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
88 } else {
89 $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
90 }
91 } else {
92 $login_thresh_qpart = "";
93 }
94
537eca40 95 // Test if the feed need a update (update interval exceeded).
2c08214a
AD
96 if (DB_TYPE == "pgsql") {
97 $update_limit_qpart = "AND ((
98 ttrss_feeds.update_interval = 0
ee0542ce 99 AND ttrss_user_prefs.value != '-1'
2c08214a
AD
100 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
101 ) OR (
102 ttrss_feeds.update_interval > 0
103 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
f08426e3
AD
104 ) OR (ttrss_feeds.last_updated IS NULL
105 AND ttrss_user_prefs.value != '-1')
106 OR (last_updated = '1970-01-01 00:00:00'
107 AND ttrss_user_prefs.value != '-1'))";
2c08214a
AD
108 } else {
109 $update_limit_qpart = "AND ((
110 ttrss_feeds.update_interval = 0
ee0542ce 111 AND ttrss_user_prefs.value != '-1'
2c08214a
AD
112 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
113 ) OR (
114 ttrss_feeds.update_interval > 0
115 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
f08426e3
AD
116 ) OR (ttrss_feeds.last_updated IS NULL
117 AND ttrss_user_prefs.value != '-1')
118 OR (last_updated = '1970-01-01 00:00:00'
119 AND ttrss_user_prefs.value != '-1'))";
2c08214a
AD
120 }
121
122 // Test if feed is currently being updated by another process.
123 if (DB_TYPE == "pgsql") {
566417c4 124 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
2c08214a 125 } else {
566417c4 126 $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
127 }
128
129 // Test if there is a limit to number of updated feeds
130 $query_limit = "";
131 if($limit) $query_limit = sprintf("LIMIT %d", $limit);
132
fce451a4 133 $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
2c08214a
AD
134 FROM
135 ttrss_feeds, ttrss_users, ttrss_user_prefs
f4ae0f05 136 WHERE
2c08214a 137 ttrss_feeds.owner_uid = ttrss_users.id
f08426e3 138 AND ttrss_user_prefs.profile IS NULL
2c08214a
AD
139 AND ttrss_users.id = ttrss_user_prefs.owner_uid
140 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
141 $login_thresh_qpart $update_limit_qpart
1c4421fc 142 $updstart_thresh_qpart
fce451a4
AD
143 ORDER BY last_updated $query_limit";
144
145 // We search for feed needing update.
146 $result = db_query($query);
2c08214a 147
ca5ff2d9 148 if($debug) _debug(sprintf("Scheduled %d feeds to update...", db_num_rows($result)));
2c08214a
AD
149
150 // Here is a little cache magic in order to minimize risk of double feed updates.
151 $feeds_to_update = array();
152 while ($line = db_fetch_assoc($result)) {
a42c55f0 153 array_push($feeds_to_update, db_escape_string($line['feed_url']));
2c08214a
AD
154 }
155
156 // We update the feed last update started date before anything else.
157 // There is no lag due to feed contents downloads
158 // It prevent an other process to update the same feed.
1c4421fc
AD
159
160 if(count($feeds_to_update) > 0) {
161 $feeds_quoted = array();
162
163 foreach ($feeds_to_update as $feed) {
a42c55f0 164 array_push($feeds_quoted, "'" . db_escape_string($feed) . "'");
1c4421fc
AD
165 }
166
a42c55f0 167 db_query(sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
1c4421fc 168 WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
2c08214a
AD
169 }
170
8292d05b 171 $nf = 0;
2d9c5684 172 $bstarted = microtime(true);
8292d05b 173
2c08214a 174 // For each feed, we call the feed update function.
1c4421fc
AD
175 foreach ($feeds_to_update as $feed) {
176 if($debug) _debug("Base feed: $feed");
2c08214a 177
a42c55f0 178 //update_rss_feed($line["id"], true);
2c08214a 179
1c4421fc
AD
180 // since we have the data cached, we can deal with other feeds with the same url
181
15c762be 182 $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
ee0542ce
AD
183 FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
184 ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
185 ttrss_users.id = ttrss_user_prefs.owner_uid AND
186 ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
f08426e3 187 ttrss_user_prefs.profile IS NULL AND
a42c55f0 188 feed_url = '".db_escape_string($feed)."' AND
ee0542ce
AD
189 (ttrss_feeds.update_interval > 0 OR
190 ttrss_user_prefs.value != '-1')
1c4421fc 191 $login_thresh_qpart
5929a0c1 192 ORDER BY ttrss_feeds.id $query_limit");
1c4421fc
AD
193
194 if (db_num_rows($tmp_result) > 0) {
88edaa93
AD
195 $rss = false;
196
1c4421fc 197 while ($tline = db_fetch_assoc($tmp_result)) {
15c762be 198 if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
f08426e3 199
2d9c5684 200 $fstarted = microtime(true);
f77c7203 201 $rss = update_rss_feed($tline["id"], true, false);
4f71d743 202 _debug_suppress(false);
2d9c5684
AD
203
204 _debug(sprintf(" %.4f (sec)", microtime(true) - $fstarted));
205
8292d05b 206 ++$nf;
1c4421fc
AD
207 }
208 }
2c08214a
AD
209 }
210
2d9c5684
AD
211 if ($nf > 0) {
212 _debug(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
213 microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
214 }
215
f1611683
AD
216 require_once "digest.php";
217
2c08214a 218 // Send feed digests by email if needed.
a42c55f0 219 send_headlines_digests($debug);
2c08214a 220
8292d05b
AD
221 return $nf;
222
2c08214a
AD
223 } // function update_daemon_common
224
6022776d
AD
225 // this is used when subscribing
226 function set_basic_feed_info($feed) {
227
228 $feed = db_escape_string($feed);
229
230 $result = db_query("SELECT feed_url,auth_pass,auth_pass_encrypted
231 FROM ttrss_feeds WHERE id = '$feed'");
232
233 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
234 0, "auth_pass_encrypted"));
235
236 $auth_login = db_fetch_result($result, 0, "auth_login");
237 $auth_pass = db_fetch_result($result, 0, "auth_pass");
238
239 if ($auth_pass_encrypted) {
240 require_once "crypt.php";
241 $auth_pass = decrypt_string($auth_pass);
242 }
243
244 $fetch_url = db_fetch_result($result, 0, "feed_url");
245
246 $feed_data = fetch_file_contents($fetch_url, false,
247 $auth_login, $auth_pass, false,
248 FEED_FETCH_TIMEOUT_TIMEOUT,
249 0);
250
251 global $fetch_curl_used;
252
253 if (!$fetch_curl_used) {
254 $tmp = @gzdecode($feed_data);
255
256 if ($tmp) $feed_data = $tmp;
257 }
258
259 $feed_data = trim($feed_data);
260
261 $rss = new FeedParser($feed_data);
262 $rss->init();
263
264 if (!$rss->error()) {
265
266 $result = db_query("SELECT title, site_url FROM ttrss_feeds WHERE id = '$feed'");
267
268 $registered_title = db_fetch_result($result, 0, "title");
269 $orig_site_url = db_fetch_result($result, 0, "site_url");
270
271 $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
272 $feed_title = db_escape_string(mb_substr($rss->get_title(), 0, 199));
273
274 if ($feed_title && (!$registered_title || $registered_title == "[Unknown]")) {
275 db_query("UPDATE ttrss_feeds SET
276 title = '$feed_title' WHERE id = '$feed'");
277 }
278
279 if ($site_url && $orig_site_url != $site_url) {
280 db_query("UPDATE ttrss_feeds SET
281 site_url = '$site_url' WHERE id = '$feed'");
282 }
283 }
284 }
285
87764a50 286 // ignore_daemon is not used
88edaa93 287 function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
2c08214a 288
2c08214a
AD
289 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
290
4f71d743 291 _debug_suppress(!$debug_enabled);
68cccafc 292 _debug("start", $debug_enabled);
2c08214a 293
6bb96beb
AD
294 $result = db_query("SELECT title FROM ttrss_feeds
295 WHERE id = '$feed'");
296 $title = db_fetch_result($result, 0, "title");
297
298 // feed was batch-subscribed or something, we need to get basic info
299 // this is not optimal currently as it fetches stuff separately TODO: optimize
300 if ($title == "[Unknown]") {
301 _debug("setting basic feed info for $feed...");
302 set_basic_feed_info($feed);
303 }
304
a42c55f0 305 $result = db_query("SELECT id,update_interval,auth_login,
5ba1ddd4 306 feed_url,auth_pass,cache_images,
5321e775 307 mark_unread_on_update, owner_uid,
90e5f4f1 308 pubsub_state, auth_pass_encrypted,
df659891 309 feed_language,
90e5f4f1
AD
310 (SELECT max(date_entered) FROM
311 ttrss_entries, ttrss_user_entries where ref_id = id AND feed_id = '$feed') AS last_article_timestamp
87764a50 312 FROM ttrss_feeds WHERE id = '$feed'");
2c08214a
AD
313
314 if (db_num_rows($result) == 0) {
68cccafc 315 _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
2c08214a
AD
316 return false;
317 }
318
90e5f4f1 319 $last_article_timestamp = @strtotime(db_fetch_result($result, 0, "last_article_timestamp"));
5d0d3887
AD
320
321 if (defined('_DISABLE_HTTP_304'))
322 $last_article_timestamp = 0;
323
2c08214a
AD
324 $owner_uid = db_fetch_result($result, 0, "owner_uid");
325 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
326 0, "mark_unread_on_update"));
2c08214a 327 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
044cff2d
AD
328 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
329 0, "auth_pass_encrypted"));
2c08214a 330
a42c55f0 331 db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
2c08214a
AD
332 WHERE id = '$feed'");
333
334 $auth_login = db_fetch_result($result, 0, "auth_login");
335 $auth_pass = db_fetch_result($result, 0, "auth_pass");
336
044cff2d
AD
337 if ($auth_pass_encrypted) {
338 require_once "crypt.php";
339 $auth_pass = decrypt_string($auth_pass);
340 }
341
2c08214a
AD
342 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
343 $fetch_url = db_fetch_result($result, 0, "feed_url");
df659891
AD
344 $feed_language = db_escape_string(mb_strtolower(db_fetch_result($result, 0, "feed_language")));
345 if (!$feed_language) $feed_language = 'english';
2c08214a 346
a42c55f0 347 $feed = db_escape_string($feed);
2c08214a 348
f074ffe9 349 $date_feed_processed = date('Y-m-d H:i');
2c08214a 350
865a3ed6 351 $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
f074ffe9 352
ee65bef4
AD
353 $pluginhost = new PluginHost();
354 $pluginhost->set_debug($debug_enabled);
355 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
356
357 $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
358 $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
359 $pluginhost->load_data();
360
88edaa93
AD
361 if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
362 _debug("using previously initialized parser object");
363 } else {
364 $rss_hash = false;
4f9cbdff 365
88edaa93 366 $force_refetch = isset($_REQUEST["force_refetch"]);
687a4f59 367
95beaa14 368 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
01465325 369 $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, $last_article_timestamp, $auth_login, $auth_pass);
95beaa14
RA
370 }
371
372 // try cache
373 if (!$feed_data &&
374 file_exists($cache_filename) &&
88edaa93
AD
375 is_readable($cache_filename) &&
376 !$auth_login && !$auth_pass &&
377 filemtime($cache_filename) > time() - 30) {
2c08214a 378
15d0528a 379 _debug("using local cache [$cache_filename].", $debug_enabled);
be574731 380
88edaa93 381 @$feed_data = file_get_contents($cache_filename);
52637d3b 382
88edaa93
AD
383 if ($feed_data) {
384 $rss_hash = sha1($feed_data);
385 }
f074ffe9 386
88edaa93
AD
387 } else {
388 _debug("local cache will not be used for this feed", $debug_enabled);
389 }
ee65bef4 390
95beaa14 391 // fetch feed from source
fd687300
AD
392 if (!$feed_data) {
393 _debug("fetching [$fetch_url]...", $debug_enabled);
394 _debug("If-Modified-Since: ".gmdate('D, d M Y H:i:s \G\M\T', $last_article_timestamp), $debug_enabled);
96f98cb0 395
