CreateZoom API

Pyramiding

This application is an asynchronous endpoint for creating deep zoom image file formats for images, for use with QUINT and other EBrains software. Main dependencies are asyncio, aiofiles, aiohttp ,fastapi and pyvips. The library libvips is installed in the image.

API Endpoints

• GET /deepzoom/health - Service health status
• POST /deepzoom - Submit image processing task
• GET /deepzoom/status/{task_id} - Check task status

Deployment

to deploy, run build in docker and the image can be deployed anywhere. There are no secrets as the token is handled in memory.

docker build -t appname .

The container defaults to eight concurrent conversion tasks and two libvips worker threads per image to keep peak memory bounded. Override PROCESS_WORKERS or VIPS_CONCURRENCY at deployment time only when the container has enough memory for the additional parallel work.

VIPS_CONCURRENCY sets software worker threads per image pipeline, not physical CPU cores. Eight jobs converting at once with VIPS_CONCURRENCY=2 can use roughly sixteen libvips worker threads, plus application and I/O threads.

Usage

{
  "path": "str",
  "target_path": "str",
  "token": "str"
}

Files paths should be submitted individually, for each file you will be assigned a task_id. You can query your tasks status with the status endpoint.

sequenceDiagram
    participant Client
    participant API
    participant TaskManager
    participant TaskStore
    participant Storage
    participant Bucket

    Client->>API: POST /deepzoom
    API->>TaskManager: Create new task
    TaskManager->>TaskStore: Store task details
    API-->>Client: Return task_id

    Note over TaskManager: Async Processing
    Bucket->>TaskManager: Download image
    TaskManager->>Storage: Create DeepZoomImage
    TaskManager->>Storage: Zip store
    TaskManager->>Bucket: Upload result
    TaskManager->>TaskStore: Update status

    Client->>API: GET /deepzoom/status/{task_id}
    API->>TaskStore: Get task status
    API-->>Client: Return task details

Loading