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
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
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
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
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
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
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
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
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
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
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
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
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
//! ## `big_int` - Arbitrary precision, arbitrary base integer arithmetic library.
//! 
//! ```
//! use big_int::*;
//! 
//! let mut a: BigInt<10> = "9000000000000000000000000000000000000000".parse().unwrap();
//! 
//! a /= 13.into();
//! assert_eq!(a, "692307692307692307692307692307692307692".parse().unwrap());
//! 
//! let mut b: BigInt<16> = a.convert();
//! assert_eq!(b, "208D59C8D8669EDC306F76344EC4EC4EC".parse().unwrap());
//! 
//! b >>= 16.into();
//! let c: BigInt<2> = b.convert();
//! assert_eq!(c, "100000100011010101100111001000110110000110011010011110110111000011".parse().unwrap());
//! ```

use std::{
    cmp::Ordering,
    collections::VecDeque,
    fmt::Display,
    ops::{
        Add, AddAssign, Div, DivAssign, Mul, MulAssign, Neg, Shl, ShlAssign, Shr, ShrAssign, Sub,
        SubAssign,
    },
    str::FromStr,
};
use thiserror::Error;

pub const STANDARD_ALPHABET: &str =
    "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz+/";
pub const BASE64_ALPHABET: &str =
    "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";

#[derive(Error, Debug, PartialEq, Eq)]
pub enum BigIntError {
    #[error("base too large: number has {0} digits, alphabet can only represent {1} digits")]
    BaseTooHigh(usize, usize),
    #[error("parsing failed: {0}")]
    ParseFailed(ParseError),
    #[error("division by zero")]
    DivisionByZero,
}

#[derive(Error, Debug, PartialEq, Eq)]
pub enum ParseError {
    #[error("unrecognized character: {0:?}")]
    UnrecognizedCharacter(char),
    #[error("not enough characters")]
    NotEnoughCharacters,
    #[error("char {0:?} is {1}; too large to be represented in base {2}")]
    DigitTooLarge(char, usize, usize),
}

/// Safely retrieve items from a collection with negative indexing.
pub trait GetBack {
    type Item;

    /// Safely retrieve items from a collection with negative indexing.
    /// Returns `None` if the index is larger than the length of the collection.
    fn get_back(&self, index: usize) -> Option<&Self::Item>;
}

impl<T> GetBack for Vec<T> {
    type Item = T;
    fn get_back(&self, index: usize) -> Option<&Self::Item> {
        self.len()
            .checked_sub(index)
            .and_then(|index| self.get(index))
    }
}

/// Safely retrieve a mutable reference from a collection with negative indexing.
pub trait GetBackMut {
    type Item;

    /// Safely retrieve a mutable reference from a collection with negative indexing.
    /// Returns `None` if the index is larger than the length of the collection.
    fn get_back_mut(&mut self, index: usize) -> Option<&mut Self::Item>;
}

impl<T> GetBackMut for Vec<T> {
    type Item = T;
    fn get_back_mut(&mut self, index: usize) -> Option<&mut Self::Item> {
        self.len()
            .checked_sub(index)
            .and_then(|index| self.get_mut(index))
    }
}

/// change these if you want to represent bases larger than 256
pub type Digit = u8;
/// this must be twice the size of Digit (for overflow prevention)
pub type DoubleDigit = u16;

