feat: do TPC-H plan validation for DuckDB with Producer (#169)

This PR reuses the plan validation that is part of the `Producer` interface in the tests that check plan validity for the TPC-H queries for DuckDB. This leads to the same result in terms of validity; however, the error class is different in the one error case, so the outcome changes from one error to another for that query. Signed-off-by: Ingo Müller <[email protected]>
substrait-io · Dec 16, 2024 · cd5ff39 · cd5ff39
1 parent 888adc7
commit cd5ff39
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 16 deletions.
diff --git a/...ans_valid/test_duckdb_substrait_plans_valid/test_tpch_sql_18/test_tpch_sql_18_outcome.txt b/...ans_valid/test_duckdb_substrait_plans_valid/test_tpch_sql_18/test_tpch_sql_18_outcome.txt
@@ -1 +1 @@
-<class 'google.protobuf.json_format.ParseError'>
+<class 'ValueError'>
diff --git a/substrait_consumer/tests/integration/test_tpch_plans_valid.py b/substrait_consumer/tests/integration/test_tpch_plans_valid.py
@@ -2,7 +2,6 @@
 
 import duckdb
 import pytest
-import substrait_validator as sv
 
 from pytest_snapshot.plugin import Snapshot
 
@@ -116,25 +115,12 @@ def test_duckdb_substrait_plans_valid(
             sql_query:
                 SQL query.
         """
-        config = sv.Config()
-
-        # Duckdb plan overrides
-        # not yet implemented: typecast validation rules are not yet implemented
-        config.override_diagnostic_level(1, "info", "info")
-        # function definition unavailable: cannot check validity of call
-        config.override_diagnostic_level(6003, "info", "info")
-        # Function Anchor to YAML file
-        config.override_diagnostic_level(3001, "info", "info")
-        # too few field names
-        config.override_diagnostic_level(4003, "info", "info")
 
         # Format the sql query by inserting all the table names
         self.duckdb_producer.setup(self.db_connection, local_files, named_tables)
 
         try:
-            proto_bytes = self.duckdb_producer.produce_substrait(sql_query)
-            # TODO: failures in the validator obstruct an otherwise passing test!
-            sv.check_plan_valid(proto_bytes, config)
+            self.duckdb_producer.produce_substrait(sql_query, validate=True)
         except BaseException as e:
             snapshot.assert_match(str(type(e)), f"{test_name}_outcome.txt")
             return
Original file line number	Diff line number	Diff line change
		@@ -1 +1 @@
		<class 'google.protobuf.json_format.ParseError'>
		<class 'ValueError'>