metona-ai-desktop/electron/harness/tools/built-in/__tests__/file-guard.test.ts

/**
 * File Guard 单元测试（P1-14 测试基线）
 * 覆盖：路径遍历防护、前缀碰撞、MEMORY.md 保护、glob 匹配、编码检测
 */

import { describe, it, expect, beforeAll, afterAll } from 'vitest';
import { mkdtempSync, rmSync, writeFileSync, mkdirSync } from 'fs';
import { tmpdir } from 'os';
import { join } from 'path';
import {
  isPathWithinWorkspace,
  isProtectedWorkspaceFile,
  safeResolvePath,
  matchGlob,
  matchAnyGlob,
  commandTouchesProtectedFile,
  decodeBufferWithDetection,
} from '../file-guard';

describe('isPathWithinWorkspace', () => {
  const ws = join(tmpdir(), 'metona-test-ws');

  beforeAll(() => {
    mkdirSync(ws, { recursive: true });
  });

  it('工作空间内的相对路径通过', () => {
    expect(isPathWithinWorkspace('src/main.ts', ws)).toBe(true);
  });

  it('工作空间内的绝对路径通过', () => {
    expect(isPathWithinWorkspace(join(ws, 'src/main.ts'), ws)).toBe(true);
  });

  it('工作空间根目录本身通过', () => {
    expect(isPathWithinWorkspace('.', ws)).toBe(true);
  });

  it('路径遍历（../）被拒绝', () => {
    expect(isPathWithinWorkspace('../etc/passwd', ws)).toBe(false);
  });

  it('多层遍历（../../..）被拒绝', () => {
    expect(isPathWithinWorkspace('../../../etc/passwd', ws)).toBe(false);
  });

  it('前缀碰撞不误判（/app-evil 不在 /app 内）', () => {
    const parent = join(tmpdir(), 'metona-prefix-app');
    mkdirSync(parent, { recursive: true });
    expect(isPathWithinWorkspace(join(tmpdir(), 'metona-prefix-app-evil/x'), parent)).toBe(false);
  });

  it('绝对路径指向工作空间外被拒绝', () => {
    expect(isPathWithinWorkspace('C:\\Windows\\System32\\cmd.exe', ws)).toBe(false);
  });
});

describe('isProtectedWorkspaceFile / safeResolvePath', () => {
  const ws = join(tmpdir(), 'metona-test-protect');

  beforeAll(() => {
    mkdirSync(join(ws, 'sub'), { recursive: true });
  });

  it('工作空间根目录的 MEMORY.md 受保护', () => {
    expect(isProtectedWorkspaceFile('MEMORY.md', ws)).toBe(true);
  });

  it('子目录的 MEMORY.md 不受保护', () => {
    expect(isProtectedWorkspaceFile(join('sub', 'MEMORY.md'), ws)).toBe(false);
  });

  it('safeResolvePath 拒绝越界路径并抛错', () => {
    expect(() => safeResolvePath('../outside.txt', ws)).toThrow(/Path traversal/);
  });

  it('safeResolvePath 拒绝根目录 MEMORY.md 并抛错', () => {
    expect(() => safeResolvePath('MEMORY.md', ws)).toThrow(/MEMORY.md/);
  });

  it('safeResolvePath 正常解析工作空间内路径', () => {
    const resolved = safeResolvePath('src/a.ts', ws);
    expect(resolved).toBe(join(ws, 'src/a.ts'));
  });
});

describe('commandTouchesProtectedFile', () => {
  it('裸引用 MEMORY.md 被拦截', () => {
    expect(commandTouchesProtectedFile('cat MEMORY.md')).toBe(true);
  });

  it('子目录 MEMORY.md 不被拦截', () => {
    expect(commandTouchesProtectedFile('cat sub/MEMORY.md')).toBe(false);
    expect(commandTouchesProtectedFile('cat sub\\MEMORY.md')).toBe(false);
  });

  it('管道/分号后的 MEMORY.md 被拦截', () => {
    expect(commandTouchesProtectedFile('echo x | cat MEMORY.md; rm file')).toBe(true);
  });

