Source

whoosh / src / whoosh / util.py

Full commit
  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
# Copyright 2007 Matt Chaput. All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
#    1. Redistributions of source code must retain the above copyright notice,
#       this list of conditions and the following disclaimer.
#
#    2. Redistributions in binary form must reproduce the above copyright
#       notice, this list of conditions and the following disclaimer in the
#       documentation and/or other materials provided with the distribution.
#
# THIS SOFTWARE IS PROVIDED BY MATT CHAPUT ``AS IS'' AND ANY EXPRESS OR
# IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
# MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
# EVENT SHALL MATT CHAPUT OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
# INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA,
# OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
# LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
# NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
# EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#
# The views and conclusions contained in the software and documentation are
# those of the authors and should not be interpreted as representing official
# policies, either expressed or implied, of Matt Chaput.

"""Miscellaneous utility functions and classes.
"""

from __future__ import with_statement
import codecs
import re
import sys
import time
from array import array
from bisect import insort, bisect_left, bisect_right
from copy import copy
from functools import wraps
from struct import pack, unpack
from threading import Lock

from whoosh.compat import xrange, u, b, string_type
from whoosh.compat import array_tobytes
from whoosh.system import pack_ushort_le, pack_uint_le
from whoosh.system import unpack_ushort_le, unpack_uint_le


if sys.platform == 'win32':
    now = time.clock
else:
    now = time.time


# Note: these functions return a tuple of (text, length), so when you call
# them, you have to add [0] on the end, e.g. str = utf8encode(unicode)[0]

utf8encode = codecs.getencoder("utf-8")
utf8decode = codecs.getdecoder("utf-8")

#utf16encode = codecs.getencoder("utf-16-be")
#utf16decode = codecs.getdecoder("utf-16-be")
#utf32encode = codecs.getencoder("utf-32-be")
#utf32decode = codecs.getdecoder("utf-32-be")


# Functions

def make_binary_tree(fn, args, **kwargs):
    """Takes a function/class that takes two positional arguments and a list of
    arguments and returns a binary tree of results/instances.
    
    >>> make_binary_tree(UnionMatcher, [matcher1, matcher2, matcher3])
    UnionMatcher(matcher1, UnionMatcher(matcher2, matcher3))
    
    Any keyword arguments given to this function are passed to the class
    initializer.
    """

    count = len(args)
    if not count:
        raise ValueError("Called make_binary_tree with empty list")
    elif count == 1:
        return args[0]

    half = count // 2
    return fn(make_binary_tree(fn, args[:half], **kwargs),
              make_binary_tree(fn, args[half:], **kwargs), **kwargs)


def make_weighted_tree(fn, ls, **kwargs):
    """Takes a function/class that takes two positional arguments and a list of
    (weight, argument) tuples and returns a huffman-like weighted tree of
    results/instances.
    """

    if not ls:
        raise ValueError("Called make_weighted_tree with empty list")

    ls.sort()
    while len(ls) > 1:
        a = ls.pop(0)
        b = ls.pop(0)
        insort(ls, (a[0] + b[0], fn(a[1], b[1])))
    return ls[0][1]


# Varint cache

# Build a cache of the varint byte sequences for the first N integers, so we
# don't have to constantly recalculate them on the fly. This makes a small but
# noticeable difference.

def _varint(i):
    a = array("B")
    while (i & ~0x7F) != 0:
        a.append((i & 0x7F) | 0x80)
        i = i >> 7
    a.append(i)
    return array_tobytes(a)


_varint_cache_size = 512
_varint_cache = []
for i in xrange(0, _varint_cache_size):
    _varint_cache.append(_varint(i))
_varint_cache = tuple(_varint_cache)


def varint(i):
    """Encodes the given integer into a string of the minimum number  of bytes.
    """
    if i < len(_varint_cache):
        return _varint_cache[i]
    return _varint(i)


def varint_to_int(vi):
    b = ord(vi[0])
    p = 1
    i = b & 0x7f
    shift = 7
    while b & 0x80 != 0:
        b = ord(vi[p])
        p += 1
        i |= (b & 0x7F) << shift
        shift += 7
    return i


def signed_varint(i):
    """Zig-zag encodes a signed integer into a varint.
    """

    if i >= 0:
        return varint(i << 1)
    return varint((i << 1) ^ (~0))


def decode_signed_varint(i):
    """Zig-zag decodes an integer value.
    """

    if not i & 1:
        return i >> 1
    return (i >> 1) ^ (~0)


def read_varint(readfn):
    """
    Reads a variable-length encoded integer.
    
