/**
* Checks to see if the parameters by which the active crawl are being
* conducted have been modified since the last time the values were put
* into queue server field variables. If so, it updates the values to
* to their new values
*/
function checkUpdateCrawlParameters()
{
crawlLog("Check for update in crawl parameters...");
$dir = CRAWL_DIR . '/cache/' . self::index_data_base_name . $this->crawl_time;
$modified_time = IndexArchiveBundle::getParamModifiedTime($dir);
if ($this->archive_modified_time == $modified_time) {
crawlLog("...none.");
return;
}
$updatable_info = array("page_range_request" => self::PAGE_RANGE_REQUEST, "max_description_len" => self::MAX_DESCRIPTION_LEN, "page_recrawl_frequency" => self::PAGE_RECRAWL_FREQUENCY, "restrict_sites_by_url" => self::RESTRICT_SITES_BY_URL, "cache_pages" => self::CACHE_PAGES, "allowed_sites" => self::ALLOWED_SITES, "disallowed_sites" => self::DISALLOWED_SITES, "page_rules" => self::PAGE_RULES, "indexed_file_types" => self::INDEXED_FILE_TYPES, "indexing_plugins" => self::INDEXING_PLUGINS, "indexing_plugins_data" => self::INDEXING_PLUGINS_DATA, "video_sources" => self::VIDEO_SOURCES);
$keys = array_keys($updatable_info);
$archive_info = IndexArchiveBundle::getArchiveInfo($dir);
$index_info = unserialize($archive_info['DESCRIPTION']);
$check_cull_fields = array("restrict_sites_by_url", "allowed_sites", "disallowed_sites");
$cull_now_non_crawlable = false;
foreach ($keys as $index_field) {
if (isset($index_info[$updatable_info[$index_field]])) {
if ($index_field == "disallowed_sites") {
$update_disallow = true;
}
if (in_array($index_field, $check_cull_fields) && (!isset($this->{$index_field}) || $this->{$index_field} != $index_info[$updatable_info[$index_field]])) {
$cull_now_non_crawlable = true;
}
$this->{$index_field} = $index_info[$updatable_info[$index_field]];
if ($this->isOnlyScheduler()) {
crawlLog("Scheduler Updating ...{$index_field}.");
} else {
crawlLog("Updating ...{$index_field}.");
}
}
}
/* We now do further processing or disallowed sites to see if any
of them are really quota sites
*/
if ($update_disallow == true) {
$this->updateDisallowedQuotaSites();
}
if ($this->isAScheduler() && $cull_now_non_crawlable) {
crawlLog("Scheduler: Allowed/Disallowed Urls have changed");
crawlLog("Scheduler: Checking if urls in queue need to be culled");
$this->cullNoncrawlableSites();
}
$this->archive_modified_time = $modified_time;
}
