[Python] Exercise fallback case on tests for parquet.read_table in case dataset is not available

[raulcd]

Describe the enhancement requested

While reviewing:

• GH-46374: [Python][Docs] Adds type checks for source in read_table #46330

I've realized that the following code path is never tested:

arrow/python/pyarrow/parquet/core.py

Lines 1807 to 1838 in e2a5b4e

	except ImportError:
	# fall back on ParquetFile for simple cases when pyarrow.dataset
	# module is not available
	if filters is not None:
	raise ValueError(
	"the 'filters' keyword is not supported when the "
	"pyarrow.dataset module is not available"
	)
	if partitioning != "hive":
	raise ValueError(
	"the 'partitioning' keyword is not supported when the "
	"pyarrow.dataset module is not available"
	)
	if schema is not None:
	raise ValueError(
	"the 'schema' argument is not supported when the "
	"pyarrow.dataset module is not available"
	)
	filesystem, path = _resolve_filesystem_and_path(source, filesystem)
	if filesystem is not None:
	source = filesystem.open_input_file(path)
	# TODO test that source is not a directory or a list
	dataset = ParquetFile(
	source, read_dictionary=read_dictionary,
	memory_map=memory_map, buffer_size=buffer_size,
	pre_buffer=pre_buffer,
	coerce_int96_timestamp_unit=coerce_int96_timestamp_unit,
	decryption_properties=decryption_properties,
	thrift_string_size_limit=thrift_string_size_limit,
	thrift_container_size_limit=thrift_container_size_limit,
	page_checksum_verification=page_checksum_verification,
	)

I've tried to enable Parquet locally on the minimal Python example build but this would require to build with snappy and others. It would be good to have a test that exercises this code path.

This was tested before due to the use of legacy_dataset flag.
See:

• GH-31303: [Python] Remove the legacy ParquetDataset custom python-based implementation #39112

Component(s)

Python

[raulcd]

cc @AlenkaF @rok