Fix ArrowEngine bug in use of clear_known_categories (#6887)

* fix bug calling clear_known_categories

* handle index in general - not just unnamed

* remove unused NONE_LABEL

* no need to target index explicitly

dask/dataframe/io/parquet/arrow.py

@@ -23,7 +23,6 @@
 preserve_ind_supported = pa.__version__ >= LooseVersion("0.15.0")
 schema_field_supported = pa.__version__ >= LooseVersion("0.15.0")
 
-
 #
 #  Private Helper Functions
 #
@@ -314,7 +313,12 @@ def _generate_dd_meta(schema, index, categories, partition_info):
 
     index_cols = index or ()
     meta = _meta_from_dtypes(all_columns, dtypes, index_cols, column_index_names)
-    meta = clear_known_categories(meta, cols=categories)
+    if categories:
+        # Make sure all categories are set to "unknown".
+        # Cannot include index names in the `cols` argument.
+        meta = clear_known_categories(
+            meta, cols=[c for c in categories if c not in meta.index.names]
+        )
 
     if partition_obj:
         for partition in partition_obj:

dask/dataframe/io/tests/test_parquet.py

@@ -914,6 +914,27 @@ def test_categories(tmpdir, engine):
         ddf2 = dd.read_parquet(fn, categories=["foo"], engine=engine)
 
 
+def test_categories_unnamed_index(tmpdir, engine):
+    # Check that we can handle an unnamed categorical index
+    # https://github.com/dask/dask/issues/6885
+
+    if engine == "pyarrow" and pa.__version__ < LooseVersion("0.15.0"):
+        pytest.skip("PyArrow>=0.15 Required.")
+
+    tmpdir = str(tmpdir)
+
+    df = pd.DataFrame(
+        data={"A": [1, 2, 3], "B": ["a", "a", "b"]}, index=["x", "y", "y"]
+    )
+    ddf = dd.from_pandas(df, npartitions=1)
+    ddf = ddf.categorize(columns=["B"])
+
+    ddf.to_parquet(tmpdir, engine=engine)
+    ddf2 = dd.read_parquet(tmpdir, engine=engine)
+
+    assert_eq(ddf.index, ddf2.index, check_divisions=False)
+
+
 def test_empty_partition(tmpdir, engine):
     fn = str(tmpdir)
     df = pd.DataFrame({"a": range(10), "b": range(10)})