我正在学习emscripten并试图更好地理解它。据我所知,用例主要用于将现有的C / C ++代码移植到Web客户端(浏览器)并从JavaScript调用C / C ++代码。
但我想知道是否可以将C ++和Emscripten用于网页(注意:这更多是出于好奇 - 我知道目前没有很多理由这样做)。我设法从C ++调用Javascript函数,并将类型为string,int,double等的参数传递给它们。但我缺少的是:从C ++调用Javascript函数并将C或C ++函数作为句柄传递。所以作为一个简单的例子:我如何编写以下Javascript代码,纯粹的C ++?
var myfun = function() { /* do something meaningful here */ }
document.onload(myfun);
答案 0 :(得分:3)
<强> TL; DR; 强>
我写了一个库:js-bind,它接受任意数量的参数来轻松完成:
using namespace std::placeholders;
using emscripten::val;
// First find the HTML object to attach the event to
auto clickme_btn = val::global("document").call<val>("getElementById", string("clickme_btn"));
// Bind the event handler for click
auto onclick = [](val event){ cout << "hello world ! " << endl; };
clickme_btn.set("onclick", js::bind(onclick, _1));
这个库是基于以下解释的一些宏元编程。
详细解答:
你有不同的可能性,比如emscripten ccall,但在我看来更容易使用的是Embind。
例如,从C ++中获取XMLHttpRequest的绑定事件处理程序。
要启用它,您必须使用:--bind -s NO_EXIT_RUNTIME=1
可以通过独立功能和单身功能轻松实现,如下所示:
#include <iostream>
#include <emscripten.h>
#include <emscripten/bind.h>
#include <emscripten/val.h>
namespace xhr {
inline emscripten::val& singleton() {
using emscripten::val;
static val instance = val::global("XMLHttpRequest").new_();
return instance;
}
void on_load(emscripten::val event) {
std::cout << "Successful Query " << std::endl;
std::cout << "response is : " << singleton()["responseText"].as<std::string>() << std::endl;
}
void on_error(emscripten::val event) {
std::cout << "Error on query " << std::endl;
}
void on_progress(emscripten::val event) {
std::cout << "Progress on query " << std::endl;
std::cout << event["lengthComputable"].as<bool>() << ": " << event["loaded"].as<unsigned int>() / event["total"].as<unsigned int>() << std::endl;
}
using namespace emscripten;
EMSCRIPTEN_BINDINGS(xhr) {
function("on_load", &on_load);
function("on_error", &on_error);
function("on_progress", &on_progress);
}
}
int main(int argc, char** argv) {
using emscripten::val;
xhr::singleton().call<val>("open", std::string("GET"), std::string("http://127.0.0.1:8080/CMakeCache.txt"), true);
// Here I set the callback to &on_load function registered via the EMSCRIPTEN_BINDINGS macro.
xhr::singleton().set("onload",val::module_property("on_load"));
xhr::singleton().set("onprogress",val::module_property("on_progress"));
xhr::singleton().set("onerror",val::module_property("on_error"));
xhr::singleton().call<val>("send");
return 0;
}
通常在C ++中,我们习惯于std :: bind回调。这也可以通过更简洁的方式以xhr为例来实现:
#include <iostream>
#include <functional>
#include <memory>
#include <emscripten.h>
#include <emscripten/bind.h>
#include <emscripten/val.h>
class MiniXhr : public std::enable_shared_from_this<MiniXhr> {
using val = emscripten::val;
using url_t = std::string;
public:
void set_url(const url_t& url) { url_ = url; }
void GET();
/**
*
* The member function to be called from javascript.
*/
void on_readystate(val event) {
std::cout << "ready " << std::endl;
std::cout << "xxhr::on_readystate: "
<< xhr["readyState"].as<size_t>() << " - " << url_ << " :: "
<< xhr["status"].as<size_t>() << ": "
<< xhr["statusText"].as<std::string>() << std::endl;
}
private:
url_t url_;
val xhr = val::global("XMLHttpRequest").new_();
};
using emscripten::class_;
EMSCRIPTEN_BINDINGS(MiniXhr) {
/**
* Binding for the class.
*/
class_<MiniXhr>("MiniXhr")
.smart_ptr<std::shared_ptr<MiniXhr>>("shared_ptr<MiniXhr>")
.function("on_readystate", &MiniXhr::on_readystate)
;
/**
* More generic binding to bind a functor with one argument (event handler get the event)
* Here std::function call operator from C++ is bound to function opcall() in JS.
