Preface to translation

Some netizens are not sure why D2JSP Can execute JavaScript The script program was surprised , So I translated this article , In the original here . This tutorial teaches you how to use SpiderMonkey Create one that can perform JavaScript The script C++ Program , And let JavaScript Scripts manipulate your C++ The internal data of the program 、 operation . You can see from this tutorial that SpiderMonkey With the help of the engine , Give Way C++ The program supports JavaScript Scripting is a very easy What happened , What's better is SpiderMonkey Can also be in Macintosh and Unix Platform use .
SpiderMonkey yes Gecko（Firefox Browser kernel ） Of JavaScript The script engine , Please refer to here .

Here is the translation .

－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－

The purpose of this tutorial is to teach you how to use JavaScript As a scripting language, make your C++ The program itself takes the initiative .

SpiderMonkey

SpiderMonkey yes Mozilla Part of the project , use C language , Is responsible for running JavaScript Script engine . The other one is called Rhino Of Java engine .

SpiderMonkey The latest version number of can be found in here download . It's published as source code , So you have to compile it yourself （ Translation notes ： In fact, there are many compiled binary version numbers on the Internet ,google once js32.dll You can find ）.Visual C++ Users can be in src Found under folder Workspace project project File to compile , The result of the compilation is a result called 'js32.dll' Of dll file .

SpiderMonkey Can also be in Macintosh and Unix Upper use , To learn how to compile on these platforms, read Readme.html.

stay C++ Run in JavaScript Program

step 1－ establish JavaScript runtime（ Execution time instance ）

Initialize a JavaScript runtime You can use JS_NewRuntime Method , The method will be runtime Allocate memory , You have to specify a number of bytes at the same time , When the memory allocation exceeds this number, the garbage collector will execute on its own initiative .

JSRuntime *rt = JS_NewRuntime(1000000L);
if ( rt == NULL )
...{
    // Do some error reporting
}

step 2－ establish context（ Context ）

Context Indicates the stack size required for script execution , The amount of private memory allocated to the script execution stack . Each script has its own context Related to .

When one context Is being used by a script or thread , Other scripts or threads cannot use this context. Just at the end of the script or thread , The context Can be reused by the next script or thread .

Create a new context You can use JS_NewContext Method .context It has to be associated with a runtime, call JS_NewContext Method must also specify the stack size .

JSContext *cx = JS_NewContext(m_rt, 8192);
if ( cx == NULL )
...{
    // Do some error reporting
}

step 3－ Initialize global objects

Before a script starts executing , It's necessary to initialize some common ones that most scripts will use JavaScript Functions and built-in （build-in） Class object .

Global objects are in a JSClass The structure describes the narrative . The structure can be initialized as follows ：

JSClass globalClass =
...{
    "Global", 0,
    JS_PropertyStub,  JS_PropertyStub,
    JS_PropertyStub, JS_PropertyStub,
    JS_EnumerateStub, JS_ResolveStub,
    JS_ConvertStub,  JS_FinalizeStub
};

 Now create and initialize this global object ：


JSObject *globalObj = JS_NewObject(cx, &globalClass, 0, 0);


JS_InitStandardClasses(cx, globalObj);

step 4－ Run script

One way to run a script is to use JS_EvaluateScript Method ：

std::string script = "var today = Date(); today.toString();"
jsval rval;
uintN lineno = 0;
JSBool ok = JS_EvaluateScript(cx, globalObj, script.c_str(), 
                              script.length(), "script", lineno, &rval);

In this script , If it works correctly, the data of the day will be saved in rval in .rval Include the result of the last run function .JS_EvaluteScript return JS_TRUE Represents successful operation , return JS_FALSE It means there is a mistake .

from rval To get the corresponding string value, you can use the following methods . I don't want to explain all the details here , For more specific information, please check by yourself API file .

JSString *str = JS_ValueToString(cx, rval);
std::cout << JS_GetStringBytes(str);

step 5－ Clean up the script engine

 You have to clean up the script engine before the end of the program ：


JS_DestroyContext(cx);


JS_DestroyRuntime(rt);

stay C++ Define one in JavaScript Classes used in

The class definitions used in this example are as follows ：

class Customer
...{
public:
    int GetAge() ...{ return m_age; }
    void SetAge(int newAge) ...{ m_age = newAge; }
    std::string GetName() ...{ return m_name; }
    void SetName(std::string newName) ...{ m_name = newName; }

private:
    int m_age;
    std::string m_name;
};

step 1－JavaScript class

from Customer Class to derive one you want in JavaScript The new ones used in C++ class , Or create one that includes Customer A new class of type member variables .

to JavaScript There has to be a class to use JSClass structure , To do that, you have to create a JSClass Static member variable of type , This variable will be used by other classes , So we have to declare it as public Variable . Other classes can use this structure to infer the type of objects （ see JS_InstanceOf API）.

// JSCustomer.h
class JSCustomer
...{
public:
    JSCustomer() : m_pCustomer(NULL) 
    ...{
    }

    ~JSCustomer()
    ...{
        delete m_pCustomer;
        m_pCustomer = NULL;
    }

    static JSClass customerClass;

protected:
    void setCustomer(Customer *customer) 
    ...{
        m_pCustomer = customer; 
    }

    Customer* getCustomer()
    ...{
        return m_pCustomer; 
    }

private:
    Customer *m_pCustomer;

};

The JSClass The structure includes JavaScript The name of the class 、 Flag bits and the name of the callback function for the script engine . For example , The script engine uses a callback function to get a property value from a class .

stay C++ Class is defined in the implementation file JSClass The structure is as follows ：

// JSCustomer.cpp
JSClass JSCustomer::customerClass = 
...{
    "Customer", JSCLASS_HAS_PRIVATE,
        JS_PropertyStub, JS_PropertyStub,
        JSCustomer::JSGetProperty, JSCustomer::JSSetProperty,
        JS_EnumerateStub, JS_ResolveStub, 
        JS_ConvertStub, JSCustomer::JSDestructor
};

The callback function used is JSCustomer::JSGetProperty,JSCustomer::JSSetProperty and JSCustomer::JSDestructor. Script engine calls JSGetProperty Get attribute value , call JSSetProperty Setting property values , call JSDestructor destructor JavaScript object .

JSCLASS_HAS_PRIVATE The flag bit causes the script engine to allocate some memory , So that you can JavaScript Add some self-defined data to the object , For example, you can use it to hold class pointers .

Callback function to C++ Class static member function mode exists ：

static JSBool JSGetProperty(JSContext *cx, JSObject *obj, jsval id, jsval *vp);
static JSBool JSSetProperty(JSContext *cx, JSObject *obj, jsval id, jsval *vp);
static JSBool JSConstructor(JSContext *cx, JSObject *obj, uintN argc, 
                            jsval *argv, jsval *rval);
static void JSDestructor(JSContext *cx, JSObject *obj);

step 2－ Initialize your JavaScript object

Create another one called JSInit Static method of , See the following example , This method will be created in the application JavaScript runtime When called .

static JSObject *JSInit(JSContext *cx, JSObject *obj, JSObject *proto);

JSInit The implementation of the method is about as follows ：

JSObject *JSCustomer::JSInit(JSContext *cx, JSObject *obj, JSObject *proto)
...{
    JSObject *newObj = JS_InitClass(cx, obj, proto, &customerClass,
        JSCustomer::JSConstructor, 0,
        JSCustomer::customer_properties, JSCustomer::customer_methods,
        NULL, NULL);
    return newObj;
}

Objects are represented in scripts （ Translation notes ：instantiated, In short, it's the object new When I came out ） When , Static methods JSConstructor Will be called . In this way... Can be used JS_SetPrivate API Attach some self-defined data to the object .

JSBool JSCustomer::JSConstructor(JSContext *cx, JSObject *obj, uintN argc, 
                                 jsval *argv, jsval *rval)
...{
    JSCustomer *p = new JSCustomer();

    p->setCustomer(new Customer());
    if ( ! JS_SetPrivate(cx, obj, p) )
        return JS_FALSE;
    *rval = OBJECT_TO_JSVAL(obj);
    return JS_TRUE;
}

JSConstructor The construction method can take multiple parameters , Used to initialize classes . So far, a pointer has been created on the heap , There has to be a way to destroy it , This can be done through JS_Destructor complete ：

void JSCustomer::JSDestructor(JSContext *cx, JSObject *obj)
...{
    JSCustomer *p = JS_GetPrivate(cx, obj);
    delete p;
    p = NULL;
}

step 3－ Add attributes

Add a JSPropertySpec Type to store attribute information , Define attributes at the same time ID Enumeration variables for .

