Spring的IOC/DI让层与层之间解耦，让接口和实现类不必强依赖，提高程序的扩展性和对需求的适应性。TeaFramework也提供了一个IOC容器，帮我们来生成实例，那么具体设计是什么样的呢？请往下看。
纯注解以其简单的优势迅速得到广大程序员的青睐，那么这里还是以注解的方式来标示bean和注入bean。TeaFramework中标示bean的注解有两种：@Component、@TeaDao。
@Component用于控制层和service层.
@TeaDao用于数据访问层
@Component包含两个属性，分别标示bean的name、生成对象的模式（单例、原型，默认是单例），代码如下：
@Target({ ElementType.TYPE })@Retention(RetentionPolicy.RUNTIME)public @interface Component { public String value() default ""; public Scope scope() default Scope.singleton;}
接下来是注入了，注入又分两种：ByName和ByType，那么注解里只需要一个参数即可搞定，当传入beanName时是按ByName注入，否则按ByType注入。
@Target({ ElementType.FIELD })@Retention(RetentionPolicy.RUNTIME)public @interface Inject { public String value() default "";}
那么这样一来我们的代码变得很简洁，控制层代码可以这样写
@Componentpublic class UserController { @Inject private UserService userService;}
service层代码可以这样写
@Componentpublic class UserServiceImpl implements UserService { @Inject private UserDao userDao; @Override public User addUser(User user) { userDao.addUser(user); return userDao.getUserById(user.getId()); }}
数据访问层代码可以这样写
@TeaDaopublic interface UserDao { @GetPrimaryKey(sql = "select s_users.nextval from dual", primaryKeyProperty = "id") @SQL("insert into users(id,username,password) values(#id#,#username#,#password#)") public void addUser(User user); @SQL("select * from users where id=#id#") public User getUserById(Long id);}
接下来看第二个问题，怎么把bean装进Bean容器呢？这里需要定义两个Map，一个是beanName和对象之间的映射，另一个是beanType和bean List的映射，具体代码如下：
public class BeanContainer { private static BeanContainer instance; private static Map
, List
> CLASS_BEAN_MAPPING = new ConcurrentHashMap
, List
>(200); private static Map
NAME_BEAN_MAPPING = new ConcurrentHashMap
(200); private BeanContainer() { } public synchronized static BeanContainer getInstance() { if (instance == null) { instance = new BeanContainer(); } return instance; } public boolean contains(String beanName) { return NAME_BEAN_MAPPING.containsKey(beanName); } public void putBean(Class
beanClass, Object object) { if (CLASS_BEAN_MAPPING.get(beanClass) == null) { List
list = new ArrayList(); list.add(object); CLASS_BEAN_MAPPING.put(beanClass, list); } else { CLASS_BEAN_MAPPING.get(beanClass).add(object); } } public void putBean(String beanName, Object object) { NAME_BEAN_MAPPING.put(beanName, object); } public
T getBeanByName(String beanName) { return (T) NAME_BEAN_MAPPING.get(beanName); } public
List
getBeansByType(Class
beanClass) { return (List
) CLASS_BEAN_MAPPING.get(beanClass); } public void inject() { for (Map.Entry
entry : NAME_BEAN_MAPPING.entrySet()) { injectProperties(entry.getValue()); } } private void injectProperties(Object bean) { List
result = new ArrayList
(); findAllField(bean.getClass(), result); for (Field field : result) { Inject inject = field.getAnnotation(Inject.class); if (inject != null) { field.setAccessible(true); Object object = null; if (!"".equals(inject.value())) {// id注入 object = NAME_BEAN_MAPPING.get(inject.value()); if (object == null) { throw new TeaInjectException("没找到beanName为" + inject.value() + "的Bean"); } } else { List
list = CLASS_BEAN_MAPPING.get(field.getGenericType());// 类型注入 if (list == null || list.size() == 0) { throw new TeaInjectException("没找到类型为" + field.getGenericType().getTypeName() + "的Bean"); } else if (list.size() > 1) { throw new TeaInjectException("存在多个类型为" + field.getGenericType().getTypeName() + "的Bean"); } object = list.get(0); } if (object == null) { throw new TeaInjectException("没找到对应bean"); } try { field.set(bean, object); } catch (Exception e) { throw new TeaInjectException(e); } } } } private void findAllField(Class
clazz, List
result) { Field[] fields = clazz.getDeclaredFields(); if (fields != null && fields.length > 0) { for (int i = 0; i < fields.length; i++) { result.add(fields[i]); } } Class
superclass = clazz.getSuperclass(); if (superclass != null && !java.lang.Object.class.equals(superclass)) { findAllField(superclass, result); } }}
上面有个重载的putBean方法，一个是把<beanName,对象>放进Map<String, Object>，另一个是对象按照类型映射到List，即Map<Class<?>, List<Object>>。
injectProperties则是注入属性的具体实现了，如果@Inject的value不为空串，则是根据beanName注入，否则则是根据type注入，ByType注入做了一个特殊处理，就是如果一个Type下有很多实例，拿第一个注入，这里有两个功能是没有实现的属性为集合时注入的场景、原型模式下产生实例的场景，后期可以补上。
这样就很简易了的实现了IOC容器和依赖注入，那么容器在启动时要扫描哪些包、对象是怎么实例化放进Map的？请关注TeaFramework——IOC容器实现（二）
项目地址：
博客：
