File: array_.py

package info (click to toggle)
python-bitstring 4.3.1-1
  • links: PTS, VCS
  • area: main
  • in suites: forky, sid
  • size: 1,312 kB
  • sloc: python: 11,397; makefile: 8; sh: 7
file content (778 lines) | stat: -rw-r--r-- 36,113 bytes parent folder | download
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
from __future__ import annotations

import math
import numbers
from collections.abc import Sized
from bitstring.exceptions import CreationError
from typing import Union, List, Iterable, Any, Optional, BinaryIO, overload, TextIO
from bitstring.bits import Bits, BitsType
from bitstring.bitarray_ import BitArray
from bitstring.dtypes import Dtype, dtype_register
from bitstring import utils
from bitstring.bitstring_options import Options, Colour
import copy
import array
import operator
import io
import sys

# The possible types stored in each element of the Array
ElementType = Union[float, str, int, bytes, bool, Bits]

options = Options()


class Array:
    """Return an Array whose elements are initialised according to the fmt string.
    The dtype string can be typecode as used in the struct module or any fixed-length bitstring
    format.

    a = Array('>H', [1, 15, 105])
    b = Array('int5', [-9, 0, 4])

    The Array data is stored compactly as a BitArray object and the Array behaves very like
    a list of items of the given format. Both the Array data and fmt properties can be freely
    modified after creation. If the data length is not a multiple of the fmt length then the
    Array will have 'trailing_bits' which will prevent some methods from appending to the
    Array.

    Methods:

    append() -- Append a single item to the end of the Array.
    byteswap() -- Change byte endianness of all items.
    count() -- Count the number of occurences of a value.
    extend() -- Append new items to the end of the Array from an iterable.
    fromfile() -- Append items read from a file object.
    insert() -- Insert an item at a given position.
    pop() -- Remove and return an item.
    pp() -- Pretty print the Array.
    reverse() -- Reverse the order of all items.
    tobytes() -- Return Array data as bytes object, padding with zero bits at the end if needed.
    tofile() -- Write Array data to a file, padding with zero bits at the end if needed.
    tolist() -- Return Array items as a list.

    Special methods:

    Also available are the operators [], ==, !=, +, *, <<, >>, &, |, ^,
    plus the mutating operators [], +=, *=, <<=, >>=, &=, |=, ^=.

    Properties:

    data -- The BitArray binary data of the Array. Can be freely modified.
    dtype -- The format string or typecode. Can be freely modified.
    itemsize -- The length *in bits* of a single item. Read only.
    trailing_bits -- If the data length is not a multiple of the fmt length, this BitArray
                     gives the leftovers at the end of the data.


    """

    def __init__(self, dtype: Union[str, Dtype], initializer: Optional[Union[int, Array, array.array, Iterable, Bits, bytes, bytearray, memoryview, BinaryIO]] = None,
                 trailing_bits: Optional[BitsType] = None) -> None:
        self.data = BitArray()
        if isinstance(dtype, Dtype) and dtype.scale == 'auto':
            if isinstance(initializer, (int, Bits, bytes, bytearray, memoryview, BinaryIO)):
                raise TypeError("An Array with an 'auto' scale factor can only be created from an iterable of values.")
            auto_scale = self._calculate_auto_scale(initializer, dtype.name, dtype.length)
            dtype = Dtype(dtype.name, dtype.length, scale=auto_scale)
        try:
            self._set_dtype(dtype)
        except ValueError as e:
            raise CreationError(e)

        if isinstance(initializer, numbers.Integral):
            self.data = BitArray(initializer * self._dtype.bitlength)
        elif isinstance(initializer, (Bits, bytes, bytearray, memoryview)):
            self.data += initializer
        elif isinstance(initializer, io.BufferedReader):
            self.fromfile(initializer)
        elif initializer is not None:
            self.extend(initializer)

        if trailing_bits is not None:
            self.data += BitArray._create_from_bitstype(trailing_bits)

