perf: more regex improvements (#12520)

Author: patak-dev

packages/vite/scripts/util.ts

@@ -17,8 +17,9 @@ export function rewriteImports(
   })
 }
 
+const windowsSlashRE = /\\/g
 export function slash(p: string): string {
-  return p.replace(/\\/g, '/')
+  return p.replace(windowsSlashRE, '/')
 }
 
 export function walkDir(dir: string, handleFile: (file: string) => void): void {

packages/vite/src/node/optimizer/esbuildDepPlugin.ts

@@ -4,6 +4,7 @@ import { CSS_LANGS_RE, KNOWN_ASSET_TYPES } from '../constants'
 import { getDepOptimizationConfig } from '..'
 import type { PackageCache, ResolvedConfig } from '..'
 import {
+  escapeRegex,
   flattenId,
   isBuiltin,
   isExternalUrl,
@@ -281,6 +282,8 @@ module.exports = Object.create(new Proxy({}, {
   }
 }
 
+const matchesEntireLine = (text: string) => `^${escapeRegex(text)}$`
+
 // esbuild doesn't transpile `require('foo')` into `import` statements if 'foo' is externalized
 // https://github.com/evanw/esbuild/issues/566#issuecomment-735551834
 export function esbuildCjsExternalPlugin(
@@ -290,9 +293,7 @@ export function esbuildCjsExternalPlugin(
   return {
     name: 'cjs-external',
     setup(build) {
-      const escape = (text: string) =>
-        `^${text.replace(/[-/\\^$*+?.()|[\]{}]/g, '\\$&')}$`
-      const filter = new RegExp(externals.map(escape).join('|'))
+      const filter = new RegExp(externals.map(matchesEntireLine).join('|'))
 
       build.onResolve({ filter: new RegExp(`^${nonFacadePrefix}`) }, (args) => {
         return {

packages/vite/src/node/optimizer/index.ts

@@ -21,6 +21,7 @@ import {
   normalizeId,
   normalizePath,
   removeDir,
+  removeLeadingSlash,
   renameDir,
   writeFile,
 } from '../utils'
@@ -41,6 +42,7 @@ const isDebugEnabled = _debug('vite:deps').enabled
 
 const jsExtensionRE = /\.js$/i
 const jsMapExtensionRE = /\.js\.map$/i
+const reExportRE = /export\s+\*\s+from/
 
 export type ExportsData = {
   hasImports: boolean
@@ -956,7 +958,7 @@ export function createIsOptimizedDepUrl(
   const depsCacheDirPrefix = depsCacheDirRelative.startsWith('../')
     ? // if the cache directory is outside root, the url prefix would be something
       // like '/@fs/absolute/path/to/node_modules/.vite'
-      `/@fs/${normalizePath(depsCacheDir).replace(/^\//, '')}`
+      `/@fs/${removeLeadingSlash(normalizePath(depsCacheDir))}`
     : // if the cache directory is inside root, the url prefix would be something
       // like '/node_modules/.vite'
       `/${depsCacheDirRelative}`
@@ -1140,7 +1142,7 @@ export async function extractExportsData(
     facade,
     hasReExports: imports.some(({ ss, se }) => {
       const exp = entryContent.slice(ss, se)
-      return /export\s+\*\s+from/.test(exp)
+      return reExportRE.test(exp)
     }),
     jsxLoader: usedJsxLoader,
   }

packages/vite/src/node/plugins/asset.ts

@@ -16,14 +16,21 @@ import {
 } from '../build'
 import type { Plugin } from '../plugin'
 import type { ResolvedConfig } from '../config'
-import { cleanUrl, getHash, joinUrlSegments, normalizePath } from '../utils'
+import {
+  cleanUrl,
+  getHash,
+  joinUrlSegments,
+  normalizePath,
+  removeLeadingSlash,
+} from '../utils'
 import { FS_PREFIX } from '../constants'
 
 export const assetUrlRE = /__VITE_ASSET__([a-z\d]+)__(?:\$_(.*?)__)?/g
 
 const rawRE = /(?:\?|&)raw(?:&|$)/
 const urlRE = /(\?|&)url(?:&|$)/
 const jsSourceMapRE = /\.[cm]?js\.map$/
+const unnededFinalQueryCharRE = /[?&]$/
 
 const assetCache = new WeakMap<ResolvedConfig, Map<string, string>>()
 
@@ -169,7 +176,7 @@ export function assetPlugin(config: ResolvedConfig): Plugin {
         return
       }
 
-      id = id.replace(urlRE, '$1').replace(/[?&]$/, '')
+      id = id.replace(urlRE, '$1').replace(unnededFinalQueryCharRE, '')
       const url = await fileToUrl(id, config, this)
       return `export default ${JSON.stringify(url)}`
     },
@@ -255,7 +262,7 @@ function fileToDevUrl(id: string, config: ResolvedConfig) {
     rtn = path.posix.join(FS_PREFIX, id)
   }
   const base = joinUrlSegments(config.server?.origin ?? '', config.base)
-  return joinUrlSegments(base, rtn.replace(/^\//, ''))
+  return joinUrlSegments(base, removeLeadingSlash(rtn))
 }
 
 export function getPublicAssetFilename(

packages/vite/src/node/plugins/clientInjections.ts

@@ -4,6 +4,9 @@ import type { ResolvedConfig } from '../config'
 import { CLIENT_ENTRY, ENV_ENTRY } from '../constants'
 import { isObject, normalizePath, resolveHostname } from '../utils'
 
+const process_env_NODE_ENV_RE =
+  /(\bglobal(This)?\.)?\bprocess\.env\.NODE_ENV\b/g
+
 // ids in transform are normalized to unix style
 const normalizedClientEntry = normalizePath(CLIENT_ENTRY)
 const normalizedEnvEntry = normalizePath(ENV_ENTRY)
@@ -86,7 +89,7 @@ export function clientInjectionsPlugin(config: ResolvedConfig): Plugin {
         // for it to avoid shimming a `process` object during dev,
         // avoiding inconsistencies between dev and build
         return code.replace(
-          /(\bglobal(This)?\.)?\bprocess\.env\.NODE_ENV\b/g,
+          process_env_NODE_ENV_RE,
           config.define?.['process.env.NODE_ENV'] ||
             JSON.stringify(process.env.NODE_ENV || config.mode),
         )

packages/vite/src/node/plugins/define.ts

@@ -1,7 +1,7 @@
 import MagicString from 'magic-string'
 import type { ResolvedConfig } from '../config'
 import type { Plugin } from '../plugin'
-import { transformStableResult } from '../utils'
+import { escapeRegex, transformStableResult } from '../utils'
 import { isCSSRequest } from './css'
 import { isHTMLRequest } from './html'
 
@@ -113,11 +113,7 @@ export function definePlugin(config: ResolvedConfig): Plugin {
           // Mustn't be preceded by a char that can be part of an identifier
           // or a '.' that isn't part of a spread operator
           '(?<![\\p{L}\\p{N}_$]|(?<!\\.\\.)\\.)(' +
-            replacementsKeys
-              .map((str) => {
-                return str.replace(/[-[\]/{}()*+?.\\^$|]/g, '\\$&')
-              })
-              .join('|') +
+            replacementsKeys.map(escapeRegex).join('|') +
             // Mustn't be followed by a char that can be part of an identifier
             // or an assignment (but allow equality operators)
             ')(?:(?<=\\.)|(?![\\p{L}\\p{N}_$]|\\s*?=[^=]))',

