feat: Add is_unique rule to dy.Column

Author: gabriel

dataframely/_base_schema.py

@@ -39,7 +39,14 @@ def _build_rules(
     # Add primary key validation to the list of rules if applicable
     primary_key = _primary_key(columns)
     if len(primary_key) > 0:
-        rules["primary_key"] = Rule(~pl.struct(primary_key).is_duplicated())
+        rules["primary_key"] = Rule(pl.struct(primary_key).is_unique())
+
+    # Add unique column validation rules
+    unique_columns = _unique_columns(columns)
+    for col_name in unique_columns:
+        # wrap the column in a struct to make `is_unique` work with list/arrays
+        # https://github.com/pola-rs/polars/issues/27286
+        rules[f"{col_name}|unique"] = Rule(pl.struct(col_name).is_unique())
 
     # Add column-specific rules
     column_rules = {
@@ -71,6 +78,10 @@ def _primary_key(columns: dict[str, Column]) -> list[str]:
     return list(k for k, col in columns.items() if col.primary_key)
 
 
+def _unique_columns(columns: dict[str, Column]) -> list[str]:
+    return list(k for k, col in columns.items() if col.unique)
+
+
 # ------------------------------------------------------------------------------------ #
 #                                      SCHEMA META                                     #
 # ------------------------------------------------------------------------------------ #
@@ -300,6 +311,11 @@ def primary_key(cls) -> list[str]:
         """The primary key columns in this schema (possibly empty)."""
         return _primary_key(cls.columns())
 
+    @classmethod
+    def unique_columns(cls) -> list[str]:
+        """The columns with unique constraints in this schema (possibly empty)."""
+        return _unique_columns(cls.columns())
+
     @classmethod
     def _validation_rules(cls, *, with_cast: bool) -> dict[str, Rule]:
         return _build_rules(

dataframely/columns/_base.py

@@ -45,6 +45,7 @@ def __init__(
         *,
         nullable: bool = False,
         primary_key: bool = False,
+        unique: bool = False,
         check: Check | None = None,
         alias: str | None = None,
         metadata: dict[str, Any] | None = None,
@@ -55,6 +56,9 @@ def __init__(
                 Explicitly set `nullable=True` if you want your column to be nullable.
             primary_key: Whether this column is part of the primary key of the schema.
                 If `True`, `nullable` is automatically set to `False`.
+            unique: Whether this column must contain unique values. Unlike ``primary_key``,
+                this checks uniqueness for this column independently. Multiple columns
+                can each have ``unique=True`` without forming a composite constraint.
             check: A custom rule or multiple rules to run for this column. This can be:
                 - A single callable that returns a non-aggregated boolean expression.
                 The name of the rule is derived from the callable name, or defaults to
@@ -78,6 +82,7 @@ def __init__(
 
         self.nullable = nullable
         self.primary_key = primary_key
+        self.unique = unique
         self.check = check
         self.alias = alias
         self.metadata = metadata
@@ -198,6 +203,7 @@ def sqlalchemy_column(self, name: str, dialect: sa.Dialect) -> sa.Column:
             self.sqlalchemy_dtype(dialect),
             nullable=self.nullable,
             primary_key=self.primary_key,
+            unique=self.unique,
             autoincrement=False,
         )
 

dataframely/columns/array.py

@@ -7,7 +7,7 @@
 import sys
 import warnings
 from collections.abc import Sequence
-from typing import Any, Literal, cast
+from typing import Any, cast
 
 import polars as pl
 
@@ -34,9 +34,8 @@ def __init__(
         shape: int | tuple[int, ...],
         *,
         nullable: bool = False,
-        # polars doesn't yet support grouping by arrays,
-        # see https://github.com/pola-rs/polars/issues/22574
-        primary_key: Literal[False] = False,
+        primary_key: bool = False,
+        unique: bool = False,
         check: Check | None = None,
         alias: str | None = None,
         metadata: dict[str, Any] | None = None,
@@ -47,7 +46,7 @@ def __init__(
             shape: The shape of the array.
             nullable: Whether this column may contain null values.
             primary_key: Whether this column is part of the primary key of the schema.
-                Not yet supported for the Array type.
+            unique: Whether this column must contain unique values.
             check: A custom rule or multiple rules to run for this column. This can be:
                 - A single callable that returns a non-aggregated boolean expression.
                 The name of the rule is derived from the callable name, or defaults to
@@ -67,7 +66,8 @@ def __init__(
         """
         super().__init__(
             nullable=nullable,
-            primary_key=False,
+            primary_key=primary_key,
+            unique=unique,
             check=check,
             alias=alias,
             metadata=metadata,

