Reduce packaged app size (#1026)

2026-05-17 23:28:09 +08:00
parent 34bfae2851
commit 3edeb8cdfa
26 changed files with 760 additions and 116 deletions
--- a/scripts/after-pack.cjs
+++ b/scripts/after-pack.cjs
@@ -155,6 +155,7 @@ const PLATFORM_NATIVE_SCOPES = {
  '@reflink': /^reflink-(darwin|linux|win32)-(x64|arm64|x64-gnu|x64-musl|arm64-gnu|arm64-musl|x64-msvc|arm64-msvc)/,
  '@node-llama-cpp': /^(mac|linux|win)-(arm64|x64|armv7l)(-metal|-cuda|-cuda-ext|-vulkan)?$/,
  '@esbuild': /^(darwin|linux|win32|android|freebsd|netbsd|openbsd|sunos|aix|openharmony)-(x64|arm64|arm|ia32|loong64|mips64el|ppc64|riscv64|s390x)/,
+  '@openai': /^codex-(darwin|linux|win32)-(x64|arm64)$/,
 };

 // Unscoped packages that follow a <name>-<platform>-<arch> convention.
@@ -173,6 +174,13 @@ function baseArch(rawArch) {
  return dash > 0 ? rawArch.slice(0, dash) : rawArch;
 }

+function matchesTargetArch(pkgArch, targetArch) {
+  if (targetArch === 'universal') {
+    return pkgArch === 'x64' || pkgArch === 'arm64' || pkgArch === 'universal';
+  }
+  return pkgArch === targetArch || pkgArch === 'universal';
+}
+
 function cleanupNativePlatformPackages(nodeModulesDir, platform, arch) {
  let removed = 0;

@@ -190,7 +198,7 @@ function cleanupNativePlatformPackages(nodeModulesDir, platform, arch) {

      const isMatch =
        pkgPlatform === platform &&
-        (pkgArch === arch || pkgArch === 'universal');
+        matchesTargetArch(pkgArch, arch);

      if (!isMatch) {
        try {
@@ -215,7 +223,7 @@ function cleanupNativePlatformPackages(nodeModulesDir, platform, arch) {

      const isMatch =
        pkgPlatform === platform &&
-        (pkgArch === arch || pkgArch === 'universal');
+        matchesTargetArch(pkgArch, arch);

      if (!isMatch) {
        try {
@@ -229,6 +237,75 @@ function cleanupNativePlatformPackages(nodeModulesDir, platform, arch) {
  return removed;
 }

+function cleanupNodeModulesRuntimeJunk(nodeModulesDir, platform, arch) {
+  let removed = 0;
+
+  const nodeWavDir = join(nodeModulesDir, 'node-wav');
+  for (const name of ['x.json', 'x.js', 'x.js~', 'file.wav']) {
+    try {
+      const target = join(nodeWavDir, name);
+      if (existsSync(target)) {
+        rmSync(target, { recursive: true, force: true });
+        removed++;
+      }
+    } catch { /* */ }
+  }
+
+  const treeSitterBashDir = join(nodeModulesDir, 'tree-sitter-bash');
+  const treeSitterSrc = join(treeSitterBashDir, 'src');
+  for (const name of ['parser.c', 'scanner.c', 'grammar.json', 'tree_sitter']) {
+    try {
+      const target = join(treeSitterSrc, name);
+      if (existsSync(target)) {
+        rmSync(target, { recursive: true, force: true });
+        removed++;
+      }
+    } catch { /* */ }
+  }
+
+  const prebuildsDir = join(treeSitterBashDir, 'prebuilds');
+  if (existsSync(prebuildsDir)) {
+    for (const entry of readdirSync(prebuildsDir)) {
+      const [entryPlatform, ...entryArchParts] = entry.split('-');
+      const entryArch = baseArch(entryArchParts.join('-'));
+      if (entryPlatform === platform && matchesTargetArch(entryArch, arch)) continue;
+      try {
+        rmSync(join(prebuildsDir, entry), { recursive: true, force: true });
+        removed++;
+      } catch { /* */ }
+    }
+  }
+
+  return removed;
+}
+
+function cleanupKnownRuntimeJunk(rootDir, platform, arch) {
+  let removed = 0;
+  const stack = [rootDir];
+
+  while (stack.length > 0) {
+    const dir = stack.pop();
+    let entries;
+    try { entries = readdirSync(normWin(dir), { withFileTypes: true }); } catch { continue; }
+
+    if (basename(dir) === 'node_modules') {
+      removed += cleanupNodeModulesRuntimeJunk(dir, platform, arch);
+    }
+
+    for (const entry of entries) {
+      if (!entry.isDirectory()) continue;
+      stack.push(join(dir, entry.name));
+    }
+  }
+
+  return removed;
+}
+
+exports.__test = {
+  cleanupNativePlatformPackages,
+  cleanupNodeModulesRuntimeJunk,
+};
+
 // ── Broken module patcher ─────────────────────────────────────────────────────
 // Some bundled packages have transpiled CJS that sets `module.exports = exports.default`
 // without ever assigning `exports.default`, leaving module.exports === undefined.
@@ -621,6 +698,10 @@ exports.default = async function afterPack(context) {
      const pluginNM = join(pluginDestDir, 'node_modules');
      cleanupUnnecessaryFiles(pluginDestDir);
      if (existsSync(pluginNM)) {
+        const pluginJunkRemoved = cleanupKnownRuntimeJunk(pluginDestDir, platform, arch);
+        if (pluginJunkRemoved > 0) {
+          console.log(`[after-pack] ✅ ${pluginId}: removed ${pluginJunkRemoved} known runtime junk files/directories.`);
+        }
        cleanupKoffi(pluginNM, platform, arch);
        cleanupNativePlatformPackages(pluginNM, platform, arch);
      }
@@ -761,7 +842,8 @@ exports.default = async function afterPack(context) {
  // 2. General cleanup on the full openclaw directory (not just node_modules)
  console.log('[after-pack] 🧹 Cleaning up unnecessary files ...');
  const removedRoot = cleanupUnnecessaryFiles(openclawRoot);
-  console.log(`[after-pack] ✅ Removed ${removedRoot} unnecessary files/directories.`);
+  const removedKnownJunk = cleanupKnownRuntimeJunk(openclawRoot, platform, arch);
+  console.log(`[after-pack] ✅ Removed ${removedRoot + removedKnownJunk} unnecessary files/directories.`);

  // 3. Platform-specific: strip koffi non-target platform binaries
  const koffiRemoved = cleanupKoffi(dest, platform, arch);
--- a/scripts/bundle-openclaw.mjs
+++ b/scripts/bundle-openclaw.mjs
@@ -177,6 +177,10 @@ echo`   Virtual store root: ${openclawVirtualNM}`;
 queue.push({ nodeModulesDir: openclawVirtualNM, skipPkg: 'openclaw' });

 const SKIP_PACKAGES = new Set([
+  // Extra bundled extensions such as @openclaw/codex can declare openclaw as a
+  // peer/optional dependency. The bundle already copies openclaw to OUTPUT root,
+  // so do not also copy a duplicate into OUTPUT/node_modules/openclaw.
+  'openclaw',
  'typescript',
  '@playwright/test',
  // @discordjs/opus is a native .node addon compiled for the system Node.js
@@ -447,6 +451,25 @@ function patchBundledExtensionPackageJsons(extensionsRoot) {

 patchBundledExtensionPackageJsons(extensionsDir);

+function bundleExternalExtension(pkgName, extensionId) {
+  const pkgLink = path.join(NODE_MODULES, ...pkgName.split('/'));
+  if (!fs.existsSync(pkgLink)) {
+    throw new Error(`Missing extension package "${pkgName}". Run pnpm install first.`);
+  }
+
+  const realPath = fs.realpathSync(pkgLink);
+  const dest = path.join(extensionsDir, extensionId);
+  fs.rmSync(normWin(dest), { recursive: true, force: true });
+  fs.mkdirSync(normWin(path.dirname(dest)), { recursive: true });
+  fs.cpSync(normWin(realPath), normWin(dest), {
+    recursive: true,
+    dereference: true,
+  });
+  echo`   Bundled external extension ${pkgName} -> dist/extensions/${extensionId}`;
+}
+
+bundleExternalExtension('@openclaw/codex', 'codex');
+
 // 6. Clean up the bundle to reduce package size
 //
 // This removes platform-agnostic waste: dev artifacts, docs, source maps,
@@ -482,6 +505,47 @@ function rmSafe(target) {
  } catch { return false; }
 }

+function cleanupNodeModulesRuntimeJunk(nodeModulesDir) {
+  let removedCount = 0;
+
+  const nodeWavDir = path.join(nodeModulesDir, 'node-wav');
+  for (const name of ['x.json', 'x.js', 'x.js~', 'file.wav']) {
+    if (rmSafe(path.join(nodeWavDir, name))) removedCount++;
+  }
+
+  // tree-sitter-bash ships C sources for rebuilding its native addon. Packaged
+  // builds use the prebuilt addon/wasm; keep node-types.json because the CJS
+  // entry exposes it as optional runtime metadata.
+  const treeSitterSrc = path.join(nodeModulesDir, 'tree-sitter-bash', 'src');
+  for (const name of ['parser.c', 'scanner.c', 'grammar.json', 'tree_sitter']) {
+    if (rmSafe(path.join(treeSitterSrc, name))) removedCount++;
+  }
+
+  return removedCount;
+}
+
+function cleanupKnownRuntimeJunk(rootDir) {
+  let removedCount = 0;
+  const stack = [rootDir];
+
+  while (stack.length > 0) {
+    const dir = stack.pop();
+    let entries;
+    try { entries = fs.readdirSync(dir, { withFileTypes: true }); } catch { continue; }
+
+    if (path.basename(dir) === 'node_modules') {
+      removedCount += cleanupNodeModulesRuntimeJunk(dir);
+    }
+
+    for (const entry of entries) {
+      if (!entry.isDirectory()) continue;
+      stack.push(path.join(dir, entry.name));
+    }
+  }
+
+  return removedCount;
+}
+
 function cleanupBundle(outputDir) {
  let removedCount = 0;
  const nm = path.join(outputDir, 'node_modules');
@@ -620,6 +684,8 @@ function cleanupBundle(outputDir) {
    if (rmSafe(path.join(outputDir, rel))) removedCount++;
  }

+  removedCount += cleanupKnownRuntimeJunk(outputDir);
+
  return removedCount;
 }

--- a/scripts/openclaw-bundle-config.mjs
+++ b/scripts/openclaw-bundle-config.mjs
@@ -28,6 +28,11 @@ export const EXTRA_BUNDLED_PACKAGES = [
  // transitive dependency graph from the app bundle context.
  'playwright-core',

+  // The Codex agent harness is published as an OpenClaw extension package.
+  // Bundle it with the runtime so openai-codex models can register the codex
+  // harness in packaged builds.
+  '@openclaw/codex',
+
  // Electron main process QR login flows resolve these files from the
  // bundled OpenClaw runtime context in packaged builds.
  'qrcode-terminal',