[vtkusers] connecting wrapped VTK and swig output

[Mathieu Malaterre]

On Feb 14, 2008 3:27 PM, Amy Squillacote <ahs at cfdrc.com> wrote:
>
> Mathieu Malaterre wrote:
> > Hi Amy !
> >
> > On Wed, Feb 13, 2008 at 7:50 PM, Amy Squillacote <ahs at cfdrc.com> wrote:
> >
> >> Hi all,
> >>
> >>  I have written some C++ code that makes use of VTK, including using VTK
> >>  objects as parameters and return values for some methods. My VTK is from
> >>  CVS on 1/22/08. It is wrapped in Tcl, Python, and Java. I am trying to
> >>  wrap the C++ code I have written into Java and Python. Since my new
> >>  classes are not subclasses of vtkObject, I don't think I can use the
> >>  wrappers in VTK. (Please correct me if I am wrong about this.) Thus, I
> >>  am trying to use swig to do the wrapping since it can output wrappers in
> >>  both Java and Python. The problem is that the swig wrapping isn't aware
> >>  of VTK, so for example, it creates objects of type
> >>  SWIGTYPE_p_vtkPolyData to hold vtkPolyData* in the wrapped languages.
> >>  I'm not sure how to make the conversion between this and the vtkPolyData
> >>  that's wrapped in Java by VTK.
> >>
> >>  Related to this, I saw this message from a few years ago on the vtkusers
> >>  list:
> >>  http://public.kitware.com/pipermail/vtkusers/2003-October/070054.html.
> >>  However, it only applies to python-wrapping, and I need java-wrapping as
> >>  well.
> >>
> >>  Does anyone have experience with mixing swig-wrapped code and
> >>  VTK-wrapped code?
> >>
> >
> > We do that quite extensively here ! The only difference is that we
> > have the gdcm library wrapped with swig, and we also have a vtkgdcm
> > library where all classes are in fact VTK class (derives from
> > vtk*filters) that are wrapped with vtkWrapPython.
> > So in summary I never tried your mixture, but I am pretty confident
> > that if you do in your swig interface:
> >
> > #include "vtkPolyData.h"
> > ...
> > %include "my_class_to_wrap_containing_vtkpolydata.h"
> >
> > Then swig should be able to do the wrapping properly. In the past I
> > have had some subbtle issues, with the ordering of the #include in the
> > swig interface file, so be sure to include them in the right order.
> >
> >
> > HTH,
> >
> Hi Mathieu!
>
> Thanks for the suggestion, but I can't get it to work. (In fact, I tried
> it yesterday.) I put #include "vtkPolyData.h" as the first include in my
> "%{ ... }%" section. (I don't want to do %include "vtkPolyData.h" since
> I don't want vtkPolyData.h wrapped by swig.) For the methods in my
> classes that use vtkPolyData* as a parameter or return type, swig uses a
> SWIGTYPE_p_vtkPolyData instead.
>
> I'm using swig 1.3.33 on Windows, if that makes any difference.

Hi Amy,

  Sorry I meant to answer, but forgot to...

Have you read this part of Wrapping/Python/README:

Creating a Python object from just the address of a VTK object
--------------------------------------------------------------

When you instantiate a class, you can provide a hexidecimal string
containing the address of an existing vtk object, e.g.
   t = vtkTransform('_1010e068_vtkTransform_p')
The string follows SWIG mangling conventions.  If a wrapper for the
specified object already exists, then that wrapper will be used rather
than a new wrapper being created.  If you want to use this feature
of vtkpython, please think twice.


HTH

-- 
Mathieu