🔖 0.8.5 (#125)

* 🐛 Fix columns missing after Join by same columns using mapping (#122)

* 🔖 0.8.5

Author: pwwang

datar/__init__.py

@@ -13,7 +13,7 @@
 )
 
 __all__ = ("f", "get_versions")
-__version__ = "0.8.4"
+__version__ = "0.8.5"
 
 apply_init_callbacks()
 

datar/dplyr/join.py

@@ -31,18 +31,20 @@ def _join(
         ret = pd.merge(newx, y, how="cross", copy=copy, suffixes=suffix)
 
     elif isinstance(by, dict):
+        left_on = list(by)
         right_on = list(by.values())
         ret = pd.merge(
             newx,
             y,
-            left_on=list(by.keys()),
+            left_on=left_on,
             right_on=right_on,
             how=how,
             copy=copy,
             suffixes=suffix,
         )
         if not keep:
-            ret.drop(columns=right_on, inplace=True)
+            to_drop = regcall(setdiff, right_on, left_on)
+            ret.drop(columns=to_drop, inplace=True)
 
     elif keep:
         if by is None:

docs/CHANGELOG.md

@@ -1,3 +1,7 @@
+## 0.8.5
+
+- 🐛 Fix columns missing after Join by same columns using mapping (#122)
+
 ## 0.8.4
 
 - ➖ Add optional deps to extras so they aren't installed by default

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "datar"
-version = "0.8.4"
+version = "0.8.5"
 description = "Port of dplyr and other related R packages in python, using pipda."
 authors = ["pwwang <pwwang@pwwang.com>"]
 readme = "README.md"

tests/dplyr/test_join.py

@@ -332,3 +332,9 @@ def test_lose_group_when_by_renamed():
     df2 = tibble(x=[2, 3, 4], z=[5, 6, 7])
     out = df1 >> inner_join(df2, keep=True)
     assert not isinstance(out, TibbleGrouped)
+
+
+def test_columns_not_missing_after_join_by_same_columns_using_mapping_GH122():
+    df1 = tibble(x=[1, 2, 3], y=[3, 4, 5])
+    out = df1 >> left_join(df1, by={"x": "x", "y": "y"})
+    assert_frame_equal(out, df1)