宜博 / lemonboss

/**
 * Unit Tests — graph-engine 的持续性存活扫描（45 号文档 Part A）
 *
 * 场景：服务器主进程没有重启，但驱动某个 agent_call 节点的 Claude CLI 会话在调用中途
 * 死掉/被打断，`graph_run_nodes.status` 永远停在 'running'。开机一次性孤儿扫描
 * （`sweepOrphanedRunningNodes`，按「早于本次进程启动时间」判定）看不到这种情况——本文件
 * 测的是它的持续性版本：`sweepStaleRunningAgentCalls`，按「running 超过一个阈值」判定，
 * 且只扫 `node_kind='agent_call'`（subgraph 节点的"运行时长"横跨多个内部调用，不适用
 * 同一固定阈值，本次不做）。
 *
 * 隔离手法与 graph-engine-orphan-sweep.test.js 一致。
 */
import { describe, it, expect, beforeAll, afterAll, beforeEach } from 'vitest';
import fs from 'fs';
import os from 'os';
import path from 'path';

describe('graph-engine 持续性存活扫描', () => {
  let wdb;
  let engine;
  let runsDir;
  const TEST_DB_PATH = '/tmp/test-graph-engine-stale-agent-call-sweep.sqlite';

  beforeAll(() => {
    for (const f of [TEST_DB_PATH, `${TEST_DB_PATH}-shm`, `${TEST_DB_PATH}-wal`]) {
      try { fs.unlinkSync(f); } catch {}
    }
    process.env.GRAPH_DB_PATH = TEST_DB_PATH;
    runsDir = fs.mkdtempSync(path.join(os.tmpdir(), 'graph-engine-stale-agent-call-sweep-'));
    process.env.GRAPH_RUNS_DIR = runsDir;
    for (const m of [
      '../../services/graph-db',
      '../../services/graph-service',
      '../../services/org-graph-service',
      '../../services/work-graph-db',
      '../../services/graph-run-sse',
      '../../services/graph-engine',
    ]) {
      delete require.cache[require.resolve(m)];
    }
    wdb = require('../../services/work-graph-db');
    engine = require('../../services/graph-engine');
  });

  afterAll(() => {
    for (const f of [TEST_DB_PATH, `${TEST_DB_PATH}-shm`, `${TEST_DB_PATH}-wal`]) {
      try { fs.unlinkSync(f); } catch {}
    }
    delete process.env.GRAPH_DB_PATH;
    try { fs.rmSync(runsDir, { recursive: true, force: true }); } catch {}
    delete process.env.GRAPH_RUNS_DIR;
  });

  beforeEach(() => {
    const db = wdb.getDb();
    db.pragma('foreign_keys = OFF');
    db.exec(`
      DELETE FROM graph_run_loop_records;
      DELETE FROM graph_run_nodes;
      DELETE FROM graph_runs;
      DELETE FROM work_graph_edges;
      DELETE FROM work_graph_nodes;
      DELETE FROM work_graphs;
      DELETE FROM system_agents;
    `);
    db.pragma('foreign_keys = ON');
    engine.setAgentExecutor(null);
  });

  let seq = 0;
  const nextCompany = () => `co-stale-${++seq}`;

  function ensureAgents(ids) {
    const db = wdb.getDb();
    const now = new Date().toISOString();
    const ins = db.prepare(`
      INSERT OR IGNORE INTO system_agents
        (id, name, scope, owner_company_id, legacy_path, status, created_at, updated_at)
      VALUES (?, ?, 'public', NULL, ?, 'active', ?, ?)
    `);
    for (const id of new Set(ids)) ins.run(id, id, `system/agents/${id}`, now, now);
  }

  function buildWorkRun({ companyId, nodes, edges }) {
    ensureAgents(nodes);
    const g = wdb.createWorkGraph({ companyId, nodes, edges });
    expect(g.errors || []).toEqual([]);
    const run = wdb.createRun({ graphKind: 'work', graphId: g.id, companyId, executionMode: 'manual' });
    engine.markRunReady(run.id);
    return { runId: run.id };
  }

  const nodeByName = (runId, name) => wdb.getRunState(runId).nodes.find((n) => n.name === name);

  function markAsRunning(runId, nodeName, startedAtIso) {
    const node = nodeByName(runId, nodeName);
    wdb.getDb()
      .prepare(`UPDATE graph_run_nodes SET status = 'running', started_at = ?, updated_at = ? WHERE id = ?`)
      .run(startedAtIso, startedAtIso, node.id);
    return node.id;
  }

  const NOW = '2026-08-31T10:00:00.000Z';

  it('agent_call 节点 running 超过阈值判定为存活超时，走 infra_flake 原地重跑', () => {
    const companyId = nextCompany();
    const { runId } = buildWorkRun({ companyId, nodes: ['w'], edges: [] });
    // 默认阈值 25 分钟；30 分钟前开始 = 超阈值
    markAsRunning(runId, 'w', '2026-08-31T09:30:00.000Z');

    const result = engine.sweepStaleRunningAgentCalls({ nowIso: NOW });

    expect(result.scanned).toBe(1);
    expect(result.swept).toHaveLength(1);
    expect(result.swept[0]).toMatchObject({ runId, infraRerun: true });
    const fresh = nodeByName(runId, 'w');
    expect(fresh.status).toBe('ready');
    expect(fresh.attempt).toBe(0);
  });

  it('running 未超过阈值的节点不受影响（大概率还在真实跑）', () => {
    const companyId = nextCompany();
    const { runId } = buildWorkRun({ companyId, nodes: ['w'], edges: [] });
    // 5 分钟前开始，远低于默认 25 分钟阈值
    markAsRunning(runId, 'w', '2026-08-31T09:55:00.000Z');

    const result = engine.sweepStaleRunningAgentCalls({ nowIso: NOW });

    expect(result.scanned).toBe(0);
    expect(nodeByName(runId, 'w').status).toBe('running');
  });

  it('阈值可调：传更短的 thresholdMs 会让原本"没超时"的节点也被扫到', () => {
    const companyId = nextCompany();
    const { runId } = buildWorkRun({ companyId, nodes: ['w'], edges: [] });
    markAsRunning(runId, 'w', '2026-08-31T09:55:00.000Z'); // 5 分钟前

    const result = engine.sweepStaleRunningAgentCalls({ nowIso: NOW, thresholdMs: 60 * 1000 }); // 阈值 1 分钟

    expect(result.scanned).toBe(1);
    expect(nodeByName(runId, 'w').status).toBe('ready');
  });

  it('subgraph 节点不在扫描范围内，即使 running 很久也不受影响', () => {
    const companyId = nextCompany();
    ensureAgents(['s']);
    const g = wdb.createWorkGraph({ companyId, nodes: [{ nodeName: 's', nodeKind: 'subgraph' }], edges: [] });
    expect(g.errors || []).toEqual([]);
    const run = wdb.createRun({ graphKind: 'work', graphId: g.id, companyId, executionMode: 'manual' });
    engine.markRunReady(run.id);
    markAsRunning(run.id, 's', '2026-08-31T08:00:00.000Z'); // 2 小时前，远超阈值

    const result = engine.sweepStaleRunningAgentCalls({ nowIso: NOW });

    expect(result.scanned).toBe(0);
    expect(nodeByName(run.id, 's').status).toBe('running');
  });

  it('没有任何超时节点时是安静的 no-op', () => {
    const companyId = nextCompany();
    buildWorkRun({ companyId, nodes: ['w'], edges: [] }); // 节点是 ready，不是 running
    const result = engine.sweepStaleRunningAgentCalls({ nowIso: NOW });
    expect(result).toEqual({ scanned: 0, swept: [] });
  });
});