Question

是否可以扩展boost::lexical_cast来处理其他数据类型，而无需实际修改这些类？

在我的情况下，我想扩展它来处理像cv::Point和cv::Point3这样的事情，采用字符串分隔的坐标列表并加载它们。所以能够做类似的事情： / p>

cv::Point mypoint = boost::lexical_cast<cv::Point>("2,4");

cv::Point类已有流操作符，但与istream和wstream不兼容，因此失败。

修改

我问这个是因为我在一个带有模板化函数get_parameter的框架中工作，它使用boost::lexical_cast将字符串（从配置文件中读取）转换为所需的数据类型。它非常适用于ints＆amp;浮动，但现在我必须多次调用它来读取2D或3D点（甚至更糟的是系数数组）。能够修改lexical_cast来处理这些情况会很好。

因此，这不是特定于OpenCV，我只选择它作为最简单的数据类型..我对一般解决方案更感兴趣。

修改2

以下是我一直在尝试的示例应用：

#include <opencv2/opencv.hpp>
#include <boost/lexical_cast.hpp>

template <typename T>
std::istream& operator>>(std::istream& stream, cv::Point_<T> &p) {
   // Eventually something will go here
   // to put stream into p
}

int main(int argc, char **argv) {
  cv::Point_<float> p = boost::lexical_cast<cv::Point_<float>>(std::string("1,2"));
  std::cout << "p = " << p << std::endl;
  return 0;
}

它失败了一个漂亮的C ++模板错误：

In file included from /home/rhand/Development/mlx/ml_3rdparty/install/boost/include/boost/lexical_cast.hpp:41:0,
             from /home/rhand/Development/experiments/lexical_Cast/test.cc:2:
/home/rhand/Development/mlx/ml_3rdparty/install/boost/include/boost/lexical_cast.hpp: In instantiation of ‘struct boost::detail::deduce_target_char_impl<boost::detail::deduce_character_type_later<cv::Point_<float> > >’:
/home/rhand/Development/mlx/ml_3rdparty/install/boost/include/boost/lexical_cast.hpp:415:89:   required from ‘struct boost::detail::deduce_target_char<cv::Point_<float> >’
/home/rhand/Development/mlx/ml_3rdparty/install/boost/include/boost/lexical_cast.hpp:674:92:   required from ‘struct boost::detail::lexical_cast_stream_traits<std::basic_string<char>, cv::Point_<float> >’
/home/rhand/Development/mlx/ml_3rdparty/install/boost/include/boost/lexical_cast.hpp:2363:19:   required from ‘static Target boost::detail::lexical_cast_do_cast<Target, Source>::lexical_cast_impl(const Source&) [with Target = cv::Point_<float>; Source = std::basic_string<char>]’
/home/rhand/Development/mlx/ml_3rdparty/install/boost/include/boost/lexical_cast.hpp:2543:50:   required from ‘Target boost::lexical_cast(const Source&) [with Target = cv::Point_<float>; Source = std::basic_string<char>]’
/home/rhand/Development/experiments/lexical_Cast/test.cc:11:82:   required from here
/home/rhand/Development/mlx/ml_3rdparty/install/boost/include/boost/static_assert.hpp:31:45: error: static assertion failed: Target type is neither std::istream`able nor std::wistream`able
 #     define BOOST_STATIC_ASSERT_MSG( ... ) static_assert(__VA_ARGS__)
                                         ^
/home/rhand/Development/mlx/ml_3rdparty/install/boost/include/boost/lexical_cast.hpp:388:13: note: in expansion of macro ‘BOOST_STATIC_ASSERT_MSG’
         BOOST_STATIC_ASSERT_MSG((result_t::value || boost::has_right_shift<std::basic_istream<wchar_t>, T >::value),
         ^
make[2]: *** [CMakeFiles/test.dir/test.cc.o] Error 1
make[1]: *** [CMakeFiles/test.dir/all] Error 2
make: *** [all] Error 2

Answer 1

如果内存服务正常，您应该能够定义istream& operator>>(istream&, cv::Point&)将使用的自由函数（lexical_cast），只要它在lexical_cast调用之前定义。

修改：有关示例，请参阅Overloading istream operator>> c++

编辑2：这是一个工作示例（VS2013，提升1.44 [是的，我需要更新！]）。你能发一个在你的情况下失败的最小例子吗？

#include <iostream>
#include <boost/lexical_cast.hpp>

template <typename ElementType>
struct Point
{
    Point() : x(0), y(0) {}
    ElementType x, y;
};

template <typename ElementType>
std::istream& operator>>(std::istream& stream, Point<ElementType> &p)
{
    stream >> p.x;
    stream.get();
    stream >> p.y;
    return stream;
}

template <typename ElementType>
std::ostream& operator<<(std::ostream& stream, const Point<ElementType> &p)
{
    stream << p.x << "," << p.y;
    return stream;
}

int main(int argc, char **argv)
{
    Point<int> p = boost::lexical_cast<Point<int>>("1,2");
    std::cout << "p=[" << p << "]";
    std::cin.get();
    return 0;
}

编辑3：使Point课程成为模板课程，因为它似乎是您的理由

编辑4：使用OpenCV 2.4.10，也适用于上述设置：

#include <iostream>
#include <boost/lexical_cast.hpp>
#include <opencv2/opencv.hpp>

template <typename ElementType>
std::istream& operator>>(std::istream& stream, cv::Point_<ElementType> &p)
{
    stream >> p.x;
    stream.get();
    stream >> p.y;
    return stream;
}

int main(int argc, char **argv)
{
    auto cv_p = boost::lexical_cast<cv::Point_<float>>(std::string("1,2"));
    std::cout << "opencv p=" << cv_p << "";
    std::cin.get();
    return 0;
}

Answer 2

您可以为要转换的类型定义boost :: lexical_cast的特化。

玩具示例：

typedef struct { int x; int y; } Point;

namespace boost {
    template<>
      std::string lexical_cast(const Point& arg) { return "2,3"; }
}

int main () {
    std::cout << boost::lexical_cast<std::string>(Point ()) << std::endl;
    }

打印2,3。

从一个字符串到一个点需要更多的工作，但你可以看到如何做。

其他类数据类型的扩展boost :: lexical_cast

2 个答案: