Building the Standards of Fundamental Astronomy (SOFA) library for pysofa

2011 Jun 29 - Brian Kloppenborg

Standards of Fundamental Astronomy (SOFA) is a FORTRAN/C library from the IAU that provides a variety of useful functions for astronomy that deal with time and astrometric calculations. Unfortunately, the library ships with a very ugly license that prohibits its modification and/or redistribution. As such, it isn’t widely used, despite being packed with very useful functions that provide:

If you wish to use SOFA from Python, you are in luck: there is a Python interface on pypi however, you will need to build SOFA as a shared library and install it on your system in order to get it to work.

System Requirements

In order to proceed you will need several things:

On Ubuntu it’s quite easy to install all of these, just do

sudo apt-get install python python-all-dev build-essential cmake

Getting SOFA

Download and extract SOFA from the official download page

Creating a Shared Library

Here is the tricky part. You can either modify the existing makefile to produce a shared library, or you can use the CMake build script and instructions I provide below. The CMake file will automatically build the library and run SWIG to create a Python interface.

After extracting SOFA, cd into the main directory and create a CMakeLists.txt file with the following content:

cmake_minimum_required(VERSION 2.6)

project(sofa_c C)
  
# Set a few variables:
set(LIBS ${LIBS} m)

# Extract all of the source files.
file(GLOB_RECURSE C_SOURCE . src/*.c)

  
# Build a shared library
add_library(sofa_c SHARED ${C_SOURCE})

  
# Now define the installation options
install(TARGETS sofa_c LIBRARY DESTINATION lib)

Now you can easily build and install SOFA as follows:

cmake .
make
make install

Installing pysofa

Here you need to install pysofa. It has three requirements (1) Python 2.5 or higher, (2) numpy, and (3) a shared library version of sofa (which we just installed). You can install all of this on the python terminal by

sudo aptitude install numpy
easy_install pysofa

And that’s it! If it all worked you can fire up the python interpreter and import pysofa and you’ll have access to all of the SOFA library functions.

Tags:
Categories: