mpz.c 37.8 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
/*
 * This file is part of the Micro Python project, http://micropython.org/
 *
 * The MIT License (MIT)
 *
 * Copyright (c) 2013, 2014 Damien P. George
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 * THE SOFTWARE.
 */

27
28
29
30
31
32
33
#include <stdint.h>
#include <stdbool.h>
#include <stdlib.h>
#include <string.h>
#include <assert.h>

#include "mpconfig.h"
34
#include "misc.h"
35
36
37
38
#include "mpz.h"

#if MICROPY_LONGINT_IMPL == MICROPY_LONGINT_IMPL_MPZ

39
#define DIG_SIZE (MPZ_DIG_SIZE)
40
41
42
#define DIG_MASK ((1L << DIG_SIZE) - 1)
#define DIG_MSB  (1L << (DIG_SIZE - 1))
#define DIG_BASE (1L << DIG_SIZE)
43
44

/*
45
46
47
48
49
50
51
52
53
54
 mpz is an arbitrary precision integer type with a public API.

 mpn functions act on non-negative integers represented by an array of generalised
 digits (eg a word per digit).  You also need to specify separately the length of the
 array.  There is no public API for mpn.  Rather, the functions are used by mpz to
 implement its features.

 Integer values are stored little endian (first digit is first in memory).

 Definition of normalise: ?
55
56
57
58
59
60
*/

/* compares i with j
   returns sign(i - j)
   assumes i, j are normalised
*/
61
STATIC int mpn_cmp(const mpz_dig_t *idig, mp_uint_t ilen, const mpz_dig_t *jdig, mp_uint_t jlen) {
62
63
64
65
    if (ilen < jlen) { return -1; }
    if (ilen > jlen) { return 1; }

    for (idig += ilen, jdig += ilen; ilen > 0; --ilen) {
66
        mpz_dbl_dig_signed_t cmp = (mpz_dbl_dig_t)*(--idig) - (mpz_dbl_dig_t)*(--jdig);
67
68
69
70
71
72
73
        if (cmp < 0) { return -1; }
        if (cmp > 0) { return 1; }
    }

    return 0;
}

74
75
/* computes i = j << n
   returns number of digits in i
76
   assumes enough memory in i; assumes normalised j; assumes n > 0
77
78
   can have i, j pointing to same memory
*/
79
80
81
STATIC mp_uint_t mpn_shl(mpz_dig_t *idig, mpz_dig_t *jdig, mp_uint_t jlen, mp_uint_t n) {
    mp_uint_t n_whole = (n + DIG_SIZE - 1) / DIG_SIZE;
    mp_uint_t n_part = n % DIG_SIZE;
82
83
84
    if (n_part == 0) {
        n_part = DIG_SIZE;
    }
85

86
87
88
    // start from the high end of the digit arrays
    idig += jlen + n_whole - 1;
    jdig += jlen - 1;
89

90
91
    // shift the digits
    mpz_dbl_dig_t d = 0;
92
    for (mp_uint_t i = jlen; i > 0; i--, idig--, jdig--) {
93
        d |= *jdig;
94
        *idig = (d >> (DIG_SIZE - n_part)) & DIG_MASK;
95
        d <<= DIG_SIZE;
96
97
    }

98
    // store remaining bits
99
    *idig = (d >> (DIG_SIZE - n_part)) & DIG_MASK;
100
    idig -= n_whole - 1;
101
    memset(idig, 0, (n_whole - 1) * sizeof(mpz_dig_t));
102
103
104
105
106

    // work out length of result
    jlen += n_whole;
    if (idig[jlen - 1] == 0) {
        jlen--;
107
108
    }

109
    // return length of result
110
111
112
    return jlen;
}

113
114
/* computes i = j >> n
   returns number of digits in i
115
   assumes enough memory in i; assumes normalised j; assumes n > 0
116
117
   can have i, j pointing to same memory
*/
118
119
120
STATIC mp_uint_t mpn_shr(mpz_dig_t *idig, mpz_dig_t *jdig, mp_uint_t jlen, mp_uint_t n) {
    mp_uint_t n_whole = n / DIG_SIZE;
    mp_uint_t n_part = n % DIG_SIZE;
121
122
123
124
125
126
127
128

    if (n_whole >= jlen) {
        return 0;
    }

    jdig += n_whole;
    jlen -= n_whole;

