Line by line display of author responsible for contribution on MediaWiki page

Question

Is there a way to show "who contributed to this line" of each line of a page of Wikipedia or of a MediaWiki site?

It would show per page and would be similar to the subversion blame tool.

Answer 1

I often need something like this as well, but it looks like there's no ready solution.

What I've done is written a script that helps me grab revisions using the MediaWiki API and import them to the Bazaar version control system. bzr qblame article.wiki, then gives a nice view of who changed what. The script is not really ready for release, but you can find it below or on Pastebin. The script adds to a mercurial repo, which can then be converted to Bazaar.

# I hereby place this script into the Public Domain!
import os, sys
import time

import mwclient

import mercurial.ui
from mercurial import localrepo
from mercurial import commands

article = 'Love'
#start_time = None
start_time = '2011-01-01T00:00:00Z'

# set up mercurial repo
ui = mercurial.ui.ui()
repo_dir = article
repo = localrepo.localrepository(ui, path=repo_dir, create = not os.path.isdir(repo_dir))
#if not os.path.isdir(article):
#   os.mkdir(article)
#os.chdir(article)
print "rep in", repo.root
content_path = os.path.join(repo.root, article + '.wiki')

site = mwclient.Site('en.wikipedia.org')
page = site.Pages[article]

for rev in page.revisions(start=start_time, limit=50,dir='newer', prop='ids|timestamp|flags|comment|user|content'):
    content = rev['*']
    timestamp = time.asctime(rev['timestamp'])
    comment = rev['comment'].encode('utf8')
    if len(comment) == 0: comment = "blank"
    print "writing revision from", timestamp

    f = open(content_path, 'wb')
    f.write(content.encode('utf8'))
    f.close()

    commands.addremove(ui, repo)
    commands.commit(ui, repo, message=comment, user=rev['user'].encode('utf8'), date=timestamp)

Answer 2

http://search.cpan.org/~daxim/Mediawiki-Blame-0.0.3/lib/Mediawiki/Blame.pm appears to support exactly this. Unfortunately, it's just a perl module, not a user-friendly command, so it's going to be a bit harder to use than one might want …

Answer 3

Inspired by previous answers, I've adapted the previous pastebin Python script to use git instead of mercurial as backend

My project is hosted on GitLab

Answer 4

Another not very user friendly solution would be the mw-to-git extension for git, which would presumably allow you to run git blame on a page.

Answer 5

You might prefer to convert the history of the page to git, and then use your favourite IDE / command lines tools to search history

#!/usr/bin/env python3
import os
import subprocess
import sys

import mwclient
from datetime import datetime
from time import mktime

print("getting page...")
site = mwclient.Site(("https", "en.wikipedia.org"))
name = sys.argv[1]
page = site.pages[name]


def conv_time(t):
    return datetime.fromtimestamp(mktime(t))


os.mkdir(name)
os.chdir(name)


subprocess.check_output(["git", "init"])
print("extracting revisions (may take a really long time, depending on the page)...")
for i, revision in enumerate(
    page.revisions(prop="ids|timestamp|flags|comment|user|content", dir="newer")
):
    if "*" in revision:
        timestamp = conv_time(revision["timestamp"])
        print(timestamp)
        with open("page", "w") as stream:
            stream.write(revision["*"])
        subprocess.check_output(["git", "add", "page"])
        subprocess.check_output(
            [
                "git",
                "commit",
                "--allow-empty",
                f"--date={timestamp.isoformat()}",
                "-m",
                f"{revision['user']} {revision['revid']}",
            ]
        )

Save this script in mwdump on your PATH then run mwdump page - this will create a directory called page containing the history of the page