conversions-inl.h 20.7 KB
Newer Older
1
// Copyright 2011 the V8 project authors. All rights reserved.
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
//     * Redistributions of source code must retain the above copyright
//       notice, this list of conditions and the following disclaimer.
//     * Redistributions in binary form must reproduce the above
//       copyright notice, this list of conditions and the following
//       disclaimer in the documentation and/or other materials provided
//       with the distribution.
//     * Neither the name of Google Inc. nor the names of its
//       contributors may be used to endorse or promote products derived
//       from this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

#ifndef V8_CONVERSIONS_INL_H_
#define V8_CONVERSIONS_INL_H_

31
#include <limits.h>        // Required for INT_MAX etc.
32
#include <math.h>
33
#include <float.h>         // Required for DBL_MAX and on Win32 for finite()
34
#include <stdarg.h>
35 36 37 38 39

// ----------------------------------------------------------------------------
// Extra POSIX/ANSI functions for Win32/MSVC.

#include "conversions.h"
40
#include "strtod.h"
41 42
#include "platform.h"

43 44
namespace v8 {
namespace internal {
45

46 47 48 49 50
static inline double JunkStringValue() {
  return std::numeric_limits<double>::quiet_NaN();
}


51
// The fast double-to-unsigned-int conversion routine does not guarantee
52 53
// rounding towards zero, or any reasonable value if the argument is larger
// than what fits in an unsigned 32-bit integer.
54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69
static inline unsigned int FastD2UI(double x) {
  // There is no unsigned version of lrint, so there is no fast path
  // in this function as there is in FastD2I. Using lrint doesn't work
  // for values of 2^31 and above.

  // Convert "small enough" doubles to uint32_t by fixing the 32
  // least significant non-fractional bits in the low 32 bits of the
  // double, and reading them from there.
  const double k2Pow52 = 4503599627370496.0;
  bool negative = x < 0;
  if (negative) {
    x = -x;
  }
  if (x < k2Pow52) {
    x += k2Pow52;
    uint32_t result;
70 71 72
    Address mantissa_ptr = reinterpret_cast<Address>(&x);
    // Copy least significant 32 bits of mantissa.
    memcpy(&result, mantissa_ptr, sizeof(result));
73 74 75 76 77 78 79
    return negative ? ~result + 1 : result;
  }
  // Large number (outside uint32 range), Infinity or NaN.
  return 0x80000000u;  // Return integer indefinite.
}


80 81 82 83 84 85 86 87 88 89 90 91 92
static inline double DoubleToInteger(double x) {
  if (isnan(x)) return 0;
  if (!isfinite(x) || x == 0) return x;
  return (x >= 0) ? floor(x) : ceil(x);
}


int32_t DoubleToInt32(double x) {
  int32_t i = FastD2I(x);
  if (FastI2D(i) == x) return i;
  static const double two32 = 4294967296.0;
  static const double two31 = 2147483648.0;
  if (!isfinite(x) || x == 0) return 0;
93
  if (x < 0 || x >= two32) x = modulo(x, two32);
94 95 96 97 98
  x = (x >= 0) ? floor(x) : ceil(x) + two32;
  return (int32_t) ((x >= two31) ? x - two32 : x);
}


99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158
template <class Iterator, class EndMark>
static bool SubStringEquals(Iterator* current,
                            EndMark end,
                            const char* substring) {
  ASSERT(**current == *substring);
  for (substring++; *substring != '\0'; substring++) {
    ++*current;
    if (*current == end || **current != *substring) return false;
  }
  ++*current;
  return true;
}


// Returns true if a nonspace character has been found and false if the
// end was been reached before finding a nonspace character.
template <class Iterator, class EndMark>
static inline bool AdvanceToNonspace(UnicodeCache* unicode_cache,
                                     Iterator* current,
                                     EndMark end) {
  while (*current != end) {
    if (!unicode_cache->IsWhiteSpace(**current)) return true;
    ++*current;
  }
  return false;
}


// Parsing integers with radix 2, 4, 8, 16, 32. Assumes current != end.
template <int radix_log_2, class Iterator, class EndMark>
static double InternalStringToIntDouble(UnicodeCache* unicode_cache,
                                        Iterator current,
                                        EndMark end,
                                        bool negative,
                                        bool allow_trailing_junk) {
  ASSERT(current != end);

