perf: scan markdown review lines with native newline search (#20245)

config/scripts/markdown-note-line-scan-benchmark.mjs

@@ -0,0 +1,81 @@
+import assert from 'node:assert/strict'
+import { execFileSync } from 'node:child_process'
+import { readFileSync } from 'node:fs'
+import { dirname, resolve } from 'node:path'
+import { performance } from 'node:perf_hooks'
+import { build } from 'esbuild'
+import { buildCounterbalancedSchedule } from './counterbalanced-benchmark-schedule.mjs'
+
+const sourcePath = 'src/renderer/src/lib/markdown-review-notes.ts'
+const baseline = process.argv[2] ?? '20ab9950654'
+async function load(contents) {
+  const result = await build({
+    stdin: { contents, loader: 'ts', resolveDir: dirname(resolve(sourcePath)) },
+    bundle: true,
+    platform: 'node',
+    format: 'esm',
+    write: false
+  })
+  return import(
+    `data:text/javascript;base64,${Buffer.from(result.outputFiles[0].text).toString('base64')}`
+  )
+}
+const before = await load(
+  execFileSync('git', ['show', `${baseline}:${sourcePath}`], { encoding: 'utf8' })
+)
+const after = await load(readFileSync(sourcePath, 'utf8'))
+const results = []
+for (const [name, lineCount, width, count, iterations] of [
+  ['small', 20, 40, 1, 1000],
+  ['long-lines', 1000, 1000, 20, 3],
+  ['many-lines', 20000, 80, 20, 3],
+  ['early-note', 20000, 80, 1, 1000]
+]) {
+  const content = Array.from({ length: lineCount }, (_, i) => `${i}: ${'x'.repeat(width)}`).join(
+    '\r\n'
+  )
+  const notes = Array.from({ length: count }, (_, i) => ({
+    id: `${i}`,
+    worktreeId: 'bench',
+    filePath: 'README.md',
+    source: 'markdown',
+    lineNumber: name === 'early-note' ? 2 : lineCount - i,
+    body: 'Clarify this line',
+    createdAt: i,
+    side: 'modified'
+  }))
+  assert.equal(
+    after.formatMarkdownReviewNotes(notes, content),
+    before.formatMarkdownReviewNotes(notes, content)
+  )
+  const arms = { before, after }
+  const samples = { before: [], after: [] }
+  function run(arm) {
+    const start = performance.now()
+    for (let i = 0; i < iterations; i++) {
+      arms[arm].formatMarkdownReviewNotes(notes, content)
+    }
+    return (performance.now() - start) / iterations
+  }
+  for (let i = 0; i < 6; i++) {
+    run('before')
+    run('after')
+  }
+  for (const pair of buildCounterbalancedSchedule(12, 'before', 'after')) {
+    for (const arm of pair) {
+      samples[arm].push(run(arm))
+    }
+  }
+  const median = (xs) => xs.sort((a, b) => a - b)[Math.floor(xs.length / 2)]
+  results.push({
+    name,
+    lineCount,
+    width,
+    count,
+    beforeMs: median(samples.before),
+    afterMs: median(samples.after)
+  })
+}
+console.log(
+  JSON.stringify({ node: process.version, platform: process.platform, baseline, results }, null, 2)
+)

src/renderer/src/lib/markdown-review-notes.test.ts

@@ -94,6 +94,44 @@ describe('markdown review notes', () => {
     expect(charCodeAt.mock.calls.length).toBeLessThan(64)
   })
 
+  it('preserves empty, trailing, and mixed line endings across excerpt ranges', () => {
+    for (const content of [
+      '',
+      'one',
+      '\n',
+      'one\n',
+      'one\r\n\r\ntwo\r',
+      'one\rtwo',
+      Array.from({ length: 15 }, (_, i) => `line ${i}`).join('\n')
+    ]) {
+      const sourceLines = content
+        .split('\n')
+        .map((line) => (line.endsWith('\r') ? line.slice(0, -1) : line))
+      for (let start = -1; start <= sourceLines.length + 1; start += 1) {
+        for (let end = start; end <= sourceLines.length + 1; end += 1) {
+          const first = Math.max(1, start)
+          const selected = sourceLines.slice(first - 1, Math.max(first, end))
+          const bounded =
+            selected.length <= 8
+              ? selected
+              : [...selected.slice(0, 4), '...', ...selected.slice(-4)]
+          expect(
+            getMarkdownReviewExcerpt(content, note({ startLine: start, lineNumber: end }))
+          ).toBe(bounded.map((line) => `> ${line}`).join('\n'))
+        }
+      }
+    }
+  })
+
+  it('finds late excerpts without checking every prefix character in JavaScript', () => {
+    const content = `${'x'.repeat(100_000)}\r\ntarget\nignored`
+    const charCodeAt = vi.spyOn(String.prototype, 'charCodeAt')
+    const excerpt = getMarkdownReviewExcerpt(content, note({ lineNumber: 2 }))
+    const checks = charCodeAt.mock.calls.length
+    expect(excerpt).toBe('> target')
+    expect(checks).toBeLessThanOrEqual(2)
+  })
+
   it('prefers exact selected text for card highlights', () => {
     const highlighted = getMarkdownReviewHighlightedText(
       'one\ntwo broad line\nthree',

src/renderer/src/lib/markdown-review-notes.ts

@@ -93,10 +93,9 @@ function forEachMarkdownReviewLine(
 ): void {
   let lineStart = 0
   let lineNumber = 1
-  for (let index = 0; index <= content.length; index += 1) {
-    if (index < content.length && content.charCodeAt(index) !== 10) {
-      continue
-    }
+  while (lineStart <= content.length) {
+    const newline = content.indexOf('\n', lineStart)
+    const index = newline === -1 ? content.length : newline
     const lineEnd = index > lineStart && content.charCodeAt(index - 1) === 13 ? index - 1 : index
     if (visit(content.slice(lineStart, lineEnd), lineNumber) === false) {
       return