packages/vite/src/node/plugins/html.ts

@@ -20,6 +20,7 @@ import {
   isExternalUrl,
   normalizePath,
   processSrcSet,
+  removeLeadingSlash,
 } from '../utils'
 import type { ResolvedConfig } from '../config'
 import { toOutputFilePathInHtml } from '../build'
@@ -537,7 +538,7 @@ export function buildHtmlPlugin(config: ResolvedConfig): Plugin {
           if (
             content !== '' && // Empty attribute
             !namedOutput.includes(content) && // Direct reference to named output
-            !namedOutput.includes(content.replace(/^\//, '')) // Allow for absolute references as named output can't be an absolute path
+            !namedOutput.includes(removeLeadingSlash(content)) // Allow for absolute references as named output can't be an absolute path
           ) {
             try {
               const url =

packages/vite/src/node/server/__tests__/pluginContainer.spec.ts

@@ -196,7 +196,7 @@ async function getPluginContainer(
   )
 
   // @ts-expect-error This plugin requires a ViteDevServer instance.
-  config.plugins = config.plugins.filter((p) => !/pre-alias/.test(p.name))
+  config.plugins = config.plugins.filter((p) => !p.name.includes('pre-alias'))
 
   resolveId = (id) => container.resolveId(id)
   const container = await createPluginContainer(config, moduleGraph)

packages/vite/src/node/server/hmr.ts

@@ -14,6 +14,8 @@ import type { ModuleNode } from './moduleGraph'
 
 export const debugHmr = createDebugger('vite:hmr')
 
+const whitespaceRE = /\s/
+
 const normalizedClientDir = normalizePath(CLIENT_DIR)
 
 export interface HmrOptions {
@@ -388,7 +390,7 @@ export function lexAcceptedHmrDeps(
         } else if (char === '`') {
           prevState = state
           state = LexerState.inTemplateString
-        } else if (/\s/.test(char)) {
+        } else if (whitespaceRE.test(char)) {
           continue
         } else {
           if (state === LexerState.inCall) {

packages/vite/src/node/server/middlewares/static.ts

@@ -14,10 +14,13 @@ import {
   isInternalRequest,
   isParentDirectory,
   isWindows,
+  removeLeadingSlash,
   shouldServeFile,
   slash,
 } from '../../utils'
 
+const knownJavascriptExtensionRE = /\.[tj]sx?$/
+
 const sirvOptions = ({
   headers,
   shouldServe,
@@ -35,7 +38,7 @@ const sirvOptions = ({
       // for the MIME type video/mp2t. In almost all cases, we can expect
       // these files to be TypeScript files, and for Vite to serve them with
       // this Content-Type.
-      if (/\.[tj]sx?$/.test(pathname)) {
+      if (knownJavascriptExtensionRE.test(pathname)) {
         res.setHeader('Content-Type', 'application/javascript')
       }
       if (headers) {
@@ -119,7 +122,7 @@ export function serveStaticMiddleware(
     }
 
     const resolvedPathname = redirectedPathname || pathname
-    let fileUrl = path.resolve(dir, resolvedPathname.replace(/^\//, ''))
+    let fileUrl = path.resolve(dir, removeLeadingSlash(resolvedPathname))
     if (resolvedPathname.endsWith('/') && !fileUrl.endsWith('/')) {
       fileUrl = fileUrl + '/'
     }

packages/vite/src/node/utils.ts

@@ -50,8 +50,9 @@ export const createFilter = _createFilter as (
   options?: { resolve?: string | false | null },
 ) => (id: string | unknown) => boolean
 
+const windowsSlashRE = /\\/g
 export function slash(p: string): string {
-  return p.replace(/\\/g, '/')
+  return p.replace(windowsSlashRE, '/')
 }
 
 /**
@@ -74,15 +75,19 @@ export function unwrapId(id: string): string {
     : id
 }
 
+const replaceSlashOrColonRE = /[/:]/g
+const replaceDotRE = /\./g
+const replaceNestedIdRE = /(\s*>\s*)/g
+const replaceHashRE = /#/g
 export const flattenId = (id: string): string =>
   id
-    .replace(/[/:]/g, '_')
-    .replace(/\./g, '__')
-    .replace(/(\s*>\s*)/g, '___')
-    .replace(/#/g, '____')
+    .replace(replaceSlashOrColonRE, '_')
+    .replace(replaceDotRE, '__')
+    .replace(replaceNestedIdRE, '___')
+    .replace(replaceHashRE, '____')
 
 export const normalizeId = (id: string): string =>
-  id.replace(/(\s*>\s*)/g, ' > ')
+  id.replace(replaceNestedIdRE, ' > ')
 
