CODEWITHSUNDEEP
pythonhttphttp-headersamazon-s3zlib
Heinrich Schmetterling
Heinrich Schmetterling

Reputation: 6884

Python zlib not decodable when returned by an http response

I'm using Amazon S3 to serve static files. When the Content-Type is just 'text/css' and I haven't compressed the file, it is returned ok. If I try to zlib.compress() the contents that will be returned and change Content-Encoding to 'gzip', the browser cannot decode the result. In Chrome, the error is

Error 330 net::ERR_CONTENT_DECODING_FAILED

in Safari,

“cannot decode raw data” (NSURLErrorDomain:-1015)

Is there something special to do with python's zlib to make sure the result can be returned and decompressed by the browser?

Upvotes: 5

Views: 1376

Answers (4)

Alfe
Alfe

Reputation: 59436

Instead of using the module zlib, (originalString = inputFile.read() and then compressedString = zlib.compress(originalString)) I am now using the module gzip:

stream = cStringIO.StringIO()
compressor = gzip.GzipFile(fileobj=stream, mode='w')
while True:  # until EOF
    chunk = inputFile.read(8192)
    if not chunk:  # EOF?
        compressor.close()
        return stream.getvalue()
    compressor.write(chunk)

The result then is compatible to gzip; I don't know if it fixes your webserver issue as well.

Upvotes: 0

Jonathan Works
Jonathan Works

Reputation: 1846

I have this same problem.

If you send the header:

Content-Encoding: gzip

Safari/Chrome show that error.

But if you instead send:

Content-Encoding: deflate

Safari/Chrome decodes the input fine.

Upvotes: 5

John Machin
John Machin

Reputation: 82934

It is decodable. The problem is that the sender is lying to the receiver -- not a good way of ensuring harmonious communication. Try calling it "zlib" instead of "gzip".

Upvotes: 0

Mark Byers
Mark Byers

Reputation: 838256

gzip is not the same as zlib.

Upvotes: 2

Related Questions