Merge pull request #205 from IQSS/feat/203-extra-param-html-to-markdown

Add extra parameter to transform or not html to markdown

docs/useCases.md

@@ -287,6 +287,8 @@ The optional `datasetVersionId` parameter can correspond to a numeric version id
 
 There is an optional third parameter called `includeDeaccessioned`, which indicates whether to consider deaccessioned versions or not in the dataset search. If not set, the default value is `false`.
 
+There is an optional fourth parameter called `keepRawFields`, which indicates whether or not to keep the metadata fields as they are and avoid the transformation to Markdown. The default value is `false`.
+
 #### Get Dataset By Private URL Token
 
 Returns a [Dataset](../src/datasets/domain/models/Dataset.ts) instance, given an associated Private URL Token.
@@ -307,6 +309,8 @@ getPrivateUrlDataset.execute(token).then((dataset: Dataset) => {
 
 _See [use case](../src/datasets/domain/useCases/GetPrivateUrlDataset.ts)_ definition.
 
+There is an optional second parameter called `keepRawFields`, which indicates whether or not to keep the metadata fields as they are and avoid the transformation to Markdown. The default value is `false`.
+
 #### Get Dataset Citation Text
 
 Returns the Dataset citation text.

src/datasets/domain/repositories/IDatasetsRepository.ts

@@ -10,15 +10,16 @@ export interface IDatasetsRepository {
   getDataset(
     datasetId: number | string,
     datasetVersionId: string,
-    includeDeaccessioned: boolean
+    includeDeaccessioned: boolean,
+    keepRawFields: boolean
   ): Promise<Dataset>
   getDatasetLocks(datasetId: number | string): Promise<DatasetLock[]>
   getDatasetCitation(
     datasetId: number,
     datasetVersionId: string,
     includeDeaccessioned: boolean
   ): Promise<string>
-  getPrivateUrlDataset(token: string): Promise<Dataset>
+  getPrivateUrlDataset(token: string, keepRawFields: boolean): Promise<Dataset>
   getAllDatasetPreviews(
     limit?: number,
     offset?: number,

src/datasets/domain/useCases/GetDataset.ts

@@ -16,17 +16,20 @@ export class GetDataset implements UseCase<Dataset> {
    * @param {number | string} [datasetId] - The dataset identifier, which can be a string (for persistent identifiers), or a number (for numeric identifiers).
    * @param {string | DatasetNotNumberedVersion} [datasetVersionId=DatasetNotNumberedVersion.LATEST] - The dataset version identifier, which can be a version-specific numeric string (for example, 1.0) or a DatasetNotNumberedVersion enum value. If this parameter is not set, the default value is: DatasetNotNumberedVersion.LATEST
    * @param {boolean} [includeDeaccessioned=false] - Indicates whether to consider deaccessioned versions in the dataset search or not. The default value is false
+   * @param {boolean} [keepRawFields=false] - Indicates whether or not the use case should keep the metadata fields as they are and avoid the transformation to markdown. The default value is false.
    * @returns {Promise<Dataset>}
    */
   async execute(
     datasetId: number | string,
     datasetVersionId: string | DatasetNotNumberedVersion = DatasetNotNumberedVersion.LATEST,
-    includeDeaccessioned = false
+    includeDeaccessioned = false,
+    keepRawFields = false
   ): Promise<Dataset> {
     return await this.datasetsRepository.getDataset(
       datasetId,
       datasetVersionId,
-      includeDeaccessioned
+      includeDeaccessioned,
+      keepRawFields
     )
   }
 }

src/datasets/domain/useCases/GetPrivateUrlDataset.ts

@@ -13,9 +13,10 @@ export class GetPrivateUrlDataset implements UseCase<Dataset> {
    * Returns a Dataset instance, given an associated Private URL Token.
    *
    * @param {string} [token] - A Private URL token.
+   * @param {boolean} [keepRawFields=false] - Indicates whether or not the use case should keep the metadata fields as they are and avoid the transformation to markdown. The default value is false.
    * @returns {Promise<Dataset>}
    */
-  async execute(token: string): Promise<Dataset> {
-    return await this.datasetsRepository.getPrivateUrlDataset(token)
+  async execute(token: string, keepRawFields = false): Promise<Dataset> {
+    return await this.datasetsRepository.getPrivateUrlDataset(token, keepRawFields)
   }
 }

src/datasets/infra/repositories/DatasetsRepository.ts

@@ -33,15 +33,15 @@ export class DatasetsRepository extends ApiRepository implements IDatasetsReposi
       })
   }
 
