Usage¶

Running MongoDB¶

To run the system, you need MongoDB available. The easiest way is to use docker:

# get the latest mongodb image
docker pull mongo:latest

# create a directory for persisting the database
mkdir mongodir

# launch a mongodb container called st-mongo
docker run -d --name st-mongo -p 27017:27017 -v $(pwd)/mongodir:/data/db mongo

MongoDB is now running.

Running the frontend¶

The easiest way to run the frontend is by calling the st_frontend script. Use the --help option for usage. This will run Flask, but as clearly stated in their documentation, this should be avoided in a real environment.

To launch the frontend in a production-like fashion, the one way is to use gunicon:

# install gunicorn and its dependencies
pip install  gunicorn

# run the frontend using gunicorn (instead of st_frontend)
# note that the actual port is setup with the gunicorn --bind option
gunicorn --preload --log-level=info --bind=:80 "swisstext.frontend.__main__:init_app()"

# You can also pass parameters to the app through the init_app call. Here is an example:
gunicorn --preload --log-level=info --bind=:80 \
    "swisstext.frontend.__main__:init_app(debug=True, mongo_host='192.131.23.10', mongo_port=27000, db='st1')"

# to use gevent and multiple workers
pip install gevent
gunicorn -k gevent -w 2 'swisstext.frontend.__main__:init_app()'

Running the backend¶

To run the backend, just call the st_scrape and st_search scripts.

If you want to customize how the system works, you can do it via a YAML configuration file. Note that you can use one single file for both scrape and search as options don’t clash with one another and no error is thrown if unknown options are set.