emul/compact/src/main/java/java/math/BigInteger.java
branchjdk7-b147
changeset 1258 724f3e1ea53e
     1.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     1.2 +++ b/emul/compact/src/main/java/java/math/BigInteger.java	Sat Sep 07 13:51:24 2013 +0200
     1.3 @@ -0,0 +1,3186 @@
     1.4 +/*
     1.5 + * Copyright (c) 1996, 2007, Oracle and/or its affiliates. All rights reserved.
     1.6 + * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
     1.7 + *
     1.8 + * This code is free software; you can redistribute it and/or modify it
     1.9 + * under the terms of the GNU General Public License version 2 only, as
    1.10 + * published by the Free Software Foundation.  Oracle designates this
    1.11 + * particular file as subject to the "Classpath" exception as provided
    1.12 + * by Oracle in the LICENSE file that accompanied this code.
    1.13 + *
    1.14 + * This code is distributed in the hope that it will be useful, but WITHOUT
    1.15 + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
    1.16 + * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
    1.17 + * version 2 for more details (a copy is included in the LICENSE file that
    1.18 + * accompanied this code).
    1.19 + *
    1.20 + * You should have received a copy of the GNU General Public License version
    1.21 + * 2 along with this work; if not, write to the Free Software Foundation,
    1.22 + * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
    1.23 + *
    1.24 + * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
    1.25 + * or visit www.oracle.com if you need additional information or have any
    1.26 + * questions.
    1.27 + */
    1.28 +
    1.29 +/*
    1.30 + * Portions Copyright (c) 1995  Colin Plumb.  All rights reserved.
    1.31 + */
    1.32 +
    1.33 +package java.math;
    1.34 +
    1.35 +import java.util.Random;
    1.36 +import java.io.*;
    1.37 +
    1.38 +/**
    1.39 + * Immutable arbitrary-precision integers.  All operations behave as if
    1.40 + * BigIntegers were represented in two's-complement notation (like Java's
    1.41 + * primitive integer types).  BigInteger provides analogues to all of Java's
    1.42 + * primitive integer operators, and all relevant methods from java.lang.Math.
    1.43 + * Additionally, BigInteger provides operations for modular arithmetic, GCD
    1.44 + * calculation, primality testing, prime generation, bit manipulation,
    1.45 + * and a few other miscellaneous operations.
    1.46 + *
    1.47 + * <p>Semantics of arithmetic operations exactly mimic those of Java's integer
    1.48 + * arithmetic operators, as defined in <i>The Java Language Specification</i>.
    1.49 + * For example, division by zero throws an {@code ArithmeticException}, and
    1.50 + * division of a negative by a positive yields a negative (or zero) remainder.
    1.51 + * All of the details in the Spec concerning overflow are ignored, as
    1.52 + * BigIntegers are made as large as necessary to accommodate the results of an
    1.53 + * operation.
    1.54 + *
    1.55 + * <p>Semantics of shift operations extend those of Java's shift operators
    1.56 + * to allow for negative shift distances.  A right-shift with a negative
    1.57 + * shift distance results in a left shift, and vice-versa.  The unsigned
    1.58 + * right shift operator ({@code >>>}) is omitted, as this operation makes
    1.59 + * little sense in combination with the "infinite word size" abstraction
    1.60 + * provided by this class.
    1.61 + *
    1.62 + * <p>Semantics of bitwise logical operations exactly mimic those of Java's
    1.63 + * bitwise integer operators.  The binary operators ({@code and},
    1.64 + * {@code or}, {@code xor}) implicitly perform sign extension on the shorter
    1.65 + * of the two operands prior to performing the operation.
    1.66 + *
    1.67 + * <p>Comparison operations perform signed integer comparisons, analogous to
    1.68 + * those performed by Java's relational and equality operators.
    1.69 + *
    1.70 + * <p>Modular arithmetic operations are provided to compute residues, perform
    1.71 + * exponentiation, and compute multiplicative inverses.  These methods always
    1.72 + * return a non-negative result, between {@code 0} and {@code (modulus - 1)},
    1.73 + * inclusive.
    1.74 + *
    1.75 + * <p>Bit operations operate on a single bit of the two's-complement
    1.76 + * representation of their operand.  If necessary, the operand is sign-
    1.77 + * extended so that it contains the designated bit.  None of the single-bit
    1.78 + * operations can produce a BigInteger with a different sign from the
    1.79 + * BigInteger being operated on, as they affect only a single bit, and the
    1.80 + * "infinite word size" abstraction provided by this class ensures that there
    1.81 + * are infinitely many "virtual sign bits" preceding each BigInteger.
    1.82 + *
    1.83 + * <p>For the sake of brevity and clarity, pseudo-code is used throughout the
    1.84 + * descriptions of BigInteger methods.  The pseudo-code expression
    1.85 + * {@code (i + j)} is shorthand for "a BigInteger whose value is
    1.86 + * that of the BigInteger {@code i} plus that of the BigInteger {@code j}."
    1.87 + * The pseudo-code expression {@code (i == j)} is shorthand for
    1.88 + * "{@code true} if and only if the BigInteger {@code i} represents the same
    1.89 + * value as the BigInteger {@code j}."  Other pseudo-code expressions are
    1.90 + * interpreted similarly.
    1.91 + *
    1.92 + * <p>All methods and constructors in this class throw
    1.93 + * {@code NullPointerException} when passed
    1.94 + * a null object reference for any input parameter.
    1.95 + *
    1.96 + * @see     BigDecimal
    1.97 + * @author  Josh Bloch
    1.98 + * @author  Michael McCloskey
    1.99 + * @since JDK1.1
   1.100 + */
   1.101 +
   1.102 +public class BigInteger extends Number implements Comparable<BigInteger> {
   1.103 +    /**
   1.104 +     * The signum of this BigInteger: -1 for negative, 0 for zero, or
   1.105 +     * 1 for positive.  Note that the BigInteger zero <i>must</i> have
   1.106 +     * a signum of 0.  This is necessary to ensures that there is exactly one
   1.107 +     * representation for each BigInteger value.
   1.108 +     *
   1.109 +     * @serial
   1.110 +     */
   1.111 +    final int signum;
   1.112 +
   1.113 +    /**
   1.114 +     * The magnitude of this BigInteger, in <i>big-endian</i> order: the
   1.115 +     * zeroth element of this array is the most-significant int of the
   1.116 +     * magnitude.  The magnitude must be "minimal" in that the most-significant
   1.117 +     * int ({@code mag[0]}) must be non-zero.  This is necessary to
   1.118 +     * ensure that there is exactly one representation for each BigInteger
   1.119 +     * value.  Note that this implies that the BigInteger zero has a
   1.120 +     * zero-length mag array.
   1.121 +     */
   1.122 +    final int[] mag;
   1.123 +
   1.124 +    // These "redundant fields" are initialized with recognizable nonsense
   1.125 +    // values, and cached the first time they are needed (or never, if they
   1.126 +    // aren't needed).
   1.127 +
   1.128 +     /**
   1.129 +     * One plus the bitCount of this BigInteger. Zeros means unitialized.
   1.130 +     *
   1.131 +     * @serial
   1.132 +     * @see #bitCount
   1.133 +     * @deprecated Deprecated since logical value is offset from stored
   1.134 +     * value and correction factor is applied in accessor method.
   1.135 +     */
   1.136 +    @Deprecated
   1.137 +    private int bitCount;
   1.138 +
   1.139 +    /**
   1.140 +     * One plus the bitLength of this BigInteger. Zeros means unitialized.
   1.141 +     * (either value is acceptable).
   1.142 +     *
   1.143 +     * @serial
   1.144 +     * @see #bitLength()
   1.145 +     * @deprecated Deprecated since logical value is offset from stored
   1.146 +     * value and correction factor is applied in accessor method.
   1.147 +     */
   1.148 +    @Deprecated
   1.149 +    private int bitLength;
   1.150 +
   1.151 +    /**
   1.152 +     * Two plus the lowest set bit of this BigInteger, as returned by
   1.153 +     * getLowestSetBit().
   1.154 +     *
   1.155 +     * @serial
   1.156 +     * @see #getLowestSetBit
   1.157 +     * @deprecated Deprecated since logical value is offset from stored
   1.158 +     * value and correction factor is applied in accessor method.
   1.159 +     */
   1.160 +    @Deprecated
   1.161 +    private int lowestSetBit;
   1.162 +
   1.163 +    /**
   1.164 +     * Two plus the index of the lowest-order int in the magnitude of this
   1.165 +     * BigInteger that contains a nonzero int, or -2 (either value is acceptable).
   1.166 +     * The least significant int has int-number 0, the next int in order of
   1.167 +     * increasing significance has int-number 1, and so forth.
   1.168 +     * @deprecated Deprecated since logical value is offset from stored
   1.169 +     * value and correction factor is applied in accessor method.
   1.170 +     */
   1.171 +    @Deprecated
   1.172 +    private int firstNonzeroIntNum;
   1.173 +
   1.174 +    /**
   1.175 +     * This mask is used to obtain the value of an int as if it were unsigned.
   1.176 +     */
   1.177 +    final static long LONG_MASK = 0xffffffffL;
   1.178 +
   1.179 +    //Constructors
   1.180 +
   1.181 +    /**
   1.182 +     * Translates a byte array containing the two's-complement binary
   1.183 +     * representation of a BigInteger into a BigInteger.  The input array is
   1.184 +     * assumed to be in <i>big-endian</i> byte-order: the most significant
   1.185 +     * byte is in the zeroth element.
   1.186 +     *
   1.187 +     * @param  val big-endian two's-complement binary representation of
   1.188 +     *         BigInteger.
   1.189 +     * @throws NumberFormatException {@code val} is zero bytes long.
   1.190 +     */
   1.191 +    public BigInteger(byte[] val) {
   1.192 +        if (val.length == 0)
   1.193 +            throw new NumberFormatException("Zero length BigInteger");
   1.194 +
   1.195 +        if (val[0] < 0) {
   1.196 +            mag = makePositive(val);
   1.197 +            signum = -1;
   1.198 +        } else {
   1.199 +            mag = stripLeadingZeroBytes(val);
   1.200 +            signum = (mag.length == 0 ? 0 : 1);
   1.201 +        }
   1.202 +    }
   1.203 +
   1.204 +    /**
   1.205 +     * This private constructor translates an int array containing the
   1.206 +     * two's-complement binary representation of a BigInteger into a
   1.207 +     * BigInteger. The input array is assumed to be in <i>big-endian</i>
   1.208 +     * int-order: the most significant int is in the zeroth element.
   1.209 +     */
   1.210 +    private BigInteger(int[] val) {
   1.211 +        if (val.length == 0)
   1.212 +            throw new NumberFormatException("Zero length BigInteger");
   1.213 +
   1.214 +        if (val[0] < 0) {
   1.215 +            mag = makePositive(val);
   1.216 +            signum = -1;
   1.217 +        } else {
   1.218 +            mag = trustedStripLeadingZeroInts(val);
   1.219 +            signum = (mag.length == 0 ? 0 : 1);
   1.220 +        }
   1.221 +    }
   1.222 +
   1.223 +    /**
   1.224 +     * Translates the sign-magnitude representation of a BigInteger into a
   1.225 +     * BigInteger.  The sign is represented as an integer signum value: -1 for
   1.226 +     * negative, 0 for zero, or 1 for positive.  The magnitude is a byte array
   1.227 +     * in <i>big-endian</i> byte-order: the most significant byte is in the
   1.228 +     * zeroth element.  A zero-length magnitude array is permissible, and will
   1.229 +     * result in a BigInteger value of 0, whether signum is -1, 0 or 1.
   1.230 +     *
   1.231 +     * @param  signum signum of the number (-1 for negative, 0 for zero, 1
   1.232 +     *         for positive).
   1.233 +     * @param  magnitude big-endian binary representation of the magnitude of
   1.234 +     *         the number.
   1.235 +     * @throws NumberFormatException {@code signum} is not one of the three
   1.236 +     *         legal values (-1, 0, and 1), or {@code signum} is 0 and
   1.237 +     *         {@code magnitude} contains one or more non-zero bytes.
   1.238 +     */
   1.239 +    public BigInteger(int signum, byte[] magnitude) {
   1.240 +        this.mag = stripLeadingZeroBytes(magnitude);
   1.241 +
   1.242 +        if (signum < -1 || signum > 1)
   1.243 +            throw(new NumberFormatException("Invalid signum value"));
   1.244 +
   1.245 +        if (this.mag.length==0) {
   1.246 +            this.signum = 0;
   1.247 +        } else {
   1.248 +            if (signum == 0)
   1.249 +                throw(new NumberFormatException("signum-magnitude mismatch"));
   1.250 +            this.signum = signum;
   1.251 +        }
   1.252 +    }
   1.253 +
   1.254 +    /**
   1.255 +     * A constructor for internal use that translates the sign-magnitude
   1.256 +     * representation of a BigInteger into a BigInteger. It checks the
   1.257 +     * arguments and copies the magnitude so this constructor would be
   1.258 +     * safe for external use.
   1.259 +     */
   1.260 +    private BigInteger(int signum, int[] magnitude) {
   1.261 +        this.mag = stripLeadingZeroInts(magnitude);
   1.262 +
   1.263 +        if (signum < -1 || signum > 1)
   1.264 +            throw(new NumberFormatException("Invalid signum value"));
   1.265 +
   1.266 +        if (this.mag.length==0) {
   1.267 +            this.signum = 0;
   1.268 +        } else {
   1.269 +            if (signum == 0)
   1.270 +                throw(new NumberFormatException("signum-magnitude mismatch"));
   1.271 +            this.signum = signum;
   1.272 +        }
   1.273 +    }
   1.274 +
   1.275 +    /**
   1.276 +     * Translates the String representation of a BigInteger in the
   1.277 +     * specified radix into a BigInteger.  The String representation
   1.278 +     * consists of an optional minus or plus sign followed by a
   1.279 +     * sequence of one or more digits in the specified radix.  The
   1.280 +     * character-to-digit mapping is provided by {@code
   1.281 +     * Character.digit}.  The String may not contain any extraneous
   1.282 +     * characters (whitespace, for example).
   1.283 +     *
   1.284 +     * @param val String representation of BigInteger.
   1.285 +     * @param radix radix to be used in interpreting {@code val}.
   1.286 +     * @throws NumberFormatException {@code val} is not a valid representation
   1.287 +     *         of a BigInteger in the specified radix, or {@code radix} is
   1.288 +     *         outside the range from {@link Character#MIN_RADIX} to
   1.289 +     *         {@link Character#MAX_RADIX}, inclusive.
   1.290 +     * @see    Character#digit
   1.291 +     */
   1.292 +    public BigInteger(String val, int radix) {
   1.293 +        int cursor = 0, numDigits;
   1.294 +        final int len = val.length();
   1.295 +
   1.296 +        if (radix < Character.MIN_RADIX || radix > Character.MAX_RADIX)
   1.297 +            throw new NumberFormatException("Radix out of range");
   1.298 +        if (len == 0)
   1.299 +            throw new NumberFormatException("Zero length BigInteger");
   1.300 +
   1.301 +        // Check for at most one leading sign
   1.302 +        int sign = 1;
   1.303 +        int index1 = val.lastIndexOf('-');
   1.304 +        int index2 = val.lastIndexOf('+');
   1.305 +        if ((index1 + index2) <= -1) {
   1.306 +            // No leading sign character or at most one leading sign character
   1.307 +            if (index1 == 0 || index2 == 0) {
   1.308 +                cursor = 1;
   1.309 +                if (len == 1)
   1.310 +                    throw new NumberFormatException("Zero length BigInteger");
   1.311 +            }
   1.312 +            if (index1 == 0)
   1.313 +                sign = -1;
   1.314 +        } else
   1.315 +            throw new NumberFormatException("Illegal embedded sign character");
   1.316 +
   1.317 +        // Skip leading zeros and compute number of digits in magnitude
   1.318 +        while (cursor < len &&
   1.319 +               Character.digit(val.charAt(cursor), radix) == 0)
   1.320 +            cursor++;
   1.321 +        if (cursor == len) {
   1.322 +            signum = 0;
   1.323 +            mag = ZERO.mag;
   1.324 +            return;
   1.325 +        }
   1.326 +
   1.327 +        numDigits = len - cursor;
   1.328 +        signum = sign;
   1.329 +
   1.330 +        // Pre-allocate array of expected size. May be too large but can
   1.331 +        // never be too small. Typically exact.
   1.332 +        int numBits = (int)(((numDigits * bitsPerDigit[radix]) >>> 10) + 1);
   1.333 +        int numWords = (numBits + 31) >>> 5;
   1.334 +        int[] magnitude = new int[numWords];
   1.335 +
   1.336 +        // Process first (potentially short) digit group
   1.337 +        int firstGroupLen = numDigits % digitsPerInt[radix];
   1.338 +        if (firstGroupLen == 0)
   1.339 +            firstGroupLen = digitsPerInt[radix];
   1.340 +        String group = val.substring(cursor, cursor += firstGroupLen);
   1.341 +        magnitude[numWords - 1] = Integer.parseInt(group, radix);
   1.342 +        if (magnitude[numWords - 1] < 0)
   1.343 +            throw new NumberFormatException("Illegal digit");
   1.344 +
   1.345 +        // Process remaining digit groups
   1.346 +        int superRadix = intRadix[radix];
   1.347 +        int groupVal = 0;
   1.348 +        while (cursor < len) {
   1.349 +            group = val.substring(cursor, cursor += digitsPerInt[radix]);
   1.350 +            groupVal = Integer.parseInt(group, radix);
   1.351 +            if (groupVal < 0)
   1.352 +                throw new NumberFormatException("Illegal digit");
   1.353 +            destructiveMulAdd(magnitude, superRadix, groupVal);
   1.354 +        }
   1.355 +        // Required for cases where the array was overallocated.
   1.356 +        mag = trustedStripLeadingZeroInts(magnitude);
   1.357 +    }
   1.358 +
   1.359 +    // Constructs a new BigInteger using a char array with radix=10
   1.360 +    BigInteger(char[] val) {
   1.361 +        int cursor = 0, numDigits;
   1.362 +        int len = val.length;
   1.363 +
   1.364 +        // Check for leading minus sign
   1.365 +        int sign = 1;
   1.366 +        if (val[0] == '-') {
   1.367 +            if (len == 1)
   1.368 +                throw new NumberFormatException("Zero length BigInteger");
   1.369 +            sign = -1;
   1.370 +            cursor = 1;
   1.371 +        } else if (val[0] == '+') {
   1.372 +            if (len == 1)
   1.373 +                throw new NumberFormatException("Zero length BigInteger");
   1.374 +            cursor = 1;
   1.375 +        }
   1.376 +
   1.377 +        // Skip leading zeros and compute number of digits in magnitude
   1.378 +        while (cursor < len && Character.digit(val[cursor], 10) == 0)
   1.379 +            cursor++;
   1.380 +        if (cursor == len) {
   1.381 +            signum = 0;
   1.382 +            mag = ZERO.mag;
   1.383 +            return;
   1.384 +        }
   1.385 +
   1.386 +        numDigits = len - cursor;
   1.387 +        signum = sign;
   1.388 +
   1.389 +        // Pre-allocate array of expected size
   1.390 +        int numWords;
   1.391 +        if (len < 10) {
   1.392 +            numWords = 1;
   1.393 +        } else {
   1.394 +            int numBits = (int)(((numDigits * bitsPerDigit[10]) >>> 10) + 1);
   1.395 +            numWords = (numBits + 31) >>> 5;
   1.396 +        }
   1.397 +        int[] magnitude = new int[numWords];
   1.398 +
   1.399 +        // Process first (potentially short) digit group
   1.400 +        int firstGroupLen = numDigits % digitsPerInt[10];
   1.401 +        if (firstGroupLen == 0)
   1.402 +            firstGroupLen = digitsPerInt[10];
   1.403 +        magnitude[numWords - 1] = parseInt(val, cursor,  cursor += firstGroupLen);
   1.404 +
   1.405 +        // Process remaining digit groups
   1.406 +        while (cursor < len) {
   1.407 +            int groupVal = parseInt(val, cursor, cursor += digitsPerInt[10]);
   1.408 +            destructiveMulAdd(magnitude, intRadix[10], groupVal);
   1.409 +        }
   1.410 +        mag = trustedStripLeadingZeroInts(magnitude);
   1.411 +    }
   1.412 +
   1.413 +    // Create an integer with the digits between the two indexes
   1.414 +    // Assumes start < end. The result may be negative, but it
   1.415 +    // is to be treated as an unsigned value.