    _largest_values = None

    @staticmethod
    def _calculate_auto_scale(initializer, name: str, length: Optional[int]) -> float:
        # Now need to find the largest power of 2 representable with this format.
        if Array._largest_values is None:
            Array._largest_values = {
                'mxint8': Bits('0b01111111').mxint8,  # 1.0 + 63.0/64.0,
                'e2m1mxfp4': Bits('0b0111').e2m1mxfp4,  # 6.0
                'e2m3mxfp6': Bits('0b011111').e2m3mxfp6,  # 7.5
                'e3m2mxfp6': Bits('0b011111').e3m2mxfp6,  # 28.0
                'e4m3mxfp8': Bits('0b01111110').e4m3mxfp8,  # 448.0
                'e5m2mxfp8': Bits('0b01111011').e5m2mxfp8,  # 57344.0
                'p4binary8': Bits('0b01111110').p4binary8,  # 224.0
                'p3binary8': Bits('0b01111110').p3binary8,  # 49152.0
                'float16': Bits('0x7bff').float16,  # 65504.0
                # The bfloat range is so large the scaling algorithm doesn't work well, so I'm disallowing it.
                # 'bfloat16': Bits('0x7f7f').bfloat16,  # 3.38953139e38,
            }
        if f'{name}{length}' in Array._largest_values.keys():
            float_values = Array('float64', initializer).tolist()
            if not float_values:
                raise ValueError("Can't calculate an 'auto' scale with an empty Array initializer.")
            max_float_value = max(abs(x) for x in float_values)
            if max_float_value == 0:
                # This special case isn't covered in the standard. I'm choosing to return no scale.
                return 1.0
            # We need to find the largest power of 2 that is less than the max value
            log2 = math.floor(math.log2(max_float_value))
            lp2 = math.floor(math.log2(Array._largest_values[f'{name}{length}']))
            lg_scale = log2 - lp2
            # Saturate at values representable in E8M0 format.
            if lg_scale > 127:
                lg_scale = 127
            elif lg_scale < -127:
                lg_scale = -127
            return 2 ** lg_scale
        else:
            raise ValueError(f"Can't calculate auto scale for format '{name}{length}'. "
                             f"This feature is only available for these formats: {list(Array._largest_values.keys())}.")

    @property
    def itemsize(self) -> int:
        return self._dtype.length

    @property
    def trailing_bits(self) -> BitArray:
        trailing_bit_length = len(self.data) % self._dtype.bitlength
        return BitArray() if trailing_bit_length == 0 else self.data[-trailing_bit_length:]

    @property
    def dtype(self) -> Dtype:
        return self._dtype

    @dtype.setter
    def dtype(self, new_dtype: Union[str, Dtype]) -> None:
        self._set_dtype(new_dtype)

    def _set_dtype(self, new_dtype: Union[str, Dtype]) -> None:
        if isinstance(new_dtype, Dtype):
            self._dtype = new_dtype
        else:
            try:
                dtype = Dtype(new_dtype)
            except ValueError:
                name_length = utils.parse_single_struct_token(new_dtype)
                if name_length is not None:
                    dtype = Dtype(name_length[0], name_length[1])
                else:
                    raise ValueError(f"Inappropriate Dtype for Array: '{new_dtype}'.")
            if dtype.length is None:
                raise ValueError(f"A fixed length format is needed for an Array, received '{new_dtype}'.")
            self._dtype = dtype
        if self._dtype.scale == 'auto':
            raise ValueError("A Dtype with an 'auto' scale factor can only be used when creating a new Array.")

    def _create_element(self, value: ElementType) -> Bits:
        """Create Bits from value according to the token_name and token_length"""
        b = self._dtype.build(value)
        if len(b) != self._dtype.length:
            raise ValueError(f"The value {value!r} has the wrong length for the format '{self._dtype}'.")
        return b

    def __len__(self) -> int:
        return len(self.data) // self._dtype.length

    @overload
    def __getitem__(self, key: slice) -> Array:
        ...

    @overload
    def __getitem__(self, key: int) -> ElementType:
        ...

    def __getitem__(self, key: Union[slice, int]) -> Union[Array, ElementType]:
        if isinstance(key, slice):
            start, stop, step = key.indices(len(self))
            if step != 1:
                d = BitArray()
                for s in range(start * self._dtype.length, stop * self._dtype.length, step * self._dtype.length):
                    d.append(self.data[s: s + self._dtype.length])
                a = self.__class__(self._dtype)
                a.data = d
                return a
            else:
                a = self.__class__(self._dtype)
                a.data = self.data[start * self._dtype.length: stop * self._dtype.length]
                return a
        else:
            if key < 0:
                key += len(self)
            if key < 0 or key >= len(self):
                raise IndexError(f"Index {key} out of range for Array of length {len(self)}.")
            return self._dtype.read_fn(self.data, start=self._dtype.length * key)

    @overload
    def __setitem__(self, key: slice, value: Iterable[ElementType]) -> None:
        ...

    @overload
    def __setitem__(self, key: int, value: ElementType) -> None:
        ...

    def __setitem__(self, key: Union[slice, int], value: Union[Iterable[ElementType], ElementType]) -> None:
        if isinstance(key, slice):
            start, stop, step = key.indices(len(self))
            if not isinstance(value, Iterable):
                raise TypeError("Can only assign an iterable to a slice.")
            if step == 1:
                new_data = BitArray()
                for x in value:
                    new_data += self._create_element(x)
                self.data[start * self._dtype.length: stop * self._dtype.length] = new_data
                return
            items_in_slice = len(range(start, stop, step))
            if not isinstance(value, Sized):
                value = list(value)
            if len(value) == items_in_slice:
                for s, v in zip(range(start, stop, step), value):
                    self.data.overwrite(self._create_element(v), s * self._dtype.length)
            else:
                raise ValueError(f"Can't assign {len(value)} values to an extended slice of length {items_in_slice}.")
        else:
            if key < 0:
                key += len(self)
            if key < 0 or key >= len(self):
                raise IndexError(f"Index {key} out of range for Array of length {len(self)}.")
            start = self._dtype.length * key
            self.data.overwrite(self._create_element(value), start)
            return

    def __delitem__(self, key: Union[slice, int]) -> None:
        if isinstance(key, slice):
            start, stop, step = key.indices(len(self))
            if step == 1:
                self.data.__delitem__(slice(start * self._dtype.length, stop * self._dtype.length))
                return
            # We need to delete from the end or the earlier positions will change
            r = reversed(range(start, stop, step)) if step > 0 else range(start, stop, step)
            for s in r:
                self.data.__delitem__(slice(s * self._dtype.length, (s + 1) * self._dtype.length))
        else:
            if key < 0:
                key += len(self)
            if key < 0 or key >= len(self):
                raise IndexError
            start = self._dtype.length * key
            del self.data[start: start + self._dtype.length]

    def __repr__(self) -> str:
        list_str = f"{self.tolist()}"
        trailing_bit_length = len(self.data) % self._dtype.length
        final_str = "" if trailing_bit_length == 0 else ", trailing_bits=" + repr(
            self.data[-trailing_bit_length:])
        return f"Array('{self._dtype}', {list_str}{final_str})"

    def astype(self, dtype: Union[str, Dtype]) -> Array:
        """Return Array with elements of new dtype, initialised from current Array."""
        new_array = self.__class__(dtype, self.tolist())
        return new_array

    def tolist(self) -> List[ElementType]:
        return [self._dtype.read_fn(self.data, start=start)
                for start in range(0, len(self.data) - self._dtype.length + 1, self._dtype.length)]

