@ConfigurationProperties 配置详解

文章转自 https://blog.csdn.net/qq_26000415/article/details/78942494

前言
新的一年到了,在这里先祝大家新年快乐.我们在上一篇spring boot 源码解析12-servlet容器的建立 中 分析 ServerProperties时,发现其类上有@ConfigurationProperties 注解,加上该注解后,就会注入在application.properties中server开头的属性,那么它是怎么生效的呢?我们这篇文章就来分析一下.这篇文章内容比较长,大家慢慢看…

@ConfigurationProperties 使用方式
我们首先声明实体类,用于属性的注入.代码如下:

public class People {

private String name;

private Integer age;

private List<String> address;

private Phone phone;

// get set 忽略,自己加上即可..
}

public class Phone {

private String number;

// get set 忽略,自己加上即可..

}
在application.properties 中加入如下配置:

com.example.demo.name=${aaa:hi}
com.example.demo.age=11
com.example.demo.address[0]=北京
com.example.demo.address[1]=上海
com.example.demo.address[2]=广州
com.example.demo.phone.number=1111111
@ConfigurationProperties 注解支持两种方式.

加在类上,需要和@Component注解,结合使用.代码如下:

@Component
@ConfigurationProperties(prefix = "com.example.demo")
public class People {

private String name;

private Integer age;

private List<String> address;

private Phone phone;
}
通过@Bean的方式进行声明,这里我们加在启动类即可,代码如下:

@SpringBootApplication
public class DemoApplication {

@Bean
@ConfigurationProperties(prefix = "com.example.demo")
public People people() {

return new People();
}

public static void main(String[] args) {
SpringApplication.run(DemoApplication.class, args);
}

}
这里我们使用第2种,原因是这样好debug,看源码…

我们再写一个Controller进行测试一下吧.代码如下:

@RestController
public class TestController {

@Autowired
private People people;

@RequestMapping("/get_name")
public String getName() {

return people.getName();
}

@RequestMapping("/get_address")
public List<String> getAddress() {

return people.getAddress();
}

@RequestMapping("/get_phone_numer")
public String getNumber() {

return people.getPhone().getNumber();
}
}
访问 /get_name,其返回值如下:

hi

访问 /get_address,其返回值如下:

[“北京”,”上海”,”广州”]

访问 get_phone_numer,其返回值如下:

1111111

使用方式就介绍完了,接下来,我们就来看看spring 是如何处理的吧.

解析
我们应该知道了@ConfigurationProperties 和 @Bean 或者 @Component 等只要能生成spring bean 的注解 结合起来使用,这样的话,当其他类注入该类时,就会触发该类的加载过程,那么在加载过程中,会调用AbstractAutowireCapableBeanFactory#applyBeanPostProcessorsBeforeInitialization.因此会触发ConfigurationPropertiesBindingPostProcessor#postProcessBeforeInitialization的调用,这里就是我们的起点.

ConfigurationPropertiesBindingPostProcessor#postProcessBeforeInitialization 代码如下:

public Object postProcessBeforeInitialization(Object bean, String beanName)
throws BeansException {
// 1. 获得类上的@ConfigurationProperties注解,如果注解存在,则调用postProcessBeforeInitialization 进行处理
ConfigurationProperties annotation = AnnotationUtils
.findAnnotation(bean.getClass(), ConfigurationProperties.class);
if (annotation != null) {
postProcessBeforeInitialization(bean, beanName, annotation);
}
// 2. 寻找工厂方法上是否有@ConfigurationProperties 注解,如果存在的话,则调用postProcessBeforeInitialization进行处理
annotation = this.beans.findFactoryAnnotation(beanName,
ConfigurationProperties.class);
if (annotation != null) {
postProcessBeforeInitialization(bean, beanName, annotation);
}
return bean;
}
2件事:

获得类上的@ConfigurationProperties注解,如果注解存在,则调用postProcessBeforeInitialization 进行处理
寻找工厂方法上是否有@ConfigurationProperties 注解,如果存在的话,则调用postProcessBeforeInitialization进行处理.对应的是@Bean的方式.
不管怎么样,最终都会调用ConfigurationPropertiesBindingPostProcessor#postProcessBeforeInitialization.代码如下:

private void postProcessBeforeInitialization(Object bean, String beanName,
ConfigurationProperties annotation) {
Object target = bean;
// 1. 实例化PropertiesConfigurationFactory,该类实现了FactoryBean, MessageSourceAware, InitializingBean 接口,并进行一些属性的设置
PropertiesConfigurationFactory<Object> factory = new PropertiesConfigurationFactory<Object>(
target);
factory.setPropertySources(this.propertySources);
factory.setValidator(determineValidator(bean));
// If no explicit conversion service is provided we add one so that (at least)
// comma-separated arrays of convertibles can be bound automatically
// 由于conversionService 一直为 null,因此会调用getDefaultConversionService
factory.setConversionService(this.conversionService == null
? getDefaultConversionService() : this.conversionService);
if (annotation != null) {
// 2. 如果注解存在,这是肯定的,不然也不会执行该方法,则根据@ConfigurationProperties的值进行配置
factory.setIgnoreInvalidFields(annotation.ignoreInvalidFields());
factory.setIgnoreUnknownFields(annotation.ignoreUnknownFields());
factory.setExceptionIfInvalid(annotation.exceptionIfInvalid());
factory.setIgnoreNestedProperties(annotation.ignoreNestedProperties());
if (StringUtils.hasLength(annotation.prefix())) {
// 2.1 如果配置了prefix,或者value 值,则设置TargetName
factory.setTargetName(annotation.prefix());
}
}
try {
// 3. 进行绑定
factory.bindPropertiesToTarget();
}
catch (Exception ex) {
String targetClass = ClassUtils.getShortName(target.getClass());
throw new BeanCreationException(beanName, "Could not bind properties to "
+ targetClass + " (" + getAnnotationDetails(annotation) + ")", ex);
}
}
3件事:

实例化PropertiesConfigurationFactory,该类实现了FactoryBean, MessageSourceAware, InitializingBean 接口,并进行一些属性的设置.

将ConfigurationPropertiesBindingPostProcessor中的propertySources赋值给PropertiesConfigurationFactory
通过调用determineValidator方法,生成Validator,并进行赋值.代码如下:

private Validator determineValidator(Object bean) {
// 1. 获得validator
Validator validator = getValidator();
// 2. 如果validator不等于null并且该Validator 支持该bean的话
boolean supportsBean = (validator != null && validator.supports(bean.getClass()));
if (ClassUtils.isAssignable(Validator.class, bean.getClass())) {// 3 如果当前类为Validator的子类
// 3.1 如果supportsBean,则实例化ChainingValidator
if (supportsBean) {
return new ChainingValidator(validator, (Validator) bean);
}
// 3.2 否则强转为Validator
return (Validator) bean;
}
// 4. 最后,如果supportsBean 则 返回Validator 否则 返回null
return (supportsBean ? validator : null);
}
4件事:

调用getValidator方法获得Validator.代码如下:

private Validator getValidator() {
// 1. 由之前可知,该validator 一直都是null.
if (this.validator != null) {
return this.validator;
}
// 2. 如果localValidator 等于null并且是jsr303环境的话,则实例化ValidatedLocalValidatorFactoryBean,并赋值给localValidator,lazy-init
// ValidatedLocalValidatorFactoryBean 实现了ValidatorFactory, ApplicationContextAware, InitializingBean, DisposableBean,SmartValidator, javax.validation.Validator
if (this.localValidator == null && isJsr303Present()) {
this.localValidator = new ValidatedLocalValidatorFactoryBean(
this.applicationContext);
}
return this.localValidator;
}
如果validator 不等于null,则直接返回.可是该validator是一直等于null.原因如下:
ConfigurationPropertiesBindingPostProcessor 实现了InitializingBean接口,因此为调用其afterPropertiesSet方法,在该方法,有如下片段:

if (this.validator == null) {
// 2. 尝试获得id 为 configurationPropertiesValidator,type为Validator 的bean,此时是没有获取到
this.validator = getOptionalBean(VALIDATOR_BEAN_NAME, Validator.class);
}
会尝试从beanFactory中获得id 为 configurationPropertiesValidator,type 为 Validator的bean,可是默认情况下,是不存在的.

如果localValidator 等于null并且是jsr303环境的话,则实例化ValidatedLocalValidatorFactoryBean,并赋值给localValidator,这是一个lazy-init,ValidatedLocalValidatorFactoryBean 实现了ValidatorFactory, ApplicationContextAware, InitializingBean, DisposableBean,SmartValidator, javax.validation.Validator接口.
如果不等于null,则直接返回
如果validator不等于null并且该Validator 支持该bean的话,则supportsBean等于true,否则为false.
如果当前类为Validator的子类

如果supportsBean为true,则实例化ChainingValidator,则初始化ChainingValidator.进行返回
否则强转为Validator,进行返回
最后,如果supportsBean 则 返回Validator 否则 返回null
如果ConfigurationPropertiesBindingPostProcessor#conversionService 等于null,则调用getDefaultConversionService获得默认的ConversionService.否则,直接将本类的conversionService 赋值给PropertiesConfigurationFactory 的ConversionService.还是由于conversionService一直为 null,因此会调用getDefaultConversionService.代码如下:

private ConversionService getDefaultConversionService() {
// 又是lazy-init 风格
// 1. 如果defaultConversionService 等于null,则意味着是第一次调用
if (this.defaultConversionService == null) {
// 1.1 实例化DefaultConversionService
DefaultConversionService conversionService = new DefaultConversionService();
// 1.2 调用autowireBean进行注入依赖,此时会注入converters,genericConverters
this.applicationContext.getAutowireCapableBeanFactory().autowireBean(this);
// 1.3 遍历converters,genericConverters 依次加入到conversionService的converters中
for (Converter<?, ?> converter : this.converters) {
conversionService.addConverter(converter);
}
for (GenericConverter genericConverter : this.genericConverters) {
conversionService.addConverter(genericConverter);
}
// 1.4 赋值给defaultConversionService
this.defaultConversionService = conversionService;
}
// 2. 如果不等于null,则直接返回
return this.defaultConversionService;
}
2件事:

如果defaultConversionService 等于null,则意味着是第一次调用,又是lazy-init 风格.

