Extra python modules

The python interpreter inside FreeCAD can easily be extended by adding new modules to your system's python installation. Those modules will be automatically detected and used by FreeCAD.

All python modules can be used from within FreeCAD, but several of them, listed below, have a special importance because they allow python programs complete access to core functionality of FreeCAD. Examples of use of those modules can also be found on the Code snippets page.

Notes:


 * Of the following modules, Pivy is now fully integrated into any FreeCAD installation package, and PyQt4 is also integrated in the Windows installation package.
 * PyQt4 will be progressively obsoleted in FreeCAD after version 0.13, in favour of PySide, which does exactly the same job but has a license (LGPL) more compatible with FreeCAD.

PyQt4
homepage: http://www.riverbankcomputing.co.uk/pyqt

PyQt (version 4) is a python bindings library which allow programs to access, create or modify Qt interfaces. Since the FreeCAD interface is built with Qt, installing PyQt4 on your system allow python programs inside FreeCAD to access all the interface, modify its parts, create new widgets or gather info from interface parts.

PyQt is released under a multiple licensing system, same system as used by Qt. To resume, there is a commercial version and a free GPL version. If you want to use it to make commercial (closed source) programs, you need to purchase the commercial license, otherwise you can simply install and use freely the GPL version.

Installation
Before installing PyQt4, you obviously need a python environment installed and working.

Linux

The simplest way to install PyQt4 is through your distribution's package manager. On Debian/Ubuntu systems, the package name is generally python-qt4, while on RPM-based systems it is named pyqt4. The necessary dependencies (Qt and SIP) will be taken care of automatically.

Windows

The program can be downloaded from here. You'll need to install the Qt and SIP libraries before installing pyqt4 (to be documented).

Usage
Once it is installed, you can check that everything is working by typing in FreeCAD python console:

import PyQt4

To access the FreeCAD interface, type:

from PyQt4 import QtCore,QtGui app = QtGui.qApp FreeCADWindow = app.activeWindow

Now you can start to explore the interface with the dir command. You can add new elements, like a custom widget, with commands like:

FreeCADWindow.addDockWidget(QtCore.Qt.RghtDockWidgetArea,my_custom_widget)

Documentation
More pyQt4 tutorials (including how to build interfaces with Qt Designer to use with python):

http://www.riverbankcomputing.co.uk/static/Docs/PyQt4/html/classes.html - the official PyQt4 API Reference

http://www.rkblog.rk.edu.pl/w/p/introduction-pyqt4/ - a simple introduction

http://www.zetcode.com/tutorials/pyqt4/ - very complete in-depth tutorial

Pivy
homepage: http://pivy.coin3d.org/

Pivy is a coin bindings library for python, officially supported by coin. Coin itself is a toolkit for building 3D applications in OpenGL. It is the toolkit that FreeCAD uses to draw its 3d Scene on the screen. Installing Pivy on your system will allow python programs to access the FreeCAD scenegraph, draw new objects on the scene and use the wide range of available Coin tools for drawing operations. Coin is based on the open Inventor scene description language. Pivy is used by the 2D drafting module of FreeCAD (and also by the complete module), so it is needed if you want to use any tool of those modules.

It is important to know that FreeCAD only uses coin for representation of objects on the screen, which is separated from the definition of objects. This means that pivy won't be able to modify existing objects, neither to create valid FreeCAD objects. But it can be used to draw all kind of temporary things on screen, such as axis, grids, manipulators, construction geometry, etc...

Pivy, as well as Coin, is released under a GPL license.

Prerequisites
I believe before compiling Pivy you will want to have Coin and SoQt installed.

I found for building on Mac it was sufficient to install the Coin3 binary package. Attempting to install coin from MacPorts was problematic: tried to add a lot of X Windows packages and ultimately crashed with a script error.

For Fedora I found an RPM with Coin3.

SoQt compiled from source fine on Mac and Linux.

Debian & Ubuntu
Starting with Debian Squeeze and Ubuntu Lucid, pivy will be available directly from the official repositories, saving us a lot of hassle. In the meantime, you can either download one of the packages we made (for debian and ubuntu karmic) availables on the Download pages, or compile it yourself.

The best way to compile pivy easily is to grab the debian source package for pivy and make a package with debuild. It is the same source code from the official pivy site, but the debian people made several bug-fixing additions. It also compiles fine on ubuntu karmic: http://packages.debian.org/squeeze/python-pivy (download the .orig.gz and the .diff.gz file, then unzip both, then apply the .diff to the source: go to the unzipped pivy source folder, and apply the .diff patch: patch -p1 < ../pivy_0.5.0~svn765-2.diff then debuild to have pivy properly built into an official installable package. Then, just install the package with gdebi.

