perf(repos): avoid quadratic icon source scans

config/scripts/repo-icon-source-href-benchmark.mjs

@@ -0,0 +1,38 @@
+import assert from 'node:assert/strict'
+import { performance } from 'node:perf_hooks'
+import { extractIconHref } from '../../src/main/repo-icon-source-href.ts'
+
+// Original production expressions, preserved for the before/after measurement.
+const html =
+  /<link\b(?=[^>]*\brel=["'](?:icon|shortcut icon)["'])(?=[^>]*\bhref=["']([^"'?]+))[^>]*>/i
+const object =
+  /(?=[^}]*\brel\s*:\s*["'](?:icon|shortcut icon)["'])(?=[^}]*\bhref\s*:\s*["']([^"'?]+))[^}]*/i
+const original = (source) => source.match(html)?.[1] ?? source.match(object)?.[1] ?? null
+
+function measure(fn, source, repetitions) {
+  const samples = []
+  for (let run = 0; run < repetitions; run++) {
+    const started = performance.now()
+    fn(source)
+    samples.push(performance.now() - started)
+  }
+  return samples.sort((a, b) => a - b)[Math.floor(samples.length / 2)]
+}
+
+const results = []
+for (const size of [8192, 16384, 32768]) {
+  for (const shape of ['no icon', 'rel without href']) {
+    const source = 'a'.repeat(size) + (shape === 'rel without href' ? ' rel:"icon"' : '')
+    assert.equal(extractIconHref(source), original(source))
+    const beforeMs = measure(original, source, 3)
+    const afterMs = measure(extractIconHref, source, 15)
+    results.push({
+      shape,
+      bytes: Buffer.byteLength(source),
+      beforeMs,
+      afterMs,
+      speedup: beforeMs / afterMs
+    })
+  }
+}
+console.log(JSON.stringify({ node: process.version, platform: process.platform, results }, null, 2))

src/main/repo-icon-file-detection.test.ts

@@ -1,4 +1,6 @@
-import { readFile, stat } from 'node:fs/promises'
+import { join } from 'node:path'
+import { tmpdir } from 'node:os'
+import { mkdtemp, mkdir, readFile, rm, stat, writeFile } from 'node:fs/promises'
 import type * as FsPromisesModule from 'node:fs/promises'
 import { describe, expect, it, vi } from 'vitest'
 import type { ExecutionHostFilesystemRoute } from './providers/execution-host-provider-dispatch'
@@ -136,3 +138,55 @@ describe('detectRepoFileIcon connection boundary', () => {
     expect(stat).toHaveBeenCalled()
   })
 })
+
+describe('declared repo icons through production filesystem routes', () => {
+  it.each([
+    ['local', false],
+    ['ssh', false],
+    ['local', true],
+    ['ssh', true]
+  ] as const)('preserves declared icon detection on %s (no icon: %s)', async (kind, noIcon) => {
+    const directory = await mkdtemp(join(tmpdir(), 'orca-icon-href-'))
+    const source = noIcon
+      ? 'a'.repeat(256 * 1024)
+      : `${'a'.repeat(32768)}{ rel: "icon", href: "/first.png", href: "/chosen.png" }`
+    try {
+      await mkdir(join(directory, 'public'))
+      await writeFile(join(directory, 'index.html'), source)
+      await writeFile(join(directory, 'public', 'chosen.png'), Buffer.from(PNG_BASE64, 'base64'))
+      const provider = remoteFilesystemProvider({
+        stat: async (path) => {
+          const info = await stat(path)
+          return {
+            type: info.isFile() ? 'file' : 'directory',
+            size: info.size,
+            mtime: info.mtimeMs
+          }
+        },
+        readFile: async (path) => {
+          const buffer = await readFile(path)
+          const isBinary = path.endsWith('.png')
+          return {
+            content: buffer.toString(isBinary ? 'base64' : 'utf8'),
+            isBinary,
+            mimeType: isBinary ? 'image/png' : 'text/html'
+          }
+        }
+      })
+      const route: ExecutionHostFilesystemRoute =
+        kind === 'local' ? { kind: 'local', hostId: 'local' } : sshRoute('icon-oracle', provider)
+      await expect(detectRepoFileIcon(directory, route)).resolves.toEqual(
+        noIcon
+          ? null
+          : {
+              type: 'image',
+              src: `data:image/png;base64,${PNG_BASE64}`,
+              source: 'file',
+              label: 'public/chosen.png'
+            }
+      )
+    } finally {
+      await rm(directory, { recursive: true, force: true })
+    }
+  })
+})

src/main/repo-icon-file-detection.ts

@@ -3,6 +3,7 @@ import { buildImageDataUri } from '../shared/image-data-uri'
 import { MAX_REPO_ICON_UPLOAD_BYTES, type RepoIcon } from '../shared/repo-icon'
 import type { ExecutionHostFilesystemRoute } from './providers/execution-host-provider-dispatch'
 import type { IFilesystemProvider } from './providers/types'
+import { extractIconHref } from './repo-icon-source-href'
 import { iconHrefCandidates } from './repo-icon-href-candidates'
 import { joinWorktreeRelativePath } from './runtime/runtime-relative-paths'
 
@@ -49,11 +50,6 @@ const REPO_ICON_SOURCE_FILE_CANDIDATES = [
 // not read large app entrypoints just to find a small favicon href.
 const MAX_REPO_ICON_SOURCE_BYTES = 256 * 1024
 
