feat: support explain decorrelated plan (#16681)

Author: xudong963

src/query/ast/src/ast/statements/explain.rs

@@ -29,6 +29,8 @@ pub enum ExplainKind {
     // `EXPLAIN RAW` and `EXPLAIN OPTIMIZED` will be deprecated in the future,
     // use explain options instead
     Raw,
+    // `EXPLAIN DECORRELATED` will show the plan after subquery decorrelation
+    Decorrelated,
     Optimized,
 
     Plan,

src/query/ast/src/ast/statements/statement.rs

@@ -405,6 +405,7 @@ impl Display for Statement {
                     ExplainKind::Fragments => write!(f, " FRAGMENTS")?,
                     ExplainKind::Raw => write!(f, " RAW")?,
                     ExplainKind::Optimized => write!(f, " Optimized")?,
+                    ExplainKind::Decorrelated => write!(f, " DECORRELATED")?,
                     ExplainKind::Plan => (),
                     ExplainKind::AnalyzePlan => write!(f, " ANALYZE")?,
                     ExplainKind::Join => write!(f, " JOIN")?,

src/query/ast/src/parser/statement.rs

@@ -56,7 +56,7 @@ pub enum CreateDatabaseOption {
 pub fn statement_body(i: Input) -> IResult<Statement> {
     let explain = map_res(
         rule! {
-            EXPLAIN ~ ( "(" ~ #comma_separated_list1(explain_option) ~ ")" )? ~ ( AST | SYNTAX | PIPELINE | JOIN | GRAPH | FRAGMENTS | RAW | OPTIMIZED | MEMO )? ~ #statement
+            EXPLAIN ~ ( "(" ~ #comma_separated_list1(explain_option) ~ ")" )? ~ ( AST | SYNTAX | PIPELINE | JOIN | GRAPH | FRAGMENTS | RAW | OPTIMIZED | MEMO | DECORRELATED)? ~ #statement
         },
         |(_, options, opt_kind, statement)| {
             Ok(Statement::Explain {
@@ -74,6 +74,7 @@ pub fn statement_body(i: Input) -> IResult<Statement> {
                     Some(TokenKind::FRAGMENTS) => ExplainKind::Fragments,
                     Some(TokenKind::RAW) => ExplainKind::Raw,
                     Some(TokenKind::OPTIMIZED) => ExplainKind::Optimized,
+                    Some(TokenKind::DECORRELATED) => ExplainKind::Decorrelated,
                     Some(TokenKind::MEMO) => ExplainKind::Memo("".to_string()),
                     Some(TokenKind::GRAPHICAL) => ExplainKind::Graphical,
                     None => ExplainKind::Plan,

src/query/ast/src/parser/token.rs

@@ -1014,6 +1014,8 @@ pub enum TokenKind {
     RAW,
     #[token("OPTIMIZED", ignore(ascii_case))]
     OPTIMIZED,
+    #[token("DECORRELATED", ignore(ascii_case))]
+    DECORRELATED,
     #[token("SCHEMA", ignore(ascii_case))]
     SCHEMA,
     #[token("SCHEMAS", ignore(ascii_case))]

src/query/service/src/interpreters/interpreter_explain.rs

@@ -90,7 +90,9 @@ impl Interpreter for ExplainInterpreter {
     #[async_backtrace::framed]
     async fn execute2(&self) -> Result<PipelineBuildResult> {
         let blocks = match &self.kind {
-            ExplainKind::Raw | ExplainKind::Optimized => self.explain_plan(&self.plan)?,
+            ExplainKind::Raw | ExplainKind::Optimized | ExplainKind::Decorrelated => {
+                self.explain_plan(&self.plan)?
+            }
             ExplainKind::Plan if self.config.logical => self.explain_plan(&self.plan)?,
             ExplainKind::Plan => match &self.plan {
                 Plan::Query {

