| 1 | // SPDX-License-Identifier: GPL-3.0-or-later |
| 2 | #pragma once |
| 3 | |
| 4 | #include <cstddef> |
| 5 | #include <mos/default_allocator.hpp> |
| 6 | #include <mos/string_view.hpp> |
| 7 | #include <mos_stdlib.hpp> |
| 8 | #include <mos_string.hpp> |
| 9 | |
| 10 | namespace mos |
| 11 | { |
| 12 | constexpr size_t short_string_capacity = 16; |
| 13 | |
| 14 | template<typename Char, typename TAllocator = mos::default_allocator<Char>> |
| 15 | class basic_string |
| 16 | { |
| 17 | bool _is_long = false; |
| 18 | size_t _length; ///< string length EXcluding null terminator |
| 19 | |
| 20 | union |
| 21 | { |
| 22 | struct _long_buffer |
| 23 | { |
| 24 | Char *_buffer; |
| 25 | size_t _capacity; ///< buffer capacity includes null terminator |
| 26 | } _long; |
| 27 | |
| 28 | struct _short_buffer |
| 29 | { |
| 30 | Char _buffer[short_string_capacity]; |
| 31 | } _short; |
| 32 | } _data; |
| 33 | |
| 34 | public: |
| 35 | static constexpr auto npos = size_t(-1); |
| 36 | |
| 37 | public: |
| 38 | basic_string(std::nullptr_t) = delete; |
| 39 | |
| 40 | basic_string() : _length{ 0 }, _is_long(false) {}; |
| 41 | |
| 42 | basic_string(const Char *buffer) |
| 43 | { |
| 44 | _init(buffer, size: generic_strlen(buffer)); |
| 45 | } |
| 46 | |
| 47 | explicit basic_string(const Char *buffer, size_t size) |
| 48 | { |
| 49 | _init(buffer, size); |
| 50 | } |
| 51 | |
| 52 | template<class StringViewLike> |
| 53 | explicit basic_string(const StringViewLike &view) |
| 54 | { |
| 55 | _init(buffer: view.data(), size: view.size()); |
| 56 | } |
| 57 | |
| 58 | basic_string(size_t size, Char c = 0) |
| 59 | { |
| 60 | _init(buffer: &c, size); |
| 61 | } |
| 62 | |
| 63 | basic_string(const basic_string &other) |
| 64 | { |
| 65 | _init(buffer: other.data(), size: other.size()); |
| 66 | } |
| 67 | |
| 68 | basic_string(basic_string &&other) |
| 69 | { |
| 70 | _length = other._length; |
| 71 | _is_long = other._is_long; |
| 72 | if (_is_long) |
| 73 | { |
| 74 | _data._long._capacity = other._data._long._capacity; |
| 75 | _data._long._buffer = other._data._long._buffer; |
| 76 | other._data._long._buffer = nullptr; // detach |
| 77 | } |
| 78 | else |
| 79 | { |
| 80 | for (size_t i = 0; i < short_string_capacity; i++) |
| 81 | _data._short._buffer[i] = other._data._short._buffer[i]; |
| 82 | } |
| 83 | } |
| 84 | |
| 85 | ~basic_string() |
| 86 | { |
| 87 | if (_is_long && _data._long._buffer) |
| 88 | TAllocator::free(_data._long._buffer); |
| 89 | } |
| 90 | |
| 91 | basic_string &operator=(const Char *buffer) |
| 92 | { |
| 93 | if (buffer == nullptr) |
| 94 | mos::__raise_null_pointer_exception(); |
| 95 | this->~basic_string(); |
| 96 | this->_length = generic_strlen(buffer); |
| 97 | this->_is_long = this->_length >= short_string_capacity; |
| 98 | if (this->_is_long) |
| 99 | { |
| 100 | this->_data._long._capacity = this->_length + 1; |
| 101 | this->_data._long._buffer = (Char *) TAllocator::allocate(sizeof(Char) * this->_data._long._capacity); |
| 102 | memcpy(this->_data._long._buffer, buffer, sizeof(Char) * this->_length); |
