NBT_IO.hpp

浏览该文件的文档.

#pragma once
 
#include <stdio.h>
#include <stdint.h>
#include <stddef.h>//size_t
#include <string.h>//memcpy
#include <fstream>
#include <vector>
#include <filesystem>
 
#include "NBT_Print.hpp"//打印输出
 
#include "vcpkg_config.h"//包含vcpkg生成的配置以确认库安装情况
 
#ifdef CJF2_NBT_CPP_USE_ZLIB
 
/*zlib*/
#define ZLIB_CONST
#include <zlib.h>
/*zlib*/
 
#endif


class NBT_IO
{
    NBT_IO(void) = delete;
    ~NBT_IO(void) = delete;
 
public:
    template <typename T = std::vector<uint8_t>>
    class DefaultInputStream
    {
    private:
        const T &tData = {};
        size_t szIndex = 0;
 
    public:
        using StreamType = T;
        using ValueType = typename T::value_type;
 
        //静态断言确保字节流与可平凡拷贝
        static_assert(sizeof(ValueType) == 1, "Error ValueType Size");
        static_assert(std::is_trivially_copyable_v<ValueType>, "ValueType Must Be Trivially Copyable");

        DefaultInputStream(const T &&_tData, size_t szStartIdx = 0) = delete;

        DefaultInputStream(const T &_tData, size_t szStartIdx = 0) :tData(_tData), szIndex(szStartIdx)
        {}

        ~DefaultInputStream(void) = default;
        DefaultInputStream(const DefaultInputStream &) = delete;
        DefaultInputStream(DefaultInputStream &&) = delete;
        DefaultInputStream &operator=(const DefaultInputStream &) = delete;
        DefaultInputStream &operator=(DefaultInputStream &&) = delete;

        const ValueType &operator[](size_t szIndex) const noexcept
        {
            return tData[szIndex];
        }

        const ValueType &GetNext() noexcept
        {
            return tData[szIndex++];
        }

        void GetRange(void *pDest, size_t szSize) noexcept
        {
            memcpy(pDest, &tData[szIndex], szSize);
            szIndex += szSize;
        }

        void UnGet() noexcept
        {
            --szIndex;
        }

        size_t SkipData(size_t szSize) noexcept
        {
            return szIndex += szSize;
        }

        size_t RewindData(size_t szSize) noexcept
        {
            return szIndex -= szSize;
        }

        bool IsEnd() const noexcept
        {
            return szIndex >= tData.size();
        }

        size_t Size() const noexcept
        {
            return tData.size();
        }

        bool HasAvailData(size_t szSize) const noexcept
        {
            return (tData.size() - szIndex) >= szSize;
        }

        void Reset() noexcept
        {
            szIndex = 0;
        }

        const size_t &Index() const noexcept
        {
            return szIndex;
        }

        size_t &Index() noexcept
        {
            return szIndex;
        }
    };

    template <typename T = std::vector<uint8_t>>
    class DefaultOutputStream
    {
    private:
        T &tData;
 
    public:
        using StreamType = T;
        using ValueType = typename T::value_type;
 
        //静态断言确保字节流与可平凡拷贝
        static_assert(sizeof(ValueType) == 1, "Error ValueType Size");
        static_assert(std::is_trivially_copyable_v<ValueType>, "ValueType Must Be Trivially Copyable");

        DefaultOutputStream(T &_tData, size_t szStartIdx = 0) :tData(_tData)//引用天生无法使用临时值构造，无需担心临时值构造导致的悬空引用
        {
            tData.resize(szStartIdx);
        }