/// `BigInt`: represents an arbitrary-size integer in base `BASE`.
///
/// `BASE` may be anywhere from 2-256.
/// If you would like to be able to represent a larger base than 65536, then increase `Digit`
/// and `DoubleDigit` as needed, as high as `u64` + `u128`.
#[derive(Clone, Debug, PartialEq, Eq)]
pub struct BigInt<const BASE: usize>(bool, Vec<Digit>);

impl<const BASE: usize> BigInt<BASE> {
    /// Create a new `BigInt` directly from a `Vec` of individual digits.
    ///
    /// Ensure the resulting int is properly normalized, and that no digits are greater than or
    /// equal to the base, to preserve soundness.
    ///
    /// To construct a negative `BigInt` from raw parts, simply apply the negation
    /// operator (`-`) afterwards.
    ///
    /// ```
    /// use big_int::*;
    ///
    /// assert_eq!(
    ///     unsafe { -BigInt::<10>::from_raw_parts(vec![1, 5]) },
    ///     (-15).into()
    /// );
    /// ```
    pub unsafe fn from_raw_parts(digits: Vec<Digit>) -> Self {
        BigInt(false, digits)
    }

    /// The constant zero represented as a `BigInt`.
    pub fn zero() -> Self {
        BigInt(false, vec![0])
    }

    /// Convert a `BigInt` to a printable string using the provided alphabet `alphabet`.
    /// `Display` uses this method with the default alphabet `STANDARD_ALPHABET`.
    ///
    /// ```
    /// use big_int::*;
    ///
    /// assert_eq!(
    ///     BigInt::<10>::from(6012).display(STANDARD_ALPHABET).unwrap(),
    ///     "6012".to_string()
    /// );
    /// ```
    pub fn display(&self, alphabet: &str) -> Result<String, BigIntError> {
        let digits = self
            .1
            .iter()
            .map(|digit| {
                alphabet
                    .chars()
                    .nth(*digit as usize)
                    .ok_or(BigIntError::BaseTooHigh(BASE, alphabet.len()))
            })
            .collect::<Result<String, _>>()?;
        if self.0 {
            Ok(format!("-{digits}"))
        } else {
            Ok(digits)
        }
    }

    /// Return a normalized version of the `BigInt`. Remove trailing zeros, and disable the parity flag
    /// if the resulting number is zero.
    ///
    /// ```
    /// use big_int::*;
    ///
    /// let n = unsafe { BigInt::<10>::from_raw_parts(vec![0, 0, 8, 3]) };
    /// assert_eq!(n.normalized(), 83.into());
    /// ```
    pub fn normalized(self) -> Self {
        match self.1.iter().position(|digit| *digit != 0) {
            None => BigInt::zero(),
            Some(pos @ 1..) => BigInt(self.0, self.1[pos..].to_vec()),
            _ => self,
        }
    }

    /// Normalize a `BigInt` in place. Remove trailing zeros, and disable the parity flag
    /// if the resulting number is zero.
    ///
    /// ```
    /// use big_int::*;
    ///
    /// let mut n = unsafe { BigInt::<10>::from_raw_parts(vec![0, 0, 8, 3]) };
    /// n.normalize();
    /// assert_eq!(n, 83.into());
    /// ```
    pub fn normalize(&mut self) {
        match self.1.iter().position(|digit| *digit != 0) {
            None => *self = BigInt(false, vec![0]),
            Some(pos @ 1..) => self.1 = self.1[pos..].to_vec(),
            _ => {}
        }
    }

    /// Parse a `BigInt` from a `value: &str`, referencing the provided `alphabet`
    /// to determine what characters represent which digits. `FromStr` uses this method
    /// with the default alphabet `STANDARD_ALPHABET`.
    ///
    /// ```
    /// use big_int::*;
    ///
    /// assert_eq!(BigInt::parse("125", STANDARD_ALPHABET), Ok(BigInt::<10>::from(125)));
    /// ```
    pub fn parse(value: &str, alphabet: &str) -> Result<Self, ParseError> {
        let mut digits = VecDeque::new();
        let (sign, chars) = match value.chars().next() {
            Some('-') => (true, value.chars().skip(1)),
            Some(_) => (false, value.chars().skip(0)),
            None => return Err(ParseError::NotEnoughCharacters),
        };
        for char in chars {
            match alphabet.chars().position(|c| c == char) {
                Some(pos) => {
                    if pos >= BASE {
                        return Err(ParseError::DigitTooLarge(char, pos, BASE));
                    } else {
                        digits.push_back(pos as Digit);
                    }
                }
                None => return Err(ParseError::UnrecognizedCharacter(char)),
            }
        }
        if digits.is_empty() {
            Err(ParseError::NotEnoughCharacters)
        } else {
            Ok(BigInt(sign, digits.into()).normalized())
        }
    }