Other linux distributions
First get the latest sources from the project's repository: hg clone http://hg.sim.no/Pivy/default Pivy As of March 2012, the latest version is Pivy-0.5.

Then you need a tool called SWIG to generate the C++ code for the Python bindings. Pivy-0.5 reports that it has only been tested with SWIG 1.3.31, 1.3.33, 1.3.35, and 1.3.40. So you can download a source tarball for one of these old versions from http://www.swig.org. Then unpack it and from a command line do (as root): ./configure make make install (or checkinstall if you use it) It takes just a few seconds to build.

Alternatively, you can try building with a more recent SWIG. As of March 2012, a typical repository version is 2.0.4. Pivy has a minor compile problem with SWIG 2.0.4 on Mac OS (see below) but seems to build fine on Fedora Core 15.

After that go to the pivy sources and call python setup.py build which creates the source files. Note that build can produce thousands of warnings, but hopefully there will be no errors.

This is probably obsolete, but you may run into a compiler error where a 'const char*' cannot be converted in a 'char*'. To fix that you just need to write a 'const' before in the appropriate lines. There are six lines to fix.

After that, install by issuing (as root): python setup.py install (or checkinstall python setup.py install) That's it, pivy is installed.

Mac OS
These instructions may not be complete. Something close to this worked for OS 10.7 as of March 2012. I use MacPorts for repositories, but other options should also work.

As for linux, get the latest source: hg clone http://hg.sim.no/Pivy/default Pivy If you don't have hg, you can get it from MacPorts: port install mercurial

Then, as above you need SWIG. It should be a matter of: port install swig

I found I needed also: port install swig-python

As of March 2012, MacPorts SWIG is version 2.0.4. As noted above for linux, you might be better off downloading an older version. SWIG 2.0.4 seems to have a bug that stops Pivy building. See first message in this digest: https://sourceforge.net/mailarchive/message.php?msg_id=28114815

This can be corrected by editing the 2 source locations to add dereferences: *arg4, *arg5 in place of arg4, arg5. Now Pivy should build: python setup.py build sudo python setup.py install

Windows
Assuming you are using Visual Studio 2005 or later you should open a command prompt with 'Visual Studio 2005 Command prompt' from the Tools menu. If the Python interpreter is not yet in the system path do set PATH=path_to_python_2.5;%PATH%

To get pivy working you should get the latest sources from the project's repository: svn co https://svn.coin3d.org/repos/Pivy/trunk Pivy Then you need a tool called SWIG to generate the C++ code for the Python bindings. It is recommended to use version 1.3.25 of SWIG, not the latest version, because at the moment pivy will only function correctly with 1.3.25. Download the binaries for 1.3.25 from http://www.swig.org. Then unpack it and from the command line add it to the system path set PATH=path_to_swig_1.3.25;%PATH% and set COINDIR to the appropriate path set COINDIR=path_to_coin

On Windows the pivy config file expects SoWin instead of SoQt as default. I didn't find an obvious way to build with SoQt, so I modified the file setup.py directly. In line 200 just remove the part 'sowin' : ('gui._sowin', 'sowin-config', 'pivy.gui.') (do not remove the closing parenthesis).

After that go to the pivy sources and call python setup.py build which creates the source files. You may run into a compiler error several header files couldn't be found. In this case adjust the INCLUDE variable set INCLUDE=%INCLUDE%;path_to_coin_include_dir and if the SoQt headers are not in the same place as the Coin headers also set INCLUDE=%INCLUDE%;path_to_soqt_include_dir and finally the Qt headers set INCLUDE=%INCLUDE%;path_to_qt4\include\Qt

If you are using the Express Edition of Visual Studio you may get a python keyerror exception. In this case you have to modify a few things in msvccompiler.py located in your python installation.

