Home | History | Annotate | Download | only in src
      1 // Copyright 2011 the V8 project authors. All rights reserved.
      2 // Use of this source code is governed by a BSD-style license that can be
      3 // found in the LICENSE file.
      4 
      5 #include <cmath>
      6 
      7 #include "include/v8stdint.h"
      8 #include "src/base/logging.h"
      9 #include "src/utils.h"
     10 
     11 #include "src/bignum-dtoa.h"
     12 
     13 #include "src/bignum.h"
     14 #include "src/double.h"
     15 
     16 namespace v8 {
     17 namespace internal {
     18 
     19 static int NormalizedExponent(uint64_t significand, int exponent) {
     20   DCHECK(significand != 0);
     21   while ((significand & Double::kHiddenBit) == 0) {
     22     significand = significand << 1;
     23     exponent = exponent - 1;
     24   }
     25   return exponent;
     26 }
     27 
     28 
     29 // Forward declarations:
     30 // Returns an estimation of k such that 10^(k-1) <= v < 10^k.
     31 static int EstimatePower(int exponent);
     32 // Computes v / 10^estimated_power exactly, as a ratio of two bignums, numerator
     33 // and denominator.
     34 static void InitialScaledStartValues(double v,
     35                                      int estimated_power,
     36                                      bool need_boundary_deltas,
     37                                      Bignum* numerator,
     38                                      Bignum* denominator,
     39                                      Bignum* delta_minus,
     40                                      Bignum* delta_plus);
     41 // Multiplies numerator/denominator so that its values lies in the range 1-10.
     42 // Returns decimal_point s.t.
     43 //  v = numerator'/denominator' * 10^(decimal_point-1)
     44 //     where numerator' and denominator' are the values of numerator and
     45 //     denominator after the call to this function.
     46 static void FixupMultiply10(int estimated_power, bool is_even,
     47                             int* decimal_point,
     48                             Bignum* numerator, Bignum* denominator,
     49                             Bignum* delta_minus, Bignum* delta_plus);
     50 // Generates digits from the left to the right and stops when the generated
     51 // digits yield the shortest decimal representation of v.
     52 static void GenerateShortestDigits(Bignum* numerator, Bignum* denominator,
     53                                    Bignum* delta_minus, Bignum* delta_plus,
     54                                    bool is_even,
     55                                    Vector<char> buffer, int* length);
     56 // Generates 'requested_digits' after the decimal point.
     57 static void BignumToFixed(int requested_digits, int* decimal_point,
     58                           Bignum* numerator, Bignum* denominator,
     59                           Vector<char>(buffer), int* length);
     60 // Generates 'count' digits of numerator/denominator.
     61 // Once 'count' digits have been produced rounds the result depending on the
     62 // remainder (remainders of exactly .5 round upwards). Might update the
     63 // decimal_point when rounding up (for example for 0.9999).
     64 static void GenerateCountedDigits(int count, int* decimal_point,
     65                                   Bignum* numerator, Bignum* denominator,
     66                                   Vector<char>(buffer), int* length);
     67 
     68 
     69 void BignumDtoa(double v, BignumDtoaMode mode, int requested_digits,
     70                 Vector<char> buffer, int* length, int* decimal_point) {
     71   DCHECK(v > 0);
     72   DCHECK(!Double(v).IsSpecial());
     73   uint64_t significand = Double(v).Significand();
     74   bool is_even = (significand & 1) == 0;
     75   int exponent = Double(v).Exponent();
     76   int normalized_exponent = NormalizedExponent(significand, exponent);
     77   // estimated_power might be too low by 1.
     78   int estimated_power = EstimatePower(normalized_exponent);
     79 
     80   // Shortcut for Fixed.
     81   // The requested digits correspond to the digits after the point. If the
     82   // number is much too small, then there is no need in trying to get any
     83   // digits.
     84   if (mode == BIGNUM_DTOA_FIXED && -estimated_power - 1 > requested_digits) {
     85     buffer[0] = '\0';
     86     *length = 0;
     87     // Set decimal-point to -requested_digits. This is what Gay does.
     88     // Note that it should not have any effect anyways since the string is
     89     // empty.
     90     *decimal_point = -requested_digits;
     91     return;
     92   }
     93 
     94   Bignum numerator;
     95   Bignum denominator;
     96   Bignum delta_minus;
     97   Bignum delta_plus;
     98   // Make sure the bignum can grow large enough. The smallest double equals
     99   // 4e-324. In this case the denominator needs fewer than 324*4 binary digits.
