v/vlib/math/unsigned/uint128.v

module unsigned

import math.bits
import encoding.binary

pub struct Uint128 {
pub mut:
	lo u64
	hi u64
}

pub const (
	uint128_zero = Uint128{}
	uint128_max  = Uint128{18446744073709551615, 18446744073709551615}
)

// is_zero returns true if u == 0
pub fn (u Uint128) is_zero() bool {
	return u == Uint128{}
}

// equals returns true if u == v.
//
// Uint128 values can be compared directly with ==, but use of the Equals method
// is preferred for consistency.
pub fn (u Uint128) equals(v Uint128) bool {
	return u == v
}

// equals_64 returns true if u == v
pub fn (u Uint128) equals_64(v u64) bool {
	return u.lo == v && u.hi == 0
}

// cmp compares u and v and returns:
//
// -1 if u <  v
//  0 if u == v
// +1 if u >  v
//
pub fn (u Uint128) cmp(v Uint128) int {
	if u == v {
		return 0
	} else if u.hi < v.hi || (u.hi == v.hi && u.lo < v.lo) {
		return -1
	} else {
		return 1
	}
}

// cmp_64 compares u and v and returns:
//
//   -1 if u <  v
//    0 if u == v
//   +1 if u >  v
//
pub fn (u Uint128) cmp_64(v u64) int {
	if u.hi == 0 && u.lo == v {
		return 0
	} else if u.hi == 0 && u.lo < v {
		return -1
	} else {
		return 1
	}
}

// and returns u & v
pub fn (u Uint128) and(v Uint128) Uint128 {
	return Uint128{u.lo & v.lo, u.hi & v.hi}
}

// and_64 rreturns u & v
pub fn (u Uint128) and_64(v u64) Uint128 {
	return Uint128{u.lo & v, u.hi & 0}
}

// or returns u | v
pub fn (u Uint128) or_(v Uint128) Uint128 {
	return Uint128{u.lo | v.lo, u.hi | v.hi}
}

// or returns u | v
pub fn (u Uint128) or_64(v u64) Uint128 {
	return Uint128{u.lo | v, u.hi | 0}
}

// xor returns u ^ v
pub fn (u Uint128) xor(v Uint128) Uint128 {
	return Uint128{u.lo ^ v.lo, u.hi ^ v.hi}
}

// xor_64 returns u ^ v
pub fn (u Uint128) xor_64(v u64) Uint128 {
	return Uint128{u.lo ^ v, u.hi ^ 0}
}

// add returns u + v with wraparound semantics
pub fn (u Uint128) add(v Uint128) Uint128 {
	lo, carry := bits.add_64(u.lo, v.lo, 0)
	hi, _ := bits.add_64(u.hi, v.hi, carry)
	return Uint128{lo, hi}
}

pub fn add_128(x Uint128, y Uint128, carry u64) (Uint128, u64) {
	mut sum := Uint128{}
	mut carry_out := u64(0)
	sum.lo, carry_out = bits.add_64(x.lo, y.lo, carry)
	sum.hi, carry_out = bits.add_64(x.hi, y.hi, carry_out)
	return sum, carry_out
}

pub fn sub_128(x Uint128, y Uint128, borrow u64) (Uint128, u64) {
	mut diff := Uint128{}
	mut borrow_out := u64(0)
	diff.lo, borrow_out = bits.sub_64(x.lo, y.lo, borrow)
	diff.hi, borrow_out = bits.sub_64(x.hi, y.hi, borrow_out)
	return diff, borrow_out
}

pub fn mul_128(x Uint128, y Uint128) (Uint128, Uint128) {
	mut lo := Uint128{}
	mut hi := Uint128{}
	lo.hi, lo.lo = bits.mul_64(x.lo, y.lo)
	hi.hi, hi.lo = bits.mul_64(x.hi, y.hi)
	t0, t1 := bits.mul_64(x.lo, y.hi)
	t2, t3 := bits.mul_64(x.hi, y.lo)

	mut c0 := u64(0)
	mut c1 := u64(0)
	lo.hi, c0 = bits.add_64(lo.hi, t1, 0)
	lo.hi, c1 = bits.add_64(lo.hi, t3, 0)
	hi.lo, c0 = bits.add_64(hi.lo, t0, c0)
	hi.lo, c1 = bits.add_64(hi.lo, t2, c1)
	hi.hi += c0 + c1
	return hi, lo
}

pub fn div_128(hi Uint128, lo Uint128, y_ Uint128) (Uint128, Uint128) {
	mut y := y_
	if y.is_zero() {
		panic('integer divide by zero')
	}

	if y.cmp(hi) <= 0 {
		panic('integer overflow')
	}

	s := u32(y.leading_zeros())
	y = y.lsh(s)

