Codeman/test/agent-codex-identity.test.ts

/** @fileoverview Strict bounded native rollout attribution without viewer heuristics. */
import { afterEach, beforeEach, describe, expect, it } from 'vitest';
import { mkdtemp, mkdir, writeFile, rm, symlink } from 'node:fs/promises';
import { join } from 'node:path';
import { tmpdir } from 'node:os';
import {
  __clearCodexAttributionMemo,
  findExactCodexConversation,
  resolveEffectiveCodexHome,
} from '../src/agent-codex-identity.js';
const roots: string[] = [];
afterEach(async () => {
  await Promise.all(roots.splice(0).map((root) => rm(root, { recursive: true, force: true })));
});
const A = '11111111-1111-4111-8111-111111111111';
const B = '22222222-2222-4222-8222-222222222222';
async function fixture() {
  const root = await mkdtemp(join(tmpdir(), 'agent-attribution-'));
  roots.push(root);
  const home = join(root, 'home');
  await mkdir(join(home, 'sessions'), { recursive: true });
  const options = { codexHome: home, sessionId: 'managed', workingDir: root };
  const put = async (id: string, payload: object = {}, text?: string) => {
    const path = join(home, 'sessions', `rollout-${id}.jsonl`);
    await writeFile(
      path,
      text ??
        JSON.stringify({
          type: 'session_meta',
          payload: { id, cwd: root, originator: 'codeman_managed', ...payload },
        }) + '\n'
    );
    return path;
  };
  return { root, home, options, put };
}
it('requires one complete native session_meta matching the rollout ID, managed originator and canonical workspace', async () => {
  const f = await fixture();
  const path = await f.put(A);
  expect(await findExactCodexConversation(f.options)).toEqual({ id: A, path });
  await f.put(A, { id: B });
  expect(await findExactCodexConversation(f.options)).toBeUndefined();
});
it('never chooses newest cwd files or an ambiguous managed originator', async () => {
  const f = await fixture();
  await f.put(A, { originator: 'codex_cli' });
  expect(await findExactCodexConversation(f.options)).toBeUndefined();
  await f.put(A);
  await f.put(B);
  expect(await findExactCodexConversation(f.options)).toBeUndefined();
});
it('allows an explicit resume identity but still requires native metadata and the workspace', async () => {
  const f = await fixture();
  const path = await f.put(A, { originator: 'original-launch' });
  expect(await findExactCodexConversation({ ...f.options, resumeSessionId: A })).toEqual({ id: A, path });
  await f.put(B);
  expect(await findExactCodexConversation({ ...f.options, resumeSessionId: A })).toBeUndefined();
  await f.put(A, { cwd: f.home });
  expect(await findExactCodexConversation({ ...f.options, resumeSessionId: A })).toBeUndefined();
});
it.each(['partial', 'duplicate', 'oversized'] as const)('refuses %s metadata', async (kind) => {
  const f = await fixture();
  const row = JSON.stringify({ type: 'session_meta', payload: { id: A, cwd: f.root, originator: 'codeman_managed' } });
  await f.put(
    A,
    {},
    kind === 'partial'
      ? row
      : kind === 'duplicate'
        ? row + '\n' + row + '\n'
        : row.slice(0, -2) + ',"instructions":"' + 'x'.repeat(140000) + '"}}\n'
  );
  expect(await findExactCodexConversation(f.options)).toBeUndefined();
});
it('enforces traversal/read caps and ignores symlinked directories', async () => {
  const f = await fixture();
  await f.put(A);
  expect(await findExactCodexConversation(f.options, { maxEntries: 0, maxBytes: 1048576 })).toBeUndefined();
  expect(await findExactCodexConversation(f.options, { maxEntries: 8192, maxBytes: 128 })).toBeUndefined();
  await mkdir(join(f.root, 'elsewhere'));
  await symlink(join(f.root, 'elsewhere'), join(f.home, 'sessions', 'alias'));
  expect((await findExactCodexConversation(f.options))?.id).toBe(A);
});
it('resolves per-session, process and default CODEX_HOME relative to the session workspace', () => {
  expect(resolveEffectiveCodexHome('custom', '/tmp/work', { CODEX_HOME: '/other' })).toBe('/tmp/work/custom');
  expect(resolveEffectiveCodexHome(undefined, '/tmp/work', { CODEX_HOME: 'process' })).toBe('/tmp/work/process');
  expect(resolveEffectiveCodexHome(undefined, '/tmp/work', { HOME: '/tmp/user' })).toBe('/tmp/user/.codex');
});

it.each([
  ['trailing-space', 'codex-home ', 'codex-home'],
  ['whitespace-only', '   ', '.codex'],
])(
  'keeps the raw %s CODEX_HOME namespace through the Session getter and attribution helper',
  async (_label, configured, wrongHome) => {
    const { Session } = await import('../src/session.js');
    const f = await fixture();
    const expectedHome = join(f.root, configured);
    const siblingHome = join(f.root, wrongHome);
    const record =
      JSON.stringify({ type: 'session_meta', payload: { id: A, cwd: f.root, originator: 'codeman_managed' } }) + '\n';
    for (const home of [expectedHome, siblingHome]) {
      await mkdir(join(home, 'sessions'), { recursive: true });
      await writeFile(join(home, 'sessions', `rollout-${A}.jsonl`), record);
    }
    const session = new Session({
      id: 'managed',
      workingDir: f.root,
      mode: 'codex',
      envOverrides: { CODEX_HOME: configured },
    });
    try {
      const codexHome = resolveEffectiveCodexHome(session.getCodexHomeOverride(), session.workingDir, { HOME: f.root });
      expect(await findExactCodexConversation({ ...f.options, codexHome })).toEqual({
        id: A,
        path: join(expectedHome, 'sessions', `rollout-${A}.jsonl`),
      });
      expect(session.getCodexHomeOverride()).toBe(configured);
    } finally {
      await session.stop(false);
    }
  }
);

it('attributes an owned rollout among 320 unrelated full-size metadata prefixes', async () => {
  const f = await fixture();
  const path = await f.put(A);
  await Promise.all(
    Array.from({ length: 320 }, (_, n) => {
      const id = `10000000-1111-4111-8111-${String(n).padStart(12, '0')}`;
      return f.put(id, { originator: 'codex_cli', instructions: 'x'.repeat(5000) });
    })
  );
  expect(await findExactCodexConversation(f.options)).toEqual({ id: A, path });
  expect(
    await findExactCodexConversation(f.options, { maxEntries: 8192, maxBytes: 1048576, maxPrefixBytes: 4096 })
  ).toBeUndefined();
});

it.each(['workspace', 'metadata'] as const)(
  'probes %s bounded past the cap before canonical filesystem reads',
  async (where) => {
    const { vi } = await import('vitest');
    const probe = await import('../src/utils/bounded-path-probe.js');
    const f = await fixture();
    const alias = join(f.root, 'workspace-alias');
    await symlink(f.root, alias);
    await f.put(A, { cwd: where === 'metadata' ? alias : f.root });
    const blocked = where === 'workspace' ? f.root : alias;
    const spy = vi
      .spyOn(probe, 'probePathKind')
      .mockImplementation(async (path) => (path === blocked ? 'unknown' : 'directory'));
    try {
      expect(await findExactCodexConversation(f.options)).toBeUndefined();
      expect(spy).toHaveBeenCalledWith(blocked, { pastCap: true });
    } finally {
      spy.mockRestore();
    }
  }
);

describe('remembered attribution on a large, never-pruned store', () => {
  beforeEach(() => __clearCodexAttributionMemo());
  const OWNED = '33333333-3333-4333-8333-333333333333';
  const NEWER = '44444444-4444-4444-8444-444444444444';
  /** codex's own layout: sessions/YYYY/MM/DD/rollout-<ts>-<uuid>.jsonl */
  async function dated(f: Awaited<ReturnType<typeof fixture>>, day: string, id: string, payload: object = {}) {
    const dir = join(f.home, 'sessions', ...day.split('/'));
    await mkdir(dir, { recursive: true });
    const path = join(dir, `rollout-${day.replaceAll('/', '-')}T00-00-00-${id}.jsonl`);
    await writeFile(
      path,
      JSON.stringify({
        type: 'session_meta',
        payload: { id, cwd: f.root, originator: 'codeman_managed', ...payload },
      }) + '\n'
    );
    return path;
  }
  /** `count` unrelated 19 KiB rollouts (a real session_meta size) spread over older days. */
  async function unrelated(f: Awaited<ReturnType<typeof fixture>>, count: number) {
    const instructions = 'x'.repeat(19 * 1024);
    for (let start = 0; start < count; start += 200) {
      await Promise.all(
        Array.from({ length: Math.min(200, count - start) }, (_, k) => {
          const n = start + k;
          const day = `2026/${String(1 + (n % 9)).padStart(2, '0')}/${String(1 + (n % 28)).padStart(2, '0')}`;
          return dated(f, day, `10000000-1111-4111-8111-${String(n).padStart(12, '0')}`, {
            originator: 'codex_cli',
            instructions,
          });
        })
      );
    }
  }
  const small = { maxEntries: 20, maxBytes: 1048576 };

  it('attributes one owned rollout beside 2,100 unrelated full-size rollouts', async () => {
    const f = await fixture();
    await unrelated(f, 2100);
    const path = await dated(f, '2026/10/24', OWNED);
    expect(await findExactCodexConversation(f.options)).toEqual({ id: OWNED, path });
  }, 60_000);

  it('re-verifies only the remembered file on a later call instead of re-walking the store', async () => {
    const f = await fixture();
    await unrelated(f, 300);
    const path = await dated(f, '2026/10/24', OWNED);
    expect(await findExactCodexConversation(f.options)).toEqual({ id: OWNED, path });
    // 20 entries cannot cover 300 rollouts in 9 months of day directories: only the
    // remembered head plus the directories not older than it fit.
    expect(await findExactCodexConversation(f.options, small)).toEqual({ id: OWNED, path });
    __clearCodexAttributionMemo();
    expect(await findExactCodexConversation(f.options, small)).toBeUndefined();
  });

  it('still catches a newer owned rollout (codex /new) after remembering one', async () => {
    const f = await fixture();
    await unrelated(f, 50);
    const path = await dated(f, '2026/10/24', OWNED);
    expect(await findExactCodexConversation(f.options)).toEqual({ id: OWNED, path });
    await dated(f, '2026/10/25', NEWER);
    expect(await findExactCodexConversation(f.options, small)).toBeUndefined();
    expect(await findExactCodexConversation(f.options)).toBeUndefined();
  });

  it('answers unattributed, without a full walk, when the remembered file verifies but its incremental scan runs out', async () => {
    const f = await fixture();
    const path = await dated(f, '2026/10/24', OWNED);
    expect(await findExactCodexConversation(f.options)).toEqual({ id: OWNED, path });
    // 30 rollouts in the remembered day itself: the incremental scan reads them all,
    // so 20 entries cannot cover it, and a full walk (a superset) could not either.
    await Promise.all(
      Array.from({ length: 30 }, (_, n) =>
        dated(f, '2026/10/24', `20000000-2222-4222-8222-${String(n).padStart(12, '0')}`, { originator: 'codex_cli' })
      )
    );
    expect(await findExactCodexConversation(f.options, small)).toBeUndefined();
    // The memo was dropped, so a call with room walks the store and attributes again.
    expect(await findExactCodexConversation(f.options)).toEqual({ id: OWNED, path });
  });

  it('drops the remembered file when it is replaced, and walks again', async () => {
    const f = await fixture();
    const path = await dated(f, '2026/10/24', OWNED);
    expect(await findExactCodexConversation(f.options)).toEqual({ id: OWNED, path });
    await rm(path);
    await dated(f, '2026/10/24', OWNED, { originator: 'codex_cli' });
    expect(await findExactCodexConversation(f.options)).toBeUndefined();
    const moved = await dated(f, '2026/10/23', NEWER);
    expect(await findExactCodexConversation(f.options)).toEqual({ id: NEWER, path: moved });
  });
});