Annotate constructors which always return None

datafold · nolar · Jan 8, 2024 · Dec 31, 2023 · Dec 31, 2023 · Dec 31, 2023
commit ce09a5c43bb3ee2264982da5d9ff8b4bfce5cdfa
diff --git a/data_diff/__main__.py b/data_diff/__main__.py
@@ -103,7 +103,7 @@ def diff_schemas(table1, table2, schema1, schema2, columns):
 
 
 class MyHelpFormatter(click.HelpFormatter):
-    def __init__(self, **kwargs):
+    def __init__(self, **kwargs) -> None:
         super().__init__(self, **kwargs)
         self.indent_increment = 6
 

diff --git a/data_diff/abcs/database_types.py b/data_diff/abcs/database_types.py
@@ -290,7 +290,7 @@ class Integer(NumericType, IKey):
     precision: int = 0
     python_type: type = int
 
-    def __attrs_post_init__(self):
+    def __attrs_post_init__(self) -> None:
         assert self.precision == 0
 
 

diff --git a/data_diff/cloud/datafold_api.py b/data_diff/cloud/datafold_api.py
@@ -185,7 +185,7 @@ class DatafoldAPI:
     host: str = "https://app.datafold.com"
     timeout: int = 30
 
-    def __attrs_post_init__(self):
+    def __attrs_post_init__(self) -> None:
         self.host = self.host.rstrip("/")
         self.headers = {
             "Authorization": f"Key {self.api_key}",

diff --git a/data_diff/databases/_connect.py b/data_diff/databases/_connect.py
@@ -100,7 +100,7 @@ class Connect:
     database_by_scheme: Dict[str, Database]
     conn_cache: MutableMapping[Hashable, Database]
 
-    def __init__(self, database_by_scheme: Dict[str, Database] = DATABASE_BY_SCHEME):
+    def __init__(self, database_by_scheme: Dict[str, Database] = DATABASE_BY_SCHEME) -> None:
         super().__init__()
         self.database_by_scheme = database_by_scheme
         self.conn_cache = weakref.WeakValueDictionary()

diff --git a/data_diff/databases/base.py b/data_diff/databases/base.py
@@ -1209,7 +1209,7 @@ class ThreadedDatabase(Database):
     _queue: Optional[ThreadPoolExecutor] = None
     thread_local: threading.local = attrs.field(factory=threading.local)
 
-    def __attrs_post_init__(self):
+    def __attrs_post_init__(self) -> None:
         self._queue = ThreadPoolExecutor(self.thread_count, initializer=self.set_conn)
         logger.info(f"[{self.name}] Starting a threadpool, size={self.thread_count}.")
 

diff --git a/data_diff/databases/bigquery.py b/data_diff/databases/bigquery.py
@@ -212,7 +212,7 @@ class BigQuery(Database):
     dataset: str
     _client: Any
 
-    def __init__(self, project, *, dataset, bigquery_credentials=None, **kw):
+    def __init__(self, project, *, dataset, bigquery_credentials=None, **kw) -> None:
         super().__init__()
         credentials = bigquery_credentials
         bigquery = import_bigquery()

diff --git a/data_diff/databases/clickhouse.py b/data_diff/databases/clickhouse.py
@@ -175,7 +175,7 @@ class Clickhouse(ThreadedDatabase):
 
     _args: Dict[str, Any]
 
-    def __init__(self, *, thread_count: int, **kw):
+    def __init__(self, *, thread_count: int, **kw) -> None:
         super().__init__(thread_count=thread_count)
 
         self._args = kw

diff --git a/data_diff/databases/databricks.py b/data_diff/databases/databricks.py
@@ -118,7 +118,7 @@ class Databricks(ThreadedDatabase):
     catalog: str
     _args: Dict[str, Any]
 
-    def __init__(self, *, thread_count, **kw):
+    def __init__(self, *, thread_count, **kw) -> None:
         super().__init__(thread_count=thread_count)
         logging.getLogger("databricks.sql").setLevel(logging.WARNING)
 

diff --git a/data_diff/databases/duckdb.py b/data_diff/databases/duckdb.py
@@ -126,7 +126,7 @@ class DuckDB(Database):
     _args: Dict[str, Any] = attrs.field(init=False)
     _conn: Any = attrs.field(init=False)
 
-    def __init__(self, **kw):
+    def __init__(self, **kw) -> None:
         super().__init__()
         self._args = kw
         self._conn = self.create_connection()

diff --git a/data_diff/databases/mssql.py b/data_diff/databases/mssql.py
@@ -168,7 +168,7 @@ class MsSQL(ThreadedDatabase):
     _args: Dict[str, Any]
     _mssql: Any
 
-    def __init__(self, host, port, user, password, *, database, thread_count, **kw):
+    def __init__(self, host, port, user, password, *, database, thread_count, **kw) -> None:
         super().__init__(thread_count=thread_count)
 
         args = dict(server=host, port=port, database=database, user=user, password=password, **kw)

diff --git a/data_diff/databases/mysql.py b/data_diff/databases/mysql.py
@@ -129,7 +129,7 @@ class MySQL(ThreadedDatabase):
 
     _args: Dict[str, Any]
 
-    def __init__(self, *, thread_count, **kw):
+    def __init__(self, *, thread_count, **kw) -> None:
         super().__init__(thread_count=thread_count)
         self._args = kw
 

diff --git a/data_diff/databases/oracle.py b/data_diff/databases/oracle.py
@@ -164,7 +164,7 @@ class Oracle(ThreadedDatabase):
     kwargs: Dict[str, Any]
     _oracle: Any
 
-    def __init__(self, *, host, database, thread_count, **kw):
+    def __init__(self, *, host, database, thread_count, **kw) -> None:
         super().__init__(thread_count=thread_count)
         self.kwargs = dict(dsn=f"{host}/{database}" if database else host, **kw)
         self.default_schema = kw.get("user").upper()

diff --git a/data_diff/databases/postgresql.py b/data_diff/databases/postgresql.py
@@ -163,7 +163,7 @@ class PostgreSQL(ThreadedDatabase):
     _args: Dict[str, Any]
     _conn: Any
 
-    def __init__(self, *, thread_count, **kw):
+    def __init__(self, *, thread_count, **kw) -> None:
         super().__init__(thread_count=thread_count)
         self._args = kw
         self.default_schema = "public"

diff --git a/data_diff/databases/presto.py b/data_diff/databases/presto.py
@@ -152,7 +152,7 @@ class Presto(Database):
 
     _conn: Any
 
-    def __init__(self, **kw):
+    def __init__(self, **kw) -> None:
         super().__init__()
         self.default_schema = "public"
         prestodb = import_presto()

diff --git a/data_diff/databases/snowflake.py b/data_diff/databases/snowflake.py
@@ -104,7 +104,7 @@ class Snowflake(Database):
 
     _conn: Any
 
-    def __init__(self, *, schema: str, key: Optional[str] = None, key_content: Optional[str] = None, **kw):
+    def __init__(self, *, schema: str, key: Optional[str] = None, key_content: Optional[str] = None, **kw) -> None:
         super().__init__()
         snowflake, serialization, default_backend = import_snowflake()
         logging.getLogger("snowflake.connector").setLevel(logging.WARNING)

diff --git a/data_diff/databases/trino.py b/data_diff/databases/trino.py
@@ -40,7 +40,7 @@ class Trino(presto.Presto):
 
     _conn: Any
 
-    def __init__(self, **kw):
+    def __init__(self, **kw) -> None:
         super().__init__()
         trino = import_trino()
 

diff --git a/data_diff/databases/vertica.py b/data_diff/databases/vertica.py
@@ -137,7 +137,7 @@ class Vertica(ThreadedDatabase):
 
     _args: Dict[str, Any]
 
-    def __init__(self, *, thread_count, **kw):
+    def __init__(self, *, thread_count, **kw) -> None:
         super().__init__(thread_count=thread_count)
         self._args = kw
         self._args["AUTOCOMMIT"] = False

diff --git a/data_diff/hashdiff_tables.py b/data_diff/hashdiff_tables.py
@@ -96,7 +96,7 @@ class HashDiffer(TableDiffer):
 
     stats: dict = attrs.field(factory=dict)
 
-    def __attrs_post_init__(self):
+    def __attrs_post_init__(self) -> None:
         # Validate options
         if self.bisection_factor >= self.bisection_threshold:
             raise ValueError("Incorrect param values (bisection factor must be lower than threshold)")

diff --git a/data_diff/lexicographic_space.py b/data_diff/lexicographic_space.py
@@ -66,7 +66,7 @@ class LexicographicSpace:
     All elements must be of the same length as the number of dimensions. (no rpadding)
     """
 
-    def __init__(self, dims: Vector):
+    def __init__(self, dims: Vector) -> None:
         super().__init__()
         self.dims = dims
 
@@ -124,7 +124,7 @@ class BoundedLexicographicSpace:
     i.e. a space resticted by a "bounding-box" between two arbitrary points.
     """
 
