ButterKnife编译时生成代码原理：butterknife-compiler源码分析

1.butterknife-compiler介绍

上篇文章：注解框架源码分析（XUtils、ButterKnife），根据代码运行流程分析了xUtils和ButterKnife，ButterKnife最终实现注解方法的代码是通过编译运行时生成的，也就是gradle依赖中butterknife-compiler实现的：

dependencies {
  implementation 'com.jakewharton:butterknife:8.8.1'
  annotationProcessor 'com.jakewharton:butterknife-compiler:8.8.1'
}

com.jakewharton:butterknife-compiler 就是自定义的注解处理器，我们在 Gradle 中注册使用它。
然而我在项目结构中找了很久也没有找到这个库的文件，有可能是在编译时才去访问的，如果需要可以在 GitHub 中找到：
butterknife-compiler

2.butterknife-compiler源码分析

我们看看ButterKnifeProcessor这个类：

@Override public synchronized void init(ProcessingEnvironment env) {
    super.init(env);

    String sdk = env.getOptions().get(OPTION_SDK_INT);
    if (sdk != null) {
      try {
        this.sdk = Integer.parseInt(sdk);
      } catch (NumberFormatException e) {
        env.getMessager()
            .printMessage(Kind.WARNING, "Unable to parse supplied minSdk option '"
                + sdk
                + "'. Falling back to API 1 support.");
      }
    }

    debuggable = !"false".equals(env.getOptions().get(OPTION_DEBUGGABLE));

    elementUtils = env.getElementUtils();
    typeUtils = env.getTypeUtils();
    filer = env.getFiler();
    try {
      trees = Trees.instance(processingEnv);
    } catch (IllegalArgumentException ignored) {
    }
  }

int()方法里面进来判断了最低的支持的sdk版本。ProcessingEnviroment参数提供很多有用的工具类Elements, Types和Filer。Types是用来处理TypeMirror的工具类，Filer用来创建生成辅助文件。至于ElementUtils嘛，其实ButterKnifeProcessor在运行的时候，会扫描所有的Java源文件，然后每一个Java源文件的每一个部分都是一个Element，比如一个包、类或者方法。

@Override public Set getSupportedAnnotationTypes() {
    Set types = new LinkedHashSet<>();
    for (Class annotation : getSupportedAnnotations()) {
      types.add(annotation.getCanonicalName());
    }
    return types;
  }

  private Set> getSupportedAnnotations() {
    Set> annotations = new LinkedHashSet<>();

    annotations.add(BindAnim.class);
    annotations.add(BindArray.class);
    annotations.add(BindBitmap.class);
    annotations.add(BindBool.class);
    annotations.add(BindColor.class);
    annotations.add(BindDimen.class);
    annotations.add(BindDrawable.class);
    annotations.add(BindFloat.class);
    annotations.add(BindFont.class);
    annotations.add(BindInt.class);
    annotations.add(BindString.class);
    annotations.add(BindView.class);
    annotations.add(BindViews.class);
    annotations.addAll(LISTENERS);

    return annotations;
  }

getSupportedAnnotationTypes()方法主要是指定ButterknifeProcessor是注册给哪些注解的。我们可以看到，在源代码里面，作者一个一个地把Class文件加到那个LinkedHashSet里面，然后再把LISTENERS也全部加进去。

其实整个类最重要的是process方法：

@Override public boolean process(Set elements, RoundEnvironment env) {
    Map bindingMap = findAndParseTargets(env);

    for (Map.Entry entry : bindingMap.entrySet()) {
      TypeElement typeElement = entry.getKey();
      BindingSet binding = entry.getValue();

      JavaFile javaFile = binding.brewJava(sdk, debuggable);
      try {
        javaFile.writeTo(filer);
      } catch (IOException e) {
        error(typeElement, "Unable to write binding for type %s: %s", typeElement, e.getMessage());
      }
    }

    return false;
  }

这个方法的作用主要是扫描、评估和处理我们程序中的注解，然后生成Java文件，也就是前面说的MainActivity_ViewBinding。首先一进这个函数就调用了findAndParseTargets方法，我们就去看看findAndParseTargets方法到底做了什么：

private Map findAndParseTargets(RoundEnvironment env) {
    Map builderMap = new LinkedHashMap<>();
    Set erasedTargetNames = new LinkedHashSet<>();

    scanForRClasses(env);

