| 4856 | |
| 4857 | |
| 4858 | def test_metadata_compat_missing_field_name(): |
| 4859 | # Combination of missing field name but with index column as metadata. |
| 4860 | # This combo occurs in the latest versions of fastparquet (0.3.2), but not |
| 4861 | # in pyarrow itself (since field_name was added in 0.8, index as metadata |
| 4862 | # only added later) |
| 4863 | |
| 4864 | a_values = [1, 2, 3, 4] |
| 4865 | b_values = ['a', 'b', 'c', 'd'] |
| 4866 | a_arrow = pa.array(a_values, type='int64') |
| 4867 | b_arrow = pa.array(b_values, type='utf8') |
| 4868 | |
| 4869 | expected = pd.DataFrame({ |
| 4870 | 'a': a_values, |
| 4871 | 'b': b_values, |
| 4872 | }, index=pd.RangeIndex(0, 8, step=2, name='qux')) |
| 4873 | table = pa.table({'a': a_arrow, 'b': b_arrow}) |
| 4874 | |
| 4875 | # metadata generated by fastparquet 0.3.2 with missing field_names |
| 4876 | table = table.replace_schema_metadata({ |
| 4877 | b'pandas': json.dumps({ |
| 4878 | 'column_indexes': [ |
| 4879 | {'field_name': None, |
| 4880 | 'metadata': None, |
| 4881 | 'name': None, |
| 4882 | 'numpy_type': 'object', |
| 4883 | 'pandas_type': 'mixed-integer'} |
| 4884 | ], |
| 4885 | 'columns': [ |
| 4886 | {'metadata': None, |
| 4887 | 'name': 'a', |
| 4888 | 'numpy_type': 'int64', |
| 4889 | 'pandas_type': 'int64'}, |
| 4890 | {'metadata': None, |
| 4891 | 'name': 'b', |
| 4892 | 'numpy_type': 'object', |
| 4893 | 'pandas_type': 'unicode'} |
| 4894 | ], |
| 4895 | 'index_columns': [ |
| 4896 | {'kind': 'range', |
| 4897 | 'name': 'qux', |
| 4898 | 'start': 0, |
| 4899 | 'step': 2, |
| 4900 | 'stop': 8} |
| 4901 | ], |
| 4902 | 'pandas_version': '0.25.0'} |
| 4903 | |
| 4904 | )}) |
| 4905 | result = table.to_pandas() |
| 4906 | tm.assert_frame_equal(result, expected) |
| 4907 | |
| 4908 | |
| 4909 | def test_metadata_index_name_not_json_serializable(): |