define_default('DAEMON_UPDATE_LOGIN_LIMIT', 30);
define_default('DAEMON_FEED_LIMIT', 500);
define_default('DAEMON_SLEEP_INTERVAL', 120);
+ define_default('_MIN_CACHE_IMAGE_SIZE', 1024);
function update_feedbrowser_cache() {
$login_thresh_qpart = "";
}
- // Test if the feed need a update (update interval exceded).
+ // Test if the feed need a update (update interval exceeded).
if (DB_TYPE == "pgsql") {
$update_limit_qpart = "AND ((
ttrss_feeds.update_interval = 0
) OR (
ttrss_feeds.update_interval > 0
AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
- ) OR ttrss_feeds.last_updated IS NULL
- OR last_updated = '1970-01-01 00:00:00')";
+ ) OR (ttrss_feeds.last_updated IS NULL
+ AND ttrss_user_prefs.value != '-1')
+ OR (last_updated = '1970-01-01 00:00:00'
+ AND ttrss_user_prefs.value != '-1'))";
} else {
$update_limit_qpart = "AND ((
ttrss_feeds.update_interval = 0
) OR (
ttrss_feeds.update_interval > 0
AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
- ) OR ttrss_feeds.last_updated IS NULL
- OR last_updated = '1970-01-01 00:00:00')";
+ ) OR (ttrss_feeds.last_updated IS NULL
+ AND ttrss_user_prefs.value != '-1')
+ OR (last_updated = '1970-01-01 00:00:00'
+ AND ttrss_user_prefs.value != '-1'))";
}
// Test if feed is currently being updated by another process.
ttrss_feeds, ttrss_users, ttrss_user_prefs
WHERE
ttrss_feeds.owner_uid = ttrss_users.id
+ AND ttrss_user_prefs.profile IS NULL
AND ttrss_users.id = ttrss_user_prefs.owner_uid
AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
$login_thresh_qpart $update_limit_qpart
// since we have the data cached, we can deal with other feeds with the same url
- $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated
+ $tmp_result = db_query("SELECT DISTINCT ttrss_feeds.id,last_updated,ttrss_feeds.owner_uid
FROM ttrss_feeds, ttrss_users, ttrss_user_prefs WHERE
ttrss_user_prefs.owner_uid = ttrss_feeds.owner_uid AND
ttrss_users.id = ttrss_user_prefs.owner_uid AND
ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL' AND
+ ttrss_user_prefs.profile IS NULL AND
feed_url = '".db_escape_string($feed)."' AND
(ttrss_feeds.update_interval > 0 OR
ttrss_user_prefs.value != '-1')
ORDER BY ttrss_feeds.id $query_limit");
if (db_num_rows($tmp_result) > 0) {
+ $rss = false;
+
while ($tline = db_fetch_assoc($tmp_result)) {
- if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"]);
- update_rss_feed($tline["id"], true);
+ if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"] . " " . $tline["owner_uid"]);
+
+ $rss = update_rss_feed($tline["id"], true, false);
+ _debug_suppress(false);
++$nf;
}
}
} // function update_daemon_common
// ignore_daemon is not used
- function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false,
- $override_url = false) {
+ function update_rss_feed($feed, $ignore_daemon = false, $no_cache = false, $rss = false) {
$debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
+ _debug_suppress(!$debug_enabled);
_debug("start", $debug_enabled);
$result = db_query("SELECT id,update_interval,auth_login,
- feed_url,auth_pass,cache_images,last_updated,
+ feed_url,auth_pass,cache_images,
mark_unread_on_update, owner_uid,
pubsub_state, auth_pass_encrypted,
(SELECT max(date_entered) FROM
return false;
}
- // set last update to now so if anything *simplepie* crashes later we won't be
- // continuously failing on the same feed
- db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
-
- $last_updated = db_fetch_result($result, 0, "last_updated");
$last_article_timestamp = @strtotime(db_fetch_result($result, 0, "last_article_timestamp"));
+
+ if (defined('_DISABLE_HTTP_304'))
+ $last_article_timestamp = 0;
+
$owner_uid = db_fetch_result($result, 0, "owner_uid");
$mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
0, "mark_unread_on_update"));
$feed = db_escape_string($feed);
- if ($override_url) $fetch_url = $override_url;
-
$date_feed_processed = date('Y-m-d H:i');
$cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
- // Ignore cache if new feed or manual update.
