pdf.py

# This script requires Wand for Python. Install using the documentation at http://docs.wand-py.org/en/0.4.1/index.html before running.

import sys, os, datetime
from wand.image import Image

list = os.listdir(os.getcwd())
tuples = []

for file in list:

	location = os.path.join(os.getcwd(), file)
	size = os.path.getsize(location)
	tuples.append((file, size))

tuples.sort(key=lambda s: s[1])

for n in tuples:

	item = n[0]
	if item.endswith(".pdf"):
		start = datetime.datetime.now()
		print 'Working on "' + item + '" (' + str(datetime.datetime.now()) + ').'
		with Image(filename=item, resolution=300) as readpdf:
			pages = len(readpdf.sequence)
			folder = os.getcwd() + '/' + item[:-4]
			if not os.path.exists(folder):
				os.makedirs(folder)
			jpgfile = folder + '/' + item[:-4] + '.jpg'
 			readpdf.compression_quality = 100
			readpdf.compression = 'no'
			readpdf.COLORSPACE_TYPES = 'srgb'
			readpdf.save(filename=jpgfile)
		os.rename(item, folder + '/' + item)
		end = datetime.datetime.now()
		print 'Took ' + str(int(round((end - start).total_seconds()))) + ' seconds for "' + item + '" (' + str(pages) + ' page(s)).'