Repository navigation
Expand file tree
/
Copy pathasync_cursor.py
More file actions
923 lines (842 loc) · 41 KB
/
Copy pathasync_cursor.py
File metadata and controls
923 lines (842 loc) · 41 KB
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
"""Native asynchronous query lifecycle, with no replay of consumed results."""
import asyncio
import logging
import time
from collections import deque
from contextlib import contextmanager
from io import BytesIO
import grpc
from .datainputstream import get_query_columns_info, read_rows_from_chunk
from .e6data_grpc import _escaper
from .exceptions import (AmbiguousSubmissionError, IncompleteResultError,
NotSupportedError, OperationalError, ProgrammingError, OAuthError)
from .server import e6x_engine_pb2 as pb
from .result_batch import ResultBatchBuffer, decode_result_batches
from .result_prefetch import ContinuousResultStream, PendingFetch
logger = logging.getLogger(__name__)
def _log_result_batch(protocol, rpc_seconds, status, response=None, decode_seconds=0.0):
chunk_count = ((len(response.resultBatches) if protocol == 'v2'
else int(bool(response.resultBatch))) if response else 0)
serialized_bytes = response.ByteSize() if response else 0
logger.debug(
'Result batch fetch protocol=%s rpc_seconds=%.6f decode_seconds=%.6f '
'chunk_count=%d serialized_bytes=%d status=%s',
protocol, rpc_seconds, decode_seconds, chunk_count, serialized_bytes, status, extra={
'result_batch_protocol': protocol,
'result_batch_rpc_seconds': rpc_seconds,
'result_batch_chunk_count': chunk_count,
'result_batch_serialized_bytes': serialized_bytes,
'result_batch_decode_seconds': decode_seconds,
'result_batch_status': status,
})
def _metadata_columns(payload):
return get_query_columns_info(BytesIO(payload))
def _operational(error):
if isinstance(error, (OperationalError, ProgrammingError, NotSupportedError, OAuthError)):
return error
return OperationalError('Query operation failed; inspect the chained cause.')
class AsyncCursor:
"""One ordinary operation at a time; different cursors may run concurrently."""
def __init__(self, connection, array_size=1000, database=None, catalog_name=None, db_name=None):
self._connection = connection
self._public_connection = connection
self._lease_guard = connection._lease_guard
self._catalog = connection.catalog if catalog_name is None else catalog_name
self._database = connection.database if database is None and db_name is None else (database if db_name is None else db_name)
self.arraysize = array_size
self._state = 'EMPTY'
self._route = None
self._rows = deque()
self._columns = None
self._description = None
self._rowcount = -1
self._rownumber = 0
self._failure = None
self._busy = False
self._operation_task = None
self._revision = 0
self._active_call = None
self._close_task = None
self._cleanup_error = None
self._result_batches = ResultBatchBuffer()
self._pending_result = None
self._result_available = None
self._retired_results = []
self._decode_owner = None
self._reset_result_protocol()
connection._cursors.add(self)
def _reset_result_protocol(self):
self._result_batch_v2 = self._connection.enable_result_batch_v2
self._result_protocol = 'v2' if self._result_batch_v2 else 'v1'
self._result_session_id = None
self._result_batches.clear()
@property
def connection(self):
return self._public_connection
@property
def arraysize(self):
return self._arraysize
@arraysize.setter
def arraysize(self, value):
if isinstance(value, bool) or not isinstance(value, int) or value <= 0:
raise ValueError('arraysize must be a positive integer.')
self._arraysize = value
@property
def rownumber(self):
return self._rownumber
@property
def lastrowid(self):
return None
@property
def rowcount(self):
return self._rowcount
@property
def description(self):
return self._description
@property
def query_id(self):
return self._route.query_id if self._route else None
@property
def cleanup_error(self):
return self._cleanup_error
def _check(self, fetch=False):
if self._lease_guard is not None:
self._lease_guard(internal=False)
self._connection._check_owner()
if self._state == 'CLOSED':
raise ProgrammingError('Cursor is closed.')
if fetch and self._failure is not None:
raise IncompleteResultError(self._failure.reason, query_id=self._failure.query_id)
if fetch and self._state in ('EMPTY', 'SUBMISSION_UNKNOWN'):
raise ProgrammingError('No complete query is available for fetching.')
@contextmanager
def _operation(self, fetch=False):
self._check(fetch=fetch)
if self._busy:
raise ProgrammingError('A cursor permits one ordinary operation at a time.')
self._busy = True
self._operation_task = asyncio.current_task()
try:
yield
finally:
self._busy = False
self._operation_task = None
def _publish(self, revision):
self._check()
if revision != self._revision:
raise asyncio.CancelledError()
async def _call(self, name, request, deadline, *, safe=False, on_dispatch=None):
revision = self._revision
response_metadata = {}
task = asyncio.create_task(self._connection._rpc(
name, request, deadline=deadline, route=self._route,
safe_retry=safe, _on_dispatch=on_dispatch, _response_metadata=response_metadata))
self._active_call = task
try:
result = await task
self._publish(revision)
if name in ('prepareStatement', 'prepareStatementV2'):
self._prepare_strategy = response_metadata['strategy']
return result
finally:
if self._active_call is task:
self._active_call = None
def _fail_result(self, reason):
self._retire_prefetch()
self._cancel_decode()
if self._failure is None:
# Never raise this template: its traceback would retain decoded inputs.
self._failure = IncompleteResultError(reason, query_id=self.query_id)
if self._state != 'CLOSED':
self._state = 'RESULT_FAILED'
self._rows.clear()
self._result_batches.clear()
return IncompleteResultError(self._failure.reason, query_id=self._failure.query_id)
def _accept_rows(self, rows):
self._rows.extend(rows)
def _take_rows(self, size):
rows = [self._rows.popleft() for _ in range(min(size, len(self._rows)))]
self._rownumber += len(rows)
return rows
async def _query_request(self, cls, deadline, *, owned=False):
if self._route is None:
raise ProgrammingError('No query handle is known.')
if self._result_batch_v2 and not self._connection._oauth_enabled and self._result_session_id:
session = self._result_session_id
elif owned:
# Owned cleanup cannot mint a replacement authentication session.
session = '' if self._connection._oauth_enabled else self._connection._session_id
if session is None:
raise OperationalError('Cleanup has no authenticated legacy session.')
else:
session = await self._connection.get_session_id(deadline=deadline)
return cls(sessionId=session, queryId=self.query_id, engineIP=self._route.engine_ip)
async def _refresh_metadata(self, deadline):
from .async_work import run_blocking
request = await self._query_request(pb.GetResultMetadataRequest, deadline)
revision = self._revision
response = await self._call('getResultMetadata', request, deadline, safe=True)
count, columns = await run_blocking(_metadata_columns, response.resultMetaData, deadline=deadline)
self._publish(revision)
description = [(column.get_name(), column.get_field_type(), None, None, None, None, True)
for column in columns]
self._rowcount, self._columns, self._description = count, columns, description
async def _execute(self, operation, parameters, deadline):
from .async_connection import QueryRoute
if self._state != 'EMPTY':
raise ProgrammingError('Clear the previous query before executing another.')
self._reset_result_protocol()
if not isinstance(operation, str) or not operation.strip():
raise ProgrammingError('operation must be a nonempty SQL string.')
