Filesystem 函数
在线手册：中文 英文
PHP手册

file_get_contents

(PHP 4 >= 4.3.0, PHP 5)

file_get_contents — 将整个文件读入一个字符串

说明

string file_get_contents ( string $filename [, bool $use_include_path [, resource $context [, int $offset [, int $maxlen ]]]] )

和 file() 一样，只除了 file_get_contents() 把文件读入一个字符串。将在参数 offset 所指定的位置开始读取长度为 maxlen 的内容。如果失败，file_get_contents() 将返回 FALSE。

file_get_contents() 函数是用来将文件的内容读入到一个字符串中的首选方法。如果操作系统支持还会使用内存映射技术来增强性能。

Note: 如果要打开有特殊字符的 URL （比如说有空格），就需要使用 urlencode() 进行 URL 编码。

Note:

context 参数可以用 NULL 来忽略。

参数

filename

Name of the file to read.

use_include_path

Note:

As of PHP 5 the FILE_USE_INCLUDE_PATH can be used to trigger include path search.

context

A valid context resource created with stream_context_create(). If you don't need to use a custom context, you can skip this parameter by NULL.

offset

The offset where the reading starts on the original stream.

Seeking (offset) is not supported with remote files. Attempting to seek on non-local files may work with small offsets, but this is unpredictable because it works on the buffered stream.

maxlen

Maximum length of data read. The default is to read until end of file is reached. Note that this parameter is applied to the stream processed by the filters.

返回值

The function returns the read data 或者在失败时返回 FALSE.

错误/异常

An E_WARNING level error is generated if either maxlength is less than zero, or if offset exceeds the length of the stream.

范例

Example #1 Get and output the source of the homepage of a website

<?php
$homepage = file_get_contents('http://www.example.com/');
echo $homepage;
?>

Example #2 Searching within the include_path

<?php
// <= PHP 5
$file = file_get_contents('./people.txt', true);
// > PHP 5
$file = file_get_contents('./people.txt', FILE_USE_INCLUDE_PATH);
?>

Example #3 Reading a section of a file

<?php
// Read 14 characters starting from the 21st character
$section = file_get_contents('./people.txt', NULL, NULL, 20, 14);
var_dump($section);
?>

以上例程的输出类似于：

string(14) "lle Bjori Ro"

Example #4 Using stream contexts

<?php
// Create a stream
$opts = array(
  'http'=>array(
    'method'=>"GET",
    'header'=>"Accept-language: en\r\n" .
              "Cookie: foo=bar\r\n"
  )
);

$context = stream_context_create($opts);

// Open the file using the HTTP headers set above
$file = file_get_contents('http://www.example.com/', false, $context);
?>

更新日志

版本 说明
5.1.0 Added the offset and maxlen parameters.
5.0.0 Added context support.

注释

Note: 此函数可安全用于二进制对象。

Tip

如已启用fopen 包装器，在此函数中， URL 可作为文件名。关于如何指定文件名详见 fopen()。各种 wapper 的不同功能请参见 Supported Protocols and Wrappers，注意其用法及其可提供的预定义变量。

Warning

使用 SSL 时，Microsoft IIS 会违反协议不发送close_notify标记就关闭连接。PHP 会在到达数据尾端时报告“SSL: Fatal Protocol Error”。 要解决此问题，error_reporting 应设定为降低级别至不包含警告。 PHP 4.3.7 及更高版本可以在使用 https:// 包装器打开流时检测出有问题的 IIS 服务器软件 并抑制警告。在使用 fsockopen() 创建 ssl:// 套接字时, 开发者需检测并抑制此警告。

参见

Filesystem 函数
在线手册：中文 英文
PHP手册
PHP手册 - N: 将整个文件读入一个字符串

用户评论:

Mobin Hosseini (16-Mar-2012 07:26)

My answer to the previous question :
file_get_contents() can only GET as you can see.
If you want to POST then the easiest solution would be the curl library!

http://php.net/manual/en/book.curl.php

Anonymous (11-Mar-2012 11:50)

Hi! How can I use file_get_contents() to send a $_FILE variable to a site trough POST method?
If that's not possible, whitch function could do that?

onglipo at hotmail dot com (08-Mar-2012 01:27)

$header = file_get_contents('http://www.example.com/faq.jsp');
echo $header;
Fails with 500 Internal Server Error.

$opts = array('http'=>array('header' => "User-Agent:MyAgent/1.0\r\n"));
$context = stream_context_create($opts);
$header = file_get_contents('http://www.example.com/faq.jsp',false,$context);
//$header = file_get_contents('http://www.polama.com/faq.jsp');
echo $header;

Works!

pperegrina (22-Dec-2011 07:30)

For those having this problem when trying to get_file_contents(url):

Warning: file_get_contents(url): failed to open stream: HTTP request failed!  in xx on line yy

If you are behind a SonicWall firewall, read this:
https://bugs.php.net/bug.php?id=40197
(this little line: uncheck a box in the internal settings of the firewall labled "Enforce Host Tag Search with for CFS")

Apparently by default SonicWall blocks any HTTP request without a "Host:" header, which is the case in the PHP get_file_contents(url) implementation.

This is why, if you try to get the same URL from the same machine with cURL our wget, it works.

I hope this will be useful to someone, it took me hours to find out :)

godwraith01 at yahoo dot com (11-Oct-2011 12:16)

I experienced a problem in using hostnames instead straight IP with some server destinations.

If i use file_get_contents("www.jbossServer.example/app1",...)
i will get an 'Invalid hostname' from the server i'm calling.

This is because file_get_contents probably will rewrite your request after getting the IP, obtaining the same thing as :
file_get_contents("xxx.yyy.www.zzz/app1",...)

And you know that many servers will deny you access if you go through IP addressing in the request.

With cURL this problem doesn't exists. It resolves the hostname leaving the request as you set it, so the server is not rude in response.

fibrefox at dynamicfiles dot de (05-Jul-2011 01:45)

If you want to insert tracking-scripts into your shopping-system, some scripts doesn't support intelligent detection of HTTPS, so i made a script i put on the server that rewrites 'http' to 'https' in the script, assuming everything has to be UTF-8 encoded (as a fallback it makes a redirect).

It is important that the HTTPS-source DOES exist!

<?php

function file_get_contents_utf8($fn) {
    $opts = array(
        'http' => array(
            'method'=>"GET",
            'header'=>"Content-Type: text/html; charset=utf-8"
        )
    );

    $context = stream_context_create($opts);
    $result = @file_get_contents($fn,false,$context);
    return $result;
}
header("Content-Type: text/html; charset=utf-8");
$tPath = "URL YOU WANT TO MODIFY";
$result = file_get_contents_utf8("http://".$tPath);

if( $result == false){
    header("Location: https://".$tPath); // fallback
    exit();
}
else{
    echo mb_ereg_replace("http","https",$result);
}
?>

andre-s at home dot nl (18-Jun-2011 08:46)

If you are getting a failed to open stream message on your windows machine check your hosts file.

127.0.0.1       localhost

must be in it and the IP6 line must be commented

#    ::1             localhost

eric at midkotasolutions dot com (04-Mar-2011 07:26)

At least as of PHP 5.3, file_get_contents no longer uses memory mapping.

See comments on this bug report:

http://bugs.php.net/bug.php?id=52802

Stas Trefilov, OpteamIS (22-Feb-2011 03:30)

here is another (maybe the easiest) way of doing POST http requests from php using its built-in capabilities. feel free to add the headers you need (notably the Host: header) to further customize the request.

note: this method does not allow file uploads. if you want to upload a file with your request you will need to modify the context parameters to provide multipart/form-data encoding (check out http://www.php.net/manual/en/context.http.php ) and build the $data_url following the guidelines on http://www.w3.org/TR/html401/interact/forms.html#h-17.13.4.2

