feat(cubesql): Top-down extractor for rewrites

Author: MazterQyou

packages/cubejs-backend-native/Cargo.lock

@@ -3217,7 +3217,7 @@ export class BaseQuery {
           '{% if filter %}\nWHERE {{ filter }}{% endif %}' +
           '{% if group_by %}\nGROUP BY {{ group_by }}{% endif %}' +
           '{% if order_by %}\nORDER BY {{ order_by | map(attribute=\'expr\') | join(\', \') }}{% endif %}' +
-          '{% if limit %}\nLIMIT {{ limit }}{% endif %}' +
+          '{% if limit is not none %}\nLIMIT {{ limit }}{% endif %}' +
           '{% if offset %}\nOFFSET {{ offset }}{% endif %}',
         group_by_exprs: '{{ group_by | map(attribute=\'index\') | join(\', \') }}',
       },

packages/cubejs-schema-compiler/src/adapter/BaseQuery.js

@@ -117,7 +117,7 @@ export class PrestodbQuery extends BaseQuery {
       '{% if group_by %} GROUP BY {{ group_by }}{% endif %}' +
       '{% if order_by %} ORDER BY {{ order_by | map(attribute=\'expr\') | join(\', \') }}{% endif %}' +
       '{% if offset %}\nOFFSET {{ offset }}{% endif %}' +
-      '{% if limit %}\nLIMIT {{ limit }}{% endif %}';
+      '{% if limit is not none %}\nLIMIT {{ limit }}{% endif %}';
     templates.expressions.extract = 'EXTRACT({{ date_part }} FROM {{ expr }})';
     templates.expressions.interval_single_date_part = 'INTERVAL \'{{ num }}\' {{ date_part }}';
     templates.expressions.timestamp_literal = 'from_iso8601_timestamp(\'{{ value }}\')';

packages/cubejs-schema-compiler/src/adapter/PrestodbQuery.ts

@@ -58,6 +58,7 @@ minijinja = { version = "1", features = ["json", "loader"] }
 lru = "0.12.1"
 sha2 = "0.10.8"
 bigdecimal = "0.4.2"
+indexmap = "1.9.3"
 
 
 [dev-dependencies]

rust/cubesql/Cargo.lock

@@ -14389,7 +14389,11 @@ ORDER BY "source"."str0" ASC
                 .wrapped_sql
                 .unwrap()
                 .sql;
-            assert!(sql.contains("\"limit\":1000"));
+            if Rewriter::top_down_extractor_enabled() {
+                assert!(sql.contains("LIMIT 1000"));
+            } else {
+                assert!(sql.contains("\"limit\":1000"));
+            }
             assert!(sql.contains("% 7"));
 
             let physical_plan = query_plan.as_physical_plan().await.unwrap();
@@ -15456,7 +15460,7 @@ ORDER BY "source"."str0" ASC
 FROM ({{ from }}) AS {{ from_alias }}
 {% if group_by %} GROUP BY {{ group_by | map(attribute='index') | join(', ') }}{% endif %}
 {% if order_by %} ORDER BY {{ order_by | map(attribute='expr') | join(', ') }}{% endif %}{% if offset %}
-OFFSET {{ offset }}{% endif %}{% if limit %}
+OFFSET {{ offset }}{% endif %}{% if limit is not none %}
 LIMIT {{ limit }}{% endif %}"#.to_string(),
                 ),
             ]
@@ -16567,7 +16571,7 @@ LIMIT {{ limit }}{% endif %}"#.to_string(),
                 measure(count) AS cnt,
                 date_trunc('month', order_date) AS dt
             FROM KibanaSampleDataEcommerce
-            WHERE order_date IN (to_timestamp('2019-01-01 00:00:00.000000', 'YYYY-MM-DD HH24:MI:SS.US'))
+            WHERE date_trunc('month', order_date) IN (to_timestamp('2019-01-01 00:00:00.000000', 'YYYY-MM-DD HH24:MI:SS.US'))
             GROUP BY 2
             ;"#
             .to_string(),
@@ -16585,10 +16589,18 @@ LIMIT {{ limit }}{% endif %}"#.to_string(),
                 time_dimensions: Some(vec![V1LoadRequestQueryTimeDimension {
                     dimension: "KibanaSampleDataEcommerce.order_date".to_string(),
                     granularity: Some("month".to_string()),
-                    date_range: Some(json!(vec![
-                        "2019-01-01T00:00:00.000Z".to_string(),
-                        "2019-01-01T00:00:00.000Z".to_string()
-                    ]))
+                    date_range: if Rewriter::top_down_extractor_enabled() {
+                        Some(json!(vec![
+                            "2019-01-01T00:00:00.000Z".to_string(),
+                            "2019-01-31T23:59:59.999Z".to_string()
+                        ]))
+                    } else {
+                        // Non-optimal variant with top down extractor disabled
+                        Some(json!(vec![
+                            "2019-01-01 00:00:00.000".to_string(),
+                            "2019-01-31 23:59:59.999".to_string()
+                        ]))
+                    }
                 }]),
                 order: Some(vec![]),
                 limit: None,
@@ -16935,4 +16947,35 @@ LIMIT {{ limit }}{% endif %}"#.to_string(),
 
         Ok(())
     }
+
+    #[tokio::test]
+    async fn test_wrapper_limit_zero() {
+        if !Rewriter::sql_push_down_enabled() {
+            return;
+        }
+        init_testing_logger();
+
+        let query_plan = convert_select_to_query_plan(
+            r#"
+            SELECT MAX(order_date) FROM KibanaSampleDataEcommerce LIMIT 0
+            "#
+            .to_string(),
+            DatabaseProtocol::PostgreSQL,
+        )
+        .await;
+
+        let logical_plan = query_plan.as_logical_plan();
+        let sql = logical_plan
+            .find_cube_scan_wrapper()
+            .wrapped_sql
+            .unwrap()
+            .sql;
+        assert!(sql.contains("LIMIT 0"));
+
+        let physical_plan = query_plan.as_physical_plan().await.unwrap();
+        println!(
+            "Physical plan: {}",
+            displayable(physical_plan.as_ref()).indent()
+        );
+    }
 }

rust/cubesql/cubesql/Cargo.toml

@@ -186,7 +186,13 @@ pub trait QueryEngine {
         let mut rewriter = Rewriter::new(finalized_graph, cube_ctx.clone());
 
         let result = rewriter
-            .find_best_plan(root, state.auth_context().unwrap(), qtrace, span_id.clone())
+            .find_best_plan(
+                root,
+                state.auth_context().unwrap(),
+                qtrace,
+                span_id.clone(),
+                self.config_ref().top_down_extractor(),
+            )
             .await
             .map_err(|e| match e.cause {
                 CubeErrorCauseType::Internal(_) => CompilationError::Internal(