Simple geospatial API for U.S. Census and other data sources, in response to the Census Area API hack request.
This application can be run in two distinct ways. When the environment variable GEO_DATASOURCE=http://forever.codeforamerica.org/Census-API/ is present, the API will serve nationwide U.S. Census geographies from the state to block level. See this in action at census.codeforamerica.org and use it freely.
Otherwise, it will use a local OGR-compatible datasource such as a shapefile or GeoJSON to serve any data and make it available for download. A sample copy with Bay Area Census data can be found at census-api-bay-area.herokuapp.com.
This is a Flask-based Python application which requires compiled geospatial libraries Shapely and GDAL to run. Directions here assume that you have an OGR-compatible datasource you’d like to use.
- Download and unpack sample Bay Area data.
- Ensure that datasource.shp and other files are located in the same directory as app.py.
- Call
python app.pyfor a test server.
To run a more robust installation using the Python WSGI HTTP server Gunicorn, prepare local data as in steps 1 & 2 above, then call:
gunicorn app:app
Compiled geospatial libraries for Heroku are available via the open source GIS Heroku buildpack. Create a new Heroku app with this buildpack:
heroku create --buildpack https://github.com/codeforamerica/heroku-buildpack-pygeo
There are two possible ways to run the API on Heroku:
-
Fork this repository, download and commit your own data as datasource.shp, and push the combined application + data repository to Heroku.
-
Use the
ZIPPED_DATA_URLsupport in heroku-buildpack-pygeo to configure a remote zip file such as bay-area-data.zip (URL linked above), making sure to install the Heroku plugin user-env-compile. Data will be automatically retrieved and expanded to datasource.shp at compile time.
Written by Michal Migurski with
Andy Hull, (c) 2013 Code for America.
See LICENSE for license information.