    // Process each @BindAnim element.
    for (Element element : env.getElementsAnnotatedWith(BindAnim.class)) {
      if (!SuperficialValidation.validateElement(element)) continue;
      try {
        parseResourceAnimation(element, builderMap, erasedTargetNames);
      } catch (Exception e) {
        logParsingError(element, BindAnim.class, e);
      }
    }
........
........
........
// Associate superclass binders with their subclass binders. This is a queue-based tree walk
    // which starts at the roots (superclasses) and walks to the leafs (subclasses).
    Deque> entries =
        new ArrayDeque<>(builderMap.entrySet());
    Map bindingMap = new LinkedHashMap<>();
    while (!entries.isEmpty()) {
      Map.Entry entry = entries.removeFirst();

      TypeElement type = entry.getKey();
      BindingSet.Builder builder = entry.getValue();

      TypeElement parentType = findParentType(type, erasedTargetNames);
      if (parentType == null) {
        bindingMap.put(type, builder.build());
      } else {
        BindingSet parentBinding = bindingMap.get(parentType);
        if (parentBinding != null) {
          builder.setParent(parentBinding);
          bindingMap.put(type, builder.build());
        } else {
          // Has a superclass binding but we haven't built it yet. Re-enqueue for later.
          entries.addLast(entry);
        }
      }
    }

    return bindingMap;

这个方法的代码非常多，这里只贴出一部分，这个方法的主要的流程如下：

扫描所有具有注解的类，然后根据这些类的信息生成BindingSet，最后生成以TypeElement为键,BindingSet为值的键值对。
循环遍历这个键值对，根据TypeElement和BindingSet里面的信息生成对应的java类。例如AnnotationActivity生成的类即为MainActivity_ViewBinding类。

这里我们可以看看BindingSet里面的代码：

final class BindingSet {
  static final ClassName UTILS = ClassName.get("butterknife.internal", "Utils");
  private static final ClassName VIEW = ClassName.get("android.view", "View");
  private static final ClassName CONTEXT = ClassName.get("android.content", "Context");
  private static final ClassName RESOURCES = ClassName.get("android.content.res", "Resources");
  private static final ClassName UI_THREAD =
      ClassName.get("android.support.annotation", "UiThread");
  private static final ClassName CALL_SUPER =
      ClassName.get("android.support.annotation", "CallSuper");
  private static final ClassName SUPPRESS_LINT =
      ClassName.get("android.annotation", "SuppressLint");
  private static final ClassName UNBINDER = ClassName.get("butterknife", "Unbinder");
  static final ClassName BITMAP_FACTORY = ClassName.get("android.graphics", "BitmapFactory");
  static final ClassName CONTEXT_COMPAT =
      ClassName.get("android.support.v4.content", "ContextCompat");
  static final ClassName ANIMATION_UTILS =
          ClassName.get("android.view.animation", "AnimationUtils");

  private final TypeName targetTypeName;
  private final ClassName bindingClassName;
  private final boolean isFinal;
  private final boolean isView;
  private final boolean isActivity;
  private final boolean isDialog;
  private final ImmutableList viewBindings;
  private final ImmutableList collectionBindings;
  private final ImmutableList resourceBindings;
  private final BindingSet parentBinding;

  private BindingSet(TypeName targetTypeName, ClassName bindingClassName, boolean isFinal,
      boolean isView, boolean isActivity, boolean isDialog, ImmutableList viewBindings,
      ImmutableList collectionBindings,
      ImmutableList resourceBindings, BindingSet parentBinding) {
    this.isFinal = isFinal;
    this.targetTypeName = targetTypeName;
    this.bindingClassName = bindingClassName;
    this.isView = isView;
    this.isActivity = isActivity;
    this.isDialog = isDialog;
    this.viewBindings = viewBindings;
    this.collectionBindings = collectionBindings;
    this.resourceBindings = resourceBindings;
    this.parentBinding = parentBinding;
  }

  JavaFile brewJava(int sdk, boolean debuggable) {
    return JavaFile.builder(bindingClassName.packageName(), createType(sdk, debuggable))
        .addFileComment("Generated code from Butter Knife. Do not modify!")
        .build();
  }

  private TypeSpec createType(int sdk, boolean debuggable) {
    TypeSpec.Builder result = TypeSpec.classBuilder(bindingClassName.simpleName())
        .addModifiers(PUBLIC);
    if (isFinal) {
      result.addModifiers(FINAL);
    }
    ........
    ........
    ........

  static final class Builder {
    private final TypeName targetTypeName;
    private final ClassName bindingClassName;
    private final boolean isFinal;
    private final boolean isView;
    private final boolean isActivity;
    private final boolean isDialog;

    private BindingSet parentBinding;

    private final Map viewIdMap = new LinkedHashMap<>();
    private final ImmutableList.Builder collectionBindings =
        ImmutableList.builder();
    private final ImmutableList.Builder resourceBindings = ImmutableList.builder();

    private Builder(TypeName targetTypeName, ClassName bindingClassName, boolean isFinal,
        boolean isView, boolean isActivity, boolean isDialog) {
      this.targetTypeName = targetTypeName;
      this.bindingClassName = bindingClassName;
      this.isFinal = isFinal;
      this.isView = isView;
      this.isActivity = isActivity;
      this.isDialog = isDialog;
    }

    void addField(Id id, FieldViewBinding binding) {
      getOrCreateViewBindings(id).setFieldBinding(binding);
    }

    void addFieldCollection(FieldCollectionViewBinding binding) {
      collectionBindings.add(binding);
    }

    boolean addMethod(
        Id id,
        ListenerClass listener,
        ListenerMethod method,
        MethodViewBinding binding) {
      ViewBinding.Builder viewBinding = getOrCreateViewBindings(id);
      if (viewBinding.hasMethodBinding(listener, method) && !"void".equals(method.returnType())) {
        return false;
      }
      viewBinding.addMethodBinding(listener, method, binding);
      return true;
    }

    void addResource(ResourceBinding binding) {
      resourceBindings.add(binding);
    }

    void setParent(BindingSet parent) {
      this.parentBinding = parent;
    }

    String findExistingBindingName(Id id) {
      ViewBinding.Builder builder = viewIdMap.get(id);
      if (builder == null) {
        return null;
      }
      FieldViewBinding fieldBinding = builder.fieldBinding;
      if (fieldBinding == null) {
        return null;
      }
      return fieldBinding.getName();
    }

    private ViewBinding.Builder getOrCreateViewBindings(Id id) {
      ViewBinding.Builder viewId = viewIdMap.get(id);
      if (viewId == null) {
        viewId = new ViewBinding.Builder(id);
        viewIdMap.put(id, viewId);
      }
      return viewId;
    }

    BindingSet build() {
      ImmutableList.Builder viewBindings = ImmutableList.builder();
      for (ViewBinding.Builder builder : viewIdMap.values()) {
        viewBindings.add(builder.build());
      }
      return new BindingSet(targetTypeName, bindingClassName, isFinal, isView, isActivity, isDialog,
          viewBindings.build(), collectionBindings.build(), resourceBindings.build(),
          parentBinding);
    }
  }
}

这个类的代码也非常多，所以我们也只贴一部分，可以自己去看看源码，这个BindingSet是管理了所有关于这个注解的一些信息还有实例本身的信息。

因为我们之前用的例子是绑定的一个View，所以我们就只贴了解析View的代码。好吧，这里遍历了所有带有@BindView的Element，然后对每一个Element进行解析，也就进入了parseBindView这个方法中：

private void parseBindView(Element element, Map builderMap,
      Set erasedTargetNames) {
    TypeElement enclosingElement = (TypeElement) element.getEnclosingElement();

    // Start by verifying common generated code restrictions.
    boolean hasError = isInaccessibleViaGeneratedCode(BindView.class, "fields", element)
        || isBindingInWrongPackage(BindView.class, element);

    // Verify that the target type extends from View.
    TypeMirror elementType = element.asType();
    if (elementType.getKind() == TypeKind.TYPEVAR) {
      TypeVariable typeVariable = (TypeVariable) elementType;
      elementType = typeVariable.getUpperBound();
    }
    Name qualifiedName = enclosingElement.getQualifiedName();
    Name simpleName = element.getSimpleName();
    if (!isSubtypeOfType(elementType, VIEW_TYPE) && !isInterface(elementType)) {
      if (elementType.getKind() == TypeKind.ERROR) {
        note(element, "@%s field with unresolved type (%s) "
                + "must elsewhere be generated as a View or interface. (%s.%s)",
            BindView.class.getSimpleName(), elementType, qualifiedName, simpleName);
      } else {
        error(element, "@%s fields must extend from View or be an interface. (%s.%s)",
            BindView.class.getSimpleName(), qualifiedName, simpleName);
        hasError = true;
      }
    }

    if (hasError) {
      return;
    }

    // Assemble information on the field.
    int id = element.getAnnotation(BindView.class).value();

