This file is indexed.

/usr/lib/python3/dist-packages/xlrd/book.py is in python3-xlrd 0.9.2-1.

This file is owned by root:root, with mode 0o644.

The actual contents of the file can be viewed below.

   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
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
# Copyright (c) 2005-2012 Stephen John Machin, Lingfo Pty Ltd
# This module is part of the xlrd package, which is released under a
# BSD-style licence.

from __future__ import print_function

from .timemachine import *
from .biffh import *
import struct; unpack = struct.unpack
import sys
import time
from . import sheet
from . import compdoc
from .formula import *
from . import formatting
if sys.version.startswith("IronPython"):
    # print >> sys.stderr, "...importing encodings"
    import encodings

empty_cell = sheet.empty_cell # for exposure to the world ...

DEBUG = 0

USE_FANCY_CD = 1

TOGGLE_GC = 0
import gc
# gc.set_debug(gc.DEBUG_STATS)

try:
    import mmap
    MMAP_AVAILABLE = 1
except ImportError:
    MMAP_AVAILABLE = 0
USE_MMAP = MMAP_AVAILABLE

MY_EOF = 0xF00BAAA # not a 16-bit number

SUPBOOK_UNK, SUPBOOK_INTERNAL, SUPBOOK_EXTERNAL, SUPBOOK_ADDIN, SUPBOOK_DDEOLE = range(5)

SUPPORTED_VERSIONS = (80, 70, 50, 45, 40, 30, 21, 20)

_code_from_builtin_name = {
    "Consolidate_Area": "\x00",
    "Auto_Open":        "\x01",
    "Auto_Close":       "\x02",
    "Extract":          "\x03",
    "Database":         "\x04",
    "Criteria":         "\x05",
    "Print_Area":       "\x06",
    "Print_Titles":     "\x07",
    "Recorder":         "\x08",
    "Data_Form":        "\x09",
    "Auto_Activate":    "\x0A",
    "Auto_Deactivate":  "\x0B",
    "Sheet_Title":      "\x0C",
    "_FilterDatabase":  "\x0D",
    }
builtin_name_from_code = {}
code_from_builtin_name = {}
for _bin, _bic in _code_from_builtin_name.items():
    _bin = UNICODE_LITERAL(_bin)
    _bic = UNICODE_LITERAL(_bic)
    code_from_builtin_name[_bin] = _bic
    builtin_name_from_code[_bic] = _bin
del _bin, _bic, _code_from_builtin_name

def open_workbook_xls(filename=None,
    logfile=sys.stdout, verbosity=0, use_mmap=USE_MMAP,
    file_contents=None,
    encoding_override=None,
    formatting_info=False, on_demand=False, ragged_rows=False,
    ):
    t0 = time.clock()
    if TOGGLE_GC:
        orig_gc_enabled = gc.isenabled()
        if orig_gc_enabled:
            gc.disable()
    bk = Book()
    try:
        bk.biff2_8_load(
            filename=filename, file_contents=file_contents,
            logfile=logfile, verbosity=verbosity, use_mmap=use_mmap,
            encoding_override=encoding_override,
            formatting_info=formatting_info,
            on_demand=on_demand,
            ragged_rows=ragged_rows,
            )
        t1 = time.clock()
        bk.load_time_stage_1 = t1 - t0
        biff_version = bk.getbof(XL_WORKBOOK_GLOBALS)
        if not biff_version:
            raise XLRDError("Can't determine file's BIFF version")
        if biff_version not in SUPPORTED_VERSIONS:
            raise XLRDError(
                "BIFF version %s is not supported"
                % biff_text_from_num[biff_version]
                )
        bk.biff_version = biff_version
        if biff_version <= 40:
            # no workbook globals, only 1 worksheet
            if on_demand:
                fprintf(bk.logfile,
                    "*** WARNING: on_demand is not supported for this Excel version.\n"
                    "*** Setting on_demand to False.\n")
                bk.on_demand = on_demand = False
            bk.fake_globals_get_sheet()
        elif biff_version == 45:
            # worksheet(s) embedded in global stream
            bk.parse_globals()
            if on_demand:
                fprintf(bk.logfile, "*** WARNING: on_demand is not supported for this Excel version.\n"
                                    "*** Setting on_demand to False.\n")
                bk.on_demand = on_demand = False
        else:
            bk.parse_globals()
            bk._sheet_list = [None for sh in bk._sheet_names]
            if not on_demand:
                bk.get_sheets()
        bk.nsheets = len(bk._sheet_list)
        if biff_version == 45 and bk.nsheets > 1:
            fprintf(bk.logfile,
                "*** WARNING: Excel 4.0 workbook (.XLW) file contains %d worksheets.\n"
                "*** Book-level data will be that of the last worksheet.\n",
                bk.nsheets
                )
        if TOGGLE_GC:
            if orig_gc_enabled:
                gc.enable()
        t2 = time.clock()
        bk.load_time_stage_2 = t2 - t1
    except:
        bk.release_resources()
        raise
    # normal exit
    if not on_demand:
        bk.release_resources()
    return bk

##
# For debugging: dump the file's BIFF records in char & hex.
# @param filename The path to the file to be dumped.
# @param outfile An open file, to which the dump is written.
# @param unnumbered If true, omit offsets (for meaningful diffs).

def dump(filename, outfile=sys.stdout, unnumbered=False):
    bk = Book()
    bk.biff2_8_load(filename=filename, logfile=outfile, )
    biff_dump(bk.mem, bk.base, bk.stream_len, 0, outfile, unnumbered)

##
# For debugging and analysis: summarise the file's BIFF records.
# I.e. produce a sorted file of (record_name, count).
# @param filename The path to the file to be summarised.
# @param outfile An open file, to which the summary is written.

def count_records(filename, outfile=sys.stdout):
    bk = Book()
    bk.biff2_8_load(filename=filename, logfile=outfile, )
    biff_count_records(bk.mem, bk.base, bk.stream_len, outfile)

##
# Information relating to a named reference, formula, macro, etc.
# <br />  -- New in version 0.6.0
# <br />  -- <i>Name information is <b>not</b> extracted from files older than
# Excel 5.0 (Book.biff_version < 50)</i>

class Name(BaseObject):

    _repr_these = ['stack']
    book = None # parent

    ##
    # 0 = Visible; 1 = Hidden
    hidden = 0

    ##
    # 0 = Command macro; 1 = Function macro. Relevant only if macro == 1
    func = 0

    ##
    # 0 = Sheet macro; 1 = VisualBasic macro. Relevant only if macro == 1
    vbasic = 0

    ##
    # 0 = Standard name; 1 = Macro name
    macro = 0

    ##
    # 0 = Simple formula; 1 = Complex formula (array formula or user defined)<br />
    # <i>No examples have been sighted.</i>
    complex = 0

    ##
    # 0 = User-defined name; 1 = Built-in name
    # (common examples: Print_Area, Print_Titles; see OOo docs for full list)
    builtin = 0

    ##
    # Function group. Relevant only if macro == 1; see OOo docs for values.
    funcgroup = 0

    ##
    # 0 = Formula definition; 1 = Binary data<br />  <i>No examples have been sighted.</i>
    binary = 0

    ##
    # The index of this object in book.name_obj_list
    name_index = 0

    ##
    # A Unicode string. If builtin, decoded as per OOo docs.
    name = UNICODE_LITERAL("")

    ##
    # An 8-bit string.
    raw_formula = b''

    ##
    # -1: The name is global (visible in all calculation sheets).<br />
    # -2: The name belongs to a macro sheet or VBA sheet.<br />
    # -3: The name is invalid.<br />
    # 0 <= scope < book.nsheets: The name is local to the sheet whose index is scope.
    scope = -1

    ##
    # The result of evaluating the formula, if any.
    # If no formula, or evaluation of the formula encountered problems,
    # the result is None. Otherwise the result is a single instance of the
    # Operand class.
    #
    result = None

