File size: 33,978 Bytes
b6068b4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
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
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
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
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
import asyncio
import asyncio.events
import functools
import inspect
import io
import numbers
import os
import re
import threading
from contextlib import contextmanager
from glob import has_magic
from typing import TYPE_CHECKING, Iterable

from .callbacks import _DEFAULT_CALLBACK
from .exceptions import FSTimeoutError
from .implementations.local import (
    LocalFileSystem,
    make_path_posix,
    trailing_sep,
    trailing_sep_maybe_asterisk,
)
from .spec import AbstractBufferedFile, AbstractFileSystem
from .utils import is_exception, other_paths

private = re.compile("_[^_]")
iothread = [None]  # dedicated fsspec IO thread
loop = [None]  # global event loop for any non-async instance
_lock = None  # global lock placeholder
get_running_loop = asyncio.get_running_loop


def get_lock():
    """Allocate or return a threading lock.

    The lock is allocated on first use to allow setting one lock per forked process.
    """
    global _lock
    if not _lock:
        _lock = threading.Lock()
    return _lock


def reset_lock():
    """Reset the global lock.

    This should be called only on the init of a forked process to reset the lock to
    None, enabling the new forked process to get a new lock.
    """
    global _lock

    iothread[0] = None
    loop[0] = None
    _lock = None


async def _runner(event, coro, result, timeout=None):
    timeout = timeout if timeout else None  # convert 0 or 0.0 to None
    if timeout is not None:
        coro = asyncio.wait_for(coro, timeout=timeout)
    try:
        result[0] = await coro
    except Exception as ex:
        result[0] = ex
    finally:
        event.set()


def sync(loop, func, *args, timeout=None, **kwargs):
    """
    Make loop run coroutine until it returns. Runs in other thread

    Examples
    --------
    >>> fsspec.asyn.sync(fsspec.asyn.get_loop(), func, *args,
                         timeout=timeout, **kwargs)
    """
    timeout = timeout if timeout else None  # convert 0 or 0.0 to None
    # NB: if the loop is not running *yet*, it is OK to submit work
    # and we will wait for it
    if loop is None or loop.is_closed():
        raise RuntimeError("Loop is not running")
    try:
        loop0 = asyncio.events.get_running_loop()
        if loop0 is loop:
            raise NotImplementedError("Calling sync() from within a running loop")
    except RuntimeError:
        pass
    coro = func(*args, **kwargs)
    result = [None]
    event = threading.Event()
    asyncio.run_coroutine_threadsafe(_runner(event, coro, result, timeout), loop)
    while True:
        # this loops allows thread to get interrupted
        if event.wait(1):
            break
        if timeout is not None:
            timeout -= 1
            if timeout < 0:
                raise FSTimeoutError

    return_result = result[0]
    if isinstance(return_result, asyncio.TimeoutError):
        # suppress asyncio.TimeoutError, raise FSTimeoutError
        raise FSTimeoutError from return_result
    elif isinstance(return_result, BaseException):
        raise return_result
    else:
        return return_result


def sync_wrapper(func, obj=None):
    """Given a function, make so can be called in async or blocking contexts

    Leave obj=None if defining within a class. Pass the instance if attaching
    as an attribute of the instance.
    """

    @functools.wraps(func)
    def wrapper(*args, **kwargs):
        self = obj or args[0]
        return sync(self.loop, func, *args, **kwargs)

    return wrapper


@contextmanager
def _selector_policy():
    original_policy = asyncio.get_event_loop_policy()
    try:
        if os.name == "nt" and hasattr(asyncio, "WindowsSelectorEventLoopPolicy"):
            asyncio.set_event_loop_policy(asyncio.WindowsSelectorEventLoopPolicy())

        yield
    finally:
        asyncio.set_event_loop_policy(original_policy)


def get_loop():
    """Create or return the default fsspec IO loop

    The loop will be running on a separate thread.
    """
    if loop[0] is None:
        with get_lock():
            # repeat the check just in case the loop got filled between the
            # previous two calls from another thread
            if loop[0] is None:
                with _selector_policy():
                    loop[0] = asyncio.new_event_loop()
                th = threading.Thread(target=loop[0].run_forever, name="fsspecIO")
                th.daemon = True
                th.start()
                iothread[0] = th
    return loop[0]


if TYPE_CHECKING:
    import resource

    ResourceError = resource.error
else:
    try:
        import resource
    except ImportError:
        resource = None
        ResourceError = OSError
    else:
        ResourceError = getattr(resource, "error", OSError)

