]> git.wh0rd.org - tt-rss.git/blame - include/rssfuncs.php
set_basic_feed_info: bail out if feed doesn't exist
[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
d77bd1a5
AD
233 if (pg_num_rows($result == 0)) return false;
234
6022776d
AD
235 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
236 0, "auth_pass_encrypted"));
237
238 $auth_login = db_fetch_result($result, 0, "auth_login");
239 $auth_pass = db_fetch_result($result, 0, "auth_pass");
240
241 if ($auth_pass_encrypted) {
242 require_once "crypt.php";
243 $auth_pass = decrypt_string($auth_pass);
244 }
245
246 $fetch_url = db_fetch_result($result, 0, "feed_url");
247
248 $feed_data = fetch_file_contents($fetch_url, false,
249 $auth_login, $auth_pass, false,
250 FEED_FETCH_TIMEOUT_TIMEOUT,
251 0);
252
253 global $fetch_curl_used;
254
255 if (!$fetch_curl_used) {
256 $tmp = @gzdecode($feed_data);
257
258 if ($tmp) $feed_data = $tmp;
259 }
260
261 $feed_data = trim($feed_data);
262
263 $rss = new FeedParser($feed_data);
264 $rss->init();
265
266 if (!$rss->error()) {
267
268 $result = db_query("SELECT title, site_url FROM ttrss_feeds WHERE id = '$feed'");
269
270 $registered_title = db_fetch_result($result, 0, "title");
271 $orig_site_url = db_fetch_result($result, 0, "site_url");
272
273 $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
274 $feed_title = db_escape_string(mb_substr($rss->get_title(), 0, 199));
275
276 if ($feed_title && (!$registered_title || $registered_title == "[Unknown]")) {
277 db_query("UPDATE ttrss_feeds SET
278 title = '$feed_title' WHERE id = '$feed'");
279 }
280
281 if ($site_url && $orig_site_url != $site_url) {
282 db_query("UPDATE ttrss_feeds SET
283 site_url = '$site_url' WHERE id = '$feed'");
284 }
285 }
286 }
287
87764a50 288 // ignore_daemon is not used
88edaa93 289 function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
2c08214a 290
2c08214a
AD
291 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
292
4f71d743 293 _debug_suppress(!$debug_enabled);
68cccafc 294 _debug("start", $debug_enabled);
2c08214a 295
6bb96beb
AD
296 $result = db_query("SELECT title FROM ttrss_feeds
297 WHERE id = '$feed'");
298 $title = db_fetch_result($result, 0, "title");
299
300 // feed was batch-subscribed or something, we need to get basic info
301 // this is not optimal currently as it fetches stuff separately TODO: optimize
302 if ($title == "[Unknown]") {
303 _debug("setting basic feed info for $feed...");
304 set_basic_feed_info($feed);
305 }
306
a42c55f0 307 $result = db_query("SELECT id,update_interval,auth_login,
5ba1ddd4 308 feed_url,auth_pass,cache_images,
5321e775 309 mark_unread_on_update, owner_uid,
90e5f4f1 310 pubsub_state, auth_pass_encrypted,
df659891 311 feed_language,
90e5f4f1
AD
312 (SELECT max(date_entered) FROM
313 ttrss_entries, ttrss_user_entries where ref_id = id AND feed_id = '$feed') AS last_article_timestamp
87764a50 314 FROM ttrss_feeds WHERE id = '$feed'");
2c08214a
AD
315
316 if (db_num_rows($result) == 0) {
68cccafc 317 _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
2c08214a
AD
318 return false;
319 }
320
90e5f4f1 321 $last_article_timestamp = @strtotime(db_fetch_result($result, 0, "last_article_timestamp"));
5d0d3887
AD
322
323 if (defined('_DISABLE_HTTP_304'))
324 $last_article_timestamp = 0;
325
2c08214a
AD
326 $owner_uid = db_fetch_result($result, 0, "owner_uid");
327 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
328 0, "mark_unread_on_update"));
2c08214a 329 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
044cff2d
AD
330 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
331 0, "auth_pass_encrypted"));
2c08214a 332
a42c55f0 333 db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
2c08214a
AD
334 WHERE id = '$feed'");
335
336 $auth_login = db_fetch_result($result, 0, "auth_login");
337 $auth_pass = db_fetch_result($result, 0, "auth_pass");
338
044cff2d
AD
339 if ($auth_pass_encrypted) {
340 require_once "crypt.php";
341 $auth_pass = decrypt_string($auth_pass);
342 }
343
2c08214a
AD
344 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
345 $fetch_url = db_fetch_result($result, 0, "feed_url");
df659891
AD
346 $feed_language = db_escape_string(mb_strtolower(db_fetch_result($result, 0, "feed_language")));
347 if (!$feed_language) $feed_language = 'english';
2c08214a 348
a42c55f0 349 $feed = db_escape_string($feed);
2c08214a 350
f074ffe9 351 $date_feed_processed = date('Y-m-d H:i');
2c08214a 352
865a3ed6 353 $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
f074ffe9 354
ee65bef4
AD
355 $pluginhost = new PluginHost();
356 $pluginhost->set_debug($debug_enabled);
357 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
358
359 $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
360 $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
361 $pluginhost->load_data();
362
88edaa93
AD
363 if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
364 _debug("using previously initialized parser object");
365 } else {
366 $rss_hash = false;
4f9cbdff 367
88edaa93 368 $force_refetch = isset($_REQUEST["force_refetch"]);
687a4f59 369
95beaa14 370 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
01465325 371 $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, $last_article_timestamp, $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);
396 _debug("If-Modified-Since: ".gmdate('D, d M Y H:i:s \G\M\T', $last_article_timestamp), $debug_enabled);
96f98cb0 397
fd687300
AD
398 $feed_data = fetch_file_contents($fetch_url, false,
399 $auth_login, $auth_pass, false,
400 $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
401 $force_refetch ? 0 : $last_article_timestamp);
3f6f0857 402
fd687300 403 global $fetch_curl_used;
3f6f0857 404
fd687300
AD
405 if (!$fetch_curl_used) {
406 $tmp = @gzdecode($feed_data);
3f6f0857 407
fd687300
AD
408 if ($tmp) $feed_data = $tmp;
409 }
1367bc3f 410
