split_pdf.py

#!/usr/bin/env python

from PyPDF2 import PdfFileWriter, PdfFileReader
import sys, getopt, os


def parse_range(page):
    start, end = page.split('-')
    start = int(start)
    end = int(end)

    return range(start, end + 1)

def parse_pages(pages):
    pages = remove_whitespace(pages)
    parsed = []
    for page in pages.split(','):
        if '-' in page:
            page_range = parse_range(page)
            parsed.extend(page_range)
        else:
            parsed.append(int(page))
    return parsed

def remove_whitespace(s):
    return "".join(s.split())

def generate_output_name(pages, inputfile):
    ext_index = inputfile.rindex('.')
    return inputfile[0:ext_index] + '_' + pages + inputfile[ext_index:]

def copy_pages(pages, inputpath, outputpath):
    if os.path.exists(outputpath):
        raise Exception('file at ' + outputpath + ' already exists')

    print 'Creating pdf \'' + outputpath + '\' from pages ' + str(pages) + ' of \'' + inputpath + '\'...'

    output = PdfFileWriter()
    input = PdfFileReader(open(inputpath, 'rb'))

    for page in pages:
        # we subtract -1 here as the library uses zero-based
        # index for the pages put we use one-based
        # page numbers as parameter
        output.addPage(input.getPage(page-1))

    out = file(outputpath, 'wb')
    output.write(out)
    out.close()
    print 'done...'

def main(argv):
    inputfile = ''
    outputfile = ''
    pages = ''

    opts, args = getopt.getopt(argv, 'i:o:p:h', ['help'])

    for opt, val in opts:
        if opt == '-h':
            print './split_pdf -i input_file [-o outputfile] -p pages'
            print 'pages is a comma separated list of indivudual page numbers or page range (e.g. 1,2,3 or 1,2,10-15)'
        elif opt == '-i':
            inputfile = val
        elif opt == '-o':
            outputfile = val
        elif opt == '-p':
            pages = val

    if not outputfile:
        outputfile = generate_output_name(pages, inputfile)
    pages = parse_pages(pages)
    copy_pages(pages, inputfile, outputfile)

    # opens the create file with system viewer
    os.system('cygstart ' + outputfile)

if __name__ == '__main__':
    main(sys.argv[1:])