    def append(self, x: ElementType) -> None:
        if len(self.data) % self._dtype.length != 0:
            raise ValueError("Cannot append to Array as its length is not a multiple of the format length.")
        self.data += self._create_element(x)

    def extend(self, iterable: Union[Array, array.array, Iterable[Any]]) -> None:
        if len(self.data) % self._dtype.length != 0:
            raise ValueError(f"Cannot extend Array as its data length ({len(self.data)} bits) is not a multiple of the format length ({self._dtype.length} bits).")
        if isinstance(iterable, Array):
            if self._dtype.name != iterable._dtype.name or self._dtype.length != iterable._dtype.length:
                raise TypeError(
                    f"Cannot extend an Array with format '{self._dtype}' from an Array of format '{iterable._dtype}'.")
            # No need to iterate over the elements, we can just append the data
            self.data.append(iterable.data)
        elif isinstance(iterable, array.array):
            # array.array types are always native-endian, hence the '='
            name_value = utils.parse_single_struct_token('=' + iterable.typecode)
            if name_value is None:
                raise ValueError(f"Cannot extend from array with typecode {iterable.typecode}.")
            other_dtype = dtype_register.get_dtype(*name_value, scale=None)
            if self._dtype.name != other_dtype.name or self._dtype.length != other_dtype.length:
                raise ValueError(
                    f"Cannot extend an Array with format '{self._dtype}' from an array with typecode '{iterable.typecode}'.")
            self.data += iterable.tobytes()
        else:
            if isinstance(iterable, str):
                raise TypeError("Can't extend an Array with a str.")
            for item in iterable:
                self.data += self._create_element(item)

    def insert(self, i: int, x: ElementType) -> None:
        """Insert a new element into the Array at position i.

        """
        i = min(i, len(self))  # Inserting beyond len of array inserts at the end (copying standard behaviour)
        self.data.insert(self._create_element(x), i * self._dtype.length)

    def pop(self, i: int = -1) -> ElementType:
        """Return and remove an element of the Array.

        Default is to return and remove the final element.

        """
        if len(self) == 0:
            raise IndexError("Can't pop from an empty Array.")
        x = self[i]
        del self[i]
        return x

    def byteswap(self) -> None:
        """Change the endianness in-place of all items in the Array.

        If the Array format is not a whole number of bytes a ValueError will be raised.

        """
        if self._dtype.length % 8 != 0:
            raise ValueError(
                f"byteswap can only be used for whole-byte elements. The '{self._dtype}' format is {self._dtype.length} bits long.")
        self.data.byteswap(self.itemsize // 8)

    def count(self, value: ElementType) -> int:
        """Return count of Array items that equal value.

        value -- The quantity to compare each Array element to. Type should be appropriate for the Array format.

        For floating point types using a value of float('nan') will count the number of elements that are NaN.

        """
        if math.isnan(value):
            return sum(math.isnan(i) for i in self)
        else:
            return sum(i == value for i in self)

    def tobytes(self) -> bytes:
        """Return the Array data as a bytes object, padding with zero bits if needed.

        Up to seven zero bits will be added at the end to byte align.

        """
        return self.data.tobytes()

    def tofile(self, f: BinaryIO) -> None:
        """Write the Array data to a file object, padding with zero bits if needed.

        Up to seven zero bits will be added at the end to byte align.

        """
        self.data.tofile(f)

    def fromfile(self, f: BinaryIO, n: Optional[int] = None) -> None:
        trailing_bit_length = len(self.data) % self._dtype.bitlength
        if trailing_bit_length != 0:
            raise ValueError(f"Cannot extend Array as its data length ({len(self.data)} bits) is not a multiple of the format length ({self._dtype.bitlength} bits).")

        new_data = Bits(f)
        max_items = len(new_data) // self._dtype.length
        items_to_append = max_items if n is None else min(n, max_items)
        self.data += new_data[0: items_to_append * self._dtype.bitlength]
        if n is not None and items_to_append < n:
            raise EOFError(f"Only {items_to_append} were appended, not the {n} items requested.")

