Skip to content

Commit

Permalink
refactor(minifier): handle big int values later (#8324)
Browse files Browse the repository at this point in the history
  • Loading branch information
@Boshen
Boshen committed Jan 7, 2025
1 parent 775a289 commit fc662b7
Showing 1 changed file with 140 additions and 131 deletions.
271 changes: 140 additions & 131 deletions crates/oxc_minifier/src/ast_passes/peephole_fold_constants.rs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -59,6 +59,8 @@ impl<'a, 'b> PeepholeFoldConstants {
UnaryOperator::LogicalNot if matches!(&e.argument, Expression::NumericLiteral(lit) if lit.value == 0.0 || lit.value == 1.0) => {
None
}
// Do not fold big int.
UnaryOperator::UnaryNegation if e.argument.is_big_int_literal() => None,
_ => ctx.eval_unary_expression(e).map(|v| ctx.value_to_expr(e.span, v)),
}
}
Expand Down Expand Up @@ -1004,79 +1006,6 @@ mod test {
test_same("'' + x === +y");
}

#[test]
#[ignore]
fn test_bigint_number_comparison() {
test("1n < 2", "true");
test("1n > 2", "false");
test("1n == 1", "true");
test("1n == 2", "false");

// comparing with decimals is allowed
test("1n < 1.1", "true");
test("1n < 1.9", "true");
test("1n < 0.9", "false");
test("-1n < -1.1", "false");
test("-1n < -1.9", "false");
test("-1n < -0.9", "true");
test("1n > 1.1", "false");
test("1n > 0.9", "true");
test("-1n > -1.1", "true");
test("-1n > -0.9", "false");

// Don't fold unsafely large numbers because there might be floating-point error
test(&format!("0n > {MAX_SAFE_INT}"), "false");
test(&format!("0n < {MAX_SAFE_INT}"), "true");
test(&format!("0n > {NEG_MAX_SAFE_INT}"), "true");
test(&format!("0n < {NEG_MAX_SAFE_INT}"), "false");
test(&format!("0n > {MAX_SAFE_FLOAT}"), "false");
test(&format!("0n < {MAX_SAFE_FLOAT}"), "true");
test(&format!("0n > {NEG_MAX_SAFE_FLOAT}"), "true");
test(&format!("0n < {NEG_MAX_SAFE_FLOAT}"), "false");

// comparing with Infinity is allowed
test("1n < Infinity", "true");
test("1n > Infinity", "false");
test("1n < -Infinity", "false");
test("1n > -Infinity", "true");

// null is interpreted as 0 when comparing with bigint
// test("1n < null", "false");
// test("1n > null", "true");
}

#[test]
#[ignore]
fn test_bigint_string_comparison() {
test("1n < '2'", "true");
test("2n > '1'", "true");
test("123n > '34'", "true");
test("1n == '1'", "true");
test("1n == '2'", "false");
test("1n != '1'", "false");
test("1n === '1'", "false");
test("1n !== '1'", "true");
}

#[test]
#[ignore]
fn test_string_bigint_comparison() {
test("'1' < 2n", "true");
test("'2' > 1n", "true");
test("'123' > 34n", "true");
test("'1' == 1n", "true");
test("'1' == 2n", "false");
test("'1' != 1n", "false");
test("'1' === 1n", "false");
test("'1' !== 1n", "true");
}

#[test]
fn test_object_bigint_comparison() {
test_same("{ valueOf: function() { return 0n; } } != 0n");
test_same("{ toString: function() { return '0'; } } != 0n");
}

#[test]
fn test_nan_comparison() {
test("NaN < 1", "false");
Expand Down Expand Up @@ -1407,64 +1336,6 @@ mod test {
test("x = 12 | NaN", "x=12");
}

