()
| 33 | |
| 34 | |
| 35 | def test_make_meta(): |
| 36 | df = pd.DataFrame( |
| 37 | {"a": [1, 2, 3], "b": list("abc"), "c": [1.0, 2.0, 3.0]}, index=[10, 20, 30] |
| 38 | ) |
| 39 | |
| 40 | # Pandas dataframe |
| 41 | meta = make_meta(df) |
| 42 | assert len(meta) == 0 |
| 43 | assert (meta.dtypes == df.dtypes).all() |
| 44 | assert isinstance(meta.index, type(df.index)) |
| 45 | # - ensure no references to original data arrays are kept |
| 46 | for col in "abc": |
| 47 | if PANDAS_GE_300 and col == "b": |
| 48 | # backed by an arrow array |
| 49 | meta_pointer = meta[col].array._pa_array.chunks[0].buffers()[1].address |
| 50 | df_pointer = df[col].array._pa_array.chunks[0].buffers()[1].address |
| 51 | else: |
| 52 | # backed by a numpy array |
| 53 | meta_pointer = meta[col].values.__array_interface__["data"][0] |
| 54 | df_pointer = df[col].values.__array_interface__["data"][0] |
| 55 | assert meta_pointer != df_pointer |
| 56 | |
| 57 | meta_index_pointer = meta.index.values.__array_interface__["data"][0] |
| 58 | df_index_pointer = df.index.values.__array_interface__["data"][0] |
| 59 | assert meta_index_pointer != df_index_pointer |
| 60 | |
| 61 | # Pandas series |
| 62 | meta = make_meta(df.a) |
| 63 | assert len(meta) == 0 |
| 64 | assert meta.dtype == df.a.dtype |
| 65 | assert isinstance(meta.index, type(df.index)) |
| 66 | # - ensure no references to original data arrays are kept |
| 67 | meta_pointer = meta.values.__array_interface__["data"][0] |
| 68 | df_pointer = df.a.values.__array_interface__["data"][0] |
| 69 | assert meta_pointer != df_pointer |
| 70 | meta_pointer = meta.index.values.__array_interface__["data"][0] |
| 71 | df_pointer = df.index.values.__array_interface__["data"][0] |
| 72 | assert meta_pointer != df_pointer |
| 73 | |
| 74 | # Pandas index |
| 75 | meta = make_meta(df.index) |
| 76 | assert isinstance(meta, type(df.index)) |
| 77 | assert len(meta) == 0 |
| 78 | # - ensure no references to original data arrays are kept |
| 79 | meta_pointer = meta.values.__array_interface__["data"][0] |
| 80 | df_pointer = df.index.values.__array_interface__["data"][0] |
| 81 | assert meta_pointer != df_pointer |
| 82 | |
| 83 | # Dask object |
| 84 | ddf = dd.from_pandas(df, npartitions=2) |
| 85 | assert make_meta(ddf) is ddf._meta |
| 86 | |
| 87 | # Dict |
| 88 | if PANDAS_GE_300: |
| 89 | meta = make_meta({"a": "i8", "b": "str", "c": "f8"}) |
| 90 | else: |
| 91 | meta = make_meta({"a": "i8", "b": "O", "c": "f8"}) |
| 92 |
nothing calls this directly
no test coverage detected
searching dependent graphs…