1 // Copyright 2009 The Go Authors. All rights reserved. 2 // Use of this source code is governed by a BSD-style 3 // license that can be found in the LICENSE file. 4 5 // Binary to decimal floating point conversion. 6 // Algorithm: 7 // 1) store mantissa in multiprecision decimal 8 // 2) shift decimal by exponent 9 // 3) read digits out & format 10 11 package strconv 12 13 import "math" 14 15 // TODO: move elsewhere? 16 type floatInfo struct { 17 mantbits uint 18 expbits uint 19 bias int 20 } 21 22 var float32info = floatInfo{23, 8, -127} 23 var float64info = floatInfo{52, 11, -1023} 24 25 // FormatFloat converts the floating-point number f to a string, 26 // according to the format fmt and precision prec. It rounds the 27 // result assuming that the original was obtained from a floating-point 28 // value of bitSize bits (32 for float32, 64 for float64). 29 // 30 // The format fmt is one of 31 // 'b' (-ddddpddd, a binary exponent), 32 // 'e' (-d.ddddedd, a decimal exponent), 33 // 'E' (-d.ddddEdd, a decimal exponent), 34 // 'f' (-ddd.dddd, no exponent), 35 // 'g' ('e' for large exponents, 'f' otherwise), or 36 // 'G' ('E' for large exponents, 'f' otherwise). 37 // 38 // The precision prec controls the number of digits 39 // (excluding the exponent) printed by the 'e', 'E', 'f', 'g', and 'G' formats. 40 // For 'e', 'E', and 'f' it is the number of digits after the decimal point. 41 // For 'g' and 'G' it is the total number of digits. 42 // The special precision -1 uses the smallest number of digits 43 // necessary such that ParseFloat will return f exactly. 44 func FormatFloat(f float64, fmt byte, prec, bitSize int) string { 45 return string(genericFtoa(make([]byte, 0, max(prec+4, 24)), f, fmt, prec, bitSize)) 46 } 47 48 // AppendFloat appends the string form of the floating-point number f, 49 // as generated by FormatFloat, to dst and returns the extended buffer. 50 func AppendFloat(dst []byte, f float64, fmt byte, prec, bitSize int) []byte { 51 return genericFtoa(dst, f, fmt, prec, bitSize) 52 } 53 54 func genericFtoa(dst []byte, val float64, fmt byte, prec, bitSize int) []byte { 55 var bits uint64 56 var flt *floatInfo 57 switch bitSize { 58 case 32: 59 bits = uint64(math.Float32bits(float32(val))) 60 flt = &float32info 61 case 64: 62 bits = math.Float64bits(val) 63 flt = &float64info 64 default: 65 panic("strconv: illegal AppendFloat/FormatFloat bitSize") 66 } 67 68 neg := bits>>(flt.expbits+flt.mantbits) != 0 69 exp := int(bits>>flt.mantbits) & (1<<flt.expbits - 1) 70 mant := bits & (uint64(1)<<flt.mantbits - 1) 71 72 switch exp { 73 case 1<<flt.expbits - 1: 74 // Inf, NaN 75 var s string 76 switch { 77 case mant != 0: 78 s = "NaN" 79 case neg: 80 s = "-Inf" 81 default: 82 s = "+Inf" 83 } 84 return append(dst, s...) 85 86 case 0: 87 // denormalized 88 exp++ 89 90 default: 91 // add implicit top bit 92 mant |= uint64(1) << flt.mantbits 93 } 94 exp += flt.bias 95 96 // Pick off easy binary format. 97 if fmt == 'b' { 98 return fmtB(dst, neg, mant, exp, flt) 99 } 100 101 if !optimize { 102 return bigFtoa(dst, prec, fmt, neg, mant, exp, flt) 103 } 104 105 var digs decimalSlice 106 ok := false 107 // Negative precision means "only as much as needed to be exact." 108 shortest := prec < 0 109 if shortest { 110 // Try Grisu3 algorithm. 111 f := new(extFloat) 112 lower, upper := f.AssignComputeBounds(mant, exp, neg, flt) 113 var buf [32]byte 114 digs.d = buf[:] 115 ok = f.ShortestDecimal(&digs, &lower, &upper) 116 if !ok { 117 return bigFtoa(dst, prec, fmt, neg, mant, exp, flt) 118 } 119 // Precision for shortest representation mode. 120 switch fmt { 121 case 'e', 'E': 122 prec = max(digs.nd-1, 0) 123 case 'f': 124 prec = max(digs.nd-digs.dp, 0) 125 case 'g', 'G': 126 prec = digs.nd 127 } 128 } else if fmt != 'f' { 129 // Fixed number of digits. 130 digits := prec 131 switch fmt { 132 case 'e', 'E': 133 digits++ 134 case 'g', 'G': 135 if prec == 0 { 136 prec = 1 137 } 138 digits = prec 139 } 140 if digits <= 15 { 141 // try fast algorithm when the number of digits is reasonable. 