1 /*
2 * Copyright (c) 1996, 2014, Oracle and/or its affiliates. All rights reserved.
3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4 *
5 * This code is free software; you can redistribute it and/or modify it
6 * under the terms of the GNU General Public License version 2 only, as
7 * published by the Free Software Foundation. Oracle designates this
8 * particular file as subject to the "Classpath" exception as provided
9 * by Oracle in the LICENSE file that accompanied this code.
10 *
11 * This code is distributed in the hope that it will be useful, but WITHOUT
12 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
13 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
14 * version 2 for more details (a copy is included in the LICENSE file that
15 * accompanied this code).
16 *
17 * You should have received a copy of the GNU General Public License version
18 * 2 along with this work; if not, write to the Free Software Foundation,
19 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
20 *
21 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
22 * or visit www.oracle.com if you need additional information or have any
23 * questions.
24 */
25
26 /*
27 * Portions Copyright (c) 1995 Colin Plumb. All rights reserved.
28 */
29
30 package java.math;
31
32 import java.io.IOException;
33 import java.io.ObjectInputStream;
34 import java.io.ObjectOutputStream;
35 import java.io.ObjectStreamField;
36 import java.util.Arrays;
37 import java.util.Random;
38 import java.util.concurrent.ThreadLocalRandom;
39 import sun.misc.DoubleConsts;
40 import sun.misc.FloatConsts;
41
42 /**
43 * Immutable arbitrary-precision integers. All operations behave as if
44 * BigIntegers were represented in two's-complement notation (like Java's
45 * primitive integer types). BigInteger provides analogues to all of Java's
46 * primitive integer operators, and all relevant methods from java.lang.Math.
47 * Additionally, BigInteger provides operations for modular arithmetic, GCD
48 * calculation, primality testing, prime generation, bit manipulation,
49 * and a few other miscellaneous operations.
50 *
51 * <p>Semantics of arithmetic operations exactly mimic those of Java's integer
52 * arithmetic operators, as defined in <i>The Java Language Specification</i>.
53 * For example, division by zero throws an {@code ArithmeticException}, and
54 * division of a negative by a positive yields a negative (or zero) remainder.
55 * All of the details in the Spec concerning overflow are ignored, as
56 * BigIntegers are made as large as necessary to accommodate the results of an
57 * operation.
58 *
59 * <p>Semantics of shift operations extend those of Java's shift operators
60 * to allow for negative shift distances. A right-shift with a negative
61 * shift distance results in a left shift, and vice-versa. The unsigned
62 * right shift operator ({@code >>>}) is omitted, as this operation makes
63 * little sense in combination with the "infinite word size" abstraction
64 * provided by this class.
65 *
66 * <p>Semantics of bitwise logical operations exactly mimic those of Java's
67 * bitwise integer operators. The binary operators ({@code and},
68 * {@code or}, {@code xor}) implicitly perform sign extension on the shorter
69 * of the two operands prior to performing the operation.
70 *
71 * <p>Comparison operations perform signed integer comparisons, analogous to
72 * those performed by Java's relational and equality operators.
73 *
74 * <p>Modular arithmetic operations are provided to compute residues, perform
75 * exponentiation, and compute multiplicative inverses. These methods always
76 * return a non-negative result, between {@code 0} and {@code (modulus - 1)},
77 * inclusive.
78 *
79 * <p>Bit operations operate on a single bit of the two's-complement
80 * representation of their operand. If necessary, the operand is sign-
81 * extended so that it contains the designated bit. None of the single-bit
82 * operations can produce a BigInteger with a different sign from the
83 * BigInteger being operated on, as they affect only a single bit, and the
84 * "infinite word size" abstraction provided by this class ensures that there
85 * are infinitely many "virtual sign bits" preceding each BigInteger.
86 *
87 * <p>For the sake of brevity and clarity, pseudo-code is used throughout the
88 * descriptions of BigInteger methods. The pseudo-code expression
89 * {@code (i + j)} is shorthand for "a BigInteger whose value is
90 * that of the BigInteger {@code i} plus that of the BigInteger {@code j}."
91 * The pseudo-code expression {@code (i == j)} is shorthand for
92 * "{@code true} if and only if the BigInteger {@code i} represents the same
93 * value as the BigInteger {@code j}." Other pseudo-code expressions are
94 * interpreted similarly.
95 *
96 * <p>All methods and constructors in this class throw
97 * {@code NullPointerException} when passed
98 * a null object reference for any input parameter.
99 *
100 * BigInteger must support values in the range
101 * -2<sup>{@code Integer.MAX_VALUE}</sup> (exclusive) to
102 * +2<sup>{@code Integer.MAX_VALUE}</sup> (exclusive)
103 * and may support values outside of that range.
104 *
105 * The range of probable prime values is limited and may be less than
106 * the full supported positive range of {@code BigInteger}.
107 * The range must be at least 1 to 2<sup>500000000</sup>.
108 *
109 * @implNote
110 * BigInteger constructors and operations throw {@code ArithmeticException} when
111 * the result is out of the supported range of
112 * -2<sup>{@code Integer.MAX_VALUE}</sup> (exclusive) to
113 * +2<sup>{@code Integer.MAX_VALUE}</sup> (exclusive).
114 *
115 * @see BigDecimal
116 * @author Josh Bloch
117 * @author Michael McCloskey
118 * @author Alan Eliasen
119 * @author Timothy Buktu
120 * @since 1.1
121 */
122
123 public class BigInteger extends Number implements Comparable<BigInteger> {
124 /**
125 * The signum of this BigInteger: -1 for negative, 0 for zero, or
126 * 1 for positive. Note that the BigInteger zero <i>must</i> have
127 * a signum of 0. This is necessary to ensures that there is exactly one
128 * representation for each BigInteger value.
129 */
130 final int signum;
131
132 /**
133 * The magnitude of this BigInteger, in <i>big-endian</i> order: the
134 * zeroth element of this array is the most-significant int of the
135 * magnitude. The magnitude must be "minimal" in that the most-significant
136 * int ({@code mag[0]}) must be non-zero. This is necessary to
137 * ensure that there is exactly one representation for each BigInteger
138 * value. Note that this implies that the BigInteger zero has a
139 * zero-length mag array.
140 */
141 final int[] mag;
142
143 // The following fields are stable variables. A stable variable's value
144 // changes at most once from the default zero value to a non-zero stable
145 // value. A stable value is calculated lazily on demand.
146
147 /**
148 * One plus the bitCount of this BigInteger. This is a stable variable.
149 *
150 * @see #bitCount
151 */
152 private int bitCountPlusOne;
153
154 /**
155 * One plus the bitLength of this BigInteger. This is a stable variable.
156 * (either value is acceptable).
157 *
158 * @see #bitLength()
159 */
160 private int bitLengthPlusOne;
161
162 /**
163 * Two plus the lowest set bit of this BigInteger. This is a stable variable.
164 *
165 * @see #getLowestSetBit
166 */
167 private int lowestSetBitPlusTwo;
168
169 /**
170 * Two plus the index of the lowest-order int in the magnitude of this
171 * BigInteger that contains a nonzero int. This is a stable variable. The
172 * least significant int has int-number 0, the next int in order of
173 * increasing significance has int-number 1, and so forth.
174 *
175 * <p>Note: never used for a BigInteger with a magnitude of zero.
176 *
177 * @see #firstNonzeroIntNum()
178 */
179 private int firstNonzeroIntNumPlusTwo;
180
181 /**
182 * This mask is used to obtain the value of an int as if it were unsigned.
183 */
184 final static long LONG_MASK = 0xffffffffL;
185
186 /**
187 * This constant limits {@code mag.length} of BigIntegers to the supported
188 * range.
189 */
190 private static final int MAX_MAG_LENGTH = Integer.MAX_VALUE / Integer.SIZE + 1; // (1 << 26)
191
192 /**
193 * Bit lengths larger than this constant can cause overflow in searchLen
194 * calculation and in BitSieve.singleSearch method.
195 */
196 private static final int PRIME_SEARCH_BIT_LENGTH_LIMIT = 500000000;
197
198 /**
199 * The threshold value for using Karatsuba multiplication. If the number
200 * of ints in both mag arrays are greater than this number, then
201 * Karatsuba multiplication will be used. This value is found
202 * experimentally to work well.
203 */
204 private static final int KARATSUBA_THRESHOLD = 80;
205
206 /**
207 * The threshold value for using 3-way Toom-Cook multiplication.
208 * If the number of ints in each mag array is greater than the
209 * Karatsuba threshold, and the number of ints in at least one of
210 * the mag arrays is greater than this threshold, then Toom-Cook
211 * multiplication will be used.
212 */
213 private static final int TOOM_COOK_THRESHOLD = 240;
214
215 /**
216 * The threshold value for using Karatsuba squaring. If the number
217 * of ints in the number are larger than this value,
218 * Karatsuba squaring will be used. This value is found
219 * experimentally to work well.
220 */
221 private static final int KARATSUBA_SQUARE_THRESHOLD = 128;
222
223 /**
224 * The threshold value for using Toom-Cook squaring. If the number
225 * of ints in the number are larger than this value,
226 * Toom-Cook squaring will be used. This value is found
227 * experimentally to work well.
228 */
229 private static final int TOOM_COOK_SQUARE_THRESHOLD = 216;
230
231 /**
232 * The threshold value for using Burnikel-Ziegler division. If the number
233 * of ints in the divisor are larger than this value, Burnikel-Ziegler
234 * division may be used. This value is found experimentally to work well.
235 */
236 static final int BURNIKEL_ZIEGLER_THRESHOLD = 80;
237
238 /**
239 * The offset value for using Burnikel-Ziegler division. If the number
240 * of ints in the divisor exceeds the Burnikel-Ziegler threshold, and the
241 * number of ints in the dividend is greater than the number of ints in the
242 * divisor plus this value, Burnikel-Ziegler division will be used. This
243 * value is found experimentally to work well.
244 */
245 static final int BURNIKEL_ZIEGLER_OFFSET = 40;
246
247 /**
248 * The threshold value for using Schoenhage recursive base conversion. If
249 * the number of ints in the number are larger than this value,
250 * the Schoenhage algorithm will be used. In practice, it appears that the
251 * Schoenhage routine is faster for any threshold down to 2, and is
252 * relatively flat for thresholds between 2-25, so this choice may be
253 * varied within this range for very small effect.
254 */
255 private static final int SCHOENHAGE_BASE_CONVERSION_THRESHOLD = 20;
256
257 /**
258 * The threshold value for using squaring code to perform multiplication
259 * of a {@code BigInteger} instance by itself. If the number of ints in
260 * the number are larger than this value, {@code multiply(this)} will
261 * return {@code square()}.
262 */
263 private static final int MULTIPLY_SQUARE_THRESHOLD = 20;
264
265 // Constructors
266
267 /**
268 * Translates a byte sub-array containing the two's-complement binary
269 * representation of a BigInteger into a BigInteger. The sub-array is
270 * specified via an offset into the array and a length. The sub-array is
271 * assumed to be in <i>big-endian</i> byte-order: the most significant
272 * byte is the element at index {@code off}. The {@code val} array is
273 * assumed to be unchanged for the duration of the constructor call.
274 *
275 * An {@code IndexOutOfBoundsException} is thrown if the length of the array
276 * {@code val} is non-zero and either {@code off} is negative, {@code len}
277 * is negative, or {@code off+len} is greater than the length of
278 * {@code val}.
279 *
280 * @param val byte array containing a sub-array which is the big-endian
281 * two's-complement binary representation of a BigInteger.
282 * @param off the start offset of the binary representation.
283 * @param len the number of bytes to use.
284 * @throws NumberFormatException {@code val} is zero bytes long.
285 * @throws IndexOutOfBoundsException if the provided array offset and
286 * length would cause an index into the byte array to be
287 * negative or greater than or equal to the array length.
288 * @since 1.9
289 */
290 public BigInteger(byte[] val, int off, int len) {
291 if (val.length == 0) {
292 throw new NumberFormatException("Zero length BigInteger");
293 } else if ((off < 0) || (off >= val.length) || (len < 0) ||
294 (len > val.length - off)) { // 0 <= off < val.length
295 throw new IndexOutOfBoundsException();
296 }
297
298 if (val[off] < 0) {
299 mag = makePositive(val, off, len);
300 signum = -1;
301 } else {
302 mag = stripLeadingZeroBytes(val, off, len);
303 signum = (mag.length == 0 ? 0 : 1);
304 }
305 if (mag.length >= MAX_MAG_LENGTH) {
306 checkRange();
307 }
308 }
309
310 /**
311 * Translates a byte array containing the two's-complement binary
312 * representation of a BigInteger into a BigInteger. The input array is
313 * assumed to be in <i>big-endian</i> byte-order: the most significant
314 * byte is in the zeroth element. The {@code val} array is assumed to be
315 * unchanged for the duration of the constructor call.
316 *
317 * @param val big-endian two's-complement binary representation of a
318 * BigInteger.
319 * @throws NumberFormatException {@code val} is zero bytes long.
320 */
321 public BigInteger(byte[] val) {
322 this(val, 0, val.length);
323 }
324
325 /**
326 * This private constructor translates an int array containing the
327 * two's-complement binary representation of a BigInteger into a
328 * BigInteger. The input array is assumed to be in <i>big-endian</i>
329 * int-order: the most significant int is in the zeroth element. The
330 * {@code val} array is assumed to be unchanged for the duration of
331 * the constructor call.
332 */
333 private BigInteger(int[] val) {
334 if (val.length == 0)
335 throw new NumberFormatException("Zero length BigInteger");
336
337 if (val[0] < 0) {
338 mag = makePositive(val);
339 signum = -1;
340 } else {
341 mag = trustedStripLeadingZeroInts(val);
342 signum = (mag.length == 0 ? 0 : 1);
343 }
344 if (mag.length >= MAX_MAG_LENGTH) {
345 checkRange();
346 }
347 }
348
349 /**
350 * Translates the sign-magnitude representation of a BigInteger into a
351 * BigInteger. The sign is represented as an integer signum value: -1 for
352 * negative, 0 for zero, or 1 for positive. The magnitude is a sub-array of
353 * a byte array in <i>big-endian</i> byte-order: the most significant byte
354 * is the element at index {@code off}. A zero value of the length
355 * {@code len} is permissible, and will result in a BigInteger value of 0,
356 * whether signum is -1, 0 or 1. The {@code magnitude} array is assumed to
357 * be unchanged for the duration of the constructor call.
358 *
359 * An {@code IndexOutOfBoundsException} is thrown if the length of the array
360 * {@code magnitude} is non-zero and either {@code off} is negative,
361 * {@code len} is negative, or {@code off+len} is greater than the length of
362 * {@code magnitude}.
363 *
364 * @param signum signum of the number (-1 for negative, 0 for zero, 1
365 * for positive).
366 * @param magnitude big-endian binary representation of the magnitude of
367 * the number.
368 * @param off the start offset of the binary representation.
369 * @param len the number of bytes to use.
370 * @throws NumberFormatException {@code signum} is not one of the three
371 * legal values (-1, 0, and 1), or {@code signum} is 0 and
372 * {@code magnitude} contains one or more non-zero bytes.
373 * @throws IndexOutOfBoundsException if the provided array offset and
374 * length would cause an index into the byte array to be
375 * negative or greater than or equal to the array length.
376 * @since 1.9
377 */
378 public BigInteger(int signum, byte[] magnitude, int off, int len) {
379 if (signum < -1 || signum > 1) {
380 throw(new NumberFormatException("Invalid signum value"));
381 } else if ((off < 0) || (len < 0) ||
382 (len > 0 &&
383 ((off >= magnitude.length) ||
384 (len > magnitude.length - off)))) { // 0 <= off < magnitude.length
385 throw new IndexOutOfBoundsException();
386 }
387
388 // stripLeadingZeroBytes() returns a zero length array if len == 0
389 this.mag = stripLeadingZeroBytes(magnitude, off, len);
390
391 if (this.mag.length == 0) {
392 this.signum = 0;
393 } else {
394 if (signum == 0)
395 throw(new NumberFormatException("signum-magnitude mismatch"));
396 this.signum = signum;
397 }
398 if (mag.length >= MAX_MAG_LENGTH) {
399 checkRange();
400 }
401 }
402
403 /**
404 * Translates the sign-magnitude representation of a BigInteger into a
405 * BigInteger. The sign is represented as an integer signum value: -1 for
406 * negative, 0 for zero, or 1 for positive. The magnitude is a byte array
407 * in <i>big-endian</i> byte-order: the most significant byte is the
408 * zeroth element. A zero-length magnitude array is permissible, and will
409 * result in a BigInteger value of 0, whether signum is -1, 0 or 1. The
410 * {@code magnitude} array is assumed to be unchanged for the duration of
411 * the constructor call.
412 *
413 * @param signum signum of the number (-1 for negative, 0 for zero, 1
414 * for positive).
415 * @param magnitude big-endian binary representation of the magnitude of
416 * the number.
417 * @throws NumberFormatException {@code signum} is not one of the three
418 * legal values (-1, 0, and 1), or {@code signum} is 0 and
419 * {@code magnitude} contains one or more non-zero bytes.
420 */
421 public BigInteger(int signum, byte[] magnitude) {
422 this(signum, magnitude, 0, magnitude.length);
423 }
424
425 /**
426 * A constructor for internal use that translates the sign-magnitude
427 * representation of a BigInteger into a BigInteger. It checks the
428 * arguments and copies the magnitude so this constructor would be
429 * safe for external use. The {@code magnitude} array is assumed to be
430 * unchanged for the duration of the constructor call.
431 */
432 private BigInteger(int signum, int[] magnitude) {
433 this.mag = stripLeadingZeroInts(magnitude);
434
435 if (signum < -1 || signum > 1)
436 throw(new NumberFormatException("Invalid signum value"));
437
438 if (this.mag.length == 0) {
439 this.signum = 0;
440 } else {
441 if (signum == 0)
442 throw(new NumberFormatException("signum-magnitude mismatch"));
443 this.signum = signum;
444 }
445 if (mag.length >= MAX_MAG_LENGTH) {
446 checkRange();
447 }
448 }
449
450 /**
451 * Translates the String representation of a BigInteger in the
452 * specified radix into a BigInteger. The String representation
453 * consists of an optional minus or plus sign followed by a
454 * sequence of one or more digits in the specified radix. The
455 * character-to-digit mapping is provided by {@code
456 * Character.digit}. The String may not contain any extraneous
457 * characters (whitespace, for example).
458 *
459 * @param val String representation of BigInteger.
460 * @param radix radix to be used in interpreting {@code val}.
461 * @throws NumberFormatException {@code val} is not a valid representation
462 * of a BigInteger in the specified radix, or {@code radix} is
463 * outside the range from {@link Character#MIN_RADIX} to
464 * {@link Character#MAX_RADIX}, inclusive.
465 * @see Character#digit
466 */
467 public BigInteger(String val, int radix) {
468 int cursor = 0, numDigits;
469 final int len = val.length();
470
471 if (radix < Character.MIN_RADIX || radix > Character.MAX_RADIX)
472 throw new NumberFormatException("Radix out of range");
473 if (len == 0)
474 throw new NumberFormatException("Zero length BigInteger");
475
476 // Check for at most one leading sign
477 int sign = 1;
478 int index1 = val.lastIndexOf('-');
479 int index2 = val.lastIndexOf('+');
480 if (index1 >= 0) {
481 if (index1 != 0 || index2 >= 0) {
482 throw new NumberFormatException("Illegal embedded sign character");
483 }
484 sign = -1;
485 cursor = 1;
486 } else if (index2 >= 0) {
487 if (index2 != 0) {
488 throw new NumberFormatException("Illegal embedded sign character");
489 }
490 cursor = 1;
491 }
492 if (cursor == len)
493 throw new NumberFormatException("Zero length BigInteger");
494
495 // Skip leading zeros and compute number of digits in magnitude
496 while (cursor < len &&
497 Character.digit(val.charAt(cursor), radix) == 0) {
498 cursor++;
499 }
500
501 if (cursor == len) {
502 signum = 0;
503 mag = ZERO.mag;
504 return;
505 }
506
507 numDigits = len - cursor;
508 signum = sign;
509
510 // Pre-allocate array of expected size. May be too large but can
511 // never be too small. Typically exact.
512 long numBits = ((numDigits * bitsPerDigit[radix]) >>> 10) + 1;
513 if (numBits + 31 >= (1L << 32)) {
514 reportOverflow();
515 }
516 int numWords = (int) (numBits + 31) >>> 5;
517 int[] magnitude = new int[numWords];
518
519 // Process first (potentially short) digit group
520 int firstGroupLen = numDigits % digitsPerInt[radix];
521 if (firstGroupLen == 0)
522 firstGroupLen = digitsPerInt[radix];
523 String group = val.substring(cursor, cursor += firstGroupLen);
524 magnitude[numWords - 1] = Integer.parseInt(group, radix);
525 if (magnitude[numWords - 1] < 0)
526 throw new NumberFormatException("Illegal digit");
527
528 // Process remaining digit groups
529 int superRadix = intRadix[radix];
530 int groupVal = 0;
531 while (cursor < len) {
532 group = val.substring(cursor, cursor += digitsPerInt[radix]);
533 groupVal = Integer.parseInt(group, radix);
534 if (groupVal < 0)
535 throw new NumberFormatException("Illegal digit");
536 destructiveMulAdd(magnitude, superRadix, groupVal);
537 }
538 // Required for cases where the array was overallocated.
539 mag = trustedStripLeadingZeroInts(magnitude);
540 if (mag.length >= MAX_MAG_LENGTH) {
541 checkRange();
542 }
543 }
544
545 /*
546 * Constructs a new BigInteger using a char array with radix=10.
547 * Sign is precalculated outside and not allowed in the val. The {@code val}
548 * array is assumed to be unchanged for the duration of the constructor
549 * call.
