Refactor: passing UploadedFileDTO directly to AddUploadedFileToDataset instead of File Blob

Author: GPortas

docs/useCases.md

@@ -917,14 +917,20 @@ This use case involves adding a file that has been previously uploaded to remote
 
 ```typescript
 import { addUploadedFileToDataset } from '@iqss/dataverse-client-javascript'
+import { UploadedFileDTO } from '@iqss/dataverse-client-javascript'
 
 /* ... */
 
 const datasetId: number | string = 123
-const file: File = new File(['content'], 'example.txt', { type: 'text/plain' })
-const storageId: string = 'some-storage-identifier'
+const uploadedFileDTO: UploadedFileDTO = {
+  fileName: 'the-file-name',
+  storageId: 'localstack1://mybucket:19121faf7e7-2c40322ff54e',
+  checksumType: 'md5',
+  checksumValue: 'ede3d3b685b4e137ba4cb2521329a75e',
+  mimeType: 'text/plain'
+}
 
-addUploadedFileToDataset.execute(datasetId, file, storageId).then(() => {
+addUploadedFileToDataset.execute(datasetId, uploadedFileDTO).then(() => {
   console.log('File added to the dataset successfully.')
 })
 
@@ -935,7 +941,7 @@ _See [use case](../src/files/domain/useCases/AddUploadedFileToDataset.ts) implem
 
 The `datasetId` parameter can be a string, for persistent identifiers, or a number, for numeric identifiers.
 
-The `file` parameter is a subclass of Blob (Binary Large Object) that represents a file.
+The `uploadedFileDTO` parameter is an instance of [UploadedFileDTO](../src/files/domain/dtos/UploadedFileDTO.ts) and includes properties related to the uploaded file. These properties should be calculated from the File Blob object previously sent through the upload file use case.
 
 The `storageId` parameter represents the storage identifier obtained after a successful call to the UploadFile use case.
 

src/files/domain/clients/IDirectUploadClient.ts

@@ -8,6 +8,4 @@ export interface IDirectUploadClient {
     abortController: AbortController,
     destination?: FileUploadDestination
   ): Promise<string>
-
-  addUploadedFileToDataset(datasetId: number | string, file: File, storageId: string): Promise<void>
 }

src/files/domain/useCases/AddUploadedFileToDataset.ts

@@ -1,11 +1,12 @@
 import { UseCase } from '../../../core/domain/useCases/UseCase'
-import { IDirectUploadClient } from '../clients/IDirectUploadClient'
+import { UploadedFileDTO } from '../dtos/UploadedFileDTO'
+import { IFilesRepository } from '../repositories/IFilesRepository'
 
 export class AddUploadedFileToDataset implements UseCase<void> {
-  private directUploadClient: IDirectUploadClient
+  private filesRepository: IFilesRepository
 
-  constructor(directUploadClient: IDirectUploadClient) {
-    this.directUploadClient = directUploadClient
+  constructor(filesRepository: IFilesRepository) {
+    this.filesRepository = filesRepository
   }
 
   /**
@@ -17,12 +18,11 @@ export class AddUploadedFileToDataset implements UseCase<void> {
    * Note: This use case can be used independently of the UploadFile use case, e.g., supporting scenarios in which the file already exists in S3 or has been uploaded via some out-of-band method.
    *
    * @param {number | string} [datasetId] - The dataset identifier, which can be a string (for persistent identifiers) or a number (for numeric identifiers).
-   * @param {File} [file] - The file object that has been uploaded.
-   * @param {string} [storageId] - The storage identifier associated with the uploaded file.
+   * @param {UploadedFileDTO} [uploadedFileDTO] - A file DTO associated with the uploaded file.
    * @returns {Promise<void>} A promise that resolves when the file has been successfully added to the dataset.
    * @throws {DirectUploadClientError} - If there are errors while performing the operation.
    */
-  async execute(datasetId: number | string, file: File, storageId: string): Promise<void> {
-    await this.directUploadClient.addUploadedFileToDataset(datasetId, file, storageId)
+  async execute(datasetId: number | string, uploadedFileDTO: UploadedFileDTO): Promise<void> {
+    await this.filesRepository.addUploadedFileToDataset(datasetId, uploadedFileDTO)
   }
 }

src/files/index.ts

@@ -25,7 +25,7 @@ const getFile = new GetFile(filesRepository)
 const getFileAndDataset = new GetFileAndDataset(filesRepository)
 const getFileCitation = new GetFileCitation(filesRepository)
 const uploadFile = new UploadFile(directUploadClient)
-const addUploadedFileToDataset = new AddUploadedFileToDataset(directUploadClient)
+const addUploadedFileToDataset = new AddUploadedFileToDataset(filesRepository)
 
 export {
   getDatasetFiles,
@@ -66,3 +66,4 @@ export {
 } from './domain/models/FileDataTable'
 export { FileDownloadSizeMode } from './domain/models/FileDownloadSizeMode'
 export { FilesSubset } from './domain/models/FilesSubset'
+export { UploadedFileDTO } from './domain/dtos/UploadedFileDTO'

src/files/infra/clients/DirectUploadClient.ts

@@ -6,12 +6,10 @@ import {
   buildRequestConfig,
   buildRequestUrl
 } from '../../../core/infra/repositories/apiConfigBuilders'
-import * as crypto from 'crypto'
 import { IFilesRepository } from '../../domain/repositories/IFilesRepository'
 import { FileUploadError } from './errors/FileUploadError'
 import { FilePartUploadError } from './errors/FilePartUploadError'
 import { MultipartCompletionError } from './errors/MultipartCompletionError'
-import { AddUploadedFileToDatasetError } from './errors/AddUploadedFileToDatasetError'
 import { UrlGenerationError } from './errors/UrlGenerationError'
 import { MultipartAbortError } from './errors/MultipartAbortError'
 import { FileUploadCancelError } from './errors/FileUploadCancelError'
@@ -21,8 +19,6 @@ export class DirectUploadClient implements IDirectUploadClient {
   private filesRepository: IFilesRepository
   private maxMultipartRetries: number
 
-  private readonly checksumAlgorithm: string = 'md5'
-
   private readonly progressAfterUrlGeneration: number = 10
   private readonly progressAfterFileUpload: number = 100
 
@@ -192,31 +188,4 @@ export class DirectUploadClient implements IDirectUploadClient {
         throw new MultipartCompletionError(fileName, datasetId, error.message)
       })
   }
-
-  public async addUploadedFileToDataset(
-    datasetId: number | string,
-    file: File,
-    storageId: string
-  ): Promise<void> {
-    const fileArrayBuffer = await file.arrayBuffer()
-    const fileBuffer = Buffer.from(fileArrayBuffer)
-    return await this.filesRepository
-      .addUploadedFileToDataset(datasetId, {
-        fileName: file.name,
-        storageId: storageId,
-        checksumType: this.checksumAlgorithm,
-        checksumValue: this.calculateBlobChecksum(fileBuffer),
-        mimeType: file.type
-      })
-      .then(() => undefined)
-      .catch((error) => {
-        throw new AddUploadedFileToDatasetError(file.name, datasetId, error.message)
-      })
-  }
-
-  private calculateBlobChecksum(blob: Buffer): string {
-    const hash = crypto.createHash(this.checksumAlgorithm)
-    hash.update(blob)
-    return hash.digest('hex')
-  }
 }

test/integration/files/DirectUploadClient.test.ts renamed to test/integration/files/DirectUpload.test.ts

@@ -21,18 +21,21 @@ import {
   createSinglepartFileBlob
 } from '../../testHelpers/files/filesHelper'
 import { FileUploadCancelError } from '../../../src/files/infra/clients/errors/FileUploadCancelError'
+import * as crypto from 'crypto'
 
-describe('DirectUploadClient', () => {
+describe('Direct Upload', () => {
   const testCollectionAlias = 'directUploadTestCollection'
   let testDataset1Ids: CreatedDatasetIdentifiers
   let testDataset2Ids: CreatedDatasetIdentifiers
 
-  const filesRepository = new FilesRepository()
-  const sut: DirectUploadClient = new DirectUploadClient(filesRepository)
+  const filesRepositorySut = new FilesRepository()
+  const directUploadSut: DirectUploadClient = new DirectUploadClient(filesRepositorySut)
 
   let singlepartFile: File
   let multipartFile: File
 
+  const checksumAlgorithm: string = 'md5'
+
   beforeEach(() => {
     jest.clearAllMocks()
   })
@@ -79,9 +82,9 @@ describe('DirectUploadClient', () => {
 
     expect(await singlepartFileExistsInBucket(singlepartFileUrl)).toBe(false)
 
-    // Test uploadFile method
+    // Test DirectUpload.uploadFile method
 
-    const actualStorageId = await sut.uploadFile(
+    const actualStorageId = await directUploadSut.uploadFile(
       testDataset1Ids.numericId,
       singlepartFile,
       progressMock,
@@ -96,9 +99,9 @@ describe('DirectUploadClient', () => {
     expect(progressMock).toHaveBeenCalledWith(100)
     expect(progressMock).toHaveBeenCalledTimes(2)
 
-    // Test addUploadedFileToDataset method
+    // Test FilesRepository.addUploadedFileToDataset method
 
-    let datasetFiles = await filesRepository.getDatasetFiles(
+    let datasetFiles = await filesRepositorySut.getDatasetFiles(
       testDataset1Ids.numericId,
       DatasetNotNumberedVersion.LATEST,
       true,
@@ -107,9 +110,20 @@ describe('DirectUploadClient', () => {
 
     expect(datasetFiles.totalFilesCount).toBe(0)
 
-    await sut.addUploadedFileToDataset(testDataset1Ids.numericId, singlepartFile, actualStorageId)
+    const fileArrayBuffer = await singlepartFile.arrayBuffer()
+    const fileBuffer = Buffer.from(fileArrayBuffer)
+
+    const uploadedFileDTO = {
+      fileName: singlepartFile.name,
+      storageId: actualStorageId,
+      checksumType: checksumAlgorithm,
+      checksumValue: calculateBlobChecksum(fileBuffer),
+      mimeType: singlepartFile.type
+    }
 
-    datasetFiles = await filesRepository.getDatasetFiles(
+    await filesRepositorySut.addUploadedFileToDataset(testDataset1Ids.numericId, uploadedFileDTO)
+
+    datasetFiles = await filesRepositorySut.getDatasetFiles(
       testDataset1Ids.numericId,
       DatasetNotNumberedVersion.LATEST,
       true,
@@ -122,7 +136,7 @@ describe('DirectUploadClient', () => {
     expect(datasetFiles.files[0].storageIdentifier).toContain('localstack1://mybucket:')
   })
 
-  test('should upload file and add it to the dataset when there are multiple destination URLs', async () => {
+  test('should upload file when there are multiple destination URLs', async () => {
     const destination = await createTestFileUploadDestination(
       multipartFile,
       testDataset2Ids.numericId
@@ -131,9 +145,9 @@ describe('DirectUploadClient', () => {
     const progressMock = jest.fn()
     const abortController = new AbortController()
 
-    // Test uploadFile method
+    // Test DirectUpload.uploadFile method
 
-    const actualStorageId = await sut.uploadFile(
+    const actualStorageId = await directUploadSut.uploadFile(
       testDataset2Ids.numericId,
       multipartFile,
       progressMock,
@@ -148,9 +162,9 @@ describe('DirectUploadClient', () => {
     expect(progressMock).toHaveBeenCalledWith(100)
     expect(progressMock).toHaveBeenCalledTimes(4)
 
-    // Test addUploadedFileToDataset method
+    // Test FilesRepository.addUploadedFileToDataset method
 
-    let datasetFiles = await filesRepository.getDatasetFiles(
+    let datasetFiles = await filesRepositorySut.getDatasetFiles(
       testDataset2Ids.numericId,
       DatasetNotNumberedVersion.LATEST,
       true,
@@ -159,9 +173,20 @@ describe('DirectUploadClient', () => {
 
     expect(datasetFiles.totalFilesCount).toBe(0)
 
-    await sut.addUploadedFileToDataset(testDataset2Ids.numericId, multipartFile, actualStorageId)
+    const fileArrayBuffer = await multipartFile.arrayBuffer()
+    const fileBuffer = Buffer.from(fileArrayBuffer)
 
-    datasetFiles = await filesRepository.getDatasetFiles(
+    const uploadedFileDTO = {
+      fileName: multipartFile.name,
+      storageId: actualStorageId,
+      checksumType: checksumAlgorithm,
+      checksumValue: calculateBlobChecksum(fileBuffer),
+      mimeType: multipartFile.type
+    }
+
+    await filesRepositorySut.addUploadedFileToDataset(testDataset2Ids.numericId, uploadedFileDTO)
+
+    datasetFiles = await filesRepositorySut.getDatasetFiles(
       testDataset2Ids.numericId,
       DatasetNotNumberedVersion.LATEST,
       true,
@@ -188,7 +213,7 @@ describe('DirectUploadClient', () => {
     }, 50)
 
     await expect(
-      sut.uploadFile(
+      directUploadSut.uploadFile(
         testDataset2Ids.numericId,
         multipartFile,
         progressMock,
@@ -221,4 +246,10 @@ describe('DirectUploadClient', () => {
         return false
       })
   }
+
+  const calculateBlobChecksum = (blob: Buffer): string => {
+    const hash = crypto.createHash(checksumAlgorithm)
+    hash.update(blob)
+    return hash.digest('hex')
+  }
 })

test/unit/files/AddUploadedFileToDataset.test.ts

@@ -1,35 +1,35 @@
-import { createSinglepartFileBlob } from '../../testHelpers/files/filesHelper'
-import { IDirectUploadClient } from '../../../src/files/domain/clients/IDirectUploadClient'
 import { DirectUploadClientError } from '../../../src/files/domain/clients/DirectUploadClientError'
 import { AddUploadedFileToDataset } from '../../../src/files/domain/useCases/AddUploadedFileToDataset'
+import { IFilesRepository } from '../../../src/files/domain/repositories/IFilesRepository'
 
 describe('execute', () => {
-  let testFile: File
-  const testStorageId = 'test'
-
-  beforeAll(async () => {
-    testFile = await createSinglepartFileBlob()
-  })
+  const testUploadedFileDTO = {
+    fileName: 'testfile',
+    storageId: 'testStorageId',
+    checksumValue: 'testChecksumValue',
+    checksumType: 'md5',
+    mimeType: 'test/type'
+  }
 
   test('should return undefined on client success', async () => {
-    const directUploadClientStub: IDirectUploadClient = {} as IDirectUploadClient
-    directUploadClientStub.addUploadedFileToDataset = jest.fn().mockResolvedValue(undefined)
+    const filesRepositoryStub: IFilesRepository = {} as IFilesRepository
+    filesRepositoryStub.addUploadedFileToDataset = jest.fn().mockResolvedValue(undefined)
 
-    const sut = new AddUploadedFileToDataset(directUploadClientStub)
+    const sut = new AddUploadedFileToDataset(filesRepositoryStub)
 
-    const actual = await sut.execute(1, testFile, testStorageId)
+    const actual = await sut.execute(1, testUploadedFileDTO)
 
     expect(actual).toEqual(undefined)
   })
 
   test('should return error on client error', async () => {
-    const directUploadClientStub: IDirectUploadClient = {} as IDirectUploadClient
-    directUploadClientStub.addUploadedFileToDataset = jest
+    const filesRepositoryStub: IFilesRepository = {} as IFilesRepository
+    filesRepositoryStub.addUploadedFileToDataset = jest
       .fn()
       .mockRejectedValue(new DirectUploadClientError('test', 'test', 'test'))
 
-    const sut = new AddUploadedFileToDataset(directUploadClientStub)
+    const sut = new AddUploadedFileToDataset(filesRepositoryStub)
 
-    await expect(sut.execute(1, testFile, testStorageId)).rejects.toThrow(DirectUploadClientError)
+    await expect(sut.execute(1, testUploadedFileDTO)).rejects.toThrow(DirectUploadClientError)
   })
 })