git.wh0rd.org / tt-rss.git / blame
| Commit | Line | Data |
|---|---|---|
| 2c08214a | 1 | <?php |
| 09e8bdfd AD |
2 | define('DAEMON_UPDATE_LOGIN_LIMIT', 30); |
| 3 | define('DAEMON_FEED_LIMIT', 100); | |
| 4 | define('DAEMON_SLEEP_INTERVAL', 120); | |
| 5 | ||
| 79178062 AD |
6 | function update_feedbrowser_cache($link) { |
| 7 | ||
| 8 | $result = db_query($link, "SELECT feed_url, site_url, title, COUNT(id) AS subscribers | |
| 9 | FROM ttrss_feeds WHERE (SELECT COUNT(id) = 0 FROM ttrss_feeds AS tf | |
| 10 | WHERE tf.feed_url = ttrss_feeds.feed_url | |
| 11 | AND (private IS true OR auth_login != '' OR auth_pass != '' OR feed_url LIKE '%:%@%/%')) | |
| 12 | GROUP BY feed_url, site_url, title ORDER BY subscribers DESC LIMIT 1000"); | |
| 13 | ||
| 14 | db_query($link, "BEGIN"); | |
| 15 | ||
| 16 | db_query($link, "DELETE FROM ttrss_feedbrowser_cache"); | |
| 17 | ||
| 18 | $count = 0; | |
| 19 | ||
| 20 | while ($line = db_fetch_assoc($result)) { | |
| 21 | $subscribers = db_escape_string($line["subscribers"]); | |
| 22 | $feed_url = db_escape_string($line["feed_url"]); | |
| 23 | $title = db_escape_string($line["title"]); | |
| 24 | $site_url = db_escape_string($line["site_url"]); | |
| 25 | ||
| 26 | $tmp_result = db_query($link, "SELECT subscribers FROM | |
| 27 | ttrss_feedbrowser_cache WHERE feed_url = '$feed_url'"); | |
| 28 | ||
| 29 | if (db_num_rows($tmp_result) == 0) { | |
| 30 | ||
| 31 | db_query($link, "INSERT INTO ttrss_feedbrowser_cache | |
| 32 | (feed_url, site_url, title, subscribers) VALUES ('$feed_url', | |
| 33 | '$site_url', '$title', '$subscribers')"); | |
| 34 | ||
| 35 | ++$count; | |
| 36 | ||
| 37 | } | |
| 38 | ||
| 39 | } | |
| 40 | ||
| 41 | db_query($link, "COMMIT"); | |
| 42 | ||
| 43 | return $count; | |
| 44 | ||
| 45 | } | |
| 46 | ||
| 47 | ||
| 2c08214a AD |
48 | /** |
| 49 | * Update a feed batch. | |
| 50 | * Used by daemons to update n feeds by run. | |
| 51 | * Only update feed needing a update, and not being processed | |
| 52 | * by another process. | |
| 53 | * | |
| 54 | * @param mixed $link Database link | |
| 55 | * @param integer $limit Maximum number of feeds in update batch. Default to DAEMON_FEED_LIMIT. | |
| 56 | * @param boolean $from_http Set to true if you call this function from http to disable cli specific code. | |
| 57 | * @param boolean $debug Set to false to disable debug output. Default to true. | |
| 58 | * @return void | |
| 59 | */ | |
| 60 | function update_daemon_common($link, $limit = DAEMON_FEED_LIMIT, $from_http = false, $debug = true) { | |
| 61 | // Process all other feeds using last_updated and interval parameters | |
| 62 | ||
| 63 | // Test if the user has loggued in recently. If not, it does not update its feeds. | |
| 09e8bdfd | 64 | if (!SINGLE_USER_MODE && DAEMON_UPDATE_LOGIN_LIMIT > 0) { |
| 2c08214a AD |
65 | if (DB_TYPE == "pgsql") { |
| 66 | $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'"; | |
| 67 | } else { | |
| 68 | $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)"; | |
| 69 | } | |
| 70 | } else { | |
| 71 | $login_thresh_qpart = ""; | |
| 72 | } | |
| 73 | ||
| 74 | // Test if the feed need a update (update interval exceded). | |
| 75 | if (DB_TYPE == "pgsql") { | |
| 76 | $update_limit_qpart = "AND (( | |
| 77 | ttrss_feeds.update_interval = 0 | |
| 78 | AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL) | |
| 79 | ) OR ( | |
| 80 | ttrss_feeds.update_interval > 0 | |
| 81 | AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL) | |
| 82 | ) OR ttrss_feeds.last_updated IS NULL)"; | |
| 83 | } else { | |
| 84 | $update_limit_qpart = "AND (( | |
| 85 | ttrss_feeds.update_interval = 0 | |
| 86 | AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE) | |
| 87 | ) OR ( | |
| 88 | ttrss_feeds.update_interval > 0 | |
| 89 | AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE) | |
| 90 | ) OR ttrss_feeds.last_updated IS NULL)"; | |
| 91 | } | |
| 92 | ||
| 93 | // Test if feed is currently being updated by another process. | |
| 94 | if (DB_TYPE == "pgsql") { | |
| 95 | $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '5 minutes')"; | |
| 96 | } else { | |
| 97 | $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 5 MINUTE))"; | |
| 98 | } | |
| 99 | ||
| 100 | // Test if there is a limit to number of updated feeds | |
| 101 | $query_limit = ""; | |
| 102 | if($limit) $query_limit = sprintf("LIMIT %d", $limit); | |
| 103 | ||
| 104 | $random_qpart = sql_random_function(); | |
| 105 | ||
| 106 | // We search for feed needing update. | |
| 107 | $result = db_query($link, "SELECT ttrss_feeds.feed_url,ttrss_feeds.id, ttrss_feeds.owner_uid, | |
| 108 | ".SUBSTRING_FOR_DATE."(ttrss_feeds.last_updated,1,19) AS last_updated, | |
| 109 | ttrss_feeds.update_interval | |
| 110 | FROM | |
| 111 | ttrss_feeds, ttrss_users, ttrss_user_prefs | |
| 112 | WHERE | |
| 113 | ttrss_feeds.owner_uid = ttrss_users.id | |
| 114 | AND ttrss_users.id = ttrss_user_prefs.owner_uid | |
| 115 | AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' | |
| 116 | $login_thresh_qpart $update_limit_qpart | |
| 117 | $updstart_thresh_qpart | |
| 118 | ORDER BY $random_qpart $query_limit"); | |
| 119 | ||
| 120 | $user_prefs_cache = array(); | |
| 121 | ||
| 122 | if($debug) _debug(sprintf("Scheduled %d feeds to update...\n", db_num_rows($result))); | |
| 123 | ||
| 124 | // Here is a little cache magic in order to minimize risk of double feed updates. | |
| 125 | $feeds_to_update = array(); | |
| 126 | while ($line = db_fetch_assoc($result)) { | |
| 127 | $feeds_to_update[$line['id']] = $line; | |
| 128 | } | |
| 129 | ||
| 130 | // We update the feed last update started date before anything else. | |
| 131 | // There is no lag due to feed contents downloads | |
| 132 | // It prevent an other process to update the same feed. | |
| 133 | $feed_ids = array_keys($feeds_to_update); | |
| 134 | if($feed_ids) { | |
| 135 | db_query($link, sprintf("UPDATE ttrss_feeds SET last_update_started = NOW() | |
| 136 | WHERE id IN (%s)", implode(',', $feed_ids))); | |
| 137 | } | |
| 138 | ||
| 3c696512 AD |
139 | expire_cached_files($debug); |
| 140 | ||
| 2c08214a AD |
141 | // For each feed, we call the feed update function. |
| 142 | while ($line = array_pop($feeds_to_update)) { | |
| 143 | ||
| 144 | if($debug) _debug("Feed: " . $line["feed_url"] . ", " . $line["last_updated"]); | |
| 145 | ||
| 146 | update_rss_feed($link, $line["id"], true); | |
| 147 | ||
| 148 | sleep(1); // prevent flood (FIXME make this an option?) | |
| 149 | } | |
| 150 | ||
| 151 | // Send feed digests by email if needed. | |
| 036cd3a4 | 152 | send_headlines_digests($link, 100, $debug); |
| 2c08214a AD |
153 | |
| 154 | } // function update_daemon_common | |
| 155 | ||
| 156 | function fetch_twitter_rss($link, $url, $owner_uid) { | |
| 157 | ||
| 158 | require_once 'lib/tmhoauth/tmhOAuth.php'; | |
| 159 | ||
| 160 | $result = db_query($link, "SELECT twitter_oauth FROM ttrss_users | |
| 161 | WHERE id = $owner_uid"); | |
| 162 | ||
| 163 | $access_token = json_decode(db_fetch_result($result, 0, 'twitter_oauth'), true); | |
| 164 | $url_escaped = db_escape_string($url); | |
| 165 | ||
| 166 | if ($access_token) { | |
| 167 | ||
| 168 | $tmhOAuth = new tmhOAuth(array( | |
| 169 | 'consumer_key' => CONSUMER_KEY, | |
| 170 | 'consumer_secret' => CONSUMER_SECRET, | |
| 171 | 'user_token' => $access_token['oauth_token'], | |
| 172 | 'user_secret' => $access_token['oauth_token_secret'], | |
| 173 | )); | |
| 174 | ||
| 175 | $code = $tmhOAuth->request('GET', $url); | |
| 176 | ||
| 177 | if ($code == 200) { | |
| 178 | ||
| 179 | $content = $tmhOAuth->response['response']; | |
| 180 | ||
| 181 | define('MAGPIE_CACHE_ON', false); | |
| 182 | ||
| 183 | $rss = new MagpieRSS($content, MAGPIE_OUTPUT_ENCODING, | |
| 184 | MAGPIE_INPUT_ENCODING, MAGPIE_DETECT_ENCODING ); | |
| 185 | ||
| 186 | return $rss; | |
| 187 | ||
| 188 | } else { | |
| 189 | ||
| 190 | db_query($link, "UPDATE ttrss_feeds | |
| 191 | SET last_error = 'OAuth authorization failed ($code).' | |
| 192 | WHERE feed_url = '$url_escaped' AND owner_uid = $owner_uid"); | |
| 193 | } | |
| 194 | ||
| 195 | } else { | |
| 196 | ||
| 197 | db_query($link, "UPDATE ttrss_feeds | |
| 198 | SET last_error = 'OAuth information not found.' | |
| 199 | WHERE feed_url = '$url_escaped' AND owner_uid = $owner_uid"); | |
| 200 | ||
| 201 | return false; | |
| 202 | } | |
| 203 | } | |
| 204 | ||
| 205 | function update_rss_feed($link, $feed, $ignore_daemon = false, $no_cache = false) { | |
| 206 | ||
| 207 | global $memcache; | |
| 208 | ||
| 209 | /* Update all feeds with the same URL to utilize memcache */ | |
| 210 | ||
| 211 | if ($memcache) { | |
| 212 | $result = db_query($link, "SELECT f1.id | |
| 213 | FROM ttrss_feeds AS f1, ttrss_feeds AS f2 | |
| 214 | WHERE f2.feed_url = f1.feed_url AND f2.id = '$feed'"); | |
| 215 | ||
| 216 | while ($line = db_fetch_assoc($result)) { | |
| 217 | update_rss_feed_real($link, $line["id"], $ignore_daemon, $no_cache); | |
| 218 | } | |
| 219 | } else { | |
| 220 | update_rss_feed_real($link, $feed, $ignore_daemon, $no_cache); | |
| 221 | } | |
| 222 | } | |
| 223 | ||
| 224 | function update_rss_feed_real($link, $feed, $ignore_daemon = false, $no_cache = false, | |
| 225 | $override_url = false) { | |
| 226 | ||
| 227 | require_once "lib/simplepie/simplepie.inc"; | |
| 228 | require_once "lib/magpierss/rss_fetch.inc"; | |
| 229 | require_once 'lib/magpierss/rss_utils.inc'; | |
| 230 | ||
| 231 | global $memcache; | |
| 232 | ||
| 233 | $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug']; | |
| 234 | ||
| 235 | if (!$_REQUEST["daemon"] && !$ignore_daemon) { | |
| 236 | return false; | |
| 237 | } | |
| 238 | ||
| 239 | if ($debug_enabled) { | |
| 240 | _debug("update_rss_feed: start"); | |
| 241 | } | |
| 242 | ||
| 243 | if (!$ignore_daemon) { | |
| 244 | ||
| 245 | if (DB_TYPE == "pgsql") { | |
| 246 | $updstart_thresh_qpart = "(ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '120 seconds')"; | |
| 247 | } else { | |
| 248 | $updstart_thresh_qpart = "(ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 120 SECOND))"; | |
| 249 | } | |
| 250 | ||
| 251 | $result = db_query($link, "SELECT id,update_interval,auth_login, | |
| 252 | auth_pass,cache_images,update_method | |
| 253 | FROM ttrss_feeds WHERE id = '$feed' AND $updstart_thresh_qpart"); | |
| 254 | ||
| 255 | } else { | |
| 256 | ||
| 257 | $result = db_query($link, "SELECT id,update_interval,auth_login, | |
| 258 | feed_url,auth_pass,cache_images,update_method,last_updated, | |
| 259 | mark_unread_on_update, owner_uid, update_on_checksum_change, | |
| 260 | pubsub_state | |
| 261 | FROM ttrss_feeds WHERE id = '$feed'"); | |
| 262 | ||
| 263 | } | |
| 264 | ||
| 265 | if (db_num_rows($result) == 0) { | |
| 266 | if ($debug_enabled) { | |
| 267 | _debug("update_rss_feed: feed $feed NOT FOUND/SKIPPED"); | |
| 268 | } | |
| 269 | return false; | |
| 270 | } | |
| 271 | ||
| 272 | $update_method = db_fetch_result($result, 0, "update_method"); | |
| 273 | $last_updated = db_fetch_result($result, 0, "last_updated"); | |
| 274 | $owner_uid = db_fetch_result($result, 0, "owner_uid"); | |
| 275 | $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result, | |
| 276 | 0, "mark_unread_on_update")); | |
| 277 | $update_on_checksum_change = sql_bool_to_bool(db_fetch_result($result, | |
| 278 | 0, "update_on_checksum_change")); | |
| 279 | $pubsub_state = db_fetch_result($result, 0, "pubsub_state"); | |
| 280 | ||
| 281 | db_query($link, "UPDATE ttrss_feeds SET last_update_started = NOW() | |
| 282 | WHERE id = '$feed'"); | |
| 283 | ||
| 284 | $auth_login = db_fetch_result($result, 0, "auth_login"); | |
| 285 | $auth_pass = db_fetch_result($result, 0, "auth_pass"); | |
| 286 | ||
| 287 | if ($update_method == 0) | |
| 288 | $update_method = DEFAULT_UPDATE_METHOD + 1; | |
| 289 | ||
| 290 | // 1 - Magpie | |
| 291 | // 2 - SimplePie | |
| 292 | // 3 - Twitter OAuth | |
| 293 | ||
| 294 | if ($update_method == 2) | |
| 295 | $use_simplepie = true; | |
| 296 | else | |
| 297 | $use_simplepie = false; | |
| 298 | ||
| 299 | if ($debug_enabled) { | |
| 300 | _debug("update method: $update_method (feed setting: $update_method) (use simplepie: $use_simplepie)\n"); | |
| 301 | } | |
| 302 | ||
| 303 | if ($update_method == 1) { | |
| 304 | $auth_login = urlencode($auth_login); | |
| 305 | $auth_pass = urlencode($auth_pass); | |
| 306 | } | |
| 307 | ||
| 308 | $update_interval = db_fetch_result($result, 0, "update_interval"); | |
| 309 | $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images")); | |
| 310 | $fetch_url = db_fetch_result($result, 0, "feed_url"); | |
| 311 | ||
| 312 | if ($update_interval < 0) { return false; } | |
| 313 | ||
| 314 | $feed = db_escape_string($feed); | |
| 315 | ||
| 316 | if ($auth_login && $auth_pass ){ | |
| 317 | $url_parts = array(); | |
| 318 | preg_match("/(^[^:]*):\/\/(.*)/", $fetch_url, $url_parts); | |
| 319 | ||
| 320 | if ($url_parts[1] && $url_parts[2]) { | |
| 321 | $fetch_url = $url_parts[1] . "://$auth_login:$auth_pass@" . $url_parts[2]; | |
| 322 | } | |
| 323 | ||
| 324 | } | |
| 325 | ||
| 326 | if ($override_url) | |
| 327 | $fetch_url = $override_url; | |
| 328 | ||
| 329 | if ($debug_enabled) { | |
| 330 | _debug("update_rss_feed: fetching [$fetch_url]..."); | |
| 331 | } | |
| 332 | ||
| 333 | $obj_id = md5("FDATA:$use_simplepie:$fetch_url"); | |
| 334 | ||
| 335 | if ($memcache && $obj = $memcache->get($obj_id)) { | |
| 336 | ||
| 337 | if ($debug_enabled) { | |
| 338 | _debug("update_rss_feed: data found in memcache."); | |
| 339 | } | |
| 340 | ||
| 341 | $rss = $obj; | |
| 342 | ||
| 343 | } else { | |
| 344 | ||
| 345 | if ($update_method == 3) { | |
| 346 | $rss = fetch_twitter_rss($link, $fetch_url, $owner_uid); | |
| 347 | } else if ($update_method == 1) { | |
| 348 | ||
| 349 | define('MAGPIE_CACHE_AGE', get_feed_update_interval($link, $feed) * 60); | |
| 350 | define('MAGPIE_CACHE_ON', !$no_cache); | |
| 351 | define('MAGPIE_FETCH_TIME_OUT', 60); | |
| 352 | define('MAGPIE_CACHE_DIR', CACHE_DIR . "/magpie"); | |
| 353 | ||
| 354 | $rss = @fetch_rss($fetch_url); | |
| 355 | } else { | |
| 356 | $simplepie_cache_dir = CACHE_DIR . "/simplepie"; | |
| 357 | ||
| 358 | if (!is_dir($simplepie_cache_dir)) { | |
| 359 | mkdir($simplepie_cache_dir); | |
| 360 | } | |
| 361 | ||
| 362 | $rss = new SimplePie(); | |
| 363 | $rss->set_useragent(SELF_USER_AGENT); | |
| 364 | # $rss->set_timeout(10); | |
| 365 | $rss->set_feed_url($fetch_url); | |
| 366 | $rss->set_output_encoding('UTF-8'); | |
| f1e511c8 | 367 | //$rss->force_feed(true); |
| 2c08214a | 368 | |
| 2c08214a AD |
369 | if ($debug_enabled) { |
| 370 | _debug("feed update interval (sec): " . | |
| 371 | get_feed_update_interval($link, $feed)*60); | |
| 372 | } | |
| 373 | ||
| 374 | $rss->enable_cache(!$no_cache); | |
| 375 | ||
| 376 | if (!$no_cache) { | |
| 377 | $rss->set_cache_location($simplepie_cache_dir); | |
| 378 | $rss->set_cache_duration(get_feed_update_interval($link, $feed) * 60); | |
| 379 | } | |
| 380 | ||
| 381 | $rss->init(); | |
| 382 | } | |
| 383 | ||
| 384 | if ($memcache && $rss) $memcache->add($obj_id, $rss, 0, 300); | |
| 385 | } | |
| 386 | ||
| 387 | // print_r($rss); | |
| 388 | ||
| 389 | if ($debug_enabled) { | |
| 390 | _debug("update_rss_feed: fetch done, parsing..."); | |
| 391 | } | |
| 392 | ||
| 393 | $feed = db_escape_string($feed); | |
| 394 | ||
| 395 | if ($update_method == 2) { | |
| 396 | $fetch_ok = !$rss->error(); | |
| 397 | } else { | |
| 398 | $fetch_ok = !!$rss; | |
| 399 | } | |
| 400 | ||
| 401 | if ($fetch_ok) { | |
| 402 | ||
| 403 | if ($debug_enabled) { | |
| 404 | _debug("update_rss_feed: processing feed data..."); | |
| 405 | } | |
| 406 | ||
| 407 | // db_query($link, "BEGIN"); | |
| 408 | ||
| 409 | $result = db_query($link, "SELECT title,icon_url,site_url,owner_uid | |
| 410 | FROM ttrss_feeds WHERE id = '$feed'"); | |
| 411 | ||
| 412 | $registered_title = db_fetch_result($result, 0, "title"); | |
| 413 | $orig_icon_url = db_fetch_result($result, 0, "icon_url"); | |
| 414 | $orig_site_url = db_fetch_result($result, 0, "site_url"); | |
| 415 | ||
| 416 | $owner_uid = db_fetch_result($result, 0, "owner_uid"); | |
| 417 | ||
| 418 | if ($use_simplepie) { | |
| 419 | $site_url = $rss->get_link(); | |
| 420 | } else { | |
| 421 | $site_url = $rss->channel["link"]; | |
| 422 | } | |
| 423 | ||
| 424 | $site_url = rewrite_relative_url($fetch_url, $site_url); | |
| 425 | ||
| 426 | if ($debug_enabled) { | |
| 427 | _debug("update_rss_feed: checking favicon..."); | |
| 428 | } | |
| 429 | ||
| 430 | check_feed_favicon($site_url, $feed, $link); | |
| 431 | ||
| 432 | if (!$registered_title || $registered_title == "[Unknown]") { | |
| 433 | ||
| 434 | if ($use_simplepie) { | |
| 435 | $feed_title = db_escape_string($rss->get_title()); | |
| 436 | } else { | |
| 437 | $feed_title = db_escape_string($rss->channel["title"]); | |
| 438 | } | |
| 439 | ||
| 440 | if ($debug_enabled) { | |
| 441 | _debug("update_rss_feed: registering title: $feed_title"); | |
| 442 | } | |
| 443 | ||
| 444 | db_query($link, "UPDATE ttrss_feeds SET | |
| 445 | title = '$feed_title' WHERE id = '$feed'"); | |
| 446 | } | |
| 447 | ||
| 448 | // weird, weird Magpie | |
| 449 | if (!$use_simplepie) { | |
| 450 | if (!$site_url) $site_url = db_escape_string($rss->channel["link_"]); | |
| 451 | } | |
| 452 | ||
| 453 | if ($site_url && $orig_site_url != db_escape_string($site_url)) { | |
| 454 | db_query($link, "UPDATE ttrss_feeds SET | |
| 455 | site_url = '$site_url' WHERE id = '$feed'"); | |
| 456 | } | |
| 457 | ||
| 458 | // print "I: " . $rss->channel["image"]["url"]; | |
| 459 | ||
| 460 | if (!$use_simplepie) { | |
| 461 | $icon_url = db_escape_string($rss->image["url"]); | |
| 462 | } else { | |
| 463 | $icon_url = db_escape_string($rss->get_image_url()); | |
| 464 | } | |
| 465 | ||
| 466 | $icon_url = substr($icon_url, 0, 250); | |
| 467 | ||
| 468 | if ($icon_url && $orig_icon_url != $icon_url) { | |
| 469 | db_query($link, "UPDATE ttrss_feeds SET icon_url = '$icon_url' WHERE id = '$feed'"); | |
| 470 | } | |
| 471 | ||
| 472 | if ($debug_enabled) { | |
| 473 | _debug("update_rss_feed: loading filters..."); | |
| 474 | } | |
| 475 | ||
| 476 | $filters = load_filters($link, $feed, $owner_uid); | |
| 477 | ||
| 478 | // if ($debug_enabled) { | |
| 479 | // print_r($filters); | |
| 480 | // } | |
| 481 | ||
| 482 | if ($use_simplepie) { | |
| 483 | $iterator = $rss->get_items(); | |
| 484 | } else { | |
| 485 | $iterator = $rss->items; | |
| 486 | if (!$iterator || !is_array($iterator)) $iterator = $rss->entries; | |