  // Skip leading 0s.
  while (*current == '0') {
    ++current;
    if (current == end) return SignedZero(negative);
  }

  int64_t number = 0;
  int exponent = 0;
  const int radix = (1 << radix_log_2);

  do {
    int digit;
    if (*current >= '0' && *current <= '9' && *current < '0' + radix) {
      digit = static_cast<char>(*current) - '0';
    } else if (radix > 10 && *current >= 'a' && *current < 'a' + radix - 10) {
      digit = static_cast<char>(*current) - 'a' + 10;
    } else if (radix > 10 && *current >= 'A' && *current < 'A' + radix - 10) {
      digit = static_cast<char>(*current) - 'A' + 10;
    } else {
      if (allow_trailing_junk ||
          !AdvanceToNonspace(unicode_cache, &current, end)) {
        break;
      } else {
159
        return JunkStringValue();
160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188
      }
    }

    number = number * radix + digit;
    int overflow = static_cast<int>(number >> 53);
    if (overflow != 0) {
      // Overflow occurred. Need to determine which direction to round the
      // result.
      int overflow_bits_count = 1;
      while (overflow > 1) {
        overflow_bits_count++;
        overflow >>= 1;
      }

      int dropped_bits_mask = ((1 << overflow_bits_count) - 1);
      int dropped_bits = static_cast<int>(number) & dropped_bits_mask;
      number >>= overflow_bits_count;
      exponent = overflow_bits_count;

      bool zero_tail = true;
      while (true) {
        ++current;
        if (current == end || !isDigit(*current, radix)) break;
        zero_tail = zero_tail && *current == '0';
        exponent += radix_log_2;
      }

      if (!allow_trailing_junk &&
          AdvanceToNonspace(unicode_cache, &current, end)) {
189
        return JunkStringValue();
190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236
      }

      int middle_value = (1 << (overflow_bits_count - 1));
      if (dropped_bits > middle_value) {
        number++;  // Rounding up.
      } else if (dropped_bits == middle_value) {
        // Rounding to even to consistency with decimals: half-way case rounds
        // up if significant part is odd and down otherwise.
        if ((number & 1) != 0 || !zero_tail) {
          number++;  // Rounding up.
        }
      }

      // Rounding up may cause overflow.
      if ((number & ((int64_t)1 << 53)) != 0) {
        exponent++;
        number >>= 1;
      }
      break;
    }
    ++current;
  } while (current != end);

  ASSERT(number < ((int64_t)1 << 53));
  ASSERT(static_cast<int64_t>(static_cast<double>(number)) == number);

  if (exponent == 0) {
    if (negative) {
      if (number == 0) return -0.0;
      number = -number;
    }
    return static_cast<double>(number);
  }

  ASSERT(number != 0);
  // The double could be constructed faster from number (mantissa), exponent
  // and sign. Assuming it's a rare case more simple code is used.
  return static_cast<double>(negative ? -number : number) * pow(2.0, exponent);
}


template <class Iterator, class EndMark>
static double InternalStringToInt(UnicodeCache* unicode_cache,
                                  Iterator current,
                                  EndMark end,
                                  int radix) {
  const bool allow_trailing_junk = true;
237
  const double empty_string_val = JunkStringValue();
238 239 240 241 242 243 244 245 246 247 248 249

  if (!AdvanceToNonspace(unicode_cache, &current, end)) {
    return empty_string_val;
  }

  bool negative = false;
  bool leading_zero = false;

  if (*current == '+') {
    // Ignore leading sign; skip following spaces.
    ++current;
    if (current == end) {
250
      return JunkStringValue();
251 252 253 254
    }
  } else if (*current == '-') {
    ++current;
    if (current == end) {
255
      return JunkStringValue();
256 257 258 259 260 261 262 263 264 265 266 267
    }
    negative = true;
  }

  if (radix == 0) {
    // Radix detection.
    if (*current == '0') {
      ++current;
      if (current == end) return SignedZero(negative);
      if (*current == 'x' || *current == 'X') {
        radix = 16;
        ++current;
268
        if (current == end) return JunkStringValue();
269 270 271 272 273 274 275 276 277 278 279 280 281 282
      } else {
        radix = 8;
        leading_zero = true;
      }
    } else {
      radix = 10;
    }
  } else if (radix == 16) {
    if (*current == '0') {
      // Allow "0x" prefix.
      ++current;
      if (current == end) return SignedZero(negative);
      if (*current == 'x' || *current == 'X') {
        ++current;
283
        if (current == end) return JunkStringValue();
284 285 286 287 288 289
      } else {
        leading_zero = true;
      }
    }
  }

