replaceAll("(?<=[^^\\p{Uppercase}])(?=[\\p{Uppercase}])"," ");

这个问题有点老了，但现在在Nuget上有一个很好的库，它可以做到这一点，以及许多其他转换到人类可读的文本。

在GitHub或Nuget上检查Humanizer。

例子

"PascalCaseInputStringIsTurnedIntoSentence".Humanize() => "Pascal case input string is turned into sentence"
"Underscored_input_string_is_turned_into_sentence".Humanize() => "Underscored input string is turned into sentence"
"Underscored_input_String_is_turned_INTO_sentence".Humanize() => "Underscored input String is turned INTO sentence"

// acronyms are left intact
"HTML".Humanize() => "HTML"

欢迎来到Unicode

所有这些解决方案对于现代文本来说本质上都是错误的。你需要使用能理解大小写的东西。由于Bob要求使用其他语言，我将为Perl提供两种语言。

我提供了四种解决方案，从最坏到最好。只有最好的人才是对的。其他人都有问题。下面是一个测试运行，向您展示什么可行，什么不可行，以及在哪里。我用了下划线，这样你们就能看到空格放在了哪里，我把所有错的地方都标记了出来。

Testing TheLoneRanger
               Worst:    The_Lone_Ranger
               Ok:       The_Lone_Ranger
               Better:   The_Lone_Ranger
               Best:     The_Lone_Ranger
Testing MountMᶜKinleyNationalPark
     [WRONG]   Worst:    Mount_MᶜKinley_National_Park
     [WRONG]   Ok:       Mount_MᶜKinley_National_Park
     [WRONG]   Better:   Mount_MᶜKinley_National_Park
               Best:     Mount_Mᶜ_Kinley_National_Park
Testing ElÁlamoTejano
     [WRONG]   Worst:    ElÁlamo_Tejano
               Ok:       El_Álamo_Tejano
               Better:   El_Álamo_Tejano
               Best:     El_Álamo_Tejano
Testing TheÆvarArnfjörðBjarmason
     [WRONG]   Worst:    TheÆvar_ArnfjörðBjarmason
               Ok:       The_Ævar_Arnfjörð_Bjarmason
               Better:   The_Ævar_Arnfjörð_Bjarmason
               Best:     The_Ævar_Arnfjörð_Bjarmason
Testing IlCaffèMacchiato
     [WRONG]   Worst:    Il_CaffèMacchiato
               Ok:       Il_Caffè_Macchiato
               Better:   Il_Caffè_Macchiato
               Best:     Il_Caffè_Macchiato
Testing MisterDženanLjubović
     [WRONG]   Worst:    MisterDženanLjubović
     [WRONG]   Ok:       MisterDženanLjubović
               Better:   Mister_Dženan_Ljubović
               Best:     Mister_Dženan_Ljubović
Testing OleKingHenryⅧ
     [WRONG]   Worst:    Ole_King_HenryⅧ
     [WRONG]   Ok:       Ole_King_HenryⅧ
     [WRONG]   Better:   Ole_King_HenryⅧ
               Best:     Ole_King_Henry_Ⅷ
Testing CarlosⅤºElEmperador
     [WRONG]   Worst:    CarlosⅤºEl_Emperador
     [WRONG]   Ok:       CarlosⅤº_El_Emperador
     [WRONG]   Better:   CarlosⅤº_El_Emperador
               Best:     Carlos_Ⅴº_El_Emperador

顺便说一下，这里几乎所有人都选择了第一种方式，即标记为“最差”的方式。少数人选择了标记为“OK”的第二种方式。但是在我之前没有人告诉过你如何做“更好”或“最好”的方法。

下面是带有四个方法的测试程序:

#!/usr/bin/env perl
use utf8;
use strict;
use warnings;

# First I'll prove these are fine variable names:
my (
    $TheLoneRanger              ,
    $MountMᶜKinleyNationalPark  ,
    $ElÁlamoTejano              ,
    $TheÆvarArnfjörðBjarmason   ,
    $IlCaffèMacchiato           ,
    $MisterDženanLjubović         ,
    $OleKingHenryⅧ              ,
    $CarlosⅤºElEmperador        ,
);

# Now I'll load up some string with those values in them:
my @strings = qw{
    TheLoneRanger
    MountMᶜKinleyNationalPark
    ElÁlamoTejano
    TheÆvarArnfjörðBjarmason
    IlCaffèMacchiato
    MisterDženanLjubović
    OleKingHenryⅧ
    CarlosⅤºElEmperador
};

my($new, $best, $ok);
my $mask = "  %10s   %-8s  %s\n";

for my $old (@strings) {
    print "Testing $old\n";
    ($best = $old) =~ s/(?<=\p{Lowercase})(?=[\p{Uppercase}\p{Lt}])/_/g;

