commonhealth-data-map

A CommonHealth data map.

Setup

Run ./scripts/setup and then ./scripts/server.

Processing data

There are two steps to processing data for the dashboard.

• Publishing processed data: Generates processed data that is committed to the repository.
• Publishing visualization data: Generates data for visualization, including vector tiles. This data is not committed to the repository.

Setup

All commands below should be run in the data-processing directory.

Build the container

data-processing> ./scripts/build

Run Jupyter Notebooks

To run jupyter notebooks, use

data-processing> ./scripts/notebook

and browse to the URL output last in the console. Notebooks are stored in notebooks. Output data should be placed in data/output folder if it's not to be published to the site (otherwise use data/published). External data that is committed to the repository should go in data. You can use data/cache as a download cache. data/output and data/cache are ignored folders.

Publishing processed data

This process takes any source data pulled in via specific notebooks in data-processing/notebooks and generates data that is committed to the repository in data-processing/data/published. Published data that needs to be refreshed continually will do so through a GitHub action.

Running papermill will also produce notebooks that have their output cells showing the results of the notebook run; these are placed in the notebook-logs folder for inspection of the latest data processing runs and committed to the repository.

Running locally

In the data-processing directory, use:

data-processing> scripts/update-data

to run the containerized workflow for updating the data. This utilizes the same process as the GitHub action, the difference being the filesystem mounted into the containers.

Publishing visualization data

This process will gather files from data-processing/data/published and place it into the site's build folder for visualization through the front-end. This step includes visualization-specific data processing steps, such as building vector tiles. Data generated through this process will not be committed to the repository, but will be built by Netlify on deploy.

Running locally

Use scripts/build-website to generate the visualization data locally. Note! All changes need to be committed for this command to work.

You can then copy build/data into public/data for usage with the dev server. Also if you make changes to GeoJSON that is directly used in the website, you can copy the data from data-processing/data/published into public/data.

The package.json contains a mechanism that will proxy requests to the staging server, so if you do not generate your own visualization data locally the requests will pull from staging data. Note that this will cause missing data files to be read as HTML responses, so you may see JSON parse errors if the data file you are trying to read in doesn't actually exist yet.

Adding new data for visualization.

If you create data for publishing to the site, create it in the data/published directory and modify the cibuild script to include moving it to the correct location or performing any secondary processing like vector tile generation.

Automated Data Update

The automated data update runs through a GitHub Actions workflow, located in .github/workflows/dataproc.yml. This runs the custom Docker action located at .github/actions/update-data, which runs the appropriate notebooks through papermill in the the data processing container in the a similar manner as described above. The workflow then checks to see if there are data updates, and if so, will push up changes, make a PR, and merge changes in if Netlify passes its build.

The below flowchart shows the broad-strokes logic of the workflow:

To test changes to the workflow, you can push to the feature/automatic-data-update branch to kick off GitHub actions. Otherwise this workflow will run against the develop branch on an hourly schedule.

The GitHub actions rely on the following GitHub secrets:

• GOOGLEAPI_REFRESH_TOKEN: The Google API refresh token to use for accessing google drive
• GOOGLEAPI_CLIENT_ID: The Google API client ID to use for accessing google drive
• GOOGLEAPI_CLIENT_SECRET: The Google API client secret to use for accessing google drive

Scripts

Name	Description
build-website	Build the application and visualization data locally.
cibuild	Build application for staging or a release.
server	Start application.
setup	Attempts to setup the project's development environment.
test	Run linters and tests.
update	Install project runtime dependencies.