实例化DefaultConversionService
调用autowireBean进行注入依赖,此时会注入converters,genericConverters
遍历converters,genericConverters 依次加入到conversionService的converters中
赋值给defaultConversionService
如果不等于null,则直接返回
如果注解存在,这是肯定的,不然也不会执行该方法,则根据@ConfigurationProperties的值进行配置

如果配置了prefix,或者value 值,则设置TargetName.这个后面解析的时候会用到该值.
调用PropertiesConfigurationFactory#bindPropertiesToTarget,进行绑定
PropertiesConfigurationFactory#bindPropertiesToTarget 代码如下:

public void bindPropertiesToTarget() throws BindException {
// 1.首先判断propertySources是否为null,如果为null的话,抛出异常.一般不会为null的
Assert.state(this.propertySources != null, "PropertySources should not be null");
try {
if (logger.isTraceEnabled()) {
logger.trace("Property Sources: " + this.propertySources);

}
// 2. 将hasBeenBound 设为true
this.hasBeenBound = true;
// 3. 调用doBindPropertiesToTarget
doBindPropertiesToTarget();
}
catch (BindException ex) {
if (this.exceptionIfInvalid) {
throw ex;
}
PropertiesConfigurationFactory.logger
.error("Failed to load Properties validation bean. "
+ "Your Properties may be invalid.", ex);
}
}
3件事:

首先判断propertySources是否为null,如果为null的话,抛出异常.一般不会为null的.因为该类在实例化的时候,已经对其进行赋值了
将hasBeenBound 设为true
调用doBindPropertiesToTarget.代码如下:

private void doBindPropertiesToTarget() throws BindException {
// 1. 初始化RelaxedDataBinder 并进行设置一下属性. // target = SpringApplication.这样RelaxedDataBinder也就持有了SpringApplication
RelaxedDataBinder dataBinder = (this.targetName != null
? new RelaxedDataBinder(this.target, this.targetName)
: new RelaxedDataBinder(this.target));
// 对于当前场景来说validator还是为null的,在 ConfigurationPropertiesBindingPostProcessor#postProcessBeforeInitialization 中,该validator为ValidatedLocalValidatorFactoryBean
if (this.validator != null
&& this.validator.supports(dataBinder.getTarget().getClass())) {
dataBinder.setValidator(this.validator);
}
if (this.conversionService != null) {
// 持有了一系列的转换器
dataBinder.setConversionService(this.conversionService);
}
dataBinder.setAutoGrowCollectionLimit(Integer.MAX_VALUE);
dataBinder.setIgnoreNestedProperties(this.ignoreNestedProperties);
dataBinder.setIgnoreInvalidFields(this.ignoreInvalidFields);
dataBinder.setIgnoreUnknownFields(this.ignoreUnknownFields);
// 2. 扩展点,空实现
customizeBinder(dataBinder);
// 3. 获得relaxedTargetNames,对于当前来说,其值为-->spring.main,也就是获得@ConfigurationProperties中配置的prefix
Iterable<String> relaxedTargetNames = getRelaxedTargetNames();
// 4. 通过遍历target的属性,这里的target为SpringApplication.然后将SpringApplication的属性按照单词划分的规则,与relaxedTargetNames进行拼接
// 举例说明:SpringApplication中有一个logStartupInfo属性,则拆分为log-startup-info,然后与spring.main拼接为
// spring.main.log-startup-info 和 spring.main_log-startup-info
// 通过拼接生成key
Set<String> names = getNames(relaxedTargetNames);
// 5. 生成PropertyValues,此时就已经将配置文件中的占位符解析完了
PropertyValues propertyValues = getPropertySourcesPropertyValues(names,
relaxedTargetNames);
// 6. 调用bind,进行绑定
dataBinder.bind(propertyValues);
if (this.validator != null) {
dataBinder.validate();
}
// 7. 检查在绑定过程中是否出现异常,如果有的话,抛出BindException
checkForBindingErrors(dataBinder);
}
7件事:

初始化RelaxedDataBinder 并进行设置一下属性,target = People.这样RelaxedDataBinder也就持有了People.

如果validator不等于null,并且validator支持该类型的话,则设置RelaxedDataBinder的Validator,对于当前场景来说,是validator.
如果conversionService 不等于null,则设置ConversionService,这样RelaxedDataBinder就持有了一系列的转换器
设置AutoGrowCollectionLimit 为Integer.MAX_VALUE,该属性在处理集合属性注入时会用到
设置是否忽略嵌套属性,默认是false.
设置是否忽略不正确的属性,默认是false
设置是否忽略未知的子弹,默认是true
调用customizeBinder,这个扩展点,默认是空实现
调用getRelaxedTargetNames,对于当前来说,其值为–>com.example.demo,也就是获得@ConfigurationProperties中配置的prefix
通过遍历target的属性,这里的target为People.然后将People的属性按照单词划分的规则,与relaxedTargetNames进行拼接.举例说明: People中有一个name属性,则拆分后为name,然后与com.example.demo拼接为com.example.demo.name
调用getPropertySourcesPropertyValues,生成PropertyValues,在这步完成了占位符解析.这个步骤很关键,我们在第4点中进行分析.
调用bind,进行绑定
检查在绑定过程中是否出现异常,如果有的话,抛出BindException
getPropertySourcesPropertyValues.代码如下:

private PropertyValues getPropertySourcesPropertyValues(Set<String> names,
Iterable<String> relaxedTargetNames) {
// 1. 根据names和relaxedTargetNames 生成PropertyNamePatternsMatcher
PropertyNamePatternsMatcher includes = getPropertyNamePatternsMatcher(names,
relaxedTargetNames);
// 2. 返回PropertySourcesPropertyValues
return new PropertySourcesPropertyValues(this.propertySources, names, includes,
this.resolvePlaceholders);
}
根据names和relaxedTargetNames 生成PropertyNamePatternsMatcher.代码如下:

private PropertyNamePatternsMatcher getPropertyNamePatternsMatcher(Set<String> names,
Iterable<String> relaxedTargetNames) {
// 1. 如果ignoreUnknownFields 并且 target 不是map的子类,则返回DefaultPropertyNamePatternsMatcher,在@ConfigurationProperties中,ignoreUnknownFields默认是true
if (this.ignoreUnknownFields && !isMapTarget()) {
// Since unknown fields are ignored we can filter them out early to save
// unnecessary calls to the PropertySource.
return new DefaultPropertyNamePatternsMatcher(EXACT_DELIMITERS, true, names);
}
// 2. 如果relaxedTargetNames 不等于null,则通过对relaxedTargetNames去重后,返回DefaultPropertyNamePatternsMatcher
if (relaxedTargetNames != null) {
// We can filter properties to those starting with the target name, but
// we can't do a complete filter since we need to trigger the
// unknown fields check
Set<String> relaxedNames = new HashSet<String>();
for (String relaxedTargetName : relaxedTargetNames) {
relaxedNames.add(relaxedTargetName);
}
return new DefaultPropertyNamePatternsMatcher(TARGET_NAME_DELIMITERS, true,
relaxedNames);
}
// Not ideal, we basically can't filter anything
// 3. 返回默认的
return PropertyNamePatternsMatcher.ALL;
}
3件事:

如果ignoreUnknownFields 并且 target 不是map的子类,则返回DefaultPropertyNamePatternsMatcher,在@ConfigurationProperties中,ignoreUnknownFields默认是true.在此时,由于target 为People,因此返回DefaultPropertyNamePatternsMatcher.
如果relaxedTargetNames 不等于null,则通过对relaxedTargetNames去重后,返回DefaultPropertyNamePatternsMatcher
返回默认的
返回PropertySourcesPropertyValues.其类图如下:

 

其构造器如下:

PropertySourcesPropertyValues(PropertySources propertySources,
Collection<String> nonEnumerableFallbackNames,
PropertyNamePatternsMatcher includes, boolean resolvePlaceholders) {
Assert.notNull(propertySources, "PropertySources must not be null");
Assert.notNull(includes, "Includes must not be null");
this.propertySources = propertySources;
this.nonEnumerableFallbackNames = nonEnumerableFallbackNames;
this.includes = includes;
this.resolvePlaceholders = resolvePlaceholders;
PropertySourcesPropertyResolver resolver = new PropertySourcesPropertyResolver(
propertySources);
for (PropertySource<?> source : propertySources) {
processPropertySource(source, resolver);
}
}
3件事:

属性赋值.
实例化PropertySourcesPropertyResolver
遍历propertySources,依次调用processPropertySource.代码如下:

private void processPropertySource(PropertySource<?> source,
PropertySourcesPropertyResolver resolver) {
if (source instanceof CompositePropertySource) {
processCompositePropertySource((CompositePropertySource) source, resolver);
}
else if (source instanceof EnumerablePropertySource) {
processEnumerablePropertySource((EnumerablePropertySource<?>) source,
resolver, this.includes);
}
else {
processNonEnumerablePropertySource(source, resolver);
}
}
如果PropertySource是CompositePropertySource的子类,则调用processCompositePropertySource方法,而该方法最终还是调用了processPropertySource,做递归处理.
如果PropertySource是EnumerablePropertySource的子类,则调用processEnumerablePropertySource.这里需要说明一下,我们是配置在application.properties中,那么其PropertySource 为 PropertiesPropertySource,是EnumerablePropertySource的子类,其继承结构如下:

因此,关于配置文件属性的注入,最终会在这里执行.

