Linux Commits Bar Chart Race

linux_commits_bar_chart_race.py

"""A script to generate a Bar Chart race graph from the linux git commits
commits.txt was generated via:

git log --format=%ai$%an > commits.txt

Pre-requisite:
==============
conda install -c conda-forge bar_chart_race ffmpeg imagemagick pandas

Instructions to run:
====================
python linux.commits.py
"""
import time
import pandas as pd
import bar_chart_race as bcr


def create_data_for_chart():
    ddf = pd.read_csv("commits.txt", sep="$", encoding="ISO-8859-1", names=['date', 'author'], header=None)
    ddf['count'] = 1
    ddf.date = pd.to_datetime(ddf.date, utc=True)
    ddf_indexed = ddf.set_index('date').sort_index()  #.loc['2005-01-01':'2022-01-01']
    ddf_grouped = ddf_indexed.groupby([pd.Grouper(freq='M'), 'author'])['count'].sum()
    ddf_new = ddf_grouped.unstack("author", fill_value=0)
    ddf_new_cumsum = ddf_new.cumsum(axis=0)
    ddf_new_cumsum.to_csv('linux-commitx.csv')
    return ddf_new_cumsum, ddf_indexed


def period_func(ddf_indexed):
    def _inner(v, _):
        total_commits_so_far = ddf_indexed.loc[:v._name]['count'].sum()
        return {
            'x': .98,
            'y': .2,
            's': f'Total Commits: {total_commits_so_far:,.0f}',
            'ha': 'right', 'size': 11
        }
    return _inner


def create_bar_chart(ddf_new_cumsum, ddf_indexed):
    bcr.bar_chart_race(
        df=ddf_new_cumsum,
        filename="linux.mp4",
        n_bars=25,
        filter_column_colors=True,
        title='Linux Committers Over Time',
        figsize=(8, 6),
        period_fmt='%B %d, %Y',
        period_label={'x': .98, 'y': .3, 'ha': 'right', 'va': 'center'},
        period_summary_func=period_func(ddf_indexed),
        dpi=500,
    )


def main():
    start = time.time()
    print("Preparing data")
    df, ddf_indexed = create_data_for_chart()
    print("Creating Bar chart")
    create_bar_chart(df, ddf_indexed)
    print(f"Time taken: {time.time() - start}")


if __name__ == '__main__':
    main()