| 103 | this->_data._long._buffer[this->_length] = 0; |
| 104 | } |
| 105 | else |
| 106 | { |
| 107 | for (size_t i = 0; i < short_string_capacity; i++) |
| 108 | this->_data._short._buffer[i] = buffer[i]; |
| 109 | } |
| 110 | return *this; |
| 111 | } |
| 112 | |
| 113 | basic_string &operator=(const mos::basic_string_view<Char> &view) |
| 114 | { |
| 115 | this->~basic_string(); |
| 116 | this->_length = view.size(); |
| 117 | this->_is_long = this->_length >= short_string_capacity; |
| 118 | if (this->_is_long) |
| 119 | { |
| 120 | this->_data._long._capacity = this->_length + 1; |
| 121 | this->_data._long._buffer = (Char *) TAllocator::allocate(sizeof(Char) * this->_data._long._capacity); |
| 122 | memcpy(this->_data._long._buffer, view.data(), sizeof(Char) * this->_length); |
| 123 | this->_data._long._buffer[this->_length] = 0; |
| 124 | } |
| 125 | else |
| 126 | { |
| 127 | for (size_t i = 0; i < short_string_capacity; i++) |
| 128 | this->_data._short._buffer[i] = view.data()[i]; |
| 129 | } |
| 130 | return *this; |
| 131 | } |
| 132 | |
| 133 | basic_string &operator=(basic_string &&other) |
| 134 | { |
| 135 | this->~basic_string(); |
| 136 | this->_length = other._length; |
| 137 | this->_is_long = other._is_long; |
| 138 | if (this->_is_long) |
| 139 | { |
| 140 | this->_data._long._capacity = other._data._long._capacity; |
| 141 | this->_data._long._buffer = other._data._long._buffer; |
| 142 | other._data._long._buffer = nullptr; // detach |
| 143 | } |
| 144 | else |
| 145 | { |
| 146 | for (size_t i = 0; i < short_string_capacity; i++) |
| 147 | this->_data._short._buffer[i] = other._data._short._buffer[i]; |
| 148 | } |
| 149 | return *this; |
| 150 | } |
| 151 | |
| 152 | basic_string &operator=(const basic_string &other) |
| 153 | { |
| 154 | _init(buffer: other.data(), size: other.size()); |
| 155 | return *this; |
| 156 | } |
| 157 | |
| 158 | friend basic_string operator+(const basic_string_view<Char> &lhs, const basic_string &rhs) |
| 159 | { |
| 160 | return basic_string(lhs) + rhs; |
| 161 | } |
| 162 | |
| 163 | void resize(size_t new_length) |
| 164 | { |
| 165 | if (new_length == _length) |
| 166 | return; |
| 167 | if (new_length < short_string_capacity) |
| 168 | _convert_to_short(); |
| 169 | else |
| 170 | _convert_to_long(new_length: _length); |
| 171 | _length = new_length; |
| 172 | } |
| 173 | |
| 174 | bool begins_with(Char c) const |
| 175 | { |
| 176 | if (_length == 0) |
| 177 | return false; |
| 178 | return _data._short._buffer[0] == c; |
| 179 | } |
| 180 | |
| 181 | bool begins_with(const basic_string_view<Char> &prefix) const |
| 182 | { |
| 183 | if (prefix.size() > _length) |
| 184 | return false; |
| 185 | return generic_strncmp(data(), prefix.data(), prefix.size()) == 0; |
| 186 | } |
| 187 | |
| 188 | bool ends_with(Char c) const |
| 189 | { |
| 190 | if (_length == 0) |
| 191 | return false; |
| 192 | return _data._short._buffer[_length - 1] == c; |
| 193 | } |
| 194 | |
| 195 | bool ends_with(const basic_string_view<Char> &suffix) const |
| 196 | { |
| 197 | if (suffix.size() > _length) |
| 198 | return false; |