    def reverse(self) -> None:
        trailing_bit_length = len(self.data) % self._dtype.length
        if trailing_bit_length != 0:
            raise ValueError(f"Cannot reverse the items in the Array as its data length ({len(self.data)} bits) is not a multiple of the format length ({self._dtype.length} bits).")
        for start_bit in range(0, len(self.data) // 2, self._dtype.length):
            start_swap_bit = len(self.data) - start_bit - self._dtype.length
            temp = self.data[start_bit: start_bit + self._dtype.length]
            self.data[start_bit: start_bit + self._dtype.length] = self.data[
                                                               start_swap_bit: start_swap_bit + self._dtype.length]
            self.data[start_swap_bit: start_swap_bit + self._dtype.length] = temp

    def pp(self, fmt: Optional[str] = None, width: int = 120,
           show_offset: bool = True, stream: TextIO = sys.stdout) -> None:
        """Pretty-print the Array contents.

        fmt -- Data format string. Defaults to current Array dtype.
        width -- Max width of printed lines in characters. Defaults to 120. A single group will always
                 be printed per line even if it exceeds the max width.
        show_offset -- If True shows the element offset in the first column of each line.
        stream -- A TextIO object with a write() method. Defaults to sys.stdout.

        """
        colour = Colour(not options.no_color)
        sep = ' '
        dtype2 = None
        tidy_fmt = None
        if fmt is None:
            fmt = self.dtype
            dtype1 = self.dtype
            tidy_fmt = "dtype='" + colour.purple + str(self.dtype) + "'" + colour.off
        else:
            token_list = utils.preprocess_tokens(fmt)
            if len(token_list) not in [1, 2]:
                raise ValueError(f"Only one or two tokens can be used in an Array.pp() format - '{fmt}' has {len(token_list)} tokens.")
            name1, length1 = utils.parse_name_length_token(token_list[0])
            dtype1 = Dtype(name1, length1)
            if len(token_list) == 2:
                name2, length2 = utils.parse_name_length_token(token_list[1])
                dtype2 = Dtype(name2, length2)

        token_length = dtype1.bitlength
        if dtype2 is not None:
            # For two types we're OK as long as they don't have different lengths given.
            if dtype1.bitlength is not None and dtype2.bitlength is not None and dtype1.bitlength != dtype2.bitlength:
                raise ValueError(f"Two different format lengths specified ('{fmt}'). Either specify just one, or two the same length.")
            if token_length is None:
                token_length = dtype2.bitlength
        if token_length is None:
            token_length = self.itemsize

        trailing_bit_length = len(self.data) % token_length
        format_sep = " : "  # String to insert on each line between multiple formats
        if tidy_fmt is None:
            tidy_fmt = colour.purple + str(dtype1) + colour.off
            if dtype2 is not None:
                tidy_fmt += ', ' + colour.blue + str(dtype2) + colour.off
            tidy_fmt = "fmt='" + tidy_fmt + "'"
        data = self.data if trailing_bit_length == 0 else self.data[0: -trailing_bit_length]
        length = len(self.data) // token_length
        len_str = colour.green + str(length) + colour.off
        stream.write(f"<{self.__class__.__name__} {tidy_fmt}, length={len_str}, itemsize={token_length} bits, total data size={(len(self.data) + 7) // 8} bytes> [\n")
        data._pp(dtype1, dtype2, token_length, width, sep, format_sep, show_offset, stream, False, token_length)
        stream.write("]")
        if trailing_bit_length != 0:
            stream.write(" + trailing_bits = " + str(self.data[-trailing_bit_length:]))
        stream.write("\n")

