]> git.wh0rd.org - tt-rss.git/blame - include/rssfuncs.php
image cache: send files as content-disposition: attachment; add .png suffix to image...
[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
5cbd1fe8
AD
182 $batch_owners = array();
183
2c08214a 184 // For each feed, we call the feed update function.
1c4421fc
AD
185 foreach ($feeds_to_update as $feed) {
186 if($debug) _debug("Base feed: $feed");
2c08214a 187
a42c55f0 188 //update_rss_feed($line["id"], true);
2c08214a 189
1c4421fc
AD
190 // since we have the data cached, we can deal with other feeds with the same url
191
15c762be 192 $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
ee0542ce
AD
193 FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
194 ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
195 ttrss_users.id = ttrss_user_prefs.owner_uid AND
196 ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
f08426e3 197 ttrss_user_prefs.profile IS NULL AND
9e84bab4
AD
198 feed_url = '".db_escape_string($feed)."'
199 $update_limit_qpart
1c4421fc 200 $login_thresh_qpart
5929a0c1 201 ORDER BY ttrss_feeds.id $query_limit");
1c4421fc
AD
202
203 if (db_num_rows($tmp_result) > 0) {
88edaa93
AD
204 $rss = false;
205
1c4421fc 206 while ($tline = db_fetch_assoc($tmp_result)) {
15c762be 207 if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
f08426e3 208
5cbd1fe8
AD
209 if (array_search($tline["owner_uid"], $batch_owners) === FALSE)
210 array_push($batch_owners, $tline["owner_uid"]);
211
2d9c5684 212 $fstarted = microtime(true);
f77c7203 213 $rss = update_rss_feed($tline["id"], true, false);
4f71d743 214 _debug_suppress(false);
2d9c5684
AD
215
216 _debug(sprintf(" %.4f (sec)", microtime(true) - $fstarted));
217
8292d05b 218 ++$nf;
1c4421fc
AD
219 }
220 }
2c08214a
AD
221 }
222
2d9c5684
AD
223 if ($nf > 0) {
224 _debug(sprintf("Processed %d feeds in %.4f (sec), %.4f (sec/feed avg)", $nf,
225 microtime(true) - $bstarted, (microtime(true) - $bstarted) / $nf));
226 }
227
5cbd1fe8
AD
228 foreach ($batch_owners as $owner_uid) {
229 _debug("Running housekeeping tasks for user $owner_uid...");
230
231 housekeeping_user($owner_uid);
232 }
233
f1611683
AD
234 require_once "digest.php";
235
2c08214a 236 // Send feed digests by email if needed.
a42c55f0 237 send_headlines_digests($debug);
2c08214a 238
8292d05b
AD
239 return $nf;
240
2c08214a
AD
241 } // function update_daemon_common
242
6022776d
AD
243 // this is used when subscribing
244 function set_basic_feed_info($feed) {
245
246 $feed = db_escape_string($feed);
247
9bd422c2 248 $result = db_query("SELECT feed_url,auth_pass,auth_login,auth_pass_encrypted
6022776d
AD
249 FROM ttrss_feeds WHERE id = '$feed'");
250
251 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
252 0, "auth_pass_encrypted"));
253
254 $auth_login = db_fetch_result($result, 0, "auth_login");
255 $auth_pass = db_fetch_result($result, 0, "auth_pass");
256
17a8e61d 257 if ($auth_pass_encrypted && function_exists("mcrypt_decrypt")) {
6022776d
AD
258 require_once "crypt.php";
259 $auth_pass = decrypt_string($auth_pass);
260 }
261
262 $fetch_url = db_fetch_result($result, 0, "feed_url");
263
264 $feed_data = fetch_file_contents($fetch_url, false,
265 $auth_login, $auth_pass, false,
3d5d2890 266 FEED_FETCH_TIMEOUT,
6022776d
AD
267 0);
268
269 global $fetch_curl_used;
270
271 if (!$fetch_curl_used) {
272 $tmp = @gzdecode($feed_data);
273
274 if ($tmp) $feed_data = $tmp;
275 }
276
277 $feed_data = trim($feed_data);
278
279 $rss = new FeedParser($feed_data);
280 $rss->init();
281
282 if (!$rss->error()) {
283
284 $result = db_query("SELECT title, site_url FROM ttrss_feeds WHERE id = '$feed'");
285
286 $registered_title = db_fetch_result($result, 0, "title");
287 $orig_site_url = db_fetch_result($result, 0, "site_url");
288
289 $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
290 $feed_title = db_escape_string(mb_substr($rss->get_title(), 0, 199));
291
292 if ($feed_title && (!$registered_title || $registered_title == "[Unknown]")) {
293 db_query("UPDATE ttrss_feeds SET
294 title = '$feed_title' WHERE id = '$feed'");
295 }
296
297 if ($site_url && $orig_site_url != $site_url) {
298 db_query("UPDATE ttrss_feeds SET
299 site_url = '$site_url' WHERE id = '$feed'");
300 }
301 }
302 }
303
87764a50 304 // ignore_daemon is not used
88edaa93 305 function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
2c08214a 306
2c08214a
AD
307 $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
308
4f71d743 309 _debug_suppress(!$debug_enabled);
68cccafc 310 _debug("start", $debug_enabled);
2c08214a 311
6bb96beb
AD
312 $result = db_query("SELECT title FROM ttrss_feeds
313 WHERE id = '$feed'");
bfe1eb4e
AD
314
315 if (db_num_rows($result) == 0) {
316 _debug("feed $feed NOT FOUND/SKIPPED", $debug_enabled);
317 user_error("Attempt to update unknown/invalid feed $feed", E_USER_WARNING);
318 return false;
319 }
320
6bb96beb
AD
321 $title = db_fetch_result($result, 0, "title");
322
323 // feed was batch-subscribed or something, we need to get basic info
324 // this is not optimal currently as it fetches stuff separately TODO: optimize
325 if ($title == "[Unknown]") {
326 _debug("setting basic feed info for $feed...");
327 set_basic_feed_info($feed);
328 }
329
a42c55f0 330 $result = db_query("SELECT id,update_interval,auth_login,
5ba1ddd4 331 feed_url,auth_pass,cache_images,
5321e775 332 mark_unread_on_update, owner_uid,
90e5f4f1 333 pubsub_state, auth_pass_encrypted,
0edf1d0d 334 feed_language
87764a50 335 FROM ttrss_feeds WHERE id = '$feed'");
2c08214a 336
2c08214a
AD
337 $owner_uid = db_fetch_result($result, 0, "owner_uid");
338 $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
339 0, "mark_unread_on_update"));
2c08214a 340 $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
044cff2d
AD
341 $auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result,
342 0, "auth_pass_encrypted"));
2c08214a 343
a42c55f0 344 db_query("UPDATE ttrss_feeds SET last_update_started = NOW()
2c08214a
AD
345 WHERE id = '$feed'");
346
347 $auth_login = db_fetch_result($result, 0, "auth_login");
348 $auth_pass = db_fetch_result($result, 0, "auth_pass");
349
17a8e61d 350 if ($auth_pass_encrypted && function_exists("mcrypt_decrypt")) {
044cff2d
AD
351 require_once "crypt.php";
352 $auth_pass = decrypt_string($auth_pass);
353 }
354
2c08214a
AD
355 $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
356 $fetch_url = db_fetch_result($result, 0, "feed_url");
df659891
AD
357 $feed_language = db_escape_string(mb_strtolower(db_fetch_result($result, 0, "feed_language")));
358 if (!$feed_language) $feed_language = 'english';
2c08214a 359
a42c55f0 360 $feed = db_escape_string($feed);
2c08214a 361
f074ffe9 362 $date_feed_processed = date('Y-m-d H:i');
2c08214a 363
865a3ed6 364 $cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
f074ffe9 365
ee65bef4
AD
366 $pluginhost = new PluginHost();
367 $pluginhost->set_debug($debug_enabled);
368 $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
369
370 $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
371 $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
372 $pluginhost->load_data();
373
88edaa93
AD
374 if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
375 _debug("using previously initialized parser object");
376 } else {
377 $rss_hash = false;
4f9cbdff 378
88edaa93 379 $force_refetch = isset($_REQUEST["force_refetch"]);
0edf1d0d 380 $feed_data = "";
687a4f59 381
95beaa14 382 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
0edf1d0d 383 $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, 0, $auth_login, $auth_pass);
95beaa14
RA
384 }
385
386 // try cache
387 if (!$feed_data &&
388 file_exists($cache_filename) &&
88edaa93
AD
389 is_readable($cache_filename) &&
390 !$auth_login && !$auth_pass &&
391 filemtime($cache_filename) > time() - 30) {
2c08214a 392
15d0528a 393 _debug("using local cache [$cache_filename].", $debug_enabled);
be574731 394
88edaa93 395 @$feed_data = file_get_contents($cache_filename);
52637d3b 396
88edaa93
AD
397 if ($feed_data) {
398 $rss_hash = sha1($feed_data);
399 }
f074ffe9 400
88edaa93
AD
401 } else {
402 _debug("local cache will not be used for this feed", $debug_enabled);
403 }
ee65bef4 404
95beaa14 405 // fetch feed from source
fd687300
AD
406 if (!$feed_data) {
407 _debug("fetching [$fetch_url]...", $debug_enabled);
312742db
AD
408
409 if (ini_get("open_basedir") && function_exists("curl_init")) {
410 _debug("not using CURL due to open_basedir restrictions");
411 }
412
fd687300
AD
413 $feed_data = fetch_file_contents($fetch_url, false,
414 $auth_login, $auth_pass, false,
415 $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT,
0edf1d0d 416 0);
3f6f0857 417
fd687300 418 global $fetch_curl_used;
3f6f0857 419
fd687300
AD
420 if (!$fetch_curl_used) {
421 $tmp = @gzdecode($feed_data);
3f6f0857 422
fd687300
AD
423 if ($tmp) $feed_data = $tmp;
424 }
1367bc3f 425
fd687300 426 $feed_data = trim($feed_data);
017401dd 427
fd687300
AD
428 _debug("fetch done.", $debug_enabled);
429
95beaa14
RA
430 // cache vanilla feed data for re-use
431 if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
432 $new_rss_hash = sha1($feed_data);
433
434 if ($new_rss_hash != $rss_hash) {
435 _debug("saving $cache_filename", $debug_enabled);
436 @file_put_contents($cache_filename, $feed_data);
437 }
438 }
4f9cbdff 439 }
017401dd 440
fd687300
AD
441 if (!$feed_data) {
442 global $fetch_last_error;
443 global $fetch_last_error_code;
f074ffe9 444
fd687300 445 _debug("unable to fetch: $fetch_last_error [$fetch_last_error_code]", $debug_enabled);
f074ffe9 446
fd687300 447 $error_escaped = '';
7a01dc77 448
fd687300
AD
449 // If-Modified-Since
450 if ($fetch_last_error_code != 304) {
451 $error_escaped = db_escape_string($fetch_last_error);
452 } else {
453 _debug("source claims data not modified, nothing to do.", $debug_enabled);
454 }
4f9cbdff 455
fd687300
AD
456 db_query(
457 "UPDATE ttrss_feeds SET last_error = '$error_escaped',
458 last_updated = NOW() WHERE id = '$feed'");
4f9cbdff 459
fd687300
AD
460 return;
461 }
f074ffe9
AD
462 }
463
1ffe3391 464 foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
6791af0c 465 $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
017401dd
AD
466 }
467
5ef84097
AD
468 // set last update to now so if anything *simplepie* crashes later we won't be
469 // continuously failing on the same feed
5de40104 470 //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
5ef84097 471
4f9cbdff 472 if (!$rss) {
cd07592c
AD
473 $rss = new FeedParser($feed_data);
474 $rss->init();
5de51df7
AD
475 }
476
2c08214a
AD
477// print_r($rss);
478
a42c55f0 479 $feed = db_escape_string($feed);
2c08214a 480
19b3992b 481 if (!$rss->error()) {
2c08214a 482
d2a421e3 483 // We use local pluginhost here because we need to load different per-user feed plugins
1ffe3391 484 $pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
4412b877 485
df659891 486 _debug("language: $feed_language", $debug_enabled);
68cccafc 487 _debug("processing feed data...", $debug_enabled);
2c08214a 488
a42c55f0 489// db_query("BEGIN");
2c08214a 490
382268c6
AD
491 if (DB_TYPE == "pgsql") {
492 $favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
493 } else {
494 $favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
495 }
496
fe4535e6 497 $result = db_query("SELECT owner_uid,favicon_avg_color,
382268c6
AD
498 (favicon_last_checked IS NULL OR $favicon_interval_qpart) AS
499 favicon_needs_check
2c08214a
AD
500 FROM ttrss_feeds WHERE id = '$feed'");
501
382268c6
AD
502 $favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0,
503 "favicon_needs_check"));
36490f11 504 $favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
2c08214a
AD
505
506 $owner_uid = db_fetch_result($result, 0, "owner_uid");
507
