Source

whoosh / src / whoosh / columns.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
 942
 943
 944
 945
 946
 947
 948
 949
 950
 951
 952
 953
 954
 955
 956
 957
 958
 959
 960
 961
 962
 963
 964
 965
 966
 967
 968
 969
 970
 971
 972
 973
 974
 975
 976
 977
 978
 979
 980
 981
 982
 983
 984
 985
 986
 987
 988
 989
 990
 991
 992
 993
 994
 995
 996
 997
 998
 999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
# Copyright 2012 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.

"""
The API and implementation of columns may change in the next version of Whoosh!

This module contains "Column" objects which you can use as the argument to a
Field object's ``sortable=`` keyword argument. Each field defines a default
column type for when the user specifies ``sortable=True`` (the object returned
by the field's ``default_column()`` method).

The default column type for most fields is ``VarBytesColumn``,
although numeric and date fields use ``NumericColumn``. Expert users may use
other field types that may be faster or more storage efficient based on the
field contents. For example, if a field always contains one of a limited number
of possible values, a ``RefBytesColumn`` will save space by only storing the
values once. If a field's values are always a fixed length, the
``FixedBytesColumn`` saves space by not storing the length of each value.

A ``Column`` object basically exists to store configuration information and
provides two important methods: ``writer()`` to return a ``ColumnWriter`` object
and ``reader()`` to return a ``ColumnReader`` object.
"""

from __future__ import division, with_statement
import struct, warnings
from array import array
from bisect import bisect_right

try:
    import zlib
except ImportError:
    zlib = None

from whoosh.compat import b, bytes_type, BytesIO
from whoosh.compat import array_tobytes, xrange
from whoosh.compat import dumps, loads
from whoosh.filedb.structfile import StructFile
from whoosh.idsets import BitSet, OnDiskBitSet
from whoosh.system import emptybytes
from whoosh.util.cache import lru_cache
from whoosh.util.numeric import typecode_max, typecode_min
from whoosh.util.numlists import GrowableArray
from whoosh.util.varints import varint


# Utility functions

def _mintype(maxn):
    if maxn < 2 ** 8:
        typecode = "B"
    elif maxn < 2 ** 16:
        typecode = "H"
    elif maxn < 2 ** 31:
        typecode = "i"
    else:
        typecode = "I"

    return typecode


# Python does not support arrays of long long see Issue 1172711
# These functions help write/read a simulated an array of q/Q using lists

def write_qsafe_array(typecode, arry, dbfile):
    if typecode == "q":
        for num in arry:
            dbfile.write_long(num)
    elif typecode == "Q":
        for num in arry:
            dbfile.write_ulong(num)
    else:
        dbfile.write_array(arry)


def read_qsafe_array(typecode, size, dbfile):
    if typecode == "q":
        arry = [dbfile.read_long() for _ in xrange(size)]
    elif typecode == "Q":
        arry = [dbfile.read_ulong() for _ in xrange(size)]
    else:
        arry = dbfile.read_array(typecode, size)

    return arry


def make_array(typecode, size=0, default=None):
    if typecode.lower() == "q":
        # Python does not support arrays of long long see Issue 1172711
        if default is not None and size:
            arry = [default] * size
        else:
            arry = []
    else:
        if default is not None and size:
            arry = array(typecode, (default for _ in xrange(size)))
        else:
            arry = array(typecode)
    return arry


# Base classes

class Column(object):
    """Represents a "column" of rows mapping docnums to document values.

    The interface requires that you store the start offset of the column, the
    length of the column data, and the number of documents (rows) separately,
    and pass them to the reader object.
    """

    reversible = False

    def writer(self, dbfile):
        """Returns a :class:`ColumnWriter` object you can use to use to create
        a column of this type on disk.

        :param dbfile: the :class:`~whoosh.filedb.structfile.StructFile` to
            write to.
        """

        return self.Writer(dbfile)

    def reader(self, dbfile, basepos, length, doccount):
        """Returns a :class:`ColumnReader` object you can use to read a column
        of this type from disk.

        :param dbfile: the :class:`~whoosh.filedb.structfile.StructFile` to
            read from.
        :param basepos: the offset within the file at which the column starts.
        :param length: the length in bytes of the column occupies in the file.
        :param doccount: the number of rows (documents) in the column.
        """

        return self.Reader(dbfile, basepos, length, doccount)

    def default_value(self):
        """Returns the default value for this column type.
        """

        return self._default

    def stores_lists(self):
        """Returns True if the column stores a list of values for each document
        instead of a single value.
        """

        return False


class ColumnWriter(object):
    def __init__(self, dbfile):
        self._dbfile = dbfile
        self._count = 0

    def fill(self, docnum):
        write = self._dbfile.write
        default = self._defaultbytes
        if docnum > self._count:
            for _ in xrange(docnum - self._count):
                write(default)

    def add(self, docnum, value):
        raise NotImplementedError

    def finish(self, docnum):
        pass


class ColumnReader(object):
    def __init__(self, dbfile, basepos, length, doccount):
        self._dbfile = dbfile
        self._basepos = basepos
        self._length = length
        self._doccount = doccount

    def __len__(self):
        return self._doccount

    def __getitem__(self, docnum):
        raise NotImplementedError

    def sort_key(self, docnum, reverse=False):
        return self[docnum]

    def __iter__(self):
        for i in xrange(self._doccount):
            yield self[i]

    def load(self):
        return list(self)


