The StarCraft II API provides access to in-game state observation and unit control. The API is a wrapper around protobuf defined protocol over a websocket connection.
While it's possible to write directly to the protocol, this library provides a C++ and class-based abstraction. You can see a simple example below.
#include <sc2api/sc2_api.h>
#include <iostream>
using namespace sc2;
class Bot : public Agent {
public:
virtual void OnGameStart() final {
std::cout << "Hello, World!" << std::endl;
}
virtual void OnStep() final {
std::cout << Observation()->GetGameLoop() << std::endl;
}
};
int main(int argc, char* argv[]) {
Coordinator coordinator;
coordinator.LoadSettings(argc, argv);
Bot bot;
coordinator.SetParticipants({
CreateParticipant(Race::Terran, &bot),
CreateComputer(Race::Zerg)
});
coordinator.LaunchStarcraft();
coordinator.StartGame(sc2::kMapBelShirVestigeLE);
while (coordinator.Update()) {
}
return 0;
}
You can find a detailed tutorial on what this code does in docs/tutorial1.md.
This library uses CMake to generate project files. It builds with Visual Studio on Windows and Makefiles on the other platforms. It relies on the following contrib packages.
- Civetweb
- Protobuf
- (optional) SDL
Follow the instructions for submodules and building in docs/building.md.
We do our best to conform to the Google C++ Style Guide with the exception that we use four space tabs instead of two space tabs.
You can find documentation for the api in docs/html, open index.html to view it in your browser. We generate our documentation using Doxygen
To generate the documentation yourself.
- Download and install doxygen
- From the root of the project run: doxygen Doxyfile
- Commit and push the generated documentation files in a seperate commit that contains documentation changes only, no code changes.
This repository only comes with a few maps for testing.
Additional maps and replays can be found here.
If you are using Visual Studio 2017 and just want precompiled library files you can download the following package:
Additional community built AI libraries can be found here.
If you are new to programming, the CommandCenter framework may be a good starting point.