fix(ai): avoid shell execution for fixture tags

packages/ai/__tests__/fetchAiMockResponses.test.ts

@@ -0,0 +1,104 @@
+import { execFileSync } from 'child_process';
+import { existsSync, readdirSync } from 'fs';
+import { rimrafSync } from 'rimraf';
+import { afterEach, describe, expect, it, jest } from '@jest/globals';
+
+import { fetchAiMockResponses } from '../../../scripts/fetch_ai_mock_responses';
+
+jest.mock('child_process', () => ({
+  execFileSync: jest.fn(),
+}));
+jest.mock('fs', () => ({
+  existsSync: jest.fn(),
+  readdirSync: jest.fn(),
+}));
+jest.mock('rimraf', () => ({
+  rimrafSync: jest.fn(),
+}));
+
+describe('fetchAiMockResponses', () => {
+  afterEach(() => {
+    jest.clearAllMocks();
+  });
+
+  it('passes a repository-controlled tag to git without shell interpretation', () => {
+    const maliciousTag = 'v17.0;id';
+    jest.mocked(existsSync).mockReturnValue(false);
+    jest
+      .mocked(execFileSync)
+      .mockReturnValueOnce(`hash refs/tags/${maliciousTag}`)
+      .mockReturnValueOnce('');
+
+    fetchAiMockResponses();
+
+    expect(execFileSync).toHaveBeenNthCalledWith(
+      1,
+      'git',
+      ['ls-remote', '--tags', '--sort=version:refname', expect.any(String)],
+      { encoding: 'utf8' },
+    );
+    expect(execFileSync).toHaveBeenNthCalledWith(2, 'git', [
+      '-c',
+      'advice.detachedHead=false',
+      'clone',
+      '--branch',
+      maliciousTag,
+      '--',
+      expect.any(String),
+      expect.any(String),
+    ]);
+    expect(rimrafSync).toHaveBeenCalledTimes(1);
+  });
+
+  it('exits early without touching git when a matching clone already exists locally', () => {
+    const exitSpy = jest.spyOn(process, 'exit').mockImplementation(() => {
+      throw new Error('process.exit called');
+    });
+    jest.mocked(existsSync).mockReturnValue(true);
+    jest
+      .mocked(readdirSync)
+      .mockReturnValue([
+        { name: 'vertexai-sdk-test-data_v17.0', isDirectory: () => true },
+      ] as unknown as ReturnType<typeof readdirSync>);
+
+    expect(() => fetchAiMockResponses()).toThrow('process.exit called');
+
+    expect(exitSpy).toHaveBeenCalledWith(0);
+    expect(execFileSync).not.toHaveBeenCalled();
+    exitSpy.mockRestore();
+  });
+
+  it('exits early without cloning when the target directory for the latest tag already exists', () => {
+    const exitSpy = jest.spyOn(process, 'exit').mockImplementation(() => {
+      throw new Error('process.exit called');
+    });
+    jest
+      .mocked(existsSync)
+      .mockReturnValueOnce(false) // no existing clone dir found yet
+      .mockReturnValueOnce(true); // target dir for the latest tag already exists
+    jest.mocked(execFileSync).mockReturnValueOnce('hash refs/tags/v18.0');
+
+    expect(() => fetchAiMockResponses()).toThrow('process.exit called');
+
+    expect(exitSpy).toHaveBeenCalledWith(0);
+    expect(execFileSync).toHaveBeenCalledTimes(1); // ls-remote only, never clone
+    exitSpy.mockRestore();
+  });
+
+  it('cleans stale fixture data while preserving TypeScript sources and the latest tag directory', () => {
+    jest.mocked(existsSync).mockReturnValue(false);
+    jest.mocked(execFileSync).mockReturnValueOnce('hash refs/tags/v19.0').mockReturnValueOnce('');
+
+    fetchAiMockResponses();
+
+    const [, rimrafOptions] = jest.mocked(rimrafSync).mock.calls[0];
+    const { filter } = rimrafOptions as unknown as {
+      filter: (path: string, stat: unknown) => boolean;
+    };
+
+    // filter returns true for entries rimraf should delete, false to keep them.
+    expect(filter('/root/keep.ts', {})).toBe(false);
+    expect(filter('/root/vertexai-sdk-test-data_v19.0/fixture.json', {})).toBe(false);
+    expect(filter('/root/vertexai-sdk-test-data_v18.0/stale.json', {})).toBe(true);
+  });
+});

scripts/fetch_ai_mock_responses.ts

@@ -16,7 +16,7 @@
 //  clone of the shared repository of Vertex AI test data.
 
 // eslint-disable-next-line @typescript-eslint/no-require-imports
-import { execSync } from 'child_process';
+import { execFileSync } from 'child_process';
 import { join } from 'path';
 import { existsSync, readdirSync } from 'fs';
 import { rimrafSync } from 'rimraf';
@@ -37,42 +37,57 @@ function findExistingCloneDirName(): string | undefined {
     .at(0);
 }
 
-const existingCloneDirName = findExistingCloneDirName();
-if (existingCloneDirName !== undefined) {
-  console.log('AI mock responses data exists locally already. Exiting fetch script.');
-  process.exit(0);
-}
+export function fetchAiMockResponses(): void {
+  const existingCloneDirName = findExistingCloneDirName();
+  if (existingCloneDirName !== undefined) {
+    console.log('AI mock responses data exists locally already. Exiting fetch script.');
+    process.exit(0);
+  }
 
-// Get tags from repository, sorted by tag name, and coerce result to a string, then trim it
-const repoTags = (
-  execSync(`git ls-remote --tags --sort=version:refname "${REPO_LINK}"`) + ''
-).trim();
+  // Get tags from repository, sorted by tag name, and coerce result to a string, then trim it
+  const repoTags = execFileSync(
+    'git',
+    ['ls-remote', '--tags', '--sort=version:refname', REPO_LINK],
+    { encoding: 'utf8' },
+  ).trim();
 
-// Fish out just the tag name from the last line (since they are sorted already)
-const latestTag = repoTags.split('/').at(-1);
-if (latestTag === undefined) {
-  console.error('Unable to determine latest test data tag.');
-  process.exit(1);
-}
+  // Fish out just the tag name from the last line (since they are sorted already)
+  const latestTag = repoTags.split('/').at(-1);
+  if (latestTag === undefined) {
+    console.error('Unable to determine latest test data tag.');
+    process.exit(1);
+  }
 
-// Create the test data directory based on the latest tag
-const cloneDirName = `${REPO_NAME}_${latestTag}`;
-const cloneDirPath = join(TEST_DATA_ROOT, cloneDirName);
+  // Create the test data directory based on the latest tag
+  const cloneDirName = `${REPO_NAME}_${latestTag}`;
+  const cloneDirPath = join(TEST_DATA_ROOT, cloneDirName);
 
-// Clean out any test data that isn't the latest test data
-rimrafSync(TEST_DATA_ROOT, {
-  preserveRoot: false,
-  filter: (path: string, _) => !path.endsWith('.ts') && !path.includes(cloneDirName),
-});
+  // Clean out any test data that isn't the latest test data
+  rimrafSync(TEST_DATA_ROOT, {
+    preserveRoot: false,
+    filter: (path: string, _) => !path.endsWith('.ts') && !path.includes(cloneDirName),
+  });
 
-// Exit if our intended latest data clone target already exists
-if (existsSync(cloneDirPath)) {
-  console.log('AI mock responses data exists locally already. Exiting fetch script.');
-  process.exit(0);
+  // Exit if our intended latest data clone target already exists
+  if (existsSync(cloneDirPath)) {
+    console.log('AI mock responses data exists locally already. Exiting fetch script.');
+    process.exit(0);
+  }
+
+  // Clone the latest test data
+  console.log(`Fetching AI mock responses data...`);
+  execFileSync('git', [
+    '-c',
+    'advice.detachedHead=false',
+    'clone',
+    '--branch',
+    latestTag,
+    '--',
+    REPO_LINK,
+    cloneDirPath,
+  ]);
 }
 
-// Clone the latest test data
-console.log(`Fetching AI mock responses data...`);
-execSync(
-  `git -c advice.detachedHead=false clone --branch ${latestTag} ${REPO_LINK} ${cloneDirPath}`,
-);
+if (require.main === module) {
+  fetchAiMockResponses();
+}