]> git.wh0rd.org - tt-rss.git/blame - include/rssfuncs.php
switch modules to new dojo definition format
[tt-rss.git] / include / rssfuncs.php
CommitLineData
2c08214a 1<?php
fce451a4
AD
2 define_default('DAEMON_UPDATE_LOGIN_LIMIT', 30);
3 define_default('DAEMON_FEED_LIMIT', 500);
e2cf81e2 4 define_default('DAEMON_SLEEP_INTERVAL', 120);
a3d50184 5 define_default('_MIN_CACHE_IMAGE_SIZE', 1024);
09e8bdfd 6
eb16bd9f 7 function calculate_article_hash($article, $pluginhost) {
af244f92
AD
8 $tmp = "";
9
10 foreach ($article as $k => $v) {
11 if ($k != "feed" && isset($v)) {
24e6ff5d
AD
12 $x = strip_tags(is_array($v) ? implode(",", $v) : $v);
13
14 //_debug("$k:" . sha1($x) . ":" . htmlspecialchars($x), true);
15
16 $tmp .= sha1("$k:" . sha1($x));
af244f92
AD
17 }
18 }
19
eb16bd9f 20 return sha1(implode(",", $pluginhost->get_plugin_names()) . $tmp);
b1840673
AD
21 }
22
6322ac79 23 function update_feedbrowser_cache() {
79178062 24
a42c55f0 25 $result = db_query("SELECT feed_url, site_url, title, COUNT(id) AS subscribers
79178062
AD
26 FROM ttrss_feeds WHERE (SELECT COUNT(id) = 0 FROM ttrss_feeds AS tf
27 WHERE tf.feed_url = ttrss_feeds.feed_url
28 AND (private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%'))
29 GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000");
30
a42c55f0 31 db_query("BEGIN");
79178062 32
a42c55f0 33 db_query("DELETE FROM ttrss_feedbrowser_cache");
79178062
AD
34
35 $count = 0;
36
37 while ($line = db_fetch_assoc($result)) {
a42c55f0
AD
38 $subscribers = db_escape_string($line["subscribers"]);
39 $feed_url = db_escape_string($line["feed_url"]);
40 $title = db_escape_string($line["title"]);
41 $site_url = db_escape_string($line["site_url"]);
79178062 42
a42c55f0 43 $tmp_result = db_query("SELECT subscribers FROM
79178062
AD
44 ttrss_feedbrowser_cache WHERE feed_url = '$feed_url'");
45
46 if (db_num_rows($tmp_result) == 0) {
47
a42c55f0 48 db_query("INSERT INTO ttrss_feedbrowser_cache
79178062
AD
49 (feed_url, site_url, title, subscribers) VALUES ('$feed_url',
50 '$site_url', '$title', '$subscribers')");
51
52 ++$count;
53
54 }
55
56 }
57
a42c55f0 58 db_query("COMMIT");
79178062
AD
59
60 return $count;
61
62 }
63
64
2c08214a
AD
65 /**
66 * Update a feed batch.
67 * Used by daemons to update n feeds by run.
68 * Only update feed needing a update, and not being processed
69 * by another process.
70 *
71 * @param mixed $link Database link
72 * @param integer $limit Maximum number of feeds in update batch. Default to DAEMON_FEED_LIMIT.
73 * @param boolean $from_http Set to true if you call this function from http to disable cli specific code.
74 * @param boolean $debug Set to false to disable debug output. Default to true.
75 * @return void
76 */
a42c55f0 77 function update_daemon_common($limit = DAEMON_FEED_LIMIT, $from_http = false, $debug = true) {
2c08214a
AD
78 // Process all other feeds using last_updated and interval parameters
79
6322ac79 80 $schema_version = get_schema_version();
857efe49
AD
81
82 if ($schema_version != SCHEMA_VERSION) {
83 die("Schema version is wrong, please upgrade the database.\n");
84 }
85
61c1812f
AD
86 define('PREFS_NO_CACHE', true);
87
2c08214a 88 // Test if the user has loggued in recently. If not, it does not update its feeds.
09e8bdfd 89 if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) {
2c08214a
AD
90 if (DB_TYPE == "pgsql") {
91 $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
92 } else {
93 $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
94 }
95 } else {
96 $login_thresh_qpart = "";
97 }
98
537eca40 99 // Test if the feed need a update (update interval exceeded).
2c08214a
AD
100 if (DB_TYPE == "pgsql") {
101 $update_limit_qpart = "AND ((
102 ttrss_feeds.update_interval = 0
ee0542ce 103 AND ttrss_user_prefs.value != '-1'
2c08214a
AD
104 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
105 ) OR (
106 ttrss_feeds.update_interval > 0
107 AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
f08426e3
AD
108 ) OR (ttrss_feeds.last_updated IS NULL
109 AND ttrss_user_prefs.value != '-1')
110 OR (last_updated = '1970-01-01 00:00:00'
111 AND ttrss_user_prefs.value != '-1'))";
2c08214a
AD
112 } else {
113 $update_limit_qpart = "AND ((
114 ttrss_feeds.update_interval = 0
ee0542ce 115 AND ttrss_user_prefs.value != '-1'
2c08214a
AD
116 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
117 ) OR (
118 ttrss_feeds.update_interval > 0
119 AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
f08426e3
AD
120 ) OR (ttrss_feeds.last_updated IS NULL
121 AND ttrss_user_prefs.value != '-1')
122 OR (last_updated = '1970-01-01 00:00:00'
123 AND ttrss_user_prefs.value != '-1'))";
2c08214a
AD
124 }
125
126 // Test if feed is currently being updated by another process.
127 if (DB_TYPE == "pgsql") {
566417c4 128 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '10 minutes')";
2c08214a 129 } else {
566417c4 130 $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 10 MINUTE))";
2c08214a
AD
131 }
132
133 // Test if there is a limit to number of updated feeds
134 $query_limit = "";
135 if($limit) $query_limit = sprintf("LIMIT %d", $limit);
136
98070db0
TK
137 // Update the least recently updated feeds first
138 $query_order = "ORDER BY last_updated";
139 if (DB_TYPE == "pgsql") $query_order .= " NULLS FIRST";
140
fce451a4 141 $query = "SELECT DISTINCT ttrss_feeds.feed_url, ttrss_feeds.last_updated
2c08214a
AD
142 FROM
143 ttrss_feeds, ttrss_users, ttrss_user_prefs
f4ae0f05 144 WHERE
2c08214a 145 ttrss_feeds.owner_uid = ttrss_users.id
f08426e3 146 AND ttrss_user_prefs.profile IS NULL
2c08214a
AD
147 AND ttrss_users.id = ttrss_user_prefs.owner_uid
148 AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
149 $login_thresh_qpart $update_limit_qpart
1c4421fc 150 $updstart_thresh_qpart
98070db0 151 $query_order $query_limit";
fce451a4
AD
152
153 // We search for feed needing update.
154 $result = db_query($query);
2c08214a 155
ca5ff2d9 156 if($debug) _debug(sprintf("Scheduled %d feeds to update...", db_num_rows($result)));
2c08214a
AD
157
158 // Here is a little cache magic in order to minimize risk of double feed updates.