- $cache_age = ($no_cache || is_null($last_updated) || strpos($last_updated, '1970-01-01') === 0) ? 30 : get_feed_update_interval($feed) * 60;
+ $pluginhost = new PluginHost();
+ $pluginhost->set_debug($debug_enabled);
+ $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
- _debug("cache filename: $cache_filename exists: " . file_exists($cache_filename), $debug_enabled);
- _debug("cache age: $cache_age; no cache: $no_cache", $debug_enabled);
+ $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
+ $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
+ $pluginhost->load_data();
- $cached_feed_data_hash = false;
+ if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
+ _debug("using previously initialized parser object");
+ } else {
+ $rss_hash = false;
- $rss = false;
- $rss_hash = false;
- $cache_timestamp = file_exists($cache_filename) ? filemtime($cache_filename) : 0;
+ $force_refetch = isset($_REQUEST["force_refetch"]);
- $force_refetch = isset($_REQUEST["force_refetch"]);
+ foreach ($pluginhost->get_hooks(PluginHost::HOOK_FETCH_FEED) as $plugin) {
+ $feed_data = $plugin->hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, $last_article_timestamp, $auth_login, $auth_pass);
+ }
- if (file_exists($cache_filename) &&
- is_readable($cache_filename) &&
- !$auth_login && !$auth_pass &&
- filemtime($cache_filename) > time() - $cache_age) {
+ // try cache
+ if (!$feed_data &&
+ file_exists($cache_filename) &&
+ is_readable($cache_filename) &&
+ !$auth_login && !$auth_pass &&
+ filemtime($cache_filename) > time() - 30) {
- _debug("using local cache.", $debug_enabled);
+ _debug("using local cache [$cache_filename].", $debug_enabled);
- if ($cache_timestamp > $last_article_timestamp) {
- @$feed_data = file_get_contents($cache_filename);
+ @$feed_data = file_get_contents($cache_filename);
- if ($feed_data) {
- $rss_hash = sha1($feed_data);
- }
- } else if (!$force_refetch) {
- _debug("local cache valid and older than last_updated, nothing to do.", $debug_enabled);
- return;
+ if ($feed_data) {
+ $rss_hash = sha1($feed_data);
}
- } else {
- _debug("local cache will not be used for this feed", $debug_enabled);
- }
- if (!$rss) {
+ } else {
+ _debug("local cache will not be used for this feed", $debug_enabled);
+ }
+ // fetch feed from source
if (!$feed_data) {
_debug("fetching [$fetch_url]...", $debug_enabled);
_debug("If-Modified-Since: ".gmdate('D, d M Y H:i:s \G\M\T', $last_article_timestamp), $debug_enabled);
_debug("fetch done.", $debug_enabled);
- if ($feed_data) {
+ /* if ($feed_data) {
$error = verify_feed_xml($feed_data);
if ($error) {
if ($error) $feed_data = '';
}
}
+ } */
+
+ // cache vanilla feed data for re-use
+ if ($feed_data && !$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
+ $new_rss_hash = sha1($feed_data);
+
+ if ($new_rss_hash != $rss_hash) {
+ _debug("saving $cache_filename", $debug_enabled);
+ @file_put_contents($cache_filename, $feed_data);
+ }
}
}
}
}
- $pluginhost = new PluginHost();
- $pluginhost->set_debug($debug_enabled, $debug_enabled);
- $user_plugins = get_pref("_ENABLED_PLUGINS", $owner_uid);
-
- $pluginhost->load(PLUGINS, PluginHost::KIND_ALL);
- $pluginhost->load($user_plugins, PluginHost::KIND_USER, $owner_uid);
- $pluginhost->load_data();
-
foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
- $feed_data = $plugin->hook_feed_fetched($feed_data);
+ $feed_data = $plugin->hook_feed_fetched($feed_data, $fetch_url, $owner_uid, $feed);
}
+ // set last update to now so if anything *simplepie* crashes later we won't be
+ // continuously failing on the same feed
+ //db_query("UPDATE ttrss_feeds SET last_updated = NOW() WHERE id = '$feed'");
+
if (!$rss) {
$rss = new FeedParser($feed_data);
$rss->init();
}
+ if (DETECT_ARTICLE_LANGUAGE) {
+ require_once "lib/languagedetect/LanguageDetect.php";
+
+ $lang = new Text_LanguageDetect();
+ $lang->setNameMode(2);
