douqi3195 2015-04-07 07:21
浏览 83

来自GitLab的私有存储库存档(tar.gz),通过GitLab的API使用PHP下载,似乎已被破坏 - 校验和无效

I want to download a private GitLab repository archive using their API. I'm using PHP and cURL to do that. I'm able to do so using the following code:

$ch = curl_init(;
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
curl_setopt($ch, CURLOPT_BINARYTRANSFER, true);

$output = curl_exec($ch);
$fh = fopen("out.tar.gz", 'w');
fwrite($fh, $output);

The code itself is nothing complicated and pretty standard.

The issue I'm facing is that once I try to unpack, using the code bellow:

    // decompress from gz
    $p = new PharData($temp_archive . '.tar.gz');
    $p->decompress($temp_archive . '.tar');

    // unarchive from the tar
    $phar = new PharData($temp_archive . '.tar');

I receive the following error:

Fatal error: Uncaught exception 'UnexpectedValueException' with message ' in phpfile.php on line 294

UnexpectedValueException: phar error: "path/to/downloaded/archive/temp_archive.tar.gz" is a corrupted tar file (checksum mismatch of file "52 comment=01607149cd460cba6f1ac3003bb7b0c9cb0e2b94
") in phpfile.php on line 294

Yes, that is the whole error, I didn't cut out the error message. The ' is the error message.

The checksum for the downloaded file is, indeed, invalid. After a few good hours of Googling and forum browsing I'm back to square one. I have no idea what's wrong with it.

If I was to untar the archive using a terminal(I'm on MAC OS Build 12E55), it will unpack normally. The files are all there and I receive no errors what-so-ever.

I tried manually downloading the archive from GitLab and unpacking it using PHP(I get the same error).

I tried creating a *.tar.gz file using a terminal and unpacking it using PHP(I receive no errors and the unpacking proceeds normally).

I tried uploading a file created using the terminal and downloading it using PHP, after which I unpacked it, again using PHP, and I receive no errors.

I can only assume that I'm either:

  • Not using their API properly
  • Not setting the right headers - although I had tried setting:
    1. Content-Type: application/octet-stream
    2. Content-Type: application/x-tar
    3. Content-Transfer-Encoding: binary...and so on. When using the binary header, the file that I put the content in was set to 'wb'. Still, no luck.

And lastly it could be that their API may not be working properly, although I doubt it.

And nudging in the right direction is greatly appreciated.

  • 写回答

1条回答 默认 最新

  • duanjiaoxi4928 2015-04-07 08:39

    This appears to be a bug in the PharData class implementation that was fixed only very recently, see

    So it seems that Github uses the pax format for tar files, instead of the ustar format. For a description on tar formats, see:

    本回答被题主选为最佳回答 , 对您是否有帮助呢?



  • ¥100 读取 IMU BNO080 数据
  • ¥15 基于RTKLIB框架写的精密单点定位-AR
  • ¥15 PFENet的预训练权重
  • ¥15 程序哪有错误怎么改?
  • ¥15 交换机和交换机之间的链路带宽以及主机带宽的理解
  • ¥15 ai创想家对战模式代码
  • ¥15 集合A由3个2行4列二维数组构成,从集合A中任意取一个二维数组元素、如果该二维数组元素的对应列位置的上、下两数都是奇数,而且仅有2个列是奇数/奇数,则该数组有意义,并放入集合B中打印输出。
  • ¥15 电信IPV6 无法外网访问吗
  • ¥15 有偿求效果比较好的遥感影像匹配的c++代码
  • ¥15 博主,你好,我下载了你的智能网联汽车辅助驾驶安全信息检测系统,现在不会运行,可以教我吗,