在 Edge 上做全文搜索

文章发布 / 更新
  → 提取标题、摘要、正文、标签
  → 构建 Orama 索引
  → gzip 压缩
  → 存入 KV
  → 记录索引版本号

搜索请求到来
  → 先看当前 Worker 进程内有没有最新版本索引
    → 有：直接搜
    → 没有：从 KV 读出索引
      → 解压
      → 加载到内存
      → 执行搜索

export const searchSchema = {
  id: "string",
  slug: "string",
  title: "string",
  summary: "string",
  content: "string",
  tags: "string[]",
} as const;

const segmenter = new Intl.Segmenter("zh-CN", { granularity: "word" });

export const chineseTokenizerConfig: Tokenizer = {
  language: "chinese",
  tokenize: (text: string) => {
    return Array.from(segmenter.segment(text))
      .filter((x) => x.isWordLike)
      .map((x) => x.segment.toLowerCase());
  },
  normalizationCache: new Map(),
};

export async function persistOramaDb(env: Env, db: MyOramaDB) {
  const raw = save(db);
  const compressed = await compressRaw(raw);
  await env.KV.put(KV_KEY, compressed);

  const newVersion = Date.now().toString();
  const meta = {
    version: newVersion,
    updatedAt: new Date().toISOString(),
    sizeInBytes: compressed.byteLength,
  };
  await env.KV.put(KV_META_KEY, JSON.stringify(meta));
  setOramaDb(db, newVersion);
  return newVersion;
}

let cachedDb: MyOramaDB | null = null;
let cachedVersion: string | null = null;
let inflight: Promise<MyOramaDB> | null = null;

if (cachedDb && cachedVersion === latestVersion) return cachedDb;
if (inflight) return inflight;

const result = await oramaSearch(db, {
  term: data.q,
  limit: Math.min(data.limit, 25),
});

return result.hits.map((hit) => {
  const { document, score } = hit;
  return {
    post: {
      id: document.id,
      slug: document.slug,
      title: document.title,
      summary: document.summary,
      tags: document.tags,
    },
    score,
    matches: {
      title: buildSnippet(...),
      summary: buildSnippet(...),
      contentSnippet: buildSnippet(...),
    },
  };
});

export async function upsert(context: { env: Env }, data: UpsertSearchDocInput) {
  const db = await getOramaDb(context.env);

  try {
    await remove(db, data.id.toString());
  } catch {}

  const plain = convertToPlainText(data.contentJson ?? null);
  const content =
    plain.length > CONTENT_SLICE ? plain.slice(0, CONTENT_SLICE) : plain;

  await insert(db, {
    id: data.id.toString(),
    slug: data.slug,
    title: data.title,
    summary: data.summary || content.slice(0, SNIPPET_SLICE),
    content,
    tags: data.tags ?? [],
  });

  await persistOramaDb(context.env, db);
}