享元模式和 C++模板

发布于 2024-09-18 08:05:28 字数 3447 浏览 2 评论 0原文

我有蝇量级模式。我有抽象类 Glyph。我有类 Letter 和从 Glyph 派生的抽象代码。我有从 Code 派生的 YusciiCode、UniCyrCode 和 UniLatCode。

我的享元工厂可以这样完成:

    template <class T>
 class CodeFactory : public AbstractCodeFactory
 {
 public:
  CodeFactory();
  virtual ~CodeFactory();
  virtual Glyph* GetFlyweight(unsigned int code);
  virtual Glyph* GetFlyweight(string letter);
 private:
  // pool of flyweights (codes or letters)
  map <unsigned int, Glyph*> my_code_map;
  map <string, Glyph*> my_letter_map;
 };

可以这样完成:

template <class key, class T>
 class CodeFactory : public AbstractCodeFactory
 {
 public:
  CodeFactory();
  virtual ~CodeFactory();
  virtual Glyph* GetFlyweight(key code);
 private:
  // pool of flyweights (codes or letters)
  map <key, Glyph*> my_code_map;
 };

在第一个示例中,GCC 链接器告诉我没有 Letter(unsigned int) 和 xxxCode(string) 构造函数。事实上,没有任何构造函数,GCC 是对的,但是有没有比定义这些构造函数更好的方法呢?

行上有一个错误

map <key, Glyph*>::iterator it;

在第二个例子中,GCC 编译器告诉我函数 GetFlyweight 的

。这种享元模式的实现方式是什么?

我需要使用它。 这是我当前的实现:

class AbstractCodeFactory
{
public:
    AbstractCodeFactory();
    virtual ~AbstractCodeFactory();
    virtual Glyph* GetFlyweight(unsigned int code) = 0;
    virtual Glyph* GetFlyweight(string letter) = 0;
};


template <class T>
    class CodeFactory : public AbstractCodeFactory
    {
    public:
        CodeFactory();
        virtual ~CodeFactory();
        virtual Glyph* GetFlyweight(unsigned int code);
        virtual Glyph* GetFlyweight(string letter);
    private:
        // pool of flyweights (codes or letters)
        map <unsigned int, Glyph*> my_code_map;
        map <string, Glyph*> my_letter_map;
    };


template <class T>
    CodeFactory<T>::CodeFactory()
    {
        // TODO Auto-generated constructor stub

    }

template <class T>
    CodeFactory<T>::~CodeFactory()
    {
        // TODO Auto-generated destructor stub
        map <unsigned int, Glyph*>::iterator it;
        map <string, Glyph*>::iterator l_it;
        for (it = my_code_map.begin(); it != my_code_map.end(); ++it)
        {
            delete it->second;
            it->second = NULL;
            my_code_map.erase(it);
        }
        for (l_it = my_letter_map.begin(); l_it != my_letter_map.end(); ++l_it)
        {
            delete l_it->second;
            l_it->second = NULL;
            my_letter_map.erase(l_it);
        }
    }

template <class T>
    Glyph* CodeFactory<T>::GetFlyweight(unsigned int code)
    {
        map <unsigned int, Glyph*>::iterator it;
        T *code_class = NULL;
        if ((it = my_code_map.find(code)) == my_code_map.end())
        {
            my_code_map.insert(pair <unsigned int, Glyph*> (code, code_class = new T(code)));
            return code_class;
        }
        else return it->second;
    }

template <class T>
    Glyph* CodeFactory<T>::GetFlyweight(string letter)
    {
        map <string, Glyph*>::iterator it;
        T *letter_class = NULL;
        if ((it = my_letter_map.find(letter)) == my_letter_map.end())
        {
            my_letter_map.insert(pair <string, Glyph*> (letter, letter_class = new T(letter)));
            return letter_class;
        }
        else return it->second;
    }

I have flyweight pattern. I have abstract class Glyph. I have class Letter and abstract Code derived from Glyph. I have YusciiCode, UniCyrCode and UniLatCode derived from Code.

My flyweight factory can be done like this:

    template <class T>
 class CodeFactory : public AbstractCodeFactory
 {
 public:
  CodeFactory();
  virtual ~CodeFactory();
  virtual Glyph* GetFlyweight(unsigned int code);
  virtual Glyph* GetFlyweight(string letter);
 private:
  // pool of flyweights (codes or letters)
  map <unsigned int, Glyph*> my_code_map;
  map <string, Glyph*> my_letter_map;
 };

It can be done like this:

template <class key, class T>
 class CodeFactory : public AbstractCodeFactory
 {
 public:
  CodeFactory();
  virtual ~CodeFactory();
  virtual Glyph* GetFlyweight(key code);
 private:
  // pool of flyweights (codes or letters)
  map <key, Glyph*> my_code_map;
 };

In the first example GCC linker tell me that there are no Letter(unsigned int) and xxxCode(string) constructor. In fact there aren't any and GCC is right, but is there a better way to do this than to define those constructors?