    :param readfn: a callable that reads a given number of bytes,
        like file.read().
    """

    b = ord(readfn(1))
    i = b & 0x7F

    shift = 7
    while b & 0x80 != 0:
        b = ord(readfn(1))
        i |= (b & 0x7F) << shift
        shift += 7
    return i


# Google Packed Ints algorithm: a set of four numbers is preceded by a "key"
# byte, which encodes how many bytes each of the next four integers use
# (stored in the byte as four 2-bit numbers)

# Number of future bytes to expect after a "key" byte value of N -- used to
# skip ahead from a key byte

_gint_lens = array("B", [4, 5, 6, 7, 5, 6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10, 5, 6,
7, 8, 6, 7, 8, 9, 7, 8, 9, 10, 8, 9, 10, 11, 6, 7, 8, 9, 7, 8, 9, 10, 8, 9, 10,
11, 9, 10, 11, 12, 7, 8, 9, 10, 8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12, 13, 5,
6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10, 8, 9, 10, 11, 6, 7, 8, 9, 7, 8, 9, 10, 8, 9,
10, 11, 9, 10, 11, 12, 7, 8, 9, 10, 8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12,
13, 8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12, 13, 11, 12, 13, 14, 6, 7, 8, 9, 7,
8, 9, 10, 8, 9, 10, 11, 9, 10, 11, 12, 7, 8, 9, 10, 8, 9, 10, 11, 9, 10, 11,
12, 10, 11, 12, 13, 8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12, 13, 11, 12, 13,
14, 9, 10, 11, 12, 10, 11, 12, 13, 11, 12, 13, 14, 12, 13, 14, 15, 7, 8, 9, 10,
8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12, 13, 8, 9, 10, 11, 9, 10, 11, 12, 10,
11, 12, 13, 11, 12, 13, 14, 9, 10, 11, 12, 10, 11, 12, 13, 11, 12, 13, 14, 12,
13, 14, 15, 10, 11, 12, 13, 11, 12, 13, 14, 12, 13, 14, 15, 13, 14, 15, 16])


def key_to_sizes(key):
    """Returns a list of the sizes of the next four numbers given a key byte.
    """

    return [(key >> (i * 2) & 3) + 1 for i in xrange(4)]


def write_gints(dbfile, nums):
    """Write the integers in the iterator nums to bytes stream dbfile.
    """

    buf = array("c")
    count = 0
    key = 0
    for v in nums:
        shift = count * 2
        if v < 256:
            buf.append(chr(v))
        elif v < 65536:
            key |= 1 << shift
            buf.extend(pack_ushort_le(v))
        elif v < 16777216:
            key |= 2 << shift
            buf.extend(pack_uint_le(v)[:3])
        else:
            key |= 3 << shift
            buf.extend(pack_uint_le(v))

        count += 1
        if count == 4:
            #print bin(key), repr(buf)
            dbfile.write(chr(key))
            dbfile.write(array_tobytes(buf))
            count = 0
            key = 0
            del buf[0:len(buf)]

    if count:
        dbfile.write(chr(key))
        dbfile.write(array_tobytes(buf))


def read_gints(dbfile, n):
    """Read N integers from the bytes stream dbfile. Expects that the file
    starts at a key byte.
    """

    count = 0
    read = dbfile.read
    for _ in xrange(n):
        if count == 0:
            key = ord(dbfile.read(1))
        code = key >> (count * 2) & 3
        if code == 0:
            yield ord(read(1))
        elif code == 1:
            yield unpack_ushort_le(read(2))[0]
        elif code == 2:
            yield unpack_uint_le(read(3) + "\x00")[0]
        else:
            yield unpack_uint_le(read(4))[0]

