Eliminate the letters or marks out only leave numbers and apostrophe

Answer 1

Use this pattern: \d+(,\d+)?.

• \d: matches digits from 0-9
• +: matches the pattern at least once, so \d+ will match at least one digit
• (,\d+)?: this indicates a group (using the parentheses) that matches the comma followed by digits and optionally matches it (due to the ? at the end of the group)

Code snippet:

string[] inputs = { "82&?", "82,9", "abse82,9>dpkg", "foobar" };
foreach (var input in inputs)
{
    Match m = Regex.Match(input, @"\d+(,\d+)?");
    if (m.Success)
    {
        Console.WriteLine(m.Value);
    }
    else
    {
        Console.WriteLine("No match!");
    }
}

Answer 2

If you are only looking for those characters and trying to rip out anything else you can do the following

    var strings = new List<string>()
                    {
                        "82&?",
                        "82,9",
                        "abse82,9>dpkg"
                    };
    var reg = new Regex("[^0-9,]*", RegexOptions.None);
    var output = new List<string>();
    foreach(var str in strings)
    {
        output.Add(reg.Replace(str, ""));
    }

Answer 3

If you are on .NET3.5+ you can use LINQ. A solution without Regex (which is faster) is the following:

var strings = new List<string>() { "82&?", "82,9", "abse82,9>dpkg" };
var result = strings.Select(s =>
    String.Join("",
        s.Where(c => char.IsNumber(c) || c == ',')
        .ToArray()) /* .ToArray() is not needed on .NET 4 */
    ).ToList();

It only select characters that are numbers or comma. But given the string 8,1aa1 it would return 8,11.

This other method is a bit slower, but it will take 8,1 from 8,1aa1 and will not take 8,a or a,a:

var strings = new List<string>() { "82&?887..2", "82,9", "abse82,9>dpkg" };
var result = strings.Select(s =>
    String.Join("",
        s.SkipWhile(c => !char.IsNumber(c))
        .TakeWhile(c => (char.IsNumber(c) || c == ','))
        .ToArray()
    )
).Where(s => char.IsNumber(s.LastOrDefault())).ToList();

Running a test (using Stopwatch) with 100,000 iterations with the methods presented on the answers I got:

Fn: BrunoLM (Method 1)
Ticks: 524999

Fn: BrunoLM (Method 2)
Ticks: 729460

Fn: Ahmad
Ticks: 1323366

Fn: Josh
Ticks: 3783158

The same test with 1000~ length string:

var strings = new List<string>() { "82&?887..2".PadRight(1000, '2'), "82,9".PadRight(1000, '1'), "abse82,9>dpkg".PadRight(1000, 'f') };

Result:

Fn: Ahmad
Ticks: 11911332

Fn: BrunoLM (Method 2)
Ticks: 28149495

Fn: Josh
Ticks: 213681541

Further reading:

• Regular Expression Matching Can Be Simple And Fast

^{regular expression and text size n}