rebuild and retest

Author: JohnMount

build/lib/data_algebra/polars_model.py

@@ -884,12 +884,18 @@ def _natural_join_step(self, op: data_algebra.data_ops_types.OperatorPlatform, *
         how = op.jointype.lower()
         if how == "full":
             how = "outer"
-        coalesce_columns = (
-            set(op.sources[0].columns_produced()).intersection(op.sources[1].columns_produced()) 
-            - set(op.on_a))
         if how != "right":
+            coalesce_columns = (
+                set(op.sources[0].columns_produced()).intersection(op.sources[1].columns_produced()) 
+                - set(op.on_a))
+            orphan_keys = [c for c in op.on_b if c not in set(op.on_a)]
+            input_right = inputs[1]
+            if len(orphan_keys) > 0:
+                input_right = input_right.with_columns([
+                    pl.col(c).alias(f"{c}_da_join_tmp_key") for c in orphan_keys
+                ])
             res = inputs[0].join(
-                inputs[1],
+                input_right,
                 left_on=op.on_a,
                 right_on=op.on_b,
                 how=how,
@@ -903,10 +909,21 @@ def _natural_join_step(self, op: data_algebra.data_ops_types.OperatorPlatform, *
                         .alias(c)
                     for c in coalesce_columns
                 ])
+            if len(orphan_keys) > 0:
+                res = res.rename({f"{c}_da_join_tmp_key": c for c in orphan_keys})
         else:
             # simulate right join with left join
+            coalesce_columns = (
+                set(op.sources[0].columns_produced()).intersection(op.sources[1].columns_produced()) 
+                - set(op.on_b))
+            orphan_keys = [c for c in op.on_a if c not in set(op.on_b)]
+            input_right = inputs[0]
+            if len(orphan_keys) > 0:
+                input_right = input_right.with_columns([
+                    pl.col(c).alias(f"{c}_da_join_tmp_key") for c in orphan_keys
+                ])
             res = inputs[1].join(
-                inputs[0],
+                input_right,
                 left_on=op.on_b,
                 right_on=op.on_a,
                 how="left",
@@ -920,6 +937,8 @@ def _natural_join_step(self, op: data_algebra.data_ops_types.OperatorPlatform, *
                         .alias(c)
                     for c in coalesce_columns
                 ])
+            if len(orphan_keys) > 0:
+                res = res.rename({f"{c}_da_join_tmp_key": c for c in orphan_keys})
         res = res.select(op.columns_produced())
         return res
 

coverage.txt

@@ -161,14 +161,14 @@ data_algebra/op_container.py             127     49    61%   46-47, 63-71, 80-81
 data_algebra/pandas_base.py              696     68    90%   54, 68, 77, 87, 92, 101, 223, 225, 239, 242, 247, 252, 427, 467, 478, 503, 506, 511, 514, 516, 528, 534-541, 548, 582-587, 617, 621, 624, 626, 663, 717, 758, 775, 795, 813, 823, 838, 886, 894, 902, 917, 928, 940, 959, 974, 1002, 1017, 1054, 1071, 1074, 1085, 1113, 1145, 1154, 1181, 1198, 1210, 1265, 1279-1281
 data_algebra/pandas_model.py              19      2    89%   32-33
 data_algebra/parse_by_lark.py            164     24    85%   71, 93, 108, 129-130, 137, 161, 171, 185-186, 188, 200, 206, 213-217, 245, 253, 263-266
-data_algebra/polars_model.py             584     72    88%   180, 189, 203, 447-453, 459-466, 484-485, 487, 570, 586, 596, 603, 616-620, 628, 630, 655, 658, 663, 666, 715, 733, 749, 816, 832-834, 879, 931, 946, 964, 982, 1002, 1014-1016, 1019, 1026, 1028, 1035-1047, 1054, 1059, 1090, 1119, 1128, 1156, 1171, 1183
+data_algebra/polars_model.py             597     74    88%   180, 189, 203, 447-453, 459-466, 484-485, 487, 570, 586, 596, 603, 616-620, 628, 630, 655, 658, 663, 666, 715, 733, 749, 816, 832-834, 879, 922, 941, 950, 965, 983, 1001, 1021, 1033-1035, 1038, 1045, 1047, 1054-1066, 1073, 1078, 1109, 1138, 1147, 1175, 1190, 1202
 data_algebra/python3_lark.py               1      0   100%
 data_algebra/solutions.py                112      3    97%   63, 308, 367
 data_algebra/sql_format_options.py        15      2    87%   61, 69
 data_algebra/test_util.py                333     62    81%   28-29, 104, 126, 136, 139, 143, 166, 169, 173, 175-178, 189, 246-247, 263-268, 272, 284, 286-294, 331, 333, 344, 352, 363, 370, 376, 388, 399, 413, 468, 472, 523-526, 528-531, 533-536, 538-541, 652-657, 662-663, 665
 data_algebra/util.py                     127     28    78%   26, 59-60, 63-64, 67-68, 71-72, 75-76, 79-80, 83-84, 87-88, 91-92, 95-96, 143, 165, 167, 182, 223, 227, 229
 --------------------------------------------------------------------
-TOTAL                                   6809    974    86%
+TOTAL                                   6822    976    86%
 
 
-======================= 363 passed in 813.20s (0:13:33) ========================
+======================= 363 passed in 832.16s (0:13:52) ========================