        count = (count + 1) % 4


def index_gints(dbfile, n, factor):
    """"Creates an index of skips over a list of g-ints. Returns a list of
    the ordinal of the number at the start of each skip block, and a
    corresponding list of the starting positions in the file of the block.
    """

    ords = array("i")
    poses = array("i")
    count = 0
    pos = dbfile.tell()
    while count < n:
        key = ord(dbfile.read(1))
        if not count % factor:
            ords.append(count)
            poses.append(pos)
        delta = _gint_lens[key]
        dbfile.seek(delta, 1)
        pos += delta + 1
        count += 4
    return ords, poses


def get_gint(dbfile, i, ords, poses):
    """Given a byte stream, the ordinal position to get, and the skip lists
    created by index_gints, returns the value of the i-th number in the
    stream.
    """

    # Find which skip block the number is in
    j = bisect_right(ords, i) - 1
    # Ordinal position within the block: the ordinal position of the start of
    # the block minus the ordinal position we're looking for
    wbi = i - ords[j]
    # Seek to the start of the block
    dbfile.seek(poses[j])

    # Skip as many 4-int sets as we can
    while wbi // 4 >= 1:
        key = ord(dbfile.read(1))
        dbfile.seek(_gint_lens[key], 1)
        wbi -= 4

    # Read enough numbers serially to get the one we're looking for
    for n in read_gints(dbfile, wbi + 1):
        pass
    # Return the last read number
    return n


# Fibonacci function

_fib_cache = {}


def fib(n):
    """Returns the nth value in the Fibonacci sequence.
    """

    if n <= 2:
        return n
    if n in _fib_cache:
        return _fib_cache[n]
    result = fib(n - 1) + fib(n - 2)
    _fib_cache[n] = result
    return result


# Float-to-byte encoding/decoding

def float_to_byte(value, mantissabits=5, zeroexp=2):
    """Encodes a floating point number in a single byte.
    """

    # Assume int size == float size

    fzero = (63 - zeroexp) << mantissabits
    bits = unpack("i", pack("f", value))[0]
    smallfloat = bits >> (24 - mantissabits)
    if smallfloat < fzero:
        # Map negative numbers and 0 to 0
        # Map underflow to next smallest non-zero number
        if bits <= 0:
            result = chr(0)
        else:
            result = chr(1)
    elif smallfloat >= fzero + 0x100:
        # Map overflow to largest number
        result = chr(255)
    else:
        result = chr(smallfloat - fzero)
    return b(result)


def byte_to_float(b, mantissabits=5, zeroexp=2):
    """Decodes a floating point number stored in a single byte.
    """
    if type(b) is not int:
        b = ord(b)
    if b == 0:
        return 0.0

    bits = (b & 0xff) << (24 - mantissabits)
    bits += (63 - zeroexp) << 24
    return unpack("f", pack("i", bits))[0]


# Length-to-byte approximation functions

# Old implementation:

#def length_to_byte(length):
#    """Returns a logarithmic approximation of the given number, in the range
#    0-255. The approximation has high precision at the low end (e.g.
#    1 -> 0, 2 -> 1, 3 -> 2 ...) and low precision at the high end. Numbers
#    equal to or greater than 108116 all approximate to 255.
#
#    This is useful for storing field lengths, where the general case is small
#    documents and very large documents are more rare.
#    """
#
#    # This encoding formula works up to 108116 -> 255, so if the length is
#    # equal to or greater than that limit, just return 255.
#    if length >= 108116:
#        return 255
#
#    # The parameters of this formula where chosen heuristically so that low
#    # numbers would approximate closely, and the byte range 0-255 would cover
#    # a decent range of document lengths (i.e. 1 to ~100000).
#    return int(round(log((length / 27.0) + 1, 1.033)))
#def _byte_to_length(n):
#    return int(round((pow(1.033, n) - 1) * 27))
#_b2l_cache = array("i", (_byte_to_length(i) for i in xrange(256)))
#byte_to_length = _b2l_cache.__getitem__

# New implementation

# Instead of computing the actual formula to get the byte for any given length,
# precompute the length associated with each byte, and use bisect to find the
# nearest value. This gives quite a large speed-up.
#
# Note that this does not give all the same answers as the old, "real"
# implementation since this implementation always "rounds down" (thanks to the
# bisect_left) while the old implementation would "round up" or "round down"
# depending on the input. Since this is a fairly gross approximation anyway,
# I don't think it matters much.

