if (($detect_language == 3 || !$language && $detect_language == 2) && $text_sample) {
try {
if ($use_cld) {
// Use PHP-CLD extension
$php_cld = 'CLD\\detect';
// in quotes to prevent PHP 5.2 parse error
$res = $php_cld($text_sample);
if (is_array($res) && count($res) > 0) {
$language = $res[0]['code'];
}
} else {
//die('what');
// Use PEAR's Text_LanguageDetect
if (!isset($l)) {
$l = new Text_LanguageDetect();
$l->setNameMode(2);
// return ISO 639-1 codes (e.g. "en")
}
$l_result = $l->detect($text_sample, 1);
if (count($l_result) > 0) {
$language = key($l_result);
}
}
} catch (Exception $e) {
//die('error: '.$e);
// do nothing
}
}
if ($language && strlen($language) < 7) {
$newitem->addElement('dc:language', $language);
}
<?php
/**
* Demonstrates how to use ISO language codes.
*
* The "name mode" changes the way languages are accepted and returned.
*/
require_once 'Text/LanguageDetect.php';
$l = new Text_LanguageDetect();
//will output the ISO 639-1 two-letter language code
// "de"
$l->setNameMode(2);
echo $l->detectSimple('Das ist ein kleiner Text') . "\n";
//will output the ISO 639-2 three-letter language code
// "deu"
$l->setNameMode(3);
echo $l->detectSimple('Das ist ein kleiner Text') . "\n";
/**
* @brief Takes a string and tries to identify the language.
*
* It uses the pear library Text_LanguageDetect and it can identify 52 human languages.
* It returns the identified languges and a confidence score for each.
*
* Strings need to have a min length config['system']['language_detect_min_length']
* and you can influence the confidence that must be met before a result will get
* returned through config['system']['language_detect_min_confidence'].
*
* @see http://pear.php.net/package/Text_LanguageDetect
* @param string $s A string to examine
* @return Language code in 2-letter ISO 639-1 (en, de, fr) format
*/
function detect_language($s)
{
require_once 'Text/LanguageDetect.php';
$min_length = get_config('system', 'language_detect_min_length');
if ($min_length === false) {
$min_length = LANGUAGE_DETECT_MIN_LENGTH;
}
$min_confidence = get_config('system', 'language_detect_min_confidence');
if ($min_confidence === false) {
$min_confidence = LANGUAGE_DETECT_MIN_CONFIDENCE;
}
// embedded apps have long base64 strings which will trip up the detector.
$naked_body = preg_replace('/\\[app\\](.*?)\\[\\/app\\]/', '', $s);
// strip off bbcode
$naked_body = preg_replace('/\\[(.+?)\\]/', '', $naked_body);
if (mb_strlen($naked_body) < intval($min_length)) {
logger('string length less than ' . intval($min_length), LOGGER_DATA);
return '';
}
$l = new Text_LanguageDetect();
try {
// return 2-letter ISO 639-1 (en) language code
$l->setNameMode(2);
$lng = $l->detectConfidence($naked_body);
logger('detect language: ' . print_r($lng, true) . $naked_body, LOGGER_DATA);
} catch (Text_LanguageDetect_Exception $e) {
logger('detect language exception: ' . $e->getMessage(), LOGGER_DATA);
}
if (!$lng || !x($lng, 'language')) {
return '';
}
if ($lng['confidence'] < (double) $min_confidence) {
logger('detect language: confidence less than ' . (double) $min_confidence, LOGGER_DATA);
return '';
}
return $lng['language'];
}
function update_rss_feed($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);
$result = db_query("SELECT id,update_interval,auth_login,\n\t\t\tfeed_url,auth_pass,cache_images,\n\t\t\tmark_unread_on_update, owner_uid,\n\t\t\tpubsub_state, auth_pass_encrypted,\n\t\t\t(SELECT max(date_entered) FROM\n\t\t\t\tttrss_entries, ttrss_user_entries where ref_id = id AND feed_id = '{$feed}') AS last_article_timestamp\n\t\t\tFROM ttrss_feeds WHERE id = '{$feed}'");
if (db_num_rows($result) == 0) {
