Dichroview

Dashboard for viewing live data, as it is collected and published to a bluesky-compatible event stream. Includes an API for notifying the viewer when new data stream and data sream events are available.

In this proof-of-concept example, live data is emulated by replaying a data stream from a databroker instance. The data is streamed to a publisher object that can be registered as a callback to a bluesky RunEngine instance. The publisher makes calls to the Dichroview API to notofify the dashboard viewer that data is available for display.

This project brings together a number of notable features:

• In-the-browser visualization of a "live" data stream
  • Uses plotly dash for rapid prototyping and customization
  • Same experience for on-site and remote beamline users
• Web-based API for push-based data stream notifications
• Websockets for subscribing multiple viewer clients
  • Push notifications go to all subscribers
• Data stream publisher that can be registered as a callback to a bluesky RunEngine

The data from this example is a series of XMCD spectra collected at ALS Beamline 4.0.2 (XMCD = x-ray magnetic circular dichroism).

Usage

1. Clone this repository

   git clone https://github.com/als-computing/dichroview.git

2. Create an isolated environment

   conda env create --file environment.yml

3. Copy 'catalogs.yml' to the path where 'intake' looks for catalogs

   cp catalogs.yml ~/opt/miniconda3/envs/dash_fastapi_2/share/intake/

4. Set up the needed ENVironment VARiables

   DATABROKER_MONGO_HOST=<HOST-NAME-OR-IP>:<27017 or PORT>
   DATABROKER_MONGO_USER=<User with read access to Mongo db>
   DATABROKER_MONGO_PASSWORD=<Password for MONGO_USER>
   
   DEBUG=False  # or True for debug mode

5. Start the DichroView server

   uvicorn dichroview:app --port 8003
   
   # or get the same result by running the 'dichroview' script
   python dichroview.py

6. Open a DichroView client in one or more browsers

   URL -- http://127.0.0.1:8003/dash
   

7. Start the data stream

   python newest_run.py
   

8. Watch data being broadcast to each client

   • If data does not display correctly, it might be because events are being streamed too quickly for the dichroview server to keep up. Try reducing the rate of the data stream.
   • In 'newest_run.py': Increase the delay time used by the ApiDelayedCallback object that publishes data to the API.

   # Default delay value is 0.1 sec.
   send_to_api = ApiDelayedCallback(port=8003, delay=0.1)
   
   # Increase the delay value to 0.2 sec.
   send_to_api = ApiDelayedCallback(port=8003, delay=0.2)

Demonstration

This video shows two browser tabs, each running a DichroView client, as data is streamed from a separate process.