From 2085a4af5654a74f2a5d8ac1fa7934a3663bf3a0 Mon Sep 17 00:00:00 2001 From: Marc Tiehuis Date: Mon, 2 May 2022 22:45:06 +1200 Subject: [PATCH] add new float-parser based on eisel-lemire algorithm The previous float-parsing method was lacking in a lot of areas. This commit introduces a state-of-the art implementation that is both accurate and fast to std. Code is derived from working repo https://github.com/tiehuis/zig-parsefloat. This includes more test-cases and performance numbers that are present in this commit. * Accuracy The primary testing regime has been using test-data found at https://github.com/tiehuis/parse-number-fxx-test-data. This is a fork of upstream with support for f128 test-cases added. This data has been verified against other independent implementations and represents accurate round-to-even IEEE-754 floating point semantics. * Performance Compared to the existing parseFloat implementation there is ~5-10x performance improvement using the above corpus. (f128 parsing excluded in below measurements). ** Old $ time ./test_all_fxx_data 3520298/5296694 succeeded (1776396 fail) ________________________________________________________ Executed in 28.68 secs fish external usr time 28.48 secs 0.00 micros 28.48 secs sys time 0.08 secs 694.00 micros 0.08 secs ** This Implementation $ time ./test_all_fxx_data 5296693/5296694 succeeded (1 fail) ________________________________________________________ Executed in 4.54 secs fish external usr time 4.37 secs 515.00 micros 4.37 secs sys time 0.10 secs 171.00 micros 0.10 secs Further performance numbers can be seen using the https://github.com/tiehuis/simple_fastfloat_benchmark/ repository, which compares against some other well-known string-to-float conversion functions. A breakdown can be found here: https://github.com/tiehuis/zig-parsefloat/blob/0d9f020f1a37ca88bf889703b397c1c41779f090/PERFORMANCE.md#commit-b15406a0d2e18b50a4b62fceb5a6a3bb60ca5706 In summary, we are within 20% of the C++ reference implementation and have about ~600-700MB/s throughput on a Intel I5-6500 3.5Ghz. * F128 Support Finally, f128 is now completely supported with full accuracy. This does use a slower path which is possible to improve in future. * Behavioural Changes There are a few behavioural changes to note. - `parseHexFloat` is now redundant and these are now supported directly in `parseFloat`. - We implement round-to-even in all parsing routines. This is as specified by IEEE-754. Previous code used different rounding mechanisms (standard was round-to-zero, hex-parsing looked to use round-up) so there may be subtle differences. Closes #2207. Fixes #11169. --- lib/std/fmt.zig | 2 +- lib/std/fmt/parse_float.zig | 499 +++-------- lib/std/fmt/parse_float/FloatInfo.zig | 131 +++ lib/std/fmt/parse_float/FloatStream.zig | 137 +++ lib/std/fmt/parse_float/common.zig | 91 ++ .../fmt/parse_float/convert_eisel_lemire.zig | 843 ++++++++++++++++++ lib/std/fmt/parse_float/convert_fast.zig | 130 +++ lib/std/fmt/parse_float/convert_hex.zig | 89 ++ lib/std/fmt/parse_float/convert_slow.zig | 114 +++ lib/std/fmt/parse_float/decimal.zig | 493 ++++++++++ lib/std/fmt/parse_float/parse.zig | 293 ++++++ lib/std/fmt/parse_float/parse_float.zig | 64 ++ lib/std/fmt/parse_hex_float.zig | 347 ------- 13 files changed, 2497 insertions(+), 736 deletions(-) create mode 100644 lib/std/fmt/parse_float/FloatInfo.zig create mode 100644 lib/std/fmt/parse_float/FloatStream.zig create mode 100644 lib/std/fmt/parse_float/common.zig create mode 100644 lib/std/fmt/parse_float/convert_eisel_lemire.zig create mode 100644 lib/std/fmt/parse_float/convert_fast.zig create mode 100644 lib/std/fmt/parse_float/convert_hex.zig create mode 100644 lib/std/fmt/parse_float/convert_slow.zig create mode 100644 lib/std/fmt/parse_float/decimal.zig create mode 100644 lib/std/fmt/parse_float/parse.zig create mode 100644 lib/std/fmt/parse_float/parse_float.zig delete mode 100644 lib/std/fmt/parse_hex_float.zig diff --git a/lib/std/fmt.zig b/lib/std/fmt.zig index 6fa8cfdfa3..96ed842154 100644 --- a/lib/std/fmt.zig +++ b/lib/std/fmt.zig @@ -1837,8 +1837,8 @@ test "parseUnsigned" { } pub const parseFloat = @import("fmt/parse_float.zig").parseFloat; +pub const parseHexFloat = @compileError("deprecated; use `parseFloat`"); pub const ParseFloatError = @import("fmt/parse_float.zig").ParseFloatError; -pub const parseHexFloat = @import("fmt/parse_hex_float.zig").parseHexFloat; test { _ = parseFloat; diff --git a/lib/std/fmt/parse_float.zig b/lib/std/fmt/parse_float.zig index 585c23ed54..44b6fb2fc0 100644 --- a/lib/std/fmt/parse_float.zig +++ b/lib/std/fmt/parse_float.zig @@ -1,386 +1,18 @@ -// Adapted from https://github.com/grzegorz-kraszewski/stringtofloat. - -// MIT License -// -// Copyright (c) 2016 Grzegorz Kraszewski -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE -// SOFTWARE. -// - -// Be aware that this implementation has the following limitations: -// -// - Is not round-trip accurate for all values -// - Only supports round-to-zero -// - Does not handle denormals +pub const parseFloat = @import("parse_float/parse_float.zig").parseFloat; +pub const ParseFloatError = @import("parse_float/parse_float.zig").ParseFloatError; const std = @import("std"); -const ascii = std.ascii; +const math = std.math; +const testing = std.testing; +const expect = testing.expect; +const expectEqual = testing.expectEqual; +const expectError = testing.expectError; +const approxEqAbs = std.math.approxEqAbs; +const epsilon = 1e-7; -// The mantissa field in FloatRepr is 64bit wide and holds only 19 digits -// without overflowing -const max_digits = 19; - -const f64_plus_zero: u64 = 0x0000000000000000; -const f64_minus_zero: u64 = 0x8000000000000000; -const f64_plus_infinity: u64 = 0x7FF0000000000000; -const f64_minus_infinity: u64 = 0xFFF0000000000000; - -const Z96 = struct { - d0: u32, - d1: u32, - d2: u32, - - // d = s >> 1 - inline fn shiftRight1(d: *Z96, s: Z96) void { - d.d0 = (s.d0 >> 1) | ((s.d1 & 1) << 31); - d.d1 = (s.d1 >> 1) | ((s.d2 & 1) << 31); - d.d2 = s.d2 >> 1; - } - - // d = s << 1 - inline fn shiftLeft1(d: *Z96, s: Z96) void { - d.d2 = (s.d2 << 1) | ((s.d1 & (1 << 31)) >> 31); - d.d1 = (s.d1 << 1) | ((s.d0 & (1 << 31)) >> 31); - d.d0 = s.d0 << 1; - } - - // d += s - inline fn add(d: *Z96, s: Z96) void { - var w = @as(u64, d.d0) + @as(u64, s.d0); - d.d0 = @truncate(u32, w); - - w >>= 32; - w += @as(u64, d.d1) + @as(u64, s.d1); - d.d1 = @truncate(u32, w); - - w >>= 32; - w += @as(u64, d.d2) + @as(u64, s.d2); - d.d2 = @truncate(u32, w); - } - - // d -= s - inline fn sub(d: *Z96, s: Z96) void { - var w = @as(u64, d.d0) -% @as(u64, s.d0); - d.d0 = @truncate(u32, w); - - w >>= 32; - w += @as(u64, d.d1) -% @as(u64, s.d1); - d.d1 = @truncate(u32, w); - - w >>= 32; - w += @as(u64, d.d2) -% @as(u64, s.d2); - d.d2 = @truncate(u32, w); - } -}; - -const FloatRepr = struct { - negative: bool, - exponent: i32, - mantissa: u64, -}; - -fn convertRepr(comptime T: type, n: FloatRepr) T { - const mask28: u32 = 0xf << 28; - - var s: Z96 = undefined; - var q: Z96 = undefined; - var r: Z96 = undefined; - - s.d0 = @truncate(u32, n.mantissa); - s.d1 = @truncate(u32, n.mantissa >> 32); - s.d2 = 0; - - var binary_exponent: i32 = 92; - var exp = n.exponent; - - while (exp > 0) : (exp -= 1) { - q.shiftLeft1(s); // q = p << 1 - r.shiftLeft1(q); // r = p << 2 - s.shiftLeft1(r); // p = p << 3 - s.add(q); // p = (p << 3) + (p << 1) - - while (s.d2 & mask28 != 0) { - q.shiftRight1(s); - binary_exponent += 1; - s = q; - } - } - - while (exp < 0) { - while (s.d2 & (1 << 31) == 0) { - q.shiftLeft1(s); - binary_exponent -= 1; - s = q; - } - - q.d2 = s.d2 / 10; - r.d1 = s.d2 % 10; - r.d2 = (s.d1 >> 8) | (r.d1 << 24); - q.d1 = r.d2 / 10; - r.d1 = r.d2 % 10; - r.d2 = ((s.d1 & 0xff) << 16) | (s.d0 >> 16) | (r.d1 << 24); - r.d0 = r.d2 / 10; - r.d1 = r.d2 % 10; - q.d1 = (q.d1 << 8) | ((r.d0 & 0x00ff0000) >> 16); - q.d0 = r.d0 << 16; - r.d2 = (s.d0 *% 0xffff) | (r.d1 << 16); - q.d0 |= r.d2 / 10; - s = q; - - exp += 1; - } - - if (s.d0 != 0 or s.d1 != 0 or s.d2 != 0) { - while (s.d2 & mask28 == 0) { - q.shiftLeft1(s); - binary_exponent -= 1; - s = q; - } - } - - binary_exponent += 1023; - - const repr: u64 = blk: { - if (binary_exponent > 2046) { - break :blk if (n.negative) f64_minus_infinity else f64_plus_infinity; - } else if (binary_exponent < 1) { - break :blk if (n.negative) f64_minus_zero else f64_plus_zero; - } else if (s.d2 != 0) { - const binexs2 = @intCast(u64, binary_exponent) << 52; - const rr = (@as(u64, s.d2 & ~mask28) << 24) | ((@as(u64, s.d1) + 128) >> 8) | binexs2; - break :blk if (n.negative) rr | (1 << 63) else rr; - } else { - break :blk 0; - } - }; - - const f = @bitCast(f64, repr); - return @floatCast(T, f); -} - -const State = enum { - MaybeSign, - LeadingMantissaZeros, - LeadingFractionalZeros, - MantissaIntegral, - MantissaFractional, - ExponentSign, - LeadingExponentZeros, - Exponent, -}; - -const ParseResult = enum { - Ok, - PlusZero, - MinusZero, - PlusInf, - MinusInf, -}; - -fn parseRepr(s: []const u8, n: *FloatRepr) !ParseResult { - var digit_index: usize = 0; - var negative_exp = false; - var exponent: i32 = 0; - - var state = State.MaybeSign; - - var i: usize = 0; - while (i < s.len) { - const c = s[i]; - - switch (state) { - .MaybeSign => { - state = .LeadingMantissaZeros; - - if (c == '+') { - i += 1; - } else if (c == '-') { - n.negative = true; - i += 1; - } else if (ascii.isDigit(c) or c == '.') { - // continue - } else { - return error.InvalidCharacter; - } - }, - .LeadingMantissaZeros => { - if (c == '0') { - i += 1; - } else if (c == '.') { - i += 1; - state = .LeadingFractionalZeros; - } else if (c == '_') { - i += 1; - } else { - state = .MantissaIntegral; - } - }, - .LeadingFractionalZeros => { - if (c == '0') { - i += 1; - if (n.exponent > std.math.minInt(i32)) { - n.exponent -= 1; - } - } else { - state = .MantissaFractional; - } - }, - .MantissaIntegral => { - if (ascii.isDigit(c)) { - if (digit_index < max_digits) { - n.mantissa *%= 10; - n.mantissa += c - '0'; - digit_index += 1; - } else if (n.exponent < std.math.maxInt(i32)) { - n.exponent += 1; - } - - i += 1; - } else if (c == '.') { - i += 1; - state = .MantissaFractional; - } else if (c == '_') { - i += 1; - } else { - state = .MantissaFractional; - } - }, - .MantissaFractional => { - if (ascii.isDigit(c)) { - if (digit_index < max_digits) { - n.mantissa *%= 10; - n.mantissa += c - '0'; - n.exponent -%= 1; - digit_index += 1; - } - - i += 1; - } else if (c == 'e' or c == 'E') { - i += 1; - state = .ExponentSign; - } else if (c == '_') { - i += 1; - } else { - state = .ExponentSign; - } - }, - .ExponentSign => { - if (c == '+') { - i += 1; - } else if (c == '_') { - return error.InvalidCharacter; - } else if (c == '-') { - negative_exp = true; - i += 1; - } - - state = .LeadingExponentZeros; - }, - .LeadingExponentZeros => { - if (c == '0') { - i += 1; - } else if (c == '_') { - i += 1; - } else { - state = .Exponent; - } - }, - .Exponent => { - if (ascii.isDigit(c)) { - if (exponent < std.math.maxInt(i32) / 10) { - exponent *= 10; - exponent += @intCast(i32, c - '0'); - } - - i += 1; - } else if (c == '_') { - i += 1; - } else { - return error.InvalidCharacter; - } - }, - } - } - - if (negative_exp) exponent = -exponent; - n.exponent += exponent; - - if (n.mantissa == 0) { - return if (n.negative) .MinusZero else .PlusZero; - } else if (n.exponent > 309) { - return if (n.negative) .MinusInf else .PlusInf; - } else if (n.exponent < -328) { - return if (n.negative) .MinusZero else .PlusZero; - } - - return .Ok; -} - -fn caseInEql(a: []const u8, b: []const u8) bool { - if (a.len != b.len) return false; - - for (a) |_, i| { - if (ascii.toUpper(a[i]) != ascii.toUpper(b[i])) { - return false; - } - } - - return true; -} - -pub const ParseFloatError = error{InvalidCharacter}; - -pub fn parseFloat(comptime T: type, s: []const u8) ParseFloatError!T { - if (s.len == 0 or (s.len == 1 and (s[0] == '+' or s[0] == '-'))) { - return error.InvalidCharacter; - } - - if (caseInEql(s, "nan")) { - return std.math.nan(T); - } else if (caseInEql(s, "inf") or caseInEql(s, "+inf")) { - return std.math.inf(T); - } else if (caseInEql(s, "-inf")) { - return -std.math.inf(T); - } - - var r = FloatRepr{ - .negative = false, - .exponent = 0, - .mantissa = 0, - }; - - return switch (try parseRepr(s, &r)) { - .Ok => convertRepr(T, r), - .PlusZero => 0.0, - .MinusZero => -@as(T, 0.0), - .PlusInf => std.math.inf(T), - .MinusInf => -std.math.inf(T), - }; -} +// See https://github.com/tiehuis/parse-number-fxx-test-data for a wider-selection of test-data. test "fmt.parseFloat" { - const testing = std.testing; - const expect = testing.expect; - const expectEqual = testing.expectEqual; - const approxEqAbs = std.math.approxEqAbs; - const epsilon = 1e-7; - inline for ([_]type{ f16, f32, f64, f128 }) |T| { const Z = std.meta.Int(.unsigned, @typeInfo(T).Float.bits); @@ -405,8 +37,8 @@ test "fmt.parseFloat" { try expect(approxEqAbs(T, try parseFloat(T, "3.141"), 3.141, epsilon)); try expect(approxEqAbs(T, try parseFloat(T, "-3.141"), -3.141, epsilon)); - try expectEqual(try parseFloat(T, "1e-700"), 0); - try expectEqual(try parseFloat(T, "1e+700"), std.math.inf(T)); + try expectEqual(try parseFloat(T, "1e-5000"), 0); + try expectEqual(try parseFloat(T, "1e+5000"), std.math.inf(T)); try expectEqual(@bitCast(Z, try parseFloat(T, "nAn")), @bitCast(Z, std.math.nan(T))); try expectEqual(try parseFloat(T, "inF"), std.math.inf(T)); @@ -415,14 +47,105 @@ test "fmt.parseFloat" { try expectEqual(try parseFloat(T, "0.4e0066999999999999999999999999999999999999999999999999999"), std.math.inf(T)); try expect(approxEqAbs(T, try parseFloat(T, "0_1_2_3_4_5_6.7_8_9_0_0_0e0_0_1_0"), @as(T, 123456.789000e10), epsilon)); - if (T != f16) { - try expect(approxEqAbs(T, try parseFloat(T, "1e-2"), 0.01, epsilon)); - try expect(approxEqAbs(T, try parseFloat(T, "1234e-2"), 12.34, epsilon)); + // underscore rule is simple and reduces to "can only occur between two digits" and multiple are not supported. + try expectError(error.InvalidCharacter, parseFloat(T, "0123456.789000e_0010")); // cannot occur immediately after exponent + try expectError(error.InvalidCharacter, parseFloat(T, "_0123456.789000e0010")); // cannot occur before any digits + try expectError(error.InvalidCharacter, parseFloat(T, "0__123456.789000e_0010")); // cannot occur twice in a row + try expectError(error.InvalidCharacter, parseFloat(T, "0123456_.789000e0010")); // cannot occur before decimal point + try expectError(error.InvalidCharacter, parseFloat(T, "0123456.789000e0010_")); // cannot occur at end of number - try expect(approxEqAbs(T, try parseFloat(T, "123142.1"), 123142.1, epsilon)); - try expect(approxEqAbs(T, try parseFloat(T, "-123142.1124"), @as(T, -123142.1124), epsilon)); - try expect(approxEqAbs(T, try parseFloat(T, "0.7062146892655368"), @as(T, 0.7062146892655368), epsilon)); - try expect(approxEqAbs(T, try parseFloat(T, "2.71828182845904523536"), @as(T, 2.718281828459045), epsilon)); - } + try expect(approxEqAbs(T, try parseFloat(T, "1e-2"), 0.01, epsilon)); + try expect(approxEqAbs(T, try parseFloat(T, "1234e-2"), 12.34, epsilon)); + + try expect(approxEqAbs(T, try parseFloat(T, "123142.1"), 123142.1, epsilon)); + try expect(approxEqAbs(T, try parseFloat(T, "-123142.1124"), @as(T, -123142.1124), epsilon)); + try expect(approxEqAbs(T, try parseFloat(T, "0.7062146892655368"), @as(T, 0.7062146892655368), epsilon)); + try expect(approxEqAbs(T, try parseFloat(T, "2.71828182845904523536"), @as(T, 2.718281828459045), epsilon)); } } + +test "fmt.parseFloat #11169" { + try expectEqual(try parseFloat(f128, "9007199254740993.0"), 9007199254740993.0); +} + +test "fmt.parseFloat hex.special" { + try testing.expect(math.isNan(try parseFloat(f32, "nAn"))); + try testing.expect(math.isPositiveInf(try parseFloat(f32, "iNf"))); + try testing.expect(math.isPositiveInf(try parseFloat(f32, "+Inf"))); + try testing.expect(math.isNegativeInf(try parseFloat(f32, "-iNf"))); +} +test "fmt.parseFloat hex.zero" { + try testing.expectEqual(@as(f32, 0.0), try parseFloat(f32, "0x0")); + try testing.expectEqual(@as(f32, 0.0), try parseFloat(f32, "-0x0")); + try testing.expectEqual(@as(f32, 0.0), try parseFloat(f32, "0x0p42")); + try testing.expectEqual(@as(f32, 0.0), try parseFloat(f32, "-0x0.00000p42")); + try testing.expectEqual(@as(f32, 0.0), try parseFloat(f32, "0x0.00000p666")); +} + +test "fmt.parseFloat hex.f16" { + try testing.expectEqual(try parseFloat(f16, "0x1p0"), 1.0); + try testing.expectEqual(try parseFloat(f16, "-0x1p-1"), -0.5); + try testing.expectEqual(try parseFloat(f16, "0x10p+10"), 16384.0); + try testing.expectEqual(try parseFloat(f16, "0x10p-10"), 0.015625); + // Max normalized value. + try testing.expectEqual(try parseFloat(f16, "0x1.ffcp+15"), math.floatMax(f16)); + try testing.expectEqual(try parseFloat(f16, "-0x1.ffcp+15"), -math.floatMax(f16)); + // Min normalized value. + try testing.expectEqual(try parseFloat(f16, "0x1p-14"), math.floatMin(f16)); + try testing.expectEqual(try parseFloat(f16, "-0x1p-14"), -math.floatMin(f16)); + // Min denormal value. + try testing.expectEqual(try parseFloat(f16, "0x1p-24"), math.floatTrueMin(f16)); + try testing.expectEqual(try parseFloat(f16, "-0x1p-24"), -math.floatTrueMin(f16)); +} + +test "fmt.parseFloat hex.f32" { + try testing.expectEqual(try parseFloat(f32, "0x1p0"), 1.0); + try testing.expectEqual(try parseFloat(f32, "-0x1p-1"), -0.5); + try testing.expectEqual(try parseFloat(f32, "0x10p+10"), 16384.0); + try testing.expectEqual(try parseFloat(f32, "0x10p-10"), 0.015625); + try testing.expectEqual(try parseFloat(f32, "0x0.ffffffp128"), 0x0.ffffffp128); + try testing.expectEqual(try parseFloat(f32, "0x0.1234570p-125"), 0x0.1234570p-125); + // Max normalized value. + try testing.expectEqual(try parseFloat(f32, "0x1.fffffeP+127"), math.floatMax(f32)); + try testing.expectEqual(try parseFloat(f32, "-0x1.fffffeP+127"), -math.floatMax(f32)); + // Min normalized value. + try testing.expectEqual(try parseFloat(f32, "0x1p-126"), math.floatMin(f32)); + try testing.expectEqual(try parseFloat(f32, "-0x1p-126"), -math.floatMin(f32)); + // Min denormal value. + try testing.expectEqual(try parseFloat(f32, "0x1P-149"), math.floatTrueMin(f32)); + try testing.expectEqual(try parseFloat(f32, "-0x1P-149"), -math.floatTrueMin(f32)); +} + +test "fmt.parseFloat hex.f64" { + try testing.expectEqual(try parseFloat(f64, "0x1p0"), 1.0); + try testing.expectEqual(try parseFloat(f64, "-0x1p-1"), -0.5); + try testing.expectEqual(try parseFloat(f64, "0x10p+10"), 16384.0); + try testing.expectEqual(try parseFloat(f64, "0x10p-10"), 0.015625); + // Max normalized value. + try testing.expectEqual(try parseFloat(f64, "0x1.fffffffffffffp+1023"), math.floatMax(f64)); + try testing.expectEqual(try parseFloat(f64, "-0x1.fffffffffffffp1023"), -math.floatMax(f64)); + // Min normalized value. + try testing.expectEqual(try parseFloat(f64, "0x1p-1022"), math.floatMin(f64)); + try testing.expectEqual(try parseFloat(f64, "-0x1p-1022"), -math.floatMin(f64)); + // Min denormalized value. + //try testing.expectEqual(try parseFloat(f64, "0x1p-1074"), math.floatTrueMin(f64)); + try testing.expectEqual(try parseFloat(f64, "-0x1p-1074"), -math.floatTrueMin(f64)); +} +test "fmt.parseFloat hex.f128" { + try testing.expectEqual(try parseFloat(f128, "0x1p0"), 1.0); + try testing.expectEqual(try parseFloat(f128, "-0x1p-1"), -0.5); + try testing.expectEqual(try parseFloat(f128, "0x10p+10"), 16384.0); + try testing.expectEqual(try parseFloat(f128, "0x10p-10"), 0.015625); + // Max normalized value. + try testing.expectEqual(try parseFloat(f128, "0xf.fffffffffffffffffffffffffff8p+16380"), math.floatMax(f128)); + try testing.expectEqual(try parseFloat(f128, "-0xf.fffffffffffffffffffffffffff8p+16380"), -math.floatMax(f128)); + // Min normalized value. + try testing.expectEqual(try parseFloat(f128, "0x1p-16382"), math.floatMin(f128)); + try testing.expectEqual(try parseFloat(f128, "-0x1p-16382"), -math.floatMin(f128)); + // // Min denormalized value. + try testing.expectEqual(try parseFloat(f128, "0x1p-16494"), math.floatTrueMin(f128)); + try testing.expectEqual(try parseFloat(f128, "-0x1p-16494"), -math.floatTrueMin(f128)); + + // NOTE: We are performing round-to-even. Previous behavior was round-up. + // try testing.expectEqual(try parseFloat(f128, "0x1.edcb34a235253948765432134674fp-1"), 0x1.edcb34a235253948765432134674fp-1); +} diff --git a/lib/std/fmt/parse_float/FloatInfo.zig b/lib/std/fmt/parse_float/FloatInfo.zig new file mode 100644 index 0000000000..8fd95963b1 --- /dev/null +++ b/lib/std/fmt/parse_float/FloatInfo.zig @@ -0,0 +1,131 @@ +const std = @import("std"); +const Self = @This(); + +// Minimum exponent that for a fast path case, or `-⌊(MANTISSA_EXPLICIT_BITS+1)/log2(5)⌋` +min_exponent_fast_path: comptime_int, + +// Maximum exponent that for a fast path case, or `⌊(MANTISSA_EXPLICIT_BITS+1)/log2(5)⌋` +max_exponent_fast_path: comptime_int, + +// Maximum exponent that can be represented for a disguised-fast path case. +// This is `MAX_EXPONENT_FAST_PATH + ⌊(MANTISSA_EXPLICIT_BITS+1)/log2(10)⌋` +max_exponent_fast_path_disguised: comptime_int, + +// Maximum mantissa for the fast-path (`1 << 53` for f64). +max_mantissa_fast_path: comptime_int, + +// Smallest decimal exponent for a non-zero value. Including subnormals. +smallest_power_of_ten: comptime_int, + +// Largest decimal exponent for a non-infinite value. +largest_power_of_ten: comptime_int, + +// The number of bits in the significand, *excluding* the hidden bit. +mantissa_explicit_bits: comptime_int, + +// Minimum exponent value `-(1 << (EXP_BITS - 1)) + 1`. +minimum_exponent: comptime_int, + +// Round-to-even only happens for negative values of q +// when q ≥ −4 in the 64-bit case and when q ≥ −17 in +// the 32-bitcase. +// +// When q ≥ 0,we have that 5^q ≤ 2m+1. In the 64-bit case,we +// have 5^q ≤ 2m+1 ≤ 2^54 or q ≤ 23. In the 32-bit case,we have +// 5^q ≤ 2m+1 ≤ 2^25 or q ≤ 10. +// +// When q < 0, we have w ≥ (2m+1)×5^−q. We must have that w < 2^64 +// so (2m+1)×5^−q < 2^64. We have that 2m+1 > 2^53 (64-bit case) +// or 2m+1 > 2^24 (32-bit case). Hence,we must have 2^53×5^−q < 2^64 +// (64-bit) and 2^24×5^−q < 2^64 (32-bit). Hence we have 5^−q < 2^11 +// or q ≥ −4 (64-bit case) and 5^−q < 2^40 or q ≥ −17 (32-bitcase). +// +// Thus we have that we only need to round ties to even when +// we have that q ∈ [−4,23](in the 64-bit case) or q∈[−17,10] +// (in the 32-bit case). In both cases,the power of five(5^|q|) +// fits in a 64-bit word. +min_exponent_round_to_even: comptime_int, +max_exponent_round_to_even: comptime_int, + +// Largest exponent value `(1 << EXP_BITS) - 1`. +infinite_power: comptime_int, + +// Following should compute based on derived calculations where possible. +pub fn from(comptime T: type) Self { + return switch (T) { + f16 => .{ + // Fast-Path + .min_exponent_fast_path = -4, + .max_exponent_fast_path = 4, + .max_exponent_fast_path_disguised = 7, + .max_mantissa_fast_path = 2 << std.math.floatMantissaBits(T), + // Slow + Eisel-Lemire + .mantissa_explicit_bits = std.math.floatMantissaBits(T), + .infinite_power = 0x1f, + // Eisel-Lemire + .smallest_power_of_ten = -26, // TODO: refine, fails one test + .largest_power_of_ten = 4, + .minimum_exponent = -15, + // w >= (2m+1) * 5^-q and w < 2^64 + // => 2m+1 > 2^11 + // => 2^11*5^-q < 2^64 + // => 5^-q < 2^53 + // => q >= -23 + .min_exponent_round_to_even = -22, + .max_exponent_round_to_even = 5, + }, + f32 => .{ + // Fast-Path + .min_exponent_fast_path = -10, + .max_exponent_fast_path = 10, + .max_exponent_fast_path_disguised = 17, + .max_mantissa_fast_path = 2 << std.math.floatMantissaBits(T), + // Slow + Eisel-Lemire + .mantissa_explicit_bits = std.math.floatMantissaBits(T), + .infinite_power = 0xff, + // Eisel-Lemire + .smallest_power_of_ten = -65, + .largest_power_of_ten = 38, + .minimum_exponent = -127, + .min_exponent_round_to_even = -17, + .max_exponent_round_to_even = 10, + }, + f64 => .{ + // Fast-Path + .min_exponent_fast_path = -22, + .max_exponent_fast_path = 22, + .max_exponent_fast_path_disguised = 37, + .max_mantissa_fast_path = 2 << std.math.floatMantissaBits(T), + // Slow + Eisel-Lemire + .mantissa_explicit_bits = std.math.floatMantissaBits(T), + .infinite_power = 0x7ff, + // Eisel-Lemire + .smallest_power_of_ten = -342, + .largest_power_of_ten = 308, + .minimum_exponent = -1023, + .min_exponent_round_to_even = -4, + .max_exponent_round_to_even = 23, + }, + f128 => .{ + // Fast-Path + .min_exponent_fast_path = -48, + .max_exponent_fast_path = 48, + .max_exponent_fast_path_disguised = 82, + .max_mantissa_fast_path = 2 << std.math.floatMantissaBits(T), + // Slow + Eisel-Lemire + .mantissa_explicit_bits = std.math.floatMantissaBits(T), + .infinite_power = 0x7fff, + // Eisel-Lemire. + // NOTE: Not yet tested (no f128 eisel-lemire implementation) + .smallest_power_of_ten = -4966, + .largest_power_of_ten = 4932, + .minimum_exponent = -16382, + // 2^113 * 5^-q < 2^128 + // 5^-q < 2^15 + // => q >= -6 + .min_exponent_round_to_even = -6, + .max_exponent_round_to_even = 49, + }, + else => unreachable, + }; +} diff --git a/lib/std/fmt/parse_float/FloatStream.zig b/lib/std/fmt/parse_float/FloatStream.zig new file mode 100644 index 0000000000..803ac65e6d --- /dev/null +++ b/lib/std/fmt/parse_float/FloatStream.zig @@ -0,0 +1,137 @@ +//! A wrapper over a byte-slice, providing useful methods for parsing string floating point values. + +const std = @import("std"); +const FloatStream = @This(); +const common = @import("common.zig"); + +slice: []const u8, +offset: usize, +underscore_count: usize, + +pub fn init(s: []const u8) FloatStream { + return .{ .slice = s, .offset = 0, .underscore_count = 0 }; +} + +// Returns the offset from the start *excluding* any underscores that were found. +pub fn offsetTrue(self: FloatStream) usize { + return self.offset - self.underscore_count; +} + +pub fn reset(self: *FloatStream) void { + self.offset = 0; + self.underscore_count = 0; +} + +pub fn len(self: FloatStream) usize { + if (self.offset > self.slice.len) { + return 0; + } + return self.slice.len - self.offset; +} + +pub fn hasLen(self: FloatStream, n: usize) bool { + return self.offset + n <= self.slice.len; +} + +pub fn firstUnchecked(self: FloatStream) u8 { + return self.slice[self.offset]; +} + +pub fn first(self: FloatStream) ?u8 { + return if (self.hasLen(1)) + return self.firstUnchecked() + else + null; +} + +pub fn isEmpty(self: FloatStream) bool { + return !self.hasLen(1); +} + +pub fn firstIs(self: FloatStream, c: u8) bool { + if (self.first()) |ok| { + return ok == c; + } + return false; +} + +pub fn firstIsLower(self: FloatStream, c: u8) bool { + if (self.first()) |ok| { + return ok | 0x20 == c; + } + return false; +} + +pub fn firstIs2(self: FloatStream, c1: u8, c2: u8) bool { + if (self.first()) |ok| { + return ok == c1 or ok == c2; + } + return false; +} + +pub fn firstIs3(self: FloatStream, c1: u8, c2: u8, c3: u8) bool { + if (self.first()) |ok| { + return ok == c1 or ok == c2 or ok == c3; + } + return false; +} + +pub fn firstIsDigit(self: FloatStream, comptime base: u8) bool { + comptime std.debug.assert(base == 10 or base == 16); + + if (self.first()) |ok| { + return common.isDigit(ok, base); + } + return false; +} + +pub fn advance(self: *FloatStream, n: usize) void { + self.offset += n; +} + +pub fn skipChars(self: *FloatStream, c: u8) void { + while (self.firstIs(c)) : (self.advance(1)) {} +} + +pub fn skipChars2(self: *FloatStream, c1: u8, c2: u8) void { + while (self.firstIs2(c1, c2)) : (self.advance(1)) {} +} + +pub fn readU64Unchecked(self: FloatStream) u64 { + return std.mem.readIntSliceLittle(u64, self.slice[self.offset..]); +} + +pub fn readU64(self: FloatStream) ?u64 { + if (self.hasLen(8)) { + return self.readU64Unchecked(); + } + return null; +} + +pub fn atUnchecked(self: *FloatStream, i: usize) u8 { + return self.slice[self.offset + i]; +} + +pub fn scanDigit(self: *FloatStream, comptime base: u8) ?u8 { + comptime std.debug.assert(base == 10 or base == 16); + + retry: while (true) { + if (self.first()) |ok| { + if ('0' <= ok and ok <= '9') { + self.advance(1); + return ok - '0'; + } else if (base == 16 and 'a' <= ok and ok <= 'f') { + self.advance(1); + return ok - 'a' + 10; + } else if (base == 16 and 'A' <= ok and ok <= 'F') { + self.advance(1); + return ok - 'A' + 10; + } else if (ok == '_') { + self.advance(1); + self.underscore_count += 1; + continue :retry; + } + } + return null; + } +} diff --git a/lib/std/fmt/parse_float/common.zig b/lib/std/fmt/parse_float/common.zig new file mode 100644 index 0000000000..c1b34a081b --- /dev/null +++ b/lib/std/fmt/parse_float/common.zig @@ -0,0 +1,91 @@ +const std = @import("std"); + +/// A custom N-bit floating point type, representing `f * 2^e`. +/// e is biased, so it be directly shifted into the exponent bits. +/// Negative exponent indicates an invalid result. +pub fn BiasedFp(comptime T: type) type { + const MantissaT = mantissaType(T); + + return struct { + const Self = @This(); + + /// The significant digits. + f: MantissaT, + /// The biased, binary exponent. + e: i32, + + pub fn zero() Self { + return .{ .f = 0, .e = 0 }; + } + + pub fn zeroPow2(e: i32) Self { + return .{ .f = 0, .e = e }; + } + + pub fn inf(comptime FloatT: type) Self { + return .{ .f = 0, .e = (1 << std.math.floatExponentBits(FloatT)) - 1 }; + } + + pub fn eql(self: Self, other: Self) bool { + return self.f == other.f and self.e == other.e; + } + + pub fn toFloat(self: Self, comptime FloatT: type, negative: bool) FloatT { + var word = self.f; + word |= @intCast(MantissaT, self.e) << std.math.floatMantissaBits(FloatT); + var f = floatFromUnsigned(FloatT, MantissaT, word); + if (negative) f = -f; + return f; + } + }; +} + +pub fn floatFromUnsigned(comptime T: type, comptime MantissaT: type, v: MantissaT) T { + return switch (T) { + f16 => @bitCast(f16, @truncate(u16, v)), + f32 => @bitCast(f32, @truncate(u32, v)), + f64 => @bitCast(f64, @truncate(u64, v)), + f128 => @bitCast(f128, v), + else => unreachable, + }; +} + +/// Represents a parsed floating point value as its components. +pub fn Number(comptime T: type) type { + return struct { + exponent: i64, + mantissa: mantissaType(T), + negative: bool, + /// More than max_mantissa digits were found during parse + many_digits: bool, + /// The number was a hex-float (e.g. 0x1.234p567) + hex: bool, + }; +} + +/// Determine if 8 bytes are all decimal digits. +/// This does not care about the order in which the bytes were loaded. +pub fn isEightDigits(v: u64) bool { + const a = v +% 0x4646_4646_4646_4646; + const b = v -% 0x3030_3030_3030_3030; + return ((a | b) & 0x8080_8080_8080_8080) == 0; +} + +pub fn isDigit(c: u8, comptime base: u8) bool { + std.debug.assert(base == 10 or base == 16); + + return if (base == 10) + '0' <= c and c <= '9' + else + '0' <= c and c <= '9' or 'a' <= c and c <= 'f' or 'A' <= c and c <= 'F'; +} + +/// Returns the underlying storage type used for the mantissa of floating-point type. +/// The output unsigned type must have at least as many bits as the input floating-point type. +pub fn mantissaType(comptime T: type) type { + return switch (T) { + f16, f32, f64 => u64, + f128 => u128, + else => unreachable, + }; +} diff --git a/lib/std/fmt/parse_float/convert_eisel_lemire.zig b/lib/std/fmt/parse_float/convert_eisel_lemire.zig new file mode 100644 index 0000000000..724be45181 --- /dev/null +++ b/lib/std/fmt/parse_float/convert_eisel_lemire.zig @@ -0,0 +1,843 @@ +const std = @import("std"); +const math = std.math; +const common = @import("common.zig"); +const FloatInfo = @import("FloatInfo.zig"); +const BiasedFp = common.BiasedFp; +const Number = common.Number; + +/// Compute a float using an extended-precision representation. +/// +/// Fast conversion of a the significant digits and decimal exponent +/// a float to an extended representation with a binary float. This +/// algorithm will accurately parse the vast majority of cases, +/// and uses a 128-bit representation (with a fallback 192-bit +/// representation). +/// +/// This algorithm scales the exponent by the decimal exponent +/// using pre-computed powers-of-5, and calculates if the +/// representation can be unambiguously rounded to the nearest +/// machine float. Near-halfway cases are not handled here, +/// and are represented by a negative, biased binary exponent. +/// +/// The algorithm is described in detail in "Daniel Lemire, Number Parsing +/// at a Gigabyte per Second" in section 5, "Fast Algorithm", and +/// section 6, "Exact Numbers And Ties", available online: +/// . +pub fn convertEiselLemire(comptime T: type, q: i64, w_: u64) ?BiasedFp(f64) { + std.debug.assert(T == f16 or T == f32 or T == f64); + var w = w_; + const float_info = FloatInfo.from(T); + + // Short-circuit if the value can only be a literal 0 or infinity. + if (w == 0 or q < float_info.smallest_power_of_ten) { + return BiasedFp(f64).zero(); + } else if (q > float_info.largest_power_of_ten) { + return BiasedFp(f64).inf(T); + } + + // Normalize our significant digits, so the most-significant bit is set. + const lz = @clz(u64, @bitCast(u64, w)); + w = math.shl(u64, w, lz); + + const r = computeProductApprox(q, w, float_info.mantissa_explicit_bits + 3); + if (r.lo == 0xffff_ffff_ffff_ffff) { + // If we have failed to approximate w x 5^-q with our 128-bit value. + // Since the addition of 1 could lead to an overflow which could then + // round up over the half-way point, this can lead to improper rounding + // of a float. + // + // However, this can only occur if q ∈ [-27, 55]. The upper bound of q + // is 55 because 5^55 < 2^128, however, this can only happen if 5^q > 2^64, + // since otherwise the product can be represented in 64-bits, producing + // an exact result. For negative exponents, rounding-to-even can + // only occur if 5^-q < 2^64. + // + // For detailed explanations of rounding for negative exponents, see + // . For detailed + // explanations of rounding for positive exponents, see + // . + const inside_safe_exponent = q >= -27 and q <= 55; + if (!inside_safe_exponent) { + return null; + } + } + + const upper_bit = @intCast(i32, r.hi >> 63); + var mantissa = math.shr(u64, r.hi, upper_bit + 64 - @intCast(i32, float_info.mantissa_explicit_bits) - 3); + var power2 = power(@intCast(i32, q)) + upper_bit - @intCast(i32, lz) - float_info.minimum_exponent; + if (power2 <= 0) { + if (-power2 + 1 >= 64) { + // Have more than 64 bits below the minimum exponent, must be 0. + return BiasedFp(f64).zero(); + } + // Have a subnormal value. + mantissa = math.shr(u64, mantissa, -power2 + 1); + mantissa += mantissa & 1; + mantissa >>= 1; + power2 = @boolToInt(mantissa >= (1 << float_info.mantissa_explicit_bits)); + return BiasedFp(f64){ .f = mantissa, .e = power2 }; + } + + // Need to handle rounding ties. Normally, we need to round up, + // but if we fall right in between and and we have an even basis, we + // need to round down. + // + // This will only occur if: + // 1. The lower 64 bits of the 128-bit representation is 0. + // IE, 5^q fits in single 64-bit word. + // 2. The least-significant bit prior to truncated mantissa is odd. + // 3. All the bits truncated when shifting to mantissa bits + 1 are 0. + // + // Or, we may fall between two floats: we are exactly halfway. + if (r.lo <= 1 and + q >= float_info.min_exponent_round_to_even and + q <= float_info.max_exponent_round_to_even and + mantissa & 3 == 1 and + math.shl(u64, mantissa, (upper_bit + 64 - @intCast(i32, float_info.mantissa_explicit_bits) - 3)) == r.hi) + { + // Zero the lowest bit, so we don't round up. + mantissa &= ~@as(u64, 1); + } + + // Round-to-even, then shift the significant digits into place. + mantissa += mantissa & 1; + mantissa >>= 1; + if (mantissa >= 2 << float_info.mantissa_explicit_bits) { + // Rounding up overflowed, so the carry bit is set. Set the + // mantissa to 1 (only the implicit, hidden bit is set) and + // increase the exponent. + mantissa = 1 << float_info.mantissa_explicit_bits; + power2 += 1; + } + + // Zero out the hidden bit + mantissa &= ~(@as(u64, 1) << float_info.mantissa_explicit_bits); + if (power2 >= float_info.infinite_power) { + // Exponent is above largest normal value, must be infinite + return BiasedFp(f64).inf(T); + } + + return BiasedFp(f64){ .f = mantissa, .e = power2 }; +} + +/// Calculate a base 2 exponent from a decimal exponent. +/// This uses a pre-computed integer approximation for +/// log2(10), where 217706 / 2^16 is accurate for the +/// entire range of non-finite decimal exponents. +fn power(q: i32) i32 { + return ((q *% (152170 + 65536)) >> 16) + 63; +} + +const U128 = struct { + lo: u64, + hi: u64, + + pub fn new(lo: u64, hi: u64) U128 { + return .{ .lo = lo, .hi = hi }; + } + + pub fn mul(a: u64, b: u64) U128 { + const x = @as(u128, a) * b; + return .{ + .hi = @truncate(u64, x >> 64), + .lo = @truncate(u64, x), + }; + } +}; + +// This will compute or rather approximate w * 5**q and return a pair of 64-bit words +// approximating the result, with the "high" part corresponding to the most significant +// bits and the low part corresponding to the least significant bits. +fn computeProductApprox(q: i64, w: u64, comptime precision: usize) U128 { + std.debug.assert(q >= eisel_lemire_smallest_power_of_five); + std.debug.assert(q <= eisel_lemire_largest_power_of_five); + std.debug.assert(precision <= 64); + + const mask = if (precision < 64) + 0xffff_ffff_ffff_ffff >> precision + else + 0xffff_ffff_ffff_ffff; + + // 5^q < 2^64, then the multiplication always provides an exact value. + // That means whenever we need to round ties to even, we always have + // an exact value. + const index = @intCast(usize, q - @intCast(i64, eisel_lemire_smallest_power_of_five)); + const pow5 = eisel_lemire_table_powers_of_five_128[index]; + + // Only need one multiplication as long as there is 1 zero but + // in the explicit mantissa bits, +1 for the hidden bit, +1 to + // determine the rounding direction, +1 for if the computed + // product has a leading zero. + var first = U128.mul(w, pow5.lo); + if (first.hi & mask == mask) { + // Need to do a second multiplication to get better precision + // for the lower product. This will always be exact + // where q is < 55, since 5^55 < 2^128. If this wraps, + // then we need to need to round up the hi product. + const second = U128.mul(w, pow5.hi); + + first.lo +%= second.hi; + if (second.hi > first.lo) { + first.hi += 1; + } + } + + return .{ .lo = first.lo, .hi = first.hi }; +} + +// Eisel-Lemire tables ~10Kb +const eisel_lemire_smallest_power_of_five = -342; +const eisel_lemire_largest_power_of_five = 308; +const eisel_lemire_table_powers_of_five_128 = [_]U128{ + U128.new(0xeef453d6923bd65a, 0x113faa2906a13b3f), // 5^-342 + U128.new(0x9558b4661b6565f8, 0x4ac7ca59a424c507), // 5^-341 + U128.new(0xbaaee17fa23ebf76, 0x5d79bcf00d2df649), // 5^-340 + U128.new(0xe95a99df8ace6f53, 0xf4d82c2c107973dc), // 5^-339 + U128.new(0x91d8a02bb6c10594, 0x79071b9b8a4be869), // 5^-338 + U128.new(0xb64ec836a47146f9, 0x9748e2826cdee284), // 5^-337 + U128.new(0xe3e27a444d8d98b7, 0xfd1b1b2308169b25), // 5^-336 + U128.new(0x8e6d8c6ab0787f72, 0xfe30f0f5e50e20f7), // 5^-335 + U128.new(0xb208ef855c969f4f, 0xbdbd2d335e51a935), // 5^-334 + U128.new(0xde8b2b66b3bc4723, 0xad2c788035e61382), // 5^-333 + U128.new(0x8b16fb203055ac76, 0x4c3bcb5021afcc31), // 5^-332 + U128.new(0xaddcb9e83c6b1793, 0xdf4abe242a1bbf3d), // 5^-331 + U128.new(0xd953e8624b85dd78, 0xd71d6dad34a2af0d), // 5^-330 + U128.new(0x87d4713d6f33aa6b, 0x8672648c40e5ad68), // 5^-329 + U128.new(0xa9c98d8ccb009506, 0x680efdaf511f18c2), // 5^-328 + U128.new(0xd43bf0effdc0ba48, 0x212bd1b2566def2), // 5^-327 + U128.new(0x84a57695fe98746d, 0x14bb630f7604b57), // 5^-326 + U128.new(0xa5ced43b7e3e9188, 0x419ea3bd35385e2d), // 5^-325 + U128.new(0xcf42894a5dce35ea, 0x52064cac828675b9), // 5^-324 + U128.new(0x818995ce7aa0e1b2, 0x7343efebd1940993), // 5^-323 + U128.new(0xa1ebfb4219491a1f, 0x1014ebe6c5f90bf8), // 5^-322 + U128.new(0xca66fa129f9b60a6, 0xd41a26e077774ef6), // 5^-321 + U128.new(0xfd00b897478238d0, 0x8920b098955522b4), // 5^-320 + U128.new(0x9e20735e8cb16382, 0x55b46e5f5d5535b0), // 5^-319 + U128.new(0xc5a890362fddbc62, 0xeb2189f734aa831d), // 5^-318 + U128.new(0xf712b443bbd52b7b, 0xa5e9ec7501d523e4), // 5^-317 + U128.new(0x9a6bb0aa55653b2d, 0x47b233c92125366e), // 5^-316 + U128.new(0xc1069cd4eabe89f8, 0x999ec0bb696e840a), // 5^-315 + U128.new(0xf148440a256e2c76, 0xc00670ea43ca250d), // 5^-314 + U128.new(0x96cd2a865764dbca, 0x380406926a5e5728), // 5^-313 + U128.new(0xbc807527ed3e12bc, 0xc605083704f5ecf2), // 5^-312 + U128.new(0xeba09271e88d976b, 0xf7864a44c633682e), // 5^-311 + U128.new(0x93445b8731587ea3, 0x7ab3ee6afbe0211d), // 5^-310 + U128.new(0xb8157268fdae9e4c, 0x5960ea05bad82964), // 5^-309 + U128.new(0xe61acf033d1a45df, 0x6fb92487298e33bd), // 5^-308 + U128.new(0x8fd0c16206306bab, 0xa5d3b6d479f8e056), // 5^-307 + U128.new(0xb3c4f1ba87bc8696, 0x8f48a4899877186c), // 5^-306 + U128.new(0xe0b62e2929aba83c, 0x331acdabfe94de87), // 5^-305 + U128.new(0x8c71dcd9ba0b4925, 0x9ff0c08b7f1d0b14), // 5^-304 + U128.new(0xaf8e5410288e1b6f, 0x7ecf0ae5ee44dd9), // 5^-303 + U128.new(0xdb71e91432b1a24a, 0xc9e82cd9f69d6150), // 5^-302 + U128.new(0x892731ac9faf056e, 0xbe311c083a225cd2), // 5^-301 + U128.new(0xab70fe17c79ac6ca, 0x6dbd630a48aaf406), // 5^-300 + U128.new(0xd64d3d9db981787d, 0x92cbbccdad5b108), // 5^-299 + U128.new(0x85f0468293f0eb4e, 0x25bbf56008c58ea5), // 5^-298 + U128.new(0xa76c582338ed2621, 0xaf2af2b80af6f24e), // 5^-297 + U128.new(0xd1476e2c07286faa, 0x1af5af660db4aee1), // 5^-296 + U128.new(0x82cca4db847945ca, 0x50d98d9fc890ed4d), // 5^-295 + U128.new(0xa37fce126597973c, 0xe50ff107bab528a0), // 5^-294 + U128.new(0xcc5fc196fefd7d0c, 0x1e53ed49a96272c8), // 5^-293 + U128.new(0xff77b1fcbebcdc4f, 0x25e8e89c13bb0f7a), // 5^-292 + U128.new(0x9faacf3df73609b1, 0x77b191618c54e9ac), // 5^-291 + U128.new(0xc795830d75038c1d, 0xd59df5b9ef6a2417), // 5^-290 + U128.new(0xf97ae3d0d2446f25, 0x4b0573286b44ad1d), // 5^-289 + U128.new(0x9becce62836ac577, 0x4ee367f9430aec32), // 5^-288 + U128.new(0xc2e801fb244576d5, 0x229c41f793cda73f), // 5^-287 + U128.new(0xf3a20279ed56d48a, 0x6b43527578c1110f), // 5^-286 + U128.new(0x9845418c345644d6, 0x830a13896b78aaa9), // 5^-285 + U128.new(0xbe5691ef416bd60c, 0x23cc986bc656d553), // 5^-284 + U128.new(0xedec366b11c6cb8f, 0x2cbfbe86b7ec8aa8), // 5^-283 + U128.new(0x94b3a202eb1c3f39, 0x7bf7d71432f3d6a9), // 5^-282 + U128.new(0xb9e08a83a5e34f07, 0xdaf5ccd93fb0cc53), // 5^-281 + U128.new(0xe858ad248f5c22c9, 0xd1b3400f8f9cff68), // 5^-280 + U128.new(0x91376c36d99995be, 0x23100809b9c21fa1), // 5^-279 + U128.new(0xb58547448ffffb2d, 0xabd40a0c2832a78a), // 5^-278 + U128.new(0xe2e69915b3fff9f9, 0x16c90c8f323f516c), // 5^-277 + U128.new(0x8dd01fad907ffc3b, 0xae3da7d97f6792e3), // 5^-276 + U128.new(0xb1442798f49ffb4a, 0x99cd11cfdf41779c), // 5^-275 + U128.new(0xdd95317f31c7fa1d, 0x40405643d711d583), // 5^-274 + U128.new(0x8a7d3eef7f1cfc52, 0x482835ea666b2572), // 5^-273 + U128.new(0xad1c8eab5ee43b66, 0xda3243650005eecf), // 5^-272 + U128.new(0xd863b256369d4a40, 0x90bed43e40076a82), // 5^-271 + U128.new(0x873e4f75e2224e68, 0x5a7744a6e804a291), // 5^-270 + U128.new(0xa90de3535aaae202, 0x711515d0a205cb36), // 5^-269 + U128.new(0xd3515c2831559a83, 0xd5a5b44ca873e03), // 5^-268 + U128.new(0x8412d9991ed58091, 0xe858790afe9486c2), // 5^-267 + U128.new(0xa5178fff668ae0b6, 0x626e974dbe39a872), // 5^-266 + U128.new(0xce5d73ff402d98e3, 0xfb0a3d212dc8128f), // 5^-265 + U128.new(0x80fa687f881c7f8e, 0x7ce66634bc9d0b99), // 5^-264 + U128.new(0xa139029f6a239f72, 0x1c1fffc1ebc44e80), // 5^-263 + U128.new(0xc987434744ac874e, 0xa327ffb266b56220), // 5^-262 + U128.new(0xfbe9141915d7a922, 0x4bf1ff9f0062baa8), // 5^-261 + U128.new(0x9d71ac8fada6c9b5, 0x6f773fc3603db4a9), // 5^-260 + U128.new(0xc4ce17b399107c22, 0xcb550fb4384d21d3), // 5^-259 + U128.new(0xf6019da07f549b2b, 0x7e2a53a146606a48), // 5^-258 + U128.new(0x99c102844f94e0fb, 0x2eda7444cbfc426d), // 5^-257 + U128.new(0xc0314325637a1939, 0xfa911155fefb5308), // 5^-256 + U128.new(0xf03d93eebc589f88, 0x793555ab7eba27ca), // 5^-255 + U128.new(0x96267c7535b763b5, 0x4bc1558b2f3458de), // 5^-254 + U128.new(0xbbb01b9283253ca2, 0x9eb1aaedfb016f16), // 5^-253 + U128.new(0xea9c227723ee8bcb, 0x465e15a979c1cadc), // 5^-252 + U128.new(0x92a1958a7675175f, 0xbfacd89ec191ec9), // 5^-251 + U128.new(0xb749faed14125d36, 0xcef980ec671f667b), // 5^-250 + U128.new(0xe51c79a85916f484, 0x82b7e12780e7401a), // 5^-249 + U128.new(0x8f31cc0937ae58d2, 0xd1b2ecb8b0908810), // 5^-248 + U128.new(0xb2fe3f0b8599ef07, 0x861fa7e6dcb4aa15), // 5^-247 + U128.new(0xdfbdcece67006ac9, 0x67a791e093e1d49a), // 5^-246 + U128.new(0x8bd6a141006042bd, 0xe0c8bb2c5c6d24e0), // 5^-245 + U128.new(0xaecc49914078536d, 0x58fae9f773886e18), // 5^-244 + U128.new(0xda7f5bf590966848, 0xaf39a475506a899e), // 5^-243 + U128.new(0x888f99797a5e012d, 0x6d8406c952429603), // 5^-242 + U128.new(0xaab37fd7d8f58178, 0xc8e5087ba6d33b83), // 5^-241 + U128.new(0xd5605fcdcf32e1d6, 0xfb1e4a9a90880a64), // 5^-240 + U128.new(0x855c3be0a17fcd26, 0x5cf2eea09a55067f), // 5^-239 + U128.new(0xa6b34ad8c9dfc06f, 0xf42faa48c0ea481e), // 5^-238 + U128.new(0xd0601d8efc57b08b, 0xf13b94daf124da26), // 5^-237 + U128.new(0x823c12795db6ce57, 0x76c53d08d6b70858), // 5^-236 + U128.new(0xa2cb1717b52481ed, 0x54768c4b0c64ca6e), // 5^-235 + U128.new(0xcb7ddcdda26da268, 0xa9942f5dcf7dfd09), // 5^-234 + U128.new(0xfe5d54150b090b02, 0xd3f93b35435d7c4c), // 5^-233 + U128.new(0x9efa548d26e5a6e1, 0xc47bc5014a1a6daf), // 5^-232 + U128.new(0xc6b8e9b0709f109a, 0x359ab6419ca1091b), // 5^-231 + U128.new(0xf867241c8cc6d4c0, 0xc30163d203c94b62), // 5^-230 + U128.new(0x9b407691d7fc44f8, 0x79e0de63425dcf1d), // 5^-229 + U128.new(0xc21094364dfb5636, 0x985915fc12f542e4), // 5^-228 + U128.new(0xf294b943e17a2bc4, 0x3e6f5b7b17b2939d), // 5^-227 + U128.new(0x979cf3ca6cec5b5a, 0xa705992ceecf9c42), // 5^-226 + U128.new(0xbd8430bd08277231, 0x50c6ff782a838353), // 5^-225 + U128.new(0xece53cec4a314ebd, 0xa4f8bf5635246428), // 5^-224 + U128.new(0x940f4613ae5ed136, 0x871b7795e136be99), // 5^-223 + U128.new(0xb913179899f68584, 0x28e2557b59846e3f), // 5^-222 + U128.new(0xe757dd7ec07426e5, 0x331aeada2fe589cf), // 5^-221 + U128.new(0x9096ea6f3848984f, 0x3ff0d2c85def7621), // 5^-220 + U128.new(0xb4bca50b065abe63, 0xfed077a756b53a9), // 5^-219 + U128.new(0xe1ebce4dc7f16dfb, 0xd3e8495912c62894), // 5^-218 + U128.new(0x8d3360f09cf6e4bd, 0x64712dd7abbbd95c), // 5^-217 + U128.new(0xb080392cc4349dec, 0xbd8d794d96aacfb3), // 5^-216 + U128.new(0xdca04777f541c567, 0xecf0d7a0fc5583a0), // 5^-215 + U128.new(0x89e42caaf9491b60, 0xf41686c49db57244), // 5^-214 + U128.new(0xac5d37d5b79b6239, 0x311c2875c522ced5), // 5^-213 + U128.new(0xd77485cb25823ac7, 0x7d633293366b828b), // 5^-212 + U128.new(0x86a8d39ef77164bc, 0xae5dff9c02033197), // 5^-211 + U128.new(0xa8530886b54dbdeb, 0xd9f57f830283fdfc), // 5^-210 + U128.new(0xd267caa862a12d66, 0xd072df63c324fd7b), // 5^-209 + U128.new(0x8380dea93da4bc60, 0x4247cb9e59f71e6d), // 5^-208 + U128.new(0xa46116538d0deb78, 0x52d9be85f074e608), // 5^-207 + U128.new(0xcd795be870516656, 0x67902e276c921f8b), // 5^-206 + U128.new(0x806bd9714632dff6, 0xba1cd8a3db53b6), // 5^-205 + U128.new(0xa086cfcd97bf97f3, 0x80e8a40eccd228a4), // 5^-204 + U128.new(0xc8a883c0fdaf7df0, 0x6122cd128006b2cd), // 5^-203 + U128.new(0xfad2a4b13d1b5d6c, 0x796b805720085f81), // 5^-202 + U128.new(0x9cc3a6eec6311a63, 0xcbe3303674053bb0), // 5^-201 + U128.new(0xc3f490aa77bd60fc, 0xbedbfc4411068a9c), // 5^-200 + U128.new(0xf4f1b4d515acb93b, 0xee92fb5515482d44), // 5^-199 + U128.new(0x991711052d8bf3c5, 0x751bdd152d4d1c4a), // 5^-198 + U128.new(0xbf5cd54678eef0b6, 0xd262d45a78a0635d), // 5^-197 + U128.new(0xef340a98172aace4, 0x86fb897116c87c34), // 5^-196 + U128.new(0x9580869f0e7aac0e, 0xd45d35e6ae3d4da0), // 5^-195 + U128.new(0xbae0a846d2195712, 0x8974836059cca109), // 5^-194 + U128.new(0xe998d258869facd7, 0x2bd1a438703fc94b), // 5^-193 + U128.new(0x91ff83775423cc06, 0x7b6306a34627ddcf), // 5^-192 + U128.new(0xb67f6455292cbf08, 0x1a3bc84c17b1d542), // 5^-191 + U128.new(0xe41f3d6a7377eeca, 0x20caba5f1d9e4a93), // 5^-190 + U128.new(0x8e938662882af53e, 0x547eb47b7282ee9c), // 5^-189 + U128.new(0xb23867fb2a35b28d, 0xe99e619a4f23aa43), // 5^-188 + U128.new(0xdec681f9f4c31f31, 0x6405fa00e2ec94d4), // 5^-187 + U128.new(0x8b3c113c38f9f37e, 0xde83bc408dd3dd04), // 5^-186 + U128.new(0xae0b158b4738705e, 0x9624ab50b148d445), // 5^-185 + U128.new(0xd98ddaee19068c76, 0x3badd624dd9b0957), // 5^-184 + U128.new(0x87f8a8d4cfa417c9, 0xe54ca5d70a80e5d6), // 5^-183 + U128.new(0xa9f6d30a038d1dbc, 0x5e9fcf4ccd211f4c), // 5^-182 + U128.new(0xd47487cc8470652b, 0x7647c3200069671f), // 5^-181 + U128.new(0x84c8d4dfd2c63f3b, 0x29ecd9f40041e073), // 5^-180 + U128.new(0xa5fb0a17c777cf09, 0xf468107100525890), // 5^-179 + U128.new(0xcf79cc9db955c2cc, 0x7182148d4066eeb4), // 5^-178 + U128.new(0x81ac1fe293d599bf, 0xc6f14cd848405530), // 5^-177 + U128.new(0xa21727db38cb002f, 0xb8ada00e5a506a7c), // 5^-176 + U128.new(0xca9cf1d206fdc03b, 0xa6d90811f0e4851c), // 5^-175 + U128.new(0xfd442e4688bd304a, 0x908f4a166d1da663), // 5^-174 + U128.new(0x9e4a9cec15763e2e, 0x9a598e4e043287fe), // 5^-173 + U128.new(0xc5dd44271ad3cdba, 0x40eff1e1853f29fd), // 5^-172 + U128.new(0xf7549530e188c128, 0xd12bee59e68ef47c), // 5^-171 + U128.new(0x9a94dd3e8cf578b9, 0x82bb74f8301958ce), // 5^-170 + U128.new(0xc13a148e3032d6e7, 0xe36a52363c1faf01), // 5^-169 + U128.new(0xf18899b1bc3f8ca1, 0xdc44e6c3cb279ac1), // 5^-168 + U128.new(0x96f5600f15a7b7e5, 0x29ab103a5ef8c0b9), // 5^-167 + U128.new(0xbcb2b812db11a5de, 0x7415d448f6b6f0e7), // 5^-166 + U128.new(0xebdf661791d60f56, 0x111b495b3464ad21), // 5^-165 + U128.new(0x936b9fcebb25c995, 0xcab10dd900beec34), // 5^-164 + U128.new(0xb84687c269ef3bfb, 0x3d5d514f40eea742), // 5^-163 + U128.new(0xe65829b3046b0afa, 0xcb4a5a3112a5112), // 5^-162 + U128.new(0x8ff71a0fe2c2e6dc, 0x47f0e785eaba72ab), // 5^-161 + U128.new(0xb3f4e093db73a093, 0x59ed216765690f56), // 5^-160 + U128.new(0xe0f218b8d25088b8, 0x306869c13ec3532c), // 5^-159 + U128.new(0x8c974f7383725573, 0x1e414218c73a13fb), // 5^-158 + U128.new(0xafbd2350644eeacf, 0xe5d1929ef90898fa), // 5^-157 + U128.new(0xdbac6c247d62a583, 0xdf45f746b74abf39), // 5^-156 + U128.new(0x894bc396ce5da772, 0x6b8bba8c328eb783), // 5^-155 + U128.new(0xab9eb47c81f5114f, 0x66ea92f3f326564), // 5^-154 + U128.new(0xd686619ba27255a2, 0xc80a537b0efefebd), // 5^-153 + U128.new(0x8613fd0145877585, 0xbd06742ce95f5f36), // 5^-152 + U128.new(0xa798fc4196e952e7, 0x2c48113823b73704), // 5^-151 + U128.new(0xd17f3b51fca3a7a0, 0xf75a15862ca504c5), // 5^-150 + U128.new(0x82ef85133de648c4, 0x9a984d73dbe722fb), // 5^-149 + U128.new(0xa3ab66580d5fdaf5, 0xc13e60d0d2e0ebba), // 5^-148 + U128.new(0xcc963fee10b7d1b3, 0x318df905079926a8), // 5^-147 + U128.new(0xffbbcfe994e5c61f, 0xfdf17746497f7052), // 5^-146 + U128.new(0x9fd561f1fd0f9bd3, 0xfeb6ea8bedefa633), // 5^-145 + U128.new(0xc7caba6e7c5382c8, 0xfe64a52ee96b8fc0), // 5^-144 + U128.new(0xf9bd690a1b68637b, 0x3dfdce7aa3c673b0), // 5^-143 + U128.new(0x9c1661a651213e2d, 0x6bea10ca65c084e), // 5^-142 + U128.new(0xc31bfa0fe5698db8, 0x486e494fcff30a62), // 5^-141 + U128.new(0xf3e2f893dec3f126, 0x5a89dba3c3efccfa), // 5^-140 + U128.new(0x986ddb5c6b3a76b7, 0xf89629465a75e01c), // 5^-139 + U128.new(0xbe89523386091465, 0xf6bbb397f1135823), // 5^-138 + U128.new(0xee2ba6c0678b597f, 0x746aa07ded582e2c), // 5^-137 + U128.new(0x94db483840b717ef, 0xa8c2a44eb4571cdc), // 5^-136 + U128.new(0xba121a4650e4ddeb, 0x92f34d62616ce413), // 5^-135 + U128.new(0xe896a0d7e51e1566, 0x77b020baf9c81d17), // 5^-134 + U128.new(0x915e2486ef32cd60, 0xace1474dc1d122e), // 5^-133 + U128.new(0xb5b5ada8aaff80b8, 0xd819992132456ba), // 5^-132 + U128.new(0xe3231912d5bf60e6, 0x10e1fff697ed6c69), // 5^-131 + U128.new(0x8df5efabc5979c8f, 0xca8d3ffa1ef463c1), // 5^-130 + U128.new(0xb1736b96b6fd83b3, 0xbd308ff8a6b17cb2), // 5^-129 + U128.new(0xddd0467c64bce4a0, 0xac7cb3f6d05ddbde), // 5^-128 + U128.new(0x8aa22c0dbef60ee4, 0x6bcdf07a423aa96b), // 5^-127 + U128.new(0xad4ab7112eb3929d, 0x86c16c98d2c953c6), // 5^-126 + U128.new(0xd89d64d57a607744, 0xe871c7bf077ba8b7), // 5^-125 + U128.new(0x87625f056c7c4a8b, 0x11471cd764ad4972), // 5^-124 + U128.new(0xa93af6c6c79b5d2d, 0xd598e40d3dd89bcf), // 5^-123 + U128.new(0xd389b47879823479, 0x4aff1d108d4ec2c3), // 5^-122 + U128.new(0x843610cb4bf160cb, 0xcedf722a585139ba), // 5^-121 + U128.new(0xa54394fe1eedb8fe, 0xc2974eb4ee658828), // 5^-120 + U128.new(0xce947a3da6a9273e, 0x733d226229feea32), // 5^-119 + U128.new(0x811ccc668829b887, 0x806357d5a3f525f), // 5^-118 + U128.new(0xa163ff802a3426a8, 0xca07c2dcb0cf26f7), // 5^-117 + U128.new(0xc9bcff6034c13052, 0xfc89b393dd02f0b5), // 5^-116 + U128.new(0xfc2c3f3841f17c67, 0xbbac2078d443ace2), // 5^-115 + U128.new(0x9d9ba7832936edc0, 0xd54b944b84aa4c0d), // 5^-114 + U128.new(0xc5029163f384a931, 0xa9e795e65d4df11), // 5^-113 + U128.new(0xf64335bcf065d37d, 0x4d4617b5ff4a16d5), // 5^-112 + U128.new(0x99ea0196163fa42e, 0x504bced1bf8e4e45), // 5^-111 + U128.new(0xc06481fb9bcf8d39, 0xe45ec2862f71e1d6), // 5^-110 + U128.new(0xf07da27a82c37088, 0x5d767327bb4e5a4c), // 5^-109 + U128.new(0x964e858c91ba2655, 0x3a6a07f8d510f86f), // 5^-108 + U128.new(0xbbe226efb628afea, 0x890489f70a55368b), // 5^-107 + U128.new(0xeadab0aba3b2dbe5, 0x2b45ac74ccea842e), // 5^-106 + U128.new(0x92c8ae6b464fc96f, 0x3b0b8bc90012929d), // 5^-105 + U128.new(0xb77ada0617e3bbcb, 0x9ce6ebb40173744), // 5^-104 + U128.new(0xe55990879ddcaabd, 0xcc420a6a101d0515), // 5^-103 + U128.new(0x8f57fa54c2a9eab6, 0x9fa946824a12232d), // 5^-102 + U128.new(0xb32df8e9f3546564, 0x47939822dc96abf9), // 5^-101 + U128.new(0xdff9772470297ebd, 0x59787e2b93bc56f7), // 5^-100 + U128.new(0x8bfbea76c619ef36, 0x57eb4edb3c55b65a), // 5^-99 + U128.new(0xaefae51477a06b03, 0xede622920b6b23f1), // 5^-98 + U128.new(0xdab99e59958885c4, 0xe95fab368e45eced), // 5^-97 + U128.new(0x88b402f7fd75539b, 0x11dbcb0218ebb414), // 5^-96 + U128.new(0xaae103b5fcd2a881, 0xd652bdc29f26a119), // 5^-95 + U128.new(0xd59944a37c0752a2, 0x4be76d3346f0495f), // 5^-94 + U128.new(0x857fcae62d8493a5, 0x6f70a4400c562ddb), // 5^-93 + U128.new(0xa6dfbd9fb8e5b88e, 0xcb4ccd500f6bb952), // 5^-92 + U128.new(0xd097ad07a71f26b2, 0x7e2000a41346a7a7), // 5^-91 + U128.new(0x825ecc24c873782f, 0x8ed400668c0c28c8), // 5^-90 + U128.new(0xa2f67f2dfa90563b, 0x728900802f0f32fa), // 5^-89 + U128.new(0xcbb41ef979346bca, 0x4f2b40a03ad2ffb9), // 5^-88 + U128.new(0xfea126b7d78186bc, 0xe2f610c84987bfa8), // 5^-87 + U128.new(0x9f24b832e6b0f436, 0xdd9ca7d2df4d7c9), // 5^-86 + U128.new(0xc6ede63fa05d3143, 0x91503d1c79720dbb), // 5^-85 + U128.new(0xf8a95fcf88747d94, 0x75a44c6397ce912a), // 5^-84 + U128.new(0x9b69dbe1b548ce7c, 0xc986afbe3ee11aba), // 5^-83 + U128.new(0xc24452da229b021b, 0xfbe85badce996168), // 5^-82 + U128.new(0xf2d56790ab41c2a2, 0xfae27299423fb9c3), // 5^-81 + U128.new(0x97c560ba6b0919a5, 0xdccd879fc967d41a), // 5^-80 + U128.new(0xbdb6b8e905cb600f, 0x5400e987bbc1c920), // 5^-79 + U128.new(0xed246723473e3813, 0x290123e9aab23b68), // 5^-78 + U128.new(0x9436c0760c86e30b, 0xf9a0b6720aaf6521), // 5^-77 + U128.new(0xb94470938fa89bce, 0xf808e40e8d5b3e69), // 5^-76 + U128.new(0xe7958cb87392c2c2, 0xb60b1d1230b20e04), // 5^-75 + U128.new(0x90bd77f3483bb9b9, 0xb1c6f22b5e6f48c2), // 5^-74 + U128.new(0xb4ecd5f01a4aa828, 0x1e38aeb6360b1af3), // 5^-73 + U128.new(0xe2280b6c20dd5232, 0x25c6da63c38de1b0), // 5^-72 + U128.new(0x8d590723948a535f, 0x579c487e5a38ad0e), // 5^-71 + U128.new(0xb0af48ec79ace837, 0x2d835a9df0c6d851), // 5^-70 + U128.new(0xdcdb1b2798182244, 0xf8e431456cf88e65), // 5^-69 + U128.new(0x8a08f0f8bf0f156b, 0x1b8e9ecb641b58ff), // 5^-68 + U128.new(0xac8b2d36eed2dac5, 0xe272467e3d222f3f), // 5^-67 + U128.new(0xd7adf884aa879177, 0x5b0ed81dcc6abb0f), // 5^-66 + U128.new(0x86ccbb52ea94baea, 0x98e947129fc2b4e9), // 5^-65 + U128.new(0xa87fea27a539e9a5, 0x3f2398d747b36224), // 5^-64 + U128.new(0xd29fe4b18e88640e, 0x8eec7f0d19a03aad), // 5^-63 + U128.new(0x83a3eeeef9153e89, 0x1953cf68300424ac), // 5^-62 + U128.new(0xa48ceaaab75a8e2b, 0x5fa8c3423c052dd7), // 5^-61 + U128.new(0xcdb02555653131b6, 0x3792f412cb06794d), // 5^-60 + U128.new(0x808e17555f3ebf11, 0xe2bbd88bbee40bd0), // 5^-59 + U128.new(0xa0b19d2ab70e6ed6, 0x5b6aceaeae9d0ec4), // 5^-58 + U128.new(0xc8de047564d20a8b, 0xf245825a5a445275), // 5^-57 + U128.new(0xfb158592be068d2e, 0xeed6e2f0f0d56712), // 5^-56 + U128.new(0x9ced737bb6c4183d, 0x55464dd69685606b), // 5^-55 + U128.new(0xc428d05aa4751e4c, 0xaa97e14c3c26b886), // 5^-54 + U128.new(0xf53304714d9265df, 0xd53dd99f4b3066a8), // 5^-53 + U128.new(0x993fe2c6d07b7fab, 0xe546a8038efe4029), // 5^-52 + U128.new(0xbf8fdb78849a5f96, 0xde98520472bdd033), // 5^-51 + U128.new(0xef73d256a5c0f77c, 0x963e66858f6d4440), // 5^-50 + U128.new(0x95a8637627989aad, 0xdde7001379a44aa8), // 5^-49 + U128.new(0xbb127c53b17ec159, 0x5560c018580d5d52), // 5^-48 + U128.new(0xe9d71b689dde71af, 0xaab8f01e6e10b4a6), // 5^-47 + U128.new(0x9226712162ab070d, 0xcab3961304ca70e8), // 5^-46 + U128.new(0xb6b00d69bb55c8d1, 0x3d607b97c5fd0d22), // 5^-45 + U128.new(0xe45c10c42a2b3b05, 0x8cb89a7db77c506a), // 5^-44 + U128.new(0x8eb98a7a9a5b04e3, 0x77f3608e92adb242), // 5^-43 + U128.new(0xb267ed1940f1c61c, 0x55f038b237591ed3), // 5^-42 + U128.new(0xdf01e85f912e37a3, 0x6b6c46dec52f6688), // 5^-41 + U128.new(0x8b61313bbabce2c6, 0x2323ac4b3b3da015), // 5^-40 + U128.new(0xae397d8aa96c1b77, 0xabec975e0a0d081a), // 5^-39 + U128.new(0xd9c7dced53c72255, 0x96e7bd358c904a21), // 5^-38 + U128.new(0x881cea14545c7575, 0x7e50d64177da2e54), // 5^-37 + U128.new(0xaa242499697392d2, 0xdde50bd1d5d0b9e9), // 5^-36 + U128.new(0xd4ad2dbfc3d07787, 0x955e4ec64b44e864), // 5^-35 + U128.new(0x84ec3c97da624ab4, 0xbd5af13bef0b113e), // 5^-34 + U128.new(0xa6274bbdd0fadd61, 0xecb1ad8aeacdd58e), // 5^-33 + U128.new(0xcfb11ead453994ba, 0x67de18eda5814af2), // 5^-32 + U128.new(0x81ceb32c4b43fcf4, 0x80eacf948770ced7), // 5^-31 + U128.new(0xa2425ff75e14fc31, 0xa1258379a94d028d), // 5^-30 + U128.new(0xcad2f7f5359a3b3e, 0x96ee45813a04330), // 5^-29 + U128.new(0xfd87b5f28300ca0d, 0x8bca9d6e188853fc), // 5^-28 + U128.new(0x9e74d1b791e07e48, 0x775ea264cf55347e), // 5^-27 + U128.new(0xc612062576589dda, 0x95364afe032a819e), // 5^-26 + U128.new(0xf79687aed3eec551, 0x3a83ddbd83f52205), // 5^-25 + U128.new(0x9abe14cd44753b52, 0xc4926a9672793543), // 5^-24 + U128.new(0xc16d9a0095928a27, 0x75b7053c0f178294), // 5^-23 + U128.new(0xf1c90080baf72cb1, 0x5324c68b12dd6339), // 5^-22 + U128.new(0x971da05074da7bee, 0xd3f6fc16ebca5e04), // 5^-21 + U128.new(0xbce5086492111aea, 0x88f4bb1ca6bcf585), // 5^-20 + U128.new(0xec1e4a7db69561a5, 0x2b31e9e3d06c32e6), // 5^-19 + U128.new(0x9392ee8e921d5d07, 0x3aff322e62439fd0), // 5^-18 + U128.new(0xb877aa3236a4b449, 0x9befeb9fad487c3), // 5^-17 + U128.new(0xe69594bec44de15b, 0x4c2ebe687989a9b4), // 5^-16 + U128.new(0x901d7cf73ab0acd9, 0xf9d37014bf60a11), // 5^-15 + U128.new(0xb424dc35095cd80f, 0x538484c19ef38c95), // 5^-14 + U128.new(0xe12e13424bb40e13, 0x2865a5f206b06fba), // 5^-13 + U128.new(0x8cbccc096f5088cb, 0xf93f87b7442e45d4), // 5^-12 + U128.new(0xafebff0bcb24aafe, 0xf78f69a51539d749), // 5^-11 + U128.new(0xdbe6fecebdedd5be, 0xb573440e5a884d1c), // 5^-10 + U128.new(0x89705f4136b4a597, 0x31680a88f8953031), // 5^-9 + U128.new(0xabcc77118461cefc, 0xfdc20d2b36ba7c3e), // 5^-8 + U128.new(0xd6bf94d5e57a42bc, 0x3d32907604691b4d), // 5^-7 + U128.new(0x8637bd05af6c69b5, 0xa63f9a49c2c1b110), // 5^-6 + U128.new(0xa7c5ac471b478423, 0xfcf80dc33721d54), // 5^-5 + U128.new(0xd1b71758e219652b, 0xd3c36113404ea4a9), // 5^-4 + U128.new(0x83126e978d4fdf3b, 0x645a1cac083126ea), // 5^-3 + U128.new(0xa3d70a3d70a3d70a, 0x3d70a3d70a3d70a4), // 5^-2 + U128.new(0xcccccccccccccccc, 0xcccccccccccccccd), // 5^-1 + U128.new(0x8000000000000000, 0x0), // 5^0 + U128.new(0xa000000000000000, 0x0), // 5^1 + U128.new(0xc800000000000000, 0x0), // 5^2 + U128.new(0xfa00000000000000, 0x0), // 5^3 + U128.new(0x9c40000000000000, 0x0), // 5^4 + U128.new(0xc350000000000000, 0x0), // 5^5 + U128.new(0xf424000000000000, 0x0), // 5^6 + U128.new(0x9896800000000000, 0x0), // 5^7 + U128.new(0xbebc200000000000, 0x0), // 5^8 + U128.new(0xee6b280000000000, 0x0), // 5^9 + U128.new(0x9502f90000000000, 0x0), // 5^10 + U128.new(0xba43b74000000000, 0x0), // 5^11 + U128.new(0xe8d4a51000000000, 0x0), // 5^12 + U128.new(0x9184e72a00000000, 0x0), // 5^13 + U128.new(0xb5e620f480000000, 0x0), // 5^14 + U128.new(0xe35fa931a0000000, 0x0), // 5^15 + U128.new(0x8e1bc9bf04000000, 0x0), // 5^16 + U128.new(0xb1a2bc2ec5000000, 0x0), // 5^17 + U128.new(0xde0b6b3a76400000, 0x0), // 5^18 + U128.new(0x8ac7230489e80000, 0x0), // 5^19 + U128.new(0xad78ebc5ac620000, 0x0), // 5^20 + U128.new(0xd8d726b7177a8000, 0x0), // 5^21 + U128.new(0x878678326eac9000, 0x0), // 5^22 + U128.new(0xa968163f0a57b400, 0x0), // 5^23 + U128.new(0xd3c21bcecceda100, 0x0), // 5^24 + U128.new(0x84595161401484a0, 0x0), // 5^25 + U128.new(0xa56fa5b99019a5c8, 0x0), // 5^26 + U128.new(0xcecb8f27f4200f3a, 0x0), // 5^27 + U128.new(0x813f3978f8940984, 0x4000000000000000), // 5^28 + U128.new(0xa18f07d736b90be5, 0x5000000000000000), // 5^29 + U128.new(0xc9f2c9cd04674ede, 0xa400000000000000), // 5^30 + U128.new(0xfc6f7c4045812296, 0x4d00000000000000), // 5^31 + U128.new(0x9dc5ada82b70b59d, 0xf020000000000000), // 5^32 + U128.new(0xc5371912364ce305, 0x6c28000000000000), // 5^33 + U128.new(0xf684df56c3e01bc6, 0xc732000000000000), // 5^34 + U128.new(0x9a130b963a6c115c, 0x3c7f400000000000), // 5^35 + U128.new(0xc097ce7bc90715b3, 0x4b9f100000000000), // 5^36 + U128.new(0xf0bdc21abb48db20, 0x1e86d40000000000), // 5^37 + U128.new(0x96769950b50d88f4, 0x1314448000000000), // 5^38 + U128.new(0xbc143fa4e250eb31, 0x17d955a000000000), // 5^39 + U128.new(0xeb194f8e1ae525fd, 0x5dcfab0800000000), // 5^40 + U128.new(0x92efd1b8d0cf37be, 0x5aa1cae500000000), // 5^41 + U128.new(0xb7abc627050305ad, 0xf14a3d9e40000000), // 5^42 + U128.new(0xe596b7b0c643c719, 0x6d9ccd05d0000000), // 5^43 + U128.new(0x8f7e32ce7bea5c6f, 0xe4820023a2000000), // 5^44 + U128.new(0xb35dbf821ae4f38b, 0xdda2802c8a800000), // 5^45 + U128.new(0xe0352f62a19e306e, 0xd50b2037ad200000), // 5^46 + U128.new(0x8c213d9da502de45, 0x4526f422cc340000), // 5^47 + U128.new(0xaf298d050e4395d6, 0x9670b12b7f410000), // 5^48 + U128.new(0xdaf3f04651d47b4c, 0x3c0cdd765f114000), // 5^49 + U128.new(0x88d8762bf324cd0f, 0xa5880a69fb6ac800), // 5^50 + U128.new(0xab0e93b6efee0053, 0x8eea0d047a457a00), // 5^51 + U128.new(0xd5d238a4abe98068, 0x72a4904598d6d880), // 5^52 + U128.new(0x85a36366eb71f041, 0x47a6da2b7f864750), // 5^53 + U128.new(0xa70c3c40a64e6c51, 0x999090b65f67d924), // 5^54 + U128.new(0xd0cf4b50cfe20765, 0xfff4b4e3f741cf6d), // 5^55 + U128.new(0x82818f1281ed449f, 0xbff8f10e7a8921a4), // 5^56 + U128.new(0xa321f2d7226895c7, 0xaff72d52192b6a0d), // 5^57 + U128.new(0xcbea6f8ceb02bb39, 0x9bf4f8a69f764490), // 5^58 + U128.new(0xfee50b7025c36a08, 0x2f236d04753d5b4), // 5^59 + U128.new(0x9f4f2726179a2245, 0x1d762422c946590), // 5^60 + U128.new(0xc722f0ef9d80aad6, 0x424d3ad2b7b97ef5), // 5^61 + U128.new(0xf8ebad2b84e0d58b, 0xd2e0898765a7deb2), // 5^62 + U128.new(0x9b934c3b330c8577, 0x63cc55f49f88eb2f), // 5^63 + U128.new(0xc2781f49ffcfa6d5, 0x3cbf6b71c76b25fb), // 5^64 + U128.new(0xf316271c7fc3908a, 0x8bef464e3945ef7a), // 5^65 + U128.new(0x97edd871cfda3a56, 0x97758bf0e3cbb5ac), // 5^66 + U128.new(0xbde94e8e43d0c8ec, 0x3d52eeed1cbea317), // 5^67 + U128.new(0xed63a231d4c4fb27, 0x4ca7aaa863ee4bdd), // 5^68 + U128.new(0x945e455f24fb1cf8, 0x8fe8caa93e74ef6a), // 5^69 + U128.new(0xb975d6b6ee39e436, 0xb3e2fd538e122b44), // 5^70 + U128.new(0xe7d34c64a9c85d44, 0x60dbbca87196b616), // 5^71 + U128.new(0x90e40fbeea1d3a4a, 0xbc8955e946fe31cd), // 5^72 + U128.new(0xb51d13aea4a488dd, 0x6babab6398bdbe41), // 5^73 + U128.new(0xe264589a4dcdab14, 0xc696963c7eed2dd1), // 5^74 + U128.new(0x8d7eb76070a08aec, 0xfc1e1de5cf543ca2), // 5^75 + U128.new(0xb0de65388cc8ada8, 0x3b25a55f43294bcb), // 5^76 + U128.new(0xdd15fe86affad912, 0x49ef0eb713f39ebe), // 5^77 + U128.new(0x8a2dbf142dfcc7ab, 0x6e3569326c784337), // 5^78 + U128.new(0xacb92ed9397bf996, 0x49c2c37f07965404), // 5^79 + U128.new(0xd7e77a8f87daf7fb, 0xdc33745ec97be906), // 5^80 + U128.new(0x86f0ac99b4e8dafd, 0x69a028bb3ded71a3), // 5^81 + U128.new(0xa8acd7c0222311bc, 0xc40832ea0d68ce0c), // 5^82 + U128.new(0xd2d80db02aabd62b, 0xf50a3fa490c30190), // 5^83 + U128.new(0x83c7088e1aab65db, 0x792667c6da79e0fa), // 5^84 + U128.new(0xa4b8cab1a1563f52, 0x577001b891185938), // 5^85 + U128.new(0xcde6fd5e09abcf26, 0xed4c0226b55e6f86), // 5^86 + U128.new(0x80b05e5ac60b6178, 0x544f8158315b05b4), // 5^87 + U128.new(0xa0dc75f1778e39d6, 0x696361ae3db1c721), // 5^88 + U128.new(0xc913936dd571c84c, 0x3bc3a19cd1e38e9), // 5^89 + U128.new(0xfb5878494ace3a5f, 0x4ab48a04065c723), // 5^90 + U128.new(0x9d174b2dcec0e47b, 0x62eb0d64283f9c76), // 5^91 + U128.new(0xc45d1df942711d9a, 0x3ba5d0bd324f8394), // 5^92 + U128.new(0xf5746577930d6500, 0xca8f44ec7ee36479), // 5^93 + U128.new(0x9968bf6abbe85f20, 0x7e998b13cf4e1ecb), // 5^94 + U128.new(0xbfc2ef456ae276e8, 0x9e3fedd8c321a67e), // 5^95 + U128.new(0xefb3ab16c59b14a2, 0xc5cfe94ef3ea101e), // 5^96 + U128.new(0x95d04aee3b80ece5, 0xbba1f1d158724a12), // 5^97 + U128.new(0xbb445da9ca61281f, 0x2a8a6e45ae8edc97), // 5^98 + U128.new(0xea1575143cf97226, 0xf52d09d71a3293bd), // 5^99 + U128.new(0x924d692ca61be758, 0x593c2626705f9c56), // 5^100 + U128.new(0xb6e0c377cfa2e12e, 0x6f8b2fb00c77836c), // 5^101 + U128.new(0xe498f455c38b997a, 0xb6dfb9c0f956447), // 5^102 + U128.new(0x8edf98b59a373fec, 0x4724bd4189bd5eac), // 5^103 + U128.new(0xb2977ee300c50fe7, 0x58edec91ec2cb657), // 5^104 + U128.new(0xdf3d5e9bc0f653e1, 0x2f2967b66737e3ed), // 5^105 + U128.new(0x8b865b215899f46c, 0xbd79e0d20082ee74), // 5^106 + U128.new(0xae67f1e9aec07187, 0xecd8590680a3aa11), // 5^107 + U128.new(0xda01ee641a708de9, 0xe80e6f4820cc9495), // 5^108 + U128.new(0x884134fe908658b2, 0x3109058d147fdcdd), // 5^109 + U128.new(0xaa51823e34a7eede, 0xbd4b46f0599fd415), // 5^110 + U128.new(0xd4e5e2cdc1d1ea96, 0x6c9e18ac7007c91a), // 5^111 + U128.new(0x850fadc09923329e, 0x3e2cf6bc604ddb0), // 5^112 + U128.new(0xa6539930bf6bff45, 0x84db8346b786151c), // 5^113 + U128.new(0xcfe87f7cef46ff16, 0xe612641865679a63), // 5^114 + U128.new(0x81f14fae158c5f6e, 0x4fcb7e8f3f60c07e), // 5^115 + U128.new(0xa26da3999aef7749, 0xe3be5e330f38f09d), // 5^116 + U128.new(0xcb090c8001ab551c, 0x5cadf5bfd3072cc5), // 5^117 + U128.new(0xfdcb4fa002162a63, 0x73d9732fc7c8f7f6), // 5^118 + U128.new(0x9e9f11c4014dda7e, 0x2867e7fddcdd9afa), // 5^119 + U128.new(0xc646d63501a1511d, 0xb281e1fd541501b8), // 5^120 + U128.new(0xf7d88bc24209a565, 0x1f225a7ca91a4226), // 5^121 + U128.new(0x9ae757596946075f, 0x3375788de9b06958), // 5^122 + U128.new(0xc1a12d2fc3978937, 