self-bench 0.3.0 → 0.3.2

package/package.json

@@ -1,25 +1,38 @@
 {
   "name": "self-bench",
-  "version": "0.3.0",
+  "version": "0.3.2",
   "description": "Turn completed repository changes into durable, private Harbor evaluations",
   "license": "MIT",
   "repository": {
     "type": "git",
     "url": "https://github.com/mupt-ai/self-bench.git"
   },
+  "publishConfig": {
+    "access": "public",
+    "registry": "https://registry.npmjs.org/"
+  },
   "type": "module",
   "bin": {
-    "selfbench": "dist/cli.js",
-    "selfbench-agent-smoke": "dist/agent-smoke-main.js",
-    "selfbench-eval": "dist/eval-main.js",
-    "selfbench-reaudit": "dist/reaudit-main.js",
-    "selfbench-repair": "dist/repair-main.js",
-    "selfbench-validate": "dist/validate-main.js"
+    "self-bench": "dist/cli.js",
+    "self-bench-agent-smoke": "dist/agent-smoke-main.js",
+    "self-bench-eval": "dist/eval-main.js",
+    "self-bench-reaudit": "dist/reaudit-main.js",
+    "self-bench-repair": "dist/repair-main.js",
+    "self-bench-validate": "dist/validate-main.js"
   },
   "files": [
     "dist",
-    "src/extensions",
-    "src/skills"
+    "src",
+    "docs",
+    "compose.yaml",
+    "Dockerfile",
+    "Dockerfile.sandbox",
+    "bun.lock",
+    "scripts",
+    "biome.json",
+    "tsconfig.json",
+    "tsconfig.build.json",
+    ".dockerignore"
   ],
   "engines": {
     "node": ">=22"
@@ -29,21 +42,21 @@
     "clean": "node -e \"require('node:fs').rmSync('dist', { recursive: true, force: true })\"",
     "cli": "bun run src/cli.ts",
     "build:review": "vite build --config review/vite.config.ts",
-    "build:server": "tsc -p tsconfig.build.json && bun build src/sandbox-author.ts --target=node --outfile=dist/sandbox-author.bundle.js && bun build src/sandbox-review.ts --target=node --outfile=dist/sandbox-review.bundle.js && bun build src/sandbox-repair.ts --target=node --outfile=dist/sandbox-repair.bundle.js && bun build src/sandbox-validation-repair.ts --target=node --outfile=dist/sandbox-validation-repair.bundle.js",
+    "build:server": "tsc -p tsconfig.build.json && bun build src/sandbox-author.ts --target=node --outfile=dist/sandbox-author.bundle.js && bun build src/sandbox-review.ts --target=node --outfile=dist/sandbox-review.bundle.js && bun build src/sandbox-repair.ts --target=node --outfile=dist/sandbox-repair.bundle.js && bun build src/sandbox-validation-repair.ts --target=node --outfile=dist/sandbox-validation-repair.bundle.js && bun run build:metadata",
+    "build:metadata": "bun scripts/write-build-metadata.ts",
+    "prepack": "bun run build",
+    "prepublishOnly": "bun run validate",
+    "verify:package": "bun scripts/verify-package.ts",
     "check": "biome check . && tsc --noEmit -p tsconfig.json && bun run typecheck:review",
     "dev:api": "tsx src/api-main.ts",
     "dev:review": "vite --config review/vite.config.ts",
     "dev:worker": "tsx src/worker-main.ts",
     "format": "biome format --write .",
-    "prepack": "bun run build",
     "start:api": "node dist/api-main.js",
     "start:worker": "node dist/worker-main.js",
     "test": "bun test tests review/src",
     "typecheck:review": "tsc --noEmit -p review/tsconfig.json",
-    "validate": "bun run check && bun run test && bun run build"
-  },
-  "publishConfig": {
-    "access": "public"
+    "validate": "bun run check && bun run test && bun run build && bun run verify:package"
   },
   "dependencies": {
     "@google-cloud/storage": "^7.17.0",

package/scripts/verify-package.ts

@@ -0,0 +1,57 @@
+import { mkdtemp, readFile, rm } from "node:fs/promises";
+import { tmpdir } from "node:os";
+import { join, resolve } from "node:path";
+
+const root = resolve(import.meta.dir, "..");
+const temporary = await mkdtemp(join(tmpdir(), "self-bench-package-"));
+async function run(command: string, args: string[], cwd = root): Promise<string> {
+  const child = Bun.spawn([command, ...args], {
+    cwd,
+    stdout: "pipe",
+    stderr: "pipe",
+  });
+  const [exitCode, stdout, stderr] = await Promise.all([
+    child.exited,
+    new Response(child.stdout).text(),
+    new Response(child.stderr).text(),
+  ]);
+  if (exitCode !== 0) {
+    throw new Error(`${command} ${args.join(" ")} failed:\n${stderr || stdout}`);
+  }
+  return stdout;
+}
+
+try {
+  await run("bun", ["pm", "pack", "--destination", temporary]);
+  const tarballs = (await Array.fromAsync(new Bun.Glob("*.tgz").scan({ cwd: temporary }))).map(
+    (name) => join(temporary, name),
+  );
+  const tarball = tarballs[0] ?? (() => {
+    throw new Error("bun pm pack did not create a tarball");
+  })();
+  const packageJson = JSON.parse(await readFile(join(root, "package.json"), "utf8")) as {
+    name: string;
+    version: string;
+  };
+  const installRoot = join(temporary, "install");
+  await run("mkdir", ["-p", installRoot]);
+  await run("bun", ["init", "--yes"], installRoot);
+  await run("bun", ["add", "--no-save", tarball], installRoot);
+  const executable = join(installRoot, "node_modules", ".bin", "self-bench");
+  const help = await run(executable, ["--help"], installRoot);
+  if (!help.includes("self-bench up")) {
+    throw new Error("installed self-bench did not print the expected CLI help");
+  }
+  for (const asset of ["compose.yaml", "Dockerfile", "Dockerfile.sandbox", "src/skills/selfbench/SKILL.md"]) {
+    await readFile(join(installRoot, "node_modules", packageJson.name, asset));
+  }
+  const installedPackage = JSON.parse(
+    await readFile(join(installRoot, "node_modules", packageJson.name, "package.json"), "utf8"),
+  ) as { name: string; version: string };
+  if (installedPackage.name !== packageJson.name || installedPackage.version !== packageJson.version) {
+    throw new Error("installed package metadata does not match the workspace package");
+  }
+  console.log(`verified ${installedPackage.name}@${installedPackage.version}`);
+} finally {
+  await rm(temporary, { recursive: true, force: true });
+}

package/scripts/write-build-metadata.ts

@@ -0,0 +1,27 @@
+import { mkdir, writeFile } from "node:fs/promises";
+import { execFile } from "node:child_process";
+import { promisify } from "node:util";
+
+const execFileAsync = promisify(execFile);
+const root = new URL("../", import.meta.url);
+const output = new URL("../dist/build-metadata.js", import.meta.url);
+const configuredCommit = process.env.SELFBENCH_BUILD_COMMIT;
+
+let commit = configuredCommit;
+if (!commit) {
+  try {
+    const result = await execFileAsync("git", ["rev-parse", "HEAD"], { cwd: root });
+    commit = result.stdout.trim();
+  } catch {
+    commit = "0".repeat(40);
+  }
+}
+if (!/^[0-9a-f]{40}$/i.test(commit)) {
+  throw new Error("SELFBENCH_BUILD_COMMIT must be a full commit SHA");
+}
+
+await mkdir(new URL("../dist/", import.meta.url), { recursive: true });
+await writeFile(
+  output,
+  `// Generated during the build.\nexport const buildCommit = ${JSON.stringify(commit.toLowerCase())};\n`,
+);