550 */
551 BigInteger(char[] val, int sign, int len) {
552 int cursor = 0, numDigits;
553
554 // Skip leading zeros and compute number of digits in magnitude
555 while (cursor < len && Character.digit(val[cursor], 10) == 0) {
556 cursor++;
557 }
558 if (cursor == len) {
559 signum = 0;
560 mag = ZERO.mag;
561 return;
562 }
563
564 numDigits = len - cursor;
565 signum = sign;
566 // Pre-allocate array of expected size
567 int numWords;
568 if (len < 10) {
569 numWords = 1;
570 } else {
571 long numBits = ((numDigits * bitsPerDigit[10]) >>> 10) + 1;
572 if (numBits + 31 >= (1L << 32)) {
573 reportOverflow();
574 }
575 numWords = (int) (numBits + 31) >>> 5;
576 }
577 int[] magnitude = new int[numWords];
578
579 // Process first (potentially short) digit group
580 int firstGroupLen = numDigits % digitsPerInt[10];
581 if (firstGroupLen == 0)
582 firstGroupLen = digitsPerInt[10];
583 magnitude[numWords - 1] = parseInt(val, cursor, cursor += firstGroupLen);
584
585 // Process remaining digit groups
586 while (cursor < len) {
587 int groupVal = parseInt(val, cursor, cursor += digitsPerInt[10]);
588 destructiveMulAdd(magnitude, intRadix[10], groupVal);
589 }
590 mag = trustedStripLeadingZeroInts(magnitude);
591 if (mag.length >= MAX_MAG_LENGTH) {
592 checkRange();
593 }
594 }
595
596 // Create an integer with the digits between the two indexes
597 // Assumes start < end. The result may be negative, but it
598 // is to be treated as an unsigned value.
599 private int parseInt(char[] source, int start, int end) {
600 int result = Character.digit(source[start++], 10);
601 if (result == -1)
602 throw new NumberFormatException(new String(source));
603
604 for (int index = start; index < end; index++) {
605 int nextVal = Character.digit(source[index], 10);
606 if (nextVal == -1)
607 throw new NumberFormatException(new String(source));
608 result = 10*result + nextVal;
609 }
610
611 return result;
612 }
613
614 // bitsPerDigit in the given radix times 1024
615 // Rounded up to avoid underallocation.
616 private static long bitsPerDigit[] = { 0, 0,
617 1024, 1624, 2048, 2378, 2648, 2875, 3072, 3247, 3402, 3543, 3672,
618 3790, 3899, 4001, 4096, 4186, 4271, 4350, 4426, 4498, 4567, 4633,
619 4696, 4756, 4814, 4870, 4923, 4975, 5025, 5074, 5120, 5166, 5210,
620 5253, 5295};
621
622 // Multiply x array times word y in place, and add word z
623 private static void destructiveMulAdd(int[] x, int y, int z) {
624 // Perform the multiplication word by word
625 long ylong = y & LONG_MASK;
626 long zlong = z & LONG_MASK;
627 int len = x.length;
628
629 long product = 0;
630 long carry = 0;
631 for (int i = len-1; i >= 0; i--) {
632 product = ylong * (x[i] & LONG_MASK) + carry;
633 x[i] = (int)product;
634 carry = product >>> 32;
635 }
636
637 // Perform the addition
638 long sum = (x[len-1] & LONG_MASK) + zlong;
639 x[len-1] = (int)sum;
640 carry = sum >>> 32;
641 for (int i = len-2; i >= 0; i--) {
642 sum = (x[i] & LONG_MASK) + carry;
643 x[i] = (int)sum;
644 carry = sum >>> 32;
645 }
646 }
647
648 /**
649 * Translates the decimal String representation of a BigInteger into a
650 * BigInteger. The String representation consists of an optional minus
651 * sign followed by a sequence of one or more decimal digits. The
652 * character-to-digit mapping is provided by {@code Character.digit}.
653 * The String may not contain any extraneous characters (whitespace, for
654 * example).
655 *
656 * @param val decimal String representation of BigInteger.
657 * @throws NumberFormatException {@code val} is not a valid representation
658 * of a BigInteger.
659 * @see Character#digit
660 */
661 public BigInteger(String val) {
662 this(val, 10);
663 }
664
665 /**
666 * Constructs a randomly generated BigInteger, uniformly distributed over
667 * the range 0 to (2<sup>{@code numBits}</sup> - 1), inclusive.
668 * The uniformity of the distribution assumes that a fair source of random
669 * bits is provided in {@code rnd}. Note that this constructor always
670 * constructs a non-negative BigInteger.
671 *
672 * @param numBits maximum bitLength of the new BigInteger.
673 * @param rnd source of randomness to be used in computing the new
674 * BigInteger.
675 * @throws IllegalArgumentException {@code numBits} is negative.
676 * @see #bitLength()
677 */
678 public BigInteger(int numBits, Random rnd) {
679 this(1, randomBits(numBits, rnd));
680 }
681
682 private static byte[] randomBits(int numBits, Random rnd) {
683 if (numBits < 0)
684 throw new IllegalArgumentException("numBits must be non-negative");
685 int numBytes = (int)(((long)numBits+7)/8); // avoid overflow
686 byte[] randomBits = new byte[numBytes];
687
688 // Generate random bytes and mask out any excess bits
689 if (numBytes > 0) {
690 rnd.nextBytes(randomBits);
691 int excessBits = 8*numBytes - numBits;
692 randomBits[0] &= (1 << (8-excessBits)) - 1;
693 }
694 return randomBits;
695 }
696
697 /**
698 * Constructs a randomly generated positive BigInteger that is probably
699 * prime, with the specified bitLength.
700 *
701 * <p>It is recommended that the {@link #probablePrime probablePrime}
702 * method be used in preference to this constructor unless there
703 * is a compelling need to specify a certainty.
704 *
705 * @param bitLength bitLength of the returned BigInteger.
706 * @param certainty a measure of the uncertainty that the caller is
707 * willing to tolerate. The probability that the new BigInteger
708 * represents a prime number will exceed
709 * (1 - 1/2<sup>{@code certainty}</sup>). The execution time of
710 * this constructor is proportional to the value of this parameter.
711 * @param rnd source of random bits used to select candidates to be
712 * tested for primality.
713 * @throws ArithmeticException {@code bitLength < 2} or {@code bitLength} is too large.
714 * @see #bitLength()
715 */
716 public BigInteger(int bitLength, int certainty, Random rnd) {
717 BigInteger prime;
718
719 if (bitLength < 2)
720 throw new ArithmeticException("bitLength < 2");
721 prime = (bitLength < SMALL_PRIME_THRESHOLD
722 ? smallPrime(bitLength, certainty, rnd)
723 : largePrime(bitLength, certainty, rnd));
724 signum = 1;
725 mag = prime.mag;
726 }
727
728 // Minimum size in bits that the requested prime number has
729 // before we use the large prime number generating algorithms.
730 // The cutoff of 95 was chosen empirically for best performance.
731 private static final int SMALL_PRIME_THRESHOLD = 95;
732
733 // Certainty required to meet the spec of probablePrime
734 private static final int DEFAULT_PRIME_CERTAINTY = 100;
735
736 /**
737 * Returns a positive BigInteger that is probably prime, with the
738 * specified bitLength. The probability that a BigInteger returned
739 * by this method is composite does not exceed 2<sup>-100</sup>.
740 *
741 * @param bitLength bitLength of the returned BigInteger.
742 * @param rnd source of random bits used to select candidates to be
743 * tested for primality.
744 * @return a BigInteger of {@code bitLength} bits that is probably prime
745 * @throws ArithmeticException {@code bitLength < 2} or {@code bitLength} is too large.
746 * @see #bitLength()
747 * @since 1.4
748 */
749 public static BigInteger probablePrime(int bitLength, Random rnd) {
750 if (bitLength < 2)
751 throw new ArithmeticException("bitLength < 2");
752
753 return (bitLength < SMALL_PRIME_THRESHOLD ?
754 smallPrime(bitLength, DEFAULT_PRIME_CERTAINTY, rnd) :
755 largePrime(bitLength, DEFAULT_PRIME_CERTAINTY, rnd));
756 }
757
758 /**
759 * Find a random number of the specified bitLength that is probably prime.
760 * This method is used for smaller primes, its performance degrades on
761 * larger bitlengths.
762 *
763 * This method assumes bitLength > 1.
764 */
765 private static BigInteger smallPrime(int bitLength, int certainty, Random rnd) {
766 int magLen = (bitLength + 31) >>> 5;
767 int temp[] = new int[magLen];
768 int highBit = 1 << ((bitLength+31) & 0x1f); // High bit of high int
769 int highMask = (highBit << 1) - 1; // Bits to keep in high int
770
771 while (true) {
772 // Construct a candidate
773 for (int i=0; i < magLen; i++)
774 temp[i] = rnd.nextInt();
775 temp[0] = (temp[0] & highMask) | highBit; // Ensure exact length
776 if (bitLength > 2)
777 temp[magLen-1] |= 1; // Make odd if bitlen > 2
778
779 BigInteger p = new BigInteger(temp, 1);
780
781 // Do cheap "pre-test" if applicable
782 if (bitLength > 6) {
783 long r = p.remainder(SMALL_PRIME_PRODUCT).longValue();
784 if ((r%3==0) || (r%5==0) || (r%7==0) || (r%11==0) ||
785 (r%13==0) || (r%17==0) || (r%19==0) || (r%23==0) ||
786 (r%29==0) || (r%31==0) || (r%37==0) || (r%41==0))
787 continue; // Candidate is composite; try another
788 }
789
790 // All candidates of bitLength 2 and 3 are prime by this point
791 if (bitLength < 4)
792 return p;
793
794 // Do expensive test if we survive pre-test (or it's inapplicable)
795 if (p.primeToCertainty(certainty, rnd))
796 return p;
797 }
798 }
799
800 private static final BigInteger SMALL_PRIME_PRODUCT
801 = valueOf(3L*5*7*11*13*17*19*23*29*31*37*41);
802
803 /**
804 * Find a random number of the specified bitLength that is probably prime.
805 * This method is more appropriate for larger bitlengths since it uses
806 * a sieve to eliminate most composites before using a more expensive
807 * test.
808 */
809 private static BigInteger largePrime(int bitLength, int certainty, Random rnd) {
810 BigInteger p;
811 p = new BigInteger(bitLength, rnd).setBit(bitLength-1);
812 p.mag[p.mag.length-1] &= 0xfffffffe;
813
814 // Use a sieve length likely to contain the next prime number
815 int searchLen = getPrimeSearchLen(bitLength);
816 BitSieve searchSieve = new BitSieve(p, searchLen);
817 BigInteger candidate = searchSieve.retrieve(p, certainty, rnd);
818
819 while ((candidate == null) || (candidate.bitLength() != bitLength)) {
820 p = p.add(BigInteger.valueOf(2*searchLen));
821 if (p.bitLength() != bitLength)
822 p = new BigInteger(bitLength, rnd).setBit(bitLength-1);
823 p.mag[p.mag.length-1] &= 0xfffffffe;
824 searchSieve = new BitSieve(p, searchLen);
825 candidate = searchSieve.retrieve(p, certainty, rnd);
826 }
827 return candidate;
828 }
829
830 /**
831 * Returns the first integer greater than this {@code BigInteger} that
832 * is probably prime. The probability that the number returned by this
833 * method is composite does not exceed 2<sup>-100</sup>. This method will
834 * never skip over a prime when searching: if it returns {@code p}, there
835 * is no prime {@code q} such that {@code this < q < p}.
836 *
837 * @return the first integer greater than this {@code BigInteger} that
838 * is probably prime.
839 * @throws ArithmeticException {@code this < 0} or {@code this} is too large.
840 * @since 1.5
841 */
842 public BigInteger nextProbablePrime() {
843 if (this.signum < 0)
844 throw new ArithmeticException("start < 0: " + this);
845
846 // Handle trivial cases
847 if ((this.signum == 0) || this.equals(ONE))
848 return TWO;
849
850 BigInteger result = this.add(ONE);
851
852 // Fastpath for small numbers
853 if (result.bitLength() < SMALL_PRIME_THRESHOLD) {
854
855 // Ensure an odd number
856 if (!result.testBit(0))
857 result = result.add(ONE);
858
859 while (true) {
860 // Do cheap "pre-test" if applicable
861 if (result.bitLength() > 6) {
862 long r = result.remainder(SMALL_PRIME_PRODUCT).longValue();
863 if ((r%3==0) || (r%5==0) || (r%7==0) || (r%11==0) ||
864 (r%13==0) || (r%17==0) || (r%19==0) || (r%23==0) ||
865 (r%29==0) || (r%31==0) || (r%37==0) || (r%41==0)) {
866 result = result.add(TWO);
867 continue; // Candidate is composite; try another
868 }
869 }
870
871 // All candidates of bitLength 2 and 3 are prime by this point
872 if (result.bitLength() < 4)
873 return result;
874
875 // The expensive test
876 if (result.primeToCertainty(DEFAULT_PRIME_CERTAINTY, null))
877 return result;
878
879 result = result.add(TWO);
880 }
881 }
882
883 // Start at previous even number
884 if (result.testBit(0))
885 result = result.subtract(ONE);
886
887 // Looking for the next large prime
888 int searchLen = getPrimeSearchLen(result.bitLength());
889
890 while (true) {
891 BitSieve searchSieve = new BitSieve(result, searchLen);
892 BigInteger candidate = searchSieve.retrieve(result,
893 DEFAULT_PRIME_CERTAINTY, null);
894 if (candidate != null)
895 return candidate;
896 result = result.add(BigInteger.valueOf(2 * searchLen));
897 }
898 }
899
900 private static int getPrimeSearchLen(int bitLength) {
901 if (bitLength > PRIME_SEARCH_BIT_LENGTH_LIMIT + 1) {
902 throw new ArithmeticException("Prime search implementation restriction on bitLength");
903 }
904 return bitLength / 20 * 64;
905 }
906
907 /**
908 * Returns {@code true} if this BigInteger is probably prime,
909 * {@code false} if it's definitely composite.
910 *
911 * This method assumes bitLength > 2.
912 *
913 * @param certainty a measure of the uncertainty that the caller is
914 * willing to tolerate: if the call returns {@code true}
915 * the probability that this BigInteger is prime exceeds
916 * {@code (1 - 1/2<sup>certainty</sup>)}. The execution time of
917 * this method is proportional to the value of this parameter.
918 * @return {@code true} if this BigInteger is probably prime,
919 * {@code false} if it's definitely composite.
920 */
921 boolean primeToCertainty(int certainty, Random random) {
922 int rounds = 0;
923 int n = (Math.min(certainty, Integer.MAX_VALUE-1)+1)/2;
924
925 // The relationship between the certainty and the number of rounds
926 // we perform is given in the draft standard ANSI X9.80, "PRIME
927 // NUMBER GENERATION, PRIMALITY TESTING, AND PRIMALITY CERTIFICATES".
928 int sizeInBits = this.bitLength();
929 if (sizeInBits < 100) {
930 rounds = 50;
931 rounds = n < rounds ? n : rounds;
932 return passesMillerRabin(rounds, random);
933 }
934
935 if (sizeInBits < 256) {
936 rounds = 27;
937 } else if (sizeInBits < 512) {
938 rounds = 15;
939 } else if (sizeInBits < 768) {
940 rounds = 8;
941 } else if (sizeInBits < 1024) {
942 rounds = 4;
943 } else {
944 rounds = 2;
945 }
946 rounds = n < rounds ? n : rounds;
947
948 return passesMillerRabin(rounds, random) && passesLucasLehmer();
949 }
950
951 /**
952 * Returns true iff this BigInteger is a Lucas-Lehmer probable prime.
953 *
954 * The following assumptions are made:
955 * This BigInteger is a positive, odd number.
956 */
957 private boolean passesLucasLehmer() {
958 BigInteger thisPlusOne = this.add(ONE);
959
960 // Step 1
961 int d = 5;
962 while (jacobiSymbol(d, this) != -1) {
963 // 5, -7, 9, -11, ...
964 d = (d < 0) ? Math.abs(d)+2 : -(d+2);
965 }
966
967 // Step 2
968 BigInteger u = lucasLehmerSequence(d, thisPlusOne, this);
969
970 // Step 3
971 return u.mod(this).equals(ZERO);
972 }
973
974 /**
975 * Computes Jacobi(p,n).
976 * Assumes n positive, odd, n>=3.
977 */
978 private static int jacobiSymbol(int p, BigInteger n) {
979 if (p == 0)
980 return 0;
981
982 // Algorithm and comments adapted from Colin Plumb's C library.
983 int j = 1;
984 int u = n.mag[n.mag.length-1];
985
986 // Make p positive
987 if (p < 0) {
988 p = -p;
989 int n8 = u & 7;
990 if ((n8 == 3) || (n8 == 7))
991 j = -j; // 3 (011) or 7 (111) mod 8
992 }
993
994 // Get rid of factors of 2 in p
995 while ((p & 3) == 0)
996 p >>= 2;
997 if ((p & 1) == 0) {
998 p >>= 1;
999 if (((u ^ (u>>1)) & 2) != 0)
1000 j = -j; // 3 (011) or 5 (101) mod 8
1001 }
1002 if (p == 1)
1003 return j;
1004 // Then, apply quadratic reciprocity
1005 if ((p & u & 2) != 0) // p = u = 3 (mod 4)?
1006 j = -j;
1007 // And reduce u mod p
1008 u = n.mod(BigInteger.valueOf(p)).intValue();
1009
1010 // Now compute Jacobi(u,p), u < p
1011 while (u != 0) {
1012 while ((u & 3) == 0)
1013 u >>= 2;
1014 if ((u & 1) == 0) {
1015 u >>= 1;
1016 if (((p ^ (p>>1)) & 2) != 0)
1017 j = -j; // 3 (011) or 5 (101) mod 8
1018 }
1019 if (u == 1)
1020 return j;
1021 // Now both u and p are odd, so use quadratic reciprocity
1022 assert (u < p);
1023 int t = u; u = p; p = t;
1024 if ((u & p & 2) != 0) // u = p = 3 (mod 4)?
1025 j = -j;
1026 // Now u >= p, so it can be reduced
1027 u %= p;
1028 }
1029 return 0;
1030 }
1031
1032 private static BigInteger lucasLehmerSequence(int z, BigInteger k, BigInteger n) {
1033 BigInteger d = BigInteger.valueOf(z);
1034 BigInteger u = ONE; BigInteger u2;
1035 BigInteger v = ONE; BigInteger v2;
1036
1037 for (int i=k.bitLength()-2; i >= 0; i--) {
1038 u2 = u.multiply(v).mod(n);
1039
1040 v2 = v.square().add(d.multiply(u.square())).mod(n);
1041 if (v2.testBit(0))
1042 v2 = v2.subtract(n);
1043
1044 v2 = v2.shiftRight(1);
1045
1046 u = u2; v = v2;
1047 if (k.testBit(i)) {
1048 u2 = u.add(v).mod(n);
1049 if (u2.testBit(0))
1050 u2 = u2.subtract(n);
1051
1052 u2 = u2.shiftRight(1);
1053 v2 = v.add(d.multiply(u)).mod(n);
1054 if (v2.testBit(0))
1055 v2 = v2.subtract(n);
1056 v2 = v2.shiftRight(1);
1057
1058 u = u2; v = v2;
1059 }
1060 }
1061 return u;
1062 }
1063
1064 /**
1065 * Returns true iff this BigInteger passes the specified number of
1066 * Miller-Rabin tests. This test is taken from the DSA spec (NIST FIPS
1067 * 186-2).
1068 *
1069 * The following assumptions are made:
1070 * This BigInteger is a positive, odd number greater than 2.
1071 * iterations<=50.
1072 */
1073 private boolean passesMillerRabin(int iterations, Random rnd) {
1074 // Find a and m such that m is odd and this == 1 + 2**a * m
1075 BigInteger thisMinusOne = this.subtract(ONE);
1076 BigInteger m = thisMinusOne;
1077 int a = m.getLowestSetBit();
1078 m = m.shiftRight(a);
1079
1080 // Do the tests
1081 if (rnd == null) {
1082 rnd = ThreadLocalRandom.current();
1083 }
1084 for (int i=0; i < iterations; i++) {
1085 // Generate a uniform random on (1, this)
1086 BigInteger b;
1087 do {
1088 b = new BigInteger(this.bitLength(), rnd);
1089 } while (b.compareTo(ONE) <= 0 || b.compareTo(this) >= 0);
1090
1091 int j = 0;
1092 BigInteger z = b.modPow(m, this);
1093 while (!((j == 0 && z.equals(ONE)) || z.equals(thisMinusOne))) {
1094 if (j > 0 && z.equals(ONE) || ++j == a)
1095 return false;
1096 z = z.modPow(TWO, this);
1097 }
1098 }
1099 return true;
1100 }
1101
1102 /**
1103 * This internal constructor differs from its public cousin
1104 * with the arguments reversed in two ways: it assumes that its
1105 * arguments are correct, and it doesn't copy the magnitude array.
1106 */
1107 BigInteger(int[] magnitude, int signum) {
1108 this.signum = (magnitude.length == 0 ? 0 : signum);
1109 this.mag = magnitude;
1110 if (mag.length >= MAX_MAG_LENGTH) {
1111 checkRange();
1112 }
1113 }
1114
1115 /**
1116 * This private constructor is for internal use and assumes that its
1117 * arguments are correct. The {@code magnitude} array is assumed to be
1118 * unchanged for the duration of the constructor call.
1119 */
1120 private BigInteger(byte[] magnitude, int signum) {
1121 this.signum = (magnitude.length == 0 ? 0 : signum);
1122 this.mag = stripLeadingZeroBytes(magnitude, 0, magnitude.length);
1123 if (mag.length >= MAX_MAG_LENGTH) {
1124 checkRange();
1125 }
1126 }
1127
1128 /**
1129 * Throws an {@code ArithmeticException} if the {@code BigInteger} would be
1130 * out of the supported range.
1131 *
1132 * @throws ArithmeticException if {@code this} exceeds the supported range.
1133 */
1134 private void checkRange() {
1135 if (mag.length > MAX_MAG_LENGTH || mag.length == MAX_MAG_LENGTH && mag[0] < 0) {
1136 reportOverflow();
1137 }
1138 }
1139
1140 private static void reportOverflow() {
1141 throw new ArithmeticException("BigInteger would overflow supported range");
1142 }
1143
1144 //Static Factory Methods
1145
1146 /**
1147 * Returns a BigInteger whose value is equal to that of the
1148 * specified {@code long}. This "static factory method" is
1149 * provided in preference to a ({@code long}) constructor
1150 * because it allows for reuse of frequently used BigIntegers.
