WPyDumps

WPyDumps is a Python module to work with dumps of Wikipedia.

It allows one to parse and extract relevant information from dump files without un-compressing them on-disk.

It works with (at least) these dumps:

• pages-meta-history….xml-….7z (“All pages with complete edit history”)
• pages-meta-current.xml.bz2

Install

pip install wpydumps

Usage

The parser uses SAX to read the files as a stream. It takes a reader or a filename and a page callback function. It parses the file and call that function with each page.

Pages are represented as wpydumps.model.Page objects. They include the pages’ details as well as their revisions (wpydumps.model.Revision). Each revision holds a reference to its contributor (wpydumps.model.Contributor) .

import wpydumps as p


def simple_page_callback(page):
    print(page.title)


# parse from a local archive
p.parse_pages_from_archive_filename("myfile.7z", simple_page_callback)

# parse from an uncompressed file
with open("myfile") as f:
    p.parse_pages_from_reader(f, simple_page_callback)

Revisions always have a text_length and diff_length int attributes. You may drop the text content by passing keep_revisions_text=False to the parser.

Examples

from wpydumps import parse_pages_from_archive_filename


def page_callback(page):
    pass  # do something with the page


# use the appropriate filename
parse_pages_from_archive_filename("frwiki-20190901-pages-meta-history1.xml-p3p1630.7z", page_callback)

Print all pages and their number of revisions

def page_callback(page):
    print(page.title, len(page.revisions))

Print all pages and their number of contributors

def page_callback(page):
    contributors = set()
    for rev in page.revisions:
        contributors.add(rev.contributor.username or rev.contributor.ip)

    print("%s: %d contributors" % (page.title, len(contributors)))