    BindingSet.Builder builder = builderMap.get(enclosingElement);
    QualifiedId qualifiedId = elementToQualifiedId(element, id);
    if (builder != null) {
      String existingBindingName = builder.findExistingBindingName(getId(qualifiedId));
      if (existingBindingName != null) {
        error(element, "Attempt to use @%s for an already bound ID %d on '%s'. (%s.%s)",
            BindView.class.getSimpleName(), id, existingBindingName,
            enclosingElement.getQualifiedName(), element.getSimpleName());
        return;
      }
    } else {
      builder = getOrCreateBindingBuilder(builderMap, enclosingElement);
    }

    String name = simpleName.toString();
    TypeName type = TypeName.get(elementType);
    boolean required = isFieldRequired(element);

    builder.addField(getId(qualifiedId), new FieldViewBinding(name, type, required));

    // Add the type-erased version to the valid binding targets set.
    erasedTargetNames.add(enclosingElement);
  }

然后这里从一进入这个方法到

int id = element.getAnnotation(BindView.class).value();

都是在拿到注解信息，然后验证注解的target的类型是否继承自view，然后上面这一行代码获得我们要绑定的View的id，再从builderMap里面取出BindingSet.Builder对象(这个BindingSet是管理了所有关于这个注解的一些信息还有实例本身的信息，其实最后是通过BindingSet来生成java代码的，上面也已经看了BindingSet的代码)，如果builderMap里面不存在的话，就在

builder = getOrCreateBindingBuilder(builderMap, enclosingElement);

这里生成一个，我们进去看一下getOrCreateBindingBuilder：

private BindingSet.Builder getOrCreateBindingBuilder(
      Map builderMap, TypeElement enclosingElement) {
    BindingSet.Builder builder = builderMap.get(enclosingElement);
    if (builder == null) {
      builder = BindingSet.newBuilder(enclosingElement);
      builderMap.put(enclosingElement, builder);
    }
    return builder;
  }

这里面其实很简单，就是获取一些这个注解所修饰的变量的一些信息，然后把这个解析后的builder加入到builderMap里面。

返回刚刚的parseBindView中，根据view的信息生成一个FieldViewBinding，最后添加到上边生成的builder实例中。这里基本完成了解析工作。最后回到findAndParseTargets中：

// Associate superclass binders with their subclass binders. This is a queue-based tree walk
    // which starts at the roots (superclasses) and walks to the leafs (subclasses).
    Deque> entries =
        new ArrayDeque<>(builderMap.entrySet());
    Map bindingMap = new LinkedHashMap<>();
    while (!entries.isEmpty()) {
      Map.Entry entry = entries.removeFirst();

      TypeElement type = entry.getKey();
      BindingSet.Builder builder = entry.getValue();

      TypeElement parentType = findParentType(type, erasedTargetNames);
      if (parentType == null) {
        bindingMap.put(type, builder.build());
      } else {
        BindingSet parentBinding = bindingMap.get(parentType);
        if (parentBinding != null) {
          builder.setParent(parentBinding);
          bindingMap.put(type, builder.build());
        } else {
          // Has a superclass binding but we haven't built it yet. Re-enqueue for later.
          entries.addLast(entry);
        }
      }
    }

这里主要的工作是建立上面的绑定的所有的实例的解绑的关系，因为我们绑定了，最后在代码中还是会解绑的。这里预先处理好了这些关系。
回到我们的process中， 现在解析完了annotation,该生成java文件了，再看看代码：

@Override public boolean process(Set elements, RoundEnvironment env) {
    Map bindingMap = findAndParseTargets(env);

    for (Map.Entry entry : bindingMap.entrySet()) {
      TypeElement typeElement = entry.getKey();
      BindingSet binding = entry.getValue();

      JavaFile javaFile = binding.brewJava(sdk, debuggable);
      try {
        javaFile.writeTo(filer);
      } catch (IOException e) {
        error(typeElement, "Unable to write binding for type %s: %s", typeElement, e.getMessage());
      }
    }

    return false;
  }

遍历刚刚得到的bindingMap，然后再一个一个地通过

javaFile.writeTo(filer);

来生成java文件。然而生成的java文件也是根据上面的信息来用字符串拼接起来的，然而这个工作在brewJava()中完成了：

JavaFile brewJava(int sdk, boolean debuggable) {
    return JavaFile.builder(bindingClassName.packageName(), createType(sdk, debuggable))
        .addFileComment("Generated code from Butter Knife. Do not modify!")
        .build();
  }

最后通过writeTo(Filer filer)生成java源文件。

自己简易打造的IOC注解框架：SteadyoungIOC