否则,调用processNonEnumerablePropertySource.
我们重点来看processEnumerablePropertySource,代码如下:

private void processEnumerablePropertySource(EnumerablePropertySource<?> source,
PropertySourcesPropertyResolver resolver,
PropertyNamePatternsMatcher includes) {
if (source.getPropertyNames().length > 0) {
for (String propertyName : source.getPropertyNames()) {
if (includes.matches(propertyName)) {// 如果存在的话,则加入到propertyValues中
Object value = getEnumerableProperty(source, resolver, propertyName);
putIfAbsent(propertyName, value, source);
}
}
}
}
思路很简单,

首先判断source中是否有属性的配置,如果有的话,则依次遍历之
在遍历过程中,会调用PropertyNamePatternsMatcher#matches 判断是否匹配.这里说明一下,这里使用的是DefaultPropertyNamePatternsMatcher,其matches 会依次遍历其内部的names 看是否与传入的propertyName 匹配,这里的names 就是在实例化时传入的com.example.demo.name等之类的东西.

如果匹配的话,则调用getEnumerableProperty 获得值.代码如下:

private Object getEnumerableProperty(EnumerablePropertySource<?> source,
PropertySourcesPropertyResolver resolver, String propertyName) {
try {
if (this.resolvePlaceholders) {
return resolver.getProperty(propertyName, Object.class);
}
}
catch (RuntimeException ex) {
// Probably could not resolve placeholders, ignore it here
}
return source.getProperty(propertyName);
}
如果resolvePlaceholders 为true,则调用PropertySourcesPropertyResolver#getProperty 处理,由于resolvePlaceholders 默认为true,因此一般都会执行这里.

否则,直接从EnumerablePropertySource 获取值即可.

调用putIfAbsent 将值,属性名,保存到propertyValues 中.
其中2.1 会调用如下代码:

public <T> T getProperty(String key, Class<T> targetValueType) {
return getProperty(key, targetValueType, true);
}
最终调用如下代码:

protected <T> T getProperty(String key, Class<T> targetValueType, boolean resolveNestedPlaceholders) {
if (this.propertySources != null) {
for (PropertySource<?> propertySource : this.propertySources) {
if (logger.isTraceEnabled()) {
logger.trace("Searching for key '" + key + "' in PropertySource '" +
propertySource.getName() + "'");
}
Object value = propertySource.getProperty(key);
if (value != null) {
if (resolveNestedPlaceholders && value instanceof String) {
value = resolveNestedPlaceholders((String) value);
}
logKeyFound(key, propertySource, value);
return convertValueIfNecessary(value, targetValueType);
}
}
}
if (logger.isDebugEnabled()) {
logger.debug("Could not find key '" + key + "' in any property source");
}
return null;
}
3件事

如果propertySources 不等于null,则依次遍历propertySources,进行处理

通过调用PropertySource#getProperty进行获取

如果获取到值的话,

如果resolveNestedPlaceholders(这个一般都是true) 并且value 为String,则调用resolveNestedPlaceholders处理占位符–>${},一般这个步骤都会执行的.
打印日志
尝试对其进行转换.
如果经过第1步处理,还是没找到的话,则直接返回null
1.1.1.1 最终会调用如下方法.代码如下:

public String resolvePlaceholders(String text) {
if (this.nonStrictHelper == null) {
this.nonStrictHelper = createPlaceholderHelper(true);
}
return doResolvePlaceholders(text, this.nonStrictHelper);
}
如果nonStrictHelper等于null,则调用createPlaceholderHelper进行实例化.lazy-init 风格.
调用AbstractPropertyResolver#doResolvePlaceholders.代码如下:

private String doResolvePlaceholders(String text, PropertyPlaceholderHelper helper) {
return helper.replacePlaceholders(text, new PropertyPlaceholderHelper.PlaceholderResolver() {
@Override
public String resolvePlaceholder(String placeholderName) {
return getPropertyAsRawString(placeholderName);
}
});
}
这里直接调用了第一步实例化的PropertyPlaceholderHelper的replacePlaceholders进行处理,同时实例化了一个PlaceholderResolver,该类在获取值的时候会用到,这个后面会有介绍.PropertyPlaceholderHelper#replacePlaceholders 代码如下:

public String replacePlaceholders(String value, PlaceholderResolver placeholderResolver) {
Assert.notNull(value, "'value' must not be null");
return parseStringValue(value, placeholderResolver, new HashSet<String>());
}
最终调用如下代码:

protected String parseStringValue(
String value, PlaceholderResolver placeholderResolver, Set<String> visitedPlaceholders) {
StringBuilder result = new StringBuilder(value);
// 1. 通过String#indexOf 获取前缀(一般都是${)的下标
int startIndex = value.indexOf(this.placeholderPrefix);
// 2 如果存在
while (startIndex != -1) {
// 2.1 获得后缀,此时获得是最小的后缀,嵌套处理
int endIndex = findPlaceholderEndIndex(result, startIndex);
if (endIndex != -1) {// 3 如果endIndex 存在,
// 3.1 通过字符串的截取获得占位符,如${placeholder},那么此时获得的是placeholder
String placeholder = result.substring(startIndex + this.placeholderPrefix.length(), endIndex);
String originalPlaceholder = placeholder;
// 3.2 进行循环引用的检查,如果存在,则抛出IllegalArgumentException
if (!visitedPlaceholders.add(originalPlaceholder)) {
throw new IllegalArgumentException(
"Circular placeholder reference '" + originalPlaceholder + "' in property definitions");
}
// Recursive invocation, parsing placeholders contained in the placeholder key.
// 3.3 递归处理
placeholder = parseStringValue(placeholder, placeholderResolver, visitedPlaceholders);
// Now obtain the value for the fully resolved key...
// 3.4 进行解析占位符
String propVal = placeholderResolver.resolvePlaceholder(placeholder);
// 3.5 如果propVal 不等于null并且 valueSeparator(默认为 :)不等于null,则此时意味有默认值,
// 那么此时调用placeholderResolver#resolvePlaceholder 进行解析,如果解析失败的话,则返回默认值
if (propVal == null && this.valueSeparator != null) {
int separatorIndex = placeholder.indexOf(this.valueSeparator);
if (separatorIndex != -1) {
String actualPlaceholder = placeholder.substring(0, separatorIndex);
String defaultValue = placeholder.substring(separatorIndex + this.valueSeparator.length());
propVal = placeholderResolver.resolvePlaceholder(actualPlaceholder);
if (propVal == null) {
propVal = defaultValue;
}
}
}
// 3.6
if (propVal != null) {
// Recursive invocation, parsing placeholders contained in the
// previously resolved placeholder value.
// 3.6.1 如果propVal 不等于null,则意味着解析成功,则继续递归处理,处理完后,进行替换,

propVal = parseStringValue(propVal, placeholderResolver, visitedPlaceholders);
// 进行替换
result.replace(startIndex, endIndex + this.placeholderSuffix.length(), propVal);
if (logger.isTraceEnabled()) {
logger.trace("Resolved placeholder '" + placeholder + "'");
}
// 重新计算startIndex
startIndex = result.indexOf(this.placeholderPrefix, startIndex + propVal.length());
}
else if (this.ignoreUnresolvablePlaceholders) {
// Proceed with unprocessed value.
// 3.6.2 如果没有解析成功并且ignoreUnresolvablePlaceholders,则重新计算startIndex
startIndex = result.indexOf(this.placeholderPrefix, endIndex + this.placeholderSuffix.length());
}
else {
// 3.6.3 抛出IllegalArgumentException
throw new IllegalArgumentException("Could not resolve placeholder '" +
placeholder + "'" + " in value \"" + value + "\"");
}
// 3.7 从visitedPlaceholders 删除.该算法有点类似dfs.
visitedPlaceholders.remove(originalPlaceholder);
}
else {
// 2.2 将startIndex 设为-1,则意味着已经处理完了
startIndex = -1;
}
}
return result.toString();
}
3件事:

通过String#indexOf 获取前缀(一般都是${)的下标
如果存在

获得后缀,此时获得是最小的后缀,嵌套处理
如果endIndex 存在,

通过字符串的截取获得占位符,如${placeholder},那么此时获得的是placeholder
进行循环引用的检查,如果存在,则抛出IllegalArgumentException
调用parseStringValue,进行递归处理.
调用PlaceholderResolver#resolvePlaceholder进行解析占位符
如果propVal 不等于null并且 valueSeparator(默认为 :)不等于null,则此时意味有默认值,那么此时调用placeholderResolver#resolvePlaceholder 进行解析,如果解析失败的话,则返回默认值
如果propVal 不等于null,则意味着解析成功,则继续递归处理,处理完后,进行替换,重新计算startIndex
如果没有解析成功并且ignoreUnresolvablePlaceholders,则重新计算startIndex
其他情况下,则抛出异常
从visitedPlaceholders 删除.该算法有点类似dfs.
如果不存在,则将startIndex 设为-1,则意味着已经处理完了.
关于占位符的处理,集合,对象导航,属性转换的处理,我们这里先不解析,我们先解析最简单的情况,为People注入name 属性.并且配置文件中的配置如下:

com.example.demo.name=hi

视线回到PropertiesConfigurationFactory#doBindPropertiesToTarget中来,此时执行完了getPropertySourcesPropertyValues,接下来就该执行第6步,调用DataBinder#bind进行绑定.这里还是假设我们只配置了com.example.demo.name=hi. 代码如下:

public void bind(PropertyValues pvs) {
MutablePropertyValues mpvs = (pvs instanceof MutablePropertyValues) ?
(MutablePropertyValues) pvs : new MutablePropertyValues(pvs);
doBind(mpvs);
}
调用DataBinder#doBind,代码如下:

protected void doBind(MutablePropertyValues mpvs) {
// 1. 检查是否存在不允许的字段存在,如果存在则删除
checkAllowedFields(mpvs);
// 2.检查是否存在Required 字段缺失的情况
checkRequiredFields(mpvs);
// 3. 进行注入
applyPropertyValues(mpvs);
}
3件事:

检查是否存在不允许的字段存在,如果存在则删除
检查是否存在Required 字段缺失的情况,如果存在,则抛出异常
调用applyPropertyValues进行注入.代码如下:

protected void applyPropertyValues(MutablePropertyValues mpvs) {
try {
// Bind request parameters onto target object.
// 1. 进行注入
getPropertyAccessor().setPropertyValues(mpvs, isIgnoreUnknownFields(), isIgnoreInvalidFields());
}
catch (PropertyBatchUpdateException ex) {
// Use bind error processor to create FieldErrors.
// 2. 如果抛出异常,则记录异常
for (PropertyAccessException pae : ex.getPropertyAccessExceptions()) {
getBindingErrorProcessor().processPropertyAccessException(pae, getInternalBindingResult());
}
}
}
2件事:

进行注入.

调用getPropertyAccessor 获得ConfigurablePropertyAccessor.代码如下:

protected ConfigurablePropertyAccessor getPropertyAccessor() {
return getInternalBindingResult().getPropertyAccessor();
}
调用getInternalBindingResult,获得AbstractPropertyBindingResult.代码如下:

protected AbstractPropertyBindingResult getInternalBindingResult() {
// 1. 同样是lazy-init,当第一次调用时 ,调用initBeanPropertyAccess 进行初始化
if (this.bindingResult == null) {
initBeanPropertyAccess();
}
return this.bindingResult;
}
同样是lazy-init,当第一次调用时 ,调用initBeanPropertyAccess 进行初始化. initBeanPropertyAccess 代码如下:

public void initBeanPropertyAccess() {
Assert.state(this.bindingResult == null,
"DataBinder is already initialized - call initBeanPropertyAccess before other configuration methods");
this.bindingResult = createBeanPropertyBindingResult();
}
调用

protected AbstractPropertyBindingResult createBeanPropertyBindingResult() {
// 1. 实例化BeanPropertyBindingResult
BeanPropertyBindingResult result = new BeanPropertyBindingResult(getTarget(),
getObjectName(), isAutoGrowNestedPaths(), getAutoGrowCollectionLimit());
if (this.conversionService != null) {
// 2. 这个步骤是一定会执行的, 进行初始化
result.initConversion(this.conversionService);
}
if (this.messageCodesResolver != null) {
// 3. 设置messageCodesResolver
result.setMessageCodesResolver(this.messageCodesResolver);
}
return result;
}
3件事:

实例化BeanPropertyBindingResult
这个步骤是一定会执行的, 进行初始化conversionService
设置messageCodesResolver
调用BeanPropertyBindingResult#getPropertyAccessor.

public final ConfigurablePropertyAccessor getPropertyAccessor() {
// 1. lazy-inits
if (this.beanWrapper == null) {
// 1.1 最终调用PropertyAccessorFactory#forBeanPropertyAccess,直接实例化了BeanWrapperImpl
this.beanWrapper = createBeanWrapper();
this.beanWrapper.setExtractOldValueForEditor(true);
this.beanWrapper.setAutoGrowNestedPaths(this.autoGrowNestedPaths);
this.beanWrapper.setAutoGrowCollectionLimit(this.autoGrowCollectionLimit);
}
return this.beanWrapper;
}
还是同样的味道,lazy-init,最终调用PropertyAccessorFactory#forBeanPropertyAccess,直接实例化了BeanWrapperImpl.

调用BeanPropertyBindingResult#setPropertyValues 进行注入.
如果在注入过程出现异常,则记录异常.

其中 1.3 BeanPropertyBindingResult#setPropertyValues ,代码如下:

public void setPropertyValues(PropertyValues pvs, boolean ignoreUnknown, boolean ignoreInvalid)
throws BeansException {

List<PropertyAccessException> propertyAccessExceptions = null;
// 1. 获得propertyValues,一般情况下,此时传入的是MutablePropertyValues,因此直接通过MutablePropertyValues#getPropertyValueList 获取即可
List<PropertyValue> propertyValues = (pvs instanceof MutablePropertyValues ?
((MutablePropertyValues) pvs).getPropertyValueList() : Arrays.asList(pvs.getPropertyValues()));
// 2. 遍历propertyValues,依次调用setPropertyValue 进行处理
for (PropertyValue pv : propertyValues) {

// 删除一些无用的try-cath,减少篇幅...
setPropertyValue(pv);

}

if (propertyAccessExceptions != null) {
PropertyAccessException[] paeArray =
propertyAccessExceptions.toArray(new PropertyAccessException[propertyAccessExceptions.size()]);
throw new PropertyBatchUpdateException(paeArray);
}
}
3件事:

获得propertyValues,一般情况下,此时传入的是MutablePropertyValues,因此直接通过MutablePropertyValues#getPropertyValueList 获取即可
遍历propertyValues,依次调用setPropertyValue 进行处理
如果propertyAccessExceptions != null,则意味在第2步处理中,出现了问题,则抛出PropertyBatchUpdateException.
其中第2步,最终调用的是AbstractNestablePropertyAccessor#setPropertyValue,代码如下:

public void setPropertyValue(PropertyValue pv) throws BeansException {
PropertyTokenHolder tokens = (PropertyTokenHolder) pv.resolvedTokens;
if (tokens == null) {
String propertyName = pv.getName();
AbstractNestablePropertyAccessor nestedPa;
try {
nestedPa = getPropertyAccessorForPropertyPath(propertyName);
}
catch (NotReadablePropertyException ex) {
throw new NotWritablePropertyException(getRootClass(), this.nestedPath + propertyName,
"Nested property in path '" + propertyName + "' does not exist", ex);
}
tokens = getPropertyNameTokens(getFinalPath(nestedPa, propertyName));
if (nestedPa == this) {
pv.getOriginalPropertyValue().resolvedTokens = tokens;
}
nestedPa.setPropertyValue(tokens, pv);
}
else {
setPropertyValue(tokens, pv);
}
}
调用getPropertyAccessorForPropertyPath,处理对象导航,还是由于此处分析的最简单的场景,因此这里返回的就是当前类.
调用getPropertyNameTokens,这里处理的是集合的情况.同样,这里先不进行分析
调用AbstractNestablePropertyAccessor#setPropertyValue,进行赋值.代码如下:

protected void setPropertyValue(PropertyTokenHolder tokens, PropertyValue pv) throws BeansException {
if (tokens.keys != null) {
processKeyedProperty(tokens, pv);
}
else {
processLocalProperty(tokens, pv);
}
}
2件事:

如果 PropertyTokenHolder 中的keys 不等于null,则意味着是要对集合进行赋值,为什么?这个我们后面有解释.
否则调用 processLocalProperty进行处理.因为我们这里分析的是最简单的情况,因此会在这里进行处理.代码如下:
private void processLocalProperty(PropertyTokenHolder tokens, PropertyValue pv) {
// 1. 获得PropertyHandler
PropertyHandler ph = getLocalPropertyHandler(tokens.actualName);
// 2. 如果ph 等于null,或者 PropertyHandler 没有set方法
if (ph == null || !ph.isWritable()) {
// 2.1 如果该属性是可选的，则打印日志,否则抛出异常
if (pv.isOptional()) {

return;
}
else {
throw createNotWritablePropertyException(tokens.canonicalName);
}
}
// 3. 进行转换处理
Object oldValue = null;
Object originalValue = pv.getValue();
Object valueToApply = originalValue;
if (!Boolean.FALSE.equals(pv.conversionNecessary)) {
if (pv.isConverted()) {
valueToApply = pv.getConvertedValue();
}
else {
if (isExtractOldValueForEditor() && ph.isReadable()) {

oldValue = ph.getValue();

