Fix: qualify all columns in the table diff source/target query scope

Author: georgesittas

sqlmesh/core/table_diff.py

@@ -12,6 +12,7 @@
 from sqlglot.helper import ensure_list
 from sqlglot.optimizer.normalize_identifiers import normalize_identifiers
 from sqlglot.optimizer.qualify_columns import quote_identifiers
+from sqlglot.optimizer.scope import find_all_in_scope
 
 from sqlmesh.utils.pydantic import PydanticModel
 
@@ -314,21 +315,27 @@ def name(e: exp.Expression) -> str:
 
             source_query = (
                 exp.select(
-                    *(exp.column(c, "s") for c in source_schema),
+                    *(exp.column(c) for c in source_schema),
                     self.source_key_expression.as_(SQLMESH_JOIN_KEY_COL),
                 )
                 .from_(self.source_table.as_("s"))
                 .where(self.where)
             )
             target_query = (
                 exp.select(
-                    *(exp.column(c, "t") for c in target_schema),
+                    *(exp.column(c) for c in target_schema),
                     self.target_key_expression.as_(SQLMESH_JOIN_KEY_COL),
                 )
                 .from_(self.target_table.as_("t"))
                 .where(self.where)
             )
 
+            # Ensure every column is qualified with the alias in the source and target queries
+            for col in find_all_in_scope(source_query, exp.Column):
+                col.set("table", exp.to_identifier("s"))
+            for col in find_all_in_scope(target_query, exp.Column):
+                col.set("table", exp.to_identifier("t"))
+
             source_table = exp.table_("__source")
             target_table = exp.table_("__target")
             stats_table = exp.table_("__stats")

tests/core/engine_adapter/integration/test_integration_bigquery.py

@@ -415,7 +415,7 @@ def test_table_diff_table_name_matches_column_name(ctx: TestContext):
     #
     # This is a problem, because we compare the columns of the source and target tables using the
     # equality operator (=), which is not defined for struct values in BigQuery, leading to an error.
-    query: exp.Query = exp.maybe_parse("SELECT 1 AS id, 2 AS source, 3 AS target")
+    query: exp.Query = exp.maybe_parse("SELECT 1 AS s, 2 AS source, 3 AS target")
 
     ctx.engine_adapter.ctas(src_table, query)
     ctx.engine_adapter.ctas(target_table, query)
@@ -424,7 +424,7 @@ def test_table_diff_table_name_matches_column_name(ctx: TestContext):
         adapter=ctx.engine_adapter,
         source=exp.table_name(src_table),
         target=exp.table_name(target_table),
-        on=["id"],
+        on=["s"],
     )
 
     row_diff = table_diff.row_diff()