_DEFAULT_BATCH_SIZE = 128
_NOFILES_DEFAULT_BATCH_SIZE = 1280


def _get_batch_size(nofiles=False):
    from fsspec.config import conf

    if nofiles:
        if "nofiles_gather_batch_size" in conf:
            return conf["nofiles_gather_batch_size"]
    else:
        if "gather_batch_size" in conf:
            return conf["gather_batch_size"]
    if nofiles:
        return _NOFILES_DEFAULT_BATCH_SIZE
    if resource is None:
        return _DEFAULT_BATCH_SIZE

    try:
        soft_limit, _ = resource.getrlimit(resource.RLIMIT_NOFILE)
    except (ImportError, ValueError, ResourceError):
        return _DEFAULT_BATCH_SIZE

    if soft_limit == resource.RLIM_INFINITY:
        return -1
    else:
        return soft_limit // 8


def running_async() -> bool:
    """Being executed by an event loop?"""
    try:
        asyncio.get_running_loop()
        return True
    except RuntimeError:
        return False


async def _run_coros_in_chunks(
    coros,
    batch_size=None,
    callback=_DEFAULT_CALLBACK,
    timeout=None,
    return_exceptions=False,
    nofiles=False,
):
    """Run the given coroutines in  chunks.

    Parameters
    ----------
    coros: list of coroutines to run
    batch_size: int or None
        Number of coroutines to submit/wait on simultaneously.
        If -1, then it will not be any throttling. If
        None, it will be inferred from _get_batch_size()
    callback: fsspec.callbacks.Callback instance
        Gets a relative_update when each coroutine completes
    timeout: number or None
        If given, each coroutine times out after this time. Note that, since
        there are multiple batches, the total run time of this function will in
        general be longer
    return_exceptions: bool
        Same meaning as in asyncio.gather
    nofiles: bool
        If inferring the batch_size, does this operation involve local files?
        If yes, you normally expect smaller batches.
    """

    if batch_size is None:
        batch_size = _get_batch_size(nofiles=nofiles)

    if batch_size == -1:
        batch_size = len(coros)

    assert batch_size > 0
    results = []
    for start in range(0, len(coros), batch_size):
        chunk = [
            asyncio.Task(asyncio.wait_for(c, timeout=timeout))
            for c in coros[start : start + batch_size]
        ]
        if callback is not _DEFAULT_CALLBACK:
            [
                t.add_done_callback(lambda *_, **__: callback.relative_update(1))
                for t in chunk
            ]
        results.extend(
            await asyncio.gather(*chunk, return_exceptions=return_exceptions),
        )
    return results


# these methods should be implemented as async by any async-able backend
async_methods = [
    "_ls",
    "_cat_file",
    "_get_file",
    "_put_file",
    "_rm_file",
    "_cp_file",
    "_pipe_file",
    "_expand_path",
    "_info",
    "_isfile",
    "_isdir",
    "_exists",
    "_walk",
    "_glob",
    "_find",
    "_du",
    "_size",
    "_mkdir",
    "_makedirs",
]


class AsyncFileSystem(AbstractFileSystem):
    """Async file operations, default implementations

    Passes bulk operations to asyncio.gather for concurrent operation.

    Implementations that have concurrent batch operations and/or async methods
    should inherit from this class instead of AbstractFileSystem. Docstrings are
    copied from the un-underscored method in AbstractFileSystem, if not given.
    """

    # note that methods do not have docstring here; they will be copied
    # for _* methods and inferred for overridden methods.

    async_impl = True
    mirror_sync_methods = True
    disable_throttling = False

    def __init__(self, *args, asynchronous=False, loop=None, batch_size=None, **kwargs):
        self.asynchronous = asynchronous
        self._pid = os.getpid()
        if not asynchronous:
            self._loop = loop or get_loop()
        else:
            self._loop = None
        self.batch_size = batch_size
        super().__init__(*args, **kwargs)

    @property
    def loop(self):
        if self._pid != os.getpid():
            raise RuntimeError("This class is not fork-safe")
        return self._loop

    async def _rm_file(self, path, **kwargs):
        raise NotImplementedError

    async def _rm(self, path, recursive=False, batch_size=None, **kwargs):
        # TODO: implement on_error
        batch_size = batch_size or self.batch_size
        path = await self._expand_path(path, recursive=recursive)
        return await _run_coros_in_chunks(
            [self._rm_file(p, **kwargs) for p in reversed(path)],
            batch_size=batch_size,
            nofiles=True,
        )

    async def _cp_file(self, path1, path2, **kwargs):
        raise NotImplementedError

    async def _copy(
        self,
        path1,
        path2,
        recursive=False,
        on_error=None,
        maxdepth=None,
        batch_size=None,
        **kwargs,
    ):
        if on_error is None and recursive:
            on_error = "ignore"
        elif on_error is None:
            on_error = "raise"

        source_is_str = isinstance(path1, str)
        paths = await self._expand_path(path1, maxdepth=maxdepth, recursive=recursive)
        if source_is_str and (not recursive or maxdepth is not None):
            # Non-recursive glob does not copy directories
            paths = [p for p in paths if not (trailing_sep(p) or await self._isdir(p))]
            if not paths:
                return

        isdir = isinstance(path2, str) and (
            trailing_sep(path2) or await self._isdir(path2)
        )
        path2 = other_paths(
            paths,
            path2,
            exists=isdir and source_is_str and not trailing_sep_maybe_asterisk(path1),
            is_dir=isdir,
            flatten=not source_is_str,
        )
        batch_size = batch_size or self.batch_size
        coros = [self._cp_file(p1, p2, **kwargs) for p1, p2 in zip(paths, path2)]
        result = await _run_coros_in_chunks(
            coros, batch_size=batch_size, return_exceptions=True, nofiles=True
        )

        for ex in filter(is_exception, result):
            if on_error == "ignore" and isinstance(ex, FileNotFoundError):
                continue
            raise ex

    async def _pipe_file(self, path, value, **kwargs):
        raise NotImplementedError

    async def _pipe(self, path, value=None, batch_size=None, **kwargs):
        if isinstance(path, str):
            path = {path: value}
        batch_size = batch_size or self.batch_size
        return await _run_coros_in_chunks(
            [self._pipe_file(k, v, **kwargs) for k, v in path.items()],
            batch_size=batch_size,
            nofiles=True,
        )

    async def _process_limits(self, url, start, end):
        """Helper for "Range"-based _cat_file"""
        size = None
        suff = False
        if start is not None and start < 0:
            # if start is negative and end None, end is the "suffix length"
            if end is None:
                end = -start
                start = ""
                suff = True
            else:
                size = size or (await self._info(url))["size"]
                start = size + start
        elif start is None:
            start = 0
        if not suff:
            if end is not None and end < 0:
                if start is not None:
                    size = size or (await self._info(url))["size"]
                    end = size + end
            elif end is None:
                end = ""
            if isinstance(end, numbers.Integral):
                end -= 1  # bytes range is inclusive
        return "bytes=%s-%s" % (start, end)

    async def _cat_file(self, path, start=None, end=None, **kwargs):
        raise NotImplementedError

    async def _cat(
        self, path, recursive=False, on_error="raise", batch_size=None, **kwargs
    ):
        paths = await self._expand_path(path, recursive=recursive)
        coros = [self._cat_file(path, **kwargs) for path in paths]
        batch_size = batch_size or self.batch_size
        out = await _run_coros_in_chunks(
            coros, batch_size=batch_size, nofiles=True, return_exceptions=True
        )
        if on_error == "raise":
            ex = next(filter(is_exception, out), False)
            if ex:
                raise ex
        if (
            len(paths) > 1
            or isinstance(path, list)
            or paths[0] != self._strip_protocol(path)
        ):
            return {
                k: v
                for k, v in zip(paths, out)
                if on_error != "omit" or not is_exception(v)
            }
        else:
            return out[0]

    async def _cat_ranges(
        self,
        paths,
        starts,
        ends,
        max_gap=None,
        batch_size=None,
        on_error="return",
        **kwargs,
    ):
        # TODO: on_error
        if max_gap is not None:
            # use utils.merge_offset_ranges
            raise NotImplementedError
        if not isinstance(paths, list):
            raise TypeError
        if not isinstance(starts, Iterable):
            starts = [starts] * len(paths)
        if not isinstance(ends, Iterable):
            ends = [starts] * len(paths)
        if len(starts) != len(paths) or len(ends) != len(paths):
            raise ValueError
        coros = [
            self._cat_file(p, start=s, end=e, **kwargs)
            for p, s, e in zip(paths, starts, ends)
        ]
        batch_size = batch_size or self.batch_size
        return await _run_coros_in_chunks(
            coros, batch_size=batch_size, nofiles=True, return_exceptions=True
        )

