ReactJS分析之入口函数render
ReactJS分析之入口函数render
前言
在使用React进行构建应用时，我们总会有一个步骤将组建或者虚拟DOM元素渲染到真实的DOM上，将任务交给浏览器，进而进行layout和paint等步骤，这个函数就是React.render()。首先看下该函数的接口定义：
ReactComponent render( ReactElement element, DOMElement container, [function callback] )
接收2-3个参数，并返回ReactComponent类型的对象，当组件被添加到DOM中后，执行回调。在这里涉及到了两个React类型--ReactComponent和ReactElement，着重分析。
ReactElement类型解读
ReactElement类型通过函数React.createElement()创建，接口定义如下：
ReactElement createElement( string/ReactClass type, [object props], [children ...] )
第一个参数可以接受字符串（如“p”，“div”等HTML的tag）或ReactClass，第二个参数为传递的参数，第三个为子元素，可以为字符串和ReactElement。
下面着重分析createElement的具体实现：
ReactElement.createElement = function(type, config, children) {
var propName;
// Reserved names are extracted
var props = {};
var key = null;
var ref = null;
if (config != null) {
ref = config.ref === undefined ? null : config.ref;
key = config.key === undefined ? null : '' + config.key;
// Remaining properties are added to a new props object
for (propName in config) {
if (config.hasOwnProperty(propName) &&
!RESERVED_PROPS.hasOwnProperty(propName)) {
props[propName] = config[propName];
}
}
}
// Children can be more than one argument, and those are transferred onto
// the newly allocated props object.
var childrenLength = arguments.length - 2;
if (childrenLength === 1) {
props.children = children;
} else if (childrenLength > 1) {
var childArray = Array(childrenLength);
for (var i = 0; i < childrenLength; i++) {
childArray[i] = arguments[i + 2];
}
props.children = childArray;
}
// Resolve default props
if (type && type.defaultProps) {
var defaultProps = type.defaultProps;
for (propName in defaultProps) {
if (typeof props[propName] === 'undefined') {
props[propName] = defaultProps[propName];
}
}
}
return new ReactElement(
type,
key,
ref,
ReactCurrentOwner.current,
ReactContext.current,
props
);
};
var ReactElement = function(type, key, ref, owner, context, props) {
// Built-in properties that belong on the element
this.type = type;
this.key = key;
this.ref = ref;
// Record the component responsible for creating this element.
this._owner = owner;
// TODO: Deprecate withContext, and then the context becomes accessible
// through the owner.
this._context = context;
if ("production" !== process.env.NODE_ENV) {
// The validation flag and props are currently mutative. We put them on
// an external backing store so that we can freeze the whole object.
// This can be replaced with a WeakMap once they are implemented in
// commonly used development environments.
this._store = {props: props, originalProps: assign({}, props)};
// To make comparing ReactElements easier for testing purposes, we make
// the validation flag non-enumerable (where possible, which should
// include every environment we run tests in), so the test framework
// ignores it.
try {
Object.defineProperty(this._store, 'validated', {
configurable: false,
enumerable: false,
writable: true
});
} catch (x) {
}
this._store.validated = false;
// We're not allowed to set props directly on the object so we early
// return and rely on the prototype membrane to forward to the backing
// store.
if (useMutationMembrane) {
Object.freeze(this);
return;
}
}
this.props = props;
};
在ReactElement.js中实现了该方法，首先保存传入的参数，其中ref和key这两个参数比较特别，ref用于父组件引用子组件的真实DOM，key用于调和算法，判断该组件是否update或remove；保存children到props中，并根据type是否有defaultProps属性对props进行mixin；最后创建ReactElement实例。其中reactElement有个实例属性_owner，用于保存所属的组件。
ReactElement的原型对象只有一个简单的方法用于判断是否是ReactElement对象，没有额外的方法。
综上，我们可以看出ReactElement有4个属性：type，ref，key，props，并且轻量，没有状态，是一个虚拟化的DOM元素。
ReactClass类型解读
React的核心是ReactElement类型，但是精髓确实ReactComponent，即组件。但是组件的创建却并不简单，我们通过React.createClass创建ReactClass类，它是ReactComponent的构造函数，不同于正常的对象创建，组件的创建由React接管，即我们无须对其实例化（new MyComponent（））。相对于ReactElement的无状态，ReactComponent是有状态的，先看接口定义：
ReactClass createClass(object specification)
传入的spec参数必须包含render方法，用于渲染虚拟DOM，render返回ReactElement类型；另外还有一些getInitialState和生命周期方法，可以根据需要定义。
下面根据createClass的实现来深入分析：
createClass: function(spec) {
var Constructor = function(props, context) {
// Wire up auto-binding
if (this.__reactAutoBindMap) {
bindAutoBindMethods(this);
}
this.props = props;
this.context = context;
this.state = null;
// ReactClasses doesn't have constructors. Instead, they use the
// getInitialState and componentWillMount methods for initialization.
var initialState = this.getInitialState ? this.getInitialState() : null;
this.state = initialState;
};
Constructor.prototype = new ReactClassComponent();
Constructor.prototype.constructor = Constructor;
injectedMixins.forEach(
mixSpecIntoComponent.bind(null, Constructor)
);
mixSpecIntoComponent(Constructor, spec);
// Initialize the defaultProps property after all mixins have been merged
if (Constructor.getDefaultProps) {
Constructor.defaultProps = Constructor.getDefaultProps();
}
// Reduce time spent doing lookups by setting these on the prototype.
