feat: add ability to set autodetect_schema query param in update_table (#2171)

* Add ability to set autodetect_schema query_param

* fixup! Add ability to set autodetect_schema query_param

* fixup! Add ability to set autodetect_schema query_param

* fixup! Add ability to set autodetect_schema query_param

---------

Co-authored-by: Chalmer Lowe <[email protected]>

Author: TheNeuralBit

google/cloud/bigquery/client.py

@@ -1389,6 +1389,7 @@ def update_table(
         self,
         table: Table,
         fields: Sequence[str],
+        autodetect_schema: bool = False,
         retry: retries.Retry = DEFAULT_RETRY,
         timeout: TimeoutType = DEFAULT_TIMEOUT,
     ) -> Table:
@@ -1419,6 +1420,10 @@ def update_table(
             fields (Sequence[str]):
                 The fields of ``table`` to change, spelled as the
                 :class:`~google.cloud.bigquery.table.Table` properties.
+            autodetect_schema (bool):
+                Specifies if the schema of the table should be autodetected when
+                updating the table from the underlying source. Only applicable
+                for external tables.
             retry (Optional[google.api_core.retry.Retry]):
                 A description of how to retry the API call.
             timeout (Optional[float]):
@@ -1438,12 +1443,18 @@ def update_table(
         path = table.path
         span_attributes = {"path": path, "fields": fields}
 
+        if autodetect_schema:
+            query_params = {"autodetect_schema": True}
+        else:
+            query_params = {}
+
         api_response = self._call_api(
             retry,
             span_name="BigQuery.updateTable",
             span_attributes=span_attributes,
             method="PATCH",
             path=path,
+            query_params=query_params,
             data=partial,
             headers=headers,
             timeout=timeout,

tests/system/test_client.py

@@ -978,6 +978,53 @@ def test_update_table_constraints(self):
         )
         self.assertIsNone(reference_table3.table_constraints, None)
 
+    def test_update_table_autodetect_schema(self):
+        dataset = self.temp_dataset(_make_dataset_id("bq_update_table_test"))
+
+        # Create an external table, restrict schema to one field
+        TABLE_NAME = "test_table"
+        set_schema = [bigquery.SchemaField("username", "STRING", mode="NULLABLE")]
+        table_arg = Table(dataset.table(TABLE_NAME))
+
+        # Create an external_config and include it in the table arguments
+        external_config = bigquery.ExternalConfig(bigquery.ExternalSourceFormat.AVRO)
+        external_config.source_uris = SOURCE_URIS_AVRO
+        external_config.reference_file_schema_uri = REFERENCE_FILE_SCHEMA_URI_AVRO
+        external_config.schema = set_schema
+        table_arg.external_data_configuration = external_config
+
+        self.assertFalse(_table_exists(table_arg))
+
+        table = helpers.retry_403(Config.CLIENT.create_table)(table_arg)
+        self.to_delete.insert(0, table)
+        self.assertTrue(_table_exists(table))
+
+        self.assertEqual(table.schema, set_schema)
+
+        # Update table with schema autodetection
+        updated_table_arg = Table(dataset.table(TABLE_NAME))
+
+        # Update the external_config and include it in the table arguments
+        updated_external_config = copy.deepcopy(external_config)
+        updated_external_config.autodetect = True
+        updated_external_config.schema = None
+        updated_table_arg.external_data_configuration = updated_external_config
+
+        # PATCH call with autodetect_schema=True to trigger schema inference
+        updated_table = Config.CLIENT.update_table(
+            updated_table_arg, ["external_data_configuration"], autodetect_schema=True
+        )
+
+        # The updated table should have a schema inferred from the reference
+        # file, which has all four fields.
+        expected_schema = [
+            bigquery.SchemaField("username", "STRING", mode="NULLABLE"),
+            bigquery.SchemaField("tweet", "STRING", mode="NULLABLE"),
+            bigquery.SchemaField("timestamp", "STRING", mode="NULLABLE"),
+            bigquery.SchemaField("likes", "INTEGER", mode="NULLABLE"),
+        ]
+        self.assertEqual(updated_table.schema, expected_schema)
+
     @staticmethod
     def _fetch_single_page(table, selected_fields=None):
         iterator = Config.CLIENT.list_rows(table, selected_fields=selected_fields)

tests/unit/test_client.py

@@ -2385,7 +2385,7 @@ def test_update_table(self):
             "resourceTags": {"123456789012/key": "value"},
         }
         conn.api_request.assert_called_once_with(
-            method="PATCH", data=sent, path="/" + path, timeout=7.5
+            method="PATCH", data=sent, path="/" + path, timeout=7.5, query_params={}
         )
         self.assertEqual(updated_table.description, table.description)
         self.assertEqual(updated_table.friendly_name, table.friendly_name)
@@ -2439,6 +2439,7 @@ def test_update_table_w_custom_property(self):
             path="/%s" % path,
             data={"newAlphaProperty": "unreleased property"},
             timeout=DEFAULT_TIMEOUT,
+            query_params={},
         )
         self.assertEqual(
             updated_table._properties["newAlphaProperty"], "unreleased property"
@@ -2475,6 +2476,7 @@ def test_update_table_only_use_legacy_sql(self):
             path="/%s" % path,
             data={"view": {"useLegacySql": True}},
             timeout=DEFAULT_TIMEOUT,
+            query_params={},
         )
         self.assertEqual(updated_table.view_use_legacy_sql, table.view_use_legacy_sql)
 
@@ -2567,9 +2569,10 @@ def test_update_table_w_query(self):
                 "schema": schema_resource,
             },
             timeout=DEFAULT_TIMEOUT,
+            query_params={},
         )
 
-    def test_update_table_w_schema_None(self):
+    def test_update_table_w_schema_None_autodetect_schema(self):
         # Simulate deleting schema:  not sure if back-end will actually
         # allow this operation, but the spec says it is optional.
         path = "projects/%s/datasets/%s/tables/%s" % (
@@ -2611,7 +2614,9 @@ def test_update_table_w_schema_None(self):
         with mock.patch(
             "google.cloud.bigquery.opentelemetry_tracing._get_final_span_attributes"
         ) as final_attributes:
-            updated_table = client.update_table(table, ["schema"])
+            updated_table = client.update_table(
+                table, ["schema"], autodetect_schema=True
+            )
 
         final_attributes.assert_called_once_with(
             {"path": "/%s" % path, "fields": ["schema"]}, client, None
@@ -2623,6 +2628,7 @@ def test_update_table_w_schema_None(self):
         sent = {"schema": {"fields": None}}
         self.assertEqual(req[1]["data"], sent)
         self.assertEqual(req[1]["path"], "/%s" % path)
+        self.assertEqual(req[1]["query_params"], {"autodetect_schema": True})
         self.assertEqual(len(updated_table.schema), 0)
 
     def test_update_table_delete_property(self):