static function getInstance()
{
if (self::$instance == null) {
self::$instance = new MultiCurl();
}
return self::$instance;
}
private static function MultiCurlExec($requestList)
{
MultiCurl::instance()->open();
MultiCurl::instance()->send($requestList);
$responseList = MultiCurl::instance()->exec();
MultiCurl::instance()->close();
return $responseList;
}
public function curl_multi_add_handle()
{
// create both cURL resources
$ch1 = new Curl();
$ch2 = new Curl();
// set URL and other appropriate options
$ch1->setUrl("http://www.example.com/")->includeHeader(false);
$ch2->setUrl("http://www.php.net/")->includeHeader(false);
//create the multiple cURL handle
$mh = new MultiCurl();
//add the two handles
$mh->add($ch1)->add($ch2);
$running = null;
//execute the handles
do {
curl_multi_exec($mh->getMultiHandle(), $running);
} while ($running > 0);
//close all the handles
unset($mh);
}
private function flatten($html)
{
$multi = MultiCurl::getInstance();
$hdls = array();
foreach (array('js', 'css') as $type) {
$hdls[$type] = array();
${$type} = '';
}
/* scripts */
preg_match_all("/<script[^<>]+src=['\"]([^<>'\"]+)['\"][^<>]*>(.*)<\\/script>/i", $html, $scripts);
$html = str_replace($scripts[0], '', $html);
foreach ($scripts[1] as $url) {
$hdl = $this->getCurlHandleForUrl($url);
$multi->add($hdl);
$hdls['js'][] = (string) $hdl;
}
/* stylesheets */
preg_match_all("/<link[^<>]+href=['\"]([^<>'\"]+.css)['\"][^<>]*[\\/]*>((<\\/link>)*)/i", $html, $links);
// This could do better by checking for the rel attribute...
$html = str_replace($links[0], '', $html);
foreach ($links[1] as $url) {
$hdl = $this->getCurlHandleForUrl($url);
$multi->add($hdl);
$hdls['css'][] = (string) $hdl;
}
/* append results */
foreach ($hdls as $key => $ids) {
foreach ($ids as $id) {
$res = $multi->get($id);
if ($res) {
${$key} .= $res;
}
}
}
if ($js) {
$html = '<script>' . $js . '</script>' . $html;
}
if ($css) {
$html = '<style>' . $css . '</style>' . $html;
}
return $html;
}
/**
* 添加到MultiCurl中去
* @param MultiCurl $multiCurl
* @return int
*/
public function addToMultiCurl(MultiCurl $multiCurl)
{
$this->multiCurlAddedTo = $multiCurl;
return curl_multi_add_handle($multiCurl->getHandle(), $this->getHandle());
}
public function actionTestMultiCurlInstantGetAndNewOne()
{
$mc = new MultiCurl();
for ($i = 0; $i < 10; $i++) {
$mc->addUrl("http://test/testMultiCurl/longTask", array('interval' => 300, 'times' => 30, 'name' => 'Jack' . $i), 'GET');
}
$mc->exec();
$mc->wait(function ($singleCurl) use($i, $mc) {
echo "URL: " . $singleCurl->getUrl() . ' <br>';
echo "CONTENT: " . $singleCurl->getResult() . ' <br>';
for (; $i < 20; $i++) {
$mc->addUrl("http://test/testMultiCurl/longTask", array('interval' => 300, 'times' => 30, 'name' => 'Tom-' . $i), 'GET');
}
});
echo "<hr> after all: <br/>";
$results = $mc->getResults();
foreach ($results as $result) {
echo "URL: " . $result->getUrl() . ' <br>';
echo "CONTENT: " . $result->getResult() . ' <br>';
}
}
function getContentFromWebPage($productUrls)
{
$options = array(CURLOPT_USERAGENT => $this->setUserAgent(), CURLOPT_HTTPHEADER => $this->setHeader(), CURLOPT_FOLLOWLOCATION => true, CURLOPT_REFERER => $this->merchantUrl);
return MultiCurl::request($productUrls, $options);
}
function _pugpig_package_download_batch($heading, $entries, $debug, $concurrent)
{
print_r("<h3>Downloading " . $heading . " - " . count($entries) . " files</h3>");
if ($debug) {
var_dump($entries);
}
// Check the URLs we're about to download are real URLs
