使用LINQ从字符串中删除字符

Question

I'm trying to brush up on my LINQ by writing some simple extension methods. Is there any better way to write such a function as below that removes a given list of characters from a string (using LINQ)?

It helps me to think of the extension methods that LINQ relies on first:

public static string Remove(this string s, IEnumerable<char> chars)
{
    string removeChars = string.Concat(chars);

    return new string(s.ToCharArray().Where(c => !removeChars.Contains(c)).ToArray());
}

But that's pretty ugly. Ergo LINQ.

The difference that I notice in the LINQ statement is that I have to use 'select' whereas with the extension method, I don't have to.

/// <summary>Strip characters out of a string.</summary>
/// <param name="chars">The characters to remove.</param>
public static string Remove(this string s, IEnumerable<char> chars)
{
    string removeChars = string.Concat(chars);

    var stripped = from c in s.ToCharArray()
                   where !removeChars.Contains(c)
                   select c;

    return new string(stripped.ToArray());
}

So I'm wondering if this (last snippet above) is the tersest LINQ statement to accomplish removal of characters.

Answer 1

I would prefer the first form with extension methods though simplified to

public static string Remove(this string s, IEnumerable<char> chars)
{
    return new string(s.Where(c => !chars.Contains(c)).ToArray());
}

As for select keyword, it's obligatory in second form. The documentation says what "A query expression must terminate with either a select clause or a group clause". That's why I would avoid LINQ syntactic sugar.

Answer 2

try this for terseness

public static string Remove(this string source, IEnumerable<char> chars) {
  return new String(source.Where(x => !chars.Contains(x)).ToArray());
}

EDIT

Updated to correct it removing duplicates from source

Answer 3

Personally I tend to use the first syntax for non relational situations. When I need to perform relational operations (join), say with Expression Trees against SQL i use the later. But, this is only because its more readable for me having used SQL for a while.

Answer 4

You get a little performance increase when using a stringBuilder instead of the new string. Below results in:

StringBuilder 00:00:13.9930633 new String 00:00:15.1495309

        string s = "ababababajjjaazsiajjsoajiojsioajlmmzaaokpdahgffaiojsia";
        var sw = new Stopwatch();
        sw.Start();
        var toRemove = new char[] { 'j', 'a', 'z' };
        for (int i = 0; i < 1000000; i++)
        {
            StringBuilder sb = new StringBuilder(s.Length, s.Length);
            foreach (var c in s) if (!toRemove.Contains(c)) sb.Append(c);
        }
        Console.WriteLine("StringBuilder " + sw.Elapsed);
        sw.Restart();
        for (int i = 0; i < 1000000; i++)
        {
            new string(s.Where(c => !toRemove.Contains(c)).ToArray());
        }
        Console.WriteLine("new String " + sw.Elapsed);