Issue with combining empty dataframes, small update to matminerfast

modnet/featurizers/featurizers.py

@@ -98,7 +98,26 @@ def featurize(self, df: pd.DataFrame) -> pd.DataFrame:
         if self.site_featurizers:
             df_site = self.featurize_site(df)
 
-        return df_composition.join(df_structure.join(df_site, lsuffix="l"), rsuffix="r")
+        def safe_join(left: pd.DataFrame, right: pd.DataFrame, lsuffix: str = "", rsuffix: str = "") -> pd.DataFrame:
+            """
+            Join two dataframes with suffixes for overlapping columns.
+            If either side is empty, return the non-empty dataframe.
+            If both are empty, return an empty DataFrame.
+            """
+            if left.empty and right.empty:
+                return pd.DataFrame([])
+            elif left.empty:
+                return right
+            elif right.empty:
+                return left
+            else:
+                return left.join(right, lsuffix=lsuffix, rsuffix=rsuffix)
+
+        # Here we use lsuffix="l" to resolve any overlapping column names.
+        tmp = safe_join(df_structure, df_site, lsuffix="l")
+
+        final_df = safe_join(df_composition, tmp, rsuffix="r")
+        return final_df
 
     def _fit_apply_featurizers(
         self,
@@ -200,11 +219,11 @@ def featurize_composition(self, df: pd.DataFrame) -> pd.DataFrame:
         """
 
         df = df.copy()
-
+        df["composition"] = df["structure"].apply(lambda s: s.composition)
         if self.composition_featurizers:
 
             LOG.info("Applying composition featurizers...")
-            df["composition"] = df["structure"].apply(lambda s: s.composition)
+
             df = self._fit_apply_featurizers(
                 df,
                 self.composition_featurizers,

modnet/featurizers/presets/__init__.py

@@ -11,6 +11,7 @@
     MatminerAll2023Featurizer,
     CompositionOnlyMatminerAll2023Featurizer,
 )
+from .matminer_2024_fast import Matminer2024FastFeaturizer
 from modnet.featurizers import MODFeaturizer
 
 DEFAULT_FEATURIZER: str = "Matminer2023"
@@ -23,4 +24,5 @@
     "MatminerAll2023": MatminerAll2023Featurizer,
     "CompositionOnlyMatminer2023": CompositionOnlyMatminer2023Featurizer,
     "CompositionOnlyMatminerAll2023": CompositionOnlyMatminerAll2023Featurizer,
+    "Matminer2024Fast":Matminer2024FastFeaturizer,
 }