feat: add table id and engine to information_schema.TABLES (#1407)

* feat: add table id and engine to informatin_schema.TABLES

* Update src/catalog/src/information_schema/tables.rs

Co-authored-by: Yingwen <realevenyag@gmail.com>

* chore: change table_engine to engine

* test: update sqlness for information schema

* test: update information_schema test in frontend::tests::instance_test.rs

* fix: github action sqlness information_schema test fail

* test: ignore table_id in information_schema

* test: support distribute and standalone have different output

---------

Co-authored-by: Yingwen <realevenyag@gmail.com>

src/catalog/src/information_schema/tables.rs

@@ -23,7 +23,7 @@ use datafusion::physical_plan::stream::RecordBatchStreamAdapter as DfRecordBatch
 use datafusion::physical_plan::SendableRecordBatchStream as DfSendableRecordBatchStream;
 use datatypes::prelude::{ConcreteDataType, ScalarVectorBuilder, VectorRef};
 use datatypes::schema::{ColumnSchema, Schema, SchemaRef};
-use datatypes::vectors::StringVectorBuilder;
+use datatypes::vectors::{StringVectorBuilder, UInt32VectorBuilder};
 use snafu::ResultExt;
 use table::metadata::TableType;
 
@@ -44,6 +44,8 @@ impl InformationSchemaTables {
             ColumnSchema::new("table_schema", ConcreteDataType::string_datatype(), false),
             ColumnSchema::new("table_name", ConcreteDataType::string_datatype(), false),
             ColumnSchema::new("table_type", ConcreteDataType::string_datatype(), false),
+            ColumnSchema::new("table_id", ConcreteDataType::uint32_datatype(), true),
+            ColumnSchema::new("engine", ConcreteDataType::string_datatype(), true),
         ]));
         Self {
             schema,
@@ -73,6 +75,8 @@ struct InformationSchemaTablesBuilder {
     schema_names: StringVectorBuilder,
     table_names: StringVectorBuilder,
     table_types: StringVectorBuilder,
+    table_ids: UInt32VectorBuilder,
+    engines: StringVectorBuilder,
 }
 
 impl InformationSchemaTablesBuilder {
@@ -85,6 +89,8 @@ impl InformationSchemaTablesBuilder {
             schema_names: StringVectorBuilder::with_capacity(42),
             table_names: StringVectorBuilder::with_capacity(42),
             table_types: StringVectorBuilder::with_capacity(42),
+            table_ids: UInt32VectorBuilder::with_capacity(42),
+            engines: StringVectorBuilder::with_capacity(42),
         }
     }
 
@@ -100,7 +106,15 @@ impl InformationSchemaTablesBuilder {
             let Some(schema) = self.catalog_provider.schema(&schema_name)? else { continue };
             for table_name in schema.table_names()? {
                 let Some(table) = schema.table(&table_name).await? else { continue };
-                self.add_table(&catalog_name, &schema_name, &table_name, table.table_type());
+                let table_info = table.table_info();
+                self.add_table(
+                    &catalog_name,
+                    &schema_name,
+                    &table_name,
+                    table.table_type(),
+                    Some(table_info.ident.table_id),
+                    Some(&table_info.meta.engine),
+                );
             }
         }
 
@@ -110,6 +124,8 @@ impl InformationSchemaTablesBuilder {
             INFORMATION_SCHEMA_NAME,
             TABLES,
             TableType::View,
+            None,
+            None,
         );
 
         self.finish()
@@ -121,6 +137,8 @@ impl InformationSchemaTablesBuilder {
         schema_name: &str,
         table_name: &str,
         table_type: TableType,
+        table_id: Option<u32>,
+        engine: Option<&str>,
     ) {
         self.catalog_names.push(Some(catalog_name));
         self.schema_names.push(Some(schema_name));
@@ -130,6 +148,8 @@ impl InformationSchemaTablesBuilder {
             TableType::View => "VIEW",
             TableType::Temporary => "LOCAL TEMPORARY",
         }));
+        self.table_ids.push(table_id);
+        self.engines.push(engine);
     }
 
     fn finish(&mut self) -> Result<RecordBatch> {
@@ -138,6 +158,8 @@ impl InformationSchemaTablesBuilder {
             Arc::new(self.schema_names.finish()),
             Arc::new(self.table_names.finish()),
             Arc::new(self.table_types.finish()),
+            Arc::new(self.table_ids.finish()),
+            Arc::new(self.engines.finish()),
         ];
         RecordBatch::new(self.schema.clone(), columns).context(CreateRecordBatchSnafu)
     }

src/frontend/src/tests/instance_test.rs

@@ -904,6 +904,7 @@ async fn test_execute_copy_from_s3(instance: Arc<dyn MockInstance>) {
 
 #[apply(both_instances_cases)]
 async fn test_information_schema(instance: Arc<dyn MockInstance>) {
+    let is_distributed_mode = instance.is_distributed_mode();
     let instance = instance.frontend();
 
     let sql = "create table another_table(i bigint time index)";
@@ -913,27 +914,55 @@ async fn test_information_schema(instance: Arc<dyn MockInstance>) {
 
