來(lái)源：jianshu.com/p/c806003a8530

前言:

在實(shí)際的開發(fā)項(xiàng)目中,一個(gè)對(duì)外暴露的接口往往會(huì)面臨很多次請(qǐng)求，我們來(lái)解釋一下冪等的概念：任意多次執(zhí)行所產(chǎn)生的影響均與一次執(zhí)行的影響相同。按照這個(gè)含義，最終的含義就是 對(duì)數(shù)據(jù)庫(kù)的影響只能是一次性的，不能重復(fù)處理。如何保證其冪等性，通常有以下手段：

1. 數(shù)據(jù)庫(kù)建立唯一性索引，可以保證最終插入數(shù)據(jù)庫(kù)的只有一條數(shù)據(jù)
2. token機(jī)制，每次接口請(qǐng)求前先獲取一個(gè)token，然后再下次請(qǐng)求的時(shí)候在請(qǐng)求的header體中加上這個(gè)token，后臺(tái)進(jìn)行驗(yàn)證，如果驗(yàn)證通過(guò)刪除token，下次請(qǐng)求再次判斷token
3. 悲觀鎖或者樂(lè)觀鎖，悲觀鎖可以保證每次for update的時(shí)候其他sql無(wú)法update數(shù)據(jù)(在數(shù)據(jù)庫(kù)引擎是innodb的時(shí)候,select的條件必須是唯一索引,防止鎖全表)
4. 先查詢后判斷，首先通過(guò)查詢數(shù)據(jù)庫(kù)是否存在數(shù)據(jù)，如果存在證明已經(jīng)請(qǐng)求過(guò)了，直接拒絕該請(qǐng)求，如果沒(méi)有存在，就證明是第一次進(jìn)來(lái)，直接放行。

redis實(shí)現(xiàn)自動(dòng)冪等的原理圖：

一：搭建redis的服務(wù)Api

1:首先是搭建redis服務(wù)器。

2:引入springboot中到的redis的stater，或者Spring封裝的jedis也可以，后面主要用到的api就是它的set方法和exists方法,這里我們使用springboot的封裝好的redisTemplate

/**
?*?redis工具類
?*/
@Component
public?class?RedisService?{

????@Autowired
????private?RedisTemplate?redisTemplate;

????/**
?????*?寫入緩存
?????*?@param?key
?????*?@param?value
?????*?@return
?????*/
????public?boolean?set(final?String?key,?Object?value)?{
????????boolean?result?=?false;
????????try?{
????????????ValueOperationsObject>?operations?=?redisTemplate.opsForValue();
????????????operations.set(key,?value);
????????????result?=?true;
????????}?catch?(Exception?e)?{
????????????e.printStackTrace();
????????}
????????return?result;
????}


????/**
?????*?寫入緩存設(shè)置時(shí)效時(shí)間
?????*?@param?key
?????*?@param?value
?????*?@return
?????*/
????public?boolean?setEx(final?String?key,?Object?value,?Long?expireTime)?{
????????boolean?result?=?false;
????????try?{
????????????ValueOperationsObject>?operations?=?redisTemplate.opsForValue();
????????????operations.set(key,?value);
????????????redisTemplate.expire(key,?expireTime,?TimeUnit.SECONDS);
????????????result?=?true;
????????}?catch?(Exception?e)?{
????????????e.printStackTrace();
????????}
????????return?result;
????}


????/**
?????*?判斷緩存中是否有對(duì)應(yīng)的value
?????*?@param?key
?????*?@return
?????*/
????public?boolean?exists(final?String?key)?{
????????return?redisTemplate.hasKey(key);
????}

????/**
?????*?讀取緩存
?????*?@param?key
?????*?@return
?????*/
????public?Object?get(final?String?key)?{
????????Object?result?=?null;
????????ValueOperationsObject>?operations?=?redisTemplate.opsForValue();
????????result?=?operations.get(key);
????????return?result;
????}

????/**
?????*?刪除對(duì)應(yīng)的value
?????*?@param?key
?????*/
????public?boolean?remove(final?String?key)?{
????????if?(exists(key))?{
????????????Boolean?delete?=?redisTemplate.delete(key);
????????????return?delete;
????????}
????????return?false;

????}

}


二：自定義注解AutoIdempotent

自定義一個(gè)注解，定義此注解的主要目的是把它添加在需要實(shí)現(xiàn)冪等的方法上，凡是某個(gè)方法注解了它，都會(huì)實(shí)現(xiàn)自動(dòng)冪等。后臺(tái)利用反射如果掃描到這個(gè)注解，就會(huì)處理這個(gè)方法實(shí)現(xiàn)自動(dòng)冪等，使用元注解ElementType.METHOD表示它只能放在方法上，etentionPolicy.RUNTIME表示它在運(yùn)行時(shí)

@Target({ElementType.METHOD})
@Retention(RetentionPolicy.RUNTIME)
public?@interface?AutoIdempotent?{
??
}


三：token創(chuàng)建和檢驗(yàn)

