Support pandas 2.2.0

Fixes a unit test to support pandas 2.2.0+. The pandas release fixes a
sorting bug with pandas-dev/pandas#54611. This
commit changes the expected results accordingly.

Also fixes a merge type mismatch introduced by upstream #1709: the
codelist metadata side was cast to StringDtype but the evaluation
dataset side was not. With pandas 2.2.0, empty columns infer as float64,
and merging float64 with string is rejected. Casting both sides to
string before the merge resolves this.

Author: filippsatverily

cdisc_rules_engine/operations/codelist_extensible.py

@@ -32,6 +32,10 @@ def _handle_multiple_versions(self) -> pd.Series:
                 "codelist_code": "string",
             }
         )
+        cast_cols = {self.params.ct_version: "string"}
+        if self.params.codelist_code in self.evaluation_dataset.columns:
+            cast_cols[self.params.codelist_code] = "string"
+        self.evaluation_dataset = self.evaluation_dataset.astype(cast_cols)
         if self.params.codelist_code in self.evaluation_dataset.columns:
             is_extensible = self.evaluation_dataset.merge(
                 ct_df.data,

cdisc_rules_engine/operations/codelist_terms.py

@@ -64,6 +64,10 @@ def _handle_multiple_versions(self) -> pd.Series:
                 "codelist_code": "string",
             }
         )
+        cast_cols = {self.params.ct_version: "string"}
+        if self.params.codelist_code in self.evaluation_dataset.columns:
+            cast_cols[self.params.codelist_code] = "string"
+        self.evaluation_dataset = self.evaluation_dataset.astype(cast_cols)
         if self.params.codelist_code in self.evaluation_dataset.columns:
             result = self.evaluation_dataset.merge(
                 ct_df.data,

pyproject.toml

@@ -25,7 +25,7 @@ dependencies = [
     "numpy >=1.26.0",
     "odmlib >=0.1.4",
     "openpyxl >=3.1.5",
-    "pandas >=2.1.4, <2.2.0",
+    "pandas >=2.1.4, <3.0.0",
     "psutil >=6.1.1",
     "pyinstaller >=6.11.0",
     "pympler >=1.1",

tests/unit/test_dataset_builders/test_dataset_metadata_define_dataset_builder.py

@@ -149,7 +149,7 @@ def test_dataset_metadata_define_dataset_builder(dataset_path):
             expected_results["dm.xpt"],
             expected_results["ae.xpt"],
         ]
-    ).astype(object)
+    ).astype(object).sort_values("dataset_location").reset_index(drop=True)
 
     result_df = result.data[expected_df.columns].reset_index(drop=True)