Sorting strings containing numbers in a user friendly way

Question

Being used to the standard way of sorting strings, I was surprised when I noticed that Windows sorts files by their names in a kind of advanced way. Let me give you an examp

Answer 1

The absolute easiest way, I found, was isolate the string you want, so in the OP's case, Path.GetFileNameWithoutExtension(), remove the non-digits, convert to int, and sort. Using LINQ and some extension methods, it's a one-liner. In my case, I was going on directories:

Directory.GetDirectories(@"a:\b\c").OrderBy(x => x.RemoveNonDigits().ToIntOrZero())

Where RemoveNonDigits and ToIntOrZero are extensions methods:

public static string RemoveNonDigits(this string value) {
    return Regex.Replace(value, "[^0-9]", string.Empty);
}

public static int ToIntOrZero(this string toConvert) {
    try {
        if (toConvert == null || toConvert.Trim() == string.Empty) return 0;            
        return int.Parse(toConvert);
    } catch (Exception) {
        return 0;
    }
}

The extension methods are common tools I use everywhere. YMMV.

Answer 2

Jeff wrote up an article about this on Coding Horror. This is called natural sorting, where you effectively treat a group of digits as a single "character". There are implementations out there in every language under the sun, but strangely it's not usually built-in to most languages' standard libraries.

Answer 3

The mother of all sorts:

ls '*.mp3' | sort --version-sort