290
  if (radix < 2 || radix > 36) return JunkStringValue();
291 292 293 294 295 296 297 298 299

  // Skip leading zeros.
  while (*current == '0') {
    leading_zero = true;
    ++current;
    if (current == end) return SignedZero(negative);
  }

  if (!leading_zero && !isDigit(*current, radix)) {
300
    return JunkStringValue();
301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347
  }

  if (IsPowerOf2(radix)) {
    switch (radix) {
      case 2:
        return InternalStringToIntDouble<1>(
            unicode_cache, current, end, negative, allow_trailing_junk);
      case 4:
        return InternalStringToIntDouble<2>(
            unicode_cache, current, end, negative, allow_trailing_junk);
      case 8:
        return InternalStringToIntDouble<3>(
            unicode_cache, current, end, negative, allow_trailing_junk);

      case 16:
        return InternalStringToIntDouble<4>(
            unicode_cache, current, end, negative, allow_trailing_junk);

      case 32:
        return InternalStringToIntDouble<5>(
            unicode_cache, current, end, negative, allow_trailing_junk);
      default:
        UNREACHABLE();
    }
  }

  if (radix == 10) {
    // Parsing with strtod.
    const int kMaxSignificantDigits = 309;  // Doubles are less than 1.8e308.
    // The buffer may contain up to kMaxSignificantDigits + 1 digits and a zero
    // end.
    const int kBufferSize = kMaxSignificantDigits + 2;
    char buffer[kBufferSize];
    int buffer_pos = 0;
    while (*current >= '0' && *current <= '9') {
      if (buffer_pos <= kMaxSignificantDigits) {
        // If the number has more than kMaxSignificantDigits it will be parsed
        // as infinity.
        ASSERT(buffer_pos < kBufferSize);
        buffer[buffer_pos++] = static_cast<char>(*current);
      }
      ++current;
      if (current == end) break;
    }

    if (!allow_trailing_junk &&
        AdvanceToNonspace(unicode_cache, &current, end)) {
348
      return JunkStringValue();
349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413
    }

    ASSERT(buffer_pos < kBufferSize);
    buffer[buffer_pos] = '\0';
    Vector<const char> buffer_vector(buffer, buffer_pos);
    return negative ? -Strtod(buffer_vector, 0) : Strtod(buffer_vector, 0);
  }

  // The following code causes accumulating rounding error for numbers greater
  // than ~2^56. It's explicitly allowed in the spec: "if R is not 2, 4, 8, 10,
  // 16, or 32, then mathInt may be an implementation-dependent approximation to
  // the mathematical integer value" (15.1.2.2).

  int lim_0 = '0' + (radix < 10 ? radix : 10);
  int lim_a = 'a' + (radix - 10);
  int lim_A = 'A' + (radix - 10);

  // NOTE: The code for computing the value may seem a bit complex at
  // first glance. It is structured to use 32-bit multiply-and-add
  // loops as long as possible to avoid loosing precision.

  double v = 0.0;
  bool done = false;
  do {
    // Parse the longest part of the string starting at index j
    // possible while keeping the multiplier, and thus the part
    // itself, within 32 bits.
    unsigned int part = 0, multiplier = 1;
    while (true) {
      int d;
      if (*current >= '0' && *current < lim_0) {
        d = *current - '0';
      } else if (*current >= 'a' && *current < lim_a) {
        d = *current - 'a' + 10;
      } else if (*current >= 'A' && *current < lim_A) {
        d = *current - 'A' + 10;
      } else {
        done = true;
        break;
      }

