chore: Add dedicated PrimaryKeyError

Author: MariusMerkleQC

diffly/__init__.py

@@ -11,6 +11,7 @@
     __version__ = "unknown"
 
 
+from ._exceptions import PrimaryKeyError
 from .comparison import compare_frames
 
-__all__ = ["compare_frames"]
+__all__ = ["PrimaryKeyError", "compare_frames"]

diffly/_exceptions.py

@@ -0,0 +1,6 @@
+# Copyright (c) QuantCo 2025-2026
+# SPDX-License-Identifier: BSD-3-Clause
+
+
+class PrimaryKeyError(ValueError):
+    """Raised when there is an issue with the primary key."""

diffly/comparison.py

@@ -14,6 +14,7 @@
 
 from ._cache import cached_method
 from ._conditions import condition_equal_columns, condition_equal_rows
+from ._exceptions import PrimaryKeyError
 from ._utils import (
     ABS_TOL_DEFAULT,
     ABS_TOL_TEMPORAL_DEFAULT,
@@ -131,23 +132,25 @@ def _init_with_validation(
         )
         if primary_key is not None:
             if len(primary_key) == 0:
-                raise ValueError("The primary key columns must not be an empty list.")
+                raise PrimaryKeyError(
+                    "The primary key columns must not be an empty list."
+                )
             if missing := (set(primary_key) - set(left_schema.names())):
-                raise ValueError(
+                raise PrimaryKeyError(
                     f"The primary key columns must be present in the left data frame, "
                     f"but the following are missing: {', '.join(missing)}."
                 )
             if missing := (set(primary_key) - set(right_schema.names())):
-                raise ValueError(
+                raise PrimaryKeyError(
                     f"The primary key columns must be present in the right data frame, "
                     f"but the following are missing: {', '.join(missing)}."
                 )
             if not is_primary_key(left, primary_key):
-                raise ValueError(
+                raise PrimaryKeyError(
                     "The columns are not a primary key for the left data frame."
                 )
             if not is_primary_key(right, primary_key):
-                raise ValueError(
+                raise PrimaryKeyError(
                     "The columns are not a primary key for the right data frame."
                 )
 
@@ -693,7 +696,7 @@ def summary(
 
     def _check_primary_key(self) -> list[str]:
         if self.primary_key is None:
-            raise ValueError(
+            raise PrimaryKeyError(
                 "`primary_key` must be provided to join `left` and `right`."
             )
         return self.primary_key

tests/test_dataframe_comparison.py

@@ -6,7 +6,7 @@
 import polars as pl
 import pytest
 
-from diffly import compare_frames
+from diffly import PrimaryKeyError, compare_frames
 
 
 @pytest.mark.parametrize("primary_key", ["name", ["name"], ("name")])
@@ -20,27 +20,27 @@ def test_primary_key_sequence_types(primary_key: str | Sequence[str]) -> None:
 def test_empty_primary_key() -> None:
     left = pl.DataFrame({"name": ["a", "b"], "value": [1, 2]})
     right = pl.DataFrame({"name": ["a", "b"], "other": [3, 4]})
-    with pytest.raises(ValueError, match="empty"):
+    with pytest.raises(PrimaryKeyError, match="empty"):
         compare_frames(left, right, primary_key=[])
 
 
 def test_missing_primary_key() -> None:
     left = pl.DataFrame({"name": ["a", "b"], "value": [1, 2]})
     right = pl.DataFrame({"name": ["a", "b"], "other": [3, 4]})
     # Primary key that neither frame has
-    with pytest.raises(ValueError, match="left.*missing.*co2_emissions"):
+    with pytest.raises(PrimaryKeyError, match="left.*missing.*co2_emissions"):
         compare_frames(left, right, primary_key=["co2_emissions"])
     # Primary key that the right frame does not have
-    with pytest.raises(ValueError, match="right.*missing.*value"):
+    with pytest.raises(PrimaryKeyError, match="right.*missing.*value"):
         compare_frames(left, right, primary_key=["value"])
 
 
 def test_pk_violation() -> None:
     df_valid = pl.DataFrame({"id": ["a", "b"], "value": [1, 2]})
     df_duplicates = pl.DataFrame({"id": ["a", "a"], "value": [1, 2]})
-    with pytest.raises(ValueError, match="primary key.*left"):
+    with pytest.raises(PrimaryKeyError, match="primary key.*left"):
         compare_frames(df_duplicates, df_valid, primary_key=["id"])
-    with pytest.raises(ValueError, match="primary key.*right"):
+    with pytest.raises(PrimaryKeyError, match="primary key.*right"):
         compare_frames(df_valid, df_duplicates, primary_key=["id"])
 
 

tests/test_fraction_same.py

@@ -14,7 +14,7 @@
     UNSIGNED_INTEGER_DTYPES,
 )
 
-from diffly import compare_frames
+from diffly import PrimaryKeyError, compare_frames
 
 from .utils import FRAME_TYPES, TYPING_FRAME_TYPES
 
@@ -23,7 +23,7 @@ def test_missing_primary_key_fraction_same() -> None:
     left = pl.DataFrame({"id": ["a", "b", "c"], "value": [1, 2, 3]})
     right = pl.DataFrame({"id": ["a", "b"], "value": [1, 2]})
     comparison = compare_frames(left, right)
-    with pytest.raises(ValueError):
+    with pytest.raises(PrimaryKeyError):
         _ = comparison.fraction_same("value")
 
 

tests/test_joined.py

@@ -5,7 +5,7 @@
 import pytest
 from polars.testing import assert_frame_equal
 
-from diffly import compare_frames
+from diffly import PrimaryKeyError, compare_frames
 
 
 def test_joined() -> None:
@@ -31,7 +31,7 @@ def test_joined_missing_primary_key() -> None:
     left = pl.DataFrame({"id": ["a", "b"], "value": [1, 2]})
     right = pl.DataFrame({"id": ["a"], "value": [1]})
     comparison = compare_frames(left, right)
-    with pytest.raises(ValueError):
+    with pytest.raises(PrimaryKeyError):
         _ = comparison.joined()