   1.416 +    private int parseInt(char[] source, int start, int end) {
   1.417 +        int result = Character.digit(source[start++], 10);
   1.418 +        if (result == -1)
   1.419 +            throw new NumberFormatException(new String(source));
   1.420 +
   1.421 +        for (int index = start; index<end; index++) {
   1.422 +            int nextVal = Character.digit(source[index], 10);
   1.423 +            if (nextVal == -1)
   1.424 +                throw new NumberFormatException(new String(source));
   1.425 +            result = 10*result + nextVal;
   1.426 +        }
   1.427 +
   1.428 +        return result;
   1.429 +    }
   1.430 +
   1.431 +    // bitsPerDigit in the given radix times 1024
   1.432 +    // Rounded up to avoid underallocation.
   1.433 +    private static long bitsPerDigit[] = { 0, 0,
   1.434 +        1024, 1624, 2048, 2378, 2648, 2875, 3072, 3247, 3402, 3543, 3672,
   1.435 +        3790, 3899, 4001, 4096, 4186, 4271, 4350, 4426, 4498, 4567, 4633,
   1.436 +        4696, 4756, 4814, 4870, 4923, 4975, 5025, 5074, 5120, 5166, 5210,
   1.437 +                                           5253, 5295};
   1.438 +
   1.439 +    // Multiply x array times word y in place, and add word z
   1.440 +    private static void destructiveMulAdd(int[] x, int y, int z) {
   1.441 +        // Perform the multiplication word by word
   1.442 +        long ylong = y & LONG_MASK;
   1.443 +        long zlong = z & LONG_MASK;
   1.444 +        int len = x.length;
   1.445 +
   1.446 +        long product = 0;
   1.447 +        long carry = 0;
   1.448 +        for (int i = len-1; i >= 0; i--) {
   1.449 +            product = ylong * (x[i] & LONG_MASK) + carry;
   1.450 +            x[i] = (int)product;
   1.451 +            carry = product >>> 32;
   1.452 +        }
   1.453 +
   1.454 +        // Perform the addition
   1.455 +        long sum = (x[len-1] & LONG_MASK) + zlong;
   1.456 +        x[len-1] = (int)sum;
   1.457 +        carry = sum >>> 32;
   1.458 +        for (int i = len-2; i >= 0; i--) {
   1.459 +            sum = (x[i] & LONG_MASK) + carry;
   1.460 +            x[i] = (int)sum;
   1.461 +            carry = sum >>> 32;
   1.462 +        }
   1.463 +    }
   1.464 +
   1.465 +    /**
   1.466 +     * Translates the decimal String representation of a BigInteger into a
   1.467 +     * BigInteger.  The String representation consists of an optional minus
   1.468 +     * sign followed by a sequence of one or more decimal digits.  The
   1.469 +     * character-to-digit mapping is provided by {@code Character.digit}.
   1.470 +     * The String may not contain any extraneous characters (whitespace, for
   1.471 +     * example).
   1.472 +     *
   1.473 +     * @param val decimal String representation of BigInteger.
   1.474 +     * @throws NumberFormatException {@code val} is not a valid representation
   1.475 +     *         of a BigInteger.
   1.476 +     * @see    Character#digit
   1.477 +     */
   1.478 +    public BigInteger(String val) {
   1.479 +        this(val, 10);
   1.480 +    }
   1.481 +
   1.482 +    /**
   1.483 +     * Constructs a randomly generated BigInteger, uniformly distributed over
   1.484 +     * the range 0 to (2<sup>{@code numBits}</sup> - 1), inclusive.
   1.485 +     * The uniformity of the distribution assumes that a fair source of random
   1.486 +     * bits is provided in {@code rnd}.  Note that this constructor always
   1.487 +     * constructs a non-negative BigInteger.
   1.488 +     *
   1.489 +     * @param  numBits maximum bitLength of the new BigInteger.
   1.490 +     * @param  rnd source of randomness to be used in computing the new
   1.491 +     *         BigInteger.
   1.492 +     * @throws IllegalArgumentException {@code numBits} is negative.
   1.493 +     * @see #bitLength()
   1.494 +     */
   1.495 +    public BigInteger(int numBits, Random rnd) {
   1.496 +        this(1, randomBits(numBits, rnd));
   1.497 +    }
   1.498 +
   1.499 +    private static byte[] randomBits(int numBits, Random rnd) {
   1.500 +        if (numBits < 0)
   1.501 +            throw new IllegalArgumentException("numBits must be non-negative");
   1.502 +        int numBytes = (int)(((long)numBits+7)/8); // avoid overflow
   1.503 +        byte[] randomBits = new byte[numBytes];
   1.504 +
   1.505 +        // Generate random bytes and mask out any excess bits
   1.506 +        if (numBytes > 0) {
   1.507 +            rnd.nextBytes(randomBits);
   1.508 +            int excessBits = 8*numBytes - numBits;
   1.509 +            randomBits[0] &= (1 << (8-excessBits)) - 1;
   1.510 +        }
   1.511 +        return randomBits;
   1.512 +    }
   1.513 +
   1.514 +    /**
   1.515 +     * Constructs a randomly generated positive BigInteger that is probably
   1.516 +     * prime, with the specified bitLength.
   1.517 +     *
   1.518 +     * <p>It is recommended that the {@link #probablePrime probablePrime}
   1.519 +     * method be used in preference to this constructor unless there
   1.520 +     * is a compelling need to specify a certainty.
   1.521 +     *
   1.522 +     * @param  bitLength bitLength of the returned BigInteger.
   1.523 +     * @param  certainty a measure of the uncertainty that the caller is
   1.524 +     *         willing to tolerate.  The probability that the new BigInteger
   1.525 +     *         represents a prime number will exceed
   1.526 +     *         (1 - 1/2<sup>{@code certainty}</sup>).  The execution time of
   1.527 +     *         this constructor is proportional to the value of this parameter.
   1.528 +     * @param  rnd source of random bits used to select candidates to be
   1.529 +     *         tested for primality.
   1.530 +     * @throws ArithmeticException {@code bitLength < 2}.
   1.531 +     * @see    #bitLength()
   1.532 +     */
   1.533 +    public BigInteger(int bitLength, int certainty, Random rnd) {
   1.534 +        BigInteger prime;
   1.535 +
   1.536 +        if (bitLength < 2)
   1.537 +            throw new ArithmeticException("bitLength < 2");
   1.538 +        // The cutoff of 95 was chosen empirically for best performance
   1.539 +        prime = (bitLength < 95 ? smallPrime(bitLength, certainty, rnd)
   1.540 +                                : largePrime(bitLength, certainty, rnd));
   1.541 +        signum = 1;
   1.542 +        mag = prime.mag;
   1.543 +    }
   1.544 +
   1.545 +    // Minimum size in bits that the requested prime number has
   1.546 +    // before we use the large prime number generating algorithms
   1.547 +    private static final int SMALL_PRIME_THRESHOLD = 95;
   1.548 +
   1.549 +    // Certainty required to meet the spec of probablePrime
   1.550 +    private static final int DEFAULT_PRIME_CERTAINTY = 100;
   1.551 +
   1.552 +    /**
   1.553 +     * Returns a positive BigInteger that is probably prime, with the
   1.554 +     * specified bitLength. The probability that a BigInteger returned
   1.555 +     * by this method is composite does not exceed 2<sup>-100</sup>.
   1.556 +     *
   1.557 +     * @param  bitLength bitLength of the returned BigInteger.
   1.558 +     * @param  rnd source of random bits used to select candidates to be
   1.559 +     *         tested for primality.
   1.560 +     * @return a BigInteger of {@code bitLength} bits that is probably prime
   1.561 +     * @throws ArithmeticException {@code bitLength < 2}.
   1.562 +     * @see    #bitLength()
   1.563 +     * @since 1.4
   1.564 +     */
   1.565 +    public static BigInteger probablePrime(int bitLength, Random rnd) {
   1.566 +        if (bitLength < 2)
   1.567 +            throw new ArithmeticException("bitLength < 2");
   1.568 +
   1.569 +        // The cutoff of 95 was chosen empirically for best performance
   1.570 +        return (bitLength < SMALL_PRIME_THRESHOLD ?
   1.571 +                smallPrime(bitLength, DEFAULT_PRIME_CERTAINTY, rnd) :
   1.572 +                largePrime(bitLength, DEFAULT_PRIME_CERTAINTY, rnd));
   1.573 +    }
   1.574 +
   1.575 +    /**
   1.576 +     * Find a random number of the specified bitLength that is probably prime.
   1.577 +     * This method is used for smaller primes, its performance degrades on
   1.578 +     * larger bitlengths.
   1.579 +     *
   1.580 +     * This method assumes bitLength > 1.
   1.581 +     */
   1.582 +    private static BigInteger smallPrime(int bitLength, int certainty, Random rnd) {
   1.583 +        int magLen = (bitLength + 31) >>> 5;
   1.584 +        int temp[] = new int[magLen];
   1.585 +        int highBit = 1 << ((bitLength+31) & 0x1f);  // High bit of high int
   1.586 +        int highMask = (highBit << 1) - 1;  // Bits to keep in high int
   1.587 +
   1.588 +        while(true) {
   1.589 +            // Construct a candidate
   1.590 +            for (int i=0; i<magLen; i++)
   1.591 +                temp[i] = rnd.nextInt();
   1.592 +            temp[0] = (temp[0] & highMask) | highBit;  // Ensure exact length
   1.593 +            if (bitLength > 2)
   1.594 +                temp[magLen-1] |= 1;  // Make odd if bitlen > 2
   1.595 +
   1.596 +            BigInteger p = new BigInteger(temp, 1);
   1.597 +
   1.598 +            // Do cheap "pre-test" if applicable
   1.599 +            if (bitLength > 6) {
   1.600 +                long r = p.remainder(SMALL_PRIME_PRODUCT).longValue();
   1.601 +                if ((r%3==0)  || (r%5==0)  || (r%7==0)  || (r%11==0) ||
   1.602 +                    (r%13==0) || (r%17==0) || (r%19==0) || (r%23==0) ||
   1.603 +                    (r%29==0) || (r%31==0) || (r%37==0) || (r%41==0))
   1.604 +                    continue; // Candidate is composite; try another
   1.605 +            }
   1.606 +
   1.607 +            // All candidates of bitLength 2 and 3 are prime by this point
   1.608 +            if (bitLength < 4)
   1.609 +                return p;
   1.610 +
   1.611 +            // Do expensive test if we survive pre-test (or it's inapplicable)
   1.612 +            if (p.primeToCertainty(certainty, rnd))
   1.613 +                return p;
   1.614 +        }
   1.615 +    }
   1.616 +
   1.617 +    private static final BigInteger SMALL_PRIME_PRODUCT
   1.618 +                       = valueOf(3L*5*7*11*13*17*19*23*29*31*37*41);
   1.619 +
   1.620 +    /**
   1.621 +     * Find a random number of the specified bitLength that is probably prime.
   1.622 +     * This method is more appropriate for larger bitlengths since it uses
   1.623 +     * a sieve to eliminate most composites before using a more expensive
   1.624 +     * test.
   1.625 +     */
   1.626 +    private static BigInteger largePrime(int bitLength, int certainty, Random rnd) {
   1.627 +        BigInteger p;
   1.628 +        p = new BigInteger(bitLength, rnd).setBit(bitLength-1);
   1.629 +        p.mag[p.mag.length-1] &= 0xfffffffe;
   1.630 +
   1.631 +        // Use a sieve length likely to contain the next prime number
   1.632 +        int searchLen = (bitLength / 20) * 64;
   1.633 +        BitSieve searchSieve = new BitSieve(p, searchLen);
   1.634 +        BigInteger candidate = searchSieve.retrieve(p, certainty, rnd);
   1.635 +
   1.636 +        while ((candidate == null) || (candidate.bitLength() != bitLength)) {
   1.637 +            p = p.add(BigInteger.valueOf(2*searchLen));
   1.638 +            if (p.bitLength() != bitLength)
   1.639 +                p = new BigInteger(bitLength, rnd).setBit(bitLength-1);
   1.640 +            p.mag[p.mag.length-1] &= 0xfffffffe;
   1.641 +            searchSieve = new BitSieve(p, searchLen);
   1.642 +            candidate = searchSieve.retrieve(p, certainty, rnd);
   1.643 +        }
   1.644 +        return candidate;
   1.645 +    }
   1.646 +
   1.647 +   /**
   1.648 +    * Returns the first integer greater than this {@code BigInteger} that
   1.649 +    * is probably prime.  The probability that the number returned by this
   1.650 +    * method is composite does not exceed 2<sup>-100</sup>. This method will
   1.651 +    * never skip over a prime when searching: if it returns {@code p}, there
   1.652 +    * is no prime {@code q} such that {@code this < q < p}.
   1.653 +    *
   1.654 +    * @return the first integer greater than this {@code BigInteger} that
   1.655 +    *         is probably prime.
   1.656 +    * @throws ArithmeticException {@code this < 0}.
   1.657 +    * @since 1.5
   1.658 +    */
   1.659 +    public BigInteger nextProbablePrime() {
   1.660 +        if (this.signum < 0)
   1.661 +            throw new ArithmeticException("start < 0: " + this);
   1.662 +
   1.663 +        // Handle trivial cases
   1.664 +        if ((this.signum == 0) || this.equals(ONE))
   1.665 +            return TWO;
   1.666 +
   1.667 +        BigInteger result = this.add(ONE);
   1.668 +
   1.669 +        // Fastpath for small numbers
   1.670 +        if (result.bitLength() < SMALL_PRIME_THRESHOLD) {
   1.671 +
   1.672 +            // Ensure an odd number
   1.673 +            if (!result.testBit(0))
   1.674 +                result = result.add(ONE);
   1.675 +
   1.676 +            while(true) {
   1.677 +                // Do cheap "pre-test" if applicable
   1.678 +                if (result.bitLength() > 6) {
   1.679 +                    long r = result.remainder(SMALL_PRIME_PRODUCT).longValue();
   1.680 +                    if ((r%3==0)  || (r%5==0)  || (r%7==0)  || (r%11==0) ||
   1.681 +                        (r%13==0) || (r%17==0) || (r%19==0) || (r%23==0) ||
   1.682 +                        (r%29==0) || (r%31==0) || (r%37==0) || (r%41==0)) {
   1.683 +                        result = result.add(TWO);
   1.684 +                        continue; // Candidate is composite; try another
   1.685 +                    }
   1.686 +                }
   1.687 +
   1.688 +                // All candidates of bitLength 2 and 3 are prime by this point
   1.689 +                if (result.bitLength() < 4)
   1.690 +                    return result;
   1.691 +
   1.692 +                // The expensive test
   1.693 +                if (result.primeToCertainty(DEFAULT_PRIME_CERTAINTY, null))
   1.694 +                    return result;
   1.695 +
   1.696 +                result = result.add(TWO);
   1.697 +            }
   1.698 +        }
   1.699 +
   1.700 +        // Start at previous even number
   1.701 +        if (result.testBit(0))
   1.702 +            result = result.subtract(ONE);
   1.703 +
   1.704 +        // Looking for the next large prime
   1.705 +        int searchLen = (result.bitLength() / 20) * 64;
   1.706 +
   1.707 +        while(true) {
   1.708 +           BitSieve searchSieve = new BitSieve(result, searchLen);
   1.709 +           BigInteger candidate = searchSieve.retrieve(result,
   1.710 +                                                 DEFAULT_PRIME_CERTAINTY, null);
   1.711 +           if (candidate != null)
   1.712 +               return candidate;
   1.713 +           result = result.add(BigInteger.valueOf(2 * searchLen));
   1.714 +        }
   1.715 +    }
   1.716 +
   1.717 +    /**
   1.718 +     * Returns {@code true} if this BigInteger is probably prime,
   1.719 +     * {@code false} if it's definitely composite.
   1.720 +     *
   1.721 +     * This method assumes bitLength > 2.
   1.722 +     *
   1.723 +     * @param  certainty a measure of the uncertainty that the caller is
   1.724 +     *         willing to tolerate: if the call returns {@code true}
   1.725 +     *         the probability that this BigInteger is prime exceeds
   1.726 +     *         {@code (1 - 1/2<sup>certainty</sup>)}.  The execution time of
   1.727 +     *         this method is proportional to the value of this parameter.
   1.728 +     * @return {@code true} if this BigInteger is probably prime,
   1.729 +     *         {@code false} if it's definitely composite.
   1.730 +     */
   1.731 +    boolean primeToCertainty(int certainty, Random random) {
   1.732 +        int rounds = 0;
   1.733 +        int n = (Math.min(certainty, Integer.MAX_VALUE-1)+1)/2;
   1.734 +
   1.735 +        // The relationship between the certainty and the number of rounds
   1.736 +        // we perform is given in the draft standard ANSI X9.80, "PRIME
   1.737 +        // NUMBER GENERATION, PRIMALITY TESTING, AND PRIMALITY CERTIFICATES".
   1.738 +        int sizeInBits = this.bitLength();
   1.739 +        if (sizeInBits < 100) {
   1.740 +            rounds = 50;
   1.741 +            rounds = n < rounds ? n : rounds;
   1.742 +            return passesMillerRabin(rounds, random);
   1.743 +        }
   1.744 +
   1.745 +        if (sizeInBits < 256) {
   1.746 +            rounds = 27;
   1.747 +        } else if (sizeInBits < 512) {
   1.748 +            rounds = 15;
   1.749 +        } else if (sizeInBits < 768) {
   1.750 +            rounds = 8;
   1.751 +        } else if (sizeInBits < 1024) {
   1.752 +            rounds = 4;
   1.753 +        } else {
   1.754 +            rounds = 2;
   1.755 +        }
   1.756 +        rounds = n < rounds ? n : rounds;
   1.757 +
   1.758 +        return passesMillerRabin(rounds, random) && passesLucasLehmer();
   1.759 +    }
   1.760 +
   1.761 +    /**
   1.762 +     * Returns true iff this BigInteger is a Lucas-Lehmer probable prime.
   1.763 +     *
   1.764 +     * The following assumptions are made:
   1.765 +     * This BigInteger is a positive, odd number.
   1.766 +     */
   1.767 +    private boolean passesLucasLehmer() {
   1.768 +        BigInteger thisPlusOne = this.add(ONE);
   1.769 +
   1.770 +        // Step 1
   1.771 +        int d = 5;
   1.772 +        while (jacobiSymbol(d, this) != -1) {
   1.773 +            // 5, -7, 9, -11, ...
   1.774 +            d = (d<0) ? Math.abs(d)+2 : -(d+2);
   1.775 +        }
   1.776 +
   1.777 +        // Step 2
   1.778 +        BigInteger u = lucasLehmerSequence(d, thisPlusOne, this);
   1.779 +
   1.780 +        // Step 3
   1.781 +        return u.mod(this).equals(ZERO);
   1.782 +    }
   1.783 +
   1.784 +    /**
   1.785 +     * Computes Jacobi(p,n).
   1.786 +     * Assumes n positive, odd, n>=3.
   1.787 +     */
   1.788 +    private static int jacobiSymbol(int p, BigInteger n) {
   1.789 +        if (p == 0)
   1.790 +            return 0;
   1.791 +
   1.792 +        // Algorithm and comments adapted from Colin Plumb's C library.
