增强JSON解析器在MacOS上转义引号,但在Linux上不行

Boosts JSON parser escapes quotation marks on MacOS but not on Linux

本文关键字:但在 Linux 转义 增强 MacOS JSON      更新时间:2023-10-16

我有一个运行在服务器上的c++程序,返回JSON格式的数据,用boost序列化。在某些情况下,JSON数据包的一个值部分是序列化为字符串的另一个JSON数据包。在我的开发机器上的XCode中运行,程序返回给我的值数据带有转义引号,在Linux系统上编译没有。有没有办法强迫linux上的boost做到这一点或一个很好的解决方案?(有一个缺失值引号的错误,我们已经修复了)。

下面是第一行的例子:

local:
"stations": "{n "area CHALLENGE_FUTURE": "{\n \"areaname CHALLENGE_FUTURE\": [...]"
online:
"stations": "{n "area CHALLENGE_FUTURE": "{\n "areaname CHALLENGE_FUTURE": [...]"

Boost版本为1.57,编译器(本地Xcode和在线Linux)为clang-linux-3.5。(参见编辑)

下面是简化后的代码:
// ----- STL
#include <iostream>
#include <string>
#include <map>
// ------ boost
#include <boost/asio.hpp>
#include <boost/foreach.hpp>

//formatter_base.hpp
//------------------------------------------------------------------------------
class formatter_base
{
protected:
    std::map<std::string, std::string> datas;
public:
    virtual ~formatter_base() {}
    void setvalue(std::string key, std::string value)
    {
        datas[key] = value;
    }
    std::string getvalue(std::string key)
    {
        return datas[key];
    }
    bool containsKey(std::string key)
    {
        return (datas.find(key) != datas.end());
    }
    virtual void deserialize(char *data, const std::size_t size) = 0;
    virtual std::string serialize(std::vector<std::string> keys) = 0;
};
//json_formatter.hpp
class json_formatter : public formatter_base
{
public:
    virtual void deserialize(char *data, const std::size_t size);
    virtual std::string serialize(std::vector<std::string> keys);
};

//datapacket.hpp
//------------------------------------------------------------------------------
class server;
extern server *tcp_server;
class datapacket
{
    static const char id[4];
public:
    enum DataFormat { BINARY = 0, JSON, XML };
    std::string ip;
    bool useHeader;
    datapacket() : useHeader(false), packet_data(NULL) {}
    datapacket(DataFormat format);
    std::vector<char> process(char *data, std::size_t size, std::string ip);
    std::string getvalue(std::string key)
    {
        return packet_data->getvalue(key);
    }
    void setvalue(std::string key, std::string value)
    {
        packet_data->setvalue(key, value);
    }
    bool containsKey(std::string key)
    {
        return packet_data->containsKey(key);
    }
    std::vector<char> serialize();
    std::string       toString();
private:
    bool deserialize(char *data, std::size_t size);
    std::string serialize_data(std::vector<std::string> keys);
    formatter_base *packet_data;
};
//datapacket.cpp
//------------------------------------------------------------------------------
#include <boost/iostreams/stream.hpp>
#include <boost/foreach.hpp>
#include <boost/property_tree/ptree.hpp>
#include <boost/property_tree/json_parser.hpp>
#include <string.h>
datapacket::datapacket(DataFormat format)
: useHeader(false)
{
    if(format == JSON)
    {
        packet_data = new json_formatter();
    }
    else
    {
        packet_data = NULL;
    }
}

std::vector<char> datapacket::process(
                                      char *data, std::size_t size, std::string ip)
{
    //std::cout << "datapacket::process" << std::endl;
    this->ip = ip;
    std::vector<char> ret;
    if (!deserialize(data, size))
        return ret;
    std::vector<std::string> keys;
    std::string result;
    /*extern void process(datapacket& gamedata);
     process(*this);*/
    ret.push_back('a');
    ret.push_back('c');
    ret.push_back('k');
    return ret;
}
bool datapacket::deserialize(char *data, std::size_t size)
{
    packet_data = new json_formatter();
    packet_data->deserialize(data, size);
    return true;
}
std::string datapacket::serialize_data(std::vector<std::string> keys)
{
    return packet_data->serialize(keys);
}
std::vector<char> datapacket::serialize()
{
    std::vector<std::string> keys;
    std::string str = serialize_data(keys);
    std::vector<char> ret;
    ret.assign(str.begin(), str.end());
    return ret;
}
std::string datapacket::toString()
{
    std::vector<std::string> keys;
    std::string str = serialize_data(keys);
    return str;
}