    ($new = $old) =~ s/(?<=[a-z])(?=[A-Z])/_/g;
    $ok = ($new ne $best) && "[WRONG]";
    printf $mask, $ok, "Worst:", $new;

    ($new = $old) =~ s/(?<=\p{Ll})(?=\p{Lu})/_/g;
    $ok = ($new ne $best) && "[WRONG]";
    printf $mask, $ok, "Ok:", $new;

    ($new = $old) =~ s/(?<=\p{Ll})(?=[\p{Lu}\p{Lt}])/_/g;
    $ok = ($new ne $best) && "[WRONG]";
    printf $mask, $ok, "Better:", $new;

    ($new = $old) =~ s/(?<=\p{Lowercase})(?=[\p{Uppercase}\p{Lt}])/_/g;
    $ok = ($new ne $best) && "[WRONG]";
    printf $mask, $ok, "Best:", $new;
}

当你能在这个数据集上得到与“最佳”相同的分数时，你就知道你做对了。在那之前，你还没有。这里没有人比“还行”做得更好，大多数人都做得“最差”。我期待看到有人发布正确的ℂ代码。

我注意到，StackOverflow的高亮代码是悲惨的笨拙再次。他们所做的一切都和这里提到的其他糟糕的方法一样(大部分但不是全部)。难道不是早就该让ASCII停止使用了吗?这已经没有意义了，假装这是你的全部是大错特错的。这会导致糟糕的代码。

我知道这是一个旧的，但这是一个扩展我使用当我需要这样做:

public static class Extensions
{
    public static string ToSentence( this string Input )
    {
        return new string(Input.SelectMany((c, i) => i > 0 && char.IsUpper(c) ? new[] { ' ', c } : new[] { c }).ToArray());
    }
}

这将允许您使用MyCasedString.ToSentence()

下面是在SQL中如何做到这一点

create  FUNCTION dbo.PascalCaseWithSpace(@pInput AS VARCHAR(MAX)) RETURNS VARCHAR(MAX)
BEGIN
    declare @output varchar(8000)

set @output = ''


Declare @vInputLength        INT
Declare @vIndex              INT
Declare @vCount              INT
Declare @PrevLetter varchar(50)
SET @PrevLetter = ''

SET @vCount = 0
SET @vIndex = 1
SET @vInputLength = LEN(@pInput)

WHILE @vIndex <= @vInputLength
BEGIN
    IF ASCII(SUBSTRING(@pInput, @vIndex, 1)) = ASCII(Upper(SUBSTRING(@pInput, @vIndex, 1)))
       begin 

        if(@PrevLetter != '' and ASCII(@PrevLetter) = ASCII(Lower(@PrevLetter)))
            SET @output = @output + ' ' + SUBSTRING(@pInput, @vIndex, 1)
            else
            SET @output = @output +  SUBSTRING(@pInput, @vIndex, 1) 

        end
    else
        begin
        SET @output = @output +  SUBSTRING(@pInput, @vIndex, 1) 

        end

set @PrevLetter = SUBSTRING(@pInput, @vIndex, 1) 

    SET @vIndex = @vIndex + 1
END


return @output
END

对于任何正在寻找回答这个问题的c++函数的人，您可以使用下面的方法。这是模仿@Binary Worrier给出的答案。这种方法只是自动保留首字母缩略词。

using namespace std;

void AddSpacesToSentence(string& testString)
        stringstream ss;
        ss << testString.at(0);
        for (auto it = testString.begin() + 1; it != testString.end(); ++it )
        {
            int index = it - testString.begin();
            char c = (*it);
            if (isupper(c))
            {
                char prev = testString.at(index - 1);
                if (isupper(prev))
                {
                    if (index < testString.length() - 1)
                    {
                        char next = testString.at(index + 1);
                        if (!isupper(next) && next != ' ')
                        {
                            ss << ' ';
                        }
                    }
                }
                else if (islower(prev)) 
                {
                   ss << ' ';
                }
            }

            ss << c;
        }

        cout << ss.str() << endl;

我为这个函数使用的测试字符串，结果是:

"helloWorld" -> "helloWorld" "HelloWorld" -> "HelloWorld" "HelloABCWorld" -> "HelloABCWorld" "HelloWorldABC" -> "HelloWorldABC" "ABCHelloWorld" -> "ABCHelloWorld" " abc hello world " -> " abc hello world " " abchelloworld " -> " abchelloworld " " a " -> " a "