In the seccond ecample GCC compiler tells me that there is an error on the line

map <key, Glyph*>::iterator it;

of the function GetFlyweight.

What is the way to implement this flyweight pattern?

I need to use it.
Here is my current implementation:

class AbstractCodeFactory
{
public:
    AbstractCodeFactory();
    virtual ~AbstractCodeFactory();
    virtual Glyph* GetFlyweight(unsigned int code) = 0;
    virtual Glyph* GetFlyweight(string letter) = 0;
};


template <class T>
    class CodeFactory : public AbstractCodeFactory
    {
    public:
        CodeFactory();
        virtual ~CodeFactory();
        virtual Glyph* GetFlyweight(unsigned int code);
        virtual Glyph* GetFlyweight(string letter);
    private:
        // pool of flyweights (codes or letters)
        map <unsigned int, Glyph*> my_code_map;
        map <string, Glyph*> my_letter_map;
    };


template <class T>
    CodeFactory<T>::CodeFactory()
    {
        // TODO Auto-generated constructor stub

    }

template <class T>
    CodeFactory<T>::~CodeFactory()
    {
        // TODO Auto-generated destructor stub
        map <unsigned int, Glyph*>::iterator it;
        map <string, Glyph*>::iterator l_it;
        for (it = my_code_map.begin(); it != my_code_map.end(); ++it)
        {
            delete it->second;
            it->second = NULL;
            my_code_map.erase(it);
        }
        for (l_it = my_letter_map.begin(); l_it != my_letter_map.end(); ++l_it)
        {
            delete l_it->second;
            l_it->second = NULL;
            my_letter_map.erase(l_it);
        }
    }

template <class T>
    Glyph* CodeFactory<T>::GetFlyweight(unsigned int code)
    {
        map <unsigned int, Glyph*>::iterator it;
        T *code_class = NULL;
        if ((it = my_code_map.find(code)) == my_code_map.end())
        {
            my_code_map.insert(pair <unsigned int, Glyph*> (code, code_class = new T(code)));
            return code_class;
        }
        else return it->second;
    }

template <class T>
    Glyph* CodeFactory<T>::GetFlyweight(string letter)
    {
        map <string, Glyph*>::iterator it;
        T *letter_class = NULL;
        if ((it = my_letter_map.find(letter)) == my_letter_map.end())
        {
            my_letter_map.insert(pair <string, Glyph*> (letter, letter_class = new T(letter)));
            return letter_class;
        }
        else return it->second;
    }

如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。

扫码二维码加入Web技术交流群

发布评论

需要 登录 才能够评论, 你可以免费 注册 一个本站的账号。

评论(1

为你拒绝所有暧昧 2024-09-25 08:05:28

由于您的享元工厂只能生成 Letter、YusciiCode、UniCyrCode 或 UniLatCode 对象,因此我会选择第二个选项(指示键类型的第二个模板参数。

编译器在声明 map::iterator it; 的缺点是编译器无法确定 map::iterator 是否引用类型或其他内容。
这是因为它取决于模板参数 key,并且您可能在成员 iterator 为 < em>不是类型。
为了帮助编译器,您必须指定您希望 map::iterator 引用类型名:

typename map<key, Glyph*>::iterator it;

As your flyweight factory can only produce either Letter, YusciiCode, UniCyrCode or UniLatCode objects, I would go with the second option (a second template parameter indicating the key type.

The problem that the compiler has with the declaration map <key, Glyph*>::iterator it; is that the compiler can"t be sure if map<key, Glyph*>::iterator refers to a type or something else.
This is because it depends on the template parameter key, and you might have a specialisation of map<> somewhere where the member iterator is not a type.
To help the compiler out, you must specify that you expect that map<ket, Glyph*>::iterator refers to a typename:

typename map<key, Glyph*>::iterator it;
~没有更多了~
我们使用 Cookies 和其他技术来定制您的体验包括您的登录状态等。通过阅读我们的 隐私政策 了解更多相关信息。 单击 接受 或继续使用网站,即表示您同意使用 Cookies 和您的相关数据。
原文