a42c55f0 508 $site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
2c08214a 509
cd07592c
AD
510 _debug("site_url: $site_url", $debug_enabled);
511 _debug("feed_title: " . $rss->get_title(), $debug_enabled);
512
687a4f59 513 if ($favicon_needs_check || $force_refetch) {
36490f11
AD
514
515 /* terrible hack: if we crash on floicon shit here, we won't check
560cbd8c 516 * the icon avgcolor again (unless the icon got updated) */
36490f11 517
560cbd8c
AD
518 $favicon_file = ICONS_DIR . "/$feed.ico";
519 $favicon_modified = @filemtime($favicon_file);
520
68cccafc 521 _debug("checking favicon...", $debug_enabled);
687a4f59 522
6f7798b6 523 check_feed_favicon($site_url, $feed);
560cbd8c
AD
524 $favicon_modified_new = @filemtime($favicon_file);
525
526 if ($favicon_modified_new > $favicon_modified)
527 $favicon_avg_color = '';
687a4f59 528
6ee0d4b0 529 if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
bc7a144d 530 require_once "colors.php";
687a4f59 531
aafd55ba
AD
532 db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE
533 id = '$feed'");
534
6322ac79 535 $favicon_color = db_escape_string(
bc7a144d 536 calculate_avg_color($favicon_file));
63c323f7 537
bc7a144d 538 $favicon_colorstring = ",favicon_avg_color = '".$favicon_color."'";
36490f11 539 } else if ($favicon_avg_color == 'fail') {
84ceb6bd 540 _debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
6ac722d5 541 }
687a4f59 542
a42c55f0 543 db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()
bc7a144d 544 $favicon_colorstring
f2798eb6
AD
545 WHERE id = '$feed'");
546 }
2c08214a 547
68cccafc 548 _debug("loading filters & labels...", $debug_enabled);
2c08214a 549
a42c55f0 550 $filters = load_filters($feed, $owner_uid);
2c08214a 551
68cccafc 552 _debug("" . count($filters) . " filters loaded.", $debug_enabled);
2c08214a 553
19b3992b 554 $items = $rss->get_items();
2c08214a 555
19b3992b 556 if (!is_array($items)) {
68cccafc 557 _debug("no articles found.", $debug_enabled);
2c08214a 558
a42c55f0 559 db_query("UPDATE ttrss_feeds
2c08214a
AD
560 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
561
562 return; // no articles
563 }
564
565 if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
566
68cccafc 567 _debug("checking for PUSH hub...", $debug_enabled);
2c08214a
AD
568
569 $feed_hub_url = false;
2c08214a 570
19b3992b 571 $links = $rss->get_links('hub');
2c08214a 572
19b3992b
AD
573 if ($links && is_array($links)) {
574 foreach ($links as $l) {
575 $feed_hub_url = $l;
576 break;
2c08214a
AD
577 }
578 }
579
68cccafc 580 _debug("feed hub url: $feed_hub_url", $debug_enabled);
2c08214a 581
d8179cb9
AD
582 $feed_self_url = $fetch_url;
583
584 $links = $rss->get_links('self');
585
586 if ($links && is_array($links)) {
587 foreach ($links as $l) {
588 $feed_self_url = $l;
589 break;
590 }
591 }
592
593 _debug("feed self url = $feed_self_url");
594
595 if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
2c08214a
AD
596 !ini_get("open_basedir")) {
597
becd215a 598 require_once 'lib/pubsubhubbub/Subscriber.php';
2c08214a
AD
599
600 $callback_url = get_self_url_prefix() .
601 "/public.php?op=pubsub&id=$feed";
602
cb3f8773 603 $s = new Pubsubhubbub\Subscriber\Subscriber($feed_hub_url, $callback_url);
2c08214a 604
d8179cb9 605 $rc = $s->subscribe($feed_self_url);
2c08214a 606
5ba1ddd4 607 _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
2c08214a 608
a42c55f0 609 db_query("UPDATE ttrss_feeds SET pubsub_state = 1
2c08214a
AD
610 WHERE id = '$feed'");
611 }
612 }
613
68cccafc 614 _debug("processing articles...", $debug_enabled);
2c08214a 615
6c9f3d4a
AD
616 $tstart = time();
617
19b3992b 618 foreach ($items as $item) {
5d56d100 619 if ($_REQUEST['xdebug'] == 3) {
2c08214a
AD
620 print_r($item);
621 }
622
6c9f3d4a
AD
623 if (ini_get("max_execution_time") > 0 && time() - $tstart >= ini_get("max_execution_time") * 0.7) {
624 _debug("looks like there's too many articles to process at once, breaking out", $debug_enabled);
625 break;
626 }
627
19b3992b
AD
628 $entry_guid = $item->get_id();
629 if (!$entry_guid) $entry_guid = $item->get_link();
630 if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
2c08214a
AD
631 if (!$entry_guid) continue;
632
3a4c8973
AD
633 $entry_guid = "$owner_uid,$entry_guid";
634
a42c55f0 635 $entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
5e3d5480 636
68cccafc 637 _debug("guid $entry_guid / $entry_guid_hashed", $debug_enabled);
5e3d5480 638
2c08214a
AD
639 $entry_timestamp = "";
640
04d2f9c8
AD
641 $entry_timestamp = $item->get_date();
642
643 _debug("orig date: " . $item->get_date(), $debug_enabled);
2c08214a 644
30123fe6 645 if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
2c08214a 646 $entry_timestamp = time();
2c08214a
AD
647 }
648
649 $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
650
68cccafc 651 _debug("date $entry_timestamp [$entry_timestamp_fmt]", $debug_enabled);
2c08214a 652
99429e57
AD
653// $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
654// $entry_title = decode_numeric_entities($entry_title);
655 $entry_title = $item->get_title();
1b35d30c 656
5d56d100 657 $entry_link = rewrite_relative_url($site_url, $item->get_link());
2c08214a 658
68cccafc
AD
659 _debug("title $entry_title", $debug_enabled);
660 _debug("link $entry_link", $debug_enabled);
2c08214a
AD
661
662 if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
663
19b3992b
AD
664 $entry_content = $item->get_content();
665 if (!$entry_content) $entry_content = $item->get_description();
2c08214a
AD
666
667 if ($_REQUEST["xdebug"] == 2) {
9ec10352 668 print "content: ";
0bc503ff 669 print htmlspecialchars($entry_content);
3c696512 670 print "\n";
2c08214a
AD
671 }
672
12ff230b
AD
673 $entry_comments = db_escape_string(mb_substr($item->get_comments_url(), 0, 245));
674 $num_comments = (int) $item->get_comments_count();
2c08214a 675
12ff230b 676 $entry_author = $item->get_author(); // escaped later
