Regex Match NC-Comments in a string with mixed C#-Code

Question

I have a textfile with mixed NC-Code and C#-Code. C#-Code starts with "<#" and ends with "#>". Now I need one regex expression to find all NC-Comments. One problem is that NC-Comments starts with ";" therefore I got some issues to distinguish NC-Comment with ";" of C#-Code.

Is it possible to achieve this with only one regular expression?

; 1. NC-Comment
FUNCT_A;
FUNCT_B;

<# // C#-Code
int temp = 42;
string var = "hello";   // C#-Comment
#>

FUNCT_C ; 2. Comment

<# // C#-Code
for(int i = 0; i <10; i++)
{
    Console.WriteLine(i.ToString());
}
#>  

; 3. Comment
FUNCT_D;

The result of the regex should be {1. NC-Comment, 2. Comment, 3. Comment}

I have played arround with following regular expressions:

1.) (;(.*?)\r?\n) --> Finds all NC-Comments but also C#-Code as comment
2.) (#>.*?<#)|(#>.*) --> Finds all NC-Code except the first NC-Code fragment
3.) #>.+?(?=<#) --> Finds all NC-Code except the first and last NC-Code fragment

One solution could be to push each "<#" to a stack and pop each "#>" from this stack. So if the stack is empty then the current string is NC-Code. Next I have to find out if this string is a NC-Comment.

Why use a complicated slow regex when you can do the same simpler and faster without regex? — Mohammad Dehghan, Mar 20 '13 at 19:30
This is not so simple. What if the C# code itself contains comment? What if you have a `<#` or `#>` in C# comments? — Mohammad Dehghan, Mar 20 '13 at 19:37

score 1 · Accepted Answer · answered Mar 21 '13 at 18:47

I rather do it without regex:

public static List<string> GetNCComments(Stream stream)
{
    using (StreamReader sr = new StreamReader(stream))
    {
        List<string> result = new List<string>();
        bool inCS = false; // are we in C# code?
        int c;
        while ((c = sr.Read()) != -1)
        {
            if (inCS)
            {
                switch ((char)c)
                {
                    case '#':
                        if (sr.Peek() == '>') // end of C# block
                        {
                            sr.Read();
                            inCS = false;
                        }
                        break;
                    case '/':
                        if (sr.Peek() == '/') // a C# comment
                            sr.ReadLine(); // skip the whole comment
                        break;
                }
            }
            else
            {
                switch ((char)c)
                {
                    case '<':
                        if (sr.Peek() == '#') // start of C# block
                        {
                            sr.Read();
                            inCS = true;
                        }
                        break;
                    case ';': // NC comment
                        string comment = sr.ReadLine();
                        if (!string.IsNullOrEmpty(comment))
                            result.Add(comment);
                        break;
                }
            }
        }
        return result;
    }
}

Usage:

var comments = GetNCComments(new FileStream(filePath, FileMode.Open, FileAccess.Read));

The code is simple and self explanatory. This also handles C# comments, but not C# strings. I mean, it works correctly if you have a #> in a C# comment. But does not work if you have the same thing a C# string (incorrectly considers it as the end of C# block). Handling this case is also easy.

Regex Match NC-Comments in a string with mixed C#-Code

1 Answers1