YDB FQ: avoid outdated syntax "SELECT * FROM cluster.db.table" (copy … (ydb-platform#6970)

Co-authored-by: Grigorii Papashvili <[email protected]>
Co-authored-by: Vitaly Isaev <[email protected]>

Author: 3 people

ydb/library/yql/providers/generic/connector/libcpp/error.cpp

@@ -37,6 +37,8 @@ namespace NYql::NConnector {
                 return NDqProto::StatusIds::StatusCode::StatusIds_StatusCode_UNSUPPORTED;
             case ::Ydb::StatusIds::StatusCode::StatusIds_StatusCode_NOT_FOUND:
                 return NDqProto::StatusIds::StatusCode::StatusIds_StatusCode_BAD_REQUEST;
+            case ::Ydb::StatusIds::StatusCode::StatusIds_StatusCode_SCHEME_ERROR:
+                return NDqProto::StatusIds::StatusCode::StatusIds_StatusCode_SCHEME_ERROR;
             default:
                 ythrow yexception() << "Unexpected YDB status code: " << ::Ydb::StatusIds::StatusCode_Name(error.status());
         }

ydb/library/yql/providers/generic/provider/yql_generic_cluster_config.cpp

@@ -107,14 +107,12 @@ namespace NYql {
                            NYql::TGenericClusterConfig& clusterConfig) {
         auto it = properties.find("database_name");
         if (it == properties.cend()) {
-            // TODO: make this property required during https://st.yandex-team.ru/YQ-2494
-            // ythrow yexception() <<  "missing 'DATABASE_NAME' value";
+            // DATABASE_NAME is a mandatory field for the most of databases,
+            // however, managed YDB does not require it, so we have to accept empty values here.
             return;
         }
 
         if (!it->second) {
-            // TODO: make this property required during https://st.yandex-team.ru/YQ-2494
-            // ythrow yexception() << "invalid 'DATABASE_NAME' value: '" << it->second << "'";
             return;
         }
 
@@ -125,14 +123,12 @@ namespace NYql {
                      NYql::TGenericClusterConfig& clusterConfig) {
         auto it = properties.find("schema");
         if (it == properties.cend()) {
-            // TODO: make this property required during https://st.yandex-team.ru/YQ-2494
-            // ythrow yexception() <<  "missing 'SCHEMA' value";
+            // SCHEMA is optional field
             return;
         }
 
         if (!it->second) {
-            // TODO: make this property required during https://st.yandex-team.ru/YQ-2494
-            // ythrow yexception() << "invalid 'SCHEMA' value: '" << it->second << "'";
+            // SCHEMA is optional field
             return;
         }
 
@@ -318,9 +314,20 @@ namespace NYql {
     }
 
     static const TSet<NConnector::NApi::EDataSourceKind> managedDatabaseKinds{
+        NConnector::NApi::EDataSourceKind::CLICKHOUSE,
+        NConnector::NApi::EDataSourceKind::GREENPLUM,
+        NConnector::NApi::EDataSourceKind::MYSQL,
         NConnector::NApi::EDataSourceKind::POSTGRESQL,
+        NConnector::NApi::EDataSourceKind::YDB,
+    };
+
+    static const TSet<NConnector::NApi::EDataSourceKind> traditionalRelationalDatabaseKinds{
         NConnector::NApi::EDataSourceKind::CLICKHOUSE,
-        NConnector::NApi::EDataSourceKind::YDB};
+        NConnector::NApi::EDataSourceKind::GREENPLUM,
+        NConnector::NApi::EDataSourceKind::MS_SQL_SERVER,
+        NConnector::NApi::EDataSourceKind::MYSQL,
+        NConnector::NApi::EDataSourceKind::POSTGRESQL,
+    };
 
     void ValidateGenericClusterConfig(
         const NYql::TGenericClusterConfig& clusterConfig,
@@ -396,6 +403,17 @@ namespace NYql {
             }
         }
 
+        // All the databases with exception to managed YDB:
+        // * DATABASE_NAME is mandatory field
+        if (traditionalRelationalDatabaseKinds.contains(clusterConfig.GetKind())) {
+            if (!clusterConfig.GetDatabaseName()) {
+                return ValidationError(
+                    clusterConfig,
+                    context,
+                    "You must provide database name explicitly");
+            }
+        }
+
         // check required fields
         if (!clusterConfig.GetName()) {
             return ValidationError(clusterConfig, context, "empty field 'Name'");

ydb/library/yql/providers/generic/provider/yql_generic_dq_integration.cpp

@@ -137,6 +137,7 @@ namespace NYql {
                     }
 
                     // prepare select
+                    Generic::TSource source;
                     auto select = source.mutable_select();
                     select->mutable_from()->set_table(TString(table));
                     select->mutable_data_source_instance()->CopyFrom(tableMeta.value()->DataSourceInstance);

ydb/library/yql/providers/generic/provider/yql_generic_load_meta.cpp

@@ -367,38 +367,8 @@ namespace NYql {
 
         void FillTablePath(NConnector::NApi::TDescribeTableRequest& request, const TGenericClusterConfig& clusterConfig,
                            const TString& tablePath) {
-            // for backward compability full path can be used (cluster_name.`db_name.table`)
-            // TODO: simplify during https://st.yandex-team.ru/YQ-2494
-            const auto dataSourceKind = clusterConfig.GetKind();
-            const auto& dbNameFromConfig = clusterConfig.GetDatabaseName();
-            TStringBuf dbNameTarget, tableName;
-            auto isFullPath = TStringBuf(tablePath).TrySplit('.', dbNameTarget, tableName);
-
-            if (!dbNameFromConfig.empty()) {
-                dbNameTarget = dbNameFromConfig;
-                if (!isFullPath) {
-                    tableName = tablePath;
-                }
-            } else if (!isFullPath) {
-                tableName = tablePath;
-                switch (dataSourceKind) {
-                    case NYql::NConnector::NApi::CLICKHOUSE:
-                        dbNameTarget = "default";
-                        break;
-                    case NYql::NConnector::NApi::POSTGRESQL:
-                        dbNameTarget = "postgres";
-                        break;
-                    case NYql::NConnector::NApi::MS_SQL_SERVER:
-                        dbNameTarget = "mssqlserver";
-                        break;
-                    default:
-                        ythrow yexception() << "You must provide database name explicitly for data source kind: '"
-                                            << NYql::NConnector::NApi::EDataSourceKind_Name(dataSourceKind) << "'";
-                }
-            } // else take database name from table path
-
-            request.mutable_data_source_instance()->set_database(TString(dbNameTarget));
-            request.set_table(TString(tableName));
+            request.mutable_data_source_instance()->set_database(clusterConfig.GetDatabaseName());
+            request.set_table(tablePath);
         }
 
     private: