feat(es/ast): Add raw to BigInt (#4218)

crates/swc_ecma_ast/src/lit.rs

@@ -62,11 +62,22 @@ bridge_lit_from!(Number, usize);
 bridge_lit_from!(BigInt, BigIntValue);
 
 #[ast_node("BigIntLiteral")]
-#[derive(Eq, Hash, EqIgnoreSpan)]
+#[derive(Eq, Hash)]
 pub struct BigInt {
     pub span: Span,
     #[cfg_attr(feature = "rkyv", with(EncodeBigInt))]
     pub value: BigIntValue,
+
+    /// Use `None` value only for transformations to avoid recalculate
+    /// characters in big integer
+    #[cfg_attr(feature = "rkyv", with(crate::EncodeJsWord))]
+    pub raw: Option<JsWord>,
+}
+
+impl EqIgnoreSpan for BigInt {
+    fn eq_ignore_span(&self, other: &Self) -> bool {
+        self.value == other.value
+    }
 }
 
 #[cfg(feature = "rkyv")]
@@ -125,8 +136,9 @@ impl<'a> arbitrary::Arbitrary<'a> for BigInt {
     fn arbitrary(u: &mut arbitrary::Unstructured<'_>) -> arbitrary::Result<Self> {
         let span = u.arbitrary()?;
         let value = u.arbitrary::<usize>()?.into();
+        let raw = Some(u.arbitrary::<String>()?.into());
 
-        Ok(Self { span, value })
+        Ok(Self { span, value, raw })
     }
 }
 
@@ -368,6 +380,7 @@ impl From<BigIntValue> for BigInt {
         BigInt {
             span: DUMMY_SP,
             value,
+            raw: None,
         }
     }
 }

crates/swc_ecma_codegen/src/lib.rs

@@ -568,8 +568,20 @@ where
     fn emit_big_lit(&mut self, v: &BigInt) -> Result {
         self.emit_leading_comments_of_span(v.span, false)?;
 
-        self.wr.write_lit(v.span, &v.value.to_string())?;
-        self.wr.write_lit(v.span, "n")?;
+        if self.cfg.minify {
+            self.wr.write_lit(v.span, &v.value.to_string())?;
+            self.wr.write_lit(v.span, "n")?;
+        } else {
+            match &v.raw {
+                Some(raw) => {
+                    self.wr.write_lit(v.span, raw)?;
+                }
+                _ => {
+                    self.wr.write_lit(v.span, &v.value.to_string())?;
+                    self.wr.write_lit(v.span, "n")?;
+                }
+            }
+        }
     }
 
     // fn emit_object_binding_pat(&mut self, node: &ObjectPat) -> Result {

crates/swc_ecma_codegen/tests/fixture/big-int/input.js

@@ -0,0 +1,14 @@
+const n1 = 9007199254740991n;
+const n2 = -9007199254740991n;
+const n3 = +9007199254740991n;
+const n4 = 100_000_000n;
+const n5 = 0n;
+const n6 = +0n;
+const n7 = -0n;
+const n8 = 0x10n;
+const n9 = 0X10n;
+const n10 = 0X10n;
+const n11 = 0b101n;
+const n12 = 0b101n;
+const n13 = 0o13n;
+const n14 = 0b101_101n;

crates/swc_ecma_codegen/tests/fixture/big-int/output.js

@@ -0,0 +1,14 @@
+const n1 = 9007199254740991n;
+const n2 = -9007199254740991n;
+const n3 = +9007199254740991n;
+const n4 = 100_000_000n;
+const n5 = 0n;
+const n6 = +0n;
+const n7 = -0n;
+const n8 = 0x10n;
+const n9 = 0X10n;
+const n10 = 0X10n;
+const n11 = 0b101n;
+const n12 = 0b101n;
+const n13 = 0o13n;
+const n14 = 0b101_101n;

crates/swc_ecma_codegen/tests/fixture/big-int/output.min.js

@@ -0,0 +1 @@
+const n1=9007199254740991n;const n2=-9007199254740991n;const n3=+9007199254740991n;const n4=100000000n;const n5=0n;const n6=+0n;const n7=-0n;const n8=16n;const n9=16n;const n10=16n;const n11=5n;const n12=5n;const n13=11n;const n14=45n

crates/swc_ecma_codegen/tests/fixture/number/input.js

@@ -61,3 +61,17 @@ const foo43 = -1000000003242;
 const foo43 = 0.0;
 const foo44 = -0.0;
 const foo45 = +0.0;
+const foo46 = 1_000_000_000
+const foo47 = 1.1_00_01
+const foo48 = 1e1_0
+const foo49 = 1e+1_0
+const foo50 = 1e-1_0
+const foo51 = 1.1e10_0
+const foo52 = 1.1e+10_0
+const foo53 = 1.1e-10_0
+const foo54 = 12_34_56
+const foo55 = 1_22_333
+const foo56 = 1_2.3_4
+const foo57 = 1_2.3_4e5_6
+const foo58 = 1_2.3_4e+5_6
+const foo59 = 1_2.3_4e-5_6

crates/swc_ecma_codegen/tests/fixture/number/output.js

@@ -63,3 +63,17 @@ const foo43 = -1000000003242;
 const foo43 = 0;
 const foo44 = -0;
 const foo45 = +0;
+const foo46 = 1000000000;
+const foo47 = 1.10001;
+const foo48 = 10000000000;
+const foo49 = 10000000000;
+const foo50 = 0.0000000001;
+const foo51 = 11000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000;
+const foo52 = 11000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000;
+const foo53 = 0.00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000011;
+const foo54 = 123456;
+const foo55 = 122333;
+const foo56 = 12.34;
+const foo57 = 1234000000000000000000000000000000000000000000000000000000;
+const foo58 = 1234000000000000000000000000000000000000000000000000000000;
+const foo59 = 0.0000000000000000000000000000000000000000000000000000001234;

crates/swc_ecma_codegen/tests/fixture/number/output.min.js

@@ -1 +1 @@
-const exp=1e3;const Exp=1e12;const negativeExp=1e-8;const huge=0xe8d4a51001;const big=100000000001;const fractional=100.23002;const numeric_separators=1e12;const one=1e3;const two=1e6;const three=-1e6;const bin=85;const oct=342391;const hex=3735928559;const fractional2=1000.0001;const identifier=_1000;const negate_identifier=-_1000;const foo=.1;const foo1=+.1;const foo2=-.1;const foo3=1050;const foo4=100500;const foo5=10005e3;const foo6=100005e4;const foo7=.1;const foo8=.01;const foo9=.001;const foo10=1e-4;const foo11=1e-5;const foo12=1e-6;const foo13=24e-6;const foo14=-24e-6;const foo15=10;const foo16=100;const foo17=1e3;const foo18=1e4;const foo19=1e5;const foo20=.1;const foo21=.01;const foo22=.001;const foo23=1e-4;const foo24=1e-5;const foo25=-.1;const foo26=-.01;const foo27=-.001;const foo28=-1e-4;const foo29=-1e-5;const foo30=.001;const foo31=.00321;const foo32=321e-6;const foo33=.1;const foo34=1e-5;const foo35=0;const foo36=-0;const foo37=+0;const foo38=1543e-8;const foo39=1543e-8;const foo40=(1543e-8);const foo41={100005e4:"foo"};const foo42=0xe8d4a51caa;const foo43=-0xe8d4a51caa;const foo43=0;const foo44=-0;const foo45=+0
+const exp=1e3;const Exp=1e12;const negativeExp=1e-8;const huge=0xe8d4a51001;const big=100000000001;const fractional=100.23002;const numeric_separators=1e12;const one=1e3;const two=1e6;const three=-1e6;const bin=85;const oct=342391;const hex=3735928559;const fractional2=1000.0001;const identifier=_1000;const negate_identifier=-_1000;const foo=.1;const foo1=+.1;const foo2=-.1;const foo3=1050;const foo4=100500;const foo5=10005e3;const foo6=100005e4;const foo7=.1;const foo8=.01;const foo9=.001;const foo10=1e-4;const foo11=1e-5;const foo12=1e-6;const foo13=24e-6;const foo14=-24e-6;const foo15=10;const foo16=100;const foo17=1e3;const foo18=1e4;const foo19=1e5;const foo20=.1;const foo21=.01;const foo22=.001;const foo23=1e-4;const foo24=1e-5;const foo25=-.1;const foo26=-.01;const foo27=-.001;const foo28=-1e-4;const foo29=-1e-5;const foo30=.001;const foo31=.00321;const foo32=321e-6;const foo33=.1;const foo34=1e-5;const foo35=0;const foo36=-0;const foo37=+0;const foo38=1543e-8;const foo39=1543e-8;const foo40=(1543e-8);const foo41={100005e4:"foo"};const foo42=0xe8d4a51caa;const foo43=-0xe8d4a51caa;const foo43=0;const foo44=-0;const foo45=+0;const foo46=1e9;const foo47=1.10001;const foo48=1e10;const foo49=1e10;const foo50=1e-10;const foo51=11e99;const foo52=11e99;const foo53=11e-101;const foo54=123456;const foo55=122333;const foo56=12.34;const foo57=1234e54;const foo58=1234e54;const foo59=1234e-58

