[Python] Selective projection of struct fields errors with use_legacy_dataset = False

[asfimport]

Selectively projecting fields from within a struct when reading from parquet files triggers an ArrowInvalid error when using the new dataset api (use_legacy_dataset=False). Passing use_legacy_dataset=True yields the expected behavior: loading only the columns enumerated in the columns argument, recursing into structs if there is a . delimeter in the field name.

Using the following test table:

df = pd.DataFrame({
 'user_id': ['abc123', 'qrs456'],
 'interaction': [{'type': 'click', 'element': 'button'}, {'type':'scroll', 'element': 'window'}]
})

table = pa.Table.from_pandas(df)

pq.write_table(table, 'example.parquet')

Using the current default datasets API:

table_latest = pq.read_table('example.parquet', columns = ['user_id', 'interaction.type'])

yields:

---------------------------------------------------------------------------
ArrowInvalid Traceback (most recent call last)
<ipython-input-25-982ca2d96075> in <module>
----> 1 table_latest = pq.read_table('/'.join([out_path, 'example.parquet']), columns = ['user_id', 'interaction.type'], filesystem = fs)
 2 table_latest

/usr/local/share/sciencebox/venv/lib/python3.6/site-packages/pyarrow/parquet.py in read_table(source, columns, use_threads, metadata, use_pandas_metadata, memory_map, read_dictionary, filesystem, filters, buffer_size, partitioning, use_legacy_dataset, ignore_prefixes, pre_buffer, coerce_int96_timestamp_unit)
 1894 
 1895 return dataset.read(columns=columns, use_threads=use_threads,
-> 1896 use_pandas_metadata=use_pandas_metadata)
 1897 
 1898 if ignore_prefixes is not None:

/usr/local/share/sciencebox/venv/lib/python3.6/site-packages/pyarrow/parquet.py in read(self, columns, use_threads, use_pandas_metadata)
 1744 table = self._dataset.to_table(
 1745 columns=columns, filter=self._filter_expression,
-> 1746 use_threads=use_threads
 1747 )
 1748 

/usr/local/share/sciencebox/venv/lib/python3.6/site-packages/pyarrow/_dataset.pyx in pyarrow._dataset.Dataset.to_table()

/usr/local/share/sciencebox/venv/lib/python3.6/site-packages/pyarrow/_dataset.pyx in pyarrow._dataset.Dataset.scanner()

/usr/local/share/sciencebox/venv/lib/python3.6/site-packages/pyarrow/_dataset.pyx in pyarrow._dataset.Scanner.from_dataset()

/usr/local/share/sciencebox/venv/lib/python3.6/site-packages/pyarrow/_dataset.pyx in pyarrow._dataset._populate_builder()

/usr/local/share/sciencebox/venv/lib/python3.6/site-packages/pyarrow/error.pxi in pyarrow.lib.check_status()

ArrowInvalid: No match for FieldRef.Name(interaction.type) in user_id: string
interaction: struct<element: string, type: string>

Whereas: 

table_legacy = pq.read_table('example.parquet', columns = ['user_id', 'interaction.type'], use_legacy_dataset = True)

Yields:

pyarrow.Table
user_id: string
interaction: struct<type: string>
 child 0, type: string

Environment: Python 3.6.9

Reporter: Mark Grey

Related issues:

• [Python] Allow to create field reference to nested field (is blocked by)

_{Note: This issue was originally created as ARROW-13798. Please see the migration documentation for further details.}

[asfimport]

Mark Grey:
After looking into the implementation, it seems this is to some extent a known limitation of the new Dataset API, perhaps arising from ARROW-11259

[asfimport]

Joris Van den Bossche / @jorisvandenbossche:
[~mgthesecond] sorry for the late reply, but this indeed a known limitation of the new Datasets API. The JIRA you linked is indeed tracking this.