# Arbitrary bytes column

class VarBytesColumn(Column):
    """Stores variable length byte strings. See also :class:`RefBytesColumn`.

    The current implementation limits the total length of all document values
    a segment to 2 GB.

    The default value (the value returned for a document that didn't have a
    value assigned to it at indexing time) is an empty bytestring (``b''``).
    """

    _default = emptybytes

    class Writer(ColumnWriter):
        def __init__(self, dbfile):
            assert isinstance(dbfile, StructFile)
            self._dbfile = dbfile
            self._count = 0
            self._lengths = GrowableArray(allow_longs=False)

        def __repr__(self):
            return "<VarBytes.Writer>"

        def fill(self, docnum):
            if docnum > self._count:
                self._lengths.extend(0 for _ in xrange(docnum - self._count))

        def add(self, docnum, v):
            self.fill(docnum)
            self._dbfile.write(v)
            self._lengths.append(len(v))
            self._count = docnum + 1

        def finish(self, doccount):
            self.fill(doccount)
            lengths = self._lengths.array

            self._dbfile.write_array(lengths)
            # Write the typecode for the lengths
            self._dbfile.write_byte(ord(lengths.typecode))

    class Reader(ColumnReader):
        def __init__(self, dbfile, basepos, length, doccount):
            self._dbfile = dbfile
            self._basepos = basepos
            self._length = length
            self._doccount = doccount

            self._read_lengths()
            # Create an array of offsets into the strings using the lengths
            offsets = array("L", (0,))
            for length in self._lengths:
                offsets.append(offsets[-1] + length)
            self._offsets = offsets

        def __repr__(self):
            return "<VarBytes.Reader>"

        def _read_lengths(self):
            dbfile = self._dbfile
            basepos = self._basepos
            length = self._length
            doccount = self._doccount

            # The end of the lengths array is the end of the data minus the
            # typecode byte
            endoflens = basepos + length - 1
            # Load the length typecode from before the key length
            typecode = chr(dbfile.get_byte(endoflens))
            # Load the length array from before the typecode
            itemsize = struct.calcsize(typecode)
            lengthsbase = endoflens - (itemsize * doccount)
            self._lengths = dbfile.get_array(lengthsbase, typecode, doccount)

        @lru_cache()
        def __getitem__(self, docnum):
            length = self._lengths[docnum]
            if not length:
                return emptybytes
            offset = self._offsets[docnum]
            return self._dbfile.get(self._basepos + offset, length)

        def __iter__(self):
            get = self._dbfile.get
            pos = self._basepos
            for length in self._lengths:
                yield get(pos, length)
                pos += length


class FixedBytesColumn(Column):
    """Stores fixed-length byte strings.
    """

    def __init__(self, fixedlen, default=None):
        """
        :param fixedlen: the fixed length of byte strings in this column.
        :param default: the default value to use for documents that don't
            specify a value. If you don't specify a default, the column will
            use ``b'\\x00' * fixedlen``.
        """

        self._fixedlen = fixedlen

        if default is None:
            default = b("\x00") * fixedlen
        elif len(default) != fixedlen:
            raise ValueError
        self._default = default

    def writer(self, dbfile):
        return self.Writer(dbfile, self._fixedlen, self._default)

    def reader(self, dbfile, basepos, length, doccount):
        return self.Reader(dbfile, basepos, length, doccount, self._fixedlen,
                           self._default)

    class Writer(ColumnWriter):
        def __init__(self, dbfile, fixedlen, default):
            self._dbfile = dbfile
            self._fixedlen = fixedlen
            self._default = self._defaultbytes = default
            self._count = 0

        def __repr__(self):
            return "<FixedBytes.Writer>"

        def add(self, docnum, v):
            if v == self._default:
                return
            if docnum > self._count:
                self.fill(docnum)
            assert len(v) == self._fixedlen
            self._dbfile.write(v)
            self._count = docnum + 1

    class Reader(ColumnReader):
        def __init__(self, dbfile, basepos, length, doccount, fixedlen,
                     default):
            self._dbfile = dbfile
            self._basepos = basepos
            self._doccount = doccount
            self._fixedlen = fixedlen
            self._default = self._defaultbytes = default
            self._count = length // fixedlen

        def __repr__(self):
            return "<FixedBytes.Reader>"

        def __getitem__(self, docnum):
            if docnum >= self._count:
                return self._defaultbytes
            pos = self._basepos + self._fixedlen * docnum
            return self._dbfile.get(pos, self._fixedlen)

        def __iter__(self):
            count = self._count
            default = self._default
            for i in xrange(self._doccount):
                if i < count:
                    yield self[i]
                else:
                    yield default


