Convert the remaining classes to attrs

Since we now use `attrs` for some classes, let's use them for them all — at least those belonging to the same hierarchies.

This will ensure that all classes are slotted and will strictly check that we define attributes properly, especially in cases of multiple inheritance.

Except for Pydantic models and Python exceptions.

Author: Sergey Vasilyev

data_diff/abcs/database_types.py

@@ -26,26 +26,33 @@ class PrecisionType(ColType):
     rounds: Union[bool, Unknown] = Unknown
 
 
+
+@attrs.define
 class Boolean(ColType):
     precision = 0
 
 
+@attrs.define
 class TemporalType(PrecisionType):
     pass
 
 
+@attrs.define
 class Timestamp(TemporalType):
     pass
 
 
+@attrs.define
 class TimestampTZ(TemporalType):
     pass
 
 
+@attrs.define
 class Datetime(TemporalType):
     pass
 
 
+@attrs.define
 class Date(TemporalType):
     pass
 
@@ -56,14 +63,17 @@ class NumericType(ColType):
     precision: int
 
 
+@attrs.define
 class FractionalType(NumericType):
     pass
 
 
+@attrs.define
 class Float(FractionalType):
     python_type = float
 
 
+@attrs.define
 class IKey(ABC):
     "Interface for ColType, for using a column as a key in table."
 
@@ -76,6 +86,7 @@ def make_value(self, value):
         return self.python_type(value)
 
 
+@attrs.define
 class Decimal(FractionalType, IKey):  # Snowflake may use Decimal as a key
     @property
     def python_type(self) -> type:
@@ -89,22 +100,27 @@ class StringType(ColType):
     python_type = str
 
 
+@attrs.define
 class ColType_UUID(ColType, IKey):
     python_type = ArithUUID
 
 
+@attrs.define
 class ColType_Alphanum(ColType, IKey):
     python_type = ArithAlphanumeric
 
 
+@attrs.define
 class Native_UUID(ColType_UUID):
     pass
 
 
+@attrs.define
 class String_UUID(ColType_UUID, StringType):
     pass
 
 
+@attrs.define
 class String_Alphanum(ColType_Alphanum, StringType):
     @staticmethod
     def test_value(value: str) -> bool:
@@ -118,6 +134,7 @@ def make_value(self, value):
         return self.python_type(value)
 
 
+@attrs.define
 class String_VaryingAlphanum(String_Alphanum):
     pass
 

data_diff/abcs/mixins.py

@@ -1,4 +1,7 @@
 from abc import ABC, abstractmethod
+
+import attrs
+
 from data_diff.abcs.database_types import (
     Array,
     TemporalType,
@@ -13,10 +16,12 @@
 from data_diff.abcs.compiler import Compilable
 
 
+@attrs.define
 class AbstractMixin(ABC):
     "A mixin for a database dialect"
 
 
+@attrs.define
 class AbstractMixin_NormalizeValue(AbstractMixin):
     @abstractmethod
     def to_comparable(self, value: str, coltype: ColType) -> str:
@@ -108,6 +113,7 @@ def normalize_value_by_type(self, value: str, coltype: ColType) -> str:
         return self.to_string(value)
 
 
+@attrs.define
 class AbstractMixin_MD5(AbstractMixin):
     """Methods for calculating an MD6 hash as an integer."""
 
@@ -116,6 +122,7 @@ def md5_as_int(self, s: str) -> str:
         "Provide SQL for computing md5 and returning an int"
 
 
+@attrs.define
 class AbstractMixin_Schema(AbstractMixin):
     """Methods for querying the database schema
 
@@ -134,6 +141,7 @@ def list_tables(self, table_schema: str, like: Compilable = None) -> Compilable:
         """
 
 
+@attrs.define
 class AbstractMixin_RandomSample(AbstractMixin):
     @abstractmethod
     def random_sample_n(self, tbl: str, size: int) -> str:
@@ -151,6 +159,7 @@ def random_sample_ratio_approx(self, tbl: str, ratio: float) -> str:
     #     """
 
 
+@attrs.define
 class AbstractMixin_TimeTravel(AbstractMixin):
     @abstractmethod
     def time_travel(
@@ -173,6 +182,7 @@ def time_travel(
         """
 
 
+@attrs.define
 class AbstractMixin_OptimizerHints(AbstractMixin):
     @abstractmethod
     def optimizer_hints(self, optimizer_hints: str) -> str:

data_diff/databases/_connect.py

@@ -93,8 +93,11 @@ def match_path(self, dsn):
 }
 
 
+@attrs.define(init=False)
 class Connect:
     """Provides methods for connecting to a supported database using a URL or connection dict."""
+    database_by_scheme: Dict[str, Database]
+    match_uri_path: Dict[str, MatchUriPath]
     conn_cache: MutableMapping[Hashable, Database]
 
     def __init__(self, database_by_scheme: Dict[str, Database] = DATABASE_BY_SCHEME):
@@ -284,6 +287,7 @@ def __make_cache_key(self, db_conf: Union[str, dict]) -> Hashable:
         return db_conf
 
