matchlab 0.1.0.dev3__tar.gz → 0.1.0.dev4__tar.gz

{matchlab-0.1.0.dev3 → matchlab-0.1.0.dev4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: matchlab
-Version: 0.1.0.dev3
+Version: 0.1.0.dev4
 Summary: A local-first library for building, running and evaluating entity resolution pipelines.
 Author: Department for Business and Trade
 Project-URL: Documentation, https://uktrade.github.io/matchlab/

{matchlab-0.1.0.dev3 → matchlab-0.1.0.dev4}/src/matchlab/models/linkers/splinklinker.py

@@ -174,21 +174,52 @@ class SplinkLinker(Linker):
     _id_dtype_l: pl.DataType
     _id_dtype_r: pl.DataType
 
+    @staticmethod
+    def _schema_summary(data: pl.DataFrame) -> str:
+        """Render a compact column-to-dtype summary for diagnostics."""
+        return (
+            ", ".join(f"{column}={dtype}" for column, dtype in data.schema.items())
+            or "<empty>"
+        )
+
+    @classmethod
+    def _conformancy_error(cls, left: pl.DataFrame, right: pl.DataFrame) -> ValueError:
+        """Build a detailed error describing how the two inputs differ."""
+        left_only = [column for column in left.columns if column not in right.columns]
+        right_only = [column for column in right.columns if column not in left.columns]
+        shared = [column for column in left.columns if column in right.columns]
+        differing_dtypes = [
+            f"{column}: left={left.schema[column]}, right={right.schema[column]}"
+            for column in shared
+            if left.schema[column] != right.schema[column]
+        ]
+
+        details = [
+            "SplinkLinker requires input data to be conformant, meaning they "
+            "share the same column names and data formats.",
+            f"Left schema: {cls._schema_summary(left)}",
+            f"Right schema: {cls._schema_summary(right)}",
+        ]
+        if left_only:
+            details.append(f"Columns only on left: {left_only}")
+        if right_only:
+            details.append(f"Columns only on right: {right_only}")
+        if differing_dtypes:
+            details.append("Differing dtypes: " + "; ".join(differing_dtypes))
+        return ValueError("\n".join(details))
+
     def prepare(self, left: pl.DataFrame, right: pl.DataFrame) -> None:
         """Build the Splink linker over left and right, and run its training functions.
 
         Runs each function in `settings.linker_training_functions`, in order, against
         the built linker. `link()` then just predicts against the trained state.
         """
-        if (set(left.columns) != set(right.columns)) or not left.dtypes == right.dtypes:
-            raise ValueError(
-                "SplinkLinker requires input data to be conformant, meaning they "
-                "share the same column names and data formats."
-            )
-
         self._id_dtype_l = left[self.left_id].dtype
         self._id_dtype_r = right[self.right_id].dtype
 
+        if (set(left.columns) != set(right.columns)) or not left.dtypes == right.dtypes:
+            raise self._conformancy_error(left, right)
+
         # Convert to pandas for Splink compatibility
         left_pd = left.with_columns(pl.col(self.left_id).cast(pl.String)).to_pandas()
         right_pd = right.with_columns(pl.col(self.right_id).cast(pl.String)).to_pandas()

{matchlab-0.1.0.dev3 → matchlab-0.1.0.dev4}/src/matchlab.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: matchlab
-Version: 0.1.0.dev3
+Version: 0.1.0.dev4
 Summary: A local-first library for building, running and evaluating entity resolution pipelines.
 Author: Department for Business and Trade
 Project-URL: Documentation, https://uktrade.github.io/matchlab/

{matchlab-0.1.0.dev3 → matchlab-0.1.0.dev4}/src/matchlab.egg-info/SOURCES.txt

@@ -178,6 +178,7 @@ test/models/test_comparison.py
 test/models/test_dedupers_deterministic.py
 test/models/test_linkers_deterministic.py
 test/models/test_linkers_probabilistic.py
+test/models/test_linkers_splink.py
 test/models/test_models.py
 test/plan/__init__.py
 test/plan/test_ground_truth.py

{matchlab-0.1.0.dev3 → matchlab-0.1.0.dev4}/src/matchlab.egg-info/scm_version.json

@@ -2,7 +2,7 @@
   "tag": "0.10.8",
   "distance": 124,
   "node": "g7e5c31cc44787c13778c651231dcaa28893314e6",
-  "dirty": false,
+  "dirty": true,
   "branch": "matchlab",
   "node_date": "2026-08-19"
 }

matchlab-0.1.0.dev4/test/models/test_linkers_splink.py

@@ -0,0 +1,42 @@
+"""Direct unit tests for `SplinkLinker` behavior.
+
+These tests cover Splink-specific schema handling and diagnostics, rather than the
+shared match/no-match behavior exercised in the other linker suites.
+"""
+
+import polars as pl
+import pytest
+from splink import SettingsCreator
+from splink import blocking_rule_library as brl
+from splink import comparison_library as cl
+
+from matchlab.models.linkers.splinklinker import SplinkLinker
+
+
+def test_splink_conformancy_error() -> None:
+    """Splink surfaces the concrete schema mismatch from the real prepare path."""
+    linker = SplinkLinker(
+        left_id="id",
+        right_id="id",
+        linker_training_functions=[],
+        linker_settings=SettingsCreator(
+            link_type="link_only",
+            blocking_rules_to_generate_predictions=[brl.block_on("id")],
+            comparisons=[cl.ExactMatch("id")],
+        ),
+        threshold=None,
+    )
+    left = pl.DataFrame({"id": [1], "company_name": ["left"]})
+    right = pl.DataFrame({"id": [2], "postcode": ["SW1A 1AA"]})
+
+    with pytest.raises(
+        ValueError,
+        match="SplinkLinker requires input data to be conformant",
+    ) as exc_info:
+        linker.prepare(left, right)
+
+    message = str(exc_info.value)
+    assert "Left schema: id=Int64, company_name=String" in message
+    assert "Right schema: id=Int64, postcode=String" in message
+    assert "Columns only on left: ['company_name']" in message
+    assert "Columns only on right: ['postcode']" in message