xiaoxia-saas/apps/web/src/test/api/uploadDedup.test.ts

/**
 * 上传去重/幂等工具单测（Issue #1714）
 */
import { describe, it, expect, vi } from "vitest"
import {
  computeFileHash,
  findDuplicateInQueue,
  HASH_FULL_READ_LIMIT,
  HASH_SAMPLE_CHUNK,
  makeClientUploadId,
  makeFileFingerprint,
} from "@/api/assets/uploadDedup"

const makeFile = (name: string, size = 100, lastModified = 1_700_000_000_000) =>
  new File([new Uint8Array(size)], name, { type: "video/mp4", lastModified })

describe("makeFileFingerprint", () => {
  it("同一文件（name+size+lastModified 相同）指纹一致", () => {
    const a = makeFile("a.mp4", 1000, 12345)
    const b = makeFile("a.mp4", 1000, 12345)
    expect(makeFileFingerprint(a)).toBe(makeFileFingerprint(b))
  })

  it("文件名/大小/修改时间任一不同指纹即不同", () => {
    const base = makeFile("a.mp4", 1000, 100)
    expect(makeFileFingerprint(base)).not.toBe(makeFileFingerprint(makeFile("b.mp4", 1000, 100)))
    expect(makeFileFingerprint(base)).not.toBe(makeFileFingerprint(makeFile("a.mp4", 1001, 100)))
    expect(makeFileFingerprint(base)).not.toBe(makeFileFingerprint(makeFile("a.mp4", 1000, 101)))
  })
})

describe("findDuplicateInQueue", () => {
  const queue = [
    { fileKey: "k1", status: "preparing" },
    { fileKey: "k2", status: "uploading" },
    { fileKey: "k3", status: "ingesting" },
    { fileKey: "k4", status: "done" },
    { fileKey: "k5", status: "error" },
  ]

  it("在途状态（preparing/uploading/ingesting/done）命中重复", () => {
    expect(findDuplicateInQueue(queue, "k1")?.status).toBe("preparing")
    expect(findDuplicateInQueue(queue, "k2")?.status).toBe("uploading")
    expect(findDuplicateInQueue(queue, "k3")?.status).toBe("ingesting")
    expect(findDuplicateInQueue(queue, "k4")?.status).toBe("done")
  })

  it("未命中返回 null", () => {
    expect(findDuplicateInQueue(queue, "missing")).toBeNull()
  })

  it("排除 error 状态后，失败项不算重复（允许重新激活）", () => {
    expect(findDuplicateInQueue(queue, "k5", ["error"])).toBeNull()
  })

  it("同时排除 done 后，已完成项也不算重复", () => {
    expect(findDuplicateInQueue(queue, "k4", ["error", "done"])).toBeNull()
    // 但在途的仍然命中
    expect(findDuplicateInQueue(queue, "k1", ["error", "done"])).not.toBeNull()
  })
})

describe("makeClientUploadId", () => {
  it("生成带前缀且互不相同的幂等 token", () => {
    const ids = new Set(Array.from({ length: 20 }, () => makeClientUploadId()))
    expect(ids.size).toBe(20)
    for (const id of ids) expect(id.startsWith("up_")).toBe(true)
  })
})

describe("computeFileHash", () => {
  it("相同内容 hash 一致、不同内容 hash 不同", async () => {
    const f1 = makeFile("a.mp4", 4096)
    const f2 = makeFile("b.mp4", 4096)
    // 两个文件都是 0 填充，内容相同 → hash 一致
    expect(await computeFileHash(f1)).toBe(await computeFileHash(f2))

    const f3 = new File([new Uint8Array(4096).fill(7)], "c.mp4", { type: "video/mp4" })
    expect(await computeFileHash(f1)).not.toBe(await computeFileHash(f3))
  })

  it("返回 64 位十六进制（SHA-256，与后端 file_hash 长度一致）", async () => {
    const hash = await computeFileHash(makeFile("a.mp4", 1024))
    expect(hash).toMatch(/^[0-9a-f]{64}$/)
  })
})

describe("computeFileHash 大文件抽样（>64MB）", () => {
  it("抽样路径正常返回 64 位 hex，且大小不同则 hash 不同", async () => {
    // mock 一个「声称」300MB 的 File：slice 返回小 buffer 即可，不真分配 300MB
    const makeBig = (declaredSize: number, head: number) => {
      const f = new File([new Uint8Array([head, 2, 3])], "big.mov", { type: "video/quicktime" })
      Object.defineProperty(f, "size", { value: declaredSize, configurable: true })
      // slice 仍按真实内容返回小片段（头尾片段内容由底层小 buffer 决定）
      return f
    }
    const h1 = await computeFileHash(makeBig(300 * 1024 * 1024, 1))
    const h2 = await computeFileHash(makeBig(301 * 1024 * 1024, 1))
    expect(h1).toMatch(/^[0-9a-f]{64}$/)
    // 声明大小不同 → 写入的 64 位 size 字段不同 → hash 必须不同（锁定 setBigUint64 路径）
    expect(h1).not.toBe(h2)
  })

  it("≤64MB 走全量读取（slice 一次覆盖整个文件）", async () => {
    const f = new File([new Uint8Array(1024).fill(9)], "full.mp4", { type: "video/mp4" })
    Object.defineProperty(f, "size", { value: HASH_FULL_READ_LIMIT, configurable: true })
    const sliceSpy = vi.spyOn(f, "slice")
    await computeFileHash(f)
    // 全量路径：唯一一次 slice 为 (0, size)
    expect(sliceSpy).toHaveBeenCalledTimes(1)
    expect(sliceSpy).toHaveBeenCalledWith(0, HASH_FULL_READ_LIMIT)
    sliceSpy.mockRestore()
  })

  it(">64MB 只读取头尾各 16MB 抽样，绝不整文件读入内存", async () => {
    const f = new File([new Uint8Array(1024).fill(9)], "big.mp4", { type: "video/mp4" })
    Object.defineProperty(f, "size", { value: HASH_FULL_READ_LIMIT + 1, configurable: true })
    const sliceSpy = vi.spyOn(f, "slice")
    await computeFileHash(f)
    // 抽样路径：两次 slice —— 头部 (0, 16MB) 与尾部 (size-16MB, size)
    expect(sliceSpy).toHaveBeenCalledTimes(2)
    expect(sliceSpy).toHaveBeenNthCalledWith(1, 0, HASH_SAMPLE_CHUNK)
    expect(sliceSpy).toHaveBeenNthCalledWith(
      2,
      HASH_FULL_READ_LIMIT + 1 - HASH_SAMPLE_CHUNK,
      HASH_FULL_READ_LIMIT + 1,
    )
    sliceSpy.mockRestore()
  })
})