    ##
    # This is a convenience method for the frequent use case where the name
    # refers to a single cell.
    # @return An instance of the Cell class.
    # @throws XLRDError The name is not a constant absolute reference
    # to a single cell.
    def cell(self):
        res = self.result
        if res:
            # result should be an instance of the Operand class
            kind = res.kind
            value = res.value
            if kind == oREF and len(value) == 1:
                ref3d = value[0]
                if (0 <= ref3d.shtxlo == ref3d.shtxhi - 1
                and      ref3d.rowxlo == ref3d.rowxhi - 1
                and      ref3d.colxlo == ref3d.colxhi - 1):
                    sh = self.book.sheet_by_index(ref3d.shtxlo)
                    return sh.cell(ref3d.rowxlo, ref3d.colxlo)
        self.dump(self.book.logfile,
            header="=== Dump of Name object ===",
            footer="======= End of dump =======",
            )
        raise XLRDError("Not a constant absolute reference to a single cell")

    ##
    # This is a convenience method for the use case where the name
    # refers to one rectangular area in one worksheet.
    # @param clipped If true (the default), the returned rectangle is clipped
    # to fit in (0, sheet.nrows, 0, sheet.ncols) -- it is guaranteed that
    # 0 <= rowxlo <= rowxhi <= sheet.nrows and that the number of usable rows
    # in the area (which may be zero) is rowxhi - rowxlo; likewise for columns.
    # @return a tuple (sheet_object, rowxlo, rowxhi, colxlo, colxhi).
    # @throws XLRDError The name is not a constant absolute reference
    # to a single area in a single sheet.
    def area2d(self, clipped=True):
        res = self.result
        if res:
            # result should be an instance of the Operand class
            kind = res.kind
            value = res.value
            if kind == oREF and len(value) == 1: # only 1 reference
                ref3d = value[0]
                if 0 <= ref3d.shtxlo == ref3d.shtxhi - 1: # only 1 usable sheet
                    sh = self.book.sheet_by_index(ref3d.shtxlo)
                    if not clipped:
                        return sh, ref3d.rowxlo, ref3d.rowxhi, ref3d.colxlo, ref3d.colxhi
                    rowxlo = min(ref3d.rowxlo, sh.nrows)
                    rowxhi = max(rowxlo, min(ref3d.rowxhi, sh.nrows))
                    colxlo = min(ref3d.colxlo, sh.ncols)
                    colxhi = max(colxlo, min(ref3d.colxhi, sh.ncols))
                    assert 0 <= rowxlo <= rowxhi <= sh.nrows
                    assert 0 <= colxlo <= colxhi <= sh.ncols
                    return sh, rowxlo, rowxhi, colxlo, colxhi
        self.dump(self.book.logfile,
            header="=== Dump of Name object ===",
            footer="======= End of dump =======",
            )
        raise XLRDError("Not a constant absolute reference to a single area in a single sheet")

##
# Contents of a "workbook".
# <p>WARNING: You don't call this class yourself. You use the Book object that
# was returned when you called xlrd.open_workbook("myfile.xls").</p>

class Book(BaseObject):

    ##
    # The number of worksheets present in the workbook file.
    # This information is available even when no sheets have yet been loaded.
    nsheets = 0

    ##
    # Which date system was in force when this file was last saved.<br />
    #    0 => 1900 system (the Excel for Windows default).<br />
    #    1 => 1904 system (the Excel for Macintosh default).<br />
    datemode = 0 # In case it's not specified in the file.

    ##
    # Version of BIFF (Binary Interchange File Format) used to create the file.
    # Latest is 8.0 (represented here as 80), introduced with Excel 97.
    # Earliest supported by this module: 2.0 (represented as 20).
    biff_version = 0

    ##
    # List containing a Name object for each NAME record in the workbook.
    # <br />  -- New in version 0.6.0
    name_obj_list = []

    ##
    # An integer denoting the character set used for strings in this file.
    # For BIFF 8 and later, this will be 1200, meaning Unicode; more precisely, UTF_16_LE.
    # For earlier versions, this is used to derive the appropriate Python encoding
    # to be used to convert to Unicode.
    # Examples: 1252 -> 'cp1252', 10000 -> 'mac_roman'
    codepage = None