    def equals(self, other: Any) -> bool:
        """Return True if format and all Array items are equal."""
        if isinstance(other, Array):
            if self._dtype.length != other._dtype.length:
                return False
            if self._dtype.name != other._dtype.name:
                return False
            if self.data != other.data:
                return False
            return True
        elif isinstance(other, array.array):
            # Assume we are comparing with an array type
            if self.trailing_bits:
                return False
            # array's itemsize is in bytes, not bits.
            if self.itemsize != other.itemsize * 8:
                return False
            if len(self) != len(other):
                return False
            if self.tolist() != other.tolist():
                return False
            return True
        return False

    def __iter__(self) -> Iterable[ElementType]:
        start = 0
        for _ in range(len(self)):
            yield self._dtype.read_fn(self.data, start=start)
            start += self._dtype.length

    def __copy__(self) -> Array:
        a_copy = self.__class__(self._dtype)
        a_copy.data = copy.copy(self.data)
        return a_copy

    def _apply_op_to_all_elements(self, op, value: Union[int, float, None], is_comparison: bool = False) -> Array:
        """Apply op with value to each element of the Array and return a new Array"""
        new_array = self.__class__('bool' if is_comparison else self._dtype)
        new_data = BitArray()
        failures = index = 0
        msg = ''
        if value is not None:
            def partial_op(a):
                return op(a, value)
        else:
            def partial_op(a):
                return op(a)
        for i in range(len(self)):
            v = self._dtype.read_fn(self.data, start=self._dtype.length * i)
            try:
                new_data.append(new_array._create_element(partial_op(v)))
            except (CreationError, ZeroDivisionError, ValueError) as e:
                if failures == 0:
                    msg = str(e)
                    index = i
                failures += 1
        if failures != 0:
            raise ValueError(f"Applying operator '{op.__name__}' to Array caused {failures} errors. "
                             f'First error at index {index} was: "{msg}"')
        new_array.data = new_data
        return new_array

    def _apply_op_to_all_elements_inplace(self, op, value: Union[int, float]) -> Array:
        """Apply op with value to each element of the Array in place."""
        # This isn't really being done in-place, but it's simpler and faster for now?
        new_data = BitArray()
        failures = index = 0
        msg = ''
        for i in range(len(self)):
            v = self._dtype.read_fn(self.data, start=self._dtype.length * i)
            try:
                new_data.append(self._create_element(op(v, value)))
            except (CreationError, ZeroDivisionError, ValueError) as e:
                if failures == 0:
                    msg = str(e)
                    index = i
                failures += 1
        if failures != 0:
            raise ValueError(f"Applying operator '{op.__name__}' to Array caused {failures} errors. "
                             f'First error at index {index} was: "{msg}"')
        self.data = new_data
        return self

    def _apply_bitwise_op_to_all_elements(self, op, value: BitsType) -> Array:
        """Apply op with value to each element of the Array as an unsigned integer and return a new Array"""
        a_copy = self[:]
        a_copy._apply_bitwise_op_to_all_elements_inplace(op, value)
        return a_copy

    def _apply_bitwise_op_to_all_elements_inplace(self, op, value: BitsType) -> Array:
        """Apply op with value to each element of the Array as an unsigned integer in place."""
        value = BitArray._create_from_bitstype(value)
        if len(value) != self._dtype.length:
            raise ValueError(f"Bitwise op needs a bitstring of length {self._dtype.length} to match format {self._dtype}.")
        for start in range(0, len(self) * self._dtype.length, self._dtype.length):
            self.data[start: start + self._dtype.length] = op(self.data[start: start + self._dtype.length], value)
        return self

