fix client.insert_dataframe() for tables with tuple columns with acceptance of numpy.ndarrays

[stankudrow]

• fixes

  • TypeError: Unsupported column type: <class 'numpy.ndarray'>. list or tuple is expected. #356
  • Problems while inserting into a table with a Tuple column #417

• reconsiders the PR fix client.insert_dataframe() for tuple columns in tables #425

Packages:

• black : 24.2.0
• clickhouse-cityhash : 1.0.2.4
• clickhouse-driver : 0.2.7
• Cython : 3.0.9
• flake8 : 7.0.0
• freezegun : 1.4.0
• lz4 : 4.3.3
• mypy : 1.9.0
• numpy : 1.26.4
• pandas : 2.2.1
• parametrized : 66.0.2 (there is the latest 66.0.3 -> to be removed in favour of pytest.mark.parametrize)
• pytest : 7.4.4
• ruff : 0.3.2
• zstd : 1.5.5.1

Checklist:

• Add tests that demonstrate the correct behaviour of the change.
• Add or update relevant docs, in the docs folder and in code -> no docs to change.
• Ensure PR doesn't contain untouched code reformatting: spaces, etc.
• Run flake8 and fix issues;
• Run pytest no tests failed. See the dev docs.
• Update CHANGELOG.md

[stankudrow]

@xzkostyan , does it look better now?

[stankudrow]

@xzkostyan , the CHANGELOG.md was updated and the PR seems to bear the solution for the #356 issue.

[xzkostyan]

Good job.

Let's keep non-numpy/numpy things separated.

[xzkostyan]

Numpy/pandas helpers should be located in clickhouse_driver/numpy/helpers.py.

I'd keep two different versions of chunks: one for non-numpy and another one for numpy.

[xzkostyan]

What about speed and memory consumption here? Did you test the new code with large dataframes?

[stankudrow]

This part is to be refined later.

[xzkostyan]

Please see how simple numpy/pandas tests are implemented: tests/numpy/test_generic.py. Skipping logic (@skipIf) is hidden under the hood. Variable PANDAS_IMPORTED is redundant in this case.