1:token服務(wù)接口

我們新建一個(gè)接口，創(chuàng)建token服務(wù)，里面主要是兩個(gè)方法，一個(gè)用來(lái)創(chuàng)建token，一個(gè)用來(lái)驗(yàn)證token。創(chuàng)建token主要產(chǎn)生的是一個(gè)字符串，檢驗(yàn)token的話主要是傳達(dá)request對(duì)象，為什么要傳request對(duì)象呢？主要作用就是獲取header里面的token,然后檢驗(yàn)，通過(guò)拋出的Exception來(lái)獲取具體的報(bào)錯(cuò)信息返回給前端

public?interface?TokenService?{

????/**
?????*?創(chuàng)建token
?????*?@return
?????*/
????public??String?createToken();

????/**
?????*?檢驗(yàn)token
?????*?@param?request
?????*?@return
?????*/
????public?boolean?checkToken(HttpServletRequest?request)?throws?Exception;

}
2:token的服務(wù)實(shí)現(xiàn)類

token引用了redis服務(wù)，創(chuàng)建token采用隨機(jī)算法工具類生成隨機(jī)uuid字符串,然后放入到redis中(為了防止數(shù)據(jù)的冗余保留,這里設(shè)置過(guò)期時(shí)間為10000秒,具體可視業(yè)務(wù)而定)，如果放入成功，最后返回這個(gè)token值。checkToken方法就是從header中獲取token到值(如果header中拿不到，就從paramter中獲取)，如若不存在,直接拋出異常。這個(gè)異常信息可以被攔截器捕捉到，然后返回給前端。

@Service
public?class?TokenServiceImpl?implements?TokenService?{

????@Autowired
????private?RedisService?redisService;


????/**
?????*?創(chuàng)建token
?????*
?????*?@return
?????*/
????@Override
????public?String?createToken()?{
????????String?str?=?RandomUtil.randomUUID();
????????StrBuilder?token?=?new?StrBuilder();
????????try?{
????????????token.append(Constant.Redis.TOKEN_PREFIX).append(str);
????????????redisService.setEx(token.toString(),?token.toString(),10000L);
????????????boolean?notEmpty?=?StrUtil.isNotEmpty(token.toString());
????????????if?(notEmpty)?{
????????????????return?token.toString();
????????????}
????????}catch?(Exception?ex){
????????????ex.printStackTrace();
????????}
????????return?null;
????}


????/**
?????*?檢驗(yàn)token
?????*
?????*?@param?request
?????*?@return
?????*/
????@Override
????public?boolean?checkToken(HttpServletRequest?request)?throws?Exception?{

????????String?token?=?request.getHeader(Constant.TOKEN_NAME);
????????if?(StrUtil.isBlank(token))?{//?header中不存在token
????????????token?=?request.getParameter(Constant.TOKEN_NAME);
????????????if?(StrUtil.isBlank(token))?{//?parameter中也不存在token
????????????????throw?new?ServiceException(Constant.ResponseCode.ILLEGAL_ARGUMENT,?100);
????????????}
????????}

????????if?(!redisService.exists(token))?{
????????????throw?new?ServiceException(Constant.ResponseCode.REPETITIVE_OPERATION,?200);
????????}

????????boolean?remove?=?redisService.remove(token);
????????if?(!remove)?{
????????????throw?new?ServiceException(Constant.ResponseCode.REPETITIVE_OPERATION,?200);
????????}
????????return?true;
????}
}


四：攔截器的配置

1:web配置類，實(shí)現(xiàn)WebMvcConfigurerAdapter，主要作用就是添加autoIdempotentInterceptor到配置類中，這樣我們到攔截器才能生效，注意使用@Configuration注解，這樣在容器啟動(dòng)是時(shí)候就可以添加進(jìn)入context中

@Configuration
public?class?WebConfiguration?extends?WebMvcConfigurerAdapter?{

????@Resource
???private?AutoIdempotentInterceptor?autoIdempotentInterceptor;

????/**
?????*?添加攔截器
?????*?@param?registry
?????*/
????@Override
????public?void?addInterceptors(InterceptorRegistry?registry)?{
????????registry.addInterceptor(autoIdempotentInterceptor);
????????super.addInterceptors(registry);
????}
}
2:攔截處理器：主要的功能是攔截掃描到AutoIdempotent到注解到方法,然后調(diào)用tokenService的checkToken()方法校驗(yàn)token是否正確，如果捕捉到異常就將異常信息渲染成json返回給前端

/**
?*?攔截器
?*/
@Component
public?class?AutoIdempotentInterceptor?implements?HandlerInterceptor?{

????@Autowired
????private?TokenService?tokenService;

????/**
?????*?預(yù)處理
?????*
?????*?@param?request
?????*?@param?response
?????*?@param?handler
?????*?@return
?????*?@throws?Exception
?????*/
????@Override
????public?boolean?preHandle(HttpServletRequest?request,?HttpServletResponse?response,?Object?handler)?throws?Exception?{

