/src/icu/source/i18n/double-conversion-strtod.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | // © 2018 and later: Unicode, Inc. and others. |
2 | | // License & terms of use: http://www.unicode.org/copyright.html |
3 | | // |
4 | | // From the double-conversion library. Original license: |
5 | | // |
6 | | // Copyright 2010 the V8 project authors. All rights reserved. |
7 | | // Redistribution and use in source and binary forms, with or without |
8 | | // modification, are permitted provided that the following conditions are |
9 | | // met: |
10 | | // |
11 | | // * Redistributions of source code must retain the above copyright |
12 | | // notice, this list of conditions and the following disclaimer. |
13 | | // * Redistributions in binary form must reproduce the above |
14 | | // copyright notice, this list of conditions and the following |
15 | | // disclaimer in the documentation and/or other materials provided |
16 | | // with the distribution. |
17 | | // * Neither the name of Google Inc. nor the names of its |
18 | | // contributors may be used to endorse or promote products derived |
19 | | // from this software without specific prior written permission. |
20 | | // |
21 | | // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
22 | | // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
23 | | // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR |
24 | | // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT |
25 | | // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, |
26 | | // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT |
27 | | // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, |
28 | | // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY |
29 | | // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
30 | | // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
31 | | // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
32 | | |
33 | | // ICU PATCH: ifdef around UCONFIG_NO_FORMATTING |
34 | | #include "unicode/utypes.h" |
35 | | #if !UCONFIG_NO_FORMATTING |
36 | | |
37 | | #include <climits> |
38 | | #include <cstdarg> |
39 | | |
40 | | // ICU PATCH: Customize header file paths for ICU. |
41 | | |
42 | | #include "double-conversion-bignum.h" |
43 | | #include "double-conversion-cached-powers.h" |
44 | | #include "double-conversion-ieee.h" |
45 | | #include "double-conversion-strtod.h" |
46 | | |
47 | | // ICU PATCH: Wrap in ICU namespace |
48 | | U_NAMESPACE_BEGIN |
49 | | |
50 | | namespace double_conversion { |
51 | | |
52 | | #if defined(DOUBLE_CONVERSION_CORRECT_DOUBLE_OPERATIONS) |
53 | | // 2^53 = 9007199254740992. |
54 | | // Any integer with at most 15 decimal digits will hence fit into a double |
55 | | // (which has a 53bit significand) without loss of precision. |
56 | | static const int kMaxExactDoubleIntegerDecimalDigits = 15; |
57 | | #endif // #if defined(DOUBLE_CONVERSION_CORRECT_DOUBLE_OPERATIONS) |
58 | | // 2^64 = 18446744073709551616 > 10^19 |
59 | | static const int kMaxUint64DecimalDigits = 19; |
60 | | |
61 | | // Max double: 1.7976931348623157 x 10^308 |
62 | | // Min non-zero double: 4.9406564584124654 x 10^-324 |
63 | | // Any x >= 10^309 is interpreted as +infinity. |
64 | | // Any x <= 10^-324 is interpreted as 0. |
65 | | // Note that 2.5e-324 (despite being smaller than the min double) will be read |
66 | | // as non-zero (equal to the min non-zero double). |
67 | | static const int kMaxDecimalPower = 309; |
68 | | static const int kMinDecimalPower = -324; |
69 | | |
70 | | // 2^64 = 18446744073709551616 |
