README

thrifter

decode/encode thrift message without IDL

Why?

  • because IDL generated model is ugly and inflexible, it is seldom used in application directly. instead we define another model, which leads to bad performance.
    • bytes need to be copied twice
    • more objects to gc
  • thrift proxy can not know all possible IDL in advance, in scenarios like api gateway, we need to decode/encode in a generic way to modify embedded header.
  • official thrift library for go is slow, verified in several benchmarks. It is even slower than json-iterator

works like encoding/json

encoding/json has a super simple api to encode/decode json. thrifter mimic the same api.

import "github.com/thrift-iterator/go"
// marshal to thrift
thriftEncodedBytes, err := thrifter.Marshal([]int{1, 2, 3})
// unmarshal back
var val []int
err = thrifter.Unmarshal(thriftEncodedBytes, &val)

even struct data binding is supported

import "github.com/thrift-iterator/go"

type NewOrderRequest struct {
    Lines []NewOrderLine `thrift:",1"`
}

type NewOrderLine struct {
    ProductId string `thrift:",1"`
    Quantity int `thrift:",2"`
}

// marshal to thrift
thriftEncodedBytes, err := thrifter.Marshal(NewOrderRequest{
	Lines: []NewOrderLine{
		{"apple", 1},
		{"orange", 2},
	}
})
// unmarshal back
var val NewOrderRequest
err = thrifter.Unmarshal(thriftEncodedBytes, &val)

without IDL

you do not need to define IDL. you do not need to use static code generation. you do not event need to define struct.

import "github.com/thrift-iterator/go"
import "github.com/thrift-iterator/go/general"

var msg general.Message
err := thrifter.Unmarshal(thriftEncodedBytes, &msg)
// the RPC call method name, type is string
fmt.Println(msg.MessageName)
// the RPC call arguments, type is general.Struct
fmt.Println(msg.Arguments)

what is general.Struct, it is defined as a map

type FieldId int16
type Struct map[FieldId]interface{}

we can extract out specific argument from deeply nested arguments using one line

productId := msg.MessageArgs.Get(
	protocol.FieldId(1), // lines of request
	0, // the first line
	protocol.FieldId(1), // product id
).(string)

You can unmarshal any thrift bytes into general objects. And you can marshal them back.

Partial decoding

fully decoding into a go struct consumes substantial resources. thrifter provide option to do partial decoding. You can modify part of the message, with untouched parts in []byte form.

import "github.com/thrift-iterator/go"
import "github.com/thrift-iterator/go/protocol"
import "github.com/thrift-iterator/go/raw"

// partial decoding
decoder := thrifter.NewDecoder(reader)
var msgHeader protocol.MessageHeader
decoder.Decode(&msgHeader)
var msgArgs raw.Struct
decoder.Decode(&msgArgs)

// modify...

// encode back
encoder := thrifter.NewEncoder(writer)
encoder.Encode(msgHeader)
encoder.Encode(msgArgs)

the definition of raw.Struct is

type StructField struct {
	Buffer []byte
	Type protocol.TType
}

type Struct map[protocol.FieldId]StructField

Performance

thrifter does not compromise performance.

gogoprotobuf

5000000	       366 ns/op	     144 B/op	      12 allocs/op

thrift

1000000	      1549 ns/op	     528 B/op	       9 allocs/op

thrifter by static codegen

5000000	       389 ns/op	     192 B/op	       6 allocs/op

thrifter by reflection

2000000	       585 ns/op	     192 B/op	       6 allocs/op

You can see the reflection implementation is not bad, much faster than the static code generated by thrift original implementation.

To have best performance, you can choose to use static code generation. The api is unchanged, just need to add extra static codegen in your build steps, and include the generated code in your package. The runtime will automatically use the generated encoder/decoder instead of reflection.

For example of static codegen, checkout https://github.com/thrift-iterator/go/blob/master/test/api/init.go

Sync IDL and Go Struct

Keep IDL and your object model is challenging. We do not always like the code generated from thrift IDL. But manually keeping the IDL and model in sync is tedious and error prone.

A separate toolchain to manipulate thrift IDL file, and keeping them bidirectionally in sync will be provided in another project.

Documentation

Index

Constants

This section is empty.