    100   // The maximum double is 1.7976931348623157e308 which needs fewer than
    101   // 308*4 binary digits.
    102   DCHECK(Bignum::kMaxSignificantBits >= 324*4);
    103   bool need_boundary_deltas = (mode == BIGNUM_DTOA_SHORTEST);
    104   InitialScaledStartValues(v, estimated_power, need_boundary_deltas,
    105                            &numerator, &denominator,
    106                            &delta_minus, &delta_plus);
    107   // We now have v = (numerator / denominator) * 10^estimated_power.
    108   FixupMultiply10(estimated_power, is_even, decimal_point,
    109                   &numerator, &denominator,
    110                   &delta_minus, &delta_plus);
    111   // We now have v = (numerator / denominator) * 10^(decimal_point-1), and
    112   //  1 <= (numerator + delta_plus) / denominator < 10
    113   switch (mode) {
    114     case BIGNUM_DTOA_SHORTEST:
    115       GenerateShortestDigits(&numerator, &denominator,
    116                              &delta_minus, &delta_plus,
    117                              is_even, buffer, length);
    118       break;
    119     case BIGNUM_DTOA_FIXED:
    120       BignumToFixed(requested_digits, decimal_point,
    121                     &numerator, &denominator,
    122                     buffer, length);
    123       break;
    124     case BIGNUM_DTOA_PRECISION:
    125       GenerateCountedDigits(requested_digits, decimal_point,
    126                             &numerator, &denominator,
    127                             buffer, length);
    128       break;
    129     default:
    130       UNREACHABLE();
    131   }
    132   buffer[*length] = '\0';
    133 }
    134 
    135 
    136 // The procedure starts generating digits from the left to the right and stops
    137 // when the generated digits yield the shortest decimal representation of v. A
    138 // decimal representation of v is a number lying closer to v than to any other
    139 // double, so it converts to v when read.
    140 //
    141 // This is true if d, the decimal representation, is between m- and m+, the
    142 // upper and lower boundaries. d must be strictly between them if !is_even.
    143 //           m- := (numerator - delta_minus) / denominator
    144 //           m+ := (numerator + delta_plus) / denominator
    145 //
    146 // Precondition: 0 <= (numerator+delta_plus) / denominator < 10.
    147 //   If 1 <= (numerator+delta_plus) / denominator < 10 then no leading 0 digit
    148 //   will be produced. This should be the standard precondition.
    149 static void GenerateShortestDigits(Bignum* numerator, Bignum* denominator,
    150                                    Bignum* delta_minus, Bignum* delta_plus,
    151                                    bool is_even,
    152                                    Vector<char> buffer, int* length) {
    153   // Small optimization: if delta_minus and delta_plus are the same just reuse
    154   // one of the two bignums.
    155   if (Bignum::Equal(*delta_minus, *delta_plus)) {
    156     delta_plus = delta_minus;
    157   }
    158   *length = 0;
    159   while (true) {
    160     uint16_t digit;
    161     digit = numerator->DivideModuloIntBignum(*denominator);
    162     DCHECK(digit <= 9);  // digit is a uint16_t and therefore always positive.
    163     // digit = numerator / denominator (integer division).
    164     // numerator = numerator % denominator.
    165     buffer[(*length)++] = digit + '0';
    166 
    167     // Can we stop already?
    168     // If the remainder of the division is less than the distance to the lower
    169     // boundary we can stop. In this case we simply round down (discarding the
    170     // remainder).
    171     // Similarly we test if we can round up (using the upper boundary).
    172     bool in_delta_room_minus;
    173     bool in_delta_room_plus;
    174     if (is_even) {
    175       in_delta_room_minus = Bignum::LessEqual(*numerator, *delta_minus);
    176     } else {
    177       in_delta_room_minus = Bignum::Less(*numerator, *delta_minus);
    178     }
    179     if (is_even) {
    180       in_delta_room_plus =
    181           Bignum::PlusCompare(*numerator, *delta_plus, *denominator) >= 0;
    182     } else {
    183       in_delta_room_plus =
    184           Bignum::PlusCompare(*numerator, *delta_plus, *denominator) > 0;
    185     }
    186     if (!in_delta_room_minus && !in_delta_room_plus) {
    187       // Prepare for next iteration.