  // v0.7.4 P2-3: 前导路径绕过根治 —— ./ .\ ~ ~/ 前缀仍指工作空间根，必须拦截
  it('./ .\\ ~ ~/ 前缀引用 MEMORY.md 被拦截（P2-3 根治）', () => {
    expect(commandTouchesProtectedFile('cat ./MEMORY.md')).toBe(true);
    expect(commandTouchesProtectedFile('cat .\\MEMORY.md')).toBe(true);
    expect(commandTouchesProtectedFile('cat ~/MEMORY.md')).toBe(true);
    expect(commandTouchesProtectedFile('cat ~/./MEMORY.md')).toBe(true);
    expect(commandTouchesProtectedFile('rm -rf ./MEMORY.md')).toBe(true);
  });

  it('子目录 MEMORY.md 仍不被拦截（路径前缀不误伤）', () => {
    expect(commandTouchesProtectedFile('cat sub/MEMORY.md')).toBe(false);
    expect(commandTouchesProtectedFile('cat sub\\MEMORY.md')).toBe(false);
    expect(commandTouchesProtectedFile('cat ./sub/MEMORY.md')).toBe(false);
    expect(commandTouchesProtectedFile('cat ~/sub/MEMORY.md')).toBe(false);
  });

  it('无关命令不误判', () => {
    expect(commandTouchesProtectedFile('npm run test')).toBe(false);
    expect(commandTouchesProtectedFile('git status')).toBe(false);
  });

  // v0.7.4 P2-3 修正: 括号/子 shell/命令替换/重定向无空格/反引号形态
  it('子 shell/括号/命令替换/重定向/反引号引用 MEMORY.md 被拦截（P2-3 修正）', () => {
    expect(commandTouchesProtectedFile('$(cat MEMORY.md)')).toBe(true);
    expect(commandTouchesProtectedFile('(cat MEMORY.md)')).toBe(true);
    expect(commandTouchesProtectedFile('cat <MEMORY.md')).toBe(true);
    expect(commandTouchesProtectedFile('echo `cat MEMORY.md`')).toBe(true);
    expect(commandTouchesProtectedFile('python -c "open(\'MEMORY.md\')"')).toBe(true);
    expect(commandTouchesProtectedFile('$(cat ./MEMORY.md)')).toBe(true);
  });
});

describe('matchGlob / matchAnyGlob', () => {
  it('单 glob 匹配', () => {
    expect(matchGlob('main.ts', '*.ts')).toBe(true);
    expect(matchGlob('main.js', '*.ts')).toBe(false);
  });

  it('? 单字符匹配', () => {
    expect(matchGlob('test1.js', 'test?.js')).toBe(true);
    expect(matchGlob('test12.js', 'test?.js')).toBe(false);
  });

  it('逗号分隔多 glob 任一匹配', () => {
    expect(matchAnyGlob('a.ts', '*.ts,*.js,*.tsx')).toBe(true);
    expect(matchAnyGlob('a.jsx', '*.ts,*.js,*.tsx')).toBe(false);
  });

  it('空 glob 字符串匹配所有', () => {
    expect(matchAnyGlob('anything.txt', '')).toBe(true);
  });
});

describe('decodeBufferWithDetection', () => {
  it('UTF-8 无 BOM 正确解码', () => {
    const buf = Buffer.from('你好 world', 'utf-8');
    const { content, encoding } = decodeBufferWithDetection(buf);
    expect(content).toBe('你好 world');
    expect(encoding).toBe('utf-8');
  });

  it('UTF-8 BOM 被剥离并识别', () => {
    const body = Buffer.from('hello', 'utf-8');
    const buf = Buffer.concat([Buffer.from([0xef, 0xbb, 0xbf]), body]);
    const { content, encoding } = decodeBufferWithDetection(buf);
    expect(content).toBe('hello');
    expect(encoding).toBe('utf-8-bom');
  });

  it('UTF-16 LE BOM 正确解码', () => {
    const body = '你好';
    const buf = Buffer.concat([Buffer.from([0xff, 0xfe]), Buffer.from(body, 'utf16le')]);
    const { content, encoding } = decodeBufferWithDetection(buf);
    expect(content).toBe(body);
    expect(encoding).toBe('utf-16le');
  });

  it('UTF-16 BE BOM 正确解码（字节交换）', () => {
    const body = '你好';
    const le = Buffer.from(body, 'utf16le');
    const be = Buffer.from(le);
    be.swap16();
    const buf = Buffer.concat([Buffer.from([0xfe, 0xff]), be]);
    const { content, encoding } = decodeBufferWithDetection(buf);
    expect(content).toBe(body);
    expect(encoding).toBe('utf-16be');
  });

  it('空 Buffer 返回空内容', () => {
    const { content, encoding } = decodeBufferWithDetection(Buffer.alloc(0));
    expect(content).toBe('');
    expect(encoding).toBe('utf-8');
  });
});

describe('workspace 文件读取场景（临时目录）', () => {
  let ws: string;

  beforeAll(() => {
    ws = mkdtempSync(join(tmpdir(), 'metona-guard-'));
    writeFileSync(join(ws, 'file.txt'), 'content', 'utf-8');
  });

  afterAll(() => {
    rmSync(ws, { recursive: true, force: true });
  });

  it('工作空间内文件路径通过校验', () => {
    expect(isPathWithinWorkspace('file.txt', ws)).toBe(true);
    expect(isPathWithinWorkspace(join(ws, 'file.txt'), ws)).toBe(true);
  });
});

// ===== v0.7.4: 表格化扩充（用例数翻倍） =====

describe('commandTouchesProtectedFile — 拦截矩阵（v0.7.4 扩充）', () => {
  it.each([
    ['裸引用', 'cat MEMORY.md'],
    ['./ 前缀', 'cat ./MEMORY.md'],
    ['.\\ 前缀', 'cat .\\MEMORY.md'],
    ['~/ 前缀', 'cat ~/MEMORY.md'],
    ['~/./ 组合', 'cat ~/./MEMORY.md'],
    ['分号后', 'echo a; cat MEMORY.md'],
    ['管道后', 'echo a | cat MEMORY.md'],
    ['& 后', 'echo a & cat MEMORY.md'],
    ['> 重定向', 'cat MEMORY.md > out'],
    ['< 重定向无空格', 'cat <MEMORY.md'],
    ['子 shell 括号', '(cat MEMORY.md)'],
    ['命令替换', '$(cat MEMORY.md)'],
    ['命令替换带前缀', '$(cat ./MEMORY.md)'],
    ['反引号', 'echo `cat MEMORY.md`'],
    ['引号包裹', "cat 'MEMORY.md'"],
    ['双引号包裹', 'cat "MEMORY.md"'],
    ['cat 后直接', 'cat MEMORY.md'],
    ['rm 引用', 'rm -rf MEMORY.md'],
    ['大写在 cmd 中', 'cat memory.MD'],
  ])('%s 被拦截', (_label, cmd) => {
    expect(commandTouchesProtectedFile(cmd)).toBe(true);
  });

  it.each([
    ['子目录正斜杠', 'cat sub/MEMORY.md'],
    ['子目录反斜杠', 'cat sub\\MEMORY.md'],
    ['./ 子目录', 'cat ./sub/MEMORY.md'],
    ['~/ 子目录', 'cat ~/sub/MEMORY.md'],
    ['无关 npm', 'npm run test'],
    ['无关 git', 'git status'],
    ['无关 node', 'node server.js'],
    ['无关 tsc', 'npx tsc --noEmit'],
  ])('%s 放行', (_label, cmd) => {
    expect(commandTouchesProtectedFile(cmd)).toBe(false);
  });
});

describe('matchGlob — 边界矩阵（v0.7.4 扩充）', () => {
  it.each([
    ['普通后缀', 'main.ts', '*.ts', true],
    ['多字符前缀', 'test-file.js', 'test-*.js', true],
    ['? 单字符', 'test1.js', 'test?.js', true],
    ['? 多字符不匹配', 'test12.js', 'test?.js', false],
    ['大小写不敏感', 'MAIN.TS', '*.ts', true],
    ['无通配', 'exact.ts', 'exact.ts', true],
    ['通配不匹配', 'main.ts', '*.js', false],
    ['空模式匹配所有', 'anything', '', false],
    ['尾点', 'file.txt', 'file.*', true],
    ['无扩展名', 'README', 'README', true],
  ])('%s: %s vs %s → %j', (_label, name, pattern, expected) => {
    expect(matchGlob(name, pattern)).toBe(expected);
  });
});

describe('matchAnyGlob — 多 glob 矩阵（v0.7.4 扩充）', () => {
  it.each([
    ['任一匹配', 'a.ts', '*.js,*.ts', true],
    ['逗号带空格', 'b.ts', '*.js, *.ts', true],
    ['全部不匹配', 'c.py', '*.js,*.ts', false],
    ['空串匹配所有', 'x', '', true],
    ['纯空白匹配所有', 'x', '   ', true],
    ['单 glob', 'd.ts', '*.ts', true],
  ])('%s: %s vs %s → %j', (_label, name, pattern, expected) => {
    expect(matchAnyGlob(name, pattern)).toBe(expected);
  });
});

describe('decodeBufferWithDetection — 编码矩阵（v0.7.4 扩充）', () => {
  it('GBK 编码中文正确解码', () => {
    // GBK 编码的"中文"（使用 iconv 等价字节：UTF-8 转 GBK 后字节）
    const gbkBytes = Buffer.from([0xd6, 0xd0, 0xce, 0xc4]); // "中文" GBK
    const { content, encoding } = decodeBufferWithDetection(gbkBytes);
    expect(content).toBe('中文');
    expect(encoding).toBe('gbk');
  });

  it('UTF-8 多字节中文 strict 解码', () => {
    const utf8 = Buffer.from('你好世界', 'utf-8');
    const { content, encoding } = decodeBufferWithDetection(utf8);
    expect(content).toBe('你好世界');
    expect(encoding).toBe('utf-8');
  });

  it('UTF-16 LE 带 BOM 解码', () => {
    const buf = Buffer.concat([Buffer.from([0xff, 0xfe]), Buffer.from('ab', 'utf16le')]);
    const { content, encoding } = decodeBufferWithDetection(buf);
    expect(content).toBe('ab');
    expect(encoding).toBe('utf-16le');
  });

  it('UTF-16 BE 带 BOM 解码（字节交换）', () => {
    const le = Buffer.from('ab', 'utf16le');
    const be = Buffer.from([le[1], le[0], le[3], le[2]]);
    const full = Buffer.concat([Buffer.from([0xfe, 0xff]), be]);
    const { content, encoding } = decodeBufferWithDetection(full);
    expect(content).toBe('ab');
    expect(encoding).toBe('utf-16be');
  });

  it('损坏 UTF-8 降级 GBK 再降级 loose', () => {
    // 无效 UTF-8 序列（0xFF 0xFE 非 BOM 场景）→ 最终 loose
    const bad = Buffer.from([0x80, 0x81, 0x82]);
    const { encoding } = decodeBufferWithDetection(bad);
    expect(['gbk', 'utf-8-loose']).toContain(encoding);
  });

  it('单字节 ASCII 走 utf-8', () => {
    const { content, encoding } = decodeBufferWithDetection(Buffer.from('hello', 'ascii'));
    expect(content).toBe('hello');
    expect(encoding).toBe('utf-8');
  });
});