1151 *
1152 * @param val value of the BigInteger to return.
1153 * @return a BigInteger with the specified value.
1154 */
1155 public static BigInteger valueOf(long val) {
1156 // If -MAX_CONSTANT < val < MAX_CONSTANT, return stashed constant
1157 if (val == 0)
1158 return ZERO;
1159 if (val > 0 && val <= MAX_CONSTANT)
1160 return posConst[(int) val];
1161 else if (val < 0 && val >= -MAX_CONSTANT)
1162 return negConst[(int) -val];
1163
1164 return new BigInteger(val);
1165 }
1166
1167 /**
1168 * Constructs a BigInteger with the specified value, which may not be zero.
1169 */
1170 private BigInteger(long val) {
1171 if (val < 0) {
1172 val = -val;
1173 signum = -1;
1174 } else {
1175 signum = 1;
1176 }
1177
1178 int highWord = (int)(val >>> 32);
1179 if (highWord == 0) {
1180 mag = new int[1];
1181 mag[0] = (int)val;
1182 } else {
1183 mag = new int[2];
1184 mag[0] = highWord;
1185 mag[1] = (int)val;
1186 }
1187 }
1188
1189 /**
1190 * Returns a BigInteger with the given two's complement representation.
1191 * Assumes that the input array will not be modified (the returned
1192 * BigInteger will reference the input array if feasible).
1193 */
1194 private static BigInteger valueOf(int val[]) {
1195 return (val[0] > 0 ? new BigInteger(val, 1) : new BigInteger(val));
1196 }
1197
1198 // Constants
1199
1200 /**
1201 * Initialize static constant array when class is loaded.
1202 */
1203 private final static int MAX_CONSTANT = 16;
1204 private static BigInteger posConst[] = new BigInteger[MAX_CONSTANT+1];
1205 private static BigInteger negConst[] = new BigInteger[MAX_CONSTANT+1];
1206
1207 /**
1208 * The cache of powers of each radix. This allows us to not have to
1209 * recalculate powers of radix^(2^n) more than once. This speeds
1210 * Schoenhage recursive base conversion significantly.
1211 */
1212 private static volatile BigInteger[][] powerCache;
1213
1214 /** The cache of logarithms of radices for base conversion. */
1215 private static final double[] logCache;
1216
1217 /** The natural log of 2. This is used in computing cache indices. */
1218 private static final double LOG_TWO = Math.log(2.0);
1219
1220 static {
1221 for (int i = 1; i <= MAX_CONSTANT; i++) {
1222 int[] magnitude = new int[1];
1223 magnitude[0] = i;
1224 posConst[i] = new BigInteger(magnitude, 1);
1225 negConst[i] = new BigInteger(magnitude, -1);
1226 }
1227
1228 /*
1229 * Initialize the cache of radix^(2^x) values used for base conversion
1230 * with just the very first value. Additional values will be created
1231 * on demand.
1232 */
1233 powerCache = new BigInteger[Character.MAX_RADIX+1][];
1234 logCache = new double[Character.MAX_RADIX+1];
1235
1236 for (int i=Character.MIN_RADIX; i <= Character.MAX_RADIX; i++) {
1237 powerCache[i] = new BigInteger[] { BigInteger.valueOf(i) };
1238 logCache[i] = Math.log(i);
1239 }
1240 }
1241
1242 /**
1243 * The BigInteger constant zero.
1244 *
1245 * @since 1.2
1246 */
1247 public static final BigInteger ZERO = new BigInteger(new int[0], 0);
1248
1249 /**
1250 * The BigInteger constant one.
1251 *
1252 * @since 1.2
1253 */
1254 public static final BigInteger ONE = valueOf(1);
1255
1256 /**
1257 * The BigInteger constant two. (Not exported.)
1258 */
1259 private static final BigInteger TWO = valueOf(2);
1260
1261 /**
1262 * The BigInteger constant -1. (Not exported.)
1263 */
1264 private static final BigInteger NEGATIVE_ONE = valueOf(-1);
1265
1266 /**
1267 * The BigInteger constant ten.
1268 *
1269 * @since 1.5
1270 */
1271 public static final BigInteger TEN = valueOf(10);
1272
1273 // Arithmetic Operations
1274
1275 /**
1276 * Returns a BigInteger whose value is {@code (this + val)}.
1277 *
1278 * @param val value to be added to this BigInteger.
1279 * @return {@code this + val}
1280 */
1281 public BigInteger add(BigInteger val) {
1282 if (val.signum == 0)
1283 return this;
1284 if (signum == 0)
1285 return val;
1286 if (val.signum == signum)
1287 return new BigInteger(add(mag, val.mag), signum);
1288
1289 int cmp = compareMagnitude(val);
1290 if (cmp == 0)
1291 return ZERO;
1292 int[] resultMag = (cmp > 0 ? subtract(mag, val.mag)
1293 : subtract(val.mag, mag));
1294 resultMag = trustedStripLeadingZeroInts(resultMag);
1295
1296 return new BigInteger(resultMag, cmp == signum ? 1 : -1);
1297 }
1298
1299 /**
1300 * Package private methods used by BigDecimal code to add a BigInteger
1301 * with a long. Assumes val is not equal to INFLATED.
1302 */
1303 BigInteger add(long val) {
1304 if (val == 0)
1305 return this;
1306 if (signum == 0)
1307 return valueOf(val);
1308 if (Long.signum(val) == signum)
1309 return new BigInteger(add(mag, Math.abs(val)), signum);
1310 int cmp = compareMagnitude(val);
1311 if (cmp == 0)
1312 return ZERO;
1313 int[] resultMag = (cmp > 0 ? subtract(mag, Math.abs(val)) : subtract(Math.abs(val), mag));
1314 resultMag = trustedStripLeadingZeroInts(resultMag);
1315 return new BigInteger(resultMag, cmp == signum ? 1 : -1);
1316 }
1317
1318 /**
1319 * Adds the contents of the int array x and long value val. This
1320 * method allocates a new int array to hold the answer and returns
1321 * a reference to that array. Assumes x.length > 0 and val is
1322 * non-negative
1323 */
1324 private static int[] add(int[] x, long val) {
1325 int[] y;
1326 long sum = 0;
1327 int xIndex = x.length;
1328 int[] result;
1329 int highWord = (int)(val >>> 32);
1330 if (highWord == 0) {
1331 result = new int[xIndex];
1332 sum = (x[--xIndex] & LONG_MASK) + val;
1333 result[xIndex] = (int)sum;
1334 } else {
1335 if (xIndex == 1) {
1336 result = new int[2];
1337 sum = val + (x[0] & LONG_MASK);
1338 result[1] = (int)sum;
1339 result[0] = (int)(sum >>> 32);
1340 return result;
1341 } else {
1342 result = new int[xIndex];
1343 sum = (x[--xIndex] & LONG_MASK) + (val & LONG_MASK);
1344 result[xIndex] = (int)sum;
1345 sum = (x[--xIndex] & LONG_MASK) + (highWord & LONG_MASK) + (sum >>> 32);
1346 result[xIndex] = (int)sum;
1347 }
1348 }
1349 // Copy remainder of longer number while carry propagation is required
1350 boolean carry = (sum >>> 32 != 0);
1351 while (xIndex > 0 && carry)
1352 carry = ((result[--xIndex] = x[xIndex] + 1) == 0);
1353 // Copy remainder of longer number
1354 while (xIndex > 0)
1355 result[--xIndex] = x[xIndex];
1356 // Grow result if necessary
1357 if (carry) {
1358 int bigger[] = new int[result.length + 1];
1359 System.arraycopy(result, 0, bigger, 1, result.length);
1360 bigger[0] = 0x01;
1361 return bigger;
1362 }
1363 return result;
1364 }
1365
1366 /**
1367 * Adds the contents of the int arrays x and y. This method allocates
1368 * a new int array to hold the answer and returns a reference to that
1369 * array.
1370 */
1371 private static int[] add(int[] x, int[] y) {
1372 // If x is shorter, swap the two arrays
1373 if (x.length < y.length) {
1374 int[] tmp = x;
1375 x = y;
1376 y = tmp;
1377 }
1378
1379 int xIndex = x.length;
1380 int yIndex = y.length;
1381 int result[] = new int[xIndex];
1382 long sum = 0;
1383 if (yIndex == 1) {
1384 sum = (x[--xIndex] & LONG_MASK) + (y[0] & LONG_MASK) ;
1385 result[xIndex] = (int)sum;
1386 } else {
1387 // Add common parts of both numbers
1388 while (yIndex > 0) {
1389 sum = (x[--xIndex] & LONG_MASK) +
1390 (y[--yIndex] & LONG_MASK) + (sum >>> 32);
1391 result[xIndex] = (int)sum;
1392 }
1393 }
1394 // Copy remainder of longer number while carry propagation is required
1395 boolean carry = (sum >>> 32 != 0);
1396 while (xIndex > 0 && carry)
1397 carry = ((result[--xIndex] = x[xIndex] + 1) == 0);
1398
1399 // Copy remainder of longer number
1400 while (xIndex > 0)
1401 result[--xIndex] = x[xIndex];
1402
1403 // Grow result if necessary
1404 if (carry) {
1405 int bigger[] = new int[result.length + 1];
1406 System.arraycopy(result, 0, bigger, 1, result.length);
1407 bigger[0] = 0x01;
1408 return bigger;
1409 }
1410 return result;
1411 }
1412
1413 private static int[] subtract(long val, int[] little) {
1414 int highWord = (int)(val >>> 32);
1415 if (highWord == 0) {
1416 int result[] = new int[1];
1417 result[0] = (int)(val - (little[0] & LONG_MASK));
1418 return result;
1419 } else {
1420 int result[] = new int[2];
1421 if (little.length == 1) {
1422 long difference = ((int)val & LONG_MASK) - (little[0] & LONG_MASK);
1423 result[1] = (int)difference;
1424 // Subtract remainder of longer number while borrow propagates
1425 boolean borrow = (difference >> 32 != 0);
1426 if (borrow) {
1427 result[0] = highWord - 1;
1428 } else { // Copy remainder of longer number
1429 result[0] = highWord;
1430 }
1431 return result;
1432 } else { // little.length == 2
1433 long difference = ((int)val & LONG_MASK) - (little[1] & LONG_MASK);
1434 result[1] = (int)difference;
1435 difference = (highWord & LONG_MASK) - (little[0] & LONG_MASK) + (difference >> 32);
1436 result[0] = (int)difference;
1437 return result;
1438 }
1439 }
1440 }
1441
1442 /**
1443 * Subtracts the contents of the second argument (val) from the
1444 * first (big). The first int array (big) must represent a larger number
1445 * than the second. This method allocates the space necessary to hold the
1446 * answer.
1447 * assumes val >= 0
1448 */
1449 private static int[] subtract(int[] big, long val) {
1450 int highWord = (int)(val >>> 32);
1451 int bigIndex = big.length;
1452 int result[] = new int[bigIndex];
1453 long difference = 0;
1454
1455 if (highWord == 0) {
1456 difference = (big[--bigIndex] & LONG_MASK) - val;
1457 result[bigIndex] = (int)difference;
1458 } else {
1459 difference = (big[--bigIndex] & LONG_MASK) - (val & LONG_MASK);
1460 result[bigIndex] = (int)difference;
1461 difference = (big[--bigIndex] & LONG_MASK) - (highWord & LONG_MASK) + (difference >> 32);
1462 result[bigIndex] = (int)difference;
1463 }
1464
1465 // Subtract remainder of longer number while borrow propagates
1466 boolean borrow = (difference >> 32 != 0);
1467 while (bigIndex > 0 && borrow)
1468 borrow = ((result[--bigIndex] = big[bigIndex] - 1) == -1);
1469
1470 // Copy remainder of longer number
1471 while (bigIndex > 0)
1472 result[--bigIndex] = big[bigIndex];
1473
1474 return result;
1475 }
1476
1477 /**
1478 * Returns a BigInteger whose value is {@code (this - val)}.
1479 *
1480 * @param val value to be subtracted from this BigInteger.
1481 * @return {@code this - val}
1482 */
1483 public BigInteger subtract(BigInteger val) {
1484 if (val.signum == 0)
1485 return this;
1486 if (signum == 0)
1487 return val.negate();
1488 if (val.signum != signum)
1489 return new BigInteger(add(mag, val.mag), signum);
1490
1491 int cmp = compareMagnitude(val);
1492 if (cmp == 0)
1493 return ZERO;
1494 int[] resultMag = (cmp > 0 ? subtract(mag, val.mag)
1495 : subtract(val.mag, mag));
1496 resultMag = trustedStripLeadingZeroInts(resultMag);
1497 return new BigInteger(resultMag, cmp == signum ? 1 : -1);
1498 }
1499
1500 /**
1501 * Subtracts the contents of the second int arrays (little) from the
1502 * first (big). The first int array (big) must represent a larger number
1503 * than the second. This method allocates the space necessary to hold the
1504 * answer.
1505 */
1506 private static int[] subtract(int[] big, int[] little) {
1507 int bigIndex = big.length;
1508 int result[] = new int[bigIndex];
1509 int littleIndex = little.length;
1510 long difference = 0;
1511
1512 // Subtract common parts of both numbers
1513 while (littleIndex > 0) {
1514 difference = (big[--bigIndex] & LONG_MASK) -
1515 (little[--littleIndex] & LONG_MASK) +
1516 (difference >> 32);
1517 result[bigIndex] = (int)difference;
1518 }
1519
1520 // Subtract remainder of longer number while borrow propagates
1521 boolean borrow = (difference >> 32 != 0);
1522 while (bigIndex > 0 && borrow)
1523 borrow = ((result[--bigIndex] = big[bigIndex] - 1) == -1);
1524
1525 // Copy remainder of longer number
1526 while (bigIndex > 0)
1527 result[--bigIndex] = big[bigIndex];
1528
1529 return result;
1530 }
1531
1532 /**
1533 * Returns a BigInteger whose value is {@code (this * val)}.
1534 *
1535 * @implNote An implementation may offer better algorithmic
1536 * performance when {@code val == this}.
1537 *
1538 * @param val value to be multiplied by this BigInteger.
1539 * @return {@code this * val}
1540 */
1541 public BigInteger multiply(BigInteger val) {
1542 if (val.signum == 0 || signum == 0)
1543 return ZERO;
1544
1545 int xlen = mag.length;
1546
1547 if (val == this && xlen > MULTIPLY_SQUARE_THRESHOLD) {
1548 return square();
1549 }
1550
1551 int ylen = val.mag.length;
1552
1553 if ((xlen < KARATSUBA_THRESHOLD) || (ylen < KARATSUBA_THRESHOLD)) {
1554 int resultSign = signum == val.signum ? 1 : -1;
1555 if (val.mag.length == 1) {
1556 return multiplyByInt(mag,val.mag[0], resultSign);
1557 }
1558 if (mag.length == 1) {
1559 return multiplyByInt(val.mag,mag[0], resultSign);
1560 }
1561 int[] result = multiplyToLen(mag, xlen,
1562 val.mag, ylen, null);
1563 result = trustedStripLeadingZeroInts(result);
1564 return new BigInteger(result, resultSign);
1565 } else {
1566 if ((xlen < TOOM_COOK_THRESHOLD) && (ylen < TOOM_COOK_THRESHOLD)) {
1567 return multiplyKaratsuba(this, val);
1568 } else {
1569 return multiplyToomCook3(this, val);
1570 }
1571 }
1572 }
1573
1574 private static BigInteger multiplyByInt(int[] x, int y, int sign) {
1575 if (Integer.bitCount(y) == 1) {
1576 return new BigInteger(shiftLeft(x,Integer.numberOfTrailingZeros(y)), sign);
1577 }
1578 int xlen = x.length;
1579 int[] rmag = new int[xlen + 1];
1580 long carry = 0;
1581 long yl = y & LONG_MASK;
1582 int rstart = rmag.length - 1;
1583 for (int i = xlen - 1; i >= 0; i--) {
1584 long product = (x[i] & LONG_MASK) * yl + carry;
1585 rmag[rstart--] = (int)product;
1586 carry = product >>> 32;
1587 }
1588 if (carry == 0L) {
1589 rmag = java.util.Arrays.copyOfRange(rmag, 1, rmag.length);
1590 } else {
1591 rmag[rstart] = (int)carry;
1592 }
1593 return new BigInteger(rmag, sign);
1594 }
1595
1596 /**
1597 * Package private methods used by BigDecimal code to multiply a BigInteger
1598 * with a long. Assumes v is not equal to INFLATED.
1599 */
1600 BigInteger multiply(long v) {
1601 if (v == 0 || signum == 0)
1602 return ZERO;
1603 if (v == BigDecimal.INFLATED)
1604 return multiply(BigInteger.valueOf(v));
1605 int rsign = (v > 0 ? signum : -signum);
1606 if (v < 0)
1607 v = -v;
1608 long dh = v >>> 32; // higher order bits
1609 long dl = v & LONG_MASK; // lower order bits
1610
1611 int xlen = mag.length;
1612 int[] value = mag;
1613 int[] rmag = (dh == 0L) ? (new int[xlen + 1]) : (new int[xlen + 2]);
1614 long carry = 0;
1615 int rstart = rmag.length - 1;
1616 for (int i = xlen - 1; i >= 0; i--) {
1617 long product = (value[i] & LONG_MASK) * dl + carry;
1618 rmag[rstart--] = (int)product;
1619 carry = product >>> 32;
1620 }
1621 rmag[rstart] = (int)carry;
1622 if (dh != 0L) {
1623 carry = 0;
1624 rstart = rmag.length - 2;
1625 for (int i = xlen - 1; i >= 0; i--) {
1626 long product = (value[i] & LONG_MASK) * dh +
1627 (rmag[rstart] & LONG_MASK) + carry;
1628 rmag[rstart--] = (int)product;
1629 carry = product >>> 32;
1630 }
1631 rmag[0] = (int)carry;
1632 }
1633 if (carry == 0L)
1634 rmag = java.util.Arrays.copyOfRange(rmag, 1, rmag.length);
1635 return new BigInteger(rmag, rsign);
1636 }
1637
1638 /**
1639 * Multiplies int arrays x and y to the specified lengths and places
1640 * the result into z. There will be no leading zeros in the resultant array.
1641 */
1642 private int[] multiplyToLen(int[] x, int xlen, int[] y, int ylen, int[] z) {
1643 int xstart = xlen - 1;
1644 int ystart = ylen - 1;
1645
1646 if (z == null || z.length < (xlen+ ylen))
1647 z = new int[xlen+ylen];
1648
1649 long carry = 0;
1650 for (int j=ystart, k=ystart+1+xstart; j >= 0; j--, k--) {
1651 long product = (y[j] & LONG_MASK) *
1652 (x[xstart] & LONG_MASK) + carry;
1653 z[k] = (int)product;
1654 carry = product >>> 32;
1655 }
1656 z[xstart] = (int)carry;
1657
1658 for (int i = xstart-1; i >= 0; i--) {
1659 carry = 0;
1660 for (int j=ystart, k=ystart+1+i; j >= 0; j--, k--) {
1661 long product = (y[j] & LONG_MASK) *
1662 (x[i] & LONG_MASK) +
1663 (z[k] & LONG_MASK) + carry;
1664 z[k] = (int)product;
1665 carry = product >>> 32;
1666 }
1667 z[i] = (int)carry;
1668 }
1669 return z;
1670 }
1671
1672 /**
1673 * Multiplies two BigIntegers using the Karatsuba multiplication
1674 * algorithm. This is a recursive divide-and-conquer algorithm which is
1675 * more efficient for large numbers than what is commonly called the
1676 * "grade-school" algorithm used in multiplyToLen. If the numbers to be
1677 * multiplied have length n, the "grade-school" algorithm has an
1678 * asymptotic complexity of O(n^2). In contrast, the Karatsuba algorithm
1679 * has complexity of O(n^(log2(3))), or O(n^1.585). It achieves this
1680 * increased performance by doing 3 multiplies instead of 4 when
1681 * evaluating the product. As it has some overhead, should be used when
1682 * both numbers are larger than a certain threshold (found
1683 * experimentally).
1684 *
1685 * See: http://en.wikipedia.org/wiki/Karatsuba_algorithm
1686 */
1687 private static BigInteger multiplyKaratsuba(BigInteger x, BigInteger y) {
1688 int xlen = x.mag.length;
1689 int ylen = y.mag.length;
1690
1691 // The number of ints in each half of the number.
1692 int half = (Math.max(xlen, ylen)+1) / 2;
1693
1694 // xl and yl are the lower halves of x and y respectively,
1695 // xh and yh are the upper halves.
1696 BigInteger xl = x.getLower(half);
1697 BigInteger xh = x.getUpper(half);
1698 BigInteger yl = y.getLower(half);
1699 BigInteger yh = y.getUpper(half);
1700
1701 BigInteger p1 = xh.multiply(yh); // p1 = xh*yh
1702 BigInteger p2 = xl.multiply(yl); // p2 = xl*yl
1703
1704 // p3=(xh+xl)*(yh+yl)
1705 BigInteger p3 = xh.add(xl).multiply(yh.add(yl));
1706
1707 // result = p1 * 2^(32*2*half) + (p3 - p1 - p2) * 2^(32*half) + p2
1708 BigInteger result = p1.shiftLeft(32*half).add(p3.subtract(p1).subtract(p2)).shiftLeft(32*half).add(p2);
1709
1710 if (x.signum != y.signum) {
1711 return result.negate();
1712 } else {
1713 return result;
1714 }
1715 }
1716
1717 /**
1718 * Multiplies two BigIntegers using a 3-way Toom-Cook multiplication
1719 * algorithm. This is a recursive divide-and-conquer algorithm which is
1720 * more efficient for large numbers than what is commonly called the
1721 * "grade-school" algorithm used in multiplyToLen. If the numbers to be
1722 * multiplied have length n, the "grade-school" algorithm has an
1723 * asymptotic complexity of O(n^2). In contrast, 3-way Toom-Cook has a
1724 * complexity of about O(n^1.465). It achieves this increased asymptotic
1725 * performance by breaking each number into three parts and by doing 5
1726 * multiplies instead of 9 when evaluating the product. Due to overhead
1727 * (additions, shifts, and one division) in the Toom-Cook algorithm, it
1728 * should only be used when both numbers are larger than a certain
1729 * threshold (found experimentally). This threshold is generally larger
1730 * than that for Karatsuba multiplication, so this algorithm is generally
1731 * only used when numbers become significantly larger.
1732 *
1733 * The algorithm used is the "optimal" 3-way Toom-Cook algorithm outlined
1734 * by Marco Bodrato.
1735 *
1736 * See: http://bodrato.it/toom-cook/
1737 * http://bodrato.it/papers/#WAIFI2007
1738 *
1739 * "Towards Optimal Toom-Cook Multiplication for Univariate and
1740 * Multivariate Polynomials in Characteristic 2 and 0." by Marco BODRATO;
1741 * In C.Carlet and B.Sunar, Eds., "WAIFI'07 proceedings", p. 116-133,
1742 * LNCS #4547. Springer, Madrid, Spain, June 21-22, 2007.
1743 *
1744 */
1745 private static BigInteger multiplyToomCook3(BigInteger a, BigInteger b) {
1746 int alen = a.mag.length;
1747 int blen = b.mag.length;
1748
1749 int largest = Math.max(alen, blen);
1750
1751 // k is the size (in ints) of the lower-order slices.
1752 int k = (largest+2)/3; // Equal to ceil(largest/3)
1753
1754 // r is the size (in ints) of the highest-order slice.
1755 int r = largest - 2*k;
1756
1757 // Obtain slices of the numbers. a2 and b2 are the most significant
1758 // bits of the numbers a and b, and a0 and b0 the least significant.
1759 BigInteger a0, a1, a2, b0, b1, b2;
1760 a2 = a.getToomSlice(k, r, 0, largest);
1761 a1 = a.getToomSlice(k, r, 1, largest);
1762 a0 = a.getToomSlice(k, r, 2, largest);
1763 b2 = b.getToomSlice(k, r, 0, largest);
1764 b1 = b.getToomSlice(k, r, 1, largest);
1765 b0 = b.getToomSlice(k, r, 2, largest);
1766
1767 BigInteger v0, v1, v2, vm1, vinf, t1, t2, tm1, da1, db1;
1768
1769 v0 = a0.multiply(b0);
1770 da1 = a2.add(a0);
1771 db1 = b2.add(b0);
1772 vm1 = da1.subtract(a1).multiply(db1.subtract(b1));
1773 da1 = da1.add(a1);
1774 db1 = db1.add(b1);
1775 v1 = da1.multiply(db1);
1776 v2 = da1.add(a2).shiftLeft(1).subtract(a0).multiply(
1777 db1.add(b2).shiftLeft(1).subtract(b0));
1778 vinf = a2.multiply(b2);
1779
1780 // The algorithm requires two divisions by 2 and one by 3.
1781 // All divisions are known to be exact, that is, they do not produce
1782 // remainders, and all results are positive. The divisions by 2 are
1783 // implemented as right shifts which are relatively efficient, leaving
1784 // only an exact division by 3, which is done by a specialized
1785 // linear-time algorithm.
1786 t2 = v2.subtract(vm1).exactDivideBy3();
1787 tm1 = v1.subtract(vm1).shiftRight(1);
1788 t1 = v1.subtract(v0);
1789 t2 = t2.subtract(t1).shiftRight(1);
1790 t1 = t1.subtract(tm1).subtract(vinf);
1791 t2 = t2.subtract(vinf.shiftLeft(1));
1792 tm1 = tm1.subtract(t2);
1793
1794 // Number of bits to shift left.
1795 int ss = k*32;
1796
1797 BigInteger result = vinf.shiftLeft(ss).add(t2).shiftLeft(ss).add(t1).shiftLeft(ss).add(tm1).shiftLeft(ss).add(v0);
1798
1799 if (a.signum != b.signum) {
1800 return result.negate();
1801 } else {
1802 return result;
1803 }
1804 }
1805
1806
1807 /**
1808 * Returns a slice of a BigInteger for use in Toom-Cook multiplication.
1809 *
1810 * @param lowerSize The size of the lower-order bit slices.
1811 * @param upperSize The size of the higher-order bit slices.
1812 * @param slice The index of which slice is requested, which must be a
1813 * number from 0 to size-1. Slice 0 is the highest-order bits, and slice
1814 * size-1 are the lowest-order bits. Slice 0 may be of different size than
1815 * the other slices.
1816 * @param fullsize The size of the larger integer array, used to align
1817 * slices to the appropriate position when multiplying different-sized
1818 * numbers.
1819 */
1820 private BigInteger getToomSlice(int lowerSize, int upperSize, int slice,
1821 int fullsize) {
1822 int start, end, sliceSize, len, offset;
1823
1824 len = mag.length;
1825 offset = fullsize - len;
1826
1827 if (slice == 0) {
1828 start = 0 - offset;
1829 end = upperSize - 1 - offset;
1830 } else {
1831 start = upperSize + (slice-1)*lowerSize - offset;
1832 end = start + lowerSize - 1;
1833 }
1834
1835 if (start < 0) {
1836 start = 0;
1837 }
1838 if (end < 0) {
1839 return ZERO;
1840 }
1841
1842 sliceSize = (end-start) + 1;
1843
1844 if (sliceSize <= 0) {
1845 return ZERO;
1846 }
1847
1848 // While performing Toom-Cook, all slices are positive and
1849 // the sign is adjusted when the final number is composed.
1850 if (start == 0 && sliceSize >= len) {
1851 return this.abs();
1852 }
1853
1854 int intSlice[] = new int[sliceSize];
1855 System.arraycopy(mag, start, intSlice, 0, sliceSize);
1856
1857 return new BigInteger(trustedStripLeadingZeroInts(intSlice), 1);
1858 }
1859
1860 /**
1861 * Does an exact division (that is, the remainder is known to be zero)
1862 * of the specified number by 3. This is used in Toom-Cook
1863 * multiplication. This is an efficient algorithm that runs in linear
1864 * time. If the argument is not exactly divisible by 3, results are
1865 * undefined. Note that this is expected to be called with positive
1866 * arguments only.
1867 */
1868 private BigInteger exactDivideBy3() {
1869 int len = mag.length;
1870 int[] result = new int[len];
1871 long x, w, q, borrow;
1872 borrow = 0L;
1873 for (int i=len-1; i >= 0; i--) {
1874 x = (mag[i] & LONG_MASK);
1875 w = x - borrow;
1876 if (borrow > x) { // Did we make the number go negative?
1877 borrow = 1L;
1878 } else {
1879 borrow = 0L;
1880 }
1881
1882 // 0xAAAAAAAB is the modular inverse of 3 (mod 2^32). Thus,
1883 // the effect of this is to divide by 3 (mod 2^32).
1884 // This is much faster than division on most architectures.
1885 q = (w * 0xAAAAAAABL) & LONG_MASK;
1886 result[i] = (int) q;
1887
1888 // Now check the borrow. The second check can of course be
1889 // eliminated if the first fails.
1890 if (q >= 0x55555556L) {
1891 borrow++;
1892 if (q >= 0xAAAAAAABL)
1893 borrow++;
1894 }
1895 }
1896 result = trustedStripLeadingZeroInts(result);
1897 return new BigInteger(result, signum);
1898 }
1899
1900 /**
1901 * Returns a new BigInteger representing n lower ints of the number.
1902 * This is used by Karatsuba multiplication and Karatsuba squaring.
1903 */
1904 private BigInteger getLower(int n) {
1905 int len = mag.length;
1906
1907 if (len <= n) {
1908 return abs();
1909 }
1910
1911 int lowerInts[] = new int[n];
1912 System.arraycopy(mag, len-n, lowerInts, 0, n);
1913
1914 return new BigInteger(trustedStripLeadingZeroInts(lowerInts), 1);
1915 }
1916
1917 /**
1918 * Returns a new BigInteger representing mag.length-n upper
1919 * ints of the number. This is used by Karatsuba multiplication and
1920 * Karatsuba squaring.
1921 */
1922 private BigInteger getUpper(int n) {
1923 int len = mag.length;
1924
1925 if (len <= n) {
1926 return ZERO;
1927 }
1928
1929 int upperLen = len - n;
1930 int upperInts[] = new int[upperLen];
1931 System.arraycopy(mag, 0, upperInts, 0, upperLen);
1932
1933 return new BigInteger(trustedStripLeadingZeroInts(upperInts), 1);
1934 }
1935
1936 // Squaring
1937
1938 /**
1939 * Returns a BigInteger whose value is {@code (this<sup>2</sup>)}.
1940 *
1941 * @return {@code this<sup>2</sup>}
1942 */
1943 private BigInteger square() {
1944 if (signum == 0) {
1945 return ZERO;
1946 }
1947 int len = mag.length;
1948
1949 if (len < KARATSUBA_SQUARE_THRESHOLD) {
1950 int[] z = squareToLen(mag, len, null);
1951 return new BigInteger(trustedStripLeadingZeroInts(z), 1);
1952 } else {
1953 if (len < TOOM_COOK_SQUARE_THRESHOLD) {
1954 return squareKaratsuba();
1955 } else {
1956 return squareToomCook3();
1957 }
1958 }
1959 }
1960
1961 /**
1962 * Squares the contents of the int array x. The result is placed into the
1963 * int array z. The contents of x are not changed.
1964 */
1965 private static final int[] squareToLen(int[] x, int len, int[] z) {
1966 int zlen = len << 1;
1967 if (z == null || z.length < zlen)
1968 z = new int[zlen];
1969
1970 // Execute checks before calling intrinsified method.
1971 implSquareToLenChecks(x, len, z, zlen);
1972 return implSquareToLen(x, len, z, zlen);
1973 }
1974
1975 /**
1976 * Parameters validation.
1977 */
1978 private static void implSquareToLenChecks(int[] x, int len, int[] z, int zlen) throws RuntimeException {
1979 if (len < 1) {
1980 throw new IllegalArgumentException("invalid input length: " + len);
1981 }
1982 if (len > x.length) {
1983 throw new IllegalArgumentException("input length out of bound: " +
1984 len + " > " + x.length);
1985 }
1986 if (len * 2 > z.length) {
1987 throw new IllegalArgumentException("input length out of bound: " +
1988 (len * 2) + " > " + z.length);
1989 }
1990 if (zlen < 1) {
1991 throw new IllegalArgumentException("invalid input length: " + zlen);
1992 }
1993 if (zlen > z.length) {
1994 throw new IllegalArgumentException("input length out of bound: " +
1995 len + " > " + z.length);
1996 }
1997 }
1998
1999 /**
2000 * Java Runtime may use intrinsic for this method.
2001 */
2002 private static final int[] implSquareToLen(int[] x, int len, int[] z, int zlen) {
2003 /*
2004 * The algorithm used here is adapted from Colin Plumb's C library.
2005 * Technique: Consider the partial products in the multiplication
2006 * of "abcde" by itself:
2007 *
2008 * a b c d e
2009 * * a b c d e
2010 * ==================
2011 * ae be ce de ee
2012 * ad bd cd dd de
2013 * ac bc cc cd ce
2014 * ab bb bc bd be
2015 * aa ab ac ad ae
2016 *
2017 * Note that everything above the main diagonal:
2018 * ae be ce de = (abcd) * e
2019 * ad bd cd = (abc) * d
2020 * ac bc = (ab) * c
2021 * ab = (a) * b
2022 *
2023 * is a copy of everything below the main diagonal:
2024 * de
2025 * cd ce
2026 * bc bd be
2027 * ab ac ad ae
2028 *
2029 * Thus, the sum is 2 * (off the diagonal) + diagonal.
2030 *
2031 * This is accumulated beginning with the diagonal (which
2032 * consist of the squares of the digits of the input), which is then
2033 * divided by two, the off-diagonal added, and multiplied by two
2034 * again. The low bit is simply a copy of the low bit of the
2035 * input, so it doesn't need special care.
2036 */
2037
2038 // Store the squares, right shifted one bit (i.e., divided by 2)
2039 int lastProductLowWord = 0;
2040 for (int j=0, i=0; j < len; j++) {
2041 long piece = (x[j] & LONG_MASK);
2042 long product = piece * piece;
2043 z[i++] = (lastProductLowWord << 31) | (int)(product >>> 33);
2044 z[i++] = (int)(product >>> 1);
2045 lastProductLowWord = (int)product;
2046 }
2047
2048 // Add in off-diagonal sums
2049 for (int i=len, offset=1; i > 0; i--, offset+=2) {
2050 int t = x[i-1];
2051 t = mulAdd(z, x, offset, i-1, t);
2052 addOne(z, offset-1, i, t);
2053 }
2054
2055 // Shift back up and set low bit
2056 primitiveLeftShift(z, zlen, 1);
2057 z[zlen-1] |= x[len-1] & 1;
2058
2059 return z;
2060 }
2061
2062 /**
2063 * Squares a BigInteger using the Karatsuba squaring algorithm. It should
2064 * be used when both numbers are larger than a certain threshold (found
2065 * experimentally). It is a recursive divide-and-conquer algorithm that
2066 * has better asymptotic performance than the algorithm used in
2067 * squareToLen.
2068 */
2069 private BigInteger squareKaratsuba() {
2070 int half = (mag.length+1) / 2;
2071
2072 BigInteger xl = getLower(half);
2073 BigInteger xh = getUpper(half);
2074
2075 BigInteger xhs = xh.square(); // xhs = xh^2
2076 BigInteger xls = xl.square(); // xls = xl^2
2077
2078 // xh^2 << 64 + (((xl+xh)^2 - (xh^2 + xl^2)) << 32) + xl^2
2079 return xhs.shiftLeft(half*32).add(xl.add(xh).square().subtract(xhs.add(xls))).shiftLeft(half*32).add(xls);
2080 }
2081
2082 /**
2083 * Squares a BigInteger using the 3-way Toom-Cook squaring algorithm. It
2084 * should be used when both numbers are larger than a certain threshold
2085 * (found experimentally). It is a recursive divide-and-conquer algorithm
2086 * that has better asymptotic performance than the algorithm used in
2087 * squareToLen or squareKaratsuba.
2088 */
2089 private BigInteger squareToomCook3() {
2090 int len = mag.length;
2091
2092 // k is the size (in ints) of the lower-order slices.
2093 int k = (len+2)/3; // Equal to ceil(largest/3)
2094
2095 // r is the size (in ints) of the highest-order slice.
2096 int r = len - 2*k;
2097
2098 // Obtain slices of the numbers. a2 is the most significant
2099 // bits of the number, and a0 the least significant.
2100 BigInteger a0, a1, a2;
2101 a2 = getToomSlice(k, r, 0, len);
2102 a1 = getToomSlice(k, r, 1, len);
2103 a0 = getToomSlice(k, r, 2, len);
2104 BigInteger v0, v1, v2, vm1, vinf, t1, t2, tm1, da1;
2105
2106 v0 = a0.square();
2107 da1 = a2.add(a0);
2108 vm1 = da1.subtract(a1).square();
2109 da1 = da1.add(a1);
2110 v1 = da1.square();
2111 vinf = a2.square();
2112 v2 = da1.add(a2).shiftLeft(1).subtract(a0).square();
2113
2114 // The algorithm requires two divisions by 2 and one by 3.
2115 // All divisions are known to be exact, that is, they do not produce
2116 // remainders, and all results are positive. The divisions by 2 are
2117 // implemented as right shifts which are relatively efficient, leaving
2118 // only a division by 3.
2119 // The division by 3 is done by an optimized algorithm for this case.
2120 t2 = v2.subtract(vm1).exactDivideBy3();
2121 tm1 = v1.subtract(vm1).shiftRight(1);
2122 t1 = v1.subtract(v0);
2123 t2 = t2.subtract(t1).shiftRight(1);
2124 t1 = t1.subtract(tm1).subtract(vinf);
2125 t2 = t2.subtract(vinf.shiftLeft(1));
2126 tm1 = tm1.subtract(t2);
2127
2128 // Number of bits to shift left.
2129 int ss = k*32;
2130
2131 return vinf.shiftLeft(ss).add(t2).shiftLeft(ss).add(t1).shiftLeft(ss).add(tm1).shiftLeft(ss).add(v0);
2132 }
2133
2134 // Division
2135
2136 /**
2137 * Returns a BigInteger whose value is {@code (this / val)}.
2138 *
2139 * @param val value by which this BigInteger is to be divided.
2140 * @return {@code this / val}
2141 * @throws ArithmeticException if {@code val} is zero.
2142 */
2143 public BigInteger divide(BigInteger val) {
2144 if (val.mag.length < BURNIKEL_ZIEGLER_THRESHOLD ||
2145 mag.length - val.mag.length < BURNIKEL_ZIEGLER_OFFSET) {
2146 return divideKnuth(val);
2147 } else {
2148 return divideBurnikelZiegler(val);
2149 }
2150 }
2151
2152 /**
2153 * Returns a BigInteger whose value is {@code (this / val)} using an O(n^2) algorithm from Knuth.
2154 *
2155 * @param val value by which this BigInteger is to be divided.
2156 * @return {@code this / val}
2157 * @throws ArithmeticException if {@code val} is zero.
2158 * @see MutableBigInteger#divideKnuth(MutableBigInteger, MutableBigInteger, boolean)
2159 */
2160 private BigInteger divideKnuth(BigInteger val) {
2161 MutableBigInteger q = new MutableBigInteger(),
2162 a = new MutableBigInteger(this.mag),
2163 b = new MutableBigInteger(val.mag);
2164
2165 a.divideKnuth(b, q, false);
2166 return q.toBigInteger(this.signum * val.signum);
2167 }
2168
2169 /**
2170 * Returns an array of two BigIntegers containing {@code (this / val)}
2171 * followed by {@code (this % val)}.
2172 *
2173 * @param val value by which this BigInteger is to be divided, and the
2174 * remainder computed.
2175 * @return an array of two BigIntegers: the quotient {@code (this / val)}
2176 * is the initial element, and the remainder {@code (this % val)}
2177 * is the final element.
2178 * @throws ArithmeticException if {@code val} is zero.
2179 */
2180 public BigInteger[] divideAndRemainder(BigInteger val) {
2181 if (val.mag.length < BURNIKEL_ZIEGLER_THRESHOLD ||
2182 mag.length - val.mag.length < BURNIKEL_ZIEGLER_OFFSET) {
2183 return divideAndRemainderKnuth(val);
2184 } else {
2185 return divideAndRemainderBurnikelZiegler(val);
2186 }
2187 }
2188
2189 /** Long division */
2190 private BigInteger[] divideAndRemainderKnuth(BigInteger val) {
2191 BigInteger[] result = new BigInteger[2];
2192 MutableBigInteger q = new MutableBigInteger(),
2193 a = new MutableBigInteger(this.mag),
2194 b = new MutableBigInteger(val.mag);
2195 MutableBigInteger r = a.divideKnuth(b, q);
2196 result[0] = q.toBigInteger(this.signum == val.signum ? 1 : -1);
2197 result[1] = r.toBigInteger(this.signum);
2198 return result;
2199 }
2200
2201 /**
2202 * Returns a BigInteger whose value is {@code (this % val)}.
2203 *
2204 * @param val value by which this BigInteger is to be divided, and the
2205 * remainder computed.
2206 * @return {@code this % val}
2207 * @throws ArithmeticException if {@code val} is zero.
2208 */
2209 public BigInteger remainder(BigInteger val) {
2210 if (val.mag.length < BURNIKEL_ZIEGLER_THRESHOLD ||
2211 mag.length - val.mag.length < BURNIKEL_ZIEGLER_OFFSET) {
2212 return remainderKnuth(val);
2213 } else {
2214 return remainderBurnikelZiegler(val);
2215 }
2216 }
2217
2218 /** Long division */
2219 private BigInteger remainderKnuth(BigInteger val) {
2220 MutableBigInteger q = new MutableBigInteger(),
2221 a = new MutableBigInteger(this.mag),
2222 b = new MutableBigInteger(val.mag);
2223
2224 return a.divideKnuth(b, q).toBigInteger(this.signum);
2225 }
2226
2227 /**
2228 * Calculates {@code this / val} using the Burnikel-Ziegler algorithm.
2229 * @param val the divisor
2230 * @return {@code this / val}
2231 */
2232 private BigInteger divideBurnikelZiegler(BigInteger val) {
2233 return divideAndRemainderBurnikelZiegler(val)[0];
2234 }
2235
2236 /**
2237 * Calculates {@code this % val} using the Burnikel-Ziegler algorithm.
2238 * @param val the divisor
2239 * @return {@code this % val}
2240 */
2241 private BigInteger remainderBurnikelZiegler(BigInteger val) {
2242 return divideAndRemainderBurnikelZiegler(val)[1];
2243 }
2244
2245 /**
2246 * Computes {@code this / val} and {@code this % val} using the
2247 * Burnikel-Ziegler algorithm.
2248 * @param val the divisor
2249 * @return an array containing the quotient and remainder
2250 */
2251 private BigInteger[] divideAndRemainderBurnikelZiegler(BigInteger val) {
2252 MutableBigInteger q = new MutableBigInteger();
2253 MutableBigInteger r = new MutableBigInteger(this).divideAndRemainderBurnikelZiegler(new MutableBigInteger(val), q);
2254 BigInteger qBigInt = q.isZero() ? ZERO : q.toBigInteger(signum*val.signum);
2255 BigInteger rBigInt = r.isZero() ? ZERO : r.toBigInteger(signum);
2256 return new BigInteger[] {qBigInt, rBigInt};
2257 }
2258
2259 /**
2260 * Returns a BigInteger whose value is <tt>(this<sup>exponent</sup>)</tt>.
2261 * Note that {@code exponent} is an integer rather than a BigInteger.
2262 *
2263 * @param exponent exponent to which this BigInteger is to be raised.
2264 * @return <tt>this<sup>exponent</sup></tt>
2265 * @throws ArithmeticException {@code exponent} is negative. (This would
2266 * cause the operation to yield a non-integer value.)
2267 */
2268 public BigInteger pow(int exponent) {
2269 if (exponent < 0) {
2270 throw new ArithmeticException("Negative exponent");
2271 }
2272 if (signum == 0) {
2273 return (exponent == 0 ? ONE : this);
2274 }
2275
2276 BigInteger partToSquare = this.abs();
2277
2278 // Factor out powers of two from the base, as the exponentiation of
2279 // these can be done by left shifts only.
2280 // The remaining part can then be exponentiated faster. The
2281 // powers of two will be multiplied back at the end.
2282 int powersOfTwo = partToSquare.getLowestSetBit();
2283 long bitsToShift = (long)powersOfTwo * exponent;
2284 if (bitsToShift > Integer.MAX_VALUE) {
2285 reportOverflow();
2286 }
2287
2288 int remainingBits;
2289
2290 // Factor the powers of two out quickly by shifting right, if needed.
2291 if (powersOfTwo > 0) {
2292 partToSquare = partToSquare.shiftRight(powersOfTwo);
2293 remainingBits = partToSquare.bitLength();
2294 if (remainingBits == 1) { // Nothing left but +/- 1?
2295 if (signum < 0 && (exponent&1) == 1) {
2296 return NEGATIVE_ONE.shiftLeft(powersOfTwo*exponent);
2297 } else {
2298 return ONE.shiftLeft(powersOfTwo*exponent);
2299 }
2300 }
2301 } else {
2302 remainingBits = partToSquare.bitLength();
2303 if (remainingBits == 1) { // Nothing left but +/- 1?
2304 if (signum < 0 && (exponent&1) == 1) {
2305 return NEGATIVE_ONE;
2306 } else {
2307 return ONE;
2308 }
2309 }
2310 }
2311
2312 // This is a quick way to approximate the size of the result,
2313 // similar to doing log2[n] * exponent. This will give an upper bound
2314 // of how big the result can be, and which algorithm to use.
2315 long scaleFactor = (long)remainingBits * exponent;
2316
2317 // Use slightly different algorithms for small and large operands.
2318 // See if the result will safely fit into a long. (Largest 2^63-1)
2319 if (partToSquare.mag.length == 1 && scaleFactor <= 62) {
2320 // Small number algorithm. Everything fits into a long.
2321 int newSign = (signum <0 && (exponent&1) == 1 ? -1 : 1);
2322 long result = 1;
2323 long baseToPow2 = partToSquare.mag[0] & LONG_MASK;
2324
2325 int workingExponent = exponent;
2326
2327 // Perform exponentiation using repeated squaring trick
2328 while (workingExponent != 0) {
2329 if ((workingExponent & 1) == 1) {
2330 result = result * baseToPow2;
2331 }
2332
2333 if ((workingExponent >>>= 1) != 0) {
2334 baseToPow2 = baseToPow2 * baseToPow2;
2335 }
2336 }
2337
2338 // Multiply back the powers of two (quickly, by shifting left)
2339 if (powersOfTwo > 0) {
2340 if (bitsToShift + scaleFactor <= 62) { // Fits in long?
2341 return valueOf((result << bitsToShift) * newSign);
2342 } else {
2343 return valueOf(result*newSign).shiftLeft((int) bitsToShift);
2344 }
2345 }
2346 else {
2347 return valueOf(result*newSign);
2348 }
2349 } else {
2350 // Large number algorithm. This is basically identical to
2351 // the algorithm above, but calls multiply() and square()
2352 // which may use more efficient algorithms for large numbers.
2353 BigInteger answer = ONE;
2354
2355 int workingExponent = exponent;
2356 // Perform exponentiation using repeated squaring trick
2357 while (workingExponent != 0) {
2358 if ((workingExponent & 1) == 1) {
2359 answer = answer.multiply(partToSquare);
2360 }
2361
2362 if ((workingExponent >>>= 1) != 0) {
2363 partToSquare = partToSquare.square();
2364 }
2365 }
2366 // Multiply back the (exponentiated) powers of two (quickly,
2367 // by shifting left)
2368 if (powersOfTwo > 0) {
2369 answer = answer.shiftLeft(powersOfTwo*exponent);
2370 }
2371
2372 if (signum < 0 && (exponent&1) == 1) {
2373 return answer.negate();
2374 } else {
2375 return answer;
2376 }
2377 }
2378 }
2379
2380 /**
2381 * Returns a BigInteger whose value is the greatest common divisor of
2382 * {@code abs(this)} and {@code abs(val)}. Returns 0 if
2383 * {@code this == 0 && val == 0}.
2384 *
2385 * @param val value with which the GCD is to be computed.
2386 * @return {@code GCD(abs(this), abs(val))}
2387 */
2388 public BigInteger gcd(BigInteger val) {
2389 if (val.signum == 0)
2390 return this.abs();
2391 else if (this.signum == 0)
2392 return val.abs();
2393
2394 MutableBigInteger a = new MutableBigInteger(this);
2395 MutableBigInteger b = new MutableBigInteger(val);
2396
2397 MutableBigInteger result = a.hybridGCD(b);
2398
2399 return result.toBigInteger(1);
2400 }
2401
2402 /**
2403 * Package private method to return bit length for an integer.
2404 */
2405 static int bitLengthForInt(int n) {
2406 return 32 - Integer.numberOfLeadingZeros(n);
2407 }
2408
2409 /**
2410 * Left shift int array a up to len by n bits. Returns the array that
2411 * results from the shift since space may have to be reallocated.
2412 */
2413 private static int[] leftShift(int[] a, int len, int n) {
2414 int nInts = n >>> 5;
2415 int nBits = n&0x1F;
2416 int bitsInHighWord = bitLengthForInt(a[0]);
2417
2418 // If shift can be done without recopy, do so
2419 if (n <= (32-bitsInHighWord)) {
2420 primitiveLeftShift(a, len, nBits);
2421 return a;
2422 } else { // Array must be resized
2423 if (nBits <= (32-bitsInHighWord)) {
2424 int result[] = new int[nInts+len];
2425 System.arraycopy(a, 0, result, 0, len);
2426 primitiveLeftShift(result, result.length, nBits);
2427 return result;
2428 } else {
2429 int result[] = new int[nInts+len+1];
2430 System.arraycopy(a, 0, result, 0, len);
2431 primitiveRightShift(result, result.length, 32 - nBits);
2432 return result;
2433 }
2434 }
2435 }
2436
2437 // shifts a up to len right n bits assumes no leading zeros, 0<n<32
2438 static void primitiveRightShift(int[] a, int len, int n) {
2439 int n2 = 32 - n;
2440 for (int i=len-1, c=a[i]; i > 0; i--) {
2441 int b = c;
2442 c = a[i-1];
2443 a[i] = (c << n2) | (b >>> n);
2444 }
2445 a[0] >>>= n;
2446 }
2447
2448 // shifts a up to len left n bits assumes no leading zeros, 0<=n<32
2449 static void primitiveLeftShift(int[] a, int len, int n) {
2450 if (len == 0 || n == 0)
2451 return;
2452
2453 int n2 = 32 - n;
2454 for (int i=0, c=a[i], m=i+len-1; i < m; i++) {
2455 int b = c;
2456 c = a[i+1];
2457 a[i] = (b << n) | (c >>> n2);
2458 }
2459 a[len-1] <<= n;
2460 }
2461
2462 /**
2463 * Calculate bitlength of contents of the first len elements an int array,
2464 * assuming there are no leading zero ints.
2465 */
2466 private static int bitLength(int[] val, int len) {
2467 if (len == 0)
2468 return 0;
2469 return ((len - 1) << 5) + bitLengthForInt(val[0]);
2470 }
2471
2472 /**
2473 * Returns a BigInteger whose value is the absolute value of this
2474 * BigInteger.
2475 *
2476 * @return {@code abs(this)}
2477 */
2478 public BigInteger abs() {
2479 return (signum >= 0 ? this : this.negate());
2480 }
2481
2482 /**
2483 * Returns a BigInteger whose value is {@code (-this)}.
2484 *
2485 * @return {@code -this}
2486 */
2487 public BigInteger negate() {
2488 return new BigInteger(this.mag, -this.signum);
2489 }
2490
2491 /**
2492 * Returns the signum function of this BigInteger.
2493 *
2494 * @return -1, 0 or 1 as the value of this BigInteger is negative, zero or
2495 * positive.
2496 */
2497 public int signum() {
2498 return this.signum;
2499 }
2500
2501 // Modular Arithmetic Operations
2502
2503 /**
2504 * Returns a BigInteger whose value is {@code (this mod m}). This method
2505 * differs from {@code remainder} in that it always returns a
2506 * <i>non-negative</i> BigInteger.
2507 *
2508 * @param m the modulus.
2509 * @return {@code this mod m}
2510 * @throws ArithmeticException {@code m} ≤ 0
2511 * @see #remainder
2512 */
2513 public BigInteger mod(BigInteger m) {
2514 if (m.signum <= 0)
2515 throw new ArithmeticException("BigInteger: modulus not positive");
2516
2517 BigInteger result = this.remainder(m);
2518 return (result.signum >= 0 ? result : result.add(m));
2519 }
2520
2521 /**
2522 * Returns a BigInteger whose value is
2523 * <tt>(this<sup>exponent</sup> mod m)</tt>. (Unlike {@code pow}, this
2524 * method permits negative exponents.)
2525 *
2526 * @param exponent the exponent.
2527 * @param m the modulus.
2528 * @return <tt>this<sup>exponent</sup> mod m</tt>
2529 * @throws ArithmeticException {@code m} ≤ 0 or the exponent is
2530 * negative and this BigInteger is not <i>relatively
2531 * prime</i> to {@code m}.
2532 * @see #modInverse
2533 */
2534 public BigInteger modPow(BigInteger exponent, BigInteger m) {
2535 if (m.signum <= 0)
2536 throw new ArithmeticException("BigInteger: modulus not positive");
2537
2538 // Trivial cases
2539 if (exponent.signum == 0)
2540 return (m.equals(ONE) ? ZERO : ONE);
2541
2542 if (this.equals(ONE))
2543 return (m.equals(ONE) ? ZERO : ONE);
2544
2545 if (this.equals(ZERO) && exponent.signum >= 0)
2546 return ZERO;
2547
2548 if (this.equals(negConst[1]) && (!exponent.testBit(0)))
2549 return (m.equals(ONE) ? ZERO : ONE);
2550
2551 boolean invertResult;
2552 if ((invertResult = (exponent.signum < 0)))
2553 exponent = exponent.negate();
2554
2555 BigInteger base = (this.signum < 0 || this.compareTo(m) >= 0
2556 ? this.mod(m) : this);
2557 BigInteger result;
2558 if (m.testBit(0)) { // odd modulus
2559 result = base.oddModPow(exponent, m);
2560 } else {
2561 /*
2562 * Even modulus. Tear it into an "odd part" (m1) and power of two
2563 * (m2), exponentiate mod m1, manually exponentiate mod m2, and
2564 * use Chinese Remainder Theorem to combine results.
2565 */
2566
2567 // Tear m apart into odd part (m1) and power of 2 (m2)
2568 int p = m.getLowestSetBit(); // Max pow of 2 that divides m
2569
2570 BigInteger m1 = m.shiftRight(p); // m/2**p
2571 BigInteger m2 = ONE.shiftLeft(p); // 2**p
2572
2573 // Calculate new base from m1
2574 BigInteger base2 = (this.signum < 0 || this.compareTo(m1) >= 0
2575 ? this.mod(m1) : this);
2576
2577 // Caculate (base ** exponent) mod m1.
2578 BigInteger a1 = (m1.equals(ONE) ? ZERO :
2579 base2.oddModPow(exponent, m1));
2580
2581 // Calculate (this ** exponent) mod m2
2582 BigInteger a2 = base.modPow2(exponent, p);
2583
2584 // Combine results using Chinese Remainder Theorem
2585 BigInteger y1 = m2.modInverse(m1);
2586 BigInteger y2 = m1.modInverse(m2);
2587
2588 if (m.mag.length < MAX_MAG_LENGTH / 2) {
2589 result = a1.multiply(m2).multiply(y1).add(a2.multiply(m1).multiply(y2)).mod(m);
2590 } else {
2591 MutableBigInteger t1 = new MutableBigInteger();
2592 new MutableBigInteger(a1.multiply(m2)).multiply(new MutableBigInteger(y1), t1);
2593 MutableBigInteger t2 = new MutableBigInteger();
2594 new MutableBigInteger(a2.multiply(m1)).multiply(new MutableBigInteger(y2), t2);
2595 t1.add(t2);
2596 MutableBigInteger q = new MutableBigInteger();
2597 result = t1.divide(new MutableBigInteger(m), q).toBigInteger();
2598 }
2599 }
2600
2601 return (invertResult ? result.modInverse(m) : result);
2602 }
2603
2604 static int[] bnExpModThreshTable = {7, 25, 81, 241, 673, 1793,
2605 Integer.MAX_VALUE}; // Sentinel
2606
2607 /**
2608 * Returns a BigInteger whose value is x to the power of y mod z.
2609 * Assumes: z is odd && x < z.
2610 */
2611 private BigInteger oddModPow(BigInteger y, BigInteger z) {
2612 /*
2613 * The algorithm is adapted from Colin Plumb's C library.
2614 *
2615 * The window algorithm:
2616 * The idea is to keep a running product of b1 = n^(high-order bits of exp)
2617 * and then keep appending exponent bits to it. The following patterns
2618 * apply to a 3-bit window (k = 3):
2619 * To append 0: square
2620 * To append 1: square, multiply by n^1
2621 * To append 10: square, multiply by n^1, square
2622 * To append 11: square, square, multiply by n^3
2623 * To append 100: square, multiply by n^1, square, square
2624 * To append 101: square, square, square, multiply by n^5
2625 * To append 110: square, square, multiply by n^3, square
2626 * To append 111: square, square, square, multiply by n^7
2627 *
2628 * Since each pattern involves only one multiply, the longer the pattern
2629 * the better, except that a 0 (no multiplies) can be appended directly.
2630 * We precompute a table of odd powers of n, up to 2^k, and can then
2631 * multiply k bits of exponent at a time. Actually, assuming random
2632 * exponents, there is on average one zero bit between needs to
2633 * multiply (1/2 of the time there's none, 1/4 of the time there's 1,
2634 * 1/8 of the time, there's 2, 1/32 of the time, there's 3, etc.), so
2635 * you have to do one multiply per k+1 bits of exponent.
2636 *
2637 * The loop walks down the exponent, squaring the result buffer as
2638 * it goes. There is a wbits+1 bit lookahead buffer, buf, that is
2639 * filled with the upcoming exponent bits. (What is read after the
2640 * end of the exponent is unimportant, but it is filled with zero here.)
2641 * When the most-significant bit of this buffer becomes set, i.e.
2642 * (buf & tblmask) != 0, we have to decide what pattern to multiply
2643 * by, and when to do it. We decide, remember to do it in future
2644 * after a suitable number of squarings have passed (e.g. a pattern
2645 * of "100" in the buffer requires that we multiply by n^1 immediately;
2646 * a pattern of "110" calls for multiplying by n^3 after one more
2647 * squaring), clear the buffer, and continue.
2648 *
2649 * When we start, there is one more optimization: the result buffer
2650 * is implcitly one, so squaring it or multiplying by it can be
2651 * optimized away. Further, if we start with a pattern like "100"
2652 * in the lookahead window, rather than placing n into the buffer
2653 * and then starting to square it, we have already computed n^2
2654 * to compute the odd-powers table, so we can place that into
2655 * the buffer and save a squaring.
2656 *
2657 * This means that if you have a k-bit window, to compute n^z,
2658 * where z is the high k bits of the exponent, 1/2 of the time
2659 * it requires no squarings. 1/4 of the time, it requires 1
2660 * squaring, ... 1/2^(k-1) of the time, it reqires k-2 squarings.
2661 * And the remaining 1/2^(k-1) of the time, the top k bits are a
2662 * 1 followed by k-1 0 bits, so it again only requires k-2
2663 * squarings, not k-1. The average of these is 1. Add that
2664 * to the one squaring we have to do to compute the table,
2665 * and you'll see that a k-bit window saves k-2 squarings
2666 * as well as reducing the multiplies. (It actually doesn't
2667 * hurt in the case k = 1, either.)
2668 */
2669 // Special case for exponent of one
2670 if (y.equals(ONE))
2671 return this;
2672
2673 // Special case for base of zero
2674 if (signum == 0)
2675 return ZERO;
2676
2677 int[] base = mag.clone();
2678 int[] exp = y.mag;
2679 int[] mod = z.mag;
2680 int modLen = mod.length;
2681
2682 // Select an appropriate window size
2683 int wbits = 0;
2684 int ebits = bitLength(exp, exp.length);
2685 // if exponent is 65537 (0x10001), use minimum window size
2686 if ((ebits != 17) || (exp[0] != 65537)) {
2687 while (ebits > bnExpModThreshTable[wbits]) {
2688 wbits++;
2689 }
2690 }
2691
2692 // Calculate appropriate table size
2693 int tblmask = 1 << wbits;
2694
2695 // Allocate table for precomputed odd powers of base in Montgomery form
2696 int[][] table = new int[tblmask][];
2697 for (int i=0; i < tblmask; i++)
2698 table[i] = new int[modLen];
2699
2700 // Compute the modular inverse
2701 int inv = -MutableBigInteger.inverseMod32(mod[modLen-1]);
2702
2703 // Convert base to Montgomery form
2704 int[] a = leftShift(base, base.length, modLen << 5);
2705
2706 MutableBigInteger q = new MutableBigInteger(),
2707 a2 = new MutableBigInteger(a),
2708 b2 = new MutableBigInteger(mod);
2709
2710 MutableBigInteger r= a2.divide(b2, q);
2711 table[0] = r.toIntArray();
2712
2713 // Pad table[0] with leading zeros so its length is at least modLen
2714 if (table[0].length < modLen) {
2715 int offset = modLen - table[0].length;
2716 int[] t2 = new int[modLen];
2717 for (int i=0; i < table[0].length; i++)
2718 t2[i+offset] = table[0][i];
2719 table[0] = t2;
2720 }
2721
2722 // Set b to the square of the base
2723 int[] b = squareToLen(table[0], modLen, null);
2724 b = montReduce(b, mod, modLen, inv);
2725
2726 // Set t to high half of b
2727 int[] t = Arrays.copyOf(b, modLen);
2728
2729 // Fill in the table with odd powers of the base
2730 for (int i=1; i < tblmask; i++) {
2731 int[] prod = multiplyToLen(t, modLen, table[i-1], modLen, null);
2732 table[i] = montReduce(prod, mod, modLen, inv);
2733 }
2734
2735 // Pre load the window that slides over the exponent
2736 int bitpos = 1 << ((ebits-1) & (32-1));
2737
2738 int buf = 0;
2739 int elen = exp.length;
2740 int eIndex = 0;
2741 for (int i = 0; i <= wbits; i++) {
2742 buf = (buf << 1) | (((exp[eIndex] & bitpos) != 0)?1:0);
2743 bitpos >>>= 1;
2744 if (bitpos == 0) {
2745 eIndex++;
2746 bitpos = 1 << (32-1);
2747 elen--;
2748 }
2749 }
2750
2751 int multpos = ebits;
2752
2753 // The first iteration, which is hoisted out of the main loop
2754 ebits--;
2755 boolean isone = true;
2756
2757 multpos = ebits - wbits;
2758 while ((buf & 1) == 0) {
2759 buf >>>= 1;
2760 multpos++;
2761 }
2762
2763 int[] mult = table[buf >>> 1];
2764
2765 buf = 0;
2766 if (multpos == ebits)
2767 isone = false;
2768
2769 // The main loop
2770 while (true) {
2771 ebits--;
2772 // Advance the window
2773 buf <<= 1;
2774
2775 if (elen != 0) {
2776 buf |= ((exp[eIndex] & bitpos) != 0) ? 1 : 0;
2777 bitpos >>>= 1;
2778 if (bitpos == 0) {
2779 eIndex++;
2780 bitpos = 1 << (32-1);
2781 elen--;
2782 }
2783 }
2784
2785 // Examine the window for pending multiplies
2786 if ((buf & tblmask) != 0) {
2787 multpos = ebits - wbits;
2788 while ((buf & 1) == 0) {
2789 buf >>>= 1;
2790 multpos++;
2791 }
2792 mult = table[buf >>> 1];
2793 buf = 0;
2794 }
2795
2796 // Perform multiply
2797 if (ebits == multpos) {
2798 if (isone) {
2799 b = mult.clone();
2800 isone = false;
2801 } else {
2802 t = b;
2803 a = multiplyToLen(t, modLen, mult, modLen, a);
2804 a = montReduce(a, mod, modLen, inv);
2805 t = a; a = b; b = t;
2806 }
2807 }
2808
2809 // Check if done
2810 if (ebits == 0)
2811 break;
2812
2813 // Square the input
2814 if (!isone) {
2815 t = b;
2816 a = squareToLen(t, modLen, a);
2817 a = montReduce(a, mod, modLen, inv);
2818 t = a; a = b; b = t;
2819 }
2820 }
2821
2822 // Convert result out of Montgomery form and return
2823 int[] t2 = new int[2*modLen];
2824 System.arraycopy(b, 0, t2, modLen, modLen);
2825
2826 b = montReduce(t2, mod, modLen, inv);
2827
2828 t2 = Arrays.copyOf(b, modLen);
2829
2830 return new BigInteger(1, t2);
2831 }
2832
2833 /**
2834 * Montgomery reduce n, modulo mod. This reduces modulo mod and divides
2835 * by 2^(32*mlen). Adapted from Colin Plumb's C library.
2836 */
2837 private static int[] montReduce(int[] n, int[] mod, int mlen, int inv) {
2838 int c=0;
2839 int len = mlen;
2840 int offset=0;
2841
2842 do {
2843 int nEnd = n[n.length-1-offset];
2844 int carry = mulAdd(n, mod, offset, mlen, inv * nEnd);
2845 c += addOne(n, offset, mlen, carry);
2846 offset++;
2847 } while (--len > 0);
2848
2849 while (c > 0)
2850 c += subN(n, mod, mlen);
2851
2852 while (intArrayCmpToLen(n, mod, mlen) >= 0)
2853 subN(n, mod, mlen);
2854
2855 return n;
2856 }
2857
2858
2859 /*
2860 * Returns -1, 0 or +1 as big-endian unsigned int array arg1 is less than,
2861 * equal to, or greater than arg2 up to length len.
2862 */
2863 private static int intArrayCmpToLen(int[] arg1, int[] arg2, int len) {
2864 for (int i=0; i < len; i++) {
2865 long b1 = arg1[i] & LONG_MASK;
2866 long b2 = arg2[i] & LONG_MASK;
2867 if (b1 < b2)
2868 return -1;
2869 if (b1 > b2)
2870 return 1;
2871 }
2872 return 0;
2873 }
2874
2875 /**
2876 * Subtracts two numbers of same length, returning borrow.