    async def _put_file(self, lpath, rpath, **kwargs):
        raise NotImplementedError

    async def _put(
        self,
        lpath,
        rpath,
        recursive=False,
        callback=_DEFAULT_CALLBACK,
        batch_size=None,
        maxdepth=None,
        **kwargs,
    ):
        """Copy file(s) from local.

        Copies a specific file or tree of files (if recursive=True). If rpath
        ends with a "/", it will be assumed to be a directory, and target files
        will go within.

        The put_file method will be called concurrently on a batch of files. The
        batch_size option can configure the amount of futures that can be executed
        at the same time. If it is -1, then all the files will be uploaded concurrently.
        The default can be set for this instance by passing "batch_size" in the
        constructor, or for all instances by setting the "gather_batch_size" key
        in ``fsspec.config.conf``, falling back to 1/8th of the system limit .
        """
        source_is_str = isinstance(lpath, str)
        if source_is_str:
            lpath = make_path_posix(lpath)
        fs = LocalFileSystem()
        lpaths = fs.expand_path(lpath, recursive=recursive, maxdepth=maxdepth)
        if source_is_str and (not recursive or maxdepth is not None):
            # Non-recursive glob does not copy directories
            lpaths = [p for p in lpaths if not (trailing_sep(p) or fs.isdir(p))]
            if not lpaths:
                return

        isdir = isinstance(rpath, str) and (
            trailing_sep(rpath) or await self._isdir(rpath)
        )
        rpath = self._strip_protocol(rpath)
        rpaths = other_paths(
            lpaths,
            rpath,
            exists=isdir and source_is_str and not trailing_sep_maybe_asterisk(lpath),
            is_dir=isdir,
            flatten=not source_is_str,
        )

        is_dir = {l: os.path.isdir(l) for l in lpaths}
        rdirs = [r for l, r in zip(lpaths, rpaths) if is_dir[l]]
        file_pairs = [(l, r) for l, r in zip(lpaths, rpaths) if not is_dir[l]]

        await asyncio.gather(*[self._makedirs(d, exist_ok=True) for d in rdirs])
        batch_size = batch_size or self.batch_size

        coros = []
        callback.set_size(len(file_pairs))
        for lfile, rfile in file_pairs:
            callback.branch(lfile, rfile, kwargs)
            coros.append(self._put_file(lfile, rfile, **kwargs))

        return await _run_coros_in_chunks(
            coros, batch_size=batch_size, callback=callback
        )

    async def _get_file(self, rpath, lpath, **kwargs):
        raise NotImplementedError

    async def _get(
        self,
        rpath,
        lpath,
        recursive=False,
        callback=_DEFAULT_CALLBACK,
        maxdepth=None,
        **kwargs,
    ):
        """Copy file(s) to local.

        Copies a specific file or tree of files (if recursive=True). If lpath
        ends with a "/", it will be assumed to be a directory, and target files
        will go within. Can submit a list of paths, which may be glob-patterns
        and will be expanded.

        The get_file method will be called concurrently on a batch of files. The
        batch_size option can configure the amount of futures that can be executed
        at the same time. If it is -1, then all the files will be uploaded concurrently.
        The default can be set for this instance by passing "batch_size" in the
        constructor, or for all instances by setting the "gather_batch_size" key
        in ``fsspec.config.conf``, falling back to 1/8th of the system limit .
        """
        source_is_str = isinstance(rpath, str)
        # First check for rpath trailing slash as _strip_protocol removes it.
        source_not_trailing_sep = source_is_str and not trailing_sep_maybe_asterisk(
            rpath
        )
        rpath = self._strip_protocol(rpath)
        rpaths = await self._expand_path(rpath, recursive=recursive)
        if source_is_str and (not recursive or maxdepth is not None):
            # Non-recursive glob does not copy directories
            rpaths = [
                p for p in rpaths if not (trailing_sep(p) or await self._isdir(p))
            ]
            if not rpaths:
                return

