arazzo: 1.0.1
info:
  title: Sensible Extract From URL And Poll
  summary: Kick off an asynchronous extraction from a document URL with a chosen config, poll until it completes, then read the parsed results.
  description: >-
    The canonical asynchronous single-document extraction pattern for Sensible.
    The workflow submits a publicly accessible (or presigned) document URL to
    the extract-from-url endpoint for a chosen document type and config,
    receives an extraction id, polls the Retrieve extraction by ID endpoint
    until Sensible reports a COMPLETE status, and then surfaces the
    parsed_document and coverage. Every step spells out its request inline,
    including the Bearer authorization, so the flow can be read and executed
    without opening the underlying OpenAPI description.
  version: 1.0.0
sourceDescriptions:
- name: extractionsApi
  url: ../openapi/sensible-extractions-api-openapi.yml
  type: openapi
workflows:
- workflowId: extract-from-url-and-poll
  summary: Asynchronously extract data from a document at a URL with a specified config and poll for the completed result.
  description: >-
    Submits a document URL for extraction under the supplied document type and
    config, then polls the extraction id until the status is COMPLETE and
    returns the extracted fields.
  inputs:
    type: object
    required:
    - apiKey
    - documentType
    - configName
    - documentUrl
    properties:
      apiKey:
        type: string
        description: Sensible API key used as the Bearer token.
      documentType:
        type: string
        description: The document type to extract from (e.g. senseml_basics).
      configName:
        type: string
        description: The config to use for extraction.
      documentUrl:
        type: string
        description: A publicly accessible or presigned URL returning the document bytes.
      documentName:
        type: string
        description: Optional filename echoed back in the extraction response.
  steps:
  - stepId: submitExtraction
    description: >-
      Submit the document URL for asynchronous extraction under the chosen
      document type and config, and capture the returned extraction id.
    operationId: provide-a-download-url-with-config
    parameters:
    - name: Authorization
      in: header
      value: "Bearer $inputs.apiKey"
    - name: document_type
      in: path
      value: $inputs.documentType
    - name: config_name
      in: path
      value: $inputs.configName
    requestBody:
      contentType: application/json
      payload:
        document_url: $inputs.documentUrl
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      extractionId: $response.body#/id
      initialStatus: $response.body#/status
  - stepId: pollStatus
    description: >-
      Poll the extraction by id until Sensible reports the COMPLETE status,
      retrying while the extraction is still WAITING or PROCESSING.
    operationId: retrieving-results
    parameters:
    - name: Authorization
      in: header
      value: "Bearer $inputs.apiKey"
    - name: id
      in: path
      value: $steps.submitExtraction.outputs.extractionId
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      status: $response.body#/status
      parsedDocument: $response.body#/parsed_document
      coverage: $response.body#/coverage
      validationSummary: $response.body#/validation_summary
    onSuccess:
    - name: extractionComplete
      type: end
      criteria:
      - context: $response.body
        condition: $.status == "COMPLETE"
        type: jsonpath
    - name: keepPolling
      type: goto
      stepId: pollStatus
      criteria:
      - context: $response.body
        condition: $.status == "WAITING" || $.status == "PROCESSING"
        type: jsonpath
  outputs:
    extractionId: $steps.submitExtraction.outputs.extractionId
    status: $steps.pollStatus.outputs.status
    parsedDocument: $steps.pollStatus.outputs.parsedDocument
    coverage: $steps.pollStatus.outputs.coverage