Cookie extractor from Set-Cookie response.

[danielatdattrixdotcom]

I needed this for working with an API that uses Django auth/sessions to work with CSRF cookie and X-CSRFToken header.

[svanoort-jenkins]

Can one of the admins verify this patch?

[svanoort]

Excellent! Glad you picked up and added this.

[svanoort]

@danielatdattrixdotcom Thank you, this is a feature that I know is in fairly high demand. I've added a couple comments (small changes) before this is ready to be merged. It also needs some sort of small unit test in test_validators.py.

This weekend I'm going to play with it a bit more and look more closely at the format of output (it might make sense to add a few options).

You get kudos by the way for looking closely at how the other extractors are implemented and following that (and adding documentation as well)!

[danielatdattrixdotcom]

@svanoort Let me know on the args/options to extractors in the YAML/JSON files and your thoughts on that and I'll make the updates and tests.

[alswl]

Oh no, I had write the same feature code, but @danielatdattrixdotcom 's documents is nice.
😂

[svanoort]

Fair, but we need to do the following for python 3 compatibility, since the library changed in Py3 😢 https://docs.python.org/2/library/cookie.html

if 'sys' is not in imports, add it:

PYTHON_MAJOR_VERSION = sys.version_info[0]
if PYTHON_MAJOR_VERSION > 2:
    import http.cookies as Cookie
else:
   import Cookie

[jimi-c]

Even easier, just do this:

from .six.moves import http_cookies as Cookie

I'm testing this patch with the above and it works perfectly well on both Py2 and Py3:

$ python2 $(which pyresttest) https://localhost:8000 test/api.yml 
Test Group Default SUCCEEDED: : 5/5 Tests Passed!

$ python3 $(which pyresttest) https://localhost:8000 test/api.yml 
Test Group Default SUCCEEDED: : 5/5 Tests Passed!

[svanoort]

@danielatdattrixdotcom Coming back to this now that the crunch period at work is settling down (spoiler: Jenkins, and specifically work on Pipeline Stage View plugin): I think it's not worth adding options to the extractor right now - I had a notion of eventually allowing use of regexes or other matchers in extraction, but it's not useful yet. We need support for templating with collections (or transformations on collections), and/or some embedded transformation language, if we're going to go down that route. It's planned in the future, but too far off to worry about now.

Added one comment about a Python 3 compatibility challenge (just an import change).

So, from my point of view, to get to merge, what we need are automated tests, covering a couple key edge cases: cookie doesn't exist, cookie exists with multiple values, normal cookie case. This means a unit test in test_validators.py, and a functional test in functionaltest.py -- the unit test just covers the extractor parsing and behavior (you can use some cookie info taken from a browser session).

The functional test needs to work with real HTTP requests; I'd prefer to add cookie-setting to the test Django-tastypie app (probably easiest done with the sessions middleware option), but if that proves hard we can just invoke a public API or site, as long as the request is trivial. I've been using GitHub for this in the past.