# Variable/fixed length reference (enum) column

class RefBytesColumn(Column):
    """Stores variable-length or fixed-length byte strings, similar to
    :class:`VarBytesColumn` and :class:`FixedBytesColumn`. However, where those
    columns stores a value for each document, this column keeps a list of all
    the unique values in the field, and for each document stores a short
    pointer into the unique list. For fields where the number of possible
    values is smaller than the number of documents (for example,
    "category" or "chapter"), this saves significant space.

    This column type supports a maximum of 65535 unique values across all
    documents in a segment. You should generally use this column type where the
    number of unique values is in no danger of approaching that number (for
    example, a "tags" field). If you try to index too many unique values, the
    column will convert additional unique values to the default value and issue
    a warning using the ``warnings`` module (this will usually be preferable to
    crashing the indexer and potentially losing indexed documents).
    """

    # NOTE that RefBytes is reversible within a single column (we could just
    # negate the reference number), but it's NOT reversible ACROSS SEGMENTS
    # (since different segments can have different uniques values in their
    # columns), so we have to say that the column type is not reversible
    reversible = False

    def __init__(self, fixedlen=0, default=None):
        """
        :param fixedlen: an optional fixed length for the values. If you
            specify a number other than 0, the column will require all values
            to be the specified length.
        :param default: a default value to use for documents that don't specify
            one. If you don't specify a default, the column will use an empty
            bytestring (``b''``), or if you specify a fixed length,
            ``b'\\x00' * fixedlen``.
        """

        self._fixedlen = fixedlen

        if default is None:
            default = b("\x00") * fixedlen if fixedlen else emptybytes
        elif fixedlen and len(default) != fixedlen:
            raise ValueError
        self._default = default

    def writer(self, dbfile):
        return self.Writer(dbfile, self._fixedlen, self._default)

    def reader(self, dbfile, basepos, length, doccount):
        return self.Reader(dbfile, basepos, length, doccount, self._fixedlen)

    class Writer(ColumnWriter):
        def __init__(self, dbfile, fixedlen, default):
            self._dbfile = dbfile
            self._fixedlen = fixedlen
            self._default = default

            # At first we'll buffer refs in a byte array. If the number of
            # uniques stays below 256, we can just write the byte array. As
            # soon as the ref count goes above 255, we know we're going to have
            # to write shorts, so we'll switch to writing directly.
            self._refs = array("B")
            self._uniques = {default: 0}
            self._count = 0

        def __repr__(self):
            return "<RefBytes.Writer>"

        def fill(self, docnum):
            if docnum > self._count:
                if self._refs is not None:
                    self._refs.extend(0 for _ in xrange(docnum - self._count))
                else:
                    dbfile = self._dbfile
                    for _ in xrange(docnum - self._count):
                        dbfile.write_ushort(0)

        def add(self, docnum, v):
            dbfile = self._dbfile
            refs = self._refs
            self.fill(docnum)

            uniques = self._uniques
            try:
                ref = uniques[v]
            except KeyError:
                uniques[v] = ref = len(uniques)
                if refs is not None and ref >= 256:
                    # We won't be able to use bytes, we have to switch to
                    # writing unbuffered ushorts
                    for n in refs:
                        dbfile.write_ushort(n)
                    refs = self._refs = None

            if refs is not None:
                self._refs.append(ref)
            else:
                if ref > 65535:
                    warnings.warn("RefBytesColumn dropped unique value %r" % v,
                                  UserWarning)
                    ref = 0
                dbfile.write_ushort(ref)

            self._count = docnum + 1

        def _write_uniques(self, typecode):
            dbfile = self._dbfile
            fixedlen = self._fixedlen
            uniques = self._uniques

            dbfile.write_varint(len(uniques))
            # Sort unique values by position
            vs = sorted(uniques.keys(), key=lambda key: uniques[key])
            for v in vs:
                if not fixedlen:
                    dbfile.write_varint(len(v))
                dbfile.write(v)

        def finish(self, doccount):
            dbfile = self._dbfile
            refs = self._refs
            self.fill(doccount)

            typecode = "H"
            if refs is not None:
                dbfile.write_array(refs)
                typecode = refs.typecode

            self._write_uniques(typecode)
            dbfile.write_byte(ord(typecode))