_debug("feed {$feed} NOT FOUND/SKIPPED", $debug_enabled);
return false;
}
$last_article_timestamp = @strtotime(db_fetch_result($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"));
$pubsub_state = db_fetch_result($result, 0, "pubsub_state");
$auth_pass_encrypted = sql_bool_to_bool(db_fetch_result($result, 0, "auth_pass_encrypted"));
db_query("UPDATE ttrss_feeds SET last_update_started = NOW()\n\t\t\tWHERE id = '{$feed}'");
$auth_login = db_fetch_result($result, 0, "auth_login");
$auth_pass = db_fetch_result($result, 0, "auth_pass");
if ($auth_pass_encrypted) {
require_once "crypt.php";
$auth_pass = decrypt_string($auth_pass);
}
$cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
$fetch_url = db_fetch_result($result, 0, "feed_url");
$feed = db_escape_string($feed);
$date_feed_processed = date('Y-m-d H:i');
$cache_filename = CACHE_DIR . "/simplepie/" . sha1($fetch_url) . ".xml";
$pluginhost = new PluginHost();
$pluginhost->set_debug($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);
$pluginhost->load_data();
if ($rss && is_object($rss) && get_class($rss) == "FeedParser") {
_debug("using previously initialized parser object");
} else {
$rss_hash = false;
$force_refetch = isset($_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, $auth_pass);
}
// try cache
if (!$feed_data && file_exists($cache_filename) && is_readable($cache_filename) && !$auth_login && !$auth_pass && filemtime($cache_filename) > time() - 30) {
_debug("using local cache [{$cache_filename}].", $debug_enabled);
@($feed_data = file_get_contents($cache_filename));
if ($feed_data) {
$rss_hash = sha1($feed_data);
}
} 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);
$feed_data = fetch_file_contents($fetch_url, false, $auth_login, $auth_pass, false, $no_cache ? FEED_FETCH_NO_CACHE_TIMEOUT : FEED_FETCH_TIMEOUT, $force_refetch ? 0 : $last_article_timestamp);
global $fetch_curl_used;
if (!$fetch_curl_used) {
$tmp = @gzdecode($feed_data);
if ($tmp) {
$feed_data = $tmp;
}
}
$feed_data = trim($feed_data);
_debug("fetch done.", $debug_enabled);
/* if ($feed_data) {
$error = verify_feed_xml($feed_data);
if ($error) {
_debug("error verifying XML, code: " . $error->code, $debug_enabled);
if ($error->code == 26) {
_debug("got error 26, trying to decode entities...", $debug_enabled);
$feed_data = html_entity_decode($feed_data, ENT_COMPAT, 'UTF-8');
$error = verify_feed_xml($feed_data);
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, $feed_data);
}
}
}
if (!$feed_data) {
global $fetch_last_error;
global $fetch_last_error_code;
_debug("unable to fetch: {$fetch_last_error} [{$fetch_last_error_code}]", $debug_enabled);
$error_escaped = '';
// If-Modified-Since
if ($fetch_last_error_code != 304) {
$error_escaped = db_escape_string($fetch_last_error);
} else {
_debug("source claims data not modified, nothing to do.", $debug_enabled);
}
db_query("UPDATE ttrss_feeds SET last_error = '{$error_escaped}',\n\t\t\t\t\t\tlast_updated = NOW() WHERE id = '{$feed}'");
return;
}
}
foreach ($pluginhost->get_hooks(PluginHost::HOOK_FEED_FETCHED) as $plugin) {
$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()) {
// 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("processing feed data...", $debug_enabled);
// db_query("BEGIN");
if (DB_TYPE == "pgsql") {
$favicon_interval_qpart = "favicon_last_checked < NOW() - INTERVAL '12 hour'";
} else {
$favicon_interval_qpart = "favicon_last_checked < DATE_SUB(NOW(), INTERVAL 12 HOUR)";
}
$result = db_query("SELECT title,site_url,owner_uid,favicon_avg_color,\n\t\t\t\t(favicon_last_checked IS NULL OR {$favicon_interval_qpart}) AS\n\t\t\t\t\t\tfavicon_needs_check\n\t\t\t\tFROM ttrss_feeds WHERE id = '{$feed}'");