+ }
+
// print_r($rss);
$feed = db_escape_string($feed);
if (!$rss->error()) {
- // cache data for later
- if (!$auth_pass && !$auth_login && is_writable(CACHE_DIR . "/simplepie")) {
- $new_rss_hash = sha1($rss_data);
-
- if ($new_rss_hash != $rss_hash && count($rss->get_items()) > 0 ) {
- _debug("saving $cache_filename", $debug_enabled);
- @file_put_contents($cache_filename, $feed_data);
- }
- }
-
// We use local pluginhost here because we need to load different per-user feed plugins
$pluginhost->run_hooks(PluginHost::HOOK_FEED_PARSED, "hook_feed_parsed", $rss);
_debug("checking favicon...", $debug_enabled);
- check_feed_favicon($site_url, $feed, $link);
+ check_feed_favicon($site_url, $feed);
$favicon_modified_new = @filemtime($favicon_file);
if ($favicon_modified_new > $favicon_modified)
_debug("feed hub url: $feed_hub_url", $debug_enabled);
- if ($feed_hub_url && function_exists('curl_init') &&
+ $feed_self_url = $fetch_url;
+
+ $links = $rss->get_links('self');
+
+ if ($links && is_array($links)) {
+ foreach ($links as $l) {
+ $feed_self_url = $l;
+ break;
+ }
+ }
+
+ _debug("feed self url = $feed_self_url");
+
+ if ($feed_hub_url && $feed_self_url && function_exists('curl_init') &&
!ini_get("open_basedir")) {
require_once 'lib/pubsubhubbub/subscriber.php';
$s = new Subscriber($feed_hub_url, $callback_url);
- $rc = $s->subscribe($fetch_url);
+ $rc = $s->subscribe($feed_self_url);
- _debug("feed hub url found, subscribe request sent.", $debug_enabled);
+ _debug("feed hub url found, subscribe request sent. [rc=$rc]", $debug_enabled);
db_query("UPDATE ttrss_feeds SET pubsub_state = 1
WHERE id = '$feed'");
if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
$entry_timestamp = time();
- $no_orig_date = 'true';
- } else {
- $no_orig_date = 'false';
}
$entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
print "\n";
}
+ $entry_language = "";
+
+ if (DETECT_ARTICLE_LANGUAGE) {
+ $entry_language = $lang->detect($entry_title . " " . $entry_content, 1);
+
+ if (count($entry_language) > 0) {
+ $entry_language = array_keys($entry_language);
+
+ // the fuck?
+ if (is_array($entry_language))
+ $entry_language = "";
+ else
+ $entry_language = db_escape_string(substr($entry_language[0], 0, 2));
+
+ _debug("detected language: $entry_language", $debug_enabled);
+ } else {
+ $entry_language = "";
+ }
+ }
+
$entry_comments = $item->get_comments_url();
$entry_author = $item->get_author();
"tags" => $entry_tags,
"plugin_data" => $entry_plugin_data,
"author" => $entry_author,
- "stored" => $stored_article);
+ "stored" => $stored_article,
+ "feed" => array("id" => $feed,
+ "fetch_url" => $fetch_url,
+ "site_url" => $site_url)
+ );
foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
$article = $plugin->hook_article_filter($article);
+
+ $article["stored"] = array("title" => $article["title"],
+ "content" => $article["content"],
+ "link" => $article["link"],
+ "tags" => $article["tags"],
+ "author" => $article["author"]);
}
$entry_tags = $article["tags"];
$entry_plugin_data = db_escape_string($article["plugin_data"]);
$entry_content = $article["content"]; // escaped below
-
_debug("plugin data: $entry_plugin_data", $debug_enabled);
if ($cache_images && is_writable(CACHE_DIR . '/images'))
updated,
content,
content_hash,
- cached_content,
no_orig_date,
date_updated,
date_entered,
comments,
num_comments,
plugin_data,
+ lang,
author)
VALUES
('$entry_title',
'$entry_timestamp_fmt',
'$entry_content',
'$content_hash',
- '',
- $no_orig_date,
+ false,
NOW(),
'$date_feed_processed',
'$entry_comments',
'$num_comments',
'$entry_plugin_data',
+ '$entry_language',
'$entry_author')");
$article_labels = array();
// now it should exist, if not - bad luck then
$result = db_query("SELECT
- id,content_hash,no_orig_date,title,plugin_data,guid,
- ".SUBSTRING_FOR_DATE."(date_updated,1,19) as date_updated,