 
+@attrs.define(init=False)
 class Connect_SetUTC(Connect):
     """Provides methods for connecting to a supported database using a URL or connection dict.
 

data_diff/databases/base.py

@@ -5,7 +5,7 @@
 import math
 import sys
 import logging
-from typing import Any, Callable, Dict, Generator, Tuple, Optional, Sequence, Type, List, Union, TypeVar
+from typing import Any, Callable, ClassVar, Dict, Generator, Tuple, Optional, Sequence, Type, List, Union, TypeVar
 from functools import partial, wraps
 from concurrent.futures import ThreadPoolExecutor
 import threading
@@ -156,15 +156,14 @@ def _one(seq):
     return x
 
 
+@attrs.define
 class ThreadLocalInterpreter:
     """An interpeter used to execute a sequence of queries within the same thread and cursor.
 
     Useful for cursor-sensitive operations, such as creating a temporary table.
     """
-
-    def __init__(self, compiler: Compiler, gen: Generator):
-        self.gen = gen
-        self.compiler = compiler
+    compiler: Compiler
+    gen: Generator
 
     def apply_queries(self, callback: Callable[[str], Any]):
         q: Expr = next(self.gen)
@@ -189,6 +188,7 @@ def apply_query(callback: Callable[[str], Any], sql_code: Union[str, ThreadLocal
         return callback(sql_code)
 
 
+@attrs.define
 class Mixin_Schema(AbstractMixin_Schema):
     def table_information(self) -> Compilable:
         return table("information_schema", "tables")
@@ -205,6 +205,7 @@ def list_tables(self, table_schema: str, like: Compilable = None) -> Compilable:
         )
 
 
+@attrs.define
 class Mixin_RandomSample(AbstractMixin_RandomSample):
     def random_sample_n(self, tbl: ITable, size: int) -> ITable:
         # TODO use a more efficient algorithm, when the table count is known
@@ -214,15 +215,17 @@ def random_sample_ratio_approx(self, tbl: ITable, ratio: float) -> ITable:
         return tbl.where(Random() < ratio)
 
 
+@attrs.define
 class Mixin_OptimizerHints(AbstractMixin_OptimizerHints):
     def optimizer_hints(self, hints: str) -> str:
         return f"/*+ {hints} */ "
 
 
+@attrs.define
 class BaseDialect(abc.ABC):
     SUPPORTS_PRIMARY_KEY = False
     SUPPORTS_INDEXES = False
-    TYPE_CLASSES: Dict[str, type] = {}
+    TYPE_CLASSES: ClassVar[Dict[str, Type[ColType]]] = {}
     MIXINS = frozenset()
 
     PLACEHOLDER_TABLE = None  # Used for Oracle
@@ -522,7 +525,7 @@ def render_select(self, parent_c: Compiler, elem: Select) -> str:
 
     def render_join(self, parent_c: Compiler, elem: Join) -> str:
         tables = [
-            t if isinstance(t, TableAlias) else TableAlias(source_table=t, name=parent_c.new_unique_name()) for t in elem.source_tables
+            t if isinstance(t, TableAlias) else TableAlias(source_table_=t, name=parent_c.new_unique_name()) for t in elem.source_tables
         ]
         c = parent_c.add_table_context(*tables, in_join=True, in_select=False)
         op = " JOIN " if elem.op is None else f" {elem.op} JOIN "
@@ -823,6 +826,7 @@ def __getitem__(self, i):
         return self.rows[i]
 
 
+@attrs.define
 class Database(abc.ABC):
     """Base abstract class for databases.
 
@@ -1099,6 +1103,7 @@ def is_autocommit(self) -> bool:
         "Return whether the database autocommits changes. When false, COMMIT statements are skipped."
 
