Vue3源码解析之 compiler（二）
站长
· 阅读数 2
前言
上一篇我们分析了 compiler 编译器中 template 是如何转换为 ast 对象的，本篇我们先来分析下 transform 函数是如何将 AST 转换为 Javascript AST 的。
案例
首先引入 compile 函数，声明 template 模板，通过 compile 函数将模板编译成 render 函数且打印该结果。
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<title>Document</title>
<script src="../../../dist/vue.global.js"></script>
</head>
<body>
<div id="app"></div>
<script>
const { compile } = Vue
const template = `<div>hello world</div>`
const renderFn = compile(template)
console.log(renderFn)
</script>
</body>
</html>
Transform 函数
transform 函数作用是将 AST 对象转换为 Javascript AST，该方法定义在 packages/compiler-core/src/transform.ts 文件中：
export function transform(root: RootNode, options: TransformOptions) {
const context = createTransformContext(root, options)
traverseNode(root, context)
if (options.hoistStatic) {
hoistStatic(root, context)
}
if (!options.ssr) {
createRootCodegen(root, context)
}
// finalize meta information
root.helpers = [...context.helpers.keys()]
root.components = [...context.components]
root.directives = [...context.directives]
root.imports = context.imports
root.hoists = context.hoists
root.temps = context.temps
root.cached = context.cached
if (__COMPAT__) {
root.filters = [...context.filters!]
}
}
该方法接收两个参数，root 为模板解析后的 ast 对象，options 配置参数，传递了一堆转换函数，我们只需关注 nodeTransforms 属性，包含 12 个转换函数：
同 baseParse 类似，通过 createTransformContext 方法创建上下文对象 context：
export function createTransformContext(
root: RootNode,
{
filename = '',
prefixIdentifiers = false,
hoistStatic = false,
cacheHandlers = false,
// 省略
}: TransformOptions
): TransformContext {
const nameMatch = filename.replace(/\?.*$/, '').match(/([^/\\]+)\.\w+$/)
const context: TransformContext = {
// options
// 省略
nodeTransforms,
directiveTransforms,
transformHoist,
isBuiltInComponent,
isCustomElement,
expressionPlugins,
scopeId,
// 省略
// methods
helper(name) {
const count = context.helpers.get(name) || 0
context.helpers.set(name, count + 1)
return name
},
// 省略
}
// 省略
return context
}
接着执行 traverseNode 函数，该函数为 transform 核心：
export function traverseNode(
node: RootNode | TemplateChildNode,
context: TransformContext
) {
context.currentNode = node
// apply transform plugins
const { nodeTransforms } = context
const exitFns = []
for (let i = 0; i < nodeTransforms.length; i++) {
const onExit = nodeTransforms[i](node, context)
if (onExit) {
if (isArray(onExit)) {
exitFns.push(...onExit)
} else {
exitFns.push(onExit)
}
}
if (!context.currentNode) {
// node was removed
return
} else {
// node may have been replaced
node = context.currentNode
}
}
switch (node.type) {
case NodeTypes.COMMENT:
if (!context.ssr) {
// inject import for the Comment symbol, which is needed for creating
// comment nodes with `createVNode`
context.helper(CREATE_COMMENT)
}
break
case NodeTypes.INTERPOLATION:
// no need to traverse, but we need to inject toString helper
if (!context.ssr) {
context.helper(TO_DISPLAY_STRING)
}
break
// for container types, further traverse downwards
case NodeTypes.IF:
for (let i = 0; i < node.branches.length; i++) {
traverseNode(node.branches[i], context)
}
break
case NodeTypes.IF_BRANCH:
case NodeTypes.FOR:
case NodeTypes.ELEMENT:
case NodeTypes.ROOT:
traverseChildren(node, context)
break
}
// exit transforms
context.currentNode = node
let i = exitFns.length
while (i--) {
exitFns[i]()
}
}
将上下文的当前节点设置为当前处理的节点 context.currentNode = node ，之后从上下文对象中解构出转换函数数组 nodeTransforms，遍历转换函数，我们取其一看下该转换方法：
export const transformOnce: NodeTransform = (node, context) => {
if (node.type === NodeTypes.ELEMENT && findDir(node, 'once', true)) {
if (seen.has(node) || context.inVOnce) {
return
}
seen.add(node)
context.inVOnce = true
context.helper(SET_BLOCK_TRACKING)
return () => {
context.inVOnce = false
const cur = context.currentNode as ElementNode | IfNode | ForNode
if (cur.codegenNode) {
cur.codegenNode = context.cache(cur.codegenNode, true /* isVNode */)
}
}
}
}
可见该方法满足条件会返回一个匿名函数，所以 exitFns 只会存在两个值，一个是 undefined，一个是 匿名函数，该函数用到了闭包概念。
执行完遍历，我们再查看 exitFns 值：
此时存在两个转换函数，接着执行 switch 逻辑，当前节点类型为 0 即 ROOT 根节点，执行 traverseChildren 方法：
export function traverseChildren(
parent: ParentNode,
context: TransformContext
) {
let i = 0
const nodeRemoved = () => {
i--
}
for (; i < parent.children.length; i++) {
const child = parent.children[i]
if (isString(child)) continue
context.parent = parent
context.childIndex = i
context.onNodeRemoved = nodeRemoved
traverseNode(child, context)
}
}
该方法会遍历所有子节点，触发 traverseNode 函数，这是一个递归执行的过程。 一直找到最后层级的文本节点：
当前节点 type = 2，不再触发 switch 中的 traverseChildren 方法，说明当前处于最底层。之后遍历执行转换函数 exitFns：
执行 postTransformElement 方法，它由 transformElement 函数返回，该方法定义在 packages/compilers-core/src/transforms/transformElement.ts 文件中：
export const transformElement: NodeTransform = (node, context) => {
// perform the work on exit, after all child expressions have been
// processed and merged.
return function postTransformElement() {
node = context.currentNode!
if (
!(
node.type === NodeTypes.ELEMENT &&
(node.tagType === ElementTypes.ELEMENT ||
node.tagType === ElementTypes.COMPONENT)
)
) {
return
}
const { tag, props } = node
const isComponent = node.tagType === ElementTypes.COMPONENT
// The goal of the transform is to create a codegenNode implementing the
// VNodeCall interface.
let vnodeTag = isComponent
? resolveComponentType(node as ComponentNode, context)
: `"${tag}"`
const isDynamicComponent =
isObject(vnodeTag) && vnodeTag.callee === RESOLVE_DYNAMIC_COMPONENT
let vnodeProps: VNodeCall['props']
let vnodeChildren: VNodeCall['children']
let vnodePatchFlag: VNodeCall['patchFlag']
let patchFlag: number = 0
let vnodeDynamicProps: VNodeCall['dynamicProps']
let dynamicPropNames: string[] | undefined
let vnodeDirectives: VNodeCall['directives']
let shouldUseBlock =
// dynamic component may resolve to plain elements
isDynamicComponent ||
vnodeTag === TELEPORT ||
vnodeTag === SUSPENSE ||
(!isComponent &&
// <svg> and <foreignObject> must be forced into blocks so that block
// updates inside get proper isSVG flag at runtime. (#639, #643)
// This is technically web-specific, but splitting the logic out of core
// leads to too much unnecessary complexity.
(tag === 'svg' || tag === 'foreignObject'))
// props
if (props.length > 0) {
const propsBuildResult = buildProps(
node,
context,
undefined,
isComponent,
isDynamicComponent
)
vnodeProps = propsBuildResult.props
patchFlag = propsBuildResult.patchFlag
dynamicPropNames = propsBuildResult.dynamicPropNames
const directives = propsBuildResult.directives
vnodeDirectives =
directives && directives.length
? (createArrayExpression(
directives.map(dir => buildDirectiveArgs(dir, context))
) as DirectiveArguments)
: undefined
if (propsBuildResult.shouldUseBlock) {
shouldUseBlock = true
}
}
// children
if (node.children.length > 0) {
if (vnodeTag === KEEP_ALIVE) {
// Although a built-in component, we compile KeepAlive with raw children
// instead of slot functions so that it can be used inside Transition
// or other Transition-wrapping HOCs.
// To ensure correct updates with block optimizations, we need to:
// 1. Force keep-alive into a block. This avoids its children being
// collected by a parent block.
shouldUseBlock = true
// 2. Force keep-alive to always be updated, since it uses raw children.
patchFlag |= PatchFlags.DYNAMIC_SLOTS
if (__DEV__ && node.children.length > 1) {
context.onError(
createCompilerError(ErrorCodes.X_KEEP_ALIVE_INVALID_CHILDREN, {
start: node.children[0].loc.start,
end: node.children[node.children.length - 1].loc.end,
source: ''
})
)
}
}
const shouldBuildAsSlots =
isComponent &&
// Teleport is not a real component and has dedicated runtime handling
vnodeTag !== TELEPORT &&
// explained above.
vnodeTag !== KEEP_ALIVE
if (shouldBuildAsSlots) {
const { slots, hasDynamicSlots } = buildSlots(node, context)
vnodeChildren = slots
if (hasDynamicSlots) {
patchFlag |= PatchFlags.DYNAMIC_SLOTS
}
} else if (node.children.length === 1 && vnodeTag !== TELEPORT) {
const child = node.children[0]
const type = child.type
// check for dynamic text children
const hasDynamicTextChild =
type === NodeTypes.INTERPOLATION ||
type === NodeTypes.COMPOUND_EXPRESSION
if (
hasDynamicTextChild &&
getConstantType(child, context) === ConstantTypes.NOT_CONSTANT
) {
patchFlag |= PatchFlags.TEXT
}
// pass directly if the only child is a text node
// (plain / interpolation / expression)
if (hasDynamicTextChild || type === NodeTypes.TEXT) {
vnodeChildren = child as TemplateTextChildNode
} else {
vnodeChildren = node.children
}
} else {
vnodeChildren = node.children
}
}
// patchFlag & dynamicPropNames
if (patchFlag !== 0) {
if (__DEV__) {
if (patchFlag < 0) {
// special flags (negative and mutually exclusive)
vnodePatchFlag = patchFlag + ` /* ${PatchFlagNames[patchFlag]} */`
} else {
// bitwise flags
const flagNames = Object.keys(PatchFlagNames)
.map(Number)
.filter(n => n > 0 && patchFlag & n)
.map(n => PatchFlagNames[n])
.join(`, `)
vnodePatchFlag = patchFlag + ` /* ${flagNames} */`
}
} else {
vnodePatchFlag = String(patchFlag)
}
if (dynamicPropNames && dynamicPropNames.length) {
vnodeDynamicProps = stringifyDynamicPropNames(dynamicPropNames)
}
}
node.codegenNode = createVNodeCall(
context,
vnodeTag,
vnodeProps,
vnodeChildren,
vnodePatchFlag,
vnodeDynamicProps,
vnodeDirectives,
!!shouldUseBlock,
false /* disableTracking */,
isComponent,
node.loc
)
}
}
条件未满足，之后返回上一层，当前节点为 div 元素节点，此时 exitFns 存在两个转换函数 transformText 和 transformElement。先执行 transformText 返回的匿名函数，该方法定义在 packages/compilers-core/src/transforms/transformText.ts 文件中：
export const transformText: NodeTransform = (node, context) => {
if (
node.type === NodeTypes.ROOT ||
node.type === NodeTypes.ELEMENT ||
node.type === NodeTypes.FOR ||
node.type === NodeTypes.IF_BRANCH
) {
// perform the transform on node exit so that all expressions have already
// been processed.
return () => {
const children = node.children
let currentContainer: CompoundExpressionNode | undefined = undefined
let hasText = false
for (let i = 0; i < children.length; i++) {
const child = children[i]
if (isText(child)) {
hasText = true
for (let j = i + 1; j < children.length; j++) {
const next = children[j]
if (isText(next)) {
if (!currentContainer) {
currentContainer = children[i] = createCompoundExpression(
[child],
child.loc
)
}
// merge adjacent text node into current
currentContainer.children.push(` + `, next)
children.splice(j, 1)
j--
} else {
currentContainer = undefined
break
}
}
}
}
if (
!hasText ||
// if this is a plain element with a single text child, leave it
// as-is since the runtime has dedicated fast path for this by directly
// setting textContent of the element.
// for component root it's always normalized anyway.
(children.length === 1 &&
(node.type === NodeTypes.ROOT ||
(node.type === NodeTypes.ELEMENT &&
node.tagType === ElementTypes.ELEMENT &&
// #3756
// custom directives can potentially add DOM elements arbitrarily,
// we need to avoid setting textContent of the element at runtime
// to avoid accidentally overwriting the DOM elements added
// by the user through custom directives.
!node.props.find(
p =>
p.type === NodeTypes.DIRECTIVE &&
!context.directiveTransforms[p.name]
) &&
// in compat mode, <template> tags with no special directives
// will be rendered as a fragment so its children must be
// converted into vnodes.
!(__COMPAT__ && node.tag === 'template'))))
) {
return
}
// pre-convert text nodes into createTextVNode(text) calls to avoid
// runtime normalization.
for (let i = 0; i < children.length; i++) {
const child = children[i]
if (isText(child) || child.type === NodeTypes.COMPOUND_EXPRESSION) {
const callArgs: CallExpression['arguments'] = []
// createTextVNode defaults to single whitespace, so if it is a
// single space the code could be an empty call to save bytes.
if (child.type !== NodeTypes.TEXT || child.content !== ' ') {
callArgs.push(child)
}
// mark dynamic text with flag so it gets patched inside a block
if (
!context.ssr &&
getConstantType(child, context) === ConstantTypes.NOT_CONSTANT
) {
callArgs.push(
PatchFlags.TEXT +
(__DEV__ ? ` /* ${PatchFlagNames[PatchFlags.TEXT]} */` : ``)
)
}
children[i] = {
type: NodeTypes.TEXT_CALL,
content: child,
loc: child.loc,
codegenNode: createCallExpression(
context.helper(CREATE_TEXT),
callArgs
)
}
}
}
}
}
}
当前 children 为文本节点：
我们再看下这么一段逻辑：
for (let i = 0; i < children.length; i++) {
const child = children[i]
if (isText(child)) {
hasText = true
for (let j = i + 1; j < children.length; j++) {
const next = children[j]
if (isText(next)) {
if (!currentContainer) {
currentContainer = children[i] = createCompoundExpression(
[child],
child.loc
)
}
// merge adjacent text node into current
currentContainer.children.push(` + `, next)
children.splice(j, 1)
j--
} else {
currentContainer = undefined
break
}
}
}
}
假如这样一个模板 <div>a {{ b }}</div>，存在两个子节点 a 和 {{ b }}。而对于 {{ b }} 为响应式数据，所以说我们需要一个函数根据 b 来获取真实的值，之后将 a 和 b 获取到的值通过 加号 连接起来变成真实的文本内容。
由于上述条件未满足，直接 break，继续遍历执行下一个转换函数 postTransformElement，当前 node 节点为 div 元素节点：
最后执行 createVNodeCall 方法将返回的对象赋值给节点的 codegenNode 属性上：
export function createVNodeCall(
context: TransformContext | null,
tag: VNodeCall['tag'],
props?: VNodeCall['props'],
children?: VNodeCall['children'],
patchFlag?: VNodeCall['patchFlag'],
dynamicProps?: VNodeCall['dynamicProps'],
directives?: VNodeCall['directives'],
isBlock: VNodeCall['isBlock'] = false,
disableTracking: VNodeCall['disableTracking'] = false,
isComponent: VNodeCall['isComponent'] = false,
loc = locStub
): VNodeCall {
if (context) {
if (isBlock) {
context.helper(OPEN_BLOCK)
context.helper(getVNodeBlockHelper(context.inSSR, isComponent))
} else {
context.helper(getVNodeHelper(context.inSSR, isComponent))
}
if (directives) {
context.helper(WITH_DIRECTIVES)
}
}
return {
type: NodeTypes.VNODE_CALL,
tag,
props,
children,
patchFlag,
dynamicProps,
directives,
isBlock,
disableTracking,
isComponent,
loc
}
}
context.helper 为函数，在创建上下文方法 createTransformContext 时已声明：
helper(name) {
const count = context.helpers.get(name) || 0
context.helpers.set(name, count + 1)
return name
},
我们再看下 getVNodeHelper 方法：
export function getVNodeHelper(ssr: boolean, isComponent: boolean) {
return ssr || isComponent ? CREATE_VNODE : CREATE_ELEMENT_VNODE
}
当前不是 ssr 也不是 组件，返回 CREATE_ELEMENT_VNODE，该值是一个 Symbol ：
此时成功将函数名 createElementVNode 放入到 context 中，之后返回对象即 codegenNode 对象。traverseNode 函数执行完成，该方法主要做了两件事：一是生成 codegenNode 对象，二是将文本节点利用 transformText 进行合并。
接着执行 createRootCodegen 方法：
function createRootCodegen(root: RootNode, context: TransformContext) {
const { helper } = context
const { children } = root
if (children.length === 1) {
const child = children[0]
// if the single child is an element, turn it into a block.
if (isSingleElementRoot(root, child) && child.codegenNode) {
// single element root is never hoisted so codegenNode will never be
// SimpleExpressionNode
const codegenNode = child.codegenNode
if (codegenNode.type === NodeTypes.VNODE_CALL) {
makeBlock(codegenNode, context)
}
root.codegenNode = codegenNode
} else {
// - single <slot/>, IfNode, ForNode: already blocks.
// - single text node: always patched.
// root codegen falls through via genNode()
root.codegenNode = child
}
} else if (children.length > 1) {
// root has multiple nodes - return a fragment block.
let patchFlag = PatchFlags.STABLE_FRAGMENT
let patchFlagText = PatchFlagNames[PatchFlags.STABLE_FRAGMENT]
// check if the fragment actually contains a single valid child with
// the rest being comments
if (
__DEV__ &&
children.filter(c => c.type !== NodeTypes.COMMENT).length === 1
) {
patchFlag |= PatchFlags.DEV_ROOT_FRAGMENT
patchFlagText += `, ${PatchFlagNames[PatchFlags.DEV_ROOT_FRAGMENT]}`
}
root.codegenNode = createVNodeCall(
context,
helper(FRAGMENT),
undefined,
root.children,
patchFlag + (__DEV__ ? ` /* ${patchFlagText} */` : ``),
undefined,
undefined,
true,
undefined,
false /* isComponent */
)
} else {
// no children = noop. codegen will return null.
}
}
之后将子节点的 codegenNode 赋值给根节点 Root 的 codegenNode 上：
该方法执行完毕，接着返回 transform 继续执行，root.helpers = [...context.helpers.keys()]，将 helpers 对应的转换函数名赋值给根节点的 helpers 中：
接着执行后续根节点属性赋值逻辑，至此 transform 函数执行完毕。
总结
transform函数核心是执行traverseNode方法。traverseNode方法主要做了两件事：一是生成codegenNode对象，二是将文本节点利用transformText进行合并。