When packaging software for Fedora, we run tests in an unpacked archive without initialized git repository for the codebase because we are testing the installed version. Would it make sense to make the tests compatible with this way of running them?

Maybe the tests can initialize their own repositories in a tmpdir or something like that.

The current log I have is:

python -m pytest
===================================== test session starts =====================================
platform linux -- Python 3.10.7, pytest-7.1.3, pluggy-1.0.0
rootdir: /home/lbalhar/Software/databooks
collected 118 items                                                                           

tests/test_affirm.py ......................                                             [ 18%]
tests/test_cli.py ..........FF.FFF.F.F.FF                                               [ 38%]
tests/test_common.py ..                                                                 [ 39%]
tests/test_conflicts.py .........                                                       [ 47%]
tests/test_git_utils.py F.                                                              [ 49%]
tests/test_metadata.py ..........                                                       [ 57%]
tests/test_recipes.py ...............                                                   [ 70%]
tests/test_tui.py ..................                                                    [ 85%]
tests/test_data_models/test_base.py .                                                   [ 86%]
tests/test_data_models/test_notebook.py ................                                [100%]

========================================== FAILURES ===========================================
__________________________________________ test_meta __________________________________________

tmpdir = local('/tmp/pytest-of-lbalhar/pytest-1/test_meta0')

    def test_meta(tmpdir: LocalPath) -> None:
        """Remove notebook metadata."""
        read_path = tmpdir.mkdir("notebooks") / "test_meta_nb.ipynb"  # type: ignore
        TestJupyterNotebook().jupyter_notebook.write(read_path)
    
        nb_read = JupyterNotebook.parse_file(path=read_path)
        result = runner.invoke(app, ["meta", str(read_path), "--yes"])
        nb_write = JupyterNotebook.parse_file(path=read_path)
    
>       assert result.exit_code == 0
E       assert 1 == 0
E        +  where 1 = <Result InvalidGitRepositoryError()>.exit_code

tests/test_cli.py:50: AssertionError
______________________________________ test_meta__check _______________________________________

tmpdir = local('/tmp/pytest-of-lbalhar/pytest-1/test_meta__check0')
caplog = <_pytest.logging.LogCaptureFixture object at 0x7fca4a63baf0>

    def test_meta__check(tmpdir: LocalPath, caplog: LogCaptureFixture) -> None:
        """Report on existing notebook metadata (both when it is and isn't present)."""
        caplog.set_level(logging.INFO)
    
        read_path = tmpdir.mkdir("notebooks") / "test_meta_nb.ipynb"  # type: ignore
        TestJupyterNotebook().jupyter_notebook.write(read_path)
    
        nb_read = JupyterNotebook.parse_file(path=read_path)
        result = runner.invoke(app, ["meta", str(read_path), "--check"])
        nb_write = JupyterNotebook.parse_file(path=read_path)
    
        logs = list(caplog.records)
        assert result.exit_code == 1
>       assert len(logs) == 1
E       assert 0 == 1
E        +  where 0 = len([])

tests/test_cli.py:83: AssertionError
______________________________________ test_meta__script ______________________________________

tmpdir = local('/tmp/pytest-of-lbalhar/pytest-1/test_meta__script0')

    def test_meta__script(tmpdir: LocalPath) -> None:
        """Raise `typer.BadParameter` when passing a script instead of a notebook."""
        py_path = tmpdir.mkdir("files") / "a_script.py"  # type: ignore
        py_path.write_text("# some python code", encoding="utf-8")
    
        result = runner.invoke(app, ["meta", str(py_path)])
>       assert result.exit_code == 2
E       assert 1 == 2
E        +  where 1 = <Result InvalidGitRepositoryError()>.exit_code

tests/test_cli.py:139: AssertionError
____________________________________ test_meta__no_confirm ____________________________________

tmpdir = local('/tmp/pytest-of-lbalhar/pytest-1/test_meta__no_confirm0')

    def test_meta__no_confirm(tmpdir: LocalPath) -> None:
        """Don't make any changes without confirmation to overwrite files (prompt)."""
        nb_path = tmpdir.mkdir("notebooks") / "test_meta_nb.ipynb"  # type: ignore
        TestJupyterNotebook().jupyter_notebook.write(nb_path)
    
        result = runner.invoke(app, ["meta", str(nb_path)])
    
        assert result.exit_code == 1
        assert JupyterNotebook.parse_file(nb_path) == TestJupyterNotebook().jupyter_notebook
>       assert result.output == (
            "1 files will be overwritten (no prefix nor suffix was passed)."
            " Continue? [y/n]: \nAborted!\n"
        )
E       AssertionError: assert '' == '1 files will... \nAborted!\n'
E         - 1 files will be overwritten (no prefix nor suffix was passed). Continue? [y/n]: 
E         - Aborted!

