Remove unused code: random sampling

Author: Sergey Vasilyev

data_diff/abcs/mixins.py

@@ -141,24 +141,6 @@ def list_tables(self, table_schema: str, like: Compilable = None) -> Compilable:
         """
 
 
-@attrs.define(frozen=False)
-class AbstractMixin_RandomSample(AbstractMixin):
-    @abstractmethod
-    def random_sample_n(self, tbl: str, size: int) -> str:
-        """Take a random sample of the given size, i.e. return 'size' amount of rows"""
-
-    @abstractmethod
-    def random_sample_ratio_approx(self, tbl: str, ratio: float) -> str:
-        """Take a random sample of the approximate size determined by the ratio (0..1), where 0 means no rows, and 1 means all rows
-
-        i.e. the actual mount of rows returned may vary by standard deviation.
-        """
-
-    # def random_sample_ratio(self, table: ITable, ratio: float):
-    #     """Take a random sample of the size determined by the ratio (0..1), where 0 means no rows, and 1 means all rows
-    #     """
-
-
 @attrs.define(frozen=False)
 class AbstractMixin_TimeTravel(AbstractMixin):
     @abstractmethod

data_diff/databases/base.py

@@ -77,7 +77,6 @@
 from data_diff.abcs.mixins import AbstractMixin_TimeTravel, Compilable
 from data_diff.abcs.mixins import (
     AbstractMixin_Schema,
-    AbstractMixin_RandomSample,
     AbstractMixin_NormalizeValue,
     AbstractMixin_OptimizerHints,
 )
@@ -218,16 +217,6 @@ def list_tables(self, table_schema: str, like: Compilable = None) -> Compilable:
         )
 
 
-@attrs.define(frozen=False)
-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
-        return tbl.order_by(Random()).limit(size)
-
-    def random_sample_ratio_approx(self, tbl: ITable, ratio: float) -> ITable:
-        return tbl.where(Random() < ratio)
-
-
 @attrs.define(frozen=False)
 class Mixin_OptimizerHints(AbstractMixin_OptimizerHints):
     def optimizer_hints(self, hints: str) -> str:

data_diff/databases/duckdb.py

@@ -20,7 +20,6 @@
 from data_diff.abcs.mixins import (
     AbstractMixin_MD5,
     AbstractMixin_NormalizeValue,
-    AbstractMixin_RandomSample,
 )
 from data_diff.databases.base import (
     Database,
@@ -44,7 +43,7 @@ def import_duckdb():
 
 
 @attrs.define(frozen=False)
-class Dialect(BaseDialect, Mixin_Schema, AbstractMixin_MD5, AbstractMixin_NormalizeValue, AbstractMixin_RandomSample):
+class Dialect(BaseDialect, Mixin_Schema, AbstractMixin_MD5, AbstractMixin_NormalizeValue):
     name = "DuckDB"
     ROUNDS_ON_PREC_LOSS = False
     SUPPORTS_PRIMARY_KEY = True
@@ -120,12 +119,6 @@ def normalize_number(self, value: str, coltype: FractionalType) -> str:
     def normalize_boolean(self, value: str, _coltype: Boolean) -> str:
         return self.to_string(f"{value}::INTEGER")
 
-    def random_sample_n(self, tbl: ITable, size: int) -> ITable:
-        return code("SELECT * FROM ({tbl}) USING SAMPLE {size};", tbl=tbl, size=size)
-
-    def random_sample_ratio_approx(self, tbl: ITable, ratio: float) -> ITable:
-        return code("SELECT * FROM ({tbl}) USING SAMPLE {percent}%;", tbl=tbl, percent=int(100 * ratio))
-
 
 @attrs.define(frozen=False, init=False, kw_only=True)
 class DuckDB(Database):