Adjust ToCamelCase Extension

[cwuethrich]

Adjust ToCamelCase Extension of string to match the behavior of System.Text.Json and Newtonsoft.Json.
Bug: #8273

[flobernd]

Suggested change

	if (char.IsSeparator(current))
	if (!char.IsLetter(current))

[flobernd]

Thanks for the PR 🙂

I suggested a few adjustments.

One suggestion is a minor performance optimization using a stack allocated Span instead of allocating a new array on the heap.

The other suggestion is a functional one. I think we want to handle e.g. digits the same way as separators so that for instance SHA1 is converted to sha1 instead of shA1. To save some checks, I reversed the logic to check for char.IsLetter() (or !char.IsLetter()) instead of explicitly excluding separators, digits, etc.

Edit:
In this usecase, we don't expect to ever encounter "whitespace" characters as they are invalid to use in .NET field/property-names, but if we want to be even more accurate, we would have to add special handling for separators like e.g. _. For example Camel_Case should end up as camelCase as camel-case does not allow such separators.

[flobernd]

Suggested change

	if (!char.IsUpper(next) && !char.IsSeparator(next))
	if (char.IsLetter(next) && !char.IsUpper(next))

[flobernd]

Suggested change

	var chars = s.ToCharArray();
	#if NET6_0_OR_GREATER
	Span<char> chars = stackalloc char[s.Length];
	s.CopyTo(chars);
	#else
	var chars = s.ToCharArray();
	#endif

[flobernd]

This requires an additional:

#if NET6_0_OR_GREATER
using System;
#endif

[flobernd]

Actually, let's just copy the System.Text.Json implementation (MIT license):

    internal static string ToCamelCase3(this string name)
    {
        if (string.IsNullOrEmpty(name) || !char.IsUpper(name[0]))
        {
            return name;
        }

#if NET
        return string.Create(name.Length, name, (chars, name) =>
        {
            name.CopyTo(chars);
            FixCasing(chars);
        });
#else
            char[] chars = name.ToCharArray();
            FixCasing(chars);
            return new string(chars);
#endif
    }

    private static void FixCasing(Span<char> chars)
    {
        for (var i = 0; i < chars.Length; i++)
        {
            if (i == 1 && !char.IsUpper(chars[i]))
            {
                break;
            }

            var hasNext = (i + 1 < chars.Length);

            // Stop when next char is already lowercase.
            if (i > 0 && hasNext && !char.IsUpper(chars[i + 1]))
            {
                // If the next char is a space, lowercase current char before exiting.
                if (chars[i + 1] == ' ')
                {
                    chars[i] = char.ToLowerInvariant(chars[i]);
                }

                break;
            }

            chars[i] = char.ToLowerInvariant(chars[i]);
        }
    }

[cwuethrich]

Wasn't sure about the license. But if we are allowed to use it, it would be the best. Then the properties for sure gonna match. Sad they didn't make theirs classes public.

Edit: Do you integrate it, or should I go for a try?

[flobernd]

@cwuethrich I can open a new PR, if you don't mind 🙂 Thanks for creating this one!