crates/swc_ecma_parser/src/lexer/mod.rs

@@ -186,7 +186,10 @@ impl<'a, I: Input> Lexer<'a, I> {
                         .read_number(true)
                         .map(|v| match v {
                             Left(v) => Num(v),
-                            Right(v) => BigInt(v),
+                            Right((value, raw)) => BigInt {
+                                value,
+                                raw: raw.into(),
+                            },
                         })
                         .map(Some);
                 }
@@ -269,7 +272,10 @@ impl<'a, I: Input> Lexer<'a, I> {
                             .read_number(false)
                             .map(|v| match v {
                                 Left(v) => Num(v),
-                                Right(v) => BigInt(v),
+                                Right((value, raw)) => BigInt {
+                                    value,
+                                    raw: raw.into(),
+                                },
                             })
                             .map(Some)
                     }
@@ -278,16 +284,23 @@ impl<'a, I: Input> Lexer<'a, I> {
                 return bigint
                     .map(|v| match v {
                         Left(v) => Num(v),
-                        Right(v) => BigInt(v),
+                        Right((value, raw)) => BigInt {
+                            value,
+                            raw: raw.into(),
+                        },
                     })
                     .map(Some);
             }
+
             '1'..='9' => {
                 return self
                     .read_number(false)
                     .map(|v| match v {
                         Left(v) => Num(v),
-                        Right(v) => BigInt(v),
+                        Right((value, raw)) => BigInt {
+                            value,
+                            raw: raw.into(),
+                        },
                     })
                     .map(Some)
             }

crates/swc_ecma_parser/src/lexer/number.rs

@@ -33,8 +33,9 @@ impl<'a, I: Input> Lexer<'a, I> {
     pub(super) fn read_number(
         &mut self,
         starts_with_dot: bool,
-    ) -> LexResult<Either<f64, BigIntValue>> {
+    ) -> LexResult<Either<f64, (BigIntValue, String)>> {
         debug_assert!(self.cur().is_some());
+
         if starts_with_dot {
             debug_assert_eq!(
                 self.cur(),
@@ -42,6 +43,7 @@ impl<'a, I: Input> Lexer<'a, I> {
                 "read_number(starts_with_dot = true) expects current char to be '.'"
             );
         }
+
         let start = self.cur_pos();
 
         let starts_with_zero = self.cur().unwrap() == '0';
@@ -52,14 +54,25 @@ impl<'a, I: Input> Lexer<'a, I> {
             0f64
         } else {
             // Use read_number_no_dot to support long numbers.
-            let (val, s, not_octal) = self
+            let (val, s, mut raw, not_octal) = self
                 .read_number_no_dot_as_str::<10, { lexical::NumberFormatBuilder::from_radix(10) }>(
                 )?;
-            if self.input.cur() == Some('n') {
-                self.input.bump();
-                return Ok(Either::Right(s.into_value()));
+
+            if self.eat(b'n') {
+                raw.push('n');
+
+                return Ok(Either::Right((s.into_value(), raw)));
             }
+
+            if self.input.cur() == Some('n') {
+                raw.push('n');
+                self.input.bump();
+
+                return Ok(Either::Right((s.into_value(), raw)));
+            }
+
             write!(raw_val, "{}", &s.value).unwrap();
+
             if starts_with_zero {
                 // TODO: I guess it would be okay if I don't use -ffast-math
                 // (or something like that), but needs review.
@@ -117,7 +130,9 @@ impl<'a, I: Input> Lexer<'a, I> {
         // `.1.a`, `.1e-4.a` are valid,
         if self.cur() == Some('.') {
             raw_val.push('.');
+
             self.bump();
+
             if starts_with_dot {
                 debug_assert!(self.cur().is_some());
                 debug_assert!(self.cur().unwrap().is_digit(10));
@@ -126,12 +141,15 @@ impl<'a, I: Input> Lexer<'a, I> {
             let mut raw = Raw(Some(String::new()));
             // Read numbers after dot
             let dec_val = self.read_int::<10>(0, &mut raw)?;
+
             val = {
                 if let Some(..) = dec_val {
                     raw_val.push_str(raw.0.as_ref().unwrap());
                 }
 