    188       numerator->Times10();
    189       delta_minus->Times10();
    190       // We optimized delta_plus to be equal to delta_minus (if they share the
    191       // same value). So don't multiply delta_plus if they point to the same
    192       // object.
    193       if (delta_minus != delta_plus) {
    194         delta_plus->Times10();
    195       }
    196     } else if (in_delta_room_minus && in_delta_room_plus) {
    197       // Let's see if 2*numerator < denominator.
    198       // If yes, then the next digit would be < 5 and we can round down.
    199       int compare = Bignum::PlusCompare(*numerator, *numerator, *denominator);
    200       if (compare < 0) {
    201         // Remaining digits are less than .5. -> Round down (== do nothing).
    202       } else if (compare > 0) {
    203         // Remaining digits are more than .5 of denominator. -> Round up.
    204         // Note that the last digit could not be a '9' as otherwise the whole
    205         // loop would have stopped earlier.
    206         // We still have an assert here in case the preconditions were not
    207         // satisfied.
    208         DCHECK(buffer[(*length) - 1] != '9');
    209         buffer[(*length) - 1]++;
    210       } else {
    211         // Halfway case.
    212         // TODO(floitsch): need a way to solve half-way cases.
    213         //   For now let's round towards even (since this is what Gay seems to
    214         //   do).
    215 
    216         if ((buffer[(*length) - 1] - '0') % 2 == 0) {
    217           // Round down => Do nothing.
    218         } else {
    219           DCHECK(buffer[(*length) - 1] != '9');
    220           buffer[(*length) - 1]++;
    221         }
    222       }
    223       return;
    224     } else if (in_delta_room_minus) {
    225       // Round down (== do nothing).
    226       return;
    227     } else {  // in_delta_room_plus
    228       // Round up.
    229       // Note again that the last digit could not be '9' since this would have
    230       // stopped the loop earlier.
    231       // We still have an DCHECK here, in case the preconditions were not
    232       // satisfied.
    233       DCHECK(buffer[(*length) -1] != '9');
    234       buffer[(*length) - 1]++;
    235       return;
    236     }
    237   }
    238 }
    239 
    240 
    241 // Let v = numerator / denominator < 10.
    242 // Then we generate 'count' digits of d = x.xxxxx... (without the decimal point)
    243 // from left to right. Once 'count' digits have been produced we decide wether
    244 // to round up or down. Remainders of exactly .5 round upwards. Numbers such
    245 // as 9.999999 propagate a carry all the way, and change the
    246 // exponent (decimal_point), when rounding upwards.
    247 static void GenerateCountedDigits(int count, int* decimal_point,
    248                                   Bignum* numerator, Bignum* denominator,
    249                                   Vector<char>(buffer), int* length) {
    250   DCHECK(count >= 0);
    251   for (int i = 0; i < count - 1; ++i) {
    252     uint16_t digit;
    253     digit = numerator->DivideModuloIntBignum(*denominator);
    254     DCHECK(digit <= 9);  // digit is a uint16_t and therefore always positive.
    255     // digit = numerator / denominator (integer division).
    256     // numerator = numerator % denominator.
    257     buffer[i] = digit + '0';
    258     // Prepare for next iteration.
    259     numerator->Times10();
    260   }
    261   // Generate the last digit.
    262   uint16_t digit;
    263   digit = numerator->DivideModuloIntBignum(*denominator);
    264   if (Bignum::PlusCompare(*numerator, *numerator, *denominator) >= 0) {
    265     digit++;
    266   }
    267   buffer[count - 1] = digit + '0';
    268   // Correct bad digits (in case we had a sequence of '9's). Propagate the
    269   // carry until we hat a non-'9' or til we reach the first digit.
    270   for (int i = count - 1; i > 0; --i) {
    271     if (buffer[i] != '0' + 10) break;
    272     buffer[i] = '0';
    273     buffer[i - 1]++;
    274   }
    275   if (buffer[0] == '0' + 10) {
    276     // Propagate a carry past the top place.
    277     buffer[0] = '1';
    278     (*decimal_point)++;
    279   }
    280   *length = count;
    281 }
    282 
    283 
    284 // Generates 'requested_digits' after the decimal point. It might omit
    285 // trailing '0's. If the input number is too small then no digits at all are
    286 // generated (ex.: 2 fixed digits for 0.00001).
