I have test to for connection error:
context 'unsuccessful API request' do
it 'responds like 500 if connection error is raised' do
http_stub = instance_double(Net::HTTP)
allow(Net::HTTP).to receive(:new).and_return(http_stub)
allow(http_stub).to receive(:request).and_raise(Errno::ECONNREFUSED)
api_client = ApiClient.new
api_client.create_implementer('Wayfarer')
expect(api_client.response_status).to eq(500)
expect(api_client.response_body).to eq({ issue' => [{ 'details' => { 'text' => 'Connection error' }}]})
end
end
And when I run it, it fails like it's supposed to but fails in the method instead of in the test so the test for the failure fails:
Failure/Error: response = http.request(request)
Errno::ECONNREFUSED:
Connection refused
Which is the correct error but it's failing in the method:
def http_request(request, uri)
http = Net::HTTP.new(uri.host, uri.port)
response = http.request(request)
@response_status = response.code.to_i
if response_successful?
@response_body = parsed_response(response)
else
@response_body = response.body rescue Errno::ECONNREFUSED
connection_error
end
end
The issue is it's stopping in the middle of the http_request
method at the line for response = http.request(request)
. It doesn't follow through with the error handling.
If I put a binding.pry
after that line it isn't reached and the method just bails.
You need to expect the error to be raised in your spec
expect { api_client.create_implementer('Wayfarer') }.to raise_error(Errno::ECONNREFUSED)