Schlangengrube

Mit dem folgenden Skript (sicherlich noch ausbaufähig!) kann man eine PDF-Datei ins 2-Up Format umwandeln (d.h. 2 Seiten auf eine):

#!/bin/env/python
# -*- coding: utf-8 -*-

# Basiert auf der pyPdf-Library von Mathieu Fenniak
# Siehe http://pybrary.net/pyPdf
#
# (C) Henning von Bargen 2007
#
import os
import sys
import unittest
from cStringIO import StringIO

from pyPdf import PdfFileWriter, PdfFileReader

from reportlab.lib.units import cm, mm, inch
from reportlab.lib import pagesizes
from reportlab.pdfgen import canvas

from pyPdf.pdf import *
from pyPdf.generic import *    


PAGESIZE = pagesizes.landscape(pagesizes.A4)

def framePage(canvas, title):
    canvas.saveState()
    canvas.setFont('Times-BoldItalic',12)
    canvas.drawCentredString(PAGESIZE[0] * 1/2, 200*mm, title)
    canvas.setFont('Times-Roman',10)
    canvas.drawCentredString(PAGESIZE[0] * 1/4, 10*mm,
                            'Page %d' % (canvas.getPageNumber() * 2 - 1))
    canvas.drawCentredString(PAGESIZE[0] * 3/4, 10*mm,
                            'Page %d' % (canvas.getPageNumber() * 2))
    canvas.restoreState()
    #canvas.doForm("frame") # Hier wäre Platz für ein Logo

def makeEmptyPagesPdf(nPages):
    buffer = StringIO()
    c = canvas.Canvas(None)
    c.setPageSize(PAGESIZE)
    #c.setPageCompression(0)
    #c.setPageCallBack(pageCallBack)
    #framePageForm(c) # define the frame form
    c.showOutline()
    for page in range(nPages):
        print page
        framePage(c, "This is a title")
        c.showPage()
    buffer.write(c.getpdfdata())
    buffer.seek(0)
    return buffer


def _pushTransformPopGS(contents, offset, scale, pdf):
    # adds a graphics state "push" and "pop" to the beginning and end
    # of a content stream.  This isolates it from changes such as 
    # transformation matricies.
    print "offset:",offset
    print "scale:",scale
    stream = ContentStream(contents, pdf)
    stream.operations.insert(0, [[], "q"])
    arr = [scale[0],0,0,scale[1],offset[0],offset[1]]
    arr = [FloatObject(str(x)) for x in arr]
    stream.operations.insert(1, [arr, "cm"])
    stream.operations.append([[], "Q"])
    return stream
    
def mergePage(page1, page2, offset, scale):

    newResources = DictionaryObject()
    rename = {}
    originalResources = page1["/Resources"].getObject()
    page2Resources = page2["/Resources"].getObject()

    for res in "/ExtGState", "/Font", "/XObject", "/ColorSpace", "/Pattern", "/Shading":
        new, newrename = PageObject._mergeResources(originalResources, page2Resources, res)
        if new:
            newResources[NameObject(res)] = new
            rename.update(newrename)

    # Combine /ProcSet sets.
    newResources[NameObject("/ProcSet")] = ArrayObject(
        ImmutableSet(originalResources.get("/ProcSet", ArrayObject()).getObject()).union(
            ImmutableSet(page2Resources.get("/ProcSet", ArrayObject()).getObject())
        )
    )

    newContentArray = ArrayObject()

    originalContent = page1["/Contents"].getObject()
    newContentArray.append(PageObject._pushPopGS(originalContent, page1.pdf))

    page2Content = page2['/Contents'].getObject()
    page2Content = PageObject._contentStreamRename(page2Content, rename, page1.pdf)
    page2Content = _pushTransformPopGS(page2Content, offset, scale, page1.pdf)
    newContentArray.append(page2Content)

    page1[NameObject('/Contents')] = ContentStream(newContentArray, page1.pdf)
    page1[NameObject('/Resources')] = newResources

def write2Up(infname, outfname):

    output = PdfFileWriter()
    input = PdfFileReader(open(infname, "rb"))
    nPages = input.getNumPages()
    print "nPages=",nPages
    nNupPages = int((nPages+1)/2.0)
    print "nNupPages=",nNupPages
    emptyPages = PdfFileReader(makeEmptyPagesPdf(nNupPages))
    
    page = None
    nupindx = 0
    for indx in range(nPages):
        print "processing page", indx
        if indx % 2 == 0:
            if page is not None:
                output.addPage(page)
            print "nupindx=",nupindx
            page = emptyPages.getPage(nupindx)
            nupindx += 1
        mergePage(page, input.getPage(indx),
                  (PAGESIZE[0]/2 * (indx%2), 0), # offset x,y
                  (0.7, 0.7)   # scaling x,y
                 )
    if page is not None:
        output.addPage(page)
    outputStream = open(outfname, "wb")
    output.write(outputStream)


if __name__ == "__main__":
    infname = sys.argv[1]
    if len(sys.argv) > 2:
        outfname = sys.argv[2]
    else:
        outfname = os.path.splitext(infname)[0] + "-new.pdf"
    write2Up(infname, outfname)