Error on invalid field format being sent to Validator (#462)

Author: mallport

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "dapla-toolbelt-pseudo"
-version = "4.5.0"
+version = "5.0.0"
 description = "Pseudonymization extensions for Dapla"
 authors = ["Dapla Developers <dapla-platform-developers@ssb.no>"]
 license = "MIT"

src/dapla_pseudo/v1/validation.py

@@ -110,6 +110,8 @@ def validate_map_to_stable_id(
             Returns:
                 Result: Containing a result dataframe with associated metadata.
             """
+            Validator._ensure_field_valid(self._field, self._dataframe)
+
             response: requests.Response = _client()._post_to_sid_endpoint(
                 "sid/lookup/batch",
                 self._dataframe[self._field].to_list(),
@@ -142,3 +144,40 @@ def validate_map_to_stable_id(
                     ],
                 )
             )
+
+    @staticmethod
+    def _ensure_field_valid(field: str, dataframe: pl.DataFrame) -> None:
+        """Ensure that all values are numeric and valid.
+
+        This is necessary for SID mapping.
+
+        Args:
+            field (str): The identifier field.
+            dataframe (pl.DataFrame): The dataframe to validate.
+
+        Raises:
+            ValueError: If the field does not exist in the dataframe.
+        """
+        if field not in dataframe.columns:
+            raise ValueError(f"Field '{field}' does not exist in the dataframe.")
+
+        if dataframe.select(pl.col(field)).to_series().has_nulls():
+            raise ValueError(
+                f"Field '{field}' contains None/NaN values which are invalid for SID mapping."
+            )
+
+        allowed_pattern = r"^\d+$"  # only numeric
+        invalid_entries = (
+            dataframe.select(
+                pl.col(field).str.contains(allowed_pattern).alias("is_valid"),
+                pl.col(field),
+            )
+            .filter(~pl.col("is_valid"))
+            .select(pl.col(field))
+        )
+
+        if not invalid_entries.is_empty():
+            invalid_values = invalid_entries.select(pl.col(field)).to_series().to_list()
+            raise ValueError(
+                f"Field '{field}' contains non-numeric values which are invalid for SID mapping: {invalid_values}"
+            )

tests/v1/integration/test_validate.py

@@ -28,3 +28,12 @@ def test_validate_not_valid(df_personer: pl.DataFrame) -> None:
     )
     pl_result = result.to_polars()
     assert sorted(pl_result["fnr"].to_list()) == sorted(expected_result)
+
+
+@pytest.mark.usefixtures("setup")
+@pytest.mark.parametrize("invalid_input", ["", "abc", None])
+@integration_test()
+def test_validate_invalid_input(invalid_input: str | None) -> None:
+    df_invalid = pl.DataFrame({"fnr": [invalid_input]})
+    with pytest.raises(ValueError):
+        Validator.from_polars(df_invalid).on_field("fnr").validate_map_to_stable_id()