}
valueToApply = convertForProperty(
tokens.canonicalName, oldValue, originalValue, ph.toTypeDescriptor());
}
pv.getOriginalPropertyValue().conversionNecessary = (valueToApply != originalValue);
}
// 4. 进行赋值
ph.setValue(this.wrappedObject, valueToApply);
}
}
4件事

获得PropertyHandler,注意,这里调用的是BeanWrapperImpl.getLocalPropertyHandler代码如下:

protected BeanPropertyHandler getLocalPropertyHandler(String propertyName) {
PropertyDescriptor pd = getCachedIntrospectionResults().getPropertyDescriptor(propertyName);
if (pd != null) {
return new BeanPropertyHandler(pd);
}
return null;
}
调用getCachedIntrospectionResults 获得CachedIntrospectionResults.代码如下:

private CachedIntrospectionResults getCachedIntrospectionResults() {
Assert.state(getWrappedInstance() != null, "BeanWrapper does not hold a bean instance");
if (this.cachedIntrospectionResults == null) {
// lazy-init,第一次调用时初始化
this.cachedIntrospectionResults = CachedIntrospectionResults.forClass(getWrappedClass());
}
return this.cachedIntrospectionResults;
}
同样的调调,lazy-init,调用CachedIntrospectionResults#forClass获得CachedIntrospectionResults. 注意,这里传入的是target,也就是 People.class.代码如下:

static CachedIntrospectionResults forClass(Class<?> beanClass) throws BeansException {
// 1. 尝试从strongClassCache,softClassCache中获取,如果不为空,则直接返回.
CachedIntrospectionResults results = strongClassCache.get(beanClass);
if (results != null) {
return results;
}
results = softClassCache.get(beanClass);
if (results != null) {
return results;
}
// 2. 初始化
results = new CachedIntrospectionResults(beanClass);
ConcurrentMap<Class<?>, CachedIntrospectionResults> classCacheToUse;
// 3. 如果当前给定的类是否是给定的ClassLoader 或者是其父ClassLoader 加载的 或者 判断给定的classLoader 是否是acceptedClassLoaders的子classLoader
// 一般都是这个了,那么就使用strongClassCache,否则使用softClassCache
if (ClassUtils.isCacheSafe(beanClass, CachedIntrospectionResults.class.getClassLoader()) ||
isClassLoaderAccepted(beanClass.getClassLoader())) {
classCacheToUse = strongClassCache;
}
else {
if (logger.isDebugEnabled()) {
logger.debug("Not strongly caching class [" + beanClass.getName() + "] because it is not cache-safe");
}
classCacheToUse = softClassCache;
}
// 4. 加入缓存中
CachedIntrospectionResults existing = classCacheToUse.putIfAbsent(beanClass, results);
return (existing != null ? existing : results);
}

4件事:

尝试从strongClassCache,softClassCache中获取,如果不为空,则直接返回.
否则,进行初始化.CachedIntrospectionResults.其中,有如下代码:

beanInfo = (shouldIntrospectorIgnoreBeaninfoClasses ?
Introspector.getBeanInfo(beanClass, Introspector.IGNORE_ALL_BEANINFO) :
Introspector.getBeanInfo(beanClass));
这里调用了java.beans.Introspector 获取BeanInfo,而CachedIntrospectionResults只是对BeanInfo包装而已,关于CachedIntrospectionResults的初始化,这里就不继续深入了,也没有必要.

如果当前给定的类是否是给定的ClassLoader 或者是其父ClassLoader 加载的 或者 判断给定的classLoader 是否是acceptedClassLoaders的子classLoader,那么就使用strongClassCache,否则使用softClassCache.一般就是strongClassCache
加入缓存中
调用CachedIntrospectionResults#getPropertyDescriptor 获得PropertyDescriptor.注意,这里返回的是java.beans.PropertyDescriptor.是关于java反射的.不懂的可以百度一下.
如果PropertyDescriptor 不等于null,就意味着在target 也就是 People 中找到了对应的属性.因此,直接返回BeanPropertyHandler.
否则,返回null.
如果ph 等于null,或者 PropertyHandler set方法不存在或者不是public的,如果该属性是可选的，则打印日志,否则抛出异常.
调用convertForProperty,进行属性的转换,这里就是真正属性转换的地方,同样,后面有解释
调用BeanPropertyHandler#setValue进行赋值.代码如下:

public void setValue(final Object object, Object valueToApply) throws Exception {
// 1. 获得该属性对应的set方法
final Method writeMethod = (this.pd instanceof GenericTypeAwarePropertyDescriptor ?
((GenericTypeAwarePropertyDescriptor) this.pd).getWriteMethodForActualAccess() :
this.pd.getWriteMethod());
// 2. 如果该方法为私有的,则通过反射的方式,设置为可访问的
if (!Modifier.isPublic(writeMethod.getDeclaringClass().getModifiers()) && !writeMethod.isAccessible()) {
if (System.getSecurityManager() != null) {
AccessController.doPrivileged(new PrivilegedAction<Object>() {
@Override
public Object run() {
writeMethod.setAccessible(true);
return null;
}
});
}
else {
writeMethod.setAccessible(true);
}
}
// 3. 进行赋值
final Object value = valueToApply;
if (System.getSecurityManager() != null) {
try {
AccessController.doPrivileged(new PrivilegedExceptionAction<Object>() {
@Override
public Object run() throws Exception {
writeMethod.invoke(object, value);
return null;
}
}, acc);
}
catch (PrivilegedActionException ex) {
throw ex.getException();
}
}
else {
writeMethod.invoke(getWrappedInstance(), value);
}
}
}
3件事:

获得该属性对应的set方法
如果该方法为私有的,则通过反射的方式,设置为可访问的
通过writeMethod#invoke的方式调用set 方法 进行赋值.
至此,关于简单属性的注入(String类型)就分析完了,接下来就占位符,集合,对象导航,属性转换来分别做处理.

占位符处理
在之前的分析过程中,我们跳过了占位符处理工程的分析,这里我们将配置文件改为如下:

com.example.demo.name=${aaa:hi}
1
还是分析对People name 属性的注入. 之前的步骤同之前的分析的一样.只不过在PropertyPlaceholderHelper# parseStringValue处开始了对占位符的处理.代码如下:

protected String parseStringValue(
String value, PlaceholderResolver placeholderResolver, Set<String> visitedPlaceholders) {

StringBuilder result = new StringBuilder(value);

// 1. 通过String#indexOf 获取前缀(一般都是${)的下标
int startIndex = value.indexOf(this.placeholderPrefix);
// 2 如果存在
while (startIndex != -1) {
// 2.1 获得后缀,此时获得是最小的后缀,嵌套处理
int endIndex = findPlaceholderEndIndex(result, startIndex);
if (endIndex != -1) {// 3 如果endIndex 存在,
// 3.1 通过字符串的截取获得占位符,如${placeholder},那么此时获得的是placeholder
String placeholder = result.substring(startIndex + this.placeholderPrefix.length(), endIndex);
String originalPlaceholder = placeholder;
// 3.2 进行循环引用的检查,如果存在,则抛出IllegalArgumentException
if (!visitedPlaceholders.add(originalPlaceholder)) {
throw new IllegalArgumentException(
"Circular placeholder reference '" + originalPlaceholder + "' in property definitions");
}
// Recursive invocation, parsing placeholders contained in the placeholder key.
// 3.3 递归处理
placeholder = parseStringValue(placeholder, placeholderResolver, visitedPlaceholders);
// Now obtain the value for the fully resolved key...
// 3.4 进行解析占位符
String propVal = placeholderResolver.resolvePlaceholder(placeholder);
// 3.5 如果propVal 不等于null并且 valueSeparator(默认为 :)不等于null,则此时意味有默认值,
// 那么此时调用placeholderResolver#resolvePlaceholder 进行解析,如果解析失败的话,则返回默认值
if (propVal == null && this.valueSeparator != null) {
int separatorIndex = placeholder.indexOf(this.valueSeparator);
if (separatorIndex != -1) {
String actualPlaceholder = placeholder.substring(0, separatorIndex);
String defaultValue = placeholder.substring(separatorIndex + this.valueSeparator.length());
propVal = placeholderResolver.resolvePlaceholder(actualPlaceholder);
if (propVal == null) {
propVal = defaultValue;
}
}
}