   1.793 +        int j = 1;
   1.794 +        int u = n.mag[n.mag.length-1];
   1.795 +
   1.796 +        // Make p positive
   1.797 +        if (p < 0) {
   1.798 +            p = -p;
   1.799 +            int n8 = u & 7;
   1.800 +            if ((n8 == 3) || (n8 == 7))
   1.801 +                j = -j; // 3 (011) or 7 (111) mod 8
   1.802 +        }
   1.803 +
   1.804 +        // Get rid of factors of 2 in p
   1.805 +        while ((p & 3) == 0)
   1.806 +            p >>= 2;
   1.807 +        if ((p & 1) == 0) {
   1.808 +            p >>= 1;
   1.809 +            if (((u ^ (u>>1)) & 2) != 0)
   1.810 +                j = -j; // 3 (011) or 5 (101) mod 8
   1.811 +        }
   1.812 +        if (p == 1)
   1.813 +            return j;
   1.814 +        // Then, apply quadratic reciprocity
   1.815 +        if ((p & u & 2) != 0)   // p = u = 3 (mod 4)?
   1.816 +            j = -j;
   1.817 +        // And reduce u mod p
   1.818 +        u = n.mod(BigInteger.valueOf(p)).intValue();
   1.819 +
   1.820 +        // Now compute Jacobi(u,p), u < p
   1.821 +        while (u != 0) {
   1.822 +            while ((u & 3) == 0)
   1.823 +                u >>= 2;
   1.824 +            if ((u & 1) == 0) {
   1.825 +                u >>= 1;
   1.826 +                if (((p ^ (p>>1)) & 2) != 0)
   1.827 +                    j = -j;     // 3 (011) or 5 (101) mod 8
   1.828 +            }
   1.829 +            if (u == 1)
   1.830 +                return j;
   1.831 +            // Now both u and p are odd, so use quadratic reciprocity
   1.832 +            assert (u < p);
   1.833 +            int t = u; u = p; p = t;
   1.834 +            if ((u & p & 2) != 0) // u = p = 3 (mod 4)?
   1.835 +                j = -j;
   1.836 +            // Now u >= p, so it can be reduced
   1.837 +            u %= p;
   1.838 +        }
   1.839 +        return 0;
   1.840 +    }
   1.841 +
   1.842 +    private static BigInteger lucasLehmerSequence(int z, BigInteger k, BigInteger n) {
   1.843 +        BigInteger d = BigInteger.valueOf(z);
   1.844 +        BigInteger u = ONE; BigInteger u2;
   1.845 +        BigInteger v = ONE; BigInteger v2;
   1.846 +
   1.847 +        for (int i=k.bitLength()-2; i>=0; i--) {
   1.848 +            u2 = u.multiply(v).mod(n);
   1.849 +
   1.850 +            v2 = v.square().add(d.multiply(u.square())).mod(n);
   1.851 +            if (v2.testBit(0))
   1.852 +                v2 = v2.subtract(n);
   1.853 +
   1.854 +            v2 = v2.shiftRight(1);
   1.855 +
   1.856 +            u = u2; v = v2;
   1.857 +            if (k.testBit(i)) {
   1.858 +                u2 = u.add(v).mod(n);
   1.859 +                if (u2.testBit(0))
   1.860 +                    u2 = u2.subtract(n);
   1.861 +
   1.862 +                u2 = u2.shiftRight(1);
   1.863 +                v2 = v.add(d.multiply(u)).mod(n);
   1.864 +                if (v2.testBit(0))
   1.865 +                    v2 = v2.subtract(n);
   1.866 +                v2 = v2.shiftRight(1);
   1.867 +
   1.868 +                u = u2; v = v2;
   1.869 +            }
   1.870 +        }
   1.871 +        return u;
   1.872 +    }
   1.873 +
   1.874 +    private static volatile Random staticRandom;
   1.875 +
   1.876 +    private static Random getSecureRandom() {
   1.877 +        if (staticRandom == null) {
   1.878 +            staticRandom = new java.security.SecureRandom();
   1.879 +        }
   1.880 +        return staticRandom;
   1.881 +    }
   1.882 +
   1.883 +    /**
   1.884 +     * Returns true iff this BigInteger passes the specified number of
   1.885 +     * Miller-Rabin tests. This test is taken from the DSA spec (NIST FIPS
   1.886 +     * 186-2).
   1.887 +     *
   1.888 +     * The following assumptions are made:
   1.889 +     * This BigInteger is a positive, odd number greater than 2.
   1.890 +     * iterations<=50.
   1.891 +     */
   1.892 +    private boolean passesMillerRabin(int iterations, Random rnd) {
   1.893 +        // Find a and m such that m is odd and this == 1 + 2**a * m
   1.894 +        BigInteger thisMinusOne = this.subtract(ONE);
   1.895 +        BigInteger m = thisMinusOne;
   1.896 +        int a = m.getLowestSetBit();
   1.897 +        m = m.shiftRight(a);
   1.898 +
   1.899 +        // Do the tests
   1.900 +        if (rnd == null) {
   1.901 +            rnd = getSecureRandom();
   1.902 +        }
   1.903 +        for (int i=0; i<iterations; i++) {
   1.904 +            // Generate a uniform random on (1, this)
   1.905 +            BigInteger b;
   1.906 +            do {
   1.907 +                b = new BigInteger(this.bitLength(), rnd);
   1.908 +            } while (b.compareTo(ONE) <= 0 || b.compareTo(this) >= 0);
   1.909 +
   1.910 +            int j = 0;
   1.911 +            BigInteger z = b.modPow(m, this);
   1.912 +            while(!((j==0 && z.equals(ONE)) || z.equals(thisMinusOne))) {
   1.913 +                if (j>0 && z.equals(ONE) || ++j==a)
   1.914 +                    return false;
   1.915 +                z = z.modPow(TWO, this);
   1.916 +            }
   1.917 +        }
   1.918 +        return true;
   1.919 +    }
   1.920 +
   1.921 +    /**
   1.922 +     * This internal constructor differs from its public cousin
   1.923 +     * with the arguments reversed in two ways: it assumes that its
   1.924 +     * arguments are correct, and it doesn't copy the magnitude array.
   1.925 +     */
   1.926 +    BigInteger(int[] magnitude, int signum) {
   1.927 +        this.signum = (magnitude.length==0 ? 0 : signum);
   1.928 +        this.mag = magnitude;
   1.929 +    }
   1.930 +
   1.931 +    /**
   1.932 +     * This private constructor is for internal use and assumes that its
   1.933 +     * arguments are correct.
   1.934 +     */
   1.935 +    private BigInteger(byte[] magnitude, int signum) {
   1.936 +        this.signum = (magnitude.length==0 ? 0 : signum);
   1.937 +        this.mag = stripLeadingZeroBytes(magnitude);
   1.938 +    }
   1.939 +
   1.940 +    //Static Factory Methods
   1.941 +
   1.942 +    /**
   1.943 +     * Returns a BigInteger whose value is equal to that of the
   1.944 +     * specified {@code long}.  This "static factory method" is
   1.945 +     * provided in preference to a ({@code long}) constructor
   1.946 +     * because it allows for reuse of frequently used BigIntegers.
   1.947 +     *
   1.948 +     * @param  val value of the BigInteger to return.
   1.949 +     * @return a BigInteger with the specified value.
   1.950 +     */
   1.951 +    public static BigInteger valueOf(long val) {
   1.952 +        // If -MAX_CONSTANT < val < MAX_CONSTANT, return stashed constant
   1.953 +        if (val == 0)
   1.954 +            return ZERO;
   1.955 +        if (val > 0 && val <= MAX_CONSTANT)
   1.956 +            return posConst[(int) val];
   1.957 +        else if (val < 0 && val >= -MAX_CONSTANT)
   1.958 +            return negConst[(int) -val];
   1.959 +
   1.960 +        return new BigInteger(val);
   1.961 +    }
   1.962 +
   1.963 +    /**
   1.964 +     * Constructs a BigInteger with the specified value, which may not be zero.
   1.965 +     */
   1.966 +    private BigInteger(long val) {
   1.967 +        if (val < 0) {
   1.968 +            val = -val;
   1.969 +            signum = -1;
   1.970 +        } else {
   1.971 +            signum = 1;
   1.972 +        }
   1.973 +
   1.974 +        int highWord = (int)(val >>> 32);
   1.975 +        if (highWord==0) {
   1.976 +            mag = new int[1];
   1.977 +            mag[0] = (int)val;
   1.978 +        } else {
   1.979 +            mag = new int[2];
   1.980 +            mag[0] = highWord;
   1.981 +            mag[1] = (int)val;
   1.982 +        }
   1.983 +    }
   1.984 +
   1.985 +    /**
   1.986 +     * Returns a BigInteger with the given two's complement representation.
   1.987 +     * Assumes that the input array will not be modified (the returned
   1.988 +     * BigInteger will reference the input array if feasible).
   1.989 +     */
   1.990 +    private static BigInteger valueOf(int val[]) {
   1.991 +        return (val[0]>0 ? new BigInteger(val, 1) : new BigInteger(val));
   1.992 +    }
   1.993 +
   1.994 +    // Constants
   1.995 +
   1.996 +    /**
   1.997 +     * Initialize static constant array when class is loaded.
   1.998 +     */
   1.999 +    private final static int MAX_CONSTANT = 16;
  1.1000 +    private static BigInteger posConst[] = new BigInteger[MAX_CONSTANT+1];
  1.1001 +    private static BigInteger negConst[] = new BigInteger[MAX_CONSTANT+1];
  1.1002 +    static {
  1.1003 +        for (int i = 1; i <= MAX_CONSTANT; i++) {
  1.1004 +            int[] magnitude = new int[1];
  1.1005 +            magnitude[0] = i;
  1.1006 +            posConst[i] = new BigInteger(magnitude,  1);
  1.1007 +            negConst[i] = new BigInteger(magnitude, -1);
  1.1008 +        }
  1.1009 +    }
  1.1010 +
  1.1011 +    /**
  1.1012 +     * The BigInteger constant zero.
  1.1013 +     *
  1.1014 +     * @since   1.2
  1.1015 +     */
  1.1016 +    public static final BigInteger ZERO = new BigInteger(new int[0], 0);
  1.1017 +
  1.1018 +    /**
  1.1019 +     * The BigInteger constant one.
  1.1020 +     *
  1.1021 +     * @since   1.2
  1.1022 +     */
  1.1023 +    public static final BigInteger ONE = valueOf(1);
  1.1024 +
  1.1025 +    /**
  1.1026 +     * The BigInteger constant two.  (Not exported.)
  1.1027 +     */
  1.1028 +    private static final BigInteger TWO = valueOf(2);
  1.1029 +
  1.1030 +    /**
  1.1031 +     * The BigInteger constant ten.
  1.1032 +     *
  1.1033 +     * @since   1.5
  1.1034 +     */
  1.1035 +    public static final BigInteger TEN = valueOf(10);
  1.1036 +
  1.1037 +    // Arithmetic Operations
  1.1038 +
  1.1039 +    /**
  1.1040 +     * Returns a BigInteger whose value is {@code (this + val)}.
  1.1041 +     *
  1.1042 +     * @param  val value to be added to this BigInteger.
  1.1043 +     * @return {@code this + val}
  1.1044 +     */
  1.1045 +    public BigInteger add(BigInteger val) {
  1.1046 +        if (val.signum == 0)
  1.1047 +            return this;
  1.1048 +        if (signum == 0)
  1.1049 +            return val;
  1.1050 +        if (val.signum == signum)
  1.1051 +            return new BigInteger(add(mag, val.mag), signum);
  1.1052 +
  1.1053 +        int cmp = compareMagnitude(val);
  1.1054 +        if (cmp == 0)
  1.1055 +            return ZERO;
  1.1056 +        int[] resultMag = (cmp > 0 ? subtract(mag, val.mag)
  1.1057 +                           : subtract(val.mag, mag));
  1.1058 +        resultMag = trustedStripLeadingZeroInts(resultMag);
  1.1059 +
  1.1060 +        return new BigInteger(resultMag, cmp == signum ? 1 : -1);
  1.1061 +    }
  1.1062 +
  1.1063 +    /**
  1.1064 +     * Adds the contents of the int arrays x and y. This method allocates
  1.1065 +     * a new int array to hold the answer and returns a reference to that
  1.1066 +     * array.
  1.1067 +     */
  1.1068 +    private static int[] add(int[] x, int[] y) {
  1.1069 +        // If x is shorter, swap the two arrays
  1.1070 +        if (x.length < y.length) {
  1.1071 +            int[] tmp = x;
  1.1072 +            x = y;
  1.1073 +            y = tmp;
  1.1074 +        }
  1.1075 +
  1.1076 +        int xIndex = x.length;
  1.1077 +        int yIndex = y.length;
  1.1078 +        int result[] = new int[xIndex];
  1.1079 +        long sum = 0;
  1.1080 +
  1.1081 +        // Add common parts of both numbers
  1.1082 +        while(yIndex > 0) {
  1.1083 +            sum = (x[--xIndex] & LONG_MASK) +
  1.1084 +                  (y[--yIndex] & LONG_MASK) + (sum >>> 32);
  1.1085 +            result[xIndex] = (int)sum;
  1.1086 +        }
  1.1087 +
  1.1088 +        // Copy remainder of longer number while carry propagation is required
  1.1089 +        boolean carry = (sum >>> 32 != 0);
  1.1090 +        while (xIndex > 0 && carry)
  1.1091 +            carry = ((result[--xIndex] = x[xIndex] + 1) == 0);
  1.1092 +
  1.1093 +        // Copy remainder of longer number
  1.1094 +        while (xIndex > 0)
  1.1095 +            result[--xIndex] = x[xIndex];
  1.1096 +
  1.1097 +        // Grow result if necessary
  1.1098 +        if (carry) {
  1.1099 +            int bigger[] = new int[result.length + 1];
  1.1100 +            System.arraycopy(result, 0, bigger, 1, result.length);
  1.1101 +            bigger[0] = 0x01;
  1.1102 +            return bigger;
  1.1103 +        }
  1.1104 +        return result;
  1.1105 +    }
  1.1106 +
  1.1107 +    /**
  1.1108 +     * Returns a BigInteger whose value is {@code (this - val)}.
  1.1109 +     *
  1.1110 +     * @param  val value to be subtracted from this BigInteger.
  1.1111 +     * @return {@code this - val}
  1.1112 +     */
  1.1113 +    public BigInteger subtract(BigInteger val) {
  1.1114 +        if (val.signum == 0)
  1.1115 +            return this;
  1.1116 +        if (signum == 0)
  1.1117 +            return val.negate();
  1.1118 +        if (val.signum != signum)
  1.1119 +            return new BigInteger(add(mag, val.mag), signum);
  1.1120 +
  1.1121 +        int cmp = compareMagnitude(val);
  1.1122 +        if (cmp == 0)
  1.1123 +            return ZERO;
  1.1124 +        int[] resultMag = (cmp > 0 ? subtract(mag, val.mag)
  1.1125 +                           : subtract(val.mag, mag));
  1.1126 +        resultMag = trustedStripLeadingZeroInts(resultMag);
  1.1127 +        return new BigInteger(resultMag, cmp == signum ? 1 : -1);
  1.1128 +    }
  1.1129 +
  1.1130 +    /**
  1.1131 +     * Subtracts the contents of the second int arrays (little) from the
  1.1132 +     * first (big).  The first int array (big) must represent a larger number
  1.1133 +     * than the second.  This method allocates the space necessary to hold the
  1.1134 +     * answer.
  1.1135 +     */
  1.1136 +    private static int[] subtract(int[] big, int[] little) {
  1.1137 +        int bigIndex = big.length;
  1.1138 +        int result[] = new int[bigIndex];
  1.1139 +        int littleIndex = little.length;
  1.1140 +        long difference = 0;
  1.1141 +
  1.1142 +        // Subtract common parts of both numbers
  1.1143 +        while(littleIndex > 0) {
  1.1144 +            difference = (big[--bigIndex] & LONG_MASK) -
  1.1145 +                         (little[--littleIndex] & LONG_MASK) +
  1.1146 +                         (difference >> 32);
  1.1147 +            result[bigIndex] = (int)difference;
  1.1148 +        }
  1.1149 +
  1.1150 +        // Subtract remainder of longer number while borrow propagates
  1.1151 +        boolean borrow = (difference >> 32 != 0);
  1.1152 +        while (bigIndex > 0 && borrow)
  1.1153 +            borrow = ((result[--bigIndex] = big[bigIndex] - 1) == -1);
  1.1154 +
  1.1155 +        // Copy remainder of longer number
  1.1156 +        while (bigIndex > 0)
  1.1157 +            result[--bigIndex] = big[bigIndex];
  1.1158 +
  1.1159 +        return result;
  1.1160 +    }
  1.1161 +
  1.1162 +    /**
  1.1163 +     * Returns a BigInteger whose value is {@code (this * val)}.
  1.1164 +     *
  1.1165 +     * @param  val value to be multiplied by this BigInteger.
  1.1166 +     * @return {@code this * val}
  1.1167 +     */
  1.1168 +    public BigInteger multiply(BigInteger val) {
  1.1169 +        if (val.signum == 0 || signum == 0)
  1.1170 +            return ZERO;
  1.1171 +
  1.1172 +        int[] result = multiplyToLen(mag, mag.length,
  1.1173 +                                     val.mag, val.mag.length, null);
  1.1174 +        result = trustedStripLeadingZeroInts(result);
  1.1175 +        return new BigInteger(result, signum == val.signum ? 1 : -1);
  1.1176 +    }
  1.1177 +
  1.1178 +    /**
  1.1179 +     * Package private methods used by BigDecimal code to multiply a BigInteger
  1.1180 +     * with a long. Assumes v is not equal to INFLATED.
  1.1181 +     */
  1.1182 +    BigInteger multiply(long v) {
  1.1183 +        if (v == 0 || signum == 0)
  1.1184 +          return ZERO;
  1.1185 +        if (v == BigDecimal.INFLATED)
  1.1186 +            return multiply(BigInteger.valueOf(v));
  1.1187 +        int rsign = (v > 0 ? signum : -signum);
  1.1188 +        if (v < 0)
  1.1189 +            v = -v;
  1.1190 +        long dh = v >>> 32;      // higher order bits
  1.1191 +        long dl = v & LONG_MASK; // lower order bits
  1.1192 +
  1.1193 +        int xlen = mag.length;
  1.1194 +        int[] value = mag;
  1.1195 +        int[] rmag = (dh == 0L) ? (new int[xlen + 1]) : (new int[xlen + 2]);
  1.1196 +        long carry = 0;
  1.1197 +        int rstart = rmag.length - 1;
  1.1198 +        for (int i = xlen - 1; i >= 0; i--) {
  1.1199 +            long product = (value[i] & LONG_MASK) * dl + carry;
  1.1200 +            rmag[rstart--] = (int)product;
  1.1201 +            carry = product >>> 32;
  1.1202 +        }
  1.1203 +        rmag[rstart] = (int)carry;
  1.1204 +        if (dh != 0L) {
  1.1205 +            carry = 0;
  1.1206 +            rstart = rmag.length - 2;
  1.1207 +            for (int i = xlen - 1; i >= 0; i--) {
  1.1208 +                long product = (value[i] & LONG_MASK) * dh +
  1.1209 +                    (rmag[rstart] & LONG_MASK) + carry;
  1.1210 +                rmag[rstart--] = (int)product;
  1.1211 +                carry = product >>> 32;
  1.1212 +            }
  1.1213 +            rmag[0] = (int)carry;
  1.1214 +        }
  1.1215 +        if (carry == 0L)
  1.1216 +            rmag = java.util.Arrays.copyOfRange(rmag, 1, rmag.length);
  1.1217 +        return new BigInteger(rmag, rsign);
  1.1218 +    }
  1.1219 +
  1.1220 +    /**
  1.1221 +     * Multiplies int arrays x and y to the specified lengths and places
  1.1222 +     * the result into z. There will be no leading zeros in the resultant array.