????????if?(!(handler?instanceof?HandlerMethod))?{
????????????return?true;
????????}
????????HandlerMethod?handlerMethod?=?(HandlerMethod)?handler;
????????Method?method?=?handlerMethod.getMethod();
????????//被ApiIdempotment標(biāo)記的掃描
????????AutoIdempotent?methodAnnotation?=?method.getAnnotation(AutoIdempotent.class);
????????if?(methodAnnotation?!=?null)?{
????????????try?{
????????????????return?tokenService.checkToken(request);//?冪等性校驗(yàn),?校驗(yàn)通過(guò)則放行,?校驗(yàn)失敗則拋出異常,?并通過(guò)統(tǒng)一異常處理返回友好提示
????????????}catch?(Exception?ex){
????????????????ResultVo?failedResult?=?ResultVo.getFailedResult(101,?ex.getMessage());
????????????????writeReturnJson(response,?JSONUtil.toJsonStr(failedResult));
????????????????throw?ex;
????????????}
????????}
????????//必須返回true,否則會(huì)被攔截一切請(qǐng)求
????????return?true;
????}


????@Override
????public?void?postHandle(HttpServletRequest?request,?HttpServletResponse?response,?Object?handler,?ModelAndView?modelAndView)?throws?Exception?{

????}

????@Override
????public?void?afterCompletion(HttpServletRequest?request,?HttpServletResponse?response,?Object?handler,?Exception?ex)?throws?Exception?{

????}

????/**
?????*?返回的json值
?????*?@param?response
?????*?@param?json
?????*?@throws?Exception
?????*/
????private?void?writeReturnJson(HttpServletResponse?response,?String?json)?throws?Exception{
????????PrintWriter?writer?=?null;
????????response.setCharacterEncoding("UTF-8");
????????response.setContentType("text/html;?charset=utf-8");
????????try?{
????????????writer?=?response.getWriter();
????????????writer.print(json);

????????}?catch?(IOException?e)?{
????????}?finally?{
????????????if?(writer?!=?null)
????????????????writer.close();
????????}
????}

}


五：測(cè)試用例

1:模擬業(yè)務(wù)請(qǐng)求類

首先我們需要通過(guò)/get/token路徑通過(guò)getToken()方法去獲取具體的token，然后我們調(diào)用testIdempotence方法，這個(gè)方法上面注解了@AutoIdempotent，攔截器會(huì)攔截所有的請(qǐng)求，當(dāng)判斷到處理的方法上面有該注解的時(shí)候，就會(huì)調(diào)用TokenService中的checkToken()方法，如果捕獲到異常會(huì)將異常拋出調(diào)用者，下面我們來(lái)模擬請(qǐng)求一下：

@RestController
public?class?BusinessController?{


????@Resource
????private?TokenService?tokenService;

????@Resource
????private?TestService?testService;


????@PostMapping("/get/token")
????public?String??getToken(){
????????String?token?=?tokenService.createToken();
????????if?(StrUtil.isNotEmpty(token))?{
????????????ResultVo?resultVo?=?new?ResultVo();
????????????resultVo.setCode(Constant.code_success);
????????????resultVo.setMessage(Constant.SUCCESS);
????????????resultVo.setData(token);
????????????return?JSONUtil.toJsonStr(resultVo);
????????}
????????return?StrUtil.EMPTY;
????}


????@AutoIdempotent
????@PostMapping("/test/Idempotence")
????public?String?testIdempotence()?{
????????String?businessResult?=?testService.testIdempotence();
????????if?(StrUtil.isNotEmpty(businessResult))?{
????????????ResultVo?successResult?=?ResultVo.getSuccessResult(businessResult);
????????????return?JSONUtil.toJsonStr(successResult);
????????}
????????return?StrUtil.EMPTY;
????}
}
2:使用postman請(qǐng)求

首先訪問(wèn)get/token路徑獲取到具體到token：

利用獲取到到token,然后放到具體請(qǐng)求到header中,可以看到第一次請(qǐng)求成功，接著我們請(qǐng)求第二次：

第二次請(qǐng)求，返回到是重復(fù)性操作，可見(jiàn)重復(fù)性驗(yàn)證通過(guò)，再多次請(qǐng)求到時(shí)候我們只讓其第一次成功，第二次就是失?。?/p>

六：總結(jié)

本篇博客介紹了使用springboot和攔截器、redis來(lái)優(yōu)雅的實(shí)現(xiàn)接口冪等，對(duì)于冪等在實(shí)際的開發(fā)過(guò)程中是十分重要的，因?yàn)橐粋€(gè)接口可能會(huì)被無(wú)數(shù)的客戶端調(diào)用，如何保證其不影響后臺(tái)的業(yè)務(wù)處理，如何保證其只影響數(shù)據(jù)一次是非常重要的，它可以防止產(chǎn)生臟數(shù)據(jù)或者亂數(shù)據(jù)，也可以減少并發(fā)量，實(shí)乃十分有益的一件事。而傳統(tǒng)的做法是每次判斷數(shù)據(jù)，這種做法不夠智能化和自動(dòng)化，比較麻煩。而今天的這種自動(dòng)化處理也可以提升程序的伸縮性。