Variables

View Source
var DefaultConfig = Config{Protocol: ProtocolBinary, StaticCodegen: false}.Froze()

Functions

func Marshal

func Marshal(obj interface{}) ([]byte, error)

func MarshalMessage

func MarshalMessage(msg general.Message) ([]byte, error)

    MarshalMessage is just a shortcut to demonstrate message decoded by UnmarshalMessage can be encoded back

    func NewIterator

    func NewIterator(reader io.Reader, buf []byte) spi.Iterator

    func NewStream

    func NewStream(writer io.Writer, buf []byte) spi.Stream

    func ToJSON

    func ToJSON(buf []byte) (string, error)

      ToJSON convert the thrift message to JSON string

      func Unmarshal

      func Unmarshal(buf []byte, obj interface{}) error

      func UnmarshalMessage

      func UnmarshalMessage(buf []byte) (general.Message, error)

        UnmarshalMessage demonstrate how to decode thrift binary without IDL into a general message struct

        Types

        type API

        type API interface {
        	// NewStream is low level streaming api
        	NewStream(writer io.Writer, buf []byte) spi.Stream
        	// NewIterator is low level streaming api
        	NewIterator(reader io.Reader, buf []byte) spi.Iterator
        	// Unmarshal from []byte
        	Unmarshal(buf []byte, obj interface{}) error
        	// UnmarshalMessage from []byte
        	UnmarshalMessage(buf []byte) (general.Message, error)
        	// Marshal to []byte
        	Marshal(obj interface{}) ([]byte, error)
        	// ToJSON convert thrift message to JSON string
        	ToJSON(buf []byte) (string, error)
        	// MarshalMessage to []byte
        	MarshalMessage(msg general.Message) ([]byte, error)
        	// NewDecoder to unmarshal from []byte or io.Reader
        	NewDecoder(reader io.Reader, buf []byte) *Decoder
        	// NewEncoder to marshal to io.Writer
        	NewEncoder(writer io.Writer) *Encoder
        	// WillDecodeFromBuffer should only be used in generic.Declare
        	WillDecodeFromBuffer(sample ...interface{})
        	// WillDecodeFromReader should only be used in generic.Declare
        	WillDecodeFromReader(sample ...interface{})
        	// WillEncode should only be used in generic.Declare
        	WillEncode(sample ...interface{})
        }

        type Config

        type Config struct {
        	Protocol      Protocol
        	StaticCodegen bool
        	Extensions    spi.Extensions
        }

        func (Config) AddExtension

        func (cfg Config) AddExtension(extension spi.Extension) Config

        func (Config) Froze

        func (cfg Config) Froze() API

        type Decoder

        type Decoder struct {
        	// contains filtered or unexported fields
        }

        func NewDecoder

        func NewDecoder(reader io.Reader, buf []byte) *Decoder

        func (*Decoder) Decode

        func (decoder *Decoder) Decode(val interface{}) error

        func (*Decoder) DecodeMessage

        func (decoder *Decoder) DecodeMessage() (general.Message, error)

        func (*Decoder) DecodeMessageArguments

        func (decoder *Decoder) DecodeMessageArguments() (general.Struct, error)

        func (*Decoder) DecodeMessageHeader

        func (decoder *Decoder) DecodeMessageHeader() (protocol.MessageHeader, error)

        func (*Decoder) Reset

        func (decoder *Decoder) Reset(reader io.Reader, buf []byte)

        type Encoder

        type Encoder struct {
        	// contains filtered or unexported fields
        }

        func NewEncoder

        func NewEncoder(writer io.Writer) *Encoder

        func (*Encoder) Buffer

        func (encoder *Encoder) Buffer() []byte

        func (*Encoder) Encode

        func (encoder *Encoder) Encode(val interface{}) error

        func (*Encoder) EncodeMessage

        func (encoder *Encoder) EncodeMessage(msg general.Message) error

        func (*Encoder) EncodeMessageArguments

        func (encoder *Encoder) EncodeMessageArguments(msgArgs general.Struct) error

        func (*Encoder) EncodeMessageHeader

        func (encoder *Encoder) EncodeMessageHeader(msgHeader protocol.MessageHeader) error

        func (*Encoder) Reset

        func (encoder *Encoder) Reset(writer io.Writer)

        type Protocol

        type Protocol int
        var ProtocolBinary Protocol = 1
        var ProtocolCompact Protocol = 2