Numbers

Decode little-endian integers, IEEE 754 floats, and Boolean flags.

Numeric schemas consume fixed-width little-endian values.

Integers

Schema	Bytes	Domain	Result
`u8()`	1	0 to 255	`number`
`u16()`	2	0 to 65,535	`number`
`u24()`	3	0 to 16,777,215	`number`
`u32()`	4	0 to 4,294,967,295	`number`
`u64()`	8	0 to 2^64 - 1	`bigint`
`i8()`	1	-2^{7 to 2}7 - 1	`number`
`i16()`	2	-2^{15 to 2}15 - 1	`number`
`i24()`	3	-2^{23 to 2}23 - 1	`number`
`i32()`	4	-2^{31 to 2}31 - 1	`number`
`i64()`	8	-2^{63 to 2}63 - 1	`bigint`

import { i16, u16, u64 } from "@marceloclp/bsd";

u16().decode(Uint8Array.of(0x34, 0x12)); // 0x1234
i16().decode(Uint8Array.of(0xfe, 0xff)); // -2
u64().decode(Uint8Array.of(1, 0, 0, 0, 0, 0, 0, 0)); // 1n

Floats

Schema	Bytes	Format	Approximate precision	Result
`f32()`	4	IEEE 754 binary32	7 digits	`number`
`f64()`	8	IEEE 754 binary64	15–16 digits	`number`

import { f32 } from "@marceloclp/bsd";

const input = new Uint8Array(4);
new DataView(input.buffer).setFloat32(0, 21.5, true);

f32().decode(input); // 21.5

The width is part of the binary contract. Floating-point results retain IEEE 754 behavior, including rounding error, NaN, and infinities.

Booleans

Schema	Bytes	Domain	Result
`bool()`	1	0 or 1	`boolean`

import { bool } from "@marceloclp/bsd";

bool().decode(Uint8Array.of(1)); // true

Constraints

All numeric schemas expose the same predicates. Failure throws BsdIssue.

Method	Predicate
`.gt(x)`	`value > x`
`.gte(x)`	`value >= x`
`.lt(x)`	`value < x`
`.lte(x)`	`value <= x`
`.positive()`	`value > 0`
`.negative()`	`value < 0`

import { f32, i16, u16 } from "@marceloclp/bsd";

const Port = u16().gte(1).lte(65_535);
const NegativeDelta = i16().negative();
const Probability = f32().gte(0).lte(1);

.is(value) validates and narrows one literal. .in(values) accepts an array, Set, or Map and narrows to its members.

import { u8, u16 } from "@marceloclp/bsd";

const Version: Bsd<3> = u16().is(3);
const Priority: Bsd<0 | 1 | 2> = u8().in([0, 1, 2]);