$registered_title = db_fetch_result($result, 0, "title");
$orig_site_url = db_fetch_result($result, 0, "site_url");
$favicon_needs_check = sql_bool_to_bool(db_fetch_result($result, 0, "favicon_needs_check"));
$favicon_avg_color = db_fetch_result($result, 0, "favicon_avg_color");
$owner_uid = db_fetch_result($result, 0, "owner_uid");
$site_url = db_escape_string(mb_substr(rewrite_relative_url($fetch_url, $rss->get_link()), 0, 245));
_debug("site_url: {$site_url}", $debug_enabled);
_debug("feed_title: " . $rss->get_title(), $debug_enabled);
if ($favicon_needs_check || $force_refetch) {
/* terrible hack: if we crash on floicon shit here, we won't check
* the icon avgcolor again (unless the icon got updated) */
$favicon_file = ICONS_DIR . "/{$feed}.ico";
$favicon_modified = @filemtime($favicon_file);
_debug("checking favicon...", $debug_enabled);
check_feed_favicon($site_url, $feed);
$favicon_modified_new = @filemtime($favicon_file);
if ($favicon_modified_new > $favicon_modified) {
$favicon_avg_color = '';
}
if (file_exists($favicon_file) && function_exists("imagecreatefromstring") && $favicon_avg_color == '') {
require_once "colors.php";
db_query("UPDATE ttrss_feeds SET favicon_avg_color = 'fail' WHERE\n\t\t\t\t\t\t\tid = '{$feed}'");
$favicon_color = db_escape_string(calculate_avg_color($favicon_file));
$favicon_colorstring = ",favicon_avg_color = '" . $favicon_color . "'";
} else {
if ($favicon_avg_color == 'fail') {
_debug("floicon failed on this file, not trying to recalculate avg color", $debug_enabled);
}
}
db_query("UPDATE ttrss_feeds SET favicon_last_checked = NOW()\n\t\t\t\t\t{$favicon_colorstring}\n\t\t\t\t\tWHERE id = '{$feed}'");
}
if (!$registered_title || $registered_title == "[Unknown]") {
$feed_title = db_escape_string(mb_substr($rss->get_title(), 0, 199));
if ($feed_title) {
_debug("registering title: {$feed_title}", $debug_enabled);
db_query("UPDATE ttrss_feeds SET\n\t\t\t\t\t\ttitle = '{$feed_title}' WHERE id = '{$feed}'");
}
}
if ($site_url && $orig_site_url != $site_url) {
db_query("UPDATE ttrss_feeds SET\n\t\t\t\t\tsite_url = '{$site_url}' WHERE id = '{$feed}'");
}
_debug("loading filters & labels...", $debug_enabled);
$filters = load_filters($feed, $owner_uid);
$labels = get_all_labels($owner_uid);
_debug("" . count($filters) . " filters loaded.", $debug_enabled);
$items = $rss->get_items();
if (!is_array($items)) {
_debug("no articles found.", $debug_enabled);
db_query("UPDATE ttrss_feeds\n\t\t\t\t\tSET last_updated = NOW(), last_error = '' WHERE id = '{$feed}'");
return;
// no articles
}
if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
_debug("checking for PUSH hub...", $debug_enabled);
$feed_hub_url = false;
$links = $rss->get_links('hub');
if ($links && is_array($links)) {
foreach ($links as $l) {
$feed_hub_url = $l;
break;
}
}
_debug("feed hub url: {$feed_hub_url}", $debug_enabled);
$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';
$callback_url = get_self_url_prefix() . "/public.php?op=pubsub&id={$feed}";
$s = new Subscriber($feed_hub_url, $callback_url);
$rc = $s->subscribe($feed_self_url);
_debug("feed hub url found, subscribe request sent. [rc={$rc}]", $debug_enabled);
db_query("UPDATE ttrss_feeds SET pubsub_state = 1\n\t\t\t\t\t\tWHERE id = '{$feed}'");
}
}
_debug("processing articles...", $debug_enabled);
foreach ($items as $item) {
if ($_REQUEST['xdebug'] == 3) {
print_r($item);
}
$entry_guid = $item->get_id();
if (!$entry_guid) {
$entry_guid = $item->get_link();
}
if (!$entry_guid) {
$entry_guid = make_guid_from_title($item->get_title());
}
if (!$entry_guid) {
continue;
}
$entry_guid = "{$owner_uid},{$entry_guid}";
