Match regex between curly braces

Question

I need to match with regular expression the tokens from text:

Hello {FullName}, I wanted to inform that your product {ProductName} is ready.
Please come to our address {Address} to get it!

How can I match the specific tokens in my text and fill the values using regex?

Also I need to do it in a safe way and avoid every possible issues where the tokens my by misspelled or have something wrong, like bellow:

**Hello {Full{Name}, I { wanted to inform that your product {{ProductName} is ready.
Please come to our } address {Addr{Street}ess} to get it!**

P.S.

I tried this: {([^}]+)}

But if I have for example :

 {FullName}

it works, but it also work if I have

{Full{Name} ...

P.S. 2:

I tried this: {=[^{^=^}]*=} but I have to use another character instead of just curly braces ... is it possible to adjust it so that this will work without the equal character?

 {=FullName=}    - this works
 {=Full{Name=}   - this doesn't work

So basically the token is between {=Token=} instead of {Token}

Define "avoid" every possible issue. Do you want to skip those, or have some magic parser decide that it is good anyway? — Bart Friederichs, Aug 08 '13 at 13:54
Does it _have_ to be regex? With your "error conditions" a custom parser would be a better option. — Oded, Aug 08 '13 at 13:54
Do you mean that `{Full{Name}` shall be found and interpreted correctly by the regex? — oddparity, Aug 08 '13 at 13:55
Why not use a parser? You've gone out of your way to implement this nice token-based system. Tokens beg for parsers! — Jesse Smith, Aug 08 '13 at 13:56
@smerny the braces don't match because I wanted to point out wrong cases. — David Dury, Aug 08 '13 at 13:57
In your example `{Full{Name}` you need to consider `{Name}` as valid or invalid? — Sriram Sakthivel, Aug 08 '13 at 13:59
@JesseSmith you mean something like this: http://stackoverflow.com/questions/711753/a-better-way-to-replace-many-strings-obfuscation-in-c-sharp — David Dury, Aug 08 '13 at 14:00
@DavidDury That's something you need to research on your own. A lot of computer science is based on parsers, compilers, lexers, linkers, etc. and literally hundreds of books have been written on the subject. — Jesse Smith, Aug 08 '13 at 14:00
@DavidDury More like this: http://blog.onderweg.eu/2011/02/simple-parsing-with-csharp/ — Jesse Smith, Aug 08 '13 at 14:01
@DavidDury You wouldn't have to write the parser on your own, you just have to define the syntax (see http://www.antlr.org/) — oddparity, Aug 08 '13 at 14:02

Michael · Answer 1 · 2013-08-08T14:59:06.463

This might give you a starting point. Handle the exceptions however you'd like.

The method travels through the input string, setting an 'open' flag and index when it finds the OpenToken. When the 'open' flag is true, and a CloseToken is found, it extracts a substring based on the index and current position.

If the ThrowOnError property is set to true, and a token is found in an unexpected location, it will throw an exception.

This code could easily be modified to handle the unexpected tokens differently... such as skipping that match entirely, adding the match as-is, or whatever you desire.

public class CustomTokenParser
{
    public char OpenToken { get; set; }
    public char CloseToken { get; set; }

    public bool ThrowOnError { get; set; }

    public CustomTokenParser()
    {
        OpenToken = '{';
        CloseToken = '}';
        ThrowOnError = true;
    }

    public CustomTokenParser(char openToken, char closeToken, bool throwOnError)
    {
        this.OpenToken = openToken;
        this.CloseToken = closeToken;
        this.ThrowOnError = throwOnError;
    }        

    public string[] Parse(string input)
    {
        bool open = false;
        int openIndex = -1;
        List<string> matches = new List<string>();

        for (int i = 0; i < input.Length; i++)
        {
            if (!open && input[i] == OpenToken)
            {
                open = true;
                openIndex = i;
            }
            else if (open && input[i] == CloseToken)
            {
                open = false;
                string match = input.Substring(openIndex + 1, i - openIndex - 1);
                matches.Add(match);
            }
            else if (open && input[i] == OpenToken && ThrowOnError)
                throw new Exception("Open token found while match is open");
            else if (!open && input[i] == CloseToken && ThrowOnError)
                throw new Exception("Close token found while match is not open");
        }

        return matches.ToArray();
    }
}

+1 That's the only way to parse a complex format. Regex only works for patterns, not for validating a format. For the very same reasons regex can't be used to "parse HTML". — Tseng, Aug 08 '13 at 14:54

score 2 · Answer 2 · answered Aug 08 '13 at 15:03

You may use Balancing Group Definitions:

class Program
{
    static void Main(string[] args)
    {
        string rawInput = @"**Hello {Full{Name}, I { wanted to 
            inform that your product {{ProductName} is ready.
            Please come to our } address {Addr{Street}ess} to get it!**";

        string pattern = "^[^{}]*" +
                       "(" +
                       "((?'Open'{)[^{}]*)+" +
                       "((?'Close-Open'})[^{}]*)+" +
                       ")*" +
                       "(?(Open)(?!))$";

        var tokens = Regex.Match(
            Regex.Match(rawInput, @"{[\s\S]*}").Value,
            pattern,
            RegexOptions.Multiline)
                .Groups["Close"]
                .Captures
                .Cast<Capture>()
                .Where(c =>
                    !c.Value.Contains('{') &&
                    !c.Value.Contains('}'))
                .ToList();

        tokens.ForEach(c =>
        {
            Console.WriteLine(c.Value);
        });
    }
}

The above outputs:

ProductName
Street

score 0 · Answer 3 · answered Aug 08 '13 at 14:41

I kinda made it work 90% using this regex:

Regex rx = new Regex("{=[^{^=^}^<^>]*=}");

however, this matches my token between {= =} instead of just { }

If I have a token like this {=FullName=}, it will be replaces with the actual name, but if the token is {=Full{Name=} it will not be replaced because is incorrect and will be ignored ... this is the idea ... now, how can I use just { } ?

Match regex between curly braces

3 Answers3