71 | | static const uint64_t kMaxUint64 = DOUBLE_CONVERSION_UINT64_2PART_C(0xFFFFFFFF, FFFFFFFF); |
72 | | |
73 | | |
74 | | #if defined(DOUBLE_CONVERSION_CORRECT_DOUBLE_OPERATIONS) |
75 | | static const double exact_powers_of_ten[] = { |
76 | | 1.0, // 10^0 |
77 | | 10.0, |
78 | | 100.0, |
79 | | 1000.0, |
80 | | 10000.0, |
81 | | 100000.0, |
82 | | 1000000.0, |
83 | | 10000000.0, |
84 | | 100000000.0, |
85 | | 1000000000.0, |
86 | | 10000000000.0, // 10^10 |
87 | | 100000000000.0, |
88 | | 1000000000000.0, |
89 | | 10000000000000.0, |
90 | | 100000000000000.0, |
91 | | 1000000000000000.0, |
92 | | 10000000000000000.0, |
93 | | 100000000000000000.0, |
94 | | 1000000000000000000.0, |
95 | | 10000000000000000000.0, |
96 | | 100000000000000000000.0, // 10^20 |
97 | | 1000000000000000000000.0, |
98 | | // 10^22 = 0x21e19e0c9bab2400000 = 0x878678326eac9 * 2^22 |
99 | | 10000000000000000000000.0 |
100 | | }; |
101 | | static const int kExactPowersOfTenSize = DOUBLE_CONVERSION_ARRAY_SIZE(exact_powers_of_ten); |
102 | | #endif // #if defined(DOUBLE_CONVERSION_CORRECT_DOUBLE_OPERATIONS) |
103 | | |
104 | | // Maximum number of significant digits in the decimal representation. |
105 | | // In fact the value is 772 (see conversions.cc), but to give us some margin |
106 | | // we round up to 780. |
107 | | static const int kMaxSignificantDecimalDigits = 780; |
108 | | |
109 | 0 | static Vector<const char> TrimLeadingZeros(Vector<const char> buffer) { |
110 | 0 | for (int i = 0; i < buffer.length(); i++) { |
111 | 0 | if (buffer[i] != '0') { |
112 | 0 | return buffer.SubVector(i, buffer.length()); |
113 | 0 | } |
114 | 0 | } |
115 | 0 | return Vector<const char>(buffer.start(), 0); |
116 | 0 | } |
117 | | |
118 | | static void CutToMaxSignificantDigits(Vector<const char> buffer, |
119 | | int exponent, |
120 | | char* significant_buffer, |
121 | 0 | int* significant_exponent) { |
122 | 0 | for (int i = 0; i < kMaxSignificantDecimalDigits - 1; ++i) { |
123 | 0 | significant_buffer[i] = buffer[i]; |
124 | 0 | } |
125 | | // The input buffer has been trimmed. Therefore the last digit must be |
126 | | // different from '0'. |
127 | 0 | DOUBLE_CONVERSION_ASSERT(buffer[buffer.length() - 1] != '0'); |
128 | | // Set the last digit to be non-zero. This is sufficient to guarantee |
129 | | // correct rounding. |
130 | 0 | significant_buffer[kMaxSignificantDecimalDigits - 1] = '1'; |
131 | 0 | *significant_exponent = |
132 | 0 | exponent + (buffer.length() - kMaxSignificantDecimalDigits); |
133 | 0 | } |
134 | | |
135 | | |
136 | | // Trims the buffer and cuts it to at most kMaxSignificantDecimalDigits. |
137 | | // If possible the input-buffer is reused, but if the buffer needs to be |
138 | | // modified (due to cutting), then the input needs to be copied into the |
139 | | // buffer_copy_space. |
140 | | static void TrimAndCut(Vector<const char> buffer, int exponent, |
141 | | char* buffer_copy_space, int space_size, |
142 | 0 | Vector<const char>* trimmed, int* updated_exponent) { |
143 | 0 | Vector<const char> left_trimmed = TrimLeadingZeros(buffer); |
144 | 0 | Vector<const char> right_trimmed = TrimTrailingZeros(left_trimmed); |
145 | 0 | exponent += left_trimmed.length() - right_trimmed.length(); |
146 | 0 | if (right_trimmed.length() > kMaxSignificantDecimalDigits) { |
147 | 0 | (void) space_size; // Mark variable as used. |
148 | 0 | DOUBLE_CONVERSION_ASSERT(space_size >= kMaxSignificantDecimalDigits); |
149 | 0 | CutToMaxSignificantDigits(right_trimmed, exponent, |
150 | 0 | buffer_copy_space, updated_exponent); |
151 | 0 | *trimmed = Vector<const char>(buffer_copy_space, |
152 | 0 | kMaxSignificantDecimalDigits); |
