/**
* 初始化采集器
* @param $entranceUrl
* @param $savePath
* @param array $allowCaptureHosts
* @param $whitelistUrls
* @param $blacklistUrls
* @param $urlPatterns
* @param $onlyCaptureUrlPatterns
*/
protected function initializeCollector($entranceUrl, $savePath, array $allowCaptureHosts, array $whitelistUrls, array $blacklistUrls, array $urlPatterns, $onlyCaptureUrlPatterns)
{
$this->collector->setRawEntranceUrl($entranceUrl);
$this->collector->setSavePath($savePath);
if (!empty($allowCaptureHosts)) {
$this->collector->setAllowedCaptureHosts($allowCaptureHosts);
}
if (!empty($whitelistUrls)) {
$this->collector->setWhitelistUrls($whitelistUrls);
}
if (!empty($blacklistUrls)) {
$this->collector->setBlacklistUrls($blacklistUrls);
}
if (!empty($urlPatterns)) {
$this->collector->setUrlPatterns($urlPatterns);
}
$this->collector->setOnlyCaptureUrlPatterns($onlyCaptureUrlPatterns);
}
function testRun()
{
$this->collector->setRawEntranceUrl('http://minimal.ondrejsvestka.cz/1-3-3/index.html');
$this->collector->run();
$this->assertNotEmpty('1');
}
<?php
include __DIR__ . '/../vendor/autoload.php';
use Slince\Collector\Collector;
use Slince\Event\Event;
$collector = new Collector(__DIR__ . '/html', 'http://demo.sc.chinaz.com/Files/DownLoad/moban/201604/moban1178/index.html');
$collector->getDispatcher()->bind(Collector::EVENT_CAPTURE_URL_REPOSITORY, function (Event $event) {
$repository = $event->getArgument('repository');
echo 'Begin Capture ', $repository->getUrl()->getUrlString(), "\r\n";
});
$collector->getDispatcher()->bind(Collector::EVENT_CAPTURED_URL_REPOSITORY, function (Event $event) {
$repository = $event->getArgument('repository');
echo $repository->getUrl()->getUrlString() . " Captured OK!\r\n";
});
$collector->run();