    /// Divide one `BigInt` by another, returning the quotient & remainder as a pair,
    /// or an error if dividing by zero.
    ///
    /// `b` - base\
    /// `d` - number of digits in quotient\
    /// Time complexity: `O(d * b)`\
    /// Memory complexity: `O(d)`\
    ///
    /// ```
    /// use big_int::*;
    ///
    /// let a: BigInt<10> = 999_999_999.into();
    /// let b = 56_789.into();
    /// assert_eq!(a.div_rem_lowmem(b), Ok((17_609.into(), 2_498.into())));
    /// ```
    pub fn div_rem_lowmem(mut self, mut other: Self) -> Result<(Self, Self), BigIntError> {
        if other.clone().normalized() == BigInt::zero() {
            return Err(BigIntError::DivisionByZero);
        }
        if other.1.len() > self.1.len() {
            return Ok((BigInt(false, vec![0]), self));
        }
        let sign = self.0 != other.0;
        self.0 = false;
        other.0 = false;
        let quot_digits = self.1.len() - other.1.len() + 1;
        let mut quot = unsafe { BigInt::from_raw_parts(vec![0; quot_digits]) };
        let mut addend =
            unsafe { BigInt::from_raw_parts([other.1, vec![0; quot_digits - 1]].concat()) };
        let mut prod = BigInt::zero();

        for digit in 0..quot.1.len() {
            for digit_value in 0..BASE {
                let new_prod = prod.clone() + addend.clone();
                if new_prod > self {
                    quot.1[digit] = digit_value as Digit;
                    break;
                } else {
                    prod = new_prod;
                }
            }
            addend.1.pop();
        }

        quot.0 = sign;
        let mut rem = self - prod;
        if rem != BigInt::zero() {
            rem.0 = sign;
        }

        Ok((quot.normalized(), rem))
    }

    /// Divide one `BigInt` by another, returning the quotient & remainder as a pair,
    /// or an error if dividing by zero. This algorithm has a different time complexity
    /// than `BigInt::div_rem_lowmem` which makes it faster for most use cases, but also uses more memory.
    ///
    /// `b` - base\
    /// `d` - number of digits in quotient\
    /// Time complexity: `O(d * log(b))`\
    /// Memory complexity: `O(d^2)`\
    ///
    /// ```
    /// use big_int::*;
    ///
    /// let a: BigInt<10> = 999_999_999.into();
    /// let b = 56_789.into();
    /// assert_eq!(a.div_rem(b), Ok((17_609.into(), 2_498.into())));
    /// ```
    pub fn div_rem(mut self, mut other: Self) -> Result<(Self, Self), BigIntError> {
        if other.clone().normalized() == BigInt::zero() {
            return Err(BigIntError::DivisionByZero);
        }
        if other.1.len() > self.1.len() {
            return Ok((BigInt(false, vec![0]), self));
        }
        let sign = self.0 != other.0;
        self.0 = false;
        other.0 = false;
        let quot_digits = self.1.len() - other.1.len() + 1;
        let mut quot = unsafe { BigInt::from_raw_parts(vec![0; quot_digits]) };
        let mut prod = BigInt::zero();
        let mut addend: BigInt<BASE> =
            unsafe { BigInt::from_raw_parts([other.1, vec![0; quot_digits - 1]].concat()) };
        let mut addends = Vec::new();
        let mut power = 1;
        while power < BASE {
            addends.push(addend.clone());
            addend += addend.clone();
            power <<= 1;
        }

        for digit in 0..quot.1.len() {
            let mut digit_value = 0;
            for power in (0..addends.len()).rev() {
                let new_prod = prod.clone() + addends[power].clone();
                if new_prod <= self {
                    digit_value += 1 << power;
                    prod = new_prod;
                }
                addends[power].1.pop();
            }
            quot.1[digit] = digit_value;
        }

        quot.0 = sign;
        let mut rem = self - prod;
        if rem != BigInt::zero() {
            rem.0 = sign;
        }

