Repeating pattern using regex in C#

Question

I have a string of words:

word dark king glow we end hello bye low wing

I need to find words where last letter of first word matches first letter of following word (example: worD Dark).

I wrote a regex expression:

\b\w*(\w)\W\1\w*\b

Currently it successfully finds 2 words in a row (Regex.Matches[0].Value = "word dark" ; Regex.Matches[1].Value = "king glow" etc.)

I need a regex expression which would read it as a pattern (Regex.Matches[0].Value = "word dark king glow we end" ; Regex.Matches[1].Value = "low wing").

How should I approach this?

Answer 1

For the record here a very expressive non regex version. I does't require picture ;)

static IEnumerable<(string W1, string W2)> GetPairs1(string input)
{
    var words = input.Split(' ', StringSplitOptions.RemoveEmptyEntries);

    if (!words.Any()) yield break;

    for( int i = 1; i < words.Length; i++) 
        if(words[i][0] == words[i-1][words[i-1].Length-1]) 
            yield return (words[i-1], words[i]);
}

Test

public static async Task Main()
{
    var input = "word dark king glow we end hello bye low wing";

    foreach (var p in GetPairs1(input)) 
        Console.WriteLine($"{p.W1} {p.W2}");
}

Output

word dark
dark king
king glow
glow we
we end
low wing

Answer 2

I would also capture the last word character, check inside a lookahead if it matches the first character of the next word, put all into a group for repetition and if the condition succeeded, match following word.

(?i)(?:\b\w*(\w) +(?=\1))+\w+

See this demo at regex101

Used with caseless flag (?i) for captured a matching A in the following word.

• if you expect non word characters besides space, use negated upper \W instead
• for even digits and underscores between, use [a-z] instead of \w and negation
• if working with unicode, use \p{L} for letter and negated upper \P{L} for non letter

Answer 3

Good question and there is good answer here.

With Positive Lookahead

I guess,

(?is)\w*(\w)(?= (\1)\w*)

might be somewhat closer, there might be edge cases though, for which you'd probably want to look into the positive lookahead here in this block:

(?= (\1)\w*)

RegEx Demo 1

---

With Positive Lookbehind

You can also lookbehind, and capture things, if/as you wish and code, maybe with some expression similar to:

(?is)(?<=([a-z])\s)(\1)([a-z]*)

RegEx Demo 2

Test

using System;
using System.Text.RegularExpressions;

public class Example
{
    public static void Main()
    {
        string pattern = @"(?is)\w*(\w)(?= (\1)\w*)";
        string input = @"word dark king glow we end hello bye low wing
word Dark King Glow We End hello bye LoW wing";

        foreach (Match m in Regex.Matches(input, pattern))
        {
            Console.WriteLine("'{0}' found at index {1}.", m.Value, m.Index);
        }
    }
}

---

If you wish to simplify/modify/explore the expression, it's been explained on the top right panel of regex101.com. If you'd like, you can also watch in this link, how it would match against some sample inputs.

---

---

Complexity

Lookarounds in general are not really complexity-friendly methods, yet I can't think of a better way now.