| 199 | return generic_strncmp(data() + _length - suffix.size(), suffix.data(), suffix.size()) == 0; |
| 200 | } |
| 201 | |
| 202 | void clear() |
| 203 | { |
| 204 | if (_is_long) |
| 205 | { |
| 206 | TAllocator::free(_data._long._buffer); |
| 207 | _data._long._buffer = nullptr; |
| 208 | _is_long = false; |
| 209 | } |
| 210 | _length = 0; |
| 211 | } |
| 212 | |
| 213 | basic_string operator+(const basic_string &other) const |
| 214 | { |
| 215 | auto copy = *this; |
| 216 | return copy += other; |
| 217 | } |
| 218 | |
| 219 | basic_string operator+(Char c) const |
| 220 | { |
| 221 | return operator+(basic_string_view(&c, 1)); |
| 222 | } |
| 223 | |
| 224 | void push_back(Char c) |
| 225 | { |
| 226 | operator+=(c); |
| 227 | } |
| 228 | |
| 229 | basic_string &operator+=(const basic_string_view<Char> &other) |
| 230 | { |
| 231 | const auto new_length = _length + other.size(); |
| 232 | if (new_length < short_string_capacity) |
| 233 | { |
| 234 | memcpy(_data._short._buffer + _length, other.data(), sizeof(Char) * other.size()); |
| 235 | _data._short._buffer[new_length] = 0; |
| 236 | _length = new_length; |
| 237 | } |
| 238 | else |
| 239 | { |
| 240 | _convert_to_long(new_length); |
| 241 | memcpy(_data._long._buffer + _length, other.data(), sizeof(Char) * other.size()); |
| 242 | _data._long._buffer[new_length] = 0; |
| 243 | _length = new_length; |
| 244 | } |
| 245 | |
| 246 | return *this; |
| 247 | } |
| 248 | |
| 249 | basic_string &operator+=(Char c) |
| 250 | { |
| 251 | return operator+=(basic_string_view(&c, 1)); |
| 252 | } |
| 253 | |
| 254 | const Char *data() const |
| 255 | { |
| 256 | return _is_long ? _data._long._buffer : _data._short._buffer; |
| 257 | } |
| 258 | |
| 259 | const Char *c_str() const |
| 260 | { |
| 261 | return data(); |
| 262 | } |
| 263 | |
| 264 | Char &operator[](size_t index) |
| 265 | { |
| 266 | return _is_long ? _data._long._buffer[index] : _data._short._buffer[index]; |
| 267 | } |
| 268 | |
| 269 | const Char &operator[](size_t index) const |
| 270 | { |
| 271 | return _is_long ? _data._long._buffer[index] : _data._short._buffer[index]; |
| 272 | } |
| 273 | |
| 274 | size_t size() const |
| 275 | { |
| 276 | return _length; |
| 277 | } |
| 278 | |
| 279 | bool empty() const |
| 280 | { |
| 281 | return _length == 0; |
| 282 | } |
| 283 | |
| 284 | ssize_t copy(Char *buffer, size_t size, size_t pos = 0) const |
| 285 | { |
| 286 | if (pos > _length) |
| 287 | return -1; |
| 288 | if (pos + size > _length) |
| 289 | size = _length - pos; |
| 290 | memcpy(buffer, data() + pos, sizeof(Char) * size); |
| 291 | return size; |
| 292 | } |
| 293 | |
| 294 | basic_string_view<Char> value_or(basic_string_view<Char> other) const |
| 295 | { |
| 296 | return empty() ? other : *this; |
| 297 | } |
| 298 | |
| 299 | Char *begin() |
| 300 | { |
| 301 | return _is_long ? &_data._long._buffer[0] : &_data._short._buffer[0]; |
| 302 | } |
| 303 | const Char *begin() const |
| 304 | { |
| 305 | return _is_long ? &_data._long._buffer[0] : &_data._short._buffer[0]; |
| 306 | } |
| 307 | |
| 308 | Char *end() |
| 309 | { |