  1.1223 +     */
  1.1224 +    private int[] multiplyToLen(int[] x, int xlen, int[] y, int ylen, int[] z) {
  1.1225 +        int xstart = xlen - 1;
  1.1226 +        int ystart = ylen - 1;
  1.1227 +
  1.1228 +        if (z == null || z.length < (xlen+ ylen))
  1.1229 +            z = new int[xlen+ylen];
  1.1230 +
  1.1231 +        long carry = 0;
  1.1232 +        for (int j=ystart, k=ystart+1+xstart; j>=0; j--, k--) {
  1.1233 +            long product = (y[j] & LONG_MASK) *
  1.1234 +                           (x[xstart] & LONG_MASK) + carry;
  1.1235 +            z[k] = (int)product;
  1.1236 +            carry = product >>> 32;
  1.1237 +        }
  1.1238 +        z[xstart] = (int)carry;
  1.1239 +
  1.1240 +        for (int i = xstart-1; i >= 0; i--) {
  1.1241 +            carry = 0;
  1.1242 +            for (int j=ystart, k=ystart+1+i; j>=0; j--, k--) {
  1.1243 +                long product = (y[j] & LONG_MASK) *
  1.1244 +                               (x[i] & LONG_MASK) +
  1.1245 +                               (z[k] & LONG_MASK) + carry;
  1.1246 +                z[k] = (int)product;
  1.1247 +                carry = product >>> 32;
  1.1248 +            }
  1.1249 +            z[i] = (int)carry;
  1.1250 +        }
  1.1251 +        return z;
  1.1252 +    }
  1.1253 +
  1.1254 +    /**
  1.1255 +     * Returns a BigInteger whose value is {@code (this<sup>2</sup>)}.
  1.1256 +     *
  1.1257 +     * @return {@code this<sup>2</sup>}
  1.1258 +     */
  1.1259 +    private BigInteger square() {
  1.1260 +        if (signum == 0)
  1.1261 +            return ZERO;
  1.1262 +        int[] z = squareToLen(mag, mag.length, null);
  1.1263 +        return new BigInteger(trustedStripLeadingZeroInts(z), 1);
  1.1264 +    }
  1.1265 +
  1.1266 +    /**
  1.1267 +     * Squares the contents of the int array x. The result is placed into the
  1.1268 +     * int array z.  The contents of x are not changed.
  1.1269 +     */
  1.1270 +    private static final int[] squareToLen(int[] x, int len, int[] z) {
  1.1271 +        /*
  1.1272 +         * The algorithm used here is adapted from Colin Plumb's C library.
  1.1273 +         * Technique: Consider the partial products in the multiplication
  1.1274 +         * of "abcde" by itself:
  1.1275 +         *
  1.1276 +         *               a  b  c  d  e
  1.1277 +         *            *  a  b  c  d  e
  1.1278 +         *          ==================
  1.1279 +         *              ae be ce de ee
  1.1280 +         *           ad bd cd dd de
  1.1281 +         *        ac bc cc cd ce
  1.1282 +         *     ab bb bc bd be
  1.1283 +         *  aa ab ac ad ae
  1.1284 +         *
  1.1285 +         * Note that everything above the main diagonal:
  1.1286 +         *              ae be ce de = (abcd) * e
  1.1287 +         *           ad bd cd       = (abc) * d
  1.1288 +         *        ac bc             = (ab) * c
  1.1289 +         *     ab                   = (a) * b
  1.1290 +         *
  1.1291 +         * is a copy of everything below the main diagonal:
  1.1292 +         *                       de
  1.1293 +         *                 cd ce
  1.1294 +         *           bc bd be
  1.1295 +         *     ab ac ad ae
  1.1296 +         *
  1.1297 +         * Thus, the sum is 2 * (off the diagonal) + diagonal.
  1.1298 +         *
  1.1299 +         * This is accumulated beginning with the diagonal (which
  1.1300 +         * consist of the squares of the digits of the input), which is then
  1.1301 +         * divided by two, the off-diagonal added, and multiplied by two
  1.1302 +         * again.  The low bit is simply a copy of the low bit of the
  1.1303 +         * input, so it doesn't need special care.
  1.1304 +         */
  1.1305 +        int zlen = len << 1;
  1.1306 +        if (z == null || z.length < zlen)
  1.1307 +            z = new int[zlen];
  1.1308 +
  1.1309 +        // Store the squares, right shifted one bit (i.e., divided by 2)
  1.1310 +        int lastProductLowWord = 0;
  1.1311 +        for (int j=0, i=0; j<len; j++) {
  1.1312 +            long piece = (x[j] & LONG_MASK);
  1.1313 +            long product = piece * piece;
  1.1314 +            z[i++] = (lastProductLowWord << 31) | (int)(product >>> 33);
  1.1315 +            z[i++] = (int)(product >>> 1);
  1.1316 +            lastProductLowWord = (int)product;
  1.1317 +        }
  1.1318 +
  1.1319 +        // Add in off-diagonal sums
  1.1320 +        for (int i=len, offset=1; i>0; i--, offset+=2) {
  1.1321 +            int t = x[i-1];
  1.1322 +            t = mulAdd(z, x, offset, i-1, t);
  1.1323 +            addOne(z, offset-1, i, t);
  1.1324 +        }
  1.1325 +
  1.1326 +        // Shift back up and set low bit
  1.1327 +        primitiveLeftShift(z, zlen, 1);
  1.1328 +        z[zlen-1] |= x[len-1] & 1;
  1.1329 +
  1.1330 +        return z;
  1.1331 +    }
  1.1332 +
  1.1333 +    /**
  1.1334 +     * Returns a BigInteger whose value is {@code (this / val)}.
  1.1335 +     *
  1.1336 +     * @param  val value by which this BigInteger is to be divided.
  1.1337 +     * @return {@code this / val}
  1.1338 +     * @throws ArithmeticException if {@code val} is zero.
  1.1339 +     */
  1.1340 +    public BigInteger divide(BigInteger val) {
  1.1341 +        MutableBigInteger q = new MutableBigInteger(),
  1.1342 +                          a = new MutableBigInteger(this.mag),
  1.1343 +                          b = new MutableBigInteger(val.mag);
  1.1344 +
  1.1345 +        a.divide(b, q);
  1.1346 +        return q.toBigInteger(this.signum == val.signum ? 1 : -1);
  1.1347 +    }
  1.1348 +
  1.1349 +    /**
  1.1350 +     * Returns an array of two BigIntegers containing {@code (this / val)}
  1.1351 +     * followed by {@code (this % val)}.
  1.1352 +     *
  1.1353 +     * @param  val value by which this BigInteger is to be divided, and the
  1.1354 +     *         remainder computed.
  1.1355 +     * @return an array of two BigIntegers: the quotient {@code (this / val)}
  1.1356 +     *         is the initial element, and the remainder {@code (this % val)}
  1.1357 +     *         is the final element.
  1.1358 +     * @throws ArithmeticException if {@code val} is zero.
  1.1359 +     */
  1.1360 +    public BigInteger[] divideAndRemainder(BigInteger val) {
  1.1361 +        BigInteger[] result = new BigInteger[2];
  1.1362 +        MutableBigInteger q = new MutableBigInteger(),
  1.1363 +                          a = new MutableBigInteger(this.mag),
  1.1364 +                          b = new MutableBigInteger(val.mag);
  1.1365 +        MutableBigInteger r = a.divide(b, q);
  1.1366 +        result[0] = q.toBigInteger(this.signum == val.signum ? 1 : -1);
  1.1367 +        result[1] = r.toBigInteger(this.signum);
  1.1368 +        return result;
  1.1369 +    }
  1.1370 +
  1.1371 +    /**
  1.1372 +     * Returns a BigInteger whose value is {@code (this % val)}.
  1.1373 +     *
  1.1374 +     * @param  val value by which this BigInteger is to be divided, and the
  1.1375 +     *         remainder computed.
  1.1376 +     * @return {@code this % val}
  1.1377 +     * @throws ArithmeticException if {@code val} is zero.
  1.1378 +     */
  1.1379 +    public BigInteger remainder(BigInteger val) {
  1.1380 +        MutableBigInteger q = new MutableBigInteger(),
  1.1381 +                          a = new MutableBigInteger(this.mag),
  1.1382 +                          b = new MutableBigInteger(val.mag);
  1.1383 +
  1.1384 +        return a.divide(b, q).toBigInteger(this.signum);
  1.1385 +    }
  1.1386 +
  1.1387 +    /**
  1.1388 +     * Returns a BigInteger whose value is <tt>(this<sup>exponent</sup>)</tt>.
  1.1389 +     * Note that {@code exponent} is an integer rather than a BigInteger.
  1.1390 +     *
  1.1391 +     * @param  exponent exponent to which this BigInteger is to be raised.
  1.1392 +     * @return <tt>this<sup>exponent</sup></tt>
  1.1393 +     * @throws ArithmeticException {@code exponent} is negative.  (This would
  1.1394 +     *         cause the operation to yield a non-integer value.)
  1.1395 +     */
  1.1396 +    public BigInteger pow(int exponent) {
  1.1397 +        if (exponent < 0)
  1.1398 +            throw new ArithmeticException("Negative exponent");
  1.1399 +        if (signum==0)
  1.1400 +            return (exponent==0 ? ONE : this);
  1.1401 +
  1.1402 +        // Perform exponentiation using repeated squaring trick
  1.1403 +        int newSign = (signum<0 && (exponent&1)==1 ? -1 : 1);
  1.1404 +        int[] baseToPow2 = this.mag;
  1.1405 +        int[] result = {1};
  1.1406 +
  1.1407 +        while (exponent != 0) {
  1.1408 +            if ((exponent & 1)==1) {
  1.1409 +                result = multiplyToLen(result, result.length,
  1.1410 +                                       baseToPow2, baseToPow2.length, null);
  1.1411 +                result = trustedStripLeadingZeroInts(result);
  1.1412 +            }
  1.1413 +            if ((exponent >>>= 1) != 0) {
  1.1414 +                baseToPow2 = squareToLen(baseToPow2, baseToPow2.length, null);
  1.1415 +                baseToPow2 = trustedStripLeadingZeroInts(baseToPow2);
  1.1416 +            }
  1.1417 +        }
  1.1418 +        return new BigInteger(result, newSign);
  1.1419 +    }
  1.1420 +
  1.1421 +    /**
  1.1422 +     * Returns a BigInteger whose value is the greatest common divisor of
  1.1423 +     * {@code abs(this)} and {@code abs(val)}.  Returns 0 if
  1.1424 +     * {@code this==0 && val==0}.
  1.1425 +     *
  1.1426 +     * @param  val value with which the GCD is to be computed.
  1.1427 +     * @return {@code GCD(abs(this), abs(val))}
  1.1428 +     */
  1.1429 +    public BigInteger gcd(BigInteger val) {
  1.1430 +        if (val.signum == 0)
  1.1431 +            return this.abs();
  1.1432 +        else if (this.signum == 0)
  1.1433 +            return val.abs();
  1.1434 +
  1.1435 +        MutableBigInteger a = new MutableBigInteger(this);
  1.1436 +        MutableBigInteger b = new MutableBigInteger(val);
  1.1437 +
  1.1438 +        MutableBigInteger result = a.hybridGCD(b);
  1.1439 +
  1.1440 +        return result.toBigInteger(1);
  1.1441 +    }
  1.1442 +
  1.1443 +    /**
  1.1444 +     * Package private method to return bit length for an integer.
  1.1445 +     */
  1.1446 +    static int bitLengthForInt(int n) {
  1.1447 +        return 32 - Integer.numberOfLeadingZeros(n);
  1.1448 +    }
  1.1449 +
  1.1450 +    /**
  1.1451 +     * Left shift int array a up to len by n bits. Returns the array that
  1.1452 +     * results from the shift since space may have to be reallocated.
  1.1453 +     */
  1.1454 +    private static int[] leftShift(int[] a, int len, int n) {
  1.1455 +        int nInts = n >>> 5;
  1.1456 +        int nBits = n&0x1F;
  1.1457 +        int bitsInHighWord = bitLengthForInt(a[0]);
  1.1458 +
  1.1459 +        // If shift can be done without recopy, do so
  1.1460 +        if (n <= (32-bitsInHighWord)) {
  1.1461 +            primitiveLeftShift(a, len, nBits);
  1.1462 +            return a;
  1.1463 +        } else { // Array must be resized
  1.1464 +            if (nBits <= (32-bitsInHighWord)) {
  1.1465 +                int result[] = new int[nInts+len];
  1.1466 +                for (int i=0; i<len; i++)
  1.1467 +                    result[i] = a[i];
  1.1468 +                primitiveLeftShift(result, result.length, nBits);
  1.1469 +                return result;
  1.1470 +            } else {
  1.1471 +                int result[] = new int[nInts+len+1];
  1.1472 +                for (int i=0; i<len; i++)
  1.1473 +                    result[i] = a[i];
  1.1474 +                primitiveRightShift(result, result.length, 32 - nBits);
  1.1475 +                return result;
  1.1476 +            }
  1.1477 +        }
  1.1478 +    }
  1.1479 +
  1.1480 +    // shifts a up to len right n bits assumes no leading zeros, 0<n<32
  1.1481 +    static void primitiveRightShift(int[] a, int len, int n) {
  1.1482 +        int n2 = 32 - n;
  1.1483 +        for (int i=len-1, c=a[i]; i>0; i--) {
  1.1484 +            int b = c;
  1.1485 +            c = a[i-1];
  1.1486 +            a[i] = (c << n2) | (b >>> n);
  1.1487 +        }
  1.1488 +        a[0] >>>= n;
  1.1489 +    }
  1.1490 +
  1.1491 +    // shifts a up to len left n bits assumes no leading zeros, 0<=n<32
  1.1492 +    static void primitiveLeftShift(int[] a, int len, int n) {
  1.1493 +        if (len == 0 || n == 0)
  1.1494 +            return;
  1.1495 +
  1.1496 +        int n2 = 32 - n;
  1.1497 +        for (int i=0, c=a[i], m=i+len-1; i<m; i++) {
  1.1498 +            int b = c;
  1.1499 +            c = a[i+1];
  1.1500 +            a[i] = (b << n) | (c >>> n2);
  1.1501 +        }
  1.1502 +        a[len-1] <<= n;
  1.1503 +    }
  1.1504 +
  1.1505 +    /**
  1.1506 +     * Calculate bitlength of contents of the first len elements an int array,
  1.1507 +     * assuming there are no leading zero ints.
  1.1508 +     */
  1.1509 +    private static int bitLength(int[] val, int len) {
  1.1510 +        if (len == 0)
  1.1511 +            return 0;
  1.1512 +        return ((len - 1) << 5) + bitLengthForInt(val[0]);
  1.1513 +    }
  1.1514 +
  1.1515 +    /**
  1.1516 +     * Returns a BigInteger whose value is the absolute value of this
  1.1517 +     * BigInteger.
  1.1518 +     *
  1.1519 +     * @return {@code abs(this)}
  1.1520 +     */
  1.1521 +    public BigInteger abs() {
  1.1522 +        return (signum >= 0 ? this : this.negate());
  1.1523 +    }
  1.1524 +
  1.1525 +    /**
  1.1526 +     * Returns a BigInteger whose value is {@code (-this)}.
  1.1527 +     *
  1.1528 +     * @return {@code -this}
  1.1529 +     */
  1.1530 +    public BigInteger negate() {
  1.1531 +        return new BigInteger(this.mag, -this.signum);
  1.1532 +    }
  1.1533 +
  1.1534 +    /**
  1.1535 +     * Returns the signum function of this BigInteger.
  1.1536 +     *
  1.1537 +     * @return -1, 0 or 1 as the value of this BigInteger is negative, zero or
  1.1538 +     *         positive.
  1.1539 +     */
  1.1540 +    public int signum() {
  1.1541 +        return this.signum;
  1.1542 +    }
  1.1543 +
  1.1544 +    // Modular Arithmetic Operations
  1.1545 +
  1.1546 +    /**
  1.1547 +     * Returns a BigInteger whose value is {@code (this mod m}).  This method
  1.1548 +     * differs from {@code remainder} in that it always returns a
  1.1549 +     * <i>non-negative</i> BigInteger.
  1.1550 +     *
  1.1551 +     * @param  m the modulus.
  1.1552 +     * @return {@code this mod m}
  1.1553 +     * @throws ArithmeticException {@code m} &le; 0
  1.1554 +     * @see    #remainder
  1.1555 +     */
  1.1556 +    public BigInteger mod(BigInteger m) {
  1.1557 +        if (m.signum <= 0)
  1.1558 +            throw new ArithmeticException("BigInteger: modulus not positive");
  1.1559 +
  1.1560 +        BigInteger result = this.remainder(m);
  1.1561 +        return (result.signum >= 0 ? result : result.add(m));
  1.1562 +    }
  1.1563 +
  1.1564 +    /**
  1.1565 +     * Returns a BigInteger whose value is
  1.1566 +     * <tt>(this<sup>exponent</sup> mod m)</tt>.  (Unlike {@code pow}, this
  1.1567 +     * method permits negative exponents.)
  1.1568 +     *
  1.1569 +     * @param  exponent the exponent.
  1.1570 +     * @param  m the modulus.
  1.1571 +     * @return <tt>this<sup>exponent</sup> mod m</tt>
  1.1572 +     * @throws ArithmeticException {@code m} &le; 0 or the exponent is
  1.1573 +     *         negative and this BigInteger is not <i>relatively
  1.1574 +     *         prime</i> to {@code m}.
  1.1575 +     * @see    #modInverse
  1.1576 +     */
  1.1577 +    public BigInteger modPow(BigInteger exponent, BigInteger m) {
  1.1578 +        if (m.signum <= 0)
  1.1579 +            throw new ArithmeticException("BigInteger: modulus not positive");
  1.1580 +
  1.1581 +        // Trivial cases
  1.1582 +        if (exponent.signum == 0)
  1.1583 +            return (m.equals(ONE) ? ZERO : ONE);
  1.1584 +
  1.1585 +        if (this.equals(ONE))
  1.1586 +            return (m.equals(ONE) ? ZERO : ONE);
  1.1587 +
  1.1588 +        if (this.equals(ZERO) && exponent.signum >= 0)
  1.1589 +            return ZERO;
  1.1590 +
  1.1591 +        if (this.equals(negConst[1]) && (!exponent.testBit(0)))
  1.1592 +            return (m.equals(ONE) ? ZERO : ONE);
  1.1593 +
  1.1594 +        boolean invertResult;
  1.1595 +        if ((invertResult = (exponent.signum < 0)))
  1.1596 +            exponent = exponent.negate();
  1.1597 +
  1.1598 +        BigInteger base = (this.signum < 0 || this.compareTo(m) >= 0
  1.1599 +                           ? this.mod(m) : this);
  1.1600 +        BigInteger result;
  1.1601 +        if (m.testBit(0)) { // odd modulus
  1.1602 +            result = base.oddModPow(exponent, m);
  1.1603 +        } else {
  1.1604 +            /*
  1.1605 +             * Even modulus.  Tear it into an "odd part" (m1) and power of two
  1.1606 +             * (m2), exponentiate mod m1, manually exponentiate mod m2, and
  1.1607 +             * use Chinese Remainder Theorem to combine results.
  1.1608 +             */
  1.1609 +
  1.1610 +            // Tear m apart into odd part (m1) and power of 2 (m2)
  1.1611 +            int p = m.getLowestSetBit();   // Max pow of 2 that divides m
  1.1612 +
  1.1613 +            BigInteger m1 = m.shiftRight(p);  // m/2**p
  1.1614 +            BigInteger m2 = ONE.shiftLeft(p); // 2**p
  1.1615 +
  1.1616 +            // Calculate new base from m1
  1.1617 +            BigInteger base2 = (this.signum < 0 || this.compareTo(m1) >= 0
  1.1618 +                                ? this.mod(m1) : this);
  1.1619 +
  1.1620 +            // Caculate (base ** exponent) mod m1.