$entry_guid_hashed = db_escape_string('SHA1:' . sha1($entry_guid));
_debug("guid {$entry_guid} / {$entry_guid_hashed}", $debug_enabled);
$entry_timestamp = "";
$entry_timestamp = $item->get_date();
_debug("orig date: " . $item->get_date(), $debug_enabled);
if ($entry_timestamp == -1 || !$entry_timestamp || $entry_timestamp > time()) {
$entry_timestamp = time();
}
$entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
_debug("date {$entry_timestamp} [{$entry_timestamp_fmt}]", $debug_enabled);
// $entry_title = html_entity_decode($item->get_title(), ENT_COMPAT, 'UTF-8');
// $entry_title = decode_numeric_entities($entry_title);
$entry_title = $item->get_title();
$entry_link = rewrite_relative_url($site_url, $item->get_link());
_debug("title {$entry_title}", $debug_enabled);
_debug("link {$entry_link}", $debug_enabled);
if (!$entry_title) {
$entry_title = date("Y-m-d H:i:s", $entry_timestamp);
}
$entry_content = $item->get_content();
if (!$entry_content) {
$entry_content = $item->get_description();
}
if ($_REQUEST["xdebug"] == 2) {
print "content: ";
print $entry_content;
print "\n";
}
$entry_language = "";
if (DETECT_ARTICLE_LANGUAGE) {
$entry_language = $lang->detect($entry_title . " " . $entry_content, 1);
if (count($entry_language) > 0) {
$possible = array_keys($entry_language);
$entry_language = $possible[0];
_debug("detected language: {$entry_language}", $debug_enabled);
} else {
$entry_language = "";
}
}
$entry_comments = $item->get_comments_url();
$entry_author = $item->get_author();
$entry_guid = db_escape_string(mb_substr($entry_guid, 0, 245));
$entry_comments = db_escape_string(mb_substr(trim($entry_comments), 0, 245));
$entry_author = db_escape_string(mb_substr(trim($entry_author), 0, 245));
$num_comments = (int) $item->get_comments_count();
_debug("author {$entry_author}", $debug_enabled);
_debug("num_comments: {$num_comments}", $debug_enabled);
_debug("looking for tags...", $debug_enabled);
// parse <category> entries into tags
$additional_tags = array();
$additional_tags_src = $item->get_categories();
if (is_array($additional_tags_src)) {
foreach ($additional_tags_src as $tobj) {
array_push($additional_tags, $tobj);
}
}
$entry_tags = array_unique($additional_tags);
for ($i = 0; $i < count($entry_tags); $i++) {
$entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
}
_debug("tags found: " . join(",", $entry_tags), $debug_enabled);
_debug("done collecting data.", $debug_enabled);
$result = db_query("SELECT id, content_hash FROM ttrss_entries\n\t\t\t\t\tWHERE guid = '" . db_escape_string($entry_guid) . "' OR guid = '{$entry_guid_hashed}'");
if (db_num_rows($result) != 0) {
$base_entry_id = db_fetch_result($result, 0, "id");
$entry_stored_hash = db_fetch_result($result, 0, "content_hash");
} else {
$base_entry_id = false;
$entry_stored_hash = "";
}
$article = array("owner_uid" => $owner_uid, "guid" => $entry_guid, "title" => $entry_title, "content" => $entry_content, "link" => $entry_link, "tags" => $entry_tags, "author" => $entry_author, "language" => $entry_language, "feed" => array("id" => $feed, "fetch_url" => $fetch_url, "site_url" => $site_url));
$entry_plugin_data = "";
$entry_current_hash = calculate_article_hash($article, $pluginhost);
_debug("article hash: {$entry_current_hash} [stored={$entry_stored_hash}]", $debug_enabled);
if ($entry_current_hash == $entry_stored_hash && !isset($_REQUEST["force_rehash"])) {
_debug("stored article seems up to date [IID: {$base_entry_id}], updating timestamp only", $debug_enabled);
// we keep encountering the entry in feeds, so we need to
// update date_updated column so that we don't get horrible
// dupes when the entry gets purged and reinserted again e.g.