                 raw_val
+                    // Remove number separator from number
+                    .replace('_', "")
                     .parse()
                     .expect("failed to parse float using rust's impl")
             };
@@ -176,7 +194,10 @@ impl<'a, I: Input> Lexer<'a, I> {
                 write!(raw_val, "{}", exp).unwrap();
 
                 // TODO:
-                raw_val.parse().expect("failed to parse float literal")
+                raw_val
+                    .replace('_', "")
+                    .parse()
+                    .expect("failed to parse float literal")
             }
         }
 
@@ -188,7 +209,7 @@ impl<'a, I: Input> Lexer<'a, I> {
     /// Returns `Left(value)` or `Right(BigInt)`
     pub(super) fn read_radix_number<const RADIX: u8, const FORMAT: u128>(
         &mut self,
-    ) -> LexResult<Either<f64, BigIntValue>> {
+    ) -> LexResult<Either<f64, (BigIntValue, String)>> {
         debug_assert!(
             RADIX == 2 || RADIX == 8 || RADIX == 16,
             "radix should be one of 2, 8, 16, but got {}",
@@ -196,17 +217,36 @@ impl<'a, I: Input> Lexer<'a, I> {
         );
         debug_assert_eq!(self.cur(), Some('0'));
 
-        self.bump(); // 0
-        self.bump(); // x
+        self.with_buf(|l, buf| {
+            l.bump();
+            buf.push('0');
 
-        let (val, s, _) = self.read_number_no_dot_as_str::<RADIX, FORMAT>()?;
-        if self.eat(b'n') {
-            return Ok(Either::Right(s.into_value()));
-        }
+            let c = match l.input.cur() {
+                Some(c) => {
+                    l.bump();
 
-        self.ensure_not_ident()?;
+                    c
+                }
+                _ => {
+                    unreachable!();
+                }
+            };
 
-        Ok(Either::Left(val))
+            buf.push(c);
+
+            let (val, s, raw, _) = l.read_number_no_dot_as_str::<RADIX, FORMAT>()?;
+
+            if l.eat(b'n') {
+                buf.push_str(&raw);
+                buf.push('n');
+
+                return Ok(Either::Right((s.into_value(), (&**buf).into())));
+            }
+
+            l.ensure_not_ident()?;
+
+            Ok(Either::Left(val))
+        })
     }
 
     /// This can read long integers like
@@ -224,6 +264,7 @@ impl<'a, I: Input> Lexer<'a, I> {
         let res = self.read_digits::<_, f64, RADIX>(
             |total, radix, v| {
                 read_any = true;
+
                 Ok((f64::mul_add(total, radix as f64, v as f64), true))
             },
             &mut Raw(None),
@@ -243,18 +284,19 @@ impl<'a, I: Input> Lexer<'a, I> {
     /// Returned bool is `true` is there was `8` or `9`.
     fn read_number_no_dot_as_str<const RADIX: u8, const FORMAT: u128>(
         &mut self,
-    ) -> LexResult<(f64, LazyBigInt<RADIX>, bool)> {
+    ) -> LexResult<(f64, LazyBigInt<RADIX>, String, bool)> {
         debug_assert!(
             RADIX == 2 || RADIX == 8 || RADIX == 10 || RADIX == 16,
             "radix for read_number_no_dot should be one of 2, 8, 10, 16, but got {}",
             RADIX
         );
         let start = self.cur_pos();
-        let mut non_octal = false;
 
