gzip decodes only first line of the file about sdk

[lrhn]

Seems related to #47244

[devoncarew]

I suspect the archive itself is corrupt or somehow non-standard. On macos, I see:

Closing as I think the issue is with the archive and not the GZipCodec class. If on further investigation of the archive you believe its well-formed / something GZipCodec should parse, please re-open.

[meowofficial]

@devoncarew I updated the file. Now it opens with the standard macos archive utility. Could you check it again?

[a-siva]

//cc @brianquinlan

[brianquinlan]

I can repro. Python is able to decompress this file:

>>> s = open('test.csv.gz', 'rb').read()
>>> import gzip
>>> t = gzip.decompress(s)
>>> len(t)
68541

The bytes that Dart actually decodes are:

>>> x = [105, 100, 83, 117, 98, 67, 97, 109, 112, 97, 105, 103, 110, 84, 105, 116, 108, 101, 44, 105, 100, 83, 117, 98, 65, 100, 83, 101, 116, 84, 105, 116, 108, 101, 44, 105, 100, 83, 117, 98, 67, 97, 109, 112, 97, 105, 103, 110, 44, 105, 100, 67, 97, 109, 112, 97, 105, 103, 110, 84, 105, 116, 108, 101, 10]
>>> bytes(x)
b'idSubCampaignTitle,idSubAdSetTitle,idSubCampaign,idCampaignTitle\n'

Which is the first line of the file. If I understand correctly, the GZIP file format consists of concatenated compressed data sets. So maybe we are only decoding the first data set?

[brianquinlan]

I get the same output as Dart when use the zpipe example after changing:

-    ret = inflateInit(&strm);
+    ret = inflateInit2(&strm, 32 + 15);

The Python implementation looks very similar to ours.

[brianquinlan]

If I extract the file and recompress it with gzip, both Dart and zpipe and decompress the file. How did you generate this archive?

[meowofficial]

@brianquinlan This archive is from raw data export API response of https://docs.tracker.my.com/api/export-api/raw/about

[brianquinlan]

Yep. OK, I missed that Python deals with gzip data starting in Python code:
https://github.com/python/cpython/blob/fc21c7f7a731d64f7e4f0e82469f78fa9c104bbd/Lib/gzip.py#L622

I also found an example on how to handle concatenated gzip streams in C from Mark Adler himself:
https://stackoverflow.com/questions/17820664/is-this-a-bug-in-this-gzip-inflate-method/17822217#17822217

[brianquinlan]

I have a straightforward fix for this but it will take a while for me to convince myself that it always works.