// in the case of SLOW SLOW OMG SLOW updating feeds
$base_entry_id = db_fetch_result($result, 0, "id");
db_query("UPDATE ttrss_entries SET date_updated = NOW()\n\t\t\t\t\t\tWHERE id = '{$base_entry_id}'");
// if we allow duplicate posts, we have to continue to
// create the user entries for this feed
if (!get_pref("ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
continue;
}
}
_debug("hash differs, applying plugin filters:", $debug_enabled);
foreach ($pluginhost->get_hooks(PluginHost::HOOK_ARTICLE_FILTER) as $plugin) {
_debug("... " . get_class($plugin), $debug_enabled);
$start = microtime(true);
$article = $plugin->hook_article_filter($article);
_debug("=== " . sprintf("%.4f (sec)", microtime(true) - $start), $debug_enabled);
$entry_plugin_data .= mb_strtolower(get_class($plugin)) . ",";
}
$entry_plugin_data = db_escape_string($entry_plugin_data);
_debug("plugin data: {$entry_plugin_data}", $debug_enabled);
$entry_tags = $article["tags"];
$entry_guid = db_escape_string($entry_guid);
$entry_title = db_escape_string($article["title"]);
$entry_author = db_escape_string($article["author"]);
$entry_link = db_escape_string($article["link"]);
$entry_content = $article["content"];
// escaped below
if ($cache_images && is_writable(CACHE_DIR . '/images')) {
cache_images($entry_content, $site_url, $debug_enabled);
}
$entry_content = db_escape_string($entry_content, false);
db_query("BEGIN");
$result = db_query("SELECT id FROM\tttrss_entries\n\t\t\t\t\tWHERE (guid = '{$entry_guid}' OR guid = '{$entry_guid_hashed}')");
if (db_num_rows($result) == 0) {
_debug("base guid [{$entry_guid}] not found", $debug_enabled);
// base post entry does not exist, create it
$result = db_query("INSERT INTO ttrss_entries\n\t\t\t\t\t\t\t(title,\n\t\t\t\t\t\t\tguid,\n\t\t\t\t\t\t\tlink,\n\t\t\t\t\t\t\tupdated,\n\t\t\t\t\t\t\tcontent,\n\t\t\t\t\t\t\tcontent_hash,\n\t\t\t\t\t\t\tno_orig_date,\n\t\t\t\t\t\t\tdate_updated,\n\t\t\t\t\t\t\tdate_entered,\n\t\t\t\t\t\t\tcomments,\n\t\t\t\t\t\t\tnum_comments,\n\t\t\t\t\t\t\tplugin_data,\n\t\t\t\t\t\t\tlang,\n\t\t\t\t\t\t\tauthor)\n\t\t\t\t\t\tVALUES\n\t\t\t\t\t\t\t('{$entry_title}',\n\t\t\t\t\t\t\t'{$entry_guid_hashed}',\n\t\t\t\t\t\t\t'{$entry_link}',\n\t\t\t\t\t\t\t'{$entry_timestamp_fmt}',\n\t\t\t\t\t\t\t'{$entry_content}',\n\t\t\t\t\t\t\t'{$entry_current_hash}',\n\t\t\t\t\t\t\tfalse,\n\t\t\t\t\t\t\tNOW(),\n\t\t\t\t\t\t\t'{$date_feed_processed}',\n\t\t\t\t\t\t\t'{$entry_comments}',\n\t\t\t\t\t\t\t'{$num_comments}',\n\t\t\t\t\t\t\t'{$entry_plugin_data}',\n\t\t\t\t\t\t\t'{$entry_language}',\n\t\t\t\t\t\t\t'{$entry_author}')");
$article_labels = array();
} else {
$base_entry_id = db_fetch_result($result, 0, "id");
$article_labels = get_article_labels($base_entry_id, $owner_uid);
}
// now it should exist, if not - bad luck then