159 $feeds_to_update = array();
160 while ($line = db_fetch_assoc($result)) {
a42c55f0 161 array_push($feeds_to_update, db_escape_string($line['feed_url']));
2c08214a
AD
162 }
163
164 // We update the feed last update started date before anything else.
165 // There is no lag due to feed contents downloads
166 // It prevent an other process to update the same feed.
1c4421fc
AD
167
168 if(count($feeds_to_update) > 0) {
169 $feeds_quoted = array();
170
171 foreach ($feeds_to_update as $feed) {
a42c55f0 172 array_push($feeds_quoted, "'" . db_escape_string($feed) . "'");
1c4421fc
AD
173 }
174
a42c55f0 175 db_query(sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
1c4421fc 176 WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
2c08214a
AD
177 }
178
8292d05b 179 $nf = 0;
2d9c5684 180 $bstarted = microtime(true);
8292d05b 181
2c08214a 182 // For each feed, we call the feed update function.
1c4421fc
AD
183 foreach ($feeds_to_update as $feed) {
184 if($debug) _debug("Base feed: $feed");
2c08214a 185
a42c55f0 186 //update_rss_feed($line["id"], true);
2c08214a 187
1c4421fc
AD
188 // since we have the data cached, we can deal with other feeds with the same url
189
15c762be 190 $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
ee0542ce
AD
191 FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
192 ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
193 ttrss_users.id = ttrss_user_prefs.owner_uid AND
194 ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
f08426e3 195 ttrss_user_prefs.profile IS NULL AND
9e84bab4
AD
196 feed_url = '".db_escape_string($feed)."'
197 $update_limit_qpart
1c4421fc 198 $login_thresh_qpart
5929a0c1 199 ORDER BY ttrss_feeds.id $query_limit");
1c4421fc
AD
200
201 if (db_num_rows($tmp_result) > 0) {
88edaa93
AD
202 $rss = false;
203
1c4421fc 204 while ($tline = db_fetch_assoc($tmp_result)) {
15c762be 205 if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
f08426e3 206
2d9c5684 207 $fstarted = microtime(true);
f77c7203 208 $rss = update_rss_feed($tline["id"], true, false);
4f71d743 209 _debug_suppress(false);
2d9c5684
AD
210
211 _debug(sprintf(" %.4f (sec)", microtime(true) - $fstarted));
212
8292d05b 213 ++$nf;
1c4421fc
AD
214 }
215 }
2c08214a
AD
216 }
217
2d9c5684
AD
218 if ($nf > 0) {
219 _debug(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
220 microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
221 }
222
f1611683
AD
223 require_once "digest.php";
224
2c08214a 225 // Send feed digests by email if needed.
a42c55f0 226 send_headlines_digests($debug);
2c08214a 227
8292d05b
AD
228 return $nf;
229
2c08214a
AD
230 } // function update_daemon_common
231
6022776d
AD
232 // this is used when subscribing
233 function set_basic_feed_info($feed) {
234
235 $feed = db_escape_string($feed);
236
9bd422c2 237 $result = db_query("SELECT feed_url,auth_pass,auth_login,auth_pass_encrypted
6022776d
AD
238 FROM ttrss_feeds WHERE id = '$feed'");
239
240 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
241 0, "auth_pass_encrypted"));
242
243 $auth_login = db_fetch_result($result, 0, "auth_login");
244 $auth_pass = db_fetch_result($result, 0, "auth_pass");
245
246 if ($auth_pass_encrypted) {
247 require_once "crypt.php";
248 $auth_pass = decrypt_string($auth_pass);
249 }
250
251 $fetch_url = db_fetch_result($result, 0, "feed_url");
252
253 $feed_data = fetch_file_contents($fetch_url, false,
254 $auth_login, $auth_pass, false,
255 FEED_FETCH_TIMEOUT_TIMEOUT,
256 0);
257
258 global $fetch_curl_used;
259
260 if (!$fetch_curl_used) {
261 $tmp = @gzdecode($feed_data);
262
263 if ($tmp) $feed_data = $tmp;
264 }
265
266 $feed_data = trim($feed_data);
267
268 $rss = new FeedParser($feed_data);
269 $rss->init();
270
271 if (!$rss->error()) {
272
273 $result = db_query("SELECT title, site_url FROM ttrss_feeds WHERE id = '$feed'");
274
275 $registered_title = db_fetch_result($result, 0, "title");
276 $orig_site_url = db_fetch_result($result, 0, "site_url");
277
278 $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
279 $feed_title = db_escape_string(mb_substr($rss->get_title(), 0, 199));
280
281 if ($feed_title && (!$registered_title || $registered_title == "[Unknown]")) {
282 db_query("UPDATE ttrss_feeds SET
283 title = '$feed_title' WHERE id = '$feed'");
284 }
285
286 if ($site_url && $orig_site_url != $site_url) {
287 db_query("UPDATE ttrss_feeds SET
288 site_url = '$site_url' WHERE id = '$feed'");
289 }
290 }
291 }
292
87764a50 293 // ignore_daemon is not used
88edaa93 294 function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
2c08214a 295
2c08214a
AD
296 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
297
4f71d743 298 _debug_suppress(!$debug_enabled);
68cccafc 299 _debug("start", $debug_enabled);
2c08214a 300
6bb96beb
AD
301 $result = db_query("SELECT title FROM ttrss_feeds
302 WHERE id = '$feed'");
303 $title = db_fetch_result($result, 0, "title");
304
305 // feed was batch-subscribed or something, we need to get basic info
306 // this is not optimal currently as it fetches stuff separately TODO: optimize
307 if ($title == "[Unknown]") {
308 _debug("setting basic feed info for $feed...");
309 set_basic_feed_info($feed);
310 }
311
a42c55f0 312 $result = db_query("SELECT id,update_interval,auth_login,
5ba1ddd4 313 feed_url,auth_pass,cache_images,
5321e775 314 mark_unread_on_update, owner_uid,
90e5f4f1 315 pubsub_state, auth_pass_encrypted,
0edf1d0d 316 feed_language
87764a50 317 FROM ttrss_feeds WHERE id = '$feed'");
2c08214a
AD
318
319 if (db_num_rows($result) == 0) {
68cccafc 320 _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
638fdf73 321 user_error("Attempt to update unknown/invalid feed $feed", E_USER_WARNING);
2c08214a
AD
322 return false;
323 }
324
2c08214a
AD
325 $owner_uid = db_fetch_result($result, 0, "owner_uid");
326 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
327 0, "mark_unread_on_update"));
2c08214a 328 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
044cff2d
AD
329 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
330 0, "auth_pass_encrypted"));
2c08214a 331
a42c55f0 332 db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
2c08214a
AD
333 WHERE id = '$feed'");
334
335 $auth_login = db_fetch_result($result, 0, "auth_login");
336 $auth_pass = db_fetch_result($result, 0, "auth_pass");
337
044cff2d
AD
338 if ($auth_pass_encrypted) {
339 require_once "crypt.php";
340 $auth_pass = decrypt_string($auth_pass);
341 }
342
2c08214a
AD
343 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
344 $fetch_url = db_fetch_result($result, 0, "feed_url");
df659891
AD
345 $feed_language = db_escape_string(mb_strtolower(db_fetch_result($result, 0, "feed_language")));
346 if (!$feed_language) $feed_language = 'english';
2c08214a 347
a42c55f0 348 $feed = db_escape_string($feed);
2c08214a 349
f074ffe9 350 $date_feed_processed = date('Y-m-d H:i');
2c08214a 351
865a3ed6 352 $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
f074ffe9 353
ee65bef4
AD
354 $pluginhost = new PluginHost();
355 $pluginhost->set_debug($debug_enabled);
356 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
357
358 $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
359 $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
360 $pluginhost->load_data();
361
88edaa93
AD
362 if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
363 _debug("using previously initialized parser object");
364 } else {
365 $rss_hash = false;
4f9cbdff 366
