source: trunk/sample/EncodeString.hpp @ 25

#ifndef OTOCO_SAMPLE_ENCODE_STRING_H
#define OTOCO_SAMPLE_ENCODE_STRING_H

#include <vector>
#include <unordered_map>
#include <string>
#include <exception>

class EncodeStringException : public std::exception
{
    const std::string message;
    EncodeStringException() = delete;
    EncodeStringException &operator =(const EncodeStringException &) = delete;
public:
    explicit EncodeStringException(const std::string &mess) : std::exception(), message(mess) {}
    EncodeStringException(const EncodeStringException &src) : std::exception(), message(src.message) {}
    virtual ~EncodeStringException() throw() {}
    virtual char const *what() const throw() { return message.c_str(); }
};

enum charset_t {
    chset_Utf8, chset_C99, chset_Java,
    chset_Ucs2, chset_Ucs2Be, chset_Ucs2Le,
    chset_Ucs4, chset_Ucs4Be, chset_Ucs4Le,
    chset_Utf16, chset_Utf16Be, chset_Utf16Le,
    chset_Utf32, chset_Utf32Be, chset_Utf32Le,
    chset_Utf7,
    chset_EucJp, chset_EucJis0213,
    chset_Iso2022Jp, chset_Iso2022Jp2, chset_Iso2022Jp1, chset_Iso2022Jp3,
    chset_ShiftJis, chset_Cp932, chset_ShiftJisX0213,
};

namespace std {
template <>
    inline size_t
    hash<charset_t>::operator()(charset_t val) const
{
    return static_cast<size_t>(val);
}
}

class EncodeStringImpl
{
    typedef std::unordered_map<charset_t, char const *> cnmap_t;
    static cnmap_t const CharsetNames;
    static size_t const TerminateNullCharNum = 4;   ///< 文字列終端のヌル文字に使用するバイト数。
protected:
    std::vector<char> result;
    size_t result_length;
    EncodeStringImpl();
    virtual ~EncodeStringImpl() {}
    void encode(void const *src, size_t src_length, size_t chr_size, charset_t from_charset, charset_t to_charset);
};

template <typename ic_t, charset_t internal_charset>
class EncodeString : private EncodeStringImpl
{
    std::basic_string<ic_t> source;
    EncodeString() = delete;
    EncodeString(const EncodeString &) = delete;
    EncodeString &operator =(const EncodeString &) = delete;
public:
    EncodeString(const std::basic_string<ic_t> &src, charset_t charset) : EncodeStringImpl(), source{src}
    { encode(source.c_str(), source.size(), sizeof(ic_t), internal_charset, charset); }
    EncodeString(const std::basic_string<ic_t> &&src, charset_t charset) : source{src}
    { encode(source.c_str(), source.size(), sizeof(ic_t), internal_charset, charset); }
    ~EncodeString() {}
    const char *getCharArray() const { return &result[0]; }
    const size_t &getLength() const { return result_length; }
};

#endif //OTOCO_SAMPLE_ENCODE_STRING_H