-    def __init__(self, min_bound: Vector, max_bound: Vector):
+    def __init__(self, min_bound: Vector, max_bound: Vector) -> None:
         super().__init__()
 
         dims = tuple(mx - mn for mn, mx in safezip(min_bound, max_bound))

diff --git a/data_diff/queries/ast_classes.py b/data_diff/queries/ast_classes.py
@@ -523,7 +523,7 @@ class GroupBy(ExprNode, ITable, Root):
     values: Optional[Sequence[Expr]] = None
     having_exprs: Optional[Sequence[Expr]] = None
 
-    def __attrs_post_init__(self):
+    def __attrs_post_init__(self) -> None:
         assert self.keys or self.values
 
     def having(self, *exprs) -> Self:

diff --git a/data_diff/table_segment.py b/data_diff/table_segment.py
@@ -126,7 +126,7 @@ class TableSegment:
     case_sensitive: Optional[bool] = True
     _schema: Optional[Schema] = None
 
-    def __attrs_post_init__(self):
+    def __attrs_post_init__(self) -> None:
         if not self.update_column and (self.min_update or self.max_update):
             raise ValueError("Error: the min_update/max_update feature requires 'update_column' to be set.")
 

diff --git a/data_diff/thread_utils.py b/data_diff/thread_utils.py
@@ -34,7 +34,7 @@ class PriorityThreadPoolExecutor(ThreadPoolExecutor):
     XXX WARNING: Might break in future versions of Python
     """
 
-    def __init__(self, *args):
+    def __init__(self, *args) -> None:
         super().__init__(*args)
         self._work_queue = AutoPriorityQueue()
 
@@ -58,7 +58,7 @@ class ThreadedYielder(Iterable):
     _exception: Optional[None]
     yield_list: bool
 
-    def __init__(self, max_workers: Optional[int] = None, yield_list: bool = False):
+    def __init__(self, max_workers: Optional[int] = None, yield_list: bool = False) -> None:
         super().__init__()
         self._pool = PriorityThreadPoolExecutor(max_workers)
         self._futures = deque()

diff --git a/data_diff/utils.py b/data_diff/utils.py
@@ -80,7 +80,7 @@ def new(self, initial=()) -> Self:
 
 
 class CaseInsensitiveDict(CaseAwareMapping):
-    def __init__(self, initial):
+    def __init__(self, initial) -> None:
         super().__init__()
         self._dict = {k.lower(): (k, v) for k, v in dict(initial).items()}
 
@@ -241,7 +241,7 @@ class ArithAlphanumeric(ArithString):
     _str: str
     _max_len: Optional[int] = None
 
-    def __attrs_post_init__(self):
+    def __attrs_post_init__(self) -> None:
         if self._str is None:
             raise ValueError("Alphanum string cannot be None")
         if self._max_len and len(self._str) > self._max_len:
@@ -540,7 +540,7 @@ class LogStatusHandler(logging.Handler):
     This log handler can be used to update a rich.status every time a log is emitted.
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.status = Status("")
         self.prefix = ""

diff --git a/tests/test_database_types.py b/tests/test_database_types.py
@@ -366,7 +366,7 @@ class PaginatedTable:
     # much memory.
     RECORDS_PER_BATCH = 1000000
 
-    def __init__(self, table_path, conn):
+    def __init__(self, table_path, conn) -> None:
         super().__init__()
         self.table_path = table_path
         self.conn = conn
@@ -398,7 +398,7 @@ class DateTimeFaker:
         datetime.fromisoformat("2022-06-01 15:10:05.009900"),
     ]
 
-    def __init__(self, max):
+    def __init__(self, max) -> None:
         super().__init__()
         self.max = max
 
@@ -414,7 +414,7 @@ def __len__(self):
 class IntFaker:
     MANUAL_FAKES = [127, -3, -9, 37, 15, 0]
 
-    def __init__(self, max):
+    def __init__(self, max) -> None:
         super().__init__()
         self.max = max
 
@@ -430,7 +430,7 @@ def __len__(self):
 class BooleanFaker:
     MANUAL_FAKES = [False, True, True, False]
 
-    def __init__(self, max):
+    def __init__(self, max) -> None:
         super().__init__()
         self.max = max
 
@@ -461,7 +461,7 @@ class FloatFaker:
         3.141592653589793,
     ]
 
-    def __init__(self, max):
+    def __init__(self, max) -> None:
         super().__init__()
         self.max = max
 
@@ -475,7 +475,7 @@ def __len__(self):
 
 
 class UUID_Faker:
-    def __init__(self, max):
+    def __init__(self, max) -> None:
         super().__init__()
         self.max = max
 
@@ -491,7 +491,7 @@ class JsonFaker:
         '{"keyText": "text", "keyInt": 3, "keyFloat": 5.4445, "keyBoolean": true}',
     ]
 
-    def __init__(self, max):
+    def __init__(self, max) -> None:
         super().__init__()
         self.max = max