sql = operation.strip().removesuffix(';')
if parameters is not None:
sql = sql % _escaper.escape_args(parameters)
await self._connection._ensure_decoder(deadline)
session = await self._connection.get_session_id(deadline=deadline)
v2 = bool(self._catalog)
request_type = pb.PrepareStatementV2Request if v2 else pb.PrepareStatementRequest
fields = dict(sessionId=session, schema=self._database or '', queryString=sql)
if v2:
fields['catalog'] = self._catalog
request = request_type(**fields)
dispatched = False
def mark_dispatch():
nonlocal dispatched
dispatched = True
try:
try:
response = await self._call('prepareStatementV2' if v2 else 'prepareStatement',
request, deadline, safe=True, on_dispatch=mark_dispatch)
except grpc.RpcError as error:
if (self._connection.auto_resume and error.code() == grpc.StatusCode.UNAVAILABLE
and error.details() == 'status: 503, cluster is suspended'):
# This exact response explicitly denies preparation admission.
dispatched = False
await self._connection._resume_cluster(deadline)
response = await self._call('prepareStatementV2' if v2 else 'prepareStatement',
request, deadline, safe=True, on_dispatch=mark_dispatch)
else:
raise
if not response.queryId or not response.engineIP:
raise OperationalError('Preparation returned an incomplete query handle.')
route = QueryRoute(self._connection.target, response.queryId, response.engineIP,
self._prepare_strategy)
self._connection._register_route(route)
self._route = route
self._state = 'ACTIVE'
self._rows.clear()
self._columns = self._description = None
self._rowcount, self._rownumber = -1, 0
dispatched = False
execute_type = pb.ExecuteStatementV2Request if v2 else pb.ExecuteStatementRequest
execute_request = await self._query_request(execute_type, deadline)
await self._call('executeStatementV2' if v2 else 'executeStatement', execute_request,
deadline, on_dispatch=mark_dispatch)
# Execution succeeded; subsequent metadata failure must retain this known query.
dispatched = False
await self._refresh_metadata(deadline)
return self.query_id
except asyncio.CancelledError:
if dispatched:
if self._state != 'CLOSED':
self._state = 'SUBMISSION_UNKNOWN'
self._connection._ambiguous_submissions.add(self)
raise
except Exception as error:
denied = isinstance(error, grpc.RpcError) and (
error.code() in (grpc.StatusCode.UNAUTHENTICATED, grpc.StatusCode.PERMISSION_DENIED,
grpc.StatusCode.INVALID_ARGUMENT)
or (error.code() == grpc.StatusCode.UNKNOWN and error.details() == 'status: 456')
or (error.code() == grpc.StatusCode.UNAVAILABLE and error.details() == 'status: 503, cluster is suspended'))
if dispatched and not denied:
if self._state != 'CLOSED':
self._state = 'SUBMISSION_UNKNOWN'
self._connection._ambiguous_submissions.add(self)
raise AmbiguousSubmissionError(query_id=self.query_id) from error
if isinstance(error, (ProgrammingError, OperationalError, OAuthError)):
raise
raise _operational(error) from error
async def execute(self, operation, parameters=None, *, timeout=None, **kwargs):
if kwargs:
raise ProgrammingError('Unsupported execute options.')
with self._operation():
return await self._execute(operation, parameters, self._connection._deadline(timeout))
async def executemany(self, operation, seq_of_parameters, *, timeout=None):
with self._operation():
deadline = self._connection._deadline(timeout)
for index, parameters in enumerate(seq_of_parameters):
try:
if index:
await self._clear(deadline)
await self._execute(operation, parameters, deadline)
except Exception as error:
error.parameter_index = index
raise
async def _decode_batch(self, payload, reservation, revision):
try:
rows = await reservation.run(read_rows_from_chunk, self._columns, payload, True)
except Exception as error:
raise self._fail_result('decode_failed') from error
self._publish(revision)
if rows is None:
self._state = 'EXHAUSTED'
return rows
async def _next_batch(self, deadline):
from .async_work import reserve_work
if self._result_batch_v2:
return await self._next_v2_batch(deadline)
if self._state == 'EXHAUSTED':
return None
if self._columns is None:
await self._refresh_metadata(deadline)
request = await self._query_request(pb.GetNextResultBatchRequest, deadline)
reservation = await reserve_work(deadline=deadline)
consumed = False
revision = self._revision
def mark_dispatch():
nonlocal consumed
consumed = True
started = time.monotonic()
response = None
decode_seconds = 0.0
try:
try:
response = await self._call('getNextResultBatch', request, deadline, on_dispatch=mark_dispatch)
finally:
rpc_seconds = time.monotonic() - started
payload = response.resultBatch
if not payload:
self._state = 'EXHAUSTED'
return None
decode_started = time.monotonic()
try:
return await self._decode_batch(payload, reservation, revision)
finally:
decode_seconds = time.monotonic() - decode_started
except asyncio.CancelledError:
if consumed:
self._fail_result('ambiguous_result')
raise
except grpc.RpcError:
if consumed:
self._fail_result('ambiguous_result')
raise
except IncompleteResultError:
raise
except Exception as error:
if consumed:
raise self._fail_result('ambiguous_result') from error
raise
finally:
reservation.release()
if consumed:
status = 'ok' if response is not None and self._failure is None else 'error'
_log_result_batch('v1', rpc_seconds,
status, response if status == 'ok' else None, decode_seconds if status == 'ok' else 0.0)
def _retain_result_session(self, response, revision):
self._publish(revision)
if self._result_batch_v2 and not self._connection._oauth_enabled and response.sessionId:
pending = self._pending_result[0] if self._pending_result is not None else None
self._result_session_id = (pending.session_id if isinstance(pending, ContinuousResultStream)
and pending.session_id else response.sessionId)
async def _decode_v2_response(self, response, reservation, revision, deadline=None):
self._retain_result_session(response, revision)
owner = None
try:
if self._connection._decoder_started:
owner = self._connection._decoder_lease.new_owner()
self._decode_owner = owner
chunks = await reservation.run(
self._connection._decoder_lease.decode, tuple(self._columns),
tuple(response.resultBatches), deadline, owner)
else:
chunks = await reservation.run(decode_result_batches, self._columns, response.resultBatches)
except asyncio.CancelledError:
self._cancel_decode()
raise
except Exception as error:
raise self._fail_result('decode_failed') from error
finally:
if self._decode_owner is owner:
self._decode_owner = None
self._publish(revision)
self._result_batches.accept(chunks, response.endOfStream)
def _cancel_decode(self):
if self._decode_owner is not None:
self._connection._decoder_lease.cancel(self._decode_owner)
def _retire_prefetch(self):
if self._pending_result is not None:
pending, _, _ = self._pending_result
self._pending_result = None
self._result_available = None
pending.retire()
self._retired_results.append(pending)
async def _dispatch_result(self, deadline, permit=None):
"""Freeze identity in the owner; the background task only runs transport."""
revision, route = self._revision, self._route
request = await self._query_request(pb.GetNextResultBatchRequest, deadline)
metadata = await self._connection._metadata(deadline, route)
self._publish(revision)
self._connection._remaining(deadline)
method = 'getNextResultBatchV2' if self._result_protocol == 'v2' else 'getNextResultBatch'
task = asyncio.create_task(self._connection._raw_result_rpc(method, request, tuple(metadata), deadline))
pending = PendingFetch(task, permit, session_id=request.sessionId, deadline=deadline)
self._pending_result = (pending, revision, route)
async def _take_pending_result(self, deadline):
pending, revision, route = self._pending_result
if isinstance(pending, ContinuousResultStream):
event = self._result_available
while not pending.ready:
event.clear()
if pending.ready:
break
async with asyncio.timeout_at(deadline):
await event.wait()
else:
handle = pending.handle
if handle is not None and not pending.settled:
done, _ = await asyncio.wait({handle}, timeout=self._connection._remaining(deadline))
if not done:
raise TimeoutError('Result transport exceeded the operation deadline.')
self._publish(revision)
if route != self._route:
raise asyncio.CancelledError()
if not self._connection._oauth_enabled and pending.session_id:
self._result_session_id = pending.session_id
if not isinstance(pending, ContinuousResultStream):
self._pending_result = None
response = pending.take()
new_strategy = getattr(response, 'new_strategy', None)
if new_strategy in ('blue', 'green'):
self._connection._pending_strategy = new_strategy
self._connection._apply_pending_strategy()
return response, pending.elapsed
async def _start_prefetch(self, response, deadline):
if (self._result_protocol != 'v2' or response.endOfStream
or self._pending_result is not None):
return
revision, route = self._revision, self._route
try:
self._publish(revision)
self._connection._remaining(deadline)
except BaseException:
logger.debug('Continuous result download preparation failed.', extra={
'result_batch_download_status': 'preparation_failed'})
raise
event = asyncio.Event()
stream = ContinuousResultStream(session_id=self._result_session_id, notify=event.set)
producer = asyncio.create_task(self._download_results(stream, revision, route))
stream.set_handle(producer)
self._result_available = event
self._pending_result = (stream, revision, route)
logger.debug('Continuous result download started.', extra={'result_batch_download_status': 'started'})
async def _download_results(self, stream, revision, route):
"""Download ordered envelopes without using any decode/work reservation."""
budget = self._connection.operation_timeout
no_progress_deadline = time.monotonic() + budget
backoff = .01
rpc_started = None
try:
while not stream.retired:
deadline = min(time.monotonic() + budget, no_progress_deadline)
rpc_started = None
metadata = await self._connection._metadata(deadline, route)
if stream.retired or revision != self._revision or route != self._route:
stream.retire()
return
if self._lease_guard is not None:
self._lease_guard(internal=True)
self._connection._check_owner(_internal=True)
self._connection._remaining(deadline)
session = '' if self._connection._oauth_enabled else stream.session_id
if not self._connection._oauth_enabled and not session:
raise ProgrammingError('An active query session is required to download results.')
request = pb.GetNextResultBatchRequest(
sessionId=session, queryId=route.query_id, engineIP=route.engine_ip)
rpc_started = time.monotonic()
transport = asyncio.create_task(self._connection._raw_result_rpc(
'getNextResultBatchV2', request, tuple(metadata), deadline))
stream.set_transport(transport)
try:
response = await asyncio.shield(transport)
finally:
if transport.done():
stream.clear_transport(transport)
stream.push(response, time.monotonic() - rpc_started)
if stream.retired or response.endOfStream:
return
if any(response.resultBatches):
no_progress_deadline = time.monotonic() + budget
backoff = .01
else:
await asyncio.sleep(min(backoff, self._connection._remaining(no_progress_deadline)))
backoff = min(backoff * 2, .1)
except asyncio.CancelledError:
stream.retire()
raise
except Exception as error:
stream.fail(error, 0.0 if rpc_started is None else time.monotonic() - rpc_started)
finally:
stream.finish()
async def _next_v2_batch(self, deadline):
from .async_work import reserve_work
batch = self._result_batches.pop()
if batch is not None:
return batch
if self._result_batches.finished or self._state == 'EXHAUSTED':
self._state = 'EXHAUSTED'
return None
consumed = self._pending_result is not None
backoff = .01
try:
await self._connection._ensure_decoder(deadline)
if self._columns is None:
await self._refresh_metadata(deadline)
while True:
consumed = consumed or self._pending_result is not None
reservation = None
revision = self._revision
protocol = self._result_protocol
previously_consumed = consumed
dispatched = False
response = None
status = 'error'
rpc_seconds = decode_seconds = 0.0
started = time.monotonic()
prefetched = self._pending_result is not None
transport_record = None
transport_status = 'error'
def mark_dispatch():
nonlocal consumed, dispatched
consumed = dispatched = True
try:
try:
if self._pending_result is None:
await self._dispatch_result(deadline)
transport_record = self._pending_result[0]
mark_dispatch()
response, rpc_seconds = await self._take_pending_result(deadline)
transport_status = 'ok'
except grpc.RpcError as error:
if protocol == 'v2' and error.code() == grpc.StatusCode.UNIMPLEMENTED:
consumed = previously_consumed
self._retire_prefetch()
