[reportlab-users] Bug in SimpleDocTemplate? --> Automated PDF printing

[Henning von Bargen]

Topic should be: automated PDF printing

If you want to have automatic (no user-interaction)
printing of PDF files ON WINDOWS, things get complicated.

And BTW this is off-topic here, perhaps you should post
to the ghostscript mailing list.

Anyway:

GhostScript has its own problems here.
This is, the mswinpr2 device and other special devices
(for PS or PCL printers) work quite differently according
to landscape printing.

If you want professional PDF batch PDF printing, then you
might want to try the PDF Printer API from the Swiss company
http://www.pdf-tools.ch .
For our commercial application, I used this API and created
a small wrapper around it with SWIG.
You can also use the command-line version, they have a trial
download available at http://www.pdf-tools.com/asp/products.asp?name=prns 

In our application, we also used Ghostscript first,
but we had to use some complicated code to select the correct
device first, and the mswinpr2 device is terrible slow.

For automatically detecting and "correcting" landscape PDFs, we used
the following code fragments:

def isLandscape (filename):
    """
    Prüft, ob es sich um eine Datei im Querformat handelt.
    """
    landscape = False
    fin = open(filename, "rb", 32768)
    inPage = False
    try:
        for zeile in fin:
            if zeile.split() == ["/Type", "/Page"]:
                inPage = True
            elif zeile.split() == [">>"]:
                inPage = False
            elif zeile.startswith("/MediaBox") and inPage:
                arr = zeile[11:zeile.find("]")].split()
                if int(arr[2]) > int(arr[3]):
                    landscape = True
                    break
                inPage = False
    finally:
        fin.close()
    return landscape


def rotatePDF (filename, newFilename):
    """
    Rotiert eine PDF-Datei um 90 Grad, so dass sie korrekt ausgedruckt werden kann.
    Voraussetzung dafür ist, dass die Datei keine /Rotate Einträge enthält
    und eigentlich Querformat haben sollte (zu erkennen an /MediaBox [0 0 AAA BBB]
    mit AAA BBB üblicherweise = 842 595
    
    Da die Datei Zeilenweise gelesen wird, dauert diese Operation evtl. etwas länger.
    """
    
    INSERT_STRING = "/Rotate 90\n"
    
    fin = open(filename, "rb", 32768)
    fout = open(newFilename, "wb", 32768)
    xref = {}
    xref["0"] = {}
    maxObj = 0
    olen = 0
    rotateFound = False
    inXREF = False
    modus = None
    try:
        for zeile in fin:
            #print zeile,
            if zeile.endswith(" obj\n") or zeile.endswith(" obj\r\n"):
                # OBJ erkannt
                modus = "obj"
                objType = None
                # Position (neu) merken
                (objnum, updnum, _) = zeile.split()
                try:
                    xref[updnum][objnum] = olen    
                except KeyError:
                    xref[updnum] = { objnum: olen }
            elif zeile.startswith("/Type "):
                objType = zeile.split()[1]
                if objType == "/Page":
                    rotateFound = False
                    landscape = False
            elif zeile.startswith("/Rotate ") and objType == "/Page":
                rotateFound = True
            elif zeile.startswith("/MediaBox") and objType == "/Page":
                arr = zeile[11:zeile.find("]")].split()
                if int(arr[2]) > int(arr[3]):
                    landscape = True
            elif zeile == ">>\n" or zeile == ">>\r\n":
                # Objekt-Ende
                if objType == "/Page" and (landscape and not rotateFound):
                    # Seitendefinition: /Rotate 90 Zeile hinzufügen
                    fout.write (INSERT_STRING)
                    olen += len(INSERT_STRING)
                modus = None
            elif zeile == "xref\n" or zeile =="xref\r\n":
                print xref
                modus = "xref"
                firstObj = None
                xrefOffset = olen
                xrefObjNum = None
            elif zeile == "startxref\n" or zeile == "startxref\r\n":
                modus = "startxref"
            # Zeile ausgeben
            if modus == "xref":
                #print "in xref, %r" % zeile
                if len(zeile) == 20:
                    if zeile.split()[-1] == "f":
                        # Free object
                        fout.write (zeile)
                        olen += len(zeile)
                    else:
                        #print "object in use"
                        generation = int(zeile.split()[1])
                        fout.write ("%010d %05d n \n" % (xref[str(generation)][str(xrefObjNum)], generation))
                        olen += 20
                    xrefObjNum += 1                    
                else:
                    if len(zeile.split()) == 2:
                        # Neue subsection
                        xrefObjNum = int(zeile.split()[0])
                    else:
                        if xrefObjNum is not None:
                            modus = None
                    fout.write (zeile)
                    olen += len(zeile)
            elif modus == "startxref" and not zeile.startswith(modus):
                fout.write ("%d\n" % xrefOffset)
                modus = None
            else:
                fout.write (zeile)
                olen += len(zeile)
    finally:
        fin.close()
        fout.close()

------
But be aware that this code has been tested only for Landscape PDFs
generated with Oracle Reports (which has nothing to do with ReportLab).

HTH
Henning