1
0
mirror of https://github.com/vlang/v.git synced 2023-08-10 21:13:21 +03:00
v/vlib/strconv/utilities.v

428 lines
8.0 KiB
V
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

module strconv
/*
f32/f64 to string utilities
Copyright (c) 2019-2021 Dario Deledda. All rights reserved.
Use of this source code is governed by an MIT license
that can be found in the LICENSE file.
This file contains the f32/f64 to string utilities functions
These functions are based on the work of:
Publication:PLDI 2018: Proceedings of the 39th ACM SIGPLAN
Conference on Programming Language Design and ImplementationJune 2018
Pages 270282 https://doi.org/10.1145/3192366.3192369
inspired by the Go version here:
https://github.com/cespare/ryu/tree/ba56a33f39e3bbbfa409095d0f9ae168a595feea
*/
import math.bits
//import math
/*
General Utilities
*/
fn assert1(t bool, msg string) {
if !t {
panic(msg)
}
}
[inline]
fn bool_to_int(b bool) int {
if b {
return 1
}
return 0
}
[inline]
fn bool_to_u32(b bool) u32 {
if b {
return u32(1)
}
return u32(0)
}
[inline]
fn bool_to_u64(b bool) u64 {
if b {
return u64(1)
}
return u64(0)
}
fn get_string_special(neg bool, expZero bool, mantZero bool) string {
if !mantZero {
return "nan"
}
if !expZero {
if neg {
return "-inf"
} else {
return "+inf"
}
}
if neg {
return "-0e+00"
}
return "0e+00"
}
/*
32 bit functions
*/
fn decimal_len_32(u u32) int {
// Function precondition: u is not a 10-digit number.
// (9 digits are sufficient for round-tripping.)
// This benchmarked faster than the log2 approach used for u64.
assert1(u < 1000000000, "too big")
if u >= 100000000 { return 9 }
else if u >= 10000000 { return 8 }
else if u >= 1000000 { return 7 }
else if u >= 100000 { return 6 }
else if u >= 10000 { return 5 }
else if u >= 1000 { return 4 }
else if u >= 100 { return 3 }
else if u >= 10 { return 2 }
return 1
}
fn mul_shift_32(m u32, mul u64, ishift int) u32 {
// QTODO
//assert ishift > 32
hi, lo := bits.mul_64(u64(m), mul)
shifted_sum := (lo >> u64(ishift)) + (hi << u64(64-ishift))
assert1(shifted_sum <= 2147483647, "shiftedSum <= math.max_u32")
return u32(shifted_sum)
}
fn mul_pow5_invdiv_pow2(m u32, q u32, j int) u32 {
return mul_shift_32(m, pow5_inv_split_32[q], j)
}
fn mul_pow5_div_pow2(m u32, i u32, j int) u32 {
return mul_shift_32(m, pow5_split_32[i], j)
}
fn pow5_factor_32(i_v u32) u32 {
mut v := i_v
for n := u32(0); ; n++ {
q := v/5
r := v%5
if r != 0 {
return n
}
v = q
}
return v
}
// multiple_of_power_of_five_32 reports whether v is divisible by 5^p.
fn multiple_of_power_of_five_32(v u32, p u32) bool {
return pow5_factor_32(v) >= p
}
// multiple_of_power_of_two_32 reports whether v is divisible by 2^p.
fn multiple_of_power_of_two_32(v u32, p u32) bool {
return u32(bits.trailing_zeros_32(v)) >= p
}
// log10_pow2 returns floor(log_10(2^e)).
fn log10_pow2(e int) u32 {
// The first value this approximation fails for is 2^1651
// which is just greater than 10^297.
assert1(e >= 0, "e >= 0")
assert1(e <= 1650, "e <= 1650")
return (u32(e) * 78913) >> 18
}
// log10_pow5 returns floor(log_10(5^e)).
fn log10_pow5(e int) u32 {
// The first value this approximation fails for is 5^2621
// which is just greater than 10^1832.
assert1(e >= 0, "e >= 0")
assert1(e <= 2620, "e <= 2620")
return (u32(e) * 732923) >> 20
}
// pow5_bits returns ceil(log_2(5^e)), or else 1 if e==0.
fn pow5_bits(e int) int {
// This approximation works up to the point that the multiplication
// overflows at e = 3529. If the multiplication were done in 64 bits,
// it would fail at 5^4004 which is just greater than 2^9297.
assert1(e >= 0, "e >= 0")
assert1(e <= 3528, "e <= 3528")
return int( ((u32(e)*1217359)>>19) + 1)
}
/*
64 bit functions
*/
fn decimal_len_64(u u64) int {
// http://graphics.stanford.edu/~seander/bithacks.html#IntegerLog10
log2 := 64 - bits.leading_zeros_64(u) - 1
t := (log2 + 1) * 1233 >> 12
return t - bool_to_int(u < powers_of_10[t]) + 1
}
fn shift_right_128(v Uint128, shift int) u64 {
// The shift value is always modulo 64.
// In the current implementation of the 64-bit version
// of Ryu, the shift value is always < 64.
// (It is in the range [2, 59].)
// Check this here in case a future change requires larger shift
// values. In this case this function needs to be adjusted.
assert1(shift < 64, "shift < 64")
return (v.hi << u64(64 - shift)) | (v.lo >> u32(shift))
}
fn mul_shift_64(m u64, mul Uint128, shift int) u64 {
hihi, hilo := bits.mul_64(m, mul.hi)
lohi, _ := bits.mul_64(m, mul.lo)
mut sum := Uint128{lo: lohi + hilo,hi: hihi}
if sum.lo < lohi {
sum.hi++ // overflow
}
return shift_right_128(sum, shift-64)
}
fn pow5_factor_64(v_i u64) u32 {
mut v := v_i
for n := u32(0); ; n++ {
q := v/5
r := v%5
if r != 0 {
return n
}
v = q
}
return u32(0)
}
fn multiple_of_power_of_five_64(v u64, p u32) bool {
return pow5_factor_64(v) >= p
}
fn multiple_of_power_of_two_64(v u64, p u32) bool {
return u32(bits.trailing_zeros_64(v)) >= p
}
/*
f64 to string with string format
*/
// f32_to_str_l return a string with the f32 converted in a string in decimal notation
pub fn f32_to_str_l(f f64) string {
return f64_to_str_l(f32(f))
}
// f64_to_str_l return a string with the f64 converted in a string in decimal notation
pub fn f64_to_str_l(f f64) string {
s := f64_to_str(f,18)
// check for +inf -inf Nan
if s.len > 2 && (s[0] == `n` || s[1] == `i`) {
return s
}
m_sgn_flag := false
mut sgn := 1
mut b := [26]byte{}
mut d_pos := 1
mut i := 0
mut i1 := 0
mut exp := 0
mut exp_sgn := 1
// get sign and decimal parts
for c in s {
if c == `-` {
sgn = -1
i++
} else if c == `+` {
sgn = 1
i++
}
else if c >= `0` && c <= `9` {
b[i1] = c
i1++
i++
} else if c == `.` {
if sgn > 0 {
d_pos = i
} else {
d_pos = i-1
}
i++
} else if c == `e` {
i++
break
} else {
return "Float conversion error!!"
}
}
b[i1] = 0
// get exponent
if s[i] == `-` {
exp_sgn = -1
i++
} else if s[i] == `+` {
exp_sgn = 1
i++
}
for c in s[i..] {
exp = exp * 10 + int(c-`0`)
}
// allocate exp+32 chars for the return string
mut res := []byte{len: exp+32, init: 0}
mut r_i := 0 // result string buffer index
//println("s:${sgn} b:${b[0]} es:${exp_sgn} exp:${exp}")
if sgn == 1 {
if m_sgn_flag {
res[r_i] = `+`
r_i++
}
} else {
res[r_i] = `-`
r_i++
}
i = 0
if exp_sgn >= 0 {
for b[i] != 0 {
res[r_i] = b[i]
r_i++
i++
if i >= d_pos && exp >= 0 {
if exp == 0 {
res[r_i] = `.`
r_i++
}
exp--
}
}
for exp >= 0 {
res[r_i] = `0`
r_i++
exp--
}
} else {
mut dot_p := true
for exp > 0 {
res[r_i] = `0`
r_i++
exp--
if dot_p {
res[r_i] = `.`
r_i++
dot_p = false
}
}
for b[i] != 0 {
res[r_i] = b[i]
r_i++
i++
}
}
res[r_i] = 0
return unsafe { tos(res.data,r_i) }
}
// dec_digits return the number of decimal digit of an u64
pub fn dec_digits(n u64) int {
if n <= 9_999_999_999 { // 1-10
if n <= 99_999 { // 5
if n <= 99 { // 2
if n <= 9 { // 1
return 1
} else {
return 2
}
} else {
if n <= 999 { // 3
return 3
} else {
if n <= 9999 { // 4
return 4
} else {
return 5
}
}
}
} else {
if n <= 9_999_999 { // 7
if n <= 999_999 { // 6
return 6
} else {
return 7
}
} else {
if n <= 99_999_999 { //8
return 8
} else {
if n <= 999_999_999 { // 9
return 9
}
return 10
}
}
}
} else {
if n <= 999_999_999_999_999 { // 5
if n <= 999_999_999_999 { // 2
if n <= 99_999_999_999 { // 1
return 11
} else {
return 12
}
} else {
if n <= 9_999_999_999_999 { // 3
return 13
} else {
if n <= 99_999_999_999_999 { // 4
return 14
} else {
return 15
}
}
}
} else {
if n <= 99_999_999_999_999_999 { // 7
if n <= 9_999_999_999_999_999 { // 6
return 16
} else {
return 17
}
} else {
if n <= 999_999_999_999_999_999 { //8
return 18
} else {
if n <= 9_999_999_999_999_999_999 { // 9
return 19
}
return 20
}
}
}
}
}