        Ok((quot.normalized(), rem))
    }

    /// Convert a `BigInt` from its own base to another target base using the provided division function.
    /// You should prefer to use one of either `BigInt::convert` or `BigInt::convert_lowmem` instead of this.
    fn convert_with<const TO: usize>(
        mut self,
        div_fn: impl Fn(BigInt<BASE>, BigInt<BASE>) -> Result<(BigInt<BASE>, BigInt<BASE>), BigIntError>,
    ) -> BigInt<TO> {
        let sign = self.0;
        self.0 = false;
        let mut digits = VecDeque::new();
        let to_base = BigInt::<BASE>::from(TO);
        while self >= to_base {
            let (quot, rem) = div_fn(self, to_base.clone()).unwrap();
            self = quot;
            digits.push_front(Digit::from(rem));
        }
        digits.push_front(Digit::from(self));
        BigInt::<TO>(sign, digits.into()).normalized()
    }

    /// Convert a `BigInt` from its own base to another target base.
    ///
    /// ```
    /// use big_int::*;
    ///
    /// assert_eq!(
    ///     BigInt::<10>::from(99825).convert(),
    ///     BigInt::<16>::from(99825)
    /// );
    /// ```
    pub fn convert<const TO: usize>(self) -> BigInt<TO> {
        self.convert_with(BigInt::div_rem)
    }

    /// Convert a `BigInt` from its own base to another target base using `BigInt::div_rem_lowmem`.
    /// Has lower memory usage, but greater time complexity.
    ///
    /// ```
    /// use big_int::*;
    ///
    /// assert_eq!(
    ///     BigInt::<10>::from(99825).convert_lowmem(),
    ///     BigInt::<16>::from(99825)
    /// );
    /// ```
    pub fn convert_lowmem<const TO: usize>(self) -> BigInt<TO> {
        self.convert_with(BigInt::div_rem_lowmem)
    }
}

impl<const BASE: usize> Default for BigInt<BASE> {
    fn default() -> Self {
        BigInt::zero()
    }
}

impl<const BASE: usize> FromStr for BigInt<BASE> {
    type Err = BigIntError;

    fn from_str(s: &str) -> Result<Self, Self::Err> {
        Self::parse(s, STANDARD_ALPHABET).map_err(BigIntError::ParseFailed)
    }
}

impl<const BASE: usize> From<u128> for BigInt<BASE> {
    fn from(mut value: u128) -> Self {
        let base = BASE as u128;
        let mut result = VecDeque::new();
        while value >= base {
            let (new_value, rem) = (value / base, value % base);
            value = new_value;
            result.push_front(rem as Digit);
        }
        result.push_front(value as Digit);
        BigInt(false, result.into()).normalized()
    }
}

impl<const BASE: usize> From<i128> for BigInt<BASE> {
    fn from(value: i128) -> Self {
        if value < 0 {
            let mut bigint = BigInt::<BASE>::from((-value) as u128);
            bigint.0 = true;
            bigint
        } else {
            BigInt::<BASE>::from(value as u128)
        }
    }
}

macro_rules! bigint_from_int {
    ($b:ident; $($i:ident),*) => {
        $(
            impl<const BASE: usize> From<$i> for BigInt<BASE> {
                fn from(value: $i) -> Self {
                    (value as $b).into()
                }
            }
        )*
    };
}

bigint_from_int!(i128; i8, i16, i32, i64, isize);
bigint_from_int!(u128; u8, u16, u32, u64, usize);

macro_rules! int_from_bigint {
    ($(($i:ident, $u:ident)),*) => {
        $(
            impl<const BASE: usize> From<BigInt<BASE>> for $i {
                fn from(value: BigInt<BASE>) -> Self {
                    let mut digits = value.1;
                    let mut total: $i = 0;
                    let mut place: $i = 1;
                    while let Some(digit) = digits.pop() {
                        total += (digit as $i) * place;
                        place *= BASE as $i;
                    }
                    if value.0 {
                        total = -total;
                    }
                    total
                }
            }

            impl<const BASE: usize> From<BigInt<BASE>> for $u {
                fn from(value: BigInt<BASE>) -> Self {
                    $i::from(value) as $u
                }
            }
        )*
    };
}

int_from_bigint!(
    (i128, u128),
    (i64, u64),
    (i32, u32),
    (i16, u16),
    (i8, u8),
    (isize, usize)
);

impl<const BASE: usize> Display for BigInt<BASE> {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        write!(
            f,
            "{}",
            self.display(STANDARD_ALPHABET)
                .map_err(|_| std::fmt::Error)?
        )
    }
}

impl<const BASE: usize> Neg for BigInt<BASE> {
    type Output = Self;

    fn neg(self) -> Self::Output {
        BigInt(!self.0, self.1)
    }
}

impl<const BASE: usize> Add for BigInt<BASE> {
    type Output = Self;

    fn add(self, rhs: Self) -> Self::Output {
        if self.0 != rhs.0 {
            self - (-rhs)
        } else {
            let mut carry = 0;
            let mut result = VecDeque::with_capacity(self.1.len().max(rhs.1.len()) + 1);
            for i in 1.. {
                match (self.1.get_back(i), rhs.1.get_back(i), carry) {
                    (None, None, 0) => break,
                    (left_digit, right_digit, carry_in) => {
                        let left_digit = left_digit.copied().unwrap_or_default() as DoubleDigit;
                        let right_digit = right_digit.copied().unwrap_or_default() as DoubleDigit;
                        let mut sum = left_digit + right_digit + carry_in;
                        if sum >= BASE as DoubleDigit {
                            sum -= BASE as DoubleDigit;
                            carry = 1;
                        } else {
                            carry = 0;
                        }
                        result.push_front(sum as Digit);
                    }
                }
            }
            BigInt(self.0, result.into()).normalized()
        }
    }
}

impl<const BASE: usize> AddAssign for BigInt<BASE> {
    fn add_assign(&mut self, rhs: Self) {
        if self.0 != rhs.0 {
            *self -= -rhs;
        } else {
            let self_len = self.1.len();
            let mut carry = 0;
            for i in 1.. {
                match (self.1.get_back(i), rhs.1.get_back(i), carry) {
                    (None, None, 0) => break,
                    (left_digit, right_digit, carry_in) => {
                        let left_digit = left_digit.copied().unwrap_or_default() as DoubleDigit;
                        let right_digit = right_digit.copied().unwrap_or_default() as DoubleDigit;
                        let mut sum = left_digit + right_digit + carry_in;
                        if sum >= BASE as DoubleDigit {
                            sum -= BASE as DoubleDigit;
                            carry = 1;
                        } else {
                            carry = 0;
                        }
                        if i <= self_len {
                            self.1[self_len - i] = sum as Digit;
                        } else {
                            self.1.insert(0, sum as Digit);
                        }
                    }
                }
            }
        }
    }
}

impl<const BASE: usize> Sub for BigInt<BASE> {
    type Output = Self;