129
    for (mp_uint_t i = jlen; i > 0; i--, idig++, jdig++) {
130
        mpz_dbl_dig_t d = *jdig;
131
        if (i > 1) {
132
            d |= (mpz_dbl_dig_t)jdig[1] << DIG_SIZE;
133
        }
134
135
136
137
138
        d >>= n_part;
        *idig = d & DIG_MASK;
    }

    if (idig[-1] == 0) {
139
        jlen--;
140
141
142
143
144
145
146
147
148
149
    }

    return jlen;
}

/* computes i = j + k
   returns number of digits in i
   assumes enough memory in i; assumes normalised j, k; assumes jlen >= klen
   can have i, j, k pointing to same memory
*/
150
STATIC mp_uint_t mpn_add(mpz_dig_t *idig, const mpz_dig_t *jdig, mp_uint_t jlen, const mpz_dig_t *kdig, mp_uint_t klen) {
151
152
153
154
155
156
    mpz_dig_t *oidig = idig;
    mpz_dbl_dig_t carry = 0;

    jlen -= klen;

    for (; klen > 0; --klen, ++idig, ++jdig, ++kdig) {
157
        carry += (mpz_dbl_dig_t)*jdig + (mpz_dbl_dig_t)*kdig;
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
        *idig = carry & DIG_MASK;
        carry >>= DIG_SIZE;
    }

    for (; jlen > 0; --jlen, ++idig, ++jdig) {
        carry += *jdig;
        *idig = carry & DIG_MASK;
        carry >>= DIG_SIZE;
    }

    if (carry != 0) {
        *idig++ = carry;
    }

    return idig - oidig;
}

/* computes i = j - k
   returns number of digits in i
   assumes enough memory in i; assumes normalised j, k; assumes j >= k
   can have i, j, k pointing to same memory
*/
180
STATIC mp_uint_t mpn_sub(mpz_dig_t *idig, const mpz_dig_t *jdig, mp_uint_t jlen, const mpz_dig_t *kdig, mp_uint_t klen) {
181
182
183
184
185
186
    mpz_dig_t *oidig = idig;
    mpz_dbl_dig_signed_t borrow = 0;

    jlen -= klen;

    for (; klen > 0; --klen, ++idig, ++jdig, ++kdig) {
187
        borrow += (mpz_dbl_dig_t)*jdig - (mpz_dbl_dig_t)*kdig;
188
189
190
191
        *idig = borrow & DIG_MASK;
        borrow >>= DIG_SIZE;
    }

192
    for (; jlen > 0; --jlen, ++idig, ++jdig) {
193
194
195
196
197
198
199
200
201
202
203
        borrow += *jdig;
        *idig = borrow & DIG_MASK;
        borrow >>= DIG_SIZE;
    }

    for (--idig; idig >= oidig && *idig == 0; --idig) {
    }

    return idig + 1 - oidig;
}

204
205
206
207
208
/* computes i = j & k
   returns number of digits in i
   assumes enough memory in i; assumes normalised j, k; assumes jlen >= klen
   can have i, j, k pointing to same memory
*/
209
STATIC mp_uint_t mpn_and(mpz_dig_t *idig, const mpz_dig_t *jdig, mp_uint_t jlen, const mpz_dig_t *kdig, mp_uint_t klen) {
210
211
212
213
214
215
    mpz_dig_t *oidig = idig;

    for (; klen > 0; --klen, ++idig, ++jdig, ++kdig) {
        *idig = *jdig & *kdig;
    }

216
    // remove trailing zeros
Damien George's avatar
Damien George committed
217
    for (--idig; idig >= oidig && *idig == 0; --idig) {
218
219
    }

Damien George's avatar
Damien George committed
220
    return idig + 1 - oidig;
221
222
}

223
224
225
226
227
/* computes i = j & -k = j & (~k + 1)
   returns number of digits in i
   assumes enough memory in i; assumes normalised j, k
   can have i, j, k pointing to same memory
*/
228
STATIC mp_uint_t mpn_and_neg(mpz_dig_t *idig, const mpz_dig_t *jdig, mp_uint_t jlen, const mpz_dig_t *kdig, mp_uint_t klen) {
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
    mpz_dig_t *oidig = idig;
    mpz_dbl_dig_t carry = 1;

    for (; jlen > 0 && klen > 0; --jlen, --klen, ++idig, ++jdig, ++kdig) {
        carry += *kdig ^ DIG_MASK;
        *idig = (*jdig & carry) & DIG_MASK;
        carry >>= DIG_SIZE;
    }

    for (; jlen > 0; --jlen, ++idig, ++jdig) {
        carry += DIG_MASK;
        *idig = (*jdig & carry) & DIG_MASK;
        carry >>= DIG_SIZE;
    }

    if (carry != 0) {
        *idig = carry;
    } else {
        // remove trailing zeros
        for (--idig; idig >= oidig && *idig == 0; --idig) {
        }
    }