fd687300
AD
396 $feed_data = fetch_file_contents($fetch_url, false,
397 $auth_login, $auth_pass, false,
398 $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
399 $force_refetch ? 0 : $last_article_timestamp);
3f6f0857 400
fd687300 401 global $fetch_curl_used;
3f6f0857 402
fd687300
AD
403 if (!$fetch_curl_used) {
404 $tmp = @gzdecode($feed_data);
3f6f0857 405
fd687300
AD
406 if ($tmp) $feed_data = $tmp;
407 }
1367bc3f 408
fd687300 409 $feed_data = trim($feed_data);
017401dd 410
fd687300
AD
411 _debug("fetch done.", $debug_enabled);
412
95beaa14
RA
413 // cache vanilla feed data for re-use
414 if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
415 $new_rss_hash = sha1($feed_data);
416
417 if ($new_rss_hash != $rss_hash) {
418 _debug("saving $cache_filename", $debug_enabled);
419 @file_put_contents($cache_filename, $feed_data);
420 }
421 }
4f9cbdff 422 }
017401dd 423
fd687300
AD
424 if (!$feed_data) {
425 global $fetch_last_error;
426 global $fetch_last_error_code;
f074ffe9 427
fd687300 428 _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
f074ffe9 429
fd687300 430 $error_escaped = '';
7a01dc77 431
fd687300
AD
432 // If-Modified-Since
433 if ($fetch_last_error_code != 304) {
434 $error_escaped = db_escape_string($fetch_last_error);
435 } else {
436 _debug("source claims data not modified, nothing to do.", $debug_enabled);
437 }
4f9cbdff 438
fd687300
AD
439 db_query(
440 "UPDATE ttrss_feeds SET last_error = '$error_escaped',
441 last_updated = NOW() WHERE id = '$feed'");
4f9cbdff 442
fd687300
AD
443 return;
444 }
f074ffe9
AD
445 }
446
1ffe3391 447 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
6791af0c 448 $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
017401dd
AD
449 }
450
5ef84097
AD
451 // set last update to now so if anything *simplepie* crashes later we won't be
452 // continuously failing on the same feed
5de40104 453 //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
5ef84097 454
4f9cbdff 455 if (!$rss) {
cd07592c
AD
456 $rss = new FeedParser($feed_data);
457 $rss->init();
5de51df7
AD
458 }
459
2c08214a
AD
460// print_r($rss);
461
a42c55f0 462 $feed = db_escape_string($feed);
2c08214a 463
19b3992b 464 if (!$rss->error()) {
2c08214a 465
d2a421e3 466 // We use local pluginhost here because we need to load different per-user feed plugins
1ffe3391 467 $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
4412b877 468
df659891 469 _debug("language: $feed_language", $debug_enabled);
68cccafc 470 _debug("processing feed data...", $debug_enabled);
2c08214a 471
a42c55f0 472// db_query("BEGIN");
2c08214a 473
382268c6
AD
474 if (DB_TYPE == "pgsql") {
475 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
476 } else {
477 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
478 }
479
fe4535e6 480 $result = db_query("SELECT owner_uid,favicon_avg_color,
382268c6
AD
481 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
482 favicon_needs_check
2c08214a
AD
483 FROM ttrss_feeds WHERE id = '$feed'");
484
382268c6
AD
485 $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
486 "favicon_needs_check"));
36490f11 487 $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
2c08214a
AD
488
489 $owner_uid = db_fetch_result($result, 0, "owner_uid");
490
a42c55f0 491 $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
2c08214a 492
cd07592c
AD
493 _debug("site_url: $site_url", $debug_enabled);
494 _debug("feed_title: " . $rss->get_title(), $debug_enabled);
495
687a4f59 496 if ($favicon_needs_check || $force_refetch) {
36490f11
AD
497
498 /* terrible hack: if we crash on floicon shit here, we won't check
560cbd8c 499 * the icon avgcolor again (unless the icon got updated) */
36490f11 500
560cbd8c
AD
501 $favicon_file = ICONS_DIR . "/$feed.ico";
502 $favicon_modified = @filemtime($favicon_file);
503
68cccafc 504 _debug("checking favicon...", $debug_enabled);
687a4f59 505
6f7798b6 506 check_feed_favicon($site_url, $feed);
560cbd8c
AD
507 $favicon_modified_new = @filemtime($favicon_file);
508
509 if ($favicon_modified_new > $favicon_modified)
510 $favicon_avg_color = '';
687a4f59 511
6ee0d4b0 512 if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
bc7a144d 513 require_once "colors.php";
687a4f59 514
aafd55ba
AD
515 db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
516 id = '$feed'");
517
6322ac79 518 $favicon_color = db_escape_string(
bc7a144d 519 calculate_avg_color($favicon_file));
63c323f7 520
bc7a144d 521 $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
36490f11 522 } else if ($favicon_avg_color == 'fail') {
84ceb6bd 523 _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
6ac722d5 524 }
687a4f59 525
a42c55f0 526 db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
bc7a144d 527 $favicon_colorstring
f2798eb6
AD
528 WHERE id = '$feed'");
529 }
2c08214a 530
68cccafc 531 _debug("loading filters & labels...", $debug_enabled);
2c08214a 532
a42c55f0 533 $filters = load_filters($feed, $owner_uid);
2c08214a 534
68cccafc 535 _debug("" . count($filters) . " filters loaded.", $debug_enabled);
2c08214a 536
19b3992b 537 $items = $rss->get_items();
2c08214a 538
19b3992b 539 if (!is_array($items)) {
68cccafc 540 _debug("no articles found.", $debug_enabled);
2c08214a 541
a42c55f0 542 db_query("UPDATE ttrss_feeds
2c08214a
AD
543 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
544
545 return; // no articles
546 }
547
548 if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
549
68cccafc 550 _debug("checking for PUSH hub...", $debug_enabled);
2c08214a
AD
551
552 $feed_hub_url = false;
2c08214a 553
19b3992b 554 $links = $rss->get_links('hub');
2c08214a 555
19b3992b
AD
556 if ($links && is_array($links)) {
557 foreach ($links as $l) {
558 $feed_hub_url = $l;
559 break;
2c08214a
AD
560 }
561 }
562
68cccafc 563 _debug("feed hub url: $feed_hub_url", $debug_enabled);
2c08214a 564
d8179cb9
AD
565 $feed_self_url = $fetch_url;
566
567 $links = $rss->get_links('self');
568
569 if ($links && is_array($links)) {
570 foreach ($links as $l) {
571 $feed_self_url = $l;
572 break;
573 }
574 }
575
576 _debug("feed self url = $feed_self_url");
577
578 if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
2c08214a
AD
579 !ini_get("open_basedir")) {
580
581 require_once 'lib/pubsubhubbub/subscriber.php';
582
583 $callback_url = get_self_url_prefix() .
584 "/public.php?op=pubsub&id=$feed";
585
586 $s = new Subscriber($feed_hub_url, $callback_url);
587
d8179cb9 588 $rc = $s->subscribe($feed_self_url);
2c08214a 589
5ba1ddd4 590 _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
2c08214a 591
a42c55f0 592 db_query("UPDATE ttrss_feeds SET pubsub_state = 1
2c08214a
AD
593 WHERE id = '$feed'");
594 }
595 }
596
68cccafc 597 _debug("processing articles...", $debug_enabled);
2c08214a 598
6c9f3d4a
AD
599 $tstart = time();
600
19b3992b 601 foreach ($items as $item) {
5d56d100 602 if ($_REQUEST['xdebug'] == 3) {
2c08214a
AD
603 print_r($item);
604 }
605
6c9f3d4a
AD
606 if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
607 _debug("looks like there's too many articles to process at once, breaking out", $debug_enabled);
608 break;
609 }
610
19b3992b
AD
611 $entry_guid = $item->get_id();
612 if (!$entry_guid) $entry_guid = $item->get_link();
613 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
2c08214a
AD
614 if (!$entry_guid) continue;
615
3a4c8973
AD
616 $entry_guid = "$owner_uid,$entry_guid";
617
a42c55f0 618 $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
5e3d5480 619
68cccafc 620 _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
5e3d5480 621
2c08214a
AD
622 $entry_timestamp = "";
623
04d2f9c8
AD
624 $entry_timestamp = $item->get_date();
625
626 _debug("orig date: " . $item->get_date(), $debug_enabled);
2c08214a 627
30123fe6 628 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
2c08214a 629 $entry_timestamp = time();
2c08214a
AD
630 }
631
632 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
633
68cccafc 634 _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
2c08214a 635
99429e57
AD
636// $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
637// $entry_title = decode_numeric_entities($entry_title);
638 $entry_title = $item->get_title();
1b35d30c 639
5d56d100 640 $entry_link = rewrite_relative_url($site_url, $item->get_link());
2c08214a 641
68cccafc
AD
642 _debug("title $entry_title", $debug_enabled);
643 _debug("link $entry_link", $debug_enabled);
2c08214a
AD
644
645 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
646
19b3992b
AD
647 $entry_content = $item->get_content();
648 if (!$entry_content) $entry_content = $item->get_description();
2c08214a
AD
649
650 if ($_REQUEST["xdebug"] == 2) {
9ec10352 651 print "content: ";
487f0750 652 print $entry_content;
3c696512 653 print "\n";
2c08214a
AD
654 }
655
cd07592c
AD
656 $entry_comments = $item->get_comments_url();
657 $entry_author = $item->get_author();
2c08214a 658
a42c55f0 659 $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
2c08214a 660
ce2b219c
AD
661 $entry_comments = db_escape_string(mb_substr(trim($entry_comments), 0, 245));
662 $entry_author = db_escape_string(mb_substr(trim($entry_author), 0, 245));
2c08214a 663
d4992d6b 664 $num_comments = (int) $item->get_comments_count();
2c08214a 665
68cccafc
AD
666 _debug("author $entry_author", $debug_enabled);
667 _debug("num_comments: $num_comments", $debug_enabled);
ee78f81c 668 _debug("looking for tags...", $debug_enabled);
2c08214a
AD
669
670 // parse <category> entries into tags
671
672 $additional_tags = array();
673
19b3992b 674 $additional_tags_src = $item->get_categories();
2c08214a 675
19b3992b
AD
676 if (is_array($additional_tags_src)) {
677 foreach ($additional_tags_src as $tobj) {
cd07592c 678 array_push($additional_tags, $tobj);
2c08214a 679 }
19b3992b 680 }
2c08214a 681
fa6fbd36 682 $entry_tags = array_unique($additional_tags);
2c08214a
AD
683
684 for ($i = 0; $i < count($entry_tags); $i++)
685 $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
686
ee78f81c
AD
687 _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
688
68cccafc 689 _debug("done collecting data.", $debug_enabled);
2c08214a 690
3318d324 691 $result = db_query("SELECT id, content_hash, lang FROM ttrss_entries
b1840673 692 WHERE guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed'");
b30abdad
AD
693
694 if (db_num_rows($result) != 0) {
b1840673
AD
695 $base_entry_id = db_fetch_result($result, 0, "id");
696 $entry_stored_hash = db_fetch_result($result, 0, "content_hash");
a29fe121 697 $article_labels = get_article_labels($base_entry_id, $owner_uid);
3318d324 698 $entry_language = db_fetch_result($result, 0, "lang");
b30abdad 699 } else {
b1840673
AD
700 $base_entry_id = false;
701 $entry_stored_hash = "";
a29fe121 702 $article_labels = array();
3318d324 703 $entry_language = "";
b30abdad
AD
704 }
705
455b1401 706 $article = array("owner_uid" => $owner_uid, // read only
b30abdad 707 "guid" => $entry_guid, // read only
59e83455 708 "guid_hashed" => $entry_guid_hashed, // read only
19b3992b
AD
709 "title" => $entry_title,
710 "content" => $entry_content,
711 "link" => $entry_link,
a29fe121 712 "labels" => $article_labels, // current limitation: can add labels to article, can't remove them
19b3992b 713 "tags" => $entry_tags,
e02555c1 714 "author" => $entry_author,
c9299c28 715 "force_catchup" => false, // ugly hack for the time being
6de3a1be 716 "score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
3318d324 717 "language" => $entry_language,
f73e03e0
AD
718 "feed" => array("id" => $feed,
719 "fetch_url" => $fetch_url,
720 "site_url" => $site_url)
721 );
cc85704f 722
b1840673 723 $entry_plugin_data = "";
eb16bd9f 724 $entry_current_hash = calculate_article_hash($article, $pluginhost);
b1840673
AD
725
726 _debug("article hash: $entry_current_hash [stored=$entry_stored_hash]", $debug_enabled);
727
522e8b35 728 if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
b1840673
AD
729 _debug("stored article seems up to date [IID: $base_entry_id], updating timestamp only", $debug_enabled);
730
731 // we keep encountering the entry in feeds, so we need to
732 // update date_updated column so that we don't get horrible
733 // dupes when the entry gets purged and reinserted again e.g.
734 // in the case of SLOW SLOW OMG SLOW updating feeds
735
736 $base_entry_id = db_fetch_result($result, 0, "id");
737
738 db_query("UPDATE ttrss_entries SET date_updated = NOW()
739 WHERE id = '$base_entry_id'");
740
e44280a0 741 // if we allow duplicate posts, we have to continue to
da7e3e91
GP
742 // create the user entries for this feed
743 if (!get_pref("ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
744 continue;
745 }
b1840673
AD
746 }
747
748 _debug("hash differs, applying plugin filters:", $debug_enabled);
749
1ffe3391 750 foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
b1840673
AD
751 _debug("... " . get_class($plugin), $debug_enabled);
752
753 $start = microtime(true);
19b3992b 754 $article = $plugin->hook_article_filter($article);
0084f0d1 755
b1840673
AD
756 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
757
758 $entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
cc85704f
AD
759 }
760
b1840673
AD
761 $entry_plugin_data = db_escape_string($entry_plugin_data);
762
763 _debug("plugin data: $entry_plugin_data", $debug_enabled);
764
35c12dc4
AD
765 // Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
766 if (DB_TYPE == "mysql") {
767 foreach ($article as $k => $v) {
35c37354
AD
768
769 // i guess we'll have to take the risk of 4byte unicode labels & tags here
770 if (!is_array($article[$k])) {
771 $article[$k] = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $v);
772 }
35c12dc4
AD
773 }
774 }
775
b8774453
AD
776 /* Collect article tags here so we could filter by them: */
777
778 $article_filters = get_article_filters($filters, $article["title"],
779 $article["content"], $article["link"], 0, $article["author"],
780 $article["tags"]);
781
782 if ($debug_enabled) {
783 _debug("article filters: ", $debug_enabled);
784 if (count($article_filters) != 0) {
785 print_r($article_filters);
786 }
787 }
788
789 $plugin_filter_names = find_article_filters($article_filters, "plugin");
790 $plugin_filter_actions = $pluginhost->get_filter_actions();
791
792 if (count($plugin_filter_names) > 0) {
793 _debug("applying plugin filter actions...", $debug_enabled);
794
795 foreach ($plugin_filter_names as $pfn) {
796 list($pfclass,$pfaction) = explode(":", $pfn["param"]);
797
798 if (isset($plugin_filter_actions[$pfclass])) {
799 $plugin = $pluginhost->get_plugin($pfclass);
800
801 _debug("... $pfclass: $pfaction", $debug_enabled);
802
803 if ($plugin) {
804 $start = microtime(true);
805 $article = $plugin->hook_article_filter_action($article, $pfaction);
806
807 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
808 } else {
809 _debug("??? $pfclass: plugin object not found.");
810 }
811 } else {
812 _debug("??? $pfclass: filter plugin not registered.");
813 }
814 }
815 }
816
19b3992b 817 $entry_tags = $article["tags"];
a42c55f0
AD
818 $entry_guid = db_escape_string($entry_guid);
819 $entry_title = db_escape_string($article["title"]);
820 $entry_author = db_escape_string($article["author"]);
821 $entry_link = db_escape_string($article["link"]);
f935d98e 822 $entry_content = $article["content"]; // escaped below
c9299c28 823 $entry_force_catchup = $article["force_catchup"];
a29fe121 824 $article_labels = $article["labels"];
6de3a1be 825 $entry_score_modifier = (int) $article["score_modifier"];
3318d324 826 $entry_language = db_escape_string($article["language"]);
a29fe121
AD
827
828 if ($debug_enabled) {
829 _debug("article labels:", $debug_enabled);
830 print_r($article_labels);
831 }
c9299c28
AD
832
833 _debug("force catchup: $entry_force_catchup");
f935d98e 834
0a3fd79b 835 if ($cache_images && is_writable(CACHE_DIR . '/images'))
f0bd8e65 836 cache_images($entry_content, $site_url, $debug_enabled);
0a3fd79b 837
a42c55f0 838 $entry_content = db_escape_string($entry_content, false);
1f45c857 839
07877caf 840 //db_query("BEGIN");
2c08214a 841
a42c55f0 842 $result = db_query("SELECT id FROM ttrss_entries
5e3d5480 843 WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
9e222305 844
2c08214a
AD
845 if (db_num_rows($result) == 0) {
846
68cccafc 847 _debug("base guid [$entry_guid] not found", $debug_enabled);
2c08214a
AD
848
849 // base post entry does not exist, create it
850
6322ac79 851 $result = db_query(
2c08214a
AD
852 "INSERT INTO ttrss_entries
853 (title,
854 guid,
855 link,
856 updated,
857 content,
858 content_hash,
859 no_orig_date,
860 date_updated,
861 date_entered,
862 comments,
863 num_comments,
b30abdad 864 plugin_data,
6b461797 865 lang,
2c08214a
AD
866 author)
867 VALUES
868 ('$entry_title',
5e3d5480 869 '$entry_guid_hashed',
2c08214a
AD
870 '$entry_link',
871 '$entry_timestamp_fmt',
872 '$entry_content',
b1840673 873 '$entry_current_hash',
5ba1ddd4 874 false,
2c08214a 875 NOW(),
be574731 876 '$date_feed_processed',
2c08214a
AD
877 '$entry_comments',
878 '$num_comments',
b30abdad 879 '$entry_plugin_data',
6b461797 880 '$entry_language',
2c08214a 881 '$entry_author')");
e8291805 882
2c08214a 883 } else {
2c08214a 884 $base_entry_id = db_fetch_result($result, 0, "id");
2c08214a
AD
885 }
886
887 // now it should exist, if not - bad luck then
888
b1840673 889 $result = db_query("SELECT id FROM ttrss_entries
5e3d5480 890 WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
2c08214a
AD
891
892 $entry_ref_id = 0;
893 $entry_int_id = 0;
894
895 if (db_num_rows($result) == 1) {
896
68cccafc 897 _debug("base guid found, checking for user record", $debug_enabled);
2c08214a 898
2c08214a
AD
899 $ref_id = db_fetch_result($result, 0, "id");
900 $entry_ref_id = $ref_id;
901
5e3d5480
AD
902 /* $stored_guid = db_fetch_result($result, 0, "guid");
903 if ($stored_guid != $entry_guid_hashed) {
68cccafc 904 if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
5e3d5480 905
a42c55f0 906 db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
5e3d5480
AD
907 id = '$ref_id'");
908 } */
909
2c08214a
AD
910 // check for user post link to main table
911
912 // do we allow duplicate posts with same GUID in different feeds?
a42c55f0 913 if (get_pref("ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
2c08214a
AD
914 $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
915 } else {
916 $dupcheck_qpart = "";
917 }
918
2c08214a 919 if (find_article_filter($article_filters, "filter")) {
07877caf 920 //db_query("COMMIT"); // close transaction in progress
2c08214a
AD
921 continue;
922 }
923
6de3a1be 924 $score = calculate_article_score($article_filters) + $entry_score_modifier;
2c08214a 925
6de3a1be 926 _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
2c08214a
AD
927
928 $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
929 ref_id = '$ref_id' AND owner_uid = '$owner_uid'
930 $dupcheck_qpart";
931
932// if ($_REQUEST["xdebug"]) print "$query\n";
933
a42c55f0 934 $result = db_query($query);
2c08214a
AD
935
936 // okay it doesn't exist - create user entry
937 if (db_num_rows($result) == 0) {
938
68cccafc 939 _debug("user record not found, creating...", $debug_enabled);
2c08214a 940
c9299c28 941 if ($score >= -500 && !find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
2c08214a
AD
942 $unread = 'true';
943 $last_read_qpart = 'NULL';
944 } else {
945 $unread = 'false';
946 $last_read_qpart = 'NOW()';
947 }
948
949 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
950 $marked = 'true';
951 } else {
952 $marked = 'false';
953 }
954
955 if (find_article_filter($article_filters, 'publish')) {
956 $published = 'true';
957 } else {
958 $published = 'false';
959 }
960
2ea9bbfd
AD
961 // N-grams
962
c9299c28 963 /* if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
2ea9bbfd 964
a42c55f0 965 $result = db_query("SELECT COUNT(*) AS similar FROM
2ea9bbfd
AD
966 ttrss_entries,ttrss_user_entries
967 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
968 AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
969 AND owner_uid = $owner_uid");
970
971 $ngram_similar = db_fetch_result($result, 0, "similar");
972
68cccafc 973 _debug("N-gram similar results: $ngram_similar", $debug_enabled);
2ea9bbfd
AD
974
975 if ($ngram_similar > 0) {
976 $unread = 'false';
977 }
c9299c28 978 } */
2ea9bbfd 979
7873d588
AD
980 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
981 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
982
6322ac79 983 $result = db_query(
2c08214a
AD
984 "INSERT INTO ttrss_user_entries
985 (ref_id, owner_uid, feed_id, unread, last_read, marked,
7873d588
AD
986 published, score, tag_cache, label_cache, uuid,
987 last_marked, last_published)
2c08214a 988 VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
7873d588
AD
989 $last_read_qpart, $marked, $published, '$score', '', '',
990 '', $last_marked, $last_published)");
2c08214a
AD
991
992 if (PUBSUBHUBBUB_HUB && $published == 'true') {
993 $rss_link = get_self_url_prefix() .
994 "/public.php?op=rss&id=-2&key=" .
a42c55f0 995 get_feed_access_key(-2, false, $owner_uid);
2c08214a
AD
996
997 $p = new Publisher(PUBSUBHUBBUB_HUB);
998
5ba1ddd4 999 /* $pubsub_result = */ $p->publish_update($rss_link);
2c08214a
AD
1000 }
1001
6322ac79 1002 $result = db_query(
2c08214a
AD
1003 "SELECT int_id FROM ttrss_user_entries WHERE
1004 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
1005 feed_id = '$feed' LIMIT 1");
1006
1007 if (db_num_rows($result) == 1) {
1008 $entry_int_id = db_fetch_result($result, 0, "int_id");
1009 }
1010 } else {
68cccafc 1011 _debug("user record FOUND", $debug_enabled);
2c08214a
AD
1012
1013 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
1014 $entry_int_id = db_fetch_result($result, 0, "int_id");
1015 }
1016
68cccafc 1017 _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
2c08214a 1018
e854442e 1019 if (DB_TYPE == "pgsql") {
978989bb 1020 $tsvector_combined = db_escape_string(mb_substr($entry_title . ' ' . strip_tags(str_replace('<', ' <', $entry_content)),
e854442e
AD
1021 0, 1000000));
1022
df659891 1023 $tsvector_qpart = "tsvector_combined = to_tsvector('$feed_language', '$tsvector_combined'),";
e854442e
AD
1024
1025 } else {
1026 $tsvector_qpart = "";
1027 }
1028
b1840673
AD
1029 db_query("UPDATE ttrss_entries
1030 SET title = '$entry_title',
1031 content = '$entry_content',
1032 content_hash = '$entry_current_hash',
1033 updated = '$entry_timestamp_fmt',
e854442e 1034 $tsvector_qpart
b1840673
AD
1035 num_comments = '$num_comments',
1036 plugin_data = '$entry_plugin_data',
1037 author = '$entry_author',
1038 lang = '$entry_language'
1039 WHERE id = '$ref_id'");
1040
59e83455
AD
1041 // update aux data
1042 db_query("UPDATE ttrss_user_entries
1043 SET score = '$score' WHERE ref_id = '$ref_id'");
1044
b1840673
AD
1045 if ($mark_unread_on_update) {
1046 db_query("UPDATE ttrss_user_entries
1047 SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
2c08214a
AD
1048 }
1049 }
1050
07877caf 1051 //db_query("COMMIT");
2c08214a 1052
a29fe121
AD
1053 _debug("assigning labels [other]...", $debug_enabled);
1054
1055 foreach ($article_labels as $label) {
1056 label_add_article($entry_ref_id, $label[1], $owner_uid);
1057 }
1058
1059 _debug("assigning labels [filters]...", $debug_enabled);
2c08214a 1060
a42c55f0 1061 assign_article_to_label_filters($entry_ref_id, $article_filters,
b24504b1 1062 $owner_uid, $article_labels);
2c08214a 1063
68cccafc 1064 _debug("looking for enclosures...", $debug_enabled);
2c08214a
AD
1065
1066 // enclosures
1067
1068 $enclosures = array();
1069
19b3992b 1070 $encs = $item->get_enclosures();
2c08214a 1071
19b3992b
AD
1072 if (is_array($encs)) {
1073 foreach ($encs as $e) {
1074 $e_item = array(
523bd90b 1075 $e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
2c08214a 1076 array_push($enclosures, $e_item);
2c08214a
AD
1077 }
1078 }
1079
2c08214a 1080 if ($debug_enabled) {
68cccafc 1081 _debug("article enclosures:", $debug_enabled);
2c08214a
AD
1082 print_r($enclosures);
1083 }
1084
dab229cd 1085 //db_query("BEGIN");
2c08214a 1086
5c54e683
AD
1087// debugging
1088// db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
1089
2c08214a 1090 foreach ($enclosures as $enc) {
a42c55f0
AD
1091 $enc_url = db_escape_string($enc[0]);
1092 $enc_type = db_escape_string($enc[1]);
