Show large files in Git repository

git-large-files

#!/usr/bin/env python
#
# Print list of large files in Git
# Copyright (c) 2015 Adam Strzelecki
#
# Installation:
#   Place it somewhere in your path, then use with `git large-files'
#
import os, sys, re

def usage():
  print '''usage: git large-files size<k|m> [rev-range]

examples:
  git large-files 1m
  git large-files 10m master^^^..
'''

def large_files(min_size, rev_range='HEAD'):
  """Print large files"""
  big_files = []
  for revision in os.popen('git rev-list %s' % rev_range):
    for f in os.popen('git ls-tree -zrl %s' % revision).read().split('\0'):
      if f:
        mode, type, commit, size, path = f.split(None, 4)
        size = int(size)
        if size > min_size:
          big_files.append((int(size), commit, path))

  for f in sorted(set(big_files), reverse=True):
    print '%-8s  %-40s  %s' % (sizeof_fmt(f[0]), f[1], f[2])

from math import log

def sizeof_fmt(num):
  """Human friendly file size"""
  unit_list = zip(['bytes', 'kB', 'MB', 'GB', 'TB', 'PB'], [0, 0, 1, 2, 2, 2])
  if num > 1:
    exponent = min(int(log(num, 1024)), len(unit_list) - 1)
    quotient = float(num) / 1024**exponent
    unit, num_decimals = unit_list[exponent]
    format_string = '{:.%sf} {}' % (num_decimals)
    return format_string.format(quotient, unit)
  elif num == 0:
    return '0 b'
  elif num == 1:
    return '1 b'

def parse_size(str):
  """Parse size into a number"""
  match = re.match(r'(\d+)([km])?', str.lower())
  unit_values = {None:1, 'k': 1024, 'm': 1024*1024}
  if match:
    size = int(match.group(1))
    unit_size = unit_values[match.group(2)]
    return size * unit_size
  else:
    return None

if len(sys.argv) >= 2 and len(sys.argv) <= 3:
  size = parse_size(sys.argv[1].lower())
  if size:
    if len(sys.argv) >= 3:
      large_files(size, sys.argv[2])
    else:
      large_files(size)
  else:
    usage()
else:
  usage()