88edaa93 367 $force_refetch = isset($_REQUEST["force_refetch"]);
0edf1d0d 368 $feed_data = "";
687a4f59 369
95beaa14 370 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
0edf1d0d 371 $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, 0, $auth_login, $auth_pass);
95beaa14
RA
372 }
373
374 // try cache
375 if (!$feed_data &&
376 file_exists($cache_filename) &&
88edaa93
AD
377 is_readable($cache_filename) &&
378 !$auth_login && !$auth_pass &&
379 filemtime($cache_filename) > time() - 30) {
2c08214a 380
15d0528a 381 _debug("using local cache [$cache_filename].", $debug_enabled);
be574731 382
88edaa93 383 @$feed_data = file_get_contents($cache_filename);
52637d3b 384
88edaa93
AD
385 if ($feed_data) {
386 $rss_hash = sha1($feed_data);
387 }
f074ffe9 388
88edaa93
AD
389 } else {
390 _debug("local cache will not be used for this feed", $debug_enabled);
391 }
ee65bef4 392
95beaa14 393 // fetch feed from source
fd687300
AD
394 if (!$feed_data) {
395 _debug("fetching [$fetch_url]...", $debug_enabled);
312742db
AD
396
397 if (ini_get("open_basedir") && function_exists("curl_init")) {
398 _debug("not using CURL due to open_basedir restrictions");
399 }
400
fd687300
AD
401 $feed_data = fetch_file_contents($fetch_url, false,
402 $auth_login, $auth_pass, false,
403 $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
0edf1d0d 404 0);
3f6f0857 405
fd687300 406 global $fetch_curl_used;
3f6f0857 407
fd687300
AD
408 if (!$fetch_curl_used) {
409 $tmp = @gzdecode($feed_data);
3f6f0857 410
fd687300
AD
411 if ($tmp) $feed_data = $tmp;
412 }
1367bc3f 413
fd687300 414 $feed_data = trim($feed_data);
017401dd 415
fd687300
AD
416 _debug("fetch done.", $debug_enabled);
417
95beaa14
RA
418 // cache vanilla feed data for re-use
419 if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
420 $new_rss_hash = sha1($feed_data);
421
422 if ($new_rss_hash != $rss_hash) {
423 _debug("saving $cache_filename", $debug_enabled);
424 @file_put_contents($cache_filename, $feed_data);
425 }
426 }
4f9cbdff 427 }
017401dd 428
fd687300
AD
429 if (!$feed_data) {
430 global $fetch_last_error;
431 global $fetch_last_error_code;
f074ffe9 432
fd687300 433 _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
f074ffe9 434
fd687300 435 $error_escaped = '';
7a01dc77 436
fd687300
AD
437 // If-Modified-Since
438 if ($fetch_last_error_code != 304) {
439 $error_escaped = db_escape_string($fetch_last_error);
440 } else {
441 _debug("source claims data not modified, nothing to do.", $debug_enabled);
442 }
4f9cbdff 443
fd687300
AD
444 db_query(
445 "UPDATE ttrss_feeds SET last_error = '$error_escaped',
446 last_updated = NOW() WHERE id = '$feed'");
4f9cbdff 447
fd687300
AD
448 return;
449 }
f074ffe9
AD
450 }
451
1ffe3391 452 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
6791af0c 453 $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
017401dd
AD
454 }
455
5ef84097
AD
456 // set last update to now so if anything *simplepie* crashes later we won't be
457 // continuously failing on the same feed
5de40104 458 //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
5ef84097 459
4f9cbdff 460 if (!$rss) {
cd07592c
AD
461 $rss = new FeedParser($feed_data);
462 $rss->init();
5de51df7
AD
463 }
464
2c08214a
AD
465// print_r($rss);
466
a42c55f0 467 $feed = db_escape_string($feed);
2c08214a 468
19b3992b 469 if (!$rss->error()) {
2c08214a 470
d2a421e3 471 // We use local pluginhost here because we need to load different per-user feed plugins
1ffe3391 472 $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
4412b877 473
df659891 474 _debug("language: $feed_language", $debug_enabled);
68cccafc 475 _debug("processing feed data...", $debug_enabled);
2c08214a 476
a42c55f0 477// db_query("BEGIN");
2c08214a 478
382268c6
AD
479 if (DB_TYPE == "pgsql") {
480 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
481 } else {
482 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
483 }
484
fe4535e6 485 $result = db_query("SELECT owner_uid,favicon_avg_color,
382268c6
AD
486 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
487 favicon_needs_check
2c08214a
AD
488 FROM ttrss_feeds WHERE id = '$feed'");
489
382268c6
AD
490 $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
491 "favicon_needs_check"));
36490f11 492 $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
2c08214a
AD
493
494 $owner_uid = db_fetch_result($result, 0, "owner_uid");
495
a42c55f0 496 $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
2c08214a 497
cd07592c
AD
498 _debug("site_url: $site_url", $debug_enabled);
499 _debug("feed_title: " . $rss->get_title(), $debug_enabled);
500
687a4f59 501 if ($favicon_needs_check || $force_refetch) {
36490f11
AD
502
503 /* terrible hack: if we crash on floicon shit here, we won't check
560cbd8c 504 * the icon avgcolor again (unless the icon got updated) */
36490f11 505
560cbd8c
AD
506 $favicon_file = ICONS_DIR . "/$feed.ico";
507 $favicon_modified = @filemtime($favicon_file);
508
68cccafc 509 _debug("checking favicon...", $debug_enabled);
687a4f59 510
6f7798b6 511 check_feed_favicon($site_url, $feed);
560cbd8c
AD
512 $favicon_modified_new = @filemtime($favicon_file);
513
514 if ($favicon_modified_new > $favicon_modified)
515 $favicon_avg_color = '';
687a4f59 516
6ee0d4b0 517 if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
bc7a144d 518 require_once "colors.php";
687a4f59 519
aafd55ba
AD
520 db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
521 id = '$feed'");
522
6322ac79 523 $favicon_color = db_escape_string(
bc7a144d 524 calculate_avg_color($favicon_file));
63c323f7 525
bc7a144d 526 $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
36490f11 527 } else if ($favicon_avg_color == 'fail') {
84ceb6bd 528 _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
6ac722d5 529 }
687a4f59 530
a42c55f0 531 db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
bc7a144d 532 $favicon_colorstring
f2798eb6
AD
533 WHERE id = '$feed'");
534 }
2c08214a 535
68cccafc 536 _debug("loading filters & labels...", $debug_enabled);
2c08214a 537
a42c55f0 538 $filters = load_filters($feed, $owner_uid);
2c08214a 539
68cccafc 540 _debug("" . count($filters) . " filters loaded.", $debug_enabled);
2c08214a 541
19b3992b 542 $items = $rss->get_items();
2c08214a 543
19b3992b 544 if (!is_array($items)) {
68cccafc 545 _debug("no articles found.", $debug_enabled);
2c08214a 546
a42c55f0 547 db_query("UPDATE ttrss_feeds
2c08214a
AD
548 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
549
550 return; // no articles
551 }
552
553 if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
554
68cccafc 555 _debug("checking for PUSH hub...", $debug_enabled);
2c08214a
AD
556
557 $feed_hub_url = false;
2c08214a 558
19b3992b 559 $links = $rss->get_links('hub');
2c08214a 560
19b3992b
AD
561 if ($links && is_array($links)) {
562 foreach ($links as $l) {
563 $feed_hub_url = $l;
564 break;
2c08214a
AD
565 }
566 }
567
68cccafc 568 _debug("feed hub url: $feed_hub_url", $debug_enabled);
2c08214a 569
d8179cb9
AD
570 $feed_self_url = $fetch_url;
571
572 $links = $rss->get_links('self');
573
574 if ($links && is_array($links)) {
575 foreach ($links as $l) {
576 $feed_self_url = $l;
577 break;
578 }
579 }
580
581 _debug("feed self url = $feed_self_url");
582
583 if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
2c08214a
AD
584 !ini_get("open_basedir")) {
585
586 require_once 'lib/pubsubhubbub/subscriber.php';
587
588 $callback_url = get_self_url_prefix() .
