Azure Synapse Analytics Deploy and Execute Spark Job Definition

arazzo: 1.0.1
info:
  title: Azure Synapse Analytics Deploy and Execute Spark Job Definition
  summary: Publish a Spark job definition, execute it, and poll the resulting batch.
  description: >-
    A Spark job definition is a reusable, named Spark application stored in the
    workspace. This workflow publishes the definition, confirms it persisted,
    executes it to launch a Spark batch job, and then polls that batch job
    through the Livy API until it reaches a terminal result. 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: sparkJobDefinitionApi
  url: ../openapi/azure-synapse-analytics-spark-job-definition-openapi.yml
  type: openapi
- name: sparkJobApi
  url: ../openapi/azure-synapse-analytics-spark-job-openapi.yml
  type: openapi
workflows:
- workflowId: deploy-execute-spark-job-definition
  summary: Publish a Spark job definition, run it, and poll the launched batch.
  description: >-
    Creates or updates a Spark job definition, verifies it, executes it to start
    a Spark batch job, then polls the batch by id until the livy result is
    terminal.
  inputs:
    type: object
    required:
    - apiVersion
    - sparkJobDefinitionName
    - sparkJobDefinition
    - sparkPoolName
    properties:
      apiVersion:
        type: string
        description: The Synapse data plane API version (e.g. 2020-12-01).
      sparkJobDefinitionName:
        type: string
        description: The name of the Spark job definition to create or update.
      sparkJobDefinition:
        type: object
        description: The SparkJobDefinitionResource definition to publish.
      sparkPoolName:
        type: string
        description: >-
          The Spark pool name used to poll the launched batch job via the Livy
          API.
  steps:
  - stepId: deployDefinition
    description: >-
      Create or update the Spark job definition in the workspace.
    operationId: SparkJobDefinition_CreateOrUpdateSparkJobDefinition
    parameters:
    - name: api-version
      in: query
      value: $inputs.apiVersion
    - name: sparkJobDefinitionName
      in: path
      value: $inputs.sparkJobDefinitionName
    requestBody:
      contentType: application/json
      payload: $inputs.sparkJobDefinition
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      definitionId: $response.body#/id
  - stepId: confirmDefinition
    description: >-
      Read the Spark job definition back to confirm it was persisted before
      executing it.
    operationId: SparkJobDefinition_GetSparkJobDefinition
    parameters:
    - name: api-version
      in: query
      value: $inputs.apiVersion
    - name: sparkJobDefinitionName
      in: path
      value: $inputs.sparkJobDefinitionName
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      confirmedName: $response.body#/name
  - stepId: executeDefinition
    description: >-
      Execute the Spark job definition, which submits a Spark batch job and
      returns the batch identifier.
    operationId: SparkJobDefinition_ExecuteSparkJobDefinition
    parameters:
    - name: api-version
      in: query
      value: $inputs.apiVersion
    - name: sparkJobDefinitionName
      in: path
      value: $inputs.sparkJobDefinitionName
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      batchId: $response.body#/id
  - stepId: pollBatchJob
    description: >-
      Poll the launched Spark batch job by id. Loop while the livy result is
      still Uncertain, and end once a terminal result is reported.
    operationId: SparkBatch_GetSparkBatchJob
    parameters:
    - name: sparkPoolName
      in: path
      value: $inputs.sparkPoolName
    - name: batchId
      in: path
      value: $steps.executeDefinition.outputs.batchId
    - name: detailed
      in: query
      value: true
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      result: $response.body#/result
    onSuccess:
    - name: stillRunning
      type: goto
      stepId: pollBatchJob
      criteria:
      - context: $response.body
        condition: $.result == "Uncertain"
        type: jsonpath
    - name: finished
      type: end
      criteria:
      - context: $response.body
        condition: $.result != "Uncertain"
        type: jsonpath
  outputs:
    definitionId: $steps.deployDefinition.outputs.definitionId
    batchId: $steps.executeDefinition.outputs.batchId
    result: $steps.pollBatchJob.outputs.result