    287 //
    288 // Input verifies:  1 <= (numerator + delta) / denominator < 10.
    289 static void BignumToFixed(int requested_digits, int* decimal_point,
    290                           Bignum* numerator, Bignum* denominator,
    291                           Vector<char>(buffer), int* length) {
    292   // Note that we have to look at more than just the requested_digits, since
    293   // a number could be rounded up. Example: v=0.5 with requested_digits=0.
    294   // Even though the power of v equals 0 we can't just stop here.
    295   if (-(*decimal_point) > requested_digits) {
    296     // The number is definitively too small.
    297     // Ex: 0.001 with requested_digits == 1.
    298     // Set decimal-point to -requested_digits. This is what Gay does.
    299     // Note that it should not have any effect anyways since the string is
    300     // empty.
    301     *decimal_point = -requested_digits;
    302     *length = 0;
    303     return;
    304   } else if (-(*decimal_point) == requested_digits) {
    305     // We only need to verify if the number rounds down or up.
    306     // Ex: 0.04 and 0.06 with requested_digits == 1.
    307     DCHECK(*decimal_point == -requested_digits);
    308     // Initially the fraction lies in range (1, 10]. Multiply the denominator
    309     // by 10 so that we can compare more easily.
    310     denominator->Times10();
    311     if (Bignum::PlusCompare(*numerator, *numerator, *denominator) >= 0) {
    312       // If the fraction is >= 0.5 then we have to include the rounded
    313       // digit.
    314       buffer[0] = '1';
    315       *length = 1;
    316       (*decimal_point)++;
    317     } else {
    318       // Note that we caught most of similar cases earlier.
    319       *length = 0;
    320     }
    321     return;
    322   } else {
    323     // The requested digits correspond to the digits after the point.
    324     // The variable 'needed_digits' includes the digits before the point.
    325     int needed_digits = (*decimal_point) + requested_digits;
    326     GenerateCountedDigits(needed_digits, decimal_point,
    327                           numerator, denominator,
    328                           buffer, length);
    329   }
    330 }
    331 
    332 
    333 // Returns an estimation of k such that 10^(k-1) <= v < 10^k where
    334 // v = f * 2^exponent and 2^52 <= f < 2^53.
    335 // v is hence a normalized double with the given exponent. The output is an
    336 // approximation for the exponent of the decimal approimation .digits * 10^k.
    337 //
    338 // The result might undershoot by 1 in which case 10^k <= v < 10^k+1.
    339 // Note: this property holds for v's upper boundary m+ too.
    340 //    10^k <= m+ < 10^k+1.
    341 //   (see explanation below).
    342 //
    343 // Examples:
    344 //  EstimatePower(0)   => 16
    345 //  EstimatePower(-52) => 0
    346 //
    347 // Note: e >= 0 => EstimatedPower(e) > 0. No similar claim can be made for e<0.
    348 static int EstimatePower(int exponent) {
    349   // This function estimates log10 of v where v = f*2^e (with e == exponent).
    350   // Note that 10^floor(log10(v)) <= v, but v <= 10^ceil(log10(v)).
    351   // Note that f is bounded by its container size. Let p = 53 (the double's
    352   // significand size). Then 2^(p-1) <= f < 2^p.
    353   //
    354   // Given that log10(v) == log2(v)/log2(10) and e+(len(f)-1) is quite close
    355   // to log2(v) the function is simplified to (e+(len(f)-1)/log2(10)).
    356   // The computed number undershoots by less than 0.631 (when we compute log3
    357   // and not log10).
    358   //
    359   // Optimization: since we only need an approximated result this computation
    360   // can be performed on 64 bit integers. On x86/x64 architecture the speedup is
    361   // not really measurable, though.
    362   //
    363   // Since we want to avoid overshooting we decrement by 1e10 so that
    364   // floating-point imprecisions don't affect us.
    365   //
    366   // Explanation for v's boundary m+: the computation takes advantage of
    367   // the fact that 2^(p-1) <= f < 2^p. Boundaries still satisfy this requirement
    368   // (even for denormals where the delta can be much more important).
    369 
    370   const double k1Log10 = 0.30102999566398114;  // 1/lg(10)
    371 
    372   // For doubles len(f) == 53 (don't forget the hidden bit).
    373   const int kSignificandSize = 53;
    374   double estimate =
    375       std::ceil((exponent + kSignificandSize - 1) * k1Log10 - 1e-10);
    376   return static_cast<int>(estimate);
    377 }
    378 
    379 
    380 // See comments for InitialScaledStartValues.
    381 static void InitialScaledStartValuesPositiveExponent(
    382     double v, int estimated_power, bool need_boundary_deltas,
    383     Bignum* numerator, Bignum* denominator,
    384     Bignum* delta_minus, Bignum* delta_plus) {
    385   // A positive exponent implies a positive power.
    386   DCHECK(estimated_power >= 0);
    387   // Since the estimated_power is positive we simply multiply the denominator
    388   // by 10^estimated_power.
    389 
    390   // numerator = v.
    391   numerator->AssignUInt64(Double(v).Significand());
    392   numerator->ShiftLeft(Double(v).Exponent());
    393   // denominator = 10^estimated_power.
    394   denominator->AssignPowerUInt16(10, estimated_power);
    395 
    396   if (need_boundary_deltas) {
    397     // Introduce a common denominator so that the deltas to the boundaries are
    398     // integers.
    399     denominator->ShiftLeft(1);
    400     numerator->ShiftLeft(1);
    401     // Let v = f * 2^e, then m+ - v = 1/2 * 2^e; With the common
    402     // denominator (of 2) delta_plus equals 2^e.
    403     delta_plus->AssignUInt16(1);
    404     delta_plus->ShiftLeft(Double(v).Exponent());
    405     // Same for delta_minus (with adjustments below if f == 2^p-1).
    406     delta_minus->AssignUInt16(1);
    407     delta_minus->ShiftLeft(Double(v).Exponent());
    408 
    409     // If the significand (without the hidden bit) is 0, then the lower
    410     // boundary is closer than just half a ulp (unit in the last place).
    411     // There is only one exception: if the next lower number is a denormal then
    412     // the distance is 1 ulp. This cannot be the case for exponent >= 0 (but we
    413     // have to test it in the other function where exponent < 0).
    414     uint64_t v_bits = Double(v).AsUint64();
    415     if ((v_bits & Double::kSignificandMask) == 0) {
    416       // The lower boundary is closer at half the distance of "normal" numbers.
    417       // Increase the common denominator and adapt all but the delta_minus.
    418       denominator->ShiftLeft(1);  // *2
    419       numerator->ShiftLeft(1);    // *2
    420       delta_plus->ShiftLeft(1);   // *2
    421     }
    422   }
    423 }
    424 
    425 
    426 // See comments for InitialScaledStartValues
    427 static void InitialScaledStartValuesNegativeExponentPositivePower(
    428     double v, int estimated_power, bool need_boundary_deltas,
    429     Bignum* numerator, Bignum* denominator,
    430     Bignum* delta_minus, Bignum* delta_plus) {
    431   uint64_t significand = Double(v).Significand();
    432   int exponent = Double(v).Exponent();
    433   // v = f * 2^e with e < 0, and with estimated_power >= 0.
    434   // This means that e is close to 0 (have a look at how estimated_power is
    435   // computed).
    436 
    437   // numerator = significand
    438   //  since v = significand * 2^exponent this is equivalent to
    439   //  numerator = v * / 2^-exponent
    440   numerator->AssignUInt64(significand);
    441   // denominator = 10^estimated_power * 2^-exponent (with exponent < 0)
    442   denominator->AssignPowerUInt16(10, estimated_power);
    443   denominator->ShiftLeft(-exponent);
    444 
    445   if (need_boundary_deltas) {
    446     // Introduce a common denominator so that the deltas to the boundaries are
    447     // integers.
    448     denominator->ShiftLeft(1);
    449     numerator->ShiftLeft(1);
    450     // Let v = f * 2^e, then m+ - v = 1/2 * 2^e; With the common
    451     // denominator (of 2) delta_plus equals 2^e.
    452     // Given that the denominator already includes v's exponent the distance
    453     // to the boundaries is simply 1.
    454     delta_plus->AssignUInt16(1);
    455     // Same for delta_minus (with adjustments below if f == 2^p-1).
    456     delta_minus->AssignUInt16(1);
    457 
    458     // If the significand (without the hidden bit) is 0, then the lower
    459     // boundary is closer than just one ulp (unit in the last place).
