Did you know you can write functions in C and then call them directly from Python? Isn't that cool? Let's skip all the background and the "why woul...
For further actions, you may consider blocking this person and/or reporting abuse
Great example! Do you know if there is any tool to help you with using it in a package? So that if you install the package (using setup.py) it automatically compiles the C code and links paths to shared objects. Maybe even writes the wrapper function in Python. Something like Rcpp for R.
Hi Jan! Off the top of my head, I don't know of anything, but have a look at Fernando B (@kodaman2 ) comment a couple comments down. He mentions SIP that can turn C or C++ bindings. If you can't find the comment, this is the link he provided: pypi.org/project/SIP/
I dunno if sip is a fully automated solution, have a look at this project I believe you have to manually write the sip file (wrapper), but I could be wrong. I've never actually successfully compiled with sip.
github.com/dimv36/QCustomPlot-PyQt5
There is quite a few ways to interface Python with C/C++. I hadn't heard of SIP before, but I think SWIG and Shiboken should be fairly similar in creating bindings automagically.
Personally I'm a big fan of Cython and pybind11. Cython was created to be able to transpile a python-like syntax to C to then compile it, but it can also be used to wrap C libraries or to interface with C code.
Pybind11 is similar to BoostPython as in that is was meant to expose C functionality to Python similar to what is being done in this article, but it also helps with converting types such as lists and tuples and makes it easier to manipulate the GIL.
If you are interested in this topic I have a talk on embedding Python into C++. Pretty much the exact opposite of this :P
This is a really neat article, I didn't know it was this easy! Because I'm not familiar with the
ctypesmodule, andfrom ctypes import *makes it a little tough to see which functionality is coming fromctypes, is it just theCDLLclass that you're using from there?Yes! and thank you for pointing out the lack of clarity. the
CDLLis the only thing in the Python script coming from the ctypes moduleHi!, great example. It was very useful. I have one question... I don't know if is only me but when my number is 13 (or bigger) the code from C doesn't work. Anyone has the same problem?
it's probably because 13! is 6.2 billion and it's possible your compiler only sets aside enough memory for 4.2 billion for
long. trylong longand see if you get the same problemHi Erik, thanks for the reply. I did your recommendation but it doesn't work. I was thinking that the problem is with gcc but I am not sure. Do you have any other idea?
I tried it out too. It's definitely weird. It's "working" for me but it's not giving me the right answer. I never tested that high. I'm at work now and can't really dig into it but if I get some time I'll let you know.
Sure, thanks again!
Not sure if this thread will be active again, but I found that on the c side of the program (using long long) it is accurate up to 20! . I found this by adding a printf statement in cfactorial.c . So it seems that at some point in the process of python and c communicating, the true value is lost.
interesting. perhaps
ints in C are represented differently in memory than in Python. maybe the best way is to parse theintinto achararray and send it to Python as a string?That does sound promising because I imagine it could scale almost indefinitely if its not bound by the long or long long limits
noob question: Is it going to have the same performance as the performance of c programming language code.
the work that the C function is doing will be equally as fast. running The C function though the Python script will add a little latency because the Python script is doing a couple things before calling the function, but the difference in time should be negligible
One more question, is it going to be the same for c++ code?
I am not sure if this will work exactly the same with C++. the first difference that comes to mind is using
g++instead ofccthe ctypes library seems to be optimized for C specifically but I bet you can use c++ as well. I'm not in a place where I can test this out right now, but play around with it and let me know what you learn
Thank you so much. I will try it out and would share the results. Thanks once again, Mr. Erik.
Hi Erik,
Thanks for this nice explanation.
Can you also please share the example wherein the C program function will return char* and how to reinterpret this data in python code.
Thanks in advance.
Great read, have you used sip? pypi.org/project/SIP/ it can turn full c or c++ libs to python bindings. I've never had the need to use it, but looks promising.
interesting, I've not heard of this before. will check it out!
This is really helpful for times when you can't find python libraries to do what you need or if you need C because it runs faster. Thanks for sharing!