Spring Boot JDBC：加载DataSource过程的源码分析及yml中DataSource的配置

装载至：https://www.cnblogs.com/storml/p/8611388.html

Spring Boot实现了自动加载DataSource及相关配置。当然，使用时加上@EnableAutoConfiguration注解是必须的。下面就是对这一部分的源码分析。

（1）Spring Boot启动后会调用org.springframework.boot.autoconfigure.jdbc.DataSourceAutoConfiguration。下面是部分源码。

 1 @Configuration
 2 @ConditionalOnClass({ DataSource.class, EmbeddedDatabaseType.class })
 3 @EnableConfigurationProperties(DataSourceProperties.class)
 4 @Import({ DataSourcePoolMetadataProvidersConfiguration.class,
 5         DataSourceInitializationConfiguration.class })
 6 public class DataSourceAutoConfiguration {
 7
 8     @Configuration
 9     @Conditional(EmbeddedDatabaseCondition.class)
10     @ConditionalOnMissingBean({ DataSource.class, XADataSource.class })
11     @Import(EmbeddedDataSourceConfiguration.class)
12     protected static class EmbeddedDatabaseConfiguration {
13     }
14
15     @Configuration
16     @Conditional(PooledDataSourceCondition.class)
17     @ConditionalOnMissingBean({ DataSource.class, XADataSource.class })
18     @Import({ DataSourceConfiguration.Hikari.class, DataSourceConfiguration.Tomcat.class,
19             DataSourceConfiguration.Dbcp2.class, DataSourceConfiguration.Generic.class,
20             DataSourceJmxConfiguration.class })
21     protected static class PooledDataSourceConfiguration {
22     }
23     ......
24 }

我们从中可以看出，DataSourceAutoConfiguration中有两个嵌套类，一个是EmbeddedDatabaseConfiguration，另一个是PooledDataSourceConfiguration。

EmbeddedDatabaseConfiguration表示已经嵌入Spring Boot的DataSource，除了Maven中加入相应的Driver，可以不做其他额外配置就能使用。从EmbeddedDatabaseType类可以看出，Spring Boot的内嵌DataSource支持HSQL，H2，DERBY这三种DB。

PooledDataSourceConfiguration表示Spring Boot还支持一些实现Pool的DataSource。从org.springframework.boot.jdbc.DataSourceBuilder中可以看出，当前版本的Spring Boot（2.0）只支持com.zaxxer.hikari.HikariDataSource，org.apache.tomcat.jdbc.pool.DataSource，org.apache.commons.dbcp2.BasicDataSource。其中，性能更加优秀的HikariDataSource是Spring Boot的默认选择（DataSourceBuilder中DATA_SOURCE_TYPE_NAMES[0] = com.zaxxer.hikari.HikariDataSource）。所以，当application.yml文件中做如下配置时，Spring Boot默认使用HikariDataSource数据库连接池。

spring:
  datasource:
    url: jdbc:mysql://localhost:3306/sas
    username: root
    password: ****
    driver-class-name: com.mysql.jdbc.Driver
    #type: com.zaxxer.hikari.HikariDataSource

（2）我们以HikariDataSource举例，接下来调用PooledDataSourceConfiguration中org.springframework.boot.autoconfigure.jdbc.DataSourceConfiguration抽象类的Hikari嵌套类（DataSourceConfiguration抽象类的一个实现类）。

 1 abstract class DataSourceConfiguration {
 2
 3     @SuppressWarnings("unchecked")
 4     protected <T> T createDataSource(DataSourceProperties properties,
 5             Class<? extends DataSource> type) {
 6         return (T) properties.initializeDataSourceBuilder().type(type).build();
 7     }
 8
 9     /* Omit Tomcat Pool DataSource configuration.*/
10     /**
11      * Hikari DataSource configuration.
12      */
13     @ConditionalOnClass(HikariDataSource.class)
14     @ConditionalOnProperty(name = "spring.datasource.type", havingValue = "com.zaxxer.hikari.HikariDataSource", matchIfMissing = true)
15     static class Hikari extends DataSourceConfiguration {
16
17         @Bean
18         @ConfigurationProperties(prefix = "spring.datasource.hikari")
19         public HikariDataSource dataSource(DataSourceProperties properties) {
20             HikariDataSource dataSource = createDataSource(properties,
21                     HikariDataSource.class);
22             if (StringUtils.hasText(properties.getName())) {
23                 dataSource.setPoolName(properties.getName());
24             }
25             return dataSource;
26         }
27     }
28     /* Omit DBCP DataSource configuration.*/
29 }

