forked from graphite-project/whisper
-
Notifications
You must be signed in to change notification settings - Fork 0
/
whisper.py
941 lines (757 loc) · 31.1 KB
/
whisper.py
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
# Copyright 2009-Present The Graphite Development Team
# Copyright 2008 Orbitz WorldWide
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
#
# This module is an implementation of the Whisper database API
# Here is the basic layout of a whisper data file
#
# File = Header,Data
# Header = Metadata,ArchiveInfo+
# Metadata = aggregationType,maxRetention,xFilesFactor,archiveCount
# ArchiveInfo = Offset,SecondsPerPoint,Points
# Data = Archive+
# Archive = Point+
# Point = timestamp,value
import itertools
import operator
import os
import re
import struct
import sys
import time
izip = getattr(itertools, 'izip', zip)
ifilter = getattr(itertools, 'ifilter', filter)
if sys.version_info >= (3, 0):
xrange = range
try:
import fcntl
CAN_LOCK = True
except ImportError:
CAN_LOCK = False
try:
import ctypes
import ctypes.util
CAN_FALLOCATE = True
except ImportError:
CAN_FALLOCATE = False
fallocate = None
if CAN_FALLOCATE:
libc_name = ctypes.util.find_library('c')
libc = ctypes.CDLL(libc_name)
c_off64_t = ctypes.c_int64
c_off_t = ctypes.c_int
try:
_fallocate = libc.posix_fallocate64
_fallocate.restype = ctypes.c_int
_fallocate.argtypes = [ctypes.c_int, c_off64_t, c_off64_t]
except AttributeError:
try:
_fallocate = libc.posix_fallocate
_fallocate.restype = ctypes.c_int
_fallocate.argtypes = [ctypes.c_int, c_off_t, c_off_t]
except AttributeError:
CAN_FALLOCATE = False
if CAN_FALLOCATE:
def _py_fallocate(fd, offset, len_):
res = _fallocate(fd.fileno(), offset, len_)
if res != 0:
raise IOError(res, 'fallocate')
fallocate = _py_fallocate
del libc
del libc_name
LOCK = True and CAN_LOCK
CACHE_HEADERS = False
AUTOFLUSH = False
__headerCache = {}
longFormat = "!L"
longSize = struct.calcsize(longFormat)
floatFormat = "!f"
floatSize = struct.calcsize(floatFormat)
valueFormat = "!d"
valueSize = struct.calcsize(valueFormat)
pointFormat = "!Ld"
pointSize = struct.calcsize(pointFormat)
metadataFormat = "!2LfL"
metadataSize = struct.calcsize(metadataFormat)
archiveInfoFormat = "!3L"
archiveInfoSize = struct.calcsize(archiveInfoFormat)
aggregationTypeToMethod = dict({
1: 'average',
2: 'sum',
3: 'last',
4: 'max',
5: 'min',
6: 'avg_zero'
})
aggregationMethodToType = dict([[v, k] for k, v in aggregationTypeToMethod.items()])
aggregationMethods = aggregationTypeToMethod.values()
debug = startBlock = endBlock = lambda *a, **k: None
UnitMultipliers = {
'seconds': 1,
'minutes': 60,
'hours': 3600,
'days': 86400,
'weeks': 86400 * 7,
'years': 86400 * 365
}
def getUnitString(s):
for value in ('seconds', 'minutes', 'hours', 'days', 'weeks', 'years'):
if value.startswith(s):
return value
raise ValueError("Invalid unit '%s'" % s)
def parseRetentionDef(retentionDef):
(precision, points) = retentionDef.strip().split(':', 1)
if precision.isdigit():
precision = int(precision) * UnitMultipliers[getUnitString('s')]
else:
precision_re = re.compile(r'^(\d+)([a-z]+)$')
match = precision_re.match(precision)
if match:
precision = int(match.group(1)) * UnitMultipliers[getUnitString(match.group(2))]
else:
raise ValueError("Invalid precision specification '%s'" % precision)
if points.isdigit():
points = int(points)
else:
points_re = re.compile(r'^(\d+)([a-z]+)$')
match = points_re.match(points)
if match:
points = int(match.group(1)) * UnitMultipliers[getUnitString(match.group(2))] // precision
else:
raise ValueError("Invalid retention specification '%s'" % points)
return (precision, points)
class WhisperException(Exception):
"""Base class for whisper exceptions."""