  1.1621 +            BigInteger a1 = (m1.equals(ONE) ? ZERO :
  1.1622 +                             base2.oddModPow(exponent, m1));
  1.1623 +
  1.1624 +            // Calculate (this ** exponent) mod m2
  1.1625 +            BigInteger a2 = base.modPow2(exponent, p);
  1.1626 +
  1.1627 +            // Combine results using Chinese Remainder Theorem
  1.1628 +            BigInteger y1 = m2.modInverse(m1);
  1.1629 +            BigInteger y2 = m1.modInverse(m2);
  1.1630 +
  1.1631 +            result = a1.multiply(m2).multiply(y1).add
  1.1632 +                     (a2.multiply(m1).multiply(y2)).mod(m);
  1.1633 +        }
  1.1634 +
  1.1635 +        return (invertResult ? result.modInverse(m) : result);
  1.1636 +    }
  1.1637 +
  1.1638 +    static int[] bnExpModThreshTable = {7, 25, 81, 241, 673, 1793,
  1.1639 +                                                Integer.MAX_VALUE}; // Sentinel
  1.1640 +
  1.1641 +    /**
  1.1642 +     * Returns a BigInteger whose value is x to the power of y mod z.
  1.1643 +     * Assumes: z is odd && x < z.
  1.1644 +     */
  1.1645 +    private BigInteger oddModPow(BigInteger y, BigInteger z) {
  1.1646 +    /*
  1.1647 +     * The algorithm is adapted from Colin Plumb's C library.
  1.1648 +     *
  1.1649 +     * The window algorithm:
  1.1650 +     * The idea is to keep a running product of b1 = n^(high-order bits of exp)
  1.1651 +     * and then keep appending exponent bits to it.  The following patterns
  1.1652 +     * apply to a 3-bit window (k = 3):
  1.1653 +     * To append   0: square
  1.1654 +     * To append   1: square, multiply by n^1
  1.1655 +     * To append  10: square, multiply by n^1, square
  1.1656 +     * To append  11: square, square, multiply by n^3
  1.1657 +     * To append 100: square, multiply by n^1, square, square
  1.1658 +     * To append 101: square, square, square, multiply by n^5
  1.1659 +     * To append 110: square, square, multiply by n^3, square
  1.1660 +     * To append 111: square, square, square, multiply by n^7
  1.1661 +     *
  1.1662 +     * Since each pattern involves only one multiply, the longer the pattern
  1.1663 +     * the better, except that a 0 (no multiplies) can be appended directly.
  1.1664 +     * We precompute a table of odd powers of n, up to 2^k, and can then
  1.1665 +     * multiply k bits of exponent at a time.  Actually, assuming random
  1.1666 +     * exponents, there is on average one zero bit between needs to
  1.1667 +     * multiply (1/2 of the time there's none, 1/4 of the time there's 1,
  1.1668 +     * 1/8 of the time, there's 2, 1/32 of the time, there's 3, etc.), so
  1.1669 +     * you have to do one multiply per k+1 bits of exponent.
  1.1670 +     *
  1.1671 +     * The loop walks down the exponent, squaring the result buffer as
  1.1672 +     * it goes.  There is a wbits+1 bit lookahead buffer, buf, that is
  1.1673 +     * filled with the upcoming exponent bits.  (What is read after the
  1.1674 +     * end of the exponent is unimportant, but it is filled with zero here.)
  1.1675 +     * When the most-significant bit of this buffer becomes set, i.e.
  1.1676 +     * (buf & tblmask) != 0, we have to decide what pattern to multiply
  1.1677 +     * by, and when to do it.  We decide, remember to do it in future
  1.1678 +     * after a suitable number of squarings have passed (e.g. a pattern
  1.1679 +     * of "100" in the buffer requires that we multiply by n^1 immediately;
  1.1680 +     * a pattern of "110" calls for multiplying by n^3 after one more
  1.1681 +     * squaring), clear the buffer, and continue.
  1.1682 +     *
  1.1683 +     * When we start, there is one more optimization: the result buffer
  1.1684 +     * is implcitly one, so squaring it or multiplying by it can be
  1.1685 +     * optimized away.  Further, if we start with a pattern like "100"
  1.1686 +     * in the lookahead window, rather than placing n into the buffer
  1.1687 +     * and then starting to square it, we have already computed n^2
  1.1688 +     * to compute the odd-powers table, so we can place that into
  1.1689 +     * the buffer and save a squaring.
  1.1690 +     *
  1.1691 +     * This means that if you have a k-bit window, to compute n^z,
  1.1692 +     * where z is the high k bits of the exponent, 1/2 of the time
  1.1693 +     * it requires no squarings.  1/4 of the time, it requires 1
  1.1694 +     * squaring, ... 1/2^(k-1) of the time, it reqires k-2 squarings.
  1.1695 +     * And the remaining 1/2^(k-1) of the time, the top k bits are a
  1.1696 +     * 1 followed by k-1 0 bits, so it again only requires k-2
  1.1697 +     * squarings, not k-1.  The average of these is 1.  Add that
  1.1698 +     * to the one squaring we have to do to compute the table,
  1.1699 +     * and you'll see that a k-bit window saves k-2 squarings
  1.1700 +     * as well as reducing the multiplies.  (It actually doesn't
  1.1701 +     * hurt in the case k = 1, either.)
  1.1702 +     */
  1.1703 +        // Special case for exponent of one
  1.1704 +        if (y.equals(ONE))
  1.1705 +            return this;
  1.1706 +
  1.1707 +        // Special case for base of zero
  1.1708 +        if (signum==0)
  1.1709 +            return ZERO;
  1.1710 +
  1.1711 +        int[] base = mag.clone();
  1.1712 +        int[] exp = y.mag;
  1.1713 +        int[] mod = z.mag;
  1.1714 +        int modLen = mod.length;
  1.1715 +
  1.1716 +        // Select an appropriate window size
  1.1717 +        int wbits = 0;
  1.1718 +        int ebits = bitLength(exp, exp.length);
  1.1719 +        // if exponent is 65537 (0x10001), use minimum window size
  1.1720 +        if ((ebits != 17) || (exp[0] != 65537)) {
  1.1721 +            while (ebits > bnExpModThreshTable[wbits]) {
  1.1722 +                wbits++;
  1.1723 +            }
  1.1724 +        }
  1.1725 +
  1.1726 +        // Calculate appropriate table size
  1.1727 +        int tblmask = 1 << wbits;
  1.1728 +
  1.1729 +        // Allocate table for precomputed odd powers of base in Montgomery form
  1.1730 +        int[][] table = new int[tblmask][];
  1.1731 +        for (int i=0; i<tblmask; i++)
  1.1732 +            table[i] = new int[modLen];
  1.1733 +
  1.1734 +        // Compute the modular inverse
  1.1735 +        int inv = -MutableBigInteger.inverseMod32(mod[modLen-1]);
  1.1736 +
  1.1737 +        // Convert base to Montgomery form
  1.1738 +        int[] a = leftShift(base, base.length, modLen << 5);
  1.1739 +
  1.1740 +        MutableBigInteger q = new MutableBigInteger(),
  1.1741 +                          a2 = new MutableBigInteger(a),
  1.1742 +                          b2 = new MutableBigInteger(mod);
  1.1743 +
  1.1744 +        MutableBigInteger r= a2.divide(b2, q);
  1.1745 +        table[0] = r.toIntArray();
  1.1746 +
  1.1747 +        // Pad table[0] with leading zeros so its length is at least modLen
  1.1748 +        if (table[0].length < modLen) {
  1.1749 +           int offset = modLen - table[0].length;
  1.1750 +           int[] t2 = new int[modLen];
  1.1751 +           for (int i=0; i<table[0].length; i++)
  1.1752 +               t2[i+offset] = table[0][i];
  1.1753 +           table[0] = t2;
  1.1754 +        }
  1.1755 +
  1.1756 +        // Set b to the square of the base
  1.1757 +        int[] b = squareToLen(table[0], modLen, null);
  1.1758 +        b = montReduce(b, mod, modLen, inv);
  1.1759 +
  1.1760 +        // Set t to high half of b
  1.1761 +        int[] t = new int[modLen];
  1.1762 +        for(int i=0; i<modLen; i++)
  1.1763 +            t[i] = b[i];
  1.1764 +
  1.1765 +        // Fill in the table with odd powers of the base
  1.1766 +        for (int i=1; i<tblmask; i++) {
  1.1767 +            int[] prod = multiplyToLen(t, modLen, table[i-1], modLen, null);
  1.1768 +            table[i] = montReduce(prod, mod, modLen, inv);
  1.1769 +        }
  1.1770 +
  1.1771 +        // Pre load the window that slides over the exponent
  1.1772 +        int bitpos = 1 << ((ebits-1) & (32-1));
  1.1773 +
  1.1774 +        int buf = 0;
  1.1775 +        int elen = exp.length;
  1.1776 +        int eIndex = 0;
  1.1777 +        for (int i = 0; i <= wbits; i++) {
  1.1778 +            buf = (buf << 1) | (((exp[eIndex] & bitpos) != 0)?1:0);
  1.1779 +            bitpos >>>= 1;
  1.1780 +            if (bitpos == 0) {
  1.1781 +                eIndex++;
  1.1782 +                bitpos = 1 << (32-1);
  1.1783 +                elen--;
  1.1784 +            }
  1.1785 +        }
  1.1786 +
  1.1787 +        int multpos = ebits;
  1.1788 +
  1.1789 +        // The first iteration, which is hoisted out of the main loop
  1.1790 +        ebits--;
  1.1791 +        boolean isone = true;
  1.1792 +
  1.1793 +        multpos = ebits - wbits;
  1.1794 +        while ((buf & 1) == 0) {
  1.1795 +            buf >>>= 1;
  1.1796 +            multpos++;
  1.1797 +        }
  1.1798 +
  1.1799 +        int[] mult = table[buf >>> 1];
  1.1800 +
  1.1801 +        buf = 0;
  1.1802 +        if (multpos == ebits)
  1.1803 +            isone = false;
  1.1804 +
  1.1805 +        // The main loop
  1.1806 +        while(true) {
  1.1807 +            ebits--;
  1.1808 +            // Advance the window
  1.1809 +            buf <<= 1;
  1.1810 +
  1.1811 +            if (elen != 0) {
  1.1812 +                buf |= ((exp[eIndex] & bitpos) != 0) ? 1 : 0;
  1.1813 +                bitpos >>>= 1;
  1.1814 +                if (bitpos == 0) {
  1.1815 +                    eIndex++;
  1.1816 +                    bitpos = 1 << (32-1);
  1.1817 +                    elen--;
  1.1818 +                }
  1.1819 +            }
  1.1820 +
  1.1821 +            // Examine the window for pending multiplies
  1.1822 +            if ((buf & tblmask) != 0) {
  1.1823 +                multpos = ebits - wbits;
  1.1824 +                while ((buf & 1) == 0) {
  1.1825 +                    buf >>>= 1;
  1.1826 +                    multpos++;
  1.1827 +                }
  1.1828 +                mult = table[buf >>> 1];
  1.1829 +                buf = 0;
  1.1830 +            }
  1.1831 +
  1.1832 +            // Perform multiply
  1.1833 +            if (ebits == multpos) {
  1.1834 +                if (isone) {
  1.1835 +                    b = mult.clone();
  1.1836 +                    isone = false;
  1.1837 +                } else {
  1.1838 +                    t = b;
  1.1839 +                    a = multiplyToLen(t, modLen, mult, modLen, a);
  1.1840 +                    a = montReduce(a, mod, modLen, inv);
  1.1841 +                    t = a; a = b; b = t;
  1.1842 +                }
  1.1843 +            }
  1.1844 +
  1.1845 +            // Check if done
  1.1846 +            if (ebits == 0)
  1.1847 +                break;
  1.1848 +
  1.1849 +            // Square the input
  1.1850 +            if (!isone) {
  1.1851 +                t = b;
  1.1852 +                a = squareToLen(t, modLen, a);
  1.1853 +                a = montReduce(a, mod, modLen, inv);
  1.1854 +                t = a; a = b; b = t;
  1.1855 +            }
  1.1856 +        }
  1.1857 +
  1.1858 +        // Convert result out of Montgomery form and return
  1.1859 +        int[] t2 = new int[2*modLen];
  1.1860 +        for(int i=0; i<modLen; i++)
  1.1861 +            t2[i+modLen] = b[i];
  1.1862 +
  1.1863 +        b = montReduce(t2, mod, modLen, inv);
  1.1864 +
  1.1865 +        t2 = new int[modLen];
  1.1866 +        for(int i=0; i<modLen; i++)
  1.1867 +            t2[i] = b[i];
  1.1868 +
  1.1869 +        return new BigInteger(1, t2);
  1.1870 +    }
  1.1871 +
  1.1872 +    /**
  1.1873 +     * Montgomery reduce n, modulo mod.  This reduces modulo mod and divides
  1.1874 +     * by 2^(32*mlen). Adapted from Colin Plumb's C library.
  1.1875 +     */
  1.1876 +    private static int[] montReduce(int[] n, int[] mod, int mlen, int inv) {
  1.1877 +        int c=0;
  1.1878 +        int len = mlen;
  1.1879 +        int offset=0;
  1.1880 +
  1.1881 +        do {
  1.1882 +            int nEnd = n[n.length-1-offset];
  1.1883 +            int carry = mulAdd(n, mod, offset, mlen, inv * nEnd);
  1.1884 +            c += addOne(n, offset, mlen, carry);
  1.1885 +            offset++;
  1.1886 +        } while(--len > 0);
  1.1887 +
  1.1888 +        while(c>0)
  1.1889 +            c += subN(n, mod, mlen);
  1.1890 +
  1.1891 +        while (intArrayCmpToLen(n, mod, mlen) >= 0)
  1.1892 +            subN(n, mod, mlen);
  1.1893 +
  1.1894 +        return n;
  1.1895 +    }
  1.1896 +
  1.1897 +
  1.1898 +    /*
  1.1899 +     * Returns -1, 0 or +1 as big-endian unsigned int array arg1 is less than,
  1.1900 +     * equal to, or greater than arg2 up to length len.
  1.1901 +     */
  1.1902 +    private static int intArrayCmpToLen(int[] arg1, int[] arg2, int len) {
  1.1903 +        for (int i=0; i<len; i++) {
  1.1904 +            long b1 = arg1[i] & LONG_MASK;
  1.1905 +            long b2 = arg2[i] & LONG_MASK;
  1.1906 +            if (b1 < b2)
  1.1907 +                return -1;
  1.1908 +            if (b1 > b2)
  1.1909 +                return 1;
  1.1910 +        }
  1.1911 +        return 0;
  1.1912 +    }
  1.1913 +
  1.1914 +    /**
  1.1915 +     * Subtracts two numbers of same length, returning borrow.
  1.1916 +     */
  1.1917 +    private static int subN(int[] a, int[] b, int len) {
  1.1918 +        long sum = 0;
  1.1919 +
  1.1920 +        while(--len >= 0) {
  1.1921 +            sum = (a[len] & LONG_MASK) -
  1.1922 +                 (b[len] & LONG_MASK) + (sum >> 32);
  1.1923 +            a[len] = (int)sum;
  1.1924 +        }
  1.1925 +
  1.1926 +        return (int)(sum >> 32);
  1.1927 +    }
  1.1928 +
  1.1929 +    /**
  1.1930 +     * Multiply an array by one word k and add to result, return the carry
  1.1931 +     */
  1.1932 +    static int mulAdd(int[] out, int[] in, int offset, int len, int k) {
  1.1933 +        long kLong = k & LONG_MASK;
  1.1934 +        long carry = 0;
  1.1935 +
  1.1936 +        offset = out.length-offset - 1;
  1.1937 +        for (int j=len-1; j >= 0; j--) {
  1.1938 +            long product = (in[j] & LONG_MASK) * kLong +
  1.1939 +                           (out[offset] & LONG_MASK) + carry;
  1.1940 +            out[offset--] = (int)product;
  1.1941 +            carry = product >>> 32;
  1.1942 +        }
  1.1943 +        return (int)carry;
  1.1944 +    }
  1.1945 +
  1.1946 +    /**
  1.1947 +     * Add one word to the number a mlen words into a. Return the resulting
  1.1948 +     * carry.
  1.1949 +     */
  1.1950 +    static int addOne(int[] a, int offset, int mlen, int carry) {
  1.1951 +        offset = a.length-1-mlen-offset;
  1.1952 +        long t = (a[offset] & LONG_MASK) + (carry & LONG_MASK);
  1.1953 +
  1.1954 +        a[offset] = (int)t;
  1.1955 +        if ((t >>> 32) == 0)
  1.1956 +            return 0;
  1.1957 +        while (--mlen >= 0) {
  1.1958 +            if (--offset < 0) { // Carry out of number
  1.1959 +                return 1;
  1.1960 +            } else {
  1.1961 +                a[offset]++;
  1.1962 +                if (a[offset] != 0)
  1.1963 +                    return 0;
  1.1964 +            }
  1.1965 +        }
  1.1966 +        return 1;
  1.1967 +    }
  1.1968 +
  1.1969 +    /**
  1.1970 +     * Returns a BigInteger whose value is (this ** exponent) mod (2**p)
  1.1971 +     */
  1.1972 +    private BigInteger modPow2(BigInteger exponent, int p) {
  1.1973 +        /*
  1.1974 +         * Perform exponentiation using repeated squaring trick, chopping off
  1.1975 +         * high order bits as indicated by modulus.
  1.1976 +         */
  1.1977 +        BigInteger result = valueOf(1);
  1.1978 +        BigInteger baseToPow2 = this.mod2(p);
  1.1979 +        int expOffset = 0;
  1.1980 +
  1.1981 +        int limit = exponent.bitLength();
  1.1982 +
  1.1983 +        if (this.testBit(0))
  1.1984 +           limit = (p-1) < limit ? (p-1) : limit;
  1.1985 +
  1.1986 +        while (expOffset < limit) {
  1.1987 +            if (exponent.testBit(expOffset))
  1.1988 +                result = result.multiply(baseToPow2).mod2(p);
  1.1989 +            expOffset++;
  1.1990 +            if (expOffset < limit)
  1.1991 +                baseToPow2 = baseToPow2.square().mod2(p);
  1.1992 +        }
  1.1993 +
  1.1994 +        return result;
  1.1995 +    }
  1.1996 +
  1.1997 +    /**
  1.1998 +     * Returns a BigInteger whose value is this mod(2**p).
  1.1999 +     * Assumes that this {@code BigInteger >= 0} and {@code p > 0}.
  1.2000 +     */
  1.2001 +    private BigInteger mod2(int p) {
  1.2002 +        if (bitLength() <= p)
  1.2003 +            return this;
  1.2004 +
  1.2005 +        // Copy remaining ints of mag
  1.2006 +        int numInts = (p + 31) >>> 5;
  1.2007 +        int[] mag = new int[numInts];
  1.2008 +        for (int i=0; i<numInts; i++)
  1.2009 +            mag[i] = this.mag[i + (this.mag.length - numInts)];
  1.2010 +
  1.2011 +        // Mask out any excess bits
  1.2012 +        int excessBits = (numInts << 5) - p;
  1.2013 +        mag[0] &= (1L << (32-excessBits)) - 1;
  1.2014 +
  1.2015 +        return (mag[0]==0 ? new BigInteger(1, mag) : new BigInteger(mag, 1));
  1.2016 +    }
  1.2017 +
  1.2018 +    /**
  1.2019 +     * Returns a BigInteger whose value is {@code (this}<sup>-1</sup> {@code mod m)}.
  1.2020 +     *
  1.2021 +     * @param  m the modulus.
  1.2022 +     * @return {@code this}<sup>-1</sup> {@code mod m}.
  1.2023 +     * @throws ArithmeticException {@code  m} &le; 0, or this BigInteger
  1.2024 +     *         has no multiplicative inverse mod m (that is, this BigInteger
  1.2025 +     *         is not <i>relatively prime</i> to m).
