OpenMensa Parsers

OpenMensa is a free canteen database. It depends on external parsers that provide the meal information for the specific canteens.

This repository contains a large collection of parsers for different canteens all over Germany - mostly university canteens provided by student unions.

Before you continue you may want to read the feed documentation describing the exchange format between parsers and OpenMensa.

Contribute

Corrections Welcome: As I do not use most of the parsers myself it is likely that I miss some parsing issues. Feel free to report an issue or even better provide a pull request.

Hosting Provided: Your canteen is missing? You could write a parser for your canteen (in Python)? But you do not know where to host the parser? I can host your parser at omfeeds.devtation.de. Please provide a PR with the new parser.

Overall Structure

The parsers itself are independent. But there is a small framework handling common tasks like URL parsing and output generation.

Each provider has it's own Python module. A provider represents a collection of canteens which are organisationally dependent and therefore can be parsed by the same process. The module itself has to implement a parse_url(canteenidentifier, today=False) method. This method is called to parse and return the feed for a specific canteen. What the canteenidentifier is exactly is up to the provider - mostly they are URL parts or URL suffixes.

The config.py contains a list of all known providers and it's canteens (plus the canteenidentifier that is passed to the parse_url method). The structure is hopefully self explaining. If not, please open an issue.

Common implementation details of providers

At the moment all providers are using PyOpenMensa (documentation, repo) to generate the XML feed and for some help for the parsing itself.

As many meal information are only available online as HTML, Beautiful Soup 4 is used as a robust but easy to use HTML parser.

Get started

1. Clone the source code

    git clone --recurse-submodules git://github.com/mswart/openmensa-parsers
   

2. Install the dependencies:

   • Python 3
   • Beautiful Soup 4 - needed for most parsers/providers.
   • python-lxml Some parsers use the lxml backend of Beautiful Soup, so you might need the Python lxml module/extension.

3. Try some parsers

    python3 parse.py magdeburg ovgu-unten full.xml
   

   general:

    python3 parse.py <provider name> <canteen name> <feed name>.xml
   

   Almost all parsers implement a feed called full including all available menu information. Most parsers implement also a today feed returning primarily the menu for today.

Running the full server

You can either use travis or use this similiar approach:

Prequesites: Apache, Unix (Ubuntu for the following lines)

sudo apt-get install libapache2-mod-wsgi-py3 python3-pip
sudo pip3 install bs4 lxml

Create new file wsgi.conf for apache2.

sudo touch /etc/apache2/conf-available/wsgi.conf

Insert those lines (depending on your path)

WSGIPythonPath /var/www/html/openmensa-parsers/
WSGICallableObject handler
WSGIScriptAlias /get /var/www/html/openmensa-parsers/wsgihandler.py

Restart Apache2

sudo systemctl restart apache2

Check if it works (should return xml links)

wget -O - 127.0.0.1/dresden/index.json

Tips for adding a new provider

1. Search where the meal information are accessible online. JSON or CSV downloads are mostly the best, HTML sites are also possible, but PDF is tricky.
2. Maybe take some look on other parsers how they solve the problem / which libraries they use.
3. Create the new provider
4. Register your provider with its canteens in config.py
5. Submit a PR
6. Wait until the PR is reviewed and merged
7. Register the new canteens on openmensa with the feed from http://omfeeds.devtation.de/<provider identifier>/<canteen identifier>.xml and (optional) today feed from http://omfeeds.devtation.de/<provider identifier>/<canteen identifier>/today.xml

Further questions

• Read the OpenMensa Documentation
• Ask at the OpenMensa IRC channel freenode#openmensa
• The add howto for developers of new parser issue may be helpful.
• Open an issue