 
+@attrs.define(init=False, slots=False)
 class ThreadedDatabase(Database):
     """Access the database through singleton threads.
 

data_diff/databases/bigquery.py

@@ -1,5 +1,8 @@
 import re
 from typing import Any, List, Union
+
+import attrs
+
 from data_diff.abcs.database_types import (
     ColType,
     Array,
@@ -50,11 +53,13 @@ def import_bigquery_service_account():
     return service_account
 
 
+@attrs.define
 class Mixin_MD5(AbstractMixin_MD5):
     def md5_as_int(self, s: str) -> str:
         return f"cast(cast( ('0x' || substr(TO_HEX(md5({s})), 18)) as int64) as numeric)"
 
 
+@attrs.define
 class Mixin_NormalizeValue(AbstractMixin_NormalizeValue):
     def normalize_timestamp(self, value: str, coltype: TemporalType) -> str:
         if coltype.rounds:
@@ -99,6 +104,7 @@ def normalize_struct(self, value: str, _coltype: Struct) -> str:
         return f"to_json_string({value})"
 
 
+@attrs.define
 class Mixin_Schema(AbstractMixin_Schema):
     def list_tables(self, table_schema: str, like: Compilable = None) -> Compilable:
         return (
@@ -112,6 +118,7 @@ def list_tables(self, table_schema: str, like: Compilable = None) -> Compilable:
         )
 
 
+@attrs.define
 class Mixin_TimeTravel(AbstractMixin_TimeTravel):
     def time_travel(
         self,
@@ -139,6 +146,7 @@ def time_travel(
         )
 
 
+@attrs.define
 class Dialect(BaseDialect, Mixin_Schema, Mixin_MD5, Mixin_NormalizeValue, AbstractMixin_MD5, AbstractMixin_NormalizeValue):
     name = "BigQuery"
     ROUNDS_ON_PREC_LOSS = False  # Technically BigQuery doesn't allow implicit rounding or truncation

data_diff/databases/clickhouse.py

@@ -1,5 +1,7 @@
 from typing import Optional, Type
 
+import attrs
+
 from data_diff.databases.base import (
     MD5_HEXDIGITS,
     CHECKSUM_HEXDIGITS,
@@ -35,12 +37,14 @@ def import_clickhouse():
     return clickhouse_driver
 
 
+@attrs.define
 class Mixin_MD5(AbstractMixin_MD5):
     def md5_as_int(self, s: str) -> str:
         substr_idx = 1 + MD5_HEXDIGITS - CHECKSUM_HEXDIGITS
         return f"reinterpretAsUInt128(reverse(unhex(lowerUTF8(substr(hex(MD5({s})), {substr_idx})))))"
 
 
+@attrs.define
 class Mixin_NormalizeValue(AbstractMixin_NormalizeValue):
     def normalize_number(self, value: str, coltype: FractionalType) -> str:
         # If a decimal value has trailing zeros in a fractional part, when casting to string they are dropped.
@@ -99,6 +103,7 @@ def normalize_timestamp(self, value: str, coltype: TemporalType) -> str:
         return f"rpad({value}, {TIMESTAMP_PRECISION_POS + 6}, '0')"
 
 
+@attrs.define
 class Dialect(BaseDialect, Mixin_MD5, Mixin_NormalizeValue, AbstractMixin_MD5, AbstractMixin_NormalizeValue):
     name = "Clickhouse"
     ROUNDS_ON_PREC_LOSS = False
@@ -163,6 +168,7 @@ def current_timestamp(self) -> str:
         return "now()"
 
 
+@attrs.define
 class Clickhouse(ThreadedDatabase):
     dialect = Dialect()
     CONNECT_URI_HELP = "clickhouse://<user>:<password>@<host>/<database>"

data_diff/databases/databricks.py

@@ -2,6 +2,8 @@
 from typing import Dict, Sequence
 import logging
 
+import attrs
+
 from data_diff.abcs.database_types import (
     Integer,
     Float,
@@ -34,11 +36,13 @@ def import_databricks():
     return databricks
 
 
+@attrs.define
 class Mixin_MD5(AbstractMixin_MD5):
     def md5_as_int(self, s: str) -> str:
         return f"cast(conv(substr(md5({s}), {1+MD5_HEXDIGITS-CHECKSUM_HEXDIGITS}), 16, 10) as decimal(38, 0))"
 
 
+@attrs.define
 class Mixin_NormalizeValue(AbstractMixin_NormalizeValue):
     def normalize_timestamp(self, value: str, coltype: TemporalType) -> str:
         """Databricks timestamp contains no more than 6 digits in precision"""
@@ -60,6 +64,7 @@ def normalize_boolean(self, value: str, _coltype: Boolean) -> str:
         return self.to_string(f"cast ({value} as int)")
 
 
+@attrs.define
 class Dialect(BaseDialect, Mixin_MD5, Mixin_NormalizeValue, AbstractMixin_MD5, AbstractMixin_NormalizeValue):
     name = "Databricks"
     ROUNDS_ON_PREC_LOSS = True
@@ -99,6 +104,7 @@ def parse_table_name(self, name: str) -> DbPath:
         return tuple(i for i in path if i is not None)
 
 
+@attrs.define(init=False)
 class Databricks(ThreadedDatabase):
     dialect = Dialect()
     CONNECT_URI_HELP = "databricks://:<access_token>@<server_hostname>/<http_path>"