	un32 := hi.lsh(s).or_(lo.rsh(128 - s))
	un10 := lo.lsh(s)
	mut q1, rhat := un32.quo_rem_64(y.hi)
	mut r1 := uint128_from_64(rhat)
	tmp := Uint128{r1.lo, un10.hi}
	for q1.hi != 0 || q1.mul_64(y.lo).cmp(tmp) > 0 {
		q1 = q1.sub_64(1)
		r1 = r1.add_64(y.hi)
		if r1.hi != 0 {
			break
		}
	}

	un21 := Uint128{un32.lo, un10.hi}.sub(q1.mul(y))
	mut q0, rhat2 := un21.quo_rem_64(y.hi)
	mut r0 := uint128_from_64(rhat2)
	tmp2 := Uint128{r0.lo, un10.lo}
	for q0.hi != 0 || q0.mul_64(y.lo).cmp(tmp2) > 0 {
		q0 = q0.sub_64(1)
		r0 = r0.add_64(y.hi)
		if r0.hi != 0 {
			break
		}
	}

	return Uint128{q1.lo, q0.lo}, Uint128{un21.lo, un10.lo}.sub(q0.mul(y)).rsh(s)
}

// add_64 returns u + v with wraparound semantics
pub fn (u Uint128) add_64(v u64) Uint128 {
	lo, carry := bits.add_64(u.lo, v, 0)
	hi := u.hi + carry
	return Uint128{lo, hi}
}

// sub returns u - v with wraparound semantics
pub fn (u Uint128) sub(v Uint128) Uint128 {
	lo, borrow := bits.sub_64(u.lo, v.lo, 0)
	hi, _ := bits.sub_64(u.hi, v.hi, borrow)
	return Uint128{lo, hi}
}

// sub_64 returns u - v with wraparound semantics
pub fn (u Uint128) sub_64(v u64) Uint128 {
	lo, borrow := bits.sub_64(u.lo, v, 0)
	hi := u.hi - borrow
	return Uint128{lo, hi}
}

// mul returns u * v with wraparound semantics
pub fn (u Uint128) mul(v Uint128) Uint128 {
	mut hi, lo := bits.mul_64(u.lo, v.lo)
	hi += u.hi * v.lo + u.lo * v.hi
	return Uint128{lo, hi}
}

// mul_64 returns u * v with wraparound semantics
pub fn (u Uint128) mul_64(v u64) Uint128 {
	mut hi, lo := bits.mul_64(u.lo, v)
	hi += u.hi * v
	return Uint128{lo, hi}
}

pub fn (u Uint128) overflowing_mul_64(v u64) (Uint128, bool) {
	hi, lo := bits.mul_64(u.lo, v)
	p0, p1 := bits.mul_64(u.hi, v)
	hi2, c0 := bits.add_64(hi, p1, 0)

	return Uint128{lo, hi2}, p0 != 0 || c0 != 0
}

pub fn (u Uint128) overflowing_add_64(v u64) (Uint128, u64) {
	lo, carry := bits.add_64(u.lo, v, 0)
	hi, carry2 := bits.add_64(u.hi, 0, carry)

	return Uint128{lo, hi}, carry2
}

// div returns u / v
pub fn (u Uint128) div(v Uint128) Uint128 {
	q, _ := u.quo_rem(v)
	return q
}

// mod returns r = u % v
pub fn (u Uint128) mod(v Uint128) Uint128 {
	_, r := u.quo_rem(v)
	return r
}

// mod_64 returns r = u % v
pub fn (u Uint128) mod_64(v u64) u64 {
	_, r := u.quo_rem_64(v)
	return r
}

// quo_rem_64 returns q = u/v and r = u%v
pub fn (u Uint128) quo_rem_64(v u64) (Uint128, u64) {
	if u.hi < v {
		mut r := u64(0)
		mut q := Uint128{0, 0}
		q.lo, r = bits.div_64(u.hi, u.lo, v)

		return q, r
	} else {
		mut q := Uint128{0, 0}
		mut r := u64(0)
		mut r2 := u64(0)
		q.hi, r = bits.div_64(0, u.hi, v)
		q.lo, r2 = bits.div_64(r, u.lo, v)
		return q, r2
	}
}

// quo_rem returns q = u/v and r = u%v
pub fn (u Uint128) quo_rem(v Uint128) (Uint128, Uint128) {
	mut q := Uint128{}
	mut r := Uint128{}
	if v.hi == 0 {
		mut r64 := u64(0)
		q, r64 = u.quo_rem_64(v.lo)
		r = Uint128{r64, 0}
	} else {
		n := bits.leading_zeros_64(v.hi)
		v1 := v.lsh(u32(n))
		u1 := v.rsh(1)

		mut tq, _ := bits.div_64(u1.hi, u1.lo, v1.hi)
		tq >>= u64(64 - n)
		if tq != 0 {
			tq -= 1
		}

		q = Uint128{tq, 0}
		r = u - v.mul_64(tq)