src/query/sql/src/planner/optimizer/optimizer.rs

@@ -204,6 +204,41 @@ pub async fn optimize(mut opt_ctx: OptimizerContext, plan: Plan) -> Result<Plan>
             ExplainKind::Ast(_) | ExplainKind::Syntax(_) => {
                 Ok(Plan::Explain { config, kind, plan })
             }
+            ExplainKind::Decorrelated => {
+                if let Plan::Query {
+                    s_expr,
+                    metadata,
+                    bind_context,
+                    rewrite_kind,
+                    formatted_ast,
+                    ignore_result,
+                } = *plan
+                {
+                    let mut s_expr = s_expr;
+                    if s_expr.contain_subquery() {
+                        s_expr = Box::new(decorrelate_subquery(
+                            opt_ctx.metadata.clone(),
+                            *s_expr.clone(),
+                        )?);
+                    }
+                    Ok(Plan::Explain {
+                        kind,
+                        config,
+                        plan: Box::new(Plan::Query {
+                            s_expr,
+                            bind_context,
+                            metadata,
+                            rewrite_kind,
+                            formatted_ast,
+                            ignore_result,
+                        }),
+                    })
+                } else {
+                    Err(ErrorCode::BadArguments(
+                        "Cannot use EXPLAIN DECORRELATED with a non-query statement",
+                    ))
+                }
+            }
             ExplainKind::Memo(_) => {
                 if let box Plan::Query { ref s_expr, .. } = plan {
                     let memo = get_optimized_memo(opt_ctx, *s_expr.clone()).await?;

tests/sqllogictests/suites/mode/standalone/explain/subquery.test

@@ -553,5 +553,72 @@ EvalScalar
         ├── push downs: [filters: [], limit: NONE]
         └── estimated rows: 4.00
 
+query T
+explain optimized select i, exists(select * from t where i > 10) from t;
+----
+EvalScalar
+├── scalars: [t.i (#0) AS (#0), exists_scalar (#4) AS (#2)]
+└── Join(Cross)
+    ├── build keys: []
+    ├── probe keys: []
+    ├── other filters: []
+    ├── Scan
+    │   ├── table: default.t
+    │   ├── filters: []
+    │   ├── order by: []
+    │   └── limit: NONE
+    └── EvalScalar
+        ├── scalars: [eq(count(*) (#3), 1) AS (#4)]
+        └── Aggregate(Final)
+            ├── group items: []
+            ├── aggregate functions: [count(*) (#3)]
+            └── Aggregate(Partial)
+                ├── group items: []
+                ├── aggregate functions: [count(*) (#3)]
+                └── Limit
+                    ├── limit: [1]
+                    ├── offset: [0]
+                    └── Filter
+                        ├── filters: [gt(t.i (#1), 10)]
+                        └── Scan
+                            ├── table: default.t
+                            ├── filters: [gt(t.i (#1), 10)]
+                            ├── order by: []
+                            └── limit: NONE
+
+
+query T
+explain decorrelated select i, exists(select * from t where i > 10) from t;
+----
+EvalScalar
+├── scalars: [t.i (#0) AS (#0), exists_scalar (#4) AS (#2)]
+└── Join(Cross)
+    ├── build keys: []
+    ├── probe keys: []
+    ├── other filters: []
+    ├── Scan
+    │   ├── table: default.t
+    │   ├── filters: []
+    │   ├── order by: []
+    │   └── limit: NONE
+    └── EvalScalar
+        ├── scalars: [eq(count(*) (#3), 1) AS (#4)]
+        └── Aggregate(Initial)
+            ├── group items: []
+            ├── aggregate functions: [count(*) (#3)]
+            └── Limit
+                ├── limit: [1]
+                ├── offset: [0]
+                └── EvalScalar
+                    ├── scalars: [t.i (#1) AS (#1)]
+                    └── Filter
+                        ├── filters: [gt(t.i (#1), 10)]
+                        └── Scan
+                            ├── table: default.t
+                            ├── filters: []
+                            ├── order by: []
+                            └── limit: NONE
+
+
 statement ok
 drop table if exists t;