module 'pyarrow' has no attribute 'from_pandas'

[DanielKerrigan]

I'm running into the following error when using VegaFusion 2.0.1 with pandas 2.1.4, Altair 5.5.0, and pyarrow 18.1.0:

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
File ~/miniconda3/envs/version-test/lib/python3.10/site-packages/IPython/core/formatters.py:1036, in MimeBundleFormatter.__call__(self, obj, include, exclude)
   1033     method = get_real_method(obj, self.print_method)
   1035     if method is not None:
-> 1036         return method(include=include, exclude=exclude)
   1037     return None
   1038 else:

File ~/miniconda3/envs/version-test/lib/python3.10/site-packages/altair/vegalite/v5/api.py:3682, in TopLevelMixin._repr_mimebundle_(self, *args, **kwds)
   3680 else:
   3681     if renderer := renderers.get():
-> 3682         return renderer(dct)

File ~/miniconda3/envs/version-test/lib/python3.10/site-packages/altair/utils/display.py:232, in HTMLRenderer.__call__(self, spec, **metadata)
    230 kwargs = self.kwargs.copy()
    231 kwargs.update(**metadata, output_div=self.output_div)
--> 232 return spec_to_mimebundle(spec, format="html", **kwargs)

File ~/miniconda3/envs/version-test/lib/python3.10/site-packages/altair/utils/mimebundle.py:129, in spec_to_mimebundle(spec, format, mode, vega_version, vegaembed_version, vegalite_version, embed_options, engine, **kwargs)
    127 internal_mode: Literal["vega-lite", "vega"] = mode
    128 if using_vegafusion():
--> 129     spec = compile_with_vegafusion(spec)
    130     internal_mode = "vega"
    132 # Default to the embed options set by alt.renderers.set_embed_options

File ~/miniconda3/envs/version-test/lib/python3.10/site-packages/altair/utils/_vegafusion_data.py:273, in compile_with_vegafusion(vegalite_spec)
    271 # Pre-evaluate transforms in vega spec with vegafusion
    272 row_limit = data_transformers.options.get("max_rows", None)
--> 273 transformed_vega_spec, warnings = vf.runtime.pre_transform_spec(
    274     vega_spec,
    275     vf.get_local_tz(),
    276     inline_datasets=inline_tables,
    277     row_limit=row_limit,
    278 )
    280 # Check from row limit warning and convert to MaxRowsError
    281 handle_row_limit_exceeded(row_limit, warnings)

File ~/miniconda3/envs/version-test/lib/python3.10/site-packages/vegafusion/runtime.py:408, in VegaFusionRuntime.pre_transform_spec(self, spec, local_tz, default_input_tz, row_limit, preserve_interactivity, inline_datasets, keep_signals, keep_datasets)
    347 """
    348 Evaluate supported transforms in an input Vega specification
    349 
   (...)
    405         were eligible for pre-transforming
    406 """
    407 local_tz = local_tz or get_local_tz()
--> 408 imported_inline_dataset = self._import_inline_datasets(
    409     inline_datasets, get_inline_column_usage(spec)
    410 )
    412 new_spec, warnings = self.runtime.pre_transform_spec(
    413     spec,
    414     local_tz=local_tz,
   (...)
    420     keep_datasets=parse_variables(keep_datasets),
    421 )
    423 return new_spec, warnings

File ~/miniconda3/envs/version-test/lib/python3.10/site-packages/vegafusion/runtime.py:303, in VegaFusionRuntime._import_inline_datasets(self, inline_datasets, inline_dataset_usage)
    300         imported_inline_datasets[name] = Table(inner_value)
    301     else:
    302         # Older pandas, convert through pyarrow
--> 303         imported_inline_datasets[name] = Table(pa.from_pandas(inner_value))
    304 elif isinstance(value, dict):
    305     # Let narwhals import the dict using a default backend
    306     df_nw = nw.from_dict(value, native_namespace=_get_default_namespace())

AttributeError: module 'pyarrow' has no attribute 'from_pandas'

To reproduce this error, I used the following Conda environment:

name: version-test
channels:
  - conda-forge
dependencies:
  - python=3.10
  - pandas=2.1.4
  - notebook
  - altair-all=5.5.0

and then ran a notebook containing this example:

import altair as alt
import pandas as pd

alt.data_transformers.enable("vegafusion")

source = pd.DataFrame({
    'a': ['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I'],
    'b': [28, 55, 43, 91, 81, 53, 19, 87, 52]
})

alt.Chart(source).mark_bar().encode(
    x='a',
    y='b'
)

[MarcoGorelli]

thanks for the report

vegafusion/vegafusion-python/vegafusion/runtime.py

Lines 302 to 303 in 8e0f7e2

	# Older pandas, convert through pyarrow
	imported_inline_datasets[name] = Table(pa.from_pandas(inner_value))

this should probably be pa.Table.from_pandas?

[jonmmease]

Ahh, yeah that's a good point. Thanks for looking into it @MarcoGorelli