    return idig + 1 - oidig;
}

255
256
257
258
259
/* computes i = j | k
   returns number of digits in i
   assumes enough memory in i; assumes normalised j, k; assumes jlen >= klen
   can have i, j, k pointing to same memory
*/
260
STATIC mp_uint_t mpn_or(mpz_dig_t *idig, const mpz_dig_t *jdig, mp_uint_t jlen, const mpz_dig_t *kdig, mp_uint_t klen) {
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
    mpz_dig_t *oidig = idig;

    jlen -= klen;

    for (; klen > 0; --klen, ++idig, ++jdig, ++kdig) {
        *idig = *jdig | *kdig;
    }

    for (; jlen > 0; --jlen, ++idig, ++jdig) {
        *idig = *jdig;
    }

    return idig - oidig;
}

/* computes i = j ^ k
   returns number of digits in i
   assumes enough memory in i; assumes normalised j, k; assumes jlen >= klen
   can have i, j, k pointing to same memory
*/
281
STATIC mp_uint_t mpn_xor(mpz_dig_t *idig, const mpz_dig_t *jdig, mp_uint_t jlen, const mpz_dig_t *kdig, mp_uint_t klen) {
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
    mpz_dig_t *oidig = idig;

    jlen -= klen;

    for (; klen > 0; --klen, ++idig, ++jdig, ++kdig) {
        *idig = *jdig ^ *kdig;
    }

    for (; jlen > 0; --jlen, ++idig, ++jdig) {
        *idig = *jdig;
    }

    return idig - oidig;
}

297
298
299
300
/* computes i = i * d1 + d2
   returns number of digits in i
   assumes enough memory in i; assumes normalised i; assumes dmul != 0
*/
301
STATIC mp_uint_t mpn_mul_dig_add_dig(mpz_dig_t *idig, mp_uint_t ilen, mpz_dig_t dmul, mpz_dig_t dadd) {
302
303
304
305
    mpz_dig_t *oidig = idig;
    mpz_dbl_dig_t carry = dadd;

    for (; ilen > 0; --ilen, ++idig) {
306
        carry += (mpz_dbl_dig_t)*idig * (mpz_dbl_dig_t)dmul; // will never overflow so long as DIG_SIZE <= 8*sizeof(mpz_dbl_dig_t)/2
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
        *idig = carry & DIG_MASK;
        carry >>= DIG_SIZE;
    }

    if (carry != 0) {
        *idig++ = carry;
    }

    return idig - oidig;
}

/* computes i = j * k
   returns number of digits in i
   assumes enough memory in i; assumes i is zeroed; assumes normalised j, k
   can have j, k point to same memory
*/
323
STATIC mp_uint_t mpn_mul(mpz_dig_t *idig, mpz_dig_t *jdig, mp_uint_t jlen, mpz_dig_t *kdig, mp_uint_t klen) {
324
    mpz_dig_t *oidig = idig;
325
    mp_uint_t ilen = 0;
326
327
328
329
330

    for (; klen > 0; --klen, ++idig, ++kdig) {
        mpz_dig_t *id = idig;
        mpz_dbl_dig_t carry = 0;

331
        mp_uint_t jl = jlen;
332
        for (mpz_dig_t *jd = jdig; jl > 0; --jl, ++jd, ++id) {
333
            carry += (mpz_dbl_dig_t)*id + (mpz_dbl_dig_t)*jd * (mpz_dbl_dig_t)*kdig; // will never overflow so long as DIG_SIZE <= 8*sizeof(mpz_dbl_dig_t)/2
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
            *id = carry & DIG_MASK;
            carry >>= DIG_SIZE;
        }

        if (carry != 0) {
            *id++ = carry;
        }

        ilen = id - oidig;
    }

    return ilen;
}

/* natural_div - quo * den + new_num = old_num (ie num is replaced with rem)
   assumes den != 0
   assumes num_dig has enough memory to be extended by 1 digit
   assumes quo_dig has enough memory (as many digits as num)
   assumes quo_dig is filled with zeros
   modifies den_dig memory, but restors it to original state at end
*/

356
STATIC void mpn_div(mpz_dig_t *num_dig, mp_uint_t *num_len, mpz_dig_t *den_dig, mp_uint_t den_len, mpz_dig_t *quo_dig, mp_uint_t *quo_len) {
357
358
359
360
361
362
363
    mpz_dig_t *orig_num_dig = num_dig;
    mpz_dig_t *orig_quo_dig = quo_dig;
    mpz_dig_t norm_shift = 0;
    mpz_dbl_dig_t lead_den_digit;