fd687300 411 $feed_data = trim($feed_data);
017401dd 412
fd687300
AD
413 _debug("fetch done.", $debug_enabled);
414
95beaa14
RA
415 // cache vanilla feed data for re-use
416 if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
417 $new_rss_hash = sha1($feed_data);
418
419 if ($new_rss_hash != $rss_hash) {
420 _debug("saving $cache_filename", $debug_enabled);
421 @file_put_contents($cache_filename, $feed_data);
422 }
423 }
4f9cbdff 424 }
017401dd 425
fd687300
AD
426 if (!$feed_data) {
427 global $fetch_last_error;
428 global $fetch_last_error_code;
f074ffe9 429
fd687300 430 _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
f074ffe9 431
fd687300 432 $error_escaped = '';
7a01dc77 433
fd687300
AD
434 // If-Modified-Since
435 if ($fetch_last_error_code != 304) {
436 $error_escaped = db_escape_string($fetch_last_error);
437 } else {
438 _debug("source claims data not modified, nothing to do.", $debug_enabled);
439 }
4f9cbdff 440
fd687300
AD
441 db_query(
442 "UPDATE ttrss_feeds SET last_error = '$error_escaped',
443 last_updated = NOW() WHERE id = '$feed'");
4f9cbdff 444
fd687300
AD
445 return;
446 }
f074ffe9
AD
447 }
448
1ffe3391 449 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
6791af0c 450 $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
017401dd
AD
451 }
452
5ef84097
AD
453 // set last update to now so if anything *simplepie* crashes later we won't be
454 // continuously failing on the same feed
5de40104 455 //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
5ef84097 456
4f9cbdff 457 if (!$rss) {
cd07592c
AD
458 $rss = new FeedParser($feed_data);
459 $rss->init();
5de51df7
AD
460 }
461
2c08214a
AD
462// print_r($rss);
463
a42c55f0 464 $feed = db_escape_string($feed);
2c08214a 465
19b3992b 466 if (!$rss->error()) {
2c08214a 467
d2a421e3 468 // We use local pluginhost here because we need to load different per-user feed plugins
1ffe3391 469 $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
4412b877 470
df659891 471 _debug("language: $feed_language", $debug_enabled);
68cccafc 472 _debug("processing feed data...", $debug_enabled);
2c08214a 473
a42c55f0 474// db_query("BEGIN");
2c08214a 475
382268c6
AD
476 if (DB_TYPE == "pgsql") {
477 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
478 } else {
479 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
480 }
481
fe4535e6 482 $result = db_query("SELECT owner_uid,favicon_avg_color,
382268c6
AD
483 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
484 favicon_needs_check
2c08214a
AD
485 FROM ttrss_feeds WHERE id = '$feed'");
486
382268c6
AD
487 $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
488 "favicon_needs_check"));
36490f11 489 $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
2c08214a
AD
490
491 $owner_uid = db_fetch_result($result, 0, "owner_uid");
492
a42c55f0 493 $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
2c08214a 494
cd07592c
AD
495 _debug("site_url: $site_url", $debug_enabled);
496 _debug("feed_title: " . $rss->get_title(), $debug_enabled);
497
687a4f59 498 if ($favicon_needs_check || $force_refetch) {
36490f11
AD
499
500 /* terrible hack: if we crash on floicon shit here, we won't check
560cbd8c 501 * the icon avgcolor again (unless the icon got updated) */
36490f11 502
560cbd8c
AD
503 $favicon_file = ICONS_DIR . "/$feed.ico";
504 $favicon_modified = @filemtime($favicon_file);
505
68cccafc 506 _debug("checking favicon...", $debug_enabled);
687a4f59 507
6f7798b6 508 check_feed_favicon($site_url, $feed);
560cbd8c
AD
509 $favicon_modified_new = @filemtime($favicon_file);
510
511 if ($favicon_modified_new > $favicon_modified)
512 $favicon_avg_color = '';
687a4f59 513
6ee0d4b0 514 if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
bc7a144d 515 require_once "colors.php";
687a4f59 516
aafd55ba
AD
517 db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
518 id = '$feed'");
519
6322ac79 520 $favicon_color = db_escape_string(
bc7a144d 521 calculate_avg_color($favicon_file));
63c323f7 522
bc7a144d 523 $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
36490f11 524 } else if ($favicon_avg_color == 'fail') {
84ceb6bd 525 _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
6ac722d5 526 }
687a4f59 527
a42c55f0 528 db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
bc7a144d 529 $favicon_colorstring
f2798eb6
AD
530 WHERE id = '$feed'");
531 }
2c08214a 532
68cccafc 533 _debug("loading filters & labels...", $debug_enabled);
2c08214a 534
a42c55f0 535 $filters = load_filters($feed, $owner_uid);
2c08214a 536
68cccafc 537 _debug("" . count($filters) . " filters loaded.", $debug_enabled);
2c08214a 538
19b3992b 539 $items = $rss->get_items();
2c08214a 540
19b3992b 541 if (!is_array($items)) {
68cccafc 542 _debug("no articles found.", $debug_enabled);
2c08214a 543
a42c55f0 544 db_query("UPDATE ttrss_feeds
2c08214a
AD
545 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
546
547 return; // no articles
548 }
549
550 if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
551
68cccafc 552 _debug("checking for PUSH hub...", $debug_enabled);
2c08214a
AD
553
554 $feed_hub_url = false;
2c08214a 555
19b3992b 556 $links = $rss->get_links('hub');
2c08214a 557
19b3992b
AD
558 if ($links && is_array($links)) {
559 foreach ($links as $l) {
560 $feed_hub_url = $l;
561 break;
2c08214a
AD
562 }
563 }
564
68cccafc 565 _debug("feed hub url: $feed_hub_url", $debug_enabled);
2c08214a 566
d8179cb9
AD
567 $feed_self_url = $fetch_url;
568
569 $links = $rss->get_links('self');
570
571 if ($links && is_array($links)) {
572 foreach ($links as $l) {
573 $feed_self_url = $l;
574 break;
575 }
576 }
577
578 _debug("feed self url = $feed_self_url");
579
580 if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
2c08214a
AD
581 !ini_get("open_basedir")) {
582
583 require_once 'lib/pubsubhubbub/subscriber.php';
584
585 $callback_url = get_self_url_prefix() .