153 | 0 | } else { |
154 | 0 | *trimmed = right_trimmed; |
155 | 0 | *updated_exponent = exponent; |
156 | 0 | } |
157 | 0 | } |
158 | | |
159 | | |
160 | | // Reads digits from the buffer and converts them to a uint64. |
161 | | // Reads in as many digits as fit into a uint64. |
162 | | // When the string starts with "1844674407370955161" no further digit is read. |
163 | | // Since 2^64 = 18446744073709551616 it would still be possible read another |
164 | | // digit if it was less or equal than 6, but this would complicate the code. |
165 | | static uint64_t ReadUint64(Vector<const char> buffer, |
166 | 0 | int* number_of_read_digits) { |
167 | 0 | uint64_t result = 0; |
168 | 0 | int i = 0; |
169 | 0 | while (i < buffer.length() && result <= (kMaxUint64 / 10 - 1)) { |
170 | 0 | int digit = buffer[i++] - '0'; |
171 | 0 | DOUBLE_CONVERSION_ASSERT(0 <= digit && digit <= 9); |
172 | 0 | result = 10 * result + digit; |
173 | 0 | } |
174 | 0 | *number_of_read_digits = i; |
175 | 0 | return result; |
176 | 0 | } |
177 | | |
178 | | |
179 | | // Reads a DiyFp from the buffer. |
180 | | // The returned DiyFp is not necessarily normalized. |
181 | | // If remaining_decimals is zero then the returned DiyFp is accurate. |
182 | | // Otherwise it has been rounded and has error of at most 1/2 ulp. |
183 | | static void ReadDiyFp(Vector<const char> buffer, |
184 | | DiyFp* result, |
185 | 0 | int* remaining_decimals) { |
186 | 0 | int read_digits; |
187 | 0 | uint64_t significand = ReadUint64(buffer, &read_digits); |
188 | 0 | if (buffer.length() == read_digits) { |
189 | 0 | *result = DiyFp(significand, 0); |
190 | 0 | *remaining_decimals = 0; |
191 | 0 | } else { |
192 | | // Round the significand. |
193 | 0 | if (buffer[read_digits] >= '5') { |
194 | 0 | significand++; |
195 | 0 | } |
196 | | // Compute the binary exponent. |
197 | 0 | int exponent = 0; |
198 | 0 | *result = DiyFp(significand, exponent); |
199 | 0 | *remaining_decimals = buffer.length() - read_digits; |
200 | 0 | } |
201 | 0 | } |
202 | | |
203 | | |
204 | | static bool DoubleStrtod(Vector<const char> trimmed, |
205 | | int exponent, |
206 | 0 | double* result) { |
207 | | #if !defined(DOUBLE_CONVERSION_CORRECT_DOUBLE_OPERATIONS) |
208 | | // Avoid "unused parameter" warnings |
209 | | (void) trimmed; |
210 | | (void) exponent; |
211 | | (void) result; |
212 | | // On x86 the floating-point stack can be 64 or 80 bits wide. If it is |
213 | | // 80 bits wide (as is the case on Linux) then double-rounding occurs and the |
214 | | // result is not accurate. |
215 | | // We know that Windows32 uses 64 bits and is therefore accurate. |
216 | | return false; |
217 | | #else |
218 | 0 | if (trimmed.length() <= kMaxExactDoubleIntegerDecimalDigits) { |
219 | 0 | int read_digits; |
220 | | // The trimmed input fits into a double. |
221 | | // If the 10^exponent (resp. 10^-exponent) fits into a double too then we |
222 | | // can compute the result-double simply by multiplying (resp. dividing) the |
223 | | // two numbers. |
224 | | // This is possible because IEEE guarantees that floating-point operations |
225 | | // return the best possible approximation. |
226 | 0 | if (exponent < 0 && -exponent < kExactPowersOfTenSize) { |
227 | | // 10^-exponent fits into a double. |
228 | 0 | *result = static_cast<double>(ReadUint64(trimmed, &read_digits)); |
229 | 0 | DOUBLE_CONVERSION_ASSERT(read_digits == trimmed.length()); |
230 | 0 | *result /= exact_powers_of_ten[-exponent]; |
231 | 0 | return true; |
232 | 0 | } |
233 | 0 | if (0 <= exponent && exponent < kExactPowersOfTenSize) { |
234 | | // 10^exponent fits into a double. |
235 | 0 | *result = static_cast<double>(ReadUint64(trimmed, &read_digits)); |
236 | 0 | DOUBLE_CONVERSION_ASSERT(read_digits == trimmed.length()); |
237 | 0 | *result *= exact_powers_of_ten[exponent]; |
238 | 0 | return true; |
239 | 0 | } |
240 | 0 | int remaining_digits = |
241 | 0 | kMaxExactDoubleIntegerDecimalDigits - trimmed.length(); |
242 | 0 | if ((0 <= exponent) && |
243 | 0 | (exponent - remaining_digits < kExactPowersOfTenSize)) { |
244 | | // The trimmed string was short and we can multiply it with |
245 | | // 10^remaining_digits. As a result the remaining exponent now fits |
246 | | // into a double too. |
247 | 0 | *result = static_cast<double>(ReadUint64(trimmed, &read_digits)); |
248 | 0 | DOUBLE_CONVERSION_ASSERT(read_digits == trimmed.length()); |
249 | 0 | *result *= exact_powers_of_ten[remaining_digits]; |
250 | 0 | *result *= exact_powers_of_ten[exponent - remaining_digits]; |
251 | 0 | return true; |
252 | 0 | } |
253 | 0 | } |
254 | 0 | return false; |
255 | 0 | #endif |
256 | 0 | } |
257 | | |
258 | | |
259 | | // Returns 10^exponent as an exact DiyFp. |
260 | | // The given exponent must be in the range [1; kDecimalExponentDistance[. |
261 | 0 | static DiyFp AdjustmentPowerOfTen(int exponent) { |
262 | 0 | DOUBLE_CONVERSION_ASSERT(0 < exponent); |
263 | 0 | DOUBLE_CONVERSION_ASSERT(exponent < PowersOfTenCache::kDecimalExponentDistance); |
264 | | // Simply hardcode the remaining powers for the given decimal exponent |
265 | | // distance. |
266 | 0 | DOUBLE_CONVERSION_ASSERT(PowersOfTenCache::kDecimalExponentDistance == 8); |
267 | 0 | switch (exponent) { |
268 | 0 | case 1: return DiyFp(DOUBLE_CONVERSION_UINT64_2PART_C(0xa0000000, 00000000), -60); |
269 | 0 | case 2: return DiyFp(DOUBLE_CONVERSION_UINT64_2PART_C(0xc8000000, 00000000), -57); |
270 | 0 | case 3: return DiyFp(DOUBLE_CONVERSION_UINT64_2PART_C(0xfa000000, 00000000), -54); |
271 | 0 | case 4: return DiyFp(DOUBLE_CONVERSION_UINT64_2PART_C(0x9c400000, 00000000), -50); |
272 | 0 | case 5: return DiyFp(DOUBLE_CONVERSION_UINT64_2PART_C(0xc3500000, 00000000), -47); |
273 | 0 | case 6: return DiyFp(DOUBLE_CONVERSION_UINT64_2PART_C(0xf4240000, 00000000), -44); |
274 | 0 | case 7: return DiyFp(DOUBLE_CONVERSION_UINT64_2PART_C(0x98968000, 00000000), -40); |
275 | 0 | default: |
276 | 0 | DOUBLE_CONVERSION_UNREACHABLE(); |
277 | 0 | } |
278 | 0 | } |
279 | | |
280 | | |
281 | | // If the function returns true then the result is the correct double. |
282 | | // Otherwise it is either the correct double or the double that is just below |
283 | | // the correct double. |
284 | | static bool DiyFpStrtod(Vector<const char> buffer, |
285 | | int exponent, |
286 | 0 | double* result) { |
287 | 0 | DiyFp input; |
288 | 0 | int remaining_decimals; |
289 | 0 | ReadDiyFp(buffer, &input, &remaining_decimals); |
290 | | // Since we may have dropped some digits the input is not accurate. |
291 | | // If remaining_decimals is different than 0 than the error is at most |
292 | | // .5 ulp (unit in the last place). |
293 | | // We don't want to deal with fractions and therefore keep a common |
294 | | // denominator. |
295 | 0 | const int kDenominatorLog = 3; |
296 | 0 | const int kDenominator = 1 << kDenominatorLog; |
297 | | // Move the remaining decimals into the exponent. |
298 | 0 | exponent += remaining_decimals; |
299 | 0 | uint64_t error = (remaining_decimals == 0 ? 0 : kDenominator / 2); |
300 | |
|
301 | 0 | int old_e = input.e(); |
302 | 0 | input.Normalize(); |
303 | 0 | error <<= old_e - input.e(); |
304 | |
|