    class Reader(ColumnReader):
        def __init__(self, dbfile, basepos, length, doccount, fixedlen):
            self._dbfile = dbfile
            self._basepos = basepos
            self._doccount = doccount
            self._fixedlen = fixedlen

            self._typecode = chr(dbfile.get_byte(basepos + length - 1))

            st = struct.Struct("!" + self._typecode)
            self._unpack = st.unpack
            self._itemsize = st.size

            dbfile.seek(basepos + doccount * self._itemsize)
            self._uniques = self._read_uniques()

        def __repr__(self):
            return "<RefBytes.Reader>"

        def _read_uniques(self):
            dbfile = self._dbfile
            fixedlen = self._fixedlen

            ucount = dbfile.read_varint()
            length = fixedlen
            uniques = []
            for _ in xrange(ucount):
                if not fixedlen:
                    length = dbfile.read_varint()
                uniques.append(dbfile.read(length))
            return uniques

        def __getitem__(self, docnum):
            pos = self._basepos + docnum * self._itemsize
            ref = self._unpack(self._dbfile.get(pos, self._itemsize))[0]
            return self._uniques[ref]

        def __iter__(self):
            get = self._dbfile.get
            basepos = self._basepos
            uniques = self._uniques
            unpack = self._unpack
            itemsize = self._itemsize

            for i in xrange(self._doccount):
                pos = basepos + i * itemsize
                ref = unpack(get(pos, itemsize))[0]
                yield uniques[ref]


# Numeric column

class NumericColumn(FixedBytesColumn):
    """Stores numbers (integers and floats) as compact binary.
    """

    reversible = True

    def __init__(self, typecode, default=0):
        """
        :param typecode: a typecode character (as used by the ``struct``
            module) specifying the number type. For example, ``"i"`` for
            signed integers.
        :param default: the default value to use for documents that don't
            specify one.
        """

        self._typecode = typecode
        self._default = default

    def writer(self, dbfile):
        return self.Writer(dbfile, self._typecode, self._default)

    def reader(self, dbfile, basepos, length, doccount):
        return self.Reader(dbfile, basepos, length, doccount, self._typecode,
                           self._default)

    class Writer(FixedBytesColumn.Writer):
        def __init__(self, dbfile, typecode, default):
            self._dbfile = dbfile
            self._pack = struct.Struct("!" + typecode).pack
            self._default = default
            self._defaultbytes = self._pack(default)
            self._fixedlen = struct.calcsize(typecode)
            self._count = 0

        def __repr__(self):
            return "<Numeric.Writer>"

        def add(self, docnum, v):
            if v == self._default:
                return
            if docnum > self._count:
                self.fill(docnum)
            self._dbfile.write(self._pack(v))
            self._count = docnum + 1

    class Reader(FixedBytesColumn.Reader):
        def __init__(self, dbfile, basepos, length, doccount, typecode,
                     default):
            self._dbfile = dbfile
            self._basepos = basepos
            self._doccount = doccount
            self._default = default

            self._typecode = typecode
            self._unpack = struct.Struct("!" + typecode).unpack
            self._defaultbytes = struct.pack("!" + typecode, default)
            self._fixedlen = struct.calcsize(typecode)
            self._count = length // self._fixedlen

        def __repr__(self):
            return "<Numeric.Reader>"

        def __getitem__(self, docnum):
            s = FixedBytesColumn.Reader.__getitem__(self, docnum)
            return self._unpack(s)[0]

        def sort_key(self, docnum, reverse=False):
            key = self[docnum]
            if reverse:
                key = 0 - key
            return key

        def load(self):
            if self._typecode in "qQ":
                return list(self)
            else:
                return array(self._typecode, self)


# Column of boolean values

class BitColumn(Column):
    """Stores a column of True/False values compactly.
    """

    reversible = True
    _default = False

    def __init__(self, compress_at=2048):
        """
        :param compress_at: columns with this number of values or fewer will
            be saved compressed on disk, and loaded into RAM for reading. Set
            this to 0 to disable compression.
        """

        self._compressat = compress_at

    def writer(self, dbfile):
        return self.Writer(dbfile, self._compressat)

    class Writer(ColumnWriter):
        def __init__(self, dbfile, compressat):
            self._dbfile = dbfile
            self._compressat = compressat
            self._bitset = BitSet()

        def __repr__(self):
            return "<Bit.Writer>"

        def add(self, docnum, value):
            if value:
                self._bitset.add(docnum)

        def finish(self, doccount):
            dbfile = self._dbfile
            bits = self._bitset.bits

            if zlib and len(bits) <= self._compressat:
                compressed = zlib.compress(array_tobytes(bits), 3)
                dbfile.write(compressed)
                dbfile.write_byte(1)
            else:
                dbfile.write_array(bits)
                dbfile.write_byte(0)