586 "/public.php?op=pubsub&id=$feed";
587
588 $s = new Subscriber($feed_hub_url, $callback_url);
589
d8179cb9 590 $rc = $s->subscribe($feed_self_url);
2c08214a 591
5ba1ddd4 592 _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
2c08214a 593
a42c55f0 594 db_query("UPDATE ttrss_feeds SET pubsub_state = 1
2c08214a
AD
595 WHERE id = '$feed'");
596 }
597 }
598
68cccafc 599 _debug("processing articles...", $debug_enabled);
2c08214a 600
6c9f3d4a
AD
601 $tstart = time();
602
19b3992b 603 foreach ($items as $item) {
5d56d100 604 if ($_REQUEST['xdebug'] == 3) {
2c08214a
AD
605 print_r($item);
606 }
607
6c9f3d4a
AD
608 if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
609 _debug("looks like there's too many articles to process at once, breaking out", $debug_enabled);
610 break;
611 }
612
19b3992b
AD
613 $entry_guid = $item->get_id();
614 if (!$entry_guid) $entry_guid = $item->get_link();
615 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
2c08214a
AD
616 if (!$entry_guid) continue;
617
3a4c8973
AD
618 $entry_guid = "$owner_uid,$entry_guid";
619
a42c55f0 620 $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
5e3d5480 621
68cccafc 622 _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
5e3d5480 623
2c08214a
AD
624 $entry_timestamp = "";
625
04d2f9c8
AD
626 $entry_timestamp = $item->get_date();
627
628 _debug("orig date: " . $item->get_date(), $debug_enabled);
2c08214a 629
30123fe6 630 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
2c08214a 631 $entry_timestamp = time();
2c08214a
AD
632 }
633
634 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
635
68cccafc 636 _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
2c08214a 637
99429e57
AD
638// $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
639// $entry_title = decode_numeric_entities($entry_title);
640 $entry_title = $item->get_title();
1b35d30c 641
5d56d100 642 $entry_link = rewrite_relative_url($site_url, $item->get_link());
2c08214a 643
68cccafc
AD
644 _debug("title $entry_title", $debug_enabled);
645 _debug("link $entry_link", $debug_enabled);
2c08214a
AD
646
647 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
648
19b3992b
AD
649 $entry_content = $item->get_content();
650 if (!$entry_content) $entry_content = $item->get_description();
2c08214a
AD
651
652 if ($_REQUEST["xdebug"] == 2) {
9ec10352 653 print "content: ";
487f0750 654 print $entry_content;
3c696512 655 print "\n";
2c08214a
AD
656 }
657
cd07592c
AD
658 $entry_comments = $item->get_comments_url();
659 $entry_author = $item->get_author();
2c08214a 660
a42c55f0 661 $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
2c08214a 662
ce2b219c
AD
663 $entry_comments = db_escape_string(mb_substr(trim($entry_comments), 0, 245));
664 $entry_author = db_escape_string(mb_substr(trim($entry_author), 0, 245));
2c08214a 665
d4992d6b 666 $num_comments = (int) $item->get_comments_count();
2c08214a 667
68cccafc
AD
668 _debug("author $entry_author", $debug_enabled);
669 _debug("num_comments: $num_comments", $debug_enabled);
ee78f81c 670 _debug("looking for tags...", $debug_enabled);
2c08214a
AD
671
672 // parse <category> entries into tags
673
674 $additional_tags = array();
675
19b3992b 676 $additional_tags_src = $item->get_categories();
2c08214a 677
19b3992b
AD
678 if (is_array($additional_tags_src)) {
679 foreach ($additional_tags_src as $tobj) {
cd07592c 680 array_push($additional_tags, $tobj);
2c08214a 681 }
19b3992b 682 }
2c08214a 683
fa6fbd36 684 $entry_tags = array_unique($additional_tags);
2c08214a
AD
685
686 for ($i = 0; $i < count($entry_tags); $i++)
687 $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
688
ee78f81c
AD
689 _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
690
68cccafc 691 _debug("done collecting data.", $debug_enabled);
2c08214a 692
3318d324 693 $result = db_query("SELECT id, content_hash, lang FROM ttrss_entries
b1840673 694 WHERE guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed'");
b30abdad
AD
695
696 if (db_num_rows($result) != 0) {
b1840673
AD
697 $base_entry_id = db_fetch_result($result, 0, "id");
698 $entry_stored_hash = db_fetch_result($result, 0, "content_hash");
a29fe121 699 $article_labels = get_article_labels($base_entry_id, $owner_uid);
3318d324 700 $entry_language = db_fetch_result($result, 0, "lang");
b30abdad 701 } else {
b1840673
AD
702 $base_entry_id = false;
703 $entry_stored_hash = "";
a29fe121 704 $article_labels = array();
3318d324 705 $entry_language = "";
b30abdad
AD
706 }
707
455b1401 708 $article = array("owner_uid" => $owner_uid, // read only
b30abdad 709 "guid" => $entry_guid, // read only
59e83455 710 "guid_hashed" => $entry_guid_hashed, // read only
19b3992b
AD
711 "title" => $entry_title,
712 "content" => $entry_content,
713 "link" => $entry_link,
a29fe121 714 "labels" => $article_labels, // current limitation: can add labels to article, can't remove them
19b3992b 715 "tags" => $entry_tags,
e02555c1 716 "author" => $entry_author,
c9299c28 717 "force_catchup" => false, // ugly hack for the time being
6de3a1be 718 "score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
3318d324 719 "language" => $entry_language,
f73e03e0
AD
720 "feed" => array("id" => $feed,
721 "fetch_url" => $fetch_url,
722 "site_url" => $site_url)
723 );
cc85704f 724
b1840673 725 $entry_plugin_data = "";
eb16bd9f 726 $entry_current_hash = calculate_article_hash($article, $pluginhost);
b1840673
AD
727
728 _debug("article hash: $entry_current_hash [stored=$entry_stored_hash]", $debug_enabled);
729
522e8b35 730 if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
b1840673
AD
731 _debug("stored article seems up to date [IID: $base_entry_id], updating timestamp only", $debug_enabled);
732
733 // we keep encountering the entry in feeds, so we need to
734 // update date_updated column so that we don't get horrible
735 // dupes when the entry gets purged and reinserted again e.g.
