maestro/src/engine/tools/msg.test.ts

import { describe, it, expect, beforeEach, afterEach } from 'vitest';
import * as fs from 'fs';
import * as os from 'os';
import * as path from 'path';
import { fileURLToPath } from 'url';
import {
  formatAddress,
  stripHtml,
  selectMsgBody,
  sanitizeAttachmentName,
  formatMsgOutput,
  assembleMsgOutput,
  pickEmail,
  isParsedMsgValid,
  executeReadMsg,
  type MsgView,
} from './msg.js';
import type { ToolContext } from './core.js';
import { executeTool as officeExecuteTool, TOOL_DEFS as OFFICE_TOOL_DEFS } from './office.js';

const FIXTURE = path.join(
  path.dirname(fileURLToPath(import.meta.url)),
  '__fixtures__',
  'attachmentFiles.msg',
);

describe('formatAddress', () => {
  it('renders name and email together', () => {
    expect(formatAddress({ name: 'Alice', email: 'alice@example.com' })).toBe(
      'Alice <alice@example.com>',
    );
  });

  it('renders name only when email is missing', () => {
    expect(formatAddress({ name: 'Alice' })).toBe('Alice');
  });

  it('renders email only when name is missing', () => {
    expect(formatAddress({ email: 'alice@example.com' })).toBe('alice@example.com');
  });

  it('falls back to a placeholder when both are missing', () => {
    expect(formatAddress({})).toBe('(unknown)');
  });
});

describe('stripHtml', () => {
  it('removes tags and decodes entities', () => {
    expect(stripHtml('<p>Hello&nbsp;<b>world</b> &amp; co</p>')).toBe('Hello world & co');
  });

  it('drops script and style content', () => {
    const html = '<style>.x{color:red}</style><p>Keep</p><script>alert(1)</script>';
    expect(stripHtml(html)).toBe('Keep');
  });

  it('turns block boundaries into newlines', () => {
    expect(stripHtml('<div>line1</div><div>line2</div>')).toBe('line1\nline2');
  });

  it('decodes valid numeric entities', () => {
    expect(stripHtml('<p>&#65;&#x42;</p>')).toBe('AB');
  });

  it('does not throw on out-of-range numeric entities', () => {
    expect(() => stripHtml('<p>&#999999999;</p>')).not.toThrow();
    expect(stripHtml('A&#x110000;B')).toBe('A&#x110000;B');
  });
});

describe('selectMsgBody', () => {
  it('prefers the plain-text body', () => {
    expect(selectMsgBody({ body: 'plain text', bodyHtml: '<p>html</p>' })).toEqual({
      text: 'plain text',
      format: 'plain',
    });
  });

  it('falls back to stripped HTML when no plain body exists', () => {
    expect(selectMsgBody({ bodyHtml: '<p>html body</p>' })).toEqual({
      text: 'html body',
      format: 'html',
    });
  });

  it('reports none when no body is present', () => {
    expect(selectMsgBody({})).toEqual({ text: '', format: 'none' });
  });

  it('decodes PidTagHtml (html) when body and bodyHtml are absent', () => {
    const html = new TextEncoder().encode('<p>from pidtag</p>');
    expect(selectMsgBody({ html })).toEqual({ text: 'from pidtag', format: 'html' });
  });

  it('prefers plain body over the PidTagHtml field', () => {
    const html = new TextEncoder().encode('<p>html</p>');
    expect(selectMsgBody({ body: 'plain', html })).toEqual({ text: 'plain', format: 'plain' });
  });

  it('falls back to PidTagHtml when bodyHtml is empty/whitespace', () => {
    const html = new TextEncoder().encode('<p>pidtag body</p>');
    expect(selectMsgBody({ bodyHtml: '   ', html })).toEqual({
      text: 'pidtag body',
      format: 'html',
    });
  });
});

describe('pickEmail', () => {
  it('prefers a real SMTP address over a legacy EX DN', () => {
    expect(pickEmail('/O=EX/OU=x/CN=alice', 'alice@example.com')).toBe('alice@example.com');
    expect(pickEmail('alice@example.com', '/O=EX/OU=x/CN=alice')).toBe('alice@example.com');
  });

  it('falls back to the EX DN when no SMTP-looking address exists', () => {
    expect(pickEmail(undefined, '/O=EX/OU=x/CN=alice')).toBe('/O=EX/OU=x/CN=alice');
  });

  it('returns undefined when nothing usable is provided', () => {
    expect(pickEmail(undefined, undefined)).toBeUndefined();
    expect(pickEmail('', '  ')).toBeUndefined();
  });
});

describe('isParsedMsgValid', () => {
  it('accepts a parsed Outlook message', () => {
    expect(isParsedMsgValid({ dataType: 'msg' })).toBe(true);
  });

  it('rejects an unsupported CFBF result (old .doc/.xls, corrupted compound file)', () => {
    expect(isParsedMsgValid({ error: 'Unsupported file type!', dataType: null })).toBe(false);
    expect(isParsedMsgValid({ dataType: null })).toBe(false);
    expect(isParsedMsgValid({ dataType: 'attachment' })).toBe(false);
  });

  it('treats a whitespace-only plain body as empty and uses HTML', () => {
    expect(selectMsgBody({ body: '   \n  ', bodyHtml: '<p>real</p>' })).toEqual({
      text: 'real',
      format: 'html',
    });
  });
});

describe('sanitizeAttachmentName', () => {
  it('keeps a normal filename unchanged', () => {
    expect(sanitizeAttachmentName('report.pdf', 0)).toBe('report.pdf');
  });

  it('strips directory components to prevent path traversal', () => {
    expect(sanitizeAttachmentName('../../etc/passwd', 0)).toBe('passwd');
    expect(sanitizeAttachmentName('foo/bar/baz.txt', 0)).toBe('baz.txt');
    expect(sanitizeAttachmentName('a\\b\\c.doc', 0)).toBe('c.doc');
  });

  it('removes control characters and null bytes', () => {
    expect(sanitizeAttachmentName('na\x00me.txt', 0)).toBe('name.txt');
    expect(sanitizeAttachmentName('tab\tname.txt', 0)).toBe('tabname.txt');
  });

  it('preserves spaces inside the filename', () => {
    expect(sanitizeAttachmentName('my report.pdf', 0)).toBe('my report.pdf');
  });

  it('falls back to an indexed name when the result is empty', () => {
    expect(sanitizeAttachmentName('', 2)).toBe('attachment-3');
    expect(sanitizeAttachmentName('...', 0)).toBe('attachment-1');
  });
});

describe('formatMsgOutput', () => {
  const baseView: MsgView = {
    subject: 'Quarterly report',
    from: { name: 'Alice', email: 'alice@example.com' },
    to: [{ name: 'Bob', email: 'bob@example.com' }],
    cc: [],
    date: 'Mon, 1 Jun 2026 10:00:00 +0900',
    body: { text: 'See attached.', format: 'plain' },
    attachments: [],
  };

  it('renders the header block and body', () => {
    const out = formatMsgOutput(baseView);
    expect(out).toContain('Subject: Quarterly report');
    expect(out).toContain('From: Alice <alice@example.com>');
    expect(out).toContain('To: Bob <bob@example.com>');
    expect(out).toContain('Date: Mon, 1 Jun 2026 10:00:00 +0900');
    expect(out).toContain('See attached.');
  });

  it('lists saved attachments with their paths and sizes', () => {
    const out = formatMsgOutput({
      ...baseView,
      attachments: [{ fileName: 'report.pdf', contentLength: 2048, savedPath: 'input/report.pdf' }],
    });
    expect(out).toContain('Attachments (1)');
    expect(out).toContain('report.pdf');
    expect(out).toContain('input/report.pdf');
    expect(out).toContain('2048');
  });

  it('shows a skip reason for attachments that were not saved', () => {
    const out = formatMsgOutput({
      ...baseView,
      attachments: [{ fileName: 'huge.bin', skipped: 'exceeds size limit' }],
    });
    expect(out).toContain('huge.bin');
    expect(out).toContain('exceeds size limit');
  });

  it('notes when the body could not be extracted', () => {
    const out = formatMsgOutput({ ...baseView, body: { text: '', format: 'none' } });
    expect(out).toContain('(no text body)');
  });

  it('omits the CC line when there are no CC recipients', () => {
    expect(formatMsgOutput(baseView)).not.toContain('Cc:');
  });