tests/test_cli.py:155: AssertionError
_____________________________________ test_meta__confirm ______________________________________

tmpdir = local('/tmp/pytest-of-lbalhar/pytest-1/test_meta__confirm0')

    def test_meta__confirm(tmpdir: LocalPath) -> None:
        """Make changes when confirming overwrite via the prompt."""
        nb_path = tmpdir.mkdir("notebooks") / "test_meta_nb.ipynb"  # type: ignore
        TestJupyterNotebook().jupyter_notebook.write(nb_path)
    
        result = runner.invoke(app, ["meta", str(nb_path)], input="y")
    
>       assert result.exit_code == 0
E       assert 1 == 0
E        +  where 1 = <Result InvalidGitRepositoryError()>.exit_code

tests/test_cli.py:168: AssertionError
_________________________________________ test_assert _________________________________________

caplog = <_pytest.logging.LogCaptureFixture object at 0x7fca4a66dff0>

    def test_assert(caplog: LogCaptureFixture) -> None:
        """Assert that notebook has sequential and increasing cell execution."""
        caplog.set_level(logging.INFO)
    
        exprs = (
            "[c.execution_count for c in exec_cells] == list(range(1, len(exec_cells) + 1))"
        )
        recipe = "seq-increase"
        with resources.path("tests.files", "demo.ipynb") as nb_path:
            result = runner.invoke(
                app, ["assert", str(nb_path), "--expr", exprs, "--recipe", recipe]
            )
    
        logs = list(caplog.records)
>       assert result.exit_code == 0
E       assert 1 == 0
E        +  where 1 = <Result InvalidGitRepositoryError()>.exit_code

tests/test_cli.py:203: AssertionError
__________________________________________ test_fix ___________________________________________

tmpdir = local('/tmp/pytest-of-lbalhar/pytest-1/test_fix0')

    def test_fix(tmpdir: LocalPath) -> None:
        """Fix notebook conflicts."""
        # Setup
        nb_path = Path("test_conflicts_nb.ipynb")
        notebook_1 = TestJupyterNotebook().jupyter_notebook
        notebook_2 = TestJupyterNotebook().jupyter_notebook
    
        notebook_1.metadata = NotebookMetadata(
            kernelspec=dict(
                display_name="different_kernel_display_name", name="kernel_name"
            ),
            field_to_remove=["Field to remove"],
            another_field_to_remove="another field",
        )
    
        extra_cell = BaseCell(
            cell_type="raw",
            metadata=CellMetadata(random_meta=["meta"]),
            source="extra",
        )
        notebook_2.cells = notebook_2.cells + [extra_cell]
        notebook_2.nbformat += 1
        notebook_2.nbformat_minor += 1
    
        git_repo = init_repo_conflicts(
            tmpdir=tmpdir,
            filename=nb_path,
            contents_main=notebook_1.json(),
            contents_other=notebook_2.json(),
            commit_message_main="Notebook from main",
            commit_message_other="Notebook from other",
        )
    
        conflict_files = get_conflict_blobs(repo=git_repo)
        id_main = conflict_files[0].first_log
        id_other = conflict_files[0].last_log
    