    class Reader(ColumnReader):
        def __init__(self, dbfile, basepos, length, doccount):
            self._dbfile = dbfile
            self._basepos = basepos
            self._length = length
            self._doccount = doccount

            compressed = dbfile.get_byte(basepos + (length - 1))
            if compressed:
                bbytes = zlib.decompress(dbfile.get(basepos, length - 1))
                bitset = BitSet.from_bytes(bbytes)
            else:
                dbfile.seek(basepos)
                bitset = OnDiskBitSet(dbfile, basepos, length - 1)
            self._bitset = bitset

        def id_set(self):
            return self._bitset

        def __repr__(self):
            return "<Bit.Reader>"

        def __getitem__(self, i):
            return i in self._bitset

        def sort_key(self, docnum, reverse=False):
            return int(self[docnum] ^ reverse)

        def __iter__(self):
            i = 0
            for num in self._bitset:
                if num > i:
                    for _ in xrange(num - i):
                        yield False
                yield True
                i = num + 1
            if self._doccount > i:
                for _ in xrange(self._doccount - i):
                    yield False

        def load(self):
            if isinstance(self._bitset, OnDiskBitSet):
                bs = self._dbfile.get_array(self._basepos, "B",
                                            self._length - 1)
                self._bitset = BitSet.from_bytes(bs)
            return self


# Compressed variants

class CompressedBytesColumn(Column):
    """Stores variable-length byte strings compressed using deflate (by
    default).
    """

    def __init__(self, level=3, module="zlib"):
        """
        :param level: the compression level to use.
        :param module: a string containing the name of the compression module
            to use. The default is "zlib". The module should export "compress"
            and "decompress" functions.
        """

        self._level = level
        self._module = module

    def writer(self, dbfile):
        return self.Writer(dbfile, self._level, self._module)

    def reader(self, dbfile, basepos, length, doccount):
        return self.Reader(dbfile, basepos, length, doccount, self._module)

    class Writer(VarBytesColumn.Writer):
        def __init__(self, dbfile, level, module):
            VarBytesColumn.Writer.__init__(self, dbfile)
            self._level = level
            self._compress = __import__(module).compress

        def __repr__(self):
            return "<CompressedBytes.Writer>"

        def add(self, docnum, v):
            v = self._compress(v, self._level)
            VarBytesColumn.Writer.add(self, docnum, v)

    class Reader(VarBytesColumn.Reader):
        def __init__(self, dbfile, basepos, length, doccount, module):
            VarBytesColumn.Reader.__init__(self, dbfile, basepos, length,
                                           doccount)
            self._decompress = __import__(module).decompress

        def __repr__(self):
            return "<CompressedBytes.Reader>"

        def __getitem__(self, docnum):
            v = VarBytesColumn.Reader.__getitem__(self, docnum)
            if v:
                v = self._decompress(v)
            return v

        def __iter__(self):
            for v in VarBytesColumn.Reader.__iter__(self):
                yield self._decompress(v)

        def load(self):
            return list(self)


class CompressedBlockColumn(Column):
    """An experimental column type that compresses and decompresses blocks of
    values at a time. This can lead to high compression and decent performance
    for columns with lots of very short values, but random access times are
    usually terrible.
    """

    def __init__(self, level=3, blocksize=32, module="zlib"):
        """
        :param level: the compression level to use.
        :param blocksize: the size (in KB) of each compressed block.
        :param module: a string containing the name of the compression module
            to use. The default is "zlib". The module should export "compress"
            and "decompress" functions.
        """

        self._level = level
        self._blocksize = blocksize
        self._module = module

    def writer(self, dbfile):
        return self.Writer(dbfile, self._level, self._blocksize, self._module)

    def reader(self, dbfile, basepos, length, doccount):
        return self.Reader(dbfile, basepos, length, doccount, self._module)

    class Writer(ColumnWriter):
        def __init__(self, dbfile, level, blocksize, module):
            self._dbfile = dbfile
            self._blocksize = blocksize * 1024
            self._level = level
            self._compress = __import__(module).compress

            self._reset()

        def __repr__(self):
            return "<CompressedBlock.Writer>"

        def _reset(self):
            self._startdoc = None
            self._block = emptybytes
            self._lengths = []

        def _emit(self):
            dbfile = self._dbfile
            block = self._compress(self._block, self._level)
            header = (self._startdoc, self._lastdoc, len(block),
                      tuple(self._lengths))
            dbfile.write_pickle(header)
            dbfile.write(block)

        def add(self, docnum, v):
            if self._startdoc is None:
                self._startdoc = docnum
            self._lengths.append((docnum, len(v)))
            self._lastdoc = docnum

            self._block += v
            if len(self._block) >= self._blocksize:
                self._emit()
                self._reset()

