Apache Nutch Admin API

openapi: 3.2.0
info:
  title: Apache Nutch REST Admin API
  description: REST API for managing Apache Nutch crawl jobs, configurations, seed lists, database queries, and data readers.
  version: 1.0.0
  license:
    name: Apache 2.0
    identifier: Apache-2.0
  contact:
    name: Apache Nutch
    url: https://nutch.apache.org
servers:
- url: '{protocol}://localhost:{port}'
  description: Nutch REST server
  variables:
    protocol:
      default: http
      enum:
      - http
      - https
      description: The protocol used to access the Nutch server.
    port:
      default: '8081'
      description: The port the Nutch server listens on. Configurable via the --port command-line argument.
security:
- basicAuth: []
tags:
- name: Admin
  description: Server administration operations
paths:
  /admin/:
    get:
      tags:
      - Admin
      summary: Apache Nutch Get Server Status
      description: Returns the current status of the Nutch server including start date, known configurations, all jobs, and currently running jobs.
      operationId: getServerStatus
      responses:
        '200':
          description: Server status retrieved successfully.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/NutchServerInfo'
        '401':
          $ref: '#/components/responses/Unauthorized'
        '500':
          $ref: '#/components/responses/InternalServerError'
      x-microcks-operation:
        delay: 0
        dispatcher: FALLBACK
  /admin/stop:
    get:
      tags:
      - Admin
      summary: Apache Nutch Stop the Nutch Server
      description: Initiates a graceful shutdown of the Nutch server. If jobs are still running and force is not set, the server will refuse to stop.
      operationId: stopServer
      parameters:
      - name: force
        in: query
        required: false
        description: If true, kills any running jobs before stopping the server.
        schema:
          type: boolean
          default: false
      responses:
        '200':
          description: Shutdown status message.
          content:
            application/json:
              schema:
                type: string
              examples:
                stopping:
                  value: Stopping in server on port 8081
                busy:
                  value: Jobs still running -- Cannot stop server now
        '401':
          $ref: '#/components/responses/Unauthorized'
        '500':
          $ref: '#/components/responses/InternalServerError'
      x-microcks-operation:
        delay: 0
        dispatcher: FALLBACK
components:
  responses:
    Unauthorized:
      description: Unauthorized. Basic authentication credentials are missing or invalid.
      content:
        application/json:
          schema:
            type: object
            properties:
              message:
                type: string
            example:
              message: Authentication required.
    InternalServerError:
      description: An unexpected server error occurred.
      content:
        text/plain:
          schema:
            type: string
          example: Internal server error.
  schemas:
    State:
      type: string
      description: The current state of a job.
      enum:
      - IDLE
      - RUNNING
      - FINISHED
      - FAILED
      - KILLED
      - STOPPING
      - KILLING
      - ANY
    JobInfo:
      type: object
      description: Information about a crawl job.
      required:
      - type
      - state
      properties:
        id:
          type: string
          description: The unique job identifier.
        type:
          $ref: '#/components/schemas/JobType'
        confId:
          type: string
          description: The configuration ID used for this job.
        args:
          type: object
          additionalProperties: true
          description: Arguments passed to the job.
        result:
          type: object
          additionalProperties: true
          description: Result data returned after job completion.
        state:
          $ref: '#/components/schemas/State'
        msg:
          type: string
          description: A human-readable status or error message.
        crawlId:
          type: string
          description: The crawl identifier associated with this job.
    JobType:
      type: string
      description: The type of Nutch crawl job.
      enum:
      - INJECT
      - GENERATE
      - FETCH
      - PARSE
      - UPDATEDB
      - INDEX
      - READDB
      - CLASS
      - INVERTLINKS
      - DEDUP
    NutchServerInfo:
      type: object
      description: Status information about the running Nutch server.
      required:
      - configuration
      - jobs
      - runningJobs
      properties:
        startDate:
          type: string
          format: date-time
          description: The date and time the server was started.
        configuration:
          type: array
          items:
            type: string
          uniqueItems: true
          description: Set of known configuration IDs.
        jobs:
          type: array
          items:
            $ref: '#/components/schemas/JobInfo'
          description: All jobs (any state).
        runningJobs:
          type: array
          items:
            $ref: '#/components/schemas/JobInfo'
          description: Currently running jobs.
  securitySchemes:
    basicAuth:
      type: http
      scheme: basic
      description: HTTP Basic Authentication.