        ~DefaultOutputStream(void) = default;
        DefaultOutputStream(const DefaultOutputStream &) = delete;
        DefaultOutputStream(DefaultOutputStream &&) = delete;
        DefaultOutputStream &operator=(const DefaultOutputStream &) = delete;
        DefaultOutputStream &operator=(DefaultOutputStream &&) = delete;

        const ValueType &operator[](size_t szIndex) const noexcept
        {
            return tData[szIndex];
        }

        template<typename V>
        requires(std::is_constructible_v<ValueType, V &&>)
        void PutOnce(V &&c)
        {
            tData.push_back(std::forward<V>(c));
        }

        void PutRange(const ValueType *pData, size_t szSize)
        {
            //tData.insert(tData.end(), &pData[0], &pData[szSize]);
 
            //使用更高效的实现，而不是迭代器写入
            size_t szCurSize = tData.size();
            tData.resize(szCurSize + szSize);
            memcpy(&tData.data()[szCurSize], &pData[0], szSize);
        }

        void AddReserve(size_t szAddSize)
        {
            tData.reserve(tData.size() + szAddSize);
        }

        void UnPut(void) noexcept
        {
            tData.pop_back();
        }

        size_t RemoveData(size_t szSize) noexcept
        {
            tData.resize(tData.size() - szSize);
            return tData.size();
        }

        size_t Size(void) const noexcept
        {
            return tData.size();
        }

        void Reset(void) noexcept
        {
            tData.clear();
        }
    };
 
 
public:
    template<typename T = std::vector<uint8_t>, typename InfoFunc = NBT_Print>
    requires (sizeof(typename T::value_type) == 1 && std::is_trivially_copyable_v<typename T::value_type>)
    static bool WriteFile(const std::filesystem::path &pathFileName, const T &tData, InfoFunc funcInfo = InfoFunc{}) noexcept
    {
        try
        {
            std::error_code ec;
 
            //存在性检查
            bool bExists = std::filesystem::exists(pathFileName, ec);
            if (ec || bExists)
            {
                if (ec)
                {
                    funcInfo(NBT_Print_Level::Err, "Error: Failed to check existence of [{}]: {}\n", pathFileName.string(), ec.message());
                    return false;
                }
                else
                {
                    //已存在，必须为普通文件
                    bool bRegularFile = std::filesystem::is_regular_file(pathFileName, ec);
                    if (ec)
                    {
                        funcInfo(NBT_Print_Level::Err, "Error: Failed to check file type of [{}]: {}\n", pathFileName.string(), ec.message());
                        return false;
                    }
 
                    //不是普通文件，出错
                    if (!bRegularFile)
                    {
                        funcInfo(NBT_Print_Level::Err, "Error: [{}] exists but is not a regular file.\n", pathFileName.string());
                        return false;
                    }
                }
            }
 
            //检查通过，继续写入
            std::fstream fWrite;
            fWrite.open(pathFileName, std::ios_base::binary | std::ios_base::out | std::ios_base::trunc);
            if (!fWrite)
            {
                funcInfo(NBT_Print_Level::Err, "Error: Cannot open file [{}] for writing.\n", pathFileName.string());
                return false;
            }
 
            //获取文件大小并写出
            uint64_t qwFileSize = tData.size();
            if (!fWrite.write((const char *)tData.data(), sizeof(tData[0]) * qwFileSize))
            {
                funcInfo(NBT_Print_Level::Err, "Error: Failed to write data to file [{}].\n", pathFileName.string());
                return false;
            }
 
            //完成，关闭文件
            fWrite.close();
 
            return true;
        }
        catch (const std::bad_alloc &e)
        {
            funcInfo(NBT_Print_Level::Err, "std::bad_alloc:[{}]\n", e.what());
            return false;
        }
        catch (const std::exception &e)
        {
            funcInfo(NBT_Print_Level::Err, "std::exception:[{}]\n", e.what());
            return false;
        }
        catch (...)
        {
            funcInfo(NBT_Print_Level::Err, "Unknown Error\n");
            return false;
        }
    }