# Values generated using the formula from the "old" implementation above
_length_byte_cache = array('i', [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 13, 14,
16, 17, 18, 20, 21, 23, 25, 26, 28, 30, 32, 34, 36, 38, 40, 42, 45, 47, 49, 52,
54, 57, 60, 63, 66, 69, 72, 75, 79, 82, 86, 89, 93, 97, 101, 106, 110, 114,
119, 124, 129, 134, 139, 145, 150, 156, 162, 169, 175, 182, 189, 196, 203, 211,
219, 227, 235, 244, 253, 262, 271, 281, 291, 302, 313, 324, 336, 348, 360, 373,
386, 399, 414, 428, 443, 459, 475, 491, 508, 526, 544, 563, 583, 603, 623, 645,
667, 690, 714, 738, 763, 789, 816, 844, 873, 903, 933, 965, 998, 1032, 1066,
1103, 1140, 1178, 1218, 1259, 1302, 1345, 1391, 1438, 1486, 1536, 1587, 1641,
1696, 1753, 1811, 1872, 1935, 1999, 2066, 2135, 2207, 2280, 2356, 2435, 2516,
2600, 2687, 2777, 2869, 2965, 3063, 3165, 3271, 3380, 3492, 3608, 3728, 3852,
3980, 4112, 4249, 4390, 4536, 4686, 4842, 5002, 5168, 5340, 5517, 5700, 5889,
6084, 6286, 6494, 6709, 6932, 7161, 7398, 7643, 7897, 8158, 8428, 8707, 8995,
9293, 9601, 9918, 10247, 10586, 10936, 11298, 11671, 12057, 12456, 12868,
13294, 13733, 14187, 14656, 15141, 15641, 16159, 16693, 17244, 17814, 18403,
19011, 19640, 20289, 20959, 21652, 22367, 23106, 23869, 24658, 25472, 26314,
27183, 28081, 29009, 29967, 30957, 31979, 33035, 34126, 35254, 36418, 37620,
38863, 40146, 41472, 42841, 44256, 45717, 47227, 48786, 50397, 52061, 53780,
55556, 57390, 59285, 61242, 63264, 65352, 67510, 69739, 72041, 74419, 76876,
79414, 82035, 84743, 87541, 90430, 93416, 96499, 99684, 102975, 106374])


def length_to_byte(length):
    if length is None:
        return 0
    if length >= 106374:
        return 255
    else:
        return bisect_left(_length_byte_cache, length)

byte_to_length = _length_byte_cache.__getitem__


# Prefix encoding functions

def first_diff(a, b):
    """Returns the position of the first differing character in the strings
    a and b. For example, first_diff('render', 'rending') == 4. This function
    limits the return value to 255 so the difference can be encoded in a single
    byte.
    """

    i = 0
    for i in xrange(0, len(a)):
        if a[i] != b[i] or i == 255:
            break
    return i


def prefix_encode(a, b):
    """Compresses string b as an integer (encoded in a byte) representing
    the prefix it shares with a, followed by the suffix encoded as UTF-8.
    """
    i = first_diff(a, b)
    return chr(i) + b[i:].encode("utf-8")


def prefix_encode_all(ls):
    """Compresses the given list of (unicode) strings by storing each string
    (except the first one) as an integer (encoded in a byte) representing
    the prefix it shares with its predecessor, followed by the suffix encoded
    as UTF-8.
    """

    last = u('')
    for w in ls:
        i = first_diff(last, w)
        yield chr(i) + w[i:].encode("utf-8")
        last = w


def prefix_decode_all(ls):
    """Decompresses a list of strings compressed by prefix_encode().
    """

    last = u('')
    for w in ls:
        i = ord(w[0])
        decoded = last[:i] + w[1:].decode("utf-8")
        yield decoded
        last = decoded


# Natural key sorting function

_nkre = re.compile(r"\D+|\d+", re.UNICODE)


def _nkconv(i):
    try:
        return int(i)
    except ValueError:
        return i.lower()


def natural_key(s):
    """Converts string ``s`` into a tuple that will sort "naturally" (i.e.,
    ``name5`` will come before ``name10`` and ``1`` will come before ``A``).
    This function is designed to be used as the ``key`` argument to sorting
    functions.
    