*/
class_<std::function<void(emscripten::val)>>("VoidValFunctor")
.constructor<>()
.function("opcall", &std::function<void(emscripten::val)>::operator());
}
/**
*
* Finally the interesting part : binding the member function on_readystate to the readystatechange event of XMLHttpRequest.
*
*/
void MiniXhr::GET() {
/**
* Here this lambda could be put as function in a library, to do an JS(std::bind),
* it should just be overloaded for different argument count. (Im on it).
*/
auto jsbind = [](val& target, const char* property, auto bind_expression ) {
// Create an std::function from the bind expression
std::function<void(emscripten::val)> functor = bind_expression;
// We ensure the correct object will always be bound to the this of the function
auto functor_adapter = val(functor)["opcall"].call<val>("bind", val(functor));
// Finally we simply set the eventhandler
target.set(property, functor_adapter);
};
// Here we could bind as many member function as we want.
// jsbind(xhr, "onload", std::bind(&MiniXhr::on_load, shared_from_this(), std::placeholders::_1));
// jsbind(xhr, "onerror", std::bind(&MiniXhr::on_error, shared_from_this(), std::placeholders::_1));
// jsbind(xhr, "onprogress", std::bind(&MiniXhr::on_progress, shared_from_this(), std::placeholders::_1));
jsbind(xhr, "onreadystatechange", std::bind(&MiniXhr::on_readystate, shared_from_this(), std::placeholders::_1));
// Note that we bind with shared_from_this(), as the scope where the class was instantiated may be dead
// and only later our callback will come back.
xhr.call<val>("open", std::string("GET"), url_, true);
xhr.call<val>("send");
}
int main(int argc, char** argv) {
auto x = std::make_shared<MiniXhr>();
x->set_url("notfound.json");
x->GET();
return 0;
}
答案 1 :(得分:1)
我不确定emscripten,但总而言之,我明白你需要知道的是如何将C ++函数作为另一个C ++函数的句柄传递。我希望我可以帮忙。
JavaScript,PHP和其他更灵活的语言,允许传递函数对象。在C和C ++中,它稍有不同,您必须将函数指针作为参数传递给其他函数。在C中,这个名称是Callback,而不是句柄。
例如:
/* This function takes a single callback as a parameter. */
//here we say that the parameter, that we name numberSource, is a function that receives no parameters itself (void), and return an int
void printNumber(int (*numberSource)(void)) {
printf("%d", numberSource());
}
/* A possible callback */
int oneExampleFunction(void) {
return 100;
}
/* Another possible callback. */
int otherExampleFunction(void) {
return 200;
}
/* This is how we would call printNumber with three different callbacks. */
//with "&" we are referencing the memory address of the function,
//since thats what printNumber is expecting
printNumber(&oneExampleFunction);
printNumber(&otherExampleFunction);
printNumber(&rand); //where are using rand(), a system function, that works as well.
通常的做法是为参数创建自定义类型,因此您不需要像int (*numberSource)(void)
那样使用丑陋的东西。它将类似于:
//Function pointer called CallbackType that takes a float
//and returns an int
typedef int (*NameYouWantForTheType)(void);
所以printNumber函数是这样的:
void printNumber(NameYouWantForTheType numberSource ) {
printf("%d", numberSource());
}
所以,在你的情况下, 如果你想翻译这个JS代码
var myfun = function() { /* do something meaningful here */ }
document.onload(myfun);
到C,你有一个名为“document”的C对象,它接收一个执行其他动作的函数,你的C代码将是:
void myfun (void) {
/* do something meaningful here */
}
document.onload(&myfun);
答案 2 :(得分:1)
这是我在C代码中修补Emscripten时使用了很久的东西:
void myfun(void(*f)(void)) { (*f)() }
然后这里是JavaScript:
var theparty = Runtime.addFunction(function() { print("Will there be confetti?") });
Module.ccall("myfun", "number", ["number"], [theparty]);
Runtime.removeFunction(theparty); // output => Will there be confetti?
我总是删除执行后不再需要的函数以保留内存。 这是使代码位协同工作的简单而无缝的方法。显然,您可以将其修改为除了打印信息之外的任何其他操作。 :P