class InvalidConfiguration(WhisperException):
"""Invalid configuration."""
class InvalidAggregationMethod(WhisperException):
"""Invalid aggregation method."""
class InvalidTimeInterval(WhisperException):
"""Invalid time interval."""
class TimestampNotCovered(WhisperException):
"""Timestamp not covered by any archives in this database."""
class CorruptWhisperFile(WhisperException):
def __init__(self, error, path):
Exception.__init__(self, error)
self.error = error
self.path = path
def __repr__(self):
return "<CorruptWhisperFile[%s] %s>" % (self.path, self.error)
def __str__(self):
return "%s (%s)" % (self.error, self.path)
def enableDebug():
global open, debug, startBlock, endBlock
class open(file):
def __init__(self, *args, **kwargs):
file.__init__(self, *args, **kwargs)
self.writeCount = 0
self.readCount = 0
def write(self, data):
self.writeCount += 1
debug('WRITE %d bytes #%d' % (len(data), self.writeCount))
return file.write(self, data)
def read(self, bytes):
self.readCount += 1
debug('READ %d bytes #%d' % (bytes, self.readCount))
return file.read(self, bytes)
def debug(message):
print('DEBUG :: %s' % message)
__timingBlocks = {}
def startBlock(name):
__timingBlocks[name] = time.time()
def endBlock(name):
debug("%s took %.5f seconds" % (name, time.time() - __timingBlocks.pop(name)))
def __readHeader(fh):
if CACHE_HEADERS:
info = __headerCache.get(fh.name)
if info:
return info
originalOffset = fh.tell()
fh.seek(0)
packedMetadata = fh.read(metadataSize)
try:
(aggregationType, maxRetention, xff, archiveCount) = struct.unpack(metadataFormat, packedMetadata)
except:
raise CorruptWhisperFile("Unable to read header", fh.name)
archives = []
for i in xrange(archiveCount):
packedArchiveInfo = fh.read(archiveInfoSize)
try:
(offset, secondsPerPoint, points) = struct.unpack(archiveInfoFormat, packedArchiveInfo)
except (struct.error, ValueError, TypeError):
raise CorruptWhisperFile("Unable to read archive%d metadata" % i, fh.name)
archiveInfo = {
'offset': offset,
'secondsPerPoint': secondsPerPoint,
'points': points,
'retention': secondsPerPoint * points,
'size': points * pointSize,
}
archives.append(archiveInfo)
fh.seek(originalOffset)
info = {
'aggregationMethod': aggregationTypeToMethod.get(aggregationType, 'average'),
'maxRetention': maxRetention,
'xFilesFactor': xff,
'archives': archives,
}
if CACHE_HEADERS:
__headerCache[fh.name] = info
return info
def setAggregationMethod(path, aggregationMethod, xFilesFactor=None):
"""setAggregationMethod(path,aggregationMethod,xFilesFactor=None)
path is a string
aggregationMethod specifies the method to use when propagating data (see ``whisper.aggregationMethods``)
xFilesFactor specifies the fraction of data points in a propagation interval that must have known values for a propagation to occur. If None, the existing xFilesFactor in path will not be changed
"""
with open(path, 'r+b') as fh:
if LOCK:
fcntl.flock(fh.fileno(), fcntl.LOCK_EX)
packedMetadata = fh.read(metadataSize)
try:
(aggregationType, maxRetention, xff, archiveCount) = struct.unpack(metadataFormat, packedMetadata)
except (struct.error, ValueError):
raise CorruptWhisperFile("Unable to read header", fh.name)
try:
newAggregationType = struct.pack(longFormat, aggregationMethodToType[aggregationMethod])
except KeyError:
raise InvalidAggregationMethod("Unrecognized aggregation method: %s" %
aggregationMethod)
if xFilesFactor is not None:
# Use specified xFilesFactor
xff = struct.pack(floatFormat, float(xFilesFactor))
else:
# Retain old value
xff = struct.pack(floatFormat, xff)
# Repack the remaining header information
maxRetention = struct.pack(longFormat, maxRetention)
archiveCount = struct.pack(longFormat, archiveCount)
packedMetadata = newAggregationType + maxRetention + xff + archiveCount
fh.seek(0)
# fh.write(newAggregationType)
fh.write(packedMetadata)
if AUTOFLUSH:
fh.flush()
os.fsync(fh.fileno())
if CACHE_HEADERS and fh.name in __headerCache:
del __headerCache[fh.name]
return aggregationTypeToMethod.get(aggregationType, 'average')
def validateArchiveList(archiveList):
""" Validates an archiveList.
