php网页爬虫

<?php

/*网页爬虫*/

#最简单的使用,属性都采用默认值

/*

$curl=curl_init('http://www.baidu.com');

$output=curl_exec($curl);

curl_close($curl);

echo $output;

*/

#稍微复杂一点的，对页面进行操作

/*

$curl=curl_init();

curl_setopt($curl, CURLOPT_URL, 'http://www.baidu.com');//可以动态改变网址

curl_setopt($curl, CURLOPT_RETURNTRANSFER, true);//不直接打印在浏览器上

$output=curl_exec($curl);//执行

curl_close($curl);//关闭

echo str_replace('百度', '虎哥', $output);

*/

#更复杂一点的,模拟POST请求

/*

$data='theCityName=北京';

$curl=curl_init();

curl_setopt($curl, CURLOPT_URL,'http://www.webxml.com.cn/WebServices/WeatherWebService.asmx/getWeatherbyCityName');

curl_setopt($curl,CURLOPT_HEADER, 0);

curl_setopt($curl, CURLOPT_RETURNTRANSFER, 1);

curl_setopt($curl,CURLOPT_POST, 1);

curl_setopt($curl,CURLOPT_POSTFIELDS,$data);

curl_setopt($curl,CURLOPT_HTTPHEADER,array('application/x-www-form-rulencode;charset=utf-8','Content-length'.strlen($data)));

$output=curl_exec($curl);

if(!curl_errno($curl)){

//$info=curl_getinfo($curl);

//print_r($info);

header('Content-type:text/html;charset=utf-8;');

echo $output;

}else{

echo 'Curl Error:'.curl_error($curl);

}

curl_close($curl);

*/

#抓取需要登录的页面

/*

$data='username=2232839753@qq.com&password=19900417&remember=1';

$curl=curl_init();

curl_setopt($curl,CURLOPT_URL,'http://www.imooc.com/user/login');

curl_setopt($curl,CURLOPT_RETURNTRANSFER, true);//执行之后不直接打印出来

//Cookie相关设置，这部分设置需要在所有回话开始之前设置

date_default_timezone_set('PRC');//使用cookie时必须设置时区

curl_setopt($curl,CURLOPT_COOKIESESSION, true);

curl_setopt($curl,CURLOPT_COOKIEFILE,'cookiefile');

curl_setopt($curl,CURLOPT_COOKIEJAR,'cookiefile');

curl_setopt($curl,CURLOPT_COOKIE, session_name().'='.session_id());

curl_setopt($curl,CURLOPT_HEADER,0);//禁止输出头信息

curl_setopt($curl,CURLOPT_FOLLOWLOCATION, 1);

curl_setopt($curl,CURLOPT_POST, 1);

curl_setopt($curl,CURLOPT_POSTFIELDS, $data);

curl_setopt($curl,CURLOPT_HTTPHEADER,array('application/x-www-form-rulencode;charset=utf-8','Content-length:'.strlen($data)));

curl_exec($curl);//执行

curl_setopt($curl,CURLOPT_URL,'http://www.imooc.com/space/index');

curl_setopt($curl,CURLOPT_POST,0);//下载页面不是post请求，故置0

curl_setopt($curl,CURLOPT_HTTPHEADER,array('Content-type:text/xml'));

$output=curl_exec($curl);

echo $output;

*/

#从ftp服务器上下载一个文件

/*

$curl=curl_init();

curl_setopt($curl,CURLOPT_URL,'ftp://10.14.12.22:21/test.txt');

curl_setopt($curl,CURLOPT_HEADER,0);//禁止输出头信息，调试时，可以输出

curl_setopt($curl,CURLOPT_RETURNTRANSFER, 1);

curl_setopt($curl,CURLOPT_TIMEOUT,300);//设置超时时间，单位秒

curl_setopt($curl,CURLOPT_USERPWD,"admin:123");//ftp用户名，密码

$outfile=fopen('mytest.txt', 'wb');//保存到本地的文件名

curl_setopt($curl,CURLOPT_FILE,$outfile);

$output=curl_exec($curl);

fclose($outfile);

if(!curl_errno($curl)){

//$info=curl_getinfo($curl);

//print_r($info);

echo "RETURN:".$output;

}else{

echo 'Curl error:'.curl_error($curl);

}

curl_close($curl);

*/

#ftp文件上传

/*

$curl=curl_init();

$uploadfile='demo1.php';

$fp=fopen($uploadfile,'r');

curl_setopt($curl, CURLOPT_URL, 'ftp://10.14.12.22:21/uploadfile.php');

curl_setopt($curl, CURLOPT_HEADER,0);

curl_setopt($curl,CURLOPT_RETURNTRANSFER,1);

curl_setopt($curl, CURLOPT_TIMEOUT,300);

curl_setopt($curl,CURLOPT_USERPWD,'admin:123');

curl_setopt($curl,CURLOPT_UPLOAD,1);

curl_setopt($curl, CURLOPT_INFILE, $fp);

curl_setopt($curl,CURLOPT_INFILESIZE,filesize($uploadfile));

$output=curl_exec($curl);

fclose($fp);

if(!curl_errno($curl)){

echo 'uploadfile successfully';

}else{

echo 'uploadfile error:'.curl_error($curl);

}

curl_close($curl);

*/

#curl访问https上的资源

/*

$curl=curl_init();

curl_setopt($curl,CURLOPT_URL,'https://ajax.aspnetcdn.com/ajax/jquery.validate/1.12.0/jquery.validate.js');

curl_setopt($curl,CURLOPT_RETURNTRANSFER,true);

date_default_timezone_set('PRC');

curl_setopt($curl,CURLOPT_SSL_VERIFYPEER,0);//终止从服务端进行验证

$output=curl_exec($curl);

curl_close($curl);

echo $output;

*/

//php官方文档网址：http://php.net/manual/zh/book.curl.php

?>