$result = db_query("SELECT id FROM ttrss_entries\n\t\t\t\t\tWHERE guid = '{$entry_guid}' OR guid = '{$entry_guid_hashed}'");
$entry_ref_id = 0;
$entry_int_id = 0;
if (db_num_rows($result) == 1) {
_debug("base guid found, checking for user record", $debug_enabled);
$ref_id = db_fetch_result($result, 0, "id");
$entry_ref_id = $ref_id;
/* $stored_guid = db_fetch_result($result, 0, "guid");
if ($stored_guid != $entry_guid_hashed) {
if ($debug_enabled) _debug("upgrading compat guid to hashed one", $debug_enabled);
db_query("UPDATE ttrss_entries SET guid = '$entry_guid_hashed' WHERE
id = '$ref_id'");
} */
// check for user post link to main table
// do we allow duplicate posts with same GUID in different feeds?
if (get_pref("ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
$dupcheck_qpart = "AND (feed_id = '{$feed}' OR feed_id IS NULL)";
} else {
$dupcheck_qpart = "";
}
/* Collect article tags here so we could filter by them: */
$article_filters = get_article_filters($filters, $entry_title, $entry_content, $entry_link, $entry_timestamp, $entry_author, $entry_tags);
if ($debug_enabled) {
_debug("article filters: ", $debug_enabled);
if (count($article_filters) != 0) {
print_r($article_filters);
}
}
if (find_article_filter($article_filters, "filter")) {
db_query("COMMIT");
// close transaction in progress
continue;
}
$score = calculate_article_score($article_filters);
_debug("initial score: {$score}", $debug_enabled);
$query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE\n\t\t\t\t\t\t\tref_id = '{$ref_id}' AND owner_uid = '{$owner_uid}'\n\t\t\t\t\t\t\t{$dupcheck_qpart}";
// if ($_REQUEST["xdebug"]) print "$query\n";
$result = db_query($query);
// okay it doesn't exist - create user entry
if (db_num_rows($result) == 0) {
_debug("user record not found, creating...", $debug_enabled);
if ($score >= -500 && !find_article_filter($article_filters, 'catchup')) {
$unread = 'true';
$last_read_qpart = 'NULL';
} else {
$unread = 'false';
$last_read_qpart = 'NOW()';
}
if (find_article_filter($article_filters, 'mark') || $score > 1000) {
$marked = 'true';
} else {
$marked = 'false';
}
if (find_article_filter($article_filters, 'publish')) {
$published = 'true';
} else {
$published = 'false';
}
// N-grams
if (DB_TYPE == "pgsql" and defined('_NGRAM_TITLE_DUPLICATE_THRESHOLD')) {
$result = db_query("SELECT COUNT(*) AS similar FROM\n\t\t\t\t\t\t\t\t\tttrss_entries,ttrss_user_entries\n\t\t\t\t\t\t\t\tWHERE ref_id = id AND updated >= NOW() - INTERVAL '7 day'\n\t\t\t\t\t\t\t\t\tAND similarity(title, '{$entry_title}') >= " . _NGRAM_TITLE_DUPLICATE_THRESHOLD . "\n\t\t\t\t\t\t\t\t\tAND owner_uid = {$owner_uid}");
$ngram_similar = db_fetch_result($result, 0, "similar");
_debug("N-gram similar results: {$ngram_similar}", $debug_enabled);
if ($ngram_similar > 0) {
$unread = 'false';
}
}
$last_marked = $marked == 'true' ? 'NOW()' : 'NULL';