589 "/public.php?op=pubsub&id=$feed";
590
591 $s = new Subscriber($feed_hub_url, $callback_url);
592
d8179cb9 593 $rc = $s->subscribe($feed_self_url);
2c08214a 594
5ba1ddd4 595 _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
2c08214a 596
a42c55f0 597 db_query("UPDATE ttrss_feeds SET pubsub_state = 1
2c08214a
AD
598 WHERE id = '$feed'");
599 }
600 }
601
68cccafc 602 _debug("processing articles...", $debug_enabled);
2c08214a 603
6c9f3d4a
AD
604 $tstart = time();
605
19b3992b 606 foreach ($items as $item) {
5d56d100 607 if ($_REQUEST['xdebug'] == 3) {
2c08214a
AD
608 print_r($item);
609 }
610
6c9f3d4a
AD
611 if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
612 _debug("looks like there's too many articles to process at once, breaking out", $debug_enabled);
613 break;
614 }
615
19b3992b
AD
616 $entry_guid = $item->get_id();
617 if (!$entry_guid) $entry_guid = $item->get_link();
618 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
2c08214a
AD
619 if (!$entry_guid) continue;
620
3a4c8973
AD
621 $entry_guid = "$owner_uid,$entry_guid";
622
a42c55f0 623 $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
5e3d5480 624
68cccafc 625 _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
5e3d5480 626
2c08214a
AD
627 $entry_timestamp = "";
628
04d2f9c8
AD
629 $entry_timestamp = $item->get_date();
630
631 _debug("orig date: " . $item->get_date(), $debug_enabled);
2c08214a 632
30123fe6 633 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
2c08214a 634 $entry_timestamp = time();
2c08214a
AD
635 }
636
637 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
638
68cccafc 639 _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
2c08214a 640
99429e57
AD
641// $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
642// $entry_title = decode_numeric_entities($entry_title);
643 $entry_title = $item->get_title();
1b35d30c 644
5d56d100 645 $entry_link = rewrite_relative_url($site_url, $item->get_link());
2c08214a 646
68cccafc
AD
647 _debug("title $entry_title", $debug_enabled);
648 _debug("link $entry_link", $debug_enabled);
2c08214a
AD
649
650 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
651
19b3992b
AD
652 $entry_content = $item->get_content();
653 if (!$entry_content) $entry_content = $item->get_description();
2c08214a
AD
654
655 if ($_REQUEST["xdebug"] == 2) {
9ec10352 656 print "content: ";
0bc503ff 657 print htmlspecialchars($entry_content);
3c696512 658 print "\n";
2c08214a
AD
659 }
660
cd07592c
AD
661 $entry_comments = $item->get_comments_url();
662 $entry_author = $item->get_author();
2c08214a 663
a42c55f0 664 $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
2c08214a 665
ce2b219c
AD
666 $entry_comments = db_escape_string(mb_substr(trim($entry_comments), 0, 245));
667 $entry_author = db_escape_string(mb_substr(trim($entry_author), 0, 245));
2c08214a 668
d4992d6b 669 $num_comments = (int) $item->get_comments_count();
2c08214a 670
68cccafc
AD
671 _debug("author $entry_author", $debug_enabled);
672 _debug("num_comments: $num_comments", $debug_enabled);
ee78f81c 673 _debug("looking for tags...", $debug_enabled);
2c08214a
AD
674
675 // parse <category> entries into tags
676
677 $additional_tags = array();
678
19b3992b 679 $additional_tags_src = $item->get_categories();
2c08214a 680
19b3992b
AD
681 if (is_array($additional_tags_src)) {
682 foreach ($additional_tags_src as $tobj) {
cd07592c 683 array_push($additional_tags, $tobj);
2c08214a 684 }
19b3992b 685 }
2c08214a 686
fa6fbd36 687 $entry_tags = array_unique($additional_tags);
2c08214a
AD
688
689 for ($i = 0; $i < count($entry_tags); $i++)
690 $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
691
ee78f81c
AD
692 _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
693
68cccafc 694 _debug("done collecting data.", $debug_enabled);
2c08214a 695
3318d324 696 $result = db_query("SELECT id, content_hash, lang FROM ttrss_entries
b1840673 697 WHERE guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed'");
b30abdad
AD
698
699 if (db_num_rows($result) != 0) {
b1840673
AD
700 $base_entry_id = db_fetch_result($result, 0, "id");
701 $entry_stored_hash = db_fetch_result($result, 0, "content_hash");
a29fe121 702 $article_labels = get_article_labels($base_entry_id, $owner_uid);
3318d324 703 $entry_language = db_fetch_result($result, 0, "lang");
a8ac7661 704
040aa540 705 $existing_tags = get_article_tags($base_entry_id, $owner_uid);
a8ac7661
AD
706 $entry_tags = array_unique(array_merge($entry_tags, $existing_tags));
707
b30abdad 708 } else {
b1840673
AD
709 $base_entry_id = false;
710 $entry_stored_hash = "";
a29fe121 711 $article_labels = array();
3318d324 712 $entry_language = "";
b30abdad
AD
713 }
714
455b1401 715 $article = array("owner_uid" => $owner_uid, // read only
b30abdad 716 "guid" => $entry_guid, // read only
59e83455 717 "guid_hashed" => $entry_guid_hashed, // read only
19b3992b
AD
718 "title" => $entry_title,
719 "content" => $entry_content,
720 "link" => $entry_link,
a29fe121 721 "labels" => $article_labels, // current limitation: can add labels to article, can't remove them
19b3992b 722 "tags" => $entry_tags,
e02555c1 723 "author" => $entry_author,
c9299c28 724 "force_catchup" => false, // ugly hack for the time being
6de3a1be 725 "score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
3318d324 726 "language" => $entry_language,
f73e03e0
AD
727 "feed" => array("id" => $feed,
728 "fetch_url" => $fetch_url,
babfadbf
J
729 "site_url" => $site_url,
730 "cache_images" => $cache_images)
f73e03e0 731 );
cc85704f 732
b1840673 733 $entry_plugin_data = "";
eb16bd9f 734 $entry_current_hash = calculate_article_hash($article, $pluginhost);
b1840673
AD
735
736 _debug("article hash: $entry_current_hash [stored=$entry_stored_hash]", $debug_enabled);
737
522e8b35 738 if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
b1840673
AD
739 _debug("stored article seems up to date [IID: $base_entry_id], updating timestamp only", $debug_enabled);
740
741 // we keep encountering the entry in feeds, so we need to
742 // update date_updated column so that we don't get horrible
743 // dupes when the entry gets purged and reinserted again e.g.
