| 1 | #include "duckdb/common/types.hpp" |
| 2 | |
| 3 | #include "duckdb/common/exception.hpp" |
| 4 | #include "duckdb/common/serializer.hpp" |
| 5 | #include "duckdb/common/string_util.hpp" |
| 6 | #include "duckdb/common/types/string_type.hpp" |
| 7 | |
| 8 | #include <cmath> |
| 9 | |
| 10 | using namespace std; |
| 11 | |
| 12 | namespace duckdb { |
| 13 | |
| 14 | const SQLType SQLType::SQLNULL = SQLType(SQLTypeId::SQLNULL); |
| 15 | const SQLType SQLType::BOOLEAN = SQLType(SQLTypeId::BOOLEAN); |
| 16 | const SQLType SQLType::TINYINT = SQLType(SQLTypeId::TINYINT); |
| 17 | const SQLType SQLType::SMALLINT = SQLType(SQLTypeId::SMALLINT); |
| 18 | const SQLType SQLType::INTEGER = SQLType(SQLTypeId::INTEGER); |
| 19 | const SQLType SQLType::BIGINT = SQLType(SQLTypeId::BIGINT); |
| 20 | const SQLType SQLType::FLOAT = SQLType(SQLTypeId::FLOAT); |
| 21 | const SQLType SQLType::DOUBLE = SQLType(SQLTypeId::DOUBLE); |
| 22 | const SQLType SQLType::DATE = SQLType(SQLTypeId::DATE); |
| 23 | const SQLType SQLType::TIMESTAMP = SQLType(SQLTypeId::TIMESTAMP); |
| 24 | const SQLType SQLType::TIME = SQLType(SQLTypeId::TIME); |
| 25 | |
| 26 | const SQLType SQLType::VARCHAR = SQLType(SQLTypeId::VARCHAR); |
| 27 | const SQLType SQLType::VARBINARY = SQLType(SQLTypeId::VARBINARY); |
| 28 | |
| 29 | const SQLType SQLType::BLOB = SQLType(SQLTypeId::BLOB); |
| 30 | |
| 31 | // TODO these are incomplete and should maybe not exist as such |
| 32 | const SQLType SQLType::STRUCT = SQLType(SQLTypeId::STRUCT); |
| 33 | const SQLType SQLType::LIST = SQLType(SQLTypeId::LIST); |
| 34 | |
| 35 | const SQLType SQLType::ANY = SQLType(SQLTypeId::ANY); |
| 36 | |
| 37 | const vector<SQLType> SQLType::NUMERIC = { |
| 38 | SQLType::TINYINT, SQLType::SMALLINT, SQLType::INTEGER, SQLType::BIGINT, |
| 39 | SQLType::FLOAT, SQLType::DOUBLE, SQLType(SQLTypeId::DECIMAL)}; |
| 40 | |
| 41 | const vector<SQLType> SQLType::INTEGRAL = {SQLType::TINYINT, SQLType::SMALLINT, SQLType::INTEGER, SQLType::BIGINT}; |
| 42 | |
| 43 | const vector<SQLType> SQLType::ALL_TYPES = { |
| 44 | SQLType::BOOLEAN, SQLType::TINYINT, SQLType::SMALLINT, SQLType::INTEGER, SQLType::BIGINT, |
| 45 | SQLType::DATE, SQLType::TIMESTAMP, SQLType::DOUBLE, SQLType::FLOAT, SQLType(SQLTypeId::DECIMAL), |
| 46 | SQLType::VARCHAR, SQLType::BLOB}; |
| 47 | // TODO add LIST/STRUCT here |
| 48 | |
| 49 | const TypeId ROW_TYPE = TypeId::INT64; |
| 50 | |
| 51 | string TypeIdToString(TypeId type) { |
| 52 | switch (type) { |
| 53 | case TypeId::BOOL: |
| 54 | return "BOOL" ; |
| 55 | case TypeId::INT8: |
| 56 | return "INT8" ; |
| 57 | case TypeId::INT16: |
| 58 | return "INT16" ; |
| 59 | case TypeId::INT32: |
| 60 | return "INT32" ; |
| 61 | case TypeId::INT64: |
| 62 | return "INT64" ; |
| 63 | case TypeId::HASH: |
| 64 | return "HASH" ; |
| 65 | case TypeId::POINTER: |
| 66 | return "POINTER" ; |
| 67 | case TypeId::FLOAT: |
| 68 | return "FLOAT" ; |
| 69 | case TypeId::DOUBLE: |
| 70 | return "DOUBLE" ; |
| 71 | case TypeId::VARCHAR: |
| 72 | return "VARCHAR" ; |
| 73 | case TypeId::VARBINARY: |