736 // in the case of SLOW SLOW OMG SLOW updating feeds
737
738 $base_entry_id = db_fetch_result($result, 0, "id");
739
740 db_query("UPDATE ttrss_entries SET date_updated = NOW()
741 WHERE id = '$base_entry_id'");
742
e44280a0 743 // if we allow duplicate posts, we have to continue to
da7e3e91
GP
744 // create the user entries for this feed
745 if (!get_pref("ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
746 continue;
747 }
b1840673
AD
748 }
749
750 _debug("hash differs, applying plugin filters:", $debug_enabled);
751
1ffe3391 752 foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
b1840673
AD
753 _debug("... " . get_class($plugin), $debug_enabled);
754
755 $start = microtime(true);
19b3992b 756 $article = $plugin->hook_article_filter($article);
0084f0d1 757
b1840673
AD
758 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
759
760 $entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
cc85704f
AD
761 }
762
b1840673
AD
763 $entry_plugin_data = db_escape_string($entry_plugin_data);
764
765 _debug("plugin data: $entry_plugin_data", $debug_enabled);
766
35c12dc4
AD
767 // Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
768 if (DB_TYPE == "mysql") {
769 foreach ($article as $k => $v) {
35c37354
AD
770
771 // i guess we'll have to take the risk of 4byte unicode labels & tags here
772 if (!is_array($article[$k])) {
773 $article[$k] = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $v);
774 }
35c12dc4
AD
775 }
776 }
777
b8774453
AD
778 /* Collect article tags here so we could filter by them: */
779
780 $article_filters = get_article_filters($filters, $article["title"],
781 $article["content"], $article["link"], 0, $article["author"],
782 $article["tags"]);
783
784 if ($debug_enabled) {
785 _debug("article filters: ", $debug_enabled);
786 if (count($article_filters) != 0) {
787 print_r($article_filters);
788 }
789 }
790
791 $plugin_filter_names = find_article_filters($article_filters, "plugin");
792 $plugin_filter_actions = $pluginhost->get_filter_actions();
793
794 if (count($plugin_filter_names) > 0) {
795 _debug("applying plugin filter actions...", $debug_enabled);
796
797 foreach ($plugin_filter_names as $pfn) {
798 list($pfclass,$pfaction) = explode(":", $pfn["param"]);
799
800 if (isset($plugin_filter_actions[$pfclass])) {
801 $plugin = $pluginhost->get_plugin($pfclass);
802
803 _debug("... $pfclass: $pfaction", $debug_enabled);
804
805 if ($plugin) {
806 $start = microtime(true);
807 $article = $plugin->hook_article_filter_action($article, $pfaction);
808
809 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
810 } else {
811 _debug("??? $pfclass: plugin object not found.");
812 }
813 } else {
814 _debug("??? $pfclass: filter plugin not registered.");
815 }
816 }
817 }
818
19b3992b 819 $entry_tags = $article["tags"];
a42c55f0
AD
820 $entry_guid = db_escape_string($entry_guid);
821 $entry_title = db_escape_string($article["title"]);
822 $entry_author = db_escape_string($article["author"]);
823 $entry_link = db_escape_string($article["link"]);
f935d98e 824 $entry_content = $article["content"]; // escaped below
c9299c28 825 $entry_force_catchup = $article["force_catchup"];
a29fe121 826 $article_labels = $article["labels"];
6de3a1be 827 $entry_score_modifier = (int) $article["score_modifier"];
3318d324 828 $entry_language = db_escape_string($article["language"]);
a29fe121
AD
829
830 if ($debug_enabled) {
831 _debug("article labels:", $debug_enabled);
832 print_r($article_labels);
833 }
c9299c28
AD
834
835 _debug("force catchup: $entry_force_catchup");
f935d98e 836
0a3fd79b 837 if ($cache_images && is_writable(CACHE_DIR . '/images'))
f0bd8e65 838 cache_images($entry_content, $site_url, $debug_enabled);
0a3fd79b 839
a42c55f0 840 $entry_content = db_escape_string($entry_content, false);
1f45c857 841
07877caf 842 //db_query("BEGIN");
2c08214a 843
a42c55f0 844 $result = db_query("SELECT id FROM ttrss_entries
5e3d5480 845 WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
9e222305 846
2c08214a
AD
847 if (db_num_rows($result) == 0) {
848
68cccafc 849 _debug("base guid [$entry_guid] not found", $debug_enabled);
2c08214a
AD
850
851 // base post entry does not exist, create it
852
6322ac79 853 $result = db_query(
2c08214a
AD
854 "INSERT INTO ttrss_entries
855 (title,
856 guid,
857 link,
858 updated,
859 content,
860 content_hash,
861 no_orig_date,
862 date_updated,
863 date_entered,
864 comments,
