基于Golang实现Redis协议解析器
作者：CSGOPHER 发布时间：2024-04-27 15:37:41
标签：Golang,Redis,协议,解析
本文实现Redis的协议层，协议层负责解析指令，然后将指令交给核心database执行
echo database用来测试协议层的代码
https://github.com/csgopher/go-redis
RESP协议
RESP是客户端与服务端通信的协议，格式有五种：
正常回复：以“+”开头，以“\r\n”结尾的字符串形式
错误回复：以“-”开头，以“\r\n”结尾的字符串形式
整数：以“:”开头，以“\r\n”结尾的字符串形式
多行字符串：以“$”开头，后跟实际发送字节数，再以“\r\n”开头和结尾
$3\r\nabc\r\n
数组：以“*”开头，后跟成员个数
SET key value
*3\r\n$3\r\nSET\r\n$3\r\nkey\r\n$5\r\nvalue\r\n
客户端和服务器发送的命令或数据一律以 \r\n （CRLF）作为换行符。
当我们输入*3\r\n$3\r\nSET\r\n$3\r\nkey\r\n$5\r\nvalue\r\n这样一串命令，服务端接收到的是如下的命令：
*3\r\n
$3\r\n
SET\r\n
$3\r\n
key\r\n
$5\r\n
value\r\n
interface/resp/conn.go
type Connection interface {
Write([]byte) error
GetDBIndex() int
SelectDB(int)
}
interface/resp/reply.go
type Reply interface {
ToBytes() []byte
}
Connection接口：Redis客户端的一个连接
Write：给客户端回复消息
GetDBIndex：Redis有16个DB
Reply接口：响应接口
resp/reply/consts.go
type PongReply struct{}
var pongBytes = []byte("+PONG\r\n")
func (r *PongReply) ToBytes() []byte {
return pongBytes
}
var thePongReply = new(PongReply)
func MakePongReply() *PongReply {
return thePongReply
}
type OkReply struct{}
var okBytes = []byte("+OK\r\n")
func (r *OkReply) ToBytes() []byte {
return okBytes
}
var theOkReply = new(OkReply)
func MakeOkReply() *OkReply {
return theOkReply
}
var nullBulkBytes = []byte("$-1\r\n")
type NullBulkReply struct{}
func (r *NullBulkReply) ToBytes() []byte {
return nullBulkBytes
}
func MakeNullBulkReply() *NullBulkReply {
return &NullBulkReply{}
}
var emptyMultiBulkBytes = []byte("*0\r\n")
type EmptyMultiBulkReply struct{}
func (r *EmptyMultiBulkReply) ToBytes() []byte {
return emptyMultiBulkBytes
}
type NoReply struct{}
var noBytes = []byte("")
func (r *NoReply) ToBytes() []byte {
return noBytes
}
定义五种回复：回复pong，ok，null，空数组，空
resp/reply/reply.go
type ErrorReply interface {
Error() string
ToBytes() []byte
}
ErrorReply：定义错误接口
resp/reply/errors.go
type UnknownErrReply struct{}
var unknownErrBytes = []byte("-Err unknown\r\n")
func (r *UnknownErrReply) ToBytes() []byte {
return unknownErrBytes
}
func (r *UnknownErrReply) Error() string {
return "Err unknown"
}
type ArgNumErrReply struct {
Cmd string
}
func (r *ArgNumErrReply) ToBytes() []byte {
return []byte("-ERR wrong number of arguments for '" + r.Cmd + "' command\r\n")
}
func (r *ArgNumErrReply) Error() string {
return "ERR wrong number of arguments for '" + r.Cmd + "' command"
}
func MakeArgNumErrReply(cmd string) *ArgNumErrReply {
return &ArgNumErrReply{
Cmd: cmd,
}
}
type SyntaxErrReply struct{}
var syntaxErrBytes = []byte("-Err syntax error\r\n")
var theSyntaxErrReply = &SyntaxErrReply{}
func MakeSyntaxErrReply() *SyntaxErrReply {
return theSyntaxErrReply
}
func (r *SyntaxErrReply) ToBytes() []byte {
return syntaxErrBytes
}
func (r *SyntaxErrReply) Error() string {
return "Err syntax error"
}
type WrongTypeErrReply struct{}