| 74 | return "VARBINARY" ; |
| 75 | case TypeId::STRUCT: |
| 76 | return "STRUCT<?>" ; |
| 77 | case TypeId::LIST: |
| 78 | return "LIST<?>" ; |
| 79 | default: |
| 80 | throw ConversionException("Invalid TypeId %d" , type); |
| 81 | } |
| 82 | } |
| 83 | |
| 84 | idx_t GetTypeIdSize(TypeId type) { |
| 85 | switch (type) { |
| 86 | case TypeId::BOOL: |
| 87 | return sizeof(bool); |
| 88 | case TypeId::INT8: |
| 89 | return sizeof(int8_t); |
| 90 | case TypeId::INT16: |
| 91 | return sizeof(int16_t); |
| 92 | case TypeId::INT32: |
| 93 | return sizeof(int32_t); |
| 94 | case TypeId::INT64: |
| 95 | return sizeof(int64_t); |
| 96 | case TypeId::FLOAT: |
| 97 | return sizeof(float); |
| 98 | case TypeId::DOUBLE: |
| 99 | return sizeof(double); |
| 100 | case TypeId::HASH: |
| 101 | return sizeof(hash_t); |
| 102 | case TypeId::POINTER: |
| 103 | return sizeof(uintptr_t); |
| 104 | case TypeId::VARCHAR: |
| 105 | return sizeof(string_t); |
| 106 | case TypeId::STRUCT: |
| 107 | return 0; // no own payload |
| 108 | case TypeId::LIST: |
| 109 | return 16; // offset + len |
| 110 | case TypeId::VARBINARY: |
| 111 | return sizeof(blob_t); |
| 112 | default: |
| 113 | throw ConversionException("Invalid TypeId %d" , type); |
| 114 | } |
| 115 | } |
| 116 | |
| 117 | SQLType SQLTypeFromInternalType(TypeId type) { |
| 118 | switch (type) { |
| 119 | case TypeId::BOOL: |
| 120 | return SQLType(SQLTypeId::BOOLEAN); |
| 121 | case TypeId::INT8: |
| 122 | return SQLType::TINYINT; |
| 123 | case TypeId::INT16: |
| 124 | return SQLType::SMALLINT; |
| 125 | case TypeId::INT32: |
| 126 | return SQLType::INTEGER; |
| 127 | case TypeId::INT64: |
| 128 | return SQLType::BIGINT; |
| 129 | case TypeId::FLOAT: |
| 130 | return SQLType::FLOAT; |
| 131 | case TypeId::DOUBLE: |
| 132 | return SQLType::DOUBLE; |
| 133 | case TypeId::VARCHAR: |
| 134 | return SQLType::VARCHAR; |
| 135 | case TypeId::VARBINARY: |
| 136 | return SQLType(SQLTypeId::VARBINARY); |
| 137 | case TypeId::STRUCT: |
| 138 | return SQLType(SQLTypeId::STRUCT); // TODO we do not know the child types here |
| 139 | case TypeId::LIST: |
| 140 | return SQLType(SQLTypeId::LIST); |
| 141 | default: |
| 142 | throw ConversionException("Invalid TypeId %d" , type); |
| 143 | } |
| 144 | } |
| 145 | |
| 146 | bool TypeIsConstantSize(TypeId type) { |
| 147 | return (type >= TypeId::BOOL && type <= TypeId::DOUBLE) || |
| 148 | (type >= TypeId::FIXED_SIZE_BINARY && type <= TypeId::DECIMAL) || type == TypeId::HASH || |
| 149 | type == TypeId::POINTER; |
| 150 | } |
| 151 | bool TypeIsIntegral(TypeId type) { |
| 152 | return (type >= TypeId::UINT8 && type <= TypeId::INT64) || type == TypeId::HASH || type == TypeId::POINTER; |
| 153 | } |
| 154 | bool TypeIsNumeric(TypeId type) { |
| 155 | return type >= TypeId::UINT8 && type <= TypeId::DOUBLE; |
| 156 | } |
| 157 | bool TypeIsInteger(TypeId type) { |
| 158 | return type >= TypeId::UINT8 && type <= TypeId::INT64; |
| 159 | } |
| 160 | |
| 161 | void SQLType::Serialize(Serializer &serializer) { |
| 162 | serializer.Write(id); |
