trouble using ExportFormFields in uno API; PDF form controls are exported even when set to False

[Dan Connolly]

Binary package hint: openoffice.org

The spec for the PDF export dialog says

"ExportFormFields" oor:type="xs:boolean" default:true Specifies whether form fields are exported as widgets or only their fixed print representation is exported.

http://specs.openoffice.org/appwide/pdf_export/PDFExportDialog.odt

but it doesn't seem to actually work.

I took the ooextract.py example
http://udk.openoffice.org/python/samples/ooextract.py
and tried to use ExportFormFields, without luck.

I'm not the only one having this problem; in a Dec 2006 forum post, I see:

         So, has anyone been able to use OOo's writer_pdf_export
         filter and the "ExportFormFields" property with success?
         I have not had success with the following code:
         -- http://www.oooforum.org/forum/viewtopic.phtml?t=50504&highlight=flatten+pdf

Then I found an Aug 2006 suggestion from Giuseppe Castagno
http://<email address hidden>/msg03726.html

Castagno's code shows a FilterData indirection, so my code now looks like:

        filterOptions = (
            PropertyValue( "ExportFormFields", 0, False, 0),
            )

        outProps = (
            PropertyValue( "FilterName" , 0, filterName , 0 ),
            PropertyValue( "FilterData", 0, filterOptions, 0 ),
     PropertyValue( "Overwrite" , 0, True , 0 ),
            PropertyValue( "OutputStream", 0, OutputStream(), 0),
 )

but the resulting .pdf still has the form controls. It's distinctly
different from what I get via the GUI, which has different shaped
X's in the checkbox controls.

Help?

The forum post shows work-arounds involving post-processing
the PDF with some iText Java library, but I'd really rather not go there.

[Dan Connolly (connolly)]

attaching the whole modified ooexport.py

[Dan Connolly (connolly)]

Here's a simple document with one form control to help reproduce the problem.

[Dan Connolly (connolly)]

This is the sort of PDF that I want; it's produced using the PDF options dialog, manually.

[Dan Connolly (connolly)]

This is the output of:

$ python ooextract.py --pdf formcontrol.odt

it exhibits the problem: I want it to have no form controls,
but it has form controls.

[Chris Cheney (ccheney)]

On openoffice.org 1:2.4.0~rc2-1ubuntu3. For me they all look like your formcontrol.pdf, gui ooextract.py from the website and the ooextract.py that you modified. They all have the stylized X mark. So does that mean they are all broken for what you are trying to do?

[Dan Connolly (connolly)]

yes, if it's like formcontrol.pdf, it's broken. What I expect is formcontrol-gui.pdf.

[Dan Connolly (connolly)]

I got a clue from somebody...

"Your problem is that you need to pass an array as a value of a PropertyValue, and a PropertyValue can receive anything. So you have to tell the COM bridge what kind of value you are passing."

second comment on http://www.oooforum.org/forum/viewtopic.phtml?t=61144

[Bruce van der Kooij (brucevdk)]

That somebody was me by the way, I stumbled upon this issue while researching another property (PageRange) which is also a sub property of FilterData. Took me about 2 hours before I found B Marcelly's post mentioned above. I initially had some trouble finding out there was a PageRange property at all (for example it's not described in the spec linked above), however all properties seem to be listed in ImpPDFTabDialog::GetFilterData (see: http://lxr.go-oo.org/source/framework/filter/source/pdf/impdialog.cxx).

I was going to try and summarize the issue but since my experience with the UNO stuff is severely limited (to today) there's nothing I can add. Thought I'd at least take the time to post a patch for the previously attached ooextract.py. Here's the important snippet:

filterOptions = uno.Any("[]com.sun.star.beans.PropertyValue", (
    PropertyValue( "ExportFormFields", 0, False, 0),
),)

Notice how the uno.Any function is used to wrap the filterOptions tuple, or set the type to []com.sun.star.beans.PropertyValue (which I assume means an array of com.sun.star.beans.PropertyValue). B Marcelly explains the reason for this wrapping in this specific post over at the OOo forums (http://www.oooforum.org/forum/viewtopic.phtml?t=58792#233845).

I'll add a couple of keywords to this post so hopefully other people will stumble on this bug report when they're facing the same problems.

Keywords: writer_pdf_Export, PageRange, PropertyValue, filters, PyUNO

P.S.

Thanks to cl over at #dev.openoffice.org for his assistance with this issue.

[Chris Cheney (ccheney)]

Setting to alpha-4 so I don't forget to get this bug fixed.

[Chris Cheney (ccheney)]

12:48 < calc> Brucevdk: thanks for the update, i'll get it into the next upload to intrepid
12:49 < Brucevdk> calc: intrepid? If you're talking about patching ooextract.py, don't.
12:50 < Brucevdk> calc: the patch was for Dan's modified version, not the one at openoffice.org (that one deals with a generic
12:50 < Brucevdk> pdf exporter)
12:52 < Brucevdk> calc: my conclusion is that it's not a bug at all, should probably be set INVALID