dataframely/columns/categorical.py

@@ -23,6 +23,7 @@ def __init__(
         *,
         nullable: bool = False,
         primary_key: bool = False,
+        unique: bool = False,
         check: Check | None = None,
         alias: str | None = None,
         metadata: dict[str, Any] | None = None,
@@ -35,6 +36,7 @@ def __init__(
                 is not specified.
             primary_key: Whether this column is part of the primary key of the schema.
                 If `True`, `nullable` is automatically set to `False`.
+            unique: Whether this column must contain unique values.
             check: A custom rule or multiple rules to run for this column. This can be:
                 - A single callable that returns a non-aggregated boolean expression.
                 The name of the rule is derived from the callable name, or defaults to
@@ -55,6 +57,7 @@ def __init__(
         super().__init__(
             nullable=nullable,
             primary_key=primary_key,
+            unique=unique,
             check=check,
             alias=alias,
             metadata=metadata,

dataframely/columns/datetime.py

@@ -37,6 +37,7 @@ def __init__(
         *,
         nullable: bool = False,
         primary_key: bool = False,
+        unique: bool = False,
         min: dt.date | None = None,
         min_exclusive: dt.date | None = None,
         max: dt.date | None = None,
@@ -54,6 +55,7 @@ def __init__(
                 is not specified.
             primary_key: Whether this column is part of the primary key of the schema.
                 If `True`, `nullable` is automatically set to `False`.
+            unique: Whether this column must contain unique values.
             min: The minimum date for dates in this column (inclusive).
             min_exclusive: Like `min` but exclusive. May not be specified if `min`
                 is specified and vice versa.
@@ -101,6 +103,7 @@ def __init__(
         super().__init__(
             nullable=nullable,
             primary_key=primary_key,
+            unique=unique,
             min=min,
             min_exclusive=min_exclusive,
             max=max,
@@ -170,6 +173,7 @@ def __init__(
         *,
         nullable: bool = False,
         primary_key: bool = False,
+        unique: bool = False,
         min: dt.time | None = None,
         min_exclusive: dt.time | None = None,
         max: dt.time | None = None,
@@ -187,6 +191,7 @@ def __init__(
                 is not specified.
             primary_key: Whether this column is part of the primary key of the schema.
                 If `True`, `nullable` is automatically set to `False`.
+            unique: Whether this column must contain unique values.
             min: The minimum time for times in this column (inclusive).
             min_exclusive: Like `min` but exclusive. May not be specified if `min`
                 is specified and vice versa.
@@ -234,6 +239,7 @@ def __init__(
         super().__init__(
             nullable=nullable,
             primary_key=primary_key,
+            unique=unique,
             min=min,
             min_exclusive=min_exclusive,
             max=max,
@@ -309,6 +315,7 @@ def __init__(
         *,
         nullable: bool = False,
         primary_key: bool = False,
+        unique: bool = False,
         min: dt.datetime | None = None,
         min_exclusive: dt.datetime | None = None,
         max: dt.datetime | None = None,
@@ -328,6 +335,7 @@ def __init__(
                 is not specified.
             primary_key: Whether this column is part of the primary key of the schema.
                 If `True`, `nullable` is automatically set to `False`.
+            unique: Whether this column must contain unique values.
             min: The minimum datetime for datetimes in this column (inclusive).
             min_exclusive: Like `min` but exclusive. May not be specified if `min`
                 is specified and vice versa.
@@ -375,6 +383,7 @@ def __init__(
         super().__init__(
             nullable=nullable,
             primary_key=primary_key,
+            unique=unique,
             min=min,
             min_exclusive=min_exclusive,
             max=max,
@@ -472,6 +481,7 @@ def __init__(
         *,
         nullable: bool = False,
         primary_key: bool = False,
+        unique: bool = False,
         min: dt.timedelta | None = None,
         min_exclusive: dt.timedelta | None = None,
         max: dt.timedelta | None = None,
@@ -490,6 +500,7 @@ def __init__(
                 is not specified.
             primary_key: Whether this column is part of the primary key of the schema.
                 If `True`, `nullable` is automatically set to `False`.
+            unique: Whether this column must contain unique values.
             min: The minimum duration for durations in this column (inclusive).
             min_exclusive: Like `min` but exclusive. May not be specified if `min`
                 is specified and vice versa.
@@ -534,6 +545,7 @@ def __init__(
         super().__init__(
             nullable=nullable,
             primary_key=primary_key,
+            unique=unique,
             min=min,
             min_exclusive=min_exclusive,
             max=max,

