Regex.Matches to get substrings between <>: I need add ( and ) also as substring in the result

Question

I'm developing a C# library with .NET Framework 4.7.1 and Visual Studio 2017.

I have to split these three strings:

"<expr><op><expr>"
"(<expr><op><expr>)"
"<pre_op>(<expr>)"

into

string[] { "<expr>", "<op>", "<expr>" }
string[] { "(", "<expr>", "<op>", "<expr>", ")"}
string[] { "<pre_op>", "(", "<expr>", ")" }

The following sentence works fine, but it doesn't handle ( or ):

string[] result1 = Regex
    .Matches("<expr><op><expr>", @"<.*?>")
    .Cast<Match>()
    .Select(m => m.Value)
    .ToArray();
string[] result2 = Regex
    .Matches("(<expr><op><expr>)", @"<.*?>")
    .Cast<Match>()
    .Select(m => m.Value)
    .ToArray();

But result1 and result2 are equal?

How can I handle ( and ) to get them as substrings also?

You want to make your regex ungreedy. Or else it will grab between `()` without finding matches in between. — user3483203, Mar 27 '18 at 16:20
@chrisz Thanks. I will look for what ungreedy means (and some examples). Thanks, again. — VansFannel, Mar 27 '18 at 16:24
@revo Yep I like yours better for the first character group (`[()]`), not sure about `[^<>]` though. Not quite sure about performance, but to me `<.*?>` is more readable. — Arian Motamedi, Mar 27 '18 at 16:33

score 0 · Answer 1 · answered Mar 28 '18 at 06:25

string[] result1 = Regex
    .Matches("<expr><op><expr>", **@"(\(|<.*?>|\))"**)
    .Cast<Match>()
    .Select(m => m.Value)
    .ToArray();
            string[] result2 = Regex
                .Matches("(<expr><op><expr>)", @"(\(|<.*?>|\))")
                .Cast<Match>()
                .Select(m => m.Value)
                .ToArray();
            string[] result3 = Regex
                .Matches("<pre_op>(<expr>)", @"(\(|<.*?>|\))")
                .Cast<Match>()
                .Select(m => m.Value)
                .ToArray();

wp78de · Answer 2 · 2018-03-29T06:29:21.740

This is arguably a nice use case for .NET balancing groups to ensure all round and angle brackets are balanced, see also What are regular expression Balancing Groups?

The first pattern does just that

(?<B>\()+[^()]+(?<-B>\))(?(B)(?!))|(?<A><)+[^<>]+(?<-A>>)+(?(A)(?!))

matching <tags> and (everything between brackets) The second regex then is used to tokenize the perfectly balanced parts.

\(|<.*?>|\)

Sample code:

using System;
using System.Text.RegularExpressions;
public class Example
{
    public static void Main()
    {
        string input = @"<expr><op><expr>
(<expr><op><expr>)
<pre_op>(<expr>)
(<expr>)<pre_op>(<expr>)";

         Regex rxBalanced = new Regex(
            @"(?<B>\()+[^()]+(?<-B>\))(?(B)(?!))|(?<A><)+[^<>]+(?<-A>>)+(?(A)(?!))",
            RegexOptions.Multiline
            | RegexOptions.CultureInvariant
            | RegexOptions.IgnorePatternWhitespace
            | RegexOptions.Compiled
        );
        Regex rxTokens = new Regex(
            @"\(|<.*?>|\)",
            RegexOptions.Multiline
            | RegexOptions.CultureInvariant
            | RegexOptions.IgnorePatternWhitespace
            | RegexOptions.Compiled
        );

        foreach (Match match in rxBalanced.Matches(input))
        {
            foreach (Match token in rxTokens.Matches(match.Value))
            {
                Console.WriteLine(token.Value);
            }
        }
    }
}

An alternative pattern that would allow you to check both balancing groups at the same time could look like this

(?<B>(\())*((?<A><)+[^<>]+(?<-A>>)+(?(A)(?!)))+(?<-B>(\)))*(?(B)(?!))

Unfortunately, it's much harder to get all the needed values from the resulting nested collection. However, I found this problem interesting enough to create a LinQ-query that does all the black magic:

var regex = new Regex("(?<B>(\\())*((?<A><)+[^<>]+(?<-A>>)+(?(A)(?!)))+(?<-B>(\\)))*(?(B)(?!))",
    RegexOptions.Multiline | RegexOptions.CultureInvariant | RegexOptions.Compiled);

var x = (from Match m in regex.Matches("(<x><y><z>)<expr>(<a><b><c>)<d>")
        select new
        {
            result = m.Groups[1].Value.StartsWith("(") ?
                        (new List<string> { "(" }
                            .Concat(m.Groups[2].Captures.Count > 1 ?
                                (from Capture c in m.Groups[2].Captures select c.Value).ToList()
                                : new List<string> { m.Groups[2].Value }
                            )
                            .Concat(new List<string> { ")" })
                        )
                        : new List<string> { m.Value }
        }).SelectMany(r => r.result);

Regex.Matches to get substrings between <>: I need add ( and ) also as substring in the result

2 Answers2