<?php
/**
make an http POST request and return the response content and headers
@param string $url    url of the requested script
@param array $data    hash array of request variables
@return returns a hash array with response content and headers in the following form:
    array ('content'=>'<html></html>'
        , 'headers'=>array ('HTTP/1.1 200 OK', 'Connection: close', ...)
        )
*/
function http_post ($url, $data)
{
    $data_url = http_build_query ($data);
    $data_len = strlen ($data_url);

    return array ('content'=>file_get_contents ($url, false, stream_context_create (array ('http'=>array ('method'=>'POST'
            , 'header'=>"Connection: close\r\nContent-Length: $data_len\r\n"
            , 'content'=>$data_url
            ))))
        , 'headers'=>$http_response_header
        );
}
?>

Yoga Wibowo Aji (08-Feb-2011 05:10)

read text per line and convert to array

for example, the input file is input.txt
the input file containt text below

one
two
three
four
five

++++++++++++++++++++++++++

read value per line

<?php
$data = file_get_contents("input.txt"); //read the file
$convert = explode("\n", $data); //create array separate by new line

for ($i=0;$i<count($convert);$i++) 
{
    echo $convert[$i].', '; //write value by index
}
?>

++++++++++++++++++++++++++

Output :

one, two, three, four, five,

john_schlick at hotmail dot com (27-Jan-2011 08:25)

For those that need to know the default of $maxLen, it's defined in the php source code as PHP_STREAM_COPY_ALL, and because that is not available to us mere mortal users, that constant is defined as  ((size_t)-1) or -1

contact at webapp dot fr (31-Dec-2010 11:57)

When using a URI with a login / password (HTTP or FTP, for an example), you may need to urlencode the password if it contains special characters.
Do not urlencode the whole URI, just the password.

Don't do :
urlencode('ftp://login:mdp%?special@host/dir/file')

Do :
'ftp://login:' . urlencode('mdp%?special') . '@host/dir/file';

Might seem obvious, but is worth noting.

slymak at gmail dot com (06-Dec-2010 11:07)

If working file is bigger than 64kb and you getting deadlock. Your buffer is overflow. Here are two way how to avoid that.
1) use temporary file for descriptor
<?php
  $descriptorspec = array(
   0 => array("file", "/tmp/ens/a.ens","r"),  // stdin is a pipe that the child will read from
   1 => array("file", "/tmp/ens/a.html","w"),  // stdout is a pipe that the child will write to
   2 => array("file", "/tmp/ens/error-output.txt", "a") // stderr is a file to write to
  );
?>

2) inline read using stream_set_blocking. PHP doesn't proper handle last part of file.
<?php
  $READ_LEN = 64*1024;
  $MAX_BUF_LEN = 2*$READ_LEN;

  $url = "http://some.domain.com:5984/".$db."/".$member."/contents";
  $src = fopen($url,"r");

  $cwd = '/tmp';
  $cmd['enscript'] = "/usr/bin/enscript";
  $cmd['enscript-options'] = " -q --language=html --color -Ejcl -o -";
  $descriptorspec = array(
    0 => array("pipe", "r"),  // stdin is a pipe that the child will read from
    1 => array("pipe", "w")   // stdout is a pipe that the child will write to
  );

  $ph=proc_open($cmd['enscript']." ".$cmd['enscript-options'],$descriptorspec,$pipes,$cwd);

  stream_set_blocking($src,0);
  stream_set_blocking($pipes[0],0);
  stream_set_blocking($pipes[1],0);

  $CMD_OUT_OPEN = TRUE; $k = 0;
  while (!feof($pipes[1]) || !feof($src) || $k > 0) {
    if (!feof($src) && $k+$READ_LEN <= $MAX_BUF_LEN) {
      $input .= fread($src,$READ_LEN);
      $k = strlen($input);
    }
    if ($k > 0) {
      $l = fwrite($pipes[0],$input);
      $k -= $l;
      $input = substr($input,$l);
    }
    if ($CMD_OUT_OPEN && $k == 0 && feof($src)) {
      fclose($pipes[0]);
      $CMD_OUT_OPEN = FALSE;
    }
    $output = fread($pipes[1],$READ_LEN);
    $outputn = str_replace("<H1>(stdin)</H1>","",$output);
        echo $outputn;
  }

  fclose($pipes[1]);

  $return_value = proc_close($ph);
?>

Anonymous (04-Dec-2010 03:33)

The offset is 0 based.  Setting it to 1 will skip the first character of the stream.

kam (26-Nov-2010 11:18)

Recently I experienced unexpected behaviour (for me) of file_get_contents('https://......')
Script gets images and file_get_contents('https://......') is in loop operator.
After several iterations Apache Server FastCGI timed out with Internal Server Error in ~30 seconds even if max_execution_time was set to 0. With fopen() fread() pair result was the same.
In different environment Apache mod_php the same script worked without problems.
I solved this changing https requests to http

KrisWebDev (21-Nov-2010 09:17)

If your file_get_contents freezes during several seconds, here is maybe your answer:

Beware that the default keepalive timeout of Apache 2.0 httpd is 15 seconds. This is true for HTTP/1.1 connections, which is not the default behavior of file_get_contents but you can force it, especially if you are trying to act as a web browser. I don't know if this is also the case for HTTP/1.0 connections.

Forcing the server to close the connection would make you gain those 15 seconds in your script:

<?php
$context = stream_context_create(array('http' => array('header'=>'Connection: close')));
$content = file_get_contents("http://www.example.com/test.html");
?>

Another way of resolving slowness issues is to use cURL or fsockopen. Bear in mind that contrary to the behavior of web browsers, file_get_contents doesn't return the result when the web page is fully downloaded (i.e. HTTP payload length = value of the response HTTP "Content-Length" header) but when the TCP connection is closed.
I hope this behavior will change in future releases of PHP.
This has been experienced with PHP 5.3.3.

vlad dot wing at gmail dot com (27-Mar-2010 10:03)

If you want to check if the function returned error, in case of a HTTP request an, it's not sufficient to test it against false. It may happen the return for that HTTP request was empty. In this case it's better to check if the return value is a bool.

<?php
$result=file_get_contents("http://www.example.com");
if ($result === false)
{
    // treat error
} else {
    // handle good case
}
?>

[EDIT BY thiago: Has enhacements from an anonymous user]

rutger at webjin dot nl (14-Jan-2010 02:22)

Sometimes you might get an error opening an http URL.
even though you have set "allow_url_fopen = On" in php.ini

For me the the solution was to also set "user_agent" to something.

ken at wetken dot net (28-Oct-2009 01:00)

On Centos 5, and maybe other Red Hat based systems, any attempt to use file_get_contents to access a URL on an http  port other than 80 (e.g. "http://www.example.com:8040/page") may fail with a permissions violation (error 13) unless the box you are running php on has its seLinux set to 'permissive' not 'enforcing' . Otherwise the request doesn't even get out of the box, i.e. the permissions violation is generated locally by seLinux.

corey at effim dot com (09-Jun-2009 10:35)

In my dev environment with a relatively low-speed drive (standard SATA 7200RPM) reading a 25MB zip file in 10 times...

<?php

$data = `cat /tmp/test.zip`;
// 1.05 seconds

$fh = fopen('/tmp/test.zip', 'r');
$data = fread($fh, filesize('/tmp/test.zip'));
fclose($fh);
// 1.31 seconds

$data = file_get_contents('/tmp/test.zip');
// 1.33 seconds

?>

However, on a 21k text file running 100 iterations...

<?php

$data = `cat /tmp/test.txt`;
// 1.98 seconds

$fh = fopen('/tmp/test.txt', 'r');
$data = fread($fh, filesize('/tmp/test.txt'));
fclose($fh);
// 0.00082 seconds

$data = file_get_contents('/tmp/test.txt');
// 0.0069 seconds

?>

Despite the comment about file_get_contents being faster do to memory mapping, file_get_contents is slowest in both of the above examples. If you need the best performance out of your production box, you might want to throw together a script to check out which method is fastest for what size files on that particular machine, then optimize your code to check the file size and use the appropriate function for it.

colnector bla-at_bla colnect.com (10-Aug-2008 10:34)

A UTF-8 issue I've encountered is that of reading a URL with a non-UTF-8 encoding that is later displayed improperly since file_get_contents() related to it as UTF-8. This small function should show you how to address this issue:

<?php
function file_get_contents_utf8($fn) {
     $content = file_get_contents($fn);
      return mb_convert_encoding($content, 'UTF-8',
          mb_detect_encoding($content, 'UTF-8, ISO-8859-1', true));
}
?>

EOD (07-Aug-2008 01:34)

if $filename has a relative path file_get_contents returns the uninterpreted sourcecode of the php-file with all comments etc.

I don't know whether this is a bug or intented or caused by server-configuration.

I think this behaviour should be included in the description of the function.

daniele dot ricci at staff dot dada dot net (05-Aug-2008 12:13)

I recently upgraded my server to Slackware 12.0.

After this, a program of mine stopped working: the call to file_get_contents (to an URL served by a custom HTTP server) was returning false without generating any error!

After some investigations I saw this: my custom HTTP server closes the connection at the end of the content. This (without the header "Connection: close") seems to cause the problem I described.

To solve the problem I simply added that header to the answer of my custom HTTP server.

joachimb at gmail dot com (15-Apr-2008 10:38)

Setting the timeout properly without messing with ini values:

<?php
$ctx = stream_context_create(array(
    'http' => array(
        'timeout' => 1
        )
    )
);
file_get_contents("http://example.com/", 0, $ctx);
?>

3n1gm4 [at] gmail [dot] com (02-Apr-2008 09:12)

This is a nice and simple substitute to get_file_contents() using curl, it returns FALSE if $contents is empty.

<?php
function curl_get_file_contents($URL)
    {
        $c = curl_init();
        curl_setopt($c, CURLOPT_RETURNTRANSFER, 1);
        curl_setopt($c, CURLOPT_URL, $URL);
        $contents = curl_exec($c);
        curl_close($c);

        if ($contents) return $contents;
            else return FALSE;
    }
?>

Hope this help, if there is something wrong or something you don't understand let me know :)

php [spat] hm2k.org (15-Jan-2008 11:58)

I decided to make a similar function to this, called file_post_contents, it uses POST instead of GET to call, kinda handy...

<?php
function file_post_contents($url,$headers=false) {
    $url = parse_url($url);

    if (!isset($url['port'])) {
      if ($url['scheme'] == 'http') { $url['port']=80; }
      elseif ($url['scheme'] == 'https') { $url['port']=443; }
    }
    $url['query']=isset($url['query'])?$url['query']:'';

    $url['protocol']=$url['scheme'].'://';
    $eol="\r\n";

    $headers =  "POST ".$url['protocol'].$url['host'].$url['path']." HTTP/1.0".$eol.
                "Host: ".$url['host'].$eol.
                "Referer: ".$url['protocol'].$url['host'].$url['path'].$eol.
                "Content-Type: application/x-www-form-urlencoded".$eol.
                "Content-Length: ".strlen($url['query']).$eol.
                $eol.$url['query'];
    $fp = fsockopen($url['host'], $url['port'], $errno, $errstr, 30);
    if($fp) {
      fputs($fp, $headers);
      $result = '';
      while(!feof($fp)) { $result .= fgets($fp, 128); }
      fclose($fp);
      if (!$headers) {
        //removes headers
        $pattern="/^.*\r\n\r\n/s";
        $result=preg_replace($pattern,'',$result);
      }
      return $result;
    }
}
?>

francois hill (03-Dec-2007 04:56)

Seems file looks for the file inside the current working (executing) directory before looking in the include path, even with the FILE_USE_INCLUDE_PATH flag specified.

Same behavior as include actually.

By the way I feel the doc is not entirely clear on the exact order of inclusion (see include). It seems to say the include_path is the first location to be searched, but I have come across at least one case where the directory containing the file including was actually the first to be searched.

Drat.

bearachute at gmail dot com (11-Jul-2007 10:38)

If you're having problems with binary and hex data:

I had a problem when trying to read information from a ttf, which is primarily hex data. A binary-safe file read automatically replaces byte values with their corresponding ASCII characters, so I thought that I could use the binary string when I needed readable ASCII strings, and bin2hex() when I needed hex strings.

However, this became a problem when I tried to pass those ASCII strings into other functions (namely gd functions). var_dump showed that a 5-character string contained 10 characters, but they weren't visible. A binary-to-"normal" string conversion function didn't seem to exist and I didn't want to have to convert every single character in hex using chr().

I used unpack with "c*" as the format flag to see what was going on, and found that every other character was null data (ordinal 0). To solve it, I just did

str_replace(chr(0), "", $string);

which did the trick.

This took forever to figure out so I hope this helps people reading from hex data!

tobsn at php dot net (02-May-2007 05:26)

you'll find the http response headers in: $http_response_header

;o)

Greg Ambrose (greg at catalina-it dot com dot au) (17-Apr-2007 05:37)

[Editors note: As of PHP 5.2.1 you can specify `timeout` context option and pass the context to file_get_contents()]

The only way I could get get_file_contents() to wait for a very slow http request was to set the socket timeout as follows.

 ini_set('default_socket_timeout',    120);   
$a = file_get_contents("http://abcxyz.com");

Other times like execution time and input time had no effect.

siegfri3d at gmail dot com (05-Dec-2006 08:52)

Use the previous example if you want to request the server for a special part of the content, IF and only if the server accepts the method.
If you want a simple example to ask the server for all the content, but only save a portion of it, do it this way:
<?php
$content=file_get_contents("http://www.google.com",FALSE,NULL,0,20);
echo $content;
?>

This will echo the 20 first bytes of the google.com source code.

fcicqbbs at gmail dot com (04-Aug-2006 09:55)

the bug #36857 was fixed.
http://bugs.php.net/36857

Now you may use this code,to fetch the partial content like this:
<?php
$context=array('http' => array ('header'=> 'Range: bytes=1024-', ),);
$xcontext = stream_context_create($context);
$str=file_get_contents("http://www.fcicq.net/wp/",FALSE,$xcontext);
?>
that's all.

richard dot quadling at bandvulc dot co dot uk (15-Nov-2005 10:47)

If, like me, you are on a Microsoft network with ISA server and require NTLM authentication, certain applications will not get out of the network. SETI@Home Classic and PHP are just 2 of them.

The workaround is fairly simple.

First you need to use an NTLM Authentication Proxy Server. There is one written in Python and is available from http://apserver.sourceforge.net/. You will need Python from http://www.python.org/.

Both sites include excellent documentation.

Python works a bit like PHP. Human readable code is handled without having to produce a compiled version. You DO have the opportunity of compiling the code (from a .py file to a .pyc file).

Once compiled, I installed this as a service (instsrv and srvany - parts of the Windows Resource Kit), so when the server is turned on (not logged in), the Python based NTLM Authentication Proxy Server is running.

Then, and here is the bit I'm really interested in, you need to tell PHP you intend to route http/ftp requests through the NTLM APS.

To do this, you use contexts.

Here is an example.

<?php

// Define a context for HTTP.
$aContext = array(
    'http' => array(
        'proxy' => 'tcp://127.0.0.1:8080', // This needs to be the server and the port of the NTLM Authentication Proxy Server.
        'request_fulluri' => True,
        ),
    );
$cxContext = stream_context_create($aContext);

// Now all file stream functions can use this context.

$sFile = file_get_contents("http://www.php.net", False, $cxContext);

echo $sFile;
?>

Hopefully this helps SOMEONE!!!

aidan at php dot net (31-Jan-2005 07:23)

This functionality is now implemented in the PEAR package PHP_Compat.

More information about using this function without upgrading your version of PHP can be found on the below link:

http://pear.php.net/package/PHP_Compat