Merge pull request duckdb#11464 from Tishj/arrow_invalid_struct

[Arrow] Throw on invalid STRUCT type

src/function/table/arrow.cpp

@@ -125,6 +125,10 @@ static unique_ptr<ArrowType> GetArrowLogicalTypeNoDictionary(ArrowSchema &schema
 	} else if (format == "+s") {
 		child_list_t<LogicalType> child_types;
 		vector<unique_ptr<ArrowType>> children;
+		if (schema.n_children == 0) {
+			throw InvalidInputException(
+			    "Attempted to convert a STRUCT with no fields to DuckDB which is not supported");
+		}
 		for (idx_t type_idx = 0; type_idx < (idx_t)schema.n_children; type_idx++) {
 			children.emplace_back(ArrowTableFunction::GetArrowLogicalType(*schema.children[type_idx]));
 			child_types.emplace_back(schema.children[type_idx]->name, children.back()->GetDuckType());
@@ -144,6 +148,9 @@ static unique_ptr<ArrowType> GetArrowLogicalTypeNoDictionary(ArrowSchema &schema
 
 		child_list_t<LogicalType> members;
 		vector<unique_ptr<ArrowType>> children;
+		if (schema.n_children == 0) {
+			throw InvalidInputException("Attempted to convert a UNION with no fields to DuckDB which is not supported");
+		}
 		for (idx_t type_idx = 0; type_idx < (idx_t)schema.n_children; type_idx++) {
 			auto type = schema.children[type_idx];
 

tools/pythonpkg/tests/fast/arrow/test_arrow_types.py

@@ -1,27 +1,47 @@
 import duckdb
+import pytest
 
-try:
-    import pyarrow as pa
-    import pyarrow.dataset as ds
-
-    can_run = True
-except:
-    can_run = False
+pa = pytest.importorskip("pyarrow")
+ds = pytest.importorskip("pyarrow.dataset")
 
 
 class TestArrowTypes(object):
     def test_null_type(self, duckdb_cursor):
-        if not can_run:
-            return
         schema = pa.schema([("data", pa.null())])
         inputs = [pa.array([None, None, None], type=pa.null())]
         arrow_table = pa.Table.from_arrays(inputs, schema=schema)
-        duckdb_conn = duckdb.connect()
-        duckdb_conn.register("testarrow", arrow_table)
+        duckdb_cursor.register("testarrow", arrow_table)
         rel = duckdb.from_arrow(arrow_table).arrow()
         # We turn it to an array of int32 nulls
         schema = pa.schema([("data", pa.int32())])
         inputs = [pa.array([None, None, None], type=pa.null())]
         arrow_table = pa.Table.from_arrays(inputs, schema=schema)
 
         assert rel['data'] == arrow_table['data']
+
+    def test_invalid_struct(self, duckdb_cursor):
+        empty_struct_type = pa.struct([])
+
+        # Create an empty array with the defined struct type
+        empty_array = pa.array([], type=empty_struct_type)
+        arrow_table = pa.Table.from_arrays([empty_array], schema=pa.schema([("data", empty_struct_type)]))
+        with pytest.raises(
+            duckdb.InvalidInputException,
+            match='Attempted to convert a STRUCT with no fields to DuckDB which is not supported',
+        ):
+            duckdb_cursor.register('invalid_struct', arrow_table)
+
+    def test_invalid_union(self, duckdb_cursor):
+        # Create a sparse union array from dense arrays
+        types = pa.array([0, 1, 1], type=pa.int8())
+        sparse_union_array = pa.UnionArray.from_sparse(types, [], type_codes=[])
+
+        arrow_table = pa.Table.from_arrays([sparse_union_array], schema=pa.schema([("data", sparse_union_array.type)]))
+        with pytest.raises(
+            duckdb.InvalidInputException,
+            match='Attempted to convert a UNION with no fields to DuckDB which is not supported',
+        ):
+            duckdb_cursor.register('invalid_union', arrow_table)
+
+            res = duckdb_cursor.sql("select * from invalid_union").fetchall()
+            print(res)