| 487 | if (!$iterator || !is_array($iterator)) $iterator = $rss; | |
| 488 | } | |
| 489 | ||
| 490 | if (!is_array($iterator)) { | |
| 491 | /* db_query($link, "UPDATE ttrss_feeds | |
| 492 | SET last_error = 'Parse error: can\'t find any articles.' | |
| 493 | WHERE id = '$feed'"); */ | |
| 494 | ||
| 495 | // clear any errors and mark feed as updated if fetched okay | |
| 496 | // even if it's blank | |
| 497 | ||
| 498 | if ($debug_enabled) { | |
| 499 | _debug("update_rss_feed: entry iterator is not an array, no articles?"); | |
| 500 | } | |
| 501 | ||
| 502 | db_query($link, "UPDATE ttrss_feeds | |
| 503 | SET last_updated = NOW(), last_error = '' WHERE id = '$feed'"); | |
| 504 | ||
| 505 | return; // no articles | |
| 506 | } | |
| 507 | ||
| 508 | if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) { | |
| 509 | ||
| 510 | if ($debug_enabled) _debug("update_rss_feed: checking for PUSH hub..."); | |
| 511 | ||
| 512 | $feed_hub_url = false; | |
| 513 | if ($use_simplepie) { | |
| 514 | $links = $rss->get_links('hub'); | |
| 515 | ||
| 516 | if ($links && is_array($links)) { | |
| 517 | foreach ($links as $l) { | |
| 518 | $feed_hub_url = $l; | |
| 519 | break; | |
| 520 | } | |
| 521 | } | |
| 522 | ||
| 523 | } else { | |
| 524 | $atom = $rss->channel['atom']; | |
| 525 | ||
| 526 | if ($atom) { | |
| 527 | if ($atom['link@rel'] == 'hub') { | |
| 528 | $feed_hub_url = $atom['link@href']; | |
| 529 | } | |
| 530 | ||
| 531 | if (!$feed_hub_url && $atom['link#'] > 1) { | |
| 532 | for ($i = 2; $i <= $atom['link#']; $i++) { | |
| 533 | if ($atom["link#$i@rel"] == 'hub') { | |
| 534 | $feed_hub_url = $atom["link#$i@href"]; | |
| 535 | break; | |
| 536 | } | |
| 537 | } | |
| 538 | } | |
| 539 | } else { | |
| 540 | $feed_hub_url = $rss->channel['link_hub']; | |
| 541 | } | |
| 542 | } | |
| 543 | ||
| 544 | if ($debug_enabled) _debug("update_rss_feed: feed hub url: $feed_hub_url"); | |
| 545 | ||
| 546 | if ($feed_hub_url && function_exists('curl_init') && | |
| 547 | !ini_get("open_basedir")) { | |
| 548 | ||
| 549 | require_once 'lib/pubsubhubbub/subscriber.php'; | |
| 550 | ||
| 551 | $callback_url = get_self_url_prefix() . | |
| 552 | "/public.php?op=pubsub&id=$feed"; | |
| 553 | ||
| 554 | $s = new Subscriber($feed_hub_url, $callback_url); | |
| 555 | ||
| 556 | $rc = $s->subscribe($fetch_url); | |
| 557 | ||
| 558 | if ($debug_enabled) | |
| 559 | _debug("update_rss_feed: feed hub url found, subscribe request sent."); | |
| 560 | ||
| 561 | db_query($link, "UPDATE ttrss_feeds SET pubsub_state = 1 | |
| 562 | WHERE id = '$feed'"); | |
| 563 | } | |
| 564 | } | |
| 565 | ||
| 566 | if ($debug_enabled) { | |
| 567 | _debug("update_rss_feed: processing articles..."); | |
| 568 | } | |
| 569 | ||
| 570 | foreach ($iterator as $item) { | |
| 571 | ||
| 572 | if ($_REQUEST['xdebug'] == 2) { | |
| 573 | print_r($item); | |
| 574 | } | |
| 575 | ||
| 576 | if ($use_simplepie) { | |
| 577 | $entry_guid = $item->get_id(); | |
| 578 | if (!$entry_guid) $entry_guid = $item->get_link(); | |
| 579 | if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title()); | |
| 580 | ||
| 581 | } else { | |
| 582 | ||
| 583 | $entry_guid = $item["id"]; | |
| 584 | ||
| 585 | if (!$entry_guid) $entry_guid = $item["guid"]; | |
| 586 | if (!$entry_guid) $entry_guid = $item["about"]; | |
| 587 | if (!$entry_guid) $entry_guid = $item["link"]; | |
| 588 | if (!$entry_guid) $entry_guid = make_guid_from_title($item["title"]); | |
| 589 | } | |
| 590 | ||
| 591 | if ($debug_enabled) { | |
| 592 | _debug("update_rss_feed: guid $entry_guid"); | |
| 593 | } | |
| 594 | ||
| 595 | if (!$entry_guid) continue; | |
| 596 | ||
| 597 | $entry_timestamp = ""; | |
| 598 | ||
| 599 | if ($use_simplepie) { | |
| 600 | $entry_timestamp = strtotime($item->get_date()); | |
| 601 | } else { | |
| 602 | $rss_2_date = $item['pubdate']; | |
| 603 | $rss_1_date = $item['dc']['date']; | |
| 604 | $atom_date = $item['issued']; | |
| 605 | if (!$atom_date) $atom_date = $item['updated']; | |
| 606 | ||
| 607 | if ($atom_date != "") $entry_timestamp = parse_w3cdtf($atom_date); | |
| 608 | if ($rss_1_date != "") $entry_timestamp = parse_w3cdtf($rss_1_date); | |
| 609 | if ($rss_2_date != "") $entry_timestamp = strtotime($rss_2_date); | |
| 610 | ||
| 611 | } | |
| 612 | ||
| 613 | if ($entry_timestamp == "" || $entry_timestamp == -1 || !$entry_timestamp) { | |
| 614 | $entry_timestamp = time(); | |
| 615 | $no_orig_date = 'true'; | |
| 616 | } else { | |
| 617 | $no_orig_date = 'false'; | |
| 618 | } | |
| 619 | ||
| 620 | $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp); | |
| 621 | ||
| 622 | if ($debug_enabled) { | |
| 623 | _debug("update_rss_feed: date $entry_timestamp [$entry_timestamp_fmt]"); | |
| 624 | } | |
| 625 | ||
| 626 | if ($use_simplepie) { | |
| 627 | $entry_title = $item->get_title(); | |
| 628 | } else { | |
| 629 | $entry_title = trim(strip_tags($item["title"])); | |
| 630 | } | |
| 631 | ||
| 632 | if ($use_simplepie) { | |
| 633 | $entry_link = $item->get_link(); | |
| 634 | } else { | |
| 635 | // strange Magpie workaround | |
| 636 | $entry_link = $item["link_"]; | |
| 637 | if (!$entry_link) $entry_link = $item["link"]; | |
| 638 | } | |
| 639 | ||
| 640 | $entry_link = rewrite_relative_url($site_url, $entry_link); | |
| 641 | ||
| 642 | if ($debug_enabled) { | |
| 643 | _debug("update_rss_feed: title $entry_title"); | |
| 644 | _debug("update_rss_feed: link $entry_link"); | |
| 645 | } | |
| 646 | ||
| 647 | if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);; | |
| 648 | ||
| 649 | $entry_link = strip_tags($entry_link); | |
| 650 | ||
| 651 | if ($use_simplepie) { | |
| 652 | $entry_content = $item->get_content(); | |
| 653 | if (!$entry_content) $entry_content = $item->get_description(); | |
| 654 | } else { | |
| 655 | $entry_content = $item["content:escaped"]; | |
| 656 | ||
| 657 | if (!$entry_content) $entry_content = $item["content:encoded"]; | |
| 658 | if (!$entry_content) $entry_content = $item["content"]["encoded"]; | |
| 659 | if (!$entry_content) $entry_content = $item["content"]; | |
| 660 | ||
| 661 | if (is_array($entry_content)) $entry_content = $entry_content[0]; | |
| 662 | ||
