Fix: set utf-8 encoding in most open calls (#2503)

Author: georgesittas

sqlmesh/core/loader.py

@@ -247,7 +247,7 @@ def _load_scripts(self) -> t.Tuple[MacroRegistry, JinjaMacroRegistry]:
                     if macros_max_mtime
                     else macro_file_mtime
                 )
-                with open(path, mode="r", encoding="utf-8") as file:
+                with open(path, "r", encoding="utf-8") as file:
                     jinja_macros.add_macros(extractor.extract(file.read()))
 
         self._macros_max_mtime = macros_max_mtime

sqlmesh/core/model/seed.py

@@ -102,5 +102,5 @@ def reader(self, dialect: str = "", settings: t.Optional[CsvSettings] = None) ->
 
 
 def create_seed(path: str | Path) -> Seed:
-    with open(Path(path), "r") as fd:
+    with open(Path(path), "r", encoding="utf-8") as fd:
         return Seed(content=fd.read())

sqlmesh/integrations/github/cicd/controller.py

@@ -215,7 +215,7 @@ def from_obj(cls, obj: t.Dict[str, t.Any]) -> GithubEvent:
 
     @classmethod
     def from_path(cls, path: t.Union[str, pathlib.Path]) -> GithubEvent:
-        with open(pathlib.Path(path)) as f:
+        with open(pathlib.Path(path), "r", encoding="utf-8") as f:
             return cls.from_obj(json.load(f))
 
     @classmethod
@@ -435,7 +435,7 @@ def _append_output(cls, key: str, value: str) -> None:
         Appends the given key/value to output so they can be read by following steps
         """
         logger.debug(f"Setting output. Key: {key}, Value: {value}")
-        with open(os.environ["GITHUB_OUTPUT"], "a") as fh:
+        with open(os.environ["GITHUB_OUTPUT"], "a", encoding="utf-8") as fh:
             print(f"{key}={value}", file=fh)
 
     def get_plan_summary(self, plan: Plan) -> str:

sqlmesh/magics.py

@@ -528,7 +528,7 @@ def dag(self, context: Context, line: str) -> None:
         args = parse_argstring(self.dag, line)
         dag = context.get_dag(args.select_model)
         if args.file:
-            with open(args.file, "w") as file:
+            with open(args.file, "w", encoding="utf-8") as file:
                 file.write(str(dag))
         # TODO: Have this go through console instead of calling display directly
         self.display(dag)

sqlmesh/schedulers/airflow/operators/databricks.py

@@ -91,7 +91,7 @@ def execute(self, context: Context) -> None:
         command_payload = self._target.serialized_command_payload(context)
         with tempfile.TemporaryDirectory() as tmp:
             local_payload_path = os.path.join(tmp, commands.COMMAND_PAYLOAD_FILE_NAME)
-            with open(local_payload_path, "w") as payload_fd:
+            with open(local_payload_path, "w", encoding="utf-8") as payload_fd:
                 payload_fd.write(command_payload)
             remote_payload_path = os.path.join(
                 self._dbfs_location, commands.COMMAND_PAYLOAD_FILE_NAME

sqlmesh/schedulers/airflow/operators/spark_submit.py

@@ -76,7 +76,7 @@ def execute(self, context: Context) -> None:
         command_payload = self._target.serialized_command_payload(context)
         with tempfile.TemporaryDirectory() as tmp:
             payload_file_path = os.path.join(tmp, commands.COMMAND_PAYLOAD_FILE_NAME)
-            with open(payload_file_path, "w") as payload_fd:
+            with open(payload_file_path, "w", encoding="utf-8") as payload_fd:
                 payload_fd.write(command_payload)
 
             if self._hook is None:

tests/cli/test_cli.py

@@ -36,7 +36,7 @@ def create_example_project(temp_dir) -> None:
         - Overwriting the config.yaml file so the duckdb database file will be created in the temp_dir directory
     """
     init_example_project(temp_dir, "duckdb")
-    with open(temp_dir / "config.yaml", "w") as f:
+    with open(temp_dir / "config.yaml", "w", encoding="utf-8") as f:
         f.write(
             f"""gateways:
     local:
@@ -53,7 +53,7 @@ def create_example_project(temp_dir) -> None:
 
 
 def update_incremental_model(temp_dir) -> None:
-    with open(temp_dir / "models" / "incremental_model.sql", "w") as f:
+    with open(temp_dir / "models" / "incremental_model.sql", "w", encoding="utf-8") as f:
         f.write(
             """
 MODEL (
@@ -80,7 +80,7 @@ def update_incremental_model(temp_dir) -> None:
 
 
 def update_full_model(temp_dir) -> None:
-    with open(temp_dir / "models" / "full_model.sql", "w") as f:
+    with open(temp_dir / "models" / "full_model.sql", "w", encoding="utf-8") as f:
         f.write(
             """
 MODEL (

