Skip to content
Merged
Show file tree
Hide file tree
Changes from 1 commit
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
File renamed without changes.
File renamed without changes.
311 changes: 311 additions & 0 deletions docs/serialization-format.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,311 @@
# Neo Serialization Format

This document describes the binary serialization format used by the Neo blockchain platform. The format is designed for efficient serialization and deserialization of blockchain data structures.

## Overview

Neo uses a custom binary serialization format that supports:
- Primitive data types (integers, booleans, bytes)
- Variable-length integers (VarInt)
- Strings (fixed and variable length)
- Arrays and collections
- Custom serializable objects
- Nullable objects

## Core Interfaces

### ISerializable

All serializable objects in Neo implement the `ISerializable` interface:

```csharp
public interface ISerializable
{
int Size { get; }

void Serialize(BinaryWriter writer);

void Deserialize(ref MemoryReader reader);
}
```

- `Size`: Returns the serialized size in bytes
- `Serialize`: Writes the object to a BinaryWriter
- `Deserialize`: Reads the object from a MemoryReader

## Primitive Data Types

### Integers

Neo supports both little-endian and big-endian integer formats:

| Type | Size | Endianness | Description |
|------|------|------------|-------------|
| `sbyte` | 1 byte | N/A | Signed 8-bit integer |
| `byte` | 1 byte | N/A | Unsigned 8-bit integer |
| `short` | 2 bytes | Little-endian | Signed 16-bit integer |
| `ushort` | 2 bytes | Little-endian | Unsigned 16-bit integer |
| `int` | 4 bytes | Little-endian | Signed 32-bit integer |
| `uint` | 4 bytes | Little-endian | Unsigned 32-bit integer |
| `long` | 8 bytes | Little-endian | Signed 64-bit integer |
| `ulong` | 8 bytes | Little-endian | Unsigned 64-bit integer |

Big-endian variants are available for `short`, `ushort`, `int`, `uint`, `long`, and `ulong`.

### Boolean

Booleans are serialized as single bytes:
- `false` → `0x00`
- `true` → `0x01`
- Any other value throws `FormatException`

### Variable-Length Integers (VarInt)

Neo uses a compact variable-length integer format:

| Value Range | Format | Size |
|-------------|--------|------|
| 0-252 | Direct value | 1 byte |
| 253-65535 | `0xFD` + 2-byte little-endian | 3 bytes |
| 65536-4294967295 | `0xFE` + 4-byte little-endian | 5 bytes |
| 4294967296+ | `0xFF` + 8-byte little-endian | 9 bytes |

**Serialization:**
```csharp
if (value < 0xFD)
{
writer.Write((byte)value);
}
else if (value <= 0xFFFF)
{
writer.Write((byte)0xFD);
writer.Write((ushort)value);
}
else if (value <= 0xFFFFFFFF)
{
writer.Write((byte)0xFE);
writer.Write((uint)value);
}
else
{
writer.Write((byte)0xFF);
writer.Write(value);
}
```

**Deserialization:**
```csharp
var b = ReadByte();
var value = b switch
{
0xfd => ReadUInt16(),
0xfe => ReadUInt32(),
0xff => ReadUInt64(),
_ => b
};
```

## Strings

### Fixed-Length Strings

Fixed-length strings are padded with null bytes:

**Format:** `[UTF-8 bytes][zero padding]`

**Serialization:**
```csharp
var bytes = value.ToStrictUtf8Bytes();
if (bytes.Length > length)
throw new ArgumentException();
writer.Write(bytes);
if (bytes.Length < length)
writer.Write(new byte[length - bytes.Length]);
```

**Deserialization:**
```csharp
var end = currentOffset + length;
var offset = currentOffset;
while (offset < end && _span[offset] != 0) offset++;
var data = _span[currentOffset..offset];
for (; offset < end; offset++)
if (_span[offset] != 0)
throw new FormatException();
currentOffset = end;
return data.ToStrictUtf8String();
```

