_get_resp_headers() is failing with KeyError: 'content-length' with 1.5.2 for unsuccessful response, this works with 1.5.1

[dbasunag]

This was seen with kubernetes client calls returning failure status, this is no longer raising up correct error response/status etc. and is now failing with KeyError: 'content-length' (with 1.5.2 only, works fine with 1.5.1):
Seems like with f85ae1f, this issue was introduced. Do you have any suggestions on how to work around this? Any suggestions or helps would be highly appreciated.
{noformat}
Traceback (most recent call last):
File "/.venv/lib/python3.9/site-packages/kubernetes/stream/ws_client.py", line 521, in websocket_call
client = WSClient(configuration, url, headers, capture_all)
File "/.venv/lib/python3.9/site-packages/kubernetes/stream/ws_client.py", line 65, in init
self.sock = create_websocket(configuration, url, headers)
File ".venv/lib/python3.9/site-packages/kubernetes/stream/ws_client.py", line 487, in create_websocket
websocket.connect(url, **connect_opt)
File "/.venv/lib/python3.9/site-packages/websocket/_core.py", line 253, in connect
self.handshake_response = handshake(self.sock, url, *addrs, **options)
File "/.venv/lib/python3.9/site-packages/websocket/_handshake.py", line 57, in handshake
status, resp = _get_resp_headers(sock)
File "/.venv/lib/python3.9/site-packages/websocket/_handshake.py", line 147, in _get_resp_headers
response_body = sock.recv(int(resp_headers['content-length'])) # read the body of the HTTP error message response and include it in the exception
KeyError: 'content-length'

File "/.venv/lib/python3.9/site-packages/kubernetes/client/api/core_v1_api.py", line 994, in connect_get_namespaced_pod_exec
return self.connect_get_namespaced_pod_exec_with_http_info(name, namespace, **kwargs) # noqa: E501
File "/.venv/lib/python3.9/site-packages/kubernetes/client/api/core_v1_api.py", line 1101, in connect_get_namespaced_pod_exec_with_http_info
return self.api_client.call_api(
File "/.venv/lib/python3.9/site-packages/kubernetes/client/api_client.py", line 348, in call_api
return self.__call_api(resource_path, method,
File "/.venv/lib/python3.9/site-packages/kubernetes/client/api_client.py", line 180, in __call_api
response_data = self.request(
File "/.venv/lib/python3.9/site-packages/kubernetes/stream/ws_client.py", line 527, in websocket_call
raise ApiException(status=0, reason=str(e))
kubernetes.client.exceptions.ApiException: (0)
Reason: 'content-length'
{noformat}

[engn33r]

I suspect the issue is that this line that was added to fix #900
response_body = sock.recv(int(resp_headers['content-length']))

and the HTTP message may not have a content-length header. Could you share the HTTP message, or at least the headers, of the message that is causing the issue? Wireshark may be useful to capture it. I did not realize this header might be optional, I'll have to check the HTTP specs.

[dbasunag]

This is a get call that I am expecting to fail with 403 forbidden error.
Here is the details
header: {'audit-id': '<something>', 'cache-control': 'no-cache, private', 'content-type': 'application/json', 'strict-transport-security': 'max-age=31536000; includeSubDomains; preload', 'date': 'Fri, 02 Jun 2023 01:25:36 GMT', 'transfer-encoding': 'chunked'} status: 403 message: Forbidden

[oliviermartin]

I can see there is already a PR for this issue: #911

[engn33r]

This should be fixed in 1.5.3