+        let mut non_octal = false;
         let mut read_any = false;
 
         let mut raw = Raw(Some(String::new()));
+
         self.read_digits::<_, f64, RADIX>(
             |total, radix, v| {
                 read_any = true;
@@ -274,13 +316,17 @@ impl<'a, I: Input> Lexer<'a, I> {
         }
 
         let raw_str = raw.0.take().unwrap();
+        // Remove number separator from number
+        let raw_number_str = raw_str.replace('_', "");
+
         Ok((
             lexical::parse_with_options::<f64, _, FORMAT>(
-                raw_str.as_bytes(),
+                raw_number_str.as_bytes(),
                 &lexical::parse_float_options::Options::from_radix(RADIX),
             )
             .expect("failed to parse float using lexical"),
-            LazyBigInt::new(raw_str),
+            LazyBigInt::new(raw_number_str),
+            raw_str,
             non_octal,
         ))
     }
@@ -372,22 +418,24 @@ impl<'a, I: Input> Lexer<'a, I> {
             "radix for read_int should be one of 2, 8, 10, 16, but got {}",
             RADIX
         );
+
         if cfg!(feature = "debug") {
             trace!("read_digits(radix = {}), cur = {:?}", RADIX, self.cur());
         }
 
         let start = self.cur_pos();
-
         let mut total: Ret = Default::default();
-
         let mut prev = None;
+
         while let Some(c) = self.cur() {
             if allow_num_separator && c == '_' {
                 let is_allowed = |c: Option<char>| {
                     if c.is_none() {
                         return false;
                     }
+
                     let c = c.unwrap();
+
                     c.is_digit(RADIX as _)
                 };
                 let is_forbidden = |c: Option<char>| {
@@ -413,6 +461,8 @@ impl<'a, I: Input> Lexer<'a, I> {
 
                 // Ignore this _ character
                 self.input.bump();
+                raw.push(c);
+
                 continue;
             }
 
@@ -426,11 +476,15 @@ impl<'a, I: Input> Lexer<'a, I> {
             raw.push(c);
 
             self.bump();
+
             let (t, cont) = op(total, RADIX, val)?;
+
             total = t;
+
             if !cont {
                 return Ok(total);
             }
+
             prev = Some(c);
         }
 
@@ -612,9 +666,12 @@ mod tests {
                 "10000000000000000000000000000000000000000000000000000n",
                 |l| l.read_number(false).unwrap().right().unwrap()
             ),
-            "10000000000000000000000000000000000000000000000000000"
-                .parse::<BigIntValue>()
-                .unwrap(),
+            (
+                "10000000000000000000000000000000000000000000000000000"
+                    .parse::<BigIntValue>()
+                    .unwrap(),
+                String::from("10000000000000000000000000000000000000000000000000000n")
+            ),
         );
     }
 

crates/swc_ecma_parser/src/lexer/state.rs

@@ -104,7 +104,7 @@ impl<'a> From<&'a Token> for TokenType {
                         | Token::Word(Word::Ident(..))
                         | Token::DollarLBrace
                         | Token::Regex(..)
-                        | Token::BigInt(..)
+                        | Token::BigInt { .. }
                         | Token::JSXText { .. }
                         | Token::RBrace
                 ),

crates/swc_ecma_parser/src/parser/expr.rs

@@ -317,7 +317,7 @@ impl<'a, I: Tokens> Parser<I> {
                 | tok!("true")
                 | tok!("false")
                 | Token::Num(..)
-                | Token::BigInt(..)
+                | Token::BigInt { .. }
                 | Token::Str { .. } => {
                     return Ok(Box::new(Expr::Lit(self.parse_lit()?)));
                 }
@@ -1834,10 +1834,11 @@ impl<'a, I: Tokens> Parser<I> {
                 }),
                 _ => unreachable!(),
             },
-            Token::BigInt(..) => match bump!(self) {
-                Token::BigInt(value) => Lit::BigInt(BigInt {
+            Token::BigInt { .. } => match bump!(self) {
+                Token::BigInt { value, raw } => Lit::BigInt(BigInt {
                     span: span!(self, start),
                     value,
+                    raw: Some(raw),
                 }),
                 _ => unreachable!(),
             },