# Settle V2 before V1 can advance the same query session.
async with asyncio.timeout_at(deadline):
await self._settle_retired_results()
self._publish(revision)
self._result_protocol = 'v1'
status = 'unimplemented'
logger.debug('Result batch compatibility fallback.',
extra={'result_batch_fallback': True})
continue
raise
finally:
wait_seconds = time.monotonic() - started
rpc_seconds = (transport_record.elapsed
if transport_record is not None and (
isinstance(transport_record, ContinuousResultStream)
or transport_record.settled)
else wait_seconds)
if prefetched:
logger.debug('Result prefetched RPC seconds=%.6f wait_seconds=%.6f status=%s',
rpc_seconds, wait_seconds, transport_status, extra={
'result_batch_prefetched_rpc_seconds': rpc_seconds,
'result_batch_prefetch_wait_seconds': wait_seconds,
'result_batch_prefetch_status': transport_status})
decode_started = time.monotonic()
try:
if protocol == 'v2':
await self._start_prefetch(response, deadline)
reservation = await reserve_work(deadline=deadline)
await self._decode_v2_response(response, reservation, revision, deadline)
else:
self._retain_result_session(response, revision)
if response.resultBatch:
reservation = await reserve_work(deadline=deadline)
batch = await self._decode_batch(response.resultBatch, reservation, revision)
else:
self._state = 'EXHAUSTED'
batch = None
finally:
decode_seconds = time.monotonic() - decode_started
status = 'ok'
if protocol == 'v1':
return batch
batch = self._result_batches.pop()
if batch is not None:
return batch
if self._result_batches.finished:
self._state = 'EXHAUSTED'
return None
finally:
if reservation is not None:
reservation.release()
if dispatched:
_log_result_batch(protocol, rpc_seconds, status,
response if status == 'ok' else None,
decode_seconds if status == 'ok' else 0.0)
if not isinstance(transport_record, ContinuousResultStream):
await asyncio.sleep(min(backoff, self._connection._remaining(deadline)))
backoff = min(backoff * 2, .1)
except asyncio.CancelledError:
if consumed:
self._fail_result('ambiguous_result')
raise
except grpc.RpcError:
if consumed:
self._fail_result('ambiguous_result')
raise
except IncompleteResultError:
raise
except Exception as error:
if consumed:
raise self._fail_result('ambiguous_result') from error
raise
async def _fetchmany(self, size, deadline):
while len(self._rows) < size and self._state != 'EXHAUSTED':
rows = await self._next_batch(deadline)
if rows is None:
break
try:
self._accept_rows(rows)
except Exception as error:
raise self._fail_result('aggregation_failed') from error
return self._take_rows(size)
async def fetchmany(self, size=None, *, timeout=None):
with self._operation(fetch=True):
size = self.arraysize if size is None else size
if isinstance(size, bool) or not isinstance(size, int) or size < 0:
raise ProgrammingError('Fetch size must be a nonnegative integer.')
try:
return await self._fetchmany(size, self._connection._deadline(timeout))
except (asyncio.CancelledError, grpc.RpcError, OperationalError, ProgrammingError, OAuthError):
raise
except Exception as error:
raise _operational(error) from error
async def fetchone(self, *, timeout=None):
rows = await self.fetchmany(1, timeout=timeout)
return rows or None
async def fetch_batch(self, *, timeout=None):
with self._operation(fetch=True):
if self._rows:
return self._take_rows(len(self._rows))
try:
rows = await self._next_batch(self._connection._deadline(timeout))
if rows is not None:
self._rownumber += len(rows)
return rows
except (asyncio.CancelledError, grpc.RpcError, OperationalError, ProgrammingError, OAuthError):
raise
except Exception as error:
raise _operational(error) from error
async def fetchall(self, *, timeout=None):
with self._operation(fetch=True):
deadline = self._connection._deadline(timeout)
rows = []
try:
while True:
if self._rows:
rows.extend(self._take_rows(len(self._rows)))
batch = await self._next_batch(deadline)
if batch is None:
return rows
rows.extend(batch)
self._rownumber += len(batch)
except grpc.RpcError:
raise
except asyncio.CancelledError:
if rows:
self._fail_result('aggregation_failed')
raise
except Exception as error:
if rows or self._failure is not None:
raise self._fail_result('aggregation_failed') from error
raise _operational(error) from error
async def fetchall_buffer(self, query_id=None, *, timeout=None):
if query_id is not None and query_id != self.query_id:
raise ProgrammingError('Query handle does not belong to this cursor.')