a42c55f0 677 $entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
2c08214a 678
68cccafc
AD
679 _debug("author $entry_author", $debug_enabled);
680 _debug("num_comments: $num_comments", $debug_enabled);
ee78f81c 681 _debug("looking for tags...", $debug_enabled);
2c08214a
AD
682
683 // parse <category> entries into tags
684
685 $additional_tags = array();
686
19b3992b 687 $additional_tags_src = $item->get_categories();
2c08214a 688
19b3992b
AD
689 if (is_array($additional_tags_src)) {
690 foreach ($additional_tags_src as $tobj) {
cd07592c 691 array_push($additional_tags, $tobj);
2c08214a 692 }
19b3992b 693 }
2c08214a 694
fa6fbd36 695 $entry_tags = array_unique($additional_tags);
2c08214a
AD
696
697 for ($i = 0; $i < count($entry_tags); $i++)
698 $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
699
ee78f81c
AD
700 _debug("tags found: " . join(",", $entry_tags), $debug_enabled);
701
68cccafc 702 _debug("done collecting data.", $debug_enabled);
2c08214a 703
3318d324 704 $result = db_query("SELECT id, content_hash, lang FROM ttrss_entries
b1840673 705 WHERE guid = '".db_escape_string($entry_guid)."' OR guid = '$entry_guid_hashed'");
b30abdad
AD
706
707 if (db_num_rows($result) != 0) {
b1840673
AD
708 $base_entry_id = db_fetch_result($result, 0, "id");
709 $entry_stored_hash = db_fetch_result($result, 0, "content_hash");
a29fe121 710 $article_labels = get_article_labels($base_entry_id, $owner_uid);
3318d324 711 $entry_language = db_fetch_result($result, 0, "lang");
a8ac7661 712
040aa540 713 $existing_tags = get_article_tags($base_entry_id, $owner_uid);
a8ac7661
AD
714 $entry_tags = array_unique(array_merge($entry_tags, $existing_tags));
715
b30abdad 716 } else {
b1840673
AD
717 $base_entry_id = false;
718 $entry_stored_hash = "";
a29fe121 719 $article_labels = array();
3318d324 720 $entry_language = "";
b30abdad
AD
721 }
722
455b1401 723 $article = array("owner_uid" => $owner_uid, // read only
b30abdad 724 "guid" => $entry_guid, // read only
59e83455 725 "guid_hashed" => $entry_guid_hashed, // read only
19b3992b
AD
726 "title" => $entry_title,
727 "content" => $entry_content,
728 "link" => $entry_link,
a29fe121 729 "labels" => $article_labels, // current limitation: can add labels to article, can't remove them
19b3992b 730 "tags" => $entry_tags,
e02555c1 731 "author" => $entry_author,
c9299c28 732 "force_catchup" => false, // ugly hack for the time being
6de3a1be 733 "score_modifier" => 0, // no previous value, plugin should recalculate score modifier based on content if needed
3318d324 734 "language" => $entry_language,
f73e03e0
AD
735 "feed" => array("id" => $feed,
736 "fetch_url" => $fetch_url,
babfadbf
J
737 "site_url" => $site_url,
738 "cache_images" => $cache_images)
f73e03e0 739 );
cc85704f 740
b1840673 741 $entry_plugin_data = "";
eb16bd9f 742 $entry_current_hash = calculate_article_hash($article, $pluginhost);
b1840673
AD
743
744 _debug("article hash: $entry_current_hash [stored=$entry_stored_hash]", $debug_enabled);
745
522e8b35 746 if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
b1840673
AD
747 _debug("stored article seems up to date [IID: $base_entry_id], updating timestamp only", $debug_enabled);
748
749 // we keep encountering the entry in feeds, so we need to
750 // update date_updated column so that we don't get horrible
751 // dupes when the entry gets purged and reinserted again e.g.
752 // in the case of SLOW SLOW OMG SLOW updating feeds
753
754 $base_entry_id = db_fetch_result($result, 0, "id");
755
756 db_query("UPDATE ttrss_entries SET date_updated = NOW()
757 WHERE id = '$base_entry_id'");
758
5bdcb8fd 759 continue;
b1840673
AD
760 }
761
762 _debug("hash differs, applying plugin filters:", $debug_enabled);
763
1ffe3391 764 foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
b1840673
AD
765 _debug("... " . get_class($plugin), $debug_enabled);
766
767 $start = microtime(true);
19b3992b 768 $article = $plugin->hook_article_filter($article);
0084f0d1 769
b1840673
AD
770 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
771
772 $entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
cc85704f
AD
773 }
774
0bc503ff
AD
775 if ($_REQUEST["xdebug"] == 2) {
776 print "processed content: ";
777 print htmlspecialchars($article["content"]);
778 print "\n";
779 }
780
b1840673
AD
781 $entry_plugin_data = db_escape_string($entry_plugin_data);
782
783 _debug("plugin data: $entry_plugin_data", $debug_enabled);
784
35c12dc4
AD
785 // Workaround: 4-byte unicode requires utf8mb4 in MySQL. See https://tt-rss.org/forum/viewtopic.php?f=1&t=3377&p=20077#p20077
786 if (DB_TYPE == "mysql") {
787 foreach ($article as $k => $v) {
35c37354
AD
788
789 // i guess we'll have to take the risk of 4byte unicode labels & tags here
dae16f72 790 if (is_string($article[$k])) {
35c37354
AD
791 $article[$k] = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $v);
792 }
35c12dc4
AD
793 }
794 }
795
b8774453
AD
796 /* Collect article tags here so we could filter by them: */
797
557d86fe
AD
798 $matched_rules = array();
799
b8774453
AD
800 $article_filters = get_article_filters($filters, $article["title"],
801 $article["content"], $article["link"], 0, $article["author"],
557d86fe 802 $article["tags"], $matched_rules);
b8774453
AD
803
804 if ($debug_enabled) {
557d86fe
AD
805 _debug("matched filter rules: ", $debug_enabled);
806
807 if (count($matched_rules) != 0) {
808 print_r($matched_rules);
809 }
810
811 _debug("filter actions: ", $debug_enabled);
812
b8774453
AD
813 if (count($article_filters) != 0) {
814 print_r($article_filters);
815 }
816 }
817
818 $plugin_filter_names = find_article_filters($article_filters, "plugin");
819 $plugin_filter_actions = $pluginhost->get_filter_actions();
820
821 if (count($plugin_filter_names) > 0) {
822 _debug("applying plugin filter actions...", $debug_enabled);
823
824 foreach ($plugin_filter_names as $pfn) {
