1-DAV-202 Data Management 2023/24
Previously 2-INF-185 Data Source Integration
Difference between revisions of "Lcpp"
Jump to navigation
Jump to search
(Created page with "<!-- NOTEX --> HWweb <!-- /NOTEX -->") |
|||
(4 intermediate revisions by 2 users not shown) | |||
Line 1: | Line 1: | ||
<!-- NOTEX --> | <!-- NOTEX --> | ||
− | [[ | + | [[HWcpp]] |
<!-- /NOTEX --> | <!-- /NOTEX --> | ||
+ | |||
+ | Python is a very flexible language with many libraries, but programs in Python can be slow and take a lot of memory compared with programs written for example in C++. In this lecture we will see how to combine C++ and Python so that you write core algorithmic parts in C++ but use Python to do other less time-sensitive computations, such as parsing inputs. We will be using [https://pybind11.readthedocs.io/en/stable/basics.html#creating-bindings-for-a-simple-function pybind11 library] to do that. | ||
+ | |||
+ | Here is an example C++ code, which can be converted as an extension and used in Python: | ||
+ | |||
+ | <syntaxhighlight lang="C++"> | ||
+ | #include <pybind11/pybind11.h> | ||
+ | #include <pybind11/stl.h> | ||
+ | #include <vector> | ||
+ | #include <string> | ||
+ | |||
+ | using std::vector; | ||
+ | using std::string; | ||
+ | |||
+ | |||
+ | // Here is your class code | ||
+ | struct MyClass { | ||
+ | MyClass() {} | ||
+ | int f(vector<string> arr) { | ||
+ | int ss = 0; | ||
+ | for (auto &s: arr) { | ||
+ | ss += s.size(); | ||
+ | } | ||
+ | return ss; | ||
+ | } | ||
+ | }; | ||
+ | |||
+ | |||
+ | // This is the binding code, every function you want to export, must be declared here | ||
+ | PYBIND11_MODULE(example, m) { | ||
+ | m.doc() = "pybind11 example plugin"; // optional module docstring | ||
+ | |||
+ | pybind11::class_<MyClass>(m, "MyClass") | ||
+ | .def(pybind11::init()) | ||
+ | .def("f", &MyClass::f); | ||
+ | } | ||
+ | </syntaxhighlight> | ||
+ | |||
+ | We can then compile this code to be used as Python module: | ||
+ | <syntaxhighlight lang="bash"> | ||
+ | c++ -O3 -Wall -shared -std=c++11 -fPIC $(python3 -m pybind11 --includes) example.cc -o example$(python3-config --extension-suffix) | ||
+ | </syntaxhighlight> | ||
+ | |||
+ | And then we can import it in Python (we need to be in same directory as compiled output): | ||
+ | |||
+ | <syntaxhighlight lang="Python"> | ||
+ | import example | ||
+ | |||
+ | c = example.MyClass() | ||
+ | print(c.f(["aaa", "bb"])) | ||
+ | print(c.f([])) | ||
+ | </syntaxhighlight> |
Latest revision as of 13:17, 30 April 2024
Python is a very flexible language with many libraries, but programs in Python can be slow and take a lot of memory compared with programs written for example in C++. In this lecture we will see how to combine C++ and Python so that you write core algorithmic parts in C++ but use Python to do other less time-sensitive computations, such as parsing inputs. We will be using pybind11 library to do that.
Here is an example C++ code, which can be converted as an extension and used in Python:
#include <pybind11/pybind11.h>
#include <pybind11/stl.h>
#include <vector>
#include <string>
using std::vector;
using std::string;
// Here is your class code
struct MyClass {
MyClass() {}
int f(vector<string> arr) {
int ss = 0;
for (auto &s: arr) {
ss += s.size();
}
return ss;
}
};
// This is the binding code, every function you want to export, must be declared here
PYBIND11_MODULE(example, m) {
m.doc() = "pybind11 example plugin"; // optional module docstring
pybind11::class_<MyClass>(m, "MyClass")
.def(pybind11::init())
.def("f", &MyClass::f);
}
We can then compile this code to be used as Python module:
c++ -O3 -Wall -shared -std=c++11 -fPIC $(python3 -m pybind11 --includes) example.cc -o example$(python3-config --extension-suffix)
And then we can import it in Python (we need to be in same directory as compiled output):
import example
c = example.MyClass()
print(c.f(["aaa", "bb"]))
print(c.f([]))