c# Nested quantifier - multiple "?" in the pattern

Question

I'm trying to find if at least one file within a directory matches patterns (that use only "?" and "*" wildcards), but some combinations keep throwing the nested qualifier error. For example - TestCashFile_10_12-25-2016????????.c?? doesn't work.

The patterns come from the non-technical users (who're educated in the basic usage of these two wildcards) so the "?" and "*" can go pretty much anywhere in the filename and I don't have much control.

What is wrong with these patterns?

This is the C# code snippet that runs this regex -

string fileName = C:\TestFiles\TestCashFile_10_12-25-2016????????.c??'
string directory = Path.GetDirectoryName(fileName);
string[] temp = fileName.Split('\\');
string file = temp[temp.Length - 1];
var found = Directory.GetFiles(directory).Any(p => Regex.Match(p, file).Success);

Update - The question has been resolved but in case it helps someone else looking for something similar, just to clarify - In this case, I wanted "?" to mean that there must be exactly one element (as opposed to zero or one element).

Answer 1

The ? operator specifies that the previous element can occur 0 or 1 time.

https://msdn.microsoft.com/en-us/library/az24scfc(v=vs.110).aspx

? Matches the previous element zero or one time. "rai?n" "ran", "rain"

If you use the wildcards built into Directory.GetFiles like @Ed Plunkett said, it should work similar to what you are looking for.

If you still want to use your current method with RegEx, do something like the following:

• .* - any number of characters
• .{n} - replace n with the number of expected characters
• .{m,n} - replace m with the min number of expected characters, and n with the max number of expected characters.

Answer 2

If you need "??" to match exactly two of any character then you're right, you'll have to use regexes. Filesystem wildcarding treats "?" as "zero or one of any character".

But you can't do it the way you tried to, because you're asking your users for filesystem wildcards -- you're just altering the semantics a bit. You'll have to turn the strings from the user into the regex you want:

a???.*

Has to become

a.?.?.?\..*

• Each question mark becomes ".": Exactly one of any character.
• Each "." becomes ".", because unescaped "." is a special character in a regex.
• Each "" has to become ".": Zero or more of any character (guessing on this one).

Do that to the file string, and your .Any(p => Regex.Match(p, file).Success); should work.

You might want to compile the regex though, if things get a little slow at runtime:

file = TranslateWildcardsToRegex(file);
var re = new Regex(file);

var found = Directory.GetFiles("").Any(p => re.IsMatch(p));

I think this is right for TranslateWildcardsToRegex():

public static String TranslateWildcardsToRegex(String s)
{
    StringBuilder sb = new StringBuilder();

    foreach (var ch in s)
    {
        switch (ch)
        {
            case '?':
                sb.Append(".");
                break;

            case '*':
                sb.Append(".*");
                break;

            //  Escape a variety of characters that 
            //  mean something special in a regex
            case '(':
            case ')':
            case '{':
            case '}':
            case '[':
            case ']':
            case '.':
                sb.Append("\\" + ch);
                break;

            default:
                sb.Append(ch);
                break;
        }
    }

    return sb.ToString();
}

UPDATE

In comments @spender offers a much nicer and cleaner way to do the same thing:

var reStr = Regex.Escape(someWildcardThing).Replace(@"\?", ".").Replace(@"\*", ".*")

I have no good excuse for not doing it that way myself, other than still being a recovering C programmer, after all these years.