    // handle simple cases
    {
364
        int cmp = mpn_cmp(num_dig, *num_len, den_dig, den_len);
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
        if (cmp == 0) {
            *num_len = 0;
            quo_dig[0] = 1;
            *quo_len = 1;
            return;
        } else if (cmp < 0) {
            // numerator remains the same
            *quo_len = 0;
            return;
        }
    }

    // count number of leading zeros in leading digit of denominator
    {
        mpz_dig_t d = den_dig[den_len - 1];
380
        while ((d & DIG_MSB) == 0) {
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
            d <<= 1;
            ++norm_shift;
        }
    }

    // normalise denomenator (leading bit of leading digit is 1)
    for (mpz_dig_t *den = den_dig, carry = 0; den < den_dig + den_len; ++den) {
        mpz_dig_t d = *den;
        *den = ((d << norm_shift) | carry) & DIG_MASK;
        carry = d >> (DIG_SIZE - norm_shift);
    }

    // now need to shift numerator by same amount as denominator
    // first, increase length of numerator in case we need more room to shift
    num_dig[*num_len] = 0;
    ++(*num_len);
    for (mpz_dig_t *num = num_dig, carry = 0; num < num_dig + *num_len; ++num) {
        mpz_dig_t n = *num;
        *num = ((n << norm_shift) | carry) & DIG_MASK;
        carry = n >> (DIG_SIZE - norm_shift);
    }

    // cache the leading digit of the denominator
    lead_den_digit = den_dig[den_len - 1];

    // point num_dig to last digit in numerator
    num_dig += *num_len - 1;

    // calculate number of digits in quotient
    *quo_len = *num_len - den_len;

    // point to last digit to store for quotient
    quo_dig += *quo_len - 1;

    // keep going while we have enough digits to divide
    while (*num_len > den_len) {
417
        mpz_dbl_dig_t quo = ((mpz_dbl_dig_t)*num_dig << DIG_SIZE) | num_dig[-1];
418
419
420
421

        // get approximate quotient
        quo /= lead_den_digit;

422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
        // Multiply quo by den and subtract from num to get remainder.
        // We have different code here to handle different compile-time
        // configurations of mpz:
        //
        //   1. DIG_SIZE is stricly less than half the number of bits
        //      available in mpz_dbl_dig_t.  In this case we can use a
        //      slightly more optimal (in time and space) routine that
        //      uses the extra bits in mpz_dbl_dig_signed_t to store a
        //      sign bit.
        //
        //   2. DIG_SIZE is exactly half the number of bits available in
        //      mpz_dbl_dig_t.  In this (common) case we need to be careful
        //      not to overflow the borrow variable.  And the shifting of
        //      borrow needs some special logic (it's a shift right with
        //      round up).

        if (DIG_SIZE < 8 * sizeof(mpz_dbl_dig_t) / 2) {
439
440
441
            mpz_dbl_dig_signed_t borrow = 0;

            for (mpz_dig_t *n = num_dig - den_len, *d = den_dig; n < num_dig; ++n, ++d) {
442
                borrow += (mpz_dbl_dig_t)*n - (mpz_dbl_dig_t)quo * (mpz_dbl_dig_t)*d; // will overflow if DIG_SIZE >= 8*sizeof(mpz_dbl_dig_t)/2
443
444
445
                *n = borrow & DIG_MASK;
                borrow >>= DIG_SIZE;
            }
446
            borrow += *num_dig; // will overflow if DIG_SIZE >= 8*sizeof(mpz_dbl_dig_t)/2
447
448
449
450
451
452
453
            *num_dig = borrow & DIG_MASK;
            borrow >>= DIG_SIZE;

            // adjust quotient if it is too big
            for (; borrow != 0; --quo) {
                mpz_dbl_dig_t carry = 0;
                for (mpz_dig_t *n = num_dig - den_len, *d = den_dig; n < num_dig; ++n, ++d) {
454
                    carry += (mpz_dbl_dig_t)*n + (mpz_dbl_dig_t)*d;
455
456
457
458
459
460
461
462
463
                    *n = carry & DIG_MASK;
                    carry >>= DIG_SIZE;
                }
                carry += *num_dig;
                *num_dig = carry & DIG_MASK;
                carry >>= DIG_SIZE;

                borrow += carry;
            }
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
        } else { // DIG_SIZE == 8 * sizeof(mpz_dbl_dig_t) / 2
            mpz_dbl_dig_t borrow = 0;

