Scrape Swiss referendum vote results from opendata.swiss

This is an unofficial scraper developed to power the website Predikon.ch, which provides real-time prediction of referendum results.

Clone the repo and install the library:

git clone [email protected]:victorkristof/swissvotes.git
cd swissvotes
pip install -r requirements.txt
pip install .

The library provides a client to opendata.swiss's real-time referendum vote results:

from swissvotes import Client

client = Client()

The client enables you to query the API. Usually, you would start by fetching the vote URLs as a dictionary {date: url}

urls = client.available_votes()
url = urls['2020-02-09']
# url --> https://www.bfs.admin.ch/bfsstatic/dam/assets/11708082/master

You can retrieve the votes' metadata for a given URL by calling:

metadata = client.get_vote_metadata(url)
# metadata --> list of dict

Each dictionary in metadata contains details about the votes (identified uniquely by the ogd_id key).

Finally, you can scrape the results at a given URL by doing:

results = client.get_results(url, level='municipality')
# results --> dict of list

The keys of results are the vote OGD ids, and the values are a list of results. Each result is a dictionary with data such as the number of "yes", the number of voters, the name of the municipality, and a timestamp. The level argument of client.get_results(...) enables you to choose the level at which you would like to scrape the results. It accepts the following keys:

• canton: results in the 26 cantons (i.e., the Swiss states)
• district: results in the Swiss districts ("bezirke")
• zhdistrict: results in the counting districts ("zaehlkreise") of Zurich and Winterthur
• municipality: results in the ~2200 municipalities
• municipality+zhdistrict: results in the municipalities and the counting districts The last key is useful to have a finer level of granularity of the results in Zurich and Winterthur, two the largest municipalities in Switzerland (the results in the counting districts are released sequentially as soon as they are available, whereas the aggregate results in the whole municipality is released much later).

The two methods client.get_vote_metadata(...) and client.get_results(...) optionally takes an argument file='path/to/data.json', which enables you to save the raw JSON data to the given path.

You can then process the raw JSON by using the Parser class:

import json
from swissvotes import Parser

with open('path/to/data.json') as f:
    data = json.load(f)

parser = Parser()
results = parser.parse_results(data, level='municipality')