将 boost::hash 与 boost::property_tree 一起使用？

Question

我正在尝试为包含 boost::property_tree (boost::property_tree::basic_ptree<std::basic_string<char>, std::basic_string<char> >) 的对象生成哈希值。在 boost 头文件中搜索属性树，我找不到任何已定义的 hash_value 函数。我想要实现的基本示例：

class MyClass{
public:
    friend std::size_t hash_value(const MyClass & obj);
private:
    boost::property_tree m_data;
}

inline std::size_t hash_value(const MyClass & obj){
    std::size_t seed = 0;
    boost::hash_combine(seed,obj.m_data);
    return seed;
}

此代码将无法编译： "no matching function for call to 'hash_value(const boost::property_tree &)'"

我的问题：在我没有找到的某个头文件中是否为 boost::property_tree 定义了 hash_value。如果不是，通过遍历 property_tree 来散列它的 boost 惯用方法是什么？

我应该使用 ptree 序列化转换为 std::string 并对其进行哈希处理，还是手动遍历树并创建递归哈希？

Answer 1

只需将 hash<> 专门用于 ptree:

#include <boost/property_tree/ptree.hpp>
#include <boost/functional/hash.hpp>

namespace boost {
    template<typename Key, typename Data, typename KeyCompare>
    struct hash<boost::property_tree::basic_ptree<Key, Data, KeyCompare> > {
        size_t operator()(boost::property_tree::basic_ptree<Key, Data, KeyCompare> const& pt) const {
            std::size_t seed = 0;
            boost::hash_combine(seed, pt.template get_value<std::string>());
            boost::hash_range(seed, pt.begin(), pt.end());
            return seed;
        }
    };
}

够了！这是一个小的 MyClass，它从类似 json 的文字中读取：

#include <boost/property_tree/json_parser.hpp>

class MyClass{
  public:
      MyClass(std::string const& json) {
          std::istringstream iss(json);
          read_json(iss, m_data);
      }
  private:
    boost::property_tree::ptree m_data;

    friend inline std::size_t hash_value(const MyClass& obj){
        std::size_t seed = 0;
        boost::hash_combine(seed, obj.m_data);
        return seed;
    }
};

现在可以测试了

Live On Coliru

#include <iostream>

int main() {
    for (std::string const data : {
            R"({"a":[1,2,3],"b":{"nest":"hello","more":"world"}})",
            R"({"b":{"nest":"hello","more":"world"},"a":[1,2,3]})",
            R"({ })",
            R"({})",
        })
    {
        MyClass o(data);
        std::cout << "object hash: " << hash_value(o) << " " << data << "\n";
    }
}

打印：

object hash: 3573231694259656572 {"a":[1,2,3],"b":{"nest":"hello","more":"world"}}
object hash: 11176663460548092204 {"b":{"nest":"hello","more":"world"},"a":[1,2,3]}
object hash: 3864292196 { }
object hash: 3864292196 {}

注意：哈希和相等性

对于许多容器，hash<> 假设有一个 对应的 相等比较器。如果它们不匹配，你会得到 Undefined Behaviour.

您可能想根据 ptree 的有序（关联）接口来定义 hash_range：

boost::hash_range(seed, pt.ordered_begin(), pt.not_found()); // CAUTION

这样做的好处是 {"a":1,"b":2} 可以匹配 {"b":2,"a":1}。

不要这样做，除非你知道自己在做什么。具体来说，您需要将兼容的相等比较器传递给每个使用此散列的 container/algorithm。

如果你这样写，并使用像这样的驱动程序进行测试：

int main() {
    MyClass a{ R"({"a":[1,2,3],"b":{"nest":"hello","more":"world"}})" },
            b{R"({"b":{"nest":"hello","more":"world"},"a":[1,2,3]})" },
            c{R"({ })" },
            d{R"({})" };

    for (auto& lhs : {a,b,c,d})
    for (auto& rhs : {a,b,c,d})
    {
        std::cout << "hash: " << hash_value(lhs) << " " << hash_value(rhs) << " - equality: " << std::boolalpha << (lhs==rhs) << "\n";
        if ((hash_value(lhs) == hash_value(rhs)) != (lhs==rhs))
            std::cout << " -- MISMATCH\n";
    }
}

它将打印：

hash: 10737438301360613971 10737438301360613971 - equality: true
hash: 10737438301360613971 10737438301360613971 - equality: false
 -- MISMATCH
hash: 10737438301360613971 3864292196 - equality: false
hash: 10737438301360613971 3864292196 - equality: false
hash: 10737438301360613971 10737438301360613971 - equality: false
 -- MISMATCH
hash: 10737438301360613971 10737438301360613971 - equality: true
hash: 10737438301360613971 3864292196 - equality: false
hash: 10737438301360613971 3864292196 - equality: false
hash: 3864292196 10737438301360613971 - equality: false
hash: 3864292196 10737438301360613971 - equality: false
hash: 3864292196 3864292196 - equality: true
hash: 3864292196 3864292196 - equality: true
hash: 3864292196 10737438301360613971 - equality: false
hash: 3864292196 10737438301360613971 - equality: false
hash: 3864292196 3864292196 - equality: true
hash: 3864292196 3864292196 - equality: true

MISMATCH 警告表明相等性和散列不一致。

如果您运行具有原始哈希（首屏）的测试驱动程序，它将打印：

hash: 3573231694259656572 3573231694259656572 - equality: true
hash: 3573231694259656572 11176663460548092204 - equality: false
hash: 3573231694259656572 3864292196 - equality: false
hash: 3573231694259656572 3864292196 - equality: false
hash: 11176663460548092204 3573231694259656572 - equality: false
hash: 11176663460548092204 11176663460548092204 - equality: true
hash: 11176663460548092204 3864292196 - equality: false
hash: 11176663460548092204 3864292196 - equality: false
hash: 3864292196 3573231694259656572 - equality: false
hash: 3864292196 11176663460548092204 - equality: false
hash: 3864292196 3864292196 - equality: true
hash: 3864292196 3864292196 - equality: true
hash: 3864292196 3573231694259656572 - equality: false
hash: 3864292196 11176663460548092204 - equality: false
hash: 3864292196 3864292196 - equality: true
hash: 3864292196 3864292196 - equality: true

将 boost::hash 与 boost::property_tree 一起使用？

Using boost::hash with boost::property_tree?

hash

boost

boost-propertytree

注意：哈希和相等性