0x52d6b1641c83ae), // 5^123 + U128.new(0xf209787bb47d6b84, 0xc0678c5dbd23a49a), // 5^124 + U128.new(0x9745eb4d50ce6332, 0xf840b7ba963646e0), // 5^125 + U128.new(0xbd176620a501fbff, 0xb650e5a93bc3d898), // 5^126 + U128.new(0xec5d3fa8ce427aff, 0xa3e51f138ab4cebe), // 5^127 + U128.new(0x93ba47c980e98cdf, 0xc66f336c36b10137), // 5^128 + U128.new(0xb8a8d9bbe123f017, 0xb80b0047445d4184), // 5^129 + U128.new(0xe6d3102ad96cec1d, 0xa60dc059157491e5), // 5^130 + U128.new(0x9043ea1ac7e41392, 0x87c89837ad68db2f), // 5^131 + U128.new(0xb454e4a179dd1877, 0x29babe4598c311fb), // 5^132 + U128.new(0xe16a1dc9d8545e94, 0xf4296dd6fef3d67a), // 5^133 + U128.new(0x8ce2529e2734bb1d, 0x1899e4a65f58660c), // 5^134 + U128.new(0xb01ae745b101e9e4, 0x5ec05dcff72e7f8f), // 5^135 + U128.new(0xdc21a1171d42645d, 0x76707543f4fa1f73), // 5^136 + U128.new(0x899504ae72497eba, 0x6a06494a791c53a8), // 5^137 + U128.new(0xabfa45da0edbde69, 0x487db9d17636892), // 5^138 + U128.new(0xd6f8d7509292d603, 0x45a9d2845d3c42b6), // 5^139 + U128.new(0x865b86925b9bc5c2, 0xb8a2392ba45a9b2), // 5^140 + U128.new(0xa7f26836f282b732, 0x8e6cac7768d7141e), // 5^141 + U128.new(0xd1ef0244af2364ff, 0x3207d795430cd926), // 5^142 + U128.new(0x8335616aed761f1f, 0x7f44e6bd49e807b8), // 5^143 + U128.new(0xa402b9c5a8d3a6e7, 0x5f16206c9c6209a6), // 5^144 + U128.new(0xcd036837130890a1, 0x36dba887c37a8c0f), // 5^145 + U128.new(0x802221226be55a64, 0xc2494954da2c9789), // 5^146 + U128.new(0xa02aa96b06deb0fd, 0xf2db9baa10b7bd6c), // 5^147 + U128.new(0xc83553c5c8965d3d, 0x6f92829494e5acc7), // 5^148 + U128.new(0xfa42a8b73abbf48c, 0xcb772339ba1f17f9), // 5^149 + U128.new(0x9c69a97284b578d7, 0xff2a760414536efb), // 5^150 + U128.new(0xc38413cf25e2d70d, 0xfef5138519684aba), // 5^151 + U128.new(0xf46518c2ef5b8cd1, 0x7eb258665fc25d69), // 5^152 + U128.new(0x98bf2f79d5993802, 0xef2f773ffbd97a61), // 5^153 + U128.new(0xbeeefb584aff8603, 0xaafb550ffacfd8fa), // 5^154 + U128.new(0xeeaaba2e5dbf6784, 0x95ba2a53f983cf38), // 5^155 + U128.new(0x952ab45cfa97a0b2, 0xdd945a747bf26183), // 5^156 + U128.new(0xba756174393d88df, 0x94f971119aeef9e4), // 5^157 + U128.new(0xe912b9d1478ceb17, 0x7a37cd5601aab85d), // 5^158 + U128.new(0x91abb422ccb812ee, 0xac62e055c10ab33a), // 5^159 + U128.new(0xb616a12b7fe617aa, 0x577b986b314d6009), // 5^160 + U128.new(0xe39c49765fdf9d94, 0xed5a7e85fda0b80b), // 5^161 + U128.new(0x8e41ade9fbebc27d, 0x14588f13be847307), // 5^162 + U128.new(0xb1d219647ae6b31c, 0x596eb2d8ae258fc8), // 5^163 + U128.new(0xde469fbd99a05fe3, 0x6fca5f8ed9aef3bb), // 5^164 + U128.new(0x8aec23d680043bee, 0x25de7bb9480d5854), // 5^165 + U128.new(0xada72ccc20054ae9, 0xaf561aa79a10ae6a), // 5^166 + U128.new(0xd910f7ff28069da4, 0x1b2ba1518094da04), // 5^167 + U128.new(0x87aa9aff79042286, 0x90fb44d2f05d0842), // 5^168 + U128.new(0xa99541bf57452b28, 0x353a1607ac744a53), // 5^169 + U128.new(0xd3fa922f2d1675f2, 0x42889b8997915ce8), // 5^170 + U128.new(0x847c9b5d7c2e09b7, 0x69956135febada11), // 5^171 + U128.new(0xa59bc234db398c25, 0x43fab9837e699095), // 5^172 + U128.new(0xcf02b2c21207ef2e, 0x94f967e45e03f4bb), // 5^173 + U128.new(0x8161afb94b44f57d, 0x1d1be0eebac278f5), // 5^174 + U128.new(0xa1ba1ba79e1632dc, 0x6462d92a69731732), // 5^175 + U128.new(0xca28a291859bbf93, 0x7d7b8f7503cfdcfe), // 5^176 + U128.new(0xfcb2cb35e702af78, 0x5cda735244c3d43e), // 5^177 + U128.new(0x9defbf01b061adab, 0x3a0888136afa64a7), // 5^178 + U128.new(0xc56baec21c7a1916, 0x88aaa1845b8fdd0), // 5^179 + U128.new(0xf6c69a72a3989f5b, 0x8aad549e57273d45), // 5^180 + U128.new(0x9a3c2087a63f6399, 0x36ac54e2f678864b), // 5^181 + U128.new(0xc0cb28a98fcf3c7f, 0x84576a1bb416a7dd), // 5^182 + U128.new(0xf0fdf2d3f3c30b9f, 0x656d44a2a11c51d5), // 5^183 + U128.new(0x969eb7c47859e743, 0x9f644ae5a4b1b325), // 5^184 + U128.new(0xbc4665b596706114, 0x873d5d9f0dde1fee), // 5^185 + U128.new(0xeb57ff22fc0c7959, 0xa90cb506d155a7ea), // 5^186 + U128.new(0x9316ff75dd87cbd8, 0x9a7f12442d588f2), // 5^187 + U128.new(0xb7dcbf5354e9bece, 0xc11ed6d538aeb2f), // 5^188 + U128.new(0xe5d3ef282a242e81, 0x8f1668c8a86da5fa), // 5^189 + U128.new(0x8fa475791a569d10, 0xf96e017d694487bc), // 5^190 + U128.new(0xb38d92d760ec4455, 0x37c981dcc395a9ac), // 5^191 + U128.new(0xe070f78d3927556a, 0x85bbe253f47b1417), // 5^192 + U128.new(0x8c469ab843b89562, 0x93956d7478ccec8e), // 5^193 + U128.new(0xaf58416654a6babb, 0x387ac8d1970027b2), // 5^194 + U128.new(0xdb2e51bfe9d0696a, 0x6997b05fcc0319e), // 5^195 + U128.new(0x88fcf317f22241e2, 0x441fece3bdf81f03), // 5^196 + U128.new(0xab3c2fddeeaad25a, 0xd527e81cad7626c3), // 5^197 + U128.new(0xd60b3bd56a5586f1, 0x8a71e223d8d3b074), // 5^198 + U128.new(0x85c7056562757456, 0xf6872d5667844e49), // 5^199 + U128.new(0xa738c6bebb12d16c, 0xb428f8ac016561db), // 5^200 + U128.new(0xd106f86e69d785c7, 0xe13336d701beba52), // 5^201 + U128.new(0x82a45b450226b39c, 0xecc0024661173473), // 5^202 + U128.new(0xa34d721642b06084, 0x27f002d7f95d0190), // 5^203 + U128.new(0xcc20ce9bd35c78a5, 0x31ec038df7b441f4), // 5^204 + U128.new(0xff290242c83396ce, 0x7e67047175a15271), // 5^205 + U128.new(0x9f79a169bd203e41, 0xf0062c6e984d386), // 5^206 + U128.new(0xc75809c42c684dd1, 0x52c07b78a3e60868), // 5^207 + U128.new(0xf92e0c3537826145, 0xa7709a56ccdf8a82), // 5^208 + U128.new(0x9bbcc7a142b17ccb, 0x88a66076400bb691), // 5^209 + U128.new(0xc2abf989935ddbfe, 0x6acff893d00ea435), // 5^210 + U128.new(0xf356f7ebf83552fe, 0x583f6b8c4124d43), // 5^211 + U128.new(0x98165af37b2153de, 0xc3727a337a8b704a), // 5^212 + U128.new(0xbe1bf1b059e9a8d6, 0x744f18c0592e4c5c), // 5^213 + U128.new(0xeda2ee1c7064130c, 0x1162def06f79df73), // 5^214 + U128.new(0x9485d4d1c63e8be7, 0x8addcb5645ac2ba8), // 5^215 + U128.new(0xb9a74a0637ce2ee1, 0x6d953e2bd7173692), // 5^216 + U128.new(0xe8111c87c5c1ba99, 0xc8fa8db6ccdd0437), // 5^217 + U128.new(0x910ab1d4db9914a0, 0x1d9c9892400a22a2), // 5^218 + U128.new(0xb54d5e4a127f59c8, 0x2503beb6d00cab4b), // 5^219 + U128.new(0xe2a0b5dc971f303a, 0x2e44ae64840fd61d), // 5^220 + U128.new(0x8da471a9de737e24, 0x5ceaecfed289e5d2), // 5^221 + U128.new(0xb10d8e1456105dad, 0x7425a83e872c5f47), // 5^222 + U128.new(0xdd50f1996b947518, 0xd12f124e28f77719), // 5^223 + U128.new(0x8a5296ffe33cc92f, 0x82bd6b70d99aaa6f), // 5^224 + U128.new(0xace73cbfdc0bfb7b, 0x636cc64d1001550b), // 5^225 + U128.new(0xd8210befd30efa5a, 0x3c47f7e05401aa4e), // 5^226 + U128.new(0x8714a775e3e95c78, 0x65acfaec34810a71), // 5^227 + U128.new(0xa8d9d1535ce3b396, 0x7f1839a741a14d0d), // 5^228 + U128.new(0xd31045a8341ca07c, 0x1ede48111209a050), // 5^229 + U128.new(0x83ea2b892091e44d, 0x934aed0aab460432), // 5^230 + U128.new(0xa4e4b66b68b65d60, 0xf81da84d5617853f), // 5^231 + U128.new(0xce1de40642e3f4b9, 0x36251260ab9d668e), // 5^232 + U128.new(0x80d2ae83e9ce78f3, 0xc1d72b7c6b426019), // 5^233 + U128.new(0xa1075a24e4421730, 0xb24cf65b8612f81f), // 5^234 + U128.new(0xc94930ae1d529cfc, 0xdee033f26797b627), // 5^235 + U128.new(0xfb9b7cd9a4a7443c, 0x169840ef017da3b1), // 5^236 + U128.new(0x9d412e0806e88aa5, 0x8e1f289560ee864e), // 5^237 + U128.new(0xc491798a08a2ad4e, 0xf1a6f2bab92a27e2), // 5^238 + U128.new(0xf5b5d7ec8acb58a2, 0xae10af696774b1db), // 5^239 + U128.new(0x9991a6f3d6bf1765, 0xacca6da1e0a8ef29), // 5^240 + U128.new(0xbff610b0cc6edd3f, 0x17fd090a58d32af3), // 5^241 + U128.new(0xeff394dcff8a948e, 0xddfc4b4cef07f5b0), // 5^242 + U128.new(0x95f83d0a1fb69cd9, 0x4abdaf101564f98e), // 5^243 + U128.new(0xbb764c4ca7a4440f, 0x9d6d1ad41abe37f1), // 5^244 + U128.new(0xea53df5fd18d5513, 0x84c86189216dc5ed), // 5^245 + U128.new(0x92746b9be2f8552c, 0x32fd3cf5b4e49bb4), // 5^246 + U128.new(0xb7118682dbb66a77, 0x3fbc8c33221dc2a1), // 5^247 + U128.new(0xe4d5e82392a40515, 0xfabaf3feaa5334a), // 5^248 + U128.new(0x8f05b1163ba6832d, 0x29cb4d87f2a7400e), // 5^249 + U128.new(0xb2c71d5bca9023f8, 0x743e20e9ef511012), // 5^250 + U128.new(0xdf78e4b2bd342cf6, 0x914da9246b255416), // 5^251 + U128.new(0x8bab8eefb6409c1a, 0x1ad089b6c2f7548e), // 5^252 + U128.new(0xae9672aba3d0c320, 0xa184ac2473b529b1), // 5^253 + U128.new(0xda3c0f568cc4f3e8, 0xc9e5d72d90a2741e), // 5^254 + U128.new(0x8865899617fb1871, 0x7e2fa67c7a658892), // 5^255 + U128.new(0xaa7eebfb9df9de8d, 0xddbb901b98feeab7), // 5^256 + U128.new(0xd51ea6fa85785631, 0x552a74227f3ea565), // 5^257 + U128.new(0x8533285c936b35de, 0xd53a88958f87275f), // 5^258 + U128.new(0xa67ff273b8460356, 0x8a892abaf368f137), // 5^259 + U128.new(0xd01fef10a657842c, 0x2d2b7569b0432d85), // 5^260 + U128.new(0x8213f56a67f6b29b, 0x9c3b29620e29fc73), // 5^261 + U128.new(0xa298f2c501f45f42, 0x8349f3ba91b47b8f), // 5^262 + U128.new(0xcb3f2f7642717713, 0x241c70a936219a73), // 5^263 + U128.new(0xfe0efb53d30dd4d7, 0xed238cd383aa0110), // 5^264 + U128.new(0x9ec95d1463e8a506, 0xf4363804324a40aa), // 5^265 + U128.new(0xc67bb4597ce2ce48, 0xb143c6053edcd0d5), // 5^266 + U128.new(0xf81aa16fdc1b81da, 0xdd94b7868e94050a), // 5^267 + U128.new(0x9b10a4e5e9913128, 0xca7cf2b4191c8326), // 5^268 + U128.new(0xc1d4ce1f63f57d72, 0xfd1c2f611f63a3f0), // 5^269 + U128.new(0xf24a01a73cf2dccf, 0xbc633b39673c8cec), // 5^270 + U128.new(0x976e41088617ca01, 0xd5be0503e085d813), // 5^271 + U128.new(0xbd49d14aa79dbc82, 0x4b2d8644d8a74e18), // 5^272 + U128.new(0xec9c459d51852ba2, 0xddf8e7d60ed1219e), // 5^273 + U128.new(0x93e1ab8252f33b45, 0xcabb90e5c942b503), // 5^274 + U128.new(0xb8da1662e7b00a17, 0x3d6a751f3b936243), // 5^275 + U128.new(0xe7109bfba19c0c9d, 0xcc512670a783ad4), // 5^276 + U128.new(0x906a617d450187e2, 0x27fb2b80668b24c5), // 5^277 + U128.new(0xb484f9dc9641e9da, 0xb1f9f660802dedf6), // 5^278 + U128.new(0xe1a63853bbd26451, 0x5e7873f8a0396973), // 5^279 + U128.new(0x8d07e33455637eb2, 0xdb0b487b6423e1e8), // 5^280 + U128.new(0xb049dc016abc5e5f, 0x91ce1a9a3d2cda62), // 5^281 + U128.new(0xdc5c5301c56b75f7, 0x7641a140cc7810fb), // 5^282 + U128.new(0x89b9b3e11b6329ba, 0xa9e904c87fcb0a9d), // 5^283 + U128.new(0xac2820d9623bf429, 0x546345fa9fbdcd44), // 5^284 + U128.new(0xd732290fbacaf133, 0xa97c177947ad4095), // 5^285 + U128.new(0x867f59a9d4bed6c0, 0x49ed8eabcccc485d), // 5^286 + U128.new(0xa81f301449ee8c70, 0x5c68f256bfff5a74), // 5^287 + U128.new(0xd226fc195c6a2f8c, 0x73832eec6fff3111), // 5^288 + U128.new(0x83585d8fd9c25db7, 0xc831fd53c5ff7eab), // 5^289 + U128.new(0xa42e74f3d032f525, 0xba3e7ca8b77f5e55), // 5^290 + U128.new(0xcd3a1230c43fb26f, 0x28ce1bd2e55f35eb), // 5^291 + U128.new(0x80444b5e7aa7cf85, 0x7980d163cf5b81b3), // 5^292 + U128.new(0xa0555e361951c366, 0xd7e105bcc332621f), // 5^293 + U128.new(0xc86ab5c39fa63440, 0x8dd9472bf3fefaa7), // 5^294 + U128.new(0xfa856334878fc150, 0xb14f98f6f0feb951), // 5^295 + U128.new(0x9c935e00d4b9d8d2, 0x6ed1bf9a569f33d3), // 5^296 + U128.new(0xc3b8358109e84f07, 0xa862f80ec4700c8), // 5^297 + U128.new(0xf4a642e14c6262c8, 0xcd27bb612758c0fa), // 5^298 + U128.new(0x98e7e9cccfbd7dbd, 0x8038d51cb897789c), // 5^299 + U128.new(0xbf21e44003acdd2c, 0xe0470a63e6bd56c3), // 5^300 + U128.new(0xeeea5d5004981478, 0x1858ccfce06cac74), // 5^301 + U128.new(0x95527a5202df0ccb, 0xf37801e0c43ebc8), // 5^302 + U128.new(0xbaa718e68396cffd, 0xd30560258f54e6ba), // 5^303 + U128.new(0xe950df20247c83fd, 0x47c6b82ef32a2069), // 5^304 + U128.new(0x91d28b7416cdd27e, 0x4cdc331d57fa5441), // 5^305 + U128.new(0xb6472e511c81471d, 0xe0133fe4adf8e952), // 5^306 + U128.new(0xe3d8f9e563a198e5, 0x58180fddd97723a6), // 5^307 + U128.new(0x8e679c2f5e44ff8f, 0x570f09eaa7ea7648), // 5^308 +}; diff --git a/lib/std/fmt/parse_float/convert_fast.zig b/lib/std/fmt/parse_float/convert_fast.zig new file mode 100644 index 0000000000..75875156fd --- /dev/null +++ b/lib/std/fmt/parse_float/convert_fast.zig @@ -0,0 +1,130 @@ +//! Representation of a float as the signficant digits and exponent. +//! The fast path algorithm using machine-sized integers and floats. +//! +//! This only works if both the mantissa and the exponent can be exactly +//! represented as a machine float, since IEE-754 guarantees no rounding +//! will occur. +//! +//! There is an exception: disguised fast-path cases, where we can shift +//! powers-of-10 from the exponent to the significant digits. + +const std = @import("std"); +const math = std.math; +const common = @import("common.zig"); +const FloatInfo = @import("FloatInfo.zig"); +const Number = common.Number; +const floatFromU64 = common.floatFromU64; + +fn isFastPath(comptime T: type, n: Number(T)) bool { + const info = FloatInfo.from(T); + + return info.min_exponent_fast_path <= n.exponent and + n.exponent <= info.max_exponent_fast_path_disguised and + n.mantissa <= info.max_mantissa_fast_path and + !n.many_digits; +} + +// upper bound for tables is floor(mantissaDigits(T) / log2(5)) +// for f64 this is floor(53 / log2(5)) = 22. +// +// Must have max_disguised_fast_path - max_exponent_fast_path entries. (82 - 48 = 34 for f128) +fn fastPow10(comptime T: type, i: usize) T { + return switch (T) { + f16 => ([8]f16{ + 1e0, 1e1, 1e2, 1e3, 1e4, 0, 0, 0, + })[i & 7], + + f32 => ([16]f32{ + 1e0, 1e1, 1e2, 1e3, 1e4, 1e5, 1e6, 1e7, + 1e8, 1e9, 1e10, 0, 0, 0, 0, 0, + })[i & 15], + + f64 => ([32]f64{ + 1e0, 1e1, 1e2, 1e3, 1e4, 1e5, 1e6, 1e7, + 1e8, 1e9, 1e10, 1e11, 1e12, 1e13, 1e14, 1e15, + 1e16, 1e17, 1e18, 1e19, 1e20, 1e21, 1e22, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + })[i & 31], + + f128 => ([64]f128{ + 1e0, 1e1, 1e2, 1e3, 1e4, 1e5, 1e6, 1e7, + 1e8, 1e9, 1e10, 1e11, 1e12, 1e13, 1e14, 1e15, + 1e16, 1e17, 1e18, 1e19, 1e20, 1e21, 1e22, 1e23, + 1e24, 1e25, 1e26, 1e27, 1e28, 1e29, 1e30, 1e31, + 1e32, 1e33, 1e34, 1e35, 1e36, 1e37, 1e38, 1e39, + 1e40, 1e41, 1e42, 1e43, 1e44, 1e45, 1e46, 1e47, + 1e48, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + })[i & 63], + + else => unreachable, + }; +} + +fn fastIntPow10(comptime T: type, i: usize) T { + return switch (T) { + u64 => ([16]u64{ + 1, 10, 100, 1000, + 10000, 100000, 1000000, 10000000, + 100000000, 1000000000, 10000000000, 100000000000, + 1000000000000, 10000000000000, 100000000000000, 1000000000000000, + })[i], + + u128 => ([35]u128{ + 1, 10, + 100, 1000, + 10000, 100000, + 1000000, 10000000, + 100000000, 1000000000, + 10000000000, 100000000000, + 1000000000000, 10000000000000, + 100000000000000, 1000000000000000, + 10000000000000000, 100000000000000000, + 1000000000000000000, 10000000000000000000, + 100000000000000000000, 1000000000000000000000, + 10000000000000000000000, 100000000000000000000000, + 1000000000000000000000000, 10000000000000000000000000, + 100000000000000000000000000, 1000000000000000000000000000, + 10000000000000000000000000000, 100000000000000000000000000000, + 1000000000000000000000000000000, 10000000000000000000000000000000, + 100000000000000000000000000000000, 1000000000000000000000000000000000, + 10000000000000000000000000000000000, + })[i], + + else => unreachable, + }; +} + +pub fn convertFast(comptime T: type, n: Number(T)) ?T { + const MantissaT = common.mantissaType(T); + + if (!isFastPath(T, n)) { + return null; + } + + // TODO: x86 (no SSE/SSE2) requires x87 FPU to be setup correctly with fldcw + const info = FloatInfo.from(T); + + var value: T = 0; + if (n.exponent <= info.max_exponent_fast_path) { + // normal fast path + value = @intToFloat(T, n.mantissa); + value = if (n.exponent < 0) + value / fastPow10(T, @intCast(usize, -n.exponent)) + else + value * fastPow10(T, @intCast(usize, n.exponent)); + } else { + // disguised fast path + const shift = n.exponent - info.max_exponent_fast_path; + const mantissa = math.mul(MantissaT, n.mantissa, fastIntPow10(MantissaT, @intCast(usize, shift))) catch return null; + if (mantissa > info.max_mantissa_fast_path) { + return null; + } + value = @intToFloat(T, mantissa) * fastPow10(T, info.max_exponent_fast_path); + } + + if (n.negative) { + value = -value; + } + return value; +} diff --git a/lib/std/fmt/parse_float/convert_hex.zig b/lib/std/fmt/parse_float/convert_hex.zig new file mode 100644 index 0000000000..3b3f797216 --- /dev/null +++ b/lib/std/fmt/parse_float/convert_hex.zig @@ -0,0 +1,89 @@ +//! Conversion of hex-float representation into an accurate value. +// +// Derived from golang strconv/atof.go. + +const std = @import("std"); +const math = std.math; +const common = @import("common.zig"); +const Number = common.Number; +const floatFromUnsigned = common.floatFromUnsigned; + +// converts the form 0xMMM.NNNpEEE. +// +// MMM.NNN = mantissa +// EEE = exponent +// +// MMM.NNN is stored as an integer, the exponent is offset. +pub fn convertHex(comptime T: type, n_: Number(T)) T { + const MantissaT = common.mantissaType(T); + var n = n_; + + if (n.mantissa == 0) { + return if (n.negative) -0.0 else 0.0; + } + + const max_exp = math.floatExponentMax(T); + const min_exp = math.floatExponentMin(T); + const mantissa_bits = math.floatMantissaBits(T); + const exp_bits = math.floatExponentBits(T); + const exp_bias = min_exp - 1; + + // mantissa now implicitly divided by 2^mantissa_bits + n.exponent += mantissa_bits; + + // Shift mantissa and exponent to bring representation into float range. + // Eventually we want a mantissa with a leading 1-bit followed by mantbits other bits. + // For rounding, we need two more, where the bottom bit represents + // whether that bit or any later bit was non-zero. + // (If the mantissa has already lost non-zero bits, trunc is true, + // and we OR in a 1 below after shifting left appropriately.) + while (n.mantissa != 0 and n.mantissa >> (mantissa_bits + 2) == 0) { + n.mantissa <<= 1; + n.exponent -= 1; + } + if (n.many_digits) { + n.mantissa |= 1; + } + while (n.mantissa >> (1 + mantissa_bits + 2) != 0) { + n.mantissa = (n.mantissa >> 1) | (n.mantissa & 1); + n.exponent += 1; + } + + // If exponent is too negative, + // denormalize in hopes of making it representable. + // (The -2 is for the rounding bits.) + while (n.mantissa > 1 and n.exponent < min_exp - 2) { + n.mantissa = (n.mantissa >> 1) | (n.mantissa & 1); + n.exponent += 1; + } + + // Round using two bottom bits. + var round = n.mantissa & 3; + n.mantissa >>= 2; + round |= n.mantissa & 1; // round to even (round up if mantissa is odd) + n.exponent += 2; + if (round == 3) { + n.mantissa += 1; + if (n.mantissa == 1 << (1 + mantissa_bits)) { + n.mantissa >>= 1; + n.exponent += 1; + } + } + + // Denormal or zero + if (n.mantissa >> mantissa_bits == 0) { + n.exponent = exp_bias; + } + + // Infinity and range error + if (n.exponent > max_exp) { + return math.inf(T); + } + + var bits = n.mantissa & ((1 << mantissa_bits) - 1); + bits |= @intCast(MantissaT, (n.exponent - exp_bias) & ((1 << exp_bits) - 1)) << mantissa_bits; + if (n.negative) { + bits |= 1 << (mantissa_bits + exp_bits); + } + return floatFromUnsigned(T, MantissaT, bits); +} diff --git a/lib/std/fmt/parse_float/convert_slow.zig b/lib/std/fmt/parse_float/convert_slow.zig new file mode 100644 index 0000000000..225a1e208c --- /dev/null +++ b/lib/std/fmt/parse_float/convert_slow.zig @@ -0,0 +1,114 @@ +const std = @import("std"); +const math = std.math; +const common = @import("common.zig"); +const BiasedFp = common.BiasedFp; +const Decimal = @import("decimal.zig").Decimal; +const mantissaType = common.mantissaType; + +const max_shift = 60; +const num_powers = 19; +const powers = [_]u8{ 0, 3, 6, 9, 13, 16, 19, 23, 26, 29, 33, 36, 39, 43, 46, 49, 53, 56, 59 }; + +pub fn getShift(n: usize) usize { + return if (n < num_powers) powers[n] else max_shift; +} + +/// Parse the significant digits and biased, binary exponent of a float. +/// +/// This is a fallback algorithm that uses a big-integer representation +/// of the float, and therefore is considerably slower than faster +/// approximations. However, it will always determine how to round +/// the significant digits to the nearest machine float, allowing +/// use to handle near half-way cases. +/// +/// Near half-way cases are halfway between two consecutive machine floats. +/// For example, the float `16777217.0` has a bitwise representation of +/// `100000000000000000000000 1`. Rounding to a single-precision float, +/// the trailing `1` is truncated. Using round-nearest, tie-even, any +/// value above `16777217.0` must be rounded up to `16777218.0`, while +/// any value before or equal to `16777217.0` must be rounded down +/// to `16777216.0`. These near-halfway conversions therefore may require +/// a large number of digits to unambiguously determine how to round. +/// +/// The algorithms described here are based on "Processing Long Numbers Quickly", +/// available here: . +pub fn convertSlow(comptime T: type, s: []const u8) BiasedFp(T) { + const MantissaT = mantissaType(T); + const min_exponent = -(1 << (math.floatExponentBits(T) - 1)) + 1; + const infinite_power = (1 << math.floatExponentBits(T)) - 1; + const mantissa_explicit_bits = math.floatMantissaBits(T); + + var d = Decimal(T).parse(s); // no need to recheck underscores + if (d.num_digits == 0 or d.decimal_point < Decimal(T).min_exponent) { + return BiasedFp(T).zero(); + } else if (d.decimal_point >= Decimal(T).max_exponent) { + return BiasedFp(T).inf(T); + } + + var exp2: i32 = 0; + // Shift right toward (1/2 .. 1] + while (d.decimal_point > 0) { + const n = @intCast(usize, d.decimal_point); + const shift = getShift(n); + d.rightShift(shift); + if (d.decimal_point < -Decimal(T).decimal_point_range) { + return BiasedFp(T).zero(); + } + exp2 += @intCast(i32, shift); + } + // Shift left toward (1/2 .. 1] + while (d.decimal_point <= 0) { + const shift = blk: { + if (d.decimal_point == 0) { + break :blk switch (d.digits[0]) { + 5...9 => break, + 0, 1 => @as(usize, 2), + else => 1, + }; + } else { + const n = @intCast(usize, -d.decimal_point); + break :blk getShift(n); + } + }; + d.leftShift(shift); + if (d.decimal_point > Decimal(T).decimal_point_range) { + return BiasedFp(T).inf(T); + } + exp2 -= @intCast(i32, shift); + } + // We are now in the range [1/2 .. 1] but the binary format uses [1 .. 2] + exp2 -= 1; + while (min_exponent + 1 > exp2) { + var n = @intCast(usize, (min_exponent + 1) - exp2); + if (n > max_shift) { + n = max_shift; + } + d.rightShift(n); + exp2 += @intCast(i32, n); + } + if (exp2 - min_exponent >= infinite_power) { + return BiasedFp(T).inf(T); + } + + // Shift the decimal to the hidden bit, and then round the value + // to get the high mantissa+1 bits. + d.leftShift(mantissa_explicit_bits + 1); + var mantissa = d.round(); + if (mantissa >= (@as(MantissaT, 1) << (mantissa_explicit_bits + 1))) { + // Rounding up overflowed to the carry bit, need to + // shift back to the hidden bit. + d.rightShift(1); + exp2 += 1; + mantissa = d.round(); + if ((exp2 - min_exponent) >= infinite_power) { + return BiasedFp(T).inf(T); + } + } + var power2 = exp2 - min_exponent; + if (mantissa < (@as(MantissaT, 1) << mantissa_explicit_bits)) { + power2 -= 1; + } + // Zero out all the bits above the explicit mantissa bits. + mantissa &= (@as(MantissaT, 1) << mantissa_explicit_bits) - 1; + return .{ .f = mantissa, .e = power2 }; +} diff --git a/lib/std/fmt/parse_float/decimal.zig b/lib/std/fmt/parse_float/decimal.zig new file mode 100644 index 0000000000..1a4c7ebb0d --- /dev/null +++ b/lib/std/fmt/parse_float/decimal.zig @@ -0,0 +1,493 @@ +const std = @import("std"); +const math = std.math; +const common = @import("common.zig"); +const FloatStream = @import("FloatStream.zig"); +const isEightDigits = @import("common.zig").isEightDigits; +const mantissaType = common.mantissaType; + +// Arbitrary-precision decimal class for fallback algorithms. +// +// This is only used if the fast-path (native floats) and +// the Eisel-Lemire algorithm are unable to unambiguously +// determine the float. +// +// The technique used is "Simple Decimal Conversion", developed +// by Nigel Tao and Ken Thompson. A detailed description of the +// algorithm can be found in "ParseNumberF64 by Simple Decimal Conversion", +// available online: . +// +// Big-decimal implementation. We do not use the big.Int routines since we only require a maximum +// fixed region of memory. Further, we require only a small subset of operations. +// +// This accepts a floating point parameter and will generate a Decimal which can correctly parse +// the input with sufficient accuracy. Internally this means either a u64 mantissa (f16, f32 or f64) +// or a u128 mantissa (f128). +pub fn Decimal(comptime T: type) type { + const MantissaT = mantissaType(T); + std.debug.assert(MantissaT == u64 or MantissaT == u128); + + return struct { + const Self = @This(); + + /// The maximum number of digits required to unambiguously round a float. + /// + /// For a double-precision IEEE-754 float, this required 767 digits, + /// so we store the max digits + 1. + /// + /// We can exactly represent a float in radix `b` from radix 2 if + /// `b` is divisible by 2. This function calculates the exact number of + /// digits required to exactly represent that float. + /// + /// According to the "Handbook of Floating Point Arithmetic", + /// for IEEE754, with emin being the min exponent, p2 being the + /// precision, and b being the radix, the number of digits follows as: + /// + /// `−emin + p2 + ⌊(emin + 1) log(2, b) − log(1 − 2^(−p2), b)⌋` + /// + /// For f32, this follows as: + /// emin = -126 + /// p2 = 24 + /// + /// For f64, this follows as: + /// emin = -1022 + /// p2 = 53 + /// + /// For f128, this follows as: + /// emin = -16383 + /// p2 = 112 + /// + /// In Python: + /// `-emin + p2 + math.floor((emin+ 1)*math.log(2, b)-math.log(1-2**(-p2), b))` + pub const max_digits = if (MantissaT == u64) 768 else 11564; + /// The max digits that can be exactly represented in a 64-bit integer. + pub const max_digits_without_overflow = if (MantissaT == u64) 19 else 38; + pub const decimal_point_range = if (MantissaT == u64) 2047 else 32767; + pub const min_exponent = if (MantissaT == u64) -324 else -4966; + pub const max_exponent = if (MantissaT == u64) 310 else 4933; + pub const max_decimal_digits = if (MantissaT == u64) 18 else 37; + + /// The number of significant digits in the decimal. + num_digits: usize, + /// The offset of the decimal point in the significant digits. + decimal_point: i32, + /// If the number of significant digits stored in the decimal is truncated. + truncated: bool, + /// buffer of the raw digits, in the range [0, 9]. + digits: [max_digits]u8, + + pub fn new() Self { + return .{ + .num_digits = 0, + .decimal_point = 0, + .truncated = false, + .digits = [_]u8{0} ** max_digits, + }; + } + + /// Append a digit to the buffer + pub fn tryAddDigit(self: *Self, digit: u8) void { + if (self.num_digits < max_digits) { + self.digits[self.num_digits] = digit; + } + self.num_digits += 1; + } + + /// Trim trailing zeroes from the buffer + pub fn trim(self: *Self) void { + // All of the following calls to `Self::trim` can't panic because: + // + // 1. `parse_decimal` sets `num_digits` to a max of `max_digits`. + // 2. `right_shift` sets `num_digits` to `write_index`, which is bounded by `num_digits`. + // 3. `left_shift` `num_digits` to a max of `max_digits`. + // + // Trim is only called in `right_shift` and `left_shift`. + std.debug.assert(self.num_digits <= max_digits); + while (self.num_digits != 0 and self.digits[self.num_digits - 1] == 0) { + self.num_digits -= 1; + } + } + + pub fn round(self: *Self) MantissaT { + if (self.num_digits == 0 or self.decimal_point < 0) { + return 0; + } else if (self.decimal_point > max_decimal_digits) { + return math.maxInt(MantissaT); + } + + const dp = @intCast(usize, self.decimal_point); + var n: MantissaT = 0; + + var i: usize = 0; + while (i < dp) : (i += 1) { + n *= 10; + if (i < self.num_digits) { + n += @as(MantissaT, self.digits[i]); + } + } + + var round_up = false; + if (dp < self.num_digits) { + round_up = self.digits[dp] >= 5; + if (self.digits[dp] == 5 and dp + 1 == self.num_digits) { + round_up = self.truncated or ((dp != 0) and (1 & self.digits[dp - 1] != 0)); + } + } + if (round_up) { + n += 1; + } + return n; + } + + /// Computes decimal * 2^shift. + pub fn leftShift(self: *Self, shift: usize) void { + if (self.num_digits == 0) { + return; + } + const num_new_digits = self.numberOfDigitsLeftShift(shift); + var read_index = self.num_digits; + var write_index = self.num_digits + num_new_digits; + var n: MantissaT = 0; + while (read_index != 0) { + read_index -= 1; + write_index -= 1; + n += math.shl(MantissaT, self.digits[read_index], shift); + + const quotient = n / 10; + const remainder = n - (10 * quotient); + if (write_index < max_digits) { + self.digits[write_index] = @intCast(u8, remainder); + } else if (remainder > 0) { + self.truncated = true; + } + n = quotient; + } + while (n > 0) { + write_index -= 1; + + const quotient = n / 10; + const remainder = n - (10 * quotient); + if (write_index < max_digits) { + self.digits[write_index] = @intCast(u8, remainder); + } else if (remainder > 0) { + self.truncated = true; + } + n = quotient; + } + + self.num_digits += num_new_digits; + if (self.num_digits > max_digits) { + self.num_digits = max_digits; + } + self.decimal_point += @intCast(i32, num_new_digits); + self.trim(); + } + + /// Computes decimal * 2^-shift. + pub fn rightShift(self: *Self, shift: usize) void { + var read_index: usize = 0; + var write_index: usize = 0; + var n: MantissaT = 0; + while (math.shr(MantissaT, n, shift) == 0) { + if (read_index < self.num_digits) { + n = (10 * n) + self.digits[read_index]; + read_index += 1; + } else if (n == 0) { + return; + } else { + while (math.shr(MantissaT, n, shift) == 0) { + n *= 10; + read_index += 1; + } + break; + } + } + + self.decimal_point -= @intCast(i32, read_index) - 1; + if (self.decimal_point < -decimal_point_range) { + self.num_digits = 0; + self.decimal_point = 0; + self.truncated = false; + return; + } + + const mask = math.shl(MantissaT, 1, shift) - 1; + while (read_index < self.num_digits) { + const new_digit = @intCast(u8, math.shr(MantissaT, n, shift)); + n = (10 * (n & mask)) + self.digits[read_index]; + read_index += 1; + self.digits[write_index] = new_digit; + write_index += 1; + } + while (n > 0) { + const new_digit = @intCast(u8, math.shr(MantissaT, n, shift)); + n = 10 * (n & mask); + if (write_index < max_digits) { + self.digits[write_index] = new_digit; + write_index += 1; + } else if (new_digit > 0) { + self.truncated = true; + } + } + self.num_digits = write_index; + self.trim(); + } + + /// Parse a bit integer representation of the float as a decimal. + // We do not verify underscores in this path since these will have been verified + // via parse.parseNumber so can assume the number is well-formed. + // This code-path does not have to handle hex-floats since these will always be handled via another + // function prior to this. + pub fn parse(s: []const u8) Self { + var d = Self.new(); + var stream = FloatStream.init(s); + + stream.skipChars2('0', '_'); + while (stream.scanDigit(10)) |digit| { + d.tryAddDigit(digit); + } + + if (stream.firstIs('.')) { + stream.advance(1); + const marker = stream.offsetTrue(); + + // Skip leading zeroes + if (d.num_digits == 0) { + stream.skipChars('0'); + } + + while (stream.hasLen(8) and d.num_digits + 8 < max_digits) { + const v = stream.readU64Unchecked(); + if (!isEightDigits(v)) { + break; + } + std.mem.writeIntSliceLittle(u64, d.digits[d.num_digits..], v - 0x3030_3030_3030_3030); + d.num_digits += 8; + stream.advance(8); + } + + while (stream.scanDigit(10)) |digit| { + d.tryAddDigit(digit); + } + d.decimal_point = @intCast(i32, marker) - @intCast(i32, stream.offsetTrue()); + } + if (d.num_digits != 0) { + // Ignore trailing zeros if any + var n_trailing_zeros: usize = 0; + var i = stream.offsetTrue() - 1; + while (true) { + if (s[i] == '0') { + n_trailing_zeros += 1; + } else if (s[i] != '.') { + break; + } + + i -= 1; + if (i == 0) break; + } + d.decimal_point += @intCast(i32, n_trailing_zeros); + d.num_digits -= n_trailing_zeros; + d.decimal_point += @intCast(i32, d.num_digits); + if (d.num_digits > max_digits) { + d.truncated = true; + d.num_digits = max_digits; + } + } + if (stream.firstIsLower('e')) { + stream.advance(1); + var neg_exp = false; + if (stream.firstIs('-')) { + neg_exp = true; + stream.advance(1); + } else if (stream.firstIs('+')) { + stream.advance(1); + } + var exp_num: i32 = 0; + while (stream.scanDigit(10)) |digit| { + if (exp_num < 0x10000) { + exp_num = 10 * exp_num + digit; + } + } + d.decimal_point += if (neg_exp) -exp_num else exp_num; + } + + var i = d.num_digits; + while (i < max_digits_without_overflow) : (i += 1) { + d.digits[i] = 0; + } + + return d; + } + + // Compute the number decimal digits introduced by a base-2 shift. This is performed + // by storing the leading digits of 1/2^i = 5^i and using these along with the cut-off + // value to quickly determine the decimal shift from binary. + // + // See also https://github.com/golang/go/blob/go1.15.3/src/strconv/decimal.go#L163 for + // another description of the method. + pub fn numberOfDigitsLeftShift(self: *Self, shift: usize) usize { + const ShiftCutoff = struct { + delta: u8, + cutoff: []const u8, + }; + + // Leading digits of 1/2^i = 5^i. + // + // ``` + // import math + // + // bits = 128 + // for i in range(bits): + // log2 = math.log(2)/math.log(10) + // print(f'.{{ .delta = {int(log2*i+1)}, .cutoff = "{5**i}" }}, // {2**i}') + // ``` + const pow2_to_pow5_table = [_]ShiftCutoff{ + .{ .delta = 0, .cutoff = "" }, + .{ .delta = 1, .cutoff = "5" }, // 2 + .{ .delta = 1, .cutoff = "25" }, // 4 + .{ .delta = 1, .cutoff = "125" }, // 8 + .{ .delta = 2, .cutoff = "625" }, // 16 + .{ .delta = 2, .cutoff = "3125" }, // 32 + .{ .delta = 2, .cutoff = "15625" }, // 64 + .