目錄

一、數(shù)據(jù)庫設(shè)計(jì)

1.1、數(shù)據(jù)庫選擇

1.2、環(huán)境配置

1.3、建庫建表接口實(shí)現(xiàn)

1.4、封裝數(shù)據(jù)庫操作

1.5、針對(duì) DataBaseManager 進(jìn)行單元測(cè)試

---

一、數(shù)據(jù)庫設(shè)計(jì)

---

1.1、數(shù)據(jù)庫選擇

MySQL 是我們最熟悉的數(shù)據(jù)庫，但是這里我們選擇使用 SQLite，原因如下：

1. SQLite 比 MySQL 更輕量：一個(gè)完整的 SQLite 數(shù)據(jù)庫，只有一個(gè)單獨(dú)的可執(zhí)行文件（不到 1M）.
2. SQLite 操作簡(jiǎn)便：SQLite 只是一個(gè)本地?cái)?shù)據(jù)庫，相當(dāng)于是直接操作本地的硬盤.
3. SQLite 應(yīng)用也非常廣泛：在一些性能不高的設(shè)備上，SQLite 是數(shù)據(jù)庫的首選，尤其是移動(dòng)端和嵌入式設(shè)備（Android 系統(tǒng)就是內(nèi)置的 SQLite）.

1.2、環(huán)境配置

在 java?中直接使用 maven 把 SQLite 依賴引入即可（版本自行考慮）~

        <!-- https://mvnrepository.com/artifact/org.xerial/sqlite-jdbc --><dependency><groupId>org.xerial</groupId><artifactId>sqlite-jdbc</artifactId><version>3.41.2.1</version></dependency>

配置如下

spring:datasource:url: jdbc:sqlite:./data/meta.dbusername:password:driver-class-name: org.sqlite.JDBC

url：SQLite 的工作路徑，用來存儲(chǔ)數(shù)據(jù)在某個(gè)指定的文件中.

username & password：對(duì)于 SQLite 來說，不需要使用 用戶名密碼.? MySQL 是一個(gè)客戶端服務(wù)器結(jié)構(gòu)的程序，而 SQLite 則不是客戶端服務(wù)器結(jié)構(gòu)的程序，只有本地主機(jī)能訪問.

Ps：SQLite 雖然和 MySQL 不太一樣，但是都可以通過 MyBatis 這樣的框架來使用.

1.3、建庫建表接口實(shí)現(xiàn)

存儲(chǔ)的數(shù)據(jù)就是：交換機(jī)、隊(duì)列、綁定.

這里我們使用 MyBatis 來完成相關(guān)的 CRUD.

mapper 接口中提供三個(gè)建庫建表操作和針對(duì)這三個(gè)庫表進(jìn)行 CRUD 的操作.

@Mapper
public interface MetaMapper {//三個(gè)核心建表方法void createExchangeTable();void createQueueTable();void createBindingTable();//基于上述三個(gè)表，進(jìn)行 插入、刪除、查詢 操作void insertExchange(Exchange exchange);List<Exchange> selectAllExchange();void deleteExchange(String exchangeName);void insertQueue(MSGQueue queue);List<MSGQueue> selectAllQueue();void deleteQueue(String queueName);void insertBinding(Binding binding);List<Binding> selectAllBinding();void deleteBinding(Binding binding);}

