2016-12-27 14:30:54 +13:00
|
|
|
<?php
|
|
|
|
|
/**
|
|
|
|
|
* @package WP Static HTML Output
|
|
|
|
|
*
|
|
|
|
|
* Copyright (c) 2011 Leon Stafford
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
class StaticHtmlOutput_UrlRequest
|
|
|
|
|
{
|
|
|
|
|
protected $_url;
|
|
|
|
|
|
|
|
|
|
protected $_response;
|
|
|
|
|
|
2016-12-27 22:04:57 +13:00
|
|
|
public function __construct($url, $cleanMeta = false)
|
2016-12-27 14:30:54 +13:00
|
|
|
{
|
2017-03-16 21:41:25 +13:00
|
|
|
$this->_url = filter_var(trim($url), FILTER_VALIDATE_URL);
|
2016-12-27 22:04:57 +13:00
|
|
|
$this->_cleanMeta = $cleanMeta;
|
2017-03-16 21:41:25 +13:00
|
|
|
|
2016-12-27 14:30:54 +13:00
|
|
|
$response = wp_remote_get($this->_url,array('timeout'=>300)); //set a long time out
|
2017-03-16 21:41:25 +13:00
|
|
|
|
|
|
|
|
$this->_response = '';
|
|
|
|
|
|
|
|
|
|
if (is_wp_error($response)) {
|
|
|
|
|
error_log('WP_ERROR');
|
|
|
|
|
error_log(print_r($response, true));
|
|
|
|
|
$this->_response = 'FAIL';
|
|
|
|
|
} else {
|
|
|
|
|
$this->_response = $response;
|
|
|
|
|
}
|
|
|
|
|
|
2016-12-27 14:30:54 +13:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
public function getUrl()
|
|
|
|
|
{
|
|
|
|
|
return $this->_url;
|
|
|
|
|
}
|
2017-03-16 21:41:25 +13:00
|
|
|
|
|
|
|
|
public function checkResponse()
|
|
|
|
|
{
|
|
|
|
|
return $this->_response;
|
|
|
|
|
}
|
2016-12-27 14:30:54 +13:00
|
|
|
|
|
|
|
|
public function setResponseBody($newBody)
|
|
|
|
|
{
|
|
|
|
|
if (is_array($this->_response))
|
|
|
|
|
{
|
|
|
|
|
$this->_response['body'] = $newBody;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
public function getResponseBody()
|
|
|
|
|
{
|
|
|
|
|
return isset($this->_response['body']) ? $this->_response['body'] : '';
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
public function getContentType()
|
|
|
|
|
{
|
|
|
|
|
return isset($this->_response['headers']['content-type']) ? $this->_response['headers']['content-type'] : null;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
public function isHtml()
|
|
|
|
|
{
|
|
|
|
|
return stripos($this->getContentType(), 'html') !== false;
|
|
|
|
|
}
|
2017-03-22 21:46:26 +13:00
|
|
|
|
|
|
|
|
public function isRewritable()
|
|
|
|
|
{
|
|
|
|
|
$contentType = $this->getContentType();
|
|
|
|
|
return (stripos($contentType, 'html') !== false) || (stripos($contentType, 'text') !== false);
|
|
|
|
|
}
|
2016-12-27 14:30:54 +13:00
|
|
|
|
|
|
|
|
public function cleanup()
|
|
|
|
|
{
|
|
|
|
|
if ($this->isHtml())
|
|
|
|
|
{
|
2016-12-27 22:04:57 +13:00
|
|
|
if ($this->_cleanMeta)
|
|
|
|
|
{
|
|
|
|
|
$responseBody = preg_replace('/<link rel=["\' ](pingback|alternate|EditURI|wlwmanifest|index|profile|prev)["\' ](.*?)>/si', '', $this->getResponseBody());
|
|
|
|
|
$responseBody = preg_replace('/<meta name=["\' ]generator["\' ](.*?)>/si', '', $responseBody);
|
|
|
|
|
$this->setResponseBody($responseBody);
|
|
|
|
|
}
|
2016-12-27 14:30:54 +13:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
public function extractAllUrls($baseUrl)
|
|
|
|
|
{
|
|
|
|
|
$allUrls = array();
|
2017-03-21 17:47:30 +13:00
|
|
|
|
|
|
|
|
|
|
|
|
|
// TODO: will this follow urls for JS/CSS easily by adjusting?
|
2016-12-27 14:30:54 +13:00
|
|
|
if ($this->isHtml() && preg_match_all('/' . str_replace('/', '\/', $baseUrl) . '[^"\'#\? ]+/i', $this->_response['body'], $matches))
|
|
|
|
|
{
|
|
|
|
|
$allUrls = array_unique($matches[0]);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return $allUrls;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
public function replaceBaseUlr($oldBaseUrl, $newBaseUrl)
|
|
|
|
|
{
|
2017-03-22 21:46:26 +13:00
|
|
|
if ($this->isRewritable())
|
2016-12-27 14:30:54 +13:00
|
|
|
{
|
|
|
|
|
$responseBody = str_replace($oldBaseUrl, $newBaseUrl, $this->getResponseBody());
|
|
|
|
|
$responseBody = str_replace('<head>', "<head>\n<base href=\"" . esc_attr($newBaseUrl) . "\" />\n", $responseBody);
|
2017-03-22 21:03:22 +13:00
|
|
|
|
|
|
|
|
/* fix for cases where URL has been escaped/modified, ie
|
|
|
|
|
http:\/\/banana.com\/
|
|
|
|
|
//banana.com
|
|
|
|
|
https:// -> http:// */
|
|
|
|
|
$oldDomain = parse_url($oldBaseUrl)['host'];
|
|
|
|
|
$newDomain = parse_url($newBaseUrl)['host'];
|
|
|
|
|
$responseBody = str_replace($oldDomain, $newDomain, $responseBody);
|
|
|
|
|
|
2016-12-27 14:30:54 +13:00
|
|
|
$this->setResponseBody($responseBody);
|
|
|
|
|
}
|
|
|
|
|
}
|
2016-12-27 22:04:57 +13:00
|
|
|
}
|
