/**
* Mark containers that this parser might be interested in as unparsed.
*
* @uses blcContainerHelper::mark_as_unsynched_where()
*
* @param bool $only_return If true, just return the list of formats and container types without actually modifying any synch. records.
* @return void|array Either nothing or an array in the form [ [format1=>timestamp1, ...], [container_type1=>timestamp1, ...] ]
*/
function resynch_relevant_containers($only_return = false)
{
global $blclog;
$blclog->log(sprintf('...... Parser "%s" is marking relevant items as unsynched', $this->module_id));
$last_deactivated = $this->module_manager->get_last_deactivation_time($this->module_id);
$formats = array();
foreach ($this->supported_formats as $format) {
$formats[$format] = $last_deactivated;
}
$container_types = array();
foreach ($this->supported_containers as $container_type) {
$container_types[$container_type] = $last_deactivated;
}
if ($only_return) {
return array($formats, $container_types);
} else {
blcContainerHelper::mark_as_unsynched_where($formats, $container_types);
}
}
/**
* Delete synch. records, instances and links that refer to missing or invalid items.
*
* @return void
*/
function blc_cleanup_database()
{
global $blclog;
//Delete synch. records for container types that don't exist
$blclog->info('... Deleting invalid container records');
blcContainerHelper::cleanup_containers();
//Delete invalid instances
$blclog->info('... Deleting invalid link instances');
blc_cleanup_instances();
//Delete orphaned links
$blclog->info('... Deleting orphaned links');
blc_cleanup_links();
}
/**
* When a bookmark is deleted, remove the related DB records.
*
* @param int $link_id
* @return void
*/
function hook_delete_link($link_id)
{
//Get the container object.
$container = blcContainerHelper::get_container(array($this->container_type, $link_id));
//Get the link(s) associated with it.
$links = $container->get_links();
//Remove synch. record & instance records.
$container->delete();
//Clean up links associated with this bookmark (there's probably only one)
$link_ids = array();
foreach ($links as $link) {
$link_ids[] = $link->link_id;
}
blc_cleanup_links($link_ids);
}
/**
* Get the container object associated with this link instance
*
* @return blcContainer|null
*/
function &get_container()
{
if (is_null($this->_container)) {
$this->_container =& blcContainerHelper::get_container(array($this->container_type, $this->container_id));
}
return $this->_container;
}
/**
* The main worker function that does all kinds of things.
*
* @return void
*/
function work()
{
global $wpdb;
if (!$this->acquire_lock()) {
//FB::warn("Another instance of BLC is already working. Stop.");
return;
}
if ($this->server_too_busy()) {
//FB::warn("Server is too busy. Stop.");
return;
}
$this->start_timer();
$max_execution_time = $this->conf->options['max_execution_time'];
/*****************************************
Preparation
******************************************/
// Check for safe mode
if (blcUtility::is_safe_mode()) {
// Do it the safe mode way - obey the existing max_execution_time setting
$t = ini_get('max_execution_time');
if ($t && $t < $max_execution_time) {
$max_execution_time = $t - 1;
}
} else {
// Do it the regular way
@set_time_limit($max_execution_time * 2);
//x2 should be plenty, running any longer would mean a glitch.
}
//Don't stop the script when the connection is closed
ignore_user_abort(true);
//Close the connection as per http://www.php.net/manual/en/features.connection-handling.php#71172
//This reduces resource usage and may solve the mysterious slowdowns certain users have
//encountered when activating the plugin.
//(Disable when debugging or you won't get the FirePHP output)
if (!headers_sent() && (!defined('BLC_DEBUG') || !constant('BLC_DEBUG'))) {
@ob_end_clean();
//Discard the existing buffer, if any
header("Connection: close");
ob_start();
echo 'Connection closed';
//This could be anything
$size = ob_get_length();
header("Content-Length: {$size}");
ob_end_flush();
// Strange behaviour, will not work
flush();
// Unless both are called !