    template<typename T = std::vector<uint8_t>, typename InfoFunc = NBT_Print>
    requires (sizeof(typename T::value_type) == 1 && std::is_trivially_copyable_v<typename T::value_type>)
    static bool ReadFile(const std::filesystem::path &pathFileName, T &tData, InfoFunc funcInfo = InfoFunc{}) noexcept
    {
        try
        {
            std::error_code ec;
 
            bool bRegularFile = std::filesystem::is_regular_file(pathFileName, ec);
            if (ec || !bRegularFile)//必须是普通文件
            {
                if (ec)
                {
                    funcInfo(NBT_Print_Level::Err, "Error: Failed to check file type of [{}]: {}.\n", pathFileName.string(), ec.message());
                }
                else
                {
                    funcInfo(NBT_Print_Level::Err, "Error: [{}] is not a regular file.\n", pathFileName.string());
                }
                return false;
            }
 
            //获取文件大小
            uintmax_t umFileSize = std::filesystem::file_size(pathFileName, ec);
            if (ec || umFileSize > (uintmax_t)std::numeric_limits<size_t>::max())//大小超过size_t范围，无法读取
            {
                if (ec)
                {
                    funcInfo(NBT_Print_Level::Err, "Error: Cannot get file size of [{}]: {}\n", pathFileName.string(), ec.message());
                }
                else
                {
                    funcInfo(NBT_Print_Level::Err, "Error: File [{}](size: [{}] bytes) is too large to fit into memory(max: [{}] bytes).\n", pathFileName.string(), umFileSize, std::numeric_limits<size_t>::max());
                }
                return false;
            }
 
            //转换为size_t
            size_t szFileSize = (size_t)umFileSize;
 
            //打开文件
            std::fstream fRead;
            fRead.open(pathFileName, std::ios_base::binary | std::ios_base::in);
            if (!fRead)
            {
                funcInfo(NBT_Print_Level::Err, "Error: Cannot open file [{}] for reading.\n", pathFileName.string());
                return false;
            }
 
            //直接给数据塞string里
            tData.resize(szFileSize);//设置长度 c++23用resize_and_overwrite
            if (!fRead.read((char *)tData.data(), sizeof(tData[0]) * szFileSize))//直接读入data
            {
                funcInfo(NBT_Print_Level::Err, "Error: Failed to read data from file [{}].\n", pathFileName.string());
                return false;
            }
 
            //完成，关闭文件
            fRead.close();
 
            return true;
        }
        catch (const std::bad_alloc &e)
        {
            funcInfo(NBT_Print_Level::Err, "std::bad_alloc:[{}]\n", e.what());
            return false;
        }
        catch (const std::exception &e)
        {
            funcInfo(NBT_Print_Level::Err, "std::exception:[{}]\n", e.what());
            return false;
        }
        catch (...)
        {
            funcInfo(NBT_Print_Level::Err, "Unknown Error\n");
            return false;
        }
    }

    static bool IsFileExist(const std::filesystem::path &pathFileName)
    {
        std::error_code ec;//判断这东西是不是true确定有没有error
        bool bExists = std::filesystem::exists(pathFileName, ec);
 
        return !ec && bExists;//没有错误并且存在
    }
 
#ifdef CJF2_NBT_CPP_USE_ZLIB

    static bool IsZlib(uint8_t u8DataFirst, uint8_t u8DataSecond)
    {
        return u8DataFirst == (uint8_t)0x78 &&
            (
                u8DataSecond == (uint8_t)0x9C ||
                u8DataSecond == (uint8_t)0x01 ||
                u8DataSecond == (uint8_t)0xDA ||
                u8DataSecond == (uint8_t)0x5E
            );
    }

    static bool IsGzip(uint8_t u8DataFirst, uint8_t u8DataSecond)
    {
        return u8DataFirst == (uint8_t)0x1F && u8DataSecond == (uint8_t)0x8B;
    }

    template<typename T>
    requires (sizeof(typename T::value_type) == 1 && std::is_trivially_copyable_v<typename T::value_type>)
    static bool IsDataZipped(const T &tData)
    {
        if (tData.size() <= 2)
        {
            return false;
        }
 