        def finish(self, doccount):
            # If there's still a pending block, write it out
            if self._startdoc is not None:
                self._emit()

    class Reader(ColumnReader):
        def __init__(self, dbfile, basepos, length, doccount, module):
            ColumnReader.__init__(self, dbfile, basepos, length, doccount)
            self._decompress = __import__(module).decompress

            self._blocks = []
            dbfile.seek(basepos)
            pos = 0
            while pos < length:
                startdoc, enddoc, blocklen, lengths = dbfile.read_pickle()
                here = dbfile.tell()
                self._blocks.append((startdoc, enddoc, here, blocklen,
                                     lengths))
                dbfile.seek(blocklen, 1)
                pos = here + blocklen

        def __repr__(self):
            return "<CompressedBlock.Reader>"

        def _find_block(self, docnum):
            # TODO: use binary search instead of linear
            for i, b in enumerate(self._blocks):
                if docnum < b[0]:
                    return None
                elif docnum <= b[1]:
                    return i
            return None

        def _get_block(self, blocknum):
            block = self._blocks[blocknum]
            pos = block[2]
            blocklen = block[3]
            lengths = block[4]

            data = self._decompress(self._dbfile.get(self._basepos + pos,
                                                     blocklen))
            values = {}
            base = 0
            for docnum, vlen in lengths:
                values[docnum] = data[base:base + vlen]
                base += vlen
            return values

        def __getitem__(self, docnum):
            i = self._find_block(docnum)
            if i is None:
                return emptybytes
            return self._get_block(i)[docnum]

        def __iter__(self):
            last = -1
            for i, block in enumerate(self._blocks):
                startdoc = block[0]
                enddoc = block[1]
                if startdoc > (last + 1):
                    for _ in xrange(startdoc - last):
                        yield emptybytes
                values = self._get_block(i)
                for docnum in xrange(startdoc, enddoc + 1):
                    if docnum in values:
                        yield values[docnum]
                    else:
                        yield emptybytes
                last = enddoc
            if enddoc < self._doccount - 1:
                for _ in xrange(self._doccount - enddoc):
                    yield emptybytes


class StructColumn(FixedBytesColumn):
    def __init__(self, spec, default):
        self._spec = spec
        self._fixedlen = struct.calcsize(spec)
        self._default = default

    def writer(self, dbfile):
        return self.Writer(dbfile, self._spec, self._default)

    def reader(self, dbfile, basepos, length, doccount):
        return self.Reader(dbfile, basepos, length, doccount, self._spec,
                           self._default)

    class Writer(FixedBytesColumn.Writer):
        def __init__(self, dbfile, spec, default):
            self._dbfile = dbfile
            self._struct = struct.Struct(spec)
            self._fixedlen = self._struct.size
            self._default = default
            self._defaultbytes = self._struct.pack(*default)
            self._count = 0

        def __repr__(self):
            return "<Struct.Writer>"

        def add(self, docnum, v):
            b = self._struct.pack(*v)
            FixedBytesColumn.Writer.add(self, docnum, b)

    class Reader(FixedBytesColumn.Reader):
        def __init__(self, dbfile, basepos, length, doccount, spec, default):
            self._dbfile = dbfile
            self._basepos = basepos
            self._doccount = doccount
            self._struct = struct.Struct(spec)
            self._fixedlen = self._struct.size
            self._default = default
            self._defaultbytes = self._struct.pack(*default)
            self._count = length // self._fixedlen

        def __repr__(self):
            return "<Struct.Reader>"

        def __getitem__(self, docnum):
            v = FixedBytesColumn.Reader.__getitem__(self, docnum)
            return self._struct.unpack(v)


# Utility readers

class EmptyColumnReader(ColumnReader):
    """Acts like a reader for a column with no stored values. Always returns
    the default.
    """

    def __init__(self, default, doccount):
        """
        :param default: the value to return for all "get" requests.
        :param doccount: the number of documents in the nominal column.
        """

        self._default = default
        self._doccount = doccount

    def __getitem__(self, docnum):
        return self._default

    def __iter__(self):
        return (self._default for _ in xrange(self._doccount))

    def load(self):
        return self


class MultiColumnReader(ColumnReader):
    """Serializes access to multiple column readers, making them appear to be
    one large column.
    """

    def __init__(self, readers):
        """
        :param readers: a sequence of column reader objects.
        """

        self._readers = readers

        self._doc_offsets = []
        self._doccount = 0
        for r in readers:
            self._doc_offsets.append(self._doccount)
            self._doccount += len(r)

    def _document_reader(self, docnum):
        return max(0, bisect_right(self._doc_offsets, docnum) - 1)

    def _reader_and_docnum(self, docnum):
        rnum = self._document_reader(docnum)
        offset = self._doc_offsets[rnum]
        return rnum, docnum - offset

    def __getitem__(self, docnum):
        x, y = self._reader_and_docnum(docnum)
        return self._readers[x][y]

    def __iter__(self):
        for r in self._readers:
            for v in r:
                yield v


class TranslatingColumnReader(ColumnReader):
    """Calls a function to "translate" values from an underlying column reader
    object before returning them.