825 list($pfclass,$pfaction) = explode(":", $pfn["param"]);
826
827 if (isset($plugin_filter_actions[$pfclass])) {
828 $plugin = $pluginhost->get_plugin($pfclass);
829
830 _debug("... $pfclass: $pfaction", $debug_enabled);
831
832 if ($plugin) {
833 $start = microtime(true);
834 $article = $plugin->hook_article_filter_action($article, $pfaction);
835
836 _debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
837 } else {
838 _debug("??? $pfclass: plugin object not found.");
839 }
840 } else {
841 _debug("??? $pfclass: filter plugin not registered.");
842 }
843 }
844 }
845
19b3992b 846 $entry_tags = $article["tags"];
a42c55f0
AD
847 $entry_guid = db_escape_string($entry_guid);
848 $entry_title = db_escape_string($article["title"]);
12ff230b 849 $entry_author = db_escape_string(mb_substr($article["author"], 0, 245));
a42c55f0 850 $entry_link = db_escape_string($article["link"]);
f935d98e 851 $entry_content = $article["content"]; // escaped below
c9299c28 852 $entry_force_catchup = $article["force_catchup"];
a29fe121 853 $article_labels = $article["labels"];
6de3a1be 854 $entry_score_modifier = (int) $article["score_modifier"];
3318d324 855 $entry_language = db_escape_string($article["language"]);
a29fe121
AD
856
857 if ($debug_enabled) {
858 _debug("article labels:", $debug_enabled);
557d86fe
AD
859
860 if (count($article_labels) != 0) {
861 print_r($article_labels);
862 }
a29fe121 863 }
c9299c28
AD
864
865 _debug("force catchup: $entry_force_catchup");
f935d98e 866
0a3fd79b 867 if ($cache_images && is_writable(CACHE_DIR . '/images'))
f0bd8e65 868 cache_images($entry_content, $site_url, $debug_enabled);
0a3fd79b 869
a42c55f0 870 $entry_content = db_escape_string($entry_content, false);
1f45c857 871
07877caf 872 //db_query("BEGIN");
2c08214a 873
a42c55f0 874 $result = db_query("SELECT id FROM ttrss_entries
5e3d5480 875 WHERE (guid = '$entry_guid' OR guid = '$entry_guid_hashed')");
9e222305 876
2c08214a
AD
877 if (db_num_rows($result) == 0) {
878
68cccafc 879 _debug("base guid [$entry_guid] not found", $debug_enabled);
2c08214a
AD
880
881 // base post entry does not exist, create it
882
6322ac79 883 $result = db_query(
2c08214a
AD
884 "INSERT INTO ttrss_entries
885 (title,
886 guid,
887 link,
888 updated,
889 content,
890 content_hash,
891 no_orig_date,
892 date_updated,
893 date_entered,
894 comments,
895 num_comments,
b30abdad 896 plugin_data,
6b461797 897 lang,
2c08214a
AD
898 author)
899 VALUES
900 ('$entry_title',
5e3d5480 901 '$entry_guid_hashed',
2c08214a
AD
902 '$entry_link',
903 '$entry_timestamp_fmt',
904 '$entry_content',
b1840673 905 '$entry_current_hash',
5ba1ddd4 906 false,
2c08214a 907 NOW(),
be574731 908 '$date_feed_processed',
2c08214a
AD
909 '$entry_comments',
910 '$num_comments',
b30abdad 911 '$entry_plugin_data',
6b461797 912 '$entry_language',
2c08214a 913 '$entry_author')");
e8291805 914
2c08214a 915 } else {
2c08214a 916 $base_entry_id = db_fetch_result($result, 0, "id");
2c08214a
AD
917 }
918
919 // now it should exist, if not - bad luck then
920
b1840673 921 $result = db_query("SELECT id FROM ttrss_entries
5e3d5480 922 WHERE guid = '$entry_guid' OR guid = '$entry_guid_hashed'");
2c08214a
AD
923
924 $entry_ref_id = 0;
925 $entry_int_id = 0;
926
927 if (db_num_rows($result) == 1) {
928
68cccafc 929 _debug("base guid found, checking for user record", $debug_enabled);
2c08214a 930
2c08214a
AD
931 $ref_id = db_fetch_result($result, 0, "id");
932 $entry_ref_id = $ref_id;
933
5e3d5480
AD
934 /* $stored_guid = db_fetch_result($result, 0, "guid");
935 if ($stored_guid != $entry_guid_hashed) {
68cccafc 936 if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
5e3d5480 937
a42c55f0 938 db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
5e3d5480
AD
939 id = '$ref_id'");
940 } */
941
2c08214a 942 if (find_article_filter($article_filters, "filter")) {
07877caf 943 //db_query("COMMIT"); // close transaction in progress
2c08214a
AD
944 continue;
945 }
946
6de3a1be 947 $score = calculate_article_score($article_filters) + $entry_score_modifier;
2c08214a 948
6de3a1be 949 _debug("initial score: $score [including plugin modifier: $entry_score_modifier]", $debug_enabled);
2c08214a 950
4f186b1f
AD
951 // check for user post link to main table
952
2c08214a 953 $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
5bdcb8fd 954 ref_id = '$ref_id' AND owner_uid = '$owner_uid'";
2c08214a
AD
955
956// if ($_REQUEST["xdebug"]) print "$query\n";
957
a42c55f0 958 $result = db_query($query);
2c08214a
AD
959
960 // okay it doesn't exist - create user entry
961 if (db_num_rows($result) == 0) {
962
68cccafc 963 _debug("user record not found, creating...", $debug_enabled);
2c08214a 964
c9299c28 965 if ($score >= -500 && !find_article_filter($article_filters, 'catchup') && !$entry_force_catchup) {
2c08214a
AD
966 $unread = 'true';
967 $last_read_qpart = 'NULL';
968 } else {
969 $unread = 'false';
970 $last_read_qpart = 'NOW()';
971 }
972
973 if (find_article_filter($article_filters, 'mark') || $score > 1000) {
974 $marked = 'true';
975 } else {
976 $marked = 'false';
977 }
978
979 if (find_article_filter($article_filters, 'publish')) {
980 $published = 'true';
981 } else {
982 $published = 'false';
983 }
984
7873d588
AD
985 $last_marked = ($marked == 'true') ? 'NOW()' : 'NULL';
986 $last_published = ($published == 'true') ? 'NOW()' : 'NULL';
987
6322ac79 988 $result = db_query(
2c08214a
AD
989 "INSERT INTO ttrss_user_entries
990 (ref_id, owner_uid, feed_id, unread, last_read, marked,
7873d588
AD
991 published, score, tag_cache, label_cache, uuid,
992 last_marked, last_published)
2c08214a 993 VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
7873d588
AD
994 $last_read_qpart, $marked, $published, '$score', '', '',
995 '', $last_marked, $last_published)");
2c08214a
AD
996
997 if (PUBSUBHUBBUB_HUB && $published == 'true') {
998 $rss_link = get_self_url_prefix() .