        uint8_t u8DataFirst = (uint8_t)tData[0];
        uint8_t u8DataSecond = (uint8_t)tData[1];
 
        return IsZlib(u8DataFirst, u8DataSecond) || IsGzip(u8DataFirst, u8DataSecond);
    }

    template<typename I, typename O>
    requires (sizeof(typename I::value_type) == 1 && std::is_trivially_copyable_v<typename I::value_type> &&
              sizeof(typename O::value_type) == 1 && std::is_trivially_copyable_v<typename O::value_type>)
    static void DecompressData(O &oData, const I &iData)
    {
        if (std::addressof(oData) == std::addressof(iData))
        {
            throw std::runtime_error("The oData object cannot be the iData object");
        }
 
        if (iData.empty())
        {
            oData.clear();
            return;
        }
 
        z_stream zs
        {
            .next_in = Z_NULL,
            .avail_in = 0,
            .total_in = 0,
 
            .next_out = Z_NULL,
            .avail_out = 0,
            .total_out = 0,
 
            .msg = Z_NULL,
            .state = Z_NULL,
 
            .zalloc = (alloc_func)Z_NULL,
            .zfree = (free_func)Z_NULL,
            .opaque = (voidpf)Z_NULL,
 
            .data_type = Z_BINARY,
 
            .adler = 0,
            .reserved = {},
        };
 
        if (inflateInit2(&zs, 32 + 15) != Z_OK)//32+15自动判断是gzip还是zlib
        {
            throw std::runtime_error("Failed to initialize zlib decompression");
        }
 
        //对于大于uint_max的数据来说，切分为多个uint_max的块作为流依次输入
        //注意zlib在实现内会移动指针，所以对于大于一定字节的情况下，只需要更新
        //avail_in，而无须更新next_in。这里的容器保证不变，所以地址不会失效
        zs.next_in = (z_const Bytef *)iData.data();
 
        //默认解压大小为压缩大小，后续扩容
        oData.resize(iData.size());
 
        //两个变量用于记录已经压缩的大小和剩余数据大小
        size_t szDecompressedSize = 0;
        size_t szRemainingSize = iData.size();
        int iRet = Z_OK;
        do
        {
            //首先计算剩余可用空间，然后决定是否扩容
            size_t szOut = oData.size() - szDecompressedSize;
            if (szOut == 0)
            {
                oData.resize(oData.size() * 2);
                szOut = oData.size() - szDecompressedSize;
            }
 
            //虽然next_out也会在内部实现被移动，但是oData是容器，会被扩容
            //一旦扩容触发，那么实际上的地址就可能会改变，所以必须每次重新获取
            //切记上面与这里的代码顺序不可改变，必须在上面计算并扩容完成后获取地址
            zs.next_out = (Bytef *)(&oData.data()[szDecompressedSize]);
 
            //如果输入被消耗完，重新赋值
            if (zs.avail_in == 0)
            {
                constexpr uInt uIntMax = (uInt)-1;
                zs.avail_in = szRemainingSize > (size_t)uIntMax ? uIntMax : (uInt)szRemainingSize;
                szRemainingSize -= zs.avail_in;//缩小剩余待处理大小
            }
            
            //如果输出大小耗尽，重新赋值
            if (zs.avail_out == 0)
            {
                constexpr uInt uIntMax = (uInt)-1;
                zs.avail_out = szOut > (size_t)uIntMax ? uIntMax : (uInt)szOut;
                //这里不对szOut处理，因为会在开头与结尾计算
            }
 
            //解压，如果剩余不为0则代表分块了，使用Z_NO_FLUSH，否则Z_FINISH
            iRet = inflate(&zs, szRemainingSize != 0 ? Z_NO_FLUSH : Z_FINISH);
 