- ".SUBSTRING_FOR_DATE."(updated,1,19) as updated,
+ id,content_hash,title,plugin_data,guid,
num_comments
FROM
ttrss_entries
$orig_content_hash = db_fetch_result($result, 0, "content_hash");
$orig_title = db_fetch_result($result, 0, "title");
$orig_num_comments = db_fetch_result($result, 0, "num_comments");
- $orig_date_updated = strtotime(db_fetch_result($result,
- 0, "date_updated"));
$orig_plugin_data = db_fetch_result($result, 0, "plugin_data");
$ref_id = db_fetch_result($result, 0, "id");
$p = new Publisher(PUBSUBHUBBUB_HUB);
- $pubsub_result = $p->publish_update($rss_link);
+ /* $pubsub_result = */ $p->publish_update($rss_link);
}
$result = db_query(
if (is_array($encs)) {
foreach ($encs as $e) {
$e_item = array(
- $e->link, $e->type, $e->length);
+ $e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
array_push($enclosures, $e_item);
}
}
db_query("BEGIN");
+// debugging
+// db_query("DELETE FROM ttrss_enclosures WHERE post_id = '$entry_ref_id'");
+
foreach ($enclosures as $enc) {
$enc_url = db_escape_string($enc[0]);
$enc_type = db_escape_string($enc[1]);
$enc_dur = db_escape_string($enc[2]);
+ $enc_title = db_escape_string($enc[3]);
+ $enc_width = intval($enc[4]);
+ $enc_height = intval($enc[5]);
$result = db_query("SELECT id FROM ttrss_enclosures
WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
if (db_num_rows($result) == 0) {
db_query("INSERT INTO ttrss_enclosures
- (content_url, content_type, title, duration, post_id) VALUES
- ('$enc_url', '$enc_type', '', '$enc_dur', '$entry_ref_id')");
+ (content_url, content_type, title, duration, post_id, width, height) VALUES
+ ('$enc_url', '$enc_type', '$enc_title', '$enc_dur', '$entry_ref_id', $enc_width, $enc_height)");
}
}
_debug("article processed", $debug_enabled);
}
- if (!$last_updated) {
- _debug("new feed, catching it up...", $debug_enabled);
- catchup_feed($feed, false, $owner_uid);
- }
-
_debug("purging feed...", $debug_enabled);
purge_feed($feed, 0, $debug_enabled);
$error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
- _debug("error fetching feed: $error_msg", $debug_enabled);
+ _debug("fetch error: $error_msg", $debug_enabled);
+
+ if (count($rss->errors()) > 1) {
+ foreach ($rss->errors() as $error) {
+ _debug("+ $error");
+ }
+ }
db_query(
"UPDATE ttrss_feeds SET last_error = '$error_msg',
- last_updated = NOW() WHERE id = '$feed'");
- }
+ last_updated = NOW() WHERE id = '$feed'");
- unset($rss);
+ unset($rss);
+ }
_debug("done", $debug_enabled);
+
+ return $rss;
}
function cache_images($html, $site_url, $debug) {
- $cache_dir = CACHE_DIR . "/images";
-
libxml_use_internal_errors(true);
$charset_hack = '<head>
if (!file_exists($local_filename)) {
$file_content = fetch_file_contents($src);
- if ($file_content && strlen($file_content) > 1024) {
+ if ($file_content && strlen($file_content) > _MIN_CACHE_IMAGE_SIZE) {
file_put_contents($local_filename, $file_content);
}
}
- if (file_exists($local_filename)) {
+ /* if (file_exists($local_filename)) {
$entry->setAttribute('src', SELF_URL_PATH . '/image.php?url=' .
base64_encode($src));
- }
+ } */
}
}
- $node = $doc->getElementsByTagName('body')->item(0);
-
- return $doc->saveXML($node);
+ //$node = $doc->getElementsByTagName('body')->item(0);
+ //return $doc->saveXML($node);
}
function expire_error_log($debug) {
mb_strtolower(strip_tags($title), 'utf-8'));
}
- function verify_feed_xml($feed_data) {
+ /* function verify_feed_xml($feed_data) {
libxml_use_internal_errors(true);
$doc = new DOMDocument();
$doc->loadXML($feed_data);
$error = libxml_get_last_error();
libxml_clear_errors();
return $error;
- }
+ } */
function housekeeping_common($debug) {
expire_cached_files($debug);
$rc = cleanup_tags( 14, 50000);
_debug("Cleaned $rc cached tags.");
+
+ PluginHost::getInstance()->run_hooks(PluginHost::HOOK_HOUSE_KEEPING, "hook_house_keeping", "");
+
}
?>
git.wh0rd.org / tt-rss.git / blobdiff