Remove (white) background from pdfs generated by browsers

removebg.py

#!/usr/bin/python3

# https://tex.stackexchange.com/questions/50069/how-to-change-white-background-of-an-included-pdf-to-transparent

# use :
# python3 removebg.py filewithbackground.pdf
# the script is not smart at all : it resizes every “full page” element that is a background 

import os
import sys
import math
import subprocess

# toggle interactive mode
interactive = False

# preset: A3, no bleed, two pages
preset = {
  "width": 29.7,
  "height": 42.0,
  "bleed": 0.0,
  "pages": 2
}

def gather_info():
  if not interactive:
    return preset.values()
  width = int(input("Page width? (millimeters)")) / 10
  height = int(input("Page height? (millimeters)")) / 10
  bleed = int(input("Any bleed? (millimeters)")) / 10
  pages = int(input("Number of pages? "))
  return (width, height, bleed, pages)

fn = sys.argv[1]
if os.path.exists(fn):

  # file and temp files
  bn = os.path.basename(fn)
  print(u"Processing {}".format(bn))

  index = bn.rfind(".")
  fn_qdf=u"{}.readable.pdf".format(fn)
  fn_tmp=u"{}.tempfile.pdf".format(fn)

  width, height, bleed, pages = gather_info() 

  print(u"fn = {}, fn_tmp = {}, fn_qdf = {}, width = {}, height = {}, bleed = {}, pages = {}".format(fn, fn_tmp, fn_qdf, width, height, bleed, pages))

  # compute white rect width and height
  w = math.ceil((96/2.54) * (width + bleed))
  h = math.ceil((96/2.54) * (height + bleed)) * pages

  print(u"w = {}, h = {}".format(w, h))

  # converts to readable format
  cmd = "qpdf -qdf '{}' '{}'".format(fn, fn_qdf)
  os.system(cmd)
  
  # resize full page rectangle to 0 × 0
  cmd = "cat '{}' | sed 's/{} {}/0 0/g' > {}".format(fn_qdf, w, h, fn_tmp)
  os.system(cmd)

  # fix pdf 
  cmd = "fix-qdf '{}' > {}".format(fn_tmp, fn)
  os.system(cmd)

  # remove temp files
  os.remove(fn_qdf)
  os.remove(fn_tmp)