IdentityShroud/IdentityShroud.Core/Helpers/SlugHelper.cs

using System.Globalization;
using System.Security.Cryptography;
using System.Text;
using Microsoft.AspNetCore.WebUtilities;

namespace IdentityShroud.Core.Helpers;

public static class SlugHelper
{
    public static string GenerateSlug(string text, int maxLength = 40)
    {
        if (string.IsNullOrWhiteSpace(text))
            return string.Empty;

        // Normalize to decomposed form (separates accents from letters)
        string normalized = text.Normalize(NormalizationForm.FormD);
        
        StringBuilder sb = new StringBuilder(normalized.Length);
        bool lastWasHyphen = false;
        
        foreach (char c in normalized)
        {
            // Skip diacritics (accents)
            if (CharUnicodeInfo.GetUnicodeCategory(c) == UnicodeCategory.NonSpacingMark)
                continue;
            
            char lower = char.ToLowerInvariant(c);
            
            // Convert valid characters
            if ((lower >= 'a' && lower <= 'z') || (lower >= '0' && lower <= '9'))
            {
                sb.Append(lower);
                lastWasHyphen = false;
            }
            // Convert spaces, underscores, and other separators to hyphen
            else if (char.IsWhiteSpace(c) || c == '_' || c == '-')
            {
                if (!lastWasHyphen && sb.Length > 0)
                {
                    sb.Append('-');
                    lastWasHyphen = true;
                }
            }
            // Skip all other characters
        }
        
        // Trim trailing hyphen if any
        if (sb.Length > 0 && sb[sb.Length - 1] == '-')
            sb.Length--;
        
        string slug = sb.ToString();
        
        // Handle truncation with hash suffix for long strings
        if (slug.Length > maxLength)
        {
            // Generate hash of original text
            string hashSuffix = GenerateHashSuffix(text);
            int contentLength = maxLength - hashSuffix.Length;
            
            // Truncate at word boundary if possible
            int cutPoint = contentLength;
            int lastHyphen = slug.LastIndexOf('-', contentLength - 1);
            
            if (lastHyphen > contentLength / 2)
                cutPoint = lastHyphen;
            
            slug = slug.Substring(0, cutPoint).TrimEnd('-') + hashSuffix;
        }
        
        return slug;
    }
    
    private static string GenerateHashSuffix(string text)
    {
        using (var md5 = MD5.Create())
        {
            byte[] hash = md5.ComputeHash(Encoding.UTF8.GetBytes(text));
            
            // Take first 4 bytes (will become ~5-6 base64url chars)
            string base64Url = WebEncoders.Base64UrlEncode(hash, 0, 4);
            return "-" + base64Url;
        }
    }
}