            //计算本次解压的大小
            szDecompressedSize += szOut - zs.avail_out;
        } while (iRet == Z_OK || iRet == Z_BUF_ERROR);//只要没错误（缓冲区不够大除外）或者没到结尾就继续运行
 
        inflateEnd(&zs);//结束解压
        oData.resize(szDecompressedSize);//设置解压大小
 
        //错误处理
        if (iRet != Z_STREAM_END)
        {
            if (zs.msg != NULL)//错误消息不为null则抛出带消息异常
            {
                throw std::runtime_error(std::string("Zlib decompression failed with error message: ") + std::string(zs.msg));
            }
            else//如果为null直接使用错误码
            {
                throw std::runtime_error(std::string("Zlib decompression failed with error code: ") + std::to_string(iRet));
            }
        }
    }

    template<typename I, typename O>
    requires (sizeof(typename I::value_type) == 1 && std::is_trivially_copyable_v<typename I::value_type> &&
              sizeof(typename O::value_type) == 1 && std::is_trivially_copyable_v<typename O::value_type>)
    static void CompressData(O &oData, const I &iData, int iLevel = Z_DEFAULT_COMPRESSION)
    {
        if (std::addressof(oData) == std::addressof(iData))
        {
            throw std::runtime_error("The oData object cannot be the iData object");
        }
 
        if (iData.empty())
        {
            oData.clear();
            return;
        }
 
        z_stream zs
        {
            .next_in = Z_NULL,
            .avail_in = 0,
            .total_in = 0,
 
            .next_out = Z_NULL,
            .avail_out = 0,
            .total_out = 0,
 
            .msg = Z_NULL,
            .state = Z_NULL,
 
            .zalloc = (alloc_func)Z_NULL,
            .zfree = (free_func)Z_NULL,
            .opaque = (voidpf)Z_NULL,
 
            .data_type = Z_BINARY,
 
            .adler = 0,
            .reserved = {},
        };
 
        if (deflateInit2(&zs, iLevel, Z_DEFLATED, 16 + 15, 8, Z_DEFAULT_STRATEGY) != Z_OK)//16+15使用gzip（mojang默认格式）
        {
            throw std::runtime_error("Failed to initialize zlib compression");
        }
 
        //与上方解压例程相同，不做重复解释
        zs.next_in = (z_const Bytef *)iData.data();
 
        /*
            如果范围溢出，则设置为比原始数据的大小加12字节大0.1%的一半
            这样做的目的是为了尽可能缩小一开始的体积
            比如实际上压缩率非常低的情况下可能根本用不到一半
            但是如果实际上压缩率很高，那么下面二倍扩容一次后刚好就是最坏情况
            这样基本上性能较优，下面zlib文档说明的最坏情况：
            
            Upon entry, destLen is the total size of the
            destination buffer, which must be at least 0.1%
            larger than sourceLen plus 12 bytes.
        */
 
        constexpr uLong uLongMax = (uLong)-1;
        if (iData.size() > (size_t)uLongMax)
        {
            size_t szNeedSize = iData.size() + 12;//先比原始数据大12byte
            //注意这里使用了向上取整的整数除法
            //加等于自身的0.1%相当于比原先的自己大0.1%，这里的1/1000就是0.1/100
            szNeedSize += (szNeedSize + (1000 - 1)) / 1000;
            //设置目标大小
            oData.resize((szNeedSize + (2 - 1)) / 2);//向上取整除以二
        }
        else
        {
            //在范围未溢出的情况下，进行预测
            //把压缩大小设置为预测的压缩大小
            oData.resize(deflateBound(&zs, (uLong)iData.size()));
        }
        
        //设置压缩后大小与待处理大小
        size_t szCompressedSize = 0;
        size_t szRemainingSize = iData.size();
        int iRet = Z_OK;
        do
        {
            //计算剩余大小并在不足时扩容
            size_t szOut = oData.size() - szCompressedSize;
            if (szOut == 0)
            {
                oData.resize(oData.size() * 2);
                szOut = oData.size() - szCompressedSize;
            }
 