305 | 0 | DOUBLE_CONVERSION_ASSERT(exponent <= PowersOfTenCache::kMaxDecimalExponent); |
306 | 0 | if (exponent < PowersOfTenCache::kMinDecimalExponent) { |
307 | 0 | *result = 0.0; |
308 | 0 | return true; |
309 | 0 | } |
310 | 0 | DiyFp cached_power; |
311 | 0 | int cached_decimal_exponent; |
312 | 0 | PowersOfTenCache::GetCachedPowerForDecimalExponent(exponent, |
313 | 0 | &cached_power, |
314 | 0 | &cached_decimal_exponent); |
315 | |
|
316 | 0 | if (cached_decimal_exponent != exponent) { |
317 | 0 | int adjustment_exponent = exponent - cached_decimal_exponent; |
318 | 0 | DiyFp adjustment_power = AdjustmentPowerOfTen(adjustment_exponent); |
319 | 0 | input.Multiply(adjustment_power); |
320 | 0 | if (kMaxUint64DecimalDigits - buffer.length() >= adjustment_exponent) { |
321 | | // The product of input with the adjustment power fits into a 64 bit |
322 | | // integer. |
323 | 0 | DOUBLE_CONVERSION_ASSERT(DiyFp::kSignificandSize == 64); |
324 | 0 | } else { |
325 | | // The adjustment power is exact. There is hence only an error of 0.5. |
326 | 0 | error += kDenominator / 2; |
327 | 0 | } |
328 | 0 | } |
329 | |
|
330 | 0 | input.Multiply(cached_power); |
331 | | // The error introduced by a multiplication of a*b equals |
332 | | // error_a + error_b + error_a*error_b/2^64 + 0.5 |
333 | | // Substituting a with 'input' and b with 'cached_power' we have |
334 | | // error_b = 0.5 (all cached powers have an error of less than 0.5 ulp), |
335 | | // error_ab = 0 or 1 / kDenominator > error_a*error_b/ 2^64 |
336 | 0 | int error_b = kDenominator / 2; |
337 | 0 | int error_ab = (error == 0 ? 0 : 1); // We round up to 1. |
338 | 0 | int fixed_error = kDenominator / 2; |
339 | 0 | error += error_b + error_ab + fixed_error; |
340 | |
|
341 | 0 | old_e = input.e(); |
342 | 0 | input.Normalize(); |
343 | 0 | error <<= old_e - input.e(); |
344 | | |
345 | | // See if the double's significand changes if we add/subtract the error. |
346 | 0 | int order_of_magnitude = DiyFp::kSignificandSize + input.e(); |
347 | 0 | int effective_significand_size = |
348 | 0 | Double::SignificandSizeForOrderOfMagnitude(order_of_magnitude); |
349 | 0 | int precision_digits_count = |
350 | 0 | DiyFp::kSignificandSize - effective_significand_size; |
351 | 0 | if (precision_digits_count + kDenominatorLog >= DiyFp::kSignificandSize) { |
352 | | // This can only happen for very small denormals. In this case the |
353 | | // half-way multiplied by the denominator exceeds the range of an uint64. |
354 | | // Simply shift everything to the right. |
355 | 0 | int shift_amount = (precision_digits_count + kDenominatorLog) - |
356 | 0 | DiyFp::kSignificandSize + 1; |
357 | 0 | input.set_f(input.f() >> shift_amount); |
358 | 0 | input.set_e(input.e() + shift_amount); |
359 | | // We add 1 for the lost precision of error, and kDenominator for |
360 | | // the lost precision of input.f(). |
361 | 0 | error = (error >> shift_amount) + 1 + kDenominator; |
362 | 0 | precision_digits_count -= shift_amount; |
363 | 0 | } |
364 | | // We use uint64_ts now. This only works if the DiyFp uses uint64_ts too. |
365 | 0 | DOUBLE_CONVERSION_ASSERT(DiyFp::kSignificandSize == 64); |
366 | 0 | DOUBLE_CONVERSION_ASSERT(precision_digits_count < 64); |
367 | 0 | uint64_t one64 = 1; |
368 | 0 | uint64_t precision_bits_mask = (one64 << precision_digits_count) - 1; |
369 | 0 | uint64_t precision_bits = input.f() & precision_bits_mask; |
370 | 0 | uint64_t half_way = one64 << (precision_digits_count - 1); |
371 | 0 | precision_bits *= kDenominator; |
372 | 0 | half_way *= kDenominator; |
373 | 0 | DiyFp rounded_input(input.f() >> precision_digits_count, |
374 | 0 | input.e() + precision_digits_count); |
375 | 0 | if (precision_bits >= half_way + error) { |
376 | 0 | rounded_input.set_f(rounded_input.f() + 1); |
377 | 0 | } |
378 | | // If the last_bits are too close to the half-way case than we are too |
379 | | // inaccurate and round down. In this case we return false so that we can |
380 | | // fall back to a more precise algorithm. |
381 | |
|
382 | 0 | *result = Double(rounded_input).value(); |
383 | 0 | if (half_way - error < precision_bits && precision_bits < half_way + error) { |
384 | | // Too imprecise. The caller will have to fall back to a slower version. |
385 | | // However the returned number is guaranteed to be either the correct |
386 | | // double, or the next-lower double. |
387 | 0 | return false; |
388 | 0 | } else { |
389 | 0 | return true; |
390 | 0 | } |
391 | 0 | } |
392 | | |
393 | | |
394 | | // Returns |
395 | | // - -1 if buffer*10^exponent < diy_fp. |
396 | | // - 0 if buffer*10^exponent == diy_fp. |
397 | | // - +1 if buffer*10^exponent > diy_fp. |
398 | | // Preconditions: |
399 | | // buffer.length() + exponent <= kMaxDecimalPower + 1 |
400 | | // buffer.length() + exponent > kMinDecimalPower |
401 | | // buffer.length() <= kMaxDecimalSignificantDigits |
402 | | static int CompareBufferWithDiyFp(Vector<const char> buffer, |
403 | | int exponent, |
404 | 0 | DiyFp diy_fp) { |
405 | 0 | DOUBLE_CONVERSION_ASSERT(buffer.length() + exponent <= kMaxDecimalPower + 1); |
406 | 0 | DOUBLE_CONVERSION_ASSERT(buffer.length() + exponent > kMinDecimalPower); |
407 | 0 | DOUBLE_CONVERSION_ASSERT(buffer.length() <= kMaxSignificantDecimalDigits); |
408 | | // Make sure that the Bignum will be able to hold all our numbers. |
409 | | // Our Bignum implementation has a separate field for exponents. Shifts will |
410 | | // consume at most one bigit (< 64 bits). |
411 | | // ln(10) == 3.3219... |
412 | 0 | DOUBLE_CONVERSION_ASSERT(((kMaxDecimalPower + 1) * 333 / 100) < Bignum::kMaxSignificantBits); |
413 | 0 | Bignum buffer_bignum; |
414 | 0 | Bignum diy_fp_bignum; |
415 | 0 | buffer_bignum.AssignDecimalString(buffer); |
416 | 0 | diy_fp_bignum.AssignUInt64(diy_fp.f()); |
417 | 0 | if (exponent >= 0) { |
418 | 0 | buffer_bignum.MultiplyByPowerOfTen(exponent); |
419 | 0 | } else { |
420 | 0 | diy_fp_bignum.MultiplyByPowerOfTen(-exponent); |
421 | 0 | } |
422 | 0 | if (diy_fp.e() > 0) { |
423 | 0 | diy_fp_bignum.ShiftLeft(diy_fp.e()); |
424 | 0 | } else { |
425 | 0 | buffer_bignum.ShiftLeft(-diy_fp.e()); |
426 | 0 | } |
427 | 0 | return Bignum::Compare(buffer_bignum, diy_fp_bignum); |
428 | 0 | } |
429 | | |
430 | | |
431 | | // Returns true if the guess is the correct double. |
432 | | // Returns false, when guess is either correct or the next-lower double. |
433 | | static bool ComputeGuess(Vector<const char> trimmed, int exponent, |
434 | 0 | double* guess) { |
435 | 0 | if (trimmed.length() == 0) { |
436 | 0 | *guess = 0.0; |
437 | 0 | return true; |
438 | 0 | } |
439 | 0 | if (exponent + trimmed.length() - 1 >= kMaxDecimalPower) { |
440 | 0 | *guess = Double::Infinity(); |
441 | 0 | return true; |
442 | 0 | } |
443 | 0 | if (exponent + trimmed.length() <= kMinDecimalPower) { |
444 | 0 | *guess = 0.0; |
445 | 0 | return true; |
446 | 0 | } |
447 | | |
448 | 0 | if (DoubleStrtod(trimmed, exponent, guess) || |
449 | 0 | DiyFpStrtod(trimmed, exponent, guess)) { |
450 | 0 | return true; |
451 | 0 | } |
452 | 0 | if (*guess == Double::Infinity()) { |
453 | 0 | return true; |
454 | 0 | } |
455 | 0 | return false; |
456 | 0 | } |
457 | | |
