diff --git a/src/executor.rs b/src/executor.rs
index 91348ac49f2e4757c018f01a49cc0facd472db02..e673374aeb121a6aaefbb502cb3b9195fac39404 100644
--- a/src/executor.rs
+++ b/src/executor.rs
@@ -53,7 +53,7 @@ pub mod vtable;
impl Plan {
/// Apply optimization rules to the plan.
- fn optimize(&mut self) -> Result<(), QueryPlannerError> {
+ pub(crate) fn optimize(&mut self) -> Result<(), QueryPlannerError> {
self.replace_in_operator()?;
self.split_columns()?;
self.set_dnf()?;
diff --git a/src/ir/explain.rs b/src/ir/explain.rs
index f0a2f26e52362ccb0147776db60c5dd23e85413f..13e9ee5fdc6e311a53ea8ff847bb1d1701f86aa6 100644
--- a/src/ir/explain.rs
+++ b/src/ir/explain.rs
@@ -7,6 +7,9 @@ use traversal::DftPost;
use crate::errors::QueryPlannerError;
use crate::ir::expression::Expression;
use crate::ir::operator::Relational;
+use crate::ir::transformation::redistribution::{
+ DataGeneration, MotionPolicy as IrMotionPolicy, Target as IrTarget,
+};
use crate::ir::Plan;
use super::operator::{Bool, Unary};
@@ -359,6 +362,78 @@ impl Display for SubQuery {
}
}
+#[derive(Debug, Serialize)]
+struct Motion {
+ policy: MotionPolicy,
+ generation: DataGeneration,
+}
+
+impl Motion {
+ fn new(policy: MotionPolicy, generation: DataGeneration) -> Self {
+ Motion { policy, generation }
+ }
+}
+
+impl Display for Motion {
+ fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+ write!(
+ f,
+ "motion [policy: {}, generation: {}]",
+ &self.policy, &self.generation
+ )
+ }
+}
+
+#[derive(Debug, Serialize)]
+enum MotionPolicy {
+ Full,
+ Segment(MotionKey),
+ Local,
+}
+
+impl Display for MotionPolicy {
+ fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+ match &self {
+ MotionPolicy::Full => write!(f, "full"),
+ MotionPolicy::Segment(mk) => write!(f, "segment({})", mk),
+ MotionPolicy::Local => write!(f, "local"),
+ }
+ }
+}
+
+#[derive(Debug, Serialize)]
+struct MotionKey {
+ pub targets: Vec<Target>,
+}
+
+impl Display for MotionKey {
+ fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+ let targets = &self
+ .targets
+ .iter()
+ .map(ToString::to_string)
+ .collect::<Vec<String>>()
+ .join(", ");
+
+ write!(f, "[{}]", targets)
+ }
+}
+
+#[derive(Debug, Serialize)]
+pub enum Target {
+ Reference(String),
+ Value(Value),
+}
+
+impl Display for Target {
+ fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+ match &self {
+ Target::Reference(s) => write!(f, "ref({})", s),
+ Target::Value(v) => write!(f, "value({})", v),
+ }
+ }
+}
+
#[derive(Debug, Serialize)]
#[allow(dead_code)]
enum ExplainNode {
@@ -368,6 +443,7 @@ enum ExplainNode {
Selection(Selection),
UnionAll,
SubQuery(SubQuery),
+ Motion(Motion),
}
impl Display for ExplainNode {
@@ -379,6 +455,7 @@ impl Display for ExplainNode {
ExplainNode::Selection(s) => format!("selection {}", s),
ExplainNode::UnionAll => "union all".to_string(),
ExplainNode::SubQuery(s) => s.to_string(),
+ ExplainNode::Motion(m) => m.to_string(),
};
write!(f, "{}", s)
@@ -551,8 +628,61 @@ impl FullExplain {
let s = SubQuery::new(alias.as_ref().map(ToString::to_string));
Some(ExplainNode::SubQuery(s))
}
+ Relational::Motion {
+ children,
+ policy,
+ generation,
+ ..
+ } => {
+ let child = stack.pop().ok_or_else(|| {
+ QueryPlannerError::CustomError(
+ "Motion node must have exactly one child".into(),
+ )
+ })?;
+ current_node.children.push(child);
+
+ let p = match policy {
+ IrMotionPolicy::Segment(s) => {
+ let child_id = children.first().ok_or_else(|| {
+ QueryPlannerError::CustomError(
+ "Current node should have exactly one child".to_string(),
+ )
+ })?;
+
+ let child_output_id = ir.get_relation_node(*child_id)?.output();
+ let col_list =
+ ir.get_expression_node(child_output_id)?.get_row_list()?;
+
+ let targets = (&s.targets)
+ .iter()
+ .map(|r| match r {
+ IrTarget::Reference(pos) => {
+ let col_id = *col_list.get(*pos).ok_or_else(|| {
+ QueryPlannerError::CustomError(String::from(
+ "Invalid position in list",
+ ))
+ })?;
+ let col_name = ir
+ .get_expression_node(col_id)?
