data.world files API

openapi: 3.1.0
info:
  contact:
    name: Contact Us
    url: https://data.world/company/contact-us
  termsOfService: https://data.world/terms-policies
  title: data.world Public catalog relationships files API
  version: '0'
  description: Manage relationships between catalog resources
servers:
- url: /v0
security:
- bearerAuth: []
tags:
- name: files
paths:
  /datasets/{owner}/{id}/files:
    delete:
      description: 'Delete one or more files from a dataset.


        **Batching**

        Note that the `name` parameter can be include multiple times in the query string, once for each file that is to be deleted together in a single request.'
      operationId: deleteFilesAndSyncSources
      parameters:
      - in: path
        name: owner
        required: true
        schema:
          type: string
          minLength: 1
      - in: path
        name: id
        required: true
        schema:
          type: string
          minLength: 1
      - in: query
        name: name
        required: true
        schema:
          type: array
          items:
            type: string
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/SuccessMessage'
          description: default response
      summary: Delete files
      tags:
      - files
    post:
      description: 'Add files from URLs to a dataset. This method allows files published on the web to be added to a

        data.world dataset via their URL. This method can also be used to retrieve data via web APIs, with

        advanced options for http method, request payload and authentication.


        The source URL will be stored so you can easily update your file anytime it changes via the *fetch latest*

        link on the [data.world](https://data.world/) dataset page or by triggering the `GET:/sync` endpoint.'
      operationId: addFilesBySource
      parameters:
      - in: path
        name: owner
        required: true
        schema:
          type: string
          minLength: 1
      - in: path
        name: id
        required: true
        schema:
          type: string
          minLength: 1
      requestBody:
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/FileBatchUpdateRequest'
        required: true
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/SuccessMessage'
          description: default response
      summary: Add files from URLs
      tags:
      - files
  /datasets/{owner}/{id}/files/{file}:
    delete:
      description: Delete a single file from a dataset.
      operationId: deleteFileAndSyncSource
      parameters:
      - in: path
        name: owner
        required: true
        schema:
          type: string
          minLength: 1
      - in: path
        name: id
        required: true
        schema:
          type: string
          minLength: 1
      - description: Filename
        in: path
        name: file
        required: true
        schema:
          type: string
          maxLength: 128
          minLength: 1
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/SuccessMessage'
          description: default response
      summary: Delete a file
      tags:
      - files
  /datasets/{owner}/{id}/files/{file}/metadata:
    get:
      description: Get file description and labels
      operationId: getFileMetadata
      parameters:
      - in: path
        name: owner
        required: true
        schema:
          type: string
          minLength: 1
      - in: path
        name: id
        required: true
        schema:
          type: string
          minLength: 1
      - in: path
        name: file
        required: true
        schema:
          type: string
          minLength: 1
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/FileMetadataResponse'
          description: default response
      summary: Get file description and labels
      tags:
      - files
    patch:
      description: 'Update file description and labels. Only elements included in the request will be updated. All omitted

        elements will remain untouched. Requesting change in a list replaces the entire list. Requesting an element

        to be null deletes all values on the element.'
      operationId: patchFileMetadata
      parameters:
      - in: path
        name: owner
        required: true
        schema:
          type: string
          minLength: 1
      - in: path
        name: id
        required: true
        schema:
          type: string
          minLength: 1
      - in: path
        name: file
        required: true
        schema:
          type: string
          minLength: 1
      requestBody:
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/FileMetadataUpdateRequest'
        required: true
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/SuccessMessage'
          description: default response
      summary: Update file description and labels
      tags:
      - files
    put:
      description: Replace file description and labels
      operationId: putFileMetadata
      parameters:
      - in: path
        name: owner
        required: true
        schema:
          type: string
          minLength: 1
      - in: path
        name: id
        required: true
        schema:
          type: string
          minLength: 1
      - in: path
        name: file
        required: true
        schema:
          type: string
          minLength: 1
      requestBody:
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/FileMetadataUpdateRequest'
        required: true
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/SuccessMessage'
          description: default response
      summary: Replace file description and labels
      tags:
      - files
  /datasets/{owner}/{id}/sync:
    get:
      description: Same as `POST:/dataset/{owner}/{id}/sync`. The id parameter can either be a dataset id or project id.
      operationId: syncViaGet
      parameters:
      - in: path
        name: owner
        required: true
        schema:
          type: string
          minLength: 1
      - in: path
        name: id
        required: true
        schema:
          type: string
          minLength: 1
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/SuccessMessage'
          description: default response
      summary: Sync files (via GET)
      tags:
      - files
    post:
      description: 'Sync files within a dataset or project. This method will process the latest data available for files added

