feat: Improve table creation logic and add test for rerunning stats on DuckDB

Author: luisggoncalves

sqlcompare/db/connection.py

@@ -152,22 +152,27 @@ def create_table_from_file(self, table_name: str, file_path: str | Path) -> None
             raise ValueError(f"Unsupported file type: {suffix}")
 
         tmp = f"tmp_{uuid.uuid4().hex}"
-        self.conn.execute(text(f"CREATE TABLE {tmp} AS {read_query}"))
-        cols = [
-            row[1]
-            for row in self.conn.execute(text(f"PRAGMA table_info('{tmp}')")).fetchall()
-        ]
-        exprs = []
-        for c in cols:
-            t = self._infer_column_type(tmp, c)
-            if t == "VARCHAR":
-                exprs.append(f'"{c}"')
-            else:
-                exprs.append(f'CAST("{c}" AS {t}) AS "{c}"')
-        self.conn.execute(
-            text(f"CREATE TABLE {table_name} AS SELECT {', '.join(exprs)} FROM {tmp}")
-        )
-        self.conn.execute(text(f"DROP TABLE {tmp}"))
+        try:
+            self.conn.execute(text(f"CREATE TABLE {tmp} AS {read_query}"))
+            cols = [
+                row[1]
+                for row in self.conn.execute(
+                    text(f"PRAGMA table_info('{tmp}')")
+                ).fetchall()
+            ]
+            exprs = []
+            for c in cols:
+                t = self._infer_column_type(tmp, c)
+                if t == "VARCHAR":
+                    exprs.append(f'"{c}"')
+                else:
+                    exprs.append(f'CAST("{c}" AS {t}) AS "{c}"')
+            self.conn.execute(text(f"DROP TABLE IF EXISTS {table_name}"))
+            self.conn.execute(
+                text(f"CREATE TABLE {table_name} AS SELECT {', '.join(exprs)} FROM {tmp}")
+            )
+        finally:
+            self.conn.execute(text(f"DROP TABLE IF EXISTS {tmp}"))
 
     @property
     def conn(self) -> Connection:

sqlcompare/stats/runner.py

@@ -163,9 +163,8 @@ def compare_table_stats(
 
     def prepare_inputs(db: DBConnection) -> None:
         if spec_prev.kind == "file" and spec_new.kind == "file":
-            if connection is None and connection_id == "duckdb:///:memory:":
-                db.create_table_from_file(table1_name, spec_prev.value)
-                db.create_table_from_file(table2_name, spec_new.value)
+            db.create_table_from_file(table1_name, spec_prev.value)
+            db.create_table_from_file(table2_name, spec_new.value)
         materialize_sql_inputs(
             db,
             previous_spec=spec_prev,

tests/test_stats_comparison.py

@@ -83,6 +83,51 @@ def test_table_command_with_stats_from_files() -> None:
     assert "0 (no change)" in result.output
 
 
+def test_table_command_with_stats_from_files_can_rerun_on_same_duckdb(
+    tmp_path, monkeypatch
+) -> None:
+    db_path = tmp_path / "sqlcompare.duckdb"
+    config_dir = tmp_path / "config"
+    set_cli_env(
+        monkeypatch,
+        config_dir,
+        "duckdb_test",
+        f"duckdb:///{db_path}",
+        schema="analysis_schema",
+    )
+    runner = CliRunner()
+    base = Path(__file__).resolve().parent
+    prev_path = base / "datasets" / "stats_compare" / "previous.csv"
+    curr_path = base / "datasets" / "stats_compare" / "current.csv"
+
+    first = runner.invoke(
+        app,
+        [
+            "run",
+            "stats",
+            str(prev_path),
+            str(curr_path),
+            "--connection",
+            "duckdb_test",
+        ],
+    )
+    second = runner.invoke(
+        app,
+        [
+            "run",
+            "stats",
+            str(prev_path),
+            str(curr_path),
+            "--connection",
+            "duckdb_test",
+        ],
+    )
+
+    assert first.exit_code == 0, first.output
+    assert second.exit_code == 0, second.output
+    assert "already exists" not in second.output.lower()
+
+
 def test_stats_command_with_inline_sql(tmp_path, monkeypatch) -> None:
     db_path = tmp_path / "sqlcompare.duckdb"
     seed_duckdb(db_path)