MapStruct分析

MapStruct在编译时会通过CheckStyle检查代码格式，一旦不符合要求，就会编译终止；所以为了在MapStruct加一些日志方便分析，可以先把CheckStley插件去掉。
位于processor模块：

<!--            <plugin>-->
<!--                <groupId>org.apache.maven.plugins</groupId>-->
<!--                <artifactId>maven-checkstyle-plugin</artifactId>-->
<!--                <executions>-->
<!--                    <execution>-->
<!--                        <id>check-style</id>-->
<!--                        <phase>verify</phase>-->
<!--                        <goals>-->
<!--                            <goal>checkstyle</goal>-->
<!--                        </goals>-->
<!--                    </execution>-->
<!--                </executions>-->
<!--            </plugin>-->

MapStruct-DEMO

使用MapStruct时，我们只定义了Interface，具体的Impl是有MapStruct帮我们实现并编译的，这些在我们代码编译时由MapStruct完成。所以要看MapStruct的工作日志，先要写一个简单的Demo。
注意:

将demo的MapStruct依赖换成我们拉取代码上的版本
将MapStruct和Demo的JDK设置成一样的，否则Demo编译的时候会报NPE
MapStruct Demo将FromDTO转成ToDTO

//FromDTO
public class FromDTO {public String name;
}
//ToDTO
public class ToDTO {public String name;
}
//mapper
@Mapper
public interface From2To {From2To from2to = Mappers.getMapper(From2To.class);ToDTO f2t(FromDTO from);
}

MapStruct入口

MapStruct时通过SPI的方式注册、暴露接口的

javax.annotation.processing.Processor(处理入口)

org.mapstruct.ap.MappingProcessor

org.mapstruct.ap.internal.processor.ModelElementProcessor

org.mapstruct.ap.internal.processor.CdiComponentProcessor
org.mapstruct.ap.internal.processor.Jsr330ComponentProcessor
org.mapstruct.ap.internal.processor.MapperCreationProcessor
org.mapstruct.ap.internal.processor.MapperRenderingProcessor
org.mapstruct.ap.internal.processor.MethodRetrievalProcessor
org.mapstruct.ap.internal.processor.SpringComponentProcessor
org.mapstruct.ap.internal.processor.MapperServiceProcessor

MapStruct的MapperCreationProcessor加了一些调试日志，编译demo输出如下

➜  demo mvn clean compile -q
MethodRetrievalProcessor test
MapperCreationProcessor test
From2To
[com.example.map.ToDTO f2t(com.example.map.FromDTO from)]
[com.example.map.ToDTO f2t(com.example.map.FromDTO from)]
MapperRenderingProcessor test
MapperServiceProcessor test

MapperProcessor

Processor		Priority
MethodRetrievalProcessor		1
MapperCreationProcessor		1000
SpringComponentProcessor	extends AnnotationBasedComponentModelProcessor	1100
Jsr330ComponentProcessor	同上	1100
CdiComponentProcessor	同上	1100
MapperRenderingProcessor		9999
MapperServiceProcessor	将生成的Impl写入文件	10000

MappingProcessor

MappingProcessor是处理入口，通过ServiceLoader加载各Mapper处理器，并按优先级排序，打日志看各处理器执行阶段。

➜  demo mvn clean compile -q
MethodRetrievalProcessor priority=1
MapperCreationProcessor  priority=1000
MapperRenderingProcessor priority=9999
MapperServiceProcessor   priority=10000

ServiceLoader加载各Mapper处理器：

    /*** Retrieves all model element processors, ordered by their priority value* (with the method retrieval processor having the lowest priority value (1)* and the code generation processor the highest priority value.** @return A list with all model element processors.*/private Iterable<ModelElementProcessor<?, ?>> getProcessors() {// TODO Re-consider which class loader to use in case processors are//类加载器加载的注解处理器// loaded from other modules, too@SuppressWarnings("rawtypes")Iterator<ModelElementProcessor> processorIterator = ServiceLoader.load(ModelElementProcessor.class,MappingProcessor.class.getClassLoader()).iterator();List<ModelElementProcessor<?, ?>> processors = new ArrayList<>();while ( processorIterator.hasNext() ) {processors.add( processorIterator.next() );}//按优先级排序Collections.sort( processors, new ProcessorComparator() );return processors;}

按优先级依次执行各MapperProcessor

    /*** Applies all registered {@link ModelElementProcessor}s to the given mapper* type.** @param context The processor context.* @param mapperTypeElement The mapper type element.*/private void processMapperTypeElement(ProcessorContext context, TypeElement mapperTypeElement) {Object model = null;//按优先级依次执行各processorfor ( ModelElementProcessor<?, ?> processor : getProcessors() ) {try {model = process( context, processor, mapperTypeElement, model );}catch ( AnnotationProcessingException e ) {processingEnv.getMessager().printMessage(Kind.ERROR,e.getMessage(),e.getElement(),e.getAnnotationMirror(),e.getAnnotationValue());break;}}}private <P, R> R process(ProcessorContext context, ModelElementProcessor<P, R> processor,TypeElement mapperTypeElement, Object modelElement) {@SuppressWarnings("unchecked")P sourceElement = (P) modelElement;return processor.process( context, mapperTypeElement, sourceElement );}

参考

深入理解SPI机制
你应该了解的 Java SPI 机制
mapstruct源码分析
注解入坑笔记：关于注解使用必须了解的——Annotation、AbstraceProcessor、APT
注解处理器是干嘛的
编译时注解（一）AbstractProcessor实战