for (var methodName in ReactClassInterface) {
if (!Constructor.prototype[methodName]) {
Constructor.prototype[methodName] = null;
}
}
// Legacy hook
Constructor.type = Constructor;
return Constructor;
}
// Constructor的原型
var ReactClassComponent = function() {};
// assign类似于mixin
assign(
ReactClassComponent.prototype,
ReactComponent.prototype,
ReactClassMixin
);
// mixin到Constructor的原型上
function mixSpecIntoComponent(Constructor, spec) {
if (!spec) {
return;
}
var proto = Constructor.prototype;
// By handling mixins before any other properties, we ensure the same
// chaining order is applied to methods with DEFINE_MANY policy, whether
// mixins are listed before or after these methods in the spec.
if (spec.hasOwnProperty(MIXINS_KEY)) {
RESERVED_SPEC_KEYS.mixins(Constructor, spec.mixins);
}
for (var name in spec) {
if (!spec.hasOwnProperty(name)) {
continue;
}
if (name === MIXINS_KEY) {
// We have already handled mixins in a special case above
continue;
}
var property = spec[name];
validateMethodOverride(proto, name);
if (RESERVED_SPEC_KEYS.hasOwnProperty(name)) {
RESERVED_SPEC_KEYS[name](Constructor, property);
} else {
// Setup methods on prototype:
// The following member methods should not be automatically bound:
// 1. Expected ReactClass methods (in the "interface").
// 2. Overridden methods (that were mixed in).
var isReactClassMethod =
ReactClassInterface.hasOwnProperty(name);
var isAlreadyDefined = proto.hasOwnProperty(name);
var markedDontBind = property && property.__reactDontBind;
var isFunction = typeof property === 'function';
var shouldAutoBind =
isFunction &&
!isReactClassMethod &&
!isAlreadyDefined &&
!markedDontBind;
if (shouldAutoBind) {
if (!proto.__reactAutoBindMap) {
proto.__reactAutoBindMap = {};
}
proto.__reactAutoBindMap[name] = property;
proto[name] = property;
} else {
if (isAlreadyDefined) {
var specPolicy = ReactClassInterface[name];
// For methods which are defined more than once, call the existing
// methods before calling the new property, merging if appropriate.
if (specPolicy === SpecPolicy.DEFINE_MANY_MERGED) {
proto[name] = createMergedResultFunction(proto[name], property);
} else if (specPolicy === SpecPolicy.DEFINE_MANY) {
proto[name] = createChainedFunction(proto[name], property);
}
} else {
proto[name] = property;
if ("production" !== process.env.NODE_ENV) {
// Add verbose displayName to the function, which helps when looking
// at profiling tools.
if (typeof property === 'function' && spec.displayName) {
proto[name].displayName = spec.displayName + '_' + name;
}
}
}
}
}
}
}
createClass返回一个Constructor构造函数，它的原型是new ReactClassComponent()对象，该对象有mixin的组件的方法(在spec对象中的mixins属性的对象的方法)和ReactComponent的方法（setState和forceUpdate），并且在mixSpecIntoComponent(Constructor, spec)方法中将spec中实现的方法绑定到Constructor的原型上，在这里对于非React提供的方法（即个人实现的一些功能函数或者事件处理函数）保存在原型的__reactAutoBindMap的属性上。最后再设置Constructor的defaultProps和type（Constructor.type = Constructor）。
在上节中提到了createElement的第一个参数可以是ReactClass，因此在Constructor实现上赋予了type和defaultProps属性。
React的入口—React.render()
React.render的实现是在ReactMount中，我们通过源码进行进一步的分析。
render: function(nextElement, container, callback) {
var prevComponent = instancesByReactRootID[getReactRootID(container)];
if (prevComponent) {
var prevElement = prevComponent._currentElement;
if (shouldUpdateReactComponent(prevElement, nextElement)) {
return ReactMount._updateRootComponent(
prevComponent,
nextElement,
container,
callback
).getPublicInstance();
} else {
ReactMount.unmountComponentAtNode(container);
}
}
var reactRootElement = getReactRootElementInContainer(container);
var containerHasReactMarkup =
reactRootElement && ReactMount.isRenderedByReact(reactRootElement);
var shouldReuseMarkup = containerHasReactMarkup && !prevComponent;
var component = ReactMount._renderNewRootComponent(
nextElement,
container,
shouldReuseMarkup
).getPublicInstance();
if (callback) {
callback.call(component);
}
return component;
}
如果是第一次挂载该ReactElement，直接添加即可；如果之前已挂载过，则通过instancesByReactRootID获取渲染之前container的旧组件，即prevComponent，具体通过获取container的firstChild，并根据缓存获取该对象对应的id，并根据id得到prevComponent。每个component对象都有对应的虚拟DOM，即ReactElement，通过shouldUpdateReactComponent(prevElement, nextElement）进行判断对组件进行update还是delete。
具体shouldUpdateReactComponent的比较算法是：如果prevElement类型为string或者number，那么nextElement类型为string或number时为true；如果prevElement和nextElement为object，并且key和type属性相同，则prevElement._owner == nextElement._owner相等时为true，否则为false。
如果需要更新，则调用ReactMount.._updateRootComponent函数进行Reconciliation，并返回该组件；否则删除该组件，具体操作则是删除container的所有子元素。然后判断shouldReuseMarkup，对于初次挂载的ReactElement而言，该标记为false。最后通过调用_renderNewRootComponent方法将ReactElement渲染到DOM上，并获取对应的ReactComponent对象，最后执行回调并返回组件对象。
对于_renderNewRootComponent方法，通过调用instantiateReactComponent(nextElement, null)来实例化组件，并在ReactMount的缓存中注册组件，批量执行更新ReactUpdates.batchedUpdates，最终通过_mountImageIntoNode方法将虚拟节点插入到DOM中。
至此，React中比较重要的方法讲解完毕。下一步计划是分析组件的实例化过程，敬请期待。