$last_published = $published == 'true' ? 'NOW()' : 'NULL';
$result = db_query("INSERT INTO ttrss_user_entries\n\t\t\t\t\t\t\t\t(ref_id, owner_uid, feed_id, unread, last_read, marked,\n\t\t\t\t\t\t\t\tpublished, score, tag_cache, label_cache, uuid,\n\t\t\t\t\t\t\t\tlast_marked, last_published)\n\t\t\t\t\t\t\tVALUES ('{$ref_id}', '{$owner_uid}', '{$feed}', {$unread},\n\t\t\t\t\t\t\t\t{$last_read_qpart}, {$marked}, {$published}, '{$score}', '', '',\n\t\t\t\t\t\t\t\t'', {$last_marked}, {$last_published})");
if (PUBSUBHUBBUB_HUB && $published == 'true') {
$rss_link = get_self_url_prefix() . "/public.php?op=rss&id=-2&key=" . get_feed_access_key(-2, false, $owner_uid);
$p = new Publisher(PUBSUBHUBBUB_HUB);
/* $pubsub_result = */
$p->publish_update($rss_link);
}
$result = db_query("SELECT int_id FROM ttrss_user_entries WHERE\n\t\t\t\t\t\t\t\tref_id = '{$ref_id}' AND owner_uid = '{$owner_uid}' AND\n\t\t\t\t\t\t\t\tfeed_id = '{$feed}' LIMIT 1");
if (db_num_rows($result) == 1) {
$entry_int_id = db_fetch_result($result, 0, "int_id");
}
} else {
_debug("user record FOUND", $debug_enabled);
$entry_ref_id = db_fetch_result($result, 0, "ref_id");
$entry_int_id = db_fetch_result($result, 0, "int_id");
}
_debug("RID: {$entry_ref_id}, IID: {$entry_int_id}", $debug_enabled);
db_query("UPDATE ttrss_entries\n\t\t\t\t\t\tSET title = '{$entry_title}',\n\t\t\t\t\t\t\tcontent = '{$entry_content}',\n\t\t\t\t\t\t\tcontent_hash = '{$entry_current_hash}',\n\t\t\t\t\t\t\tupdated = '{$entry_timestamp_fmt}',\n\t\t\t\t\t\t\tnum_comments = '{$num_comments}',\n\t\t\t\t\t\t\tplugin_data = '{$entry_plugin_data}',\n\t\t\t\t\t\t\tauthor = '{$entry_author}',\n\t\t\t\t\t\t\tlang = '{$entry_language}'\n\t\t\t\t\t\tWHERE id = '{$ref_id}'");
if ($mark_unread_on_update) {
db_query("UPDATE ttrss_user_entries\n\t\t\t\t\t\t\tSET last_read = null, unread = true WHERE ref_id = '{$ref_id}'");
}
}
db_query("COMMIT");
_debug("assigning labels...", $debug_enabled);
assign_article_to_label_filters($entry_ref_id, $article_filters, $owner_uid, $article_labels);
_debug("looking for enclosures...", $debug_enabled);
// enclosures
$enclosures = array();
$encs = $item->get_enclosures();
if (is_array($encs)) {
foreach ($encs as $e) {
$e_item = array($e->link, $e->type, $e->length, $e->title, $e->width, $e->height);
array_push($enclosures, $e_item);
}
}
if ($debug_enabled) {
_debug("article enclosures:", $debug_enabled);
print_r($enclosures);
}
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]);
$result = db_query("SELECT id FROM ttrss_enclosures\n\t\t\t\t\t\tWHERE content_url = '{$enc_url}' AND post_id = '{$entry_ref_id}'");
if (db_num_rows($result) == 0) {
db_query("INSERT INTO ttrss_enclosures\n\t\t\t\t\t\t\t(content_url, content_type, title, duration, post_id, width, height) VALUES\n\t\t\t\t\t\t\t('{$enc_url}', '{$enc_type}', '{$enc_title}', '{$enc_dur}', '{$entry_ref_id}', {$enc_width}, {$enc_height})");