2877 */
2878 private static int subN(int[] a, int[] b, int len) {
2879 long sum = 0;
2880
2881 while (--len >= 0) {
2882 sum = (a[len] & LONG_MASK) -
2883 (b[len] & LONG_MASK) + (sum >> 32);
2884 a[len] = (int)sum;
2885 }
2886
2887 return (int)(sum >> 32);
2888 }
2889
2890 /**
2891 * Multiply an array by one word k and add to result, return the carry
2892 */
2893 static int mulAdd(int[] out, int[] in, int offset, int len, int k) {
2894 implMulAddCheck(out, in, offset, len, k);
2895 return implMulAdd(out, in, offset, len, k);
2896 }
2897
2898 /**
2899 * Parameters validation.
2900 */
2901 private static void implMulAddCheck(int[] out, int[] in, int offset, int len, int k) {
2902 if (len > in.length) {
2903 throw new IllegalArgumentException("input length is out of bound: " + len + " > " + in.length);
2904 }
2905 if (offset < 0) {
2906 throw new IllegalArgumentException("input offset is invalid: " + offset);
2907 }
2908 if (offset > (out.length - 1)) {
2909 throw new IllegalArgumentException("input offset is out of bound: " + offset + " > " + (out.length - 1));
2910 }
2911 if (len > (out.length - offset)) {
2912 throw new IllegalArgumentException("input len is out of bound: " + len + " > " + (out.length - offset));
2913 }
2914 }
2915
2916 /**
2917 * Java Runtime may use intrinsic for this method.
2918 */
2919 private static int implMulAdd(int[] out, int[] in, int offset, int len, int k) {
2920 long kLong = k & LONG_MASK;
2921 long carry = 0;
2922
2923 offset = out.length-offset - 1;
2924 for (int j=len-1; j >= 0; j--) {
2925 long product = (in[j] & LONG_MASK) * kLong +
2926 (out[offset] & LONG_MASK) + carry;
2927 out[offset--] = (int)product;
2928 carry = product >>> 32;
2929 }
2930 return (int)carry;
2931 }
2932
2933 /**
2934 * Add one word to the number a mlen words into a. Return the resulting
2935 * carry.
2936 */
2937 static int addOne(int[] a, int offset, int mlen, int carry) {
2938 offset = a.length-1-mlen-offset;
2939 long t = (a[offset] & LONG_MASK) + (carry & LONG_MASK);
2940
2941 a[offset] = (int)t;
2942 if ((t >>> 32) == 0)
2943 return 0;
2944 while (--mlen >= 0) {
2945 if (--offset < 0) { // Carry out of number
2946 return 1;
2947 } else {
2948 a[offset]++;
2949 if (a[offset] != 0)
2950 return 0;
2951 }
2952 }
2953 return 1;
2954 }
2955
2956 /**
2957 * Returns a BigInteger whose value is (this ** exponent) mod (2**p)
2958 */
2959 private BigInteger modPow2(BigInteger exponent, int p) {
2960 /*
2961 * Perform exponentiation using repeated squaring trick, chopping off
2962 * high order bits as indicated by modulus.
2963 */
2964 BigInteger result = ONE;
2965 BigInteger baseToPow2 = this.mod2(p);
2966 int expOffset = 0;
2967
2968 int limit = exponent.bitLength();
2969
2970 if (this.testBit(0))
2971 limit = (p-1) < limit ? (p-1) : limit;
2972
2973 while (expOffset < limit) {
2974 if (exponent.testBit(expOffset))
2975 result = result.multiply(baseToPow2).mod2(p);
2976 expOffset++;
2977 if (expOffset < limit)
2978 baseToPow2 = baseToPow2.square().mod2(p);
2979 }
2980
2981 return result;
2982 }
2983
2984 /**
2985 * Returns a BigInteger whose value is this mod(2**p).
2986 * Assumes that this {@code BigInteger >= 0} and {@code p > 0}.
2987 */
2988 private BigInteger mod2(int p) {
2989 if (bitLength() <= p)
2990 return this;
2991
2992 // Copy remaining ints of mag
2993 int numInts = (p + 31) >>> 5;
2994 int[] mag = new int[numInts];
2995 System.arraycopy(this.mag, (this.mag.length - numInts), mag, 0, numInts);
2996
2997 // Mask out any excess bits
2998 int excessBits = (numInts << 5) - p;
2999 mag[0] &= (1L << (32-excessBits)) - 1;
3000
3001 return (mag[0] == 0 ? new BigInteger(1, mag) : new BigInteger(mag, 1));
3002 }
3003
3004 /**
3005 * Returns a BigInteger whose value is {@code (this}<sup>-1</sup> {@code mod m)}.
3006 *
3007 * @param m the modulus.
3008 * @return {@code this}<sup>-1</sup> {@code mod m}.
3009 * @throws ArithmeticException {@code m} ≤ 0, or this BigInteger
3010 * has no multiplicative inverse mod m (that is, this BigInteger
3011 * is not <i>relatively prime</i> to m).
3012 */
3013 public BigInteger modInverse(BigInteger m) {
3014 if (m.signum != 1)
3015 throw new ArithmeticException("BigInteger: modulus not positive");
3016
3017 if (m.equals(ONE))
3018 return ZERO;
3019
3020 // Calculate (this mod m)
3021 BigInteger modVal = this;
3022 if (signum < 0 || (this.compareMagnitude(m) >= 0))
3023 modVal = this.mod(m);
3024
3025 if (modVal.equals(ONE))
3026 return ONE;
3027
3028 MutableBigInteger a = new MutableBigInteger(modVal);
3029 MutableBigInteger b = new MutableBigInteger(m);
3030
3031 MutableBigInteger result = a.mutableModInverse(b);
3032 return result.toBigInteger(1);
3033 }
3034
3035 // Shift Operations
3036
3037 /**
3038 * Returns a BigInteger whose value is {@code (this << n)}.
3039 * The shift distance, {@code n}, may be negative, in which case
3040 * this method performs a right shift.
3041 * (Computes <tt>floor(this * 2<sup>n</sup>)</tt>.)
3042 *
3043 * @param n shift distance, in bits.
3044 * @return {@code this << n}
3045 * @see #shiftRight
3046 */
3047 public BigInteger shiftLeft(int n) {
3048 if (signum == 0)
3049 return ZERO;
3050 if (n > 0) {
3051 return new BigInteger(shiftLeft(mag, n), signum);
3052 } else if (n == 0) {
3053 return this;
3054 } else {
3055 // Possible int overflow in (-n) is not a trouble,
3056 // because shiftRightImpl considers its argument unsigned
3057 return shiftRightImpl(-n);
3058 }
3059 }
3060
3061 /**
3062 * Returns a magnitude array whose value is {@code (mag << n)}.
3063 * The shift distance, {@code n}, is considered unnsigned.
3064 * (Computes <tt>this * 2<sup>n</sup></tt>.)
3065 *
3066 * @param mag magnitude, the most-significant int ({@code mag[0]}) must be non-zero.
3067 * @param n unsigned shift distance, in bits.
3068 * @return {@code mag << n}
3069 */
3070 private static int[] shiftLeft(int[] mag, int n) {
3071 int nInts = n >>> 5;
3072 int nBits = n & 0x1f;
3073 int magLen = mag.length;
3074 int newMag[] = null;
3075
3076 if (nBits == 0) {
3077 newMag = new int[magLen + nInts];
3078 System.arraycopy(mag, 0, newMag, 0, magLen);
3079 } else {
3080 int i = 0;
3081 int nBits2 = 32 - nBits;
3082 int highBits = mag[0] >>> nBits2;
3083 if (highBits != 0) {
3084 newMag = new int[magLen + nInts + 1];
3085 newMag[i++] = highBits;
3086 } else {
3087 newMag = new int[magLen + nInts];
3088 }
3089 int j=0;
3090 while (j < magLen-1)
3091 newMag[i++] = mag[j++] << nBits | mag[j] >>> nBits2;
3092 newMag[i] = mag[j] << nBits;
3093 }
3094 return newMag;
3095 }
3096
3097 /**
3098 * Returns a BigInteger whose value is {@code (this >> n)}. Sign
3099 * extension is performed. The shift distance, {@code n}, may be
3100 * negative, in which case this method performs a left shift.
3101 * (Computes <tt>floor(this / 2<sup>n</sup>)</tt>.)
3102 *
3103 * @param n shift distance, in bits.
3104 * @return {@code this >> n}
3105 * @see #shiftLeft
3106 */
3107 public BigInteger shiftRight(int n) {
3108 if (signum == 0)
3109 return ZERO;
3110 if (n > 0) {
3111 return shiftRightImpl(n);
3112 } else if (n == 0) {
3113 return this;
3114 } else {
3115 // Possible int overflow in {@code -n} is not a trouble,
3116 // because shiftLeft considers its argument unsigned
3117 return new BigInteger(shiftLeft(mag, -n), signum);
3118 }
3119 }
3120
3121 /**
3122 * Returns a BigInteger whose value is {@code (this >> n)}. The shift
3123 * distance, {@code n}, is considered unsigned.
3124 * (Computes <tt>floor(this * 2<sup>-n</sup>)</tt>.)
3125 *
3126 * @param n unsigned shift distance, in bits.
3127 * @return {@code this >> n}
3128 */
3129 private BigInteger shiftRightImpl(int n) {
3130 int nInts = n >>> 5;
3131 int nBits = n & 0x1f;
3132 int magLen = mag.length;
3133 int newMag[] = null;
3134
3135 // Special case: entire contents shifted off the end
3136 if (nInts >= magLen)
3137 return (signum >= 0 ? ZERO : negConst[1]);
3138
3139 if (nBits == 0) {
3140 int newMagLen = magLen - nInts;
3141 newMag = Arrays.copyOf(mag, newMagLen);
3142 } else {
3143 int i = 0;
3144 int highBits = mag[0] >>> nBits;
3145 if (highBits != 0) {
3146 newMag = new int[magLen - nInts];
3147 newMag[i++] = highBits;
3148 } else {
3149 newMag = new int[magLen - nInts -1];
3150 }
3151
3152 int nBits2 = 32 - nBits;
3153 int j=0;
3154 while (j < magLen - nInts - 1)
3155 newMag[i++] = (mag[j++] << nBits2) | (mag[j] >>> nBits);
3156 }
3157
3158 if (signum < 0) {
3159 // Find out whether any one-bits were shifted off the end.
3160 boolean onesLost = false;
3161 for (int i=magLen-1, j=magLen-nInts; i >= j && !onesLost; i--)
3162 onesLost = (mag[i] != 0);
3163 if (!onesLost && nBits != 0)
3164 onesLost = (mag[magLen - nInts - 1] << (32 - nBits) != 0);
3165
3166 if (onesLost)
3167 newMag = javaIncrement(newMag);
3168 }
3169
3170 return new BigInteger(newMag, signum);
3171 }
3172
3173 int[] javaIncrement(int[] val) {
3174 int lastSum = 0;
3175 for (int i=val.length-1; i >= 0 && lastSum == 0; i--)
3176 lastSum = (val[i] += 1);
3177 if (lastSum == 0) {
3178 val = new int[val.length+1];
3179 val[0] = 1;
3180 }
3181 return val;
3182 }
3183
3184 // Bitwise Operations
3185
3186 /**
3187 * Returns a BigInteger whose value is {@code (this & val)}. (This
3188 * method returns a negative BigInteger if and only if this and val are
3189 * both negative.)
3190 *
3191 * @param val value to be AND'ed with this BigInteger.
3192 * @return {@code this & val}
3193 */
3194 public BigInteger and(BigInteger val) {
3195 int[] result = new int[Math.max(intLength(), val.intLength())];
3196 for (int i=0; i < result.length; i++)
3197 result[i] = (getInt(result.length-i-1)
3198 & val.getInt(result.length-i-1));
3199
3200 return valueOf(result);
3201 }
3202
3203 /**
3204 * Returns a BigInteger whose value is {@code (this | val)}. (This method
3205 * returns a negative BigInteger if and only if either this or val is
3206 * negative.)
3207 *
3208 * @param val value to be OR'ed with this BigInteger.
3209 * @return {@code this | val}
3210 */
3211 public BigInteger or(BigInteger val) {
3212 int[] result = new int[Math.max(intLength(), val.intLength())];
3213 for (int i=0; i < result.length; i++)
3214 result[i] = (getInt(result.length-i-1)
3215 | val.getInt(result.length-i-1));
3216
3217 return valueOf(result);
3218 }
3219
3220 /**
3221 * Returns a BigInteger whose value is {@code (this ^ val)}. (This method
3222 * returns a negative BigInteger if and only if exactly one of this and
3223 * val are negative.)
3224 *
3225 * @param val value to be XOR'ed with this BigInteger.
3226 * @return {@code this ^ val}
3227 */
3228 public BigInteger xor(BigInteger val) {
3229 int[] result = new int[Math.max(intLength(), val.intLength())];
3230 for (int i=0; i < result.length; i++)
3231 result[i] = (getInt(result.length-i-1)
3232 ^ val.getInt(result.length-i-1));
3233
3234 return valueOf(result);
3235 }
3236
3237 /**
3238 * Returns a BigInteger whose value is {@code (~this)}. (This method
3239 * returns a negative value if and only if this BigInteger is
3240 * non-negative.)
3241 *
3242 * @return {@code ~this}
3243 */
3244 public BigInteger not() {
3245 int[] result = new int[intLength()];
3246 for (int i=0; i < result.length; i++)
3247 result[i] = ~getInt(result.length-i-1);
3248
3249 return valueOf(result);
3250 }
3251
3252 /**
3253 * Returns a BigInteger whose value is {@code (this & ~val)}. This
3254 * method, which is equivalent to {@code and(val.not())}, is provided as
3255 * a convenience for masking operations. (This method returns a negative
3256 * BigInteger if and only if {@code this} is negative and {@code val} is
3257 * positive.)
3258 *
3259 * @param val value to be complemented and AND'ed with this BigInteger.
3260 * @return {@code this & ~val}
3261 */
3262 public BigInteger andNot(BigInteger val) {
3263 int[] result = new int[Math.max(intLength(), val.intLength())];
3264 for (int i=0; i < result.length; i++)
3265 result[i] = (getInt(result.length-i-1)
3266 & ~val.getInt(result.length-i-1));
3267
3268 return valueOf(result);
3269 }
3270
3271
3272 // Single Bit Operations
3273
3274 /**
3275 * Returns {@code true} if and only if the designated bit is set.
3276 * (Computes {@code ((this & (1<<n)) != 0)}.)
3277 *
3278 * @param n index of bit to test.
3279 * @return {@code true} if and only if the designated bit is set.
3280 * @throws ArithmeticException {@code n} is negative.
3281 */
3282 public boolean testBit(int n) {
3283 if (n < 0)
3284 throw new ArithmeticException("Negative bit address");
3285
3286 return (getInt(n >>> 5) & (1 << (n & 31))) != 0;
3287 }
3288
3289 /**
3290 * Returns a BigInteger whose value is equivalent to this BigInteger
3291 * with the designated bit set. (Computes {@code (this | (1<<n))}.)
3292 *
3293 * @param n index of bit to set.
3294 * @return {@code this | (1<<n)}
3295 * @throws ArithmeticException {@code n} is negative.
3296 */
3297 public BigInteger setBit(int n) {
3298 if (n < 0)
3299 throw new ArithmeticException("Negative bit address");
3300
3301 int intNum = n >>> 5;
3302 int[] result = new int[Math.max(intLength(), intNum+2)];
3303
3304 for (int i=0; i < result.length; i++)
3305 result[result.length-i-1] = getInt(i);
3306
3307 result[result.length-intNum-1] |= (1 << (n & 31));
3308
3309 return valueOf(result);
3310 }
3311
3312 /**
3313 * Returns a BigInteger whose value is equivalent to this BigInteger
3314 * with the designated bit cleared.
3315 * (Computes {@code (this & ~(1<<n))}.)
3316 *
3317 * @param n index of bit to clear.
3318 * @return {@code this & ~(1<<n)}
3319 * @throws ArithmeticException {@code n} is negative.
3320 */
3321 public BigInteger clearBit(int n) {
3322 if (n < 0)
3323 throw new ArithmeticException("Negative bit address");
3324
3325 int intNum = n >>> 5;
3326 int[] result = new int[Math.max(intLength(), ((n + 1) >>> 5) + 1)];
3327
3328 for (int i=0; i < result.length; i++)
3329 result[result.length-i-1] = getInt(i);
3330
3331 result[result.length-intNum-1] &= ~(1 << (n & 31));
3332
3333 return valueOf(result);
3334 }
3335
3336 /**
3337 * Returns a BigInteger whose value is equivalent to this BigInteger
3338 * with the designated bit flipped.
3339 * (Computes {@code (this ^ (1<<n))}.)
3340 *
3341 * @param n index of bit to flip.
3342 * @return {@code this ^ (1<<n)}
3343 * @throws ArithmeticException {@code n} is negative.
3344 */
3345 public BigInteger flipBit(int n) {
3346 if (n < 0)
3347 throw new ArithmeticException("Negative bit address");
3348
3349 int intNum = n >>> 5;
3350 int[] result = new int[Math.max(intLength(), intNum+2)];
3351
3352 for (int i=0; i < result.length; i++)
3353 result[result.length-i-1] = getInt(i);
3354
3355 result[result.length-intNum-1] ^= (1 << (n & 31));
3356
3357 return valueOf(result);
3358 }
3359
3360 /**
3361 * Returns the index of the rightmost (lowest-order) one bit in this
3362 * BigInteger (the number of zero bits to the right of the rightmost
3363 * one bit). Returns -1 if this BigInteger contains no one bits.
3364 * (Computes {@code (this == 0? -1 : log2(this & -this))}.)
3365 *
3366 * @return index of the rightmost one bit in this BigInteger.
3367 */
3368 public int getLowestSetBit() {
3369 int lsb = lowestSetBitPlusTwo - 2;
3370 if (lsb == -2) { // lowestSetBit not initialized yet
3371 lsb = 0;
3372 if (signum == 0) {
3373 lsb -= 1;
3374 } else {
3375 // Search for lowest order nonzero int
3376 int i,b;
3377 for (i=0; (b = getInt(i)) == 0; i++)
3378 ;
3379 lsb += (i << 5) + Integer.numberOfTrailingZeros(b);
3380 }
3381 lowestSetBitPlusTwo = lsb + 2;
3382 }
3383 return lsb;
3384 }
3385
3386
3387 // Miscellaneous Bit Operations
3388
3389 /**
3390 * Returns the number of bits in the minimal two's-complement
3391 * representation of this BigInteger, <i>excluding</i> a sign bit.
3392 * For positive BigIntegers, this is equivalent to the number of bits in
3393 * the ordinary binary representation. (Computes
3394 * {@code (ceil(log2(this < 0 ? -this : this+1)))}.)
3395 *
3396 * @return number of bits in the minimal two's-complement
3397 * representation of this BigInteger, <i>excluding</i> a sign bit.
3398 */
3399 public int bitLength() {
3400 int n = bitLengthPlusOne - 1;
3401 if (n == -1) { // bitLength not initialized yet
3402 int[] m = mag;
3403 int len = m.length;
3404 if (len == 0) {
3405 n = 0; // offset by one to initialize
3406 } else {
3407 // Calculate the bit length of the magnitude
3408 int magBitLength = ((len - 1) << 5) + bitLengthForInt(mag[0]);
3409 if (signum < 0) {
3410 // Check if magnitude is a power of two
3411 boolean pow2 = (Integer.bitCount(mag[0]) == 1);
3412 for (int i=1; i< len && pow2; i++)
3413 pow2 = (mag[i] == 0);
3414
3415 n = (pow2 ? magBitLength -1 : magBitLength);
3416 } else {
3417 n = magBitLength;
3418 }
3419 }
3420 bitLengthPlusOne = n + 1;
3421 }
3422 return n;
3423 }
3424
3425 /**
3426 * Returns the number of bits in the two's complement representation
3427 * of this BigInteger that differ from its sign bit. This method is
3428 * useful when implementing bit-vector style sets atop BigIntegers.
3429 *
3430 * @return number of bits in the two's complement representation
3431 * of this BigInteger that differ from its sign bit.
3432 */
3433 public int bitCount() {
3434 int bc = bitCountPlusOne - 1;
3435 if (bc == -1) { // bitCount not initialized yet
3436 bc = 0; // offset by one to initialize
3437 // Count the bits in the magnitude
3438 for (int i=0; i < mag.length; i++)
3439 bc += Integer.bitCount(mag[i]);
3440 if (signum < 0) {
3441 // Count the trailing zeros in the magnitude
3442 int magTrailingZeroCount = 0, j;
3443 for (j=mag.length-1; mag[j] == 0; j--)
3444 magTrailingZeroCount += 32;
3445 magTrailingZeroCount += Integer.numberOfTrailingZeros(mag[j]);
3446 bc += magTrailingZeroCount - 1;
3447 }
3448 bitCountPlusOne = bc + 1;
3449 }
3450 return bc;
3451 }
3452
3453 // Primality Testing
3454
3455 /**
3456 * Returns {@code true} if this BigInteger is probably prime,
3457 * {@code false} if it's definitely composite. If
3458 * {@code certainty} is ≤ 0, {@code true} is
3459 * returned.
3460 *
3461 * @param certainty a measure of the uncertainty that the caller is
3462 * willing to tolerate: if the call returns {@code true}
3463 * the probability that this BigInteger is prime exceeds
3464 * (1 - 1/2<sup>{@code certainty}</sup>). The execution time of
3465 * this method is proportional to the value of this parameter.
3466 * @return {@code true} if this BigInteger is probably prime,
3467 * {@code false} if it's definitely composite.
