arazzo: 1.0.1
info:
  title: Abstract API Screenshot, Scrape, and Image Processing
  summary: Capture a website screenshot, scrape its HTML, then optimize an image asset.
  description: >-
    A site-capture flow that takes a full-page screenshot of a URL, scrapes the
    same URL's rendered HTML, and then runs an image through the Image Processing
    API to produce an optimized, compressed asset. Note an adaptation: the
    Website Screenshot operation returns a binary image rather than a JSON URL,
    so its output cannot be chained downstream; the Image Processing step instead
    operates on a separately supplied image URL. Every step inlines the api_key
    query parameter and request so the flow reads and runs without opening the
    underlying OpenAPI descriptions.
  version: 1.0.0
sourceDescriptions:
- name: websiteScreenshotApi
  url: ../openapi/abstract-api-website-screenshot.yaml
  type: openapi
- name: webScrapingApi
  url: ../openapi/abstract-api-web-scraping.yaml
  type: openapi
- name: imageProcessingApi
  url: ../openapi/abstract-api-image-processing.yaml
  type: openapi
workflows:
- workflowId: screenshot-scrape-and-process
  summary: Screenshot a URL, scrape it, and optimize the resulting screenshot image.
  description: >-
    Captures a full-page screenshot of a URL, scrapes the same page's rendered
    HTML, and feeds the screenshot URL into image processing to produce an
    optimized asset.
  inputs:
    type: object
    required:
    - apiKey
    - url
    - imageUrl
    properties:
      apiKey:
        type: string
        description: Abstract API key valid for Website Screenshot, Web Scraping, and Image Processing.
      url:
        type: string
        description: The page URL to screenshot and scrape.
      imageUrl:
        type: string
        description: An image URL to optimize via the Image Processing API (the screenshot returns binary, so it cannot be chained).
  steps:
  - stepId: captureScreenshot
    description: >-
      Capture a full-page screenshot of the target URL. The response is the
      screenshot image binary, so there are no JSON body outputs to chain.
    operationId: captureWebsiteScreenshot
    parameters:
    - name: api_key
      in: query
      value: $inputs.apiKey
    - name: url
      in: query
      value: $inputs.url
    - name: capture_full_page
      in: query
      value: true
    successCriteria:
    - condition: $statusCode == 200
  - stepId: scrapePage
    description: >-
      Scrape the rendered HTML of the same URL with JavaScript rendering enabled.
    operationId: scrapeWebPage
    parameters:
    - name: api_key
      in: query
      value: $inputs.apiKey
    - name: url
      in: query
      value: $inputs.url
    - name: render_js
      in: query
      value: true
    successCriteria:
    - condition: $statusCode == 200
  - stepId: optimizeImage
    description: >-
      Run the supplied image URL through image processing to produce a lossy,
      compressed WebP asset and report the bytes saved.
    operationId: processImageByURL
    parameters:
    - name: api_key
      in: query
      value: $inputs.apiKey
    - name: url
      in: query
      value: $inputs.imageUrl
    - name: lossy
      in: query
      value: true
    - name: output_format
      in: query
      value: webp
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      processedUrl: $response.body#/url
      savedBytes: $response.body#/saved_bytes
      savedPercent: $response.body#/saved_percent
  outputs:
    processedUrl: $steps.optimizeImage.outputs.processedUrl
    savedBytes: $steps.optimizeImage.outputs.savedBytes