458 | | #if U_DEBUG // needed for ICU only in debug mode |
459 | | static bool IsDigit(const char d) { |
460 | | return ('0' <= d) && (d <= '9'); |
461 | | } |
462 | | |
463 | | static bool IsNonZeroDigit(const char d) { |
464 | | return ('1' <= d) && (d <= '9'); |
465 | | } |
466 | | |
467 | | #ifdef __has_cpp_attribute |
468 | | #if __has_cpp_attribute(maybe_unused) |
469 | | [[maybe_unused]] |
470 | | #endif |
471 | | #endif |
472 | | static bool AssertTrimmedDigits(const Vector<const char>& buffer) { |
473 | | for(int i = 0; i < buffer.length(); ++i) { |
474 | | if(!IsDigit(buffer[i])) { |
475 | | return false; |
476 | | } |
477 | | } |
478 | | return (buffer.length() == 0) || (IsNonZeroDigit(buffer[0]) && IsNonZeroDigit(buffer[buffer.length()-1])); |
479 | | } |
480 | | #endif // needed for ICU only in debug mode |
481 | | |
482 | 0 | double StrtodTrimmed(Vector<const char> trimmed, int exponent) { |
483 | 0 | DOUBLE_CONVERSION_ASSERT(trimmed.length() <= kMaxSignificantDecimalDigits); |
484 | 0 | DOUBLE_CONVERSION_ASSERT(AssertTrimmedDigits(trimmed)); |
485 | 0 | double guess; |
486 | 0 | const bool is_correct = ComputeGuess(trimmed, exponent, &guess); |
487 | 0 | if (is_correct) { |
488 | 0 | return guess; |
489 | 0 | } |
490 | 0 | DiyFp upper_boundary = Double(guess).UpperBoundary(); |
491 | 0 | int comparison = CompareBufferWithDiyFp(trimmed, exponent, upper_boundary); |
492 | 0 | if (comparison < 0) { |
493 | 0 | return guess; |
494 | 0 | } else if (comparison > 0) { |
495 | 0 | return Double(guess).NextDouble(); |
496 | 0 | } else if ((Double(guess).Significand() & 1) == 0) { |
497 | | // Round towards even. |
498 | 0 | return guess; |
499 | 0 | } else { |
500 | 0 | return Double(guess).NextDouble(); |
501 | 0 | } |
502 | 0 | } |
503 | | |
504 | 0 | double Strtod(Vector<const char> buffer, int exponent) { |
505 | 0 | char copy_buffer[kMaxSignificantDecimalDigits]; |
506 | 0 | Vector<const char> trimmed; |
507 | 0 | int updated_exponent; |
508 | 0 | TrimAndCut(buffer, exponent, copy_buffer, kMaxSignificantDecimalDigits, |
509 | 0 | &trimmed, &updated_exponent); |
510 | 0 | return StrtodTrimmed(trimmed, updated_exponent); |
511 | 0 | } |
512 | | |
513 | 0 | static float SanitizedDoubletof(double d) { |
514 | 0 | DOUBLE_CONVERSION_ASSERT(d >= 0.0); |
515 | | // ASAN has a sanitize check that disallows casting doubles to floats if |
516 | | // they are too big. |
517 | | // https://clang.llvm.org/docs/UndefinedBehaviorSanitizer.html#available-checks |
518 | | // The behavior should be covered by IEEE 754, but some projects use this |
519 | | // flag, so work around it. |
520 | 0 | float max_finite = 3.4028234663852885981170418348451692544e+38; |
521 | | // The half-way point between the max-finite and infinity value. |
522 | | // Since infinity has an even significand everything equal or greater than |
523 | | // this value should become infinity. |
524 | 0 | double half_max_finite_infinity = |
525 | 0 | 3.40282356779733661637539395458142568448e+38; |
526 | 0 | if (d >= max_finite) { |
527 | 0 | if (d >= half_max_finite_infinity) { |
528 | 0 | return Single::Infinity(); |
529 | 0 | } else { |
530 | 0 | return max_finite; |
531 | 0 | } |
532 | 0 | } else { |
533 | 0 | return static_cast<float>(d); |
534 | 0 | } |
535 | 0 | } |
536 | | |
537 | 0 | float Strtof(Vector<const char> buffer, int exponent) { |
538 | 0 | char copy_buffer[kMaxSignificantDecimalDigits]; |
539 | 0 | Vector<const char> trimmed; |
540 | 0 | int updated_exponent; |
541 | 0 | TrimAndCut(buffer, exponent, copy_buffer, kMaxSignificantDecimalDigits, |
542 | 0 | &trimmed, &updated_exponent); |
