mbeliaev/recorder

[beliaev-maksim]

propose beta feature for 0.22.0 to record responses to files using man in the middle

there are still steps to polish within this PR (types, docs, style, etc), but before doing this I need to confirm the idea.
General workflow looks like:

import requests
from responses import _recorder


def another():
    rsp = requests.get("https://httpstat.us/500")
    rsp = requests.get("https://httpstat.us/202")

@_recorder.record(file_path="out.yml")
def test_recorder():
    rsp = requests.get("https://httpstat.us/404")
    rsp = requests.get("https://httpstat.us/201")
    another()

which will product following out.yml file:

responses:
- response:
    auto_calculate_content_length: false
    body: ''
    content_type: text/plain
    headers: null
    method: GET
    status: 404
    url: https://httpstat.us/404
- response:
    auto_calculate_content_length: false
    body: ''
    content_type: text/plain
    headers: null
    method: GET
    status: 201
    url: https://httpstat.us/201
- response:
    auto_calculate_content_length: false
    body: ''
    content_type: text/plain
    headers: null
    method: GET
    status: 500
    url: https://httpstat.us/500
- response:
    auto_calculate_content_length: false
    body: ''
    content_type: text/plain
    headers: null
    method: GET
    status: 202
    url: https://httpstat.us/202

later I am going to extend the functionality and add possibility to load these yaml files

[codecov]

Codecov Report

Merging #545 (f7f937e) into master (d49f829) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##            master      #545   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            7         9    +2     
  Lines         2644      2730   +86     
=========================================
+ Hits          2644      2730   +86     

Impacted Files	Coverage Δ
responses/_recorder.py	100.00% <100.00%> (ø)
responses/tests/test_recorder.py	100.00% <100.00%> (ø)
responses/matchers.py	100.00% <0.00%> (ø)
responses/tests/test_matchers.py	100.00% <0.00%> (ø)

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[beliaev-maksim]

@markstory
I removed redundant code by inheriting from RequestMock
also, switched to toml format and added type hints in the code

if we are fine with the idea of implementation, then I will add tests to increase coverage

so, following code

import requests
from responses import _recorder


def another():
    rsp = requests.get("https://httpstat.us/500")
    rsp = requests.get("https://httpstat.us/202")

@_recorder.record(file_path="out.toml")
def test_recorder():
    rsp = requests.get("https://httpstat.us/404")
    rsp = requests.get("https://httpbin.org/status/wrong")
    another()

will product next output in out.toml

[[responses]]

[responses.response]
method = "GET"
url = "https://httpstat.us/404"
body = "404 Not Found"
status = 404
content_type = "text/plain"
auto_calculate_content_length = false
[[responses]]

[responses.response]
method = "GET"
url = "https://httpbin.org/status/wrong"
body = "Invalid status code"
status = 400
content_type = "text/plain"
auto_calculate_content_length = false
[[responses]]

[responses.response]
method = "GET"
url = "https://httpstat.us/500"
body = "500 Internal Server Error"
status = 500
content_type = "text/plain"
auto_calculate_content_length = false
[[responses]]

[responses.response]
method = "GET"
url = "https://httpstat.us/202"
body = "202 Accepted"
status = 202
content_type = "text/plain"
auto_calculate_content_length = false

[markstory]

Should dumping to a file be a separate thing? That would let userland code or us in the future change the serialization format that is being used to save response fixtures.

[beliaev-maksim]

I decided to create it in the registry because this will allow to dump existing tests to the file
Eg
If user ran responses.add() multiple times, then user can run responses.get_registry()._dump(*args)

That will allow safe (compared to brand new recording) transfer to the new method if required

In the future we can provide serializer argument, but at the moment I am a bit reluctant to support other serializers than one we decide

Also, user can also overwrite this method after inheriting

Or what are the other options to introduce this method?
As a separate function?

[markstory]

Couldn't dumping be on the Recorder()? I'm concerned that recording feels like it is half in the Recorder and half in the registry right now.

[beliaev-maksim]

@markstory
done!

this could be done even as a separate function. Then it will be possible to easily dump existing registries into the file

[beliaev-maksim]

@markstory
can you please review?

[beliaev-maksim]

@markstory
Can you check pls?