.NET Compatible gzuncompress

[TonyValenti]

I am working with a legacy app that has compressed files using PHP's gzcompress / gzuncompress.

From a bit of disecting, it looks like gzcompress returns content compressed with a GZipStream but without the GZip header and footer.

Is there a way to read/write these values in .NET? Does anyone have any ideas? I would be fine creating a synthetic header/footer for each record and wrapping the value from PHP inside those bytes but I'm not familiar enough with the GZip format to know what to synthesize.

Any help?

[svick]

I believe you can use ZLibStream for this, which is a new type in .Net 6.

E.g. consider this PHP code:

print(base64_encode(gzcompress("some test data")));

That code prints eJwrzs9NVShJLS5RSEksSQQAKGMFTw==. You can then read it in .Net 6 preview like this:

var inputStream = new MemoryStream(Convert.FromBase64String("eJwrzs9NVShJLS5RSEksSQQAKGMFTw=="));
using var zlibStream = new ZLibStream(inputStream, CompressionMode.Decompress);
using var streamReader = new StreamReader(zlibStream);
string output = streamReader.ReadToEnd();

(Your real code probably won't use base64, this was just an easy way for me to test this out.)

[MarkPflug]

Are you sure that GZipStream is the right format and not DeflateStream? My understanding was that ZLib, which the php gzcompress documentation calls out uses the deflate algorithm. I had to deal with something similar recently, decoding open street map pbf data, which is a protocol buffer format that uses zlib to compress blobs internally. I finally got the DeflateStream to decompress the data, but the caveat was that DeflateStream expects the stream to not contain a zlib header. The zlib header adds a two byte header to the start and a 4 byte checksum to the end of the data. By trimming these off (new MemoryStream(data, 2, data.Length - 6)) the DeflateStream was able to decode the data correctly.

[svick]

Trimming the extra bytes and using DeflateStream should work, but on .Net 6, just using ZLibStream is the easier option.