Technical Metadata API

openapi: 3.0.0
info:
  description: API for the Stanford digital repository technical metadata
  version: 1.0.0
  title: Technical Metadata API
  license:
    name: Apache 2.0
    url: 'http://www.apache.org/licenses/LICENSE-2.0.html'
servers:
  - url: 'https://technical-metadata-{env}.stanford.edu'
    description: Production service
    variables:
      env:
        default: prod
  - url: 'https://technical-metadata-{env}.stanford.edu'
    description: Staging service
    variables:
      env:
        default: stage
tags:
  - name: metadata
    description: Digital Repository Objects
paths:
  /v1/technical-metadata:
    post:
      tags:
        - metadata
      summary: Create the metadata for the object
      description: ''
      operationId: 'technical_metadata#create'
      responses:
        '200':
          description: OK
      requestBody:
        content:
          application/json:
            schema:
              type: object
              required:
                - druid
                - files
                - basepath
              properties:
                druid:
                    $ref: '#/components/schemas/Druid'
                files:
                  type: array
                  items:
                    $ref: '#/components/schemas/FileInfo'
                basepath:
                  type: string
                  description: Path to be removed from filepath to get filename
                force:
                  type: boolean
                  description: Extract the technical metadata even if a record already exists with that checksum
                lane-id:
                  type: string
                  enum:
                    - 'high'
                    - 'default'
                    - 'low'
                  default: 'default'
  /v1/technical-metadata/druid/{druid}:
    get:
      tags:
        - metadata
      summary: Get the metadata for the files associated with a druid
      description: ''
      operationId: technical_metadata#show_by_druid
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                type: array
                items:
                  $ref: '#/components/schemas/DroFile'
        '404':
          description: Druid not found
      parameters:
        - name: druid
          in: path
          description: druid for which to return the technical metadata
          required: true
          schema:
            $ref: '#/components/schemas/Druid'
  /v1/technical-metadata/audit/{druid}:
    post:
      tags:
        - metadata
      summary: Diff the technical metadata DB entries for a druid against the expected file/hash list provided to the endpoint
      description: ''
      operationId: technical_metadata#audit_by_druid
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/FileAuditResult'
        '404':
          description: Druid not found
        '401':
          description: Unauthorized
      parameters:
        - name: druid
          in: path
          description: druid for which to audit the technical metadata
          required: true
          schema:
            $ref: '#/components/schemas/Druid'
      requestBody:
        content:
          application/json:
            schema:
              type: object
              required:
                - expected_files
              properties:
                expected_files:
                  type: array
                  items:
                    $ref: '#/components/schemas/FileAuditInfo'
components:
  schemas:
    AudioMetadata:
      description: Audio-specific technical metadata
      type: object
      properties:
        format_profile:
          type: string
        codec_id:
          type: string
        channels:
          type: string
        sampling_rate:
          type: integer
        bit_depth:
          type: integer
        stream_size:
          type: integer
        mean_volume:
          type: number
          format: float
        max_volume:
          type: number
          format: float
    AvMetadata:
      description: Audio/visual-specific technical metadata
      type: object
      properties:
        video_count:
          type: integer
        audio_count:
          type: integer
        other_count:
          type: integer
        file_extension:
          type: string
        format:
          type: string
        format_profile:
          type: string
        codec_id:
          type: string
        duration:
          type: number
          format: float
        frame_rate:
          type: number
          format: float
        encoded_date:
          type: string
    DroFile:
      description: Technical metadata for a single file
      type: object
      properties:
        druid:
          $ref: '#/components/schemas/Druid'
        filename:
          type: string
        filetype:
          description: Pronom id
          type: string
        mimetype:
          type: string
        bytes:
          type: integer
        file_modification:
          type: string
          format: date-time
        image_metadata:
          $ref: '#/components/schemas/ImageMetadata'
        pdf_metdata:
          $ref: '#/components/schemas/PdfMetadata'
        av_metadata:
          $ref: '#/components/schemas/AvMetadata'
        dro_file_parts:
          type: array
          items:
            $ref: '#/components/schemas/DroFilePart'
      required:
        - druid
        - filename
    DroFilePart:
      description: Technical metadata for part of a file, e.g., a track
      type: object
      properties:
        part_type:
          type: string
          enum:
            - 'audio'
            - 'video'
            - 'text'
            - 'other'
        part_id:
          type: string
        order:
          type: integer
        format:
          type: string
        audio_metadata:
          $ref: '#/components/schemas/AudioMetadata'
        video_metadata:
          $ref: '#/components/schemas/VideoMetadata'
        other_metadata:
          $ref: '#/components/schemas/OtherAvMetadata'
      required:
        - part_type
    Druid:
      description: Digital Repository Unique Identifier (ID)
      type: string
      pattern: '^druid:[b-df-hjkmnp-tv-z]{2}[0-9]{3}[b-df-hjkmnp-tv-z]{2}[0-9]{4}$'
      example: 'druid:bc123df4567'
    FileURI:
      description: File URI for the absolute path of the file
      type: string
      pattern: '^file:\/\/\/([^\\/]+\/)*[^\\/]+$'
      example: 'file:///foo/bar/baz/quix.jp2'
    FileName:
      description: filename for a file in SDR (from Cocina metadata)
      type: string
      pattern: '^([^\\/]+\/)*[^\\/]+$'
      example: 'foo/bar/baz/quix.jp2'
    FileInfo:
      description: Information about a file.
      type: object
      properties:
        uri:
          $ref: '#/components/schemas/FileURI'
        md5:
          type: string
      required:
        - uri
        - md5
    FileAuditInfo:
      description: Information about a file in SDR for which we expect technical metadata to exist.
      type: object
      properties:
        filename:
          $ref: '#/components/schemas/FileName'
        md5:
          type: string
      required:
        - filename
        - md5
    FileAuditResult:
      description: Information about detected differences between expected technical metadata entries and actual technical metadata entries
      type: object
      properties:
        missing_filenames:
          type: array
          items:
            $ref:
              '#/components/schemas/FileName'
        unexpected_filenames:
          type: array
          items:
            $ref:
              '#/components/schemas/FileName'
        mismatched_checksum_file_infos:
          type: array
          items:
            $ref:
              '#/components/schemas/FileAuditInfo'
    ImageMetadata:
      description: Image-specific technical metadata
      type: object
      properties:
        # Some image formats, e.g., SVG return floats for height and width, hence type: number
        height:
          type: number
        width:
          type: number
    OtherAvMetadata:
      description: Other technical metadata for a file part
      type: object
      properties:
        other_type:
          type: string
        type:
          type: string
    PdfMetadata:
      description: PDF-specific technical metadata
      type: object
      properties:
        pdf_version:
          type: string
        pages:
          type: integer
        page_size:
          type: string
        tagged:
          type: boolean
        encrypted:
          type: boolean
        javascript:
          type: boolean
        form:
          type: boolean
        text:
          type: boolean
        creator:
          type: string
        producer:
          type: string
    VideoMetadata:
      description: Video-specific technical metadata
      type: object
      properties:
        format_profile:
          type: string
        codec_id:
          type: string
        height:
          type: integer
        width:
          type: integer
        display_aspect_ratio:
          type: number
          format: float
        pixel_aspect_ratio:
          type: number
          format: float
        frame_rate:
          type: number
          format: float
        color_space:
          type: string
        chroma_subsampling:
          type: string
        bit_depth:
          type: integer
        language:
          type: string
        stream_size:
          type: integer
        standard:
          type: string