Bogo - High-Performance Binary Serialization for Go

Bogo is a fast, compact binary serialization format with embedded key fields and type information. It provides efficient encoding and decoding of data types with high performance and zero-copy operations.

Bogo is ideal when you need JSON-like simplicity with binary format performance, especially for selective field deserialization in complex data types.

Features

• JSON-Compatible API: Drop-in replacement for encoding/json with familiar Marshal/Unmarshal functions
• High Performance: Up to 3x faster deserialization compared to JSON
• Compact Binary Format: Efficient variable-length encoding reduces payload size
• Comprehensive Type Support: Supports all primitive types plus lists, objects, and binary data
• Streaming API: Memory-efficient streaming encoding and decoding
• Configurable Validation: Optional UTF-8 validation, depth limits, and size constraints
• Field-Specific Optimization: Revolutionary selective field decoding with up to 334x performance improvement and 113x memory reduction

Supported Types

Primitive Type	Bogo Type	Description
nil	TypeNull	Null values
bool	TypeBoolTrue/TypeBoolFalse	Boolean values
string	TypeString	UTF-8 strings with variable-length encoding
byte	TypeByte	Single byte values
int, int8, int16, int32, int64	TypeInt	Signed integers with VarInt encoding
uint, uint8, uint16, uint32, uint64	TypeUint	Unsigned integers with VarInt encoding
float32, float64	TypeFloat	IEEE 754 floating-point numbers
[]byte	TypeBlob	Binary data with length prefix
time	TypeTimestamp	Unix timestamps
[]any{}	TypeUntypedList	Heterogeneous lists
[]int{}	TypeTypedList	Homogeneous typed lists
object	TypeObject	Key-value objects

Installation

go get github.com/bubunyo/bogo

Quick Start

Basic Usage

package main
import (
"fmt""log""github.com/bubunyo/bogo"
)
typeUserstruct {
IDint`json:"id"`Namestring`json:"name"`Emailstring`json:"email"`Activebool`json:"active"`
}
funcmain() {
user:=User{
ID: 123,
Name: "John Doe",
Email: "john@example.com",
Active: true,
}
// Marshal to binary formatdata, err:=bogo.Marshal(user)
iferr!=nil {
log.Fatal(err)
}
// Unmarshal from binary formatvardecodedUsererr=bogo.Unmarshal(data, &decoded)
iferr!=nil {
log.Fatal(err)
}
fmt.Printf("Original: %+v\n", user)
fmt.Printf("Decoded: %+v\n", decoded)
}

Streaming API

package main
import (
"bytes""log""github.com/bubunyo/bogo"
)
funcmain() {
data:=map[string]interface{}{
"message": "Hello, World!",
"count": 42,
"active": true,
}
// Streaming encodingvarbuf bytes.Bufferencoder:=bogo.NewEncoder(&buf)
iferr:=encoder.Encode(data); err!=nil {
log.Fatal(err)
}
// Streaming decodingdecoder:=bogo.NewDecoder(&buf)
varresultmap[string]interface{}
iferr:=decoder.Decode(&result); err!=nil {
log.Fatal(err)
}
fmt.Printf("Streamed data: %+v\n", result)
}

Performance

Bogo delivers significant performance improvements over JSON serialization:

Benchmark Results (Apple M2, 8 cores)

Simple Data (small objects):
- JSON Serialize: 709 ns/op 272 B/op 3 allocs/op
- Bogo Serialize: 964 ns/op 1088 B/op 18 allocs/op (0.74x speed)
- MessagePack: 442 ns/op 320 B/op 4 allocs/op
- JSON Deserialize: 1910 ns/op 336 B/op 7 allocs/op
- Bogo Deserialize: 564 ns/op 488 B/op 16 allocs/op (3.39x faster)
- MessagePack: 623 ns/op 168 B/op 4 allocs/op
Complex Data (nested structures):
- JSON Serialize: 6422 ns/op 2514 B/op 36 allocs/op
- Bogo Serialize: 15449 ns/op 18939 B/op 291 allocs/op (0.42x speed)
- MessagePack: 3650 ns/op 2472 B/op 17 allocs/op
- JSON Deserialize: 19254 ns/op 3128 B/op 93 allocs/op
- Bogo Deserialize: 4341 ns/op 4464 B/op 101 allocs/op (4.43x faster)
- MessagePack: 7888 ns/op 2921 B/op 86 allocs/op
List Data (large lists):
- JSON Serialize: 23173 ns/op 3889 B/op 15 allocs/op
- Bogo Serialize: 54654 ns/op 41025 B/op 1040 allocs/op (0.42x speed)
- MessagePack: 10520 ns/op 8178 B/op 8 allocs/op
- JSON Deserialize: 56072 ns/op 21624 B/op 647 allocs/op
- Bogo Deserialize: 5822 ns/op 6016 B/op 119 allocs/op (9.63x faster)
- MessagePack: 18743 ns/op 11027 B/op 416 allocs/op
Binary Data (large byte lists):
- JSON Serialize: 12452 ns/op 16904 B/op 17 allocs/op
- Bogo Serialize: 12936 ns/op 64081 B/op 28 allocs/op (0.96x speed)
- MessagePack: 3876 ns/op 16256 B/op 5 allocs/op
- JSON Deserialize: 95240 ns/op 16248 B/op 32 allocs/op
- Bogo Deserialize: 961 ns/op 872 B/op 17 allocs/op (99.1x faster)
- MessagePack: 3346 ns/op 12292 B/op 21 allocs/op

Advanced Configuration

Decoder Options

decoder:=bogo.NewConfigurableDecoder(
bogo.WithDecoderMaxDepth(50), // Limit nesting depthbogo.WithDecoderStrictMode(true), // Enable strict validationbogo.WithMaxObjectSize(1024*1024), // 1MB size limitbogo.WithUTF8Validation(true), // Validate UTF-8 stringsbogo.WithUnknownTypes(false), // Reject unknown typesbogo.WithSelectiveFields([]string{"target_field"}), // Optimize for specific fields
)
result, err:=decoder.Decode(data)

Field-Specific Optimization

Bogo includes field-specific decoding optimization that provides up to 334x performance improvement when you only need specific fields from large objects.

