NSNumber/NSDecimalNumber奇怪的行为 [英] NSNumber/NSDecimalNumber bizarre behavior

问题描述

以下代码为我刚选择的一个测试用例产生了奇怪的结果.操场上的代码，然后是结果

Following code produces bizarre result for a test case I just happened to choose. Code from playground, followed by result

import Cocoa

func spellDollars(_ amount: Decimal) -> String {
    let fDollars = NumberFormatter()
    fDollars.numberStyle = NumberFormatter.Style.spellOut
    let fCents = NumberFormatter()
    fCents.format = "00"
    let c = (amount * 100 as NSDecimalNumber).intValue % 100
    let cents = fCents.string(from: NSNumber(value: c))!
    let d  = (amount as NSDecimalNumber).intValue
    print(d, c, amount)
    let dollars = fDollars.string(from: NSNumber(value: d))!.capitalized
    let s = "\(dollars) and \(cents)/100"
    return s
}

print(spellDollars(1019123.08))

结果:

-825551 -32 1019123.0799999997952 负八十五二十五万五十一和-32/100

-825551 -32 1019123.0799999997952 Minus Eight Hundred Twenty-Five Thousand Five Hundred Fifty-One and -32/100

推荐答案

最重要的是，问题在于您使用Double调用了spellDollars，并将其转换为Decimal.您可以这样做:

Bottom line, the problem is that you called spellDollars with a Double, which it converted to a Decimal. You could do:

print(spellDollars(Decimal(string: "1019123.08")!))

或

print(spellDollars(Decimal(sign: .plus, exponent: -2, significand: 101912308)))

而这两个方法都可以在不将Double引入Decimal转换噪声的情况下工作.

And both of those would have worked without introducing this Double to Decimal conversion noise.

考虑一个简化的示例:

let decimal1 = Decimal(floatLiteral: 1019123.08)
let decimal2 = Decimal(string: "1019123.08")!
let decimal3 = Decimal(sign: .plus, exponent: -2, significand: 101912308)

for value in [decimal1, decimal2, decimal3] {
    print(value)
}

这将输出:

1019123.0799999997952
1019123.08
1019123.08

1019123.0799999997952
1019123.08
1019123.08

要了解为什么第一个示例中的Double转换为Decimal值1019123.0799999997952的原因，建议您参考

To understand why that Double in the first example gets converted to a Decimal value of 1019123.0799999997952, I’d suggest you refer to What Every Computer Scientist Should Know About Floating-Point Arithmetic.

关于为什么从intValue中获得-825551的原因，我想这是由于有效位数(10191230799999997952)太大导致无法以整数格式或类似形式表示而导致的一些溢出.

Regarding why you’re getting -825551 from intValue, I’d guess it’s some overflow resulting from a significand (10191230799999997952) that is so large that it can’t be represented in an integer format or something like that.

我建议您切开Gordian结，而不要使用NSDecimalNumber算术来手动舍入值.使用 NSDecimalRound 函数，例如

I’d suggest cutting the Gordian knot and not use NSDecimalNumber arithmetic to manually round values. Use the NSDecimalRound function, e.g.

func spellDollars(_ amount: Decimal) -> String {
    // instantiate formatters

    let dollarsFormatter = NumberFormatter()
    dollarsFormatter.numberStyle = .spellOut
    let centsFormatter = NumberFormatter()
    centsFormatter.minimumIntegerDigits = 2

    // get dollars

    var inputAmount = amount
    var dollars = Decimal.zero
    NSDecimalRound(&dollars, &inputAmount, 0, .down)

    // get dollars and cents

    var dollarsAndCents = Decimal.zero
    NSDecimalRound(&dollarsAndCents, &inputAmount, 2, .plain)

    // get cents

    let cents = (dollarsAndCents - dollars) * 100

    // build strings

    let centsString = centsFormatter.string(for: cents)!
    let dollarsString = dollarsFormatter.string(for: dollars)!.capitalized
    return "\(dollarsString) and \(centsString)/100"
}

这篇关于NSNumber/NSDecimalNumber奇怪的行为的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！