對(duì)應(yīng)的實(shí)現(xiàn)如下：

    <update id="createExchangeTable">create table if not exists exchange (name varchar(50) primary key,type int,durable boolean,autoDelete boolean,arguments varchar(1024));</update><update id="createQueueTable">create table if not exists queue (name varchar(50) primary key,durable boolean,exclusive boolean,autoDelete boolean,arguments varchar(1024));</update><update id="createBindingTable">create table if not exists binding (exchangeName varchar(50),queueName varchar(50),bindingKey varchar(256))</update><insert id="insertExchange" parameterType="com.example.rabbitmqproject.mqserver.core.Exchange">insert into exchange values (#{name}, #{type}, #{durable}, #{autoDelete}, #{arguments});</insert><select id="selectAllExchange" resultType="com.example.rabbitmqproject.mqserver.core.Exchange">select * from exchange;</select><delete id="deleteExchange" parameterType="com.example.rabbitmqproject.mqserver.core.Exchange">delete from exchange where name = #{name};</delete><insert id="insertQueue" parameterType="com.example.rabbitmqproject.mqserver.core.MSGQueue">insert into queue values(#{name}, #{durable}, #{exclusive}, #{autoDelete}, #{arguments});</insert><select id="selectAllQueue" resultType="com.example.rabbitmqproject.mqserver.core.MSGQueue">select * from queue;</select><delete id="deleteQueue">delete from queue where name = #{name};</delete><insert id="insertBinding">insert into binding values (#{exchangeName}, #{queueName}, #{bindingKey});</insert><select id="selectAllBinding" resultType="com.example.rabbitmqproject.mqserver.core.Binding">select * from binding;</select><delete id="deleteBinding">delete from binding where exchangeName = #{exchangeName} and queueName = #{queueName};</delete>

1.4、封裝數(shù)據(jù)庫操作

這里我們通過定制化 代碼 的方式來自動(dòng)完成建庫建表的操作（符合?RabbitMQ 中間件的設(shè)定）.

創(chuàng)建 DataBaseManager 類，來完成數(shù)據(jù)庫相關(guān)的操作，注意細(xì)節(jié)如下：

1. 初始化方法：一般談到初始化，都會(huì)用到 構(gòu)造方法，但是這里我們使用一個(gè) 普通的方法 —— init()；構(gòu)造方法一般是用來初始化類的屬性，不會(huì)涉及到太多的業(yè)務(wù)邏輯，而此處的初始化，帶有業(yè)務(wù)邏輯，還是單獨(dú)領(lǐng)出來，手動(dòng)來調(diào)用比較合適.
2. 建庫建表邏輯：這里期望，broker server 啟動(dòng)的時(shí)候做出如下邏輯判斷：
   1. 如果數(shù)據(jù)庫已經(jīng)存在（表存在），不做任何操作.
   2. 如果數(shù)據(jù)庫不存在，則建庫建表，構(gòu)造默認(rèn)數(shù)據(jù).

Ps：怎么判定數(shù)據(jù)庫存在或者不存在?就判定 meta.db 文件是否存在即可（配置文件中的 url）.

public class DataBaseManager {//這里不使用 Autowired 注解獲取，因?yàn)楫?dāng)前這個(gè)類需要我們后面手動(dòng)管理private MetaMapper metaMapper;//針對(duì)數(shù)據(jù)庫進(jìn)行初始化public void init() {//手動(dòng)獲取到 MetaMappermetaMapper = RabbitmqProjectApplication.context.getBean(MetaMapper.class);if(!checkDBExists()) {//數(shù)據(jù)庫不存在，就進(jìn)行建庫建表操作//先創(chuàng)建出目錄結(jié)構(gòu)(否則會(huì)報(bào)錯(cuò)：找不到目錄結(jié)構(gòu))File dataDir = new File("./data");dataDir.mkdirs();//創(chuàng)建數(shù)據(jù)庫createTable();//插入默認(rèn)數(shù)據(jù)createDefaultData();System.out.println("[DataBaseManager] 數(shù)據(jù)庫初始化完成!");} else {//數(shù)據(jù)庫存在，什么都不做即可System.out.println("[DataBaseManager] 數(shù)據(jù)庫已存在!");}}private boolean checkDBExists() {File file = new File("./data/meta.db");return file.exists();}private void createTable() {metaMapper.createExchangeTable();metaMapper.createQueueTable();metaMapper.createBindingTable();System.out.println("[DataBaseManager] 創(chuàng)建表完成！");}/*** 添加默認(rèn)交換機(jī)* RabbitMQ 有一個(gè)這樣的設(shè)定：帶有一個(gè) 匿名 的交換機(jī)，類型是 Direct*/private void createDefaultData() {Exchange exchange = new Exchange();exchange.setName("");exchange.setType(ExchangeType.DIRECT);exchange.setDurable(true);exchange.setAutoDelete(false);metaMapper.insertExchange(exchange);System.out.println("[DataBaseManager] 創(chuàng)建初始數(shù)據(jù)完成！");}//把數(shù)據(jù)庫中其他操作也在這里封裝一下public void insertExchange(Exchange exchange) {metaMapper.insertExchange(exchange);}public List<Exchange> selectAllExchange() {return metaMapper.selectAllExchange();}public void deleteExchange(String exchangeName) {metaMapper.deleteExchange(exchangeName);}public void insertQueue(MSGQueue queue) {metaMapper.insertQueue(queue);}public List<MSGQueue> selectAllQueue() {return metaMapper.selectAllQueue();}public void deleteQueue(String queueName) {metaMapper.deleteQueue(queueName);}public void insertBinding(Binding binding) {metaMapper.insertBinding(binding);}public List<Binding> selectAllBinding() {return metaMapper.selectAllBinding();}public void deleteBinding(Binding binding) {metaMapper.deleteBinding(binding);}public void deleteDB() {//刪除文件File file = new File("./data/meta.db");boolean res = file.delete();if(res) {System.out.println("[DataBaseManager] 數(shù)據(jù)庫文件刪除完畢!");} else {System.out.println("[DataBaseManager] 數(shù)據(jù)庫文件刪除失敗!");}//刪除目錄File dataDir = new File("./data");boolean ret = dataDir.delete();if(ret) {System.out.println("[DataBaseManager] 數(shù)據(jù)庫目錄刪除完成!");} else {System.out.println("[DataBaseManager] 數(shù)據(jù)庫目錄刪除失敗!");}}}