### Variable-Length Strings

Variable-length strings use VarInt for length prefix:

**Format:** `[VarInt length][UTF-8 bytes]`

**Serialization:**
```csharp
writer.WriteVarInt(value.Length);
writer.Write(value.ToStrictUtf8Bytes());
```

**Deserialization:**
```csharp
var length = (int)ReadVarInt((ulong)max);
EnsurePosition(length);
var data = _span.Slice(currentOffset, length);
currentOffset += length;
return data.ToStrictUtf8String();
```

## Byte Arrays

### Fixed-Length Byte Arrays

**Format:** `[raw bytes]`

### Variable-Length Byte Arrays

**Format:** `[VarInt length][raw bytes]`

**Serialization:**
```csharp
writer.WriteVarInt(value.Length);
writer.Write(value);
```

**Deserialization:**
```csharp
return ReadMemory((int)ReadVarInt((ulong)max));
```

## Collections

### Serializable Arrays

**Format:** `[VarInt count][item1][item2]...[itemN]`

**Serialization:**
```csharp
writer.WriteVarInt(value.Count);
foreach (T item in value)
{
item.Serialize(writer);
}
```

**Deserialization:**
```csharp
var array = new T[reader.ReadVarInt((ulong)max)];
for (var i = 0; i < array.Length; i++)
{
array[i] = new T();
array[i].Deserialize(ref reader);
}
return array;
```

### Nullable Arrays

**Format:** `[VarInt count][bool1][item1?][bool2][item2?]...[boolN][itemN?]`

**Serialization:**
```csharp
writer.WriteVarInt(value.Length);
foreach (var item in value)
{
var isNull = item is null;
writer.Write(!isNull);
if (isNull) continue;
item!.Serialize(writer);
}
```

**Deserialization:**
```csharp
var array = new T[reader.ReadVarInt((ulong)max)];
for (var i = 0; i < array.Length; i++)
array[i] = reader.ReadBoolean() ? reader.ReadSerializable<T>() : null;
return array;
```

## UTF-8 Encoding

Neo uses strict UTF-8 encoding with the following characteristics:

- **Strict Mode**: Invalid UTF-8 sequences throw exceptions
- **No Fallback**: No replacement characters for invalid sequences
- **Exception Handling**: Detailed error messages for debugging

**String to Bytes:**
```csharp
public static byte[] ToStrictUtf8Bytes(this string value)
{
return StrictUTF8.GetBytes(value);
}
```

**Bytes to String:**
```csharp
public static string ToStrictUtf8String(this ReadOnlySpan<byte> value)
{
return StrictUTF8.GetString(value);
}
```

## Error Handling

The serialization format includes comprehensive error handling:

- **FormatException**: Invalid data format or corrupted data
- **ArgumentNullException**: Null values where not allowed
- **ArgumentException**: Invalid arguments (e.g., string too long)
- **ArgumentOutOfRangeException**: Values outside allowed ranges
- **DecoderFallbackException**: Invalid UTF-8 sequences
- **EncoderFallbackException**: Characters that cannot be encoded

## Examples

### Simple Object Serialization

```csharp
public class SimpleData : ISerializable
{
public string Name { get; set; }
public int Value { get; set; }

public int Size => Name.GetStrictUtf8ByteCount() + sizeof(int);

public void Serialize(BinaryWriter writer)
{
writer.WriteVarString(Name);
writer.Write(Value);
}

public void Deserialize(ref MemoryReader reader)
{
Name = reader.ReadVarString();
Value = reader.ReadInt32();
}
}
```

### Array Serialization

```csharp
public class DataArray : ISerializable
{
public SimpleData[] Items { get; set; }

public int Size => Items.Sum(item => item.Size) + GetVarSize(Items.Length);

public void Serialize(BinaryWriter writer)
{
writer.Write(Items);
}

public void Deserialize(ref MemoryReader reader)
{
Items = reader.ReadSerializableArray<SimpleData>();
}
}
```