Perf/serialize (#255)

* perf: use `byteorder` on `DataValue::to_raw` & `DataValue::from_raw`

* perf: Use `bumpalo` to control memory allocation of `Sort` and `TableCodec`

* chore: simplify `Tuple::deserialize_from` & `ScalaExpression::eval` & `HepGraph::node_iter` & `HepOptimizer::apply_batch`

* perf: encode the tablename prefix in `TableCodec` into a hash, enabling it to apply rocksdb prefix range

* chore: simplify parameter on execute sql

Cargo.toml

@@ -2,7 +2,7 @@
 
 [package]
 name          = "fnck_sql"
-version       = "0.0.8"
+version       = "0.0.9"
 edition       = "2021"
 authors       = ["Kould <[email protected]>", "Xwg <[email protected]>"]
 description   = "SQL as a Function for Rust"
@@ -35,12 +35,13 @@ harness = false
 [dependencies]
 ahash                 = { version = "0.8" }
 bincode               = { version = "1" }
-bytes                 = { version = "1" }
+bumpalo               = { version = "3", features = ["allocator-api2", "collections", "std"] }
+byteorder             = { version = "1" }
 chrono                = { version = "0.4" }
 comfy-table           = { version = "7" }
 csv                   = { version = "1" }
-encode_unicode        = { version = "1" }
 dirs                  = { version = "5" }
+fixedbitset           = { version = "0.4" }
 itertools             = { version = "0.12" }
 ordered-float         = { version = "4" }
 paste                 = { version = "1" }
@@ -68,10 +69,8 @@ tokio                 = { version = "1.36", features = ["full"], optional = true
 
 
 [dev-dependencies]
-cargo-tarpaulin       = { version = "0.27" }
 criterion             = { version = "0.5", features = ["html_reports"] }
 indicatif             = { version = "0.17" }
-rand_distr            = { version = "0.4" }
 tempfile              = { version = "3.10" }
 # Benchmark
 sqlite                = { version = "0.34" }

README.md

@@ -73,13 +73,13 @@ run `cargo run -p tpcc --release` to run tpcc
 - Tips: TPC-C currently only supports single thread
 ```shell
 <90th Percentile RT (MaxRT)>
-   New-Order : 0.003  (0.012)
-     Payment : 0.001  (0.003)
-Order-Status : 0.054  (0.188)
-    Delivery : 0.021  (0.049)
- Stock-Level : 0.004  (0.006)
+   New-Order : 0.002  (0.004)
+     Payment : 0.001  (0.025)
+Order-Status : 0.053  (0.175)
+    Delivery : 0.022  (0.027)
+ Stock-Level : 0.003  (0.019)
 <TpmC>
-7345 Tpmc
+7815 tpmC
 ```
 #### 👉[check more](tpcc/README.md)
 

src/binder/aggregate.rs

@@ -3,18 +3,18 @@ use itertools::Itertools;
 use sqlparser::ast::{Expr, OrderByExpr};
 use std::collections::HashSet;
 
+use super::{Binder, QueryBindStep};
 use crate::errors::DatabaseError;
 use crate::expression::function::scala::ScalarFunction;
 use crate::planner::LogicalPlan;
 use crate::storage::Transaction;
+use crate::types::value::DataValue;
 use crate::{
     expression::ScalarExpression,
     planner::operator::{aggregate::AggregateOperator, sort::SortField},
 };
 
-use super::{Binder, QueryBindStep};
-
-impl<T: Transaction> Binder<'_, '_, T> {
+impl<T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'_, '_, T, A> {
     pub fn bind_aggregate(
         &mut self,
         children: LogicalPlan,

src/binder/alter_table.rs

@@ -11,8 +11,9 @@ use crate::planner::operator::table_scan::TableScanOperator;
 use crate::planner::operator::Operator;
 use crate::planner::{Childrens, LogicalPlan};
 use crate::storage::Transaction;
+use crate::types::value::DataValue;
 
-impl<T: Transaction> Binder<'_, '_, T> {
+impl<T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'_, '_, T, A> {
     pub(crate) fn bind_alter_table(
         &mut self,
         name: &ObjectName,

src/binder/analyze.rs

@@ -5,10 +5,11 @@ use crate::planner::operator::table_scan::TableScanOperator;
 use crate::planner::operator::Operator;
 use crate::planner::{Childrens, LogicalPlan};
 use crate::storage::Transaction;
+use crate::types::value::DataValue;
 use sqlparser::ast::ObjectName;
 use std::sync::Arc;
 