| 163 | serializer.Write(width); |
| 164 | serializer.Write(scale); |
| 165 | serializer.WriteString(collation); |
| 166 | } |
| 167 | |
| 168 | SQLType SQLType::Deserialize(Deserializer &source) { |
| 169 | auto id = source.Read<SQLTypeId>(); |
| 170 | auto width = source.Read<uint16_t>(); |
| 171 | auto scale = source.Read<uint8_t>(); |
| 172 | auto collation = source.Read<string>(); |
| 173 | return SQLType(id, width, scale, collation); |
| 174 | } |
| 175 | |
| 176 | string SQLTypeIdToString(SQLTypeId id) { |
| 177 | switch (id) { |
| 178 | case SQLTypeId::BOOLEAN: |
| 179 | return "BOOLEAN" ; |
| 180 | case SQLTypeId::TINYINT: |
| 181 | return "TINYINT" ; |
| 182 | case SQLTypeId::SMALLINT: |
| 183 | return "SMALLINT" ; |
| 184 | case SQLTypeId::INTEGER: |
| 185 | return "INTEGER" ; |
| 186 | case SQLTypeId::BIGINT: |
| 187 | return "BIGINT" ; |
| 188 | case SQLTypeId::DATE: |
| 189 | return "DATE" ; |
| 190 | case SQLTypeId::TIME: |
| 191 | return "TIME" ; |
| 192 | case SQLTypeId::TIMESTAMP: |
| 193 | return "TIMESTAMP" ; |
| 194 | case SQLTypeId::FLOAT: |
| 195 | return "FLOAT" ; |
| 196 | case SQLTypeId::DOUBLE: |
| 197 | return "DOUBLE" ; |
| 198 | case SQLTypeId::DECIMAL: |
| 199 | return "DECIMAL" ; |
| 200 | case SQLTypeId::VARCHAR: |
| 201 | return "VARCHAR" ; |
| 202 | case SQLTypeId::BLOB: |
| 203 | return "BLOB" ; |
| 204 | case SQLTypeId::VARBINARY: |
| 205 | return "VARBINARY" ; |
| 206 | case SQLTypeId::CHAR: |
| 207 | return "CHAR" ; |
| 208 | case SQLTypeId::SQLNULL: |
| 209 | return "NULL" ; |
| 210 | case SQLTypeId::ANY: |
| 211 | return "ANY" ; |
| 212 | case SQLTypeId::STRUCT: |
| 213 | return "STRUCT<?>" ; |
| 214 | case SQLTypeId::LIST: |
| 215 | return "LIST<?>" ; |
| 216 | default: |
| 217 | return "INVALID" ; |
| 218 | } |
| 219 | } |
| 220 | |
| 221 | string SQLTypeToString(SQLType type) { |
| 222 | // FIXME: display width/scale |
| 223 | switch (type.id) { |
| 224 | case SQLTypeId::STRUCT: { |
| 225 | string ret = "STRUCT<" ; |
| 226 | for (size_t i = 0; i < type.child_type.size(); i++) { |
| 227 | ret += type.child_type[i].first + ": " + SQLTypeToString(type.child_type[i].second); |
| 228 | if (i < type.child_type.size() - 1) { |
| 229 | ret += ", " ; |
| 230 | } |
| 231 | } |
| 232 | ret += ">" ; |
| 233 | return ret; |
| 234 | } |
| 235 | case SQLTypeId::LIST: { |
| 236 | if (type.child_type.size() == 0) { |
| 237 | return "LIST<?>" ; |
| 238 | } |
| 239 | if (type.child_type.size() != 1) { |
| 240 | throw Exception("List needs a single child element" ); |
| 241 | } |
| 242 | return "LIST<" + SQLTypeToString(type.child_type[0].second) + ">" ; |
| 243 | } |
| 244 | default: |
| 245 | return SQLTypeIdToString(type.id); |
| 246 | } |
| 247 | } |
| 248 | |
| 249 | SQLType TransformStringToSQLType(string str) { |
| 250 | auto lower_str = StringUtil::Lower(str); |
| 251 | // Transform column type |
| 252 | if (lower_str == "int" || lower_str == "int4" || lower_str == "signed" || lower_str == "integer" || |
| 253 | lower_str == "integral" || lower_str == "int32" ) { |
