perf: avoid rescanning emitted source in analysis guards (#18920)

config/scripts/source-string-blanking-benchmark.mjs

@@ -0,0 +1,79 @@
+import assert from 'node:assert/strict'
+import { execFileSync } from 'node:child_process'
+import { stripTypeScriptTypes } from 'node:module'
+import { performance } from 'node:perf_hooks'
+import { blankStringContents as after } from '../../src/shared/source-scan/source-tree-scan.ts'
+
+const ref = process.argv[2]
+if (!ref) {
+  throw new Error('Usage: node config/scripts/source-string-blanking-benchmark.mjs <baseline-ref>')
+}
+const source = execFileSync('git', ['show', `${ref}:src/shared/source-scan/source-tree-scan.ts`], {
+  encoding: 'utf8'
+})
+const { blankStringContents: before } = await import(
+  `data:text/javascript;base64,${Buffer.from(stripTypeScriptTypes(source)).toString('base64')}`
+)
+const tokens = [
+  'a',
+  '/',
+  '*',
+  ' ',
+  '\n',
+  '\r',
+  '\t',
+  '\u00a0',
+  '\u2028',
+  '"',
+  "'",
+  '`',
+  '${',
+  '}',
+  '{',
+  '\\',
+  '(',
+  ')',
+  '[',
+  ']',
+  '=',
+  '+',
+  '-',
+  ';'
+]
+let seed = 173
+for (let sample = 0; sample < 3000; sample++) {
+  let input = ''
+  for (let token = 0; token < 40; token++) {
+    seed = (Math.imul(seed, 1664525) + 1013904223) >>> 0
+    input += tokens[seed % tokens.length]
+  }
+  assert.equal(after(input), before(input), JSON.stringify(input))
+  assert.equal(after(input, true), before(input, true), JSON.stringify(input))
+}
+function measure(fn, input) {
+  const samples = []
+  for (let run = 0; run < 3; run++) {
+    const start = performance.now()
+    fn(input)
+    samples.push(performance.now() - start)
+  }
+  return samples.sort((a, b) => a - b)[1]
+}
+const results = []
+for (const lines of [100, 1000, 5000, 10000]) {
+  const input = 'const x = value / 2;\n'.repeat(lines)
+  assert.equal(after(input), before(input))
+  results.push({
+    lines,
+    bytes: Buffer.byteLength(input),
+    beforeMs: measure(before, input),
+    afterMs: measure(after, input)
+  })
+}
+console.log(
+  JSON.stringify(
+    { node: process.version, platform: process.platform, differentialCases: 3000, results },
+    null,
+    2
+  )
+)

src/shared/source-scan/source-tree-scan.test.ts

@@ -47,6 +47,13 @@ describe('stripComments', () => {
 })
 
 describe('blankStringContents', () => {
+  it('does not rescan the accumulated source for each division operator', () => {
+    const source = 'const x = value / 2;\n'.repeat(10000)
+    const started = performance.now()
+    expect(blankStringContents(source)).toBe(source)
+    expect(performance.now() - started).toBeLessThan(200)
+  })
+
   it('neutralises parentheses inside a string so a call is matched whole', () => {
     // A shell script embedded as a string closed the call early, so the options
     // object fell outside the match and its flags read as absent.

src/shared/source-scan/source-tree-scan.ts

@@ -179,8 +179,7 @@ export function blankStringContentsDesynced(source: string): boolean {
  * each also has a prefix reading: `!` (non-null assertion vs `!/re/.test(x)`),
  * `+` `-` `*` `%` `^` `~` (postfix `--`/`++`), and `>` `}` (JSX close).
  */
-function startsRegexLiteral(emitted: string): boolean {
-  const prev = emitted.replace(/\s+$/, '').at(-1)
+function startsRegexLiteral(prev: string | undefined): boolean {
   return prev === undefined || '(,=:[&|?;'.includes(prev)
 }
 
@@ -209,6 +208,7 @@ function findRegexLiteralEnd(source: string, start: number): number {
 
 export function blankStringContents(source: string, reportDesync = false): string {
   let out = ''
+  let lastSignificantChar: string | undefined
   let index = 0
   let quote: string | null = null
   // Brace depth per interpolation, so a `}` inside `${ { a: 1 } }` does not
@@ -220,6 +220,7 @@ export function blankStringContents(source: string, reportDesync = false): strin
       templates.push(0)
       quote = null
       out += '${'
+      lastSignificantChar = '{'
       index += 2
       continue
     }
@@ -232,6 +233,7 @@ export function blankStringContents(source: string, reportDesync = false): strin
           templates.pop()
           quote = '`'
           out += char
+          lastSignificantChar = char
           index += 1
           continue
         }
@@ -256,6 +258,7 @@ export function blankStringContents(source: string, reportDesync = false): strin
       if (char === quote) {
         quote = null
         out += char
+        lastSignificantChar = char
       } else {
         out += char === '\n' ? char : ' '
       }
@@ -270,10 +273,11 @@ export function blankStringContents(source: string, reportDesync = false): strin
     // comments first, but this runs standalone too, and at index 0 a file
     // starting with a banner comment read as one giant regex.
     const next = source[index + 1]
-    if (char === '/' && next !== '/' && next !== '*' && startsRegexLiteral(out)) {
+    if (char === '/' && next !== '/' && next !== '*' && startsRegexLiteral(lastSignificantChar)) {
       const end = findRegexLiteralEnd(source, index)
       if (end !== -1) {
         out += `/${' '.repeat(end - index - 1)}`
+        lastSignificantChar = '/'
         index = end
         continue
       }
@@ -282,6 +286,9 @@ export function blankStringContents(source: string, reportDesync = false): strin
       quote = char
     }
     out += char
+    if (/\S/.test(char)) {
+      lastSignificantChar = char
+    }
     index += 1
   }
   if (reportDesync) {