+ .get_alias_name()?
+ .to_string();
+
+ Ok(Target::Reference(col_name))
+ }
+ IrTarget::Value(v) => Ok(Target::Value(v.clone())),
+ })
+ .collect::<Result<Vec<Target>, _>>()?;
+
+ MotionPolicy::Segment(MotionKey { targets })
+ }
+ IrMotionPolicy::Full => MotionPolicy::Full,
+ IrMotionPolicy::Local => MotionPolicy::Local,
+ };
+ let m = Motion::new(p, generation.clone());
+
+ Some(ExplainNode::Motion(m))
+ }
Relational::InnerJoin { .. }
- | Relational::Motion { .. }
| Relational::Insert { .. }
| Relational::Values { .. }
| Relational::ValuesRow { .. } => {
diff --git a/src/ir/explain/tests.rs b/src/ir/explain/tests.rs
index 9164969e505cd52b322a668b184c1d189fd96cc5..685a975c7ee04a150837aad6199785f76bfec6cf 100644
--- a/src/ir/explain/tests.rs
+++ b/src/ir/explain/tests.rs
@@ -1,14 +1,14 @@
use pretty_assertions::assert_eq;
use super::*;
-use crate::ir::transformation::helpers::sql_to_ir;
+use crate::ir::transformation::helpers::sql_to_optimized_ir;
#[test]
fn simple_query_without_cond_plan() {
let query =
r#"SELECT "t"."identification_number" as "c1", "product_code" FROM "hash_testing" as "t""#;
- let plan = sql_to_ir(query, vec![]);
+ let plan = sql_to_optimized_ir(query, vec![]);
let top = &plan.get_top().unwrap();
let explain_tree = FullExplain::new(&plan, *top).unwrap();
@@ -27,7 +27,7 @@ fn simple_query_without_cond_plan() {
fn simple_query_with_cond_plan() {
let query = r#"SELECT "t"."identification_number" as "c1", "product_code" FROM "hash_testing" as "t" WHERE "t"."identification_number" = 1 AND "t"."product_code" = '222'"#;
- let plan = sql_to_ir(query, vec![]);
+ let plan = sql_to_optimized_ir(query, vec![]);
let top = &plan.get_top().unwrap();
let explain_tree = FullExplain::new(&plan, *top).unwrap();
@@ -35,7 +35,7 @@ fn simple_query_with_cond_plan() {
let mut actual_explain = String::new();
actual_explain.push_str(
r#"projection ("t"."identification_number" -> "c1", "t"."product_code" -> "product_code")
- selection ROW("t"."identification_number") = ROW(1) and ROW("t"."product_code") = ROW('222')
+ selection ROW("t"."identification_number", "t"."product_code") = ROW(1, '222')
scan "hash_testing" -> "t"
"#,
);
@@ -49,7 +49,7 @@ fn union_query_plan() {
UNION ALL
SELECT "t2"."identification_number", "product_code" FROM "hash_testing_hist" as "t2""#;
- let plan = sql_to_ir(query, vec![]);
+ let plan = sql_to_optimized_ir(query, vec![]);
let top = &plan.get_top().unwrap();
let explain_tree = FullExplain::new(&plan, *top).unwrap();
@@ -74,7 +74,7 @@ SELECT "id", "FIRST_NAME" FROM "test_space_hist" WHERE "sys_op" < 0
) as "t"
WHERE "id" = 1"#;
- let plan = sql_to_ir(query, vec![]);
+ let plan = sql_to_optimized_ir(query, vec![]);
let top = &plan.get_top().unwrap();
let explain_tree = FullExplain::new(&plan, *top).unwrap();
@@ -85,7 +85,7 @@ WHERE "id" = 1"#;
scan "t"
union all
projection ("test_space"."id" -> "id", "test_space"."FIRST_NAME" -> "FIRST_NAME")
- selection ROW("test_space"."sys_op") > ROW(0) and ROW("test_space"."sysFrom") < ROW(0)
+ selection ROW("test_space"."sysFrom") < ROW(0) and ROW("test_space"."sys_op") > ROW(0)
scan "test_space"
projection ("test_space_hist"."id" -> "id", "test_space_hist"."FIRST_NAME" -> "FIRST_NAME")
selection ROW("test_space_hist"."sys_op") < ROW(0)
@@ -109,21 +109,20 @@ WHERE "id" IN (SELECT "id"
SELECT "id", "FIRST_NAME" FROM "test_space_hist" WHERE "sys_op" < 0
) as "t2"
WHERE "t2"."id" = 4)
- AND "FIRST_NAME" IN (SELECT "FIRST_NAME" FROM "test_space" WHERE "id" = 5)
"#;
- let plan = sql_to_ir(query, vec![]);
