I'm trying to upload a zipfile to a Server using Python requests. The upload works fine. However the uploaded file cannot be opened using Windows Explorer or ark. I suppose there's some problem with mime-type or content-Length.
Oddly, uploading the file using curl, does not seem to cause the same problem. Here is my python code for the request:
s = requests.Session()
headers = {'Content-Type': 'application/zip'}
zip = open('file.zip', 'rb')
files = {'file': ('file.zip', zip, 'application/zip')}
fc = {'Content-Disposition': 'attachment; filename=file.zip'}
headers.update(fc)
r = requests.Request('POST', url, files=files, headers=headers, auth=(user, password))
prepared = r.prepare()
resp = s.send(prepared)
This is the curl code, which works flawlessly:
curl -X POST \
-ik \
-u user:password \
--data-binary '@file.zip' \
-H 'Content-Type: application/zip' \
-H "Content-Disposition: attachment; filename=file.zip" \
url
Uploading the file works in both, the Server also seems to recognize the content-type. However the file is rendered invalid when re-downloading. The zifile is readable before sending via requests or after sending with normal curl, using --data-binary. Opening the downloaded zifile with unip or file-roller works either way.
EDIT: I was uploading two files successively. Oddly the error was fixed when uploading the exact same files in reverse order. This has NOT been a python problem. When trying with standard curl I must have accidentally reversed the order, which is why it has been working.
I can not explain this behavior nor do I have a fix for it. In conclusion: Uploading the bigger file first did the trick.
All of the above seems to be applicable in curl, pycurl and python requests, so I assume it's some kind of bug in one of the curl libraries.