        # Run CLI and check conflict resolution
        result = runner.invoke(app, ["fix", str(tmpdir)])
>       fixed_notebook = JupyterNotebook.parse_file(path=tmpdir / nb_path)

tests/test_cli.py:268: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
databooks/data_models/notebook.py:250: in parse_file
    return super(JupyterNotebook, cls).parse_file(
pydantic/main.py:561: in pydantic.main.BaseModel.parse_file
    ???
pydantic/parse.py:64: in pydantic.parse.load_file
    ???
pydantic/parse.py:37: in pydantic.parse.load_str_bytes
    ???
/usr/lib64/python3.10/json/__init__.py:346: in loads
    return _default_decoder.decode(s)
/usr/lib64/python3.10/json/decoder.py:337: in decode
    obj, end = self.raw_decode(s, idx=_w(s, 0).end())
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

self = <json.decoder.JSONDecoder object at 0x7fca4be6b1f0>
s = '<<<<<<< HEAD\n{"nbformat": 4, "nbformat_minor": 4, "metadata": {"field_to_remove": ["Field to remove"], "another_fiel...xecution_count": 1}, {"metadata": {"random_meta": ["meta"]}, "source": "extra", "cell_type": "raw"}]}\n>>>>>>> other\n'
idx = 0

    def raw_decode(self, s, idx=0):
        """Decode a JSON document from ``s`` (a ``str`` beginning with
        a JSON document) and return a 2-tuple of the Python
        representation and the index in ``s`` where the document ended.
    
        This can be used to decode a JSON document from a string that may
        have extraneous data at the end.
    
        """
        try:
            obj, end = self.scan_once(s, idx)
        except StopIteration as err:
>           raise JSONDecodeError("Expecting value", s, err.value) from None
E           json.decoder.JSONDecodeError: Expecting value: line 1 column 1 (char 0)

/usr/lib64/python3.10/json/decoder.py:355: JSONDecodeError
__________________________________________ test_show __________________________________________

    def test_show() -> None:
        """Show notebook in terminal."""
        with resources.path("tests.files", "tui-demo.ipynb") as nb_path:
            result = runner.invoke(app, ["show", str(nb_path)])
>       assert result.exit_code == 0
E       assert 1 == 0
E        +  where 1 = <Result InvalidGitRepositoryError()>.exit_code

tests/test_cli.py:382: AssertionError
____________________________________ test_show_no_multiple ____________________________________

    def test_show_no_multiple() -> None:
        """Don't show multiple notebooks if not confirmed in prompt."""
        with resources.path("tests.files", "tui-demo.ipynb") as nb:
            dirpath = str(nb.parent)
    
        # Exit code is 0 if user responds to prompt with `n`
        result = runner.invoke(app, ["show", dirpath], input="n")
>       assert result.exit_code == 0
E       assert 1 == 0
E        +  where 1 = <Result InvalidGitRepositoryError()>.exit_code

tests/test_cli.py:457: AssertionError
________________________________________ test_get_repo ________________________________________

    def test_get_repo() -> None:
        """Find git repository."""
        curr_dir = Path(__file__).parent
>       repo = get_repo(curr_dir)

tests/test_git_utils.py:54: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
databooks/git_utils.py:38: in get_repo
    repo = Repo(path=repo_dir)
../../.virtualenvs/databooks/lib/python3.10/site-packages/git/repo/base.py:266: in __init__
    self.working_dir: Optional[PathLike] = self._working_tree_dir or self.common_dir
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

self = <git.repo.base.Repo ''>

    @property
    def common_dir(self) -> PathLike:
        """
        :return: The git dir that holds everything except possibly HEAD,
            FETCH_HEAD, ORIG_HEAD, COMMIT_EDITMSG, index, and logs/."""
        if self._common_dir:
            return self._common_dir
        elif self.git_dir:
            return self.git_dir
        else:
            # or could return ""
>           raise InvalidGitRepositoryError()
E           git.exc.InvalidGitRepositoryError

../../.virtualenvs/databooks/lib/python3.10/site-packages/git/repo/base.py:347: InvalidGitRepositoryError
=================================== short test summary info ===================================
FAILED tests/test_cli.py::test_meta - assert 1 == 0
FAILED tests/test_cli.py::test_meta__check - assert 0 == 1
FAILED tests/test_cli.py::test_meta__script - assert 1 == 2
FAILED tests/test_cli.py::test_meta__no_confirm - AssertionError: assert '' == '1 files will...
FAILED tests/test_cli.py::test_meta__confirm - assert 1 == 0
FAILED tests/test_cli.py::test_assert - assert 1 == 0
FAILED tests/test_cli.py::test_fix - json.decoder.JSONDecodeError: Expecting value: line 1 c...
FAILED tests/test_cli.py::test_show - assert 1 == 0
FAILED tests/test_cli.py::test_show_no_multiple - assert 1 == 0
FAILED tests/test_git_utils.py::test_get_repo - git.exc.InvalidGitRepositoryError
=============================== 10 failed, 108 passed in 0.60s ================================

The first commit drops dependency on py module and switch from LocalPath to pathlib.Path which has a different API so it required some more changes.

