Show the sizes of objects in a pptx file

pptxslidesizes.py

#!/usr/bin/env python
'''
Python script to show the size of images (bytes) on each slide

Requires pptx installed:

  pip install -U python-pptx
'''

import os
import locale
import pptx
import sys
import logging


def dictSum(d):
  t = 0
  for k in list(d.keys()):
    t += d[k]
  return t


def slideImageSizes(slide):
  images = {} #sha1 : size
  tbytes = 0
  for item in slide.shapes:
    if isinstance(item, pptx.shapes.picture.Picture):
      sha = item.image.sha1
      tbytes += len(item.image.blob)
      if sha not in images:
        images[sha] = len(item.image.blob)
  return tbytes, images


def doWork(fname):
  statinfo = os.stat(fname)
  presentation = pptx.Presentation(fname)
  slides = presentation.slides
  all_images = {}
  i = 1
  total = 0
  utotal = 0
  print("# ")
  print("# N = slide number")
  print("# T1 = total bytes of images on slide")
  print("# T2 = total bytes for slide not including images counted elsewhere")  
  print(("{0:>3} : {1:>12}  {2:>12}".format("N", "T1", "T2")))
  for slide in slides:
    uisize = 0
    tsize, images = slideImageSizes(slide)
    usize = dictSum(images)
    total += tsize
    for k in list(images.keys()):
      if k not in all_images:
        uisize += images[k]
        all_images[k] = images[k]        
    tssize = locale.format("%d", tsize, grouping=True)
    tusize = locale.format("%d", uisize, grouping=True)
    print(("{0:>3} : {1:>12}  {2:>12}".format(i, tssize, tusize)))
    i += 1
    total += tsize
  t_total = locale.format("%d", total, grouping=True)
  u_total = locale.format("%d", dictSum(all_images), grouping=True)
  print(("Total = {0:>12}  {1:>12}".format(t_total, u_total) ))
  print(("File size (compressed) = " + locale.format("%d", statinfo.st_size, grouping=True)))


if __name__ == "__main__":
  locale.setlocale(locale.LC_ALL, 'en_US')
  logging.basicConfig(level=logging.DEBUG)
  try:
    fname = sys.argv[1]
  except:
    print(("Oops. File name needed\nTry: {0} file_name.pptx".format(sys.argv[0])))
    sys.exit(0)
  doWork(fname)