| 663 | // Magpie bugs are getting ridiculous | |
| 664 | if (trim($entry_content) == "Array") $entry_content = false; | |
| 665 | ||
| 666 | if (!$entry_content) $entry_content = $item["atom_content"]; | |
| 667 | if (!$entry_content) $entry_content = $item["summary"]; | |
| 668 | ||
| 669 | if (!$entry_content || | |
| 670 | strlen($entry_content) < strlen($item["description"])) { | |
| 671 | $entry_content = $item["description"]; | |
| 672 | }; | |
| 673 | ||
| 674 | // WTF | |
| 675 | if (is_array($entry_content)) { | |
| 676 | $entry_content = $entry_content["encoded"]; | |
| 677 | if (!$entry_content) $entry_content = $entry_content["escaped"]; | |
| 678 | } | |
| 679 | } | |
| 680 | ||
| c5867798 | 681 | if ($cache_images && is_writable(CACHE_DIR . '/images')) |
| 3c696512 AD |
682 | $entry_content = cache_images($entry_content, $site_url, $debug_enabled); |
| 683 | ||
| 2c08214a AD |
684 | if ($_REQUEST["xdebug"] == 2) { |
| 685 | print "update_rss_feed: content: "; | |
| 487f0750 | 686 | print $entry_content; |
| 3c696512 | 687 | print "\n"; |
| 2c08214a AD |
688 | } |
| 689 | ||
| 690 | $entry_content_unescaped = $entry_content; | |
| 691 | ||
| 692 | if ($use_simplepie) { | |
| 693 | $entry_comments = strip_tags($item->data["comments"]); | |
| 694 | if ($item->get_author()) { | |
| 695 | $entry_author_item = $item->get_author(); | |
| 696 | $entry_author = $entry_author_item->get_name(); | |
| 697 | if (!$entry_author) $entry_author = $entry_author_item->get_email(); | |
| 698 | ||
| 699 | $entry_author = db_escape_string($entry_author); | |
| 700 | } | |
| 701 | } else { | |
| 702 | $entry_comments = strip_tags($item["comments"]); | |
| 703 | ||
| 704 | $entry_author = db_escape_string(strip_tags($item['dc']['creator'])); | |
| 705 | ||
| 706 | if ($item['author']) { | |
| 707 | ||
| 708 | if (is_array($item['author'])) { | |
| 709 | ||
| 710 | if (!$entry_author) { | |
| 711 | $entry_author = db_escape_string(strip_tags($item['author']['name'])); | |
| 712 | } | |
| 713 | ||
| 714 | if (!$entry_author) { | |
| 715 | $entry_author = db_escape_string(strip_tags($item['author']['email'])); | |
| 716 | } | |
| 717 | } | |
| 718 | ||
| 719 | if (!$entry_author) { | |
| 720 | $entry_author = db_escape_string(strip_tags($item['author'])); | |
| 721 | } | |
| 722 | } | |
| 723 | } | |
| 724 | ||
| 725 | if (preg_match('/^[\t\n\r ]*$/', $entry_author)) $entry_author = ''; | |
| 726 | ||
| 727 | $entry_guid = db_escape_string(strip_tags($entry_guid)); | |
| 728 | $entry_guid = mb_substr($entry_guid, 0, 250); | |
| 729 | ||
| 730 | $result = db_query($link, "SELECT id FROM ttrss_entries | |
| 731 | WHERE guid = '$entry_guid'"); | |
| 732 | ||
| 733 | $entry_content = db_escape_string($entry_content, false); | |
| 734 | ||
| 487f0750 | 735 | $content_hash = "SHA1:" . sha1(strip_tags($entry_content)); |
| 2c08214a AD |
736 | |
| 737 | $entry_title = db_escape_string($entry_title); | |
| 738 | $entry_link = db_escape_string($entry_link); | |
| 739 | $entry_comments = mb_substr(db_escape_string($entry_comments), 0, 250); | |
| 740 | $entry_author = mb_substr($entry_author, 0, 250); | |
| 741 | ||
| 742 | if ($use_simplepie) { | |
| 743 | $num_comments = 0; #FIXME# | |
| 744 | } else { | |
| 745 | $num_comments = db_escape_string($item["slash"]["comments"]); | |
| 746 | } | |
| 747 | ||
| 748 | if (!$num_comments) $num_comments = 0; | |
| 749 | ||
| 750 | if ($debug_enabled) { | |
| 751 | _debug("update_rss_feed: looking for tags [1]..."); | |
| 752 | } | |
| 753 | ||
| 754 | // parse <category> entries into tags | |
| 755 | ||
| 756 | $additional_tags = array(); | |
| 757 | ||
| 758 | if ($use_simplepie) { | |
| 759 | ||
| 760 | $additional_tags_src = $item->get_categories(); | |
| 761 | ||
| 762 | if (is_array($additional_tags_src)) { | |
| 763 | foreach ($additional_tags_src as $tobj) { | |
| 764 | array_push($additional_tags, $tobj->get_term()); | |
| 765 | } | |
| 766 | } | |
| 767 | ||
| 768 | if ($debug_enabled) { | |
| 769 | _debug("update_rss_feed: category tags:"); | |
| 770 | print_r($additional_tags); | |
| 771 | } | |
| 772 | ||
| 773 | } else { | |
| 774 | ||
| 775 | $t_ctr = $item['category#']; | |
| 776 | ||
| 777 | if ($t_ctr == 0) { | |
| 778 | $additional_tags = array(); | |
| 779 | } else if ($t_ctr > 0) { | |
| 780 | $additional_tags = array($item['category']); | |
| 781 | ||
| 782 | if ($item['category@term']) { | |
| 783 | array_push($additional_tags, $item['category@term']); | |
| 784 | } | |
| 785 | ||
| 786 | for ($i = 0; $i <= $t_ctr; $i++ ) { | |
| 787 | if ($item["category#$i"]) { | |
| 788 | array_push($additional_tags, $item["category#$i"]); | |
| 789 | } | |
| 790 | ||
| 791 | if ($item["category#$i@term"]) { | |
| 792 | array_push($additional_tags, $item["category#$i@term"]); | |
| 793 | } | |
| 794 | } | |
| 795 | } | |
| 796 | ||
| 797 | // parse <dc:subject> elements | |
| 798 | ||
| 799 | $t_ctr = $item['dc']['subject#']; | |
| 800 | ||
| 801 | if ($t_ctr > 0) { | |
| 802 | array_push($additional_tags, $item['dc']['subject']); | |
| 803 | ||
| 804 | for ($i = 0; $i <= $t_ctr; $i++ ) { | |
| 805 | if ($item['dc']["subject#$i"]) { | |
| 806 | array_push($additional_tags, $item['dc']["subject#$i"]); | |
| 807 | } | |
| 808 | } | |
| 809 | } | |
| 810 | } | |
| 811 | ||
| 812 | if ($debug_enabled) { | |
| 813 | _debug("update_rss_feed: looking for tags [2]..."); | |
| 814 | } | |
| 815 | ||
| 816 | /* taaaags */ | |
| 817 | // <a href="..." rel="tag">Xorg</a>, // | |
| 818 | ||
| 819 | $entry_tags = null; | |
| 820 | ||
| 821 | preg_match_all("/<a.*?rel=['\"]tag['\"].*?\>([^<]+)<\/a>/i", | |
| 822 | $entry_content_unescaped, $entry_tags); | |
| 823 | ||
| 824 | $entry_tags = $entry_tags[1]; | |
| 825 | ||
| 826 | $entry_tags = array_merge($entry_tags, $additional_tags); | |
| 827 | $entry_tags = array_unique($entry_tags); | |
| 828 | ||
| 829 | for ($i = 0; $i < count($entry_tags); $i++) | |
| 830 | $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8'); | |
| 831 | ||
| 832 | if ($debug_enabled) { | |
| 833 | _debug("update_rss_feed: unfiltered tags found:"); | |
| 834 | print_r($entry_tags); | |
| 835 | } | |
| 836 | ||
| 837 | # sanitize content | |
| 838 | ||