            for (mpz_dig_t *n = num_dig - den_len, *d = den_dig; n < num_dig; ++n, ++d) {
                mpz_dbl_dig_t x = (mpz_dbl_dig_t)quo * (mpz_dbl_dig_t)(*d);
                if (x >= *n || *n - x <= borrow) {
                    borrow += (mpz_dbl_dig_t)x - (mpz_dbl_dig_t)*n;
                    *n = (-borrow) & DIG_MASK;
                    borrow = (borrow >> DIG_SIZE) + ((borrow & DIG_MASK) == 0 ? 0 : 1); // shift-right with round-up
                } else {
                    *n = ((mpz_dbl_dig_t)*n - (mpz_dbl_dig_t)x - (mpz_dbl_dig_t)borrow) & DIG_MASK;
                    borrow = 0;
                }
            }
            if (borrow >= *num_dig) {
                borrow -= (mpz_dbl_dig_t)*num_dig;
                *num_dig = (-borrow) & DIG_MASK;
                borrow = (borrow >> DIG_SIZE) + ((borrow & DIG_MASK) == 0 ? 0 : 1); // shift-right with round-up
            } else {
                *num_dig = (*num_dig - borrow) & DIG_MASK;
                borrow = 0;
            }

            // adjust quotient if it is too big
            for (; borrow != 0; --quo) {
                mpz_dbl_dig_t carry = 0;
                for (mpz_dig_t *n = num_dig - den_len, *d = den_dig; n < num_dig; ++n, ++d) {
                    carry += (mpz_dbl_dig_t)*n + (mpz_dbl_dig_t)*d;
                    *n = carry & DIG_MASK;
                    carry >>= DIG_SIZE;
                }
                carry += (mpz_dbl_dig_t)*num_dig;
                *num_dig = carry & DIG_MASK;
                carry >>= DIG_SIZE;

                //assert(borrow >= carry); // enable this to check the logic
                borrow -= carry;
            }
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
        }

        // store this digit of the quotient
        *quo_dig = quo & DIG_MASK;
        --quo_dig;

        // move down to next digit of numerator
        --num_dig;
        --(*num_len);
    }

    // unnormalise denomenator
    for (mpz_dig_t *den = den_dig + den_len - 1, carry = 0; den >= den_dig; --den) {
        mpz_dig_t d = *den;
        *den = ((d >> norm_shift) | carry) & DIG_MASK;
        carry = d << (DIG_SIZE - norm_shift);
    }

    // unnormalise numerator (remainder now)
    for (mpz_dig_t *num = orig_num_dig + *num_len - 1, carry = 0; num >= orig_num_dig; --num) {
        mpz_dig_t n = *num;
        *num = ((n >> norm_shift) | carry) & DIG_MASK;
        carry = n << (DIG_SIZE - norm_shift);
    }

    // strip trailing zeros

    while (*quo_len > 0 && orig_quo_dig[*quo_len - 1] == 0) {
        --(*quo_len);
    }

    while (*num_len > 0 && orig_num_dig[*num_len - 1] == 0) {
        --(*num_len);
    }
}

538
#define MIN_ALLOC (2)
539

540
STATIC const uint8_t log_base2_floor[] = {
541
542
543
544
545
546
547
548
549
550
551
552
553
    0,
    0, 1, 1, 2,
    2, 2, 2, 3,
    3, 3, 3, 3,
    3, 3, 3, 4,
    4, 4, 4, 4,
    4, 4, 4, 4,
    4, 4, 4, 4,
    4, 4, 4, 5
};

void mpz_init_zero(mpz_t *z) {
    z->neg = 0;
554
555
    z->fixed_dig = 0;
    z->alloc = 0;
556
557
558
559
    z->len = 0;
    z->dig = NULL;
}

560
void mpz_init_from_int(mpz_t *z, mp_int_t val) {
561
562
563
564
    mpz_init_zero(z);
    mpz_set_from_int(z, val);
}

565
void mpz_init_fixed_from_int(mpz_t *z, mpz_dig_t *dig, mp_uint_t alloc, mp_int_t val) {
566
567
568
569
570
571
572
573
    z->neg = 0;
    z->fixed_dig = 1;
    z->alloc = alloc;
    z->len = 0;
    z->dig = dig;
    mpz_set_from_int(z, val);
}

574
void mpz_deinit(mpz_t *z) {
575
    if (z != NULL && !z->fixed_dig) {
576
577
578
579
580
581
582
583
584
585
        m_del(mpz_dig_t, z->dig, z->alloc);
    }
}

mpz_t *mpz_zero(void) {
    mpz_t *z = m_new_obj(mpz_t);
    mpz_init_zero(z);
    return z;
}

586
mpz_t *mpz_from_int(mp_int_t val) {
587
588
589
590
591
    mpz_t *z = mpz_zero();
    mpz_set_from_int(z, val);
    return z;
}