    :param s: the str/unicode string to convert.
    :rtype: tuple
    """

    # Use _nkre to split the input string into a sequence of
    # digit runs and non-digit runs. Then use _nkconv() to convert
    # the digit runs into ints and the non-digit runs to lowercase.
    return tuple(_nkconv(m) for m in _nkre.findall(s))


# Mixins and decorators

class ClosableMixin(object):
    """Mix-in for classes with a close() method to allow them to be used as a
    context manager.
    """

    def __enter__(self):
        return self

    def __exit__(self, *exc_info):
        self.close()


def protected(func):
    """Decorator for storage-access methods. This decorator (a) checks if the
    object has already been closed, and (b) synchronizes on a threading lock.
    The parent object must have 'is_closed' and '_sync_lock' attributes.
    """

    @wraps(func)
    def protected_wrapper(self, *args, **kwargs):
        if self.is_closed:
            raise Exception("%r has been closed" % self)
        with self._sync_lock:
            return func(self, *args, **kwargs)

    return protected_wrapper


def synchronized(func):
    """Decorator for storage-access methods, which synchronizes on a threading
    lock. The parent object must have 'is_closed' and '_sync_lock' attributes.
    """

    @wraps(func)
    def synchronized_wrapper(self, *args, **kwargs):
        with self._sync_lock:
            return func(self, *args, **kwargs)

    return synchronized_wrapper


def unbound_cache(func):
    """Caching decorator with an unbounded cache size.
    """

    cache = {}

    @wraps(func)
    def caching_wrapper(*args):
        try:
            return cache[args]
        except KeyError:
            result = func(*args)
            cache[args] = result
            return result

    return caching_wrapper


def lru_cache(maxsize=100):
    """Double-barrel least-recently-used cache decorator. This is a simple
    LRU algorithm that keeps a primary and secondary dict. Keys are checked
    in the primary dict, and then the secondary. Once the primary dict fills
    up, the secondary dict is cleared and the two dicts are swapped.
    
    This function duplicates (more-or-less) the protocol of the
    ``functools.lru_cache`` decorator in the Python 3.2 standard library.

    Arguments to the cached function must be hashable.

    View the cache statistics named tuple (hits, misses, maxsize, currsize)
    with f.cache_info().  Clear the cache and statistics with f.cache_clear().
    Access the underlying function with f.__wrapped__.
    """

    def decorating_function(user_function):
        # Cache1, Cache2, Pointer, Hits, Misses
        stats = [{}, {}, 0, 0, 0]

        @wraps(user_function)
        def wrapper(*args):
            ptr = stats[2]
            a = stats[ptr]
            b = stats[not ptr]
            key = args

            if key in a:
                stats[3] += 1  # Hit
                return a[key]
            elif key in b:
                stats[3] += 1  # Hit
                return b[key]
            else:
                stats[4] += 1  # Miss
                result = user_function(*args)
                a[key] = result
                if len(a) >= maxsize:
                    stats[2] = not ptr
                    b.clear()
                return result

        def cache_info():
            """Report cache statistics"""
            return (stats[3], stats[4], maxsize, len(stats[0]) + len(stats[1]))

        def cache_clear():
            """Clear the cache and cache statistics"""
            stats[0].clear()
            stats[1].clear()
            stats[3] = stats[4] = 0

        wrapper.cache_info = cache_info
        wrapper.cache_clear = cache_clear

        return wrapper
    return decorating_function


def clockface_lru_cache(maxsize=100):
    """Least-recently-used cache decorator.

    This function duplicates (more-or-less) the protocol of the
    ``functools.lru_cache`` decorator in the Python 3.2 standard library, but
    uses the clock face LRU algorithm instead of an ordered dictionary.

    If *maxsize* is set to None, the LRU features are disabled and the cache
    can grow without bound.

    Arguments to the cached function must be hashable.

    View the cache statistics named tuple (hits, misses, maxsize, currsize)
    with f.cache_info().  Clear the cache and statistics with f.cache_clear().
    Access the underlying function with f.__wrapped__.
    """

    def decorating_function(user_function):

        stats = [0, 0, 0]  # hits, misses, hand
        data = {}

        if maxsize:
            # The keys at each point on the clock face
            clock_keys = [None] * maxsize
            # The "referenced" bits at each point on the clock face
            clock_refs = array("B", (0 for _ in xrange(maxsize)))
            lock = Lock()

            @wraps(user_function)
            def wrapper(*args):
                key = args
                try:
                    with lock:
                        pos, result = data[key]
                        # The key is in the cache. Set the key's reference bit
                        clock_refs[pos] = 1
                        # Record a cache hit
                        stats[0] += 1
                except KeyError:
                    # Compute the value
                    result = user_function(*args)
                    with lock:
                        # Current position of the clock hand
                        hand = stats[2]
                        # Remember to stop here after a full revolution
                        end = hand
                        # Sweep around the clock looking for a position with
                        # the reference bit off
                        while True:
                            hand = (hand + 1) % maxsize
                            current_ref = clock_refs[hand]
                            if current_ref:
                                # This position's "referenced" bit is set. Turn
                                # the bit off and move on.
                                clock_refs[hand] = 0
                            elif not current_ref or hand == end:
                                # We've either found a position with the
                                # "reference" bit off or reached the end of the
                                # circular cache. So we'll replace this
                                # position with the new key
                                current_key = clock_keys[hand]
                                if current_key in data:
                                    del data[current_key]
                                clock_keys[hand] = key
                                clock_refs[hand] = 1
                                break
                        # Put the key and result in the cache
                        data[key] = (hand, result)
                        # Save the new hand position
                        stats[2] = hand
                        # Record a cache miss
                        stats[1] += 1
                return result

        else:
            @wraps(user_function)
            def wrapper(*args):
                key = args
                try:
                    result = data[key]
                    stats[0] += 1
                except KeyError:
                    result = user_function(*args)
                    data[key] = result
                    stats[1] += 1
                return result

        def cache_info():
            """Report cache statistics"""
            return (stats[0], stats[1], maxsize, len(data))

        def cache_clear():
            """Clear the cache and cache statistics"""
            data.clear()
            stats[0] = stats[1] = stats[2] = 0
            for i in xrange(maxsize):
                clock_keys[i] = None
                clock_refs[i] = 0

        wrapper.cache_info = cache_info
        wrapper.cache_clear = cache_clear
        return wrapper

    return decorating_function


def find_object(name, blacklist=None, whitelist=None):
    """Imports and returns an object given a fully qualified name.
    
    >>> find_object("whoosh.analysis.StopFilter")
    <class 'whoosh.analysis.StopFilter'>
    """

    if blacklist:
        for pre in blacklist:
            if name.startswith(pre):
                raise TypeError("%r: can't instantiate names starting with %r"
                                % (name, pre))
    if whitelist:
        passes = False
        for pre in whitelist:
            if name.startswith(pre):
                passes = True
                break
        if not passes:
            raise TypeError("Can't instantiate %r" % name)

    lastdot = name.rfind(".")

    assert lastdot > -1, "Name %r must be fully qualified" % name
    modname = name[:lastdot]
    clsname = name[lastdot + 1:]

    mod = __import__(modname, fromlist=[clsname])
    cls = getattr(mod, clsname)
    return cls


def rcompile(pattern, flags=0, verbose=False):
    """A wrapper for re.compile that checks whether "pattern" is a regex object
    or a string to be compiled, and automatically adds the re.UNICODE flag.
    """

    if not isinstance(pattern, string_type):
        # If it's not a string, assume it's already a compiled pattern
        return pattern
    if verbose:
        flags |= re.VERBOSE
    return re.compile(pattern, re.UNICODE | flags)