Custom character class in C# regex

Question

Is there any way to define custom character class in C# regex?

In flex it is done in very obvious way:

DIGIT    [0-9]
%%
{DIGIT}+    {printf( "An integer: %s (%d)\n", yytext, atoi( yytext ) );}

http://westes.github.io/flex/manual/Simple-Examples.html#Simple-Examples

As explained in this answer, in PHP defining a custom character class works like this:

(?(DEFINE)(?<a>[acegikmoqstz@#&]))\g<a>(?:.*\g<a>){2}

Is there a way to achieve this result in c#, without repeating the full character class definition each time it is used?

@Rawling: It's same kind of question, but the point is: How to do it (if possible) in C#. — PiotrB, Aug 17 '14 at 11:20
Reason for my *reopen vote*: The answer in the linked duplicate **does not address c#** at all, it explicitly only deals with Java and PHP. The solutions presented there are not applicable for c# (@Rawling) — HugoRune, Aug 20 '14 at 14:45
@HugoRune Good point, I thought both the answer were just language-specific versions of string concatenation but the PHP one is doing something special. There is a [C# specific question here](http://stackoverflow.com/questions/8204214/regex-reusing-subexpressions) and I expect most answers you attract will be along the same lines. — Rawling, Aug 20 '14 at 15:00
@Rawling Yes, I don't think a better solution exists either. But I was googling for this problem, and this question here seemed to be the only applicable result, so a definitive answer here should be useful to future visitors, even if it is a negative one. — HugoRune, Aug 20 '14 at 15:04
It may be possible to use named blocks and class subtraction to get the same effect, or there may be a named block that already matches the required characters — Panagiotis Kanavos, Aug 20 '14 at 15:13

score 3 · Answer 1 · answered Aug 20 '14 at 15:12

Custom character classes aren't supported in C# but you may be able to use named blocks and character class subtraction to get a similar effect.

.NET defines a large number of named blocks that correspond to Unicode character categories like math or Greek symbols. There may be a block that already matches your requirements.

Character class subtraction allows you to exclude the characters in one class or block from the characters in a broader class. The syntax is :

[ base_group -[ excluded_group ]]

The following example, copied from the linked documentation, matches all Unicode characters except whitespace, Greek characters, punctuation and newlines:

[\u0000-\uFFFF-[\s\p{P}\p{IsGreek}\x85]]

score 2 · Answer 2 · answered Aug 20 '14 at 14:59

Nope, not supported in C#. This link will give you a nice overview of the .NET Regex engine. Note that nothing really stops you from defining variables and using them to construct your Regex string:

var digit = "[0-9]";
var regex = new Regex(digit + "[A-Z]");

Custom character class in C# regex

2 Answers2