(tempdir, dataset_reader)
| 3389 | |
| 3390 | @pytest.mark.orc |
| 3391 | def test_orc_format(tempdir, dataset_reader): |
| 3392 | from pyarrow import orc |
| 3393 | table = pa.table({'a': pa.array([1, 2, 3], type="int8"), |
| 3394 | 'b': pa.array([.1, .2, .3], type="float64")}) |
| 3395 | |
| 3396 | path = str(tempdir / 'test.orc') |
| 3397 | orc.write_table(table, path) |
| 3398 | |
| 3399 | dataset = ds.dataset(path, format=ds.OrcFileFormat()) |
| 3400 | fragments = list(dataset.get_fragments()) |
| 3401 | assert isinstance(fragments[0], ds.FileFragment) |
| 3402 | result = dataset_reader.to_table(dataset) |
| 3403 | result.validate(full=True) |
| 3404 | assert result.equals(table) |
| 3405 | |
| 3406 | assert_dataset_fragment_convenience_methods(dataset) |
| 3407 | |
| 3408 | dataset = ds.dataset(path, format="orc") |
| 3409 | result = dataset_reader.to_table(dataset) |
| 3410 | result.validate(full=True) |
| 3411 | assert result.equals(table) |
| 3412 | |
| 3413 | result = dataset_reader.to_table(dataset, columns=["b"]) |
| 3414 | result.validate(full=True) |
| 3415 | assert result.equals(table.select(["b"])) |
| 3416 | |
| 3417 | result = dataset_reader.to_table( |
| 3418 | dataset, columns={"b2": ds.field("b") * 2} |
| 3419 | ) |
| 3420 | result.validate(full=True) |
| 3421 | assert result.equals( |
| 3422 | pa.table({'b2': pa.array([.2, .4, .6], type="float64")}) |
| 3423 | ) |
| 3424 | |
| 3425 | assert dataset_reader.count_rows(dataset) == 3 |
| 3426 | assert dataset_reader.count_rows(dataset, filter=ds.field("a") > 2) == 1 |
| 3427 | |
| 3428 | |
| 3429 | @pytest.mark.orc |
nothing calls this directly
no test coverage detected