    ##
    # The encoding that was derived from the codepage.
    encoding = None

    ##
    # A tuple containing the (telephone system) country code for:<br />
    #    [0]: the user-interface setting when the file was created.<br />
    #    [1]: the regional settings.<br />
    # Example: (1, 61) meaning (USA, Australia).
    # This information may give a clue to the correct encoding for an unknown codepage.
    # For a long list of observed values, refer to the OpenOffice.org documentation for
    # the COUNTRY record.
    countries = (0, 0)

    ##
    # What (if anything) is recorded as the name of the last user to save the file.
    user_name = UNICODE_LITERAL('')

    ##
    # A list of Font class instances, each corresponding to a FONT record.
    # <br /> -- New in version 0.6.1
    font_list = []

    ##
    # A list of XF class instances, each corresponding to an XF record.
    # <br /> -- New in version 0.6.1
    xf_list = []

    ##
    # A list of Format objects, each corresponding to a FORMAT record, in
    # the order that they appear in the input file.
    # It does <i>not</i> contain builtin formats.
    # If you are creating an output file using (for example) pyExcelerator,
    # use this list.
    # The collection to be used for all visual rendering purposes is format_map.
    # <br /> -- New in version 0.6.1
    format_list = []

    ##
    # The mapping from XF.format_key to Format object.
    # <br /> -- New in version 0.6.1
    format_map = {}

    ##
    # This provides access via name to the extended format information for
    # both built-in styles and user-defined styles.<br />
    # It maps <i>name</i> to (<i>built_in</i>, <i>xf_index</i>), where:<br />
    # <i>name</i> is either the name of a user-defined style,
    # or the name of one of the built-in styles. Known built-in names are
    # Normal, RowLevel_1 to RowLevel_7,
    # ColLevel_1 to ColLevel_7, Comma, Currency, Percent, "Comma [0]",
    # "Currency [0]", Hyperlink, and "Followed Hyperlink".<br />
    # <i>built_in</i> 1 = built-in style, 0 = user-defined<br />
    # <i>xf_index</i> is an index into Book.xf_list.<br />
    # References: OOo docs s6.99 (STYLE record); Excel UI Format/Style
    # <br /> -- New in version 0.6.1; since 0.7.4, extracted only if
    # open_workbook(..., formatting_info=True)
    style_name_map = {}

    ##
    # This provides definitions for colour indexes. Please refer to the
    # above section "The Palette; Colour Indexes" for an explanation
    # of how colours are represented in Excel.<br />
    # Colour indexes into the palette map into (red, green, blue) tuples.
    # "Magic" indexes e.g. 0x7FFF map to None.
    # <i>colour_map</i> is what you need if you want to render cells on screen or in a PDF
    # file. If you are writing an output XLS file, use <i>palette_record</i>.
    # <br /> -- New in version 0.6.1. Extracted only if open_workbook(..., formatting_info=True)
    colour_map = {}

    ##
    # If the user has changed any of the colours in the standard palette, the XLS
    # file will contain a PALETTE record with 56 (16 for Excel 4.0 and earlier)
    # RGB values in it, and this list will be e.g. [(r0, b0, g0), ..., (r55, b55, g55)].
    # Otherwise this list will be empty. This is what you need if you are
    # writing an output XLS file. If you want to render cells on screen or in a PDF
    # file, use colour_map.
    # <br /> -- New in version 0.6.1. Extracted only if open_workbook(..., formatting_info=True)
    palette_record = []

    ##
    # Time in seconds to extract the XLS image as a contiguous string (or mmap equivalent).
    load_time_stage_1 = -1.0

    ##
    # Time in seconds to parse the data from the contiguous string (or mmap equivalent).
    load_time_stage_2 = -1.0

    ##
    # @return A list of all sheets in the book.
    # All sheets not already loaded will be loaded.
    def sheets(self):
        for sheetx in xrange(self.nsheets):
            if not self._sheet_list[sheetx]:
                self.get_sheet(sheetx)
        return self._sheet_list[:]

    ##
    # @param sheetx Sheet index in range(nsheets)
    # @return An object of the Sheet class
    def sheet_by_index(self, sheetx):
        return self._sheet_list[sheetx] or self.get_sheet(sheetx)

    ##
    # @param sheet_name Name of sheet required
    # @return An object of the Sheet class
    def sheet_by_name(self, sheet_name):
        try:
            sheetx = self._sheet_names.index(sheet_name)
        except ValueError:
            raise XLRDError('No sheet named <%r>' % sheet_name)
        return self.sheet_by_index(sheetx)

