Multipart Form Data Headers

[ghost]

After testing #2382 it seems that the request headers are not being updated correctly.

filename = 'test.tar'
data = {"file": (Path(filename).name, open(filename, "rb"), "application/x-tar")}
resp = httpx.request("POST", 'URL', files=data)

The request has the following header {'Content-Type': 'multipart/form-data; boundary=b194f2c9bc744ebf40c3fa03d6d53987'}

However if using an initialized client:

client = httpx.Client(headers={"Content-Type": "application/json"})
filename = 'test.tar'
data = {"file": (Path(filename).name, open(filename, "rb"), "application/x-tar")}
resp = client.post('URL', files=data)

The request is still using {'Content-Type': 'application/json'}
Error message from our API: '{"code":"API_MALFORMED_BODY","message":"Malformed JSON"}'

Shouldn't the POST change the headers if using multipart forms?

Specifying the following also does not work because the boundaries are not included in the headers:
resp = client.post('URL', files=data, headers={'Content-Type': 'multipart/form-data'})

[lovelydinosaur]

Okay, it's a little bit ambiguous what behavior you'd expect in those cases.
But, yes, I can see why you might like the client-default Content-Type to be overridden there.

I don't think this is related to #2382, it looks like the same behaviour that we've had for a while...

import httpx

# Client with Content-Type headers set.
client = httpx.Client(headers={"Content-Type": "application/json"})

# An `application/x-www-form-urlencoded` POST request.
# Essentially the same case as your multipart example, but just a little more simple.
request = client.build_request("POST", "https://www.example.com", data={"a": "b"})

print(request.headers)
print(request.content)

Across a few different version of httpx...

# 0.20.0
Headers({'host': 'www.example.com', 'accept': '*/*', 'accept-encoding': 'gzip, deflate', 'connection': 'keep-alive', 'user-agent': 'python-httpx/0.20.0', 'content-type': 'application/json', 'content-length': '3'})
b'a=b'

# 0.21.0
Headers({'host': 'www.example.com', 'accept': '*/*', 'accept-encoding': 'gzip, deflate', 'connection': 'keep-alive', 'user-agent': 'python-httpx/0.21.0', 'content-type': 'application/json', 'content-length': '3'})
b'a=b'

# 0.22.0
Headers({'host': 'www.example.com', 'accept': '*/*', 'accept-encoding': 'gzip, deflate', 'connection': 'keep-alive', 'user-agent': 'python-httpx/0.22.0', 'content-type': 'application/json', 'content-length': '3'})

# latest
Headers({'host': 'www.example.com', 'accept': '*/*', 'accept-encoding': 'gzip, deflate', 'connection': 'keep-alive', 'user-agent': 'python-httpx/0.23.2', 'content-type': 'application/json', 'content-length': '3'})
b'a=b'

All the cases have Content-Type: application/json set, but have a form-encoded body.

The quickest solution to the issue would be to stop doing this...

# Not needed. You'll get this header if you use `json=...` anyways.
# Specifying a `Content-Type` header on the client instance is almost always going to
# be a bad idea. Eg. you'll also get odd looking GET requests which have a `Content-Type`
# header but no content.
client = httpx.Client(headers={"Content-Type": "application/json"})

But, yeah, there's other options too. I can see that we might be able to change the priority rules so that client headers only get applied if they're not overriding the auto-generated request headers.

Or... we could guard against setting certain headers on the client. In particular we could prevent setting the Content-Type, Content-Length, Host and Transfer-Encoding headers, which are all auto-populated based on the request.