  1.2026 +     */
  1.2027 +    public BigInteger modInverse(BigInteger m) {
  1.2028 +        if (m.signum != 1)
  1.2029 +            throw new ArithmeticException("BigInteger: modulus not positive");
  1.2030 +
  1.2031 +        if (m.equals(ONE))
  1.2032 +            return ZERO;
  1.2033 +
  1.2034 +        // Calculate (this mod m)
  1.2035 +        BigInteger modVal = this;
  1.2036 +        if (signum < 0 || (this.compareMagnitude(m) >= 0))
  1.2037 +            modVal = this.mod(m);
  1.2038 +
  1.2039 +        if (modVal.equals(ONE))
  1.2040 +            return ONE;
  1.2041 +
  1.2042 +        MutableBigInteger a = new MutableBigInteger(modVal);
  1.2043 +        MutableBigInteger b = new MutableBigInteger(m);
  1.2044 +
  1.2045 +        MutableBigInteger result = a.mutableModInverse(b);
  1.2046 +        return result.toBigInteger(1);
  1.2047 +    }
  1.2048 +
  1.2049 +    // Shift Operations
  1.2050 +
  1.2051 +    /**
  1.2052 +     * Returns a BigInteger whose value is {@code (this << n)}.
  1.2053 +     * The shift distance, {@code n}, may be negative, in which case
  1.2054 +     * this method performs a right shift.
  1.2055 +     * (Computes <tt>floor(this * 2<sup>n</sup>)</tt>.)
  1.2056 +     *
  1.2057 +     * @param  n shift distance, in bits.
  1.2058 +     * @return {@code this << n}
  1.2059 +     * @throws ArithmeticException if the shift distance is {@code
  1.2060 +     *         Integer.MIN_VALUE}.
  1.2061 +     * @see #shiftRight
  1.2062 +     */
  1.2063 +    public BigInteger shiftLeft(int n) {
  1.2064 +        if (signum == 0)
  1.2065 +            return ZERO;
  1.2066 +        if (n==0)
  1.2067 +            return this;
  1.2068 +        if (n<0) {
  1.2069 +            if (n == Integer.MIN_VALUE) {
  1.2070 +                throw new ArithmeticException("Shift distance of Integer.MIN_VALUE not supported.");
  1.2071 +            } else {
  1.2072 +                return shiftRight(-n);
  1.2073 +            }
  1.2074 +        }
  1.2075 +
  1.2076 +        int nInts = n >>> 5;
  1.2077 +        int nBits = n & 0x1f;
  1.2078 +        int magLen = mag.length;
  1.2079 +        int newMag[] = null;
  1.2080 +
  1.2081 +        if (nBits == 0) {
  1.2082 +            newMag = new int[magLen + nInts];
  1.2083 +            for (int i=0; i<magLen; i++)
  1.2084 +                newMag[i] = mag[i];
  1.2085 +        } else {
  1.2086 +            int i = 0;
  1.2087 +            int nBits2 = 32 - nBits;
  1.2088 +            int highBits = mag[0] >>> nBits2;
  1.2089 +            if (highBits != 0) {
  1.2090 +                newMag = new int[magLen + nInts + 1];
  1.2091 +                newMag[i++] = highBits;
  1.2092 +            } else {
  1.2093 +                newMag = new int[magLen + nInts];
  1.2094 +            }
  1.2095 +            int j=0;
  1.2096 +            while (j < magLen-1)
  1.2097 +                newMag[i++] = mag[j++] << nBits | mag[j] >>> nBits2;
  1.2098 +            newMag[i] = mag[j] << nBits;
  1.2099 +        }
  1.2100 +
  1.2101 +        return new BigInteger(newMag, signum);
  1.2102 +    }
  1.2103 +
  1.2104 +    /**
  1.2105 +     * Returns a BigInteger whose value is {@code (this >> n)}.  Sign
  1.2106 +     * extension is performed.  The shift distance, {@code n}, may be
  1.2107 +     * negative, in which case this method performs a left shift.
  1.2108 +     * (Computes <tt>floor(this / 2<sup>n</sup>)</tt>.)
  1.2109 +     *
  1.2110 +     * @param  n shift distance, in bits.
  1.2111 +     * @return {@code this >> n}
  1.2112 +     * @throws ArithmeticException if the shift distance is {@code
  1.2113 +     *         Integer.MIN_VALUE}.
  1.2114 +     * @see #shiftLeft
  1.2115 +     */
  1.2116 +    public BigInteger shiftRight(int n) {
  1.2117 +        if (n==0)
  1.2118 +            return this;
  1.2119 +        if (n<0) {
  1.2120 +            if (n == Integer.MIN_VALUE) {
  1.2121 +                throw new ArithmeticException("Shift distance of Integer.MIN_VALUE not supported.");
  1.2122 +            } else {
  1.2123 +                return shiftLeft(-n);
  1.2124 +            }
  1.2125 +        }
  1.2126 +
  1.2127 +        int nInts = n >>> 5;
  1.2128 +        int nBits = n & 0x1f;
  1.2129 +        int magLen = mag.length;
  1.2130 +        int newMag[] = null;
  1.2131 +
  1.2132 +        // Special case: entire contents shifted off the end
  1.2133 +        if (nInts >= magLen)
  1.2134 +            return (signum >= 0 ? ZERO : negConst[1]);
  1.2135 +
  1.2136 +        if (nBits == 0) {
  1.2137 +            int newMagLen = magLen - nInts;
  1.2138 +            newMag = new int[newMagLen];
  1.2139 +            for (int i=0; i<newMagLen; i++)
  1.2140 +                newMag[i] = mag[i];
  1.2141 +        } else {
  1.2142 +            int i = 0;
  1.2143 +            int highBits = mag[0] >>> nBits;
  1.2144 +            if (highBits != 0) {
  1.2145 +                newMag = new int[magLen - nInts];
  1.2146 +                newMag[i++] = highBits;
  1.2147 +            } else {
  1.2148 +                newMag = new int[magLen - nInts -1];
  1.2149 +            }
  1.2150 +
  1.2151 +            int nBits2 = 32 - nBits;
  1.2152 +            int j=0;
  1.2153 +            while (j < magLen - nInts - 1)
  1.2154 +                newMag[i++] = (mag[j++] << nBits2) | (mag[j] >>> nBits);
  1.2155 +        }
  1.2156 +
  1.2157 +        if (signum < 0) {
  1.2158 +            // Find out whether any one-bits were shifted off the end.
  1.2159 +            boolean onesLost = false;
  1.2160 +            for (int i=magLen-1, j=magLen-nInts; i>=j && !onesLost; i--)
  1.2161 +                onesLost = (mag[i] != 0);
  1.2162 +            if (!onesLost && nBits != 0)
  1.2163 +                onesLost = (mag[magLen - nInts - 1] << (32 - nBits) != 0);
  1.2164 +
  1.2165 +            if (onesLost)
  1.2166 +                newMag = javaIncrement(newMag);
  1.2167 +        }
  1.2168 +
  1.2169 +        return new BigInteger(newMag, signum);
  1.2170 +    }
  1.2171 +
  1.2172 +    int[] javaIncrement(int[] val) {
  1.2173 +        int lastSum = 0;
  1.2174 +        for (int i=val.length-1;  i >= 0 && lastSum == 0; i--)
  1.2175 +            lastSum = (val[i] += 1);
  1.2176 +        if (lastSum == 0) {
  1.2177 +            val = new int[val.length+1];
  1.2178 +            val[0] = 1;
  1.2179 +        }
  1.2180 +        return val;
  1.2181 +    }
  1.2182 +
  1.2183 +    // Bitwise Operations
  1.2184 +
  1.2185 +    /**
  1.2186 +     * Returns a BigInteger whose value is {@code (this & val)}.  (This
  1.2187 +     * method returns a negative BigInteger if and only if this and val are
  1.2188 +     * both negative.)
  1.2189 +     *
  1.2190 +     * @param val value to be AND'ed with this BigInteger.
  1.2191 +     * @return {@code this & val}
  1.2192 +     */
  1.2193 +    public BigInteger and(BigInteger val) {
  1.2194 +        int[] result = new int[Math.max(intLength(), val.intLength())];
  1.2195 +        for (int i=0; i<result.length; i++)
  1.2196 +            result[i] = (getInt(result.length-i-1)
  1.2197 +                         & val.getInt(result.length-i-1));
  1.2198 +
  1.2199 +        return valueOf(result);
  1.2200 +    }
  1.2201 +
  1.2202 +    /**
  1.2203 +     * Returns a BigInteger whose value is {@code (this | val)}.  (This method
  1.2204 +     * returns a negative BigInteger if and only if either this or val is
  1.2205 +     * negative.)
  1.2206 +     *
  1.2207 +     * @param val value to be OR'ed with this BigInteger.
  1.2208 +     * @return {@code this | val}
  1.2209 +     */
  1.2210 +    public BigInteger or(BigInteger val) {
  1.2211 +        int[] result = new int[Math.max(intLength(), val.intLength())];
  1.2212 +        for (int i=0; i<result.length; i++)
  1.2213 +            result[i] = (getInt(result.length-i-1)
  1.2214 +                         | val.getInt(result.length-i-1));
  1.2215 +
  1.2216 +        return valueOf(result);
  1.2217 +    }
  1.2218 +
  1.2219 +    /**
  1.2220 +     * Returns a BigInteger whose value is {@code (this ^ val)}.  (This method
  1.2221 +     * returns a negative BigInteger if and only if exactly one of this and
  1.2222 +     * val are negative.)
  1.2223 +     *
  1.2224 +     * @param val value to be XOR'ed with this BigInteger.
  1.2225 +     * @return {@code this ^ val}
  1.2226 +     */
  1.2227 +    public BigInteger xor(BigInteger val) {
  1.2228 +        int[] result = new int[Math.max(intLength(), val.intLength())];
  1.2229 +        for (int i=0; i<result.length; i++)
  1.2230 +            result[i] = (getInt(result.length-i-1)
  1.2231 +                         ^ val.getInt(result.length-i-1));
  1.2232 +
  1.2233 +        return valueOf(result);
  1.2234 +    }
  1.2235 +
  1.2236 +    /**
  1.2237 +     * Returns a BigInteger whose value is {@code (~this)}.  (This method
  1.2238 +     * returns a negative value if and only if this BigInteger is
  1.2239 +     * non-negative.)
  1.2240 +     *
  1.2241 +     * @return {@code ~this}
  1.2242 +     */
  1.2243 +    public BigInteger not() {
  1.2244 +        int[] result = new int[intLength()];
  1.2245 +        for (int i=0; i<result.length; i++)
  1.2246 +            result[i] = ~getInt(result.length-i-1);
  1.2247 +
  1.2248 +        return valueOf(result);
  1.2249 +    }
  1.2250 +
  1.2251 +    /**
  1.2252 +     * Returns a BigInteger whose value is {@code (this & ~val)}.  This
  1.2253 +     * method, which is equivalent to {@code and(val.not())}, is provided as
  1.2254 +     * a convenience for masking operations.  (This method returns a negative
  1.2255 +     * BigInteger if and only if {@code this} is negative and {@code val} is
  1.2256 +     * positive.)
  1.2257 +     *
  1.2258 +     * @param val value to be complemented and AND'ed with this BigInteger.
  1.2259 +     * @return {@code this & ~val}
  1.2260 +     */
  1.2261 +    public BigInteger andNot(BigInteger val) {
  1.2262 +        int[] result = new int[Math.max(intLength(), val.intLength())];
  1.2263 +        for (int i=0; i<result.length; i++)
  1.2264 +            result[i] = (getInt(result.length-i-1)
  1.2265 +                         & ~val.getInt(result.length-i-1));
  1.2266 +
  1.2267 +        return valueOf(result);
  1.2268 +    }
  1.2269 +
  1.2270 +
  1.2271 +    // Single Bit Operations
  1.2272 +
  1.2273 +    /**
  1.2274 +     * Returns {@code true} if and only if the designated bit is set.
  1.2275 +     * (Computes {@code ((this & (1<<n)) != 0)}.)
  1.2276 +     *
  1.2277 +     * @param  n index of bit to test.
  1.2278 +     * @return {@code true} if and only if the designated bit is set.
  1.2279 +     * @throws ArithmeticException {@code n} is negative.
  1.2280 +     */
  1.2281 +    public boolean testBit(int n) {
  1.2282 +        if (n<0)
  1.2283 +            throw new ArithmeticException("Negative bit address");
  1.2284 +
  1.2285 +        return (getInt(n >>> 5) & (1 << (n & 31))) != 0;
  1.2286 +    }
  1.2287 +
  1.2288 +    /**
  1.2289 +     * Returns a BigInteger whose value is equivalent to this BigInteger
  1.2290 +     * with the designated bit set.  (Computes {@code (this | (1<<n))}.)
  1.2291 +     *
  1.2292 +     * @param  n index of bit to set.
  1.2293 +     * @return {@code this | (1<<n)}
  1.2294 +     * @throws ArithmeticException {@code n} is negative.
  1.2295 +     */
  1.2296 +    public BigInteger setBit(int n) {
  1.2297 +        if (n<0)
  1.2298 +            throw new ArithmeticException("Negative bit address");
  1.2299 +
  1.2300 +        int intNum = n >>> 5;
  1.2301 +        int[] result = new int[Math.max(intLength(), intNum+2)];
  1.2302 +
  1.2303 +        for (int i=0; i<result.length; i++)
  1.2304 +            result[result.length-i-1] = getInt(i);
  1.2305 +
  1.2306 +        result[result.length-intNum-1] |= (1 << (n & 31));
  1.2307 +
  1.2308 +        return valueOf(result);
  1.2309 +    }
  1.2310 +
  1.2311 +    /**
  1.2312 +     * Returns a BigInteger whose value is equivalent to this BigInteger
  1.2313 +     * with the designated bit cleared.
  1.2314 +     * (Computes {@code (this & ~(1<<n))}.)
  1.2315 +     *
  1.2316 +     * @param  n index of bit to clear.
  1.2317 +     * @return {@code this & ~(1<<n)}
  1.2318 +     * @throws ArithmeticException {@code n} is negative.
  1.2319 +     */
  1.2320 +    public BigInteger clearBit(int n) {
  1.2321 +        if (n<0)
  1.2322 +            throw new ArithmeticException("Negative bit address");
  1.2323 +
  1.2324 +        int intNum = n >>> 5;
  1.2325 +        int[] result = new int[Math.max(intLength(), ((n + 1) >>> 5) + 1)];
  1.2326 +
  1.2327 +        for (int i=0; i<result.length; i++)
  1.2328 +            result[result.length-i-1] = getInt(i);
  1.2329 +
  1.2330 +        result[result.length-intNum-1] &= ~(1 << (n & 31));
  1.2331 +
  1.2332 +        return valueOf(result);
  1.2333 +    }
  1.2334 +
  1.2335 +    /**
  1.2336 +     * Returns a BigInteger whose value is equivalent to this BigInteger
  1.2337 +     * with the designated bit flipped.
  1.2338 +     * (Computes {@code (this ^ (1<<n))}.)
  1.2339 +     *
  1.2340 +     * @param  n index of bit to flip.
  1.2341 +     * @return {@code this ^ (1<<n)}
  1.2342 +     * @throws ArithmeticException {@code n} is negative.
  1.2343 +     */
  1.2344 +    public BigInteger flipBit(int n) {
  1.2345 +        if (n<0)
  1.2346 +            throw new ArithmeticException("Negative bit address");
  1.2347 +
  1.2348 +        int intNum = n >>> 5;
  1.2349 +        int[] result = new int[Math.max(intLength(), intNum+2)];
  1.2350 +
  1.2351 +        for (int i=0; i<result.length; i++)
  1.2352 +            result[result.length-i-1] = getInt(i);
  1.2353 +
  1.2354 +        result[result.length-intNum-1] ^= (1 << (n & 31));
  1.2355 +
  1.2356 +        return valueOf(result);
  1.2357 +    }
  1.2358 +
  1.2359 +    /**
  1.2360 +     * Returns the index of the rightmost (lowest-order) one bit in this
  1.2361 +     * BigInteger (the number of zero bits to the right of the rightmost
  1.2362 +     * one bit).  Returns -1 if this BigInteger contains no one bits.
  1.2363 +     * (Computes {@code (this==0? -1 : log2(this & -this))}.)
  1.2364 +     *
  1.2365 +     * @return index of the rightmost one bit in this BigInteger.
  1.2366 +     */
  1.2367 +    public int getLowestSetBit() {
  1.2368 +        @SuppressWarnings("deprecation") int lsb = lowestSetBit - 2;
  1.2369 +        if (lsb == -2) {  // lowestSetBit not initialized yet
  1.2370 +            lsb = 0;
  1.2371 +            if (signum == 0) {
  1.2372 +                lsb -= 1;
  1.2373 +            } else {
  1.2374 +                // Search for lowest order nonzero int
  1.2375 +                int i,b;
  1.2376 +                for (i=0; (b = getInt(i))==0; i++)
  1.2377 +                    ;
  1.2378 +                lsb += (i << 5) + Integer.numberOfTrailingZeros(b);
  1.2379 +            }
  1.2380 +            lowestSetBit = lsb + 2;
  1.2381 +        }
  1.2382 +        return lsb;
  1.2383 +    }
  1.2384 +
  1.2385 +
  1.2386 +    // Miscellaneous Bit Operations
  1.2387 +
  1.2388 +    /**
  1.2389 +     * Returns the number of bits in the minimal two's-complement
  1.2390 +     * representation of this BigInteger, <i>excluding</i> a sign bit.
  1.2391 +     * For positive BigIntegers, this is equivalent to the number of bits in
  1.2392 +     * the ordinary binary representation.  (Computes
  1.2393 +     * {@code (ceil(log2(this < 0 ? -this : this+1)))}.)
  1.2394 +     *
  1.2395 +     * @return number of bits in the minimal two's-complement
  1.2396 +     *         representation of this BigInteger, <i>excluding</i> a sign bit.
  1.2397 +     */
  1.2398 +    public int bitLength() {
  1.2399 +        @SuppressWarnings("deprecation") int n = bitLength - 1;
  1.2400 +        if (n == -1) { // bitLength not initialized yet
  1.2401 +            int[] m = mag;
  1.2402 +            int len = m.length;
  1.2403 +            if (len == 0) {
  1.2404 +                n = 0; // offset by one to initialize
  1.2405 +            }  else {
  1.2406 +                // Calculate the bit length of the magnitude
  1.2407 +                int magBitLength = ((len - 1) << 5) + bitLengthForInt(mag[0]);
  1.2408 +                 if (signum < 0) {
  1.2409 +                     // Check if magnitude is a power of two
  1.2410 +                     boolean pow2 = (Integer.bitCount(mag[0]) == 1);
  1.2411 +                     for(int i=1; i< len && pow2; i++)
  1.2412 +                         pow2 = (mag[i] == 0);
  1.2413 +
  1.2414 +                     n = (pow2 ? magBitLength -1 : magBitLength);
  1.2415 +                 } else {
  1.2416 +                     n = magBitLength;
  1.2417 +                 }
  1.2418 +            }
  1.2419 +            bitLength = n + 1;
  1.2420 +        }
  1.2421 +        return n;
  1.2422 +    }
  1.2423 +
  1.2424 +    /**
  1.2425 +     * Returns the number of bits in the two's complement representation
  1.2426 +     * of this BigInteger that differ from its sign bit.  This method is
  1.2427 +     * useful when implementing bit-vector style sets atop BigIntegers.
  1.2428 +     *
  1.2429 +     * @return number of bits in the two's complement representation
  1.2430 +     *         of this BigInteger that differ from its sign bit.