 //TODO: revisit later to see if the edge case that "compiling using node v12 code to be run in node v16 in the server" is what we intend to support.
 const builtins = new Set([
@@ -300,10 +305,14 @@ export function removeDirectQuery(url: string): string {
   return url.replace(directRequestRE, '$1').replace(trailingSeparatorRE, '')
 }
 
+const replacePercentageRE = /%/g
 export function injectQuery(url: string, queryToInject: string): string {
   // encode percents for consistent behavior with pathToFileURL
   // see #2614 for details
-  const resolvedUrl = new URL(url.replace(/%/g, '%25'), 'relative:///')
+  const resolvedUrl = new URL(
+    url.replace(replacePercentageRE, '%25'),
+    'relative:///',
+  )
   const { search, hash } = resolvedUrl
   let pathname = cleanUrl(url)
   pathname = isWindows ? slash(pathname) : pathname
@@ -659,13 +668,12 @@ export function processSrcSetSync(
   )
 }
 
+const cleanSrcSetRE =
+  /(?:url|image|gradient|cross-fade)\([^)]*\)|"([^"]|(?<=\\)")*"|'([^']|(?<=\\)')*'/g
 function splitSrcSet(srcs: string) {
   const parts: string[] = []
   // There could be a ',' inside of url(data:...), linear-gradient(...) or "data:..."
-  const cleanedSrcs = srcs.replace(
-    /(?:url|image|gradient|cross-fade)\([^)]*\)|"([^"]|(?<=\\)")*"|'([^']|(?<=\\)')*'/g,
-    blankReplacer,
-  )
+  const cleanedSrcs = srcs.replace(cleanSrcSetRE, blankReplacer)
   let startIndex = 0
   let splitIndex: number
   do {
@@ -678,22 +686,26 @@ function splitSrcSet(srcs: string) {
   return parts
 }
 
+const windowsDriveRE = /^[A-Z]:/
+const replaceWindowsDriveRE = /^([A-Z]):\//
+const linuxAbsolutePathRE = /^\/[^/]/
 function escapeToLinuxLikePath(path: string) {
-  if (/^[A-Z]:/.test(path)) {
-    return path.replace(/^([A-Z]):\//, '/windows/$1/')
+  if (windowsDriveRE.test(path)) {
+    return path.replace(replaceWindowsDriveRE, '/windows/$1/')
   }
-  if (/^\/[^/]/.test(path)) {
+  if (linuxAbsolutePathRE.test(path)) {
     return `/linux${path}`
   }
   return path
 }
 
+const revertWindowsDriveRE = /^\/windows\/([A-Z])\//
 function unescapeToLinuxLikePath(path: string) {
   if (path.startsWith('/linux/')) {
     return path.slice('/linux'.length)
   }
   if (path.startsWith('/windows/')) {
-    return path.replace(/^\/windows\/([A-Z])\//, '$1:/')
+    return path.replace(revertWindowsDriveRE, '$1:/')
   }
   return path
 }
@@ -1222,6 +1234,10 @@ export function joinUrlSegments(a: string, b: string): string {
   return a + b
 }
 
+export function removeLeadingSlash(str: string): string {
+  return str[0] === '/' ? str.slice(1) : str
+}
+
 export function stripBase(path: string, base: string): string {
   if (path === base) {
     return '/'
@@ -1246,3 +1262,8 @@ export function evalValue<T = any>(rawValue: string): T {
   `)
   return fn()
 }
+
+const escapeRegexRE = /[-/\\^$*+?.()|[\]{}]/g
+export function escapeRegex(str: string): string {
+  return str.replace(escapeRegexRE, '\\$&')
+}