如何使用php的curl库并行下载一个文件的多个部分?


How can I download multiple parts of a file in parallel with PHP's curl library?

我决定开始一个关于在PHP中使用curl加速下载的项目,使用curl_multi函数

下面是我的代码:
set_time_limit(0);
error_reporting(E_ALL);
$fileurl = "http://hq-scenes.com/tv.exe";
$filename = basename($fileurl);
$size = getFileSize($fileurl);
$splits = range(0, $size, round($size/5));
$megaconnect = curl_multi_init();
$partnames = array();
for ($i = 0; $i < sizeof($splits); $i++) {
    $ch[$i] = curl_init();
    curl_setopt($ch[$i], CURLOPT_URL, $fileurl);
    curl_setopt($ch[$i], CURLOPT_RETURNTRANSFER, 0); 
    curl_setopt($ch[$i], CURLOPT_FOLLOWLOCATION, 0); 
    curl_setopt($ch[$i], CURLOPT_VERBOSE, 1);
    curl_setopt($ch[$i], CURLOPT_BINARYTRANSFER, 1);
    curl_setopt($ch[$i], CURLOPT_FRESH_CONNECT, 0);
    curl_setopt($ch[$i], CURLOPT_CONNECTTIMEOUT, 10);
    $partnames[$i] = $filename . $i;
    $bh[$i] = fopen(getcwd(). '/' . $partnames[$i], 'w+');
    curl_setopt($ch[$i], CURLOPT_FILE, $bh[$i]);
        $x = ($i == 0 ? 0 : $splits[$i]+1);
        $y = ($i == sizeof($splits)-1 ? $size : $splits[$i+1]);
        $range = $x.'-'.$y;
    curl_setopt($ch[$i], CURLOPT_RANGE, $range);
    curl_setopt($ch[$i], CURLOPT_USERAGENT, "Mozilla/5.0 (Windows NT 5.1) AppleWebKit/535.1 (KHTML, like Gecko) Chrome/14.0.835.29 Safari/535.1");
    curl_multi_add_handle($megaconnect, $ch[$i]); 
}
$active = null;
do {
    $mrc = curl_multi_exec($megaconnect, $active);
} while ($mrc == CURLM_CALL_MULTI_PERFORM);
while ($active && $mrc == CURLM_OK) {
    if (curl_multi_select($megaconnect) != -1) {
        do {
            $mrc = curl_multi_exec($megaconnect, $active);
        } while ($mrc == CURLM_CALL_MULTI_PERFORM);
    }
}
$final = fopen($filename, "w+");
for ($i = 0; $i < sizeof($splits); $i++) {
    $contents = fread($bh[$i], filesize($partnames[$i]));
    fclose($bh[$i]);
    fwrite($final, $contents);
}
fclose($final);
function getFileSize($url) {
    $ch = curl_init($url);
    curl_setopt($ch, CURLOPT_NOBODY, true);
    curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
    curl_setopt($ch, CURLOPT_HEADER, true);
    curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
    $h = fopen('header', "w+");
    curl_setopt($ch, CURLOPT_WRITEHEADER, $h);
    $data = curl_exec($ch);
    curl_close($ch);
    if (preg_match('/Content-Length: ('d+)/', $data, $matches)) {
        return $contentLength = (int)$matches[1];
    }
    else return false;
}

一切顺利,除了一件事:

最后一个零件文件没有到达文件的末尾。实际文件大小为:3279848字节

范围是:

0-655970
655971-1311940
1311941-1967910
1967911-2623880
2623881-3279848

部件文件大小

tv.exe0 655360
tv.exe1 655360
tv.exe2 655360
tv.exe3 655360
tv.exe4 655360

使最终文件长度为3276800字节,但必须是3279848字节。当然,可执行文件不工作:(

注意,部件文件具有相同的大小。即使是最后一个,它应该有更多的字节。所以问题是在下载范围或其他地方,而不是在合并过程中。

我做错了什么?

我建议您在fclose($final);之后添加这个,以删除不再需要的文件部分!

foreach($partnames as $files_to_delete){ 
    unlink($files_to_delete); 
}

必须在读取前将文件指针设置为0。从end读取xy字节为0字节;)

$final = fopen($filename, "w+");
for ($i = 0; $i < sizeof($splits); $i++) {
  fseek($bh[$i], 0, SEEK_SET);
  $contents = fread($bh[$i], filesize($partnames[$i]));
  fclose($bh[$i]);
  fwrite($final, $contents);
}

$size未设置为任何值

设置为预期的大小后

655971 17 Aug 22:59 tv.exe0
655970 17 Aug 22:59 tv.exe1
655970 17 Aug 22:59 tv.exe2
655970 17 Aug 22:59 tv.exe3
655967 17 Aug 22:59 tv.exe4

您需要使用ceil()而不是round。Round可以Round DOWN,这将切断文件的末端。CEIL将四舍五入,保证指定的范围覆盖整个文件:

$splits = range(0, $size, ceil($size/5));
                          ^^^^

。如果文件的大小是12,而你取整(13/5),你将得到2。