#[test]
fn test_fold_bitwise_op_with_big_int() {
test("x = 1n & 1n", "x = 1n");
test("x = 1n & 2n", "x = 0n");
test("x = 3n & 1n", "x = 1n");
test("x = 3n & 3n", "x = 3n");

test("x = 1n | 1n", "x = 1n");
test("x = 1n | 2n", "x = 3n");
test("x = 1n | 3n", "x = 3n");
test("x = 3n | 1n", "x = 3n");
test("x = 3n | 3n", "x = 3n");
test("x = 1n | 4n", "x = 5n");

test("x = 1n ^ 1n", "x = 0n");
test("x = 1n ^ 2n", "x = 3n");
test("x = 3n ^ 1n", "x = 2n");
test("x = 3n ^ 3n", "x = 0n");

test("x = -1n & 0n", "x = 0n");
test("x = 0n & -1n", "x = 0n");
test("x = 1n & 4n", "x = 0n");
test("x = 2n & 3n", "x = 2n");

test("x = 1n & 3000000000n", "x = 0n");
test("x = 3000000000n & 1n", "x = 0n");

// bitwise OR does not affect the sign of a bigint
test("x = 1n | 3000000001n", "x = 3000000001n");
test("x = 4294967295n | 0n", "x = 4294967295n");

test("x = y & 1n & 1n", "x = y & 1n");
test("x = y & 1n & 2n", "x = y & 0n");
test("x = y & 3n & 1n", "x = y & 1n");
test("x = 3n & y & 1n", "x = y & 1n");
test("x = y & 3n & 3n", "x = y & 3n");
test("x = 3n & y & 3n", "x = y & 3n");

test("x = y | 1n | 1n", "x = y | 1n");
test("x = y | 1n | 2n", "x = y | 3n");
test("x = y | 3n | 1n", "x = y | 3n");
test("x = 3n | y | 1n", "x = y | 3n");
test("x = y | 3n | 3n", "x = y | 3n");
test("x = 3n | y | 3n", "x = y | 3n");

test("x = y ^ 1n ^ 1n", "x = y ^ 0n");
test("x = y ^ 1n ^ 2n", "x = y ^ 3n");
test("x = y ^ 3n ^ 1n", "x = y ^ 2n");
test("x = 3n ^ y ^ 1n", "x = y ^ 2n");
test("x = y ^ 3n ^ 3n", "x = y ^ 0n");
test("x = 3n ^ y ^ 3n", "x = y ^ 0n");

// TypeError: Cannot mix BigInt and other types
test_same("1n & 1");
test_same("1n | 1");
test_same("1n ^ 1");
}

#[test]
fn test_fold_bitwise_op_additional() {
test("x = null & 1", "x = 0");
Expand Down Expand Up @@ -1822,4 +1693,142 @@ mod test {
test("typeof foo.bar === typeof foo.bar", "typeof foo.bar == typeof foo.bar");
test("typeof foo.bar !== typeof foo.bar", "typeof foo.bar != typeof foo.bar");
}

// TODO: All big ints are rare and difficult to handle.
mod bigint {
use super::{
test, test_same, MAX_SAFE_FLOAT, MAX_SAFE_INT, NEG_MAX_SAFE_FLOAT, NEG_MAX_SAFE_INT,
};

#[test]
fn test_fold_bitwise_op_with_big_int() {
test("x = 1n & 1n", "x = 1n");
test("x = 1n & 2n", "x = 0n");
test("x = 3n & 1n", "x = 1n");
test("x = 3n & 3n", "x = 3n");

test("x = 1n | 1n", "x = 1n");
test("x = 1n | 2n", "x = 3n");
test("x = 1n | 3n", "x = 3n");
test("x = 3n | 1n", "x = 3n");
test("x = 3n | 3n", "x = 3n");
test("x = 1n | 4n", "x = 5n");

test("x = 1n ^ 1n", "x = 0n");
test("x = 1n ^ 2n", "x = 3n");
test("x = 3n ^ 1n", "x = 2n");
test("x = 3n ^ 3n", "x = 0n");

// test("x = -1n & 0n", "x = 0n");
// test("x = 0n & -1n", "x = 0n");
test("x = 1n & 4n", "x = 0n");
test("x = 2n & 3n", "x = 2n");

test("x = 1n & 3000000000n", "x = 0n");
test("x = 3000000000n & 1n", "x = 0n");

