YQ-3221 Result set row is empty with empty schema (ydb-platform#4607)

maximyurchuk · May 17, 2024 · ea6a93e · ea6a93e
1 parent 05b8833
commit ea6a93e
Show file tree

Hide file tree

Showing 5 changed files with 48 additions and 2 deletions.
diff --git a/ydb/core/external_sources/object_storage.cpp b/ydb/core/external_sources/object_storage.cpp
@@ -248,7 +248,7 @@ struct TObjectStorageExternalSource : public IExternalSource {
             ++realSchemaColumnsCount;
         }
 
-        if (realSchemaColumnsCount > 1) {
+        if (realSchemaColumnsCount != 1) {
             issues.AddIssue(MakeErrorIssue(Ydb::StatusIds::BAD_REQUEST, TStringBuilder{} << TStringBuilder() << "Only one column in schema supported in raw format (you have " 
                 << realSchemaColumnsCount << " fields)"));
         }

diff --git a/ydb/library/yql/providers/common/provider/yql_provider.cpp b/ydb/library/yql/providers/common/provider/yql_provider.cpp
@@ -1532,7 +1532,7 @@ bool ValidateFormatForInput(
             ++realSchemaColumnsCount;
         }
 
-        if (realSchemaColumnsCount > 1) {
+        if (realSchemaColumnsCount != 1) {
             ctx.AddError(TIssue(TStringBuilder() << "Only one column in schema supported in raw format (you have "
                 << realSchemaColumnsCount << " fields)"));
             return false;

diff --git a/ydb/tests/fq/s3/test_bindings.py b/ydb/tests/fq/s3/test_bindings.py
@@ -645,3 +645,17 @@ def test_ast_in_failed_query_compilation(self, kikimr, s3, client, unique_prefix
 
         ast = client.describe_query(query_id).result.query.ast.data
         assert "(\'columns \'(\'\"some_unknown_column\"))" in ast, "Invalid query ast"
+
+    @yq_all
+    @pytest.mark.parametrize("client", [{"folder_id": "my_folder"}], indirect=True)
+    def test_raw_empty_schema_binding(self, kikimr, client, unique_prefix):
+        kikimr.control_plane.wait_bootstrap(1)
+        connection_response = client.create_storage_connection(unique_prefix + "fruitbucket", "fbucket")
+        binding_response = client.create_object_storage_binding(name=unique_prefix + "my_binding",
+                                                                path="fruits.csv",
+                                                                format="raw",
+                                                                connection_id=connection_response.result.connection_id,
+                                                                columns=[],
+                                                                check_issues=False)
+        assert "Only one column in schema supported in raw format" in str(binding_response.issues), str(
+            binding_response.issues)
diff --git a/ydb/tests/fq/s3/test_formats.py b/ydb/tests/fq/s3/test_formats.py
@@ -492,3 +492,20 @@ def test_precompute(self, kikimr, s3, client, unique_prefix):
         assert len(result_set.rows) == 1
         assert result_set.rows[0].items[0].bytes_value == b"Pear"
         assert result_set.rows[0].items[1].int32_value == 15
+
+    @yq_all
+    @pytest.mark.parametrize("client", [{"folder_id": "my_folder"}], indirect=True)
+    def test_raw_empty_schema_query(self, kikimr, s3, client, unique_prefix):
+        self.create_bucket_and_upload_file("test.parquet", s3, kikimr)
+        storage_connection_name = unique_prefix + "fruitbucket"
+        client.create_storage_connection(storage_connection_name, "fbucket")
+        sql = f'''
+            SELECT * FROM `{storage_connection_name}`.`*`
+            WITH (format=raw, SCHEMA ());
+            '''
+
+        query_id = client.create_query("test_raw_empty_schema", sql, type=fq.QueryContent.QueryType.ANALYTICS).result.query_id
+        client.wait_query_status(query_id, fq.QueryMeta.FAILED)
+        describe_result = client.describe_query(query_id).result
+        describe_string = "{}".format(describe_result)
+        assert r"Only one column in schema supported in raw format" in describe_string
diff --git a/ydb/tests/fq/yds/test_yds_bindings.py b/ydb/tests/fq/yds/test_yds_bindings.py
@@ -55,3 +55,18 @@ def test_yds_insert(self, client):
         assert result_set.rows[0].items[1].text_value == 'xxx'
         assert result_set.rows[1].items[0].int32_value == 456
         assert result_set.rows[1].items[1].text_value == 'yyy'
+
+    @yq_v1
+    def test_raw_empty_schema_binding(self, kikimr, client, yq_version):
+        self.init_topics(f"pq_test_raw_empty_schema_binding_{yq_version}")
+        connection_response = client.create_yds_connection("myyds2", os.getenv("YDB_DATABASE"),
+                                                           os.getenv("YDB_ENDPOINT"))
+        assert not connection_response.issues, str(connection_response.issues)
+        binding_response = client.create_yds_binding(name="my_binding",
+                                                     stream=self.input_topic,
+                                                     format="raw",
+                                                     connection_id=connection_response.result.connection_id,
+                                                     columns=[],
+                                                     check_issues=False)
+        assert "Only one column in schema supported in raw format" in str(binding_response.issues), str(
+            binding_response.issues)