var wrongTypeErrBytes = []byte("-WRONGTYPE Operation against a key holding the wrong kind of value\r\n")
func (r *WrongTypeErrReply) ToBytes() []byte {
return wrongTypeErrBytes
}
func (r *WrongTypeErrReply) Error() string {
return "WRONGTYPE Operation against a key holding the wrong kind of value"
}
type ProtocolErrReply struct {
Msg string
}
func (r *ProtocolErrReply) ToBytes() []byte {
return []byte("-ERR Protocol error: '" + r.Msg + "'\r\n")
}
func (r *ProtocolErrReply) Error() string {
return "ERR Protocol error: '" + r.Msg
}
errors定义5种错误：UnknownErrReply 未知错误，ArgNumErrReply 参数个数错误，SyntaxErrReply 语法错误，WrongTypeErrReply 数据类型错误，ProtocolErrReply 协议错误
resp/reply/reply.go
var (
nullBulkReplyBytes = []byte("$-1")
// 协议的结尾
CRLF = "\r\n"
)
type BulkReply struct {
Arg []byte
}
func MakeBulkReply(arg []byte) *BulkReply {
return &BulkReply{
Arg: arg,
}
}
func (r *BulkReply) ToBytes() []byte {
if len(r.Arg) == 0 {
return nullBulkReplyBytes
}
return []byte("$" + strconv.Itoa(len(r.Arg)) + CRLF + string(r.Arg) + CRLF)
}
type MultiBulkReply struct {
Args [][]byte
}
func (r *MultiBulkReply) ToBytes() []byte {
argLen := len(r.Args)
var buf bytes.Buffer
buf.WriteString("*" + strconv.Itoa(argLen) + CRLF)
for _, arg := range r.Args {
if arg == nil {
buf.WriteString("$-1" + CRLF)
} else {
buf.WriteString("$" + strconv.Itoa(len(arg)) + CRLF + string(arg) + CRLF)
}
}
return buf.Bytes()
}
func MakeMultiBulkReply(args [][]byte) *MultiBulkReply {
return &MultiBulkReply{
Args: args,
}
}
type StatusReply struct {
Status string
}
func MakeStatusReply(status string) *StatusReply {
return &StatusReply{
Status: status,
}
}
func (r *StatusReply) ToBytes() []byte {
return []byte("+" + r.Status + CRLF)
}
type IntReply struct {
Code int64
}
func MakeIntReply(code int64) *IntReply {
return &IntReply{
Code: code,
}
}
func (r *IntReply) ToBytes() []byte {
return []byte(":" + strconv.FormatInt(r.Code, 10) + CRLF)
}
type StandardErrReply struct {
Status string
}
func (r *StandardErrReply) ToBytes() []byte {
return []byte("-" + r.Status + CRLF)
}
func (r *StandardErrReply) Error() string {
return r.Status
}
func MakeErrReply(status string) *StandardErrReply {
return &StandardErrReply{
Status: status,
}
}
func IsErrorReply(reply resp.Reply) bool {
return reply.ToBytes()[0] == '-'
}
BulkReply：回复一个字符串
MultiBulkReply：回复字符串数组
StatusReply：状态回复
IntReply：数字回复
StandardErrReply：标准错误回复
IsErrorReply：判断是否为错误回复
ToBytes：将字符串转成RESP协议规定的格式
resp/parser/parser.go
type Payload struct {
Data resp.Reply
Err error
}
type readState struct {
readingMultiLine bool
expectedArgsCount int
msgType byte
args [][]byte
bulkLen int64
}
func (s *readState) finished() bool {
return s.expectedArgsCount > 0 && len(s.args) == s.expectedArgsCount
}
func ParseStream(reader io.Reader) <-chan *Payload {
ch := make(chan *Payload)
go parse0(reader, ch)
return ch
}
func parse0(reader io.Reader, ch chan<- *Payload) {
......