//json_formatter.cpp
//------------------------------------------------------------------------------
using namespace boost::property_tree;
void json_formatter::deserialize(char *data, const std::size_t size)
{
    std::stringstream ss;
    ss.write(data, size);
    //  std::cout << "ss: " << ss.str() << std::endl;
    ptree pt;
    read_json(ss, pt);
    BOOST_FOREACH(ptree::value_type &v, pt)
    {
        //log all received data
        //std::cout << v.first.data() << ": " << v.second.data() << std::endl;
        datas[v.first.data()] = v.second.data();
    }
}
///-workaround 1.57 json
template <typename T>
struct my_id_translator
{
    typedef T internal_type;
    typedef T external_type;
    boost::optional<T> get_value(const T &v) { return  v.substr(1, v.size() - 2) ; }
    boost::optional<T> put_value(const T &v) { return '"' + v + '"'; }
};
///
std::string json_formatter::serialize(std::vector<std::string> keys)
{

    ptree pt;
    if(keys.empty())
    {
        typedef std::map<std::string, std::string> mapType;
        BOOST_FOREACH(const mapType::value_type& myPair, datas)
        {
            //workaround for wrong formatted string
            if((BOOST_VERSION == 105700) && (BOOST_OS_LINUX))
            {
                //1.57
                pt.put(myPair.first, myPair.second, my_id_translator<std::string>());
            }
            else
            {
                //1.54
                pt.put(myPair.first, myPair.second);
            }
            //std::cout << myPair.first << ": " << myPair.second << std::endl;
        }
    }
    else
    {
        BOOST_FOREACH(std::string key, keys)
        {
            //workaround for wrong formatted string
            if(BOOST_VERSION == 105700)
            {
#if BOOST_OS_LINUX
                pt.put(key, """ + datas[key] + """, my_id_translator<std::string>());
#else
                pt.put(key, datas[key], my_id_translator<std::string>());
#endif
            }
            else
            {
                pt.put(key, datas[key]);
            }
            //      std::cout << key << ": " << datas[key] << std::endl;
        }
    }
    std::stringstream ss;
    write_json(ss, pt);
    std::string str = ss.str();
    // WORKAROUND
    // Replace all escaped backslashes
    // This was because some clients couldn't interpret "command\/read"
    std::string oldStr = "\/";
    std::string newStr = "/";
    std::size_t pos = 0;
    while((pos = str.find(oldStr)) != std::string::npos){
        str = str.replace(pos, oldStr.length(), newStr);
        pos += newStr.length();
    }
    // /WORKAROUND
    //std::cout << "Serialize message:" << std::endl;
    //std::cout << str << std::endl;
    return str;
}

//main.cpp
//------------------------------------------------------------------------------
class dataClass
{
public:
    dataClass() {}
    std::string name;
};
class innerDataClass
{
public:
    innerDataClass() {}
    std::string name;
    int         score;
    std::string baseClassName;
};
using boost::asio::ip::tcp;

namespace stdpatch
{
    template <typename T> std::string to_string(const T& n)
    {
        std::ostringstream stm;
        stm << n;
        return stm.str();
    }
}
std::map<std::string, dataClass>        listDC;
std::map<std::string, innerDataClass>   listIDC;
void Init()
{
    //Some initial values
    dataClass d1; d1.name = "dataClass1"; listDC["d1"] = d1;
    dataClass d2; d2.name = "dataClass2"; listDC["d2"] = d2;
    innerDataClass i1; i1.name = "innerClass1"; i1.baseClassName = "dataClass1";
    i1.score = 5; listIDC["i1"] = i1;
    innerDataClass i2; i2.name = "innerClass2"; i2.baseClassName = "dataClass1";
    i1.score = 21; listIDC["i2"] = i2;
    innerDataClass i3; i3.name = "innerClass3"; i1.baseClassName = "dataClass2";
    i1.score = 1; listIDC["i3"] = i3;
}
//returns JSON
datapacket GetJSON()
{
    std::pair<std::string, dataClass>       baseClassPair;
    std::pair<std::string, innerDataClass>  innerClassPair;
    datapacket baseClasses (datapacket::JSON);
    baseClasses.setvalue("comment", "this holds all the base classes");
    BOOST_FOREACH(baseClassPair, listDC)
    {
        datapacket baseClassData (datapacket::JSON);
        baseClassData.setvalue("dataName", baseClassPair.first);
        BOOST_FOREACH(innerClassPair, listIDC)
        {
            if (innerClassPair.second.baseClassName == baseClassPair.second.name)
            {
                datapacket innerClassData (datapacket::JSON);
                innerClassData.setvalue(
                                        "name", innerClassPair.second.name);
                innerClassData.setvalue(
                                        "score", stdpatch::to_string(innerClassPair.second.score));
                baseClassData.setvalue(
                                       "inner " + innerClassPair.first, innerClassData.toString());
            }
        }
        baseClasses.setvalue("base " + baseClassPair.first, baseClassData.toString());
    }
    datapacket packet (datapacket::JSON);
    packet.setvalue("comment", "this is the base-packet");
    packet.setvalue("baseClasses", baseClasses.toString());
    return packet;
}

int main(int argc, char* argv[])
{
    Init();
    datapacket packet (datapacket::JSON);
    packet = GetJSON();
    std::cout << std::endl << std::endl
    << "------------- RESULT --------------"
    << std::endl << std::endl;
    std::cout << packet.toString() << std::endl;
    return 0;
}