-  public async getPrivateUrlDataset(token: string): Promise<Dataset> {
+  public async getPrivateUrlDataset(token: string, keepRawFields: boolean): Promise<Dataset> {
     return this.doGet(
       this.buildApiEndpoint(this.datasetsResourceName, `privateUrlDatasetVersion/${token}`),
       false,
       {
         returnOwners: true
       }
     )
-      .then((response) => transformVersionResponseToDataset(response))
+      .then((response) => transformVersionResponseToDataset(response, keepRawFields))
       .catch((error) => {
         throw error
       })
@@ -50,7 +50,8 @@ export class DatasetsRepository extends ApiRepository implements IDatasetsReposi
   public async getDataset(
     datasetId: number | string,
     datasetVersionId: string,
-    includeDeaccessioned: boolean
+    includeDeaccessioned: boolean,
+    keepRawFields: boolean
   ): Promise<Dataset> {
     return this.doGet(
       this.buildApiEndpoint(this.datasetsResourceName, `versions/${datasetVersionId}`, datasetId),
@@ -61,7 +62,7 @@ export class DatasetsRepository extends ApiRepository implements IDatasetsReposi
         returnOwners: true
       }
     )
-      .then((response) => transformVersionResponseToDataset(response))
+      .then((response) => transformVersionResponseToDataset(response, keepRawFields))
       .catch((error) => {
         throw error
       })

src/datasets/infra/repositories/transformers/datasetTransformers.ts

@@ -209,12 +209,18 @@ export const transformMetadataChildFieldValueToRequestPayload = (
   return metadataChildFieldRequestPayload
 }
 
-export const transformVersionResponseToDataset = (response: AxiosResponse): Dataset => {
+export const transformVersionResponseToDataset = (
+  response: AxiosResponse,
+  keepRawFields: boolean
+): Dataset => {
   const versionPayload = response.data.data
-  return transformVersionPayloadToDataset(versionPayload)
+  return transformVersionPayloadToDataset(versionPayload, keepRawFields)
 }
 
-export const transformVersionPayloadToDataset = (versionPayload: DatasetPayload): Dataset => {
+export const transformVersionPayloadToDataset = (
+  versionPayload: DatasetPayload,
+  keepRawFields: boolean
+): Dataset => {
   const datasetModel: Dataset = {
     id: versionPayload.datasetId,
     versionId: versionPayload.id,
@@ -227,7 +233,10 @@ export const transformVersionPayloadToDataset = (versionPayload: DatasetPayload)
       lastUpdateTime: new Date(versionPayload.lastUpdateTime),
       releaseTime: new Date(versionPayload.releaseTime)
     },
-    metadataBlocks: transformPayloadToDatasetMetadataBlocks(versionPayload.metadataBlocks),
+    metadataBlocks: transformPayloadToDatasetMetadataBlocks(
+      versionPayload.metadataBlocks,
+      keepRawFields
+    ),
     ...(versionPayload.isPartOf && {
       isPartOf: transformPayloadToOwnerNode(versionPayload.isPartOf)
     })
@@ -260,25 +269,28 @@ const transformPayloadToDatasetLicense = (licensePayload: LicensePayload): Datas
 }
 
 const transformPayloadToDatasetMetadataBlocks = (
-  metadataBlocksPayload: MetadataBlocksPayload
+  metadataBlocksPayload: MetadataBlocksPayload,
+  keepRawFields: boolean
 ): DatasetMetadataBlocks => {
   return Object.keys(metadataBlocksPayload).map((metadataBlockKey) => {
     const metadataBlock = metadataBlocksPayload[metadataBlockKey]
     return {
       name: metadataBlock.name,
-      fields: transformPayloadToDatasetMetadataFields(metadataBlock.fields)
+      fields: transformPayloadToDatasetMetadataFields(metadataBlock.fields, keepRawFields)
     }
   }) as DatasetMetadataBlocks
 }
 
 const transformPayloadToDatasetMetadataFields = (
-  metadataFieldsPayload: MetadataFieldPayload[]
+  metadataFieldsPayload: MetadataFieldPayload[],
+  keepRawFields: boolean
 ): DatasetMetadataFields => {
   return metadataFieldsPayload.reduce(
     (datasetMetadataFieldsMap: DatasetMetadataFields, field: MetadataFieldPayload) => {
       datasetMetadataFieldsMap[field.typeName] = transformPayloadToDatasetMetadataFieldValue(
         field.value,
-        field.typeClass
+        field.typeClass,
+        keepRawFields
       )
       return datasetMetadataFieldsMap
     },
@@ -288,7 +300,8 @@ const transformPayloadToDatasetMetadataFields = (
 
 const transformPayloadToDatasetMetadataFieldValue = (
   metadataFieldValuePayload: MetadataFieldValuePayload,
-  typeClass: string
+  typeClass: string,
+  keepRawFields: boolean
 ): DatasetMetadataFieldValue => {
   function isArrayOfSubfieldValue(
     array: (string | MetadataSubfieldValuePayload)[]
@@ -301,29 +314,37 @@ const transformPayloadToDatasetMetadataFieldValue = (
   }
 
   if (typeof metadataFieldValuePayload === 'string') {
+    if (keepRawFields) {
+      return metadataFieldValuePayload
+    }
     return transformHtmlToMarkdown(metadataFieldValuePayload)
   } else if (Array.isArray(metadataFieldValuePayload)) {
     if (isArrayOfSubfieldValue(metadataFieldValuePayload)) {
       return metadataFieldValuePayload.map((metadataSubfieldValuePayload) =>
-        transformPayloadToDatasetMetadataSubfieldValue(metadataSubfieldValuePayload)
+        transformPayloadToDatasetMetadataSubfieldValue(metadataSubfieldValuePayload, keepRawFields)
       )
     } else {
+      if (keepRawFields) {
+        return metadataFieldValuePayload
+      }
       return metadataFieldValuePayload.map(transformHtmlToMarkdown)
     }
   } else {
     return transformPayloadToDatasetMetadataSubfieldValue(
-      metadataFieldValuePayload as MetadataSubfieldValuePayload
+      metadataFieldValuePayload as MetadataSubfieldValuePayload,
+      keepRawFields
     )
   }
 }
 
 const transformPayloadToDatasetMetadataSubfieldValue = (
-  metadataSubfieldValuePayload: MetadataSubfieldValuePayload
+  metadataSubfieldValuePayload: MetadataSubfieldValuePayload,
+  keepRawFields: boolean
 ): DatasetMetadataSubField => {
   const result: DatasetMetadataSubField = {}
   Object.keys(metadataSubfieldValuePayload).forEach((key) => {
     const subFieldValue = metadataSubfieldValuePayload[key].value
-    result[key] = transformHtmlToMarkdown(subFieldValue)
+    result[key] = keepRawFields ? subFieldValue : transformHtmlToMarkdown(subFieldValue)
   })
   return result
 }

src/files/infra/repositories/transformers/fileTransformers.ts

@@ -27,7 +27,7 @@ export const transformFileResponseToFile = (
   if (returnDatasetVersion) {
     return [
       transformFilePayloadToFile(filePayload),
-      transformVersionPayloadToDataset(filePayload.datasetVersion)
+      transformVersionPayloadToDataset(filePayload.datasetVersion, false)
     ]
   }
   return transformFilePayloadToFile(filePayload)