{ .delta = 3, .cutoff = "78125" }, // 128 + .{ .delta = 3, .cutoff = "390625" }, // 256 + .{ .delta = 3, .cutoff = "1953125" }, // 512 + .{ .delta = 4, .cutoff = "9765625" }, // 1024 + .{ .delta = 4, .cutoff = "48828125" }, // 2048 + .{ .delta = 4, .cutoff = "244140625" }, // 4096 + .{ .delta = 4, .cutoff = "1220703125" }, // 8192 + .{ .delta = 5, .cutoff = "6103515625" }, // 16384 + .{ .delta = 5, .cutoff = "30517578125" }, // 32768 + .{ .delta = 5, .cutoff = "152587890625" }, // 65536 + .{ .delta = 6, .cutoff = "762939453125" }, // 131072 + .{ .delta = 6, .cutoff = "3814697265625" }, // 262144 + .{ .delta = 6, .cutoff = "19073486328125" }, // 524288 + .{ .delta = 7, .cutoff = "95367431640625" }, // 1048576 + .{ .delta = 7, .cutoff = "476837158203125" }, // 2097152 + .{ .delta = 7, .cutoff = "2384185791015625" }, // 4194304 + .{ .delta = 7, .cutoff = "11920928955078125" }, // 8388608 + .{ .delta = 8, .cutoff = "59604644775390625" }, // 16777216 + .{ .delta = 8, .cutoff = "298023223876953125" }, // 33554432 + .{ .delta = 8, .cutoff = "1490116119384765625" }, // 67108864 + .{ .delta = 9, .cutoff = "7450580596923828125" }, // 134217728 + .{ .delta = 9, .cutoff = "37252902984619140625" }, // 268435456 + .{ .delta = 9, .cutoff = "186264514923095703125" }, // 536870912 + .{ .delta = 10, .cutoff = "931322574615478515625" }, // 1073741824 + .{ .delta = 10, .cutoff = "4656612873077392578125" }, // 2147483648 + .{ .delta = 10, .cutoff = "23283064365386962890625" }, // 4294967296 + .{ .delta = 10, .cutoff = "116415321826934814453125" }, // 8589934592 + .{ .delta = 11, .cutoff = "582076609134674072265625" }, // 17179869184 + .{ .delta = 11, .cutoff = "2910383045673370361328125" }, // 34359738368 + .{ .delta = 11, .cutoff = "14551915228366851806640625" }, // 68719476736 + .{ .delta = 12, .cutoff = "72759576141834259033203125" }, // 137438953472 + .{ .delta = 12, .cutoff = "363797880709171295166015625" }, // 274877906944 + .{ .delta = 12, .cutoff = "1818989403545856475830078125" }, // 549755813888 + .{ .delta = 13, .cutoff = "9094947017729282379150390625" }, // 1099511627776 + .{ .delta = 13, .cutoff = "45474735088646411895751953125" }, // 2199023255552 + .{ .delta = 13, .cutoff = "227373675443232059478759765625" }, // 4398046511104 + .{ .delta = 13, .cutoff = "1136868377216160297393798828125" }, // 8796093022208 + .{ .delta = 14, .cutoff = "5684341886080801486968994140625" }, // 17592186044416 + .{ .delta = 14, .cutoff = "28421709430404007434844970703125" }, // 35184372088832 + .{ .delta = 14, .cutoff = "142108547152020037174224853515625" }, // 70368744177664 + .{ .delta = 15, .cutoff = "710542735760100185871124267578125" }, // 140737488355328 + .{ .delta = 15, .cutoff = "3552713678800500929355621337890625" }, // 281474976710656 + .{ .delta = 15, .cutoff = "17763568394002504646778106689453125" }, // 562949953421312 + .{ .delta = 16, .cutoff = "88817841970012523233890533447265625" }, // 1125899906842624 + .{ .delta = 16, .cutoff = "444089209850062616169452667236328125" }, // 2251799813685248 + .{ .delta = 16, .cutoff = "2220446049250313080847263336181640625" }, // 4503599627370496 + .{ .delta = 16, .cutoff = "11102230246251565404236316680908203125" }, // 9007199254740992 + .{ .delta = 17, .cutoff = "55511151231257827021181583404541015625" }, // 18014398509481984 + .{ .delta = 17, .cutoff = "277555756156289135105907917022705078125" }, // 36028797018963968 + .{ .delta = 17, .cutoff = "1387778780781445675529539585113525390625" }, // 72057594037927936 + .{ .delta = 18, .cutoff = "6938893903907228377647697925567626953125" }, // 144115188075855872 + .{ .delta = 18, .cutoff = "34694469519536141888238489627838134765625" }, // 288230376151711744 + .{ .delta = 18, .cutoff = "173472347597680709441192448139190673828125" }, // 576460752303423488 + .{ .delta = 19, .cutoff = "867361737988403547205962240695953369140625" }, // 1152921504606846976 + .{ .delta = 19, .cutoff = "4336808689942017736029811203479766845703125" }, // 2305843009213693952 + .{ .delta = 19, .cutoff = "21684043449710088680149056017398834228515625" }, // 4611686018427387904 + .{ .delta = 19, .cutoff = "108420217248550443400745280086994171142578125" }, // 9223372036854775808 + .{ .delta = 20, .cutoff = "542101086242752217003726400434970855712890625" }, // 18446744073709551616 + .{ .delta = 20, .cutoff = "2710505431213761085018632002174854278564453125" }, // 36893488147419103232 + .{ .delta = 20, .cutoff = "13552527156068805425093160010874271392822265625" }, // 73786976294838206464 + .{ .delta = 21, .cutoff = "67762635780344027125465800054371356964111328125" }, // 147573952589676412928 + .{ .delta = 21, .cutoff = "338813178901720135627329000271856784820556640625" }, // 295147905179352825856 + .{ .delta = 21, .cutoff = "1694065894508600678136645001359283924102783203125" }, // 590295810358705651712 + .{ .delta = 22, .cutoff = "8470329472543003390683225006796419620513916015625" }, // 1180591620717411303424 + .{ .delta = 22, .cutoff = "42351647362715016953416125033982098102569580078125" }, // 2361183241434822606848 + .{ .delta = 22, .cutoff = "211758236813575084767080625169910490512847900390625" }, // 4722366482869645213696 + .{ .delta = 22, .cutoff = "1058791184067875423835403125849552452564239501953125" }, // 9444732965739290427392 + .{ .delta = 23, .cutoff = "5293955920339377119177015629247762262821197509765625" }, // 18889465931478580854784 + .{ .delta = 23, .cutoff = "26469779601696885595885078146238811314105987548828125" }, // 37778931862957161709568 + .{ .delta = 23, .cutoff = "132348898008484427979425390731194056570529937744140625" }, // 75557863725914323419136 + .{ .delta = 24, .cutoff = "661744490042422139897126953655970282852649688720703125" }, // 151115727451828646838272 + .{ .delta = 24, .cutoff = "3308722450212110699485634768279851414263248443603515625" }, // 302231454903657293676544 + .{ .delta = 24, .cutoff = "16543612251060553497428173841399257071316242218017578125" }, // 604462909807314587353088 + .{ .delta = 25, .cutoff = "82718061255302767487140869206996285356581211090087890625" }, // 1208925819614629174706176 + .{ .delta = 25, .cutoff = "413590306276513837435704346034981426782906055450439453125" }, // 2417851639229258349412352 + .{ .delta = 25, .cutoff = "2067951531382569187178521730174907133914530277252197265625" }, // 4835703278458516698824704 + .{ .delta = 25, .cutoff = "10339757656912845935892608650874535669572651386260986328125" }, // 9671406556917033397649408 + .{ .delta = 26, .cutoff = "51698788284564229679463043254372678347863256931304931640625" }, // 19342813113834066795298816 + .{ .delta = 26, .cutoff = "258493941422821148397315216271863391739316284656524658203125" }, // 38685626227668133590597632 + .{ .delta = 26, .cutoff = "1292469707114105741986576081359316958696581423282623291015625" }, // 77371252455336267181195264 + .{ .delta = 27, .cutoff = "6462348535570528709932880406796584793482907116413116455078125" }, // 154742504910672534362390528 + .{ .delta = 27, .cutoff = "32311742677852643549664402033982923967414535582065582275390625" }, // 309485009821345068724781056 + .{ .delta = 27, .cutoff = "161558713389263217748322010169914619837072677910327911376953125" }, // 618970019642690137449562112 + .{ .delta = 28, .cutoff = "807793566946316088741610050849573099185363389551639556884765625" }, // 1237940039285380274899124224 + .{ .delta = 28, .cutoff = "4038967834731580443708050254247865495926816947758197784423828125" }, // 2475880078570760549798248448 + .{ .delta = 28, .cutoff = "20194839173657902218540251271239327479634084738790988922119140625" }, // 4951760157141521099596496896 + .{ .delta = 28, .cutoff = "100974195868289511092701256356196637398170423693954944610595703125" }, // 9903520314283042199192993792 + .{ .delta = 29, .cutoff = "504870979341447555463506281780983186990852118469774723052978515625" }, // 19807040628566084398385987584 + .{ .delta = 29, .cutoff = "2524354896707237777317531408904915934954260592348873615264892578125" }, // 39614081257132168796771975168 + .{ .delta = 29, .cutoff = "12621774483536188886587657044524579674771302961744368076324462890625" }, // 79228162514264337593543950336 + .{ .delta = 30, .cutoff = "63108872417680944432938285222622898373856514808721840381622314453125" }, // 158456325028528675187087900672 + .{ .delta = 30, .cutoff = "315544362088404722164691426113114491869282574043609201908111572265625" }, // 316912650057057350374175801344 + .{ .delta = 30, .cutoff = "1577721810442023610823457130565572459346412870218046009540557861328125" }, // 633825300114114700748351602688 + .{ .delta = 31, .cutoff = "7888609052210118054117285652827862296732064351090230047702789306640625" }, // 1267650600228229401496703205376 + .{ .delta = 31, .cutoff = "39443045261050590270586428264139311483660321755451150238513946533203125" }, // 2535301200456458802993406410752 + .{ .delta = 31, .cutoff = "197215226305252951352932141320696557418301608777255751192569732666015625" }, // 5070602400912917605986812821504 + .{ .delta = 32, .cutoff = "986076131526264756764660706603482787091508043886278755962848663330078125" }, // 10141204801825835211973625643008 + .{ .delta = 32, .cutoff = "4930380657631323783823303533017413935457540219431393779814243316650390625" }, // 20282409603651670423947251286016 + .{ .delta = 32, .cutoff = "24651903288156618919116517665087069677287701097156968899071216583251953125" }, // 40564819207303340847894502572032 + .{ .delta = 32, .cutoff = "123259516440783094595582588325435348386438505485784844495356082916259765625" }, // 81129638414606681695789005144064 + .{ .delta = 33, .cutoff = "616297582203915472977912941627176741932192527428924222476780414581298828125" }, // 162259276829213363391578010288128 + .{ .delta = 33, .cutoff = "3081487911019577364889564708135883709660962637144621112383902072906494140625" }, // 324518553658426726783156020576256 + .{ .delta = 33, .cutoff = "15407439555097886824447823540679418548304813185723105561919510364532470703125" }, // 649037107316853453566312041152512 + .{ .delta = 34, .cutoff = "77037197775489434122239117703397092741524065928615527809597551822662353515625" }, // 1298074214633706907132624082305024 + .{ .delta = 34, .cutoff = "385185988877447170611195588516985463707620329643077639047987759113311767578125" }, // 2596148429267413814265248164610048 + .{ .delta = 34, .cutoff = "1925929944387235853055977942584927318538101648215388195239938795566558837890625" }, // 5192296858534827628530496329220096 + .{ .delta = 35, .cutoff = "9629649721936179265279889712924636592690508241076940976199693977832794189453125" }, // 10384593717069655257060992658440192 + .{ .delta = 35, .cutoff = "48148248609680896326399448564623182963452541205384704880998469889163970947265625" }, // 20769187434139310514121985316880384 + .{ .delta = 35, .cutoff = "240741243048404481631997242823115914817262706026923524404992349445819854736328125" }, // 41538374868278621028243970633760768 + .{ .delta = 35, .cutoff = "1203706215242022408159986214115579574086313530134617622024961747229099273681640625" }, // 83076749736557242056487941267521536 + .{ .delta = 36, .cutoff = "6018531076210112040799931070577897870431567650673088110124808736145496368408203125" }, // 166153499473114484112975882535043072 + .{ .delta = 36, .cutoff = "30092655381050560203999655352889489352157838253365440550624043680727481842041015625" }, // 332306998946228968225951765070086144 + .{ .delta = 36, .cutoff = "150463276905252801019998276764447446760789191266827202753120218403637409210205078125" }, // 664613997892457936451903530140172288 + .{ .delta = 37, .cutoff = "752316384526264005099991383822237233803945956334136013765601092018187046051025390625" }, // 1329227995784915872903807060280344576 + .{ .delta = 37, .cutoff = "3761581922631320025499956919111186169019729781670680068828005460090935230255126953125" }, // 2658455991569831745807614120560689152 + .{ .delta = 37, .cutoff = "18807909613156600127499784595555930845098648908353400344140027300454676151275634765625" }, // 5316911983139663491615228241121378304 + .{ .delta = 38, .cutoff = "94039548065783000637498922977779654225493244541767001720700136502273380756378173828125" }, // 10633823966279326983230456482242756608 + .{ .delta = 38, .cutoff = "470197740328915003187494614888898271127466222708835008603500682511366903781890869140625" }, // 21267647932558653966460912964485513216 + .{ .delta = 38, .cutoff = "2350988701644575015937473074444491355637331113544175043017503412556834518909454345703125" }, // 42535295865117307932921825928971026432 + .{ .delta = 38, .cutoff = "11754943508222875079687365372222456778186655567720875215087517062784172594547271728515625" }, // 85070591730234615865843651857942052864 + .{ .delta = 39, .cutoff = "58774717541114375398436826861112283890933277838604376075437585313920862972736358642578125" }, // 170141183460469231731687303715884105728 + }; + + std.debug.assert(shift < pow2_to_pow5_table.len); + const x = pow2_to_pow5_table[shift]; + + // Compare leading digits of current to check if lexicographically less than cutoff. + for (x.cutoff) |p5, i| { + if (i >= self.num_digits) { + return x.delta - 1; + } else if (self.digits[i] == p5 - '0') { // digits are stored as integers + continue; + } else if (self.digits[i] < p5 - '0') { + return x.delta - 1; + } else { + return x.delta; + } + return x.delta; + } + return x.delta; + } + }; +} diff --git a/lib/std/fmt/parse_float/parse.zig b/lib/std/fmt/parse_float/parse.zig new file mode 100644 index 0000000000..104052c3b3 --- /dev/null +++ b/lib/std/fmt/parse_float/parse.zig @@ -0,0 +1,293 @@ +const std = @import("std"); +const common = @import("common.zig"); +const FloatStream = @import("FloatStream.zig"); +const isEightDigits = common.isEightDigits; +const Number = common.Number; + +/// Parse 8 digits, loaded as bytes in little-endian order. +/// +/// This uses the trick where every digit is in [0x030, 0x39], +/// and therefore can be parsed in 3 multiplications, much +/// faster than the normal 8. +/// +/// This is based off the algorithm described in "Fast numeric string to +/// int", available here: . +fn parse8Digits(v_: u64) u64 { + var v = v_; + const mask = 0x0000_00ff_0000_00ff; + const mul1 = 0x000f_4240_0000_0064; + const mul2 = 0x0000_2710_0000_0001; + v -= 0x3030_3030_3030_3030; + v = (v * 10) + (v >> 8); // will not overflow, fits in 63 bits + const v1 = (v & mask) *% mul1; + const v2 = ((v >> 16) & mask) *% mul2; + return @as(u64, @truncate(u32, (v1 +% v2) >> 32)); +} + +/// Parse digits until a non-digit character is found. +fn tryParseDigits(comptime T: type, stream: *FloatStream, x: *T, comptime base: u8) void { + // Try to parse 8 digits at a time, using an optimized algorithm. + // This only supports decimal digits. + if (base == 10) { + while (stream.hasLen(8)) { + const v = stream.readU64Unchecked(); + if (!isEightDigits(v)) { + break; + } + + x.* = x.* *% 1_0000_0000 +% parse8Digits(v); + stream.advance(8); + } + } + + while (stream.scanDigit(base)) |digit| { + x.* *%= base; + x.* +%= digit; + } +} + +fn min_n_digit_int(comptime T: type, digit_count: usize) T { + var n: T = 1; + var i: usize = 1; + while (i < digit_count) : (i += 1) n *= 10; + return n; +} + +/// Parse up to N digits +fn tryParseNDigits(comptime T: type, stream: *FloatStream, x: *T, comptime base: u8, comptime n: usize) void { + while (x.* < min_n_digit_int(T, n)) { + if (stream.scanDigit(base)) |digit| { + x.* *%= base; + x.* +%= digit; + } else { + break; + } + } +} + +/// Parse the scientific notation component of a float. +fn parseScientific(stream: *FloatStream) ?i64 { + var exponent: i64 = 0; + var negative = false; + + if (stream.first()) |c| { + negative = c == '-'; + if (c == '-' or c == '+') { + stream.advance(1); + } + } + if (stream.firstIsDigit(10)) { + while (stream.scanDigit(10)) |digit| { + // no overflows here, saturate well before overflow + if (exponent < 0x1000_0000) { + exponent = 10 * exponent + digit; + } + } + + return if (negative) -exponent else exponent; + } + + return null; +} + +const ParseInfo = struct { + // 10 or 16 + base: u8, + // 10^19 fits in u64, 16^16 fits in u64 + max_mantissa_digits: usize, + // e.g. e or p (E and P also checked) + exp_char_lower: u8, +}; + +fn parsePartialNumberBase(comptime T: type, stream: *FloatStream, negative: bool, n: *usize, comptime info: ParseInfo) ?Number(T) { + const MantissaT = common.mantissaType(T); + + // parse initial digits before dot + var mantissa: MantissaT = 0; + tryParseDigits(MantissaT, stream, &mantissa, info.base); + var int_end = stream.offsetTrue(); + var n_digits = @intCast(isize, stream.offsetTrue()); + + // handle dot with the following digits + var exponent: i64 = 0; + if (stream.firstIs('.')) { + stream.advance(1); + const marker = stream.offsetTrue(); + tryParseDigits(MantissaT, stream, &mantissa, info.base); + const n_after_dot = stream.offsetTrue() - marker; + exponent = -@intCast(i64, n_after_dot); + n_digits += @intCast(isize, n_after_dot); + } + + // adjust required shift to offset mantissa for base-16 (2^4) + if (info.base == 16) { + exponent *= 4; + } + + if (n_digits == 0) { + return null; + } + + // handle scientific format + var exp_number: i64 = 0; + if (stream.firstIsLower(info.exp_char_lower)) { + stream.advance(1); + exp_number = parseScientific(stream) orelse return null; + exponent += exp_number; + } + + const len = stream.offset; // length must be complete parsed length + n.