goerz / bmconverter.py Goto Github PK

#!/usr/bin/env python
# -*- coding: UTF-8 -*-

import os, sys


def usage():
    print 'Usage: python pdftk2pdfmarks.py PDFTK-bookmark-file-input PDFMARKS-bookmark-file-output'


def cleanup(s):
#   cleanup pairs: replace odd with even (many things not considered)
    i = ['\xad','-', ':',' -', '(','[', ')',']', '\n','']
    for j in range(len(i)/2):
        s = s.replace(i[2*j-2], i[2*j-1])
    return s


def main(data):

    if len(data) < 2:
        usage()
    else:

#       open pkftk bookmark file (from `pdftk <inputpdf> dump_data_utf8 <bookmarks file>`)
        f = open(data[0], 'r')

#       open pdfmarks bookmark file for writing/appending
        pdfmarks = open(data[1], 'a')

        try:
            l = []
            for line in f:

#               put node levels in list l
                if line[:13] == 'BookmarkLevel':
                    l.append(int(line.split(':')[1]))

            j = [0] * len(l)

#           most important part: convert levels (pdftk "notation") to subordinate count (pdfmarks "notation")
            for u, v in enumerate(l):
                for other in l[u+1:]:
                    if other == v+1:
                        j[u] += 1
                    elif other == v:
                        break

#           now write to pdfmarks file (self-explainatory)
            i = 0
            f.seek(0)
            for line in f:

                m = line.split(':')[0]
                n = line[len(m)+2:-1]

                if m[:8] == 'Bookmark':

                    if i == -1 : pdfmarks.write(' /DOCINFO pdfmark'); i += 1

                    if m == 'BookmarkTitle':
                        pdfmarks.write('\n[ /Title (%s)' % (cleanup(n)))

                    if m == 'BookmarkLevel':
                        if not j[i] == 0:
                            pdfmarks.write(' /Count %s' % (j[i]))
                        i += 1

                    if m == 'BookmarkPageNumber':
                        pdfmarks.write(' /Page %s /OUT pdfmark' % (n))

                else:
                    if m == 'InfoKey':
                        if i == 0:
                            pdfmarks.write('[')
                        pdfmarks.write(' /%s' % (n))
                    if m == 'InfoValue':
                        pdfmarks.write(' (%s)\n' % (cleanup(n))); i = -1

        except Exception, e:
            print e

        pdfmarks.close()
        f.close()


if __name__ == "__main__":

    try:
        main(sys.argv[1:])
    except:
        usage()