    ``IndexReader`` objects can wrap a column reader with this object to call
    ``FieldType.from_column_value`` on the stored column value before returning
    it the the user.
    """

    def __init__(self, reader, translate):
        """
        :param reader: the underlying ColumnReader object to get values from.
        :param translate: a function that takes a value from the underlying
            reader and returns a translated value.
        """

        self._reader = reader
        self._translate = translate

    def raw_column(self):
        """Returns the underlying column reader.
        """

        return self._reader

    def __len__(self):
        return len(self._reader)

    def __getitem__(self, docnum):
        return self._translate(self._reader[docnum])

    def sort_key(self, docnum, reverse=False):
        return self._reader.sort_key(docnum, reverse=reverse)

    def __iter__(self):
        translate = self._translate
        return (translate(v) for v in self._reader)


# Column wrappers

class WrappedColumn(Column):
    def __init__(self, child):
        self._child = child

    def writer(self, *args, **kwargs):
        return self.Writer(self._child.writer(*args, **kwargs))

    def reader(self, *args, **kwargs):
        return self.Reader(self._child.reader(*args, **kwargs))

    def stores_lists(self):
        return self._child.stores_lists()


class WrappedColumnWriter(ColumnWriter):
    def __init__(self, child):
        self._child = child

    def fill(self, docnum):
        return self._child.fill(docnum)

    def add(self, docnum, value):
        return self._child.add(docnum, value)

    def finish(self, docnum):
        return self._child.finish(docnum)


class WrappedColumnReader(ColumnReader):
    def __init__(self, child):
        self._child = child

    def __len__(self):
        return len(self._child)

    def __getitem__(self, docnum):
        return self._child[docnum]

    def sort_key(self, docnum, reverse=False):
        return self._child.sort_key(docnum, reverse=reverse)

    def __iter__(self):
        return iter(self._child)

    def load(self):
        return list(self)


class ClampedNumericColumn(WrappedColumn):
    """An experimental wrapper type for NumericColumn that clamps out-of-range
    values instead of raising an exception.
    """

    def reader(self, *args, **kwargs):
        return self._child.reader(*args, **kwargs)

    class Writer(WrappedColumnWriter):
        def __init__(self, child):
            self._child = child
            self._min = typecode_min[child._typecode]
            self._max = typecode_max[child._typecode]

        def add(self, docnum, v):
            v = min(v, self._min)
            v = max(v, self._max)
            self._child.add(docnum, v)


class PickleColumn(WrappedColumn):
    """Converts arbitrary objects to pickled bytestrings and stores them using
    the wrapped column (usually a :class:`VarBytesColumn` or
    :class:`CompressedBytesColumn`).

    If you can express the value you want to store as a number or bytestring,
    you should use the appropriate column type to avoid the time and size
    overhead of pickling and unpickling.
    """

    class Writer(WrappedColumnWriter):
        def __repr__(self):
            return "<PickleWriter>"

        def add(self, docnum, v):
            if v is None:
                v = emptybytes
            else:
                v = dumps(v, -1)
            self._child.add(docnum, v)

    class Reader(WrappedColumnReader):
        def __repr__(self):
            return "<PickleReader>"

        def __getitem__(self, docnum):
            v = self._child[docnum]
            if not v:
                return None
            else:
                return loads(v)

        def __iter__(self):
            for v in self._child:
                if not v:
                    yield None
                else:
                    yield loads(v)


# List columns

class ListColumn(WrappedColumn):
    def stores_lists(self):
        return True


class ListColumnReader(ColumnReader):
    def sort_key(self, docnum, reverse=False):
        return self[docnum][0]

    def __iter__(self):
        for docnum in xrange(len(self)):
            yield self[docnum]


class VarBytesListColumn(ListColumn):
    def __init__(self):
        self._child = VarBytesColumn()

    class Writer(WrappedColumnWriter):
        def add(self, docnum, ls):
            out = [varint(len(ls))]
            for v in ls:
                assert isinstance(v, bytes_type)
                out.append(varint(len(v)))
                out.append(v)
            self._child.add(emptybytes.join(out))

    class Reader(WrappedColumnReader, ListColumnReader):
        def __getitem__(self, docnum):
            bio = BytesIO(self._child[docnum])
            count = bio.read_varint()
            out = []
            for _ in xrange(count):
                vlen = bio.read_varint()
                v = bio.read(vlen)
                out.append(v)
            return out


