| 1 | /*------------------------------------------------------------------------- |
| 2 | * |
| 3 | * numutils.c |
| 4 | * utility functions for I/O of built-in numeric types. |
| 5 | * |
| 6 | * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group |
| 7 | * Portions Copyright (c) 1994, Regents of the University of California |
| 8 | * |
| 9 | * |
| 10 | * IDENTIFICATION |
| 11 | * src/backend/utils/adt/numutils.c |
| 12 | * |
| 13 | *------------------------------------------------------------------------- |
| 14 | */ |
| 15 | #include "postgres.h" |
| 16 | |
| 17 | #include <math.h> |
| 18 | #include <limits.h> |
| 19 | #include <ctype.h> |
| 20 | |
| 21 | #include "common/int.h" |
| 22 | #include "utils/builtins.h" |
| 23 | |
| 24 | /* |
| 25 | * pg_atoi: convert string to integer |
| 26 | * |
| 27 | * allows any number of leading or trailing whitespace characters. |
| 28 | * |
| 29 | * 'size' is the sizeof() the desired integral result (1, 2, or 4 bytes). |
| 30 | * |
| 31 | * c, if not 0, is a terminator character that may appear after the |
| 32 | * integer (plus whitespace). If 0, the string must end after the integer. |
| 33 | * |
| 34 | * Unlike plain atoi(), this will throw ereport() upon bad input format or |
| 35 | * overflow. |
| 36 | */ |
| 37 | int32 |
| 38 | pg_atoi(const char *s, int size, int c) |
| 39 | { |
| 40 | long l; |
| 41 | char *badp; |
| 42 | |
| 43 | /* |
| 44 | * Some versions of strtol treat the empty string as an error, but some |
| 45 | * seem not to. Make an explicit test to be sure we catch it. |
| 46 | */ |
| 47 | if (s == NULL) |
| 48 | elog(ERROR, "NULL pointer" ); |
| 49 | if (*s == 0) |
| 50 | ereport(ERROR, |
| 51 | (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), |
| 52 | errmsg("invalid input syntax for type %s: \"%s\"" , |
| 53 | "integer" , s))); |
| 54 | |
| 55 | errno = 0; |
| 56 | l = strtol(s, &badp, 10); |
| 57 | |
| 58 | /* We made no progress parsing the string, so bail out */ |
| 59 | if (s == badp) |
| 60 | ereport(ERROR, |
| 61 | (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), |
| 62 | errmsg("invalid input syntax for type %s: \"%s\"" , |
| 63 | "integer" , s))); |
| 64 | |
| 65 | switch (size) |
| 66 | { |
| 67 | case sizeof(int32): |
| 68 | if (errno == ERANGE |
| 69 | #if defined(HAVE_LONG_INT_64) |
| 70 | /* won't get ERANGE on these with 64-bit longs... */ |
| 71 | || l < INT_MIN || l > INT_MAX |
| 72 | #endif |
| 73 | ) |
| 74 | ereport(ERROR, |
| 75 | (errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE), |
| 76 | errmsg("value \"%s\" is out of range for type %s" , s, |
| 77 | "integer" ))); |
| 78 | break; |
| 79 | case sizeof(int16): |
| 80 | if (errno == ERANGE || l < SHRT_MIN || l > SHRT_MAX) |
| 81 | ereport(ERROR, |
| 82 | (errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE), |
| 83 | errmsg("value \"%s\" is out of range for type %s" , s, |
| 84 | "smallint" ))); |
| 85 | break; |
| 86 | case sizeof(int8): |
| 87 | if (errno == ERANGE || l < SCHAR_MIN || l > SCHAR_MAX) |
| 88 | ereport(ERROR, |
| 89 | (errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE), |
| 90 | errmsg("value \"%s\" is out of range for 8-bit integer" , s))); |
| 91 | break; |
| 92 | default: |
| 93 | elog(ERROR, "unsupported result size: %d" , size); |
| 94 | } |
| 95 | |
| 96 | /* |
| 97 | * Skip any trailing whitespace; if anything but whitespace remains before |
| 98 | * the terminating character, bail out |
| 99 | */ |
| 100 | while (*badp && *badp != c && isspace((unsigned char) *badp)) |
| 101 | badp++; |
| 102 | |
| 103 | if (*badp && *badp != c) |
| 104 | ereport(ERROR, |
| 105 | (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), |
| 106 | errmsg("invalid input syntax for type %s: \"%s\"" , |
| 107 | "integer" , s))); |
| 108 | |
| 109 | return (int32) l; |
| 110 | } |
| 111 | |
| 112 | /* |
| 113 | * Convert input string to a signed 16 bit integer. |
| 114 | * |
| 115 | * Allows any number of leading or trailing whitespace characters. Will throw |
| 116 | * ereport() upon bad input format or overflow. |
| 117 | * |
| 118 | * NB: Accumulate input as a negative number, to deal with two's complement |
| 119 | * representation of the most negative number, which can't be represented as a |
| 120 | * positive number. |
| 121 | */ |
| 122 | int16 |
| 123 | pg_strtoint16(const char *s) |
| 124 | { |
| 125 | const char *ptr = s; |
| 126 | int16 tmp = 0; |
| 127 | bool neg = false; |
| 128 | |
| 129 | /* skip leading spaces */ |
| 130 | while (likely(*ptr) && isspace((unsigned char) *ptr)) |
| 131 | ptr++; |
| 132 | |
| 133 | /* handle sign */ |
| 134 | if (*ptr == '-') |
| 135 | { |
| 136 | ptr++; |
| 137 | neg = true; |
| 138 | } |
| 139 | else if (*ptr == '+') |
| 140 | ptr++; |
| 141 | |
| 142 | /* require at least one digit */ |
| 143 | if (unlikely(!isdigit((unsigned char) *ptr))) |
| 144 | goto invalid_syntax; |
| 145 | |
| 146 | /* process digits */ |
| 147 | while (*ptr && isdigit((unsigned char) *ptr)) |
| 148 | { |
| 149 | int8 digit = (*ptr++ - '0'); |
| 150 | |
| 151 | if (unlikely(pg_mul_s16_overflow(tmp, 10, &tmp)) || |
| 152 | unlikely(pg_sub_s16_overflow(tmp, digit, &tmp))) |
| 153 | goto out_of_range; |
| 154 | } |
| 155 | |
| 156 | /* allow trailing whitespace, but not other trailing chars */ |
| 157 | while (*ptr != '\0' && isspace((unsigned char) *ptr)) |
| 158 | ptr++; |
| 159 | |
| 160 | if (unlikely(*ptr != '\0')) |
| 161 | goto invalid_syntax; |
| 162 | |
| 163 | if (!neg) |
| 164 | { |
| 165 | /* could fail if input is most negative number */ |
| 166 | if (unlikely(tmp == PG_INT16_MIN)) |
| 167 | goto out_of_range; |
| 168 | tmp = -tmp; |
| 169 | } |
| 170 | |
| 171 | return tmp; |
| 172 | |
| 173 | out_of_range: |
| 174 | ereport(ERROR, |
| 175 | (errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE), |
| 176 | errmsg("value \"%s\" is out of range for type %s" , |
| 177 | s, "smallint" ))); |
| 178 | |
| 179 | invalid_syntax: |
| 180 | ereport(ERROR, |
| 181 | (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), |
| 182 | errmsg("invalid input syntax for type %s: \"%s\"" , |
| 183 | "smallint" , s))); |
| 184 | |
| 185 | return 0; /* keep compiler quiet */ |
| 186 | } |
| 187 | |
| 188 | /* |
| 189 | * Convert input string to a signed 32 bit integer. |
| 190 | * |
| 191 | * Allows any number of leading or trailing whitespace characters. Will throw |
| 192 | * ereport() upon bad input format or overflow. |
| 193 | * |
| 194 | * NB: Accumulate input as a negative number, to deal with two's complement |
| 195 | * representation of the most negative number, which can't be represented as a |
| 196 | * positive number. |
| 197 | */ |
| 198 | int32 |
| 199 | pg_strtoint32(const char *s) |
| 200 | { |
| 201 | const char *ptr = s; |
| 202 | int32 tmp = 0; |
| 203 | bool neg = false; |
| 204 | |
| 205 | /* skip leading spaces */ |
| 206 | while (likely(*ptr) && isspace((unsigned char) *ptr)) |
| 207 | ptr++; |
| 208 | |
| 209 | /* handle sign */ |
| 210 | if (*ptr == '-') |
| 211 | { |
| 212 | ptr++; |
| 213 | neg = true; |
| 214 | } |
| 215 | else if (*ptr == '+') |
| 216 | ptr++; |
| 217 | |
| 218 | /* require at least one digit */ |
| 219 | if (unlikely(!isdigit((unsigned char) *ptr))) |
| 220 | goto invalid_syntax; |
| 221 | |
| 222 | /* process digits */ |
| 223 | while (*ptr && isdigit((unsigned char) *ptr)) |
| 224 | { |
| 225 | int8 digit = (*ptr++ - '0'); |
| 226 | |
| 227 | if (unlikely(pg_mul_s32_overflow(tmp, 10, &tmp)) || |
| 228 | unlikely(pg_sub_s32_overflow(tmp, digit, &tmp))) |
| 229 | goto out_of_range; |
| 230 | } |
| 231 | |
| 232 | /* allow trailing whitespace, but not other trailing chars */ |
| 233 | while (*ptr != '\0' && isspace((unsigned char) *ptr)) |
| 234 | ptr++; |
| 235 | |
| 236 | if (unlikely(*ptr != '\0')) |
| 237 | goto invalid_syntax; |
| 238 | |
| 239 | if (!neg) |
| 240 | { |
| 241 | /* could fail if input is most negative number */ |
| 242 | if (unlikely(tmp == PG_INT32_MIN)) |
| 243 | goto out_of_range; |
| 244 | tmp = -tmp; |
| 245 | } |
| 246 | |
| 247 | return tmp; |
| 248 | |
| 249 | out_of_range: |
| 250 | ereport(ERROR, |
| 251 | (errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE), |
| 252 | errmsg("value \"%s\" is out of range for type %s" , |
| 253 | s, "integer" ))); |
| 254 | |
| 255 | invalid_syntax: |
| 256 | ereport(ERROR, |
| 257 | (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), |
| 258 | errmsg("invalid input syntax for type %s: \"%s\"" , |
| 259 | "integer" , s))); |
| 260 | |
| 261 | return 0; /* keep compiler quiet */ |
| 262 | } |
| 263 | |
| 264 | /* |
| 265 | * pg_itoa: converts a signed 16-bit integer to its string representation |
| 266 | * |
| 267 | * Caller must ensure that 'a' points to enough memory to hold the result |
| 268 | * (at least 7 bytes, counting a leading sign and trailing NUL). |
| 269 | * |
| 270 | * It doesn't seem worth implementing this separately. |
| 271 | */ |
| 272 | void |
| 273 | pg_itoa(int16 i, char *a) |
| 274 | { |
| 275 | pg_ltoa((int32) i, a); |
| 276 | } |
| 277 | |
| 278 | /* |
| 279 | * pg_ltoa: converts a signed 32-bit integer to its string representation |
| 280 | * |
| 281 | * Caller must ensure that 'a' points to enough memory to hold the result |
| 282 | * (at least 12 bytes, counting a leading sign and trailing NUL). |
| 283 | */ |
| 284 | void |
| 285 | pg_ltoa(int32 value, char *a) |
| 286 | { |
| 287 | char *start = a; |
| 288 | bool neg = false; |
| 289 | |
| 290 | /* |
| 291 | * Avoid problems with the most negative integer not being representable |
| 292 | * as a positive integer. |
| 293 | */ |
| 294 | if (value == PG_INT32_MIN) |
| 295 | { |
| 296 | memcpy(a, "-2147483648" , 12); |
| 297 | return; |
| 298 | } |
| 299 | else if (value < 0) |
| 300 | { |
| 301 | value = -value; |
| 302 | neg = true; |
| 303 | } |
| 304 | |
| 305 | /* Compute the result string backwards. */ |
| 306 | do |
| 307 | { |
| 308 | int32 remainder; |
| 309 | int32 oldval = value; |
| 310 | |
| 311 | value /= 10; |
| 312 | remainder = oldval - value * 10; |
| 313 | *a++ = '0' + remainder; |
| 314 | } while (value != 0); |
| 315 | |
| 316 | if (neg) |
| 317 | *a++ = '-'; |
| 318 | |
| 319 | /* Add trailing NUL byte, and back up 'a' to the last character. */ |
| 320 | *a-- = '\0'; |
| 321 | |
| 322 | /* Reverse string. */ |
| 323 | while (start < a) |
| 324 | { |
| 325 | char swap = *start; |
| 326 | |
| 327 | *start++ = *a; |
| 328 | *a-- = swap; |
| 329 | } |
| 330 | } |
| 331 | |
| 332 | /* |
| 333 | * pg_lltoa: convert a signed 64-bit integer to its string representation |
| 334 | * |
| 335 | * Caller must ensure that 'a' points to enough memory to hold the result |