            //获取新的地址
            zs.next_out = (Bytef *)(&oData.data()[szCompressedSize]);
 
            //如果输入被消耗完，重新赋值
            if (zs.avail_in == 0)
            {
                constexpr uInt uIntMax = (uInt)-1;
                zs.avail_in = szRemainingSize > (size_t)uIntMax ? uIntMax : (uInt)szRemainingSize;
                szRemainingSize -= zs.avail_in;//缩小剩余待处理大小
            }
 
            //如果输出大小耗尽，重新赋值
            if (zs.avail_out == 0)
            {
                constexpr uInt uIntMax = (uInt)-1;
                zs.avail_out = szOut > (size_t)uIntMax ? uIntMax : (uInt)szOut;
                //这里不对szOut处理，因为会在开头与结尾计算
            }
            
            //压缩，如果剩余不为0则代表分块了，使用Z_NO_FLUSH，否则Z_FINISH
            iRet = deflate(&zs, szRemainingSize != 0 ? Z_NO_FLUSH : Z_FINISH);
 
            //计算本次压缩的大小
            szCompressedSize += szOut - zs.avail_out;
        } while (iRet == Z_OK || iRet == Z_BUF_ERROR);//只要没错误（缓冲区不够大除外）或者没到结尾就继续运行
 
        //结束并设置大小
        deflateEnd(&zs);
        oData.resize(szCompressedSize);
 
        //错误处理
        if (iRet != Z_STREAM_END)
        {
            if (zs.msg != NULL)
            {
                throw std::runtime_error(std::string("Zlib compression failed with error message: ") + std::string(zs.msg));
            }
            else
            {
                throw std::runtime_error(std::string("Zlib compression failed with error code: ") + std::to_string(iRet));
            }
        }
    }

    template<typename I, typename O, typename InfoFunc = NBT_Print>
    requires (sizeof(typename I::value_type) == 1 && std::is_trivially_copyable_v<typename I::value_type> &&
              sizeof(typename O::value_type) == 1 && std::is_trivially_copyable_v<typename O::value_type>)
    static bool DecompressDataNoThrow(O &oData, const I &iData, InfoFunc funcInfo = InfoFunc{}) noexcept
    {
        try
        {
            DecompressData(oData, iData);
            return true;
        }
        catch (const std::bad_alloc &e)
        {
            funcInfo(NBT_Print_Level::Err, "std::bad_alloc:[{}]\n", e.what());
            return false;
        }
        catch (const std::exception &e)
        {
            funcInfo(NBT_Print_Level::Err, "std::exception:[{}]\n", e.what());
            return false;
        }
        catch (...)
        {
            funcInfo(NBT_Print_Level::Err, "Unknown Error\n");
            return false;
        }
    }

    template<typename I, typename O, typename InfoFunc = NBT_Print>
    requires (sizeof(typename I::value_type) == 1 && std::is_trivially_copyable_v<typename I::value_type> &&
              sizeof(typename O::value_type) == 1 && std::is_trivially_copyable_v<typename O::value_type>)
    static bool CompressDataNoThrow(O &oData, const I &iData, int iLevel = Z_DEFAULT_COMPRESSION, InfoFunc funcInfo = InfoFunc{}) noexcept
    {
        try
        {
            CompressData(oData, iData, iLevel);
            return true;
        }
        catch (const std::bad_alloc &e)
        {
            funcInfo(NBT_Print_Level::Err, "std::bad_alloc:[{}]\n", e.what());
            return false;
        }
        catch (const std::exception &e)
        {
            funcInfo(NBT_Print_Level::Err, "std::exception:[{}]\n", e.what());
            return false;
        }
        catch (...)
        {
            funcInfo(NBT_Print_Level::Err, "Unknown Error\n");
            return false;
        }
    }
#endif
 
};