		if r.cmp(v) >= 0 {
			q = q.add_64(1)
			r = r - v
		}
	}

	return q, r
}

// lsh returns u << n
pub fn (u Uint128) lsh(n u32) Uint128 {
	mut s := Uint128{}
	if n > 64 {
		s.lo = 0
		s.hi = u.lo << (n - 64)
	} else {
		s.lo = u.lo << n
		s.hi = u.hi << n | u.lo >> (64 - n)
	}

	return s
}

// rsh returns u >> n
pub fn (u Uint128) rsh(n u32) Uint128 {
	mut s := Uint128{}
	if n > 64 {
		s.hi = 0
		s.lo = u.hi << (n - 64)
	} else {
		s.lo = u.lo >> n | u.hi << (64 - n)
		s.hi = u.hi >> n
	}

	return s
}

// leading_zeros returns the number of leading zero bits in u; the result is 128
// for u == 0.
pub fn (u Uint128) leading_zeros() int {
	if u.hi > 0 {
		return bits.leading_zeros_64(u.hi)
	}
	return 64 + bits.leading_zeros_64(u.lo)
}

// trailing_zeros returns the number of trailing zero bits in u; the result is
// 128 for u == 0.
pub fn (u Uint128) trailing_zeros() int {
	if u.lo > 0 {
		return bits.trailing_zeros_64(u.lo)
	}
	return 64 + bits.trailing_zeros_64(u.hi)
}

// ones_count returns the number of one bits ("population count" in u)
pub fn (u Uint128) ones_count() int {
	return bits.ones_count_64(u.hi) + bits.ones_count_64(u.lo)
}

// rotate_left returns the value of u rotated left by (k mod 128) bits.
pub fn (u Uint128) rotate_left(k int) Uint128 {
	n := u32(128)
	s := u32(k) & (n - 1)
	return u.lsh(s).or_(u.rsh(n - s))
}

// rotate_right returns the value of u rotated right by (k mod 128) bits.
pub fn (u Uint128) rotate_right(k int) Uint128 {
	return u.rotate_left(-k)
}

// reverse returns the value of u with its bits in reversed order.
pub fn (u Uint128) reverse() Uint128 {
	return Uint128{bits.reverse_64(u.hi), bits.reverse_64(u.lo)}
}

// reverse_bytes returns the value of u with its bytes in reversed order.
pub fn (u Uint128) reverse_bytes() Uint128 {
	return Uint128{bits.reverse_bytes_64(u.hi), bits.reverse_bytes_64(u.lo)}
}

pub fn (u Uint128) not() Uint128 {
	return Uint128{~u.lo, ~u.hi}
}

// len returns the minimum number of bits required to represent u; the result is
// 0 for u == 0.
pub fn (u Uint128) len() int {
	return 128 - u.leading_zeros()
}

// string returns the base-10 representation of u as a string
pub fn (u_ Uint128) str() string {
	mut u := u_
	if u.is_zero() {
		return '0'
	}

	mut buf := '0000000000000000000000000000000000000000'.bytes() // log10(2^128) < 40

	for i := buf.len; true; i -= 19 {
		q, mut r := u.quo_rem_64(u64(1e19))

		mut n := int(0)
		for ; r != 0; r /= 10 {
			n++
			buf[i - n] += byte(r % 10)
		}
		if q.is_zero() {
			return buf[i - n..].bytestr()
		}
		u = q
	}

	return ''
}

// put_bytes stores u in b in little-endian order
pub fn (u Uint128) put_bytes(mut b []byte) {
	binary.little_endian_put_u64(mut b, u.lo)
	binary.little_endian_put_u64(mut b, u.hi)
}

// uint128_from_64 converts v to a Uint128 value
pub fn uint128_from_64(v u64) Uint128 {
	return uint128_new(v, 0)
}

pub fn uint128_new(lo u64, hi u64) Uint128 {
	return Uint128{lo, hi}
}

pub fn uint128_from_dec_str(value string) ?Uint128 {
	mut res := unsigned.uint128_zero
	for b_ in value.bytes() {
		b := b_ - '0'.bytes()[0]
		if b > 9 {
			return error('invalid character "$b"')
		}

		r, overflow := res.overflowing_mul_64(10)

		if overflow {
			return error('invalid length')
		}
		r2, overflow2 := r.overflowing_add_64(u64(b))

		if overflow2 > 0 {
			return error('invalid length')
		}

		res = r2
	}
	return res
}

pub fn (u Uint128) / (v Uint128) Uint128 {
	return u.div(v)
}

pub fn (u Uint128) % (v Uint128) Uint128 {
	return u.mod(v)
}

pub fn (u Uint128) + (v Uint128) Uint128 {
	return u.add(v)
}

pub fn (u Uint128) - (v Uint128) Uint128 {
	return u.sub(v)
}

pub fn (u Uint128) * (v Uint128) Uint128 {
	return u.mul(v)
}

pub fn (u Uint128) < (v Uint128) bool {
	return u.cmp(v) == -1
}