    fn sub(mut self, rhs: Self) -> Self::Output {
        if self.0 != rhs.0 {
            self + (-rhs)
        } else if rhs > self {
            -(rhs - self)
        } else {
            let self_len = self.1.len();
            let mut result = VecDeque::with_capacity(self_len.max(rhs.1.len()) + 1);
            for i in 1.. {
                match (self.1.get_back(i), rhs.1.get_back(i)) {
                    (None, None) => break,
                    (left_digit, right_digit) => {
                        let mut left_digit = left_digit.copied().unwrap_or_default() as DoubleDigit;
                        let right_digit = right_digit.copied().unwrap_or_default() as DoubleDigit;
                        if left_digit < right_digit {
                            for j in i + 1.. {
                                match self.1.get_back_mut(j) {
                                    None => unreachable!("`BigInt` subtraction with overflow"),
                                    Some(digit @ 0) => *digit = (BASE - 1) as Digit,
                                    Some(digit) => {
                                        *digit -= 1;
                                        break;
                                    }
                                }
                            }
                            left_digit += BASE as DoubleDigit;
                        }
                        let difference = left_digit - right_digit;
                        result.push_front(difference as Digit);
                    }
                }
            }
            BigInt(self.0, result.into()).normalized()
        }
    }
}

impl<const BASE: usize> SubAssign for BigInt<BASE> {
    fn sub_assign(&mut self, mut rhs: Self) {
        if self.0 != rhs.0 {
            *self += -rhs;
        } else if rhs > *self {
            rhs -= self.clone();
            *self = -rhs;
        } else {
            let self_len = self.1.len();
            for i in 1.. {
                match (self.1.get_back(i), rhs.1.get_back(i)) {
                    (None, None) => break,
                    (left_digit, right_digit) => {
                        let mut left_digit = left_digit.copied().unwrap_or_default() as DoubleDigit;
                        let right_digit = right_digit.copied().unwrap_or_default() as DoubleDigit;
                        if left_digit < right_digit {
                            for j in i + 1.. {
                                match self.1.get_back_mut(j) {
                                    None => unreachable!("subtraction with overflow"),
                                    Some(digit @ 0) => *digit = (BASE - 1) as Digit,
                                    Some(digit) => {
                                        *digit -= 1;
                                        break;
                                    }
                                }
                            }
                            left_digit += BASE as DoubleDigit;
                        }
                        self.1[self_len - i] = (left_digit - right_digit) as Digit;
                    }
                }
            }
        }
        self.normalize();
    }
}

impl<const BASE: usize> Mul for BigInt<BASE> {
    type Output = Self;

    fn mul(mut self, mut rhs: Self) -> Self::Output {
        let sign = self.0 != rhs.0;
        self.0 = false;
        rhs.0 = false;
        let mut result = BigInt::from(0);
        for i in 1.. {
            if let Some(&digit) = self.1.get_back(i) {
                for _ in 0..digit {
                    result += rhs.clone();
                }
                rhs.1.push(0);
            } else {
                break;
            }
        }
        result.0 = sign;
        result.normalized()
    }
}

impl<const BASE: usize> MulAssign for BigInt<BASE> {
    fn mul_assign(&mut self, rhs: Self) {
        *self = self.clone() * rhs;
    }
}

impl<const BASE: usize> Div for BigInt<BASE> {
    type Output = Self;

    fn div(self, rhs: Self) -> Self::Output {
        self.div_rem(rhs).unwrap().0
    }
}

impl<const BASE: usize> DivAssign for BigInt<BASE> {
    fn div_assign(&mut self, rhs: Self) {
        *self = self.clone() / rhs;
    }
}

impl<const BASE: usize> Shl for BigInt<BASE> {
    type Output = Self;

    /// Shifts a `BigInt` left by multiples of its `BASE` (not by 2).
    fn shl(self, rhs: Self) -> Self::Output {
        BigInt(self.0, [self.1, vec![0; rhs.into()]].concat())
    }
}

impl<const BASE: usize> ShlAssign for BigInt<BASE> {
    /// Shifts a `BigInt` left by multiples of its `BASE` (not by 2).
    fn shl_assign(&mut self, rhs: Self) {
        *self = self.clone() << rhs;
    }
}

impl<const BASE: usize> Shr for BigInt<BASE> {
    type Output = Self;

