Nanonets Async URL Predict Poll and Approve

arazzo: 1.0.1
info:
  title: Nanonets Async URL Predict Poll and Approve
  summary: Async-predict a document by URL, poll until ready, then approve the file.
  description: >-
    An end-to-end URL-driven extraction flow. The workflow submits a publicly
    accessible document URL to a Nanonets OCR model in async mode, polls the
    inference request until the prediction is available, and then marks the
    finished file as approved. Every step spells out its request inline so the
    flow can be read and executed without opening the underlying OpenAPI
    description.
  version: 1.0.0
sourceDescriptions:
- name: ocrApi
  url: ../openapi/nanonets-ocr-api-openapi.yml
  type: openapi
- name: fileManagementApi
  url: ../openapi/nanonets-file-management-api-openapi.yml
  type: openapi
workflows:
- workflowId: async-url-predict-poll-approve
  summary: Async predict by URL, poll for completion, and approve the resulting file.
  description: >-
    Submits a document URL for async OCR, polls the inference request until a
    result is present, and approves the file.
  inputs:
    type: object
    required:
    - authorization
    - modelId
    - urls
    properties:
      authorization:
        type: string
        description: HTTP Basic credential header value (Basic <base64 of apiKey:>).
      modelId:
        type: string
        description: Unique identifier for the Nanonets OCR model.
      urls:
        type: array
        description: Publicly accessible URL(s) of the document(s) to extract.
        items:
          type: string
      requestMetadata:
        type: string
        description: Free-form identifier echoed back in the prediction response.
  steps:
  - stepId: submitAsyncUrl
    description: Submit the document URL(s) to the model in async mode.
    operationId: ocrModelLabelUrlsAsyncByModelIdPost
    parameters:
    - name: Authorization
      in: header
      value: $inputs.authorization
    - name: model_id
      in: path
      value: $inputs.modelId
    requestBody:
      contentType: application/x-www-form-urlencoded
      payload:
        urls: $inputs.urls
        async: true
        request_metadata: $inputs.requestMetadata
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      requestFileId: $response.body#/result/0/request_file_id
  - stepId: pollPrediction
    description: >-
      Poll the inference request for the file. Loop back while no page has been
      moderated or unmoderated yet, otherwise continue to approval.
    operationId: ocrModelGetPredictionFileByFileId
    parameters:
    - name: Authorization
      in: header
      value: $inputs.authorization
    - name: model_id
      in: path
      value: $inputs.modelId
    - name: request_file_id
      in: path
      value: $steps.submitAsyncUrl.outputs.requestFileId
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      moderatedCount: $response.body#/moderated_images_count
      unmoderatedCount: $response.body#/unmoderated_images_count
    onSuccess:
    - name: stillPending
      type: goto
      stepId: pollPrediction
      criteria:
      - context: $response.body
        condition: $.moderated_images_count == 0 && $.unmoderated_images_count == 0
        type: jsonpath
    - name: ready
      type: goto
      stepId: approveFile
  - stepId: approveFile
    description: Approve the finished file once the prediction is available.
    operationId: verifyFile
    parameters:
    - name: Authorization
      in: header
      value: $inputs.authorization
    - name: model_id
      in: path
      value: $inputs.modelId
    - name: request_file_id
      in: path
      value: $steps.submitAsyncUrl.outputs.requestFileId
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      approveResponse: $response.body
  outputs:
    requestFileId: $steps.submitAsyncUrl.outputs.requestFileId
    approveResponse: $steps.approveFile.outputs.approveResponse