An ArchiveList must:
1. Have at least one archive config. Example: (60, 86400)
2. No archive may be a duplicate of another.
3. Higher precision archives' precision must evenly divide all lower precision archives' precision.
4. Lower precision archives must cover larger time intervals than higher precision archives.
5. Each archive must have at least enough points to consolidate to the next archive
Returns True or False
"""
if not archiveList:
raise InvalidConfiguration("You must specify at least one archive configuration!")
archiveList.sort(key=lambda a: a[0]) # Sort by precision (secondsPerPoint)
for i, archive in enumerate(archiveList):
if i == len(archiveList) - 1:
break
nextArchive = archiveList[i+1]
if not archive[0] < nextArchive[0]:
raise InvalidConfiguration("A Whisper database may not be configured having "
"two archives with the same precision (archive%d: %s, archive%d: %s)" %
(i, archive, i + 1, nextArchive))
if nextArchive[0] % archive[0] != 0:
raise InvalidConfiguration("Higher precision archives' precision "
"must evenly divide all lower precision archives' precision "
"(archive%d: %s, archive%d: %s)" %
(i, archive[0], i + 1, nextArchive[0]))
retention = archive[0] * archive[1]
nextRetention = nextArchive[0] * nextArchive[1]
if not nextRetention > retention:
raise InvalidConfiguration("Lower precision archives must cover "
"larger time intervals than higher precision archives "
"(archive%d: %s seconds, archive%d: %s seconds)" %
(i, retention, i + 1, nextRetention))
archivePoints = archive[1]
pointsPerConsolidation = nextArchive[0] // archive[0]
if not archivePoints >= pointsPerConsolidation:
raise InvalidConfiguration("Each archive must have at least enough points "
"to consolidate to the next archive (archive%d consolidates %d of "
"archive%d's points but it has only %d total points)" %
(i + 1, pointsPerConsolidation, i, archivePoints))
def create(path, archiveList, xFilesFactor=None, aggregationMethod=None, sparse=False, useFallocate=False):
"""create(path,archiveList,xFilesFactor=0.5,aggregationMethod='average')
path is a string
archiveList is a list of archives, each of which is of the form (secondsPerPoint,numberOfPoints)
xFilesFactor specifies the fraction of data points in a propagation interval that must have known values for a propagation to occur
aggregationMethod specifies the function to use when propagating data (see ``whisper.aggregationMethods``)
"""
# Set default params
if xFilesFactor is None:
xFilesFactor = 0.5
if aggregationMethod is None:
aggregationMethod = 'average'
