librsvg oxidation chart

svg_burndown.py

import json
import os
import re
import pandas
import matplotlib.pyplot as plt
import dateutil

from subprocess import run, PIPE

repo = "/home/jess/repo/librsvg"
tmp_csv = "/tmp/burndown.csv"

rx1 = re.compile("(\d+) insertions")
rx2 = re.compile("(\d+) deletions")


def inspect(info):
    proc = run(["git", "show", "-s", "--format=%ci", "HEAD"], stdout=PIPE)
    date = dateutil.parser.parse(proc.stdout)
    date = date.replace(tzinfo=None)

    proc = run(["git", "diff", "HEAD^", "HEAD", "--shortstat"], stdout=PIPE)

    m1 = rx1.search(proc.stdout.decode("utf8"))
    if m1:
        addd = int(m1.groups(1)[0])
    else:
        addd = 0

    m2 = rx2.search(proc.stdout.decode("utf8"))
    if m2:
        deld = int(m2.groups(1)[0])
    else:
        deld = 0

    proc = run(["tokei", "-o", "json", "-t", "Rust,C,C Header"], stdout=PIPE)
    out = json.loads(proc.stdout)
    del out["C"]["stats"]
    del out["CHeader"]["stats"]

    try:
        del out["Rust"]["stats"]
    except KeyError:
        # add the first non-rusty stats then reraise to catch in parent
        info.append((date, out["C"]["lines"], 0, out["CHeader"]["lines"], addd, deld))
        raise

    info.append(
        (
            date,
            out["C"]["lines"],
            out["Rust"]["lines"],
            out["CHeader"]["lines"],
            addd,
            deld,
        )
    )


def postprocess(df):
    df.sort_index(inplace=True)
    df.to_csv(tmp_csv)
    ax = df[["C", "Rust", "CHeader"]].plot(title="librsvg burndown", drawstyle="steps")
    ax.set_ylabel("LOC")

    # df["Commit"] = 1
    # df2 = df[["Added", "Deleted", "Commit"]].resample("D").sum().rolling("7D").mean()
    # df2.plot(ax=ax, )
    plt.show()


def new_run():
    info = []
    ct = 0
    os.chdir(repo)
    proc = run(["git", "checkout", "master"])
    while True:
        try:
            inspect(info)
        except KeyError:
            break
        ct += 1
        print(ct)
        _proc = run(["git", "checkout", "HEAD^"], stderr=PIPE)
        assert _proc.returncode == 0
    df = pandas.DataFrame.from_records(
        info, "Date", columns=["Date", "C", "Rust", "CHeader", "Added", "Deleted"]
    )
    proc = run(["git", "checkout", "master"])
    return df


if __name__ == "__main__":
    try:
        # delete tmp_csv file to re-run everything
        df = pandas.read_csv(tmp_csv, index_col="Date", parse_dates=["Date"])
    except:
        df = new_run()

    postprocess(df)

Hi there - yes of course, it's all your hard work, do whatever you like with it. I feel a little bit like I've stolen some unearned credit!

In theory, to run the script, all you need to do is pip3 install --user matplotlib pandas dateutil and change the repo location on line 10

Thanks! I'll try to get my env set up for running it in the future.

May I cannibalize the code to generate a similar chart for Remacs? We like the chart over there and are working on tracking something similar for that project.