构建文档版本存储

项目概述

本项目实现一个面向 document versioning 的完整系统，核心要解决的问题是 keeping every revision of a mutable document addressable without storing full copies for each edit。整体由 Content-addressed blob、Revision tree、Diff serializer、Snapshot materializer、Garbage collector 五个模块组成，每个模块单一职责、可独立测试，最后通过集成测试验证端到端正确性。设计上把 a long branch chain whose materialization requires walking thousands of diffs to reconstruct the current state 作为第一类需求先行处理，而非事后再补。

架构与关键决策

数据模型为 a tree of revision nodes pointing at content-addressed blobs, where each node stores a diff against its parent。Content-addressed blob 定义其余阶段共同依赖的数据形状并在边界处校验输入；Revision tree 承载 document versioning 的正确性逻辑；Diff serializer 专门处理 a long branch chain whose materialization requires walking thousands of diffs to reconstruct the current state，在失败路径上给出明确的拒绝、重试或降级策略；Snapshot materializer 提供可观测性，每次操作输出带关联 id 的结构化事件；Garbage collector 在过载时按既定方式降级，避免过载演变为数据损坏。配置与代码分离，使风险变更可经 diff 审阅。

实施与测试

按五个阶段依次实现 Content-addressed blob、Revision tree、Diff serializer、Snapshot materializer、Garbage collector，每阶段配套聚焦单测。集成测试组合各模块并驱动真实 document versioning 场景：先验证正常路径，再刻意触发 a long branch chain whose materialization requires walking thousands of diffs to reconstruct the current state 并断言系统降级而非损坏。所有依赖均在进程内或被桩替换，用例在 CI 中确定可复现。

限制与后续

• 单区域设计，多区域复制超出范围。
• Content-addressed blob 默认进程内，未水平共享。
• 无鉴权层，假定部署在强制身份的边缘之后。

代码示例

# setup for document-version-store
pnpm init
pnpm add express blake3 level
pnpm add -D typescript tsx vitest

document_version_store/
├── src/index.ts
├── src/content-addressed-blob.ts
├── src/revision-tree.ts
├── tests/document_version_store.test.ts
└── package.json

// Content-addressed blob: document versioning core
export function contentaddressedblob(input: ContentaddressedblobInput): ContentaddressedblobResult {
  // Build the content-addressed blob that anchors document versioning and defines the core data shape.
  return processContentaddressedblob(input);
}

// Revision tree — handles document versioning Add the revision tree so
export class Revisiontree {
  constructor(private opts: RevisiontreeOptions) {}
  async run(ctx: Context): Promise<RevisiontreeOutput> {
    return this.execute(ctx);
  }
}

// Diff serializer: document versioning stage
export const diffserializer = (next: Next) => async (req: Request) => {
  const out = await transform(req);
  return next(out);
};

// Snapshot materializer for document versioning
export async function snapshotmaterializer(stream: AsyncIterable<Chunk>): Promise<Aggregate> {
  let acc = init();
  for await (const c of stream) acc = merge(acc, c);
  return acc;
}

// Garbage collector: document versioning finalization
export function garbagecollector(state: State): Result {
  validate(state);
  return finalize(state);
}

# run the document versioning test suite
pnpm vitest run