# Validate archive configurations...
validateArchiveList(archiveList)
# Looks good, now we create the file and write the header
if os.path.exists(path):
raise InvalidConfiguration("File %s already exists!" % path)
with open(path, 'wb') as fh:
try:
if LOCK:
fcntl.flock(fh.fileno(), fcntl.LOCK_EX)
aggregationType = struct.pack(longFormat, aggregationMethodToType.get(aggregationMethod, 1))
oldest = max([secondsPerPoint * points for secondsPerPoint, points in archiveList])
maxRetention = struct.pack(longFormat, oldest)
xFilesFactor = struct.pack(floatFormat, float(xFilesFactor))
archiveCount = struct.pack(longFormat, len(archiveList))
packedMetadata = aggregationType + maxRetention + xFilesFactor + archiveCount
fh.write(packedMetadata)
headerSize = metadataSize + (archiveInfoSize * len(archiveList))
archiveOffsetPointer = headerSize
for secondsPerPoint, points in archiveList:
archiveInfo = struct.pack(archiveInfoFormat, archiveOffsetPointer, secondsPerPoint, points)
fh.write(archiveInfo)
archiveOffsetPointer += (points * pointSize)
# If configured to use fallocate and capable of fallocate use that, else
# attempt sparse if configure or zero pre-allocate if sparse isn't configured.
if CAN_FALLOCATE and useFallocate:
remaining = archiveOffsetPointer - headerSize
fallocate(fh, headerSize, remaining)
elif sparse:
fh.seek(archiveOffsetPointer - 1)
fh.write(b'\x00')
else:
remaining = archiveOffsetPointer - headerSize
chunksize = 16384
zeroes = b'\x00' * chunksize
while remaining > chunksize:
fh.write(zeroes)
remaining -= chunksize
fh.write(zeroes[:remaining])
if AUTOFLUSH:
fh.flush()
os.fsync(fh.fileno())
# Explicitly close the file to catch IOError on close()
fh.close()
except IOError:
# if we got an IOError above, the file is either empty or half created.
# Better off deleting it to avoid surprises later
os.unlink(fh.name)
raise
def aggregate(aggregationMethod, knownValues, neighborValues=None):
if aggregationMethod == 'average':
return float(sum(knownValues)) / float(len(knownValues))
elif aggregationMethod == 'sum':
return float(sum(knownValues))
elif aggregationMethod == 'last':
return knownValues[len(knownValues)-1]
elif aggregationMethod == 'max':
return max(knownValues)
elif aggregationMethod == 'min':
return min(knownValues)
elif aggregationMethod == 'avg_zero':
if not neighborValues:
raise InvalidAggregationMethod("Using avg_zero without neighborValues")
values = [x or 0 for x in neighborValues]
return float(sum(values)) / float(len(values))
else:
raise InvalidAggregationMethod("Unrecognized aggregation method %s" %
aggregationMethod)
def __propagate(fh, header, timestamp, higher, lower):
aggregationMethod = header['aggregationMethod']
xff = header['xFilesFactor']
lowerIntervalStart = timestamp - (timestamp % lower['secondsPerPoint'])
lowerIntervalEnd = lowerIntervalStart + lower['secondsPerPoint']
fh.seek(higher['offset'])
packedPoint = fh.read(pointSize)
(higherBaseInterval, higherBaseValue) = struct.unpack(pointFormat, packedPoint)
if higherBaseInterval == 0:
higherFirstOffset = higher['offset']
else:
timeDistance = lowerIntervalStart - higherBaseInterval
pointDistance = timeDistance // higher['secondsPerPoint']
byteDistance = pointDistance * pointSize
higherFirstOffset = higher['offset'] + (byteDistance % higher['size'])
higherPoints = lower['secondsPerPoint'] // higher['secondsPerPoint']
higherSize = higherPoints * pointSize
relativeFirstOffset = higherFirstOffset - higher['offset']
relativeLastOffset = (relativeFirstOffset + higherSize) % higher['size']
higherLastOffset = relativeLastOffset + higher['offset']
fh.seek(higherFirstOffset)
if higherFirstOffset < higherLastOffset: # We don't wrap the archive
seriesString = fh.read(higherLastOffset - higherFirstOffset)
else: # We do wrap the archive
higherEnd = higher['offset'] + higher['size']
seriesString = fh.read(higherEnd - higherFirstOffset)
fh.seek(higher['offset'])
seriesString += fh.read(higherLastOffset - higher['offset'])
# Now we unpack the series data we just read
byteOrder, pointTypes = pointFormat[0], pointFormat[1:]
points = len(seriesString) // pointSize
seriesFormat = byteOrder + (pointTypes * points)
unpackedSeries = struct.unpack(seriesFormat, seriesString)
# And finally we construct a list of values
neighborValues = [None] * points
currentInterval = lowerIntervalStart
step = higher['secondsPerPoint']
for i in xrange(0, len(unpackedSeries), 2):
pointTime = unpackedSeries[i]
if pointTime == currentInterval:
neighborValues[i//2] = unpackedSeries[i+1]
currentInterval += step
# Propagate aggregateValue to propagate from neighborValues if we have enough known points
knownValues = [v for v in neighborValues if v is not None]
if not knownValues:
return False
knownPercent = float(len(knownValues)) / float(len(neighborValues))
if knownPercent >= xff: # We have enough data to propagate a value!