-const LINK_ICON_HTML_RE =
-  /<link\b(?=[^>]*\brel=["'](?:icon|shortcut icon)["'])(?=[^>]*\bhref=["']([^"'?]+))[^>]*>/i
-const LINK_ICON_OBJECT_RE =
-  /(?=[^}]*\brel\s*:\s*["'](?:icon|shortcut icon)["'])(?=[^}]*\bhref\s*:\s*["']([^"'?]+))[^}]*/i
-
 type DetectedImageFormat = {
   mimeType: 'image/png' | 'image/webp'
 }
@@ -98,10 +94,6 @@ function detectImageFormat(buffer: Buffer): DetectedImageFormat | null {
   return null
 }
 
-function extractIconHref(source: string): string | null {
-  return source.match(LINK_ICON_HTML_RE)?.[1] ?? source.match(LINK_ICON_OBJECT_RE)?.[1] ?? null
-}
-
 function repoIconFromImageBuffer(buffer: Buffer, relativePath: string): RepoIcon | null {
   const format = detectImageFormat(buffer)
   if (!format) {

src/main/repo-icon-source-href.test.ts

@@ -0,0 +1,52 @@
+import { describe, expect, it } from 'vitest'
+import { extractIconHref } from './repo-icon-source-href'
+
+// Original production expressions are the compatibility oracle.
+const HTML_RE =
+  /<link\b(?=[^>]*\brel=["'](?:icon|shortcut icon)["'])(?=[^>]*\bhref=["']([^"'?]+))[^>]*>/i
+const OBJECT_RE =
+  /(?=[^}]*\brel\s*:\s*["'](?:icon|shortcut icon)["'])(?=[^}]*\bhref\s*:\s*["']([^"'?]+))[^}]*/i
+
+export function originalIconHref(source: string): string | null {
+  return source.match(HTML_RE)?.[1] ?? source.match(OBJECT_RE)?.[1] ?? null
+}
+
+describe('repo icon source href compatibility', () => {
+  it.each([
+    '',
+    'plain source without icon properties',
+    '{ rel: "icon", href: "/first.png", href: "/last.png" }',
+    '{ href: "/first.png", rel: "icon", rel: "stylesheet" }',
+    '{ rel: "icon" } { href: "/unrelated.png" }',
+    '{ rel: "icon", href: "/first.png" } { rel: "icon", href: "/last.png" }',
+    '{ rel: "icon", href: "/object.png" } <link href="/html.png" rel="icon">',
+    '{ rel: "ICON", href: "/UPPER.png" }',
+    '}\n\n{href: "/line.png",\nrel : "shortcut icon"}',
+    '{ rel: "icon", href: "/unterminated}after brace',
+    '{ rel: "icon", href: "?query" }',
+    '{ rel: "icon", href: "/before?query" }',
+    '<link rel="stylesheet" href="/no.png">',
+    '<link rel="icon" href="/yes.png"',
+    '{rel:"icon",href:"/nested{brace}.png"}',
+    'xrel:"icon", xhref:"/no.png"'
+  ])('preserves original selection for %s', (source) => {
+    expect(extractIconHref(source)).toBe(originalIconHref(source))
+  })
+
+  it('matches the original across generated malformed property sequences', () => {
+    const tokens = ['}', '{', ' ', 'rel:"icon"', 'href:"a"', 'href:"b"', 'rel:"other"', 'x', '\n']
+    let seed = 97
+    for (let sample = 0; sample < 3000; sample++) {
+      let source = ''
+      for (let token = 0; token < 12; token++) {
+        seed = (Math.imul(seed, 1664525) + 1013904223) >>> 0
+        source += tokens[seed % tokens.length]
+      }
+      expect(extractIconHref(source), source).toBe(originalIconHref(source))
+    }
+  })
+
+  it('handles a maximum-size icon-free entrypoint', () => {
+    expect(extractIconHref('a'.repeat(256 * 1024))).toBeNull()
+  })
+})

src/main/repo-icon-source-href.ts

@@ -0,0 +1,26 @@
+const LINK_ICON_HTML_RE =
+  /<link\b(?=[^>]*\brel=["'](?:icon|shortcut icon)["'])(?=[^>]*\bhref=["']([^"'?]+))[^>]*>/i
+const LINK_ICON_OBJECT_RE =
+  /(?=[^}]*\brel\s*:\s*["'](?:icon|shortcut icon)["'])(?=[^}]*\bhref\s*:\s*["']([^"'?]+))[^}]*/iy
+
+export function extractIconHref(source: string): string | null {
+  const htmlHref = source.match(LINK_ICON_HTML_RE)?.[1]
+  if (htmlHref !== undefined) {
+    return htmlHref
+  }
+  let start = 0
+  while (start <= source.length) {
+    // Every suffix before the next closing brace sees the same candidate properties.
+    LINK_ICON_OBJECT_RE.lastIndex = start
+    const href = LINK_ICON_OBJECT_RE.exec(source)?.[1]
+    if (href !== undefined) {
+      return href
+    }
+    const closingBrace = source.indexOf('}', start)
+    if (closingBrace === -1) {
+      return null
+    }
+    start = closingBrace + 1
+  }
+  return null
+}