543 | 0 | exponent = updated_exponent; |
544 | 0 | return StrtofTrimmed(trimmed, exponent); |
545 | 0 | } |
546 | | |
547 | 0 | float StrtofTrimmed(Vector<const char> trimmed, int exponent) { |
548 | 0 | DOUBLE_CONVERSION_ASSERT(trimmed.length() <= kMaxSignificantDecimalDigits); |
549 | 0 | DOUBLE_CONVERSION_ASSERT(AssertTrimmedDigits(trimmed)); |
550 | |
|
551 | 0 | double double_guess; |
552 | 0 | bool is_correct = ComputeGuess(trimmed, exponent, &double_guess); |
553 | |
|
554 | 0 | float float_guess = SanitizedDoubletof(double_guess); |
555 | 0 | if (float_guess == double_guess) { |
556 | | // This shortcut triggers for integer values. |
557 | 0 | return float_guess; |
558 | 0 | } |
559 | | |
560 | | // We must catch double-rounding. Say the double has been rounded up, and is |
561 | | // now a boundary of a float, and rounds up again. This is why we have to |
562 | | // look at previous too. |
563 | | // Example (in decimal numbers): |
564 | | // input: 12349 |
565 | | // high-precision (4 digits): 1235 |
566 | | // low-precision (3 digits): |
567 | | // when read from input: 123 |
568 | | // when rounded from high precision: 124. |
569 | | // To do this we simply look at the neighbors of the correct result and see |
570 | | // if they would round to the same float. If the guess is not correct we have |
571 | | // to look at four values (since two different doubles could be the correct |
572 | | // double). |
573 | | |
574 | 0 | double double_next = Double(double_guess).NextDouble(); |
575 | 0 | double double_previous = Double(double_guess).PreviousDouble(); |
576 | |
|
577 | 0 | float f1 = SanitizedDoubletof(double_previous); |
578 | 0 | float f2 = float_guess; |
579 | 0 | float f3 = SanitizedDoubletof(double_next); |
580 | 0 | float f4; |
581 | 0 | if (is_correct) { |
582 | 0 | f4 = f3; |
583 | 0 | } else { |
584 | 0 | double double_next2 = Double(double_next).NextDouble(); |
585 | 0 | f4 = SanitizedDoubletof(double_next2); |
586 | 0 | } |
587 | 0 | (void) f2; // Mark variable as used. |
588 | 0 | DOUBLE_CONVERSION_ASSERT(f1 <= f2 && f2 <= f3 && f3 <= f4); |
589 | | |
590 | | // If the guess doesn't lie near a single-precision boundary we can simply |
591 | | // return its float-value. |
592 | 0 | if (f1 == f4) { |
593 | 0 | return float_guess; |
594 | 0 | } |
595 | | |
596 | 0 | DOUBLE_CONVERSION_ASSERT((f1 != f2 && f2 == f3 && f3 == f4) || |
597 | 0 | (f1 == f2 && f2 != f3 && f3 == f4) || |
598 | 0 | (f1 == f2 && f2 == f3 && f3 != f4)); |
599 | | |
600 | | // guess and next are the two possible candidates (in the same way that |
601 | | // double_guess was the lower candidate for a double-precision guess). |
602 | 0 | float guess = f1; |
603 | 0 | float next = f4; |
604 | 0 | DiyFp upper_boundary; |
605 | 0 | if (guess == 0.0f) { |
606 | 0 | float min_float = 1e-45f; |
607 | 0 | upper_boundary = Double(static_cast<double>(min_float) / 2).AsDiyFp(); |
608 | 0 | } else { |
609 | 0 | upper_boundary = Single(guess).UpperBoundary(); |
610 | 0 | } |
611 | 0 | int comparison = CompareBufferWithDiyFp(trimmed, exponent, upper_boundary); |
612 | 0 | if (comparison < 0) { |
613 | 0 | return guess; |
614 | 0 | } else if (comparison > 0) { |
615 | 0 | return next; |
616 | 0 | } else if ((Single(guess).Significand() & 1) == 0) { |
617 | | // Round towards even. |
618 | 0 | return guess; |
619 | 0 | } else { |
620 | 0 | return next; |
621 | 0 | } |
622 | 0 | } |
623 | | |
624 | | } // namespace double_conversion |
625 | | |
626 | | // ICU PATCH: Close ICU namespace |
627 | | U_NAMESPACE_END |
628 | | #endif // ICU PATCH: close #if !UCONFIG_NO_FORMATTING |