}
}
db_query("COMMIT");
// check for manual tags (we have to do it here since they're loaded from filters)
foreach ($article_filters as $f) {
if ($f["type"] == "tag") {
$manual_tags = trim_array(explode(",", $f["param"]));
foreach ($manual_tags as $tag) {
if (tag_is_valid($tag)) {
array_push($entry_tags, $tag);
}
}
}
}
// Skip boring tags
$boring_tags = trim_array(explode(",", mb_strtolower(get_pref('BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
$filtered_tags = array();
$tags_to_cache = array();
if ($entry_tags && is_array($entry_tags)) {
foreach ($entry_tags as $tag) {
if (array_search($tag, $boring_tags) === false) {
array_push($filtered_tags, $tag);
}
}
}
$filtered_tags = array_unique($filtered_tags);
if ($debug_enabled) {
_debug("filtered article tags:", $debug_enabled);
print_r($filtered_tags);
}
// Save article tags in the database
if (count($filtered_tags) > 0) {
db_query("BEGIN");
foreach ($filtered_tags as $tag) {
$tag = sanitize_tag($tag);
$tag = db_escape_string($tag);
if (!tag_is_valid($tag)) {
continue;
}
$result = db_query("SELECT id FROM ttrss_tags\n\t\t\t\t\t\t\tWHERE tag_name = '{$tag}' AND post_int_id = '{$entry_int_id}' AND\n\t\t\t\t\t\t\towner_uid = '{$owner_uid}' LIMIT 1");
if ($result && db_num_rows($result) == 0) {
db_query("INSERT INTO ttrss_tags\n\t\t\t\t\t\t\t\t\t(owner_uid,tag_name,post_int_id)\n\t\t\t\t\t\t\t\t\tVALUES ('{$owner_uid}','{$tag}', '{$entry_int_id}')");
}
array_push($tags_to_cache, $tag);
}
/* update the cache */
$tags_to_cache = array_unique($tags_to_cache);
$tags_str = db_escape_string(join(",", $tags_to_cache));
db_query("UPDATE ttrss_user_entries\n\t\t\t\t\t\tSET tag_cache = '{$tags_str}' WHERE ref_id = '{$entry_ref_id}'\n\t\t\t\t\t\tAND owner_uid = {$owner_uid}");
db_query("COMMIT");
}
if (get_pref("AUTO_ASSIGN_LABELS", $owner_uid, false)) {
_debug("auto-assigning labels...", $debug_enabled);
foreach ($labels as $label) {
$caption = preg_quote($label["caption"]);
if ($caption && preg_match("/\\b{$caption}\\b/i", "{$tags_str} " . strip_tags($entry_content) . " {$entry_title}")) {
if (!labels_contains_caption($article_labels, $caption)) {
label_add_article($entry_ref_id, $caption, $owner_uid);
}
}
}
}
_debug("article processed", $debug_enabled);
}
_debug("purging feed...", $debug_enabled);
purge_feed($feed, 0, $debug_enabled);
db_query("UPDATE ttrss_feeds\n\t\t\t\tSET last_updated = NOW(), last_error = '' WHERE id = '{$feed}'");
// db_query("COMMIT");
} else {
$error_msg = db_escape_string(mb_substr($rss->error(), 0, 245));
_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 = '{$error_msg}',\n\t\t\t\tlast_updated = NOW() WHERE id = '{$feed}'");
unset($rss);
}
_debug("done", $debug_enabled);
return $rss;
}
