A spatial API for your GitHub-hosted GeoJSON.
GitHub now supports rendering GeoJSON files. This makes it easy to collaborate on geo data. GitSpatial provides a spatial API for GeoJSON hosted at GitHub.
- Visit http://gitspatial.herokuapp.com
- Authorize the site to access your GitHub repos
- Sync the repos that have GeoJSON in them
- Sync individual GeoJSON files within these repos
- Utilize the API to query your features
Find all parks within a map view:
http://gitspatial.herokuapp.com/api/v1/JasonSanford/mecklenburg-gis-opendata/parks?bbox=-81.025,35.023,-80.685,35.487
Find all parks within 3000 meters of a point:
http://gitspatial.herokuapp.com/api/v1/JasonSanford/mecklenburg-gis-opendata/parks?lat=35.255&lon=-80.855&distance=3000
GitSpatial is a Django app that uses PostGIS as a backend to GeoDjango. So, you might run into trouble getting dependencies installed.
- GEOS
- GDAL
- Proj
Create a virtual environment. Python 2.7 is ideal as this is what Heroku uses. I drop the virtual envrionment within the root of the project. The venv directory is in .gitignore.
virtualenv venv
Source to the virtual environment.
source venv/bin/activate
Install dependencies defined in requirements.txt.
pip install -r requirements.txt
To keep secret keys, passwords and such out of version control, we store them in environment variables. Below are the variables required to run.
AWS_ACCESS_KEY_ID=my-aws-key
AWS_SECRET_ACCESS_KEY=super-secret-aws-secret-access-key
DATABASE_URL=postgis://user:password@host:port/db_name
GITHUB_APP_ID=not-a-secret
GITHUB_API_SECRET=super-secret
LD_LIBRARY_PATH=/app/.heroku/vendor/lib:vendor/geos/geos/lib:vendor/proj/proj/lib:vendor/gdal/gdal/lib
LIBRARY_PATH=/app/.heroku/vendor/lib:vendor/geos/geos/lib:vendor/proj/proj/lib:vendor/gdal/gdal/lib
Assuming you have a database already created and credentials are in the DATABASE_URL envrionment variable, we just need to syncdb to get the necessary tables created. Do not create a superuser if prompted.
python manage.py syncdb
The django web server (gunicorn) and the celery process are defined in Procfile and Procfile-dev and point to production and development settings respectively. Run with Foreman. Heroku will look for Procfile, while we'll use Procfile-dev locally.
foreman start -f Procfile-dev
The application is currently hosted at heroku. Deployments are handled by git push'ing to a heroku endpoint. The current git remote is [email protected]:gitspatial.git
Add this remote to enable pushing.
git remote add heroku [email protected]:gitspatial.git
Then, deploying the latest code is just git push heroku master.
A custom buildpack is used to include the GEOS, GDAL and Proj binaries necessary for Shapely and GeoDjango. To force deployment to use a custom buildpack, add the following heroku config.
heroku config:set BUILDPACK_URL=git://github.com/JasonSanford/heroku-buildpack-python-geos.git
Additionally, the LIBRARY_PATH and LD_LIBRARY_PATH enverionment variables must be updated so that Shapely can locate the necessary binaries.
heroku config:set LIBRARY_PATH=/app/.heroku/vendor/lib:vendor/geos/geos/lib:vendor/proj/proj/lib:vendor/gdal/gdal/lib
heroku config:set LD_LIBRARY_PATH=/app/.heroku/vendor/lib:vendor/geos/geos/lib:vendor/proj/proj/lib:vendor/gdal/gdal/lib
Static file deployment is handled by the collectstatic method when pushing to Heroku. We're using a combination of django-store and boto to automatically collect/push static files to Amazon S3 during deployment.
All static files should be placed in the /static directory if they need to be deployed.
Note! - With the custom buildpack, collectstatic doesn't seem to be running automatically. To run manually:
heroku run python manage.py collectstatic
Use one command to test all apps
python manage.py test gitspatial api user geojson