        lpath = make_path_posix(lpath)
        isdir = isinstance(lpath, str) and (
            trailing_sep(lpath) or LocalFileSystem().isdir(lpath)
        )
        lpaths = other_paths(
            rpaths,
            lpath,
            exists=isdir and source_not_trailing_sep,
            is_dir=isdir,
            flatten=not source_is_str,
        )
        [os.makedirs(os.path.dirname(lp), exist_ok=True) for lp in lpaths]
        batch_size = kwargs.pop("batch_size", self.batch_size)

        coros = []
        callback.set_size(len(lpaths))
        for lpath, rpath in zip(lpaths, rpaths):
            callback.branch(rpath, lpath, kwargs)
            coros.append(self._get_file(rpath, lpath, **kwargs))
        return await _run_coros_in_chunks(
            coros, batch_size=batch_size, callback=callback
        )

    async def _isfile(self, path):
        try:
            return (await self._info(path))["type"] == "file"
        except:  # noqa: E722
            return False

    async def _isdir(self, path):
        try:
            return (await self._info(path))["type"] == "directory"
        except OSError:
            return False

    async def _size(self, path):
        return (await self._info(path)).get("size", None)

    async def _sizes(self, paths, batch_size=None):
        batch_size = batch_size or self.batch_size
        return await _run_coros_in_chunks(
            [self._size(p) for p in paths], batch_size=batch_size
        )

    async def _exists(self, path):
        try:
            await self._info(path)
            return True
        except FileNotFoundError:
            return False

    async def _info(self, path, **kwargs):
        raise NotImplementedError

    async def _ls(self, path, detail=True, **kwargs):
        raise NotImplementedError

    async def _walk(self, path, maxdepth=None, **kwargs):
        if maxdepth is not None and maxdepth < 1:
            raise ValueError("maxdepth must be at least 1")

        path = self._strip_protocol(path)
        full_dirs = {}
        dirs = {}
        files = {}

        detail = kwargs.pop("detail", False)
        try:
            listing = await self._ls(path, detail=True, **kwargs)
        except (FileNotFoundError, OSError):
            if detail:
                yield path, {}, {}
            else:
                yield path, [], []
            return

        for info in listing:
            # each info name must be at least [path]/part , but here
            # we check also for names like [path]/part/
            pathname = info["name"].rstrip("/")
            name = pathname.rsplit("/", 1)[-1]
            if info["type"] == "directory" and pathname != path:
                # do not include "self" path
                full_dirs[name] = pathname
                dirs[name] = info
            elif pathname == path:
                # file-like with same name as give path
                files[""] = info
            else:
                files[name] = info

        if detail:
            yield path, dirs, files
        else:
            yield path, list(dirs), list(files)

        if maxdepth is not None:
            maxdepth -= 1
            if maxdepth < 1:
                return

        for d in dirs:
            async for _ in self._walk(
                full_dirs[d], maxdepth=maxdepth, detail=detail, **kwargs
            ):
                yield _

    async def _glob(self, path, **kwargs):
        import re

        ends = path.endswith("/")
        path = self._strip_protocol(path)
        indstar = path.find("*") if path.find("*") >= 0 else len(path)
        indques = path.find("?") if path.find("?") >= 0 else len(path)
        indbrace = path.find("[") if path.find("[") >= 0 else len(path)

        ind = min(indstar, indques, indbrace)

        detail = kwargs.pop("detail", False)

        if not has_magic(path):
            root = path
            depth = 1
            if ends:
                path += "/*"
            elif await self._exists(path):
                if not detail:
                    return [path]
                else:
                    return {path: await self._info(path)}
            else:
                if not detail:
                    return []  # glob of non-existent returns empty
                else:
                    return {}
        elif "/" in path[:ind]:
            ind2 = path[:ind].rindex("/")
            root = path[: ind2 + 1]
            depth = None if "**" in path else path[ind2 + 1 :].count("/") + 1
        else:
            root = ""
            depth = None if "**" in path else path[ind + 1 :].count("/") + 1