    ##
    # @return A list of the names of all the worksheets in the workbook file.
    # This information is available even when no sheets have yet been loaded.
    def sheet_names(self):
        return self._sheet_names[:]

    ##
    # @param sheet_name_or_index Name or index of sheet enquired upon
    # @return true if sheet is loaded, false otherwise
    # <br />  -- New in version 0.7.1
    def sheet_loaded(self, sheet_name_or_index):
        # using type(1) because int won't work with Python 2.1
        if isinstance(sheet_name_or_index, type(1)):
            sheetx = sheet_name_or_index
        else:
            try:
                sheetx = self._sheet_names.index(sheet_name_or_index)
            except ValueError:
                raise XLRDError('No sheet named <%r>' % sheet_name_or_index)
        return self._sheet_list[sheetx] and True or False # Python 2.1 again

    ##
    # @param sheet_name_or_index Name or index of sheet to be unloaded.
    # <br />  -- New in version 0.7.1
    def unload_sheet(self, sheet_name_or_index):
        # using type(1) because int won't work with Python 2.1
        if isinstance(sheet_name_or_index, type(1)):
            sheetx = sheet_name_or_index
        else:
            try:
                sheetx = self._sheet_names.index(sheet_name_or_index)
            except ValueError:
                raise XLRDError('No sheet named <%r>' % sheet_name_or_index)
        self._sheet_list[sheetx] = None
        
    ##
    # This method has a dual purpose. You can call it to release
    # memory-consuming objects and (possibly) a memory-mapped file
    # (mmap.mmap object) when you have finished loading sheets in
    # on_demand mode, but still require the Book object to examine the
    # loaded sheets. It is also called automatically (a) when open_workbook
    # raises an exception and (b) if you are using a "with" statement, when 
    # the "with" block is exited. Calling this method multiple times on the 
    # same object has no ill effect.
    def release_resources(self):
        self._resources_released = 1
        if hasattr(self.mem, "close"):
            # must be a mmap.mmap object
            self.mem.close()
        self.mem = None
        if hasattr(self.filestr, "close"):
            self.filestr.close()
        self.filestr = None
        self._sharedstrings = None
        self._rich_text_runlist_map = None
    
    def __enter__(self):
        return self
        
    def __exit__(self, exc_type, exc_value, exc_tb):
        self.release_resources()
        # return false        

    ##
    # A mapping from (lower_case_name, scope) to a single Name object.
    # <br />  -- New in version 0.6.0
    name_and_scope_map = {}

    ##
    # A mapping from lower_case_name to a list of Name objects. The list is
    # sorted in scope order. Typically there will be one item (of global scope)
    # in the list.
    # <br />  -- New in version 0.6.0
    name_map = {}

    def __init__(self):
        self._sheet_list = []
        self._sheet_names = []
        self._sheet_visibility = [] # from BOUNDSHEET record
        self.nsheets = 0
        self._sh_abs_posn = [] # sheet's absolute position in the stream
        self._sharedstrings = []
        self._rich_text_runlist_map = {}
        self.raw_user_name = False
        self._sheethdr_count = 0 # BIFF 4W only
        self.builtinfmtcount = -1 # unknown as yet. BIFF 3, 4S, 4W
        self.initialise_format_info()
        self._all_sheets_count = 0 # includes macro & VBA sheets
        self._supbook_count = 0
        self._supbook_locals_inx = None
        self._supbook_addins_inx = None
        self._all_sheets_map = [] # maps an all_sheets index to a calc-sheets index (or -1)
        self._externsheet_info = []
        self._externsheet_type_b57 = []
        self._extnsht_name_from_num = {}
        self._sheet_num_from_name = {}
        self._extnsht_count = 0
        self._supbook_types = []
        self._resources_released = 0
        self.addin_func_names = []
        self.name_obj_list = []
        self.colour_map = {}
        self.palette_record = []
        self.xf_list = []
        self.style_name_map = {}
        self.mem = b''
        self.filestr = b''

