The wrap library wraps the GTSAM library into a Python library or MATLAB toolbox.
It was designed to be more general than just wrapping GTSAM. For notes on creating a wrap interface, see gtsam.h
for what features can be wrapped into a toolbox, as well as the current state of the toolbox for GTSAM.
Pybind11
and pyparsing
pybind11
, which is included as a subdirectory in GTSAM.interface_parser.py
in this library uses pyparsing
to parse the interface file gtsam.h
. Please install it first in your current Python environment before attempting the build.python3 -m pip install pyparsing
Clone this repository to your local machine and perform the standard CMake install:
mkdir build && cd build
cmake ..
make install # use sudo if needed
Using wrap
in your project is straightforward from here. In your CMakeLists.txt
file, you just need to add the following:
find_package(gtwrap)
set(interface_files ${PROJECT_SOURCE_DIR}/cpp/${PROJECT_NAME}.h)
pybind_wrap(${PROJECT_NAME}_py # target
"${interface_files}" # list of interface header files
"${PROJECT_NAME}.cpp" # the generated cpp
"${PROJECT_NAME}" # module_name
"${PROJECT_MODULE_NAME}" # top namespace in the cpp file e.g. gtsam
"${ignore}" # ignore classes
${PROJECT_BINARY_DIR}/${PROJECT_NAME}.tpl # the wrapping template file
${PROJECT_NAME} # libs
"${PROJECT_NAME}" # dependencies
ON # use boost serialization
)
For more information, please follow our tutorial.
Documentation for wrapping C++ code can be found here.
WARNING: On macOS, you have to statically build GTSAM to use the wrapper.
Set GTSAM_BUILD_PYTHON=ON
while configuring the build with cmake
.
What you can do in the build
folder:
Just run python then import GTSAM and play around:
import gtsam
gtsam.__dir__()
Run the unittests:
python -m unittest discover
Edit the unittests in python/gtsam/*.py
and simply rerun the test.
They were symlinked to <build_folder>/gtsam/*.py
to facilitate fast development.
python -m unittest gtsam/tests/test_Pose3.py
- NOTE: You might need to re-run cmake ..
if files are deleted or added.
Do make install
and cd <gtsam_install_folder>/python
. Here, you can:
python setup.py test
gtsam
to your current Python environment.
python setup.py install
In the CMake, simply include the MatlabWrap.cmake
file.
include(MatlabWrap)
This cmake file defines functions for generating MATLAB wrappers.
wrap_and_install_library(interfaceHeader linkLibraries extraIncludeDirs extraMexFlags)
Generates wrap code and compiles the wrapper.Usage example:
`wrap_and_install_library("lba.h" "" "" "")`
Arguments:
interfaceHeader
: The relative or absolute path to the wrapper interface definition file.linkLibraries
: Any additional libraries to link. Your project library
(e.g. lba
), libraries it depends on, and any necessary
MATLAB libraries will be linked automatically. So normally,
leave this empty.extraIncludeDirs
: Any additional include paths required by dependent
libraries that have not already been added by
include_directories. Again, normally, leave this empty.extraMexFlags
: Any additional flags to pass to the compiler when building
the wrap code. Normally, leave this empty.*WARNING*: Running the ./update_wrap.sh script from the GTSAM repo creates 2 new commits in GTSAM. Be sure to NOT push these directly to master/develop. Preferably, open up a new PR with these updates (see below).
The wrap library is included in GTSAM as a git subtree. This means that sometimes the wrap library can have new features or changes that are not yet reflected in GTSAM. There are two options to get the most up-to-date versions of wrap:
wrap
rather than the one pre-packaged with GTSAM../update_wrap.sh
from the root of GTSAM's repository to pull in the newest version of wrap to your local GTSAM installation. See the warning above about this script automatically creating commits.To make a PR on GTSAM with the most recent wrap updates, create a new branch/fork then pull in the most recent wrap changes using ./update_wrap.sh
. You should find that two new commits have been made: a squash and a merge from master. You can push these (to the non-develop branch) and open a PR.
For any code contributions to the wrap project, please make them on the wrap repository.