aggregateValue = aggregate(aggregationMethod, knownValues, neighborValues)
myPackedPoint = struct.pack(pointFormat, lowerIntervalStart, aggregateValue)
fh.seek(lower['offset'])
packedPoint = fh.read(pointSize)
(lowerBaseInterval, lowerBaseValue) = struct.unpack(pointFormat, packedPoint)
if lowerBaseInterval == 0: # First propagated update to this lower archive
fh.seek(lower['offset'])
fh.write(myPackedPoint)
else: # Not our first propagated update to this lower archive
timeDistance = lowerIntervalStart - lowerBaseInterval
pointDistance = timeDistance // lower['secondsPerPoint']
byteDistance = pointDistance * pointSize
lowerOffset = lower['offset'] + (byteDistance % lower['size'])
fh.seek(lowerOffset)
fh.write(myPackedPoint)
return True
else:
return False
def update(path, value, timestamp=None):
"""
update(path, value, timestamp=None)
path is a string
value is a float
timestamp is either an int or float
"""
value = float(value)
with open(path, 'r+b') as fh:
return file_update(fh, value, timestamp)
def file_update(fh, value, timestamp):
if LOCK:
fcntl.flock(fh.fileno(), fcntl.LOCK_EX)
header = __readHeader(fh)
now = int(time.time())
if timestamp is None:
timestamp = now
timestamp = int(timestamp)
diff = now - timestamp
if not ((diff < header['maxRetention']) and diff >= 0):
raise TimestampNotCovered("Timestamp not covered by any archives in "
"this database.")
for i, archive in enumerate(header['archives']): # Find the highest-precision archive that covers timestamp
if archive['retention'] < diff: continue
lowerArchives = header['archives'][i+1:] # We'll pass on the update to these lower precision archives later
break
# First we update the highest-precision archive
myInterval = timestamp - (timestamp % archive['secondsPerPoint'])
myPackedPoint = struct.pack(pointFormat, myInterval, value)
fh.seek(archive['offset'])
packedPoint = fh.read(pointSize)
(baseInterval, baseValue) = struct.unpack(pointFormat, packedPoint)
if baseInterval == 0: # This file's first update
fh.seek(archive['offset'])
fh.write(myPackedPoint)
baseInterval, baseValue = myInterval, value
else: # Not our first update
timeDistance = myInterval - baseInterval
pointDistance = timeDistance // archive['secondsPerPoint']
byteDistance = pointDistance * pointSize
myOffset = archive['offset'] + (byteDistance % archive['size'])
fh.seek(myOffset)
fh.write(myPackedPoint)
# Now we propagate the update to lower-precision archives
higher = archive
for lower in lowerArchives:
if not __propagate(fh, header, myInterval, higher, lower):
break
higher = lower
if AUTOFLUSH:
fh.flush()
os.fsync(fh.fileno())
def update_many(path, points):
"""update_many(path,points)
path is a string
points is a list of (timestamp,value) points
"""
if not points: return
points = [(int(t), float(v)) for (t, v) in points]
points.sort(key=lambda p: p[0], reverse=True) # Order points by timestamp, newest first
with open(path, 'r+b') as fh:
return file_update_many(fh, points)
def file_update_many(fh, points):
if LOCK:
fcntl.flock(fh.fileno(), fcntl.LOCK_EX)
header = __readHeader(fh)
now = int(time.time())
archives = iter(header['archives'])
currentArchive = next(archives)
currentPoints = []
for point in points:
age = now - point[0]
while currentArchive['retention'] < age: # We can't fit any more points in this archive
if currentPoints: # Commit all the points we've found that it can fit
currentPoints.reverse() # Put points in chronological order
__archive_update_many(fh, header, currentArchive, currentPoints)
currentPoints = []
try:
currentArchive = next(archives)
except StopIteration:
currentArchive = None
break
if not currentArchive:
break # Drop remaining points that don't fit in the database
currentPoints.append(point)
if currentArchive and currentPoints: # Don't forget to commit after we've checked all the archives
currentPoints.reverse()
__archive_update_many(fh, header, currentArchive, currentPoints)
if AUTOFLUSH:
fh.flush()
os.fsync(fh.fileno())
def __archive_update_many(fh, header, archive, points):
step = archive['secondsPerPoint']
alignedPoints = [(timestamp - (timestamp % step), value)
for (timestamp, value) in points]
# Create a packed string for each contiguous sequence of points
packedStrings = []
previousInterval = None
currentString = b""
lenAlignedPoints = len(alignedPoints)
for i in xrange(0, lenAlignedPoints):
# Take last point in run of points with duplicate intervals
if i+1 < lenAlignedPoints and alignedPoints[i][0] == alignedPoints[i+1][0]:
continue
(interval, value) = alignedPoints[i]
if (not previousInterval) or (interval == previousInterval + step):
currentString += struct.pack(pointFormat, interval, value)
previousInterval = interval
else:
numberOfPoints = len(currentString) // pointSize
startInterval = previousInterval - (step * (numberOfPoints-1))
packedStrings.append((startInterval, currentString))
currentString = struct.pack(pointFormat, interval, value)
previousInterval = interval
if currentString:
numberOfPoints = len(currentString) // pointSize
startInterval = previousInterval - (step * (numberOfPoints-1))
packedStrings.append((startInterval, currentString))
# Read base point and determine where our writes will start
fh.seek(archive['offset'])
packedBasePoint = fh.read(pointSize)
(baseInterval, baseValue) = struct.unpack(pointFormat, packedBasePoint)
if baseInterval == 0: # This file's first update
baseInterval = packedStrings[0][0] # Use our first string as the base, so we start at the start
# Write all of our packed strings in locations determined by the baseInterval
for (interval, packedString) in packedStrings:
timeDistance = interval - baseInterval
pointDistance = timeDistance // step
byteDistance = pointDistance * pointSize
myOffset = archive['offset'] + (byteDistance % archive['size'])
fh.seek(myOffset)
archiveEnd = archive['offset'] + archive['size']
bytesBeyond = (myOffset + len(packedString)) - archiveEnd
if bytesBeyond > 0:
fh.write(packedString[:-bytesBeyond])
assert fh.tell() == archiveEnd, "archiveEnd=%d fh.tell=%d bytesBeyond=%d len(packedString)=%d" % (archiveEnd, fh.tell(), bytesBeyond, len(packedString))
fh.seek(archive['offset'])
fh.write(packedString[-bytesBeyond:]) # Safe because it can't exceed the archive (retention checking logic above)
else:
fh.write(packedString)
# Now we propagate the updates to lower-precision archives
higher = archive
lowerArchives = [arc for arc in header['archives'] if arc['secondsPerPoint'] > archive['secondsPerPoint']]
for lower in lowerArchives:
fit = lambda i: i - (i % lower['secondsPerPoint'])
lowerIntervals = [fit(p[0]) for p in alignedPoints]
uniqueLowerIntervals = set(lowerIntervals)
propagateFurther = False
for interval in uniqueLowerIntervals:
if __propagate(fh, header, interval, higher, lower):
propagateFurther = True
if not propagateFurther:
break
higher = lower
def info(path):
"""
info(path)
path is a string
"""
try:
with open(path, 'rb') as fh:
return __readHeader(fh)
except (IOError, OSError):
pass
return None
def fetch(path, fromTime, untilTime=None, now=None):
"""fetch(path,fromTime,untilTime=None)
path is a string
fromTime is an epoch time
untilTime is also an epoch time, but defaults to now.
Returns a tuple of (timeInfo, valueList)
where timeInfo is itself a tuple of (fromTime, untilTime, step)
Returns None if no data can be returned
"""
with open(path, 'rb') as fh:
return file_fetch(fh, fromTime, untilTime, now)
def file_fetch(fh, fromTime, untilTime, now=None):
header = __readHeader(fh)
if now is None:
now = int(time.time())
if untilTime is None:
untilTime = now
fromTime = int(fromTime)
untilTime = int(untilTime)
# Here we try and be flexible and return as much data as we can.
# If the range of data is from too far in the past or fully in the future, we
# return nothing
if fromTime > untilTime:
raise InvalidTimeInterval("Invalid time interval: from time '%s' is after until time '%s'" % (fromTime, untilTime))
oldestTime = now - header['maxRetention']
# Range is in the future
if fromTime > now:
return None
# Range is beyond retention
if untilTime < oldestTime:
return None
# Range requested is partially beyond retention, adjust
if fromTime < oldestTime:
fromTime = oldestTime
# Range is partially in the future, adjust
if untilTime > now:
untilTime = now
diff = now - fromTime
for archive in header['archives']:
if archive['retention'] >= diff:
break
return __archive_fetch(fh, archive, fromTime, untilTime)
def __archive_fetch(fh, archive, fromTime, untilTime):
"""
Fetch data from a single archive. Note that checks for validity of the time
period requested happen above this level so it's possible to wrap around the
archive on a read and request data older than the archive's retention
"""
fromInterval = int(fromTime - (fromTime % archive['secondsPerPoint'])) + archive['secondsPerPoint']
untilInterval = int(untilTime - (untilTime % archive['secondsPerPoint'])) + archive['secondsPerPoint']
fh.seek(archive['offset'])
packedPoint = fh.read(pointSize)
(baseInterval, baseValue) = struct.unpack(pointFormat, packedPoint)
if baseInterval == 0:
step = archive['secondsPerPoint']
points = (untilInterval - fromInterval) // step
timeInfo = (fromInterval, untilInterval, step)
valueList = [None] * points
return (timeInfo, valueList)
# Determine fromOffset
timeDistance = fromInterval - baseInterval
pointDistance = timeDistance // archive['secondsPerPoint']
byteDistance = pointDistance * pointSize
fromOffset = archive['offset'] + (byteDistance % archive['size'])
# Determine untilOffset
timeDistance = untilInterval - baseInterval
pointDistance = timeDistance // archive['secondsPerPoint']
byteDistance = pointDistance * pointSize
untilOffset = archive['offset'] + (byteDistance % archive['size'])
# Read all the points in the interval
fh.seek(fromOffset)
if fromOffset < untilOffset: # If we don't wrap around the archive
seriesString = fh.read(untilOffset - fromOffset)
else: # We do wrap around the archive, so we need two reads
archiveEnd = archive['offset'] + archive['size']
seriesString = fh.read(archiveEnd - fromOffset)
fh.seek(archive['offset'])
seriesString += fh.read(untilOffset - archive['offset'])
# Now we unpack the series data we just read (anything faster than unpack?)
byteOrder, pointTypes = pointFormat[0], pointFormat[1:]
points = len(seriesString) // pointSize
seriesFormat = byteOrder + (pointTypes * points)
unpackedSeries = struct.unpack(seriesFormat, seriesString)
# And finally we construct a list of values (optimize this!)
valueList = [None] * points # Pre-allocate entire list for speed
currentInterval = fromInterval
step = archive['secondsPerPoint']
for i in xrange(0, len(unpackedSeries), 2):
pointTime = unpackedSeries[i]
if pointTime == currentInterval:
pointValue = unpackedSeries[i+1]
valueList[i//2] = pointValue # In-place reassignment is faster than append()
currentInterval += step
timeInfo = (fromInterval, untilInterval, step)
return (timeInfo, valueList)
def merge(path_from, path_to):
""" Merges the data from one whisper file into another. Each file must have
the same archive configuration
"""
# Python 2.7 will allow the following commented line
# with open(path_from, 'rb') as fh_from, open(path_to, 'rb+') as fh_to:
# But with Python 2.6 we need to use this (I prefer not to introduce
# contextlib.nested just for this):
with open(path_from, 'rb') as fh_from:
with open(path_to, 'rb+') as fh_to:
return file_merge(fh_from, fh_to)
def file_merge(fh_from, fh_to):
headerFrom = __readHeader(fh_from)
headerTo = __readHeader(fh_to)
if headerFrom['archives'] != headerTo['archives']:
raise NotImplementedError("%s and %s archive configurations are unalike. " \
"Resize the input before merging" % (fh_from.name, fh_to.name))
archives = headerFrom['archives']
archives.sort(key=operator.itemgetter('retention'))
now = int(time.time())
untilTime = now
for archive in archives:
fromTime = now - archive['retention']
(timeInfo, values) = __archive_fetch(fh_from, archive, fromTime, untilTime)
(start, end, archive_step) = timeInfo
pointsToWrite = list(ifilter(
lambda points: points[1] is not None,
izip(xrange(start, end, archive_step), values)))
__archive_update_many(fh_to, headerTo, archive, pointsToWrite)
untilTime = fromTime
def diff(path_from, path_to, ignore_empty=False):
""" Compare two whisper databases. Each file must have the same archive configuration """
with open(path_from, 'rb') as fh_from:
with open(path_to, 'rb+') as fh_to:
return file_diff(fh_from, fh_to, ignore_empty)
def file_diff(fh_from, fh_to, ignore_empty=False):
headerFrom = __readHeader(fh_from)
headerTo = __readHeader(fh_to)
if headerFrom['archives'] != headerTo['archives']:
# TODO: Add specific whisper-resize commands to right size things
raise NotImplementedError("%s and %s archive configurations are unalike. " \
"Resize the input before diffing" % (fh_from.name, fh_to.name))
archives = headerFrom['archives']
archives.sort(key=operator.itemgetter('retention'))
archive_diffs = []
now = int(time.time())
untilTime = now
for archive_number, archive in enumerate(archives):
diffs = []
startTime = now - archive['retention']
(fromTimeInfo, fromValues) = __archive_fetch(fh_from, archive, startTime, untilTime)
(toTimeInfo, toValues) = __archive_fetch(fh_to, archive, startTime, untilTime)
(start, end, archive_step) = (min(fromTimeInfo[0], toTimeInfo[0]), max(fromTimeInfo[1], toTimeInfo[1]), min(fromTimeInfo[2], toTimeInfo[2]))
points = map(lambda s: (s * archive_step + start, fromValues[s], toValues[s]), xrange(0, (end - start) // archive_step))
if ignore_empty:
points = [p for p in points if p[1] is not None and p[2] is not None]
else:
points = [p for p in points if p[1] is not None or p[2] is not None]
diffs = [p for p in points if p[1] != p[2]]
archive_diffs.append((archive_number, diffs, points.__len__()))
untilTime = startTime
return archive_diffs