744 // in the case of SLOW SLOW OMG SLOW updating feeds
745
746 $base_entry_id = db_fetch_result($result, 0, "id");
747
748 db_query("UPDATE ttrss_entries SET date_updated = NOW()
749 WHERE id = '$base_entry_id'");
750
5bdcb8fd 751 continue;
b1840673
AD
752 }
753
754 _debug("hash differs, applying plugin filters:", $debug_enabled);
755
1ffe3391 756 foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
b1840673
AD
757 _debug("... " . get_class($plugin), $debug_enabled);
758
759 $start = microtime(true);
19b3992b 760 $article = $plugin->hook_article_filter($article);
0084f0d1 761
b1840673
AD
762 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
763
764 $entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
cc85704f
AD
765 }
766
0bc503ff
AD
767 if ($_REQUEST["xdebug"] == 2) {
768 print "processed content: ";
769 print htmlspecialchars($article["content"]);
770 print "\n";
771 }
772
b1840673
AD
773 $entry_plugin_data = db_escape_string($entry_plugin_data);
774
775 _debug("plugin data: $entry_plugin_data", $debug_enabled);
776
35c12dc4
AD
777 // Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
778 if (DB_TYPE == "mysql") {
779 foreach ($article as $k => $v) {
35c37354
AD
780
781 // i guess we'll have to take the risk of 4byte unicode labels & tags here
dae16f72 782 if (is_string($article[$k])) {
35c37354
AD
783 $article[$k] = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $v);
784 }
35c12dc4
AD
785 }
786 }
787
b8774453
AD
788 /* Collect article tags here so we could filter by them: */
789
790 $article_filters = get_article_filters($filters, $article["title"],
791 $article["content"], $article["link"], 0, $article["author"],
792 $article["tags"]);
793
794 if ($debug_enabled) {
795 _debug("article filters: ", $debug_enabled);
796 if (count($article_filters) != 0) {
797 print_r($article_filters);
798 }
799 }
800
801 $plugin_filter_names = find_article_filters($article_filters, "plugin");
802 $plugin_filter_actions = $pluginhost->get_filter_actions();
803
804 if (count($plugin_filter_names) > 0) {
805 _debug("applying plugin filter actions...", $debug_enabled);
806
807 foreach ($plugin_filter_names as $pfn) {
808 list($pfclass,$pfaction) = explode(":", $pfn["param"]);
809
810 if (isset($plugin_filter_actions[$pfclass])) {
811 $plugin = $pluginhost->get_plugin($pfclass);
812
813 _debug("... $pfclass: $pfaction", $debug_enabled);
814
815 if ($plugin) {
816 $start = microtime(true);
817 $article = $plugin->hook_article_filter_action($article, $pfaction);
818
819 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
820 } else {
821 _debug("??? $pfclass: plugin object not found.");
822 }
823 } else {
824 _debug("??? $pfclass: filter plugin not registered.");
825 }
826 }
827 }
828
19b3992b 829 $entry_tags = $article["tags"];
a42c55f0
AD
830 $entry_guid = db_escape_string($entry_guid);
831 $entry_title = db_escape_string($article["title"]);
832 $entry_author = db_escape_string($article["author"]);
833 $entry_link = db_escape_string($article["link"]);
f935d98e 834 $entry_content = $article["content"]; // escaped below
c9299c28 835 $entry_force_catchup = $article["force_catchup"];
a29fe121 836 $article_labels = $article["labels"];
6de3a1be 837 $entry_score_modifier = (int) $article["score_modifier"];
3318d324 838 $entry_language = db_escape_string($article["language"]);
a29fe121
AD
839
840 if ($debug_enabled) {
841 _debug("article labels:", $debug_enabled);
842 print_r($article_labels);
843 }
c9299c28
AD
844
845 _debug("force catchup: $entry_force_catchup");
f935d98e 846
0a3fd79b 847 if ($cache_images && is_writable(CACHE_DIR . '/images'))
f0bd8e65 848 cache_images($entry_content, $site_url, $debug_enabled);
0a3fd79b 849
a42c55f0 850 $entry_content = db_escape_string($entry_content, false);
1f45c857 851
07877caf 852 //db_query("BEGIN");
2c08214a 853
a42c55f0 854 $result = db_query("SELECT id FROM ttrss_entries
5e3d5480 855 WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
9e222305 856
2c08214a
AD
857 if (db_num_rows($result) == 0) {
858
68cccafc 859 _debug("base guid [$entry_guid] not found", $debug_enabled);
2c08214a
AD
860
861 // base post entry does not exist, create it
862
6322ac79 863 $result = db_query(
2c08214a
AD
864 "INSERT INTO ttrss_entries
865 (title,
866 guid,
867 link,
868 updated,
869 content,
870 content_hash,
871 no_orig_date,
872 date_updated,
873 date_entered,
874 comments,
875 num_comments,
b30abdad 876 plugin_data,
6b461797 877 lang,
2c08214a
AD
878 author)
879 VALUES
880 ('$entry_title',
5e3d5480 881 '$entry_guid_hashed',
2c08214a
AD
882 '$entry_link',
883 '$entry_timestamp_fmt',
884 '$entry_content',
b1840673 885 '$entry_current_hash',
5ba1ddd4 886 false,
2c08214a 887 NOW(),
be574731 888 '$date_feed_processed',
2c08214a
AD
889 '$entry_comments',
890 '$num_comments',
b30abdad 891 '$entry_plugin_data',
6b461797 892 '$entry_language',
2c08214a 893 '$entry_author')");
e8291805 894
2c08214a 895 } else {
2c08214a 896 $base_entry_id = db_fetch_result($result, 0, "id");
2c08214a
AD
897 }
898
899 // now it should exist, if not - bad luck then
900
b1840673 901 $result = db_query("SELECT id FROM ttrss_entries
5e3d5480 902 WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
2c08214a
AD
903
904 $entry_ref_id = 0;
905 $entry_int_id = 0;
906
907 if (db_num_rows($result) == 1) {
908
68cccafc 909 _debug("base guid found, checking for user record", $debug_enabled);
2c08214a 910
2c08214a
AD
911 $ref_id = db_fetch_result($result, 0, "id");
912 $entry_ref_id = $ref_id;
913
5e3d5480
AD
914 /* $stored_guid = db_fetch_result($result, 0, "guid");
915 if ($stored_guid != $entry_guid_hashed) {
68cccafc 916 if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
5e3d5480 917
a42c55f0 918 db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
5e3d5480
AD
919 id = '$ref_id'");
920 } */
921
2c08214a 922 if (find_article_filter($article_filters, "filter")) {
07877caf 923 //db_query("COMMIT"); // close transaction in progress
2c08214a
AD
924 continue;
925 }
926
6de3a1be 927 $score = calculate_article_score($article_filters) + $entry_score_modifier;