}
//Load modules for this context
$moduleManager = blcModuleManager::getInstance();
$moduleManager->load_modules('work');
/*****************************************
Parse posts and bookmarks
******************************************/
$orphans_possible = false;
$still_need_resynch = $this->conf->options['need_resynch'];
if ($still_need_resynch) {
//FB::log("Looking for containers that need parsing...");
while ($containers = blcContainerHelper::get_unsynched_containers(50)) {
//FB::log($containers, 'Found containers');
foreach ($containers as $container) {
//FB::log($container, "Parsing container");
$container->synch();
//Check if we still have some execution time left
if ($this->execution_time() > $max_execution_time) {
//FB::log('The alloted execution time has run out');
blc_cleanup_links();
$this->release_lock();
return;
}
//Check if the server isn't overloaded
if ($this->server_too_busy()) {
//FB::log('Server overloaded, bailing out.');
blc_cleanup_links();
$this->release_lock();
return;
}
}
$orphans_possible = true;
}
//FB::log('No unparsed items found.');
$still_need_resynch = false;
} else {
//FB::log('Resynch not required.');
}
/******************************************
Resynch done?
*******************************************/
if ($this->conf->options['need_resynch'] && !$still_need_resynch) {
$this->conf->options['need_resynch'] = $still_need_resynch;
$this->conf->save_options();
}
/******************************************
Remove orphaned links
*******************************************/
if ($orphans_possible) {
//FB::log('Cleaning up the link table.');
blc_cleanup_links();
}
//Check if we still have some execution time left
if ($this->execution_time() > $max_execution_time) {
//FB::log('The alloted execution time has run out');
$this->release_lock();
return;
}
if ($this->server_too_busy()) {
//FB::log('Server overloaded, bailing out.');
$this->release_lock();
return;
}
/*****************************************
Check links
******************************************/
while ($links = $this->get_links_to_check(30)) {
//Some unchecked links found
//FB::log("Checking ".count($links)." link(s)");
foreach ($links as $link) {
//Does this link need to be checked? Excluded links aren't checked, but their URLs are still
//tested periodically to see if they're still on the exlusion list.
if (!$this->is_excluded($link->url)) {
//Check the link.
//FB::log($link->url, "Checking link {$link->link_id}");
$link->check(true);
} else {
//FB::info("The URL {$link->url} is excluded, skipping link {$link->link_id}.");
$link->last_check_attempt = time();
$link->save();
}
//Check if we still have some execution time left
if ($this->execution_time() > $max_execution_time) {
//FB::log('The alloted execution time has run out');
$this->release_lock();
return;
}
//Check if the server isn't overloaded
if ($this->server_too_busy()) {
//FB::log('Server overloaded, bailing out.');
$this->release_lock();
return;
}
}
}
//FB::log('No links need to be checked right now.');
$this->release_lock();
//FB::log('All done.');
}
/**
* Get the message to display after $n containers of a specific type have been moved to the trash.
*
* @see blcContainerHelper::ui_bulk_delete_message()
*
* @param string $container_type
* @param int $n
* @return string
*/
static function ui_bulk_trash_message($container_type, $n)
{
$manager = blcContainerHelper::get_manager($container_type);
if (is_null($manager)) {
return sprintf(__("Container type '%s' not recognized", 'broken-link-checker'), $container_type);
} else {
return $manager->ui_bulk_trash_message($n);
}
}
/**
* The main worker function that does all kinds of things.
*
* @return void
*/
function work()
{
global $blclog;
//Close the session to prevent lock-ups.
//PHP sessions are blocking. session_start() will wait until all other scripts that are using the same session
//are finished. As a result, a long-running script that unintentionally keeps the session open can cause
//the entire site to "lock up" for the current user/browser. WordPress itself doesn't use sessions, but some
//plugins do, so we should explicitly close the session (if any) before starting the worker.
if (session_id() != '') {
session_write_close();
}
if (!$this->acquire_lock()) {
//FB::warn("Another instance of BLC is already working. Stop.");
$blclog->info('Another instance of BLC is already working. Stop.');
return;
}
if ($this->server_too_busy()) {
//FB::warn("Server is too busy. Stop.");
$blclog->warn('Server load is too high, stopping.');
return;
}
$this->start_timer();
$blclog->info('work() starts');
$max_execution_time = $this->conf->options['max_execution_time'];
/*****************************************
Preparation
******************************************/
// Check for safe mode
if (blcUtility::is_safe_mode()) {
// Do it the safe mode way - obey the existing max_execution_time setting
$t = ini_get('max_execution_time');
if ($t && $t < $max_execution_time) {
$max_execution_time = $t - 1;
}
} else {
// Do it the regular way
@set_time_limit($max_execution_time * 2);
//x2 should be plenty, running any longer would mean a glitch.
