(tempdir)
| 4836 | |
| 4837 | |
| 4838 | def test_write_iterable(tempdir): |
| 4839 | table = pa.table([ |
| 4840 | pa.array(range(20)), pa.array(random.random() for _ in range(20)), |
| 4841 | pa.array(['a'] * 10 + ['b'] * 10) |
| 4842 | ], names=["f1", "f2", "part"]) |
| 4843 | |
| 4844 | base_dir = tempdir / 'inmemory_iterable' |
| 4845 | ds.write_dataset((batch for batch in table.to_batches()), base_dir, |
| 4846 | schema=table.schema, |
| 4847 | basename_template='dat_{i}.arrow', format="ipc") |
| 4848 | result = ds.dataset(base_dir, format="ipc").to_table() |
| 4849 | assert result.equals(table) |
| 4850 | |
| 4851 | base_dir = tempdir / 'inmemory_reader' |
| 4852 | reader = pa.RecordBatchReader.from_batches(table.schema, |
| 4853 | table.to_batches()) |
| 4854 | ds.write_dataset(reader, base_dir, basename_template='dat_{i}.arrow', format="ipc") |
| 4855 | result = ds.dataset(base_dir, format="ipc").to_table() |
| 4856 | assert result.equals(table) |
| 4857 | |
| 4858 | base_dir = tempdir / 'inmemory_pycapsule' |
| 4859 | stream = TableStreamWrapper(table) |
| 4860 | ds.write_dataset(stream, base_dir, basename_template='dat_{i}.arrow', format="ipc") |
| 4861 | result = ds.dataset(base_dir, format="ipc").to_table() |
| 4862 | assert result.equals(table) |
| 4863 | |
| 4864 | |
| 4865 | def test_write_scanner(tempdir, dataset_reader): |
nothing calls this directly
no test coverage detected