feat: add pluggable search service registry

2026-07-23 21:38:36 +08:00
parent 48e9f68211
commit 4e36b98c9e
7 changed files with 506 additions and 21 deletions
@@ -1,13 +1,74 @@
 import test from 'node:test';
 import assert from 'node:assert/strict';
 import { buildCitations, isSafeHttpUrl, validateSearchRequest } from './search-capability.mjs';
-import { MINDSEARCH_DEFAULT_CONFIG, normalizeMindSearchConfig } from './mindsearch-config.mjs';
+import {
+  MINDSEARCH_DEFAULT_CONFIG,
+  normalizeMindSearchConfig,
+  probeMindSearchService,
+  resolveMindSearchService,
+} from './mindsearch-config.mjs';
 import { buildAgentExtensionPolicy, DEFAULT_USER_CAPABILITIES } from './capabilities.mjs';
-import { searchGithubCode, searchSearxng } from './mindsearch-providers.mjs';
+import {
+  searchGithubCode,
+  searchResearchService,
+  searchSearxng,
+  startResearchTask,
+} from './mindsearch-providers.mjs';

 test('MindSearch defaults to disabled and normalizes unsafe modes', () => {
  assert.equal(MINDSEARCH_DEFAULT_CONFIG.enabled, false);
-  assert.deepEqual(normalizeMindSearchConfig({ enabled: true, mode: 'invalid', providers: { github: true } }), { enabled: false, mode: 'off', providers: { searxng: false, github: true, reader: false }, settings: { searxngEndpoint: '', maxResults: 10, timeoutMs: 8000, readerMaxChars: 12000 } });
+  const config = normalizeMindSearchConfig({ enabled: true, mode: 'invalid', providers: { github: true } });
+  assert.equal(config.enabled, false);
+  assert.equal(config.mode, 'off');
+  assert.deepEqual(config.providers, { searxng: false, github: true, reader: false });
+  assert.equal(config.services.find((service) => service.id === 'github').enabled, true);
+  assert.deepEqual(config.routes, { web: 'searxng', news: 'searxng', code: 'github', read: 'reader', research: '' });
+});
+
+test('MindSearch service registry accepts research orchestrators and rejects unsafe endpoints', () => {
+  const config = normalizeMindSearchConfig({
+    enabled: true,
+    mode: 'assist',
+    services: [
+      {
+        id: 'research-main',
+        name: 'Research',
+        kind: 'orchestrator',
+        adapter: 'research-http',
+        endpoint: 'http://127.0.0.1:20100',
+        healthPath: '/health',
+        capabilities: ['search.web', 'research.execute', 'research.execute', 'INVALID CAPABILITY'],
+        enabled: true,
+        timeoutMs: 500000,
+      },
+      { id: 'metadata', adapter: 'research-http', endpoint: 'http://169.254.169.254/latest', enabled: true },
+    ],
+    routes: { research: 'research-main', web: 'research-main' },
+  });
+  assert.equal(config.services.some((service) => service.id === 'metadata'), false);
+  assert.deepEqual(config.services.find((service) => service.id === 'research-main').capabilities, ['search.web', 'research.execute']);
+  assert.equal(config.services.find((service) => service.id === 'research-main').timeoutMs, 120000);
+  assert.equal(resolveMindSearchService(config, 'research').id, 'research-main');
+});
+
+test('MindSearch service probe supports loopback SearXNG without exposing arbitrary response data', async () => {
+  let requested;
+  const result = await probeMindSearchService({
+    id: 'searxng-local',
+    name: 'Local',
+    kind: 'provider',
+    adapter: 'searxng',
+    endpoint: 'http://127.0.0.1:8080/search',
+    enabled: true,
+  }, {
+    fetchImpl: async (url) => {
+      requested = String(url);
+      return { ok: true, status: 200, json: async () => ({ results: [{ title: 'A' }] }) };
+    },
+  });
+  assert.match(requested, /q=OpenAI/);
+  assert.equal(result.ok, true);
+  assert.equal(result.resultCount, 1);
 });

 test('search request validation and citations are deterministic', () => {
@@ -42,3 +103,20 @@ test('GitHub code adapter sends bounded queries and normalizes results', async (
  assert.equal(requested.options.headers.accept, 'application/vnd.github+json');
  assert.equal(result[0].source, 'github');
 });
+
+test('Research service adapter uses the stable HTTP contract', async () => {
+  const requests = [];
+  const fetchImpl = async (url, options) => {
+    requests.push({ url: String(url), options });
+    if (String(url).endsWith('/v1/research')) {
+      return { ok: true, json: async () => ({ task_id: 'task-1', status: 'running' }) };
+    }
+    return { ok: true, json: async () => ({ results: [{ title: 'Report', url: 'https://example.com/report', snippet: 'evidence' }] }) };
+  };
+  const results = await searchResearchService('AI Agent', { endpoint: 'http://research.local', limit: 5, fetchImpl });
+  const task = await startResearchTask('Analyze AI Agent', { endpoint: 'http://research.local', depth: 'deep', fetchImpl });
+  assert.equal(results[0].source, 'research-service');
+  assert.deepEqual(task, { taskId: 'task-1', status: 'running', service: 'research-service' });
+  assert.equal(JSON.parse(requests[0].options.body).query, 'AI Agent');
+  assert.deepEqual(JSON.parse(requests[1].options.body), { question: 'Analyze AI Agent', depth: 'deep' });
+});