using System.Text;

namespace ZB.MOM.WW.OtOpcUa.Driver.TwinCAT;

/// <summary>
///     Decodes an IEC <c>STRING(n)</c> / <c>WSTRING(n)</c> buffer read at the PLC's own declared width.
///     The driver reads string symbols as raw bytes and decodes them here instead of letting the
///     Beckhoff marshaller pick an encoding, because that marshaller falls back to the HOST's ANSI code
///     page (<c>CultureInfo.CurrentCulture.TextInfo.ANSICodePage</c>) for a non-UTF-8 target — so a
///     TwinCAT 2 string would decode differently on a Polish (1250) or Cyrillic (1251) server.
/// </summary>
/// <remarks>
///     Two rules, both ported from the ignitionoee TwinCAT driver and both found on a real PLC:
///     <list type="bullet">
///         <item>A value ENDS at its first NUL — <c>0x00</c> for a STRING, an aligned <c>0x00 0x00</c>
///         code unit for a WSTRING. Whatever follows is stale content a longer earlier value left in the
///         buffer (live on the Elma cleaner Z33856: work order <c>67504440</c> read back with a stale
///         <c>"\0" + "1"</c> tail).</item>
///         <item>A STRING is single-byte <b>Windows-1252</b> (umlauts, <c>°</c> in alarm texts), unless
///         the target reports UTF-8 string data (TwinCAT 3.1 4026+ "UTF8 Encoding for STRING").
///         ISO-8859-1 is the fallback on a runtime without code-page support — identical outside
///         0x80–0x9F.</item>
///     </list>
/// </remarks>
internal static class TwinCATStringCodec
{
    private static readonly Encoding Windows1252 = CreateWindows1252();

    private static Encoding CreateWindows1252()
    {
        try
        {
            Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
            return Encoding.GetEncoding(1252);
        }
        catch (Exception)
        {
            return Encoding.Latin1;
        }
    }

    /// <summary>Decodes an IEC <c>STRING(n)</c> buffer: ends at the first NUL, Windows-1252 unless <paramref name="utf8"/>.</summary>
    /// <param name="bytes">The raw symbol bytes (n+1 for a <c>STRING(n)</c>).</param>
    /// <param name="utf8">True when the target reports UTF-8 string data.</param>
    /// <returns>The decoded value.</returns>
    public static string DecodeString(ReadOnlySpan<byte> bytes, bool utf8)
    {
        var end = bytes.IndexOf((byte)0);
        if (end >= 0) bytes = bytes[..end];
        return (utf8 ? Encoding.UTF8 : Windows1252).GetString(bytes);
    }

    /// <summary>Decodes an IEC <c>WSTRING(n)</c> buffer (UTF-16LE): ends at the first aligned <c>U+0000</c>.</summary>
    /// <param name="bytes">The raw symbol bytes (2(n+1) for a <c>WSTRING(n)</c>).</param>
    /// <returns>The decoded value.</returns>
    public static string DecodeWString(ReadOnlySpan<byte> bytes)
    {
        var end = bytes.Length & ~1; // whole UTF-16 code units only — a trailing odd byte is ignored
        for (var i = 0; i + 1 < end; i += 2)
        {
            if (bytes[i] == 0 && bytes[i + 1] == 0)
            {
                end = i;
                break;
            }
        }
        return Encoding.Unicode.GetString(bytes[..end]);
    }
}