git-word-blame 0.0.7

git-word-blame

Show word-by-word authors of a file

Demo

Executing git word-blame git.c on the git repository we get:

• word-blame-by-commit.html
• word-blame-by-author.html
• authors_stats.tsv
• commit_stats.tsv

Installation

pip install git-word-blame

Usage

git word-blame <path/to/my/file>

It will produce the following files (by default in /tmp/git-word-blame/):

- author_stats.tsv         # top authors by number of characters attributed to them
- commit_stats.tsv          # same for commits
- word-blame-by-commit.html # hover on some text to see which commit created it
- word-blame-by-author.html # same for authors
- text-output # tokens grouped by authors as a grep-friendly output

Authorship algorithms

Two algorithms are available:

• wikiwho (default): coarse but more robust
• mwpersistence: precise but with more false-positives

To change the engine used by git word-blame:

git config --global word-blame.engine mwpersistence

HTML output

You can choose between 3 themes for the HTML output: black-and-white, solarized-dark, solarized-light.

git config --global word-blame.theme solarized-dark

By default a git command is shown when clicking on the text, you can change this behavior and make it go to the diff online (Github/Framagit) directly:

git config --global --bool word-blame.link-to-online-commit true

Limits

git-word-blame doesn't support renames for now and is only tested on files with a linear history.

There's also a default maxmum of 2000 commits processed. This can be removed with this command: git config --global word-blame.limit 0.

WikiWho does not yet, to my knowledge, support attributing white-space, it's attributed to the token coming after it.

See also

• https://github.com/wikiwho/WikiWho/ and https://github.com/wikiwho/WhoColor A word-by-word blame for Wikipedia with a well tested algorithm for prose (default algorithm) The HTML vizualisation of this project is heavily inspired by WhoColor

• https://github.com/mediawiki-utilities/python-mwpersistence/ An alternative authorship algorithm detection made also for Wikipedia

• https://github.com/d33tah/wordblame Export Wikipedia articles to git to perform a word-by-word blame, it does that by putting each word in a separate line for each file in the history

• https://github.com/lucadealfaro/authorship-tracking Another altenative not yet integrated

• https://github.com/git-persistence/git-persistence A character-by-character authorship oriented towards scoring the authors on a whole repository