rgwfuncs 0.0.9__py3-none-any.whl → 0.0.11__py3-none-any.whl

rgwfuncs/df_lib.py

@@ -1631,10 +1631,27 @@ def union_join(df1: pd.DataFrame, df2: pd.DataFrame) -> pd.DataFrame:
     if set(df1.columns) != set(df2.columns):
         raise ValueError("Both DataFrames must have the same columns for a union join")
 
-    result_df = pd.concat([df1, df2], ignore_index=True).drop_duplicates()
+    # Inspect initial columns
+    print("Initial df1 columns:", df1.columns)
+    print("Initial df2 columns:", df2.columns)
+
+    # Drop all-NA columns, if any
+    df1_clean = df1.dropna(axis=1, how='all')
+    df2_clean = df2.dropna(axis=1, how='all')
+
+    # Inspect resulting columns after dropping all-NA columns
+    print("Cleaned df1 columns:", df1_clean.columns)
+    print("Cleaned df2 columns:", df2_clean.columns)
+
+    # Ensure they still have the same columns after dropping all-NA columns
+    if set(df1_clean.columns) != set(df2_clean.columns):
+        raise ValueError("Both DataFrames must have the same columns after dropping all-NA columns")
+
+    result_df = pd.concat([df1_clean, df2_clean], ignore_index=True).drop_duplicates()
     return result_df
 
 
+
 def bag_union_join(df1: pd.DataFrame, df2: pd.DataFrame) -> pd.DataFrame:
     """
     Perform a bag union join, concatenating the two DataFrames without dropping duplicates.
@@ -1652,7 +1669,15 @@ def bag_union_join(df1: pd.DataFrame, df2: pd.DataFrame) -> pd.DataFrame:
     if set(df1.columns) != set(df2.columns):
         raise ValueError("Both DataFrames must have the same columns for a bag union join")
 
-    result_df = pd.concat([df1, df2], ignore_index=True)
+    # Drop all-NA columns, if any
+    df1_clean = df1.dropna(axis=1, how='all')
+    df2_clean = df2.dropna(axis=1, how='all')
+
+    # Ensure they still have the same columns after dropping all-NA columns
+    if set(df1_clean.columns) != set(df2_clean.columns):
+        raise ValueError("Both DataFrames must have the same columns after dropping all-NA columns")
+
+    result_df = pd.concat([df1_clean, df2_clean], ignore_index=True)
     return result_df
 
 

{rgwfuncs-0.0.9.dist-info → rgwfuncs-0.0.11.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: rgwfuncs
-Version: 0.0.9
+Version: 0.0.11
 Summary: A functional programming paradigm for mathematical modelling and data science
 Home-page: https://github.com/ryangerardwilson/rgwfunc
 Author: Ryan Gerard Wilson

rgwfuncs-0.0.11.dist-info/RECORD

@@ -0,0 +1,8 @@
+rgwfuncs/__init__.py,sha256=BP8Nh8ivyCCz8Ga-21JW3NWInJFOElKoIfRuioJRWbA,1076
+rgwfuncs/df_lib.py,sha256=N5XCe9LYMIVjh0pg2Xc1i3Md--mdNJyJ5t7NpO9x4-s,63973
+rgwfuncs-0.0.11.dist-info/LICENSE,sha256=7EI8xVBu6h_7_JlVw-yPhhOZlpY9hP8wal7kHtqKT_E,1074
+rgwfuncs-0.0.11.dist-info/METADATA,sha256=mqOIwsthIDAdpgtcFt35vW_xc99f2xtu5O_k4_Y-wQY,32059
+rgwfuncs-0.0.11.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+rgwfuncs-0.0.11.dist-info/entry_points.txt,sha256=j-c5IOPIQ0252EaOV6j6STio56sbXl2C4ym_fQ0lXx0,43
+rgwfuncs-0.0.11.dist-info/top_level.txt,sha256=aGuVIzWsKiV1f2gCb6mynx0zx5ma0B1EwPGFKVEMTi4,9
+rgwfuncs-0.0.11.dist-info/RECORD,,

rgwfuncs-0.0.9.dist-info/RECORD

@@ -1,8 +0,0 @@
-rgwfuncs/__init__.py,sha256=BP8Nh8ivyCCz8Ga-21JW3NWInJFOElKoIfRuioJRWbA,1076
-rgwfuncs/df_lib.py,sha256=3PYfu_zs8HfL56C9Sb41jzoyaG9Oc7x5MZQYvo1zy6M,62930
-rgwfuncs-0.0.9.dist-info/LICENSE,sha256=7EI8xVBu6h_7_JlVw-yPhhOZlpY9hP8wal7kHtqKT_E,1074
-rgwfuncs-0.0.9.dist-info/METADATA,sha256=tJiGspLMJbt78FGyyvf3w10ZbWCn17PJ4070wUsH1ew,32058
-rgwfuncs-0.0.9.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-rgwfuncs-0.0.9.dist-info/entry_points.txt,sha256=j-c5IOPIQ0252EaOV6j6STio56sbXl2C4ym_fQ0lXx0,43
-rgwfuncs-0.0.9.dist-info/top_level.txt,sha256=aGuVIzWsKiV1f2gCb6mynx0zx5ma0B1EwPGFKVEMTi4,9
-rgwfuncs-0.0.9.dist-info/RECORD,,