| 254 | return SQLType::INTEGER; |
| 255 | } else if (lower_str == "varchar" || lower_str == "bpchar" || lower_str == "text" || lower_str == "string" || |
| 256 | lower_str == "char" ) { |
| 257 | return SQLType::VARCHAR; |
| 258 | } else if (lower_str == "bytea" || lower_str == "blob" ) { |
| 259 | return SQLType::BLOB; |
| 260 | } else if (lower_str == "int8" || lower_str == "bigint" || lower_str == "int64" || lower_str == "long" ) { |
| 261 | return SQLType::BIGINT; |
| 262 | } else if (lower_str == "int2" || lower_str == "smallint" || lower_str == "short" || lower_str == "int16" ) { |
| 263 | return SQLType::SMALLINT; |
| 264 | } else if (lower_str == "timestamp" || lower_str == "datetime" ) { |
| 265 | return SQLType::TIMESTAMP; |
| 266 | } else if (lower_str == "bool" || lower_str == "boolean" || lower_str == "logical" ) { |
| 267 | return SQLType(SQLTypeId::BOOLEAN); |
| 268 | } else if (lower_str == "real" || lower_str == "float4" || lower_str == "float" ) { |
| 269 | return SQLType::FLOAT; |
| 270 | } else if (lower_str == "double" || lower_str == "numeric" || lower_str == "float8" ) { |
| 271 | return SQLType::DOUBLE; |
| 272 | } else if (lower_str == "tinyint" || lower_str == "int1" ) { |
| 273 | return SQLType::TINYINT; |
| 274 | } else if (lower_str == "varbinary" ) { |
| 275 | return SQLType(SQLTypeId::VARBINARY); |
| 276 | } else if (lower_str == "date" ) { |
| 277 | return SQLType::DATE; |
| 278 | } else if (lower_str == "time" ) { |
| 279 | return SQLType::TIME; |
| 280 | } else { |
| 281 | throw NotImplementedException("DataType %s not supported yet...\n" , str.c_str()); |
| 282 | } |
| 283 | } |
| 284 | |
| 285 | bool SQLType::IsIntegral() const { |
| 286 | switch (id) { |
| 287 | case SQLTypeId::TINYINT: |
| 288 | case SQLTypeId::SMALLINT: |
| 289 | case SQLTypeId::INTEGER: |
| 290 | case SQLTypeId::BIGINT: |
| 291 | return true; |
| 292 | default: |
| 293 | return false; |
| 294 | } |
| 295 | } |
| 296 | |
| 297 | bool SQLType::IsNumeric() const { |
| 298 | switch (id) { |
| 299 | case SQLTypeId::TINYINT: |
| 300 | case SQLTypeId::SMALLINT: |
| 301 | case SQLTypeId::INTEGER: |
| 302 | case SQLTypeId::BIGINT: |
| 303 | case SQLTypeId::FLOAT: |
| 304 | case SQLTypeId::DOUBLE: |
| 305 | case SQLTypeId::DECIMAL: |
| 306 | return true; |
| 307 | default: |
| 308 | return false; |
| 309 | } |
| 310 | } |
| 311 | |
| 312 | bool SQLType::IsMoreGenericThan(SQLType &other) const { |
| 313 | if (other.id == id) { |
| 314 | return false; |
| 315 | } |
| 316 | |
| 317 | if (other.id == SQLTypeId::SQLNULL) { |
| 318 | return true; |
| 319 | } |
| 320 | |
| 321 | switch (id) { |
| 322 | case SQLTypeId::SMALLINT: |
| 323 | switch (other.id) { |
| 324 | case SQLTypeId::TINYINT: |
| 325 | return true; |
| 326 | default: |
| 327 | return false; |
| 328 | } |
| 329 | case SQLTypeId::INTEGER: |
| 330 | switch (other.id) { |
| 331 | case SQLTypeId::TINYINT: |
| 332 | case SQLTypeId::SMALLINT: |
| 333 | return true; |
| 334 | default: |
| 335 | return false; |
| 336 | } |
| 337 | case SQLTypeId::BIGINT: |
