How to let ChatGPT interpret Vue3 source code

Practical combat

setup

setup Where is the function? We don’t know the name of its implementation function, so we asked ChatGPT:

ChatGPT tells me that the setup function is in the packages/runtime-core/src/component.ts file. As we all know, runtime-core is the runtime core code of Vue3. Let's go in and take a look.

According to what it said, we found the setupComponent and createComponentInstance functions, but did not find the setupRenderEffect function. ChatGPT only knows about 2021 Previous knowledge, Vue3 code has gone through a lot of changes, but it doesn't matter, this doesn't affect too much.

ChatGPT told me that the setupComponent function is executed in the createComponentInstance function. createComponentInstance looks at the name to create a component instance. Take a look at the detailed code .

Copy directly to ChatGPT:

We read the code according to ChatGPT's explanation and found that createComponentInstance just created an instance of the component and return. It does not execute setupComponent in the function as it said above, stupid ChatGPT.

Then find out where setupComponent is called.

You canpackages/runtime-core/Search for the function name and you will find it quickly. In the mountComponent function in the packages/runtime-core/src/renderer.ts file.

mountComponent is the method of mounting components. There is a bunch of custom renderer logic in front of it, which will not be discussed here.

  const mountComponent: MountComponentFn = (...args) => {
    const instance: ComponentInternalInstance =
      compatMountInstance ||
      (initialVNode.component = createComponentInstance(
        initialVNode,
        parentComponent,
        parentSuspense
      ))
    // ... 省略代码
    // resolve props and slots for setup context
    if (!(__COMPAT__ && compatMountInstance)) {
        // ...这里调用了setupComponent，传入了实例，还写了注释，感人
      setupComponent(instance)
    }
    // setupRenderEffect 居然也在这
    setupRenderEffect(
      instance,
      initialVNode,
      container,
      anchor,
      parentSuspense,
      isSVG,
      optimized
    )
  }

mountComponentThe function first calls createComponentInstance, returns a component instance, and then passes the instance as a parameter to setupComponent. By the way, we also discovered the setupRenderEffect function that ChatGPT lost here, which is used to handle some rendering side effects.

Back to the setupComponent function, Evan’s comments tell us that it handles props and slots.

export function setupComponent(
  instance: ComponentInternalInstance,
  isSSR = false
) {
  isInSSRComponentSetup = isSSR
  const { props, children } = instance.vnode
  const isStateful = isStatefulComponent(instance)
  initProps(instance, props, isStateful, isSSR)
  initSlots(instance, children)
  const setupResult = isStateful
    ? setupStatefulComponent(instance, isSSR)
    : undefined
  isInSSRComponentSetup = false
  return setupResult
}

Feed the code to ChatGPT:

setupComponent In the function, after processing the props and slots, according to whether it is stateful The component called setupStatefulComponent.

Directly feed the entire setupStatefulComponent to ChatGPT:

## is too long, probably meaning:

• Created a proxy cache accessCache, we don’t know why it is used, you can ask ChatGPT

• Create a public instance proxy object (proxy)

• Execute the component's setup()

The subsequent operation is to call

handleSetupResult and finishComponentSetup to return the rendering function. Let’s start with the rendering logic.

The above is the detailed content of How to let ChatGPT interpret Vue3 source code. For more information, please follow other related articles on the PHP Chinese website!