      // Update the value of the part as long as the multiplier fits
      // in 32 bits. When we can't guarantee that the next iteration
      // will not overflow the multiplier, we stop parsing the part
      // by leaving the loop.
      const unsigned int kMaximumMultiplier = 0xffffffffU / 36;
      uint32_t m = multiplier * radix;
      if (m > kMaximumMultiplier) break;
      part = part * radix + d;
      multiplier = m;
      ASSERT(multiplier > part);

      ++current;
      if (current == end) {
        done = true;
        break;
      }
    }

    // Update the value and skip the part in the string.
    v = v * multiplier + part;
  } while (!done);

  if (!allow_trailing_junk &&
      AdvanceToNonspace(unicode_cache, &current, end)) {
414
    return JunkStringValue();
415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462
  }

  return negative ? -v : v;
}


// Converts a string to a double value. Assumes the Iterator supports
// the following operations:
// 1. current == end (other ops are not allowed), current != end.
// 2. *current - gets the current character in the sequence.
// 3. ++current (advances the position).
template <class Iterator, class EndMark>
static double InternalStringToDouble(UnicodeCache* unicode_cache,
                                     Iterator current,
                                     EndMark end,
                                     int flags,
                                     double empty_string_val) {
  // To make sure that iterator dereferencing is valid the following
  // convention is used:
  // 1. Each '++current' statement is followed by check for equality to 'end'.
  // 2. If AdvanceToNonspace returned false then current == end.
  // 3. If 'current' becomes be equal to 'end' the function returns or goes to
  // 'parsing_done'.
  // 4. 'current' is not dereferenced after the 'parsing_done' label.
  // 5. Code before 'parsing_done' may rely on 'current != end'.
  if (!AdvanceToNonspace(unicode_cache, &current, end)) {
    return empty_string_val;
  }

  const bool allow_trailing_junk = (flags & ALLOW_TRAILING_JUNK) != 0;

  // The longest form of simplified number is: "-<significant digits>'.1eXXX\0".
  const int kBufferSize = kMaxSignificantDigits + 10;
  char buffer[kBufferSize];  // NOLINT: size is known at compile time.
  int buffer_pos = 0;

  // Exponent will be adjusted if insignificant digits of the integer part
  // or insignificant leading zeros of the fractional part are dropped.
  int exponent = 0;
  int significant_digits = 0;
  int insignificant_digits = 0;
  bool nonzero_digit_dropped = false;

  bool negative = false;

  if (*current == '+') {
    // Ignore leading sign.
    ++current;
463
    if (current == end) return JunkStringValue();
464 465
  } else if (*current == '-') {
    ++current;
466
    if (current == end) return JunkStringValue();
467 468 469 470 471 472
    negative = true;
  }

  static const char kInfinitySymbol[] = "Infinity";
  if (*current == kInfinitySymbol[0]) {
    if (!SubStringEquals(&current, end, kInfinitySymbol)) {
473
      return JunkStringValue();
474 475 476 477
    }

    if (!allow_trailing_junk &&
        AdvanceToNonspace(unicode_cache, &current, end)) {
478
      return JunkStringValue();
479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495
    }

    ASSERT(buffer_pos == 0);
    return negative ? -V8_INFINITY : V8_INFINITY;
  }

  bool leading_zero = false;
  if (*current == '0') {
    ++current;
    if (current == end) return SignedZero(negative);

    leading_zero = true;

    // It could be hexadecimal value.
    if ((flags & ALLOW_HEX) && (*current == 'x' || *current == 'X')) {
      ++current;
      if (current == end || !isDigit(*current, 16)) {
496
        return JunkStringValue();  // "0x".
497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535
      }

      return InternalStringToIntDouble<4>(unicode_cache,
                                          current,
                                          end,
                                          negative,
                                          allow_trailing_junk);
    }

    // Ignore leading zeros in the integer part.
    while (*current == '0') {
      ++current;
      if (current == end) return SignedZero(negative);
    }
  }

  bool octal = leading_zero && (flags & ALLOW_OCTALS) != 0;

