v2.0.0

Available on PyPI: https://pypi.org/project/pook/2.0.0/

See #128 for a summary of the breaking changes and how to update your code if you are affected.

• Breaking change: Remove Response::body's binary parameter and enforce a keyword argument for chunked.

  • The binary parameter is no longer needed since responses are now always byte-encoded in all cases (see below).
  • A keyword argument is now enforced for chunked to ensure unnamed arguments meant for the removed binary parameter cannot be confused as chunked arguments.

• Only return byte-encoded response bodies, matching the bahviour of all supported client libraries.

  • This is possible because for all supported client libraries, pook mocks the actual response sent to the
    client library over the wire, which will, in all cases, be bytes. Client libraries that support reading
    response content as a string or other formats will continue to work as expected, because they'll always
    be handling bytes from pook.
  • This causes no change to application code for users of pook, except for users of the standard library urllib,
    for which this also fixed a bug where non-bytes bodies were returned by pook in certain cases. This is impossible
    in real application code. If you rely on pook to mock urllib responses and have code that handles non-bytes response
    bodies, that code can be safely deleted (provided the only reason it was there was pook in the first place).

• Breaking change: Remove Mock::body's binary parameter.

  • This parameter was already unused due to a bug in the code (it was not passed through to the BodyMatcher),
    so this will not cause any changes to tests that relied on it: it didn't do anything to begin with.
  • The breaking change is simply the removal of the unused parameter, which should be deleted from tests using pook.
  • Pook's code has also been updated to remove all cases where non-bytes objects were being handled. Instead, the body
    of any interecepted request will always be stored as bytes, and only decoded when necessary for individual downstream
    matchers (JSON, XML).

• Correct documentation strings for XMLMatcher and JSONMatcher to no longer suggest they can handle regex matchers.

  • These classes never implemented regex matching.

• Add a pytest fixture to the package (kudos @wimglenn)