-impl<T: Transaction> Binder<'_, '_, T> {
+impl<T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'_, '_, T, A> {
     pub(crate) fn bind_analyze(&mut self, name: &ObjectName) -> Result<LogicalPlan, DatabaseError> {
         let table_name = Arc::new(lower_case_name(name)?);
 

src/binder/copy.rs

@@ -6,6 +6,7 @@ use super::*;
 use crate::errors::DatabaseError;
 use crate::planner::operator::copy_from_file::CopyFromFileOperator;
 use crate::planner::operator::copy_to_file::CopyToFileOperator;
+use crate::planner::operator::table_scan::TableScanOperator;
 use crate::planner::operator::Operator;
 use crate::planner::Childrens;
 use fnck_sql_serde_macros::ReferenceSerialization;
@@ -63,7 +64,7 @@ impl FromStr for ExtSource {
     }
 }
 
-impl<T: Transaction> Binder<'_, '_, T> {
+impl<T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'_, '_, T, A> {
     pub(super) fn bind_copy(
         &mut self,
         source: CopySource,
@@ -96,11 +97,10 @@ impl<T: Transaction> Binder<'_, '_, T> {
                 // COPY <source_table> TO <dest_file>
                 Ok(LogicalPlan::new(
                     Operator::CopyToFile(CopyToFileOperator {
-                        table: table.name.to_string(),
                         target: ext_source,
                         schema_ref,
                     }),
-                    Childrens::None,
+                    Childrens::Only(TableScanOperator::build(table_name, table)),
                 ))
             } else {
                 // COPY <dest_table> FROM <source_file>

src/binder/create_index.rs

@@ -7,10 +7,11 @@ use crate::planner::operator::Operator;
 use crate::planner::{Childrens, LogicalPlan};
 use crate::storage::Transaction;
 use crate::types::index::IndexType;
+use crate::types::value::DataValue;
 use sqlparser::ast::{ObjectName, OrderByExpr};
 use std::sync::Arc;
 
-impl<T: Transaction> Binder<'_, '_, T> {
+impl<T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'_, '_, T, A> {
     pub(crate) fn bind_create_index(
         &mut self,
         table_name: &ObjectName,

src/binder/create_table.rs

@@ -12,9 +12,10 @@ use crate::planner::operator::create_table::CreateTableOperator;
 use crate::planner::operator::Operator;
 use crate::planner::{Childrens, LogicalPlan};
 use crate::storage::Transaction;
+use crate::types::value::DataValue;
 use crate::types::LogicalType;
 
-impl<T: Transaction> Binder<'_, '_, T> {
+impl<T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'_, '_, T, A> {
     // TODO: TableConstraint
     pub(crate) fn bind_create_table(
         &mut self,
@@ -157,7 +158,6 @@ mod tests {
     use crate::types::LogicalType;
     use crate::utils::lru::SharedLruCache;
     use sqlparser::ast::CharLengthUnits;
-    use std::cell::RefCell;
     use std::hash::RandomState;
     use std::sync::atomic::AtomicUsize;
     use tempfile::TempDir;
@@ -173,7 +173,6 @@ mod tests {
         let table_functions = Default::default();
 
         let sql = "create table t1 (id int primary key, name varchar(10) null)";
-        let args = RefCell::new(Vec::new());
         let mut binder = Binder::new(
             BinderContext::new(
                 &table_cache,
@@ -183,7 +182,7 @@ mod tests {
                 &table_functions,
                 Arc::new(AtomicUsize::new(0)),
             ),
-            &args,
+            &[],
             None,
         );
         let stmt = crate::parser::parse_sql(sql).unwrap();

src/binder/create_view.rs

@@ -7,12 +7,13 @@ use crate::planner::operator::create_view::CreateViewOperator;
 use crate::planner::operator::Operator;
 use crate::planner::{Childrens, LogicalPlan};
 use crate::storage::Transaction;
+use crate::types::value::DataValue;
 use itertools::Itertools;
 use sqlparser::ast::{Ident, ObjectName, Query};
 use std::sync::Arc;
 use ulid::Ulid;
 