    460     // There is only one exception: if the next lower number is a denormal
    461     // then the distance is 1 ulp. Since the exponent is close to zero
    462     // (otherwise estimated_power would have been negative) this cannot happen
    463     // here either.
    464     uint64_t v_bits = Double(v).AsUint64();
    465     if ((v_bits & Double::kSignificandMask) == 0) {
    466       // The lower boundary is closer at half the distance of "normal" numbers.
    467       // Increase the denominator and adapt all but the delta_minus.
    468       denominator->ShiftLeft(1);  // *2
    469       numerator->ShiftLeft(1);    // *2
    470       delta_plus->ShiftLeft(1);   // *2
    471     }
    472   }
    473 }
    474 
    475 
    476 // See comments for InitialScaledStartValues
    477 static void InitialScaledStartValuesNegativeExponentNegativePower(
    478     double v, int estimated_power, bool need_boundary_deltas,
    479     Bignum* numerator, Bignum* denominator,
    480     Bignum* delta_minus, Bignum* delta_plus) {
    481   const uint64_t kMinimalNormalizedExponent =
    482       V8_2PART_UINT64_C(0x00100000, 00000000);
    483   uint64_t significand = Double(v).Significand();
    484   int exponent = Double(v).Exponent();
    485   // Instead of multiplying the denominator with 10^estimated_power we
    486   // multiply all values (numerator and deltas) by 10^-estimated_power.
    487 
    488   // Use numerator as temporary container for power_ten.
    489   Bignum* power_ten = numerator;
    490   power_ten->AssignPowerUInt16(10, -estimated_power);
    491 
    492   if (need_boundary_deltas) {
    493     // Since power_ten == numerator we must make a copy of 10^estimated_power
    494     // before we complete the computation of the numerator.
    495     // delta_plus = delta_minus = 10^estimated_power
    496     delta_plus->AssignBignum(*power_ten);
    497     delta_minus->AssignBignum(*power_ten);
    498   }
    499 
    500   // numerator = significand * 2 * 10^-estimated_power
    501   //  since v = significand * 2^exponent this is equivalent to
    502   // numerator = v * 10^-estimated_power * 2 * 2^-exponent.
    503   // Remember: numerator has been abused as power_ten. So no need to assign it
    504   //  to itself.
    505   DCHECK(numerator == power_ten);
    506   numerator->MultiplyByUInt64(significand);
    507 
    508   // denominator = 2 * 2^-exponent with exponent < 0.
    509   denominator->AssignUInt16(1);
    510   denominator->ShiftLeft(-exponent);
    511 
    512   if (need_boundary_deltas) {
    513     // Introduce a common denominator so that the deltas to the boundaries are
    514     // integers.
    515     numerator->ShiftLeft(1);
    516     denominator->ShiftLeft(1);
    517     // With this shift the boundaries have their correct value, since
    518     // delta_plus = 10^-estimated_power, and
    519     // delta_minus = 10^-estimated_power.
    520     // These assignments have been done earlier.
    521 
    522     // The special case where the lower boundary is twice as close.
    523     // This time we have to look out for the exception too.
    524     uint64_t v_bits = Double(v).AsUint64();
    525     if ((v_bits & Double::kSignificandMask) == 0 &&
    526         // The only exception where a significand == 0 has its boundaries at
    527         // "normal" distances:
    528         (v_bits & Double::kExponentMask) != kMinimalNormalizedExponent) {
    529       numerator->ShiftLeft(1);    // *2
    530       denominator->ShiftLeft(1);  // *2
    531       delta_plus->ShiftLeft(1);   // *2
    532     }
    533   }
    534 }
    535 
    536 
    537 // Let v = significand * 2^exponent.
    538 // Computes v / 10^estimated_power exactly, as a ratio of two bignums, numerator
    539 // and denominator. The functions GenerateShortestDigits and
    540 // GenerateCountedDigits will then convert this ratio to its decimal
    541 // representation d, with the required accuracy.
    542 // Then d * 10^estimated_power is the representation of v.
    543 // (Note: the fraction and the estimated_power might get adjusted before
    544 // generating the decimal representation.)
    545 //
    546 // The initial start values consist of:
    547 //  - a scaled numerator: s.t. numerator/denominator == v / 10^estimated_power.
    548 //  - a scaled (common) denominator.