        from URLs or via streams. The id parameter can either be a dataset id or project id.'
      operationId: sync
      parameters:
      - in: path
        name: owner
        required: true
        schema:
          type: string
          minLength: 1
      - in: path
        name: id
        required: true
        schema:
          type: string
          minLength: 1
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/SuccessMessage'
          description: default response
      summary: Sync files
      tags:
      - files
  /file_download/{owner}/{id}/{file}:
    get:
      description: 'Download a file within the dataset as originally uploaded. Prefer `POST:/sql` or `POST:/sparql` for

        retrieving clean and structured data.'
      operationId: downloadFile
      parameters:
      - in: path
        name: owner
        required: true
        schema:
          type: string
      - in: path
        name: id
        required: true
        schema:
          type: string
      - in: path
        name: file
        required: true
        schema:
          type: string
      responses:
        '200':
          content:
            '*/*': {}
          description: default response
      summary: Download a file
      tags:
      - files
  /uploads/{owner}/{id}/files:
    post:
      description: "\"Upload multiple files at once to a dataset.\n\nThis endpoint expects requests of type `multipart/form-data` and you can include one or more parts named\n`file`, each containing a different file to be uploaded.\n\nFor example, assuming that you want to upload two local files named `file1.csv` and `file2.csv` to a\nhypothetical dataset `https://data.world/awesome-user/awesome-dataset`, this is what the cURL command\nwould look like.\n\n```bash\ncurl \\\n  -H \"Authorization: Bearer <YOUR_API_TOKEN>\" \\\n  -F \"file=@file1.csv\" \\\n  -F \"file=@file2.csv\" \\\n  https://api.data.world/v0/uploads/awesome-user/awesome-dataset/files\n```\n\nSwagger clients may limit this method of upload to one file at a time. Other HTTP clients capable of\nmaking multipart/form-data requests can be used to upload multiple files in a single request.\"\n"
      operationId: uploadFiles
      parameters:
      - in: path
        name: owner
        required: true
        schema:
          type: string
      - in: path
        name: id
        required: true
        schema:
          type: string
      - in: query
        name: expandArchives
        schema:
          type: boolean
          default: false
      requestBody:
        content:
          multipart/form-data:
            encoding:
              file:
                style: form
            schema:
              type: object
              properties:
                file:
                  type: string
                  format: binary
                  description: Multipart-encoded file contents
              required:
              - file
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/SuccessMessage'
          description: default response
      summary: Upload files
      tags:
      - files
  /uploads/{owner}/{id}/files/{file}:
    put:
      description: "Upload one file at a time to a dataset.\n\nThis endpoint expects requests of type `application/octet-stream`.\n\nFor example, assuming that you want to upload a local file named `file1.csv` to a hypothetical dataset\n`https://data.world/awesome-user/awesome-dataset` and choose its name on data.world to be\n`better-name.csv`, this is what the cURL command would look like.\n\n```bash\ncurl \\\n  -H \"Authorization: Bearer <YOUR_API_TOKEN>\" \\\n  -X PUT -H \"Content-Type: application/octet-stream\" \\\n  --data-binary @file1.csv \\\n  https://api.data.world/v0/uploads/awesome-user/awesome-dataset/files/better-name.csv\n```\n\nThis method of upload is typically not supported by Swagger clients. Other HTTP clients can be used to\nsupply the contents of the file directly in the body of the request.\n"
      operationId: uploadFile
      parameters:
      - in: path
        name: owner
        required: true
        schema:
          type: string
      - in: path
        name: id
        required: true
        schema:
          type: string
      - description: Filename
        in: path
        name: file
        required: true
        schema:
          type: string
          maxLength: 128
          minLength: 1
          pattern: ^[^/]+$
      - in: query
        name: expandArchive
        schema:
          type: boolean
          default: false
      requestBody:
        content:
          '*/*':
            schema:
              type: string
              format: binary
          application/octet-stream:
            schema:
              type: string
              format: binary
        description: file data
        required: true
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/SuccessMessage'
          description: default response
      summary: Upload a file
      tags:
      - files
components:
  schemas:
    QueryExecutionDto:
      type: object
      properties:
        query:
          type: string
          maxLength: 10000
          minLength: 1
      required:
      - query
    FileMetadataUpdateRequest:
      type: object
      properties:
        description:
          type: string
        labels:
          type: array
          items:
            type: string
          uniqueItems: true
    DatabaseMetadataSpecDto:
      type: object
      properties:
        tableSpecs:
          type: array
          items:
            $ref: '#/components/schemas/TableMetadataSpecDto'
          maxItems: 100
          minItems: 0
    TableMetadataSpecDto:
      type: object
      properties:
        database:
          type: string
        schema:
          type: string
        tablePrefixes:
          type: array
          items:
            type: string
        tableTypes:
          type: array
          items:
            type: string
            enum:
            - TABLE
            - EXTERNAL_TABLE
            - VIEW
        tables:
          type: array
          items:
            type: string
    SingleTableMetadataSpecDto:
      type: object
      properties:
        database:
          type: string
        schema:
          type: string
        table:
          type: string
        tableType:
          type: string
          enum:
          - EXTRACT
          - VIRTUAL
    WebCredentials:
      type: object
      properties:
        password:
          type: string
          maxLength: 2048
          minLength: 0
        user:
          type: string
          maxLength: 2048
          minLength: 0
      required:
      - user
    FileCreateOrUpdateRequest:
      type: object
      properties:
        description:
          type: string
          maxLength: 120
          minLength: 1
        labels:
          type: array
          items:
            type: string
        mediaType:
          type: string
        mediaTypeLocked:
          type: boolean
        name:
          type: string
          maxLength: 128
          minLength: 1
          pattern: ^[^/]+$
        source:
          $ref: '#/components/schemas/FileSourceCreateOrUpdateRequest'
      required:
      - name
    ViewRequestDto:
      type: object
      properties:
        baseAgentId:
          type: string
        baseDatasetId:
          type: string
        queryId:
          type: string
          format: uuid
        queryName:
          type: string
        queryText:
          type: string
          maxLength: 10000
          minLength: 1
        queryType:
          type: string
          enum:
          - SPARQL
          - SQL
        runAsAgentId:
          type: string
        targetAgentId:
          type: string
        targetDatasetId:
          type: string
      required:
      - baseAgentId
      - baseDatasetId
      - queryName
      - queryType
      - runAsAgentId
      - targetAgentId
      - targetDatasetId
    FileSourceCreateOrUpdateRequest:
      type: object
      properties:
        authorization:
          $ref: '#/components/schemas/WebAuthorization'
        credentials:
          $ref: '#/components/schemas/WebCredentials'
        dataTables:
          type: object
          additionalProperties:
            $ref: '#/components/schemas/QueryExecutionDto'
        databaseMetadataSpec:
          $ref: '#/components/schemas/DatabaseMetadataSpecDto'
        databaseSource:
          $ref: '#/components/schemas/DatabaseSourceReference'
        dwccSpec:
          $ref: '#/components/schemas/DwccSpecDto'
        expandArchive:
          type: boolean
        method:
          type: string
          enum:
          - GET
          - POST
        oauthToken:
          $ref: '#/components/schemas/OauthTokenReference'
        requestEntity:
          type: string
          maxLength: 10000
          minLength: 0
        requestHeaders:
          type: object
          additionalProperties:
            type: string
        tableSpec:
          $ref: '#/components/schemas/SingleTableMetadataSpecDto'
        url:
          type: string
          format: uri
        useDwCredentials:
          type: boolean
        viewRequest:
          $ref: '#/components/schemas/ViewRequestDto'
    OauthTokenReference:
      type: object
      properties:
        id:
          type: string
          format: uuid
        owner:
          type: string
        site:
          type: string
      required:
      - id
      - owner
      - site
    DwccSpecDto:
      type: object
      properties:
        databaseId:
          type: string
        metadataToExclude:
          type: array
          items:
            type: string
        sourceDatabase:
          type: string
        sourceSchemas:
          type: array
          items:
            type: string
        targetCatalog:
          type: string
    SuccessMessage:
      type: object
      properties:
        message:
          type: string
          maxLength: 256
          minLength: 0
    DatabaseSourceReference:
      type: object
      properties:
        id:
          type: string
          format: uuid
        owner:
          type: string
      required:
      - id
      - owner
    WebAuthorization:
      type: object
      properties:
        credentials:
          type: string
          maxLength: 2048
          minLength: 1
        type:
          type: string
          maxLength: 50
          minLength: 0
      required:
      - type
    FileMetadataResponse:
      type: object
      properties:
        created:
          type: string
          format: date-time
        createdBy:
          type: string
        description:
          type: string
          maxLength: 120
          minLength: 1
        labels:
          type: array
          items:
            type: string
        name:
          type: string
          maxLength: 128
          minLength: 1
          pattern: ^[^/]+$
        sizeInBytes:
          type: integer
          format: int64
        updated:
          type: string
          format: date-time
        updatedBy:
          type: string
      required:
      - created
      - name
      - updated
    FileBatchUpdateRequest:
      type: object
      properties:
        files:
          type: array
          items:
            $ref: '#/components/schemas/FileCreateOrUpdateRequest'
  securitySchemes:
    bearerAuth:
      bearerFormat: JWT
      scheme: bearer
      type: http