class FixedBytesListColumn(ListColumn):
    def __init__(self, fixedlen):
        self._fixedlen = fixedlen
        self._child = VarBytesColumn()

    def writer(self, *args, **kwargs):
        return self.Writer(self._child.writer(*args, **kwargs), self._fixedlen)

    def reader(self, *args, **kwargs):
        return self.Reader(self._child.reader(*args, **kwargs), self._fixedlen)

    class Writer(WrappedColumnWriter):
        def __init__(self, child, fixedlen):
            self._child = child
            self._fixedlen = fixedlen
            self._lengths = GrowableArray()
            self._count = 0

        def add(self, docnum, ls):
            out = []
            for v in ls:
                assert len(v) == self._fixedlen
                out.append(v)
            b = emptybytes.join(out)
            self._child.add(docnum, b)

    class Reader(WrappedColumnReader, ListColumnReader):
        def __init__(self, child, fixedlen):
            self._child = child
            self._fixedlen = fixedlen

        def __getitem__(self, docnum):
            fixedlen = self._fixedlen
            v = self._child[docnum]
            if not v:
                return []
            ls = [v[i:i + fixedlen] for i in xrange(0, len(v), fixedlen)]
            return ls


#class RefListColumn(Column):
#    def __init__(self, fixedlen=0):
#        """
#        :param fixedlen: an optional fixed length for the values. If you
#            specify a number other than 0, the column will require all values
#            to be the specified length.
#        :param default: a default value to use for documents that don't specify
#            one. If you don't specify a default, the column will use an empty
#            bytestring (``b''``), or if you specify a fixed length,
#            ``b'\\x00' * fixedlen``.
#        """
#
#        self._fixedlen = fixedlen
#
#    def stores_lists(self):
#        return True
#
#    def writer(self, dbfile):
#        return self.Writer(dbfile, self._fixedlen)
#
#    def reader(self, dbfile, basepos, length, doccount):
#        return self.Reader(dbfile, basepos, length, doccount, self._fixedlen)
#
#    class Writer(ColumnWriter):
#        def __init__(self, dbfile, fixedlen):
#            self._dbfile = dbfile
#            self._fixedlen = fixedlen
#
#            self._refs = GrowableArray(allow_longs=False)
#            self._lengths = GrowableArray(allow_longs=False)
#            self._count = 0
#
#        def __repr__(self):
#            return "<RefList.Writer>"
#
#        def fill(self, docnum):
#            if docnum > self._count:
#                self._lengths.extend(0 for _ in xrange(docnum - self._count))
#
#        def add(self, docnum, ls):
#            uniques = self._uniques
#            refs = self._refs
#
#            self.fill(docnum)
#            self._lengths.append(len(ls))
#            for v in ls:
#                try:
#                    i = uniques[v]
#                except KeyError:
#                    uniques[v] = i = len(uniques)
#                refs.append(i)
#
#            self._count = docnum + 1
#
#        def finish(self, doccount):
#            dbfile = self._dbfile
#            refs = self._refs.array
#            lengths = self._lengths.array
#
#            self.fill(doccount)
#            dbfile.write_byte(ord(lengths.typecode))
#            dbfile.write_array(lengths)
#            dbfile.write_byte(ord(refs.typecode))
#            self._write_uniques(refs.typecode)
#            dbfile.write_array(refs)
#
#    class Reader(ListColumnReader):
#        def __init__(self, dbfile, basepos, length, doccount, fixedlen):
#            self._dbfile = dbfile
#            self._basepos = basepos
#            self._doccount = doccount
#            self._fixedlen = fixedlen
#
#            dbfile.seek(basepos)
#            lencode = chr(dbfile.read_byte())
#            self._lengths = dbfile.read_array(lencode, doccount)
#
#            self._typecode = chr(dbfile.read_byte())
#            refst = struct.Struct("!" + self._typecode)
#            self._unpack = refst.unpack
#            self._itemsize = refst.size
#
#            self._read_uniques()
#            self._refbase = dbfile.tell()
#
#            # Create an array of offsets into the references using the lengths
#            offsets = array("i", (0,))
#            for length in self._lengths:
#                offsets.append(offsets[-1] + length)
#            self._offsets = offsets
#
#        def __repr__(self):
#            return "<RefBytes.Reader>"
#
#        def _get_ref(self, docnum):
#            pos = self._basepos + 1 + docnum * self._itemsize
#            return self._unpack(self._dbfile.get(pos, self._itemsize))[0]
#
#        def __getitem__(self, docnum):
#            offset = self._offsets[docnum]
#            length = self._lengths[docnum]
#
#            pos = self._refbase + offset * self._itemsize
#            reflist = self._dbfile.get_array(pos, self._typecode, length)
#            return [self._uniques[ref] for ref in reflist]