}
//Don't stop the script when the connection is closed
ignore_user_abort(true);
//Close the connection as per http://www.php.net/manual/en/features.connection-handling.php#71172
//This reduces resource usage.
//(Disable when debugging or you won't get the FirePHP output)
if (!headers_sent() && (defined('DOING_AJAX') && constant('DOING_AJAX')) && (!defined('BLC_DEBUG') || !constant('BLC_DEBUG'))) {
@ob_end_clean();
//Discard the existing buffer, if any
header("Connection: close");
ob_start();
echo 'Connection closed';
//This could be anything
$size = ob_get_length();
header("Content-Length: {$size}");
ob_end_flush();
// Strange behaviour, will not work
flush();
// Unless both are called !
}
//Load modules for this context
$moduleManager = blcModuleManager::getInstance();
$moduleManager->load_modules('work');
$target_usage_fraction = $this->conf->get('target_resource_usage', 0.25);
//Target usage must be between 1% and 100%.
$target_usage_fraction = max(min($target_usage_fraction, 1), 0.01);
/*****************************************
Parse posts and bookmarks
******************************************/
$orphans_possible = false;
$still_need_resynch = $this->conf->options['need_resynch'];
if ($still_need_resynch) {
//FB::log("Looking for containers that need parsing...");
$max_containers_per_query = 50;
$start = microtime(true);
$containers = blcContainerHelper::get_unsynched_containers($max_containers_per_query);
$get_containers_time = microtime(true) - $start;
while (!empty($containers)) {
//FB::log($containers, 'Found containers');
$this->sleep_to_maintain_ratio($get_containers_time, $target_usage_fraction);
foreach ($containers as $container) {
$synch_start_time = microtime(true);
//FB::log($container, "Parsing container");
$container->synch();
$synch_elapsed_time = microtime(true) - $synch_start_time;
$blclog->info(sprintf('Parsed container %s[%s] in %.2f ms', $container->container_type, $container->container_id, $synch_elapsed_time * 1000));
//Check if we still have some execution time left
if ($this->execution_time() > $max_execution_time) {
//FB::log('The allotted execution time has run out');
blc_cleanup_links();
$this->release_lock();
return;
}
//Check if the server isn't overloaded
if ($this->server_too_busy()) {
//FB::log('Server overloaded, bailing out.');
blc_cleanup_links();
$this->release_lock();
return;
}
//Intentionally slow down parsing to reduce the load on the server. Basically,
//we work $target_usage_fraction of the time and sleep the rest of the time.
$this->sleep_to_maintain_ratio($synch_elapsed_time, $target_usage_fraction);
}
$orphans_possible = true;
$start = microtime(true);
$containers = blcContainerHelper::get_unsynched_containers($max_containers_per_query);
$get_containers_time = microtime(true) - $start;
}
//FB::log('No unparsed items found.');
$still_need_resynch = false;
} else {
//FB::log('Resynch not required.');
}
/******************************************
Resynch done?
*******************************************/
if ($this->conf->options['need_resynch'] && !$still_need_resynch) {
$this->conf->options['need_resynch'] = $still_need_resynch;
$this->conf->save_options();
}
/******************************************
Remove orphaned links
*******************************************/
if ($orphans_possible) {
$start = microtime(true);
$blclog->info('Removing orphaned links.');
blc_cleanup_links();
$get_links_time = microtime(true) - $start;
$this->sleep_to_maintain_ratio($get_links_time, $target_usage_fraction);
}
//Check if we still have some execution time left
if ($this->execution_time() > $max_execution_time) {
//FB::log('The allotted execution time has run out');
$blclog->info('The allotted execution time has run out.');
$this->release_lock();
return;
}
if ($this->server_too_busy()) {
//FB::log('Server overloaded, bailing out.');
$blclog->info('Server load too high, stopping.');
$this->release_lock();
return;
}
/*****************************************
Check links
******************************************/
$max_links_per_query = 30;
$start = microtime(true);
$links = $this->get_links_to_check($max_links_per_query);
$get_links_time = microtime(true) - $start;
while ($links) {
$this->sleep_to_maintain_ratio($get_links_time, $target_usage_fraction);
//Some unchecked links found
//FB::log("Checking ".count($links)." link(s)");
$blclog->info("Checking " . count($links) . " link(s)");
//Randomizing the array reduces the chances that we'll get several links to the same domain in a row.