    /// Shifts a `BigInt` right by multiples of its `BASE` (not by 2).
    fn shr(self, rhs: Self) -> Self::Output {
        BigInt(self.0, self.1[..self.1.len() - usize::from(rhs)].to_vec()).normalized()
    }
}

impl<const BASE: usize> ShrAssign for BigInt<BASE> {
    /// Shifts a `BigInt` right by multiples of its `BASE` (not by 2).
    fn shr_assign(&mut self, rhs: Self) {
        *self = self.clone() >> rhs;
        self.normalize()
    }
}

impl<const BASE: usize> Ord for BigInt<BASE> {
    fn cmp(&self, other: &Self) -> Ordering {
        match (self.0, other.0) {
            (false, true) => Ordering::Greater,
            (true, false) => Ordering::Less,
            _ => cmp(&self.1, &other.1),
        }
    }
}

impl<const BASE: usize> PartialOrd for BigInt<BASE> {
    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
        Some(self.cmp(other))
    }
}

/// Helper function called recursively when comparing two `BigInt`s.
fn cmp(a: &[Digit], b: &[Digit]) -> Ordering {
    if a.len() > b.len() {
        Ordering::Greater
    } else if a.len() < b.len() {
        Ordering::Less
    } else {
        match (a.split_first(), b.split_first()) {
            (None, None) => Ordering::Equal,
            (None, Some(_)) => Ordering::Less,
            (Some(_), None) => Ordering::Greater,
            (Some((a_digit, rest_a)), Some((b_digit, rest_b))) => match a_digit.cmp(b_digit) {
                Ordering::Equal => cmp(rest_a, rest_b),
                ordering => ordering,
            },
        }
    }
}

/// Encode an array of bytes into base64 data.
///
/// Note: probably slower than using a standalone
/// library to perform this conversion. However, it's very neat :3
/// 
/// Note: may fail if the data begins with zeros.
///
/// ```
/// use big_int::*;
/// assert_eq!(base64_encode(b"Hello world!"), "SGVsbG8gd29ybGQh");
/// ```
pub fn base64_encode(bytes: &[u8]) -> String {
    let mut digits = bytes
        .into_iter()
        .copied()
        .map(Digit::from)
        .collect::<Vec<_>>();
    let padding = 3 - ((digits.len() - 1) % 3) - 1;
    digits.extend(vec![0; padding]);
    let data_as_int: BigInt<256> = unsafe { BigInt::from_raw_parts(digits) };
    let base64_data: BigInt<64> = data_as_int.convert();
    let base64_string = base64_data.display(BASE64_ALPHABET).unwrap();
    base64_string[..base64_string.len() - padding].to_string()
}

/// Decode a base64 string into an array of bytes.
///
/// Note: probably slower than using a standalone
/// library to perform this conversion. However, again, it's very neat c:
///
/// Note: may fail if the data begins with zeros.
/// 
/// ```
/// use big_int::*;
/// assert_eq!(base64_decode("SGVsbG8gd29ybGQh").unwrap(), b"Hello world!");
/// ```
pub fn base64_decode(b64_string: impl Into<String>) -> Result<Vec<u8>, BigIntError> {
    let mut b64_string = b64_string.into();
    let padding = 4 - ((b64_string.len() - 1) % 4) - 1;
    b64_string.extend(vec!['A'; padding]);
    let string_as_int: BigInt<64> =
        BigInt::parse(&b64_string, BASE64_ALPHABET).map_err(BigIntError::ParseFailed)?;
    let bytes_int: BigInt<256> = string_as_int.convert();
    let bytes = bytes_int
        .1
        .into_iter()
        .map(u8::try_from)
        .collect::<Result<Vec<_>, _>>()
        .unwrap();
    let bytes = bytes[..bytes.len() - padding].to_vec();
    Ok(bytes)
}