999 "/public.php?op=rss&id=-2&key=" .
a42c55f0 1000 get_feed_access_key(-2, false, $owner_uid);
2c08214a 1001
cb3f8773 1002 $p = new pubsubhubbub\publisher\Publisher(PUBSUBHUBBUB_HUB);
2c08214a 1003
5ba1ddd4 1004 /* $pubsub_result = */ $p->publish_update($rss_link);
2c08214a
AD
1005 }
1006
6322ac79 1007 $result = db_query(
2c08214a
AD
1008 "SELECT int_id FROM ttrss_user_entries WHERE
1009 ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
1010 feed_id = '$feed' LIMIT 1");
1011
1012 if (db_num_rows($result) == 1) {
1013 $entry_int_id = db_fetch_result($result, 0, "int_id");
1014 }
1015 } else {
68cccafc 1016 _debug("user record FOUND", $debug_enabled);
2c08214a
AD
1017
1018 $entry_ref_id = db_fetch_result($result, 0, "ref_id");
1019 $entry_int_id = db_fetch_result($result, 0, "int_id");
1020 }
1021
68cccafc 1022 _debug("RID: $entry_ref_id, IID: $entry_int_id", $debug_enabled);
2c08214a 1023
e854442e 1024 if (DB_TYPE == "pgsql") {
978989bb 1025 $tsvector_combined = db_escape_string(mb_substr($entry_title . ' ' . strip_tags(str_replace('<', ' <', $entry_content)),
e854442e
AD
1026 0, 1000000));
1027
df659891 1028 $tsvector_qpart = "tsvector_combined = to_tsvector('$feed_language', '$tsvector_combined'),";
e854442e
AD
1029
1030 } else {
1031 $tsvector_qpart = "";
1032 }
1033
b1840673
AD
1034 db_query("UPDATE ttrss_entries
1035 SET title = '$entry_title',
1036 content = '$entry_content',
1037 content_hash = '$entry_current_hash',
1038 updated = '$entry_timestamp_fmt',
e854442e 1039 $tsvector_qpart
b1840673
AD
1040 num_comments = '$num_comments',
1041 plugin_data = '$entry_plugin_data',
1042 author = '$entry_author',
1043 lang = '$entry_language'
1044 WHERE id = '$ref_id'");
1045
59e83455
AD
1046 // update aux data
1047 db_query("UPDATE ttrss_user_entries
1048 SET score = '$score' WHERE ref_id = '$ref_id'");
1049
b1840673 1050 if ($mark_unread_on_update) {
24e6ff5d
AD
1051 _debug("article updated, marking unread as requested.", $debug_enabled);
1052
b1840673
AD
1053 db_query("UPDATE ttrss_user_entries
1054 SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
2c08214a
AD
1055 }
1056 }
1057
07877caf 1058 //db_query("COMMIT");
2c08214a 1059
a29fe121
AD
1060 _debug("assigning labels [other]...", $debug_enabled);
1061
1062 foreach ($article_labels as $label) {
1063 label_add_article($entry_ref_id, $label[1], $owner_uid);
1064 }
1065
1066 _debug("assigning labels [filters]...", $debug_enabled);
2c08214a 1067
a42c55f0 1068 assign_article_to_label_filters($entry_ref_id, $article_filters,
b24504b1 1069 $owner_uid, $article_labels);
2c08214a 1070
68cccafc 1071 _debug("looking for enclosures...", $debug_enabled);
2c08214a
AD
1072
1073 // enclosures
1074
1075 $enclosures = array();
1076
19b3992b 1077 $encs = $item->get_enclosures();
2c08214a 1078
19b3992b
AD
1079 if (is_array($encs)) {
1080 foreach ($encs as $e) {
1081 $e_item = array(
523bd90b 1082 $e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
2c08214a 1083 array_push($enclosures, $e_item);
2c08214a
AD
1084 }
1085 }
1086
2c08214a 1087 if ($debug_enabled) {
68cccafc 1088 _debug("article enclosures:", $debug_enabled);
2c08214a
AD
1089 print_r($enclosures);
1090 }
1091
dab229cd 1092 //db_query("BEGIN");
2c08214a 1093
5c54e683
AD
1094// debugging
1095// db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
1096
2c08214a 1097 foreach ($enclosures as $enc) {
a42c55f0
AD
1098 $enc_url = db_escape_string($enc[0]);
1099 $enc_type = db_escape_string($enc[1]);
1100 $enc_dur = db_escape_string($enc[2]);
5c54e683 1101 $enc_title = db_escape_string($enc[3]);
523bd90b
FE
1102 $enc_width = intval($enc[4]);
1103 $enc_height = intval($enc[5]);
2c08214a 1104
a42c55f0 1105 $result = db_query("SELECT id FROM ttrss_enclosures
2c08214a
AD
1106 WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
1107
1108 if (db_num_rows($result) == 0) {
a42c55f0 1109 db_query("INSERT INTO ttrss_enclosures
523bd90b
FE
1110 (content_url, content_type, title, duration, post_id, width, height) VALUES
1111 ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
2c08214a
AD
1112 }
1113 }
1114
dab229cd 1115 //db_query("COMMIT");
2c08214a
AD
1116
1117 // check for manual tags (we have to do it here since they're loaded from filters)
1118
1119 foreach ($article_filters as $f) {
6aff7845 1120 if ($f["type"] == "tag") {
2c08214a 1121
6aff7845 1122 $manual_tags = trim_array(explode(",", $f["param"]));
2c08214a
AD
1123
1124 foreach ($manual_tags as $tag) {
1125 if (tag_is_valid($tag)) {
1126 array_push($entry_tags, $tag);
1127 }
1128 }
1129 }
1130 }
1131
1132 // Skip boring tags
1133
6322ac79 1134 $boring_tags = trim_array(explode(",", mb_strtolower(get_pref(
2c08214a
AD
1135 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
1136
1137 $filtered_tags = array();
1138 $tags_to_cache = array();
1139
1140 if ($entry_tags && is_array($entry_tags)) {
1141 foreach ($entry_tags as $tag) {
1142 if (array_search($tag, $boring_tags) === false) {
1143 array_push($filtered_tags, $tag);
1144 }
1145 }
1146 }
1147
1148 $filtered_tags = array_unique($filtered_tags);
1149
1150 if ($debug_enabled) {
68cccafc 1151 _debug("filtered article tags:", $debug_enabled);
2c08214a
AD
1152 print_r($filtered_tags);
1153 }
1154
1155 // Save article tags in the database
1156
1157 if (count($filtered_tags) > 0) {
1158
dab229cd 1159 //db_query("BEGIN");
2c08214a
AD
1160
1161 foreach ($filtered_tags as $tag) {
1162
1163 $tag = sanitize_tag($tag);
a42c55f0 1164 $tag = db_escape_string($tag);
2c08214a
AD
1165
1166 if (!tag_is_valid($tag)) continue;
1167
a42c55f0 1168 $result = db_query("SELECT id FROM ttrss_tags
2c08214a
AD
1169 WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
1170 owner_uid = '$owner_uid' LIMIT 1");
1171
1172 if ($result && db_num_rows($result) == 0) {
1173
a42c55f0 1174 db_query("INSERT INTO ttrss_tags
2c08214a
AD
1175 (owner_uid,tag_name,post_int_id)
1176 VALUES ('$owner_uid','$tag', '$entry_int_id')");
1177 }
1178
1179 array_push($tags_to_cache, $tag);
1180 }
1181
1182 /* update the cache */
1183
1184 $tags_to_cache = array_unique($tags_to_cache);
1185
a42c55f0 1186 $tags_str = db_escape_string(join(",", $tags_to_cache));
2c08214a 1187
a42c55f0 1188 db_query("UPDATE ttrss_user_entries
2c08214a
AD
1189 SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
1190 AND owner_uid = $owner_uid");
1191
dab229cd 1192 //db_query("COMMIT");
2c08214a
AD
1193 }
1194
68cccafc 1195 _debug("article processed", $debug_enabled);
2c08214a
AD
1196 }
1197
68cccafc 1198 _debug("purging feed...", $debug_enabled);
2c08214a 1199
a42c55f0 1200 purge_feed($feed, 0, $debug_enabled);
2c08214a 1201
a42c55f0 1202 db_query("UPDATE ttrss_feeds
f2798eb6 1203 SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
2c08214a 1204
a42c55f0 1205// db_query("COMMIT");
2c08214a
AD
1206
1207 } else {
1208
a42c55f0 1209 $error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
2c08214a 1210
4ad04ee2
AD
1211 _debug("fetch error: $error_msg", $debug_enabled);
1212
1213 if (count($rss->errors()) > 1) {
1214 foreach ($rss->errors() as $error) {
1215 _debug("+ $error");
1216 }
1217 }
2c08214a 1218
6322ac79 1219 db_query(
2c08214a 1220 "UPDATE ttrss_feeds SET last_error = '$error_msg',
88edaa93 1221 last_updated = NOW() WHERE id = '$feed'");
2c08214a 1222
88edaa93
AD
1223 unset($rss);
1224 }
2c08214a 1225
68cccafc 1226 _debug("done", $debug_enabled);
88edaa93
AD
1227
1228 return $rss;
2c08214a
AD
1229 }
1230
3c696512 1231 function cache_images($html, $site_url, $debug) {
3c696512
AD
1232 libxml_use_internal_errors(true);
1233
1234 $charset_hack = '<head>
1235 <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
1236 </head>';
1237
1238 $doc = new DOMDocument();
1239 $doc->loadHTML($charset_hack . $html);
1240 $xpath = new DOMXPath($doc);
1241
1242 $entries = $xpath->query('(//img[@src])');
1243
1244 foreach ($entries as $entry) {
1245 if ($entry->hasAttribute('src')) {
1246 $src = rewrite_relative_url($site_url, $entry->getAttribute('src'));
1247
1248 $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png";
1249
1250 if ($debug) _debug("cache_images: downloading: $src to $local_filename");
1251
1252 if (!file_exists($local_filename)) {
1253 $file_content = fetch_file_contents($src);
1254
a3d50184 1255 if ($file_content && strlen($file_content) > _MIN_CACHE_IMAGE_SIZE) {
3c696512
AD
1256 file_put_contents($local_filename, $file_content);
1257 }
4a27966e
J
1258 } else {
1259 touch($local_filename);
3c696512 1260 }
3c696512
AD
1261 }
1262 }
3c696512
AD
1263 }
1264
a42c55f0 1265 function expire_error_log($debug) {
e2261e17
AD
1266 if ($debug) _debug("Removing old error log entries...");
1267
1268 if (DB_TYPE == "pgsql") {
a42c55f0 1269 db_query("DELETE FROM ttrss_error_log
e2261e17
AD
1270 WHERE created_at < NOW() - INTERVAL '7 days'");
1271 } else {
a42c55f0 1272 db_query("DELETE FROM ttrss_error_log
e2261e17
AD
1273 WHERE created_at < DATE_SUB(NOW(), INTERVAL 7 DAY)");
1274 }
1275
1276 }
1277
2a91b6ff 1278 function expire_lock_files($debug) {
65465085 1279 //if ($debug) _debug("Removing old lock files...");
2a91b6ff
AD
1280
1281 $num_deleted = 0;
1282
1283 if (is_writable(LOCK_DIRECTORY)) {
1284 $files = glob(LOCK_DIRECTORY . "/*.lock");
1285
1286 if ($files) {
1287 foreach ($files as $file) {
11344971 1288 if (!file_is_locked(basename($file)) && time() - filemtime($file) > 86400*2) {
2a91b6ff
AD
1289 unlink($file);
1290 ++$num_deleted;
1291 }
1292 }
1293 }
1294 }
1295
65465085 1296 if ($debug) _debug("Removed $num_deleted old lock files.");
2a91b6ff
AD
1297 }
1298
3c696512 1299 function expire_cached_files($debug) {
3306daec 1300 foreach (array("simplepie", "images", "export", "upload") as $dir) {
3c696512 1301 $cache_dir = CACHE_DIR . "/$dir";
2c08214a 1302
65465085 1303// if ($debug) _debug("Expiring $cache_dir");
2c08214a 1304
3c696512
AD
1305 $num_deleted = 0;
1306
1307 if (is_writable($cache_dir)) {
1308 $files = glob("$cache_dir/*");
1309
2a91b6ff 1310 if ($files) {
2ab20c31
AD
1311 foreach ($files as $file) {
1312 if (time() - filemtime($file) > 86400*7) {
1313 unlink($file);
3c696512 1314
2ab20c31
AD
1315 ++$num_deleted;
1316 }
3c696512
AD
1317 }
1318 }
2a91b6ff 1319 }
3c696512 1320
65465085 1321 if ($debug) _debug("$cache_dir: removed $num_deleted files.");
3c696512
AD
1322 }
1323 }
2c08214a 1324
a3e0bdcf
AD
1325 /**
1326 * Source: http://www.php.net/manual/en/function.parse-url.php#104527
1327 * Returns the url query as associative array
1328 *
1329 * @param string query
1330 * @return array params
1331 */
1332 function convertUrlQuery($query) {
1333 $queryParts = explode('&', $query);
1334
1335 $params = array();
1336
1337 foreach ($queryParts as $param) {
1338 $item = explode('=', $param);
1339 $params[$item[0]] = $item[1];
1340 }
1341
1342 return $params;
1343 }
92c14e9d 1344
557d86fe 1345 function get_article_filters($filters, $title, $content, $link, $timestamp, $author, $tags, &$matched_rules = false) {
92c14e9d
AD
1346 $matches = array();
1347
1348 foreach ($filters as $filter) {
1349 $match_any_rule = $filter["match_any_rule"];
a3a896a1 1350 $inverse = $filter["inverse"];
92c14e9d
AD
1351 $filter_match = false;
1352
1353 foreach ($filter["rules"] as $rule) {
1354 $match = false;
ffa1bd7b 1355 $reg_exp = str_replace('/', '\/', $rule["reg_exp"]);
a3a896a1 1356 $rule_inverse = $rule["inverse"];
92c14e9d
AD
1357
1358 if (!$reg_exp)
1359 continue;
1360
1361 switch ($rule["type"]) {
1362 case "title":
e3dceca6 1363 $match = @preg_match("/$reg_exp/iu", $title);
92c14e9d
AD
1364 break;
1365 case "content":
d03ae73e
AD
1366 // we don't need to deal with multiline regexps
1367 $content = preg_replace("/[\r\n\t]/", "", $content);
1368
e3dceca6 1369 $match = @preg_match("/$reg_exp/iu", $content);
92c14e9d
AD
1370 break;
1371 case "both":
d03ae73e
AD
1372 // we don't need to deal with multiline regexps
1373 $content = preg_replace("/[\r\n\t]/", "", $content);
1374
e3dceca6 1375 $match = (@preg_match("/$reg_exp/iu", $title) || @preg_match("/$reg_exp/iu", $content));
92c14e9d
AD
1376 break;
1377 case "link":
e3dceca6 1378 $match = @preg_match("/$reg_exp/iu", $link);
92c14e9d
AD
1379 break;
1380 case "author":
e3dceca6 1381 $match = @preg_match("/$reg_exp/iu", $author);
92c14e9d
AD
1382 break;
1383 case "tag":
7b80b5e1 1384 foreach ($tags as $tag) {
e3dceca6 1385 if (@preg_match("/$reg_exp/iu", $tag)) {
7b80b5e1
DK
1386 $match = true;
1387 break;
1388 }
1389 }
92c14e9d
AD
1390 break;
1391 }
1392
a3a896a1
AD
1393 if ($rule_inverse) $match = !$match;
1394
92c14e9d
AD
1395 if ($match_any_rule) {
1396 if ($match) {
1397 $filter_match = true;
1398 break;
1399 }
1400 } else {
1401 $filter_match = $match;
1402 if (!$match) {
1403 break;
1404 }
1405 }
1406 }
1407
a3a896a1
AD
1408 if ($inverse) $filter_match = !$filter_match;
1409
92c14e9d 1410 if ($filter_match) {
557d86fe
AD
1411 if (is_array($matched_rules)) array_push($matched_rules, $rule);
1412
92c14e9d
AD
1413 foreach ($filter["actions"] AS $action) {
1414 array_push($matches, $action);
5e736e45
AD
1415
1416 // if Stop action encountered, perform no further processing
fd3e5e8d 1417 if (isset($action["type"]) && $action["type"] == "stop") return $matches;
92c14e9d
AD
1418 }
1419 }
1420 }
1421
1422 return $matches;
1423 }
1424
1425 function find_article_filter($filters, $filter_name) {
1426 foreach ($filters as $f) {
1427 if ($f["type"] == $filter_name) {
1428 return $f;
1429 };
1430 }
1431 return false;
1432 }
1433
1434 function find_article_filters($filters, $filter_name) {
1435 $results = array();
1436
1437 foreach ($filters as $f) {
1438 if ($f["type"] == $filter_name) {
1439 array_push($results, $f);
1440 };
1441 }
1442 return $results;
1443 }
1444
1445 function calculate_article_score($filters) {
1446 $score = 0;
1447
1448 foreach ($filters as $f) {
1449 if ($f["type"] == "score") {
1450 $score += $f["param"];
1451 };
1452 }
1453 return $score;
1454 }
1455
b24504b1
AD
1456 function labels_contains_caption($labels, $caption) {
1457 foreach ($labels as $label) {
1458 if ($label[1] == $caption) {
1459 return true;
1460 }
1461 }
1462
1463 return false;
1464 }
1465
a42c55f0 1466 function assign_article_to_label_filters($id, $filters, $owner_uid, $article_labels) {
92c14e9d
AD
1467 foreach ($filters as $f) {
1468 if ($f["type"] == "label") {
b24504b1 1469 if (!labels_contains_caption($article_labels, $f["param"])) {
a42c55f0 1470 label_add_article($id, $f["param"], $owner_uid);
b24504b1
AD
1471 }
1472 }
92c14e9d
AD
1473 }
1474 }
87764a50 1475
87d7e850
AD
1476 function make_guid_from_title($title) {
1477 return preg_replace("/[ \"\',.:;]/", "-",
1478 mb_strtolower(strip_tags($title), 'utf-8'));
1479 }
1480
d1f3fa97 1481 /* function verify_feed_xml($feed_data) {
ebec81a6
AD
1482 libxml_use_internal_errors(true);
1483 $doc = new DOMDocument();
1484 $doc->loadXML($feed_data);
1485 $error = libxml_get_last_error();
1486 libxml_clear_errors();
1487 return $error;
d1f3fa97 1488 } */
87d7e850 1489
168cf351
AD
1490 function cleanup_counters_cache($debug) {
1491 $result = db_query("DELETE FROM ttrss_counters_cache
1492 WHERE feed_id > 0 AND
1493 (SELECT COUNT(id) FROM ttrss_feeds WHERE
1494 id = feed_id AND
1495 ttrss_counters_cache.owner_uid = ttrss_feeds.owner_uid) = 0");
1496 $frows = db_affected_rows($result);
1497
1498 $result = db_query("DELETE FROM ttrss_cat_counters_cache
1499 WHERE feed_id > 0 AND
1500 (SELECT COUNT(id) FROM ttrss_feed_categories WHERE
1501 id = feed_id AND
1502 ttrss_cat_counters_cache.owner_uid = ttrss_feed_categories.owner_uid) = 0");
1503 $crows = db_affected_rows($result);
1504
1505 _debug("Removed $frows (feeds) $crows (cats) orphaned counter cache entries.");
1506 }
1507
5cbd1fe8
AD
1508 function housekeeping_user($owner_uid) {
1509 $tmph = new PluginHost();
1510
1511 load_user_plugins($owner_uid, $tmph);
1512
1513 $tmph->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
1514 }
1515
e2cf81e2
AD
1516 function housekeeping_common($debug) {
1517 expire_cached_files($debug);
1518 expire_lock_files($debug);
1519 expire_error_log($debug);
1520
1521 $count = update_feedbrowser_cache();
1522 _debug("Feedbrowser updated, $count feeds processed.");
1523
1524 purge_orphans( true);
168cf351 1525 cleanup_counters_cache($debug);
e2cf81e2 1526
9b736a20
AD
1527 //$rc = cleanup_tags( 14, 50000);
1528 //_debug("Cleaned $rc cached tags.");
8e470220 1529
00f22824 1530 PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
e2cf81e2 1531 }
2c08214a 1532?>