static JSPropertySpec customer_properties[];
enum
...{
    name_prop,
    age_prop
};

In the implementation file, for example, initialize the array ：

JSPropertySpec JSCustomer::customer_properties[] = 
...{ 
    ...{ "name", name_prop, JSPROP_ENUMERATE },
    ...{ "age", age_prop, JSPROP_ENUMERATE },
    ...{ 0 }
};

The last element of the array must be empty , Each of these elements is an element with 3 Array of elements . The first element is for JavaScript Name used . The second element is the only element of the attribute ID, Will be passed to the callback function . The third element is the flag bit ,JSPROP_ENUMERATE Represents the script in enumerating Customer You can see all the properties of the object , You can also specify JSPROP_READONLY To indicate that the attribute does not agree to be changed by the script .

Now it's possible to implement getting and setting Call back function ：

JSBool JSCustomer::JSGetProperty(JSContext *cx, JSObject *obj, jsval id, jsval *vp)
...{
    if (JSVAL_IS_INT(id)) 
    ...{
        Customer *priv = (Customer *) JS_GetPrivate(cx, obj);
        switch(JSVAL_TO_INT(id))
        ...{
        case name_prop:

            break;
        case age_prop:
            *vp = INT_TO_JSVAL(priv->getCustomer()->GetAge());
            break;
        }
    }
    return JS_TRUE;
}


JSBool JSCustomer::JSSetProperty(JSContext *cx, JSObject *obj, jsval id, jsval *vp)
...{
    if (JSVAL_IS_INT(id)) 
    ...{
        Customer *priv = (Customer *) JS_GetPrivate(cx, obj);
        switch(JSVAL_TO_INT(id))
        ...{
        case name_prop:
            break;
        case age_prop:
            priv->getCustomer()->SetAge(JSVAL_TO_INT(*vp));
            break;
        }
    }
    return JS_TRUE;
}

It is recommended to return... In the callback function of the property JS_TRUE. Suppose we return to JS_FALSE, When the attribute is not found in the object （ The script engine ） No search .

step 4－ Add method

Create a JSFunctionSpec An array of static members of type ：

static JSFunctionSpec customer_methods[];

In the implementation file, for example, initialize the array ：

JSFunctionSpec wxJSFrame::wxFrame_methods[] = 
...{
    ...{ "computeReduction", computeReduction, 1, 0, 0 },
    ...{ 0 }
};

The last element must be empty , Each of these elements is an element with 5 Array of elements . The first element is the method name for the script . The second is the name of a global or static member function . The third is the number of parameters of the method . The last two can be ignored .

Create a static method in the class ：

static JSBool computeReduction(JSContext *cx, JSObject *obj, uintN argc, 
                               jsval *argv, jsval *rval);

This function returns the JS_TRUE, Otherwise return to JS_FALSE. Pay attention to the real JavaScript The return value of the method is stored in rval In the parameter .

An implementation example of this method ：

JSBool JSCustomer::computeReduction(JSContext *cx, JSObject *obj, uintN argc, 
                                    jsval *argv, jsval *rval)
...{
    JSCustomer *p = JS_GetPrivate(cx, obj);
    if ( p->getCustomer()->GetAge() < 25 )
        *rval = INT_TO_JSVAL(10);
    else
        *rval = INT_TO_JSVAL(5);
    return JS_TRUE;
}

Use samples

The following script uses the object created earlier ：

var c = new Customer();
c.name = "Franky";
c.age = 32;
var reduction = c.computeReduction();

Don't forget to create context Initializes JavaScript object ：

JSObject *obj = JSCustomer::JSInit(cx, global);

Class constant

JavaScript type

This chapter explains in JavaScript There are several types that will be used in ：Integer,String,Boolean,Double,Object and Function.

In the building ......

Garbage collection

In the building ......

download

main.cpp Demonstrate how to run a javascript Program .JSCustomer.h demonstration Customer Of JavaScript The definition of a class .JSCustomer.cpp demonstration JSCustomer The implementation of the .Customer.h yes Customer C++ The definition of a class .example.js Sample scripts .