fix(ingest): consider sql parsing fallback as failure (datahub-projec…

…t#11896)

metadata-ingestion/src/datahub/cli/check_cli.py

@@ -268,7 +268,9 @@ def sql_lineage(
     )
 
     logger.debug("Sql parsing debug info: %s", lineage.debug_info)
-    if lineage.debug_info.error:
+    if lineage.debug_info.table_error:
+        raise lineage.debug_info.table_error
+    elif lineage.debug_info.error:
         logger.debug("Sql parsing error details", exc_info=lineage.debug_info.error)
 
     click.echo(lineage.json(indent=4))

metadata-ingestion/src/datahub/sql_parsing/sqlglot_lineage.py

@@ -904,6 +904,15 @@ def _sqlglot_lineage_inner(
     logger.debug("Parsing lineage from sql statement: %s", sql)
     statement = parse_statement(sql, dialect=dialect)
 
+    if isinstance(statement, sqlglot.exp.Command):
+        # For unsupported syntax, sqlglot will usually fallback to parsing as a Command.
+        # This is effectively a parsing error, and we won't get any lineage from it.
+        # See https://github.com/tobymao/sqlglot/commit/3a13fdf4e597a2f0a3f9fc126a129183fe98262f
+        # and https://github.com/tobymao/sqlglot/pull/2874
+        raise UnsupportedStatementTypeError(
+            f"Got unsupported syntax for statement: {sql}"
+        )
+
     original_statement, statement = statement, statement.copy()
     # logger.debug(
     #     "Formatted sql statement: %s",

metadata-ingestion/tests/unit/sql_parsing/goldens/test_sqlite_attach_database.json

@@ -0,0 +1,12 @@
+{
+    "query_type": "UNKNOWN",
+    "query_type_props": {},
+    "query_fingerprint": null,
+    "in_tables": [],
+    "out_tables": [],
+    "column_lineage": null,
+    "debug_info": {
+        "confidence": 0.0,
+        "generalized_statement": null
+    }
+}

metadata-ingestion/tests/unit/sql_parsing/test_sqlglot_lineage.py

@@ -1268,3 +1268,14 @@ def test_bigquery_subquery_column_inference() -> None:
         dialect="bigquery",
         expected_file=RESOURCE_DIR / "test_bigquery_subquery_column_inference.json",
     )
+
+
+def test_sqlite_attach_database() -> None:
+    assert_sql_result(
+        """\
+ATTACH DATABASE ':memory:' AS aux1
+""",
+        dialect="sqlite",
+        expected_file=RESOURCE_DIR / "test_sqlite_attach_database.json",
+        allow_table_error=True,
+    )