    def _apply_op_between_arrays(self, op, other: Array, is_comparison: bool = False) -> Array:
        if len(self) != len(other):
            msg = f"Cannot operate element-wise on Arrays with different lengths ({len(self)} and {len(other)})."
            if op in [operator.add, operator.iadd]:
                msg += " Use extend() method to concatenate Arrays."
            if op in [operator.eq, operator.ne]:
                msg += " Use equals() method to compare Arrays for a single boolean result."
            raise ValueError(msg)
        if is_comparison:
            new_type = dtype_register.get_dtype('bool', 1)
        else:
            new_type = self._promotetype(self._dtype, other._dtype)
        new_array = self.__class__(new_type)
        new_data = BitArray()
        failures = index = 0
        msg = ''
        for i in range(len(self)):
            a = self._dtype.read_fn(self.data, start=self._dtype.length * i)
            b = other._dtype.read_fn(other.data, start=other._dtype.length * i)
            try:
                new_data.append(new_array._create_element(op(a, b)))
            except (CreationError, ValueError, ZeroDivisionError) as e:
                if failures == 0:
                    msg = str(e)
                    index = i
                failures += 1
        if failures != 0:
            raise ValueError(f"Applying operator '{op.__name__}' between Arrays caused {failures} errors. "
                             f'First error at index {index} was: "{msg}"')
        new_array.data = new_data
        return new_array

    @classmethod
    def _promotetype(cls, type1: Dtype, type2: Dtype) -> Dtype:
        """When combining types which one wins?

        1. We only deal with types representing floats or integers.
        2. One of the two types gets returned. We never create a new one.
        3. Floating point types always win against integer types.
        4. Signed integer types always win against unsigned integer types.
        5. Longer types win against shorter types.
        6. In a tie the first type wins against the second type.

        """
        def is_float(x): return x.return_type is float
        def is_int(x): return x.return_type is int or x.return_type is bool
        if is_float(type1) + is_int(type1) + is_float(type2) + is_int(type2) != 2:
            raise ValueError(f"Only integer and floating point types can be combined - not '{type1}' and '{type2}'.")
        # If same type choose the widest
        if type1.name == type2.name:
            return type1 if type1.length > type2.length else type2
        # We choose floats above integers, irrespective of the widths
        if is_float(type1) and is_int(type2):
            return type1
        if is_int(type1) and is_float(type2):
            return type2
        if is_float(type1) and is_float(type2):
            return type2 if type2.length > type1.length else type1
        assert is_int(type1) and is_int(type2)
        if type1.is_signed and not type2.is_signed:
            return type1
        if type2.is_signed and not type1.is_signed:
            return type2
        return type2 if type2.length > type1.length else type1

    # Operators between Arrays or an Array and scalar value

    def __add__(self, other: Union[int, float, Array]) -> Array:
        """Add int or float to all elements."""
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.add, other)
        return self._apply_op_to_all_elements(operator.add, other)

    def __iadd__(self, other: Union[int, float, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.add, other)
        return self._apply_op_to_all_elements_inplace(operator.add, other)

    def __isub__(self, other: Union[int, float, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.sub, other)
        return self._apply_op_to_all_elements_inplace(operator.sub, other)

    def __sub__(self, other: Union[int, float, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.sub, other)
        return self._apply_op_to_all_elements(operator.sub, other)

    def __mul__(self, other: Union[int, float, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.mul, other)
        return self._apply_op_to_all_elements(operator.mul, other)

    def __imul__(self, other: Union[int, float, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.mul, other)
        return self._apply_op_to_all_elements_inplace(operator.mul, other)

    def __floordiv__(self, other: Union[int, float, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.floordiv, other)
        return self._apply_op_to_all_elements(operator.floordiv, other)

    def __ifloordiv__(self, other: Union[int, float, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.floordiv, other)
        return self._apply_op_to_all_elements_inplace(operator.floordiv, other)

    def __truediv__(self, other: Union[int, float, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.truediv, other)
        return self._apply_op_to_all_elements(operator.truediv, other)

    def __itruediv__(self, other: Union[int, float, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.truediv, other)
        return self._apply_op_to_all_elements_inplace(operator.truediv, other)

    def __rshift__(self, other: Union[int, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.rshift, other)
        return self._apply_op_to_all_elements(operator.rshift, other)

    def __lshift__(self, other: Union[int, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.lshift, other)
        return self._apply_op_to_all_elements(operator.lshift, other)

    def __irshift__(self, other: Union[int, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.rshift, other)
        return self._apply_op_to_all_elements_inplace(operator.rshift, other)

    def __ilshift__(self, other: Union[int, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.lshift, other)
        return self._apply_op_to_all_elements_inplace(operator.lshift, other)

    def __mod__(self, other: Union[int, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.mod, other)
        return self._apply_op_to_all_elements(operator.mod, other)

    def __imod__(self, other: Union[int, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.mod, other)
        return self._apply_op_to_all_elements_inplace(operator.mod, other)

