PlainTextWikipedia

This revision is from 2024/10/21 07:00. You can Restore it.

Convert Wikipedia database dumps into plain text files (JSON). This can parse literally all of Wikipedia with pretty high fidelity. There's a copy available on Kaggle Datasets.

https://github.com/daveshap/PlainTextWikipedia?tab=readme-ov-file

QUICK START

  • Download and unzip a Wikipedia dump (see Data Sources below) make sure you get a monolithic XML file
  • Open up wiki_to_text.py and edit the filename to point at your XML file. Also update the savedir location
  • Run wiki_to_text.py - it should take about 2.5 days to run, with some variation based on your CPU and storage speed

Data Sources

There are two primary data sources.

Navigate into the latest dump. You're likley looking for the very first file in the download section. They will look something like this:

  • enwiki-20210401-pages-articles-multistream.xml.bz2 18.1 GB
  • simplewiki-20210401-pages-articles-multistream.xml.bz2 203.5 MB

Download and extract these to a storage directory. I usually shorten the folder name and filename.

https://en.wikipedia.org/wiki/Wikipedia:Reusing_Wikipedia_content

Wikipedia is published under Creative Commons Attribution Share-Alike license (CC-BY-SA).

My script is published under the MIT license but this does not confer the same privileges to the material you convert with it.

  

📝 📜 ⏱️ ⬆️