Funnily enough, I had that error when trying to work with the Stack Overflow API using Python.

I managed to get it working with the GzipFile object from the gzip directory, roughly like this:

import gzip

gzip_file = gzip.GzipFile(fileobj=open('abc.gz', 'rb'))

file_contents = gzip_file.read()

I just solved the “incorrect header check” problem when uncompressing gzipped data.

You need to set -WindowBits => WANT_GZIP in your call to inflateInit2 (use the 2 version)

Yes, this can be very frustrating. A typically shallow reading of the documentation presents Zlib as an API to Gzip compression, but by default (not using the gz* methods) it does not create or uncompress the Gzip format. You have to send this non-very-prominently documented flag.

You have this error:

zlib.error: Error -3 while decompressing: incorrect header check

Which is most likely because you are trying to check headers that are not there, e.g. your data follows RFC 1951 (deflate compressed format) rather than RFC 1950 (zlib compressed format) or RFC 1952 (gzip compressed format).

choosing windowBits

But zlib can decompress all those formats:

• to (de-)compress deflate format, use wbits = -zlib.MAX_WBITS
• to (de-)compress zlib format, use wbits = zlib.MAX_WBITS
• to (de-)compress gzip format, use wbits = zlib.MAX_WBITS | 16

See documentation in http://www.zlib.net/manual.html#Advanced (section inflateInit2)

examples

test data:

>>> deflate_compress = zlib.compressobj(9, zlib.DEFLATED, -zlib.MAX_WBITS)
>>> zlib_compress = zlib.compressobj(9, zlib.DEFLATED, zlib.MAX_WBITS)
>>> gzip_compress = zlib.compressobj(9, zlib.DEFLATED, zlib.MAX_WBITS | 16)
>>> 
>>> text = '''test'''
>>> deflate_data = deflate_compress.compress(text) + deflate_compress.flush()
>>> zlib_data = zlib_compress.compress(text) + zlib_compress.flush()
>>> gzip_data = gzip_compress.compress(text) + gzip_compress.flush()
>>> 

obvious test for zlib:

>>> zlib.decompress(zlib_data)
'test'

test for deflate:

>>> zlib.decompress(deflate_data)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
zlib.error: Error -3 while decompressing data: incorrect header check
>>> zlib.decompress(deflate_data, -zlib.MAX_WBITS)
'test'

test for gzip:

>>> zlib.decompress(gzip_data)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
zlib.error: Error -3 while decompressing data: incorrect header check
>>> zlib.decompress(gzip_data, zlib.MAX_WBITS|16)
'test'

the data is also compatible with gzip module:

>>> import gzip
>>> import StringIO
>>> fio = StringIO.StringIO(gzip_data)  # io.BytesIO for Python 3
>>> f = gzip.GzipFile(fileobj=fio)
>>> f.read()
'test'
>>> f.close()

automatic header detection (zlib or gzip)

adding 32 to windowBits will trigger header detection

>>> zlib.decompress(gzip_data, zlib.MAX_WBITS|32)
'test'
>>> zlib.decompress(zlib_data, zlib.MAX_WBITS|32)
'test'

using gzip instead

or you can ignore zlib and use gzip module directly; but please remember that under the hood, gzip uses zlib.

fh = gzip.open('abc.gz', 'rb')
cdata = fh.read()
fh.close()

My case was to decompress email messages that are stored in Bullhorn database. The snippet is the following:

import pyodbc
import zlib

cn = pyodbc.connect('connection string')
cursor = cn.cursor()
cursor.execute('SELECT TOP(1) userMessageID, commentsCompressed FROM BULLHORN1.BH_UserMessage WHERE DATALENGTH(commentsCompressed) > 0 ')



 for msg in cursor.fetchall():
    #magic in the second parameter, use negative value for deflate format
    decompressedMessageBody = zlib.decompress(bytes(msg.commentsCompressed), -zlib.MAX_WBITS)

Just add headers ‘Accept-Encoding’: ‘identity’

import requests

requests.get('http://gett.bike/', headers={'Accept-Encoding': 'identity'})

https://github.com/requests/requests/issues/3849

To decompress incomplete gzipped bytes that are in memory, the answer by dnozay is useful but it misses the zlib.decompressobj call which I found to be necessary:

incomplete_decompressed_content = zlib.decompressobj(wbits=zlib.MAX_WBITS | 16).decompress(incomplete_gzipped_content)

Note that zlib.MAX_WBITS | 16 is 15 | 16 which is 31. For some background about wbits, see zlib.decompress.

---

Credit: answer by Yann Vernier which notes the the zlib.decompressobj call.

This does not answer the original question, but it may help someone else that ends up here.

The zlib.error: Error -3 while decompressing: incorrect header check also occurs in the example below:

b64_encoded_bytes = base64.b64encode(zlib.compress(b'abcde'))
encoded_bytes_representation = str(b64_encoded_bytes)  # this the cause
zlib.decompress(base64.b64decode(encoded_bytes_representation))

The example is a minimal reproduction of something I encountered in some legacy Django code, where Base64 encoded bytes (from an HTTP POST) were being stored in a Django CharField (instead of a BinaryField).

When reading a CharField value from the database, str() is called on the value, without an explicit encoding, as can be seen in the Django source.

The str() documentation says:

If neither encoding nor errors is given, str(object) returns object.str(), which is the “informal” or nicely printable string representation of object. For string objects, this is the string itself. If object does not have a str() method, then str() falls back to returning repr(object).

So, in the example, we are inadvertently base64-decoding

"b'eJxLTEpOSQUABcgB8A=='"

instead of

b'eJxLTEpOSQUABcgB8A=='.

The zlib decompression in the example would succeed if an explicit encoding were used, e.g. str(b64_encoded_bytes, 'utf-8').

NOTE specific to Django:

What’s especially tricky: this issue only arises when retrieving a value from the database. See for example the test below, which passes (in Django 3.0.3):

class MyModelTests(TestCase):
    def test_bytes(self):
        my_model = MyModel.objects.create(data=b'abcde')
        self.assertIsInstance(my_model.data, bytes)  # issue does not arise
        my_model.refresh_from_db()
        self.assertIsInstance(my_model.data, str)  # issue does arise

where MyModel is

class MyModel(models.Model):
    data = models.CharField(max_length=100)