From 7eb25731786f5cc2ca794c2bac0350acc5473603 Mon Sep 17 00:00:00 2001
From: Boshen <boshenc@gmail.com>
Date: Mon, 8 Jan 2024 12:37:20 +0800
Subject: [PATCH] refactor(parser): parse BigInt lazily (#1924)

This PR partially fixes #1803 and is part of #1880.

BigInt is removed from the `Token` value, so that the token size can be
reduced once we removed all the variants.

`Token` is now also `Copy`, which removes all the `clone` and `drop`
calls.

This yields 5% performance improvement for the parser.
---
 crates/oxc_parser/src/cursor.rs        |  2 +-
 crates/oxc_parser/src/js/expression.rs | 11 ++++++-----
 crates/oxc_parser/src/lexer/mod.rs     | 19 +++++++++++--------
 crates/oxc_parser/src/lexer/token.rs   | 14 +++-----------
 4 files changed, 21 insertions(+), 25 deletions(-)

diff --git a/crates/oxc_parser/src/cursor.rs b/crates/oxc_parser/src/cursor.rs
index 02f377e200db0..ff4dbe798de14 100644
--- a/crates/oxc_parser/src/cursor.rs
+++ b/crates/oxc_parser/src/cursor.rs
@@ -236,7 +236,7 @@ impl<'a> Parser<'a> {
     pub(crate) fn checkpoint(&self) -> ParserCheckpoint<'a> {
         ParserCheckpoint {
             lexer: self.lexer.checkpoint(),
-            cur_token: self.token.clone(),
+            cur_token: self.token,
             prev_span_end: self.prev_token_end,
             errors_pos: self.errors.len(),
         }
diff --git a/crates/oxc_parser/src/js/expression.rs b/crates/oxc_parser/src/js/expression.rs
index b0dd4c0184fa1..72594a12e08c8 100644
--- a/crates/oxc_parser/src/js/expression.rs
+++ b/crates/oxc_parser/src/js/expression.rs
@@ -17,6 +17,7 @@ use super::{
 };
 use crate::{
     diagnostics,
+    lexer::parse_big_int,
     lexer::{Kind, TokenValue},
     list::SeparatedList,
     Context, Parser,
@@ -306,12 +307,12 @@ impl<'a> Parser<'a> {
             Kind::Hex => BigintBase::Hex,
             _ => return Err(self.unexpected()),
         };
-        let value = match self.cur_kind() {
-            kind if kind.is_number() => self.cur_token().value.as_bigint(),
-            _ => return Err(self.unexpected()),
-        };
+        let token = self.cur_token();
+        let src = self.cur_src().strip_suffix('n').unwrap();
+        let value = parse_big_int(src, token.kind)
+            .map_err(|err| diagnostics::InvalidNumber(err, token.span()))?;
         self.bump_any();
-        Ok(BigintLiteral { span: self.end_span(span), value, base })
+        Ok(self.ast.bigint_literal(self.end_span(span), value, base))
     }
 
     pub(crate) fn parse_literal_regexp(&mut self) -> Result<RegExpLiteral> {
diff --git a/crates/oxc_parser/src/lexer/mod.rs b/crates/oxc_parser/src/lexer/mod.rs
index 7933e6379a8d8..1b016f9101645 100644
--- a/crates/oxc_parser/src/lexer/mod.rs
+++ b/crates/oxc_parser/src/lexer/mod.rs
@@ -26,9 +26,9 @@ use oxc_syntax::{
 };
 pub use token::{RegExp, Token, TokenValue};
 
-pub use self::kind::Kind;
+pub use self::{kind::Kind, number::parse_big_int};
 use self::{
-    number::{parse_big_int, parse_float, parse_int},
+    number::{parse_float, parse_int},
     string_builder::AutoCow,
     trivia_builder::TriviaBuilder,
 };
@@ -105,7 +105,7 @@ impl<'a> Lexer<'a> {
     pub fn checkpoint(&self) -> LexerCheckpoint<'a> {
         LexerCheckpoint {
             chars: self.current.chars.clone(),
-            token: self.current.token.clone(),
+            token: self.current.token,
             errors_pos: self.errors.len(),
         }
     }
@@ -178,7 +178,9 @@ impl<'a> Lexer<'a> {
         self.current.token.kind = kind;
         self.current.token.end = self.offset();
         debug_assert!(self.current.token.start <= self.current.token.end);
-        std::mem::take(&mut self.current.token)
+        let token = self.current.token;
+        self.current.token = Token::default();
+        token
     }
 
     /// Re-tokenize the current `/` or `/=` and return `RegExp`
@@ -299,10 +301,11 @@ impl<'a> Lexer<'a> {
     fn set_numeric_value(&mut self, kind: Kind, src: &'a str) {
         let value = match kind {
             Kind::Decimal | Kind::Binary | Kind::Octal | Kind::Hex => {
-                src.strip_suffix('n').map_or_else(
-                    || parse_int(src, kind).map(TokenValue::Number),
-                    |src| parse_big_int(src, kind).map(TokenValue::BigInt),
-                )
+                if src.ends_with('n') {
+                    // BigInt is parsed lazily in the parser
+                    return;
+                }
+                parse_int(src, kind).map(TokenValue::Number)
             }
             Kind::Float | Kind::PositiveExponential | Kind::NegativeExponential => {
                 parse_float(src).map(TokenValue::Number)
diff --git a/crates/oxc_parser/src/lexer/token.rs b/crates/oxc_parser/src/lexer/token.rs
index f986e538a3e9b..b7ea930395794 100644
--- a/crates/oxc_parser/src/lexer/token.rs
+++ b/crates/oxc_parser/src/lexer/token.rs
@@ -5,7 +5,7 @@ use oxc_span::Span;
 
 use super::kind::Kind;
 
-#[derive(Debug, Clone, Default)]
+#[derive(Debug, Clone, Copy, Default)]
 pub struct Token<'a> {
     /// Token Kind
     pub kind: Kind,
@@ -38,16 +38,15 @@ impl<'a> Token<'a> {
     }
 }
 
-#[derive(Debug, Clone)]
+#[derive(Debug, Copy, Clone)]
 pub enum TokenValue<'a> {
     None,
     Number(f64),
-    BigInt(num_bigint::BigInt),
     String(&'a str),
     RegExp(RegExp<'a>),
 }
 
-#[derive(Debug, Clone)]
+#[derive(Debug, Copy, Clone)]
 pub struct RegExp<'a> {
     pub pattern: &'a str,
     pub flags: RegExpFlags,
@@ -67,13 +66,6 @@ impl<'a> TokenValue<'a> {
         }
     }
 
-    pub fn as_bigint(&self) -> num_bigint::BigInt {
-        match self {
-            Self::BigInt(s) => s.clone(),
-            _ => unreachable!("expected bigint!"),
-        }
-    }
-
     pub fn as_regex(&self) -> &RegExp<'a> {
         match self {
             Self::RegExp(regex) => regex,