(C) 2014 Jochen Schlobohm, [email protected]

========== SWig Inline C ==========

This project makes it possible to use c-code inside python code. It is inspired by weave but tries to keep it as simple as possible.

The code relies on swig and distutils to generate and compile the temporary module. Both have to be installed on your computer.

One example can be seen in swice.py when running as main-program.

Another Example:

import swice
code = '''

	for(int i = 0; i< Na(0); i++)
		a(i) = a(i)*5;
    
'''
a = np.array([1,2,3,4])
swice.inline(code, ['a'], locals(), globals())

print(a) # will print [5,10,15,20]

return_val is an integer wich will be returned when invoking inline()

In order to find out if the given code has already been used (and therefore compiled) a hash is calculated from the generated interface and code. If the temporary library is already present, it will just be loaded and not recompiled. Recompiling can be forced by setting the flag recompile. If you change additional libraries, headers or source files without altering the code or interface, the hash algorithm will not notice. Setting recompile to True will help. If the extracode is changed, the hash will change because it is simply copied into the main C-file.

By the way: Integer Arrays are mapped to stdint.h -types (e.g. uint8_t instead of unsigned char)

"__g" is a forbidden identifier for your c-code and variable names.

C variable access: Each numpy array will be accessible by a (dim1, [dim2, [...]]). The size of each dimension is stored in N[]. Scalars are copied into and from the module.

Necessary programs and libraries:

numpy swig python-dev (e.g. for ubuntu)

TODOs

• additional sorce files (not tested yet)
• includes (not tested yet)
• src-dir
• libs & libdirs (not tested yet)