  // Copy significant digits of the integer part (if any) to the buffer.
  while (*current >= '0' && *current <= '9') {
    if (significant_digits < kMaxSignificantDigits) {
      ASSERT(buffer_pos < kBufferSize);
      buffer[buffer_pos++] = static_cast<char>(*current);
      significant_digits++;
      // Will later check if it's an octal in the buffer.
    } else {
      insignificant_digits++;  // Move the digit into the exponential part.
      nonzero_digit_dropped = nonzero_digit_dropped || *current != '0';
    }
    octal = octal && *current < '8';
    ++current;
    if (current == end) goto parsing_done;
  }

  if (significant_digits == 0) {
    octal = false;
  }

  if (*current == '.') {
536
    if (octal && !allow_trailing_junk) return JunkStringValue();
537 538 539 540 541
    if (octal) goto parsing_done;

    ++current;
    if (current == end) {
      if (significant_digits == 0 && !leading_zero) {
542
        return JunkStringValue();
543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558
      } else {
        goto parsing_done;
      }
    }

    if (significant_digits == 0) {
      // octal = false;
      // Integer part consists of 0 or is absent. Significant digits start after
      // leading zeros (if any).
      while (*current == '0') {
        ++current;
        if (current == end) return SignedZero(negative);
        exponent--;  // Move this 0 into the exponent.
      }
    }

559 560
    // There is a fractional part.  We don't emit a '.', but adjust the exponent
    // instead.
561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580
    while (*current >= '0' && *current <= '9') {
      if (significant_digits < kMaxSignificantDigits) {
        ASSERT(buffer_pos < kBufferSize);
        buffer[buffer_pos++] = static_cast<char>(*current);
        significant_digits++;
        exponent--;
      } else {
        // Ignore insignificant digits in the fractional part.
        nonzero_digit_dropped = nonzero_digit_dropped || *current != '0';
      }
      ++current;
      if (current == end) goto parsing_done;
    }
  }

  if (!leading_zero && exponent == 0 && significant_digits == 0) {
    // If leading_zeros is true then the string contains zeros.
    // If exponent < 0 then string was [+-]\.0*...
    // If significant_digits != 0 the string is not equal to 0.
    // Otherwise there are no digits in the string.
581
    return JunkStringValue();
582 583 584 585
  }

  // Parse exponential part.
  if (*current == 'e' || *current == 'E') {
586
    if (octal) return JunkStringValue();
587 588 589 590 591
    ++current;
    if (current == end) {
      if (allow_trailing_junk) {
        goto parsing_done;
      } else {
592
        return JunkStringValue();
593 594 595 596 597 598 599 600 601 602
      }
    }
    char sign = '+';
    if (*current == '+' || *current == '-') {
      sign = static_cast<char>(*current);
      ++current;
      if (current == end) {
        if (allow_trailing_junk) {
          goto parsing_done;
        } else {
603
          return JunkStringValue();
604 605 606 607 608 609 610 611
        }
      }
    }

    if (current == end || *current < '0' || *current > '9') {
      if (allow_trailing_junk) {
        goto parsing_done;
      } else {
612
        return JunkStringValue();
613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635
      }
    }

    const int max_exponent = INT_MAX / 2;
    ASSERT(-max_exponent / 2 <= exponent && exponent <= max_exponent / 2);
    int num = 0;
    do {
      // Check overflow.
      int digit = *current - '0';
      if (num >= max_exponent / 10
          && !(num == max_exponent / 10 && digit <= max_exponent % 10)) {
        num = max_exponent;
      } else {
        num = num * 10 + digit;
      }
      ++current;
    } while (current != end && *current >= '0' && *current <= '9');

    exponent += (sign == '-' ? -num : num);
  }

  if (!allow_trailing_junk &&
      AdvanceToNonspace(unicode_cache, &current, end)) {
636
    return JunkStringValue();
637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661
  }

  parsing_done:
  exponent += insignificant_digits;

  if (octal) {
    return InternalStringToIntDouble<3>(unicode_cache,
                                        buffer,
                                        buffer + buffer_pos,
                                        negative,
                                        allow_trailing_junk);
  }

  if (nonzero_digit_dropped) {
    buffer[buffer_pos++] = '1';
    exponent--;
  }

  ASSERT(buffer_pos < kBufferSize);
  buffer[buffer_pos] = '\0';

  double converted = Strtod(Vector<const char>(buffer, buffer_pos), exponent);
  return negative ? -converted : converted;
}

662 663 664
} }  // namespace v8::internal

#endif  // V8_CONVERSIONS_INL_H_