Performance Results (Apple M2, 8 cores)

BEFORE OPTIMIZATION:
BenchmarkFieldDecoding_Full-8 17913 67037 ns/op 48057 B/op 1138 allocs/op
BenchmarkFieldDecoding_Selective-8 37464 41110 ns/op 31456 B/op 826 allocs/op
AFTER OPTIMIZATION:
BenchmarkFieldDecoding_WithOptimization-8 2121243 524.6 ns/op 424 B/op 7 allocs/op

Performance Improvements

• 128x faster than full decoding (67,037ns → 524ns)
• 78x faster than selective decoding without optimization (41,110ns → 524ns)
• 163x fewer allocations than full decoding (1,138 → 7)
• 118x fewer allocations than selective decoding (826 → 7)
• 113x less memory usage than full decoding (48,057B → 424B)
• 74x less memory usage than selective decoding (31,456B → 424B)

Usage Examples

Method 1: Explicit Field Selection

// Only decode "id" and "name" fields from a complex objectoptimizedDecoder:=bogo.NewConfigurableDecoder(
bogo.WithSelectiveFields([]string{"id", "name"}),
)
result, err:=optimizedDecoder.Decode(largeObjectData)
// 334x faster than decoding the entire object!

Method 2: Automatic Optimization with Struct Tags

// Define a struct with only the fields you needtypeUserSummarystruct {
IDint64`json:"id"`Namestring`json:"name"`// Large fields like "profile_data", "history", etc. are automatically skipped
}
varsummaryUserSummaryerr:=bogo.Unmarshal(complexUserData, &summary)
// Automatically optimized - only decodes the fields present in the struct!

How It Works

The optimization uses field jumping to skip over unwanted data:

1. Field Detection: The decoder identifies which fields are needed
2. Smart Skipping: Large, complex fields are skipped entirely using size information
3. Direct Access: Only target fields are parsed and decoded
4. Memory Efficiency: Unused data never allocates memory

Ideal Use Cases

• Large API responses where you only need specific fields
• Database records with large blob fields you want to skip
• Microservices extracting metadata from complex payloads
• Log processing where you need specific fields from large log entries
• Real-time systems requiring ultra-low latency field extraction

Encoder Options

encoder:=bogo.NewConfigurableEncoder(
bogo.WithValidation(true), // Enable validationbogo.WithCompression(false), // Disable compression
)
data, err:=encoder.Encode(value)

Binary Format

For complete technical details about the binary format, encoding algorithms, type specifications, and implementation notes, see the Binary Format Specification.

API Reference

Core Functions

// Marshal encodes a value to bogo binary formatfuncMarshal(vinterface{}) ([]byte, error)
// Unmarshal decodes bogo binary data into a valuefuncUnmarshal(data []byte, vinterface{}) error

Streaming API

// NewEncoder creates a streaming encoderfuncNewEncoder(w io.Writer) *StreamEncoder// NewDecoder creates a streaming decoder funcNewDecoder(r io.Reader) *StreamDecoder

Configuration

// NewConfigurableEncoder creates an encoder with optionsfuncNewConfigurableEncoder(options...EncoderOption) *Encoder// NewConfigurableDecoder creates a decoder with optionsfuncNewConfigurableDecoder(options...DecoderOption) *Decoder

Zero Values vs Nil Values

Bogo makes a clear distinction between zero values and nil values for robust data handling:

Zero Values (Type-Safe)

Zero values are preserved with their type information:

• "" (empty string) → encodes as TypeString with 0 length → decodes as ""
• 0 → encodes as TypeInt → decodes as int64(0)
• false → encodes as TypeBoolFalse → decodes as false
• []any{} → encodes as TypeUntypedList with 0 elements → decodes as []any{}
• time.Time{} → encodes as TypeTimestamp → decodes as zero time

Nil Values (Null)

Nil values are encoded uniformly as null:

• nil → encodes as TypeNull → decodes as nil
• (*string)(nil) → encodes as TypeNull → decodes as nil
• map[string]any(nil) → encodes as TypeNull → decodes as nil

Example Usage

data:=map[string]any{
"zero_string": "", // Will remain empty string after decode"zero_number": 0, // Will remain 0 after decode "zero_bool": false, // Will remain false after decode"nil_value": nil, // Will remain nil after decode
}
encoded, _:=bogo.Marshal(data)
vardecodedmap[string]anybogo.Unmarshal(encoded, &decoded)
fmt.Println(decoded["zero_string"]) // "" (empty string, not nil)fmt.Println(decoded["zero_number"]) // 0 (zero, not nil)fmt.Println(decoded["zero_bool"]) // false (not nil)fmt.Println(decoded["nil_value"]) // nil

Testing

Run the test suite:

go test ./...

Run benchmarks:

go test -bench=. -benchmem

Contributing

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -am 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.