2c08214a 928
6de3a1be 929 _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
2c08214a 930
4f186b1f
AD
931 // check for user post link to main table
932
2c08214a 933 $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
5bdcb8fd 934 ref_id = '$ref_id' AND owner_uid = '$owner_uid'";
2c08214a
AD
935
936// if ($_REQUEST["xdebug"]) print "$query\n";
937
a42c55f0 938 $result = db_query($query);
2c08214a
AD
939
940 // okay it doesn't exist - create user entry
941 if (db_num_rows($result) == 0) {
942
68cccafc 943 _debug("user record not found, creating...", $debug_enabled);
2c08214a 944
c9299c28 945 if ($score >= -500 && !find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
2c08214a
AD
946 $unread = 'true';
947 $last_read_qpart = 'NULL';
948 } else {
949 $unread = 'false';
950 $last_read_qpart = 'NOW()';
951 }
952
953 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
954 $marked = 'true';
955 } else {
956 $marked = 'false';
957 }
958
959 if (find_article_filter($article_filters, 'publish')) {
960 $published = 'true';
961 } else {
962 $published = 'false';
963 }
964
2ea9bbfd
AD
965 // N-grams
966
c9299c28 967 /* if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
2ea9bbfd 968
a42c55f0 969 $result = db_query("SELECT COUNT(*) AS similar FROM
2ea9bbfd
AD
970 ttrss_entries,ttrss_user_entries
971 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
972 AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
973 AND owner_uid = $owner_uid");
974
975 $ngram_similar = db_fetch_result($result, 0, "similar");
976
68cccafc 977 _debug("N-gram similar results: $ngram_similar", $debug_enabled);
2ea9bbfd
AD
978
979 if ($ngram_similar > 0) {
980 $unread = 'false';
981 }
c9299c28 982 } */
2ea9bbfd 983
7873d588
AD
984 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
985 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
986
6322ac79 987 $result = db_query(
2c08214a
AD
988 "INSERT INTO ttrss_user_entries
989 (ref_id, owner_uid, feed_id, unread, last_read, marked,
7873d588
AD
990 published, score, tag_cache, label_cache, uuid,
991 last_marked, last_published)
2c08214a 992 VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
7873d588
AD
993 $last_read_qpart, $marked, $published, '$score', '', '',
994 '', $last_marked, $last_published)");
2c08214a
AD
995
996 if (PUBSUBHUBBUB_HUB && $published == 'true') {
997 $rss_link = get_self_url_prefix() .
998 "/public.php?op=rss&id=-2&key=" .
a42c55f0 999 get_feed_access_key(-2, false, $owner_uid);
2c08214a
AD
1000
1001 $p = new Publisher(PUBSUBHUBBUB_HUB);
1002
5ba1ddd4 1003 /* $pubsub_result = */ $p->publish_update($rss_link);
2c08214a
AD
1004 }
1005
6322ac79 1006 $result = db_query(
2c08214a
AD
1007 "SELECT int_id FROM ttrss_user_entries WHERE
1008 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
1009 feed_id = '$feed' LIMIT 1");
1010
1011 if (db_num_rows($result) == 1) {
1012 $entry_int_id = db_fetch_result($result, 0, "int_id");
1013 }
1014 } else {
68cccafc 1015 _debug("user record FOUND", $debug_enabled);
2c08214a
AD
1016
1017 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
1018 $entry_int_id = db_fetch_result($result, 0, "int_id");
1019 }
1020
68cccafc 1021 _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
2c08214a 1022
e854442e 1023 if (DB_TYPE == "pgsql") {
978989bb 1024 $tsvector_combined = db_escape_string(mb_substr($entry_title . ' ' . strip_tags(str_replace('<', ' <', $entry_content)),
e854442e
AD
1025 0, 1000000));
1026
df659891 1027 $tsvector_qpart = "tsvector_combined = to_tsvector('$feed_language', '$tsvector_combined'),";
e854442e
AD
1028
1029 } else {
1030 $tsvector_qpart = "";
1031 }
1032
b1840673
AD
1033 db_query("UPDATE ttrss_entries
1034 SET title = '$entry_title',
1035 content = '$entry_content',
1036 content_hash = '$entry_current_hash',
1037 updated = '$entry_timestamp_fmt',
e854442e 1038 $tsvector_qpart
b1840673
AD
1039 num_comments = '$num_comments',
1040 plugin_data = '$entry_plugin_data',
1041 author = '$entry_author',
1042 lang = '$entry_language'
1043 WHERE id = '$ref_id'");
1044
59e83455
AD
1045 // update aux data
1046 db_query("UPDATE ttrss_user_entries
1047 SET score = '$score' WHERE ref_id = '$ref_id'");
1048
b1840673 1049 if ($mark_unread_on_update) {
24e6ff5d
AD
1050 _debug("article updated, marking unread as requested.", $debug_enabled);
1051
b1840673
AD
1052 db_query("UPDATE ttrss_user_entries
1053 SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
2c08214a
AD
1054 }
1055 }
1056
07877caf 1057 //db_query("COMMIT");
2c08214a 1058
a29fe121
AD
1059 _debug("assigning labels [other]...", $debug_enabled);
1060
1061 foreach ($article_labels as $label) {
1062 label_add_article($entry_ref_id, $label[1], $owner_uid);
1063 }
1064
1065 _debug("assigning labels [filters]...", $debug_enabled);
2c08214a 1066
a42c55f0 1067 assign_article_to_label_filters($entry_ref_id, $article_filters,
b24504b1 1068 $owner_uid, $article_labels);
2c08214a 1069
68cccafc 1070 _debug("looking for enclosures...", $debug_enabled);
2c08214a
AD
1071
1072 // enclosures
1073
1074 $enclosures = array();
1075
19b3992b 1076 $encs = $item->get_enclosures();
2c08214a 1077
19b3992b
AD
1078 if (is_array($encs)) {
1079 foreach ($encs as $e) {
1080 $e_item = array(
523bd90b 1081 $e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
2c08214a 1082 array_push($enclosures, $e_item);
2c08214a
AD
1083 }
1084 }
1085
2c08214a 1086 if ($debug_enabled) {
68cccafc 1087 _debug("article enclosures:", $debug_enabled);
2c08214a
AD
1088 print_r($enclosures);
1089 }
1090
dab229cd 1091 //db_query("BEGIN");
2c08214a 1092
5c54e683
AD
1093// debugging
1094// db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
1095
2c08214a 1096 foreach ($enclosures as $enc) {
a42c55f0
AD
1097 $enc_url = db_escape_string($enc[0]);
1098 $enc_type = db_escape_string($enc[1]);
1099 $enc_dur = db_escape_string($enc[2]);
5c54e683 1100 $enc_title = db_escape_string($enc[3]);
523bd90b
FE
1101 $enc_width = intval($enc[4]);
1102 $enc_height = intval($enc[5]);
2c08214a 1103