}
Payload结构体：客服端给我们发的数据
Reply：客户端与服务端互相发的数据都称为Reply
readState结构体：
readingMultiLine：解析单行还是多行数据
expectedArgsCount：应该读取的参数个数
msgType：消息类型
args：消息内容
bulkLen：数据长度
finished方法：判断解析是否完成
ParseStream方法：异步解析数据后放入管道，返回管道数据
func readLine(bufReader *bufio.Reader, state *readState) ([]byte, bool, error) {
var msg []byte
var err error
if state.bulkLen == 0 {
msg, err = bufReader.ReadBytes('\n')
if err != nil {
return nil, true, err
}
if len(msg) == 0 || msg[len(msg)-2] != '\r' {
return nil, false, errors.New("protocol error: " + string(msg))
}
} else {
msg = make([]byte, state.bulkLen+2)
_, err = io.ReadFull(bufReader, msg)
if err != nil {
return nil, true, err
}
if len(msg) == 0 || msg[len(msg)-2] != '\r' || msg[len(msg)-1] != '\n' {
return nil, false, errors.New("protocol error: " + string(msg))
}
state.bulkLen = 0
}
return msg, false, nil
}
readLine：一行一行的读取。读正常的行，以\n分隔。读正文中包含\r\n字符的行时，state.bulkLen加上换行符\r\n（state.bulkLen+2）
func parseMultiBulkHeader(msg []byte, state *readState) error {
var err error
var expectedLine uint64
expectedLine, err = strconv.ParseUint(string(msg[1:len(msg)-2]), 10, 32)
if err != nil {
return errors.New("protocol error: " + string(msg))
}
if expectedLine == 0 {
state.expectedArgsCount = 0
return nil
} else if expectedLine > 0 {
state.msgType = msg[0]
state.readingMultiLine = true
state.expectedArgsCount = int(expectedLine)
state.args = make([][]byte, 0, expectedLine)
return nil
} else {
return errors.New("protocol error: " + string(msg))
}
}
func parseBulkHeader(msg []byte, state *readState) error {
var err error
state.bulkLen, err = strconv.ParseInt(string(msg[1:len(msg)-2]), 10, 64)
if err != nil {
return errors.New("protocol error: " + string(msg))
}
if state.bulkLen == -1 { // null bulk
return nil
} else if state.bulkLen > 0 {
state.msgType = msg[0]
state.readingMultiLine = true
state.expectedArgsCount = 1
state.args = make([][]byte, 0, 1)
return nil
} else {
return errors.New("protocol error: " + string(msg))
}
}
parseMultiBulkHeader：解析数组的头部，设置期望的行数和相关参数。
parseBulkHeader：解析多行字符串的头部。
func parseSingleLineReply(msg []byte) (resp.Reply, error) {
str := strings.TrimSuffix(string(msg), "\r\n")
var result resp.Reply
switch msg[0] {
case '+': // status reply
result = reply.MakeStatusReply(str[1:])
case '-': // err reply
result = reply.MakeErrReply(str[1:])
case ':': // int reply
val, err := strconv.ParseInt(str[1:], 10, 64)
if err != nil {
return nil, errors.New("protocol error: " + string(msg))
}
result = reply.MakeIntReply(val)
}
return result, nil
}
func readBody(msg []byte, state *readState) error {
line := msg[0 : len(msg)-2]
var err error
if line[0] == '$' {
// bulk reply
state.bulkLen, err = strconv.ParseInt(string(line[1:]), 10, 64)
if err != nil {
return errors.New("protocol error: " + string(msg))
}
if state.bulkLen <= 0 { // null bulk in multi bulks
state.args = append(state.args, []byte{})
state.bulkLen = 0
}
} else {
state.args = append(state.args, line)
}
return nil
}
parseSingleLineReply：解析单行命令
readBody：读取多行的命令，如果是$开头，设置bulkLen，读取下一行时根据这个+2，不是$开头则直接添加到args
func parse0(reader io.Reader, ch chan<- *Payload) {
defer func() {
if err := recover(); err != nil {
logger.Error(string(debug.Stack()))
}
}()
bufReader := bufio.NewReader(reader)
var state readState
var err error
var msg []byte
for {
var ioErr bool
msg, ioErr, err = readLine(bufReader, &state)
if err != nil {
if ioErr {
ch <- &Payload{
Err: err,
}
close(ch)
return
}
ch <- &Payload{
Err: err,
}
state = readState{}
continue
}
if !state.readingMultiLine {
if msg[0] == '*' {
// multi bulk reply
err = parseMultiBulkHeader(msg, &state)
if err != nil {
ch <- &Payload{
Err: errors.New("protocol error: " + string(msg)),
}
