Minimal fix for categoricals

rapidsai · May 30, 2024 · 1859149 · 1859149
1 parent 9f0c30b
commit 1859149
Showing 1 changed file with 7 additions and 0 deletions.
diff --git a/python/cudf/cudf/core/frame.py b/python/cudf/cudf/core/frame.py
@@ -897,6 +897,13 @@ def from_arrow(cls, data: pa.Table) -> Self:
         # so handling indices and dictionary as two different columns.
         # This needs be removed once we have hooked libcudf dictionary32
         # with categorical.
+        if any(
+            isinstance(x.type, pa.DictionaryType)
+            and isinstance(x, pa.ChunkedArray)
+            for x in data
+        ):
+            data = data.combine_chunks()
+
         dict_indices = {}
         dict_dictionaries = {}
         dict_ordered = {}