865 num_comments,
b30abdad 866 plugin_data,
6b461797 867 lang,
2c08214a
AD
868 author)
869 VALUES
870 ('$entry_title',
5e3d5480 871 '$entry_guid_hashed',
2c08214a
AD
872 '$entry_link',
873 '$entry_timestamp_fmt',
874 '$entry_content',
b1840673 875 '$entry_current_hash',
5ba1ddd4 876 false,
2c08214a 877 NOW(),
be574731 878 '$date_feed_processed',
2c08214a
AD
879 '$entry_comments',
880 '$num_comments',
b30abdad 881 '$entry_plugin_data',
6b461797 882 '$entry_language',
2c08214a 883 '$entry_author')");
e8291805 884
2c08214a 885 } else {
2c08214a 886 $base_entry_id = db_fetch_result($result, 0, "id");
2c08214a
AD
887 }
888
889 // now it should exist, if not - bad luck then
890
b1840673 891 $result = db_query("SELECT id FROM ttrss_entries
5e3d5480 892 WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
2c08214a
AD
893
894 $entry_ref_id = 0;
895 $entry_int_id = 0;
896
897 if (db_num_rows($result) == 1) {
898
68cccafc 899 _debug("base guid found, checking for user record", $debug_enabled);
2c08214a 900
2c08214a
AD
901 $ref_id = db_fetch_result($result, 0, "id");
902 $entry_ref_id = $ref_id;
903
5e3d5480
AD
904 /* $stored_guid = db_fetch_result($result, 0, "guid");
905 if ($stored_guid != $entry_guid_hashed) {
68cccafc 906 if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
5e3d5480 907
a42c55f0 908 db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
5e3d5480
AD
909 id = '$ref_id'");
910 } */
911
2c08214a
AD
912 // check for user post link to main table
913
914 // do we allow duplicate posts with same GUID in different feeds?
a42c55f0 915 if (get_pref("ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
2c08214a
AD
916 $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
917 } else {
918 $dupcheck_qpart = "";
919 }
920
2c08214a 921 if (find_article_filter($article_filters, "filter")) {
07877caf 922 //db_query("COMMIT"); // close transaction in progress
2c08214a
AD
923 continue;
924 }
925
6de3a1be 926 $score = calculate_article_score($article_filters) + $entry_score_modifier;
2c08214a 927
6de3a1be 928 _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
2c08214a
AD
929
930 $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
931 ref_id = '$ref_id' AND owner_uid = '$owner_uid'
932 $dupcheck_qpart";
933
934// if ($_REQUEST["xdebug"]) print "$query\n";
935
a42c55f0 936 $result = db_query($query);
2c08214a
AD
937
938 // okay it doesn't exist - create user entry
939 if (db_num_rows($result) == 0) {
940
68cccafc 941 _debug("user record not found, creating...", $debug_enabled);
2c08214a 942
c9299c28 943 if ($score >= -500 && !find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
2c08214a
AD
944 $unread = 'true';
945 $last_read_qpart = 'NULL';
946 } else {
947 $unread = 'false';
948 $last_read_qpart = 'NOW()';
949 }
950
951 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
952 $marked = 'true';
953 } else {
954 $marked = 'false';
955 }
956
957 if (find_article_filter($article_filters, 'publish')) {
958 $published = 'true';
959 } else {
960 $published = 'false';
961 }
962
2ea9bbfd
AD
963 // N-grams
964
c9299c28 965 /* if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
2ea9bbfd 966
a42c55f0 967 $result = db_query("SELECT COUNT(*) AS similar FROM
2ea9bbfd
AD
968 ttrss_entries,ttrss_user_entries
969 WHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'
970 AND similarity(title, '$entry_title') >= "._NGRAM_TITLE_DUPLICATE_THRESHOLD."
971 AND owner_uid = $owner_uid");
972
973 $ngram_similar = db_fetch_result($result, 0, "similar");
974
68cccafc 975 _debug("N-gram similar results: $ngram_similar", $debug_enabled);
2ea9bbfd
AD
976
977 if ($ngram_similar > 0) {
978 $unread = 'false';
979 }
c9299c28 980 } */
2ea9bbfd 981
7873d588
AD
982 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
983 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
984
6322ac79 985 $result = db_query(
2c08214a
AD
986 "INSERT INTO ttrss_user_entries
987 (ref_id, owner_uid, feed_id, unread, last_read, marked,
7873d588
AD
988 published, score, tag_cache, label_cache, uuid,
989 last_marked, last_published)
2c08214a 990 VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
7873d588
AD
991 $last_read_qpart, $marked, $published, '$score', '', '',
992 '', $last_marked, $last_published)");
2c08214a
AD
993
994 if (PUBSUBHUBBUB_HUB && $published == 'true') {
995 $rss_link = get_self_url_prefix() .
996 "/public.php?op=rss&id=-2&key=" .
