blob: 918ccab29879da5600f49be1e929d6f4b580a83a [file] [log] [blame]
Yi Kongce81bb62020-08-31 01:21:33 +08001// Translated from C to Rust. The original C code can be found at
2// https://github.com/ulfjack/ryu and carries the following license:
3//
4// Copyright 2018 Ulf Adams
5//
6// The contents of this file may be used under the terms of the Apache License,
7// Version 2.0.
8//
9// (See accompanying file LICENSE-Apache or copy at
10// http://www.apache.org/licenses/LICENSE-2.0)
11//
12// Alternatively, the contents of this file may be used under the terms of
13// the Boost Software License, Version 1.0.
14// (See accompanying file LICENSE-Boost or copy at
15// https://www.boost.org/LICENSE_1_0.txt)
16//
17// Unless required by applicable law or agreed to in writing, this software
18// is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
19// KIND, either express or implied.
20
21use core::ptr;
22
23// Returns (lo, hi).
24#[cfg(not(integer128))]
25#[cfg_attr(feature = "no-panic", inline)]
26pub fn umul128(a: u64, b: u64) -> (u64, u64) {
27 let a_lo = a as u32;
28 let a_hi = (a >> 32) as u32;
29 let b_lo = b as u32;
30 let b_hi = (b >> 32) as u32;
31
32 let b00 = a_lo as u64 * b_lo as u64;
33 let b01 = a_lo as u64 * b_hi as u64;
34 let b10 = a_hi as u64 * b_lo as u64;
35 let b11 = a_hi as u64 * b_hi as u64;
36
37 let b00_lo = b00 as u32;
38 let b00_hi = (b00 >> 32) as u32;
39
40 let mid1 = b10 + b00_hi as u64;
41 let mid1_lo = mid1 as u32;
42 let mid1_hi = (mid1 >> 32) as u32;
43
44 let mid2 = b01 + mid1_lo as u64;
45 let mid2_lo = mid2 as u32;
46 let mid2_hi = (mid2 >> 32) as u32;
47
48 let p_hi = b11 + mid1_hi as u64 + mid2_hi as u64;
49 let p_lo = ((mid2_lo as u64) << 32) | b00_lo as u64;
50
51 (p_lo, p_hi)
52}
53
54#[cfg(not(integer128))]
55#[cfg_attr(feature = "no-panic", inline)]
56pub fn shiftright128(lo: u64, hi: u64, dist: u32) -> u64 {
57 // We don't need to handle the case dist >= 64 here (see above).
58 debug_assert!(dist > 0);
59 debug_assert!(dist < 64);
60 (hi << (64 - dist)) | (lo >> dist)
61}
62
63#[cfg_attr(feature = "no-panic", inline)]
64pub fn div5(x: u64) -> u64 {
65 x / 5
66}
67
68#[cfg_attr(feature = "no-panic", inline)]
69pub fn div10(x: u64) -> u64 {
70 x / 10
71}
72
73#[cfg_attr(feature = "no-panic", inline)]
74pub fn div100(x: u64) -> u64 {
75 x / 100
76}
77
78#[cfg_attr(feature = "no-panic", inline)]
79fn pow5_factor(mut value: u64) -> u32 {
80 let mut count = 0u32;
81 loop {
82 debug_assert!(value != 0);
83 let q = div5(value);
84 let r = (value as u32).wrapping_sub(5u32.wrapping_mul(q as u32));
85 if r != 0 {
86 break;
87 }
88 value = q;
89 count += 1;
90 }
91 count
92}
93
94// Returns true if value is divisible by 5^p.
95#[cfg_attr(feature = "no-panic", inline)]
96pub fn multiple_of_power_of_5(value: u64, p: u32) -> bool {
97 // I tried a case distinction on p, but there was no performance difference.
98 pow5_factor(value) >= p
99}
100
101// Returns true if value is divisible by 2^p.
102#[cfg_attr(feature = "no-panic", inline)]
103pub fn multiple_of_power_of_2(value: u64, p: u32) -> bool {
104 debug_assert!(value != 0);
105 debug_assert!(p < 64);
106 // __builtin_ctzll doesn't appear to be faster here.
107 (value & ((1u64 << p) - 1)) == 0
108}
109
110#[cfg(integer128)]
111#[cfg_attr(feature = "no-panic", inline)]
112pub fn mul_shift_64(m: u64, mul: &(u64, u64), j: u32) -> u64 {
113 let b0 = m as u128 * mul.0 as u128;
114 let b2 = m as u128 * mul.1 as u128;
115 (((b0 >> 64) + b2) >> (j - 64)) as u64
116}
117
118#[cfg(integer128)]
119#[cfg_attr(feature = "no-panic", inline)]
120pub unsafe fn mul_shift_all_64(
121 m: u64,
122 mul: &(u64, u64),
123 j: u32,
124 vp: *mut u64,
125 vm: *mut u64,
126 mm_shift: u32,
127) -> u64 {
128 ptr::write(vp, mul_shift_64(4 * m + 2, mul, j));
129 ptr::write(vm, mul_shift_64(4 * m - 1 - mm_shift as u64, mul, j));
130 mul_shift_64(4 * m, mul, j)
131}
132
133#[cfg(not(integer128))]
134#[cfg_attr(feature = "no-panic", inline)]
135pub unsafe fn mul_shift_all_64(
136 mut m: u64,
137 mul: &(u64, u64),
138 j: u32,
139 vp: *mut u64,
140 vm: *mut u64,
141 mm_shift: u32,
142) -> u64 {
143 m <<= 1;
144 // m is maximum 55 bits
145 let (lo, tmp) = umul128(m, mul.0);
146 let (mut mid, mut hi) = umul128(m, mul.1);
147 mid = mid.wrapping_add(tmp);
148 hi = hi.wrapping_add((mid < tmp) as u64); // overflow into hi
149
150 let lo2 = lo.wrapping_add(mul.0);
151 let mid2 = mid.wrapping_add(mul.1).wrapping_add((lo2 < lo) as u64);
152 let hi2 = hi.wrapping_add((mid2 < mid) as u64);
153 ptr::write(vp, shiftright128(mid2, hi2, j - 64 - 1));
154
155 if mm_shift == 1 {
156 let lo3 = lo.wrapping_sub(mul.0);
157 let mid3 = mid.wrapping_sub(mul.1).wrapping_sub((lo3 > lo) as u64);
158 let hi3 = hi.wrapping_sub((mid3 > mid) as u64);
159 ptr::write(vm, shiftright128(mid3, hi3, j - 64 - 1));
160 } else {
161 let lo3 = lo + lo;
162 let mid3 = mid.wrapping_add(mid).wrapping_add((lo3 < lo) as u64);
163 let hi3 = hi.wrapping_add(hi).wrapping_add((mid3 < mid) as u64);
164 let lo4 = lo3.wrapping_sub(mul.0);
165 let mid4 = mid3.wrapping_sub(mul.1).wrapping_sub((lo4 > lo3) as u64);
166 let hi4 = hi3.wrapping_sub((mid4 > mid3) as u64);
167 ptr::write(vm, shiftright128(mid4, hi4, j - 64));
168 }
169
170 shiftright128(mid, hi, j - 64 - 1)
171}