首页导航博客电子书算法众创代码随贴关于我们

curl保存cookie,https采集要用到函数取两个串之间内容替换两个串之间内容

代码正文双击正文可选择全部

1[代码][php]

<?php
$debug=1;
$debug ?	error_reporting(E_ALL ^ E_NOTICE) : error_reporting(0);
@date_default_timezone_set('Etc/GMT-8');

echo getSourcesHttps('http://218.249.164.250:2048/login','http://218.249.164.250:2048/menu',"user=20070017&pass=a147258369&url=^U");
echo '<script type="text/javascript">setTimeout(\'location.href="http://218.249.164.250:2048/login?url=http://www.cnki.net"\',3000);</script>';

/**
 * 抓取一个页面
 * @param   str		$url= "https://cn-virtual.excalibur.com.hk:5512/spadmin/english/validate.php"; 需要登陆验证的页面地址
 * @param   str		$get_content_url	采集登陆后的页面
 * @param   str		$post_fields= "username=FENGKONG&password=368382368382"; post串
 * @param   str		$User_Agent = "Mozilla/5.0 (compatible; Baiduspider/2.0; +http://www.baidu.com/search/spider.html)";   user_agent信息 如“baiduspider”或“googlebot”
 * @param   str		$Referer_Url = 'http://www.baidu.com/';  来源页
 */
function getSourcesHttps($Url='http://218.249.164.250:2048/login',$get_content_url='http://218.249.164.250:2048/menu',$post_fields= "user=20070017&pass=a147258369&url=^U",$User_Agent='Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/45.0.2454.101 Safari/537.36',$Referer_Url='https://www.baidu.com/') //抓取某个指定的页面 
{
	$cookie_file=tempnam('./temp','cookie');
	$ch = curl_init($Url);
	curl_setopt($ch, CURLOPT_HEADER, 0);
	curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
	curl_setopt($ch, CURLOPT_USERAGENT, $User_Agent); 
	curl_setopt($ch, CURLOPT_REFERER, $Referer_Url); 
	#curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false); 
	#curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, false);
	curl_setopt($ch, CURLOPT_POST, 1);
	curl_setopt($ch, CURLOPT_POSTFIELDS, $post_fields);
	curl_setopt($ch, CURLOPT_FOLLOWLOCATION,1);
	curl_setopt($ch, CURLOPT_COOKIEJAR, $cookie_file);
	$MySources1 = curl_exec ($ch);
	curl_close($ch); 
	$url=$get_content_url;
	$ch = curl_init($url);
	curl_setopt($ch, CURLOPT_HEADER, 0);
	curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
	curl_setopt($ch, CURLOPT_USERAGENT, $User_Agent); 
	curl_setopt($ch, CURLOPT_REFERER, $Url); 
	#curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false); 
	#curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, false);
	curl_setopt($ch, CURLOPT_COOKIEFILE, $cookie_file);
	$MySources2 = curl_exec($ch);
	curl_close($ch);	
	return $MySources2;
}

/**
 * 取两个串之间内容 如参数一个为空返回为空
 * @param   str		first str;
 * @param   str		second str;
 * @param   str		haystack; 其中没有first或second 返回空
 * @return  str		成功返回串，失败返回空串
 */
function sub_getstr($f,$s,$h)
{
	if(empty($f) ||empty($s) ||empty($h)) return '';
	list($part1,$part2)=explode($f,$h,2);
	empty($part2) ? $get='' : list($get,$tem)=explode($s,$part2,2);
	if($get===$part2) $get='';
	return empty($get) ? '' : $get; 
}
/**
 * 替换两个串之间内容   把1 2之间内容换成 3
 * @param   str		first str;
 * @param   str		second str;
 * @param   str		repalce str;
 * @param   str		haystack;
 * @return  str		成功返回串，失败返回空串
 */
function sub_replace($f,$s,$r,$h)
{
	return str_replace(sub_getstr($f,$s,$h),$r,$h);
}

/**
 * 正则取一个串
 * @param   str    $pattern  '/a(b)c/'
 * @return  str
 */
function preg_get($pattern,$file)
{
	preg_match($pattern,$file,$matches);
	return $matches;
}
/**
 * 正则取多个串 返回所有匹配
 * @param   str    $pattern  '/a(b)(c)d/'
 * @return  str
 */
function preg_get_all($pattern,$file)
{
	preg_match_all($pattern,$file,$matches);
	return $matches;
}

curl保存cookie,https采集要用到函数 取两个串之间内容 替换两个串之间内容

代码正文双击正文可选择全部

curl保存cookie,https采集要用到函数取两个串之间内容替换两个串之间内容