using System.Globalization;
using System.Text;

namespace DataTransfer.Client.Services;

public static class FileNameSanitizer
{
    private const int MaxLength = 255;
    private const string Fallback = "file";

    // Characters invalid in file names on Windows, macOS or Linux. Path.GetInvalidFileNameChars() returns only the
    // Unix set in WebAssembly, but the file is saved on any operating system.
    private const string InvalidCharacters = "<>:\"/\\|?*";

    /// <summary>
    ///     Makes a file name received from the peer safe to display and save. Invisible formatting characters are removed,
    ///     so for example a right-to-left override can't disguise <c>invoice&#x202E;fdp.exe</c> as <c>invoiceexe.pdf</c>.
    /// </summary>
    public static string Sanitize(string? name)
    {
        if (string.IsNullOrEmpty(name))
        {
            return Fallback;
        }

        var builder = new StringBuilder(Math.Min(name.Length, MaxLength));
        foreach (var rune in name.EnumerateRunes())
        {
            switch (Rune.GetUnicodeCategory(rune))
            {
                case UnicodeCategory.Format:
                    continue;
                case UnicodeCategory.Control:
                    Append(builder, new Rune('_'));
                    continue;
            }

            Append(builder, rune.IsAscii && InvalidCharacters.Contains((char)rune.Value) ? new Rune('_') : rune);
        }

        // Windows silently drops trailing dots and spaces, which could change the extension.
        var cleaned = builder.ToString().Trim().TrimEnd('.');
        return cleaned.Length == 0 ? Fallback : cleaned;
    }

    // Appends whole runes only, so truncation never splits a surrogate pair.
    private static void Append(StringBuilder builder, Rune rune)
    {
        if (builder.Length + rune.Utf16SequenceLength <= MaxLength)
        {
            builder.Append(rune.ToString());
        }
    }
}