592
mpz_t *mpz_from_ll(long long val, bool is_signed) {
593
    mpz_t *z = mpz_zero();
594
    mpz_set_from_ll(z, val, is_signed);
595
596
597
    return z;
}

598
mpz_t *mpz_from_str(const char *str, mp_uint_t len, bool neg, mp_uint_t base) {
599
600
601
602
603
604
605
606
607
608
609
610
    mpz_t *z = mpz_zero();
    mpz_set_from_str(z, str, len, neg, base);
    return z;
}

void mpz_free(mpz_t *z) {
    if (z != NULL) {
        m_del(mpz_dig_t, z->dig, z->alloc);
        m_del_obj(mpz_t, z);
    }
}

611
STATIC void mpz_need_dig(mpz_t *z, mp_uint_t need) {
612
    if (need < MIN_ALLOC) {
613
        need = MIN_ALLOC;
614
615
    }

616
617
618
619
620
621
622
623
    if (z->dig == NULL || z->alloc < need) {
        if (z->fixed_dig) {
            // cannot reallocate fixed buffers
            assert(0);
            return;
        }
        z->dig = m_renew(mpz_dig_t, z->dig, z->alloc, need);
        z->alloc = need;
624
625
626
627
628
629
    }
}

mpz_t *mpz_clone(const mpz_t *src) {
    mpz_t *z = m_new_obj(mpz_t);
    z->neg = src->neg;
630
631
    z->fixed_dig = 0;
    z->alloc = src->alloc;
632
633
634
635
636
637
638
639
640
641
    z->len = src->len;
    if (src->dig == NULL) {
        z->dig = NULL;
    } else {
        z->dig = m_new(mpz_dig_t, z->alloc);
        memcpy(z->dig, src->dig, src->alloc * sizeof(mpz_dig_t));
    }
    return z;
}

642
643
644
/* sets dest = src
   can have dest, src the same
*/
645
646
647
648
649
650
651
void mpz_set(mpz_t *dest, const mpz_t *src) {
    mpz_need_dig(dest, src->len);
    dest->neg = src->neg;
    dest->len = src->len;
    memcpy(dest->dig, src->dig, src->len * sizeof(mpz_dig_t));
}

652
void mpz_set_from_int(mpz_t *z, mp_int_t val) {
653
    mpz_need_dig(z, MPZ_NUM_DIG_FOR_INT);
654

655
    mp_uint_t uval;
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
    if (val < 0) {
        z->neg = 1;
        uval = -val;
    } else {
        z->neg = 0;
        uval = val;
    }

    z->len = 0;
    while (uval > 0) {
        z->dig[z->len++] = uval & DIG_MASK;
        uval >>= DIG_SIZE;
    }
}

671
void mpz_set_from_ll(mpz_t *z, long long val, bool is_signed) {
672
673
674
    mpz_need_dig(z, MPZ_NUM_DIG_FOR_LL);

    unsigned long long uval;
675
    if (is_signed && val < 0) {
676
        z->neg = 1;
677
        uval = -val;
678
679
    } else {
        z->neg = 0;
680
        uval = val;
681
682
683
    }

    z->len = 0;
684
685
686
    while (uval > 0) {
        z->dig[z->len++] = uval & DIG_MASK;
        uval >>= DIG_SIZE;
687
688
689
690
    }
}

// returns number of bytes from str that were processed
691
mp_uint_t mpz_set_from_str(mpz_t *z, const char *str, mp_uint_t len, bool neg, mp_uint_t base) {
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
    assert(base < 36);

    const char *cur = str;
    const char *top = str + len;

    mpz_need_dig(z, len * 8 / DIG_SIZE + 1);

    if (neg) {
        z->neg = 1;
    } else {
        z->neg = 0;
    }

    z->len = 0;
    for (; cur < top; ++cur) { // XXX UTF8 next char
707
708
        //mp_uint_t v = char_to_numeric(cur#); // XXX UTF8 get char
        mp_uint_t v = *cur;
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
        if ('0' <= v && v <= '9') {
            v -= '0';
        } else if ('A' <= v && v <= 'Z') {
            v -= 'A' - 10;
        } else if ('a' <= v && v <= 'z') {
            v -= 'a' - 10;
        } else {
            break;
        }
        if (v >= base) {
            break;
        }
        z->len = mpn_mul_dig_add_dig(z->dig, z->len, base, v);
    }

    return cur - str;
}

bool mpz_is_zero(const mpz_t *z) {
    return z->len == 0;
}

bool mpz_is_pos(const mpz_t *z) {
    return z->len > 0 && z->neg == 0;
}

bool mpz_is_neg(const mpz_t *z) {
    return z->len > 0 && z->neg != 0;
}

bool mpz_is_odd(const mpz_t *z) {
    return z->len > 0 && (z->dig[0] & 1) != 0;
}

bool mpz_is_even(const mpz_t *z) {
    return z->len == 0 || (z->dig[0] & 1) == 0;
}