a42c55f0 1104 $result = db_query("SELECT id FROM ttrss_enclosures
2c08214a
AD
1105 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
1106
1107 if (db_num_rows($result) == 0) {
a42c55f0 1108 db_query("INSERT INTO ttrss_enclosures
523bd90b
FE
1109 (content_url, content_type, title, duration, post_id, width, height) VALUES
1110 ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
2c08214a
AD
1111 }
1112 }
1113
dab229cd 1114 //db_query("COMMIT");
2c08214a
AD
1115
1116 // check for manual tags (we have to do it here since they're loaded from filters)
1117
1118 foreach ($article_filters as $f) {
6aff7845 1119 if ($f["type"] == "tag") {
2c08214a 1120
6aff7845 1121 $manual_tags = trim_array(explode(",", $f["param"]));
2c08214a
AD
1122
1123 foreach ($manual_tags as $tag) {
1124 if (tag_is_valid($tag)) {
1125 array_push($entry_tags, $tag);
1126 }
1127 }
1128 }
1129 }
1130
1131 // Skip boring tags
1132
6322ac79 1133 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
2c08214a
AD
1134 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1135
1136 $filtered_tags = array();
1137 $tags_to_cache = array();
1138
1139 if ($entry_tags && is_array($entry_tags)) {
1140 foreach ($entry_tags as $tag) {
1141 if (array_search($tag, $boring_tags) === false) {
1142 array_push($filtered_tags, $tag);
1143 }
1144 }
1145 }
1146
1147 $filtered_tags = array_unique($filtered_tags);
1148
1149 if ($debug_enabled) {
68cccafc 1150 _debug("filtered article tags:", $debug_enabled);
2c08214a
AD
1151 print_r($filtered_tags);
1152 }
1153
1154 // Save article tags in the database
1155
1156 if (count($filtered_tags) > 0) {
1157
dab229cd 1158 //db_query("BEGIN");
2c08214a
AD
1159
1160 foreach ($filtered_tags as $tag) {
1161
1162 $tag = sanitize_tag($tag);
a42c55f0 1163 $tag = db_escape_string($tag);
2c08214a
AD
1164
1165 if (!tag_is_valid($tag)) continue;
1166
a42c55f0 1167 $result = db_query("SELECT id FROM ttrss_tags
2c08214a
AD
1168 WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1169 owner_uid = '$owner_uid' LIMIT 1");
1170
1171 if ($result && db_num_rows($result) == 0) {
1172
a42c55f0 1173 db_query("INSERT INTO ttrss_tags
2c08214a
AD
1174 (owner_uid,tag_name,post_int_id)
1175 VALUES ('$owner_uid','$tag', '$entry_int_id')");
1176 }
1177
1178 array_push($tags_to_cache, $tag);
1179 }
1180
1181 /* update the cache */
1182
1183 $tags_to_cache = array_unique($tags_to_cache);
1184
a42c55f0 1185 $tags_str = db_escape_string(join(",", $tags_to_cache));
2c08214a 1186
a42c55f0 1187 db_query("UPDATE ttrss_user_entries
2c08214a
AD
1188 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1189 AND owner_uid = $owner_uid");
1190
dab229cd 1191 //db_query("COMMIT");
2c08214a
AD
1192 }
1193
68cccafc 1194 _debug("article processed", $debug_enabled);
2c08214a
AD
1195 }
1196
68cccafc 1197 _debug("purging feed...", $debug_enabled);
2c08214a 1198
a42c55f0 1199 purge_feed($feed, 0, $debug_enabled);
2c08214a 1200
a42c55f0 1201 db_query("UPDATE ttrss_feeds
f2798eb6 1202 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
2c08214a 1203
a42c55f0 1204// db_query("COMMIT");
2c08214a
AD
1205
1206 } else {
1207
a42c55f0 1208 $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
2c08214a 1209
4ad04ee2
AD
1210 _debug("fetch error: $error_msg", $debug_enabled);
1211
1212 if (count($rss->errors()) > 1) {
1213 foreach ($rss->errors() as $error) {
1214 _debug("+ $error");
1215 }
1216 }
2c08214a 1217
6322ac79 1218 db_query(
2c08214a 1219 "UPDATE ttrss_feeds SET last_error = '$error_msg',
88edaa93 1220 last_updated = NOW() WHERE id = '$feed'");
2c08214a 1221
88edaa93
AD
1222 unset($rss);
1223 }
2c08214a 1224
68cccafc 1225 _debug("done", $debug_enabled);
88edaa93
AD
1226
1227 return $rss;
2c08214a
AD
1228 }
1229
3c696512 1230 function cache_images($html, $site_url, $debug) {
3c696512
AD
1231 libxml_use_internal_errors(true);
1232
1233 $charset_hack = '<head>
1234 <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1235 </head>';
1236
1237 $doc = new DOMDocument();
1238 $doc->loadHTML($charset_hack . $html);
1239 $xpath = new DOMXPath($doc);
1240
1241 $entries = $xpath->query('(//img[@src])');
1242
1243 foreach ($entries as $entry) {
1244 if ($entry->hasAttribute('src')) {
1245 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1246
1247 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1248
1249 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1250
1251 if (!file_exists($local_filename)) {
1252 $file_content = fetch_file_contents($src);
1253
a3d50184 1254 if ($file_content && strlen($file_content) > _MIN_CACHE_IMAGE_SIZE) {
3c696512
AD
1255 file_put_contents($local_filename, $file_content);
1256 }
4a27966e
J
1257 } else {
1258 touch($local_filename);
3c696512 1259 }
3c696512
AD
1260 }
1261 }
3c696512
AD
1262 }
1263
a42c55f0 1264 function expire_error_log($debug) {
e2261e17
AD
1265 if ($debug) _debug("Removing old error log entries...");
1266
1267 if (DB_TYPE == "pgsql") {
a42c55f0 1268 db_query("DELETE FROM ttrss_error_log
e2261e17
AD
1269 WHERE created_at < NOW() - INTERVAL '7 days'");
1270 } else {
a42c55f0 1271 db_query("DELETE FROM ttrss_error_log
e2261e17
AD
1272 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1273 }
1274
1275 }
1276
2a91b6ff 1277 function expire_lock_files($debug) {
65465085 1278 //if ($debug) _debug("Removing old lock files...");
2a91b6ff
AD
1279
1280 $num_deleted = 0;
1281
1282 if (is_writable(LOCK_DIRECTORY)) {
1283 $files = glob(LOCK_DIRECTORY . "/*.lock");
1284
1285 if ($files) {
1286 foreach ($files as $file) {
11344971 1287 if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
2a91b6ff
AD
1288 unlink($file);
1289 ++$num_deleted;
1290 }
1291 }
1292 }
1293 }
1294
65465085 1295 if ($debug) _debug("Removed $num_deleted old lock files.");
2a91b6ff
AD
1296 }
1297
3c696512 1298 function expire_cached_files($debug) {
3306daec 1299 foreach (array("simplepie", "images", "export", "upload") as $dir) {
3c696512 1300 $cache_dir = CACHE_DIR . "/$dir";
2c08214a 1301
65465085 1302// if ($debug) _debug("Expiring $cache_dir");
2c08214a 1303