| 310 | return _is_long ? &_data._long._buffer[_length] : &_data._short._buffer[_length]; |
| 311 | } |
| 312 | const Char *end() const |
| 313 | { |
| 314 | return _is_long ? &_data._long._buffer[_length] : &_data._short._buffer[_length]; |
| 315 | } |
| 316 | |
| 317 | bool operator==(const basic_string &other) const |
| 318 | { |
| 319 | return _do_compare(other: other.data(), other_size: other.size()) == 0; |
| 320 | } |
| 321 | |
| 322 | bool operator==(const char *rhs) const |
| 323 | { |
| 324 | return _do_compare(other: rhs, other_size: generic_strlen(c: rhs)) == 0; |
| 325 | } |
| 326 | |
| 327 | operator basic_string_view<Char>() const |
| 328 | { |
| 329 | return basic_string_view<Char>(data(), _length); |
| 330 | } |
| 331 | |
| 332 | bool operator==(std::nullptr_t) const = delete; |
| 333 | |
| 334 | private: |
| 335 | void _init(const Char *buffer, size_t size) |
| 336 | { |
| 337 | if (buffer == nullptr) |
| 338 | mos::__raise_null_pointer_exception(); |
| 339 | _length = size; |
| 340 | if (_length < short_string_capacity) |
| 341 | { |
| 342 | for (size_t i = 0; i < _length; i++) |
| 343 | _data._short._buffer[i] = buffer[i]; |
| 344 | _data._short._buffer[_length] = 0; |
| 345 | _is_long = false; |
| 346 | } |
| 347 | else |
| 348 | { |
| 349 | _data._long._capacity = sizeof(Char) * _length + 1; |
| 350 | _data._long._buffer = (Char *) TAllocator::allocate(_data._long._capacity); |
| 351 | memcpy(_data._long._buffer, buffer, sizeof(Char) * size); |
| 352 | _data._long._buffer[size] = 0; |
| 353 | _is_long = true; |
| 354 | } |
| 355 | } |
| 356 | int _do_compare(const Char *other, const size_t other_size) const |
| 357 | { |
| 358 | if (_length != other_size) |
| 359 | return _length < other_size ? -1 : 1; |
| 360 | |
| 361 | return generic_strncmp(data(), other, _length); |
| 362 | } |
| 363 | |
| 364 | void _convert_to_long(size_t new_length = 0) |
| 365 | { |
| 366 | if (_is_long) |
| 367 | return; |
| 368 | |
| 369 | const auto new_capacity = sizeof(Char) * std::max(a: new_length, b: _length) + 1; //_length + 1 for null terminator |
| 370 | const auto new_buffer = (Char *) TAllocator::allocate(new_capacity); |
| 371 | memcpy(new_buffer, _data._short._buffer, std::min(a: short_string_capacity, b: _length) * sizeof(Char)); |
| 372 | new_buffer[_length] = 0; |
| 373 | |
| 374 | _data._long._buffer = new_buffer; |
| 375 | _data._long._capacity = new_capacity; |
| 376 | _is_long = true; |
| 377 | } |
| 378 | |
| 379 | void _convert_to_short() |
| 380 | { |
| 381 | if (!_is_long) |
| 382 | return; |
| 383 | |
| 384 | const auto buffer = _data._long._buffer; |
| 385 | const auto capacity = _data._long._capacity; |
| 386 | |
| 387 | memcpy(_data._short._buffer, buffer, std::min(a: short_string_capacity, b: _length) * sizeof(Char)); |
| 388 | _data._short._buffer[_length] = 0; |
| 389 | |
| 390 | TAllocator::free(buffer); |
| 391 | _is_long = false; |
| 392 | } |
| 393 | }; |
| 394 | |
| 395 | using string = mos::basic_string<char>; |
| 396 | |
| 397 | // convert a pointer to a string |
| 398 | mos::string to_string(const void *value); |
| 399 | } // namespace mos |
| 400 | |