Count Git LFS objects and their sizes, total and per file type

git_lfs_calculate_size_by_type.py

#!/usr/bin/env python
# Counts Git LFS objects and per file type, and calculates total sizes.
import os
import sys
import tempfile

print('Collecting sizes of Git LFS objects')
tmp_file = os.path.join(os.getcwd(), 'tmp_sizes.tx')
os.system('git lfs ls-files --long --size > {0}'.format(tmp_file))

print('Calculating total size per file type')
total = {}
to_bytes = {"B": 1, "KB": 10**3, "MB": 10**6, "GB": 10**9, "TB": 10**12}

with open(tmp_file, 'r') as f:
    for line in f:
        sha, line = line.split(' - ')
        name, line = line.split(' (')
        size, line = line.split(')')
        name, ext = os.path.splitext(name)
        size, unit = size.split()

        if not ext in total:
            total[ext] = { 'count': 0, 'size': 0}
        total[ext]['count'] += 1
        total[ext]['size'] += int(float(size) * to_bytes[unit])

if total:
    print('Git LFS objects summary:')
    for k, v in total.items():
        print('{}:\tcount: {}\tsize: {:.2f} MB'.format(k, v['count'], v['size'] / 1024**2))
    c = sum([ v['count'] for _,v in total.items() ])
    s = sum([ v['size'] for _,v in total.items() ]) / 1024**2
    print('Total:\tcount: {}\tsize: {:.2f} MB'.format(c, s))
else:
    print('no summary to print')

if os.path.exists(tmp_file):
    os.unlink(tmp_file)

git_lfs_list_objects_by_size.sh

#!/bin/sh

# The ls-files report human-friendly sizes.
# The poor-man sed commands convert all sizes to bytes.
git lfs ls-files --long --size \
    | awk '{print $4" "$5" "$3" "$1}' \
    | tr -d '()' \
    | sed 's/^\([0-9]\+\)\(\.[0-9]\+\)\? GB/\1000000000\2/' \
    | sed 's/^\([0-9]\+\)\(\.[0-9]\+\)\? MB/\1000000\2/' \
    | sed 's/^\([0-9]\+\)\(\.[0-9]\+\)\? KB/\1000\2/' \
    | sed 's/^\([0-9]\+\)\(\.[0-9]\+\)\? B/\1\2/' \
    | sort --key 1 --numeric-sort --reverse \
    | uniq

@massich Nice one, I wasn't aware of --human-numeric-sort. Thanks.