3468 */
3469 public boolean isProbablePrime(int certainty) {
3470 if (certainty <= 0)
3471 return true;
3472 BigInteger w = this.abs();
3473 if (w.equals(TWO))
3474 return true;
3475 if (!w.testBit(0) || w.equals(ONE))
3476 return false;
3477
3478 return w.primeToCertainty(certainty, null);
3479 }
3480
3481 // Comparison Operations
3482
3483 /**
3484 * Compares this BigInteger with the specified BigInteger. This
3485 * method is provided in preference to individual methods for each
3486 * of the six boolean comparison operators ({@literal <}, ==,
3487 * {@literal >}, {@literal >=}, !=, {@literal <=}). The suggested
3488 * idiom for performing these comparisons is: {@code
3489 * (x.compareTo(y)} <<i>op</i>> {@code 0)}, where
3490 * <<i>op</i>> is one of the six comparison operators.
3491 *
3492 * @param val BigInteger to which this BigInteger is to be compared.
3493 * @return -1, 0 or 1 as this BigInteger is numerically less than, equal
3494 * to, or greater than {@code val}.
3495 */
3496 public int compareTo(BigInteger val) {
3497 if (signum == val.signum) {
3498 switch (signum) {
3499 case 1:
3500 return compareMagnitude(val);
3501 case -1:
3502 return val.compareMagnitude(this);
3503 default:
3504 return 0;
3505 }
3506 }
3507 return signum > val.signum ? 1 : -1;
3508 }
3509
3510 /**
3511 * Compares the magnitude array of this BigInteger with the specified
3512 * BigInteger's. This is the version of compareTo ignoring sign.
3513 *
3514 * @param val BigInteger whose magnitude array to be compared.
3515 * @return -1, 0 or 1 as this magnitude array is less than, equal to or
3516 * greater than the magnitude aray for the specified BigInteger's.
3517 */
3518 final int compareMagnitude(BigInteger val) {
3519 int[] m1 = mag;
3520 int len1 = m1.length;
3521 int[] m2 = val.mag;
3522 int len2 = m2.length;
3523 if (len1 < len2)
3524 return -1;
3525 if (len1 > len2)
3526 return 1;
3527 for (int i = 0; i < len1; i++) {
3528 int a = m1[i];
3529 int b = m2[i];
3530 if (a != b)
3531 return ((a & LONG_MASK) < (b & LONG_MASK)) ? -1 : 1;
3532 }
3533 return 0;
3534 }
3535
3536 /**
3537 * Version of compareMagnitude that compares magnitude with long value.
3538 * val can't be Long.MIN_VALUE.
3539 */
3540 final int compareMagnitude(long val) {
3541 assert val != Long.MIN_VALUE;
3542 int[] m1 = mag;
3543 int len = m1.length;
3544 if (len > 2) {
3545 return 1;
3546 }
3547 if (val < 0) {
3548 val = -val;
3549 }
3550 int highWord = (int)(val >>> 32);
3551 if (highWord == 0) {
3552 if (len < 1)
3553 return -1;
3554 if (len > 1)
3555 return 1;
3556 int a = m1[0];
3557 int b = (int)val;
3558 if (a != b) {
3559 return ((a & LONG_MASK) < (b & LONG_MASK))? -1 : 1;
3560 }
3561 return 0;
3562 } else {
3563 if (len < 2)
3564 return -1;
3565 int a = m1[0];
3566 int b = highWord;
3567 if (a != b) {
3568 return ((a & LONG_MASK) < (b & LONG_MASK))? -1 : 1;
3569 }
3570 a = m1[1];
3571 b = (int)val;
3572 if (a != b) {
3573 return ((a & LONG_MASK) < (b & LONG_MASK))? -1 : 1;
3574 }
3575 return 0;
3576 }
3577 }
3578
3579 /**
3580 * Compares this BigInteger with the specified Object for equality.
3581 *
3582 * @param x Object to which this BigInteger is to be compared.
3583 * @return {@code true} if and only if the specified Object is a
3584 * BigInteger whose value is numerically equal to this BigInteger.
3585 */
3586 public boolean equals(Object x) {
3587 // This test is just an optimization, which may or may not help
3588 if (x == this)
3589 return true;
3590
3591 if (!(x instanceof BigInteger))
3592 return false;
3593
3594 BigInteger xInt = (BigInteger) x;
3595 if (xInt.signum != signum)
3596 return false;
3597
3598 int[] m = mag;
3599 int len = m.length;
3600 int[] xm = xInt.mag;
3601 if (len != xm.length)
3602 return false;
3603
3604 for (int i = 0; i < len; i++)
3605 if (xm[i] != m[i])
3606 return false;
3607
3608 return true;
3609 }
3610
3611 /**
3612 * Returns the minimum of this BigInteger and {@code val}.
3613 *
3614 * @param val value with which the minimum is to be computed.
3615 * @return the BigInteger whose value is the lesser of this BigInteger and
3616 * {@code val}. If they are equal, either may be returned.
3617 */
3618 public BigInteger min(BigInteger val) {
3619 return (compareTo(val) < 0 ? this : val);
3620 }
3621
3622 /**
3623 * Returns the maximum of this BigInteger and {@code val}.
3624 *
3625 * @param val value with which the maximum is to be computed.
3626 * @return the BigInteger whose value is the greater of this and
3627 * {@code val}. If they are equal, either may be returned.
3628 */
3629 public BigInteger max(BigInteger val) {
3630 return (compareTo(val) > 0 ? this : val);
3631 }
3632
3633
3634 // Hash Function
3635
3636 /**
3637 * Returns the hash code for this BigInteger.
3638 *
3639 * @return hash code for this BigInteger.
3640 */
3641 public int hashCode() {
3642 int hashCode = 0;
3643
3644 for (int i=0; i < mag.length; i++)
3645 hashCode = (int)(31*hashCode + (mag[i] & LONG_MASK));
3646
3647 return hashCode * signum;
3648 }
3649
3650 /**
3651 * Returns the String representation of this BigInteger in the
3652 * given radix. If the radix is outside the range from {@link
3653 * Character#MIN_RADIX} to {@link Character#MAX_RADIX} inclusive,
3654 * it will default to 10 (as is the case for
3655 * {@code Integer.toString}). The digit-to-character mapping
3656 * provided by {@code Character.forDigit} is used, and a minus
3657 * sign is prepended if appropriate. (This representation is
3658 * compatible with the {@link #BigInteger(String, int) (String,
3659 * int)} constructor.)
3660 *
3661 * @param radix radix of the String representation.
3662 * @return String representation of this BigInteger in the given radix.
3663 * @see Integer#toString
3664 * @see Character#forDigit
3665 * @see #BigInteger(java.lang.String, int)
3666 */
3667 public String toString(int radix) {
3668 if (signum == 0)
3669 return "0";
3670 if (radix < Character.MIN_RADIX || radix > Character.MAX_RADIX)
3671 radix = 10;
3672
3673 // If it's small enough, use smallToString.
3674 if (mag.length <= SCHOENHAGE_BASE_CONVERSION_THRESHOLD)
3675 return smallToString(radix);
3676
3677 // Otherwise use recursive toString, which requires positive arguments.
3678 // The results will be concatenated into this StringBuilder
3679 StringBuilder sb = new StringBuilder();
3680 if (signum < 0) {
3681 toString(this.negate(), sb, radix, 0);
3682 sb.insert(0, '-');
3683 }
3684 else
3685 toString(this, sb, radix, 0);
3686
3687 return sb.toString();
3688 }
3689
3690 /** This method is used to perform toString when arguments are small. */
3691 private String smallToString(int radix) {
3692 if (signum == 0) {
3693 return "0";
3694 }
3695
3696 // Compute upper bound on number of digit groups and allocate space
3697 int maxNumDigitGroups = (4*mag.length + 6)/7;
3698 String digitGroup[] = new String[maxNumDigitGroups];
3699
3700 // Translate number to string, a digit group at a time
3701 BigInteger tmp = this.abs();
3702 int numGroups = 0;
3703 while (tmp.signum != 0) {
3704 BigInteger d = longRadix[radix];
3705
3706 MutableBigInteger q = new MutableBigInteger(),
3707 a = new MutableBigInteger(tmp.mag),
3708 b = new MutableBigInteger(d.mag);
3709 MutableBigInteger r = a.divide(b, q);
3710 BigInteger q2 = q.toBigInteger(tmp.signum * d.signum);
3711 BigInteger r2 = r.toBigInteger(tmp.signum * d.signum);
3712
3713 digitGroup[numGroups++] = Long.toString(r2.longValue(), radix);
3714 tmp = q2;
3715 }
3716
3717 // Put sign (if any) and first digit group into result buffer
3718 StringBuilder buf = new StringBuilder(numGroups*digitsPerLong[radix]+1);
3719 if (signum < 0) {
3720 buf.append('-');
3721 }
3722 buf.append(digitGroup[numGroups-1]);
3723
3724 // Append remaining digit groups padded with leading zeros
3725 for (int i=numGroups-2; i >= 0; i--) {
3726 // Prepend (any) leading zeros for this digit group
3727 int numLeadingZeros = digitsPerLong[radix]-digitGroup[i].length();
3728 if (numLeadingZeros != 0) {
3729 buf.append(zeros[numLeadingZeros]);
3730 }
3731 buf.append(digitGroup[i]);
3732 }
3733 return buf.toString();
3734 }
3735
3736 /**
3737 * Converts the specified BigInteger to a string and appends to
3738 * {@code sb}. This implements the recursive Schoenhage algorithm
3739 * for base conversions.
3740 * <p>
3741 * See Knuth, Donald, _The Art of Computer Programming_, Vol. 2,
3742 * Answers to Exercises (4.4) Question 14.
3743 *
3744 * @param u The number to convert to a string.
3745 * @param sb The StringBuilder that will be appended to in place.
3746 * @param radix The base to convert to.
3747 * @param digits The minimum number of digits to pad to.
3748 */
3749 private static void toString(BigInteger u, StringBuilder sb, int radix,
3750 int digits) {
3751 // If we're smaller than a certain threshold, use the smallToString
3752 // method, padding with leading zeroes when necessary.
3753 if (u.mag.length <= SCHOENHAGE_BASE_CONVERSION_THRESHOLD) {
3754 String s = u.smallToString(radix);
3755
3756 // Pad with internal zeros if necessary.
3757 // Don't pad if we're at the beginning of the string.
3758 if ((s.length() < digits) && (sb.length() > 0)) {
3759 for (int i=s.length(); i < digits; i++) {
3760 sb.append('0');
3761 }
3762 }
3763
3764 sb.append(s);
3765 return;
3766 }
3767
3768 int b, n;
3769 b = u.bitLength();
3770
3771 // Calculate a value for n in the equation radix^(2^n) = u
3772 // and subtract 1 from that value. This is used to find the
3773 // cache index that contains the best value to divide u.
3774 n = (int) Math.round(Math.log(b * LOG_TWO / logCache[radix]) / LOG_TWO - 1.0);
3775 BigInteger v = getRadixConversionCache(radix, n);
3776 BigInteger[] results;
3777 results = u.divideAndRemainder(v);
3778
3779 int expectedDigits = 1 << n;
3780
3781 // Now recursively build the two halves of each number.
3782 toString(results[0], sb, radix, digits-expectedDigits);
3783 toString(results[1], sb, radix, expectedDigits);
3784 }
3785
3786 /**
3787 * Returns the value radix^(2^exponent) from the cache.
3788 * If this value doesn't already exist in the cache, it is added.
3789 * <p>
3790 * This could be changed to a more complicated caching method using
3791 * {@code Future}.
3792 */
3793 private static BigInteger getRadixConversionCache(int radix, int exponent) {
3794 BigInteger[] cacheLine = powerCache[radix]; // volatile read
3795 if (exponent < cacheLine.length) {
3796 return cacheLine[exponent];
3797 }
3798
3799 int oldLength = cacheLine.length;
3800 cacheLine = Arrays.copyOf(cacheLine, exponent + 1);
3801 for (int i = oldLength; i <= exponent; i++) {
3802 cacheLine[i] = cacheLine[i - 1].pow(2);
3803 }
3804
3805 BigInteger[][] pc = powerCache; // volatile read again
3806 if (exponent >= pc[radix].length) {
3807 pc = pc.clone();
3808 pc[radix] = cacheLine;
3809 powerCache = pc; // volatile write, publish
3810 }
3811 return cacheLine[exponent];
3812 }
3813
3814 /* zero[i] is a string of i consecutive zeros. */
3815 private static String zeros[] = new String[64];
3816 static {
3817 zeros[63] =
3818 "000000000000000000000000000000000000000000000000000000000000000";
3819 for (int i=0; i < 63; i++)
3820 zeros[i] = zeros[63].substring(0, i);
3821 }
3822
3823 /**
3824 * Returns the decimal String representation of this BigInteger.
3825 * The digit-to-character mapping provided by
3826 * {@code Character.forDigit} is used, and a minus sign is
3827 * prepended if appropriate. (This representation is compatible
3828 * with the {@link #BigInteger(String) (String)} constructor, and
3829 * allows for String concatenation with Java's + operator.)
3830 *
3831 * @return decimal String representation of this BigInteger.
3832 * @see Character#forDigit
3833 * @see #BigInteger(java.lang.String)
3834 */
3835 public String toString() {
3836 return toString(10);
3837 }
3838
3839 /**
3840 * Returns a byte array containing the two's-complement
3841 * representation of this BigInteger. The byte array will be in
3842 * <i>big-endian</i> byte-order: the most significant byte is in
3843 * the zeroth element. The array will contain the minimum number
3844 * of bytes required to represent this BigInteger, including at
3845 * least one sign bit, which is {@code (ceil((this.bitLength() +
3846 * 1)/8))}. (This representation is compatible with the
3847 * {@link #BigInteger(byte[]) (byte[])} constructor.)
3848 *
3849 * @return a byte array containing the two's-complement representation of
3850 * this BigInteger.
3851 * @see #BigInteger(byte[])
3852 */
3853 public byte[] toByteArray() {
3854 int byteLen = bitLength()/8 + 1;
3855 byte[] byteArray = new byte[byteLen];
3856
3857 for (int i=byteLen-1, bytesCopied=4, nextInt=0, intIndex=0; i >= 0; i--) {
3858 if (bytesCopied == 4) {
3859 nextInt = getInt(intIndex++);
3860 bytesCopied = 1;
3861 } else {
3862 nextInt >>>= 8;
3863 bytesCopied++;
3864 }
3865 byteArray[i] = (byte)nextInt;
3866 }
3867 return byteArray;
3868 }
3869
3870 /**
3871 * Converts this BigInteger to an {@code int}. This
3872 * conversion is analogous to a
3873 * <i>narrowing primitive conversion</i> from {@code long} to
3874 * {@code int} as defined in section 5.1.3 of
3875 * <cite>The Java™ Language Specification</cite>:
3876 * if this BigInteger is too big to fit in an
3877 * {@code int}, only the low-order 32 bits are returned.
3878 * Note that this conversion can lose information about the
3879 * overall magnitude of the BigInteger value as well as return a
3880 * result with the opposite sign.
3881 *
3882 * @return this BigInteger converted to an {@code int}.
3883 * @see #intValueExact()
3884 */
3885 public int intValue() {
3886 int result = 0;
3887 result = getInt(0);
3888 return result;
3889 }
3890
3891 /**
3892 * Converts this BigInteger to a {@code long}. This
3893 * conversion is analogous to a
3894 * <i>narrowing primitive conversion</i> from {@code long} to
3895 * {@code int} as defined in section 5.1.3 of
3896 * <cite>The Java™ Language Specification</cite>:
3897 * if this BigInteger is too big to fit in a
3898 * {@code long}, only the low-order 64 bits are returned.
3899 * Note that this conversion can lose information about the
3900 * overall magnitude of the BigInteger value as well as return a
3901 * result with the opposite sign.
3902 *
3903 * @return this BigInteger converted to a {@code long}.
3904 * @see #longValueExact()
3905 */
3906 public long longValue() {
3907 long result = 0;
3908
3909 for (int i=1; i >= 0; i--)
3910 result = (result << 32) + (getInt(i) & LONG_MASK);
3911 return result;
3912 }
3913
3914 /**
3915 * Converts this BigInteger to a {@code float}. This
3916 * conversion is similar to the
3917 * <i>narrowing primitive conversion</i> from {@code double} to
3918 * {@code float} as defined in section 5.1.3 of
3919 * <cite>The Java™ Language Specification</cite>:
3920 * if this BigInteger has too great a magnitude
3921 * to represent as a {@code float}, it will be converted to
3922 * {@link Float#NEGATIVE_INFINITY} or {@link
3923 * Float#POSITIVE_INFINITY} as appropriate. Note that even when
3924 * the return value is finite, this conversion can lose
3925 * information about the precision of the BigInteger value.
3926 *
3927 * @return this BigInteger converted to a {@code float}.
3928 */
3929 public float floatValue() {
3930 if (signum == 0) {
3931 return 0.0f;
3932 }
3933
3934 int exponent = ((mag.length - 1) << 5) + bitLengthForInt(mag[0]) - 1;
3935
3936 // exponent == floor(log2(abs(this)))
3937 if (exponent < Long.SIZE - 1) {
3938 return longValue();
3939 } else if (exponent > Float.MAX_EXPONENT) {
3940 return signum > 0 ? Float.POSITIVE_INFINITY : Float.NEGATIVE_INFINITY;
3941 }
3942
3943 /*
3944 * We need the top SIGNIFICAND_WIDTH bits, including the "implicit"
3945 * one bit. To make rounding easier, we pick out the top
3946 * SIGNIFICAND_WIDTH + 1 bits, so we have one to help us round up or
3947 * down. twiceSignifFloor will contain the top SIGNIFICAND_WIDTH + 1
3948 * bits, and signifFloor the top SIGNIFICAND_WIDTH.
3949 *
3950 * It helps to consider the real number signif = abs(this) *
3951 * 2^(SIGNIFICAND_WIDTH - 1 - exponent).
3952 */
3953 int shift = exponent - FloatConsts.SIGNIFICAND_WIDTH;
3954
3955 int twiceSignifFloor;
3956 // twiceSignifFloor will be == abs().shiftRight(shift).intValue()
3957 // We do the shift into an int directly to improve performance.
3958
3959 int nBits = shift & 0x1f;
3960 int nBits2 = 32 - nBits;
3961
3962 if (nBits == 0) {
3963 twiceSignifFloor = mag[0];
3964 } else {
3965 twiceSignifFloor = mag[0] >>> nBits;
3966 if (twiceSignifFloor == 0) {
3967 twiceSignifFloor = (mag[0] << nBits2) | (mag[1] >>> nBits);
3968 }
3969 }
3970
3971 int signifFloor = twiceSignifFloor >> 1;
3972 signifFloor &= FloatConsts.SIGNIF_BIT_MASK; // remove the implied bit
3973
3974 /*
3975 * We round up if either the fractional part of signif is strictly
3976 * greater than 0.5 (which is true if the 0.5 bit is set and any lower
3977 * bit is set), or if the fractional part of signif is >= 0.5 and
3978 * signifFloor is odd (which is true if both the 0.5 bit and the 1 bit
3979 * are set). This is equivalent to the desired HALF_EVEN rounding.
3980 */
3981 boolean increment = (twiceSignifFloor & 1) != 0
3982 && ((signifFloor & 1) != 0 || abs().getLowestSetBit() < shift);
3983 int signifRounded = increment ? signifFloor + 1 : signifFloor;
3984 int bits = ((exponent + FloatConsts.EXP_BIAS))
3985 << (FloatConsts.SIGNIFICAND_WIDTH - 1);
3986 bits += signifRounded;
3987 /*
3988 * If signifRounded == 2^24, we'd need to set all of the significand
3989 * bits to zero and add 1 to the exponent. This is exactly the behavior
3990 * we get from just adding signifRounded to bits directly. If the
3991 * exponent is Float.MAX_EXPONENT, we round up (correctly) to
3992 * Float.POSITIVE_INFINITY.
3993 */
3994 bits |= signum & FloatConsts.SIGN_BIT_MASK;
3995 return Float.intBitsToFloat(bits);
3996 }
3997
3998 /**
3999 * Converts this BigInteger to a {@code double}. This
4000 * conversion is similar to the
4001 * <i>narrowing primitive conversion</i> from {@code double} to
4002 * {@code float} as defined in section 5.1.3 of
4003 * <cite>The Java™ Language Specification</cite>:
4004 * if this BigInteger has too great a magnitude
4005 * to represent as a {@code double}, it will be converted to
4006 * {@link Double#NEGATIVE_INFINITY} or {@link
4007 * Double#POSITIVE_INFINITY} as appropriate. Note that even when
4008 * the return value is finite, this conversion can lose
4009 * information about the precision of the BigInteger value.
4010 *
4011 * @return this BigInteger converted to a {@code double}.
4012 */
4013 public double doubleValue() {
4014 if (signum == 0) {
4015 return 0.0;
4016 }
4017
4018 int exponent = ((mag.length - 1) << 5) + bitLengthForInt(mag[0]) - 1;
4019
4020 // exponent == floor(log2(abs(this))Double)
4021 if (exponent < Long.SIZE - 1) {
4022 return longValue();
4023 } else if (exponent > Double.MAX_EXPONENT) {
4024 return signum > 0 ? Double.POSITIVE_INFINITY : Double.NEGATIVE_INFINITY;
4025 }
4026
4027 /*
4028 * We need the top SIGNIFICAND_WIDTH bits, including the "implicit"
4029 * one bit. To make rounding easier, we pick out the top
4030 * SIGNIFICAND_WIDTH + 1 bits, so we have one to help us round up or
4031 * down. twiceSignifFloor will contain the top SIGNIFICAND_WIDTH + 1
4032 * bits, and signifFloor the top SIGNIFICAND_WIDTH.
4033 *
4034 * It helps to consider the real number signif = abs(this) *
4035 * 2^(SIGNIFICAND_WIDTH - 1 - exponent).
4036 */
4037 int shift = exponent - DoubleConsts.SIGNIFICAND_WIDTH;
4038
4039 long twiceSignifFloor;
4040 // twiceSignifFloor will be == abs().shiftRight(shift).longValue()
4041 // We do the shift into a long directly to improve performance.