| 336 | * (at least MAXINT8LEN+1 bytes, counting a leading sign and trailing NUL). |
| 337 | */ |
| 338 | void |
| 339 | pg_lltoa(int64 value, char *a) |
| 340 | { |
| 341 | char *start = a; |
| 342 | bool neg = false; |
| 343 | |
| 344 | /* |
| 345 | * Avoid problems with the most negative integer not being representable |
| 346 | * as a positive integer. |
| 347 | */ |
| 348 | if (value == PG_INT64_MIN) |
| 349 | { |
| 350 | memcpy(a, "-9223372036854775808" , 21); |
| 351 | return; |
| 352 | } |
| 353 | else if (value < 0) |
| 354 | { |
| 355 | value = -value; |
| 356 | neg = true; |
| 357 | } |
| 358 | |
| 359 | /* Compute the result string backwards. */ |
| 360 | do |
| 361 | { |
| 362 | int64 remainder; |
| 363 | int64 oldval = value; |
| 364 | |
| 365 | value /= 10; |
| 366 | remainder = oldval - value * 10; |
| 367 | *a++ = '0' + remainder; |
| 368 | } while (value != 0); |
| 369 | |
| 370 | if (neg) |
| 371 | *a++ = '-'; |
| 372 | |
| 373 | /* Add trailing NUL byte, and back up 'a' to the last character. */ |
| 374 | *a-- = '\0'; |
| 375 | |
| 376 | /* Reverse string. */ |
| 377 | while (start < a) |
| 378 | { |
| 379 | char swap = *start; |
| 380 | |
| 381 | *start++ = *a; |
| 382 | *a-- = swap; |
| 383 | } |
| 384 | } |
| 385 | |
| 386 | |
| 387 | /* |
| 388 | * pg_ltostr_zeropad |
| 389 | * Converts 'value' into a decimal string representation stored at 'str'. |
| 390 | * 'minwidth' specifies the minimum width of the result; any extra space |
| 391 | * is filled up by prefixing the number with zeros. |
| 392 | * |
| 393 | * Returns the ending address of the string result (the last character written |
| 394 | * plus 1). Note that no NUL terminator is written. |
| 395 | * |
| 396 | * The intended use-case for this function is to build strings that contain |
| 397 | * multiple individual numbers, for example: |
| 398 | * |
| 399 | * str = pg_ltostr_zeropad(str, hours, 2); |
| 400 | * *str++ = ':'; |
| 401 | * str = pg_ltostr_zeropad(str, mins, 2); |
| 402 | * *str++ = ':'; |
| 403 | * str = pg_ltostr_zeropad(str, secs, 2); |
| 404 | * *str = '\0'; |
| 405 | * |
| 406 | * Note: Caller must ensure that 'str' points to enough memory to hold the |
| 407 | * result. |
| 408 | */ |
| 409 | char * |
| 410 | pg_ltostr_zeropad(char *str, int32 value, int32 minwidth) |
| 411 | { |
| 412 | char *start = str; |
| 413 | char *end = &str[minwidth]; |
| 414 | int32 num = value; |
| 415 | |
| 416 | Assert(minwidth > 0); |
| 417 | |
| 418 | /* |
| 419 | * Handle negative numbers in a special way. We can't just write a '-' |
| 420 | * prefix and reverse the sign as that would overflow for INT32_MIN. |
| 421 | */ |
| 422 | if (num < 0) |
| 423 | { |
| 424 | *start++ = '-'; |
| 425 | minwidth--; |
| 426 | |
| 427 | /* |
| 428 | * Build the number starting at the last digit. Here remainder will |
| 429 | * be a negative number, so we must reverse the sign before adding '0' |
| 430 | * in order to get the correct ASCII digit. |
| 431 | */ |
| 432 | while (minwidth--) |
| 433 | { |
| 434 | int32 oldval = num; |
| 435 | int32 remainder; |
| 436 | |
| 437 | num /= 10; |
| 438 | remainder = oldval - num * 10; |
| 439 | start[minwidth] = '0' - remainder; |
| 440 | } |
| 441 | } |
| 442 | else |
| 443 | { |
| 444 | /* Build the number starting at the last digit */ |
| 445 | while (minwidth--) |
| 446 | { |
| 447 | int32 oldval = num; |
| 448 | int32 remainder; |
| 449 | |
| 450 | num /= 10; |
| 451 | remainder = oldval - num * 10; |
| 452 | start[minwidth] = '0' + remainder; |
| 453 | } |
| 454 | } |
| 455 | |