-impl<T: Transaction> Binder<'_, '_, T> {
+impl<T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'_, '_, T, A> {
     pub(crate) fn bind_create_view(
         &mut self,
         or_replace: &bool,

src/binder/delete.rs

@@ -5,11 +5,12 @@ use crate::planner::operator::table_scan::TableScanOperator;
 use crate::planner::operator::Operator;
 use crate::planner::{Childrens, LogicalPlan};
 use crate::storage::Transaction;
+use crate::types::value::DataValue;
 use itertools::Itertools;
 use sqlparser::ast::{Expr, TableAlias, TableFactor, TableWithJoins};
 use std::sync::Arc;
 
-impl<T: Transaction> Binder<'_, '_, T> {
+impl<T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'_, '_, T, A> {
     pub(crate) fn bind_delete(
         &mut self,
         from: &TableWithJoins,

src/binder/describe.rs

@@ -4,10 +4,11 @@ use crate::planner::operator::describe::DescribeOperator;
 use crate::planner::operator::Operator;
 use crate::planner::{Childrens, LogicalPlan};
 use crate::storage::Transaction;
+use crate::types::value::DataValue;
 use sqlparser::ast::ObjectName;
 use std::sync::Arc;
 
-impl<T: Transaction> Binder<'_, '_, T> {
+impl<T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'_, '_, T, A> {
     pub(crate) fn bind_describe(
         &mut self,
         name: &ObjectName,

src/binder/distinct.rs

@@ -3,8 +3,9 @@ use crate::expression::ScalarExpression;
 use crate::planner::operator::aggregate::AggregateOperator;
 use crate::planner::LogicalPlan;
 use crate::storage::Transaction;
+use crate::types::value::DataValue;
 
-impl<T: Transaction> Binder<'_, '_, T> {
+impl<T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'_, '_, T, A> {
     pub fn bind_distinct(
         &mut self,
         children: LogicalPlan,

src/binder/drop_table.rs

@@ -4,10 +4,11 @@ use crate::planner::operator::drop_table::DropTableOperator;
 use crate::planner::operator::Operator;
 use crate::planner::{Childrens, LogicalPlan};
 use crate::storage::Transaction;
+use crate::types::value::DataValue;
 use sqlparser::ast::ObjectName;
 use std::sync::Arc;
 
-impl<T: Transaction> Binder<'_, '_, T> {
+impl<T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'_, '_, T, A> {
     pub(crate) fn bind_drop_table(
         &mut self,
         name: &ObjectName,

src/binder/drop_view.rs

@@ -4,10 +4,11 @@ use crate::planner::operator::drop_view::DropViewOperator;
 use crate::planner::operator::Operator;
 use crate::planner::{Childrens, LogicalPlan};
 use crate::storage::Transaction;
+use crate::types::value::DataValue;
 use sqlparser::ast::ObjectName;
 use std::sync::Arc;
 
-impl<T: Transaction> Binder<'_, '_, T> {
+impl<T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'_, '_, T, A> {
     pub(crate) fn bind_drop_view(
         &mut self,
         name: &ObjectName,

src/binder/explain.rs

@@ -3,8 +3,9 @@ use crate::errors::DatabaseError;
 use crate::planner::operator::Operator;
 use crate::planner::{Childrens, LogicalPlan};
 use crate::storage::Transaction;
+use crate::types::value::DataValue;
 
-impl<T: Transaction> Binder<'_, '_, T> {
+impl<T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'_, '_, T, A> {
     pub(crate) fn bind_explain(&mut self, plan: LogicalPlan) -> Result<LogicalPlan, DatabaseError> {
         Ok(LogicalPlan::new(Operator::Explain, Childrens::Only(plan)))
     }

src/binder/expr.rs

@@ -40,7 +40,7 @@ macro_rules! try_default {
     };
 }
 
-impl<'a, T: Transaction> Binder<'a, '_, T> {
+impl<'a, T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'a, '_, T, A> {
     pub(crate) fn bind_expr(&mut self, expr: &Expr) -> Result<ScalarExpression, DatabaseError> {
         match expr {
             Expr::Identifier(ident) => {
@@ -50,14 +50,11 @@ impl<'a, T: Transaction> Binder<'a, '_, T> {
             Expr::BinaryOp { left, right, op } => self.bind_binary_op_internal(left, right, op),
             Expr::Value(v) => {
                 let value = if let Value::Placeholder(name) = v {
-                    let (i, _) = self
-                        .args
-                        .borrow()
+                    self.args
+                        .as_ref()
                         .iter()
-                        .enumerate()
-                        .find(|(_, (key, _))| key == name)
-                        .ok_or_else(|| DatabaseError::ParametersNotFound(name.to_string()))?;
-                    self.args.borrow_mut().remove(i).1
+                        .find_map(|(key, value)| (key == name).then(|| value.clone()))
+                        .ok_or_else(|| DatabaseError::ParametersNotFound(name.to_string()))?
                 } else {
                     v.into()
                 };

src/binder/insert.rs

@@ -12,7 +12,7 @@ use sqlparser::ast::{Expr, Ident, ObjectName};
 use std::slice;
 use std::sync::Arc;
 
-impl<T: Transaction> Binder<'_, '_, T> {
+impl<T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'_, '_, T, A> {
     pub(crate) fn bind_insert(
         &mut self,
         name: &ObjectName,

src/binder/mod.rs

@@ -19,20 +19,20 @@ mod truncate;
 mod update;
 
 use sqlparser::ast::{Ident, ObjectName, ObjectType, SetExpr, Statement};
-use std::cell::RefCell;
 use std::collections::{BTreeMap, HashMap, HashSet};
 use std::sync::atomic::{AtomicUsize, Ordering};
 use std::sync::Arc;
 
 use crate::catalog::view::View;
 use crate::catalog::{ColumnRef, TableCatalog, TableName};
-use crate::db::{Args, ScalaFunctions, TableFunctions};
+use crate::db::{ScalaFunctions, TableFunctions};
 use crate::errors::DatabaseError;
 use crate::expression::ScalarExpression;
 use crate::planner::operator::join::JoinType;
 use crate::planner::{LogicalPlan, SchemaOutput};
 use crate::storage::{TableCache, Transaction, ViewCache};
 use crate::types::tuple::SchemaRef;
+use crate::types::value::DataValue;
 
 pub enum InputRefType {
     AggCall,
@@ -313,18 +313,18 @@ impl<'a, T: Transaction> BinderContext<'a, T> {
     }
 }
 
-pub struct Binder<'a, 'b, T: Transaction> {
+pub struct Binder<'a, 'b, T: Transaction, A: AsRef<[(&'static str, DataValue)]>> {
     context: BinderContext<'a, T>,
     table_schema_buf: HashMap<TableName, Option<SchemaOutput>>,
-    args: &'a RefCell<Args>,
-    pub(crate) parent: Option<&'b Binder<'a, 'b, T>>,
+    args: &'a A,
+    pub(crate) parent: Option<&'b Binder<'a, 'b, T, A>>,
 }
 
-impl<'a, 'b, T: Transaction> Binder<'a, 'b, T> {
+impl<'a, 'b, T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'a, 'b, T, A> {
     pub fn new(
         context: BinderContext<'a, T>,
-        args: &'a RefCell<Args>,
-        parent: Option<&'b Binder<'a, 'b, T>>,
+        args: &'a A,
+        parent: Option<&'b Binder<'a, 'b, T, A>>,
     ) -> Self {
         Binder {
             context,
@@ -488,7 +488,6 @@ pub mod test {
     use crate::types::ColumnId;
     use crate::types::LogicalType::Integer;
     use crate::utils::lru::SharedLruCache;
-    use std::cell::RefCell;
     use std::hash::RandomState;
     use std::path::PathBuf;
     use std::sync::atomic::AtomicUsize;
@@ -507,7 +506,6 @@ pub mod test {
             let scala_functions = Default::default();
             let table_functions = Default::default();
             let transaction = self.storage.transaction()?;
-            let args = RefCell::new(Vec::new());
             let mut binder = Binder::new(
                 BinderContext::new(
                     &self.table_cache,
@@ -517,7 +515,7 @@ pub mod test {
                     &table_functions,
                     Arc::new(AtomicUsize::new(0)),
                 ),
-                &args,
+                &[],
                 None,
             );
             let stmt = crate::parser::parse_sql(sql)?;

src/binder/select.rs

@@ -38,7 +38,7 @@ use sqlparser::ast::{
     TableWithJoins,
 };
 
-impl<'a: 'b, 'b, T: Transaction> Binder<'a, 'b, T> {
+impl<'a: 'b, 'b, T: Transaction, A: AsRef<[(&'static str, DataValue)]>> Binder<'a, 'b, T, A> {
     pub(crate) fn bind_query(&mut self, query: &Query) -> Result<LogicalPlan, DatabaseError> {
         let origin_step = self.context.step_now();