1.5、針對(duì) DataBaseManager 進(jìn)行單元測(cè)試

設(shè)計(jì)單元測(cè)試，這里的要求就是單元測(cè)試用例和用例之間是需要相互獨(dú)立的，不會(huì)干擾，例如以下情況：

測(cè)試過程中，向數(shù)據(jù)庫中插入數(shù)據(jù) a .

在針對(duì) b 進(jìn)行測(cè)試，可能 a 這里的數(shù)據(jù)會(huì)對(duì) b 造成干擾.

Ps：這里不一定是數(shù)據(jù)庫，也可能是其他方面，例如是否搞了一個(gè)文件，是否占用了端口...

@SpringBootTest
public class DataBaseManagerTests {private DataBaseManager dataBaseManager = new DataBaseManager();@BeforeEachpublic void setUp() {RabbitmqProjectApplication.context = SpringApplication.run(RabbitmqProjectApplication.class);dataBaseManager.init();}@AfterEachpublic void setclose() {//此處不能直接刪除 數(shù)據(jù)庫文件 ，需要先關(guān)閉 context 對(duì)象//此處 context 對(duì)象持有了 MetaMapper 的實(shí)例， MetaMapper 又打開了 meta.db 數(shù)據(jù)庫//如果 meta.db 被別人打開了，此時(shí)刪除文件是不會(huì)成功的(Windows 系統(tǒng)限制, Linux 則不會(huì))//另一方面 context 會(huì)占用 8080 端口，此處的 close 也是釋放 8080 端口RabbitmqProjectApplication.context.close();dataBaseManager.deleteDB();}@Testpublic void testInitTable() {List<Exchange> exchanges = dataBaseManager.selectAllExchange();List<MSGQueue> msgQueues = dataBaseManager.selectAllQueue();List<Binding> bindings = dataBaseManager.selectAllBinding();Assertions.assertEquals(1, exchanges.size());Assertions.assertEquals("", exchanges.get(0).getName());Assertions.assertEquals(ExchangeType.DIRECT, exchanges.get(0).getType());Assertions.assertEquals(0, msgQueues.size());Assertions.assertEquals(0, bindings.size());}private Exchange createTestExchange(String exchangeName) {Exchange  exchange = new Exchange();exchange.setName(exchangeName);exchange.setType(ExchangeType.FANOUT);exchange.setDurable(true);exchange.setAutoDelete(false);exchange.setArguments("aaa", 1);exchange.setArguments("bbb", 2);return exchange;}@Testpublic void insertExhangeTest() {Exchange exchange = createTestExchange("testExchange");dataBaseManager.insertExchange(exchange);List<Exchange> exchanges = dataBaseManager.selectAllExchange();Assertions.assertEquals(2, exchanges.size());Exchange testExchange = exchanges.get(1);Assertions.assertEquals("testExchange", testExchange.getName());Assertions.assertEquals(ExchangeType.FANOUT, testExchange.getType());Assertions.assertEquals(true, testExchange.isDurable());Assertions.assertEquals(false, testExchange.isAutoDelete());Assertions.assertEquals(1, testExchange.getArguments("aaa"));Assertions.assertEquals(2, testExchange.getArguments("bbb"));}@Testpublic void deleteExchangeTest() {Exchange exchange = createTestExchange("testExchange");dataBaseManager.insertExchange(exchange);List<Exchange> exchanges = dataBaseManager.selectAllExchange();Assertions.assertEquals(2, exchanges.size());Assertions.assertEquals("testExchange", exchanges.get(1).getName());//刪除dataBaseManager.deleteExchange("testExchange");exchanges = dataBaseManager.selectAllExchange();Assertions.assertEquals(1, exchanges.size());}private MSGQueue createTestQueue(String queueName) {MSGQueue queue = new MSGQueue();queue.setName(queueName);queue.setDurable(true);queue.setExclusive(false);queue.setAutoDelete(false);queue.setArguments("aaa", 1);queue.setArguments("bbb", 2);return queue;}@Testpublic void testInsertQueue() {MSGQueue queue = createTestQueue("testQueue");dataBaseManager.insertQueue(queue);List<MSGQueue> queues = dataBaseManager.selectAllQueue();Assertions.assertEquals(1, queues.size());MSGQueue msgQueue = queues.get(0);Assertions.assertEquals("testQueue", msgQueue.getName());Assertions.assertEquals(true, msgQueue.isDurable());Assertions.assertEquals(false, msgQueue.isExclusive());Assertions.assertEquals(false, msgQueue.isAutoDelete());Assertions.assertEquals(1, msgQueue.getArguments("aaa"));Assertions.assertEquals(2, msgQueue.getArguments("bbb"));}@Testpublic void testDeleteQueue() {MSGQueue queue = createTestQueue("testQueue");dataBaseManager.insertQueue(queue);List<MSGQueue> queues = dataBaseManager.selectAllQueue();Assertions.assertEquals(1, queues.size());//刪除dataBaseManager.deleteQueue("testQueue");queues = dataBaseManager.selectAllQueue();Assertions.assertEquals(0, queues.size());}private Binding createTestBinding(String exchangeName, String queueName) {Binding binding = new Binding();binding.setExchangeName(exchangeName);binding.setQueueName(queueName);binding.setBindingKey("testBindingKey");return binding;}@Testpublic void testInsertAndDeleteBinding() {Binding binding = createTestBinding("testExchange", "testQueue");dataBaseManager.insertBinding(binding);List<Binding> bindingList = dataBaseManager.selectAllBinding();Assertions.assertEquals(1, bindingList.size());binding = bindingList.get(0);Assertions.assertEquals("testExchange", binding.getExchangeName());Assertions.assertEquals("testQueue", binding.getQueueName());Assertions.assertEquals("testBindingKey", binding.getBindingKey());//刪除dataBaseManager.deleteBinding(binding);bindingList = dataBaseManager.selectAllBinding();Assertions.assertEquals(0, bindingList.size());}}

當(dāng)然，我只是做了簡(jiǎn)單的設(shè)計(jì)測(cè)試用例，實(shí)際上站在更嚴(yán)謹(jǐn)?shù)慕嵌?#xff0c;還需要設(shè)計(jì)更豐富的測(cè)試用例~

相比于 功能/業(yè)務(wù)代碼，測(cè)試用例代碼編寫起來雖然比較無聊，但是重要性是非常大的，這些操作會(huì)大大提高整個(gè)項(xiàng)目的開發(fā)效率.

Ps：單元測(cè)試，本來就是開發(fā)要搞的活，寫代碼不可能沒有 bug，進(jìn)行周密的測(cè)試，是應(yīng)對(duì) bug 最有效的手段.

?