// bitwise OR does not affect the sign of a bigint
test("x = 1n | 3000000001n", "x = 3000000001n");
test("x = 4294967295n | 0n", "x = 4294967295n");

test("x = y & 1n & 1n", "x = y & 1n");
test("x = y & 1n & 2n", "x = y & 0n");
test("x = y & 3n & 1n", "x = y & 1n");
test("x = 3n & y & 1n", "x = y & 1n");
test("x = y & 3n & 3n", "x = y & 3n");
test("x = 3n & y & 3n", "x = y & 3n");

test("x = y | 1n | 1n", "x = y | 1n");
test("x = y | 1n | 2n", "x = y | 3n");
test("x = y | 3n | 1n", "x = y | 3n");
test("x = 3n | y | 1n", "x = y | 3n");
test("x = y | 3n | 3n", "x = y | 3n");
test("x = 3n | y | 3n", "x = y | 3n");

test("x = y ^ 1n ^ 1n", "x = y ^ 0n");
test("x = y ^ 1n ^ 2n", "x = y ^ 3n");
test("x = y ^ 3n ^ 1n", "x = y ^ 2n");
test("x = 3n ^ y ^ 1n", "x = y ^ 2n");
test("x = y ^ 3n ^ 3n", "x = y ^ 0n");
test("x = 3n ^ y ^ 3n", "x = y ^ 0n");

// TypeError: Cannot mix BigInt and other types
test_same("1n & 1");
test_same("1n | 1");
test_same("1n ^ 1");
}

#[test]
#[ignore]
fn test_bigint_number_comparison() {
test("1n < 2", "true");
test("1n > 2", "false");
test("1n == 1", "true");
test("1n == 2", "false");

// comparing with decimals is allowed
test("1n < 1.1", "true");
test("1n < 1.9", "true");
test("1n < 0.9", "false");
test("-1n < -1.1", "false");
test("-1n < -1.9", "false");
test("-1n < -0.9", "true");
test("1n > 1.1", "false");
test("1n > 0.9", "true");
test("-1n > -1.1", "true");
test("-1n > -0.9", "false");

// Don't fold unsafely large numbers because there might be floating-point error
test(&format!("0n > {MAX_SAFE_INT}"), "false");
test(&format!("0n < {MAX_SAFE_INT}"), "true");
test(&format!("0n > {NEG_MAX_SAFE_INT}"), "true");
test(&format!("0n < {NEG_MAX_SAFE_INT}"), "false");
test(&format!("0n > {MAX_SAFE_FLOAT}"), "false");
test(&format!("0n < {MAX_SAFE_FLOAT}"), "true");
test(&format!("0n > {NEG_MAX_SAFE_FLOAT}"), "true");
test(&format!("0n < {NEG_MAX_SAFE_FLOAT}"), "false");

// comparing with Infinity is allowed
test("1n < Infinity", "true");
test("1n > Infinity", "false");
test("1n < -Infinity", "false");
test("1n > -Infinity", "true");

// null is interpreted as 0 when comparing with bigint
// test("1n < null", "false");
// test("1n > null", "true");
}

#[test]
#[ignore]
fn test_bigint_string_comparison() {
test("1n < '2'", "true");
test("2n > '1'", "true");
test("123n > '34'", "true");
test("1n == '1'", "true");
test("1n == '2'", "false");
test("1n != '1'", "false");
test("1n === '1'", "false");
test("1n !== '1'", "true");
}

#[test]
#[ignore]
fn test_string_bigint_comparison() {
test("'1' < 2n", "true");
test("'2' > 1n", "true");
test("'123' > 34n", "true");
test("'1' == 1n", "true");
test("'1' == 2n", "false");
test("'1' != 1n", "false");
test("'1' === 1n", "false");
test("'1' !== 1n", "true");
}

#[test]
fn test_object_bigint_comparison() {
test_same("{ valueOf: function() { return 0n; } } != 0n");
test_same("{ toString: function() { return '0'; } } != 0n");
}
}
}

0 comments on commit fc662b7

Please sign in to comment.