期望的输出应该是:

------------- RESULT --------------
baseClasses: {
    "base d1": "{n    "dataName": "d1",n    "inner i1": "{\n    \"gameID\": \"5\",\n    \"name\": \"innerClass1\"\n}\n",n    "inner i2": "{\n    \"gameID\": \"1989860191\",\n    \"name\": \"innerClass2\"\n}\n"n}n",
    "base d2": "{n    "dataName": "d2"n}n",
    "comment": "this holds all the base classes"
}
comment: this is the base-packet
{
    "baseClasses": "{n    "base d1": "{\n    \"dataName\": \"d1\",\n    \"inner i1\": \"{\\n    \\\"gameID\\\": \\\"5\\\",\\n    \\\"name\\\": \\\"innerClass1\\\"\\n}\\n\",\n    \"inner i2\": \"{\\n    \\\"gameID\\\": \\\"1989860191\\\",\\n    \\\"name\\\": \\\"innerClass2\\\"\\n}\\n\"\n}\n",n    "base d2": "{\n    \"dataName\": \"d2\"\n}\n",n    "comment": "this holds all the base classes"n}n",
    "comment": "this is the base-packet"
}

但是引号在我的服务器中没有转义:

comment: this is the base-packet
{
    "baseClasses": "{n    "base d1": "{\n    "dataName": "d1",\n    "inner i1": "{\\n    "gameID": "5",\\n    "name": "innerClass1"\\n}\\n",\n    "inner i2": "{\\n    "gameID": "1989860191",\\n    "name": "innerClass2"\\n}\\n"\n}\n",n    "base d2": "{\n    \"dataName": "d2"\n}\n",n    "comment": "this holds all the base classes"n}n",
    "comment": "this is the base-packet"
}

为了测试问题是在write_json还是传输中,我做了一个简单的JSON包:

datapacket testData(datapacket::JSON);
testData.setvalue("text", ""world"");
testData.setvalue("inner1", testData.toString());
testData.setvalue("inner2", testData.toString());

结果如下:

XCode:
{
    "inner1": "{n    "text": "\"world\""n}n",
    "inner2": "{n    "inner1": "{\n    \"text\": \"\\\"world\\\"\"\n}\n",n    "text": "\"world\""n}n",
    "text": ""world""
}
Server:
{  
    "inner1": "{n    "text": ""world""n}n"
    "inner2": "{n    "text": ""world"",n    "inner1": "{\n    "text": ""world""\n}\n"n}n",
    "text": ""world""
}

这应该表明问题在write_json

在我的linux机器上,我已经测试了这不是一个问题:

Live On Coliru

#include<boost/property_tree/json_parser.hpp>
#include<iostream>
int main(){
    using namespace boost::property_tree;
    ptree inner;
    inner.put("area CHALLENGE_FUTURE.areaname CHALLENGE_FUTURE", "something");
    std::ostringstream inner_json;
    write_json(inner_json, inner);
    ptree outer;
    outer.put("stations", inner_json.str());
    write_json(std::cout, outer);
}

打印

{
    "stations": "{n    "area CHALLENGE_FUTURE": {n        "areaname CHALLENGE_FUTURE": "something"n    }n}n"
}

如果你问我的话,这正是应该打印的内容。


如果你想要'内部' JSON是未转义的,为什么不让它成为同一棵树的一部分?

Live On Coliru

#include<boost/property_tree/json_parser.hpp>
#include<iostream>
int main(){
    using namespace boost::property_tree;
    ptree outer;
    auto& inner = outer.add_child("stations", {});
    inner.put("area CHALLENGE_FUTURE.areaname CHALLENGE_FUTURE", "something");
    write_json(std::cout, outer);
}

打印

{
    "stations": {
        "area CHALLENGE_FUTURE": {
            "areaname CHALLENGE_FUTURE": "something"
        }
    }
}

这可能是有史以来最愚蠢的错误:经过数周的跟踪错误,我们发现我们用来在机器上构建项目的linux编译shell脚本手动更改了JSON解析行为。

相关文章: