# BioSim CZ extraction provenance
#
# Current extraction-run metadata stored on an Experiment. The prototype keeps
# only the latest extraction attempt for a draft Experiment; rerunning extraction
# replaces this object together with automatically extracted metadata.

ExtractionProvenance:
  type: object
  properties:
    engine_type:
      type: keyword
      required: true
      label:
        en: Engine Type
      help:
        en: Simulation engine family handled by the extraction workflow (e.g., gromacs).

    extractor_name:
      type: keyword
      required: true
      label:
        en: Extractor Name
      help:
        en: Name of the engine-specific extraction tool (e.g., GROMACS MetaDump).

    extractor_version:
      type: keyword
      label:
        en: Extractor Version
      help:
        en: Version of the engine-specific extraction tool.

    adaptor_name:
      type: keyword
      required: true
      label:
        en: Adaptor Name
      help:
        en: Name of the adaptor that mapped engine-native metadata into the BioSim CZ schema.

    adaptor_version:
      type: keyword
      label:
        en: Adaptor Version
      help:
        en: Version or release label of the adaptor or mapping rules.

    extracted_at:
      type: keyword
      required: true
      label:
        en: Extracted At
      help:
        en: Timestamp of the extraction run, stored as an ISO 8601 date-time string.

    status:
      type: keyword
      required: true
      enum:
        - success
        - warning
        - failed
      label:
        en: Extraction Status
      help:
        en: Overall result of the latest extraction and adaptation attempt.

    source_files:
      type: array
      label:
        en: Source Files
      help:
        en: Files selected as inputs for the extraction run.
      items:
        type: object
        properties:
          file_key:
            type: keyword
            required: true
            label:
              en: File Key
            help:
              en: Filename or repository key of the selected source file.
          semantic_type:
            type: vocabulary
            vocabulary-type: file_semantic_type
            required: true
            label:
              en: Semantic Type
            help:
              en: Semantic type assigned to the source file.
          extraction_role:
            type: keyword
            required: true
            enum:
              - primary
              - optional
            label:
              en: Extraction Role
            help:
              en: Role of the file in the extraction run.

    messages:
      type: array
      label:
        en: Extraction Messages
      help:
        en: File-level or stage-level messages from the latest extraction and adaptation attempt.
      items:
        type: object
        properties:
          stage:
            type: keyword
            required: true
            enum:
              - file_profile
              - extractor
              - adaptor
            label:
              en: Stage
            help:
              en: Processing stage that produced the message.
          severity:
            type: keyword
            required: true
            enum:
              - info
              - warning
              - error
            label:
              en: Severity
            help:
              en: Message severity.
          code:
            type: keyword
            required: true
            label:
              en: Message Code
            help:
              en: Stable code identifying the message type.
          message:
            type: fulltext
            required: true
            label:
              en: Message
            help:
              en: Human-readable extraction or adaptation message.
          file_key:
            type: keyword
            label:
              en: File Key
            help:
              en: Source file related to the message, when the message concerns one selected file.