/**
* @return string
*/
public function __toString()
{
return sprintf('%s:%s', $this->scraper->getOrigin()->getName(), $this->originalUrl);
}
/**
* @param ScraperEntity $scraperEntity
* @param bool $disableLimit
*
* @return ScraperInterface
*/
protected function createScraper(ScraperEntity $scraperEntity, $disableLimit = false)
{
if (!array_key_exists($scraperEntity->getId(), $this->scrapers)) {
$scraper = $this->factory->createScraper($scraperEntity);
if ($disableLimit) {
$limit = $scraper->getCrawler()->getRateLimit();
if ($limit instanceof EnablingRateLimitInterface) {
$limit->disable();
}
}
$this->scrapers[$scraperEntity->getId()] = $scraper;
}
return $this->scrapers[$scraperEntity->getId()];
}
/**
* Returns a unique hash for a scraper.
*
* @param Scraper $scraper
*
* @return string
*/
protected function getScraperHash(Scraper $scraper)
{
return md5('scraper' . $scraper->getId());
}
/**
* @inheritdoc
*/
public function findSourceByScraper(Scraper $scraper, $originalId)
{
// look for mapping
$params = ['scraper' => $scraper->getId(), 'originalId' => $originalId];
return $this->getRepository()->findOneBy($params);
}
/**
* @param ScraperEntity $scraper
*
* @return ParserInterface
*/
protected function createParser(ScraperEntity $scraper)
{
$options = array_merge(['scraper' => $scraper], $scraper->getParserOptions());
$parserType = $this->getParserType($scraper->getParser());
$builder = new ParserBuilder($this->eventDispatcher);
return $builder->build($parserType, $options);
}
/**
* @param Scraper $scraper
*
* @return ParserInterface
*/
protected function getParser(Scraper $scraper)
{
$parserType = $this->get('tree_house.io.scrape.scraper_factory')->getParserType($scraper->getParser());
$options = array_merge(['scraper' => $scraper], $scraper->getParserOptions());
return (new ParserBuilder())->build($parserType, $options);
}
