Add ITrashIntrospection, impl on PKM classes

new api allows for checking for trash byte metadata, makes it much easier to write a verifier now

PKHeX.Core/PKM/BK4.cs

@@ -319,4 +319,9 @@ public sealed class BK4 : G4PKM
         => StringConverter4.LoadString(data, destBuffer);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter4.SetString(destBuffer, value, maxLength, Language, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes.GetTerminatorIndex(data, StringConverter4.Terminator);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes.GetStringLength(data, StringConverter4.Terminator);
+    public override int GetBytesPerChar() => 2;
 }

PKHeX.Core/PKM/CK3.cs

@@ -237,4 +237,9 @@ public sealed class CK3(byte[] Data) : G3PKM(Data), IShadowCapture
         => StringConverter3GC.LoadString(data, destBuffer);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter3GC.SetString(destBuffer, value, maxLength, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes.GetTerminatorIndex(data, StringConverter3GC.TerminatorBigEndian);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes.GetStringLength(data, StringConverter3GC.TerminatorBigEndian);
+    public override int GetBytesPerChar() => 2;
 }

PKHeX.Core/PKM/HOME/PKH.cs

@@ -432,6 +432,11 @@ public sealed class PKH : PKM, IHandlerLanguage, IFormArgument, IHomeTrack, IBat
         => StringConverter8.LoadString(data, destBuffer);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter8.SetString(destBuffer, value, maxLength, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes.GetTerminatorIndex(data);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes.GetStringLength(data);
+    public override int GetBytesPerChar() => 2;
 
     /// <summary>
     /// Revises the Handler details of a <see cref="PKM"/> to match the current <see cref="ITrainerInfo"/>.

PKHeX.Core/PKM/PA8.cs

@@ -738,4 +738,9 @@ public sealed class PA8 : PKM, ISanityChecksum,
         => StringConverter8.LoadString(data, destBuffer);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter8.SetString(destBuffer, value, maxLength, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes.GetTerminatorIndex(data);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes.GetStringLength(data);
+    public override int GetBytesPerChar() => 2;
 }

PKHeX.Core/PKM/PB7.cs

@@ -593,4 +593,9 @@ public sealed class PB7 : G6PKM, IHyperTrain, IAwakened, IScaledSizeValue, IComb
         => StringConverter8.LoadString(data, destBuffer);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter8.SetString(destBuffer, value, maxLength, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes.GetTerminatorIndex(data);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes.GetStringLength(data);
+    public override int GetBytesPerChar() => 2;
 }

PKHeX.Core/PKM/PB8.cs

@@ -141,4 +141,9 @@ public sealed class PB8 : G8PKM
         => StringConverter8.LoadString(data, destBuffer);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter8.SetString(destBuffer, value, maxLength, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes.GetTerminatorIndex(data);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes.GetStringLength(data);
+    public override int GetBytesPerChar() => 2;
 }

PKHeX.Core/PKM/PK1.cs

@@ -254,6 +254,11 @@ public sealed class PK1 : GBPKML, IPersonalType
         => StringConverter1.LoadString(data, destBuffer, Japanese);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter1.SetString(destBuffer, value, maxLength, Japanese, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes.GetTerminatorIndex(data, StringConverter4.Terminator);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes.GetStringLength(data, StringConverter4.Terminator);
+    public override int GetBytesPerChar() => 2;
 
     /// <summary>
     /// Gets a checksum over all the entity's data using a single list to wrap all components.

PKHeX.Core/PKM/PK2.cs

@@ -279,6 +279,11 @@ public sealed class PK2 : GBPKML, ICaughtData2
             return StringConverter2KOR.SetString(destBuffer, value, maxLength, option);
         return StringConverter2.SetString(destBuffer, value, maxLength, Language, option);
     }
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => Korean ? StringConverter2KOR.GetTerminatorIndex(data) : TrashBytesGB.GetTerminatorIndex(data);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => Korean ? StringConverter2KOR.GetStringLength(data) : TrashBytesGB.GetStringLength(data);
+    public override int GetBytesPerChar() => 1;
 
     /// <summary>
     /// Gets a checksum over all the entity's data using a single list to wrap all components.

PKHeX.Core/PKM/PK3.cs

@@ -362,4 +362,9 @@ public sealed class PK3 : G3PKM, ISanityChecksum
         => StringConverter3.LoadString(data, destBuffer, Language);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter3.SetString(destBuffer, value, maxLength, Language, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes8.GetTerminatorIndex(data);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes8.GetStringLength(data);
+    public override int GetBytesPerChar() => 1;
 }

PKHeX.Core/PKM/PK4.cs

@@ -395,4 +395,9 @@ public sealed class PK4 : G4PKM
         => StringConverter4.LoadString(data, destBuffer);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter4.SetString(destBuffer, value, maxLength, Language, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes.GetTerminatorIndex(data, StringConverter4.Terminator);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes.GetStringLength(data, StringConverter4.Terminator);
+    public override int GetBytesPerChar() => 2;
 }

PKHeX.Core/PKM/PK5.cs

@@ -572,6 +572,11 @@ public sealed class PK5 : PKM, ISanityChecksum,
         => StringConverter5.LoadString(data, destBuffer);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter5.SetString(destBuffer, value, maxLength, Language, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes.GetTerminatorIndex(data, StringConverter5.Terminator);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes.GetStringLength(data, StringConverter5.Terminator);
+    public override int GetBytesPerChar() => 2;
 
     /// <inheritdoc cref="G4PKM.CheckKoreanNidoranDPPt"/>
     /// <remarks> Gen4->Gen5 chars transfer without resetting the name. Still relevant even as PK5. </remarks>

PKHeX.Core/PKM/PK6.cs

@@ -531,4 +531,9 @@ public sealed class PK6 : G6PKM, IRibbonSetEvent3, IRibbonSetEvent4, IRibbonSetC
         => StringConverter6.LoadString(data, destBuffer);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter6.SetString(destBuffer, value, maxLength, Language, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes.GetTerminatorIndex(data);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes.GetStringLength(data);
+    public override int GetBytesPerChar() => 2;
 }

PKHeX.Core/PKM/PK7.cs

@@ -556,6 +556,11 @@ public sealed class PK7 : G6PKM, IRibbonSetEvent3, IRibbonSetEvent4, IRibbonSetC
         => StringConverter7.LoadString(data, destBuffer);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter7.SetString(destBuffer, value, maxLength, Language);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes.GetTerminatorIndex(data);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes.GetStringLength(data);
+    public override int GetBytesPerChar() => 2;
 }
 
 public enum ResortEventState : byte

PKHeX.Core/PKM/PK8.cs

@@ -124,4 +124,9 @@ public sealed class PK8 : G8PKM, IHandlerUpdate
         => StringConverter8.LoadString(data, destBuffer);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter8.SetString(destBuffer, value, maxLength, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes.GetTerminatorIndex(data);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes.GetStringLength(data);
+    public override int GetBytesPerChar() => 2;
 }

PKHeX.Core/PKM/PK9.cs

@@ -677,4 +677,9 @@ public sealed class PK9 : PKM, ISanityChecksum, ITeraType, ITechRecord, IObedien
         => StringConverter8.LoadString(data, destBuffer);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter8.SetString(destBuffer, value, maxLength, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes.GetTerminatorIndex(data);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes.GetStringLength(data);
+    public override int GetBytesPerChar() => 2;
 }

PKHeX.Core/PKM/PKM.cs

@@ -11,7 +11,7 @@ namespace PKHeX.Core;
 /// Object representing a <see cref="PKM"/>'s data and derived properties.
 /// </summary>
 [DynamicallyAccessedMembers(PublicProperties | NonPublicProperties | PublicParameterlessConstructor)]
-public abstract class PKM : ISpeciesForm, ITrainerID32, IGeneration, IShiny, ILangNick, IGameValueLimit, INature, IFatefulEncounter, IStringConverter
+public abstract class PKM : ISpeciesForm, ITrainerID32, IGeneration, IShiny, ILangNick, IGameValueLimit, INature, IFatefulEncounter, IStringConverter, ITrashIntrospection
 {
     /// <summary>
     /// Valid file extensions that represent <see cref="PKM"/> data, without the leading '.'
@@ -151,6 +151,9 @@ public abstract class PKM : ISpeciesForm, ITrainerID32, IGeneration, IShiny, ILa
     public abstract string GetString(ReadOnlySpan<byte> data);
     public abstract int LoadString(ReadOnlySpan<byte> data, Span<char> text);
     public abstract int SetString(Span<byte> data, ReadOnlySpan<char> text, int length, StringConverterOption option);
+    public abstract int GetStringTerminatorIndex(ReadOnlySpan<byte> data);
+    public abstract int GetStringLength(ReadOnlySpan<byte> data);
+    public abstract int GetBytesPerChar();
 
     /// <summary>
     /// The date the Pokémon was met.

PKHeX.Core/PKM/RK4.cs

@@ -357,4 +357,9 @@ public sealed class RK4 : G4PKM
         => StringConverter4.LoadString(data, destBuffer);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter4.SetString(destBuffer, value, maxLength, Language, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes.GetTerminatorIndex(data, StringConverter4.Terminator);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes.GetStringLength(data, StringConverter4.Terminator);
+    public override int GetBytesPerChar() => 2;
 }

PKHeX.Core/PKM/SK2.cs

@@ -221,4 +221,9 @@ public sealed class SK2 : GBPKM, ICaughtData2
         => StringConverter2.LoadString(data, destBuffer, Language);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter2.SetString(destBuffer, value, maxLength, Language, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytesGB.GetTerminatorIndex(data);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytesGB.GetStringLength(data);
+    public override int GetBytesPerChar() => 1;
 }

PKHeX.Core/PKM/Strings/StringConverter2KOR.cs

@@ -76,6 +76,40 @@ public static class StringConverter2KOR
         return ctr;
     }
 
+    /// <summary>
+    /// Gets the count of characters in the data.
+    /// </summary>
+    public static int GetStringLength(ReadOnlySpan<byte> data)
+    {
+        int length = 0;
+        for (var i = 0; i < data.Length; i++)
+        {
+            byte value = data[i];
+            if (value is TableInvalid or TerminatorCode)
+                break;
+            if (value <= TableMax)
+                i++; // Korean char takes 2 bytes
+            length++;
+        }
+        return length;
+    }
+
+    /// <summary>
+    /// Gets the index of the first terminator in the data.
+    /// </summary>
+    public static int GetTerminatorIndex(ReadOnlySpan<byte> data)
+    {
+        for (var i = 0; i < data.Length; i++)
+        {
+            byte value = data[i];
+            if (value is TableInvalid or TerminatorCode)
+                return i;
+            if (value <= TableMax)
+                i++; // Korean char takes 2 bytes
+        }
+        return -1;
+    }
+
     /// <summary>
     /// Converts a string to Generation 1 encoded data.
     /// </summary>

PKHeX.Core/PKM/Strings/StringConverter3GC.cs

@@ -8,7 +8,7 @@ namespace PKHeX.Core;
 /// </summary>
 public static class StringConverter3GC
 {
-    private const char TerminatorBigEndian = (char)0; // GC
+    public const char TerminatorBigEndian = (char)0; // GC
 
     /// <summary>Converts Big Endian encoded data to decoded string.</summary>
     /// <param name="data">Encoded data</param>

PKHeX.Core/PKM/Strings/StringConverter5.cs

@@ -8,7 +8,7 @@ namespace PKHeX.Core;
 /// </summary>
 public static class StringConverter5
 {
-    private const char TerminatorFFFF = (char)0xFFFF;
+    public const char Terminator = (char)0xFFFF;
 
     /// <summary>Converts Generation 5 encoded data to decoded string.</summary>
     /// <param name="data">Encoded data</param>
@@ -31,7 +31,7 @@ public static class StringConverter5
         for (; i < data.Length; i += 2)
         {
             var value = ReadUInt16LittleEndian(data[i..]);
-            if (value == TerminatorFFFF)
+            if (value == Terminator)
                 break;
             result[ctr++] = (char)value;
         }
@@ -68,7 +68,7 @@ public static class StringConverter5
         int count = value.Length * 2;
         if (count == destBuffer.Length)
             return count;
-        WriteUInt16LittleEndian(destBuffer[count..], TerminatorFFFF);
+        WriteUInt16LittleEndian(destBuffer[count..], Terminator);
         return count + 2;
     }
 }

PKHeX.Core/PKM/Strings/Trash/ITrashIntrospection.cs

@@ -0,0 +1,10 @@
+using System;
+
+namespace PKHeX.Core;
+
+public interface ITrashIntrospection
+{
+    int GetStringTerminatorIndex(ReadOnlySpan<byte> data);
+    int GetStringLength(ReadOnlySpan<byte> data);
+    int GetBytesPerChar();
+}

PKHeX.Core/PKM/Strings/Trash/TrashBytes.cs

@@ -16,7 +16,7 @@ public static class TrashBytes
     /// <returns>Decoded index (char) of the terminator, or max length if not found.</returns>
     public static int GetStringLength(ReadOnlySpan<byte> buffer, ushort terminator = 0)
     {
-        int index = FindTerminatorIndex(buffer, terminator);
+        int index = GetTerminatorIndex(buffer, terminator);
         return index == -1 ? buffer.Length / 2 : index;
     }
 
@@ -27,7 +27,7 @@ public static class TrashBytes
     /// <param name="terminator">Terminator character to search for.</param>
     /// <returns>Decoded index (char) of the terminator, or -1 if not found.</returns>
     /// <remarks>When used on a raw string, returns the computed length of the string, assuming a terminator is present.</remarks>
-    public static int FindTerminatorIndex(ReadOnlySpan<byte> buffer, ushort terminator = 0)
+    public static int GetTerminatorIndex(ReadOnlySpan<byte> buffer, ushort terminator = 0)
     {
         var u16 = MemoryMarshal.Cast<byte, ushort>(buffer);
         return u16.IndexOf(terminator);

PKHeX.Core/PKM/Strings/Trash/TrashBytes8.cs

@@ -10,7 +10,7 @@ public static class TrashBytes8
     /// <inheritdoc cref="TrashBytes.GetStringLength"/>
     public static int GetStringLength(ReadOnlySpan<byte> buffer)
     {
-        int index = FindTerminatorIndex(buffer);
+        int index = GetTerminatorIndex(buffer);
         return index == -1 ? buffer.Length : index;
     }
 
@@ -19,6 +19,6 @@ public static class TrashBytes8
     /// </summary>
     /// <param name="buffer">Backing buffer of the string.</param>
     /// <returns>Index of the terminator, or -1 if not found.</returns>
-    public static int FindTerminatorIndex(ReadOnlySpan<byte> buffer)
+    public static int GetTerminatorIndex(ReadOnlySpan<byte> buffer)
         => buffer.IndexOf<byte>(0xFF);
 }

PKHeX.Core/PKM/Strings/Trash/TrashBytesGB.cs

@@ -11,15 +11,15 @@ public static class TrashBytesGB
     /// <inheritdoc cref="TrashBytes.GetStringLength"/>
     public static int GetStringLength(ReadOnlySpan<byte> buffer)
     {
-        int index = FindTerminatorIndex(buffer);
+        int index = GetTerminatorIndex(buffer);
         return index == -1 ? buffer.Length : index;
     }
 
     /// <summary>
-    /// Returns a 16-bit aligned index of the terminator.
+    /// Returns a 8-bit aligned index of the terminator.
     /// </summary>
     /// <param name="buffer">Backing buffer of the string.</param>
     /// <returns>Index of the terminator, or -1 if not found.</returns>
-    public static int FindTerminatorIndex(ReadOnlySpan<byte> buffer)
+    public static int GetTerminatorIndex(ReadOnlySpan<byte> buffer)
         => buffer.IndexOfAny(TerminatorZero, TerminatorCode);
 }

PKHeX.Core/PKM/XK3.cs

@@ -241,4 +241,9 @@ public sealed class XK3 : G3PKM, IShadowCapture
         => StringConverter3GC.LoadString(data, destBuffer);
     public override int SetString(Span<byte> destBuffer, ReadOnlySpan<char> value, int maxLength, StringConverterOption option)
         => StringConverter3GC.SetString(destBuffer, value, maxLength, option);
+    public override int GetStringTerminatorIndex(ReadOnlySpan<byte> data)
+        => TrashBytes.GetTerminatorIndex(data, StringConverter3GC.TerminatorBigEndian);
+    public override int GetStringLength(ReadOnlySpan<byte> data)
+        => TrashBytes.GetStringLength(data, StringConverter3GC.TerminatorBigEndian);
+    public override int GetBytesPerChar() => 2;
 }