747
748
int mpz_cmp(const mpz_t *z1, const mpz_t *z2) {
    int cmp = (int)z2->neg - (int)z1->neg;
749
750
751
752
753
754
755
756
757
758
    if (cmp != 0) {
        return cmp;
    }
    cmp = mpn_cmp(z1->dig, z1->len, z2->dig, z2->len);
    if (z1->neg != 0) {
        cmp = -cmp;
    }
    return cmp;
}

759
760
761
#if 0
// obsolete
// compares mpz with an integer that fits within DIG_SIZE bits
762
763
mp_int_t mpz_cmp_sml_int(const mpz_t *z, mp_int_t sml_int) {
    mp_int_t cmp;
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
    if (z->neg == 0) {
        if (sml_int < 0) return 1;
        if (sml_int == 0) {
            if (z->len == 0) return 0;
            return 1;
        }
        if (z->len == 0) return -1;
        assert(sml_int < (1 << DIG_SIZE));
        if (z->len != 1) return 1;
        cmp = z->dig[0] - sml_int;
    } else {
        if (sml_int > 0) return -1;
        if (sml_int == 0) {
            if (z->len == 0) return 0;
            return -1;
        }
        if (z->len == 0) return 1;
        assert(sml_int > -(1 << DIG_SIZE));
        if (z->len != 1) return -1;
        cmp = -z->dig[0] - sml_int;
    }
    if (cmp < 0) return -1;
    if (cmp > 0) return 1;
    return 0;
}
789
#endif
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866

#if 0
these functions are unused

/* returns abs(z)
*/
mpz_t *mpz_abs(const mpz_t *z) {
    mpz_t *z2 = mpz_clone(z);
    z2->neg = 0;
    return z2;
}

/* returns -z
*/
mpz_t *mpz_neg(const mpz_t *z) {
    mpz_t *z2 = mpz_clone(z);
    z2->neg = 1 - z2->neg;
    return z2;
}

/* returns lhs + rhs
   can have lhs, rhs the same
*/
mpz_t *mpz_add(const mpz_t *lhs, const mpz_t *rhs) {
    mpz_t *z = mpz_zero();
    mpz_add_inpl(z, lhs, rhs);
    return z;
}

/* returns lhs - rhs
   can have lhs, rhs the same
*/
mpz_t *mpz_sub(const mpz_t *lhs, const mpz_t *rhs) {
    mpz_t *z = mpz_zero();
    mpz_sub_inpl(z, lhs, rhs);
    return z;
}

/* returns lhs * rhs
   can have lhs, rhs the same
*/
mpz_t *mpz_mul(const mpz_t *lhs, const mpz_t *rhs) {
    mpz_t *z = mpz_zero();
    mpz_mul_inpl(z, lhs, rhs);
    return z;
}

/* returns lhs ** rhs
   can have lhs, rhs the same
*/
mpz_t *mpz_pow(const mpz_t *lhs, const mpz_t *rhs) {
    mpz_t *z = mpz_zero();
    mpz_pow_inpl(z, lhs, rhs);
    return z;
}
#endif

/* computes dest = abs(z)
   can have dest, z the same
*/
void mpz_abs_inpl(mpz_t *dest, const mpz_t *z) {
    if (dest != z) {
        mpz_set(dest, z);
    }
    dest->neg = 0;
}

/* computes dest = -z
   can have dest, z the same
*/
void mpz_neg_inpl(mpz_t *dest, const mpz_t *z) {
    if (dest != z) {
        mpz_set(dest, z);
    }
    dest->neg = 1 - dest->neg;
}

867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
/* computes dest = ~z (= -z - 1)
   can have dest, z the same
*/
void mpz_not_inpl(mpz_t *dest, const mpz_t *z) {
    if (dest != z) {
        mpz_set(dest, z);
    }
    if (dest->neg) {
        dest->neg = 0;
        mpz_dig_t k = 1;
        dest->len = mpn_sub(dest->dig, dest->dig, dest->len, &k, 1);
    } else {
        mpz_dig_t k = 1;
        dest->len = mpn_add(dest->dig, dest->dig, dest->len, &k, 1);
        dest->neg = 1;
    }
}