a42c55f0 997 get_feed_access_key(-2, false, $owner_uid);
2c08214a
AD
998
999 $p = new Publisher(PUBSUBHUBBUB_HUB);
1000
5ba1ddd4 1001 /* $pubsub_result = */ $p->publish_update($rss_link);
2c08214a
AD
1002 }
1003
6322ac79 1004 $result = db_query(
2c08214a
AD
1005 "SELECT int_id FROM ttrss_user_entries WHERE
1006 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
1007 feed_id = '$feed' LIMIT 1");
1008
1009 if (db_num_rows($result) == 1) {
1010 $entry_int_id = db_fetch_result($result, 0, "int_id");
1011 }
1012 } else {
68cccafc 1013 _debug("user record FOUND", $debug_enabled);
2c08214a
AD
1014
1015 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
1016 $entry_int_id = db_fetch_result($result, 0, "int_id");
1017 }
1018
68cccafc 1019 _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
2c08214a 1020
e854442e 1021 if (DB_TYPE == "pgsql") {
978989bb 1022 $tsvector_combined = db_escape_string(mb_substr($entry_title . ' ' . strip_tags(str_replace('<', ' <', $entry_content)),
e854442e
AD
1023 0, 1000000));
1024
df659891 1025 $tsvector_qpart = "tsvector_combined = to_tsvector('$feed_language', '$tsvector_combined'),";
e854442e
AD
1026
1027 } else {
1028 $tsvector_qpart = "";
1029 }
1030
b1840673
AD
1031 db_query("UPDATE ttrss_entries
1032 SET title = '$entry_title',
1033 content = '$entry_content',
1034 content_hash = '$entry_current_hash',
1035 updated = '$entry_timestamp_fmt',
e854442e 1036 $tsvector_qpart
b1840673
AD
1037 num_comments = '$num_comments',
1038 plugin_data = '$entry_plugin_data',
1039 author = '$entry_author',
1040 lang = '$entry_language'
1041 WHERE id = '$ref_id'");
1042
59e83455
AD
1043 // update aux data
1044 db_query("UPDATE ttrss_user_entries
1045 SET score = '$score' WHERE ref_id = '$ref_id'");
1046
b1840673
AD
1047 if ($mark_unread_on_update) {
1048 db_query("UPDATE ttrss_user_entries
1049 SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
2c08214a
AD
1050 }
1051 }
1052
07877caf 1053 //db_query("COMMIT");
2c08214a 1054
a29fe121
AD
1055 _debug("assigning labels [other]...", $debug_enabled);
1056
1057 foreach ($article_labels as $label) {
1058 label_add_article($entry_ref_id, $label[1], $owner_uid);
1059 }
1060
1061 _debug("assigning labels [filters]...", $debug_enabled);
2c08214a 1062
a42c55f0 1063 assign_article_to_label_filters($entry_ref_id, $article_filters,
b24504b1 1064 $owner_uid, $article_labels);
2c08214a 1065
68cccafc 1066 _debug("looking for enclosures...", $debug_enabled);
2c08214a
AD
1067
1068 // enclosures
1069
1070 $enclosures = array();
1071
19b3992b 1072 $encs = $item->get_enclosures();
2c08214a 1073
19b3992b
AD
1074 if (is_array($encs)) {
1075 foreach ($encs as $e) {
1076 $e_item = array(
523bd90b 1077 $e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
2c08214a 1078 array_push($enclosures, $e_item);
2c08214a
AD
1079 }
1080 }
1081
2c08214a 1082 if ($debug_enabled) {
68cccafc 1083 _debug("article enclosures:", $debug_enabled);
2c08214a
AD
1084 print_r($enclosures);
1085 }
1086
dab229cd 1087 //db_query("BEGIN");
2c08214a 1088
5c54e683
AD
1089// debugging
1090// db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
1091
2c08214a 1092 foreach ($enclosures as $enc) {
a42c55f0
AD
1093 $enc_url = db_escape_string($enc[0]);
1094 $enc_type = db_escape_string($enc[1]);
1095 $enc_dur = db_escape_string($enc[2]);
5c54e683 1096 $enc_title = db_escape_string($enc[3]);
523bd90b
FE
1097 $enc_width = intval($enc[4]);
1098 $enc_height = intval($enc[5]);
2c08214a 1099
a42c55f0 1100 $result = db_query("SELECT id FROM ttrss_enclosures
2c08214a
AD
1101 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
1102
1103 if (db_num_rows($result) == 0) {
a42c55f0 1104 db_query("INSERT INTO ttrss_enclosures
523bd90b
FE
1105 (content_url, content_type, title, duration, post_id, width, height) VALUES
1106 ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
2c08214a
AD
1107 }
1108 }
1109
dab229cd 1110 //db_query("COMMIT");
2c08214a
AD
1111
1112 // check for manual tags (we have to do it here since they're loaded from filters)
1113
1114 foreach ($article_filters as $f) {
6aff7845 1115 if ($f["type"] == "tag") {
2c08214a 1116
6aff7845 1117 $manual_tags = trim_array(explode(",", $f["param"]));
2c08214a
AD
1118
1119 foreach ($manual_tags as $tag) {
1120 if (tag_is_valid($tag)) {
1121 array_push($entry_tags, $tag);
1122 }
1123 }
1124 }
1125 }
1126
1127 // Skip boring tags
1128
6322ac79 1129 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
2c08214a
AD
1130 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1131
1132 $filtered_tags = array();
1133 $tags_to_cache = array();
1134
1135 if ($entry_tags && is_array($entry_tags)) {
1136 foreach ($entry_tags as $tag) {
1137 if (array_search($tag, $boring_tags) === false) {
1138 array_push($filtered_tags, $tag);
1139 }
1140 }
1141 }
1142
1143 $filtered_tags = array_unique($filtered_tags);
1144
1145 if ($debug_enabled) {
68cccafc 1146 _debug("filtered article tags:", $debug_enabled);
2c08214a
AD
1147 print_r($filtered_tags);
1148 }
1149
1150 // Save article tags in the database
1151
1152 if (count($filtered_tags) > 0) {
1153
dab229cd 1154 //db_query("BEGIN");
2c08214a
AD
1155
1156 foreach ($filtered_tags as $tag) {
1157