tests/core/test_config.py

@@ -28,7 +28,7 @@
 @pytest.fixture(scope="session")
 def yaml_config_path(tmp_path_factory) -> Path:
     config_path = tmp_path_factory.mktemp("yaml_config") / "config.yaml"
-    with open(config_path, "w") as fd:
+    with open(config_path, "w", encoding="utf-8") as fd:
         fd.write(
             """
 gateways:
@@ -47,7 +47,7 @@ def yaml_config_path(tmp_path_factory) -> Path:
 @pytest.fixture(scope="session")
 def python_config_path(tmp_path_factory) -> Path:
     config_path = tmp_path_factory.mktemp("python_config") / "config.py"
-    with open(config_path, "w") as fd:
+    with open(config_path, "w", encoding="utf-8") as fd:
         fd.write(
             """from sqlmesh.core.config import Config, DuckDBConnectionConfig, GatewayConfig, ModelDefaultsConfig
 config = Config(gateways=GatewayConfig(connection=DuckDBConnectionConfig()), model_defaults=ModelDefaultsConfig(dialect=''))
@@ -163,11 +163,11 @@ def test_load_config_from_paths(yaml_config_path: Path, python_config_path: Path
 
 def test_load_config_multiple_config_files_in_folder(tmp_path):
     config_a_path = tmp_path / "config.yaml"
-    with open(config_a_path, "w") as fd:
+    with open(config_a_path, "w", encoding="utf-8") as fd:
         fd.write("project: project_a")
 
     config_b_path = tmp_path / "config.yml"
-    with open(config_b_path, "w") as fd:
+    with open(config_b_path, "w", encoding="utf-8") as fd:
         fd.write("project: project_b")
 
     with pytest.raises(ConfigError, match=r"^Multiple configuration files found in folder.*"):
@@ -285,7 +285,7 @@ def test_load_config_from_env_invalid_variable_name():
 
 def test_load_config_from_python_module_missing_config(tmp_path):
     config_path = tmp_path / "missing_config.py"
-    with open(config_path, "w") as fd:
+    with open(config_path, "w", encoding="utf-8") as fd:
         fd.write("from sqlmesh.core.config import Config")
 
     with pytest.raises(ConfigError, match="Config 'config' was not found."):
@@ -294,7 +294,7 @@ def test_load_config_from_python_module_missing_config(tmp_path):
 
 def test_load_config_from_python_module_invalid_config_object(tmp_path):
     config_path = tmp_path / "invalid_config.py"
-    with open(config_path, "w") as fd:
+    with open(config_path, "w", encoding="utf-8") as fd:
         fd.write("config = None")
 
     with pytest.raises(
@@ -306,7 +306,7 @@ def test_load_config_from_python_module_invalid_config_object(tmp_path):
 
 def test_cloud_composer_scheduler_config(tmp_path_factory):
     config_path = tmp_path_factory.mktemp("yaml_config") / "config.yaml"
-    with open(config_path, "w") as fd:
+    with open(config_path, "w", encoding="utf-8") as fd:
         fd.write(
             """
 gateways:
@@ -365,7 +365,7 @@ def test_cloud_composer_scheduler_config(tmp_path_factory):
 )
 def test_environment_catalog_mapping(tmp_path_factory, mapping, expected, dialect, raise_error):
     config_path = tmp_path_factory.mktemp("yaml_config") / "config.yaml"
-    with open(config_path, "w") as fd:
+    with open(config_path, "w", encoding="utf-8") as fd:
         fd.write(
             f"""
 gateways:
@@ -395,7 +395,7 @@ def test_environment_catalog_mapping(tmp_path_factory, mapping, expected, dialec
 
 def test_load_feature_flag(tmp_path_factory):
     config_path = tmp_path_factory.mktemp("yaml_config") / "config.yaml"
-    with open(config_path, "w") as fd:
+    with open(config_path, "w", encoding="utf-8") as fd:
         fd.write(
             """
 gateways:

tests/core/test_connection_config.py

@@ -37,7 +37,7 @@ def snowflake_key_no_passphrase_path() -> str:
 
 @pytest.fixture
 def snowflake_key_no_passphrase_pem(snowflake_key_no_passphrase_path) -> str:
-    with open(snowflake_key_no_passphrase_path, "r") as key:
+    with open(snowflake_key_no_passphrase_path, "r", encoding="utf-8") as key:
         return key.read()
 
 
@@ -58,7 +58,7 @@ def snowflake_key_passphrase_path() -> str:
 
 @pytest.fixture
 def snowflake_key_passphrase_pem(snowflake_key_passphrase_path) -> str:
-    with open(snowflake_key_passphrase_path, "r") as key:
+    with open(snowflake_key_passphrase_path, "r", encoding="utf-8") as key:
         return key.read()
 
 

tests/core/test_model.py

@@ -763,14 +763,14 @@ def test_seed_model_diff(tmp_path):
     model_a_csv_path = (tmp_path / "model_a.csv").absolute()
     model_b_csv_path = (tmp_path / "model_b.csv").absolute()
 
-    with open(model_a_csv_path, "w") as fd:
+    with open(model_a_csv_path, "w", encoding="utf-8") as fd:
         fd.write(
             """key,value
 1,value_a
 """
         )
 
-    with open(model_b_csv_path, "w") as fd:
+    with open(model_b_csv_path, "w", encoding="utf-8") as fd:
         fd.write(
             """key,value
 2,value_b
@@ -860,7 +860,7 @@ def test_seed_pre_statements_only():
 def test_seed_model_custom_types(tmp_path):
     model_csv_path = (tmp_path / "model.csv").absolute()
 
-    with open(model_csv_path, "w") as fd:
+    with open(model_csv_path, "w", encoding="utf-8") as fd:
         fd.write(
             """key,ds_date,ds_timestamp,b_a,b_b,i,i_str
 123,2022-01-01,2022-01-01,false,0,321,321