1093 $enc_dur = db_escape_string($enc[2]);
5c54e683 1094 $enc_title = db_escape_string($enc[3]);
523bd90b
FE
1095 $enc_width = intval($enc[4]);
1096 $enc_height = intval($enc[5]);
2c08214a 1097
a42c55f0 1098 $result = db_query("SELECT id FROM ttrss_enclosures
2c08214a
AD
1099 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
1100
1101 if (db_num_rows($result) == 0) {
a42c55f0 1102 db_query("INSERT INTO ttrss_enclosures
523bd90b
FE
1103 (content_url, content_type, title, duration, post_id, width, height) VALUES
1104 ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
2c08214a
AD
1105 }
1106 }
1107
dab229cd 1108 //db_query("COMMIT");
2c08214a
AD
1109
1110 // check for manual tags (we have to do it here since they're loaded from filters)
1111
1112 foreach ($article_filters as $f) {
6aff7845 1113 if ($f["type"] == "tag") {
2c08214a 1114
6aff7845 1115 $manual_tags = trim_array(explode(",", $f["param"]));
2c08214a
AD
1116
1117 foreach ($manual_tags as $tag) {
1118 if (tag_is_valid($tag)) {
1119 array_push($entry_tags, $tag);
1120 }
1121 }
1122 }
1123 }
1124
1125 // Skip boring tags
1126
6322ac79 1127 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
2c08214a
AD
1128 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1129
1130 $filtered_tags = array();
1131 $tags_to_cache = array();
1132
1133 if ($entry_tags && is_array($entry_tags)) {
1134 foreach ($entry_tags as $tag) {
1135 if (array_search($tag, $boring_tags) === false) {
1136 array_push($filtered_tags, $tag);
1137 }
1138 }
1139 }
1140
1141 $filtered_tags = array_unique($filtered_tags);
1142
1143 if ($debug_enabled) {
68cccafc 1144 _debug("filtered article tags:", $debug_enabled);
2c08214a
AD
1145 print_r($filtered_tags);
1146 }
1147
1148 // Save article tags in the database
1149
1150 if (count($filtered_tags) > 0) {
1151
dab229cd 1152 //db_query("BEGIN");
2c08214a
AD
1153
1154 foreach ($filtered_tags as $tag) {
1155
1156 $tag = sanitize_tag($tag);
a42c55f0 1157 $tag = db_escape_string($tag);
2c08214a
AD
1158
1159 if (!tag_is_valid($tag)) continue;
1160
a42c55f0 1161 $result = db_query("SELECT id FROM ttrss_tags
2c08214a
AD
1162 WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1163 owner_uid = '$owner_uid' LIMIT 1");
1164
1165 if ($result && db_num_rows($result) == 0) {
1166
a42c55f0 1167 db_query("INSERT INTO ttrss_tags
2c08214a
AD
1168 (owner_uid,tag_name,post_int_id)
1169 VALUES ('$owner_uid','$tag', '$entry_int_id')");
1170 }
1171
1172 array_push($tags_to_cache, $tag);
1173 }
1174
1175 /* update the cache */
1176
1177 $tags_to_cache = array_unique($tags_to_cache);
1178
a42c55f0 1179 $tags_str = db_escape_string(join(",", $tags_to_cache));
2c08214a 1180
a42c55f0 1181 db_query("UPDATE ttrss_user_entries
2c08214a
AD
1182 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1183 AND owner_uid = $owner_uid");
1184
dab229cd 1185 //db_query("COMMIT");
2c08214a
AD
1186 }
1187
68cccafc 1188 _debug("article processed", $debug_enabled);
2c08214a
AD
1189 }
1190
68cccafc 1191 _debug("purging feed...", $debug_enabled);
2c08214a 1192
a42c55f0 1193 purge_feed($feed, 0, $debug_enabled);
2c08214a 1194
a42c55f0 1195 db_query("UPDATE ttrss_feeds
f2798eb6 1196 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
2c08214a 1197
a42c55f0 1198// db_query("COMMIT");
2c08214a
AD
1199
1200 } else {
1201
a42c55f0 1202 $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
2c08214a 1203
4ad04ee2
AD
1204 _debug("fetch error: $error_msg", $debug_enabled);
1205
1206 if (count($rss->errors()) > 1) {
1207 foreach ($rss->errors() as $error) {
1208 _debug("+ $error");
1209 }
1210 }
2c08214a 1211
6322ac79 1212 db_query(
2c08214a 1213 "UPDATE ttrss_feeds SET last_error = '$error_msg',
88edaa93 1214 last_updated = NOW() WHERE id = '$feed'");
2c08214a 1215
88edaa93
AD
1216 unset($rss);
1217 }
2c08214a 1218
68cccafc 1219 _debug("done", $debug_enabled);
88edaa93
AD
1220
1221 return $rss;
2c08214a
AD
1222 }
1223
3c696512 1224 function cache_images($html, $site_url, $debug) {
3c696512
AD
1225 libxml_use_internal_errors(true);
1226
1227 $charset_hack = '<head>
1228 <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1229 </head>';
1230
1231 $doc = new DOMDocument();
1232 $doc->loadHTML($charset_hack . $html);
1233 $xpath = new DOMXPath($doc);
1234
1235 $entries = $xpath->query('(//img[@src])');
1236
1237 foreach ($entries as $entry) {
1238 if ($entry->hasAttribute('src')) {
1239 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1240
1241 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1242
1243 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1244
1245 if (!file_exists($local_filename)) {
1246 $file_content = fetch_file_contents($src);
1247
a3d50184 1248 if ($file_content && strlen($file_content) > _MIN_CACHE_IMAGE_SIZE) {
3c696512
AD
1249 file_put_contents($local_filename, $file_content);
1250 }
4a27966e
J
1251 } else {
1252 touch($local_filename);
3c696512 1253 }
3c696512
AD
1254 }
1255 }
3c696512
AD
1256 }
1257
a42c55f0 1258 function expire_error_log($debug) {
e2261e17
AD
1259 if ($debug) _debug("Removing old error log entries...");
1260
1261 if (DB_TYPE == "pgsql") {
a42c55f0 1262 db_query("DELETE FROM ttrss_error_log
e2261e17
AD
1263 WHERE created_at < NOW() - INTERVAL '7 days'");
1264 } else {
a42c55f0 1265 db_query("DELETE FROM ttrss_error_log
e2261e17
AD
1266 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1267 }
1268
1269 }
1270
2a91b6ff 1271 function expire_lock_files($debug) {
65465085 1272 //if ($debug) _debug("Removing old lock files...");
2a91b6ff
AD
1273
1274 $num_deleted = 0;
1275
1276 if (is_writable(LOCK_DIRECTORY)) {
1277 $files = glob(LOCK_DIRECTORY . "/*.lock");
1278
1279 if ($files) {
1280 foreach ($files as $file) {
11344971 1281 if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
2a91b6ff
AD
1282 unlink($file);
1283 ++$num_deleted;
1284 }
1285 }
1286 }
1287 }
1288
65465085 1289 if ($debug) _debug("Removed $num_deleted old lock files.");
2a91b6ff
AD
1290 }
1291
3c696512 1292 function expire_cached_files($debug) {
3306daec 1293 foreach (array("simplepie", "images", "export", "upload") as $dir) {