// 3.6
if (propVal != null) {
// Recursive invocation, parsing placeholders contained in the
// previously resolved placeholder value.
// 3.6.1 如果propVal 不等于null,则意味着解析成功,则继续递归处理,处理完后,进行替换,

propVal = parseStringValue(propVal, placeholderResolver, visitedPlaceholders);
// 进行替换
result.replace(startIndex, endIndex + this.placeholderSuffix.length(), propVal);
if (logger.isTraceEnabled()) {
logger.trace("Resolved placeholder '" + placeholder + "'");
}
// 重新计算startIndex
startIndex = result.indexOf(this.placeholderPrefix, startIndex + propVal.length());
}
else if (this.ignoreUnresolvablePlaceholders) {
// Proceed with unprocessed value.
// 3.6.2 如果没有解析成功并且ignoreUnresolvablePlaceholders,则重新计算startIndex
startIndex = result.indexOf(this.placeholderPrefix, endIndex + this.placeholderSuffix.length());
}
else {
// 3.6.3 抛出IllegalArgumentException
throw new IllegalArgumentException("Could not resolve placeholder '" +
placeholder + "'" + " in value \"" + value + "\"");
}
// 3.7 从visitedPlaceholders 删除.该算法有点类似dfs.
visitedPlaceholders.remove(originalPlaceholder);
}
else {
// 2.2 将startIndex 设为-1,则意味着已经处理完了
startIndex = -1;
}
}
return result.toString();
}
首先通过String#indexOf 获得 { 的下标,这里是存在的,因此继续处理。此时调用的是PropertyPlaceholderHelper#findPlaceholderEndIndex 获的后缀.比方说如果我们配置的是  
 ${aa} 那么此时返回的就是}的下标,如果配置的是{ 的下标,这里是存在的,因此继续处理。此时调用的是PropertyPlaceholderHelper#findPlaceholderEndIndex 获的后缀.比方说如果我们配置的是   ${aa} 那么此时返回的就是}的下标,如果配置的是{aa}${bb}我们返回的就是,a后面的}的下标.代码如下:

private int findPlaceholderEndIndex(CharSequence buf, int startIndex) {
int index = startIndex + this.placeholderPrefix.length();
int withinNestedPlaceholder = 0;
while (index < buf.length()) {
if (StringUtils.substringMatch(buf, index, this.placeholderSuffix)) {
if (withinNestedPlaceholder > 0) {
withinNestedPlaceholder--;
index = index + this.placeholderSuffix.length();
}
else {
return index;
}
}
else if (StringUtils.substringMatch(buf, index, this.simplePrefix)) {
withinNestedPlaceholder++;
index = index + this.simplePrefix.length();
}
else {
index++;
}
}
return -1;
}
接下来,进行字符截取,此时我们配置的是com.example.demo.name=${aaa:hi},截取后获得的是aaa:hi,

第三步,将aaa:hi 作为参数,递归调用parseStringValue,由于此时aaa:hi 不存在${,因此直接返回的还是aaa:hi.

接下来,判断是否存在:,对于当前,是存在的,因此对其进行截取分别获得actualPlaceholder,defaultValue.对于当前, actualPlaceholder = aaa, defaultValue = hi, 然后调用PlaceholderResolver#resolvePlaceholder获得值,如果actualPlaceholder 解析失败,则将propVal 设为默认值.关于这部分对于的源码如下:

if (propVal == null && this.valueSeparator != null) {
int separatorIndex = placeholder.indexOf(this.valueSeparator);
if (separatorIndex != -1) {
String actualPlaceholder = placeholder.substring(0, separatorIndex);
String defaultValue = placeholder.substring(separatorIndex + this.valueSeparator.length());
propVal = placeholderResolver.resolvePlaceholder(actualPlaceholder);
if (propVal == null) {
propVal = defaultValue;
}
}
}
此刻,调用PlaceholderResolver#resolvePlaceholder,实际上调用的是在AbstractPropertyResolver#doResolvePlaceholders中实例化的PlaceholderResolver的实现.代码如下:

private String doResolvePlaceholders(String text, PropertyPlaceholderHelper helper) {
return helper.replacePlaceholders(text, new PropertyPlaceholderHelper.PlaceholderResolver() {
@Override
public String resolvePlaceholder(String placeholderName) {
return getPropertyAsRawString(placeholderName);
}
});
}
因此这里最终调用PropertySourcesPropertyResolver#getPropertyAsRawString,代码如下:

protected String getPropertyAsRawString(String key) {
return getProperty(key, String.class, false);
}
调用了
org.springframework.core.env.PropertySourcesPropertyResolver.getProperty(String, Class, boolean)方法,接下来的故事,就很之前一样了,这里就不在赘述了.

集合处理
接下来我们将application.properties 改为

com.example.demo.address[0]=北京
com.example.demo.address[1]=上海
com.example.demo.address[2]=广州
这里分析对People 中 address 属性的注入. 同样,前面的准备工作都一样,在对属性进行注入时,会调用AbstractNestablePropertyAccessor#setPropertyValue,代码如下:

public void setPropertyValue(String propertyName, Object value) throws BeansException {
AbstractNestablePropertyAccessor nestedPa;
try {
// 1. 生成AbstractNestablePropertyAccessor
nestedPa = getPropertyAccessorForPropertyPath(propertyName);
}
catch (NotReadablePropertyException ex) {
throw new NotWritablePropertyException(getRootClass(), this.nestedPath + propertyName,
"Nested property in path '" + propertyName + "' does not exist", ex);
}
// 2. 获得PropertyTokenHolder, getFinalPath 获得最终的PropertyName
PropertyTokenHolder tokens = getPropertyNameTokens(getFinalPath(nestedPa, propertyName));
// 3. 进行赋值
nestedPa.setPropertyValue(tokens, new PropertyValue(propertyName, value));
}
这里调用了AbstractNestablePropertyAccessor#getPropertyNameTokens,代码如下:

private PropertyTokenHolder getPropertyNameTokens(String propertyName) {
PropertyTokenHolder tokens = new PropertyTokenHolder();
String actualName = null;
List<String> keys = new ArrayList<String>(2);
int searchIndex = 0;
while (searchIndex != -1) {
// 1. 获得 [ 的下标
int keyStart = propertyName.indexOf(PROPERTY_KEY_PREFIX, searchIndex);
searchIndex = -1;
if (keyStart != -1) {
// 2 如果存在的话,则截取获得]的下标
int keyEnd = propertyName.indexOf(PROPERTY_KEY_SUFFIX, keyStart + PROPERTY_KEY_PREFIX.length());
if (keyEnd != -1) {
// 3. 如果存在的话,则截取出actualName,例如[map],那么此时就是""
if (actualName == null) {
actualName = propertyName.substring(0, keyStart);
}
// 4. 截取出key 此时就是map
String key = propertyName.substring(keyStart + PROPERTY_KEY_PREFIX.length(), keyEnd);
if (key.length() > 1 && (key.startsWith("'") && key.endsWith("'")) ||
(key.startsWith("\"") && key.endsWith("\""))) {
key = key.substring(1, key.length() - 1);
}
keys.add(key);
searchIndex = keyEnd + PROPERTY_KEY_SUFFIX.length();
}
}
}
tokens.actualName = (actualName != null ? actualName : propertyName);
tokens.canonicalName = tokens.actualName;
if (!keys.isEmpty()) {
// [ + StringUtils#collectionToDelimitedString(keys,][)+]
tokens.canonicalName += PROPERTY_KEY_PREFIX +
StringUtils.collectionToDelimitedString(keys, PROPERTY_KEY_SUFFIX + PROPERTY_KEY_PREFIX) +
PROPERTY_KEY_SUFFIX;
tokens.keys = StringUtils.toStringArray(keys);
}
return tokens;
}
步骤如下:

首先获得[ 的下标

如果存在的话,则尝试获取]的下标.

如果存在的]下标的话, 则截取出属性值,对应于当前的情况,就是address.然后加入keys中.
如果不存在,则意味着不存在集合的情况.接下来的处理就很之前一样.
注意在这里, keys 只加入了一个, 如:0,为什么呢?

因为我们是循环处理的,这点很重要.后面的步骤都是在此基础上进行的,在处理完com.example.demo.address[0]=北京 后,再处理 com.example.demo.address[1]=上海 .为啥呢? 因为在AbstractPropertyAccessor#setPropertyValues中我们是通过遍历的方式处理的,代码如下:

for (PropertyValue pv : propertyValues) {

setPropertyValue(pv);

}
接下来, 在AbstractNestablePropertyAccessor#setPropertyValue,由于此刻keys 不等于null,因此会执行processKeyedProperty.代码如下:

protected void setPropertyValue(PropertyTokenHolder tokens, PropertyValue pv) throws BeansException {
if (tokens.keys != null) {
processKeyedProperty(tokens, pv);
}
else {
processLocalProperty(tokens, pv);
}
}
processKeyedProperty 代码如下:

private void processKeyedProperty(PropertyTokenHolder tokens, PropertyValue pv) {
// 1. 获得
Object propValue = getPropertyHoldingValue(tokens);
String lastKey = tokens.keys[tokens.keys.length - 1];

if (propValue.getClass().isArray()) {
// 省略....
}

else if (propValue instanceof List) {
PropertyHandler ph = getPropertyHandler(tokens.actualName);
Class<?> requiredType = ph.getCollectionType(tokens.keys.length);
List<Object> list = (List<Object>) propValue;
int index = Integer.parseInt(lastKey);
Object oldValue = null;
if (isExtractOldValueForEditor() && index < list.size()) {
oldValue = list.get(index);
}
Object convertedValue = convertIfNecessary(tokens.canonicalName, oldValue, pv.getValue(),
requiredType, ph.nested(tokens.keys.length));
int size = list.size();
if (index >= size && index < this.autoGrowCollectionLimit) {
for (int i = size; i < index; i++) {
try {
list.add(null);
}
catch (NullPointerException ex) {
throw new InvalidPropertyException(getRootClass(), this.nestedPath + tokens.canonicalName,
"Cannot set element with index " + index + " in List of size " +
size + ", accessed using property path '" + tokens.canonicalName +
"': List does not support filling up gaps with null elements");
}
}
list.add(convertedValue);
}
else {
try {
list.set(index, convertedValue);
}
catch (IndexOutOfBoundsException ex) {
throw new InvalidPropertyException(getRootClass(), this.nestedPath + tokens.canonicalName,
"Invalid list index in property path '" + tokens.canonicalName + "'", ex);
}
}
}

else if (propValue instanceof Map) {
// 省略....
}

else {
throw new InvalidPropertyException(getRootClass(), this.nestedPath + tokens.canonicalName,
"Property referenced in indexed property path '" + tokens.canonicalName +
"' is neither an array nor a List nor a Map; returned value was [" + propValue + "]");
}
2件事:

调用getPropertyHoldingValue 获得 属性对应的对象. 对于当前,就是获得People 中address 所对应的对象实例.代码如下:

private Object getPropertyHoldingValue(PropertyTokenHolder tokens) {
// Apply indexes and map keys: fetch value for all keys but the last one.
// 1. 实例化PropertyTokenHolder
PropertyTokenHolder getterTokens = new PropertyTokenHolder();
getterTokens.canonicalName = tokens.canonicalName;
getterTokens.actualName = tokens.actualName;
getterTokens.keys = new String[tokens.keys.length - 1];
System.arraycopy(tokens.keys, 0, getterTokens.keys, 0, tokens.keys.length - 1);

Object propValue;
try {
// 2. 获得值
propValue = getPropertyValue(getterTokens);
}
catch (NotReadablePropertyException ex) {
throw new NotWritablePropertyException(getRootClass(), this.nestedPath + tokens.canonicalName,
"Cannot access indexed value in property referenced " +
"in indexed property path '" + tokens.canonicalName + "'", ex);
}

if (propValue == null) {
// null map value case
if (isAutoGrowNestedPaths()) {
int lastKeyIndex = tokens.canonicalName.lastIndexOf('[');
getterTokens.canonicalName = tokens.canonicalName.substring(0, lastKeyIndex);
propValue = setDefaultValue(getterTokens);
}
else {
throw new NullValueInNestedPathException(getRootClass(), this.nestedPath + tokens.canonicalName,
"Cannot access indexed value in property referenced " +
"in indexed property path '" + tokens.canonicalName + "': returned null");
}
}
return propValue;
}
实例化PropertyTokenHolder
调用getPropertyValue,获得该属性所对应的对象–> propValue.代码如下:

protected Object getPropertyValue(PropertyTokenHolder tokens) throws BeansException {
String propertyName = tokens.canonicalName;
String actualName = tokens.actualName;
// 1. 根据属性值获得PropertyHandler
PropertyHandler ph = getLocalPropertyHandler(actualName);
// 2. 如果PropertyHandler等于null或者没有get方法,抛出异常
if (ph == null || !ph.isReadable()) {
throw new NotReadablePropertyException(getRootClass(), this.nestedPath + propertyName);
}
// 3. 获得对应的属性值.
Object value = ph.getValue();
if (tokens.keys != null) {
// 4. 如果tokens.keys 不等于null,这里是不会执行的
if (value == null) {
// 4.1 如果autoGrowNestedPaths 值为true,则生成默认值,一般都会生成默认值的
if (isAutoGrowNestedPaths()) {
value = setDefaultValue(tokens.actualName);
}
else {
// 4.2 抛出异常
throw new NullValueInNestedPathException(getRootClass(), this.nestedPath + propertyName,
"Cannot access indexed value of property referenced in indexed " +
"property path '" + propertyName + "': returned null");
}
}
String indexedPropertyName = tokens.actualName;
// apply indexes and map keys
// 5. 依次进行遍历
for (int i = 0; i < tokens.keys.length; i++) {
String key = tokens.keys[i];
// 5.1 如果value等于null,则抛出异常
if (value == null) {
throw new NullValueInNestedPathException(getRootClass(), this.nestedPath + propertyName,
"Cannot access indexed value of property referenced in indexed " +
"property path '" + propertyName + "': returned null");
}
else if (value.getClass().isArray()) {
// 省略...
}
else if (value instanceof List) {
int index = Integer.parseInt(key);
List<Object> list = (List<Object>) value;
growCollectionIfNecessary(list, index, indexedPropertyName, ph, i + 1);
value = list.get(index);
}
else if (value instanceof Set) {
// 省略...
}
else if (value instanceof Map) {
// 省略...
}
else {
throw new InvalidPropertyException(getRootClass(), this.nestedPath + propertyName,
"Property referenced in indexed property path '" + propertyName +
"' is neither an array nor a List nor a Set nor a Map; returned value was [" + value + "]");
}
indexedPropertyName += PROPERTY_KEY_PREFIX + key + PROPERTY_KEY_SUFFIX;
}
}
return value;
}
获得调用getLocalPropertyHandler获得PropertyHandler,这个我们前面已经分析过了.
如果PropertyHandler等于null或者没有get方法,抛出NotReadablePropertyException
获得对应的属性对象,也就是People 中的address.
如果tokens.keys 不等于null,对于当前来说,keys 不等于null,因此是会执行的.

如果autoGrowNestedPaths 值为true,则生成默认值,一般都会生成默认值的,否则抛出NullValueInNestedPathException.
依次进行遍历keys,针对value的不同类型做不同的处理,这里我们只看List,处理如下：
将key 转为index,注意这里传入的是0.
通过list.get(index)的方式获得下标所对应的对象.
如果propValue 等于null,如果autoGrowNestedPaths 属性值为true,则调用setDefaultValue 进行实例化,否则抛出NullValueInNestedPathException异常,一般情况下, autoGrowNestedPaths为true.同样,该方法一般情况下都会执行的.代码如下:

private PropertyValue createDefaultPropertyValue(PropertyTokenHolder tokens) {
TypeDescriptor desc = getPropertyTypeDescriptor(tokens.canonicalName);
Class<?> type = desc.getType();
if (type == null) {
throw new NullValueInNestedPathException(getRootClass(), this.nestedPath + tokens.canonicalName,
"Could not determine property type for auto-growing a default value");
}
Object defaultValue = newValue(type, desc, tokens.canonicalName);
return new PropertyValue(tokens.canonicalName, defaultValue);
}

这样就实例化了，具体是怎么实例化的,这里就不展开了.

针对propValue的类型做不同的处理,如果该类型不是数字,List,Map,则抛出InvalidPropertyException.这里我们只分析list的情况,其他类似.

获得属性对应的对象
获得集合的泛型
获得下标
进行转换
如果下标大于集合的size,则将index - size 这段范围内,插入null值,最后在插入对应的值.否则,直接插入即可.
至此,集合的注入就分析完了.

对象导航处理
我们将配置文件改为如下:

com.example.demo.phone.number=1111111
1
这里分析对People 中 phone 的number 属性的注入. 还是同样的套路,前面的准备工作都一样,最终在进行属性注入时,调用了AbstractNestablePropertyAccessor#setPropertyValue.在该方法中调用了getPropertyAccessorForPropertyPath 用于处理嵌套属性.代码如下:

protected AbstractNestablePropertyAccessor getPropertyAccessorForPropertyPath(String propertyPath) {
// 1. 通过PropertyAccessorUtils#getFirstNestedPropertySeparatorIndex 获得下标
int pos = PropertyAccessorUtils.getFirstNestedPropertySeparatorIndex(propertyPath);
// Handle nested properties recursively.
if (pos > -1) {
// 如果存在的话,则意味着有嵌套存在,则递归处理,例如 map[my.key],
String nestedProperty = propertyPath.substring(0, pos);// nestedProperty = map[my
String nestedPath = propertyPath.substring(pos + 1); // nestedPath = key
// 3. 获得嵌套对象
AbstractNestablePropertyAccessor nestedPa = getNestedPropertyAccessor(nestedProperty);
// 4. 获取AbstractNestablePropertyAccessor,递归调用
return nestedPa.getPropertyAccessorForPropertyPath(nestedPath);
}
else {
// 如果不存在,则返回this
return this;
}
}
2件事:

通过PropertyAccessorUtils#getFirstNestedPropertySeparatorIndex 获得下标.该方法最终调用了PropertyAccessorUtils#getNestedPropertySeparatorIndex,该方法处理的逻辑很简单,看是否存在.,代码如下:

private static int getNestedPropertySeparatorIndex(String propertyPath, boolean last) {
boolean inKey = false;
int length = propertyPath.length();
int i = (last ? length - 1 : 0);// 起始下标
while (last ? i >= 0 : i < length) {
switch (propertyPath.charAt(i)) {
case PropertyAccessor.PROPERTY_KEY_PREFIX_CHAR: // [
case PropertyAccessor.PROPERTY_KEY_SUFFIX_CHAR:// ]
inKey = !inKey;
break;
case PropertyAccessor.NESTED_PROPERTY_SEPARATOR_CHAR: // .
if (!inKey) {
return i;
}
}
if (last) {
i--;
}
else {
i++;
}
}
return -1;
}
如果存在嵌套属性,则递归处理

通过字符串截取,获得nestedProperty,nestedPath .对于当前来说, nestedProperty = phone, nestedPath = number
调用getNestedPropertyAccessor 获得AbstractNestablePropertyAccessor.代码如下:

private AbstractNestablePropertyAccessor getNestedPropertyAccessor(String nestedProperty) {
// 1. 如果nestedPropertyAccessors 等于null,则实例化
if (this.nestedPropertyAccessors == null) {
this.nestedPropertyAccessors = new HashMap<String, AbstractNestablePropertyAccessor>();
}
// Get value of bean property.
// 2. 获取属性名
PropertyTokenHolder tokens = getPropertyNameTokens(nestedProperty);
String canonicalName = tokens.canonicalName;
// 3. 获得对应的值
Object value = getPropertyValue(tokens);
if (value == null || (value.getClass() == javaUtilOptionalClass && OptionalUnwrapper.isEmpty(value))) {
if (isAutoGrowNestedPaths()) {
value = setDefaultValue(tokens);
}
else {
throw new NullValueInNestedPathException(getRootClass(), this.nestedPath + canonicalName);
}
}

// Lookup cached sub-PropertyAccessor, create new one if not found.
// 4. 获得访问嵌套对象
AbstractNestablePropertyAccessor nestedPa = this.nestedPropertyAccessors.get(canonicalName);
if (nestedPa == null || nestedPa.getWrappedInstance() !=
(value.getClass() == javaUtilOptionalClass ? OptionalUnwrapper.unwrap(value) : value)) {
if (logger.isTraceEnabled()) {
logger.trace("Creating new nested " + getClass().getSimpleName() + " for property '" + canonicalName + "'");
}
// 5. 如果不存在则创建一个,实例化的是BeanWrapperImpl
nestedPa = newNestedPropertyAccessor(value, this.nestedPath + canonicalName + NESTED_PROPERTY_SEPARATOR);
// Inherit all type-specific PropertyEditors.
copyDefaultEditorsTo(nestedPa);
copyCustomEditorsTo(nestedPa, canonicalName);
// 6. 存入缓存
this.nestedPropertyAccessors.put(canonicalName, nestedPa);
}
else {
if (logger.isTraceEnabled()) {
logger.trace("Using cached nested property accessor for property '" + canonicalName + "'");
}
}
return nestedPa;
}
6件事:

如果nestedPropertyAccessors 等于null,则实例化. lazy-init
调用getPropertyNameTokens 获得PropertyTokenHolder,对于当前,获得的是phone所对应的PropertyTokenHolder.这个方法,我们之前已经分析过了。
调用getPropertyValue , 获得phone所对应的对象。关于这个方法,我们也已经分析过了,此时会将people中的phone 实例化.
尝试从nestedPropertyAccessors缓存中获得AbstractNestablePropertyAccessor. 如果没有获得,则实例化一个BeanPropertyHandler.然后进行初始化后放入nestedPropertyAccessors.
递归调用getPropertyAccessorForPropertyPath.
那我们的例子来说,第一次传入的参数是phone.number,有嵌套属性,因此会在实例化phone所对应后的AbstractNestablePropertyAccessor后,会递归调用getPropertyAccessorForPropertyPath,此时由于传入的参数是number,因此方法退出,因此该递归最终返回的是 phone所对应后的AbstractNestablePropertyAccessor.

接着,AbstractNestablePropertyAccessor#getFinalPath,获得最终的路径,代码如下:

protected String getFinalPath(AbstractNestablePropertyAccessor pa, String nestedPath) {
if (pa == this) {
return nestedPath;
}
return nestedPath.substring(PropertyAccessorUtils.getLastNestedPropertySeparatorIndex(nestedPath) + 1);
}
由于pa 不等于this,因此会调用PropertyAccessorUtils#getLastNestedPropertySeparatorIndex 方法获得最后一个. 所对应的下标,通过字符串截取后,获得属性名,此时,会获得number。

2个问题:

为什么pa 不等于 this?

还是拿例子来说话,this, 指的是people 所对应的AbstractNestablePropertyAccessor,pa 在当前来说,是phone所对应的AbstractNestablePropertyAccessor.明显不相等的.

为什么只需返回最后一个属性,就行了? 也就是

假如我们新增如下一个类型:

public class Operator {// 运营商

private String name;

// get set 忽略,自己加上即可..
}
然后将Phone 改为如下:

public class Phone {

private String number;
private Operator operator;

// get set 忽略,自己加上即可..
}

将配置文件加入如下配置:

com.example.demo.phone.operator.name=移动
1
为什么此时返回是name？

理由很简单,因为在调用AbstractNestablePropertyAccessor#getPropertyAccessorForPropertyPath时是递归处理的,该方法会首先实例化People 中的phone,接着实例化Phone 中operator所对应的Operator对象.后续的故事,就是直接赋值了,我们已经分析过了.

属性转换处理
这里,我们来看最后一个–>属性转换,将配置文件该为如下:

com.example.demo.age=11
1
之前的准备工作,就不在赘述了,在最终进行赋值时,会调用
AbstractNestablePropertyAccessor#processLocalProperty,而在该方法中的第三步,会调用AbstractNestablePropertyAccessor#convertForProperty进行转换处理,代码如下:

protected Object convertForProperty(String propertyName, Object oldValue, Object newValue, TypeDescriptor td)
throws TypeMismatchException {

return convertIfNecessary(propertyName, oldValue, newValue, td.getType(), td);
}

最终调用TypeConverterDelegate#convertIfNecessary,代码如下:

1
获得自定义的PropertyEditor
从propertyEditorRegistry 获得自定义的ConversionService,这里使用的是org.springframework.boot.bind.RelaxedConversionService
如果PropertyEditor 等于null && conversionService 不等于null,&& newValue 不等于null,&& typeDescriptor 不等于null,则调用ConversionService#convert.

…

这里由于不存在自定义的PropertyEditor,同时第2步获得的propertyEditorRegistry不等于null,因此最终会调用RelaxedConversionService#convert 进行转换,代码如下:

public Object convert(Object source, TypeDescriptor sourceType,
TypeDescriptor targetType) {
if (this.conversionService != null) {
try {
return this.conversionService.convert(source, sourceType, targetType);
}
catch (ConversionFailedException ex) {
// Ignore and try the additional converters
}
}
return this.additionalConverters.convert(source, sourceType, targetType);
}
2件事:

如果conversionService 不等于null,则调用conversionService#convert 进行转换.对于当前,会执行这里, conversionService为GenericConversionService,代码如下:

public Object convert(Object source, TypeDescriptor sourceType, TypeDescriptor targetType) {
Assert.notNull(targetType, "Target type to convert to cannot be null");
// 1. 如果sourceType 等于null,则抛出ConversionFailedException
if (sourceType == null) {
Assert.isTrue(source == null, "Source must be [null] if source type == [null]");
return handleResult(null, targetType, convertNullSource(null, targetType));
}
// 2. 如果source不等于null，并且sourceType 不是source 的类型,则抛出IllegalArgumentException
if (source != null && !sourceType.getObjectType().isInstance(source)) {
throw new IllegalArgumentException("Source to convert from must be an instance of [" +
sourceType + "]; instead it was a [" + source.getClass().getName() + "]");
}
// 3. 获得GenericConverter
GenericConverter converter = getConverter(sourceType, targetType);
if (converter != null) {
// 3.1 如果Converter,则通过ConversionUtils#invokeConverter 进行转换
Object result = ConversionUtils.invokeConverter(converter, source, sourceType, targetType);
return handleResult(sourceType, targetType, result);
}
// 4. 当Converter 没有找到时 ,进行处理
return handleConverterNotFound(source, sourceType, targetType);
}

4件事:

如果sourceType 等于null,则抛出ConversionFailedException
如果source不等于null，并且sourceType 不是source 的类型,则抛出IllegalArgumentException
获得GenericConverter,如果Converter 不等于null,则通过ConversionUtils#invokeConverter 进行转换.代码如下:

protected GenericConverter getConverter(TypeDescriptor sourceType, TypeDescriptor targetType) {
// 1. 实例化ConverterCacheKey
ConverterCacheKey key = new ConverterCacheKey(sourceType, targetType);
// 2. 尝试从converterCache 获取
GenericConverter converter = this.converterCache.get(key);
if (converter != null) {
return (converter != NO_MATCH ? converter : null);
}

// 3. 从converters 获取
converter = this.converters.find(sourceType, targetType);
if (converter == null) {
// 4. 如果还没有得到,则返回默认的Converter
converter = getDefaultConverter(sourceType, targetType);
}

if (converter != null) {
// 5. 如果不等于null,则放入缓存中
this.converterCache.put(key, converter);
return converter;
}

// 6. 如果converter 等于null,则在converterCache中放入NO_MATCH
this.converterCache.put(key, NO_MATCH);
return null;
}
6件事:

实例化ConverterCacheKey
尝试从converterCache 获取
从converters 获取
如果还没有得到,则返回默认的Converter
如果不等于null,则放入缓存中
如果converter 等于null,则在converterCache中放入NO_MATCH
对于当前,获得的是ConverterFactoryAdapter,其convert方法如下:

public Object convert(Object source, TypeDescriptor sourceType, TypeDescriptor targetType) {
if (source == null) {
return convertNullSource(sourceType, targetType);
}
return this.converterFactory.getConverter(targetType.getObjectType()).convert(source);
}
最终调用的是StringToNumber#convert 方法,代码如下:

public T convert(String source) {
if (source.isEmpty()) {
return null;
}
return NumberUtils.parseNumber(source, this.targetType);
}
至此,就将com.example.demo.age = 11 ,由原先的字符串,转换为了Integer.后面只需赋值即可了,关于这个,我们已经分析过了.

当Converter 没有找到时 ,进行处理
否则调用additionalConverters#convert 进行转换。