        allpaths = await self._find(
            root, maxdepth=depth, withdirs=True, detail=True, **kwargs
        )
        # Escape characters special to python regex, leaving our supported
        # special characters in place.
        # See https://www.gnu.org/software/bash/manual/html_node/Pattern-Matching.html
        # for shell globbing details.
        pattern = (
            "^"
            + (
                path.replace("\\", r"\\")
                .replace(".", r"\.")
                .replace("+", r"\+")
                .replace("//", "/")
                .replace("(", r"\(")
                .replace(")", r"\)")
                .replace("|", r"\|")
                .replace("^", r"\^")
                .replace("$", r"\$")
                .replace("{", r"\{")
                .replace("}", r"\}")
                .rstrip("/")
                .replace("?", ".")
            )
            + "$"
        )
        pattern = re.sub("[*]{2}", "=PLACEHOLDER=", pattern)
        pattern = re.sub("[*]", "[^/]*", pattern)
        pattern = re.compile(pattern.replace("=PLACEHOLDER=", ".*"))
        out = {
            p: allpaths[p]
            for p in sorted(allpaths)
            if pattern.match(p.replace("//", "/").rstrip("/"))
        }
        if detail:
            return out
        else:
            return list(out)

    async def _du(self, path, total=True, maxdepth=None, **kwargs):
        sizes = {}
        # async for?
        for f in await self._find(path, maxdepth=maxdepth, **kwargs):
            info = await self._info(f)
            sizes[info["name"]] = info["size"]
        if total:
            return sum(sizes.values())
        else:
            return sizes

    async def _find(self, path, maxdepth=None, withdirs=False, **kwargs):
        path = self._strip_protocol(path)
        out = dict()
        detail = kwargs.pop("detail", False)
        # async for?
        async for _, dirs, files in self._walk(path, maxdepth, detail=True, **kwargs):
            if withdirs:
                files.update(dirs)
            out.update({info["name"]: info for name, info in files.items()})
        if not out and (await self._isfile(path)):
            # walk works on directories, but find should also return [path]
            # when path happens to be a file
            out[path] = {}
        names = sorted(out)
        if not detail:
            return names
        else:
            return {name: out[name] for name in names}

    async def _expand_path(self, path, recursive=False, maxdepth=None):
        if maxdepth is not None and maxdepth < 1:
            raise ValueError("maxdepth must be at least 1")

        if isinstance(path, str):
            out = await self._expand_path([path], recursive, maxdepth)
        else:
            out = set()
            path = [self._strip_protocol(p) for p in path]
            for p in path:  # can gather here
                if has_magic(p):
                    bit = set(await self._glob(p))
                    out |= bit
                    if recursive:
                        # glob call above expanded one depth so if maxdepth is defined
                        # then decrement it in expand_path call below. If it is zero
                        # after decrementing then avoid expand_path call.
                        if maxdepth is not None and maxdepth <= 1:
                            continue
                        out |= set(
                            await self._expand_path(
                                list(bit),
                                recursive=recursive,
                                maxdepth=maxdepth - 1 if maxdepth is not None else None,
                            )
                        )
                    continue
                elif recursive:
                    rec = set(await self._find(p, maxdepth=maxdepth, withdirs=True))
                    out |= rec
                if p not in out and (recursive is False or (await self._exists(p))):
                    # should only check once, for the root
                    out.add(p)
        if not out:
            raise FileNotFoundError(path)
        return list(sorted(out))

    async def _mkdir(self, path, create_parents=True, **kwargs):
        pass  # not necessary to implement, may not have directories

    async def _makedirs(self, path, exist_ok=False):
        pass  # not necessary to implement, may not have directories

    async def open_async(self, path, mode="rb", **kwargs):
        if "b" not in mode or kwargs.get("compression"):
            raise ValueError
        raise NotImplementedError


def mirror_sync_methods(obj):
    """Populate sync and async methods for obj

    For each method will create a sync version if the name refers to an async method
    (coroutine) and there is no override in the child class; will create an async
    method for the corresponding sync method if there is no implementation.

