Custom wikitext parser to produce html, plain text fields and relevant links from wikipedia page source code.
To install the wikitext_asymptote package, simply run
pip install wikitext_asymptoteYou can then use the package as follows:
import wikitext_asymptote as wa
# Raw wikitext goes here
page = """..."""
# Parse it
parsed_page = wa.parse_page(page)
# parsed_page contains fields for html, plain text or links
print(parsed_page)
# parsed_page = {
# 'html': '...',
# 'text': '...',
# 'opening_text': '...',
# 'auxiliary_text': [...],
# 'heading': [...],
# 'links': [...]
# }This package was created to fulfill the precise needs for parsing wikitext in the context of the EPFL Graph project, which are a bit more involved than the defaults from mwparserfromhell.
However, the task of parsing wikitext taking into account all its syntax is gigantic. Already deciding, not even implementing, the parsing of each template in the myriad that's available, with all its variations, is virtually unfeasible.
Hence, the approach we have decided to commit to is that of approximation (hence Asymptote). We try to parse a number of templates, tags and other entities as correctly as possible, in a way that most cases are covered. In addition, we also have defaults, which may or may not be adequate for some cases. This implies that the parsed output is not perfect, and that is by design. From that point on, we work on a case-by-case basis to include parsing for new templates, tags or other entities, should the need arise.
Wikitext asymptote is built on top of the mwparserfromhell package, so we acknowledge and are grateful for the work of their creators.