| 839 | $entry_content = sanitize_article_content($entry_content); | |
| 840 | $entry_title = sanitize_article_content($entry_title); | |
| 841 | ||
| 842 | if ($debug_enabled) { | |
| 843 | _debug("update_rss_feed: done collecting data [TITLE:$entry_title]"); | |
| 844 | } | |
| 845 | ||
| 846 | db_query($link, "BEGIN"); | |
| 847 | ||
| 848 | if (db_num_rows($result) == 0) { | |
| 849 | ||
| 850 | if ($debug_enabled) { | |
| 851 | _debug("update_rss_feed: base guid not found"); | |
| 852 | } | |
| 853 | ||
| 854 | // base post entry does not exist, create it | |
| 855 | ||
| 856 | $result = db_query($link, | |
| 857 | "INSERT INTO ttrss_entries | |
| 858 | (title, | |
| 859 | guid, | |
| 860 | link, | |
| 861 | updated, | |
| 862 | content, | |
| 863 | content_hash, | |
| 864 | no_orig_date, | |
| 865 | date_updated, | |
| 866 | date_entered, | |
| 867 | comments, | |
| 868 | num_comments, | |
| 869 | author) | |
| 870 | VALUES | |
| 871 | ('$entry_title', | |
| 872 | '$entry_guid', | |
| 873 | '$entry_link', | |
| 874 | '$entry_timestamp_fmt', | |
| 875 | '$entry_content', | |
| 876 | '$content_hash', | |
| 877 | $no_orig_date, | |
| 878 | NOW(), | |
| 879 | NOW(), | |
| 880 | '$entry_comments', | |
| 881 | '$num_comments', | |
| 882 | '$entry_author')"); | |
| 883 | } else { | |
| 884 | // we keep encountering the entry in feeds, so we need to | |
| 885 | // update date_updated column so that we don't get horrible | |
| 886 | // dupes when the entry gets purged and reinserted again e.g. | |
| 887 | // in the case of SLOW SLOW OMG SLOW updating feeds | |
| 888 | ||
| 889 | $base_entry_id = db_fetch_result($result, 0, "id"); | |
| 890 | ||
| 891 | db_query($link, "UPDATE ttrss_entries SET date_updated = NOW() | |
| 892 | WHERE id = '$base_entry_id'"); | |
| 893 | } | |
| 894 | ||
| 895 | // now it should exist, if not - bad luck then | |
| 896 | ||
| 897 | $result = db_query($link, "SELECT | |
| 898 | id,content_hash,no_orig_date,title, | |
| 899 | ".SUBSTRING_FOR_DATE."(date_updated,1,19) as date_updated, | |
| 900 | ".SUBSTRING_FOR_DATE."(updated,1,19) as updated, | |
| 901 | num_comments | |
| 902 | FROM | |
| 903 | ttrss_entries | |
| 904 | WHERE guid = '$entry_guid'"); | |
| 905 | ||
| 906 | $entry_ref_id = 0; | |
| 907 | $entry_int_id = 0; | |
| 908 | ||
| 909 | if (db_num_rows($result) == 1) { | |
| 910 | ||
| 911 | if ($debug_enabled) { | |
| 912 | _debug("update_rss_feed: base guid found, checking for user record"); | |
| 913 | } | |
| 914 | ||
| 915 | // this will be used below in update handler | |
| 916 | $orig_content_hash = db_fetch_result($result, 0, "content_hash"); | |
| 917 | $orig_title = db_fetch_result($result, 0, "title"); | |
| 918 | $orig_num_comments = db_fetch_result($result, 0, "num_comments"); | |
| 919 | $orig_date_updated = strtotime(db_fetch_result($result, | |
| 920 | 0, "date_updated")); | |
| 921 | ||
| 922 | $ref_id = db_fetch_result($result, 0, "id"); | |
| 923 | $entry_ref_id = $ref_id; | |
| 924 | ||
| 925 | // check for user post link to main table | |
| 926 | ||
| 927 | // do we allow duplicate posts with same GUID in different feeds? | |
| 928 | if (get_pref($link, "ALLOW_DUPLICATE_POSTS", $owner_uid, false)) { | |
| 929 | $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)"; | |
| 930 | } else { | |
| 931 | $dupcheck_qpart = ""; | |
| 932 | } | |
| 933 | ||
| 934 | /* Collect article tags here so we could filter by them: */ | |
| 935 | ||
| 936 | $article_filters = get_article_filters($filters, $entry_title, | |
| 937 | $entry_content, $entry_link, $entry_timestamp, $entry_author, | |
| 938 | $entry_tags); | |
| 939 | ||
| 940 | if ($debug_enabled) { | |
| 941 | _debug("update_rss_feed: article filters: "); | |
| 942 | if (count($article_filters) != 0) { | |
| 943 | print_r($article_filters); | |
| 944 | } | |
| 945 | } | |
| 946 | ||
| 947 | if (find_article_filter($article_filters, "filter")) { | |
| 948 | db_query($link, "COMMIT"); // close transaction in progress | |
| 949 | continue; | |
| 950 | } | |
| 951 | ||
| 952 | $score = calculate_article_score($article_filters); | |
| 953 | ||
| 954 | if ($debug_enabled) { | |
| 955 | _debug("update_rss_feed: initial score: $score"); | |
| 956 | } | |
| 957 | ||
| 958 | $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE | |
| 959 | ref_id = '$ref_id' AND owner_uid = '$owner_uid' | |
| 960 | $dupcheck_qpart"; | |
| 961 | ||
| 962 | // if ($_REQUEST["xdebug"]) print "$query\n"; | |
| 963 | ||
| 964 | $result = db_query($link, $query); | |
| 965 | ||
| 966 | // okay it doesn't exist - create user entry | |
| 967 | if (db_num_rows($result) == 0) { | |
| 968 | ||
| 969 | if ($debug_enabled) { | |
| 970 | _debug("update_rss_feed: user record not found, creating..."); | |
| 971 | } | |
| 972 | ||
| 973 | if ($score >= -500 && !find_article_filter($article_filters, 'catchup')) { | |
| 974 | $unread = 'true'; | |
| 975 | $last_read_qpart = 'NULL'; | |
| 976 | } else { | |
| 977 | $unread = 'false'; | |
| 978 | $last_read_qpart = 'NOW()'; | |
| 979 | } | |
| 980 | ||
| 981 | if (find_article_filter($article_filters, 'mark') || $score > 1000) { | |
| 982 | $marked = 'true'; | |
| 983 | } else { | |
| 984 | $marked = 'false'; | |
| 985 | } | |
| 986 | ||
| 987 | if (find_article_filter($article_filters, 'publish')) { | |
| 988 | $published = 'true'; | |
| 989 | } else { | |
| 990 | $published = 'false'; | |
| 991 | } | |
| 992 | ||
| 993 | $result = db_query($link, | |
| 994 | "INSERT INTO ttrss_user_entries | |
| 995 | (ref_id, owner_uid, feed_id, unread, last_read, marked, | |
| 996 | published, score, tag_cache, label_cache, uuid) | |
| 997 | VALUES ('$ref_id', '$owner_uid', '$feed', $unread, | |
| 998 | $last_read_qpart, $marked, $published, '$score', '', '', '')"); | |
| 999 | ||
| 1000 | if (PUBSUBHUBBUB_HUB && $published == 'true') { | |
| 1001 | $rss_link = get_self_url_prefix() . | |
| 1002 | "/public.php?op=rss&id=-2&key=" . | |
| 1003 | get_feed_access_key($link, -2, false, $owner_uid); | |
| 1004 | ||
| 1005 | $p = new Publisher(PUBSUBHUBBUB_HUB); | |
| 1006 | ||
| 1007 | $pubsub_result = $p->publish_update($rss_link); | |
| 1008 | } | |
| 1009 | ||
| 1010 | $result = db_query($link, | |
| 1011 | "SELECT int_id FROM ttrss_user_entries WHERE | |
