Unittest ops + bugfix in Bucketize (#496)

* test_minmix

* updates test

* unittest ops

nvtabular/ops/bucketize.py

@@ -45,6 +45,5 @@ def transform(self, columns, gdf: cudf.DataFrame):
             val = 0
             for boundary in b:
                 val += (gdf[col] >= boundary).astype("int")
-            new_col = f"{col}_{self._id}"
-            new_gdf[new_col] = val
+            new_gdf[col] = val
         return new_gdf

nvtabular/ops/hashed_cross.py

@@ -25,16 +25,15 @@ def __init__(self, num_buckets):
         self.num_buckets = num_buckets
 
     @annotate("HashedCross_op", color="darkgreen", domain="nvt_python")
-    def op_logic(self, columns, gdf: cudf.DataFrame):
+    def transform(self, columns, gdf: cudf.DataFrame):
         new_gdf = cudf.DataFrame()
-        for cross in columns:
-            val = 0
-            for column in cross:
-                val ^= gdf[column].hash_values()  # or however we want to do this aggregation
-            # TODO: support different size buckets per cross
-            val = val % self.bucket_size
-            new_gdf["_X_".join(cross)] = val
+        val = 0
+        for column in columns:
+            val ^= gdf[column].hash_values()  # or however we want to do this aggregation
+        # TODO: support different size buckets per cross
+        val = val % self.num_buckets
+        new_gdf["_X_".join(columns)] = val
         return new_gdf
 
     def output_column_names(self, columns):
-        return ["_X_".join(cross) for cross in columns]
+        return ["_X_".join(columns)]

nvtabular/ops/join_external.py

@@ -88,7 +88,7 @@ def __init__(
         cache="host",
         **kwargs,
     ):
-        super().__init__(replace=False)
+        super(JoinExternal).__init__()
         self.on = on
         self.df_ext = df_ext
         self.on_ext = on_ext or self.on
@@ -155,9 +155,9 @@ def transform(
         return new_gdf
 
     def output_column_names(self, columns):
-        if self.ext_columns:
-            return columns + self.ext_columns
-        return columns + self._ext.columns
+        if self.columns_ext:
+            return list(set(columns + self.columns_ext))
+        return list(set(columns + list(self._ext.columns)))
 
 
 def _detect_format(data):