    549 //  optionally (used by GenerateShortestDigits to decide if it has the shortest
    550 //  decimal converting back to v):
    551 //  - v - m-: the distance to the lower boundary.
    552 //  - m+ - v: the distance to the upper boundary.
    553 //
    554 // v, m+, m-, and therefore v - m- and m+ - v all share the same denominator.
    555 //
    556 // Let ep == estimated_power, then the returned values will satisfy:
    557 //  v / 10^ep = numerator / denominator.
    558 //  v's boundarys m- and m+:
    559 //    m- / 10^ep == v / 10^ep - delta_minus / denominator
    560 //    m+ / 10^ep == v / 10^ep + delta_plus / denominator
    561 //  Or in other words:
    562 //    m- == v - delta_minus * 10^ep / denominator;
    563 //    m+ == v + delta_plus * 10^ep / denominator;
    564 //
    565 // Since 10^(k-1) <= v < 10^k    (with k == estimated_power)
    566 //  or       10^k <= v < 10^(k+1)
    567 //  we then have 0.1 <= numerator/denominator < 1
    568 //           or    1 <= numerator/denominator < 10
    569 //
    570 // It is then easy to kickstart the digit-generation routine.
    571 //
    572 // The boundary-deltas are only filled if need_boundary_deltas is set.
    573 static void InitialScaledStartValues(double v,
    574                                      int estimated_power,
    575                                      bool need_boundary_deltas,
    576                                      Bignum* numerator,
    577                                      Bignum* denominator,
    578                                      Bignum* delta_minus,
    579                                      Bignum* delta_plus) {
    580   if (Double(v).Exponent() >= 0) {
    581     InitialScaledStartValuesPositiveExponent(
    582         v, estimated_power, need_boundary_deltas,
    583         numerator, denominator, delta_minus, delta_plus);
    584   } else if (estimated_power >= 0) {
    585     InitialScaledStartValuesNegativeExponentPositivePower(
    586         v, estimated_power, need_boundary_deltas,
    587         numerator, denominator, delta_minus, delta_plus);
    588   } else {
    589     InitialScaledStartValuesNegativeExponentNegativePower(
    590         v, estimated_power, need_boundary_deltas,
    591         numerator, denominator, delta_minus, delta_plus);
    592   }
    593 }
    594 
    595 
    596 // This routine multiplies numerator/denominator so that its values lies in the
    597 // range 1-10. That is after a call to this function we have:
    598 //    1 <= (numerator + delta_plus) /denominator < 10.
    599 // Let numerator the input before modification and numerator' the argument
    600 // after modification, then the output-parameter decimal_point is such that
    601 //  numerator / denominator * 10^estimated_power ==
    602 //    numerator' / denominator' * 10^(decimal_point - 1)
    603 // In some cases estimated_power was too low, and this is already the case. We
    604 // then simply adjust the power so that 10^(k-1) <= v < 10^k (with k ==
    605 // estimated_power) but do not touch the numerator or denominator.
    606 // Otherwise the routine multiplies the numerator and the deltas by 10.
    607 static void FixupMultiply10(int estimated_power, bool is_even,
    608                             int* decimal_point,
    609                             Bignum* numerator, Bignum* denominator,
    610                             Bignum* delta_minus, Bignum* delta_plus) {
    611   bool in_range;
    612   if (is_even) {
    613     // For IEEE doubles half-way cases (in decimal system numbers ending with 5)
    614     // are rounded to the closest floating-point number with even significand.
    615     in_range = Bignum::PlusCompare(*numerator, *delta_plus, *denominator) >= 0;
    616   } else {
    617     in_range = Bignum::PlusCompare(*numerator, *delta_plus, *denominator) > 0;
    618   }
    619   if (in_range) {
    620     // Since numerator + delta_plus >= denominator we already have
    621     // 1 <= numerator/denominator < 10. Simply update the estimated_power.
    622     *decimal_point = estimated_power + 1;
    623   } else {
    624     *decimal_point = estimated_power;
    625     numerator->Times10();
    626     if (Bignum::Equal(*delta_minus, *delta_plus)) {
    627       delta_minus->Times10();
    628       delta_plus->AssignBignum(*delta_minus);
    629     } else {
    630       delta_minus->Times10();
    631       delta_plus->Times10();
    632     }
    633   }
    634 }
    635 
    636 } }  // namespace v8::internal
    637