| 1012 | ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND | |
| 1013 | feed_id = '$feed' LIMIT 1"); | |
| 1014 | ||
| 1015 | if (db_num_rows($result) == 1) { | |
| 1016 | $entry_int_id = db_fetch_result($result, 0, "int_id"); | |
| 1017 | } | |
| 1018 | } else { | |
| 1019 | if ($debug_enabled) { | |
| 1020 | _debug("update_rss_feed: user record FOUND"); | |
| 1021 | } | |
| 1022 | ||
| 1023 | $entry_ref_id = db_fetch_result($result, 0, "ref_id"); | |
| 1024 | $entry_int_id = db_fetch_result($result, 0, "int_id"); | |
| 1025 | } | |
| 1026 | ||
| 1027 | if ($debug_enabled) { | |
| 1028 | _debug("update_rss_feed: RID: $entry_ref_id, IID: $entry_int_id"); | |
| 1029 | } | |
| 1030 | ||
| 1031 | $post_needs_update = false; | |
| 1032 | $update_insignificant = false; | |
| 1033 | ||
| 1034 | if ($orig_num_comments != $num_comments) { | |
| 1035 | $post_needs_update = true; | |
| 1036 | $update_insignificant = true; | |
| 1037 | } | |
| 1038 | ||
| 1039 | if ($content_hash != $orig_content_hash) { | |
| 1040 | $post_needs_update = true; | |
| 1041 | $update_insignificant = false; | |
| 1042 | } | |
| 1043 | ||
| 1044 | if (db_escape_string($orig_title) != $entry_title) { | |
| 1045 | $post_needs_update = true; | |
| 1046 | $update_insignificant = false; | |
| 1047 | } | |
| 1048 | ||
| 1049 | // if post needs update, update it and mark all user entries | |
| 1050 | // linking to this post as updated | |
| 1051 | if ($post_needs_update) { | |
| 1052 | ||
| 1053 | if (defined('DAEMON_EXTENDED_DEBUG')) { | |
| 1054 | _debug("update_rss_feed: post $entry_guid needs update..."); | |
| 1055 | } | |
| 1056 | ||
| 1057 | // print "<!-- post $orig_title needs update : $post_needs_update -->"; | |
| 1058 | ||
| 1059 | db_query($link, "UPDATE ttrss_entries | |
| 1060 | SET title = '$entry_title', content = '$entry_content', | |
| 1061 | content_hash = '$content_hash', | |
| 1062 | updated = '$entry_timestamp_fmt', | |
| 1063 | num_comments = '$num_comments' | |
| 1064 | WHERE id = '$ref_id'"); | |
| 1065 | ||
| 1066 | if (!$update_insignificant) { | |
| 1067 | if ($mark_unread_on_update) { | |
| 1068 | db_query($link, "UPDATE ttrss_user_entries | |
| 1069 | SET last_read = null, unread = true WHERE ref_id = '$ref_id'"); | |
| 1070 | } else if ($update_on_checksum_change) { | |
| 1071 | db_query($link, "UPDATE ttrss_user_entries | |
| 1072 | SET last_read = null WHERE ref_id = '$ref_id' | |
| 1073 | AND unread = false"); | |
| 1074 | } | |
| 1075 | } | |
| 1076 | } | |
| 1077 | } | |
| 1078 | ||
| 1079 | db_query($link, "COMMIT"); | |
| 1080 | ||
| 1081 | if ($debug_enabled) { | |
| 1082 | _debug("update_rss_feed: assigning labels..."); | |
| 1083 | } | |
| 1084 | ||
| 1085 | assign_article_to_labels($link, $entry_ref_id, $article_filters, | |
| 1086 | $owner_uid); | |
| 1087 | ||
| 1088 | if ($debug_enabled) { | |
| 1089 | _debug("update_rss_feed: looking for enclosures..."); | |
| 1090 | } | |
| 1091 | ||
| 1092 | // enclosures | |
| 1093 | ||
| 1094 | $enclosures = array(); | |
| 1095 | ||
| 1096 | if ($use_simplepie) { | |
| 1097 | $encs = $item->get_enclosures(); | |
| 1098 | ||
| 1099 | if (is_array($encs)) { | |
| 1100 | foreach ($encs as $e) { | |
| 1101 | $e_item = array( | |
| 1102 | $e->link, $e->type, $e->length); | |
| 1103 | ||
| 1104 | array_push($enclosures, $e_item); | |
| 1105 | } | |
| 1106 | } | |
| 1107 | ||
| 1108 | } else { | |
| 1109 | // <enclosure> | |
| 1110 | ||
| 1111 | $e_ctr = $item['enclosure#']; | |
| 1112 | ||
| 1113 | if ($e_ctr > 0) { | |
| 1114 | $e_item = array($item['enclosure@url'], | |
| 1115 | $item['enclosure@type'], | |
| 1116 | $item['enclosure@length']); | |
| 1117 | ||
| 1118 | array_push($enclosures, $e_item); | |
| 1119 | ||
| 1120 | for ($i = 0; $i <= $e_ctr; $i++ ) { | |
| 1121 | ||
| 1122 | if ($item["enclosure#$i@url"]) { | |
| 1123 | $e_item = array($item["enclosure#$i@url"], | |
| 1124 | $item["enclosure#$i@type"], | |
| 1125 | $item["enclosure#$i@length"]); | |
| 1126 | array_push($enclosures, $e_item); | |
| 1127 | } | |
| 1128 | } | |
| 1129 | } | |
| 1130 | ||
| 1131 | // <media:content> | |
| 1132 | // can there be many of those? yes -fox | |
| 1133 | ||
| 1134 | $m_ctr = $item['media']['content#']; | |
| 1135 | ||
| 1136 | if ($m_ctr > 0) { | |
| 1137 | $e_item = array($item['media']['content@url'], | |
| 1138 | $item['media']['content@medium'], | |
| 1139 | $item['media']['content@length']); | |
| 1140 | ||
| 1141 | array_push($enclosures, $e_item); | |
| 1142 | ||
| 1143 | for ($i = 0; $i <= $m_ctr; $i++ ) { | |
| 1144 | ||
| 1145 | if ($item["media"]["content#$i@url"]) { | |
| 1146 | $e_item = array($item["media"]["content#$i@url"], | |
| 1147 | $item["media"]["content#$i@medium"], | |
| 1148 | $item["media"]["content#$i@length"]); | |
| 1149 | array_push($enclosures, $e_item); | |
| 1150 | } | |
| 1151 | } | |
| 1152 | ||
| 1153 | } | |
| 1154 | } | |
| 1155 | ||
| 1156 | ||
| 1157 | if ($debug_enabled) { | |
| 1158 | _debug("update_rss_feed: article enclosures:"); | |
| 1159 | print_r($enclosures); | |
| 1160 | } | |
| 1161 | ||
| 1162 | db_query($link, "BEGIN"); | |
| 1163 | ||
| 1164 | foreach ($enclosures as $enc) { | |
| 1165 | $enc_url = db_escape_string($enc[0]); | |
| 1166 | $enc_type = db_escape_string($enc[1]); | |
| 1167 | $enc_dur = db_escape_string($enc[2]); | |
| 1168 | ||
| 1169 | $result = db_query($link, "SELECT id FROM ttrss_enclosures | |
| 1170 | WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'"); | |
| 1171 | ||
| 1172 | if (db_num_rows($result) == 0) { | |
| 1173 | db_query($link, "INSERT INTO ttrss_enclosures | |
| 1174 | (content_url, content_type, title, duration, post_id) VALUES | |
| 1175 | ('$enc_url', '$enc_type', '', '$enc_dur', '$entry_ref_id')"); | |
| 1176 | } | |
| 1177 | } | |
| 1178 | ||
| 1179 | db_query($link, "COMMIT"); | |
| 1180 | ||
| 1181 | // check for manual tags (we have to do it here since they're loaded from filters) | |
| 1182 | ||
| 1183 | foreach ($article_filters as $f) { | |
| 1184 | if ($f[0] == "tag") { | |
| 1185 | ||
| 1186 | $manual_tags = trim_array(explode(",", $f[1])); | |
| 1187 | ||
| 1188 | foreach ($manual_tags as $tag) { | |
| 1189 | if (tag_is_valid($tag)) { | |
