(WIP) Add support for ARRAY standard SQL type

Author: plamut

bigquery/google/cloud/bigquery/schema.py

@@ -37,6 +37,7 @@
     "DATE": types.StandardSqlDataType.DATE,
     "TIME": types.StandardSqlDataType.TIME,
     "DATETIME": types.StandardSqlDataType.DATETIME,
+    # no direct conversion from ARRAY, the latter is represented by mode="REPEATED"
 }
 """String names of the legacy SQL types to integer codes of Standard SQL types."""
 
@@ -179,13 +180,25 @@ def to_standard_sql(self):
             An instance of :class:`~google.cloud.bigquery_v2.types.StandardSqlField`.
         """
         sql_type = types.StandardSqlDataType()
-        sql_type.type_kind = LEGACY_TO_STANDARD_TYPES.get(
-            self.field_type, types.StandardSqlDataType.TYPE_KIND_UNSPECIFIED
-        )
 
-        # NOTE: No need to also handle the "ARRAY" composed type, the latter
-        # does not exist in legacy SQL types.
-        if sql_type.type_kind == types.StandardSqlDataType.STRUCT:  # noqa: E721
+        if self.mode == "REPEATED":
+            sql_type.type_kind = types.StandardSqlDataType.ARRAY
+        else:
+            sql_type.type_kind = LEGACY_TO_STANDARD_TYPES.get(
+                self.field_type, types.StandardSqlDataType.TYPE_KIND_UNSPECIFIED
+            )
+
+        if sql_type.type_kind == types.StandardSqlDataType.ARRAY:  # noqa: E721
+            array_element_type = LEGACY_TO_STANDARD_TYPES.get(
+                self.field_type, types.StandardSqlDataType.TYPE_KIND_UNSPECIFIED
+            )
+            # ARRAY cannot directly contain other arrays, only scalar types and STRUCTs
+            # https://cloud.google.com/bigquery/docs/reference/standard-sql/data-types#array-type
+            if array_element_type != types.StandardSqlDataType.STRUCT:  # noqa: E721
+                sql_type.array_element_type.type_kind = array_element_type
+            else:
+                raise NotImplementedError("no support yet for array of STRUCT")
+        elif sql_type.type_kind == types.StandardSqlDataType.STRUCT:  # noqa: E721
             sql_type.struct_type.fields.extend(
                 field.to_standard_sql() for field in self.fields
             )

bigquery/tests/unit/test_schema.py

@@ -256,6 +256,25 @@ def test_to_standard_sql_struct_type(self):
 
         self.assertEqual(standard_field, expected_result)
 
+    def test_to_standard_sql_array_type_simple(self):
+        from google.cloud.bigquery_v2 import types
+
+        # construct expected result object
+        sql_type = types.StandardSqlDataType(type_kind=types.StandardSqlDataType.ARRAY)
+        sql_type.array_element_type.type_kind = types.StandardSqlDataType.INT64
+        expected_result = types.StandardSqlField(name="valid_numbers", type=sql_type)
+
+        # construct "repeated" SchemaField object and convert to standard SQL
+        schema_field = self._make_one("valid_numbers", "INT64", mode="REPEATED")
+        standard_field = schema_field.to_standard_sql()
+
+        self.assertEqual(standard_field, expected_result)
+
+    def test_to_standard_sql_array_type_struct(self):
+        # from google.cloud.bigquery_v2 import types
+        pass
+        # TODO: implement test
+
     def test_to_standard_sql_unknown_type(self):
         sql_type = self._get_standard_sql_data_type_class()
         field = self._make_one("weird_field", "TROOLEAN")