shuffle($links);
$transactionManager = TransactionManager::getInstance();
$transactionManager->start();
foreach ($links as $link) {
//Does this link need to be checked? Excluded links aren't checked, but their URLs are still
//tested periodically to see if they're still on the exclusion list.
if (!$this->is_excluded($link->url)) {
//Check the link.
//FB::log($link->url, "Checking link {$link->link_id}");
$link->check(true);
} else {
//FB::info("The URL {$link->url} is excluded, skipping link {$link->link_id}.");
$link->last_check_attempt = time();
$link->save();
}
//Check if we still have some execution time left
if ($this->execution_time() > $max_execution_time) {
//FB::log('The allotted execution time has run out');
$blclog->info('The allotted execution time has run out.');
$this->release_lock();
return;
}
//Check if the server isn't overloaded
if ($this->server_too_busy()) {
//FB::log('Server overloaded, bailing out.');
$blclog->info('Server load too high, stopping.');
$this->release_lock();
return;
}
}
$transactionManager->commit();
$start = microtime(true);
$links = $this->get_links_to_check($max_links_per_query);
$get_links_time = microtime(true) - $start;
}
//FB::log('No links need to be checked right now.');
$this->release_lock();
$blclog->info('work(): All done.');
//FB::log('All done.');
}
function hook_untrash_post_comments($post_id)
{
//Unlike with the 'trashed_post_comments' hook, WP doesn't pass the list of (un)trashed
//comments to callbacks assigned to the 'untrash_post_comments' and 'untrashed_post_comments'
//actions. Therefore, we must read it from the appropriate metadata entry.
$statuses = get_post_meta($post_id, '_wp_trash_meta_comments_status', true);
if (empty($statuses) || !is_array($statuses)) {
return;
}
foreach ($statuses as $comment_id => $comment_status) {
if ($comment_status == '1') {
//if approved
$container = blcContainerHelper::get_container(array($this->container_type, $comment_id));
$container->mark_as_unsynched();
}
}
}
/**
* When a post is restored, mark all of its custom fields as unparsed.
* Called via the 'untrashed_post' action.
*
* @param int $post_id
* @return void
*/
function post_untrashed($post_id)
{
//Get the associated container object
$container =& blcContainerHelper::get_container(array($this->container_type, intval($post_id)));
$container->mark_as_unsynched();
}
/**
* When a post is saved or modified, mark it as unparsed.
*
* @param int $post_id
* @return void
*/
function post_saved($post_id)
{
//Get the container type matching the type of the deleted post
$post = get_post($post_id);
if (!$post) {
return;
}
//Only check links in currently enabled post types
if (!in_array($post->post_type, $this->enabled_post_types)) {
return;
}
//Only check posts that have one of the allowed statuses
if (!in_array($post->post_status, $this->enabled_post_statuses)) {
return;
}
//Get the container & mark it as unparsed
$args = array($post->post_type, intval($post_id));
$post_container = blcContainerHelper::get_container($args);
$post_container->mark_as_unsynched();
}
public static function hook_post_deleted($post_id)
{
if (get_option('mainwp_linkschecker_ext_enabled') !== "Y") {
return;
}
if (!defined('BLC_ACTIVE') || !function_exists('blc_init')) {
return;
}
blc_init();
//Get the container type matching the type of the deleted post
$post = get_post($post_id);
if (!$post) {
return;
}
//Get the associated container object
$post_container = blcContainerHelper::get_container(array($post->post_type, intval($post_id)));
if ($post_container) {
//Delete it
$post_container->delete();
//Clean up any dangling links
blc_cleanup_links();
}
}