1158 $tag = sanitize_tag($tag);
a42c55f0 1159 $tag = db_escape_string($tag);
2c08214a
AD
1160
1161 if (!tag_is_valid($tag)) continue;
1162
a42c55f0 1163 $result = db_query("SELECT id FROM ttrss_tags
2c08214a
AD
1164 WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1165 owner_uid = '$owner_uid' LIMIT 1");
1166
1167 if ($result && db_num_rows($result) == 0) {
1168
a42c55f0 1169 db_query("INSERT INTO ttrss_tags
2c08214a
AD
1170 (owner_uid,tag_name,post_int_id)
1171 VALUES ('$owner_uid','$tag', '$entry_int_id')");
1172 }
1173
1174 array_push($tags_to_cache, $tag);
1175 }
1176
1177 /* update the cache */
1178
1179 $tags_to_cache = array_unique($tags_to_cache);
1180
a42c55f0 1181 $tags_str = db_escape_string(join(",", $tags_to_cache));
2c08214a 1182
a42c55f0 1183 db_query("UPDATE ttrss_user_entries
2c08214a
AD
1184 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1185 AND owner_uid = $owner_uid");
1186
dab229cd 1187 //db_query("COMMIT");
2c08214a
AD
1188 }
1189
68cccafc 1190 _debug("article processed", $debug_enabled);
2c08214a
AD
1191 }
1192
68cccafc 1193 _debug("purging feed...", $debug_enabled);
2c08214a 1194
a42c55f0 1195 purge_feed($feed, 0, $debug_enabled);
2c08214a 1196
a42c55f0 1197 db_query("UPDATE ttrss_feeds
f2798eb6 1198 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
2c08214a 1199
a42c55f0 1200// db_query("COMMIT");
2c08214a
AD
1201
1202 } else {
1203
a42c55f0 1204 $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
2c08214a 1205
4ad04ee2
AD
1206 _debug("fetch error: $error_msg", $debug_enabled);
1207
1208 if (count($rss->errors()) > 1) {
1209 foreach ($rss->errors() as $error) {
1210 _debug("+ $error");
1211 }
1212 }
2c08214a 1213
6322ac79 1214 db_query(
2c08214a 1215 "UPDATE ttrss_feeds SET last_error = '$error_msg',
88edaa93 1216 last_updated = NOW() WHERE id = '$feed'");
2c08214a 1217
88edaa93
AD
1218 unset($rss);
1219 }
2c08214a 1220
68cccafc 1221 _debug("done", $debug_enabled);
88edaa93
AD
1222
1223 return $rss;
2c08214a
AD
1224 }
1225
3c696512 1226 function cache_images($html, $site_url, $debug) {
3c696512
AD
1227 libxml_use_internal_errors(true);
1228
1229 $charset_hack = '<head>
1230 <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1231 </head>';
1232
1233 $doc = new DOMDocument();
1234 $doc->loadHTML($charset_hack . $html);
1235 $xpath = new DOMXPath($doc);
1236
1237 $entries = $xpath->query('(//img[@src])');
1238
1239 foreach ($entries as $entry) {
1240 if ($entry->hasAttribute('src')) {
1241 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1242
1243 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1244
1245 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1246
1247 if (!file_exists($local_filename)) {
1248 $file_content = fetch_file_contents($src);
1249
a3d50184 1250 if ($file_content && strlen($file_content) > _MIN_CACHE_IMAGE_SIZE) {
3c696512
AD
1251 file_put_contents($local_filename, $file_content);
1252 }
4a27966e
J
1253 } else {
1254 touch($local_filename);
3c696512 1255 }
3c696512
AD
1256 }
1257 }
3c696512
AD
1258 }
1259
a42c55f0 1260 function expire_error_log($debug) {
e2261e17
AD
1261 if ($debug) _debug("Removing old error log entries...");
1262
1263 if (DB_TYPE == "pgsql") {
a42c55f0 1264 db_query("DELETE FROM ttrss_error_log
e2261e17
AD
1265 WHERE created_at < NOW() - INTERVAL '7 days'");
1266 } else {
a42c55f0 1267 db_query("DELETE FROM ttrss_error_log
e2261e17
AD
1268 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1269 }
1270
1271 }
1272
2a91b6ff 1273 function expire_lock_files($debug) {
65465085 1274 //if ($debug) _debug("Removing old lock files...");
2a91b6ff
AD
1275
1276 $num_deleted = 0;
1277
1278 if (is_writable(LOCK_DIRECTORY)) {
1279 $files = glob(LOCK_DIRECTORY . "/*.lock");
1280
1281 if ($files) {
1282 foreach ($files as $file) {
11344971 1283 if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
2a91b6ff
AD
1284 unlink($file);
1285 ++$num_deleted;
1286 }
1287 }
1288 }
1289 }
1290
65465085 1291 if ($debug) _debug("Removed $num_deleted old lock files.");
2a91b6ff
AD
1292 }
1293
3c696512 1294 function expire_cached_files($debug) {
3306daec 1295 foreach (array("simplepie", "images", "export", "upload") as $dir) {
3c696512 1296 $cache_dir = CACHE_DIR . "/$dir";
2c08214a 1297
65465085 1298// if ($debug) _debug("Expiring $cache_dir");
2c08214a 1299
3c696512
AD
1300 $num_deleted = 0;
1301
1302 if (is_writable($cache_dir)) {
1303 $files = glob("$cache_dir/*");
1304
2a91b6ff 1305 if ($files) {
2ab20c31
AD
1306 foreach ($files as $file) {
1307 if (time() - filemtime($file) > 86400*7) {
1308 unlink($file);
3c696512 1309
2ab20c31
AD
1310 ++$num_deleted;
1311 }
3c696512
AD
1312 }
1313 }
2a91b6ff 1314 }
3c696512 1315
65465085 1316 if ($debug) _debug("$cache_dir: removed $num_deleted files.");
3c696512
AD
1317 }
1318 }
2c08214a 1319
a3e0bdcf
AD
1320 /**
1321 * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1322 * Returns the url query as associative array
1323 *
1324 * @param string query
1325 * @return array params
1326 */
1327 function convertUrlQuery($query) {
1328 $queryParts = explode('&', $query);
1329
1330 $params = array();
1331
1332 foreach ($queryParts as $param) {
1333 $item = explode('=', $param);