$format_failures = array();
foreach (array_keys($entries) as $entry) {
if (!filter_var($entry, FILTER_VALIDATE_URL)) {
$format_failures[$entry] = "Invalid URL format: " . $entry;
unset($entries[$entry]);
} else {
$path = $entries[$entry];
if (strpos($path, "?") || strpos($path, "&")) {
$format_failures[$entry] = "Invalid local disk path: " . $path;
unset($entries[$entry]);
}
}
}
_pugpig_package_show_failures($format_failures);
// Both of these should become a settings, as well as timeout value
$warning_file_size = 150 * 1024;
// 150 Kb
$mc = new MultiCurl($entries, $concurrent, $warning_file_size);
$mc->process();
$failures = $mc->getFailures();
if (count($failures) > 0) {
_pugpig_package_show_failures($failures);
_print_immediately('<b>Aborting</b><br /><a href="javascript:location.reload(true);">Refresh this page to reload and try again. (It will resume from where it last succeeded.)</a><br />');
exit;
} else {
print_r("<em>Done</em><br />");
}
return $entries;
}
public function __construct($key)
{
$this->key = $key;
$this->epiCurl = MultiCurl::getInstance();
}
function _pugpig_package_download_batch($heading, $entries, $debug = false, $concurrent = 3)
{
print_r("<h3>Downloading " . $heading . " - " . count($entries) . " files</h3>");
// Check the URLs we're about to download are real URLs
$format_failures = array();
foreach (array_keys($entries) as $entry) {
if (!_pugpig_package_validate_download_url($entry, $entries[$entry], $format_failures)) {
unset($entries[$entry]);
}
}
_pugpig_package_show_failures($format_failures);
// Both of these should become a settings, as well as timeout value
$warning_file_size = 150 * 1024;
// 150 Kb
$mc = new MultiCurl($entries, $concurrent, $warning_file_size);
$mc->process();
$successes = $mc->getSuccesses();
$failures = $mc->getFailures();
_pugpig_show_batch($successes);
// _pugpig_array_to_table($successes);
if (count($failures) > 0) {
_pugpig_package_show_failures($failures);
_print_immediately('<b>Aborting</b><br /><a href="javascript:location.reload(true);">Refresh this page to reload and try again. (It will resume from where it last succeeded.)</a><br />');
exit;
} else {
print_r("<em>Done</em><br />");
}
return $entries;
}
public function testSimple()
{
$this->markTestSkipped('Looks bad');
$multi = new MultiCurl();
$multi->setCompleteCallback(function ($curl) {
$info = $curl->getInfo();
# printf("%.2F: %s\n", $info['total_time'], $info['url']);
});
$curl = new Curl();
$curl->setUrl('http://www.golem.de/');
$response1 = $curl->getResponse();
$curl->execute();
$multi->addCurl($curl);
$curl = new Curl();
$curl->setUrl('http://www.heise.de/');
$response2 = $curl->getResponse();
$multi->addCurl($curl);
$curl = new Curl();
$curl->setUrl('http://www.example.com/');
$response3 = $curl->getResponse();
$multi->addCurl($curl);
$curl = new Curl();
$curl->setUrl('http://localhost/');
$response4 = $curl->getResponse();
$multi->addCurl($curl);
$curl = new Curl();
$curl->setUrl('dasdasdasfdf432rwedfs://localhost/');
$response5 = $curl->getResponse();
$multi->addCurl($curl);
$index = 0;
while ($multi->active) {
$index++;
$multi->step();
if ($index == 1) {
$curl = new Curl();
$curl->setUrl('http://www.example.com/1');
$multi->addCurl($curl);
}
if ($index == 1) {
$curl = new Curl();
$curl->setUrl('http://www.example.com/2');
$multi->addCurl($curl);
}
if ($index == 3) {
$curl = new Curl();
$curl->setUrl('http://www.example.com/3');
$multi->addCurl($curl);
}
}
}
