Add ConanVersionRange class and test

[johnmhoran]

Fixes:

• Add univers support for Conan #101

[TG1999]

Thanks, few nits for your consideration

[TG1999]

It should be "conan" : "conan1" since gitlab uses conan in their package slug

[johnmhoran]

Thank you @TG1999. I made the change yesterday to conan1 to address an error I was getting when running this test:

pytest -vvs tests/test_version_range.py::test_conan_gitlab_version_range_parse

I just made the change you requested and reran that test, and again I get the same error. Here's an excerpt from a large terminal output:

(venv) Fri Feb 24, 2023 10:00 AM  /home/jmh/dev/nexb/univers jmh (101-univers-support-for-conan)
$ pytest -vvs tests/test_version_range.py::test_conan_gitlab_version_range_parse

[...]

=================================================================================================== FAILURES ====================================================================================================
_______________________________________________________________________________ test_conan_gitlab_version_range_parse[test_case0] _______________________________________________________________________________

test_case = {'expected_vers': 'vers:conan1/>=5.1.0|<=5.1.1', 'native': '>=5.1.0 <=5.1.1', 'scheme': 'conan1', 'test_index': 1}

    @pytest.mark.parametrize("test_case", json.load(open("./tests/data/conan_advisory.json")))
    def test_conan_gitlab_version_range_parse(test_case):
        if test_case["expected_vers"] is None:
            with pytest.raises(InvalidVersion):
                Conan1VersionRange.from_native(string=test_case["native"])
            return
>       result = from_gitlab_native(
            gitlab_scheme=test_case["scheme"],
            string=test_case["native"],
        )

tests/test_version_range.py:364:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

gitlab_scheme = 'conan1', string = '>=5.1.0 <=5.1.1'

    def from_gitlab_native(gitlab_scheme, string):
>       purl_scheme = PURL_TYPE_BY_GITLAB_SCHEME[gitlab_scheme]
E       KeyError: 'conan1'

src/univers/version_range.py:1106: KeyError

[...]

========================================================================================= 427 failed, 1 passed in 8.87s =========================================================================================

(venv) Fri Feb 24, 2023 10:00 AM  /home/jmh/dev/nexb/univers jmh (101-univers-support-for-conan)
$

I was unable to identify the source of this error -- do you have any suggestions?

Meanwhile, I will implement your other comments and see if that resolves this set of error messages.

[TG1999]

add a parameter "gitlab_scheme" : "conan" in every test here.

[johnmhoran]

@TG1999 Very interesting -- I'm getting a better sense of the structure we need for the gitlab advisory data. I've followed your suggestion and the JSON objects now look like this:

  {
    "test_index": 1,
    "scheme": "conan1",
    "native": ">=5.1.0 <=5.1.1",
    "expected_vers": "vers:conan1/>=5.1.0|<=5.1.1",
    "gitlab_scheme": "conan"
  },

[TG1999]

Suggested change

	gitlab_scheme=test_case["scheme"],
	gitlab_scheme=test_case["gitlab_scheme"],

[johnmhoran]

Outstanding @TG1999 -- with this and the 2 other sets of changes, all 428 of the test cases pass both Conan-related tests.

Thank you. 🙂 Will commit and push shortly.

[TG1999]

LGTM!

[johnmhoran]

In src/univers/version_range.py, there's an s missing in from univers.conan.version_range import VersionRange as conan_verion_range.

[pombredanne]

Don't you need an __init__.py in the conan directory?
I would rather have a regular package than an implicit namespace package.

[pombredanne]

You are missing the ABOUT and license/notice files

[johnmhoran]

@pombredanne @TG1999 I've added an __init__.py file in the /univers/conan/ directory.

[johnmhoran]

@pombredanne @TG1999 I've also added a pair of .ABOUT/.NOTICE files for each of errors.py, version_range.py and version.py, and have hand-crafted each of the .ABOUT files. (Should the latter files somehow have been tool-generated?)

[johnmhoran]

@pombredanne @TG1999 I just committed and pushed and immediately saw that a GH check has failed --

https://github.com/nexB/univers/pull/102/checks?check_run_id=11840361414

There are 66 commits incorrectly signed off. This means that the author(s) of these commits failed to include a Signed-off-by line in their commit message.

I suspect these sign-off errors are from the commit histories of conan files added from the JFrog repo.

How do we handle this?

[johnmhoran]

It's odd that this GH check would have failed after I committed and pushed my limited changes -- the JFrog Conan code had already been added by @TG1999 , and I retrieved it with a git pull. But somehow something I introduced triggered this . . . .

[TG1999]

@johnmhoran thanks for adding the ABOUT files, we cannot do a lot for the DCO failing here because of reusing Conan commit history I have set DCO to pass.

[pombredanne]

@johnmhoran Thanks... just one last nitpicking to add ABOUT files for the test code (and headers)

[pombredanne]

LGTM! Thanks!