+ let plan = sql_to_optimized_ir(query, vec![]);
let top = &plan.get_top().unwrap();
let explain_tree = FullExplain::new(&plan, *top).unwrap();
let mut actual_explain = String::new();
actual_explain.push_str(r#"projection ("t"."id" -> "id", "t"."FIRST_NAME" -> "FIRST_NAME")
- selection ROW("t"."id") in ROW($0) and ROW("t"."FIRST_NAME") in ROW($1)
+ selection ROW("t"."id") in ROW($0)
scan "t"
union all
projection ("test_space"."id" -> "id", "test_space"."FIRST_NAME" -> "FIRST_NAME")
- selection ROW("test_space"."sys_op") > ROW(0) and ROW("test_space"."sysFrom") < ROW(0)
+ selection ROW("test_space"."sysFrom") < ROW(0) and ROW("test_space"."sys_op") > ROW(0)
scan "test_space"
projection ("test_space_hist"."id" -> "id", "test_space_hist"."FIRST_NAME" -> "FIRST_NAME")
selection ROW("test_space_hist"."sys_op") < ROW(0)
@@ -140,11 +139,6 @@ scan
projection ("test_space_hist"."id" -> "id", "test_space_hist"."FIRST_NAME" -> "FIRST_NAME")
selection ROW("test_space_hist"."sys_op") < ROW(0)
scan "test_space_hist"
-subquery $1:
-scan
- projection ("test_space"."FIRST_NAME" -> "FIRST_NAME")
- selection ROW("test_space"."id") = ROW(5)
- scan "test_space"
"#);
assert_eq!(actual_explain, explain_tree.to_string())
@@ -156,17 +150,74 @@ fn explain_except1() {
EXCEPT DISTINCT
SELECT "identification_number" FROM "hash_testing_hist""#;
- let plan = sql_to_ir(query, vec![]);
+ let plan = sql_to_optimized_ir(query, vec![]);
let top = &plan.get_top().unwrap();
let explain_tree = FullExplain::new(&plan, *top).unwrap();
let expected = format!(
- "{}\n{}\n{}\n{}\n{}\n",
+ "{}\n{}\n{}\n{}\n{}\n{}\n",
r#"except"#,
r#" projection ("t"."product_code" -> "pc")"#,
r#" scan "hash_testing" -> "t""#,
- r#" projection ("hash_testing_hist"."identification_number" -> "identification_number")"#,
- r#" scan "hash_testing_hist""#,
+ r#" motion [policy: full, generation: none]"#,
+ r#" projection ("hash_testing_hist"."identification_number" -> "identification_number")"#,
+ r#" scan "hash_testing_hist""#,
);
assert_eq!(expected, explain_tree.to_string())
}
+
+#[test]
+fn motion_subquery_plan() {
+ let query = r#"SELECT * FROM (
+SELECT "id", "FIRST_NAME" FROM "test_space" WHERE "sys_op" > 0 and "sysFrom" < 0
+UNION ALL
+SELECT "id", "FIRST_NAME" FROM "test_space_hist" WHERE "sys_op" < 0
+) as "t"
+WHERE "id" IN (SELECT "id"
+ FROM (
+ SELECT "id", "FIRST_NAME" FROM "test_space" WHERE "sys_op" > 0
+ UNION ALL
+ SELECT "id", "FIRST_NAME" FROM "test_space_hist" WHERE "sys_op" < 0
+ ) as "t2"
+ WHERE "t2"."id" = 4)
+ OR "id" IN (SELECT "identification_number" FROM "hash_testing" WHERE "identification_number" = 5 AND "product_code" = '123')
+"#;
+
+ let plan = sql_to_optimized_ir(query, vec![]);
+
+ let top = &plan.get_top().unwrap();
+ let explain_tree = FullExplain::new(&plan, *top).unwrap();
+
+ let mut actual_explain = String::new();
+ actual_explain.push_str(r#"projection ("t"."id" -> "id", "t"."FIRST_NAME" -> "FIRST_NAME")
+ selection ROW("t"."id") in ROW($0) or ROW("t"."id") in ROW($1)
+ scan "t"
+ union all
+ projection ("test_space"."id" -> "id", "test_space"."FIRST_NAME" -> "FIRST_NAME")
+ selection ROW("test_space"."sysFrom") < ROW(0) and ROW("test_space"."sys_op") > ROW(0)
+ scan "test_space"
+ projection ("test_space_hist"."id" -> "id", "test_space_hist"."FIRST_NAME" -> "FIRST_NAME")
+ selection ROW("test_space_hist"."sys_op") < ROW(0)
+ scan "test_space_hist"
+subquery $0:
+scan
+ projection ("t2"."id" -> "id")
+ selection ROW("t2"."id") = ROW(4)
+ scan "t2"
+ union all
+ projection ("test_space"."id" -> "id", "test_space"."FIRST_NAME" -> "FIRST_NAME")
+ selection ROW("test_space"."sys_op") > ROW(0)
+ scan "test_space"
+ projection ("test_space_hist"."id" -> "id", "test_space_hist"."FIRST_NAME" -> "FIRST_NAME")
+ selection ROW("test_space_hist"."sys_op") < ROW(0)
+ scan "test_space_hist"
+subquery $1:
+motion [policy: segment([ref("identification_number")]), generation: none]
+ scan
+ projection ("hash_testing"."identification_number" -> "identification_number")
+ selection ROW("hash_testing"."identification_number", "hash_testing"."product_code") = ROW(5, '123')
+ scan "hash_testing"
+"#);
+
+ assert_eq!(actual_explain, explain_tree.to_string())
+}
diff --git a/src/ir/transformation/helpers.rs b/src/ir/transformation/helpers.rs
index 89f101e9cdbb93079939f127e03b2455a90c15d2..b4d009282fb66ce0534c43473efbe1e3e0d05a3b 100644
--- a/src/ir/transformation/helpers.rs
+++ b/src/ir/transformation/helpers.rs
@@ -10,6 +10,14 @@ use crate::frontend::Ast;
use crate::ir::value::Value;
use crate::ir::Plan;
+/// Compiles an SQL query to optimized IR plan.
+#[allow(dead_code)]
+pub fn sql_to_optimized_ir(query: &str, params: Vec<Value>) -> Plan {
+ let mut plan = sql_to_ir(query, params);
+ plan.optimize().unwrap();
+ plan
+}
+
/// Compiles an SQL query to IR plan.
#[allow(dead_code)]
pub fn sql_to_ir(query: &str, params: Vec<Value>) -> Plan {
diff --git a/src/ir/transformation/redistribution.rs b/src/ir/transformation/redistribution.rs
index 58d53f0efd8e0ba778e779b206c3bcae6a062fee..f1b3010fd2915519dafe15e42ca6934c199ac0b2 100644
--- a/src/ir/transformation/redistribution.rs
+++ b/src/ir/transformation/redistribution.rs
@@ -3,6 +3,7 @@
use ahash::RandomState;
use std::cmp::Ordering;
use std::collections::{hash_map::Entry, HashMap, HashSet};
+use std::fmt::{Display, Formatter};
use serde::{Deserialize, Serialize};
use traversal::{Bft, DftPost};
@@ -79,6 +80,17 @@ pub enum DataGeneration {
ShardingColumn,
}
+impl Display for DataGeneration {
+ fn fmt(&self, f: &mut Formatter) -> std::fmt::Result {
+ let op = match &self {
+ DataGeneration::None => "none",
+ DataGeneration::ShardingColumn => "sharding_column",
+ };
+
+ write!(f, "{}", op)
+ }
+}
+
struct BoolOp {
left: usize,
op: Bool,
diff --git a/test_app/test/integration/api_test.lua b/test_app/test/integration/api_test.lua
index a93b6d33a485bf042d2b202a8ed423407d6a7fb5..080fc950911f640efc2812816ff0a6f9d95587d7 100644
--- a/test_app/test/integration/api_test.lua
+++ b/test_app/test/integration/api_test.lua
@@ -777,13 +777,26 @@ g.test_bucket_id_in_join = function()
})
end
-g.test_invalid_explain = function()
+g.test_motion_explain = function()
local api = cluster:server("api-1").net_box
- local _, err = api:call("sbroad.explain", { [[SELECT "id", "name" FROM "testing_space"
+ local r, err = api:call("sbroad.explain", { [[SELECT "id", "name" FROM "testing_space"
WHERE "id" in (SELECT "id" FROM "space_simple_shard_key_hist" WHERE "sysOp" < 0)]] })
-
- t.assert_str_contains(tostring(err), "Explain hasn't supported node Motion")
+ t.assert_equals(err, nil)
+ t.assert_equals(
+ r,
+ {
+ "projection (\"testing_space\".\"id\" -> \"id\", \"testing_space\".\"name\" -> \"name\")",
+ " selection ROW(\"testing_space\".\"id\") in ROW($0)",
+ " scan \"testing_space\"",
+ "subquery $0:",
+ "motion [policy: full, generation: none]",
+ " scan",
+ " projection (\"space_simple_shard_key_hist\".\"id\" -> \"id\")",
+ " selection ROW(\"space_simple_shard_key_hist\".\"sysOp\") < ROW(0)",
+ " scan \"space_simple_shard_key_hist\"",
+ }
+ )
end
g.test_valid_explain = function()