crates/swc_ecma_parser/src/parser/object.rs

@@ -65,10 +65,11 @@ impl<'a, I: Tokens> Parser<I> {
                     }),
                     _ => unreachable!(),
                 },
-                Token::BigInt(_) => match bump!(p) {
-                    Token::BigInt(value) => PropName::BigInt(BigInt {
+                Token::BigInt { .. } => match bump!(p) {
+                    Token::BigInt { value, raw } => PropName::BigInt(BigInt {
                         span: span!(p, start),
                         value,
+                        raw: Some(raw),
                     }),
                     _ => unreachable!(),
                 },

crates/swc_ecma_parser/src/token.rs

@@ -122,7 +122,10 @@ pub enum Token {
     Num(f64),
 
     #[kind(starts_expr)]
-    BigInt(BigIntValue),
+    BigInt {
+        value: BigIntValue,
+        raw: JsWord,
+    },
 
     JSXName {
         name: JsWord,
@@ -611,7 +614,7 @@ impl Debug for Token {
             Str { value, .. } => write!(f, "string literal ({})", value)?,
             Regex(exp, flags) => write!(f, "regexp literal ({}, {})", exp, flags)?,
             Num(..) => write!(f, "numeric literal")?,
-            BigInt(..) => write!(f, "bigint literal")?,
+            BigInt { .. } => write!(f, "bigint literal")?,
             JSXName { name } => write!(f, "jsx name ({})", name)?,
             JSXText { raw } => write!(f, "jsx text ({})", raw)?,
             JSXTagStart => write!(f, "< (jsx tag start)")?,

crates/swc_ecma_parser/tests/tsc/constructBigint.json

@@ -86,7 +86,8 @@
                 [
                   1
                 ]
-              ]
+              ],
+              "raw": "1n"
             }
           }
         ],

crates/swc_ecma_parser/tests/typescript/custom/bigint/input.ts.json

@@ -32,7 +32,8 @@
             395401161,
             37
           ]
-        ]
+        ],
+        "raw": "1000000000000000000000000000000000000000000000000000000000000000000000n"
       }
     }
   ],

crates/swc_ecma_transforms_compat/src/es2018/object_rest_spread.rs

@@ -790,7 +790,9 @@ impl ObjectRest {
                                 }))),
                             }),
                         ),
-                        PropName::BigInt(BigInt { span, ref value }) => {
+                        PropName::BigInt(BigInt {
+                            span, ref value, ..
+                        }) => {
                             let value = value.clone();
                             (
                                 key,
@@ -1003,7 +1005,7 @@ fn excluded_props(props: &[ObjectPatProp]) -> Vec<Option<ExprOrSpread>> {
                     value: format!("{}", value).into(),
                 })
                 .as_arg(),
-                PropName::BigInt(BigInt { span, value }) => Lit::Str(Str {
+                PropName::BigInt(BigInt { span, value, .. }) => Lit::Str(Str {
                     span: *span,
                     raw: None,
                     value: format!("{}", value).into(),

crates/swc_ecma_visit/src/lib.rs

@@ -986,6 +986,7 @@ define!({
     pub struct BigInt {
         pub span: Span,
         pub value: BigIntValue,
+        pub raw: Option<JsWord>,
     }
     pub struct Str {
         pub span: Span,

crates/swc_estree_ast/src/lit.rs

@@ -317,6 +317,8 @@ pub struct BigIntLiteral {
     pub base: BaseNode,
     #[serde(default)]
     pub value: String,
+    #[serde(default)]
+    pub raw: JsWord,
 }
 
 #[derive(Debug, Clone, Serialize, Deserialize, PartialEq)]

crates/swc_estree_compat/src/babelify/lit.rs

@@ -84,6 +84,11 @@ impl Babelify for BigInt {
         BigIntLiteral {
             base: ctx.base(self.span),
             value: self.value.to_string(),
+            // TODO improve me
+            raw: match self.raw {
+                Some(value) => value,
+                _ => "".into(),
+            },
         }
     }
 }

crates/swc_estree_compat/src/swcify/lit.rs

@@ -138,6 +138,8 @@ impl Swcify for BigIntLiteral {
                 .value
                 .parse()
                 .expect("failed to parse the value of BigIntLiteral"),
+            // TODO improve me
+            raw: None,
         }
     }
 }