diff --git a/source/main.cpp b/source/main.cpp index 8ac66c7..dbd64ba 100644 --- a/source/main.cpp +++ b/source/main.cpp @@ -1,5 +1,6 @@ #include #include + #include int main() diff --git a/tests/utf8_string_test.cpp b/tests/utf8_string_test.cpp index ccf5d2a..5a5a78f 100644 --- a/tests/utf8_string_test.cpp +++ b/tests/utf8_string_test.cpp @@ -61,4 +61,86 @@ TEST(utf8_string, replace_char_case_3) ustr.replace_char("مرحبا", 2); ASSERT_EQ(ustr.get_str(), "прمرحباпgт"); +} + +// --------------------------------------------------------------------------------------------------------------------- +TEST(utf8_string, ctors) +{ + // ----------------------------- const char* PART ----------------------------- + const char* c_str{ "привет" }; + + utf8::ustring ustr_from_c_str{ c_str }; + ASSERT_EQ(c_str, ustr_from_c_str.get_c_str()); + + ustr_from_c_str.replace_char("g", 0); + ASSERT_NE(c_str, ustr_from_c_str.get_c_str()); + + + // ----------------------------- std::string PART ----------------------------- + std::string cpp_str{ "пока" }; + + utf8::ustring ustr_from_cpp_str{ cpp_str }; + ASSERT_EQ(cpp_str, ustr_from_cpp_str.get_c_str()); + + ustr_from_cpp_str.replace_char("h", 0); + ASSERT_NE(cpp_str, ustr_from_cpp_str.get_c_str()); + + + // ----------------------------- std::string_view PART ----------------------------- + std::string_view cpp_str_view{ "что" }; + + utf8::ustring ustr_from_cpp_str_view{ cpp_str_view }; + ASSERT_EQ(cpp_str_view, ustr_from_cpp_str_view.get_c_str()); + + ustr_from_cpp_str_view.replace_char("L", 0); + ASSERT_NE(cpp_str_view, ustr_from_cpp_str_view.get_c_str()); + + + // ----------------------------- move ctor PART ----------------------------- + const char* move_data{ "привет мир" }; + + utf8::ustring ustr_move{ move_data }; + auto other_ustr_move{ std::move(ustr_move) }; + + ASSERT_EQ(other_ustr_move.get_c_str(), move_data); + ASSERT_NE(ustr_move.get_c_str(), move_data); + + + // ----------------------------- copy ctor PART ----------------------------- + const char* copy_data{ "хеллоу ворлд" }; + + utf8::ustring copy_data_first{ copy_data }; + utf8::ustring copy_data_second{ copy_data_first }; + + ASSERT_EQ(copy_data, copy_data_first.get_c_str()); + ASSERT_EQ(copy_data, copy_data_second.get_c_str()); + ASSERT_EQ(copy_data_first, copy_data_second); + + // TODO: write for other ctors .... +} + +// --------------------------------------------------------------------------------------------------------------------- +TEST(utf8_string, operators) +{ + // ----------------------------- move operator PART ----------------------------- + const char* move_data{ "hello world" }; + + utf8::ustring ustr_move{ move_data }; + utf8::ustring other_ustr_move; + other_ustr_move = std::move(ustr_move); + + ASSERT_EQ(other_ustr_move.get_c_str(), move_data); + ASSERT_NE(ustr_move.get_c_str(), move_data); + + + // ----------------------------- copy operator PART ----------------------------- + const char* copy_data{ "хеллоу ворлд" }; + + utf8::ustring copy_data_first{ copy_data }; + utf8::ustring copy_data_second; + copy_data_second = copy_data_first; + + ASSERT_EQ(copy_data, copy_data_first.get_c_str()); + ASSERT_EQ(copy_data, copy_data_second.get_c_str()); + ASSERT_EQ(copy_data_first, copy_data_second); } \ No newline at end of file diff --git a/utf8-string-lib/utf8_string.cpp b/utf8-string-lib/utf8_string.cpp index 67e53fb..8aa1247 100644 --- a/utf8-string-lib/utf8_string.cpp +++ b/utf8-string-lib/utf8_string.cpp @@ -14,34 +14,88 @@ namespace utf8 std::istream& operator>>(std::istream& is, ustring& ustr) { // TODO: ............................................. + throw std::runtime_error("operator >> for utf8::ustring not implemented yet !"); } // ----------------------------------------------------------------------------------------------------------------- - constexpr ustring::ustring() + ustring::ustring() { Symbols_Info.reserve(100); String.reserve(100); } // ----------------------------------------------------------------------------------------------------------------- - ustring::ustring(const char* utf8_str) + ustring::ustring(const std::size_t bytes_reserve) { - Symbols_Info.reserve(100); - String.reserve(100); + Symbols_Info.reserve(bytes_reserve); + String.reserve(bytes_reserve); + } - // TODO: redo this part - std::size_t next_symbol_position{}; // next symbol position in bytes - for (std::size_t i{}; utf8_str[i] != '\0'; ++i) - { - String.push_back(utf8_str[i]); - if (next_symbol_position == i) // if it`s beginning of a symbol - { - std::uint8_t symbol_size{ Get_Size_Of_Symbol(utf8_str[i]) }; - Symbols_Info.push_back(SSymbol_Info{ symbol_size, next_symbol_position }); + // ----------------------------------------------------------------------------------------------------------------- + ustring::ustring(const char* default_string) : ustring(100) + { + Copy_With_Metadata_To_Internal_String(default_string); + } - next_symbol_position += symbol_size; - } + // ----------------------------------------------------------------------------------------------------------------- + ustring::ustring(const char* default_string, const std::size_t