    def biff2_8_load(self, filename=None, file_contents=None,
        logfile=sys.stdout, verbosity=0, use_mmap=USE_MMAP,
        encoding_override=None,
        formatting_info=False,
        on_demand=False,
        ragged_rows=False,
        ):
        # DEBUG = 0
        self.logfile = logfile
        self.verbosity = verbosity
        self.use_mmap = use_mmap and MMAP_AVAILABLE
        self.encoding_override = encoding_override
        self.formatting_info = formatting_info
        self.on_demand = on_demand
        self.ragged_rows = ragged_rows

        if not file_contents:
            if python_version < (2, 2) and self.use_mmap:
                # need to open for update
                open_mode = "r+b"
            else:
                open_mode = "rb"
            retry = False
            f = None
            try:
                try:
                    f = open(filename, open_mode)
                except IOError:
                    e, v = sys.exc_info()[:2]
                    if open_mode == "r+b" \
                    and (v.errno == 13 or v.strerror == "Permission denied"):
                        # Maybe the file is read-only
                        retry = True
                        self.use_mmap = False
                    else:
                        raise
                if retry:
                    f = open(filename, "rb")
                f.seek(0, 2) # EOF
                size = f.tell()
                f.seek(0, 0) # BOF
                if size == 0:
                    raise XLRDError("File size is 0 bytes")
                if self.use_mmap:
                    if python_version < (2, 2):
                        self.filestr = mmap.mmap(f.fileno(), size)
                    else:
                        self.filestr = mmap.mmap(f.fileno(), size, access=mmap.ACCESS_READ)
                    self.stream_len = size
                else:
                    self.filestr = f.read()
                    self.stream_len = len(self.filestr)
            finally:
                if f: f.close()
        else:
            self.filestr = file_contents
            self.stream_len = len(file_contents)

        self.base = 0
        if self.filestr[:8] != compdoc.SIGNATURE:
            # got this one at the antique store
            self.mem = self.filestr
        else:
            cd = compdoc.CompDoc(self.filestr, logfile=self.logfile)
            if USE_FANCY_CD:
                for qname in ['Workbook', 'Book']:
                    self.mem, self.base, self.stream_len = \
                                cd.locate_named_stream(UNICODE_LITERAL(qname))
                    if self.mem: break
                else:
                    raise XLRDError("Can't find workbook in OLE2 compound document")
            else:
                for qname in ['Workbook', 'Book']:
                    self.mem = cd.get_named_stream(UNICODE_LITERAL(qname))
                    if self.mem: break
                else:
                    raise XLRDError("Can't find workbook in OLE2 compound document")
                self.stream_len = len(self.mem)
            del cd
            if self.mem is not self.filestr:
                if hasattr(self.filestr, "close"):
                    self.filestr.close()
                self.filestr = b''
        self._position = self.base
        if DEBUG:
            print("mem: %s, base: %d, len: %d" % (type(self.mem), self.base, self.stream_len), file=self.logfile)

    def initialise_format_info(self):
        # needs to be done once per sheet for BIFF 4W :-(
        self.format_map = {}
        self.format_list = []
        self.xfcount = 0
        self.actualfmtcount = 0 # number of FORMAT records seen so far
        self._xf_index_to_xl_type_map = {0: XL_CELL_NUMBER}
        self._xf_epilogue_done = 0
        self.xf_list = []
        self.font_list = []

    def get2bytes(self):
        pos = self._position
        buff_two = self.mem[pos:pos+2]
        lenbuff = len(buff_two)
        self._position += lenbuff
        if lenbuff < 2:
            return MY_EOF
        lo, hi = buff_two
        return (BYTES_ORD(hi) << 8) | BYTES_ORD(lo)

    def get_record_parts(self):
        pos = self._position
        mem = self.mem
        code, length = unpack('<HH', mem[pos:pos+4])
        pos += 4
        data = mem[pos:pos+length]
        self._position = pos + length
        return (code, length, data)

    def get_record_parts_conditional(self, reqd_record):
        pos = self._position
        mem = self.mem
        code, length = unpack('<HH', mem[pos:pos+4])
        if code != reqd_record:
            return (None, 0, b'')
        pos += 4
        data = mem[pos:pos+length]
        self._position = pos + length
        return (code, length, data)

    def get_sheet(self, sh_number, update_pos=True):
        if self._resources_released:
            raise XLRDError("Can't load sheets after releasing resources.")
        if update_pos:
            self._position = self._sh_abs_posn[sh_number]
        _unused_biff_version = self.getbof(XL_WORKSHEET)
        # assert biff_version == self.biff_version ### FAILS
        # Have an example where book is v7 but sheet reports v8!!!
        # It appears to work OK if the sheet version is ignored.
        # Confirmed by Daniel Rentz: happens when Excel does "save as"
        # creating an old version file; ignore version details on sheet BOF.
        sh = sheet.Sheet(self,
                self._position,
                self._sheet_names[sh_number],
                sh_number,
                )
        sh.read(self)
        self._sheet_list[sh_number] = sh
        return sh

    def get_sheets(self):
        # DEBUG = 0
        if DEBUG: print("GET_SHEETS:", self._sheet_names, self._sh_abs_posn, file=self.logfile)
        for sheetno in xrange(len(self._sheet_names)):
            if DEBUG: print("GET_SHEETS: sheetno =", sheetno, self._sheet_names, self._sh_abs_posn, file=self.logfile)
            self.get_sheet(sheetno)

    def fake_globals_get_sheet(self): # for BIFF 4.0 and earlier
        formatting.initialise_book(self)
        fake_sheet_name = UNICODE_LITERAL('Sheet 1')
        self._sheet_names = [fake_sheet_name]
        self._sh_abs_posn = [0]
        self._sheet_visibility = [0] # one sheet, visible
        self._sheet_list.append(None) # get_sheet updates _sheet_list but needs a None beforehand
        self.get_sheets()

    def handle_boundsheet(self, data):
        # DEBUG = 1
        bv = self.biff_version
        self.derive_encoding()
        if DEBUG:
            fprintf(self.logfile, "BOUNDSHEET: bv=%d data %r\n", bv, data);
        if bv == 45: # BIFF4W
            #### Not documented in OOo docs ...
            # In fact, the *only* data is the name of the sheet.
            sheet_name = unpack_string(data, 0, self.encoding, lenlen=1)
            visibility = 0
            sheet_type = XL_BOUNDSHEET_WORKSHEET # guess, patch later
            if len(self._sh_abs_posn) == 0:
                abs_posn = self._sheetsoffset + self.base
                # Note (a) this won't be used
                # (b) it's the position of the SHEETHDR record
                # (c) add 11 to get to the worksheet BOF record
            else:
                abs_posn = -1 # unknown
        else:
            offset, visibility, sheet_type = unpack('<iBB', data[0:6])
            abs_posn = offset + self.base # because global BOF is always at posn 0 in the stream
            if bv < BIFF_FIRST_UNICODE:
                sheet_name = unpack_string(data, 6, self.encoding, lenlen=1)
            else:
                sheet_name = unpack_unicode(data, 6, lenlen=1)

        if DEBUG or self.verbosity >= 2:
            fprintf(self.logfile,
                "BOUNDSHEET: inx=%d vis=%r sheet_name=%r abs_posn=%d sheet_type=0x%02x\n",
                self._all_sheets_count, visibility, sheet_name, abs_posn, sheet_type)
        self._all_sheets_count += 1
        if sheet_type != XL_BOUNDSHEET_WORKSHEET:
            self._all_sheets_map.append(-1)
            descr = {
                1: 'Macro sheet',
                2: 'Chart',
                6: 'Visual Basic module',
                }.get(sheet_type, 'UNKNOWN')

            if DEBUG or self.verbosity >= 1:
                fprintf(self.logfile,
                    "NOTE *** Ignoring non-worksheet data named %r (type 0x%02x = %s)\n",
                    sheet_name, sheet_type, descr)
        else:
            snum = len(self._sheet_names)
            self._all_sheets_map.append(snum)
            self._sheet_names.append(sheet_name)
            self._sh_abs_posn.append(abs_posn)
            self._sheet_visibility.append(visibility)
            self._sheet_num_from_name[sheet_name] = snum