    Uses the methods specified in
    - async_methods: the set that an implementation is expected to provide
    - default_async_methods: that can be derived from their sync version in
      AbstractFileSystem
    - AsyncFileSystem: async-specific default coroutines
    """
    from fsspec import AbstractFileSystem

    for method in async_methods + dir(AsyncFileSystem):
        if not method.startswith("_"):
            continue
        smethod = method[1:]
        if private.match(method):
            isco = inspect.iscoroutinefunction(getattr(obj, method, None))
            unsync = getattr(getattr(obj, smethod, False), "__func__", None)
            is_default = unsync is getattr(AbstractFileSystem, smethod, "")
            if isco and is_default:
                mth = sync_wrapper(getattr(obj, method), obj=obj)
                setattr(obj, smethod, mth)
                if not mth.__doc__:
                    mth.__doc__ = getattr(
                        getattr(AbstractFileSystem, smethod, None), "__doc__", ""
                    )


class FSSpecCoroutineCancel(Exception):
    pass


def _dump_running_tasks(
    printout=True, cancel=True, exc=FSSpecCoroutineCancel, with_task=False
):
    import traceback

    tasks = [t for t in asyncio.tasks.all_tasks(loop[0]) if not t.done()]
    if printout:
        [task.print_stack() for task in tasks]
    out = [
        {
            "locals": task._coro.cr_frame.f_locals,
            "file": task._coro.cr_frame.f_code.co_filename,
            "firstline": task._coro.cr_frame.f_code.co_firstlineno,
            "linelo": task._coro.cr_frame.f_lineno,
            "stack": traceback.format_stack(task._coro.cr_frame),
            "task": task if with_task else None,
        }
        for task in tasks
    ]
    if cancel:
        for t in tasks:
            cbs = t._callbacks
            t.cancel()
            asyncio.futures.Future.set_exception(t, exc)
            asyncio.futures.Future.cancel(t)
            [cb[0](t) for cb in cbs]  # cancels any dependent concurrent.futures
            try:
                t._coro.throw(exc)  # exits coro, unless explicitly handled
            except exc:
                pass
    return out


class AbstractAsyncStreamedFile(AbstractBufferedFile):
    # no read buffering, and always auto-commit
    # TODO: readahead might still be useful here, but needs async version

    async def read(self, length=-1):
        """
        Return data from cache, or fetch pieces as necessary

        Parameters
        ----------
        length: int (-1)
            Number of bytes to read; if <0, all remaining bytes.
        """
        length = -1 if length is None else int(length)
        if self.mode != "rb":
            raise ValueError("File not in read mode")
        if length < 0:
            length = self.size - self.loc
        if self.closed:
            raise ValueError("I/O operation on closed file.")
        if length == 0:
            # don't even bother calling fetch
            return b""
        out = await self._fetch_range(self.loc, self.loc + length)
        self.loc += len(out)
        return out

    async def write(self, data):
        """
        Write data to buffer.

        Buffer only sent on flush() or if buffer is greater than
        or equal to blocksize.

        Parameters
        ----------
        data: bytes
            Set of bytes to be written.
        """
        if self.mode not in {"wb", "ab"}:
            raise ValueError("File not in write mode")
        if self.closed:
            raise ValueError("I/O operation on closed file.")
        if self.forced:
            raise ValueError("This file has been force-flushed, can only close")
        out = self.buffer.write(data)
        self.loc += out
        if self.buffer.tell() >= self.blocksize:
            await self.flush()
        return out

    async def close(self):
        """Close file

        Finalizes writes, discards cache
        """
        if getattr(self, "_unclosable", False):
            return
        if self.closed:
            return
        if self.mode == "rb":
            self.cache = None
        else:
            if not self.forced:
                await self.flush(force=True)

            if self.fs is not None:
                self.fs.invalidate_cache(self.path)
                self.fs.invalidate_cache(self.fs._parent(self.path))

        self.closed = True

    async def flush(self, force=False):
        if self.closed:
            raise ValueError("Flush on closed file")
        if force and self.forced:
            raise ValueError("Force flush cannot be called more than once")
        if force:
            self.forced = True

        if self.mode not in {"wb", "ab"}:
            # no-op to flush on read-mode
            return

        if not force and self.buffer.tell() < self.blocksize:
            # Defer write on small block
            return

        if self.offset is None:
            # Initialize a multipart upload
            self.offset = 0
            try:
                await self._initiate_upload()
            except:  # noqa: E722
                self.closed = True
                raise

        if await self._upload_chunk(final=force) is not False:
            self.offset += self.buffer.seek(0, 2)
            self.buffer = io.BytesIO()

    async def __aenter__(self):
        return self

    async def __aexit__(self, exc_type, exc_val, exc_tb):
        await self.close()

    async def _fetch_range(self, start, end):
        raise NotImplementedError

    async def _initiate_upload(self):
        pass

    async def _upload_chunk(self, final=False):
        raise NotImplementedError