885
886
887
/* computes dest = lhs << rhs
   can have dest, lhs the same
*/
888
void mpz_shl_inpl(mpz_t *dest, const mpz_t *lhs, mp_int_t rhs) {
889
    if (lhs->len == 0 || rhs == 0) {
890
        mpz_set(dest, lhs);
891
892
    } else if (rhs < 0) {
        mpz_shr_inpl(dest, lhs, -rhs);
893
    } else {
894
895
896
        mpz_need_dig(dest, lhs->len + (rhs + DIG_SIZE - 1) / DIG_SIZE);
        dest->len = mpn_shl(dest->dig, lhs->dig, lhs->len, rhs);
        dest->neg = lhs->neg;
897
898
899
900
901
902
    }
}

/* computes dest = lhs >> rhs
   can have dest, lhs the same
*/
903
void mpz_shr_inpl(mpz_t *dest, const mpz_t *lhs, mp_int_t rhs) {
904
    if (lhs->len == 0 || rhs == 0) {
905
        mpz_set(dest, lhs);
906
907
    } else if (rhs < 0) {
        mpz_shl_inpl(dest, lhs, -rhs);
908
    } else {
909
910
911
912
913
        mpz_need_dig(dest, lhs->len);
        dest->len = mpn_shr(dest->dig, lhs->dig, lhs->len, rhs);
        dest->neg = lhs->neg;
        if (dest->neg) {
            // arithmetic shift right, rounding to negative infinity
914
915
            mp_uint_t n_whole = rhs / DIG_SIZE;
            mp_uint_t n_part = rhs % DIG_SIZE;
916
            mpz_dig_t round_up = 0;
917
            for (mp_uint_t i = 0; i < lhs->len && i < n_whole; i++) {
918
919
920
921
922
923
924
925
926
927
928
929
                if (lhs->dig[i] != 0) {
                    round_up = 1;
                    break;
                }
            }
            if (n_whole < lhs->len && (lhs->dig[n_whole] & ((1 << n_part) - 1)) != 0) {
                round_up = 1;
            }
            if (round_up) {
                dest->len = mpn_add(dest->dig, dest->dig, dest->len, &round_up, 1);
            }
        }
930
931
932
    }
}

933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
/* computes dest = lhs + rhs
   can have dest, lhs, rhs the same
*/
void mpz_add_inpl(mpz_t *dest, const mpz_t *lhs, const mpz_t *rhs) {
    if (mpn_cmp(lhs->dig, lhs->len, rhs->dig, rhs->len) < 0) {
        const mpz_t *temp = lhs;
        lhs = rhs;
        rhs = temp;
    }

    if (lhs->neg == rhs->neg) {
        mpz_need_dig(dest, lhs->len + 1);
        dest->len = mpn_add(dest->dig, lhs->dig, lhs->len, rhs->dig, rhs->len);
    } else {
        mpz_need_dig(dest, lhs->len);
        dest->len = mpn_sub(dest->dig, lhs->dig, lhs->len, rhs->dig, rhs->len);
    }

    dest->neg = lhs->neg;
}

/* computes dest = lhs - rhs
   can have dest, lhs, rhs the same
*/
void mpz_sub_inpl(mpz_t *dest, const mpz_t *lhs, const mpz_t *rhs) {
    bool neg = false;

    if (mpn_cmp(lhs->dig, lhs->len, rhs->dig, rhs->len) < 0) {
        const mpz_t *temp = lhs;
        lhs = rhs;
        rhs = temp;
        neg = true;
    }

    if (lhs->neg != rhs->neg) {
        mpz_need_dig(dest, lhs->len + 1);
        dest->len = mpn_add(dest->dig, lhs->dig, lhs->len, rhs->dig, rhs->len);
    } else {
        mpz_need_dig(dest, lhs->len);
        dest->len = mpn_sub(dest->dig, lhs->dig, lhs->len, rhs->dig, rhs->len);
    }

    if (neg) {
        dest->neg = 1 - lhs->neg;
    } else {
        dest->neg = lhs->neg;
    }
}

982
983
984
985
986
/* computes dest = lhs & rhs
   can have dest, lhs, rhs the same
*/
void mpz_and_inpl(mpz_t *dest, const mpz_t *lhs, const mpz_t *rhs) {
    if (lhs->neg == rhs->neg) {
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
        if (lhs->neg == 0) {
            // make sure lhs has the most digits
            if (lhs->len < rhs->len) {
                const mpz_t *temp = lhs;
                lhs = rhs;
                rhs = temp;
            }
            // do the and'ing
            mpz_need_dig(dest, rhs->len);
            dest->len = mpn_and(dest->dig, lhs->dig, lhs->len, rhs->dig, rhs->len);
            dest->neg = 0;
        } else {
            // TODO both args are negative
            assert(0);