| 1190 | array_push($entry_tags, $tag); | |
| 1191 | } | |
| 1192 | } | |
| 1193 | } | |
| 1194 | } | |
| 1195 | ||
| 1196 | // Skip boring tags | |
| 1197 | ||
| 1198 | $boring_tags = trim_array(explode(",", mb_strtolower(get_pref($link, | |
| 1199 | 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8'))); | |
| 1200 | ||
| 1201 | $filtered_tags = array(); | |
| 1202 | $tags_to_cache = array(); | |
| 1203 | ||
| 1204 | if ($entry_tags && is_array($entry_tags)) { | |
| 1205 | foreach ($entry_tags as $tag) { | |
| 1206 | if (array_search($tag, $boring_tags) === false) { | |
| 1207 | array_push($filtered_tags, $tag); | |
| 1208 | } | |
| 1209 | } | |
| 1210 | } | |
| 1211 | ||
| 1212 | $filtered_tags = array_unique($filtered_tags); | |
| 1213 | ||
| 1214 | if ($debug_enabled) { | |
| 1215 | _debug("update_rss_feed: filtered article tags:"); | |
| 1216 | print_r($filtered_tags); | |
| 1217 | } | |
| 1218 | ||
| 1219 | // Save article tags in the database | |
| 1220 | ||
| 1221 | if (count($filtered_tags) > 0) { | |
| 1222 | ||
| 1223 | db_query($link, "BEGIN"); | |
| 1224 | ||
| 1225 | foreach ($filtered_tags as $tag) { | |
| 1226 | ||
| 1227 | $tag = sanitize_tag($tag); | |
| 1228 | $tag = db_escape_string($tag); | |
| 1229 | ||
| 1230 | if (!tag_is_valid($tag)) continue; | |
| 1231 | ||
| 1232 | $result = db_query($link, "SELECT id FROM ttrss_tags | |
| 1233 | WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND | |
| 1234 | owner_uid = '$owner_uid' LIMIT 1"); | |
| 1235 | ||
| 1236 | if ($result && db_num_rows($result) == 0) { | |
| 1237 | ||
| 1238 | db_query($link, "INSERT INTO ttrss_tags | |
| 1239 | (owner_uid,tag_name,post_int_id) | |
| 1240 | VALUES ('$owner_uid','$tag', '$entry_int_id')"); | |
| 1241 | } | |
| 1242 | ||
| 1243 | array_push($tags_to_cache, $tag); | |
| 1244 | } | |
| 1245 | ||
| 1246 | /* update the cache */ | |
| 1247 | ||
| 1248 | $tags_to_cache = array_unique($tags_to_cache); | |
| 1249 | ||
| 1250 | $tags_str = db_escape_string(join(",", $tags_to_cache)); | |
| 1251 | ||
| 1252 | db_query($link, "UPDATE ttrss_user_entries | |
| 1253 | SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id' | |
| 1254 | AND owner_uid = $owner_uid"); | |
| 1255 | ||
| 1256 | db_query($link, "COMMIT"); | |
| 1257 | } | |
| 1258 | ||
| 1259 | if ($debug_enabled) { | |
| 1260 | _debug("update_rss_feed: article processed"); | |
| 1261 | } | |
| 1262 | } | |
| 1263 | ||
| 1264 | if (!$last_updated) { | |
| 1265 | if ($debug_enabled) { | |
| 1266 | _debug("update_rss_feed: new feed, catching it up..."); | |
| 1267 | } | |
| 1268 | catchup_feed($link, $feed, false, $owner_uid); | |
| 1269 | } | |
| 1270 | ||
| 1271 | if ($debug_enabled) { | |
| 1272 | _debug("purging feed..."); | |
| 1273 | } | |
| 1274 | ||
| 1275 | purge_feed($link, $feed, 0, $debug_enabled); | |
| 1276 | ||
| 1277 | db_query($link, "UPDATE ttrss_feeds | |
| 1278 | SET last_updated = NOW(), last_error = '' WHERE id = '$feed'"); | |
| 1279 | ||
| 1280 | // db_query($link, "COMMIT"); | |
| 1281 | ||
| 1282 | } else { | |
| 1283 | ||
| 1284 | if ($use_simplepie) { | |
| 1285 | $error_msg = mb_substr($rss->error(), 0, 250); | |
| 1286 | } else { | |
| 1287 | $error_msg = mb_substr(magpie_error(), 0, 250); | |
| 1288 | } | |
| 1289 | ||
| 1290 | if ($debug_enabled) { | |
| 1291 | _debug("update_rss_feed: error fetching feed: $error_msg"); | |
| 1292 | } | |
| 1293 | ||
| 1294 | $error_msg = db_escape_string($error_msg); | |
| 1295 | ||
| 1296 | db_query($link, | |
| 1297 | "UPDATE ttrss_feeds SET last_error = '$error_msg', | |
| 1298 | last_updated = NOW() WHERE id = '$feed'"); | |
| 1299 | } | |
| 1300 | ||
| 1301 | if ($use_simplepie) { | |
| 1302 | unset($rss); | |
| 1303 | } | |
| 1304 | ||
| 1305 | if ($debug_enabled) { | |
| 1306 | _debug("update_rss_feed: done"); | |
| 1307 | } | |
| 1308 | ||
| 1309 | } | |
| 1310 | ||
| 3c696512 AD |
1311 | function cache_images($html, $site_url, $debug) { |
| 1312 | $cache_dir = CACHE_DIR . "/images"; | |
| 1313 | ||
| 1314 | libxml_use_internal_errors(true); | |
| 1315 | ||
| 1316 | $charset_hack = '<head> | |
| 1317 | <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/> | |
| 1318 | </head>'; | |
| 1319 | ||
| 1320 | $doc = new DOMDocument(); | |
| 1321 | $doc->loadHTML($charset_hack . $html); | |
| 1322 | $xpath = new DOMXPath($doc); | |
| 1323 | ||
| 1324 | $entries = $xpath->query('(//img[@src])'); | |
| 1325 | ||
| 1326 | foreach ($entries as $entry) { | |
| 1327 | if ($entry->hasAttribute('src')) { | |
| 1328 | $src = rewrite_relative_url($site_url, $entry->getAttribute('src')); | |
| 1329 | ||
| 1330 | $local_filename = CACHE_DIR . "/images/" . sha1($src) . ".png"; | |
| 1331 | ||
| 1332 | if ($debug) _debug("cache_images: downloading: $src to $local_filename"); | |
| 1333 | ||
| 1334 | if (!file_exists($local_filename)) { | |
| 1335 | $file_content = fetch_file_contents($src); | |
| 1336 | ||
| 1337 | if ($file_content) { | |
| 1338 | file_put_contents($local_filename, $file_content); | |
| 1339 | } | |
| 1340 | } | |
| 1341 | ||
| 1342 | if (file_exists($local_filename)) { | |
| 1343 | $entry->setAttribute('src', SELF_URL_PATH . '/image.php?url=' . | |
| 487f0750 | 1344 | base64_encode($src)); |
| 3c696512 AD |
1345 | } |
| 1346 | } | |
| 1347 | } | |
| 1348 | ||
| 1349 | $node = $doc->getElementsByTagName('body')->item(0); | |
| 1350 | ||
| 1351 | return $doc->saveXML($node); | |
| 1352 | } | |
| 1353 | ||
| 1354 | function expire_cached_files($debug) { | |
| 1355 | foreach (array("magpie", "simplepie", "images") as $dir) { | |
| 1356 | $cache_dir = CACHE_DIR . "/$dir"; | |
| 2c08214a | 1357 | |
| 3c696512 | 1358 | if ($debug) _debug("Expiring $cache_dir"); |
| 2c08214a | 1359 | |
| 3c696512 AD |
1360 | $num_deleted = 0; |
| 1361 | ||
| 1362 | if (is_writable($cache_dir)) { | |
| 1363 | $files = glob("$cache_dir/*"); | |
| 1364 | ||
| 1365 | foreach ($files as $file) { | |
| 1366 | if (time() - filemtime($file) > 86400*7) { | |
| 1367 | unlink($file); | |
| 1368 | ||
| 1369 | ++$num_deleted; | |
| 1370 | } | |
| 1371 | } | |
| 1372 | } | |
| 1373 | ||
| 1374 | if ($debug) _debug("Removed $num_deleted files."); | |
| 1375 | } | |
| 1376 | } | |
| 2c08214a AD |
1377 | |
| 1378 | ?> |