A Python script for visualizing the size of objects in your git repo.

git-files-histogram.py

#!/usr/bin/env python3
"""
Counts file size of objects in your repository's packfile and sorts by
buckets. Defaults to buckets of 1M.
Requires third-party libraries matplotlib and numpy
Author: Jason Drew Panzer
"""

from matplotlib import pyplot as plt
from subprocess import run
import numpy as np

def packfile_histogram():
    # git verify-pack for all pack index files, filtered by blobs and sorted by size in reverse order
    objects = run('git verify-pack -v .git/objects/pack/pack-*.idx | grep blob | grep -v chain | sort -k3nr', shell=True, capture_output=True, encoding='utf-8')
    objects = objects.stdout.split('\n')[:-1]

    # reshape blob sizes for numpy.histogram
    object_sizes = [(int(x.split()[2])/(1024**2)) for x in objects]
    object_size_hist, object_size_bins = np.histogram(object_sizes, bins=50)

    #show off that histogram
    plt.bar(object_size_bins[:-1], object_size_hist, width=1)
    plt.xlabel("size in mb")
    plt.xlim(1, max(object_size_bins))
    plt.ylabel("number of objects")
    plt.ylim(0, max(object_size_hist[1:]))
    plt.show()

if __name__ == '__main__':
    packfile_histogram()