3c696512
AD
1304 $num_deleted = 0;
1305
1306 if (is_writable($cache_dir)) {
1307 $files = glob("$cache_dir/*");
1308
2a91b6ff 1309 if ($files) {
2ab20c31
AD
1310 foreach ($files as $file) {
1311 if (time() - filemtime($file) > 86400*7) {
1312 unlink($file);
3c696512 1313
2ab20c31
AD
1314 ++$num_deleted;
1315 }
3c696512
AD
1316 }
1317 }
2a91b6ff 1318 }
3c696512 1319
65465085 1320 if ($debug) _debug("$cache_dir: removed $num_deleted files.");
3c696512
AD
1321 }
1322 }
2c08214a 1323
a3e0bdcf
AD
1324 /**
1325 * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1326 * Returns the url query as associative array
1327 *
1328 * @param string query
1329 * @return array params
1330 */
1331 function convertUrlQuery($query) {
1332 $queryParts = explode('&', $query);
1333
1334 $params = array();
1335
1336 foreach ($queryParts as $param) {
1337 $item = explode('=', $param);
1338 $params[$item[0]] = $item[1];
1339 }
1340
1341 return $params;
1342 }
92c14e9d 1343
48cb2536 1344 function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
92c14e9d
AD
1345 $matches = array();
1346
1347 foreach ($filters as $filter) {
1348 $match_any_rule = $filter["match_any_rule"];
a3a896a1 1349 $inverse = $filter["inverse"];
92c14e9d
AD
1350 $filter_match = false;
1351
1352 foreach ($filter["rules"] as $rule) {
1353 $match = false;
ffa1bd7b 1354 $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
a3a896a1 1355 $rule_inverse = $rule["inverse"];
92c14e9d
AD
1356
1357 if (!$reg_exp)
1358 continue;
1359
1360 switch ($rule["type"]) {
1361 case "title":
1362 $match = @preg_match("/$reg_exp/i", $title);
1363 break;
1364 case "content":
d03ae73e
AD
1365 // we don't need to deal with multiline regexps
1366 $content = preg_replace("/[\r\n\t]/", "", $content);
1367
92c14e9d
AD
1368 $match = @preg_match("/$reg_exp/i", $content);
1369 break;
1370 case "both":
d03ae73e
AD
1371 // we don't need to deal with multiline regexps
1372 $content = preg_replace("/[\r\n\t]/", "", $content);
1373
72d1d067 1374 $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
92c14e9d
AD
1375 break;
1376 case "link":
1377 $match = @preg_match("/$reg_exp/i", $link);
1378 break;
1379 case "author":
1380 $match = @preg_match("/$reg_exp/i", $author);
1381 break;
1382 case "tag":
7b80b5e1
DK
1383 foreach ($tags as $tag) {
1384 if (@preg_match("/$reg_exp/i", $tag)) {
1385 $match = true;
1386 break;
1387 }
1388 }
92c14e9d
AD
1389 break;
1390 }
1391
a3a896a1
AD
1392 if ($rule_inverse) $match = !$match;
1393
92c14e9d
AD
1394 if ($match_any_rule) {
1395 if ($match) {
1396 $filter_match = true;
1397 break;
1398 }
1399 } else {
1400 $filter_match = $match;
1401 if (!$match) {
1402 break;
1403 }
1404 }
1405 }
1406
a3a896a1
AD
1407 if ($inverse) $filter_match = !$filter_match;
1408
92c14e9d
AD
1409 if ($filter_match) {
1410 foreach ($filter["actions"] AS $action) {
1411 array_push($matches, $action);
5e736e45
AD
1412
1413 // if Stop action encountered, perform no further processing
fd3e5e8d 1414 if (isset($action["type"]) && $action["type"] == "stop") return $matches;
92c14e9d
AD
1415 }
1416 }
1417 }
1418
1419 return $matches;
1420 }
1421
1422 function find_article_filter($filters, $filter_name) {
1423 foreach ($filters as $f) {
1424 if ($f["type"] == $filter_name) {
1425 return $f;
1426 };
1427 }
1428 return false;
1429 }
1430
1431 function find_article_filters($filters, $filter_name) {
1432 $results = array();
1433
1434 foreach ($filters as $f) {
1435 if ($f["type"] == $filter_name) {
1436 array_push($results, $f);
1437 };
1438 }
1439 return $results;
1440 }
1441
1442 function calculate_article_score($filters) {
1443 $score = 0;
1444
1445 foreach ($filters as $f) {
1446 if ($f["type"] == "score") {
1447 $score += $f["param"];
1448 };
1449 }
1450 return $score;
1451 }
1452
b24504b1
AD
1453 function labels_contains_caption($labels, $caption) {
1454 foreach ($labels as $label) {
1455 if ($label[1] == $caption) {
1456 return true;
1457 }
1458 }
1459
1460 return false;
1461 }
1462
a42c55f0 1463 function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
92c14e9d
AD
1464 foreach ($filters as $f) {
1465 if ($f["type"] == "label") {
b24504b1 1466 if (!labels_contains_caption($article_labels, $f["param"])) {
a42c55f0 1467 label_add_article($id, $f["param"], $owner_uid);
b24504b1
AD
1468 }
1469 }
92c14e9d
AD
1470 }
1471 }
87764a50 1472
87d7e850
AD
1473 function make_guid_from_title($title) {
1474 return preg_replace("/[ \"\',.:;]/", "-",
1475 mb_strtolower(strip_tags($title), 'utf-8'));
1476 }
1477
d1f3fa97 1478 /* function verify_feed_xml($feed_data) {
ebec81a6
AD
1479 libxml_use_internal_errors(true);
1480 $doc = new DOMDocument();
1481 $doc->loadXML($feed_data);
1482 $error = libxml_get_last_error();
1483 libxml_clear_errors();
1484 return $error;
d1f3fa97 1485 } */
87d7e850 1486
168cf351
AD
1487 function cleanup_counters_cache($debug) {
1488 $result = db_query("DELETE FROM ttrss_counters_cache
1489 WHERE feed_id > 0 AND
1490 (SELECT COUNT(id) FROM ttrss_feeds WHERE
1491 id = feed_id AND
1492 ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
1493 $frows = db_affected_rows($result);
1494
1495 $result = db_query("DELETE FROM ttrss_cat_counters_cache
1496 WHERE feed_id > 0 AND
1497 (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
1498 id = feed_id AND
1499 ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
1500 $crows = db_affected_rows($result);
1501
1502 _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
1503 }
1504
e2cf81e2
AD
1505 function housekeeping_common($debug) {
1506 expire_cached_files($debug);
1507 expire_lock_files($debug);
1508 expire_error_log($debug);
1509
1510 $count = update_feedbrowser_cache();
1511 _debug("Feedbrowser updated, $count feeds processed.");
1512
1513 purge_orphans( true);
168cf351 1514 cleanup_counters_cache($debug);
e2cf81e2 1515
9b736a20
AD
1516 //$rc = cleanup_tags( 14, 50000);
1517 //_debug("Cleaned $rc cached tags.");
8e470220 1518
00f22824 1519 PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
8e470220 1520
e2cf81e2 1521 }
2c08214a 1522?>