142 var buf [24]byte 143 digs.d = buf[:] 144 f := extFloat{mant, exp - int(flt.mantbits), neg} 145 ok = f.FixedDecimal(&digs, digits) 146 } 147 } 148 if !ok { 149 return bigFtoa(dst, prec, fmt, neg, mant, exp, flt) 150 } 151 return formatDigits(dst, shortest, neg, digs, prec, fmt) 152 } 153 154 // bigFtoa uses multiprecision computations to format a float. 155 func bigFtoa(dst []byte, prec int, fmt byte, neg bool, mant uint64, exp int, flt *floatInfo) []byte { 156 d := new(decimal) 157 d.Assign(mant) 158 d.Shift(exp - int(flt.mantbits)) 159 var digs decimalSlice 160 shortest := prec < 0 161 if shortest { 162 roundShortest(d, mant, exp, flt) 163 digs = decimalSlice{d: d.d[:], nd: d.nd, dp: d.dp} 164 // Precision for shortest representation mode. 165 switch fmt { 166 case 'e', 'E': 167 prec = digs.nd - 1 168 case 'f': 169 prec = max(digs.nd-digs.dp, 0) 170 case 'g', 'G': 171 prec = digs.nd 172 } 173 } else { 174 // Round appropriately. 175 switch fmt { 176 case 'e', 'E': 177 d.Round(prec + 1) 178 case 'f': 179 d.Round(d.dp + prec) 180 case 'g', 'G': 181 if prec == 0 { 182 prec = 1 183 } 184 d.Round(prec) 185 } 186 digs = decimalSlice{d: d.d[:], nd: d.nd, dp: d.dp} 187 } 188 return formatDigits(dst, shortest, neg, digs, prec, fmt) 189 } 190 191 func formatDigits(dst []byte, shortest bool, neg bool, digs decimalSlice, prec int, fmt byte) []byte { 192 switch fmt { 193 case 'e', 'E': 194 return fmtE(dst, neg, digs, prec, fmt) 195 case 'f': 196 return fmtF(dst, neg, digs, prec) 197 case 'g', 'G': 198 // trailing fractional zeros in 'e' form will be trimmed. 199 eprec := prec 200 if eprec > digs.nd && digs.nd >= digs.dp { 201 eprec = digs.nd 202 } 203 // %e is used if the exponent from the conversion 204 // is less than -4 or greater than or equal to the precision. 205 // if precision was the shortest possible, use precision 6 for this decision. 206 if shortest { 207 eprec = 6 208 } 209 exp := digs.dp - 1 210 if exp < -4 || exp >= eprec { 211 if prec > digs.nd { 212 prec = digs.nd 213 } 214 return fmtE(dst, neg, digs, prec-1, fmt+'e'-'g') 215 } 216 if prec > digs.dp { 217 prec = digs.nd 218 } 219 return fmtF(dst, neg, digs, max(prec-digs.dp, 0)) 220 } 221 222 // unknown format 223 return append(dst, '%', fmt) 224 } 225 226 // roundShortest rounds d (= mant * 2^exp) to the shortest number of digits 227 // that will let the original floating point value be precisely reconstructed. 228 func roundShortest(d *decimal, mant uint64, exp int, flt *floatInfo) { 229 // If mantissa is zero, the number is zero; stop now. 230 if mant == 0 { 231 d.nd = 0 232 return 233 } 234 235 // Compute upper and lower such that any decimal number 236 // between upper and lower (possibly inclusive) 237 // will round to the original floating point number. 238 239 // We may see at once that the number is already shortest. 240 // 241 // Suppose d is not denormal, so that 2^exp <= d < 10^dp. 242 // The closest shorter number is at least 10^(dp-nd) away. 243 // The lower/upper bounds computed below are at distance 244 // at most 2^(exp-mantbits). 245 // 246 // So the number is already shortest if 10^(dp-nd) > 2^(exp-mantbits), 247 // or equivalently log2(10)*(dp-nd) > exp-mantbits. 248 // It is true if 332/100*(dp-nd) >= exp-mantbits (log2(10) > 3.32). 249 minexp := flt.bias + 1 // minimum possible exponent 250 if exp > minexp && 332*(d.dp-d.nd) >= 100*(exp-int(flt.mantbits)) { 251 // The number is already shortest. 252 return 253 } 254 255 // d = mant << (exp - mantbits) 256 // Next highest floating point number is mant+1 << exp-mantbits. 257 // Our upper bound is halfway between, mant*2+1 << exp-mantbits-1. 258 upper := new(decimal) 259 upper.Assign(mant*2 + 1) 260 upper.Shift(exp - int(flt.mantbits) - 1) 261 262 // d = mant << (exp - mantbits) 263 // Next lowest floating point number is mant-1 << exp-mantbits, 264 // unless mant-1 drops the significant bit and exp is not the minimum exp, 265 // in which case the next lowest is mant*2-1 << exp-mantbits-1. 266 // Either way, call it mantlo << explo-mantbits. 267 // Our lower bound is halfway between, mantlo*2+1 << explo-mantbits-1. 