dataframely/columns/decimal.py

@@ -29,6 +29,7 @@ def __init__(
         *,
         nullable: bool = False,
         primary_key: bool = False,
+        unique: bool = False,
         min: decimal.Decimal | int | None = None,
         min_exclusive: decimal.Decimal | int | None = None,
         max: decimal.Decimal | int | None = None,
@@ -47,6 +48,7 @@ def __init__(
                 is not specified.
             primary_key: Whether this column is part of the primary key of the schema.
                 If `True`, `nullable` is automatically set to `False`.
+            unique: Whether this column must contain unique values.
             min: The minimum value for decimals in this column (inclusive).
             min_exclusive: Like `min` but exclusive. May not be specified if `min`
                 is specified and vice versa.
@@ -91,6 +93,7 @@ def __init__(
         super().__init__(
             nullable=nullable,
             primary_key=primary_key,
+            unique=unique,
             min=min,
             min_exclusive=min_exclusive,
             max=max,

dataframely/columns/enum.py

@@ -28,6 +28,7 @@ def __init__(
         *,
         nullable: bool = False,
         primary_key: bool = False,
+        unique: bool = False,
         check: Check | None = None,
         alias: str | None = None,
         metadata: dict[str, Any] | None = None,
@@ -42,6 +43,7 @@ def __init__(
                 is not specified.
             primary_key: Whether this column is part of the primary key of the schema.
                 If `True`, `nullable` is automatically set to `False`.
+            unique: Whether this column must contain unique values.
             check: A custom rule or multiple rules to run for this column. This can be:
                 - A single callable that returns a non-aggregated boolean expression.
                 The name of the rule is derived from the callable name, or defaults to
@@ -62,6 +64,7 @@ def __init__(
         super().__init__(
             nullable=nullable,
             primary_key=primary_key,
+            unique=unique,
             check=check,
             alias=alias,
             metadata=metadata,

dataframely/columns/float.py

@@ -29,6 +29,7 @@ def __init__(
         *,
         nullable: bool = False,
         primary_key: bool = False,
+        unique: bool = False,
         allow_inf: bool = False,
         allow_nan: bool = False,
         min: float | None = None,
@@ -47,6 +48,7 @@ def __init__(
                 is not specified.
             primary_key: Whether this column is part of the primary key of the schema.
                 If `True`, `nullable` is automatically set to `False`.
+            unique: Whether this column must contain unique values.
             allow_inf: Whether this column may contain infinity values.
             allow_nan: Whether this column may contain NaN values.
             min: The minimum value for floats in this column (inclusive).
@@ -83,6 +85,7 @@ def __init__(
         super().__init__(
             nullable=nullable,
             primary_key=primary_key,
+            unique=unique,
             min=min,
             min_exclusive=min_exclusive,
             max=max,

dataframely/columns/integer.py

@@ -26,6 +26,7 @@ def __init__(
         *,
         nullable: bool = False,
         primary_key: bool = False,
+        unique: bool = False,
         min: int | None = None,
         min_exclusive: int | None = None,
         max: int | None = None,
@@ -43,6 +44,7 @@ def __init__(
                 is not specified.
             primary_key: Whether this column is part of the primary key of the schema.
                 If `True`, `nullable` is automatically set to `False`.
+            unique: Whether this column must contain unique values.
             min: The minimum value for integers in this column (inclusive).
             min_exclusive: Like `min` but exclusive. May not be specified if `min`
                 is specified and vice versa.
@@ -80,6 +82,7 @@ def __init__(
         super().__init__(
             nullable=nullable,
             primary_key=primary_key,
+            unique=unique,
             min=min,
             min_exclusive=min_exclusive,
             max=max,

dataframely/columns/list.py

@@ -35,6 +35,7 @@ def __init__(
         *,
         nullable: bool = False,
         primary_key: bool = False,
+        unique: bool = False,
         check: Check | None = None,
         alias: str | None = None,
         min_length: int | None = None,
@@ -53,6 +54,7 @@ def __init__(
                 In a future release, `nullable=False` will be the default if `nullable`
                 is not specified.
             primary_key: Whether this column is part of the primary key of the schema.
+            unique: Whether this column must contain unique values.
             check: A custom rule or multiple rules to run for this column. This can be:
                 - A single callable that returns a non-aggregated boolean expression.
                 The name of the rule is derived from the callable name, or defaults to
@@ -73,6 +75,7 @@ def __init__(
         super().__init__(
             nullable=nullable,
             primary_key=primary_key,
+            unique=unique,
             check=check,
             alias=alias,
             metadata=metadata,