3c696512 1294 $cache_dir = CACHE_DIR . "/$dir";
2c08214a 1295
65465085 1296// if ($debug) _debug("Expiring $cache_dir");
2c08214a 1297
3c696512
AD
1298 $num_deleted = 0;
1299
1300 if (is_writable($cache_dir)) {
1301 $files = glob("$cache_dir/*");
1302
2a91b6ff 1303 if ($files) {
2ab20c31
AD
1304 foreach ($files as $file) {
1305 if (time() - filemtime($file) > 86400*7) {
1306 unlink($file);
3c696512 1307
2ab20c31
AD
1308 ++$num_deleted;
1309 }
3c696512
AD
1310 }
1311 }
2a91b6ff 1312 }
3c696512 1313
65465085 1314 if ($debug) _debug("$cache_dir: removed $num_deleted files.");
3c696512
AD
1315 }
1316 }
2c08214a 1317
a3e0bdcf
AD
1318 /**
1319 * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1320 * Returns the url query as associative array
1321 *
1322 * @param string query
1323 * @return array params
1324 */
1325 function convertUrlQuery($query) {
1326 $queryParts = explode('&', $query);
1327
1328 $params = array();
1329
1330 foreach ($queryParts as $param) {
1331 $item = explode('=', $param);
1332 $params[$item[0]] = $item[1];
1333 }
1334
1335 return $params;
1336 }
92c14e9d 1337
48cb2536 1338 function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
92c14e9d
AD
1339 $matches = array();
1340
1341 foreach ($filters as $filter) {
1342 $match_any_rule = $filter["match_any_rule"];
a3a896a1 1343 $inverse = $filter["inverse"];
92c14e9d
AD
1344 $filter_match = false;
1345
1346 foreach ($filter["rules"] as $rule) {
1347 $match = false;
ffa1bd7b 1348 $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
a3a896a1 1349 $rule_inverse = $rule["inverse"];
92c14e9d
AD
1350
1351 if (!$reg_exp)
1352 continue;
1353
1354 switch ($rule["type"]) {
1355 case "title":
1356 $match = @preg_match("/$reg_exp/i", $title);
1357 break;
1358 case "content":
d03ae73e
AD
1359 // we don't need to deal with multiline regexps
1360 $content = preg_replace("/[\r\n\t]/", "", $content);
1361
92c14e9d
AD
1362 $match = @preg_match("/$reg_exp/i", $content);
1363 break;
1364 case "both":
d03ae73e
AD
1365 // we don't need to deal with multiline regexps
1366 $content = preg_replace("/[\r\n\t]/", "", $content);
1367
72d1d067 1368 $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
92c14e9d
AD
1369 break;
1370 case "link":
1371 $match = @preg_match("/$reg_exp/i", $link);
1372 break;
1373 case "author":
1374 $match = @preg_match("/$reg_exp/i", $author);
1375 break;
1376 case "tag":
7b80b5e1
DK
1377 foreach ($tags as $tag) {
1378 if (@preg_match("/$reg_exp/i", $tag)) {
1379 $match = true;
1380 break;
1381 }
1382 }
92c14e9d
AD
1383 break;
1384 }
1385
a3a896a1
AD
1386 if ($rule_inverse) $match = !$match;
1387
92c14e9d
AD
1388 if ($match_any_rule) {
1389 if ($match) {
1390 $filter_match = true;
1391 break;
1392 }
1393 } else {
1394 $filter_match = $match;
1395 if (!$match) {
1396 break;
1397 }
1398 }
1399 }
1400
a3a896a1
AD
1401 if ($inverse) $filter_match = !$filter_match;
1402
92c14e9d
AD
1403 if ($filter_match) {
1404 foreach ($filter["actions"] AS $action) {
1405 array_push($matches, $action);
5e736e45
AD
1406
1407 // if Stop action encountered, perform no further processing
1408 if ($action["type"] == "stop") return $matches;
92c14e9d
AD
1409 }
1410 }
1411 }
1412
1413 return $matches;
1414 }
1415
1416 function find_article_filter($filters, $filter_name) {
1417 foreach ($filters as $f) {
1418 if ($f["type"] == $filter_name) {
1419 return $f;
1420 };
1421 }
1422 return false;
1423 }
1424
1425 function find_article_filters($filters, $filter_name) {
1426 $results = array();
1427
1428 foreach ($filters as $f) {
1429 if ($f["type"] == $filter_name) {
1430 array_push($results, $f);
1431 };
1432 }
1433 return $results;
1434 }
1435
1436 function calculate_article_score($filters) {
1437 $score = 0;
1438
1439 foreach ($filters as $f) {
1440 if ($f["type"] == "score") {
1441 $score += $f["param"];
1442 };
1443 }
1444 return $score;
1445 }
1446
b24504b1
AD
1447 function labels_contains_caption($labels, $caption) {
1448 foreach ($labels as $label) {
1449 if ($label[1] == $caption) {
1450 return true;
1451 }
1452 }
1453
1454 return false;
1455 }
1456
a42c55f0 1457 function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
92c14e9d
AD
1458 foreach ($filters as $f) {
1459 if ($f["type"] == "label") {
b24504b1 1460 if (!labels_contains_caption($article_labels, $f["param"])) {
a42c55f0 1461 label_add_article($id, $f["param"], $owner_uid);
b24504b1
AD
1462 }
1463 }
92c14e9d
AD
1464 }
1465 }
87764a50 1466
87d7e850
AD
1467 function make_guid_from_title($title) {
1468 return preg_replace("/[ \"\',.:;]/", "-",
1469 mb_strtolower(strip_tags($title), 'utf-8'));
1470 }
1471
d1f3fa97 1472 /* function verify_feed_xml($feed_data) {
ebec81a6
AD
1473 libxml_use_internal_errors(true);
1474 $doc = new DOMDocument();
1475 $doc->loadXML($feed_data);
1476 $error = libxml_get_last_error();
1477 libxml_clear_errors();
1478 return $error;
d1f3fa97 1479 } */
87d7e850 1480
168cf351
AD
1481 function cleanup_counters_cache($debug) {
1482 $result = db_query("DELETE FROM ttrss_counters_cache
1483 WHERE feed_id > 0 AND
1484 (SELECT COUNT(id) FROM ttrss_feeds WHERE
1485 id = feed_id AND
1486 ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
1487 $frows = db_affected_rows($result);
1488
1489 $result = db_query("DELETE FROM ttrss_cat_counters_cache
1490 WHERE feed_id > 0 AND
1491 (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
1492 id = feed_id AND
1493 ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
1494 $crows = db_affected_rows($result);
1495
1496 _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
1497 }
1498
e2cf81e2
AD
1499 function housekeeping_common($debug) {
1500 expire_cached_files($debug);
1501 expire_lock_files($debug);
1502 expire_error_log($debug);
1503
1504 $count = update_feedbrowser_cache();
1505 _debug("Feedbrowser updated, $count feeds processed.");
1506
1507 purge_orphans( true);
168cf351 1508 cleanup_counters_cache($debug);
e2cf81e2
AD
1509 $rc = cleanup_tags( 14, 50000);
1510
1511 _debug("Cleaned $rc cached tags.");
8e470220 1512
00f22824 1513 PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
8e470220 1514
e2cf81e2 1515 }
2c08214a 1516?>