while True:
rows = await self.fetch_batch(timeout=timeout)
if rows is None:
return
yield rows
def __aiter__(self):
return self
async def __anext__(self):
row = await self.fetchone()
if row is None:
raise StopAsyncIteration
return row
async def update_mete_data(self, *, timeout=None):
with self._operation():
try:
await self._refresh_metadata(self._connection._deadline(timeout))
except (asyncio.CancelledError, OperationalError, ProgrammingError, OAuthError):
raise
except Exception as error:
raise _operational(error) from error
refresh_metadata = update_mete_data
async def get_rowcount(self, *, timeout=None):
await self.update_mete_data(timeout=timeout)
return self.rowcount
async def get_description(self, *, timeout=None):
await self.update_mete_data(timeout=timeout)
return self.description
async def get_rpc_metadata(self, *, timeout=None):
with self._operation():
return await self._connection._metadata(self._connection._deadline(timeout), self._route)
async def _read_query(self, name, cls, timeout, query_id=None, safe=False):
with self._operation():
if query_id is not None and query_id != self.query_id:
raise ProgrammingError('Query handle does not belong to this cursor.')
deadline = self._connection._deadline(timeout)
try:
request = await self._query_request(cls, deadline)
return await self._call(name, request, deadline, safe=safe)
except (asyncio.CancelledError, OperationalError, ProgrammingError, OAuthError):
raise
except Exception as error:
raise _operational(error) from error
async def status(self, query_id=None, *, timeout=None):
return await self._read_query('status', pb.StatusRequest, timeout, query_id, safe=True)
async def explain(self, *, timeout=None):
return (await self._read_query('explain', pb.ExplainRequest, timeout)).explain
async def explain_analyse(self, *, timeout=None):
response = await self._read_query('explainAnalyze', pb.ExplainAnalyzeRequest, timeout)
return dict(is_cached=response.isCached, parsing_time=response.parsingTime,
queuing_time=response.queueingTime, planner=response.explainAnalyze)
async def get_tables(self, *, timeout=None):
with self._operation():
return await self._connection.get_tables(self._catalog, self._database, timeout=timeout)
async def get_columns(self, table, *, timeout=None):
with self._operation():
return await self._connection.get_columns(self._catalog, self._database, table, timeout=timeout)
async def get_schema_names(self, *, timeout=None):
with self._operation():
return await self._connection.get_schema_names(self._catalog, timeout=timeout)
async def _stop_active(self):
self._cancel_decode()
self._retire_prefetch()
self._revision += 1
task = self._operation_task or self._active_call
if task is not None and task is not asyncio.current_task():
task.cancel()
await asyncio.gather(task, return_exceptions=True)
await self._settle_retired_results()
async def _settle_retired_results(self):
for pending in tuple(self._retired_results):
handle = pending.handle
if handle is not None and not pending.settled:
await asyncio.gather(asyncio.shield(handle), return_exceptions=True)
while not pending.settled:
await asyncio.sleep(.005)
if not self._connection._oauth_enabled and pending.session_id:
self._result_session_id = pending.session_id
self._retired_results.remove(pending)
async def _clear(self, deadline, owned=False):
await self._stop_active()
if self._route is None:
if self._state == 'SUBMISSION_UNKNOWN' or self in self._connection._ambiguous_submissions:
raise AmbiguousSubmissionError()
self._reset_query_state()
return None
request = await self._query_request(pb.ClearOrCancelQueryRequest, deadline, owned=owned)
response = await self._connection._rpc('clearOrCancelQuery', request, deadline=deadline,
route=self._route, _cleanup=owned)
self._connection._release_route(self.query_id)
self._connection._ambiguous_submissions.discard(self)
self._reset_query_state()
return response
def _reset_query_state(self):
self._route = None
self._rows.clear()
self._reset_result_protocol()
self._columns = self._description = None
self._rowcount, self._rownumber = -1, 0
self._failure = self._cleanup_error = None
if self._state != 'CLOSED':
self._state = 'EMPTY'
async def clear(self, query_id=None, *, timeout=None):
self._check()
if query_id is not None and query_id != self.query_id:
raise ProgrammingError('Query handle does not belong to this cursor.')
