IntersectionObserver 可以轻松判断元素是否可见,在之前的 精读《用 React 做按需渲染》 中介绍了原生 API 的方法,这次刚好看到其 React 封装版本 react-intersection-observer,让我们看一看 React 封装思路。
react-intersection-observer 提供了 Hook useInView
判断元素是否在可视区域内,API 如下:
import React from "react";
import { useInView } from "react-intersection-observer";
const Component = () => {
const [ref, inView] = useInView();
return (
<div ref={ref}>
<h2>{`Header inside viewport ${inView}.`}</h2>
</div>
);
};
由于判断元素是否可见是基于 dom 的,所以必须将 ref
回调函数传递给 代表元素轮廓的 DOM 元素,上面的例子中,我们将 ref
传递给了最外层 DIV。
useInView
还支持下列参数:
root
:检测是否可见基于的视窗元素,默认是整个浏览器 viewport。rootMargin
:root 边距,可以在检测时提前或者推迟固定像素判断。threshold
:是否可见的阈值,范围 0 ~ 1,0 表示任意可见即为可见,1 表示完全可见即为可见。triggerOnce
:是否仅触发一次。首先从入口函数 useInView
开始解读,这是一个 Hook,利用 ref
存储上一次 DOM 实例,state
则存储 inView
元素是否可见的 boolean 值:
export function useInView(
options: IntersectionOptions = {},
): InViewHookResponse {
const ref = React.useRef<Element>()
const [state, setState] = React.useState<State>(initialState)
// 中间部分..
return [setRef, state.inView, state.entry]
}
当组件 ref 被赋值时会调用 setRef
,回调 node
是新的 DOM 节点,因此先 unobserve(ref.current)
取消旧节点的监听,再 observe(node)
对新节点进行监听,最后 ref.current = node
更新旧节点:
// 中间部分 1
const setRef = React.useCallback(
(node) => {
if (ref.current) {
unobserve(ref.current);
}
if (node) {
observe(
node,
(inView, intersection) => {
setState({ inView, entry: intersection });
if (inView && options.triggerOnce) {
// If it should only trigger once, unobserve the element after it's inView
unobserve(node);
}
},
options
);
}
// Store a reference to the node, so we can unobserve it later
ref.current = node;
},
[options.threshold, options.root, options.rootMargin, options.triggerOnce]
);
另一段是,当 ref
不存在时会清空 inView
状态,毕竟当不存在监听对象时,inView 值只有重设为默认 false 才合理:
// 中间部分 2
useEffect(() => {
if (!ref.current && state !== initialState && !options.triggerOnce) {
// If we don't have a ref, then reset the state (unless the hook is set to only `triggerOnce`)
// This ensures we correctly reflect the current state - If you aren't observing anything, then nothing is inView
setState(initialState);
}
});
这就是入口文件的逻辑,我们可以看到还有两个重要的函数 observe
与 unobserve
,这两个函数的实现在 intersection.ts 文件中,这个文件有三个核心函数:observe
、unobserve
、onChange
。
observe
:监听 element 是否在可视区域。unobserve
:取消监听。onChange
:处理 observe
变化的回调。先看 observe
,对于同一个 root 下的监听会做合并操作,因此需要生成 observerId
作为唯一标识,这个标识由 getRootId
、rootMargin
、threshold
共同决定。
对于同一个 root 的监听下,拿到 new IntersectionObserver()
创建的 observerInstance
实例,调用 observerInstance.observe
进行监听。这里存储了两个 Map - OBSERVER_MAP
与 INSTANCE_MAP
,前者是保证同一 root 下 IntersectionObserver
实例唯一,后者存储了组件 inView
以及回调等信息,在 onChange
函数使用:
export function observe(
element: Element,
callback: ObserverInstanceCallback,
options: IntersectionObserverInit = {}
) {
// IntersectionObserver needs a threshold to trigger, so set it to 0 if it's not defined.
// Modify the options object, since it's used in the onChange handler.
if (!options.threshold) options.threshold = 0;
const { root, rootMargin, threshold } = options;
// Validate that the element is not being used in another <Observer />
invariant(
!INSTANCE_MAP.has(element),
"react-intersection-observer: Trying to observe %s, but it's already being observed by another instance.\nMake sure the `ref` is only used by a single <Observer /> instance.\n\n%s"
);
/* istanbul ignore if */
if (!element) return;
// Create a unique ID for this observer instance, based on the root, root margin and threshold.
// An observer with the same options can be reused, so lets use this fact
let observerId: string =
getRootId(root) +
(rootMargin
? `${threshold.toString()}_${rootMargin}`
: threshold.toString());
let observerInstance = OBSERVER_MAP.get(observerId);
if (!observerInstance) {
observerInstance = new IntersectionObserver(onChange, options);
/* istanbul ignore else */
if (observerId) OBSERVER_MAP.set(observerId, observerInstance);
}
const instance: ObserverInstance = {
callback,
element,
inView: false,
observerId,
observer: observerInstance,
// Make sure we have the thresholds value. It's undefined on a browser like Chrome 51.
thresholds:
observerInstance.thresholds ||
(Array.isArray(threshold) ? threshold : [threshold]),
};
INSTANCE_MAP.set(element, instance);
observerInstance.observe(element);
return instance;
}
对于 onChange
函数,因为采用了多元素监听,所以需要遍历 changes
数组,并判断 intersectionRatio
超过阈值判定为 inView
状态,通过 INSTANCE_MAP
拿到对应实例,修改其 inView
状态并执行 callback
。
这个 callback
就对应了 useInView
Hook 中 observe
的第二个参数回调:
function onChange(changes: IntersectionObserverEntry[]) {
changes.forEach((intersection) => {
const { isIntersecting, intersectionRatio, target } = intersection;
const instance = INSTANCE_MAP.get(target);
// Firefox can report a negative intersectionRatio when scrolling.
/* istanbul ignore else */
if (instance && intersectionRatio >= 0) {
// If threshold is an array, check if any of them intersects. This just triggers the onChange event multiple times.
let inView = instance.thresholds.some((threshold) => {
return instance.inView
? intersectionRatio > threshold
: intersectionRatio >= threshold;
});
if (isIntersecting !== undefined) {
// If isIntersecting is defined, ensure that the element is actually intersecting.
// Otherwise it reports a threshold of 0
inView = inView && isIntersecting;
}
instance.inView = inView;
instance.callback(inView, intersection);
}
});
}
最后是 unobserve
取消监听的实现,在 useInView
setRef
灌入新 Node 节点时,会调用 unobserve
对旧节点取消监听。
首先利用 INSTANCE_MAP
找到实例,调用 observer.unobserve(element)
销毁监听。最后销毁不必要的 INSTANCE_MAP
与 ROOT_IDS
存储。
export function unobserve(element: Element | null) {
if (!element) return;
const instance = INSTANCE_MAP.get(element);
if (instance) {
const { observerId, observer } = instance;
const { root } = observer;
observer.unobserve(element);
// Check if we are still observing any elements with the same threshold.
let itemsLeft = false;
// Check if we still have observers configured with the same root.
let rootObserved = false;
/* istanbul ignore else */
if (observerId) {
INSTANCE_MAP.forEach((item, key) => {
if (key !== element) {
if (item.observerId === observerId) {
itemsLeft = true;
rootObserved = true;
}
if (item.observer.root === root) {
rootObserved = true;
}
}
});
}
if (!rootObserved && root) ROOT_IDS.delete(root);
if (observer && !itemsLeft) {
// No more elements to observe for threshold, disconnect observer
observer.disconnect();
}
// Remove reference to element
INSTANCE_MAP.delete(element);
}
}
从其实现角度来看,为了保证正确识别到子元素存在,一定要保证 ref
能持续传递给组件最外层 DOM,如果出现传递断裂,就会判定当前组件不在视图内,比如:
const Component = () => {
const [ref, inView] = useInView();
return <Child ref={ref} />;
};
const Child = ({ loading, ref }) => {
if (loading) {
// 这一步会判定为 inView:false
return <Spin />;
}
return <div ref={ref}>Child</div>;
};
如果你的代码基于 inView
做了阻止渲染的判定,那么这个组件进入 loading 后就无法改变状态了。为了避免这种情况,要么不要让 ref
的传递断掉,要么当没有拿到 ref
对象时判定 inView
为 true。
分析了这么多 React- 类的库,其核心思想有两个:
unobserve
再重新 observe
。看过 react-intersection-observer 的源码后,你觉得还有可优化的地方吗?欢迎讨论。
版权声明:自由转载-非商用-非衍生-保持署名(创意共享 3.0 许可证)