| 338 | switch (other.id) { |
| 339 | case SQLTypeId::TINYINT: |
| 340 | case SQLTypeId::SMALLINT: |
| 341 | case SQLTypeId::INTEGER: |
| 342 | return true; |
| 343 | default: |
| 344 | return false; |
| 345 | } |
| 346 | case SQLTypeId::DOUBLE: |
| 347 | switch (other.id) { |
| 348 | case SQLTypeId::TINYINT: |
| 349 | case SQLTypeId::SMALLINT: |
| 350 | case SQLTypeId::INTEGER: |
| 351 | case SQLTypeId::BIGINT: |
| 352 | return true; |
| 353 | default: |
| 354 | return false; |
| 355 | } |
| 356 | return false; |
| 357 | case SQLTypeId::DATE: |
| 358 | return false; |
| 359 | case SQLTypeId::TIMESTAMP: |
| 360 | switch (other.id) { |
| 361 | case SQLTypeId::TIME: |
| 362 | case SQLTypeId::DATE: |
| 363 | return true; |
| 364 | default: |
| 365 | return false; |
| 366 | } |
| 367 | case SQLTypeId::VARCHAR: |
| 368 | return true; |
| 369 | default: |
| 370 | return false; |
| 371 | } |
| 372 | |
| 373 | return true; |
| 374 | } |
| 375 | |
| 376 | TypeId GetInternalType(SQLType type) { |
| 377 | switch (type.id) { |
| 378 | case SQLTypeId::BOOLEAN: |
| 379 | return TypeId::BOOL; |
| 380 | case SQLTypeId::TINYINT: |
| 381 | return TypeId::INT8; |
| 382 | case SQLTypeId::SMALLINT: |
| 383 | return TypeId::INT16; |
| 384 | case SQLTypeId::SQLNULL: |
| 385 | case SQLTypeId::DATE: |
| 386 | case SQLTypeId::TIME: |
| 387 | case SQLTypeId::INTEGER: |
| 388 | return TypeId::INT32; |
| 389 | case SQLTypeId::BIGINT: |
| 390 | case SQLTypeId::TIMESTAMP: |
| 391 | return TypeId::INT64; |
| 392 | case SQLTypeId::FLOAT: |
| 393 | return TypeId::FLOAT; |
| 394 | case SQLTypeId::DOUBLE: |
| 395 | return TypeId::DOUBLE; |
| 396 | case SQLTypeId::DECIMAL: |
| 397 | // FIXME: for now |
| 398 | return TypeId::DOUBLE; |
| 399 | case SQLTypeId::VARCHAR: |
| 400 | case SQLTypeId::CHAR: |
| 401 | case SQLTypeId::BLOB: |
| 402 | return TypeId::VARCHAR; |
| 403 | case SQLTypeId::VARBINARY: |
| 404 | return TypeId::VARBINARY; |
| 405 | case SQLTypeId::STRUCT: |
| 406 | return TypeId::STRUCT; |
| 407 | case SQLTypeId::LIST: |
| 408 | return TypeId::LIST; |
| 409 | case SQLTypeId::ANY: |
| 410 | return TypeId::INVALID; |
| 411 | default: |
| 412 | throw ConversionException("Invalid SQLType %s" , SQLTypeToString(type).c_str()); |
| 413 | } |
| 414 | } |
| 415 | |
| 416 | SQLType MaxSQLType(SQLType left, SQLType right) { |
| 417 | if (left.id < right.id) { |
| 418 | return right; |
| 419 | } else if (right.id < left.id) { |
| 420 | return left; |
| 421 | } else if (left.width > right.width || left.collation > right.collation) { |
| 422 | return left; |
| 423 | } else { |
| 424 | return right; |
| 425 | } |
| 426 | } |
| 427 | |
| 428 | bool ApproxEqual(float ldecimal, float rdecimal) { |
| 429 | float epsilon = fabs(rdecimal) * 0.01; |
| 430 | return fabs(ldecimal - rdecimal) <= epsilon; |
| 431 | } |
| 432 | |
| 433 | bool ApproxEqual(double ldecimal, double rdecimal) { |
| 434 | double epsilon = fabs(rdecimal) * 0.01; |
| 435 | return fabs(ldecimal - rdecimal) <= epsilon; |
| 436 | } |
| 437 | |
| 438 | } // namespace duckdb |
| 439 | |