test/functional/datasets/GetDataset.test.ts

@@ -0,0 +1,109 @@
+import { ApiConfig, createDataset, getDataset, ReadError } from '../../../src'
+import { DataverseApiAuthMechanism } from '../../../src/core/infra/repositories/ApiConfig'
+import { DatasetDescription } from '../../../src/datasets/domain/models/Dataset'
+import { deleteUnpublishedDatasetViaApi } from '../../testHelpers/datasets/datasetHelper'
+import { TestConstants } from '../../testHelpers/TestConstants'
+
+const testNewDataset = {
+  metadataBlockValues: [
+    {
+      name: 'citation',
+      fields: {
+        title: 'Dataset created using the createDataset use case',
+        author: [
+          {
+            authorName: 'Admin, Dataverse',
+            authorAffiliation: 'Dataverse.org'
+          },
+          {
+            authorName: 'Owner, Dataverse',
+            authorAffiliation: 'Dataversedemo.org'
+          }
+        ],
+        datasetContact: [
+          {
+            datasetContactEmail: '[email protected]',
+            datasetContactName: 'Finch, Fiona'
+          }
+        ],
+        dsDescription: [
+          {
+            dsDescriptionValue: 'Hello <b>world</b>'
+          }
+        ],
+        subject: ['Medicine, Health and Life Sciences']
+      }
+    }
+  ]
+}
+
+describe('execute', () => {
+  beforeEach(async () => {
+    ApiConfig.init(
+      TestConstants.TEST_API_URL,
+      DataverseApiAuthMechanism.API_KEY,
+      process.env.TEST_API_KEY
+    )
+  })
+
+  test('should successfully get a dataset when a valid id is sent', async () => {
+    const createdDatasetIdentifiers = await createDataset.execute(testNewDataset)
+
+    const dataset = await getDataset.execute(createdDatasetIdentifiers.numericId)
+    expect(dataset).not.toBeNull()
+    expect(dataset.id).toBe(createdDatasetIdentifiers.numericId)
+
+    await deleteUnpublishedDatasetViaApi(createdDatasetIdentifiers.numericId)
+  })
+
+  test('should successfully get a dataset when a valid persistent id is sent', async () => {
+    const createdDatasetIdentifiers = await createDataset.execute(testNewDataset)
+
+    const dataset = await getDataset.execute(createdDatasetIdentifiers.persistentId)
+    expect(dataset).not.toBeNull()
+    expect(dataset.id).toBe(createdDatasetIdentifiers.numericId)
+
+    await deleteUnpublishedDatasetViaApi(createdDatasetIdentifiers.numericId)
+  })
+
+  test('should throw an error when an invalid id is sent', async () => {
+    const nonExistentTestDatasetId = 'non-existent-dataset'
+    const expectedError = new ReadError(`[400] Bad dataset ID number: ${nonExistentTestDatasetId}.`)
+
+    await expect(getDataset.execute(nonExistentTestDatasetId)).rejects.toThrow(expectedError)
+  })
+
+  test('should return metadata fields in markdown format when keepRawFields is false', async () => {
+    const createdDatasetIdentifiers = await createDataset.execute(testNewDataset)
+
+    const dataset = await getDataset.execute(
+      createdDatasetIdentifiers.numericId,
+      undefined,
+      false,
+      false
+    )
+
+    expect(
+      (dataset.metadataBlocks[0].fields.dsDescription[0] as DatasetDescription).dsDescriptionValue
+    ).toBe('Hello **world**')
+
+    await deleteUnpublishedDatasetViaApi(createdDatasetIdentifiers.numericId)
+  })
+
+  test('should not return metadata fields in markdown format when keepRawFields is true', async () => {
+    const createdDatasetIdentifiers = await createDataset.execute(testNewDataset)
+
+    const dataset = await getDataset.execute(
+      createdDatasetIdentifiers.numericId,
+      undefined,
+      false,
+      true
+    )
+
+    expect(
+      (dataset.metadataBlocks[0].fields.dsDescription[0] as DatasetDescription).dsDescriptionValue
+    ).toBe('Hello <b>world</b>')
+
+    await deleteUnpublishedDatasetViaApi(createdDatasetIdentifiers.numericId)
+  })
+})