The second commit actually removes the unnecessary subdirs. If you want to keep them, I can drop the second commit.

Hi!

I really enjoyed your talk on PyCon PL today and I've decided that it'd be great to have databook packaged in Fedora Linux. I've tried to create a package but found a problem - we have too new type in Fedora (version 0.6.1) but databook requires version < 0.5.

https://github.com/datarootsio/databooks/blob/0b1d3e67c5fd00a145179545a456786bab5b7e77/pyproject.toml#L20

Would it make sense to make the requirement less strict? If thy are following semantic versioning, something like <1.0.0 should guarantee enough stability.

LICENSE is included automatically by poetry and it's installed into .dist-info directory. This manual include caused the file to be installed also into lib/python3.11/site-packages next to the main databooks package.

This will no longer work with pytest 7.2.0 and higher. py lib is deprecated but you can fix the problem by requiring it explicitly if there is no other way now.

https://github.com/datarootsio/databooks/blob/6febdbadde115cdb6fb89eec737d00cfff1d3223/tests/test_common.py#L3

Make recipes take variables so that they could be reused more easily.

Example:

The recipe:

max_cells = RecipeInfo(
    src="len(nb.cells) < amount_max",
    description="Assert that there are less than 'amount_max' cells in the notebook.",
)

The command:

databooks assert path/to/nbs --recipe max_cells --amount_max 10

databooks assert path/to/nbs --recipe max_cells --help

Would return the arguments of that recipe

New CLI command! 🎉

A databooks assert ... command that will take an expression and evaluate against all notebooks in path (internally databooks.affirm.affirm to hopefully remind of python's assert without creating any ambiguity). The command also comes with recipes that are short-hand for some useful expressions. These can also be used as inspiration for new expressions.

Variables in scope for expression:

• nb: Jupyter notebook found in path
• raw_cells: notebook cells of raw type
• md_cells: notebook cells of markdown type
• code_cells: notebook cells of code type
• exec_cells: notebook cells of executed code type

It uses python's eval. But, since eval is really dangerous we first parse the string and only whitelist some nodes/attributes.

• Allowed nodes are in databooks.affirm._ALLOWED_NODES
• Allowed functions are in databooks.affirm._ALLOWED_BUILTINS
• Only class attributes are allowed (found Pydantic model attributes using dict(model).keys())

Check tests/test_affirm.py for examples of valid/invalid expressions.

Package uses logging functionalty but introduces a "verbose" concept where loglevel INFO suddendly becomes more verbose. This is a small PR that switches log level level to DEBUG when --verbose flag is passed by the user.

Changes:

• Add custom exception when parsing HTML to table fails
• Add try-except for that exception when parsing outputs - when there are issues, fallback to text implementation

Add rich table (HTML) rendering to notebook.

• Other HTML elements are not displayed.
• Fall back to displaying text if that's the case
• Refactor logic to simplify this
• Refactor logic to gracefully show not supported mime types (anything that is not one of image/png, text/html, text/plain

• We should not need the git repo for meta and assert commands
• Config files/repos should always be located from the filepaths, not current working directory
  • Only exception is when we run databooks diff with no paths - in that case we use the current dir to find the repo
  • Not the case for databooks fix requires paths

Closes https://github.com/datarootsio/databooks/issues/48

Enhancements:

• If diff is only on code/output, then only show the split for that section
• Show when only cell/notebook metadata changes - maybe make cell border yellow, etc...
• For cell source diffs, show lines/character changes by changing the background color (red/green)
• Add rich rendering for other outputs
  • HTML (other than tables)
  • PNG
  • SVG
  • ...

I like the way databooks is able to show the content of the notebook. It might make sense to implement a run command which can run all cells in a notebook and show the progress in the CLI interface cell by cell. We don't need to implement the execution because it can be done by nbconvert or nbclient packages.

What do you think?

I'd like to use databooks in the pre-commit Git hook, but my notebooks are in a sub-folder called "notebooks". Is there a way to specify that in the pre-commit?