sqlframe 3.14.1__tar.gz → 3.14.2__tar.gz

{sqlframe-3.14.1 → sqlframe-3.14.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: sqlframe
-Version: 3.14.1
+Version: 3.14.2
 Summary: Turning PySpark Into a Universal DataFrame API
 Home-page: https://github.com/eakmanrq/sqlframe
 Author: Ryan Eakman

{sqlframe-3.14.1 → sqlframe-3.14.2}/sqlframe/_version.py

@@ -12,5 +12,5 @@ __version__: str
 __version_tuple__: VERSION_TUPLE
 version_tuple: VERSION_TUPLE
 
-__version__ = version = '3.14.1'
-__version_tuple__ = version_tuple = (3, 14, 1)
+__version__ = version = '3.14.2'
+__version_tuple__ = version_tuple = (3, 14, 2)

{sqlframe-3.14.1 → sqlframe-3.14.2}/sqlframe/base/dataframe.py

@@ -9,6 +9,7 @@ import json
 import logging
 import sys
 import typing as t
+import uuid
 import zlib
 from copy import copy
 from dataclasses import dataclass
@@ -491,22 +492,20 @@ class BaseDataFrame(t.Generic[SESSION, WRITER, NA, STAT, GROUP_DATA]):
         with_expression = expression.args.get("with")
         if with_expression:
             existing_ctes = with_expression.expressions
-            existing_cte_counts = {x.alias_or_name: 0 for x in existing_ctes}
+            existing_cte_names = {x.alias_or_name for x in existing_ctes}
             replaced_cte_names = {}  # type: ignore
             for cte in ctes:
                 if replaced_cte_names:
                     cte = cte.transform(replace_id_value, replaced_cte_names)  # type: ignore
-                if cte.alias_or_name in existing_cte_counts:
-                    existing_cte_counts[cte.alias_or_name] += 10
+                if cte.alias_or_name in existing_cte_names:
+                    random_filter = exp.Literal.string(uuid.uuid4().hex)
                     # Add unique where filter to ensure that the hash of the CTE is unique
                     cte.set(
                         "this",
                         cte.this.where(
                             exp.EQ(
-                                this=exp.Literal.number(existing_cte_counts[cte.alias_or_name]),
-                                expression=exp.Literal.number(
-                                    existing_cte_counts[cte.alias_or_name]
-                                ),
+                                this=random_filter,
+                                expression=random_filter,
                             )
                         ),
                     )
@@ -520,7 +519,7 @@ class BaseDataFrame(t.Generic[SESSION, WRITER, NA, STAT, GROUP_DATA]):
                             new_cte_alias, dialect=self.session.input_dialect, into=exp.TableAlias
                         ),
                     )
-                    existing_cte_counts[new_cte_alias] = 0
+                    existing_cte_names.add(new_cte_alias)
                 existing_ctes.append(cte)
         else:
             existing_ctes = ctes

{sqlframe-3.14.1 → sqlframe-3.14.2}/sqlframe.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: sqlframe
-Version: 3.14.1
+Version: 3.14.2
 Summary: Turning PySpark Into a Universal DataFrame API
 Home-page: https://github.com/eakmanrq/sqlframe
 Author: Ryan Eakman

{sqlframe-3.14.1 → sqlframe-3.14.2}/tests/integration/test_int_dataframe.py

@@ -2355,3 +2355,29 @@ def test_union_common_root(
     dfs_final = dfs_1.union(dfs_2)
 
     compare_frames(df_final, dfs_final, compare_schema=False)
+
+
+# https://github.com/eakmanrq/sqlframe/issues/253
+def test_union_common_root_again(
+    pyspark_employee: PySparkDataFrame,
+    get_df: t.Callable[[str], BaseDataFrame],
+    compare_frames: t.Callable,
+):
+    df_1 = pyspark_employee.filter(F.col("age") > 40)
+    df_2 = df_1.join(
+        pyspark_employee.select("employee_id").distinct(),
+        on="employee_id",
+        how="right",
+    )
+    df_final = df_1.union(df_2).union(pyspark_employee)
+
+    employee = get_df("employee")
+    dfs_1 = employee.filter(SF.col("age") > 40)
+    dfs_2 = dfs_1.join(
+        employee.select("employee_id").distinct(),
+        on="employee_id",
+        how="right",
+    )
+    dfs_final = dfs_1.union(dfs_2).union(employee)
+
+    compare_frames(df_final, dfs_final, compare_schema=False)