* = len; + + if (stream.underscore_count > 0 and !validUnderscores(stream.slice, info.base)) { + return null; + } + + // common case with not many digits + if (n_digits <= info.max_mantissa_digits) { + return Number(T){ + .exponent = exponent, + .mantissa = mantissa, + .negative = negative, + .many_digits = false, + .hex = info.base == 16, + }; + } + + n_digits -= info.max_mantissa_digits; + var many_digits = false; + stream.reset(); // re-parse from beginning + while (stream.firstIs3('0', '.', '_')) { + // '0' = '.' + 2 + const next = stream.firstUnchecked(); + if (next != '_') { + n_digits -= @intCast(isize, next -| ('0' - 1)); + } else { + stream.underscore_count += 1; + } + stream.advance(1); + } + if (n_digits > 0) { + // at this point we have more than max_mantissa_digits significant digits, let's try again + many_digits = true; + mantissa = 0; + stream.reset(); + tryParseNDigits(MantissaT, stream, &mantissa, info.base, info.max_mantissa_digits); + + exponent = blk: { + if (mantissa >= min_n_digit_int(MantissaT, info.max_mantissa_digits)) { + // big int + break :blk @intCast(i64, int_end) - @intCast(i64, stream.offsetTrue()); + } else { + // the next byte must be present and be '.' + // We know this is true because we had more than 19 + // digits previously, so we overflowed a 64-bit integer, + // but parsing only the integral digits produced less + // than 19 digits. That means we must have a decimal + // point, and at least 1 fractional digit. + stream.advance(1); + var marker = stream.offsetTrue(); + tryParseNDigits(MantissaT, stream, &mantissa, info.base, info.max_mantissa_digits); + break :blk @intCast(i64, marker) - @intCast(i64, stream.offsetTrue()); + } + }; + // add back the explicit part + exponent += exp_number; + } + + return Number(T){ + .exponent = exponent, + .mantissa = mantissa, + .negative = negative, + .many_digits = many_digits, + .hex = info.base == 16, + }; +} + +/// Parse a partial, non-special floating point number. +/// +/// This creates a representation of the float as the +/// significant digits and the decimal exponent. +fn parsePartialNumber(comptime T: type, s: []const u8, negative: bool, n: *usize) ?Number(T) { + std.debug.assert(s.len != 0); + var stream = FloatStream.init(s); + const MantissaT = common.mantissaType(T); + + if (stream.hasLen(2) and stream.atUnchecked(0) == '0' and std.ascii.toLower(stream.atUnchecked(1)) == 'x') { + stream.advance(2); + return parsePartialNumberBase(T, &stream, negative, n, .{ + .base = 16, + .max_mantissa_digits = if (MantissaT == u64) 16 else 32, + .exp_char_lower = 'p', + }); + } else { + return parsePartialNumberBase(T, &stream, negative, n, .{ + .base = 10, + .max_mantissa_digits = if (MantissaT == u64) 19 else 38, + .exp_char_lower = 'e', + }); + } +} + +pub fn parseNumber(comptime T: type, s: []const u8, negative: bool) ?Number(T) { + var consumed: usize = 0; + if (parsePartialNumber(T, s, negative, &consumed)) |number| { + // must consume entire float (no trailing data) + if (s.len == consumed) { + return number; + } + } + return null; +} + +fn parsePartialInfOrNan(comptime T: type, s: []const u8, n: *usize) ?T { + // inf/infinity; infxxx should only consume inf. + if (std.ascii.startsWithIgnoreCase(s, "inf")) { + n.* = 3; + if (std.ascii.startsWithIgnoreCase(s[3..], "inity")) { + n.* = 8; + } + return std.math.inf(T); + } + + if (std.ascii.startsWithIgnoreCase(s, "nan")) { + n.* = 3; + return std.math.nan(T); + } + + return null; +} + +pub fn parseInfOrNan(comptime T: type, s: []const u8, negative: bool) ?T { + var consumed: usize = 0; + if (parsePartialInfOrNan(T, s, &consumed)) |special| { + if (s.len == consumed) { + if (negative) { + return -1 * special; + } + return special; + } + } + return null; +} + +pub fn validUnderscores(s: []const u8, comptime base: u8) bool { + var i: usize = 0; + while (i < s.len) : (i += 1) { + if (s[i] == '_') { + // underscore at start of end + if (i == 0 or i + 1 == s.len) { + return false; + } + // consecutive underscores + if (!common.isDigit(s[i - 1], base) or !common.isDigit(s[i + 1], base)) { + return false; + } + + // next is guaranteed a digit, skip an extra + i += 1; + } + } + + return true; +} diff --git a/lib/std/fmt/parse_float/parse_float.zig b/lib/std/fmt/parse_float/parse_float.zig new file mode 100644 index 0000000000..d781b11495 --- /dev/null +++ b/lib/std/fmt/parse_float/parse_float.zig @@ -0,0 +1,64 @@ +const std = @import("std"); +const parse = @import("parse.zig"); +const parseNumber = parse.parseNumber; +const parseInfOrNan = parse.parseInfOrNan; +const convertFast = @import("convert_fast.zig").convertFast; +const convertEiselLemire = @import("convert_eisel_lemire.zig").convertEiselLemire; +const convertSlow = @import("convert_slow.zig").convertSlow; +const convertHex = @import("convert_hex.zig").convertHex; + +const optimize = true; + +pub const ParseFloatError = error{ + InvalidCharacter, +}; + +pub fn parseFloat(comptime T: type, s: []const u8) ParseFloatError!T { + if (s.len == 0) { + return error.InvalidCharacter; + } + + var i: usize = 0; + const negative = s[i] == '-'; + if (s[i] == '-' or s[i] == '+') { + i += 1; + } + if (s.len == i) { + return error.InvalidCharacter; + } + + const n = parse.parseNumber(T, s[i..], negative) orelse { + return parse.parseInfOrNan(T, s[i..], negative) orelse error.InvalidCharacter; + }; + + if (n.hex) { + return convertHex(T, n); + } + + if (optimize) { + if (convertFast(T, n)) |f| { + return f; + } + + if (T == f16 or T == f32 or T == f64) { + // If significant digits were truncated, then we can have rounding error + // only if `mantissa + 1` produces a different result. We also avoid + // redundantly using the Eisel-Lemire algorithm if it was unable to + // correctly round on the first pass. + if (convertEiselLemire(T, n.exponent, n.mantissa)) |bf| { + if (!n.many_digits) { + return bf.toFloat(T, n.negative); + } + if (convertEiselLemire(T, n.exponent, n.mantissa + 1)) |bf2| { + if (bf.eql(bf2)) { + return bf.toFloat(T, n.negative); + } + } + } + } + } + + // Unable to correctly round the float using the Eisel-Lemire algorithm. + // Fallback to a slower, but always correct algorithm. + return convertSlow(T, s[i..]).toFloat(T, negative); +} diff --git a/lib/std/fmt/parse_hex_float.zig b/lib/std/fmt/parse_hex_float.zig deleted file mode 100644 index 3e8bc5c5d9..0000000000 --- a/lib/std/fmt/parse_hex_float.zig +++ /dev/null @@ -1,347 +0,0 @@ -// The rounding logic is inspired by LLVM's APFloat and Go's atofHex -// implementation. - -const std = @import("std"); -const ascii = std.ascii; -const fmt = std.fmt; -const math = std.math; -const testing = std.testing; - -const assert = std.debug.assert; - -pub fn parseHexFloat(comptime T: type, s: []const u8) !T { - assert(@typeInfo(T) == .Float); - - const TBits = std.meta.Int(.unsigned, @typeInfo(T).Float.bits); - - const mantissa_bits = math.floatMantissaBits(T); - const exponent_bits = math.floatExponentBits(T); - const exponent_min = math.floatExponentMin(T); - const exponent_max = math.floatExponentMax(T); - - const exponent_bias = exponent_max; - const sign_shift = mantissa_bits + exponent_bits; - - if (s.len == 0) - return error.InvalidCharacter; - - if (ascii.eqlIgnoreCase(s, "nan")) { - return math.nan(T); - } else if (ascii.eqlIgnoreCase(s, "inf") or ascii.eqlIgnoreCase(s, "+inf")) { - return math.inf(T); - } else if (ascii.eqlIgnoreCase(s, "-inf")) { - return -math.inf(T); - } - - var negative: bool = false; - var exp_negative: bool = false; - - var mantissa: u128 = 0; - var exponent: i16 = 0; - var frac_scale: i16 = 0; - - const State = enum { - MaybeSign, - Prefix, - LeadingIntegerDigit, - IntegerDigit, - MaybeDot, - LeadingFractionDigit, - FractionDigit, - ExpPrefix, - MaybeExpSign, - ExpDigit, - }; - - var state = State.MaybeSign; - - var i: usize = 0; - while (i < s.len) { - const c = s[i]; - - switch (state) { - .MaybeSign => { - state = .Prefix; - - if (c == '+') { - i += 1; - } else if (c == '-') { - negative = true; - i += 1; - } - }, - .Prefix => { - state = .LeadingIntegerDigit; - - // Match both 0x and 0X. - if (i + 2 > s.len or s[i] != '0' or s[i + 1] | 32 != 'x') - return error.InvalidCharacter; - i += 2; - }, - .LeadingIntegerDigit => { - if (c == '0') { - // Skip leading zeros. - i += 1; - } else if (c == '_') { - return error.InvalidCharacter; - } else { - state = .IntegerDigit; - } - }, - .IntegerDigit => { - if (ascii.isXDigit(c)) { - if (mantissa >= math.maxInt(u128) / 16) - return error.Overflow; - mantissa *%= 16; - mantissa += try fmt.charToDigit(c, 16); - i += 1; - } else if (c == '_') { - i += 1; - } else { - state = .MaybeDot; - } - }, - .MaybeDot => { - if (c == '.') { - state = .LeadingFractionDigit; - i += 1; - } else state = .ExpPrefix; - }, - .LeadingFractionDigit => { - if (c == '_') { - return error.InvalidCharacter; - } else state = .FractionDigit; - }, - .FractionDigit => { - if (ascii.isXDigit(c)) { - if (mantissa < math.maxInt(u128) / 16) { - mantissa *%= 16; - mantissa +%= try fmt.charToDigit(c, 16); - frac_scale += 1; - } else if (c != '0') { - return error.Overflow; - } - i += 1; - } else if (c == '_') { - i += 1; - } else { - state = .ExpPrefix; - } - }, - .ExpPrefix => { - state = .MaybeExpSign; - // Match both p and P. - if (c | 32 != 'p') - return error.InvalidCharacter; - i += 1; - }, - .MaybeExpSign => { - state = .ExpDigit; - - if (c == '+') { - i += 1; - } else if (c == '-') { - exp_negative = true; - i += 1; - } - }, - .ExpDigit => { - if (ascii.isXDigit(c)) { - if (exponent >= math.maxInt(i16) / 10) - return error.Overflow; - exponent *%= 10; - exponent +%= try fmt.charToDigit(c, 10); - i += 1; - } else if (c == '_') { - i += 1; - } else { - return error.InvalidCharacter; - } - }, - } - } - - if (exp_negative) - exponent *= -1; - - // Bring the decimal part to the left side of the decimal dot. - exponent -= frac_scale * 4; - - if (mantissa == 0) { - // Signed zero. - return if (negative) -0.0 else 0.0; - } - - // Divide by 2^mantissa_bits to right-align the mantissa in the fractional - // part. - exponent += mantissa_bits; - - // Keep around two extra bits to correctly round any value that doesn't fit - // the available mantissa bits. The result LSB serves as Guard bit, the - // following one is the Round bit and the last one is the Sticky bit, - // computed by OR-ing all the dropped bits. - - // Normalize by aligning the implicit one bit. - while (mantissa >> (mantissa_bits + 2) == 0) { - mantissa <<= 1; - exponent -= 1; - } - - // Normalize again by dropping the excess precision. - // Note that the discarded bits are folded into the Sticky bit. - while (mantissa >> (mantissa_bits + 2 + 1) != 0) { - mantissa = mantissa >> 1 | (mantissa & 1); - exponent += 1; - } - - // Very small numbers can be possibly represented as denormals, reduce the - // exponent as much as possible. - while (mantissa != 0 and exponent < exponent_min - 2) { - mantissa = mantissa >> 1 | (mantissa & 1); - exponent += 1; - } - - // Whenever the guard bit is one (G=1) and: - // - we've truncated more than 0.5ULP (R=S=1) - // - we've truncated exactly 0.5ULP (R=1 S=0) - // Were are going to increase the mantissa (round up) - const guard_bit_and_half_or_more = (mantissa & 0b110) == 0b110; - mantissa >>= 2; - exponent += 2; - - if (guard_bit_and_half_or_more) { - mantissa += 1; - } - - if (mantissa == (1 << (mantissa_bits + 1))) { - // Renormalize, if the exponent overflows we'll catch that below. - mantissa >>= 1; - exponent += 1; - } - - if (mantissa >> mantissa_bits == 0) { - // This is a denormal number, the biased exponent is zero. - exponent = -exponent_bias; - } - - if (exponent > exponent_max) { - // Overflow, return +inf. - return math.inf(T); - } - - // Remove the implicit bit. - mantissa &= @as(u128, (1 << mantissa_bits) - 1); - - const raw: TBits = - (if (negative) @as(TBits, 1) << sign_shift else 0) | - @as(TBits, @bitCast(u16, exponent + exponent_bias)) << mantissa_bits | - @truncate(TBits, mantissa); - - return @bitCast(T, raw); -} - -test "special" { - try testing.expect(math.isNan(try parseHexFloat(f32, "nAn"))); - try testing.expect(math.isPositiveInf(try parseHexFloat(f32, "iNf"))); - try testing.expect(math.isPositiveInf(try parseHexFloat(f32, "+Inf"))); - try testing.expect(math.isNegativeInf(try parseHexFloat(f32, "-iNf"))); -} -test "zero" { - try testing.expectEqual(@as(f32, 0.0), try parseHexFloat(f32, "0x0")); - try testing.expectEqual(@as(f32, 0.0), try parseHexFloat(f32, "-0x0")); - try testing.expectEqual(@as(f32, 0.0), try parseHexFloat(f32, "0x0p42")); - try testing.expectEqual(@as(f32, 0.0), try parseHexFloat(f32, "-0x0.00000p42")); - try testing.expectEqual(@as(f32, 0.0), try parseHexFloat(f32, "0x0.00000p666")); -} - -test "f16" { - const Case = struct { s: []const u8, v: f16 }; - const cases: []const Case = &[_]Case{ - .{ .s = "0x1p0", .v = 1.0 }, - .{ .s = "-0x1p-1", .v = -0.5 }, - .{ .s = "0x10p+10", .v = 16384.0 }, - .{ .s = "0x10p-10", .v = 0.015625 }, - // Max normalized value. - .{ .s = "0x1.ffcp+15", .v = math.floatMax(f16) }, - .{ .s = "-0x1.ffcp+15", .v = -math.floatMax(f16) }, - // Min normalized value. - .{ .s = "0x1p-14", .v = math.floatMin(f16) }, - .{ .s = "-0x1p-14", .v = -math.floatMin(f16) }, - // Min denormal value. - .{ .s = "0x1p-24", .v = math.floatTrueMin(f16) }, - .{ .s = "-0x1p-24", .v = -math.floatTrueMin(f16) }, - }; - - for (cases) |case| { - try testing.expectEqual(case.v, try parseHexFloat(f16, case.s)); - } -} -test "f32" { - const Case = struct { s: []const u8, v: f32 }; - const cases: []const Case = &[_]Case{ - .{ .s = "0x1p0", .v = 1.0 }, - .{ .s = "-0x1p-1", .v = -0.5 }, - .{ .s = "0x10p+10", .v = 16384.0 }, - .{ .s = "0x10p-10", .v = 0.015625 }, - .{ .s = "0x0.ffffffp128", .v = 0x0.ffffffp128 }, - .{ .s = "0x0.1234570p-125", .v = 0x0.1234570p-125 }, - // Max normalized value. - .{ .s = "0x1.fffffeP+127", .v = math.floatMax(f32) }, - .{ .s = "-0x1.fffffeP+127", .v = -math.floatMax(f32) }, - // Min normalized value. - .{ .s = "0x1p-126", .v = math.floatMin(f32) }, - .{ .s = "-0x1p-126", .v = -math.floatMin(f32) }, - // Min denormal value. - .{ .s = "0x1P-149", .v = math.floatTrueMin(f32) }, - .{ .s = "-0x1P-149", .v = -math.floatTrueMin(f32) }, - }; - - for (cases) |case| { - try testing.expectEqual(case.v, try parseHexFloat(f32, case.s)); - } -} -test "f64" { - const Case = struct { s: []const u8, v: f64 }; - const cases: []const Case = &[_]Case{ - .{ .s = "0x1p0", .v = 1.0 }, - .{ .s = "-0x1p-1", .v = -0.5 }, - .{ .s = "0x10p+10", .v = 16384.0 }, - .{ .s = "0x10p-10", .v = 0.015625 }, - // Max normalized value. - .{ .s = "0x1.fffffffffffffp+1023", .v = math.floatMax(f64) }, - .{ .s = "-0x1.fffffffffffffp1023", .v = -math.floatMax(f64) }, - // Min normalized value. - .{ .s = "0x1p-1022", .v = math.floatMin(f64) }, - .{ .s = "-0x1p-1022", .v = -math.floatMin(f64) }, - // Min denormalized value. - .{ .s = "0x1p-1074", .v = math.floatTrueMin(f64) }, - .{ .s = "-0x1p-1074", .v = -math.floatTrueMin(f64) }, - }; - - for (cases) |case| { - try testing.expectEqual(case.v, try parseHexFloat(f64, case.s)); - } -} -test "f128" { - const Case = struct { s: []const u8, v: f128 }; - const cases: []const Case = &[_]Case{ - .{ .s = "0x1p0", .v = 1.0 }, - .{ .s = "-0x1p-1", .v = -0.5 }, - .{ .s = "0x10p+10", .v = 16384.0 }, - .{ .s = "0x10p-10", .v = 0.015625 }, - // Max normalized value. - .{ .s = "0xf.fffffffffffffffffffffffffff8p+16380", .v = math.floatMax(f128) }, - .{ .s = "-0xf.fffffffffffffffffffffffffff8p+16380", .v = -math.floatMax(f128) }, - // Min normalized value. - .{ .s = "0x1p-16382", .v = math.floatMin(f128) }, - .{ .s = "-0x1p-16382", .v = -math.floatMin(f128) }, - // // Min denormalized value. - .{ .s = "0x1p-16494", .v = math.floatTrueMin(f128) }, - .{ .s = "-0x1p-16494", .v = -math.floatTrueMin(f128) }, - .{ .s = "0x1.edcb34a235253948765432134674fp-1", .v = 0x1.edcb34a235253948765432134674fp-1 }, - }; - - for (cases) |case| { - try testing.expectEqual(@bitCast(u128, case.v), @bitCast(u128, try parseHexFloat(f128, case.s))); - } -}