  1.2431 +     */
  1.2432 +    public int bitCount() {
  1.2433 +        @SuppressWarnings("deprecation") int bc = bitCount - 1;
  1.2434 +        if (bc == -1) {  // bitCount not initialized yet
  1.2435 +            bc = 0;      // offset by one to initialize
  1.2436 +            // Count the bits in the magnitude
  1.2437 +            for (int i=0; i<mag.length; i++)
  1.2438 +                bc += Integer.bitCount(mag[i]);
  1.2439 +            if (signum < 0) {
  1.2440 +                // Count the trailing zeros in the magnitude
  1.2441 +                int magTrailingZeroCount = 0, j;
  1.2442 +                for (j=mag.length-1; mag[j]==0; j--)
  1.2443 +                    magTrailingZeroCount += 32;
  1.2444 +                magTrailingZeroCount += Integer.numberOfTrailingZeros(mag[j]);
  1.2445 +                bc += magTrailingZeroCount - 1;
  1.2446 +            }
  1.2447 +            bitCount = bc + 1;
  1.2448 +        }
  1.2449 +        return bc;
  1.2450 +    }
  1.2451 +
  1.2452 +    // Primality Testing
  1.2453 +
  1.2454 +    /**
  1.2455 +     * Returns {@code true} if this BigInteger is probably prime,
  1.2456 +     * {@code false} if it's definitely composite.  If
  1.2457 +     * {@code certainty} is &le; 0, {@code true} is
  1.2458 +     * returned.
  1.2459 +     *
  1.2460 +     * @param  certainty a measure of the uncertainty that the caller is
  1.2461 +     *         willing to tolerate: if the call returns {@code true}
  1.2462 +     *         the probability that this BigInteger is prime exceeds
  1.2463 +     *         (1 - 1/2<sup>{@code certainty}</sup>).  The execution time of
  1.2464 +     *         this method is proportional to the value of this parameter.
  1.2465 +     * @return {@code true} if this BigInteger is probably prime,
  1.2466 +     *         {@code false} if it's definitely composite.
  1.2467 +     */
  1.2468 +    public boolean isProbablePrime(int certainty) {
  1.2469 +        if (certainty <= 0)
  1.2470 +            return true;
  1.2471 +        BigInteger w = this.abs();
  1.2472 +        if (w.equals(TWO))
  1.2473 +            return true;
  1.2474 +        if (!w.testBit(0) || w.equals(ONE))
  1.2475 +            return false;
  1.2476 +
  1.2477 +        return w.primeToCertainty(certainty, null);
  1.2478 +    }
  1.2479 +
  1.2480 +    // Comparison Operations
  1.2481 +
  1.2482 +    /**
  1.2483 +     * Compares this BigInteger with the specified BigInteger.  This
  1.2484 +     * method is provided in preference to individual methods for each
  1.2485 +     * of the six boolean comparison operators ({@literal <}, ==,
  1.2486 +     * {@literal >}, {@literal >=}, !=, {@literal <=}).  The suggested
  1.2487 +     * idiom for performing these comparisons is: {@code
  1.2488 +     * (x.compareTo(y)} &lt;<i>op</i>&gt; {@code 0)}, where
  1.2489 +     * &lt;<i>op</i>&gt; is one of the six comparison operators.
  1.2490 +     *
  1.2491 +     * @param  val BigInteger to which this BigInteger is to be compared.
  1.2492 +     * @return -1, 0 or 1 as this BigInteger is numerically less than, equal
  1.2493 +     *         to, or greater than {@code val}.
  1.2494 +     */
  1.2495 +    public int compareTo(BigInteger val) {
  1.2496 +        if (signum == val.signum) {
  1.2497 +            switch (signum) {
  1.2498 +            case 1:
  1.2499 +                return compareMagnitude(val);
  1.2500 +            case -1:
  1.2501 +                return val.compareMagnitude(this);
  1.2502 +            default:
  1.2503 +                return 0;
  1.2504 +            }
  1.2505 +        }
  1.2506 +        return signum > val.signum ? 1 : -1;
  1.2507 +    }
  1.2508 +
  1.2509 +    /**
  1.2510 +     * Compares the magnitude array of this BigInteger with the specified
  1.2511 +     * BigInteger's. This is the version of compareTo ignoring sign.
  1.2512 +     *
  1.2513 +     * @param val BigInteger whose magnitude array to be compared.
  1.2514 +     * @return -1, 0 or 1 as this magnitude array is less than, equal to or
  1.2515 +     *         greater than the magnitude aray for the specified BigInteger's.
  1.2516 +     */
  1.2517 +    final int compareMagnitude(BigInteger val) {
  1.2518 +        int[] m1 = mag;
  1.2519 +        int len1 = m1.length;
  1.2520 +        int[] m2 = val.mag;
  1.2521 +        int len2 = m2.length;
  1.2522 +        if (len1 < len2)
  1.2523 +            return -1;
  1.2524 +        if (len1 > len2)
  1.2525 +            return 1;
  1.2526 +        for (int i = 0; i < len1; i++) {
  1.2527 +            int a = m1[i];
  1.2528 +            int b = m2[i];
  1.2529 +            if (a != b)
  1.2530 +                return ((a & LONG_MASK) < (b & LONG_MASK)) ? -1 : 1;
  1.2531 +        }
  1.2532 +        return 0;
  1.2533 +    }
  1.2534 +
  1.2535 +    /**
  1.2536 +     * Compares this BigInteger with the specified Object for equality.
  1.2537 +     *
  1.2538 +     * @param  x Object to which this BigInteger is to be compared.
  1.2539 +     * @return {@code true} if and only if the specified Object is a
  1.2540 +     *         BigInteger whose value is numerically equal to this BigInteger.
  1.2541 +     */
  1.2542 +    public boolean equals(Object x) {
  1.2543 +        // This test is just an optimization, which may or may not help
  1.2544 +        if (x == this)
  1.2545 +            return true;
  1.2546 +
  1.2547 +        if (!(x instanceof BigInteger))
  1.2548 +            return false;
  1.2549 +
  1.2550 +        BigInteger xInt = (BigInteger) x;
  1.2551 +        if (xInt.signum != signum)
  1.2552 +            return false;
  1.2553 +
  1.2554 +        int[] m = mag;
  1.2555 +        int len = m.length;
  1.2556 +        int[] xm = xInt.mag;
  1.2557 +        if (len != xm.length)
  1.2558 +            return false;
  1.2559 +
  1.2560 +        for (int i = 0; i < len; i++)
  1.2561 +            if (xm[i] != m[i])
  1.2562 +                return false;
  1.2563 +
  1.2564 +        return true;
  1.2565 +    }
  1.2566 +
  1.2567 +    /**
  1.2568 +     * Returns the minimum of this BigInteger and {@code val}.
  1.2569 +     *
  1.2570 +     * @param  val value with which the minimum is to be computed.
  1.2571 +     * @return the BigInteger whose value is the lesser of this BigInteger and
  1.2572 +     *         {@code val}.  If they are equal, either may be returned.
  1.2573 +     */
  1.2574 +    public BigInteger min(BigInteger val) {
  1.2575 +        return (compareTo(val)<0 ? this : val);
  1.2576 +    }
  1.2577 +
  1.2578 +    /**
  1.2579 +     * Returns the maximum of this BigInteger and {@code val}.
  1.2580 +     *
  1.2581 +     * @param  val value with which the maximum is to be computed.
  1.2582 +     * @return the BigInteger whose value is the greater of this and
  1.2583 +     *         {@code val}.  If they are equal, either may be returned.
  1.2584 +     */
  1.2585 +    public BigInteger max(BigInteger val) {
  1.2586 +        return (compareTo(val)>0 ? this : val);
  1.2587 +    }
  1.2588 +
  1.2589 +
  1.2590 +    // Hash Function
  1.2591 +
  1.2592 +    /**
  1.2593 +     * Returns the hash code for this BigInteger.
  1.2594 +     *
  1.2595 +     * @return hash code for this BigInteger.
  1.2596 +     */
  1.2597 +    public int hashCode() {
  1.2598 +        int hashCode = 0;
  1.2599 +
  1.2600 +        for (int i=0; i<mag.length; i++)
  1.2601 +            hashCode = (int)(31*hashCode + (mag[i] & LONG_MASK));
  1.2602 +
  1.2603 +        return hashCode * signum;
  1.2604 +    }
  1.2605 +
  1.2606 +    /**
  1.2607 +     * Returns the String representation of this BigInteger in the
  1.2608 +     * given radix.  If the radix is outside the range from {@link
  1.2609 +     * Character#MIN_RADIX} to {@link Character#MAX_RADIX} inclusive,
  1.2610 +     * it will default to 10 (as is the case for
  1.2611 +     * {@code Integer.toString}).  The digit-to-character mapping
  1.2612 +     * provided by {@code Character.forDigit} is used, and a minus
  1.2613 +     * sign is prepended if appropriate.  (This representation is
  1.2614 +     * compatible with the {@link #BigInteger(String, int) (String,
  1.2615 +     * int)} constructor.)
  1.2616 +     *
  1.2617 +     * @param  radix  radix of the String representation.
  1.2618 +     * @return String representation of this BigInteger in the given radix.
  1.2619 +     * @see    Integer#toString
  1.2620 +     * @see    Character#forDigit
  1.2621 +     * @see    #BigInteger(java.lang.String, int)
  1.2622 +     */
  1.2623 +    public String toString(int radix) {
  1.2624 +        if (signum == 0)
  1.2625 +            return "0";
  1.2626 +        if (radix < Character.MIN_RADIX || radix > Character.MAX_RADIX)
  1.2627 +            radix = 10;
  1.2628 +
  1.2629 +        // Compute upper bound on number of digit groups and allocate space
  1.2630 +        int maxNumDigitGroups = (4*mag.length + 6)/7;
  1.2631 +        String digitGroup[] = new String[maxNumDigitGroups];
  1.2632 +
  1.2633 +        // Translate number to string, a digit group at a time
  1.2634 +        BigInteger tmp = this.abs();
  1.2635 +        int numGroups = 0;
  1.2636 +        while (tmp.signum != 0) {
  1.2637 +            BigInteger d = longRadix[radix];
  1.2638 +
  1.2639 +            MutableBigInteger q = new MutableBigInteger(),
  1.2640 +                              a = new MutableBigInteger(tmp.mag),
  1.2641 +                              b = new MutableBigInteger(d.mag);
  1.2642 +            MutableBigInteger r = a.divide(b, q);
  1.2643 +            BigInteger q2 = q.toBigInteger(tmp.signum * d.signum);
  1.2644 +            BigInteger r2 = r.toBigInteger(tmp.signum * d.signum);
  1.2645 +
  1.2646 +            digitGroup[numGroups++] = Long.toString(r2.longValue(), radix);
  1.2647 +            tmp = q2;
  1.2648 +        }
  1.2649 +
  1.2650 +        // Put sign (if any) and first digit group into result buffer
  1.2651 +        StringBuilder buf = new StringBuilder(numGroups*digitsPerLong[radix]+1);
  1.2652 +        if (signum<0)
  1.2653 +            buf.append('-');
  1.2654 +        buf.append(digitGroup[numGroups-1]);
  1.2655 +
  1.2656 +        // Append remaining digit groups padded with leading zeros
  1.2657 +        for (int i=numGroups-2; i>=0; i--) {
  1.2658 +            // Prepend (any) leading zeros for this digit group
  1.2659 +            int numLeadingZeros = digitsPerLong[radix]-digitGroup[i].length();
  1.2660 +            if (numLeadingZeros != 0)
  1.2661 +                buf.append(zeros[numLeadingZeros]);
  1.2662 +            buf.append(digitGroup[i]);
  1.2663 +        }
  1.2664 +        return buf.toString();
  1.2665 +    }
  1.2666 +
  1.2667 +    /* zero[i] is a string of i consecutive zeros. */
  1.2668 +    private static String zeros[] = new String[64];
  1.2669 +    static {
  1.2670 +        zeros[63] =
  1.2671 +            "000000000000000000000000000000000000000000000000000000000000000";
  1.2672 +        for (int i=0; i<63; i++)
  1.2673 +            zeros[i] = zeros[63].substring(0, i);
  1.2674 +    }
  1.2675 +
  1.2676 +    /**
  1.2677 +     * Returns the decimal String representation of this BigInteger.
  1.2678 +     * The digit-to-character mapping provided by
  1.2679 +     * {@code Character.forDigit} is used, and a minus sign is
  1.2680 +     * prepended if appropriate.  (This representation is compatible
  1.2681 +     * with the {@link #BigInteger(String) (String)} constructor, and
  1.2682 +     * allows for String concatenation with Java's + operator.)
  1.2683 +     *
  1.2684 +     * @return decimal String representation of this BigInteger.
  1.2685 +     * @see    Character#forDigit
  1.2686 +     * @see    #BigInteger(java.lang.String)
  1.2687 +     */
  1.2688 +    public String toString() {
  1.2689 +        return toString(10);
  1.2690 +    }
  1.2691 +
  1.2692 +    /**
  1.2693 +     * Returns a byte array containing the two's-complement
  1.2694 +     * representation of this BigInteger.  The byte array will be in
  1.2695 +     * <i>big-endian</i> byte-order: the most significant byte is in
  1.2696 +     * the zeroth element.  The array will contain the minimum number
  1.2697 +     * of bytes required to represent this BigInteger, including at
  1.2698 +     * least one sign bit, which is {@code (ceil((this.bitLength() +
  1.2699 +     * 1)/8))}.  (This representation is compatible with the
  1.2700 +     * {@link #BigInteger(byte[]) (byte[])} constructor.)
  1.2701 +     *
  1.2702 +     * @return a byte array containing the two's-complement representation of
  1.2703 +     *         this BigInteger.
  1.2704 +     * @see    #BigInteger(byte[])
  1.2705 +     */
  1.2706 +    public byte[] toByteArray() {
  1.2707 +        int byteLen = bitLength()/8 + 1;
  1.2708 +        byte[] byteArray = new byte[byteLen];
  1.2709 +
  1.2710 +        for (int i=byteLen-1, bytesCopied=4, nextInt=0, intIndex=0; i>=0; i--) {
  1.2711 +            if (bytesCopied == 4) {
  1.2712 +                nextInt = getInt(intIndex++);
  1.2713 +                bytesCopied = 1;
  1.2714 +            } else {
  1.2715 +                nextInt >>>= 8;
  1.2716 +                bytesCopied++;
  1.2717 +            }
  1.2718 +            byteArray[i] = (byte)nextInt;
  1.2719 +        }
  1.2720 +        return byteArray;
  1.2721 +    }
  1.2722 +
  1.2723 +    /**
  1.2724 +     * Converts this BigInteger to an {@code int}.  This
  1.2725 +     * conversion is analogous to a
  1.2726 +     * <i>narrowing primitive conversion</i> from {@code long} to
  1.2727 +     * {@code int} as defined in section 5.1.3 of
  1.2728 +     * <cite>The Java&trade; Language Specification</cite>:
  1.2729 +     * if this BigInteger is too big to fit in an
  1.2730 +     * {@code int}, only the low-order 32 bits are returned.
  1.2731 +     * Note that this conversion can lose information about the
  1.2732 +     * overall magnitude of the BigInteger value as well as return a
  1.2733 +     * result with the opposite sign.
  1.2734 +     *
  1.2735 +     * @return this BigInteger converted to an {@code int}.
  1.2736 +     */
  1.2737 +    public int intValue() {
  1.2738 +        int result = 0;
  1.2739 +        result = getInt(0);
  1.2740 +        return result;
  1.2741 +    }
  1.2742 +
  1.2743 +    /**
  1.2744 +     * Converts this BigInteger to a {@code long}.  This
  1.2745 +     * conversion is analogous to a
  1.2746 +     * <i>narrowing primitive conversion</i> from {@code long} to
  1.2747 +     * {@code int} as defined in section 5.1.3 of
  1.2748 +     * <cite>The Java&trade; Language Specification</cite>:
  1.2749 +     * if this BigInteger is too big to fit in a
  1.2750 +     * {@code long}, only the low-order 64 bits are returned.
  1.2751 +     * Note that this conversion can lose information about the
  1.2752 +     * overall magnitude of the BigInteger value as well as return a
  1.2753 +     * result with the opposite sign.
  1.2754 +     *
  1.2755 +     * @return this BigInteger converted to a {@code long}.
  1.2756 +     */
  1.2757 +    public long longValue() {
  1.2758 +        long result = 0;
  1.2759 +
  1.2760 +        for (int i=1; i>=0; i--)
  1.2761 +            result = (result << 32) + (getInt(i) & LONG_MASK);
  1.2762 +        return result;
  1.2763 +    }
  1.2764 +
  1.2765 +    /**
  1.2766 +     * Converts this BigInteger to a {@code float}.  This
  1.2767 +     * conversion is similar to the
  1.2768 +     * <i>narrowing primitive conversion</i> from {@code double} to
  1.2769 +     * {@code float} as defined in section 5.1.3 of
  1.2770 +     * <cite>The Java&trade; Language Specification</cite>:
  1.2771 +     * if this BigInteger has too great a magnitude
  1.2772 +     * to represent as a {@code float}, it will be converted to
  1.2773 +     * {@link Float#NEGATIVE_INFINITY} or {@link
  1.2774 +     * Float#POSITIVE_INFINITY} as appropriate.  Note that even when
  1.2775 +     * the return value is finite, this conversion can lose
  1.2776 +     * information about the precision of the BigInteger value.
  1.2777 +     *
  1.2778 +     * @return this BigInteger converted to a {@code float}.
  1.2779 +     */
  1.2780 +    public float floatValue() {
  1.2781 +        // Somewhat inefficient, but guaranteed to work.
  1.2782 +        return Float.parseFloat(this.toString());
  1.2783 +    }
  1.2784 +
  1.2785 +    /**
  1.2786 +     * Converts this BigInteger to a {@code double}.  This
  1.2787 +     * conversion is similar to the
  1.2788 +     * <i>narrowing primitive conversion</i> from {@code double} to
  1.2789 +     * {@code float} as defined in section 5.1.3 of
  1.2790 +     * <cite>The Java&trade; Language Specification</cite>:
  1.2791 +     * if this BigInteger has too great a magnitude
  1.2792 +     * to represent as a {@code double}, it will be converted to
  1.2793 +     * {@link Double#NEGATIVE_INFINITY} or {@link
  1.2794 +     * Double#POSITIVE_INFINITY} as appropriate.  Note that even when
  1.2795 +     * the return value is finite, this conversion can lose
  1.2796 +     * information about the precision of the BigInteger value.
  1.2797 +     *
  1.2798 +     * @return this BigInteger converted to a {@code double}.
  1.2799 +     */
  1.2800 +    public double doubleValue() {
  1.2801 +        // Somewhat inefficient, but guaranteed to work.
  1.2802 +        return Double.parseDouble(this.toString());
  1.2803 +    }
  1.2804 +
  1.2805 +    /**
  1.2806 +     * Returns a copy of the input array stripped of any leading zero bytes.
  1.2807 +     */
  1.2808 +    private static int[] stripLeadingZeroInts(int val[]) {
  1.2809 +        int vlen = val.length;
  1.2810 +        int keep;
  1.2811 +
  1.2812 +        // Find first nonzero byte
  1.2813 +        for (keep = 0; keep < vlen && val[keep] == 0; keep++)
  1.2814 +            ;
  1.2815 +        return java.util.Arrays.copyOfRange(val, keep, vlen);
  1.2816 +    }
  1.2817 +
  1.2818 +    /**
  1.2819 +     * Returns the input array stripped of any leading zero bytes.
  1.2820 +     * Since the source is trusted the copying may be skipped.
  1.2821 +     */
  1.2822 +    private static int[] trustedStripLeadingZeroInts(int val[]) {
  1.2823 +        int vlen = val.length;
  1.2824 +        int keep;
  1.2825 +
  1.2826 +        // Find first nonzero byte
  1.2827 +        for (keep = 0; keep < vlen && val[keep] == 0; keep++)
  1.2828 +            ;
  1.2829 +        return keep == 0 ? val : java.util.Arrays.copyOfRange(val, keep, vlen);
  1.2830 +    }
  1.2831 +
  1.2832 +    /**
  1.2833 +     * Returns a copy of the input array stripped of any leading zero bytes.