268 var mantlo uint64 269 var explo int 270 if mant > 1<<flt.mantbits || exp == minexp { 271 mantlo = mant - 1 272 explo = exp 273 } else { 274 mantlo = mant*2 - 1 275 explo = exp - 1 276 } 277 lower := new(decimal) 278 lower.Assign(mantlo*2 + 1) 279 lower.Shift(explo - int(flt.mantbits) - 1) 280 281 // The upper and lower bounds are possible outputs only if 282 // the original mantissa is even, so that IEEE round-to-even 283 // would round to the original mantissa and not the neighbors. 284 inclusive := mant%2 == 0 285 286 // Now we can figure out the minimum number of digits required. 287 // Walk along until d has distinguished itself from upper and lower. 288 for i := 0; i < d.nd; i++ { 289 var l, m, u byte // lower, middle, upper digits 290 if i < lower.nd { 291 l = lower.d[i] 292 } else { 293 l = '0' 294 } 295 m = d.d[i] 296 if i < upper.nd { 297 u = upper.d[i] 298 } else { 299 u = '0' 300 } 301 302 // Okay to round down (truncate) if lower has a different digit 303 // or if lower is inclusive and is exactly the result of rounding down. 304 okdown := l != m || (inclusive && l == m && i+1 == lower.nd) 305 306 // Okay to round up if upper has a different digit and 307 // either upper is inclusive or upper is bigger than the result of rounding up. 308 okup := m != u && (inclusive || m+1 < u || i+1 < upper.nd) 309 310 // If it's okay to do either, then round to the nearest one. 311 // If it's okay to do only one, do it. 312 switch { 313 case okdown && okup: 314 d.Round(i + 1) 315 return 316 case okdown: 317 d.RoundDown(i + 1) 318 return 319 case okup: 320 d.RoundUp(i + 1) 321 return 322 } 323 } 324 } 325 326 type decimalSlice struct { 327 d []byte 328 nd, dp int 329 neg bool 330 } 331 332 // %e: -d.dddddedd 333 func fmtE(dst []byte, neg bool, d decimalSlice, prec int, fmt byte) []byte { 334 // sign 335 if neg { 336 dst = append(dst, '-') 337 } 338 339 // first digit 340 ch := byte('0') 341 if d.nd != 0 { 342 ch = d.d[0] 343 } 344 dst = append(dst, ch) 345 346 // .moredigits 347 if prec > 0 { 348 dst = append(dst, '.') 349 i := 1 350 m := min(d.nd, prec+1) 351 if i < m { 352 dst = append(dst, d.d[i:m]...) 353 i = m 354 } 355 for ; i <= prec; i++ { 356 dst = append(dst, '0') 357 } 358 } 359 360 // e 361 dst = append(dst, fmt) 362 exp := d.dp - 1 363 if d.nd == 0 { // special case: 0 has exponent 0 364 exp = 0 365 } 366 if exp < 0 { 367 ch = '-' 368 exp = -exp 369 } else { 370 ch = '+' 371 } 372 dst = append(dst, ch) 373 374 // dd or ddd 375 switch { 376 case exp < 10: 377 dst = append(dst, '0', byte(exp)+'0') 378 case exp < 100: 379 dst = append(dst, byte(exp/10)+'0', byte(exp%10)+'0') 380 default: 381 dst = append(dst, byte(exp/100)+'0', byte(exp/10)%10+'0', byte(exp%10)+'0') 382 } 383 384 return dst 385 } 386 387 // %f: -ddddddd.ddddd 388 func fmtF(dst []byte, neg bool, d decimalSlice, prec int) []byte { 389 // sign 390 if neg { 391 dst = append(dst, '-') 392 } 393 394 // integer, padded with zeros as needed. 395 if d.dp > 0 { 396 m := min(d.nd, d.dp) 397 dst = append(dst, d.d[:m]...) 398 for ; m < d.dp; m++ { 399 dst = append(dst, '0') 400 } 401 } else { 402 dst = append(dst, '0') 403 } 404 405 // fraction 406 if prec > 0 { 407 dst = append(dst, '.') 408 for i := 0; i < prec; i++ { 409 ch := byte('0') 410 if j := d.dp + i; 0 <= j && j < d.nd { 411 ch = d.d[j] 412 } 413 dst = append(dst, ch) 414 } 415 } 416 417 return dst 418 } 419 420 // %b: -ddddddddpddd 421 func fmtB(dst []byte, neg bool, mant uint64, exp int, flt *floatInfo) []byte { 422 // sign 423 if neg { 424 dst = append(dst, '-') 425 } 426 427 // mantissa 428 dst, _ = formatBits(dst, mant, 10, false, true) 429 430 // p 431 dst = append(dst, 'p') 432 433 // exponent 434 exp -= int(flt.mantbits) 435 if exp >= 0 { 436 dst = append(dst, '+') 437 } 438 dst, _ = formatBits(dst, uint64(exp), 10, exp < 0, true) 439 440 return dst 441 } 442 443 func min(a, b int) int { 444 if a < b { 445 return a 446 } 447 return b 448 } 449 450 func max(a, b int) int { 451 if a > b { 452 return a 453 } 454 return b 455 } 456