我们从黄色部分可以看出，当application.yml文件中配置spring.datasource.type = com.zaxxer.hikari.HikariDataSource时，会使用HikariDataSource作为数据库连接池（当然上面也分析了，它是默认选择）。我们从绿色部分可以看出它的配置信息主要从两个类中读取，一个是org.springframework.boot.autoconfigure.jdbc.DataSourceProperties，另一个则是本类HikariDataSource的父类com.zaxxer.hikari.HikariConfig。

@ConfigurationProperties(prefix = "spring.datasource")
public class DataSourceProperties implements BeanClassLoaderAware, InitializingBean {}

@ConfigurationProperties(prefix = "spring.datasource.hikari")
public HikariDataSource dataSource(DataSourceProperties properties) {}

我们从@ConfigurationProperties配置及两个具体的类所包含的的域可以得出配置HikariDataSource信息。下面是例子。

spring:
  datasource:
    name: #Name of the datasource. Default to "testdb" when using an embedded database.
    driverClassName: #Fully qualified name of the JDBC driver. Auto-detected based on the URL by default.
    url: #DBC URL of the database.
    type: #Fully qualified name of the connection pool implementation to use. By default, it is auto-detected from the classpath.
    username: #Login username of the database.
    password: #Login password of the database.
    ## For more details please see DataSourceProperties.
    hikari:
      connectionTimeout:
      validationTimeout:
      maxPoolSize:
      minIdle:
      dataSourceProperties:
      ## For more details please see HikariConfig.

（3）当读完配置后，则会通过HikariDataSource.getConnection()方法创建HikariPool对象。HikariPool及其父类PoolBase做了许多复杂的工作，包括创建Pool，创建Connection，读取Config，验证等等。调用HikariDataSource.getConnection()方法最终得到了这个Connection对象。这个过程中主要做了以下几步：

① 创建HikariPool对象。

② 调用HikariPool对象的父类对象PoolBase的构造器，读取HikariConfig配置信息配置PoolBase的属性。

③ 调用PoolBase的构造器的initializeDataSource方法，利用com.zaxxer.hikari.util.DriverDataSource创建DataSource对象（这里主要指JDBC URL方式）。DriverDataSource中会把所有的DataSource信息封装到driverProperties属性中，这是为了适配java.sql.Driver的connect(String url, java.util.Properties info)方法。

 1 public final class DriverDataSource implements DataSource {
 2
 3     private final String jdbcUrl;
 4     private final Properties driverProperties;
 5     private Driver driver;
 6
 7     public DriverDataSource(String jdbcUrl, String driverClassName, Properties properties, String username, String password) {
 8         this.jdbcUrl = jdbcUrl;
 9         this.driverProperties = new Properties();
10         Iterator e = properties.entrySet().iterator();
11
12         while(e.hasNext()) {
13             Entry driverClass = (Entry)e.next();
14             this.driverProperties.setProperty(driverClass.getKey().toString(), driverClass.getValue().toString());
15         }
16
17         if(username != null) {
18             this.driverProperties.put("user", this.driverProperties.getProperty("user", username));
19         }
20
21         if(password != null) {
22             this.driverProperties.put("password", this.driverProperties.getProperty("password", password));
23         }
24     ......
25     }
26
27    @Override
28    public Connection getConnection() throws SQLException
29    {
30       return driver.connect(jdbcUrl, driverProperties);
31    }
32 }

④ 调用HikariPool对象的构造器，同样也是配置一堆线程池信息。

⑤ 返回HikariPool.getConnection()。这个过程中，做了包含PoolBase.newPoolEntry()及PoolBase.newConnection()的许多复杂方法。从PoolBase.newConnection()可以看出，最终还是调用的步骤③的getConnection()方法获取到了这个Connection对象。

 1    private Connection newConnection() throws Exception
 2    {
 3       final long start = currentTime();
 4
 5       Connection connection = null;
 6       try {
 7          String username = config.getUsername();
 8          String password = config.getPassword();
 9
10          connection = (username == null) ? dataSource.getConnection() : dataSource.getConnection(username, password);
11          if (connection == null) {
12             throw new SQLTransientConnectionException("DataSource returned null unexpectedly");
13          }
14
15          setupConnection(connection);
16          lastConnectionFailure.set(null);
17          return connection;
18       }
19       catch (Exception e) {
20          if (connection != null) {
21             quietlyCloseConnection(connection, "(Failed to create/setup connection)");
22          }
23          else if (getLastConnectionFailure() == null) {
24             LOGGER.debug("{} - Failed to create/setup connection: {}", poolName, e.getMessage());
25          }
26
27          lastConnectionFailure.set(e);
28          throw e;
29       }
30       finally {
31          // tracker will be null during failFast check
32          if (metricsTracker != null) {
33             metricsTracker.recordConnectionCreated(elapsedMillis(start));
34          }
35       }
36    }