  1.2834 +     */
  1.2835 +    private static int[] stripLeadingZeroBytes(byte a[]) {
  1.2836 +        int byteLength = a.length;
  1.2837 +        int keep;
  1.2838 +
  1.2839 +        // Find first nonzero byte
  1.2840 +        for (keep = 0; keep < byteLength && a[keep]==0; keep++)
  1.2841 +            ;
  1.2842 +
  1.2843 +        // Allocate new array and copy relevant part of input array
  1.2844 +        int intLength = ((byteLength - keep) + 3) >>> 2;
  1.2845 +        int[] result = new int[intLength];
  1.2846 +        int b = byteLength - 1;
  1.2847 +        for (int i = intLength-1; i >= 0; i--) {
  1.2848 +            result[i] = a[b--] & 0xff;
  1.2849 +            int bytesRemaining = b - keep + 1;
  1.2850 +            int bytesToTransfer = Math.min(3, bytesRemaining);
  1.2851 +            for (int j=8; j <= (bytesToTransfer << 3); j += 8)
  1.2852 +                result[i] |= ((a[b--] & 0xff) << j);
  1.2853 +        }
  1.2854 +        return result;
  1.2855 +    }
  1.2856 +
  1.2857 +    /**
  1.2858 +     * Takes an array a representing a negative 2's-complement number and
  1.2859 +     * returns the minimal (no leading zero bytes) unsigned whose value is -a.
  1.2860 +     */
  1.2861 +    private static int[] makePositive(byte a[]) {
  1.2862 +        int keep, k;
  1.2863 +        int byteLength = a.length;
  1.2864 +
  1.2865 +        // Find first non-sign (0xff) byte of input
  1.2866 +        for (keep=0; keep<byteLength && a[keep]==-1; keep++)
  1.2867 +            ;
  1.2868 +
  1.2869 +
  1.2870 +        /* Allocate output array.  If all non-sign bytes are 0x00, we must
  1.2871 +         * allocate space for one extra output byte. */
  1.2872 +        for (k=keep; k<byteLength && a[k]==0; k++)
  1.2873 +            ;
  1.2874 +
  1.2875 +        int extraByte = (k==byteLength) ? 1 : 0;
  1.2876 +        int intLength = ((byteLength - keep + extraByte) + 3)/4;
  1.2877 +        int result[] = new int[intLength];
  1.2878 +
  1.2879 +        /* Copy one's complement of input into output, leaving extra
  1.2880 +         * byte (if it exists) == 0x00 */
  1.2881 +        int b = byteLength - 1;
  1.2882 +        for (int i = intLength-1; i >= 0; i--) {
  1.2883 +            result[i] = a[b--] & 0xff;
  1.2884 +            int numBytesToTransfer = Math.min(3, b-keep+1);
  1.2885 +            if (numBytesToTransfer < 0)
  1.2886 +                numBytesToTransfer = 0;
  1.2887 +            for (int j=8; j <= 8*numBytesToTransfer; j += 8)
  1.2888 +                result[i] |= ((a[b--] & 0xff) << j);
  1.2889 +
  1.2890 +            // Mask indicates which bits must be complemented
  1.2891 +            int mask = -1 >>> (8*(3-numBytesToTransfer));
  1.2892 +            result[i] = ~result[i] & mask;
  1.2893 +        }
  1.2894 +
  1.2895 +        // Add one to one's complement to generate two's complement
  1.2896 +        for (int i=result.length-1; i>=0; i--) {
  1.2897 +            result[i] = (int)((result[i] & LONG_MASK) + 1);
  1.2898 +            if (result[i] != 0)
  1.2899 +                break;
  1.2900 +        }
  1.2901 +
  1.2902 +        return result;
  1.2903 +    }
  1.2904 +
  1.2905 +    /**
  1.2906 +     * Takes an array a representing a negative 2's-complement number and
  1.2907 +     * returns the minimal (no leading zero ints) unsigned whose value is -a.
  1.2908 +     */
  1.2909 +    private static int[] makePositive(int a[]) {
  1.2910 +        int keep, j;
  1.2911 +
  1.2912 +        // Find first non-sign (0xffffffff) int of input
  1.2913 +        for (keep=0; keep<a.length && a[keep]==-1; keep++)
  1.2914 +            ;
  1.2915 +
  1.2916 +        /* Allocate output array.  If all non-sign ints are 0x00, we must
  1.2917 +         * allocate space for one extra output int. */
  1.2918 +        for (j=keep; j<a.length && a[j]==0; j++)
  1.2919 +            ;
  1.2920 +        int extraInt = (j==a.length ? 1 : 0);
  1.2921 +        int result[] = new int[a.length - keep + extraInt];
  1.2922 +
  1.2923 +        /* Copy one's complement of input into output, leaving extra
  1.2924 +         * int (if it exists) == 0x00 */
  1.2925 +        for (int i = keep; i<a.length; i++)
  1.2926 +            result[i - keep + extraInt] = ~a[i];
  1.2927 +
  1.2928 +        // Add one to one's complement to generate two's complement
  1.2929 +        for (int i=result.length-1; ++result[i]==0; i--)
  1.2930 +            ;
  1.2931 +
  1.2932 +        return result;
  1.2933 +    }
  1.2934 +
  1.2935 +    /*
  1.2936 +     * The following two arrays are used for fast String conversions.  Both
  1.2937 +     * are indexed by radix.  The first is the number of digits of the given
  1.2938 +     * radix that can fit in a Java long without "going negative", i.e., the
  1.2939 +     * highest integer n such that radix**n < 2**63.  The second is the
  1.2940 +     * "long radix" that tears each number into "long digits", each of which
  1.2941 +     * consists of the number of digits in the corresponding element in
  1.2942 +     * digitsPerLong (longRadix[i] = i**digitPerLong[i]).  Both arrays have
  1.2943 +     * nonsense values in their 0 and 1 elements, as radixes 0 and 1 are not
  1.2944 +     * used.
  1.2945 +     */
  1.2946 +    private static int digitsPerLong[] = {0, 0,
  1.2947 +        62, 39, 31, 27, 24, 22, 20, 19, 18, 18, 17, 17, 16, 16, 15, 15, 15, 14,
  1.2948 +        14, 14, 14, 13, 13, 13, 13, 13, 13, 12, 12, 12, 12, 12, 12, 12, 12};
  1.2949 +
  1.2950 +    private static BigInteger longRadix[] = {null, null,
  1.2951 +        valueOf(0x4000000000000000L), valueOf(0x383d9170b85ff80bL),
  1.2952 +        valueOf(0x4000000000000000L), valueOf(0x6765c793fa10079dL),
  1.2953 +        valueOf(0x41c21cb8e1000000L), valueOf(0x3642798750226111L),
  1.2954 +        valueOf(0x1000000000000000L), valueOf(0x12bf307ae81ffd59L),
  1.2955 +        valueOf( 0xde0b6b3a7640000L), valueOf(0x4d28cb56c33fa539L),
  1.2956 +        valueOf(0x1eca170c00000000L), valueOf(0x780c7372621bd74dL),
  1.2957 +        valueOf(0x1e39a5057d810000L), valueOf(0x5b27ac993df97701L),
  1.2958 +        valueOf(0x1000000000000000L), valueOf(0x27b95e997e21d9f1L),
  1.2959 +        valueOf(0x5da0e1e53c5c8000L), valueOf( 0xb16a458ef403f19L),
  1.2960 +        valueOf(0x16bcc41e90000000L), valueOf(0x2d04b7fdd9c0ef49L),
  1.2961 +        valueOf(0x5658597bcaa24000L), valueOf( 0x6feb266931a75b7L),
  1.2962 +        valueOf( 0xc29e98000000000L), valueOf(0x14adf4b7320334b9L),
  1.2963 +        valueOf(0x226ed36478bfa000L), valueOf(0x383d9170b85ff80bL),
  1.2964 +        valueOf(0x5a3c23e39c000000L), valueOf( 0x4e900abb53e6b71L),
  1.2965 +        valueOf( 0x7600ec618141000L), valueOf( 0xaee5720ee830681L),
  1.2966 +        valueOf(0x1000000000000000L), valueOf(0x172588ad4f5f0981L),
  1.2967 +        valueOf(0x211e44f7d02c1000L), valueOf(0x2ee56725f06e5c71L),
  1.2968 +        valueOf(0x41c21cb8e1000000L)};
  1.2969 +
  1.2970 +    /*
  1.2971 +     * These two arrays are the integer analogue of above.
  1.2972 +     */
  1.2973 +    private static int digitsPerInt[] = {0, 0, 30, 19, 15, 13, 11,
  1.2974 +        11, 10, 9, 9, 8, 8, 8, 8, 7, 7, 7, 7, 7, 7, 7, 6, 6, 6, 6,
  1.2975 +        6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 5};
  1.2976 +
  1.2977 +    private static int intRadix[] = {0, 0,
  1.2978 +        0x40000000, 0x4546b3db, 0x40000000, 0x48c27395, 0x159fd800,
  1.2979 +        0x75db9c97, 0x40000000, 0x17179149, 0x3b9aca00, 0xcc6db61,
  1.2980 +        0x19a10000, 0x309f1021, 0x57f6c100, 0xa2f1b6f,  0x10000000,
  1.2981 +        0x18754571, 0x247dbc80, 0x3547667b, 0x4c4b4000, 0x6b5a6e1d,
  1.2982 +        0x6c20a40,  0x8d2d931,  0xb640000,  0xe8d4a51,  0x1269ae40,
  1.2983 +        0x17179149, 0x1cb91000, 0x23744899, 0x2b73a840, 0x34e63b41,
  1.2984 +        0x40000000, 0x4cfa3cc1, 0x5c13d840, 0x6d91b519, 0x39aa400
  1.2985 +    };
  1.2986 +
  1.2987 +    /**
  1.2988 +     * These routines provide access to the two's complement representation
  1.2989 +     * of BigIntegers.
  1.2990 +     */
  1.2991 +
  1.2992 +    /**
  1.2993 +     * Returns the length of the two's complement representation in ints,
  1.2994 +     * including space for at least one sign bit.
  1.2995 +     */
  1.2996 +    private int intLength() {
  1.2997 +        return (bitLength() >>> 5) + 1;
  1.2998 +    }
  1.2999 +
  1.3000 +    /* Returns sign bit */
  1.3001 +    private int signBit() {
  1.3002 +        return signum < 0 ? 1 : 0;
  1.3003 +    }
  1.3004 +
  1.3005 +    /* Returns an int of sign bits */
  1.3006 +    private int signInt() {
  1.3007 +        return signum < 0 ? -1 : 0;
  1.3008 +    }
  1.3009 +
  1.3010 +    /**
  1.3011 +     * Returns the specified int of the little-endian two's complement
  1.3012 +     * representation (int 0 is the least significant).  The int number can
  1.3013 +     * be arbitrarily high (values are logically preceded by infinitely many
  1.3014 +     * sign ints).
  1.3015 +     */
  1.3016 +    private int getInt(int n) {
  1.3017 +        if (n < 0)
  1.3018 +            return 0;
  1.3019 +        if (n >= mag.length)
  1.3020 +            return signInt();
  1.3021 +
  1.3022 +        int magInt = mag[mag.length-n-1];
  1.3023 +
  1.3024 +        return (signum >= 0 ? magInt :
  1.3025 +                (n <= firstNonzeroIntNum() ? -magInt : ~magInt));
  1.3026 +    }
  1.3027 +
  1.3028 +    /**
  1.3029 +     * Returns the index of the int that contains the first nonzero int in the
  1.3030 +     * little-endian binary representation of the magnitude (int 0 is the
  1.3031 +     * least significant). If the magnitude is zero, return value is undefined.
  1.3032 +     */
  1.3033 +     private int firstNonzeroIntNum() {
  1.3034 +         int fn = firstNonzeroIntNum - 2;
  1.3035 +         if (fn == -2) { // firstNonzeroIntNum not initialized yet
  1.3036 +             fn = 0;
  1.3037 +
  1.3038 +             // Search for the first nonzero int
  1.3039 +             int i;
  1.3040 +             int mlen = mag.length;
  1.3041 +             for (i = mlen - 1; i >= 0 && mag[i] == 0; i--)
  1.3042 +                 ;
  1.3043 +             fn = mlen - i - 1;
  1.3044 +             firstNonzeroIntNum = fn + 2; // offset by two to initialize
  1.3045 +         }
  1.3046 +         return fn;
  1.3047 +     }
  1.3048 +
  1.3049 +    /** use serialVersionUID from JDK 1.1. for interoperability */
  1.3050 +    private static final long serialVersionUID = -8287574255936472291L;
  1.3051 +
  1.3052 +    /**
  1.3053 +     * Serializable fields for BigInteger.
  1.3054 +     *
  1.3055 +     * @serialField signum  int
  1.3056 +     *              signum of this BigInteger.
  1.3057 +     * @serialField magnitude int[]
  1.3058 +     *              magnitude array of this BigInteger.
  1.3059 +     * @serialField bitCount  int
  1.3060 +     *              number of bits in this BigInteger
  1.3061 +     * @serialField bitLength int
  1.3062 +     *              the number of bits in the minimal two's-complement
  1.3063 +     *              representation of this BigInteger
  1.3064 +     * @serialField lowestSetBit int
  1.3065 +     *              lowest set bit in the twos complement representation
  1.3066 +     */
  1.3067 +    private static final ObjectStreamField[] serialPersistentFields = {
  1.3068 +        new ObjectStreamField("signum", Integer.TYPE),
  1.3069 +        new ObjectStreamField("magnitude", byte[].class),
  1.3070 +        new ObjectStreamField("bitCount", Integer.TYPE),
  1.3071 +        new ObjectStreamField("bitLength", Integer.TYPE),
  1.3072 +        new ObjectStreamField("firstNonzeroByteNum", Integer.TYPE),
  1.3073 +        new ObjectStreamField("lowestSetBit", Integer.TYPE)
  1.3074 +        };
  1.3075 +
  1.3076 +    /**
  1.3077 +     * Reconstitute the {@code BigInteger} instance from a stream (that is,
  1.3078 +     * deserialize it). The magnitude is read in as an array of bytes
  1.3079 +     * for historical reasons, but it is converted to an array of ints
  1.3080 +     * and the byte array is discarded.
  1.3081 +     * Note:
  1.3082 +     * The current convention is to initialize the cache fields, bitCount,
  1.3083 +     * bitLength and lowestSetBit, to 0 rather than some other marker value.
  1.3084 +     * Therefore, no explicit action to set these fields needs to be taken in
  1.3085 +     * readObject because those fields already have a 0 value be default since
  1.3086 +     * defaultReadObject is not being used.
  1.3087 +     */
  1.3088 +    private void readObject(java.io.ObjectInputStream s)
  1.3089 +        throws java.io.IOException, ClassNotFoundException {
  1.3090 +        /*
  1.3091 +         * In order to maintain compatibility with previous serialized forms,
  1.3092 +         * the magnitude of a BigInteger is serialized as an array of bytes.
  1.3093 +         * The magnitude field is used as a temporary store for the byte array
  1.3094 +         * that is deserialized. The cached computation fields should be
  1.3095 +         * transient but are serialized for compatibility reasons.
  1.3096 +         */
  1.3097 +
  1.3098 +        // prepare to read the alternate persistent fields
  1.3099 +        ObjectInputStream.GetField fields = s.readFields();
  1.3100 +
  1.3101 +        // Read the alternate persistent fields that we care about
  1.3102 +        int sign = fields.get("signum", -2);
  1.3103 +        byte[] magnitude = (byte[])fields.get("magnitude", null);
  1.3104 +
  1.3105 +        // Validate signum
  1.3106 +        if (sign < -1 || sign > 1) {
  1.3107 +            String message = "BigInteger: Invalid signum value";
  1.3108 +            if (fields.defaulted("signum"))
  1.3109 +                message = "BigInteger: Signum not present in stream";
  1.3110 +            throw new java.io.StreamCorruptedException(message);
  1.3111 +        }
  1.3112 +        if ((magnitude.length == 0) != (sign == 0)) {
  1.3113 +            String message = "BigInteger: signum-magnitude mismatch";
  1.3114 +            if (fields.defaulted("magnitude"))
  1.3115 +                message = "BigInteger: Magnitude not present in stream";
  1.3116 +            throw new java.io.StreamCorruptedException(message);
  1.3117 +        }
  1.3118 +
  1.3119 +        // Commit final fields via Unsafe
  1.3120 +        unsafe.putIntVolatile(this, signumOffset, sign);
  1.3121 +
  1.3122 +        // Calculate mag field from magnitude and discard magnitude
  1.3123 +        unsafe.putObjectVolatile(this, magOffset,
  1.3124 +                                 stripLeadingZeroBytes(magnitude));
  1.3125 +    }
  1.3126 +
  1.3127 +    // Support for resetting final fields while deserializing
  1.3128 +    private static final sun.misc.Unsafe unsafe = sun.misc.Unsafe.getUnsafe();
  1.3129 +    private static final long signumOffset;
  1.3130 +    private static final long magOffset;
  1.3131 +    static {
  1.3132 +        try {
  1.3133 +            signumOffset = unsafe.objectFieldOffset
  1.3134 +                (BigInteger.class.getDeclaredField("signum"));
  1.3135 +            magOffset = unsafe.objectFieldOffset
  1.3136 +                (BigInteger.class.getDeclaredField("mag"));
  1.3137 +        } catch (Exception ex) {
  1.3138 +            throw new Error(ex);
  1.3139 +        }
  1.3140 +    }
  1.3141 +
  1.3142 +    /**
  1.3143 +     * Save the {@code BigInteger} instance to a stream.
  1.3144 +     * The magnitude of a BigInteger is serialized as a byte array for
  1.3145 +     * historical reasons.
  1.3146 +     *
  1.3147 +     * @serialData two necessary fields are written as well as obsolete
  1.3148 +     *             fields for compatibility with older versions.
  1.3149 +     */
  1.3150 +    private void writeObject(ObjectOutputStream s) throws IOException {
  1.3151 +        // set the values of the Serializable fields
  1.3152 +        ObjectOutputStream.PutField fields = s.putFields();
  1.3153 +        fields.put("signum", signum);
  1.3154 +        fields.put("magnitude", magSerializedForm());
  1.3155 +        // The values written for cached fields are compatible with older
  1.3156 +        // versions, but are ignored in readObject so don't otherwise matter.
  1.3157 +        fields.put("bitCount", -1);
  1.3158 +        fields.put("bitLength", -1);
  1.3159 +        fields.put("lowestSetBit", -2);
  1.3160 +        fields.put("firstNonzeroByteNum", -2);
  1.3161 +
  1.3162 +        // save them
  1.3163 +        s.writeFields();
  1.3164 +}
  1.3165 +
  1.3166 +    /**
  1.3167 +     * Returns the mag array as an array of bytes.
  1.3168 +     */
  1.3169 +    private byte[] magSerializedForm() {
  1.3170 +        int len = mag.length;
  1.3171 +
  1.3172 +        int bitLen = (len == 0 ? 0 : ((len - 1) << 5) + bitLengthForInt(mag[0]));
  1.3173 +        int byteLen = (bitLen + 7) >>> 3;
  1.3174 +        byte[] result = new byte[byteLen];
  1.3175 +
  1.3176 +        for (int i = byteLen - 1, bytesCopied = 4, intIndex = len - 1, nextInt = 0;
  1.3177 +             i>=0; i--) {
  1.3178 +            if (bytesCopied == 4) {
  1.3179 +                nextInt = mag[intIndex--];
  1.3180 +                bytesCopied = 1;
  1.3181 +            } else {
  1.3182 +                nextInt >>>= 8;
  1.3183 +                bytesCopied++;
  1.3184 +            }
  1.3185 +            result[i] = (byte)nextInt;
  1.3186 +        }
  1.3187 +        return result;
  1.3188 +    }
  1.3189 +}