1334 $params[$item[0]] = $item[1];
1335 }
1336
1337 return $params;
1338 }
92c14e9d 1339
48cb2536 1340 function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags) {
92c14e9d
AD
1341 $matches = array();
1342
1343 foreach ($filters as $filter) {
1344 $match_any_rule = $filter["match_any_rule"];
a3a896a1 1345 $inverse = $filter["inverse"];
92c14e9d
AD
1346 $filter_match = false;
1347
1348 foreach ($filter["rules"] as $rule) {
1349 $match = false;
ffa1bd7b 1350 $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
a3a896a1 1351 $rule_inverse = $rule["inverse"];
92c14e9d
AD
1352
1353 if (!$reg_exp)
1354 continue;
1355
1356 switch ($rule["type"]) {
1357 case "title":
1358 $match = @preg_match("/$reg_exp/i", $title);
1359 break;
1360 case "content":
d03ae73e
AD
1361 // we don't need to deal with multiline regexps
1362 $content = preg_replace("/[\r\n\t]/", "", $content);
1363
92c14e9d
AD
1364 $match = @preg_match("/$reg_exp/i", $content);
1365 break;
1366 case "both":
d03ae73e
AD
1367 // we don't need to deal with multiline regexps
1368 $content = preg_replace("/[\r\n\t]/", "", $content);
1369
72d1d067 1370 $match = (@preg_match("/$reg_exp/i", $title) || @preg_match("/$reg_exp/i", $content));
92c14e9d
AD
1371 break;
1372 case "link":
1373 $match = @preg_match("/$reg_exp/i", $link);
1374 break;
1375 case "author":
1376 $match = @preg_match("/$reg_exp/i", $author);
1377 break;
1378 case "tag":
7b80b5e1
DK
1379 foreach ($tags as $tag) {
1380 if (@preg_match("/$reg_exp/i", $tag)) {
1381 $match = true;
1382 break;
1383 }
1384 }
92c14e9d
AD
1385 break;
1386 }
1387
a3a896a1
AD
1388 if ($rule_inverse) $match = !$match;
1389
92c14e9d
AD
1390 if ($match_any_rule) {
1391 if ($match) {
1392 $filter_match = true;
1393 break;
1394 }
1395 } else {
1396 $filter_match = $match;
1397 if (!$match) {
1398 break;
1399 }
1400 }
1401 }
1402
a3a896a1
AD
1403 if ($inverse) $filter_match = !$filter_match;
1404
92c14e9d
AD
1405 if ($filter_match) {
1406 foreach ($filter["actions"] AS $action) {
1407 array_push($matches, $action);
5e736e45
AD
1408
1409 // if Stop action encountered, perform no further processing
fd3e5e8d 1410 if (isset($action["type"]) && $action["type"] == "stop") return $matches;
92c14e9d
AD
1411 }
1412 }
1413 }
1414
1415 return $matches;
1416 }
1417
1418 function find_article_filter($filters, $filter_name) {
1419 foreach ($filters as $f) {
1420 if ($f["type"] == $filter_name) {
1421 return $f;
1422 };
1423 }
1424 return false;
1425 }
1426
1427 function find_article_filters($filters, $filter_name) {
1428 $results = array();
1429
1430 foreach ($filters as $f) {
1431 if ($f["type"] == $filter_name) {
1432 array_push($results, $f);
1433 };
1434 }
1435 return $results;
1436 }
1437
1438 function calculate_article_score($filters) {
1439 $score = 0;
1440
1441 foreach ($filters as $f) {
1442 if ($f["type"] == "score") {
1443 $score += $f["param"];
1444 };
1445 }
1446 return $score;
1447 }
1448
b24504b1
AD
1449 function labels_contains_caption($labels, $caption) {
1450 foreach ($labels as $label) {
1451 if ($label[1] == $caption) {
1452 return true;
1453 }
1454 }
1455
1456 return false;
1457 }
1458
a42c55f0 1459 function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
92c14e9d
AD
1460 foreach ($filters as $f) {
1461 if ($f["type"] == "label") {
b24504b1 1462 if (!labels_contains_caption($article_labels, $f["param"])) {
a42c55f0 1463 label_add_article($id, $f["param"], $owner_uid);
b24504b1
AD
1464 }
1465 }
92c14e9d
AD
1466 }
1467 }
87764a50 1468
87d7e850
AD
1469 function make_guid_from_title($title) {
1470 return preg_replace("/[ \"\',.:;]/", "-",
1471 mb_strtolower(strip_tags($title), 'utf-8'));
1472 }
1473
d1f3fa97 1474 /* function verify_feed_xml($feed_data) {
ebec81a6
AD
1475 libxml_use_internal_errors(true);
1476 $doc = new DOMDocument();
1477 $doc->loadXML($feed_data);
1478 $error = libxml_get_last_error();
1479 libxml_clear_errors();
1480 return $error;
d1f3fa97 1481 } */
87d7e850 1482
168cf351
AD
1483 function cleanup_counters_cache($debug) {
1484 $result = db_query("DELETE FROM ttrss_counters_cache
1485 WHERE feed_id > 0 AND
1486 (SELECT COUNT(id) FROM ttrss_feeds WHERE
1487 id = feed_id AND
1488 ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
1489 $frows = db_affected_rows($result);
1490
1491 $result = db_query("DELETE FROM ttrss_cat_counters_cache
1492 WHERE feed_id > 0 AND
1493 (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
1494 id = feed_id AND
1495 ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
1496 $crows = db_affected_rows($result);
1497
1498 _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
1499 }
1500
e2cf81e2
AD
1501 function housekeeping_common($debug) {
1502 expire_cached_files($debug);
1503 expire_lock_files($debug);
1504 expire_error_log($debug);
1505
1506 $count = update_feedbrowser_cache();
1507 _debug("Feedbrowser updated, $count feeds processed.");
1508
1509 purge_orphans( true);
168cf351 1510 cleanup_counters_cache($debug);
e2cf81e2
AD
1511 $rc = cleanup_tags( 14, 50000);
1512
1513 _debug("Cleaned $rc cached tags.");
8e470220 1514
00f22824 1515 PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
8e470220 1516
e2cf81e2 1517 }
2c08214a 1518?>