    # Bitwise operators

    def __and__(self, other: BitsType) -> Array:
        return self._apply_bitwise_op_to_all_elements(operator.iand, other)

    def __iand__(self, other: BitsType) -> Array:
        return self._apply_bitwise_op_to_all_elements_inplace(operator.iand, other)

    def __or__(self, other: BitsType) -> Array:
        return self._apply_bitwise_op_to_all_elements(operator.ior, other)

    def __ior__(self, other: BitsType) -> Array:
        return self._apply_bitwise_op_to_all_elements_inplace(operator.ior, other)

    def __xor__(self, other: BitsType) -> Array:
        return self._apply_bitwise_op_to_all_elements(operator.ixor, other)

    def __ixor__(self, other: BitsType) -> Array:
        return self._apply_bitwise_op_to_all_elements_inplace(operator.ixor, other)

    # Reverse operators between a scalar value and an Array

    def __rmul__(self, other: Union[int, float]) -> Array:
        return self._apply_op_to_all_elements(operator.mul, other)

    def __radd__(self, other: Union[int, float]) -> Array:
        return self._apply_op_to_all_elements(operator.add, other)

    def __rsub__(self, other: Union[int, float]) -> Array:
        # i - A == (-A) + i
        neg = self._apply_op_to_all_elements(operator.neg, None)
        return neg._apply_op_to_all_elements(operator.add, other)

    # Reverse operators between a scalar and something that can be a BitArray.

    def __rand__(self, other: BitsType) -> Array:
        return self._apply_bitwise_op_to_all_elements(operator.iand, other)

    def __ror__(self, other: BitsType) -> Array:
        return self._apply_bitwise_op_to_all_elements(operator.ior, other)

    def __rxor__(self, other: BitsType) -> Array:
        return self._apply_bitwise_op_to_all_elements(operator.ixor, other)

    # Comparison operators

    def __lt__(self, other: Union[int, float, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.lt, other, is_comparison=True)
        return self._apply_op_to_all_elements(operator.lt, other, is_comparison=True)

    def __gt__(self, other: Union[int, float, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.gt, other, is_comparison=True)
        return self._apply_op_to_all_elements(operator.gt, other, is_comparison=True)

    def __ge__(self, other: Union[int, float, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.ge, other, is_comparison=True)
        return self._apply_op_to_all_elements(operator.ge, other, is_comparison=True)

    def __le__(self, other: Union[int, float, Array]) -> Array:
        if isinstance(other, Array):
            return self._apply_op_between_arrays(operator.le, other, is_comparison=True)
        return self._apply_op_to_all_elements(operator.le, other, is_comparison=True)

    def _eq_ne(self, op, other: Any) -> Array:
        if isinstance(other, (int, float, str, Bits)):
            return self._apply_op_to_all_elements(op, other, is_comparison=True)
        other = self.__class__(self.dtype, other)
        return self._apply_op_between_arrays(op, other, is_comparison=True)

    def __eq__(self, other: Any) -> Array:
        return self._eq_ne(operator.eq, other)

    def __ne__(self, other: Any) -> Array:
        return self._eq_ne(operator.ne, other)

    # Unary operators

    def __neg__(self):
        return self._apply_op_to_all_elements(operator.neg, None)

    def __abs__(self):
        return self._apply_op_to_all_elements(operator.abs, None)