feat: api command should extract source code of api elements

[Aclrian]

Closes #1106.

Summary of Changes

added an attribute code to functions and classes that contain the unformatted source code of the function or class

Testing Instructions

run package-parser api -p sklearn -o ./out and compare the output with the original file that the element contains.
package-parser annotations -a ../data/api/sklearn__api.json -u ../data/usages/sklearn_usages_counts.json -o ./out/annotations.json can verify that no error occur if the file is loaded.

[github-actions]

🦙 MegaLinter status: ✅ SUCCESS

Descriptor	Linter	Files	Fixed	Errors	Elapsed time
✅ JSON	eslint-plugin-jsonc	1	0	0	9.24s
✅ JSON	jsonlint	1		0	0.57s
✅ JSON	prettier	1	0	0	7.27s
✅ JSON	v8r	1		0	1.96s
✅ PYTHON	bandit	3		0	0.76s
✅ PYTHON	black	3	0	0	1.42s
✅ PYTHON	flake8	3		0	0.58s
✅ PYTHON	isort	3	0	0	1.27s
✅ PYTHON	mypy	3		0	2.1s
✅ PYTHON	pylint	3		0	4.45s

See errors details in artifact MegaLinter reports on CI Job page
Set VALIDATE_ALL_CODEBASE: true in mega-linter.yml to validate all sources, not only the diff

MegaLinter is graciously provided by

[Aclrian]

I have two questions:

• Does it need tests?
• Should the data folder be updated? (regarding the BSD 3-Clause "New" or "Revised" License from scikit-learn)

[lars-reimann]

• Does it need tests?

A function like trim_code should have tests. However, in this case you might be able to simplify the code by using the as_string method of astroid. And if you're only calling a well-tested third-party function, you don't need tests.

• Should the data folder be updated? (regarding the BSD 3-Clause "New" or "Revised" License from scikit-learn)

For the sake of easy (manual) testing, it makes sense to update the data folder. Just include the license file from scikit-learn as well in the data/api folder.

[Aclrian]

However, in this case you might be able to simplify the code by using the as_string method of astroid.

I compared output of the alternatives and found out that the as_string method ignores comments in the code, but it includes decorators. With a change from fromlineno to lineno decorators would be visible at the current process. Also, it seems that as_string() rearrange lines by removing comments.

Original sklearn.datasets._base.load_boston's last lines:

    return Bunch(
        data=data,
        target=target,
        # last column is target value
        feature_names=feature_names[:-1],
        DESCR=descr_text,
        filename=data_file_name,
        data_module=DATA_MODULE,
    )

and what as_string() converts it to:

    return Bunch(data=data, target=target, feature_names=feature_names[:-1], DESCR=descr_text, filename=data_file_name, data_module=DATA_MODULE)

If the comment line were removed in a new version, there would be a few new lines when we change to as_string().
So, I think we should keep the current procedure.

[lars-reimann]

Let's discuss this tomorrow.

[lars-reimann]

Please resolve conversations yourself once you are done with them.

[github-actions]

🎉 This PR is included in version 1.67.0 🎉

The release is available on GitHub release

Your semantic-release bot 📦🚀