| 456 | /* |
| 457 | * If minwidth was not high enough to fit the number then num won't have |
| 458 | * been divided down to zero. We punt the problem to pg_ltostr(), which |
| 459 | * will generate a correct answer in the minimum valid width. |
| 460 | */ |
| 461 | if (num != 0) |
| 462 | return pg_ltostr(str, value); |
| 463 | |
| 464 | /* Otherwise, return last output character + 1 */ |
| 465 | return end; |
| 466 | } |
| 467 | |
| 468 | /* |
| 469 | * pg_ltostr |
| 470 | * Converts 'value' into a decimal string representation stored at 'str'. |
| 471 | * |
| 472 | * Returns the ending address of the string result (the last character written |
| 473 | * plus 1). Note that no NUL terminator is written. |
| 474 | * |
| 475 | * The intended use-case for this function is to build strings that contain |
| 476 | * multiple individual numbers, for example: |
| 477 | * |
| 478 | * str = pg_ltostr(str, a); |
| 479 | * *str++ = ' '; |
| 480 | * str = pg_ltostr(str, b); |
| 481 | * *str = '\0'; |
| 482 | * |
| 483 | * Note: Caller must ensure that 'str' points to enough memory to hold the |
| 484 | * result. |
| 485 | */ |
| 486 | char * |
| 487 | pg_ltostr(char *str, int32 value) |
| 488 | { |
| 489 | char *start; |
| 490 | char *end; |
| 491 | |
| 492 | /* |
| 493 | * Handle negative numbers in a special way. We can't just write a '-' |
| 494 | * prefix and reverse the sign as that would overflow for INT32_MIN. |
| 495 | */ |
| 496 | if (value < 0) |
| 497 | { |
| 498 | *str++ = '-'; |
| 499 | |
| 500 | /* Mark the position we must reverse the string from. */ |
| 501 | start = str; |
| 502 | |
| 503 | /* Compute the result string backwards. */ |
| 504 | do |
| 505 | { |
| 506 | int32 oldval = value; |
| 507 | int32 remainder; |
| 508 | |
| 509 | value /= 10; |
| 510 | remainder = oldval - value * 10; |
| 511 | /* As above, we expect remainder to be negative. */ |
| 512 | *str++ = '0' - remainder; |
| 513 | } while (value != 0); |
| 514 | } |
| 515 | else |
| 516 | { |
| 517 | /* Mark the position we must reverse the string from. */ |
| 518 | start = str; |
| 519 | |
| 520 | /* Compute the result string backwards. */ |
| 521 | do |
| 522 | { |
| 523 | int32 oldval = value; |
| 524 | int32 remainder; |
| 525 | |
| 526 | value /= 10; |
| 527 | remainder = oldval - value * 10; |
| 528 | *str++ = '0' + remainder; |
| 529 | } while (value != 0); |
| 530 | } |
| 531 | |
| 532 | /* Remember the end+1 and back up 'str' to the last character. */ |
| 533 | end = str--; |
| 534 | |
| 535 | /* Reverse string. */ |
| 536 | while (start < str) |
| 537 | { |
| 538 | char swap = *start; |
| 539 | |
| 540 | *start++ = *str; |
| 541 | *str-- = swap; |
| 542 | } |
| 543 | |
| 544 | return end; |
| 545 | } |
| 546 | |
| 547 | /* |
| 548 | * pg_strtouint64 |
| 549 | * Converts 'str' into an unsigned 64-bit integer. |
| 550 | * |
| 551 | * This has the identical API to strtoul(3), except that it will handle |
| 552 | * 64-bit ints even where "long" is narrower than that. |
| 553 | * |
| 554 | * For the moment it seems sufficient to assume that the platform has |
| 555 | * such a function somewhere; let's not roll our own. |
| 556 | */ |
| 557 | uint64 |
| 558 | pg_strtouint64(const char *str, char **endptr, int base) |
| 559 | { |
| 560 | #ifdef _MSC_VER /* MSVC only */ |
| 561 | return _strtoui64(str, endptr, base); |
| 562 | #elif defined(HAVE_STRTOULL) && SIZEOF_LONG < 8 |
| 563 | return strtoull(str, endptr, base); |
| 564 | #else |
| 565 | return strtoul(str, endptr, base); |
| 566 | #endif |
| 567 | } |
| 568 | |