state = readState{}
continue
}
if state.expectedArgsCount == 0 {
ch <- &Payload{
Data: &reply.EmptyMultiBulkReply{},
}
state = readState{}
continue
}
} else if msg[0] == '$' { // bulk reply
err = parseBulkHeader(msg, &state)
if err != nil {
ch <- &Payload{
Err: errors.New("protocol error: " + string(msg)),
}
state = readState{} // reset state
continue
}
if state.bulkLen == -1 { // null bulk reply
ch <- &Payload{
Data: &reply.NullBulkReply{},
}
state = readState{} // reset state
continue
}
} else {
// single line reply
result, err := parseSingleLineReply(msg)
ch <- &Payload{
Data: result,
Err: err,
}
state = readState{} // reset state
continue
}
} else {
// read bulk reply
err = readBody(msg, &state)
if err != nil {
ch <- &Payload{
Err: errors.New("protocol error: " + string(msg)),
}
state = readState{} // reset state
continue
}
// if sending finished
if state.finished() {
var result resp.Reply
if state.msgType == '*' {
result = reply.MakeMultiBulkReply(state.args)
} else if state.msgType == '$' {
result = reply.MakeBulkReply(state.args[0])
}
ch <- &Payload{
Data: result,
Err: err,
}
state = readState{}
}
}
}
}
parse0：解析指令，解析完成后通过channel发出去
resp/connection/conn.go
type Connection struct {
conn net.Conn
waitingReply wait.Wait
mu sync.Mutex // 避免多个协程往客户端中写
selectedDB int
}
func NewConn(conn net.Conn) *Connection {
return &Connection{
conn: conn,
}
}
func (c *Connection) RemoteAddr() net.Addr {
return c.conn.RemoteAddr()
}
func (c *Connection) Close() error {
c.waitingReply.WaitWithTimeout(10 * time.Second)
_ = c.conn.Close()
return nil
}
func (c *Connection) Write(b []byte) error {
if len(b) == 0 {
return nil
}
c.mu.Lock()
c.waitingReply.Add(1)
defer func() {
c.waitingReply.Done()
c.mu.Unlock()
}()
_, err := c.conn.Write(b)
return err
}
func (c *Connection) GetDBIndex() int {
return c.selectedDB
}
func (c *Connection) SelectDB(dbNum int) {
c.selectedDB = dbNum
}
之前写的EchoHandler是用户传过来什么，我们传回去什么。现在要写一个RespHandler来代替EchoHandler，让解析器来解析。RespHandler中要有一个管理客户端连接的结构体Connection。
Connection：客户端连接，在协议层的handler中会用到
resp/handler/handler.go
var (
unknownErrReplyBytes = []byte("-ERR unknown\r\n")
)
type RespHandler struct {
activeConn sync.Map
db databaseface.Database
closing atomic.Boolean
}
func MakeHandler() *RespHandler {
var db databaseface.Database
db = database.NewEchoDatabase()
return &RespHandler{
db: db,
}
}
func (h *RespHandler) closeClient(client *connection.Connection) {
_ = client.Close()
h.db.AfterClientClose(client)
h.activeConn.Delete(client)
}
func (h *RespHandler) Handle(ctx context.Context, conn net.Conn) {
if h.closing.Get() {
// closing handler refuse new connection
_ = conn.Close()
}
client := connection.NewConn(conn)
h.activeConn.Store(client, 1)
ch := parser.ParseStream(conn)
for payload := range ch {
if payload.Err != nil {
if payload.Err == io.EOF ||
payload.Err == io.ErrUnexpectedEOF ||
strings.Contains(payload.Err.Error(), "use of closed network connection") {
// connection closed
h.closeClient(client)
logger.Info("connection closed: " + client.RemoteAddr().String())
return
}
// protocol err
errReply := reply.MakeErrReply(payload.Err.Error())
err := client.Write(errReply.ToBytes())
if err != nil {
h.closeClient(client)
logger.Info("connection closed: " + client.RemoteAddr().String())
return
}
continue
}
if payload.Data == nil {
logger.Error("empty payload")
continue
}
r, ok := payload.Data.(*reply.MultiBulkReply)
if !ok {
logger.Error("require multi bulk reply")
continue
}
result := h.db.Exec(client, r.Args)
if result != nil {
_ = client.Write(result.ToBytes())
} else {
_ = client.Write(unknownErrReplyBytes)
}
}
}
func (h *RespHandler) Close() error {
logger.Info("handler shutting down...")
