File: whisper.py

package info (click to toggle)
python-whisper 0.9.10-1
  • links: PTS, VCS
  • area: main
  • in suites: wheezy
  • size: 176 kB
  • sloc: python: 911; makefile: 5
file content (776 lines) | stat: -rw-r--r-- 25,766 bytes parent folder | download
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
# 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 os, struct, time, operator, itertools

try:
  import fcntl
  CAN_LOCK = True
except ImportError:
  CAN_LOCK = False

LOCK = False
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'
})
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):
  if 'seconds'.startswith(s): return 'seconds'
  if 'minutes'.startswith(s): return 'minutes'
  if 'hours'.startswith(s): return 'hours'
  if 'days'.startswith(s): return 'days'
  if 'weeks'.startswith(s): return 'weeks'
  if 'years'.startswith(s): return 'years'
  raise ValueError("Invalid unit '%s'" % s)

def parseRetentionDef(retentionDef):
  import re
  (precision, points) = retentionDef.strip().split(':')

  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):
  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:
      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):
  """setAggregationMethod(path,aggregationMethod)

path is a string
aggregationMethod specifies the method to use when propogating data (see ``whisper.aggregationMethods``)
"""
  fh = open(path,'r+b')
  if LOCK:
    fcntl.flock( fh.fileno(), fcntl.LOCK_EX )

  packedMetadata = fh.read(metadataSize)

  try:
    (aggregationType,maxRetention,xff,archiveCount) = struct.unpack(metadataFormat,packedMetadata)
  except:
    raise CorruptWhisperFile("Unable to read header", fh.name)

  try:
    newAggregationType = struct.pack( longFormat, aggregationMethodToType[aggregationMethod] )
  except KeyError:
    raise InvalidAggregationMethod("Unrecognized aggregation method: %s" %
          aggregationMethod)

  fh.seek(0)
  fh.write(newAggregationType)

  if AUTOFLUSH:
    fh.flush()
    os.fsync(fh.fileno())

  if CACHE_HEADERS and fh.name in __headerCache:
    del __headerCache[fh.name]

  fh.close()

  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 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, archive[1], i + 1, nextArchive[1]))

    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):
  """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 propogating 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)

  fh = open(path,'wb')
  if LOCK:
    fcntl.flock( fh.fileno(), fcntl.LOCK_EX )

  aggregationType = struct.pack( longFormat, aggregationMethodToType.get(aggregationMethod, 1) )
  oldest = sorted([secondsPerPoint * points for secondsPerPoint,points in archiveList])[-1]
  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 sparse:
    fh.seek(archiveOffsetPointer - headerSize - 1)
    fh.write("\0")
  else:
    # If not creating the file sparsely, then fill the rest of the file with
    # zeroes.
    remaining = archiveOffsetPointer - headerSize
    chunksize = 16384
    zeroes = '\x00' * chunksize
    while remaining > chunksize:
      fh.write(zeroes)
      remaining -= chunksize
    fh.write(zeroes[:remaining])

  if AUTOFLUSH:
    fh.flush()
    os.fsync(fh.fileno())

  fh.close()

def __aggregate(aggregationMethod, knownValues):
  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)
  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)
    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)
  fh = open(path,'r+b')
  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())

  fh.close()


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
  fh = open(path,'r+b')
  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 = archives.next()
  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 = archives.next()
      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())

  fh.close()


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 = ""
  for (interval,value) in alignedPoints:
    if interval == previousInterval: continue
    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
"""
  fh = open(path,'rb')
  info = __readHeader(fh)
  fh.close()
  return info


def fetch(path,fromTime,untilTime=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
"""
  fh = open(path,'rb')
  return file_fetch(fh, fromTime, untilTime)


def file_fetch(fh, fromTime, untilTime):
  header = __readHeader(fh)
  now = int( time.time() )
  if untilTime is None:
    untilTime = now
  fromTime = int(fromTime)
  untilTime = int(untilTime)

  oldestTime = now - header['maxRetention']
  if fromTime < oldestTime:
    fromTime = oldestTime

  if not (fromTime < untilTime):
    raise InvalidTimeInterval("Invalid time interval")
  if untilTime > now:
    untilTime = now
  if untilTime < fromTime:
    untilTime = now

  diff = now - fromTime
  for archive in header['archives']:
    if archive['retention'] >= diff:
      break

  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

  fh.close()
  timeInfo = (fromInterval,untilInterval,step)
  return (timeInfo,valueList)

def merge(path_from, path_to, step=1<<12):
  headerFrom = info(path_from)

  archives = headerFrom['archives']
  archives.sort(key=operator.itemgetter('retention'), reverse=True)

  # Start from maxRetention of the oldest file, and skip forward at max 'step'
  # points at a time.
  fromTime = int(time.time()) - headerFrom['maxRetention']
  for archive in archives:
    pointsRemaining = archive['points']
    while pointsRemaining:
      pointsToRead = step
      if pointsRemaining < step:
        pointsToRead = pointsRemaining
      pointsRemaining -= pointsToRead
      untilTime = fromTime + (pointsToRead * archive['secondsPerPoint'])
      (timeInfo, values) = fetch(path_from, fromTime, untilTime)
      (start, end, archive_step) = timeInfo
      pointsToWrite = list(itertools.ifilter(
        lambda points: points[1] is not None,
        itertools.izip(xrange(start, end, archive_step), values)))
      pointsToWrite.sort(key=lambda p: p[0],reverse=True) #order points by timestamp, newest first
      update_many(path_to, pointsToWrite)
      fromTime = untilTime
#!/usr/bin/env python