deadline = self._connection._deadline(self._connection.cleanup_timeout if timeout is None else min(timeout, self._connection.cleanup_timeout))
try:
async with asyncio.timeout_at(deadline):
return await self._clear(deadline)
except (asyncio.CancelledError, OperationalError, ProgrammingError, OAuthError):
raise
except Exception as error:
raise _operational(error) from error
async def cancel(self, query_id=None, *, timeout=None):
self._check()
if query_id is not None and query_id != self.query_id:
raise ProgrammingError('Query handle does not belong to this cursor.')
if self._route is None:
raise ProgrammingError('No query handle is known.')
deadline = self._connection._deadline(timeout)
try:
async with asyncio.timeout_at(deadline):
await self._stop_active()
if self._result_batch_v2:
self._fail_result('cancelled_result')
request = await self._query_request(pb.CancelQueryRequest, deadline)
await self._connection._rpc('cancelQuery', request, deadline=deadline, route=self._route)
self._fail_result('cancelled_result')
except (asyncio.CancelledError, OperationalError, ProgrammingError, OAuthError):
raise
except Exception as error:
raise _operational(error) from error
async def _close_owned(self, deadline):
if self._state == 'CLOSED' and self._route is None and not self._retired_results:
return
unknown_without_handle = self._state == 'SUBMISSION_UNKNOWN' and self._route is None
if unknown_without_handle:
self._connection._ambiguous_submissions.add(self)
self._state = 'CLOSED'
try:
async with asyncio.timeout_at(deadline):
await self._stop_active()
if self._route is None and self in self._connection._ambiguous_submissions:
raise AmbiguousSubmissionError()
await self._clear(deadline, owned=True)
except asyncio.CancelledError:
self._cleanup_error = OperationalError('Query cleanup is unconfirmed.')
raise
except Exception:
self._cleanup_error = OperationalError('Query cleanup is unconfirmed.')
finally:
self._rows.clear()
self._result_batches.clear()
if self._route is None and not self._retired_results:
self._result_session_id = None
self._connection._cursors.discard(self)
async def close(self):
if self._lease_guard is not None:
self._lease_guard(internal=False)
self._connection._check_owner(_cleanup=True)
if self._state == 'CLOSED':
return
self._connection._check_owner()
deadline = self._connection._deadline(self._connection.cleanup_timeout)
if self._close_task is None:
self._close_task = asyncio.create_task(self._close_owned(deadline))
try:
await asyncio.shield(self._close_task)
except asyncio.CancelledError:
await asyncio.shield(self._close_task)
raise
async def __aenter__(self):
self._check()
return self
async def __aexit__(self, *exc):
await self.close()
def setinputsizes(self, sizes):
return None
def setoutputsize(self, size, column=None):
return None
async def poll(self, *args, **kwargs):
raise NotSupportedError('Query polling is not implemented.')
async def fetch_logs(self, *args, **kwargs):
raise NotSupportedError('Query log retrieval is not implemented.')