     // User can only see information schema under current catalog.
     // A necessary requirement to GreptimeCloud.
-    let sql = "select table_catalog, table_schema, table_name, table_type from information_schema.tables where table_type != 'SYSTEM VIEW' order by table_name";
+    let sql = "select table_catalog, table_schema, table_name, table_type, table_id, engine from information_schema.tables where table_type != 'SYSTEM VIEW' order by table_name";
 
     let output = execute_sql(&instance, sql).await;
-    let expected = "\
-+---------------+--------------------+------------+------------+
-| table_catalog | table_schema       | table_name | table_type |
-+---------------+--------------------+------------+------------+
-| greptime      | public             | numbers    | BASE TABLE |
-| greptime      | public             | scripts    | BASE TABLE |
-| greptime      | information_schema | tables     | VIEW       |
-+---------------+--------------------+------------+------------+";
+    let expected = match is_distributed_mode {
+        true => {
+            "\
++---------------+--------------------+------------+------------+----------+-------------+
+| table_catalog | table_schema       | table_name | table_type | table_id | engine      |
++---------------+--------------------+------------+------------+----------+-------------+
+| greptime      | public             | numbers    | BASE TABLE | 1        | test_engine |
+| greptime      | public             | scripts    | BASE TABLE | 1024     | mito        |
+| greptime      | information_schema | tables     | VIEW       |          |             |
++---------------+--------------------+------------+------------+----------+-------------+"
+        }
+        false => {
+            "\
++---------------+--------------------+------------+------------+----------+-------------+
+| table_catalog | table_schema       | table_name | table_type | table_id | engine      |
++---------------+--------------------+------------+------------+----------+-------------+
+| greptime      | public             | numbers    | BASE TABLE | 1        | test_engine |
+| greptime      | public             | scripts    | BASE TABLE | 1        | mito        |
+| greptime      | information_schema | tables     | VIEW       |          |             |
++---------------+--------------------+------------+------------+----------+-------------+"
+        }
+    };
+
     check_output_stream(output, expected).await;
 
     let output = execute_sql_with(&instance, sql, query_ctx).await;
-    let expected = "\
-+-----------------+--------------------+---------------+------------+
-| table_catalog   | table_schema       | table_name    | table_type |
-+-----------------+--------------------+---------------+------------+
-| another_catalog | another_schema     | another_table | BASE TABLE |
-| another_catalog | information_schema | tables        | VIEW       |
-+-----------------+--------------------+---------------+------------+";
+    let expected = match is_distributed_mode {
+        true => {
+            "\
++-----------------+--------------------+---------------+------------+----------+--------+
+| table_catalog   | table_schema       | table_name    | table_type | table_id | engine |
++-----------------+--------------------+---------------+------------+----------+--------+
+| another_catalog | another_schema     | another_table | BASE TABLE | 1025     | mito   |
+| another_catalog | information_schema | tables        | VIEW       |          |        |
++-----------------+--------------------+---------------+------------+----------+--------+"
+        }
+        false => {
+            "\
++-----------------+--------------------+---------------+------------+----------+--------+
+| table_catalog   | table_schema       | table_name    | table_type | table_id | engine |
++-----------------+--------------------+---------------+------------+----------+--------+
+| another_catalog | another_schema     | another_table | BASE TABLE | 1024     | mito   |
+| another_catalog | information_schema | tables        | VIEW       |          |        |
++-----------------+--------------------+---------------+------------+----------+--------+"
+        }
+    };
     check_output_stream(output, expected).await;
 }
 

src/frontend/src/tests/test_util.rs

@@ -26,18 +26,28 @@ use crate::tests::{
 
 pub(crate) trait MockInstance {
     fn frontend(&self) -> Arc<Instance>;
+
+    fn is_distributed_mode(&self) -> bool;
 }
 
 impl MockInstance for MockStandaloneInstance {
     fn frontend(&self) -> Arc<Instance> {
         self.instance.clone()
     }
+
+    fn is_distributed_mode(&self) -> bool {
+        false
+    }
 }
 
 impl MockInstance for MockDistributedInstance {
     fn frontend(&self) -> Arc<Instance> {
         self.frontend.clone()
     }
+
+    fn is_distributed_mode(&self) -> bool {
+        true
+    }
 }
 
 pub(crate) async fn standalone() -> Arc<dyn MockInstance> {

tests/cases/standalone/common/system/information_schema.result

@@ -27,18 +27,18 @@ order by table_name;
 | foo        |
 +------------+
 
-select table_catalog, table_schema, table_name, table_type
+select table_catalog, table_schema, table_name, table_type, engine
 from information_schema.tables
 where table_catalog = 'greptime'
   and table_schema != 'public'
 order by table_schema, table_name;
 
-+---------------+--------------------+------------+------------+
-| table_catalog | table_schema       | table_name | table_type |
-+---------------+--------------------+------------+------------+
-| greptime      | information_schema | tables     | VIEW       |
-| greptime      | my_db              | foo        | BASE TABLE |
-+---------------+--------------------+------------+------------+
++---------------+--------------------+------------+------------+--------+
+| table_catalog | table_schema       | table_name | table_type | engine |
++---------------+--------------------+------------+------------+--------+
+| greptime      | information_schema | tables     | VIEW       |        |
+| greptime      | my_db              | foo        | BASE TABLE | mito   |
++---------------+--------------------+------------+------------+--------+
 
 use
 public;

tests/cases/standalone/common/system/information_schema.sql

@@ -14,7 +14,7 @@ from information_schema.tables
 where table_schema = 'my_db'
 order by table_name;
 
-select table_catalog, table_schema, table_name, table_type
+select table_catalog, table_schema, table_name, table_type, engine
 from information_schema.tables
 where table_catalog = 'greptime'
   and table_schema != 'public'