  it('keeps the attachment list when the body is truncated to budget', () => {
    const longBody = 'word '.repeat(20000);
    const out = assembleMsgOutput(
      {
        ...baseView,
        body: { text: longBody, format: 'plain' },
        attachments: [{ fileName: 'a.pdf', contentLength: 10, savedPath: 'input/a.pdf' }],
      },
      100,
      'mail.msg',
    );
    expect(out).toContain('input/a.pdf');
    expect(out).toContain('Subject: Quarterly report');
    expect(out.length).toBeLessThan(longBody.length);
  });

  it('includes the CC line when CC recipients exist', () => {
    const out = formatMsgOutput({ ...baseView, cc: [{ email: 'carol@example.com' }] });
    expect(out).toContain('Cc: carol@example.com');
  });
});

describe('executeReadMsg (integration)', () => {
  let workspace: string;
  const ctx = (): ToolContext => ({ workspacePath: workspace, editAllowed: true });

  beforeEach(() => {
    workspace = fs.mkdtempSync(path.join(os.tmpdir(), 'readmsg-'));
    fs.copyFileSync(FIXTURE, path.join(workspace, 'mail.msg'));
  });

  afterEach(() => {
    fs.rmSync(workspace, { recursive: true, force: true });
  });

  it('extracts headers and body from a real .msg file', async () => {
    const result = await executeReadMsg({ file_path: 'mail.msg' }, ctx());
    expect(result.isError).toBeFalsy();
    expect(result.output).toContain('Subject: attachmentFiles');
    expect(result.output).toContain('From: hmailuser <hmailuser@hmailserver.test>');
    expect(result.output).toContain('To: hmailuser@hmailserver.test');
    expect(result.output).toContain('attachmentFiles');
  });

  it('saves attachments to input/ and lists them', async () => {
    const result = await executeReadMsg({ file_path: 'mail.msg' }, ctx());
    expect(result.output).toContain('Attachments (3)');
    for (const [name, size] of [
      ['jpg.jpg', 726],
      ['png.png', 134],
      ['tif.tif', 664],
    ] as const) {
      const saved = path.join(workspace, 'input', name);
      expect(fs.existsSync(saved)).toBe(true);
      expect(fs.statSync(saved).size).toBe(size);
      expect(result.output).toContain(path.join('input', name));
    }
  });

  it('rejects paths outside the workspace', async () => {
    const result = await executeReadMsg({ file_path: '../../etc/passwd' }, ctx());
    expect(result.isError).toBe(true);
  });

  it('reports a clear error for a non-.msg file', async () => {
    fs.writeFileSync(path.join(workspace, 'junk.msg'), 'not a real msg file');
    const result = await executeReadMsg({ file_path: 'junk.msg' }, ctx());
    expect(result.isError).toBe(true);
    expect(result.output).toContain('ReadMsg');
  });

  it('does not write attachments in a read-only phase', async () => {
    const result = await executeReadMsg(
      { file_path: 'mail.msg' },
      { workspacePath: workspace, editAllowed: false },
    );
    expect(result.isError).toBeFalsy();
    expect(fs.existsSync(path.join(workspace, 'input', 'jpg.jpg'))).toBe(false);
    expect(result.output).toContain('read-only');
  });

  it('does not overwrite an existing file in input/', async () => {
    fs.mkdirSync(path.join(workspace, 'input'), { recursive: true });
    fs.writeFileSync(path.join(workspace, 'input', 'jpg.jpg'), 'pre-existing');
    const result = await executeReadMsg({ file_path: 'mail.msg' }, ctx());
    expect(fs.readFileSync(path.join(workspace, 'input', 'jpg.jpg'), 'utf8')).toBe('pre-existing');
    expect(fs.existsSync(path.join(workspace, 'input', 'jpg-1.jpg'))).toBe(true);
    expect(result.output).toContain('jpg-1.jpg');
  });

  it('rejects files exceeding the configured size limit', async () => {
    const result = await executeReadMsg(
      { file_path: 'mail.msg' },
      { workspacePath: workspace, editAllowed: true, toolsConfig: { officeMsgMaxSizeMb: 0.001 } },
    );
    expect(result.isError).toBe(true);
    expect(result.output).toMatch(/size|limit|too large/i);
  });

  it('is registered and routed through the office module dispatch', async () => {
    expect(OFFICE_TOOL_DEFS.ReadMsg).toBeDefined();
    const result = await officeExecuteTool('ReadMsg', { file_path: 'mail.msg' }, ctx());
    expect(result?.isError).toBeFalsy();
    expect(result?.output).toContain('Subject: attachmentFiles');
  });
});