bytes_reserve) : ustring(bytes_reserve) + { + Copy_With_Metadata_To_Internal_String(default_string); + } + + // ----------------------------------------------------------------------------------------------------------------- + ustring::ustring(const std::basic_string& default_string) : ustring(default_string.c_str(), default_string.size() * 2) + { } + + // ----------------------------------------------------------------------------------------------------------------- + ustring::ustring(const std::basic_string_view& default_string) : ustring(default_string.data(), default_string.size() * 2) + { } + + // ----------------------------------------------------------------------------------------------------------------- + ustring::ustring(const ustring& other) + { + String = other.String; + Symbols_Info = other.Symbols_Info; + } + + // ----------------------------------------------------------------------------------------------------------------- + ustring::ustring(ustring&& other) noexcept + { + String = std::move(other.String); + Symbols_Info = std::move(other.Symbols_Info); + } + + // ----------------------------------------------------------------------------------------------------------------- + ustring& ustring::operator=(const ustring& other) + { + if (this != &other) + { + String = other.String; + Symbols_Info = other.Symbols_Info; } + + return *this; + } + + // ----------------------------------------------------------------------------------------------------------------- + ustring& ustring::operator=(ustring&& other) noexcept + { + String = std::move(other.String); + Symbols_Info = std::move(other.Symbols_Info); + + return *this; + } + + // ----------------------------------------------------------------------------------------------------------------- + bool ustring::operator==(const ustring& other) const + { + return (String == other.String); + } + + // ----------------------------------------------------------------------------------------------------------------- + bool ustring::operator!=(const ustring& other) const + { + return (String != other.String); } // ----------------------------------------------------------------------------------------------------------------- @@ -80,6 +134,12 @@ namespace utf8 return String; } + // ----------------------------------------------------------------------------------------------------------------- + const std::basic_string& ustring::get_c_str() const + { + return String; + } + // ----------------------------------------------------------------------------------------------------------------- std::uint8_t ustring::Get_Size_Of_Symbol(std::uint8_t symbol) { @@ -111,4 +171,21 @@ namespace utf8 Symbols_Info[i].Symbol_Offset += new_offset; } } + + // ----------------------------------------------------------------------------------------------------------------- + void ustring::Copy_With_Metadata_To_Internal_String(const char* default_string) + { + std::size_t next_symbol_position{}; // next symbol position in bytes + for (std::size_t i{}; default_string[i] != '\0'; ++i) + { + String.push_back(default_string[i]); + if (next_symbol_position == i) // if it`s beginning of a symbol + { + std::uint8_t symbol_size{ Get_Size_Of_Symbol(default_string[i]) }; + Symbols_Info.push_back(SSymbol_Info{ symbol_size, next_symbol_position }); + + next_symbol_position += symbol_size; + } + } + } } \ No newline at end of file diff --git a/utf8-string-lib/utf8_string.hpp b/utf8-string-lib/utf8_string.hpp index 9b3edc2..7d620e1 100644 --- a/utf8-string-lib/utf8_string.hpp +++ b/utf8-string-lib/utf8_string.hpp @@ -5,6 +5,7 @@ #include #include #include +#include #define ONE_BYTE 0b0000'0000 // first bit 0 #define TWO_BYTE 0b1100'0000 // first three bits 110 @@ -20,15 +21,37 @@ namespace utf8 friend std::istream& operator>>(std::istream& is, ustring& ustr); public: - constexpr ustring(); - explicit ustring(const char* utf8_str); + ustring(); + explicit ustring(const std::size_t bytes_reserve); + explicit ustring(const char* default_string); + explicit ustring(const char* default_string, const std::size_t bytes_reserve); + explicit ustring(const std::basic_string& default_string); + explicit ustring(const std::basic_string_view& default_string); + + constexpr ~ustring() = default; + + explicit ustring(const ustring& other); + explicit ustring(ustring&& other) noexcept; + + ustring& operator=(const ustring& other); + ustring& operator=(ustring&& other) noexcept; + + bool operator==(const ustring& other) const; + bool operator!=(const ustring& other) const; + + // TODO: ............................................... + char& operator[](std::size_t index) = delete; + const char& operator[](std::size_t index) const = delete; void replace_char(std::basic_string_view new_symbol, std::size_t idx); + std::basic_string& get_str(); + [[nodiscard]] const std::basic_string& get_c_str() const; private: std::uint8_t Get_Size_Of_Symbol(std::uint8_t symbol); void Recalculate_Symbols_Offset(std::size_t start_pos, std::ptrdiff_t new_offset); + void Copy_With_Metadata_To_Internal_String(const char* default_string); private: struct SSymbol_Info