我测试this code下载10 MB块大文件:当下载带有“FWRITE”二进制文件 - “0D”字节消失
/**
* Copy remote file over HTTP one small chunk at a time.
*
* @param $infile The full URL to the remote file
* @param $outfile The path where to save the file
*/
function copyfile_chunked($infile, $outfile) {
$chunksize = 10 * (1024 * 1024); // 10 Megs
/**
* parse_url breaks a part a URL into it's parts, i.e. host, path,
* query string, etc.
*/
$parts = parse_url($infile);
$i_handle = fsockopen($parts['host'], 80, $errstr, $errcode, 5);
$o_handle = fopen($outfile, 'wb');
if ($i_handle == false || $o_handle == false) {
return false;
}
if (!empty($parts['query'])) {
$parts['path'] .= '?' . $parts['query'];
}
/**
* Send the request to the server for the file
*/
$request = "GET {$parts['path']} HTTP/1.1\r\n";
$request .= "Host: {$parts['host']}\r\n";
$request .= "User-Agent: Mozilla/5.0\r\n";
$request .= "Keep-Alive: 115\r\n";
$request .= "Connection: keep-alive\r\n\r\n";
fwrite($i_handle, $request);
/**
* Now read the headers from the remote server. We'll need
* to get the content length.
*/
$headers = array();
while(!feof($i_handle)) {
$line = fgets($i_handle);
if ($line == "\r\n") break;
$headers[] = $line;
}
/**
* Look for the Content-Length header, and get the size
* of the remote file.
*/
$length = 0;
foreach($headers as $header) {
if (stripos($header, 'Content-Length:') === 0) {
$length = (int)str_replace('Content-Length: ', '', $header);
break;
}
}
/**
* Start reading in the remote file, and writing it to the
* local file one chunk at a time.
*/
$cnt = 0;
while(!feof($i_handle)) {
$buf = '';
$buf = fread($i_handle, $chunksize);
$bytes = fwrite($o_handle, $buf);
if ($bytes == false) {
return false;
}
$cnt += $bytes;
/**
* We're done reading when we've reached the conent length
*/
if ($cnt >= $length) break;
}
fclose($i_handle);
fclose($o_handle);
return $cnt;
}
我第一次一个小图像上测试该代码。图像被下载到我的帐户,但以一种破损的形式:所有的字节看起来是正确的,从下载的图像中删除“0D”字节,这使得它不可用。
为什么会发生这种情况,我该如何克服它?
谢谢!
您最终读取的字节数与“Content-Length”指示的一样多吗? – Jon
嗨,乔恩。奇怪的事情。原始图像是15444字节,该函数返回15444字节被下载,但是当我检索下载的图像时,它证明只有15397字节,可能是因为缺少“0D”字节。 – GreenBear
是否有一个特别的原因,你不使用一个工作的http客户端库?你古怪的头解码不符合HTTP。并非所有的答案都来自于分块的TE。 – mario