Go to line 122 and replace the line vsbase = r"Software\Microsoft\VisualStudio\%0.1f" % version with vsbase = r"Software\Microsoft\VCExpress\%0.1f" % version Then retry again. If you get a second error like error: Python was built with Visual Studio 2003;... you must also replace line 128 self.set_macro("FrameworkSDKDir", net, "sdkinstallrootv1.1") with self.set_macro("FrameworkSDKDir", net, "sdkinstallrootv2.0") Retry once again. If you get again an error like error: Python was built with Visual Studio version 8.0, and extensions need to be built with the same version of the compiler, but it isn't installed. then you should check the environment variables DISTUTILS_USE_SDK and MSSDK with echo %DISTUTILS_USE_SDK% echo %MSSDK% If not yet set then just set it e.g. to 1 set DISTUTILS_USE_SDK=1 set MSSDK=1

Now, you may run into a compiler error where a 'const char*' cannot be converted in a 'char*'. To fix that you just need to write a 'const' before in the appropriate lines. There are six lines to fix. After that copy the generated pivy directory to a place where the python interpreter in FreeCAD can find it.

Usage
To have Pivy access the FreeCAD scenegraph do the following:

from pivy import coin App.newDocument # Open a document and a view view = Gui.ActiveDocument.ActiveView FCSceneGraph = view.getSceneGraph # returns a pivy Python object that holds a SoSeparator, the main "container" of the Coin scenegraph FCSceneGraph.addChild(coin.SoCube) # add a box to scene

You can now explore the FCSceneGraph with the dir command.

Documentation
Unfortunately documentation about pivy is still almost inexistant on the net. But you might find Coin documentation useful, since pivy simply translate Coin functions, nodes and methods in python, everything keeps the same name and properties, keeping in mind the difference of syntax between C and python:

http://doc.coin3d.org/Coin/classes.html - Coin3D API Reference

http://www-evasion.imag.fr/~Francois.Faure/doc/inventorMentor/sgi_html/index.html - The Inventor Mentor - The "bible" of Inventor scene description language.

You can also look at the Draft.py file in the FreeCAD Mod/Draft folder, since it makes big use of pivy.

pyCollada
pyCollada is a python library that allow programs to read and write Collada (*.DAE) files. When pyCollada is installed on your system, FreeCAD will detect it and add import and export options to handle opening and saving in the Collada file format.

Installation
Pycollada is usually not yet available in linux distributions repositories, but since it is made only of python files, it doesn't require compilation, and is easy to install. You have 2 ways, or directly from the official pycollada git repository, or with the easy_install tool.

Linux
In either case, you'll need the following packages already installed on your system:

python-lxml python-numpy python-dateutil

From the git repository
git clone git://github.com/pycollada/pycollada.git pycollada cd pycollada sudo python setup.py install

With easy_install
Assuming you have a complete python installation already, the easy_install utility should be present already:

easy_install pycollada

You can check if pycollada was correctly installed by issuing in a python console:

import collada

If it returns nothing (no error message), then all is OK

Windows
To Be Documented

IfcOpenShell
IFCOpenShell is a library currently in development, that allows to import (and soon export) Industry foundation Classes (*.IFC) files. IFC is an extension to the STEP format, and is becoming the standard in BIM workflows. When ifcopenshell is correctly installed on your system, the FreeCAD Arch Module will detect it and use it to import IFC files. Since ifcopenshell is based on OpenCasCade, like FreeCAD, the quality of the import is very high, producing high-quality solid geometry.

Homepage: http://www.ifcopenshell.org

Installation
Since ifcopenshell is pretty new, you'll likely need to compile it yourself.

Linux
You will need a couple of development packages installed on your system in order to compile ifcopenshell:

liboce-*-dev python-dev swig

but since FreeCAD requires all of them too, if you can compile FreeCAD, you won't need any extra dependency to compile IfcOpenShell.

Grab the latest source code from here:

svn co https://ifcopenshell.svn.sourceforge.net/svnroot/ifcopenshell ifcopenshell

The build process is very easy:

mkdir ifcopenshell-build cd ifcopenshell-build cmake ../ifcopenshell/cmake

or, if you are using oce instead of opencascade:

cmake -DOCC_INCLUDE_DIR=/usr/include/oce ../ifcopenshell/cmake

Since ifcopenshell is made primarily for Blender, it uses python3 by default. To use it inside FreeCAD, you need to compile it against the same version of python that is used by FreeCAD. So you might need to force the python version with additional cmake parameters (adjust the python version to yours):

cmake -DOCC_INCLUDE_DIR=/usr/include/oce -DPYTHON_INCLUDE_DIR=/usr/include/python2.7 -DPYTHON_LIBRARY=/usr/lib/python2.7.so ../ifcopenshell/cmake

Then:

make sudo make install

You can check that ifcopenshell was correctly installed by issuing in a python console:

import IfcImport

If it returns nothing (no error message), then all is OK