4042
4043 int nBits = shift & 0x1f;
4044 int nBits2 = 32 - nBits;
4045
4046 int highBits;
4047 int lowBits;
4048 if (nBits == 0) {
4049 highBits = mag[0];
4050 lowBits = mag[1];
4051 } else {
4052 highBits = mag[0] >>> nBits;
4053 lowBits = (mag[0] << nBits2) | (mag[1] >>> nBits);
4054 if (highBits == 0) {
4055 highBits = lowBits;
4056 lowBits = (mag[1] << nBits2) | (mag[2] >>> nBits);
4057 }
4058 }
4059
4060 twiceSignifFloor = ((highBits & LONG_MASK) << 32)
4061 | (lowBits & LONG_MASK);
4062
4063 long signifFloor = twiceSignifFloor >> 1;
4064 signifFloor &= DoubleConsts.SIGNIF_BIT_MASK; // remove the implied bit
4065
4066 /*
4067 * We round up if either the fractional part of signif is strictly
4068 * greater than 0.5 (which is true if the 0.5 bit is set and any lower
4069 * bit is set), or if the fractional part of signif is >= 0.5 and
4070 * signifFloor is odd (which is true if both the 0.5 bit and the 1 bit
4071 * are set). This is equivalent to the desired HALF_EVEN rounding.
4072 */
4073 boolean increment = (twiceSignifFloor & 1) != 0
4074 && ((signifFloor & 1) != 0 || abs().getLowestSetBit() < shift);
4075 long signifRounded = increment ? signifFloor + 1 : signifFloor;
4076 long bits = (long) ((exponent + DoubleConsts.EXP_BIAS))
4077 << (DoubleConsts.SIGNIFICAND_WIDTH - 1);
4078 bits += signifRounded;
4079 /*
4080 * If signifRounded == 2^53, we'd need to set all of the significand
4081 * bits to zero and add 1 to the exponent. This is exactly the behavior
4082 * we get from just adding signifRounded to bits directly. If the
4083 * exponent is Double.MAX_EXPONENT, we round up (correctly) to
4084 * Double.POSITIVE_INFINITY.
4085 */
4086 bits |= signum & DoubleConsts.SIGN_BIT_MASK;
4087 return Double.longBitsToDouble(bits);
4088 }
4089
4090 /**
4091 * Returns a copy of the input array stripped of any leading zero bytes.
4092 */
4093 private static int[] stripLeadingZeroInts(int val[]) {
4094 int vlen = val.length;
4095 int keep;
4096
4097 // Find first nonzero byte
4098 for (keep = 0; keep < vlen && val[keep] == 0; keep++)
4099 ;
4100 return java.util.Arrays.copyOfRange(val, keep, vlen);
4101 }
4102
4103 /**
4104 * Returns the input array stripped of any leading zero bytes.
4105 * Since the source is trusted the copying may be skipped.
4106 */
4107 private static int[] trustedStripLeadingZeroInts(int val[]) {
4108 int vlen = val.length;
4109 int keep;
4110
4111 // Find first nonzero byte
4112 for (keep = 0; keep < vlen && val[keep] == 0; keep++)
4113 ;
4114 return keep == 0 ? val : java.util.Arrays.copyOfRange(val, keep, vlen);
4115 }
4116
4117 /**
4118 * Returns a copy of the input array stripped of any leading zero bytes.
4119 */
4120 private static int[] stripLeadingZeroBytes(byte a[], int off, int len) {
4121 int indexBound = off + len;
4122 int keep;
4123
4124 // Find first nonzero byte
4125 for (keep = off; keep < indexBound && a[keep] == 0; keep++)
4126 ;
4127
4128 // Allocate new array and copy relevant part of input array
4129 int intLength = ((indexBound - keep) + 3) >>> 2;
4130 int[] result = new int[intLength];
4131 int b = indexBound - 1;
4132 for (int i = intLength-1; i >= 0; i--) {
4133 result[i] = a[b--] & 0xff;
4134 int bytesRemaining = b - keep + 1;
4135 int bytesToTransfer = Math.min(3, bytesRemaining);
4136 for (int j=8; j <= (bytesToTransfer << 3); j += 8)
4137 result[i] |= ((a[b--] & 0xff) << j);
4138 }
4139 return result;
4140 }
4141
4142 /**
4143 * Takes an array a representing a negative 2's-complement number and
4144 * returns the minimal (no leading zero bytes) unsigned whose value is -a.
4145 */
4146 private static int[] makePositive(byte a[], int off, int len) {
4147 int keep, k;
4148 int indexBound = off + len;
4149
4150 // Find first non-sign (0xff) byte of input
4151 for (keep=off; keep < indexBound && a[keep] == -1; keep++)
4152 ;
4153
4154
4155 /* Allocate output array. If all non-sign bytes are 0x00, we must
4156 * allocate space for one extra output byte. */
4157 for (k=keep; k < indexBound && a[k] == 0; k++)
4158 ;
4159
4160 int extraByte = (k == indexBound) ? 1 : 0;
4161 int intLength = ((indexBound - keep + extraByte) + 3) >>> 2;
4162 int result[] = new int[intLength];
4163
4164 /* Copy one's complement of input into output, leaving extra
4165 * byte (if it exists) == 0x00 */
4166 int b = indexBound - 1;
4167 for (int i = intLength-1; i >= 0; i--) {
4168 result[i] = a[b--] & 0xff;
4169 int numBytesToTransfer = Math.min(3, b-keep+1);
4170 if (numBytesToTransfer < 0)
4171 numBytesToTransfer = 0;
4172 for (int j=8; j <= 8*numBytesToTransfer; j += 8)
4173 result[i] |= ((a[b--] & 0xff) << j);
4174
4175 // Mask indicates which bits must be complemented
4176 int mask = -1 >>> (8*(3-numBytesToTransfer));
4177 result[i] = ~result[i] & mask;
4178 }
4179
4180 // Add one to one's complement to generate two's complement
4181 for (int i=result.length-1; i >= 0; i--) {
4182 result[i] = (int)((result[i] & LONG_MASK) + 1);
4183 if (result[i] != 0)
4184 break;
4185 }
4186
4187 return result;
4188 }
4189
4190 /**
4191 * Takes an array a representing a negative 2's-complement number and
4192 * returns the minimal (no leading zero ints) unsigned whose value is -a.
4193 */
4194 private static int[] makePositive(int a[]) {
4195 int keep, j;
4196
4197 // Find first non-sign (0xffffffff) int of input
4198 for (keep=0; keep < a.length && a[keep] == -1; keep++)
4199 ;
4200
4201 /* Allocate output array. If all non-sign ints are 0x00, we must
4202 * allocate space for one extra output int. */
4203 for (j=keep; j < a.length && a[j] == 0; j++)
4204 ;
4205 int extraInt = (j == a.length ? 1 : 0);
4206 int result[] = new int[a.length - keep + extraInt];
4207
4208 /* Copy one's complement of input into output, leaving extra
4209 * int (if it exists) == 0x00 */
4210 for (int i = keep; i < a.length; i++)
4211 result[i - keep + extraInt] = ~a[i];
4212
4213 // Add one to one's complement to generate two's complement
4214 for (int i=result.length-1; ++result[i] == 0; i--)
4215 ;
4216
4217 return result;
4218 }
4219
4220 /*
4221 * The following two arrays are used for fast String conversions. Both
4222 * are indexed by radix. The first is the number of digits of the given
4223 * radix that can fit in a Java long without "going negative", i.e., the
4224 * highest integer n such that radix**n < 2**63. The second is the
4225 * "long radix" that tears each number into "long digits", each of which
4226 * consists of the number of digits in the corresponding element in
4227 * digitsPerLong (longRadix[i] = i**digitPerLong[i]). Both arrays have
4228 * nonsense values in their 0 and 1 elements, as radixes 0 and 1 are not
4229 * used.
4230 */
4231 private static int digitsPerLong[] = {0, 0,
4232 62, 39, 31, 27, 24, 22, 20, 19, 18, 18, 17, 17, 16, 16, 15, 15, 15, 14,
4233 14, 14, 14, 13, 13, 13, 13, 13, 13, 12, 12, 12, 12, 12, 12, 12, 12};
4234
4235 private static BigInteger longRadix[] = {null, null,
4236 valueOf(0x4000000000000000L), valueOf(0x383d9170b85ff80bL),
4237 valueOf(0x4000000000000000L), valueOf(0x6765c793fa10079dL),
4238 valueOf(0x41c21cb8e1000000L), valueOf(0x3642798750226111L),
4239 valueOf(0x1000000000000000L), valueOf(0x12bf307ae81ffd59L),
4240 valueOf( 0xde0b6b3a7640000L), valueOf(0x4d28cb56c33fa539L),
4241 valueOf(0x1eca170c00000000L), valueOf(0x780c7372621bd74dL),
4242 valueOf(0x1e39a5057d810000L), valueOf(0x5b27ac993df97701L),
4243 valueOf(0x1000000000000000L), valueOf(0x27b95e997e21d9f1L),
4244 valueOf(0x5da0e1e53c5c8000L), valueOf( 0xb16a458ef403f19L),
4245 valueOf(0x16bcc41e90000000L), valueOf(0x2d04b7fdd9c0ef49L),
4246 valueOf(0x5658597bcaa24000L), valueOf( 0x6feb266931a75b7L),
4247 valueOf( 0xc29e98000000000L), valueOf(0x14adf4b7320334b9L),
4248 valueOf(0x226ed36478bfa000L), valueOf(0x383d9170b85ff80bL),
4249 valueOf(0x5a3c23e39c000000L), valueOf( 0x4e900abb53e6b71L),
4250 valueOf( 0x7600ec618141000L), valueOf( 0xaee5720ee830681L),
4251 valueOf(0x1000000000000000L), valueOf(0x172588ad4f5f0981L),
4252 valueOf(0x211e44f7d02c1000L), valueOf(0x2ee56725f06e5c71L),
4253 valueOf(0x41c21cb8e1000000L)};
4254
4255 /*
4256 * These two arrays are the integer analogue of above.
4257 */
4258 private static int digitsPerInt[] = {0, 0, 30, 19, 15, 13, 11,
4259 11, 10, 9, 9, 8, 8, 8, 8, 7, 7, 7, 7, 7, 7, 7, 6, 6, 6, 6,
4260 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 5};
4261
4262 private static int intRadix[] = {0, 0,
4263 0x40000000, 0x4546b3db, 0x40000000, 0x48c27395, 0x159fd800,
4264 0x75db9c97, 0x40000000, 0x17179149, 0x3b9aca00, 0xcc6db61,
4265 0x19a10000, 0x309f1021, 0x57f6c100, 0xa2f1b6f, 0x10000000,
4266 0x18754571, 0x247dbc80, 0x3547667b, 0x4c4b4000, 0x6b5a6e1d,
4267 0x6c20a40, 0x8d2d931, 0xb640000, 0xe8d4a51, 0x1269ae40,
4268 0x17179149, 0x1cb91000, 0x23744899, 0x2b73a840, 0x34e63b41,
4269 0x40000000, 0x4cfa3cc1, 0x5c13d840, 0x6d91b519, 0x39aa400
4270 };
4271
4272 /**
4273 * These routines provide access to the two's complement representation
4274 * of BigIntegers.
4275 */
4276
4277 /**
4278 * Returns the length of the two's complement representation in ints,
4279 * including space for at least one sign bit.
4280 */
4281 private int intLength() {
4282 return (bitLength() >>> 5) + 1;
4283 }
4284
4285 /* Returns sign bit */
4286 private int signBit() {
4287 return signum < 0 ? 1 : 0;
4288 }
4289
4290 /* Returns an int of sign bits */
4291 private int signInt() {
4292 return signum < 0 ? -1 : 0;
4293 }
4294
4295 /**
4296 * Returns the specified int of the little-endian two's complement
4297 * representation (int 0 is the least significant). The int number can
4298 * be arbitrarily high (values are logically preceded by infinitely many
4299 * sign ints).
4300 */
4301 private int getInt(int n) {
4302 if (n < 0)
4303 return 0;
4304 if (n >= mag.length)
4305 return signInt();
4306
4307 int magInt = mag[mag.length-n-1];
4308
4309 return (signum >= 0 ? magInt :
4310 (n <= firstNonzeroIntNum() ? -magInt : ~magInt));
4311 }
4312
4313 /**
4314 * Returns the index of the int that contains the first nonzero int in the
4315 * little-endian binary representation of the magnitude (int 0 is the
4316 * least significant). If the magnitude is zero, return value is undefined.
4317 *
4318 * <p>Note: never used for a BigInteger with a magnitude of zero.
4319 * @see #getInt.
4320 */
4321 private int firstNonzeroIntNum() {
4322 int fn = firstNonzeroIntNumPlusTwo - 2;
4323 if (fn == -2) { // firstNonzeroIntNum not initialized yet
4324 // Search for the first nonzero int
4325 int i;
4326 int mlen = mag.length;
4327 for (i = mlen - 1; i >= 0 && mag[i] == 0; i--)
4328 ;
4329 fn = mlen - i - 1;
4330 firstNonzeroIntNumPlusTwo = fn + 2; // offset by two to initialize
4331 }
4332 return fn;
4333 }
4334
4335 /** use serialVersionUID from JDK 1.1. for interoperability */
4336 private static final long serialVersionUID = -8287574255936472291L;
4337
4338 /**
4339 * Serializable fields for BigInteger.
4340 *
4341 * @serialField signum int
4342 * signum of this BigInteger
4343 * @serialField magnitude byte[]
4344 * magnitude array of this BigInteger
4345 * @serialField bitCount int
4346 * appears in the serialized form for backward compatibility
4347 * @serialField bitLength int
4348 * appears in the serialized form for backward compatibility
4349 * @serialField firstNonzeroByteNum int
4350 * appears in the serialized form for backward compatibility
4351 * @serialField lowestSetBit int
4352 * appears in the serialized form for backward compatibility
4353 */
4354 private static final ObjectStreamField[] serialPersistentFields = {
4355 new ObjectStreamField("signum", Integer.TYPE),
4356 new ObjectStreamField("magnitude", byte[].class),
4357 new ObjectStreamField("bitCount", Integer.TYPE),
4358 new ObjectStreamField("bitLength", Integer.TYPE),
4359 new ObjectStreamField("firstNonzeroByteNum", Integer.TYPE),
4360 new ObjectStreamField("lowestSetBit", Integer.TYPE)
4361 };
4362
4363 /**
4364 * Reconstitute the {@code BigInteger} instance from a stream (that is,
4365 * deserialize it). The magnitude is read in as an array of bytes
4366 * for historical reasons, but it is converted to an array of ints
4367 * and the byte array is discarded.
4368 * Note:
4369 * The current convention is to initialize the cache fields, bitCountPlusOne,
4370 * bitLengthPlusOne and lowestSetBitPlusTwo, to 0 rather than some other
4371 * marker value. Therefore, no explicit action to set these fields needs to
4372 * be taken in readObject because those fields already have a 0 value by
4373 * default since defaultReadObject is not being used.
4374 */
4375 private void readObject(java.io.ObjectInputStream s)
4376 throws java.io.IOException, ClassNotFoundException {
4377 // prepare to read the alternate persistent fields
4378 ObjectInputStream.GetField fields = s.readFields();
4379
4380 // Read the alternate persistent fields that we care about
4381 int sign = fields.get("signum", -2);
4382 byte[] magnitude = (byte[])fields.get("magnitude", null);
4383
4384 // Validate signum
4385 if (sign < -1 || sign > 1) {
4386 String message = "BigInteger: Invalid signum value";
4387 if (fields.defaulted("signum"))
4388 message = "BigInteger: Signum not present in stream";
4389 throw new java.io.StreamCorruptedException(message);
4390 }
4391 int[] mag = stripLeadingZeroBytes(magnitude, 0, magnitude.length);
4392 if ((mag.length == 0) != (sign == 0)) {
4393 String message = "BigInteger: signum-magnitude mismatch";
4394 if (fields.defaulted("magnitude"))
4395 message = "BigInteger: Magnitude not present in stream";
4396 throw new java.io.StreamCorruptedException(message);
4397 }
4398
4399 // Commit final fields via Unsafe
4400 UnsafeHolder.putSign(this, sign);
4401
4402 // Calculate mag field from magnitude and discard magnitude
4403 UnsafeHolder.putMag(this, mag);
4404 if (mag.length >= MAX_MAG_LENGTH) {
4405 try {
4406 checkRange();
4407 } catch (ArithmeticException e) {
4408 throw new java.io.StreamCorruptedException("BigInteger: Out of the supported range");
4409 }
4410 }
4411 }
4412
4413 // Support for resetting final fields while deserializing
4414 private static class UnsafeHolder {
4415 private static final sun.misc.Unsafe unsafe;
4416 private static final long signumOffset;
4417 private static final long magOffset;
4418 static {
4419 try {
4420 unsafe = sun.misc.Unsafe.getUnsafe();
4421 signumOffset = unsafe.objectFieldOffset
4422 (BigInteger.class.getDeclaredField("signum"));
4423 magOffset = unsafe.objectFieldOffset
4424 (BigInteger.class.getDeclaredField("mag"));
4425 } catch (Exception ex) {
4426 throw new ExceptionInInitializerError(ex);
4427 }
4428 }
4429
4430 static void putSign(BigInteger bi, int sign) {
4431 unsafe.putInt(bi, signumOffset, sign);
4432 }
4433
4434 static void putMag(BigInteger bi, int[] magnitude) {
4435 unsafe.putObject(bi, magOffset, magnitude);
4436 }
4437 }
4438
4439 /**
4440 * Save the {@code BigInteger} instance to a stream. The magnitude of a
4441 * {@code BigInteger} is serialized as a byte array for historical reasons.
4442 * To maintain compatibility with older implementations, the integers
4443 * -1, -1, -2, and -2 are written as the values of the obsolete fields
4444 * {@code bitCount}, {@code bitLength}, {@code lowestSetBit}, and
4445 * {@code firstNonzeroByteNum}, respectively. These values are compatible
4446 * with older implementations, but will be ignored by current
4447 * implementations.
4448 */
4449 private void writeObject(ObjectOutputStream s) throws IOException {
4450 // set the values of the Serializable fields
4451 ObjectOutputStream.PutField fields = s.putFields();
4452 fields.put("signum", signum);
4453 fields.put("magnitude", magSerializedForm());
4454 // The values written for cached fields are compatible with older
4455 // versions, but are ignored in readObject so don't otherwise matter.
4456 fields.put("bitCount", -1);
4457 fields.put("bitLength", -1);
4458 fields.put("lowestSetBit", -2);
4459 fields.put("firstNonzeroByteNum", -2);
4460
4461 // save them
4462 s.writeFields();
4463 }
4464
4465 /**
4466 * Returns the mag array as an array of bytes.
4467 */
4468 private byte[] magSerializedForm() {
4469 int len = mag.length;
4470
4471 int bitLen = (len == 0 ? 0 : ((len - 1) << 5) + bitLengthForInt(mag[0]));
4472 int byteLen = (bitLen + 7) >>> 3;
4473 byte[] result = new byte[byteLen];
4474
4475 for (int i = byteLen - 1, bytesCopied = 4, intIndex = len - 1, nextInt = 0;
4476 i >= 0; i--) {
4477 if (bytesCopied == 4) {
4478 nextInt = mag[intIndex--];
4479 bytesCopied = 1;
4480 } else {
4481 nextInt >>>= 8;
4482 bytesCopied++;
4483 }
4484 result[i] = (byte)nextInt;
4485 }
4486 return result;
4487 }
4488
4489 /**
4490 * Converts this {@code BigInteger} to a {@code long}, checking
4491 * for lost information. If the value of this {@code BigInteger}
4492 * is out of the range of the {@code long} type, then an
4493 * {@code ArithmeticException} is thrown.
4494 *
4495 * @return this {@code BigInteger} converted to a {@code long}.
4496 * @throws ArithmeticException if the value of {@code this} will
4497 * not exactly fit in a {@code long}.
4498 * @see BigInteger#longValue
4499 * @since 1.8
4500 */
4501 public long longValueExact() {
4502 if (mag.length <= 2 && bitLength() <= 63)
4503 return longValue();
4504 else
4505 throw new ArithmeticException("BigInteger out of long range");
4506 }
4507
4508 /**
4509 * Converts this {@code BigInteger} to an {@code int}, checking
4510 * for lost information. If the value of this {@code BigInteger}
4511 * is out of the range of the {@code int} type, then an
4512 * {@code ArithmeticException} is thrown.
4513 *
4514 * @return this {@code BigInteger} converted to an {@code int}.
4515 * @throws ArithmeticException if the value of {@code this} will
4516 * not exactly fit in a {@code int}.
4517 * @see BigInteger#intValue
4518 * @since 1.8
4519 */
4520 public int intValueExact() {
4521 if (mag.length <= 1 && bitLength() <= 31)
4522 return intValue();
4523 else
4524 throw new ArithmeticException("BigInteger out of int range");
4525 }
4526
4527 /**
4528 * Converts this {@code BigInteger} to a {@code short}, checking
4529 * for lost information. If the value of this {@code BigInteger}
4530 * is out of the range of the {@code short} type, then an
4531 * {@code ArithmeticException} is thrown.
4532 *
4533 * @return this {@code BigInteger} converted to a {@code short}.
4534 * @throws ArithmeticException if the value of {@code this} will
4535 * not exactly fit in a {@code short}.
4536 * @see BigInteger#shortValue
4537 * @since 1.8
4538 */
4539 public short shortValueExact() {
4540 if (mag.length <= 1 && bitLength() <= 31) {
4541 int value = intValue();
4542 if (value >= Short.MIN_VALUE && value <= Short.MAX_VALUE)
4543 return shortValue();
4544 }
4545 throw new ArithmeticException("BigInteger out of short range");
4546 }
4547
4548 /**
4549 * Converts this {@code BigInteger} to a {@code byte}, checking
4550 * for lost information. If the value of this {@code BigInteger}
4551 * is out of the range of the {@code byte} type, then an
4552 * {@code ArithmeticException} is thrown.
4553 *
4554 * @return this {@code BigInteger} converted to a {@code byte}.
4555 * @throws ArithmeticException if the value of {@code this} will
4556 * not exactly fit in a {@code byte}.
4557 * @see BigInteger#byteValue
4558 * @since 1.8
4559 */
4560 public byte byteValueExact() {
4561 if (mag.length <= 1 && bitLength() <= 31) {
4562 int value = intValue();
4563 if (value >= Byte.MIN_VALUE && value <= Byte.MAX_VALUE)
4564 return byteValue();
4565 }
4566 throw new ArithmeticException("BigInteger out of byte range");
4567 }
4568 }