feat: add online FAQ/KB sync without old bridge

2026-10-23 11:29:43 +08:00
parent f526696f93
commit 81b49c9350
24 changed files with 1793 additions and 5 deletions
@@ -0,0 +1,73 @@
+#!/usr/bin/env node
+'use strict';
+
+// 在线 FAQ/KB 同步（不依赖旧 bridge）：
+//   抓官网 FAQ JSON-LD → 更新 bridge/kb/*.md → parent/child + embedding → 原子切换 kb_docs
+// 用法:
+//   node scripts/sync-kb-online.js [--force]
+// 需要环境变量:
+//   RAG_PG_URL / EMB_BASE_URL / EMB_API_KEY / EMB_MODEL / EMB_DIM
+const fs = require('fs');
+const path = require('path');
+const { loadConfig } = require('../bridge/src/config');
+const { createEmbeddingClient } = require('../bridge/src/embedding');
+const { createKbIndexer } = require('../bridge/src/kb-indexer');
+const { createFaqFetcher, loadFaqConfig } = require('../bridge/src/faq-fetcher');
+const { updateMarkdownFile } = require('../bridge/src/faq-update');
+
+const ROOT = path.resolve(__dirname, '..');
+const env = {};
+for (const line of fs.readFileSync(path.join(ROOT, '.env'), 'utf8').split('\n')) {
+  if (!line.includes('=') || line.trim().startsWith('#')) continue;
+  const i = line.indexOf('=');
+  env[line.slice(0, i).trim()] = line.slice(i + 1).trim();
+}
+const cfg = loadConfig({ ...env, ...process.env });
+const force = process.argv.includes('--force');
+const localOnly = process.argv.includes('--local-only');
+const KB_DIR = path.join(ROOT, 'bridge', 'kb');
+const FAQ_CONFIG = path.join(ROOT, 'bridge', 'config', 'faq.json');
+
+(async () => {
+  if (!cfg.ragPgUrl) throw new Error('RAG_PG_URL is not configured');
+  if (!cfg.embApiKey || !cfg.embBaseUrl) throw new Error('EMB_* is not configured');
+  const faqCfg = loadFaqConfig(FAQ_CONFIG);
+  if (!faqCfg.enabled) throw new Error('faq.json disabled');
+
+  let changedFiles = 0;
+  let added = 0;
+  let updated = 0;
+  if (!localOnly) {
+    const fetcher = createFaqFetcher(faqCfg);
+    const fetched = await fetcher.fetchAll();
+    for (const page of fetched) {
+      if (page.error || !page.questions.length) {
+        console.log(JSON.stringify({ ts: new Date().toISOString(), faq_fetch: 'skip', file: page.file, error: page.error || 'no_questions' }));
+        continue;
+      }
+      const filePath = path.join(KB_DIR, page.file);
+      const res = updateMarkdownFile(filePath, page.questions);
+      if (res.changed) changedFiles += 1;
+      added += res.added || 0;
+      updated += res.updated || 0;
+      console.log(JSON.stringify({ ts: new Date().toISOString(), faq_fetch: 'ok', file: page.file, questions: page.questions.length, added: res.added || 0, updated: res.updated || 0 }));
+    }
+  }
+
+  if (!localOnly && !force && changedFiles === 0) {
+    console.log(JSON.stringify({ ts: new Date().toISOString(), kb_sync: 'skipped', reason: 'no_faq_changes' }));
+    return;
+  }
+
+  const emb = createEmbeddingClient(cfg);
+  const indexer = createKbIndexer({ pool: emb.pool, embeddingClient: emb, kbDir: KB_DIR, dim: cfg.embDim });
+  try {
+    const result = await indexer.reindex({ note: `sync-kb-online ${new Date().toISOString().slice(0, 10)}` });
+    console.log(JSON.stringify({ ts: new Date().toISOString(), kb_sync: 'done', changed_files: changedFiles, added, updated, ...result }));
+  } finally {
+    await emb.pool.end().catch(() => {});
+  }
+})().catch((err) => {
+  console.error('[sync-kb-online]', err.message);
+  process.exit(1);
+});