Convert string to mathematical function

[Colin J. Williams]

Nick Vatamaniuc wrote:
> Jeremy,
> 
> Which method of extension are you using? For example you can pass the
> function to C++/C directly using weave, without the need to convert to
> Python first. Python has a different syntax than C++. Also I assume you
> want exponentiation in you example "x^2..." and not 'xor', well, in
> Python that would be "x**2" and in C you would have to include math.h
> and then do pow(x,2). In other words there is no easy way to map Python
> expressions to C++ (if it would be, Python could just all be mapped to
> C++ and run at C++ speed!).
> 
> Below is my attempt at using weave (I prefer pyrex myself...). Note:
> this won't work too fast if your function changes on-the-fly because
> each new version will need to be compiled (gcc will be called and so
> on...). So it would only pay off if your function gets compiled ones
> and then takes a __very__ long time to compute.
> 
> Anyway here is my example:
> --------------------------------------------------------------------------------------------
>>>> import weave
>>>> includes="""
> .... #include <math.h>
> .... """
>>>> c_code="""
> .... return_val=pow(a,N);
> .... """
>>>> a=42
>>>> N=42
>>>> ans=weave.inline(c_code, ['a','N'], support_code=includes)
> file changed
> None
> cc1plus: warning: command line option "-Wstrict-prototypes" is valid
> for Ada/C/ObjC but not for C++
>>>> ans
> 1.5013093754529656e+68
>>>> a**N
> 150130937545296572356771972164254457814047970568738777235893533016064L
>>>> #now call the function again. no compilation this time, the result was cached!
>>>> ans=weave.inline(c_code, ['a','N'], support_code=includes)
>>>> ans
> 1.5013093754529656e+68
> ----------------------------------------------------------------------------------
> Look up weave for Python (it is a part of scipy) for more examples...
> 
> Hope this helps,
> Nick Vatamaniuc
> 
> 
> jeremito wrote:
>> I am extending python with C++ and need some help.  I would like to
>> convert a string to a mathematical function and then make this a C++
>> function.  My C++ code would then refer to this function to calculate
>> what it needs.  For example I want to tell my function to calculate
>> "x^2 + 3x +2", but later on I may want: "x + 3".  Does anybody know how
>> I can get an arbitrary string in Python (but proper mathematical
>> function) turned into a C++ function?  My one idea (although I don't
>> know how to implement it, I'm still new at this) is to pass to C++ a
>> pointer to a (Python) function.  Will this work?
>> Thanks,
>> Jeremy
> 
Jeremy,

I presume that you have considered and rejected use of Python's eval 
function.

To this nonC++ user, it seems the simpler way.

A someone else pointed out, you would need to import the math module.

Colin W.