h.closing.Set(true)
// TODO: concurrent wait
h.activeConn.Range(func(key interface{}, val interface{}) bool {
client := key.(*connection.Connection)
_ = client.Close()
return true
})
h.db.Close()
return nil
}
RespHandler：和之前的echo类似，加了核心层的db.exec执行解析的指令
interface/database/database.go
type CmdLine = [][]byte
type Database interface {
Exec(client resp.Connection, args [][]byte) resp.Reply
AfterClientClose(c resp.Connection)
Close()
}
type DataEntity struct {
Data interface{}
}
Exec：核心层的执行
AfterClientClose：关闭之后的善后方法
CmdLine：二维字节数组的指令别名
DataEntity：表示Redis的数据，包括string, list, set等等
database/echo_database.go
type EchoDatabase struct {
}
func NewEchoDatabase() *EchoDatabase {
return &EchoDatabase{}
}
func (e EchoDatabase) Exec(client resp.Connection, args [][]byte) resp.Reply {
return reply.MakeMultiBulkReply(args)
}
func (e EchoDatabase) AfterClientClose(c resp.Connection) {
logger.Info("EchoDatabase AfterClientClose")
}
func (e EchoDatabase) Close() {
logger.Info("EchoDatabase Close")
}
echo_database：测试协议层
Exec：指令解析后，再使用MakeMultiBulkReply包装一下返回去
main.go
err := tcp.ListenAndServeWithSignal(
&tcp.Config{
Address: fmt.Sprintf("％s:％d",
config.Properties.Bind,
config.Properties.Port),
},
handler.MakeHandler())
if err != nil {
logger.Error(err)
}
main改成刚才写的：handler.MakeHandler()
来源：https://www.cnblogs.com/csgopher/p/17249305.html
0
投稿猜你喜欢
- 两种情况：1.带索引 2.不带索引前提介绍：方式：采用命令行的方式来模拟1.mysq由于默认是开启自动提交事务，所以首先得查看自己当前的数据
- 我们在.NET程序的开发过程中，常常需要和用户进行信息交互，比如执行某项操作是否成功，“确定”还是“取消”，以及选择“确定”或“取消”后是否
- 我就废话不多说了，大家还是直接看代码吧~'''Created on 2018-4-16'''
- Embedding词嵌入在 pytorch 中非常简单，只需要调用 torch.nn.Embedding(m, n) 就可以了，m 表示单词
介绍众所周知，计算机视觉在机器学习和人工智能领域获得了巨大的普及。图像识别技术允许计算机处理比人眼更多的信息，通常更快、更准确，或者只是在人- 版本链在InnoDB引擎表中，他们的聚簇索引记录中有两个隐藏列：trx_id：用来存储对数据进行修改时的事务idroll_pointer：每
- 一、Pyeharts简介pyecharts 是一个用于生成 Echarts 图表的类库。用 Echarts 生成的图可视化效果很不错，pye
- 1.下载Navicat for MySQL 15https://www.navicat.com.cn/download/navicat-fo
- 单线程实现多个定时器NewTimer.py#!/usr/bin/env pythonfrom heapq import *from thre
- 美化主要表现在鼠标放到菜单上后（即鼠标悬停）的效果，这里首先介绍几个经常用到的CSS属性：backgr
- 本文和大家重点讨论一下Perl哈希表的概念，Perl语言和其他编程语言各有各的特点，这里和大家分享一下Perl哈希表的概念，其实Perl哈希
- 第一点：找Python安装目录方法一：方法二：输入import sysprint（sys.path）化黑线处第二点：找到安装目录后就可以开始
- 使用memcache来同步session是还是不错的，当然也可以通过redis来保存session，可以php开启并将Session存储到R
- # -*- coding: utf-8 -*- #mysqldb &nb
- 本文实例讲述了python中readline判断文件读取结束的方法。分享给大家供大家参考。具体分析如下：大家知道，python中按行读取文件
- 你可能在使用MySQL过程中，各种意外导致数据库表的损坏，而且这些数据往往是最新的数据，通常不可能在备份数据中找到。本章将讲述如何检测MyS
- 1.创建虚拟环境首先创建一个新文件夹在PyCharm终端中切换到这个文件夹，输入 python -m venv 环境名 创建虚拟环境&nbs
- 1、demo第一个代码是多线程的简单使用，编写了线程如何执行函数和类。import threadingimport timeclass Cl
- 一、Pandoc转换1.1 问题由于我们markdown编辑器比较特殊，一般情况下，我们不太好看，如果转换成pdf的话，我们就不需要可以的去
- 需求背景用python进行文件的创建和读写操作时，我们很少关注所创建的文件的权限配置。对于一些安全性较高的系统，如果我们创建的文件权限其他用
