/**
* Initiates a crawler-process
*/
protected function initCrawlerProcess()
{
// Create working directory
$this->createWorkingDirectory();
// Setup url-cache
if ($this->url_cache_type == PHPCrawlerUrlCacheTypes::URLCACHE_SQLITE) {
$this->LinkCache = new PHPCrawlerSQLiteURLCache($this->working_directory . "urlcache.db3", true);
} else {
$this->LinkCache = new PHPCrawlerMemoryURLCache();
}
// Setup cookie-cache (use SQLite-cache if crawler runs multi-processed)
if ($this->url_cache_type == PHPCrawlerUrlCacheTypes::URLCACHE_SQLITE) {
$this->CookieCache = new PHPCrawlerSQLiteCookieCache($this->working_directory . "cookiecache.db3", true);
} else {
$this->CookieCache = new PHPCrawlerMemoryCookieCache();
}
// ProcessCommunication
$this->ProcessCommunication = new PHPCrawlerProcessCommunication($this->crawler_uniqid, $this->multiprocess_mode, $this->working_directory);
// DocumentInfo-Queue
if ($this->multiprocess_mode == PHPCrawlerMultiProcessModes::MPMODE_PARENT_EXECUTES_USERCODE) {
$this->DocumentInfoQueue = new PHPCrawlerDocumentInfoQueue($this->working_directory . "doc_queue.db3", true);
}
// Set tmp-file for PageRequest
$this->PageRequest->setTmpFile($this->working_directory . "phpcrawl_" . getmypid() . ".tmp");
// Pass url-priorities to link-cache
$this->LinkCache->addLinkPriorities($this->link_priority_array);
// Pass base-URL to the UrlFilter
$this->UrlFilter->setBaseURL($this->starting_url);
// Add the starting-URL to the url-cache
$this->LinkCache->addUrl(new PHPCrawlerURLDescriptor($this->starting_url));
}
/**
* Initiates a crawler-process
*/
protected function initCrawlerProcess()
{
// Create working directory
$this->createWorkingDirectory();
// Setup url-cache
if ($this->SQLiteAvailable && $this->url_cache_type == PHPCrawlerUrlCacheTypes::URLCACHE_SQLITE) {
$this->LinkCache = new PHPCrawlerSQLiteURLCache($this->working_directory . "urlcache.db3", true);
} elseif ($this->url_cache_type == PHPCrawlerUrlCacheTypes::URLCACHE_D7) {
$this->LinkCache = new PHPCrawlerD7URLCache($this->database_url_cache_table, $this->crawler_uniqid);
} else {
$this->LinkCache = new PHPCrawlerMemoryURLCache();
}
// Perge/cleanup SQLite-urlcache for resumed crawling-processes (only ONCE!)
if (($this->SQLiteAvailable && $this->url_cache_type == PHPCrawlerUrlCacheTypes::URLCACHE_SQLITE || $this->url_cache_type == PHPCrawlerUrlCacheTypes::URLCACHE_D7) && $this->urlcache_purged == false) {
$this->LinkCache->purgeCache();
$this->urlcache_purged = true;
}
// Setup cookie-cache (use SQLite-cache if crawler runs multi-processed)
if ($this->SQLiteAvailable && $this->url_cache_type == PHPCrawlerUrlCacheTypes::URLCACHE_SQLITE) {
$this->CookieCache = new PHPCrawlerSQLiteCookieCache($this->working_directory . "cookiecache.db3", true);
} elseif ($this->url_cache_type == PHPCrawlerUrlCacheTypes::URLCACHE_D7) {
$this->CookieCache = new PHPCrawlerD7CookieCache($this->database_cookie_cache_table, $this->crawler_uniqid);
} else {
$this->CookieCache = new PHPCrawlerMemoryCookieCache();
}
// ProcessHandler
$this->ProcessHandler = new PHPCrawlerProcessHandler($this->crawler_uniqid, $this->working_directory);
// Setup PHPCrawlerStatusHandler
$this->CrawlerStatusHandler = new PHPCrawlerStatusHandler($this->crawler_uniqid, $this->working_directory);
$this->setupCrawlerStatusHandler();
// DocumentInfo-Queue
if ($this->multiprocess_mode == PHPCrawlerMultiProcessModes::MPMODE_PARENT_EXECUTES_USERCODE) {
$this->DocumentInfoQueue = new PHPCrawlerDocumentInfoQueue($this->working_directory . "doc_queue.db3", true);
}
// Set tmp-file for PageRequest
$this->PageRequest->setTmpFile($this->working_directory . "phpcrawl_" . getmypid() . ".tmp");
// Pass url-priorities to link-cache
$this->LinkCache->addLinkPriorities($this->link_priority_array);
// Pass base-URL to the UrlFilter
$this->UrlFilter->setBaseURL($this->starting_url);
// Add the starting-URL to the url-cache
$this->LinkCache->addUrl(new PHPCrawlerURLDescriptor($this->starting_url));
}
