Fix(table_diff): Allow diffing of empty tables (#4347)

Author: erindru

sqlmesh/core/console.py

@@ -2181,6 +2181,12 @@ def show_schema_diff(self, schema_diff: SchemaDiff) -> None:
     def show_row_diff(
         self, row_diff: RowDiff, show_sample: bool = True, skip_grain_check: bool = False
     ) -> None:
+        if row_diff.empty:
+            self.console.print(
+                "\n[b][red]Neither the source nor the target table contained any records[/red][/b]"
+            )
+            return
+
         source_name = row_diff.source
         if row_diff.source_alias:
             source_name = row_diff.source_alias.upper()

sqlmesh/core/table_diff.py

@@ -83,6 +83,15 @@ def target_count(self) -> int:
         """Count of the target."""
         return int(self.stats["t_count"])
 
+    @property
+    def empty(self) -> bool:
+        return (
+            self.source_count == 0
+            and self.target_count == 0
+            and self.s_only_count == 0
+            and self.t_only_count == 0
+        )
+
     @property
     def count_pct_change(self) -> float:
         """The percentage change of the counts."""
@@ -446,7 +455,7 @@ def name(e: exp.Expression) -> str:
 
                 summary_query = exp.select(*summary_sums).from_(table)
 
-                stats_df = self.adapter.fetchdf(summary_query, quote_identifiers=True)
+                stats_df = self.adapter.fetchdf(summary_query, quote_identifiers=True).fillna(0)
                 stats_df["s_only_count"] = stats_df["s_count"] - stats_df["join_count"]
                 stats_df["t_only_count"] = stats_df["t_count"] - stats_df["join_count"]
                 stats = stats_df.iloc[0].to_dict()

tests/core/test_table_diff.py

@@ -788,3 +788,44 @@ def test_data_diff_forward_only(sushi_context_fixed_date, capsys, caplog):
         assert row_diff.stats["distinct_count_t"] == 2
         assert row_diff.s_sample.shape == (2, 2)
         assert row_diff.t_sample.shape == (2, 2)
+
+
+def test_data_diff_empty_tables():
+    engine_adapter = DuckDBConnectionConfig().create_engine_adapter()
+
+    columns_to_types_src = {
+        "key": exp.DataType.build("int"),
+        "value": exp.DataType.build("varchar"),
+    }
+    columns_to_types_target = {
+        "key": exp.DataType.build("int"),
+        "value2": exp.DataType.build("varchar"),
+    }
+
+    engine_adapter.create_table("table_diff_source", columns_to_types_src)
+    engine_adapter.create_table("table_diff_target", columns_to_types_target)
+
+    table_diff = TableDiff(
+        adapter=engine_adapter,
+        source="table_diff_source",
+        target="table_diff_target",
+        source_alias="dev",
+        target_alias="prod",
+        on=["key"],
+    )
+
+    # should show the schema diff
+    schema_diff = table_diff.schema_diff()
+    assert len(schema_diff.added) == 1
+    assert schema_diff.added[0][0] == "value2"
+    assert len(schema_diff.removed) == 1
+    assert schema_diff.removed[0][0] == "value"
+
+    # should not error on the row diff
+    row_diff = table_diff.row_diff()
+    assert row_diff.empty
+
+    output = capture_console_output("show_row_diff", row_diff=row_diff)
+    assert (
+        strip_ansi_codes(output) == "Neither the source nor the target table contained any records"
+    )