    def handle_builtinfmtcount(self, data):
        ### N.B. This count appears to be utterly useless.
        # DEBUG = 1
        builtinfmtcount = unpack('<H', data[0:2])[0]
        if DEBUG: fprintf(self.logfile, "BUILTINFMTCOUNT: %r\n", builtinfmtcount)
        self.builtinfmtcount = builtinfmtcount

    def derive_encoding(self):
        if self.encoding_override:
            self.encoding = self.encoding_override
        elif self.codepage is None:
            if self.biff_version < 80:
                fprintf(self.logfile,
                    "*** No CODEPAGE record, no encoding_override: will use 'ascii'\n")
                self.encoding = 'ascii'
            else:
                self.codepage = 1200 # utf16le
                if self.verbosity >= 2:
                    fprintf(self.logfile, "*** No CODEPAGE record; assuming 1200 (utf_16_le)\n")
        else:
            codepage = self.codepage
            if codepage in encoding_from_codepage:
                encoding = encoding_from_codepage[codepage]
            elif 300 <= codepage <= 1999:
                encoding = 'cp' + str(codepage)
            else:
                encoding = 'unknown_codepage_' + str(codepage)
            if DEBUG or (self.verbosity and encoding != self.encoding) :
                fprintf(self.logfile, "CODEPAGE: codepage %r -> encoding %r\n", codepage, encoding)
            self.encoding = encoding
        if self.codepage != 1200: # utf_16_le
            # If we don't have a codec that can decode ASCII into Unicode,
            # we're well & truly stuffed -- let the punter know ASAP.
            try:
                _unused = unicode(b'trial', self.encoding)
            except:
                ei = sys.exc_info()[:2]
                fprintf(self.logfile,
                    "ERROR *** codepage %r -> encoding %r -> %s: %s\n",
                    self.codepage, self.encoding, ei[0].__name__.split(".")[-1], ei[1])
                raise
        if self.raw_user_name:
            strg = unpack_string(self.user_name, 0, self.encoding, lenlen=1)
            strg = strg.rstrip()
            # if DEBUG:
            #     print "CODEPAGE: user name decoded from %r to %r" % (self.user_name, strg)
            self.user_name = strg
            self.raw_user_name = False
        return self.encoding

    def handle_codepage(self, data):
        # DEBUG = 0
        codepage = unpack('<H', data[0:2])[0]
        self.codepage = codepage
        self.derive_encoding()

    def handle_country(self, data):
        countries = unpack('<HH', data[0:4])
        if self.verbosity: print("Countries:", countries, file=self.logfile)
        # Note: in BIFF7 and earlier, country record was put (redundantly?) in each worksheet.
        assert self.countries == (0, 0) or self.countries == countries
        self.countries = countries

    def handle_datemode(self, data):
        datemode = unpack('<H', data[0:2])[0]
        if DEBUG or self.verbosity:
            fprintf(self.logfile, "DATEMODE: datemode %r\n", datemode)
        assert datemode in (0, 1)
        self.datemode = datemode

    def handle_externname(self, data):
        blah = DEBUG or self.verbosity >= 2
        if self.biff_version >= 80:
            option_flags, other_info =unpack("<HI", data[:6])
            pos = 6
            name, pos = unpack_unicode_update_pos(data, pos, lenlen=1)
            extra = data[pos:]
            if self._supbook_types[-1] == SUPBOOK_ADDIN:
                self.addin_func_names.append(name)
            if blah:
                fprintf(self.logfile,
                    "EXTERNNAME: sbktype=%d oflags=0x%04x oinfo=0x%08x name=%r extra=%r\n",
                    self._supbook_types[-1], option_flags, other_info, name, extra)

    def handle_externsheet(self, data):
        self.derive_encoding() # in case CODEPAGE record missing/out of order/wrong
        self._extnsht_count += 1 # for use as a 1-based index
        blah1 = DEBUG or self.verbosity >= 1
        blah2 = DEBUG or self.verbosity >= 2
        if self.biff_version >= 80:
            num_refs = unpack("<H", data[0:2])[0]
            bytes_reqd = num_refs * 6 + 2
            while len(data) < bytes_reqd:
                if blah1:
                    fprintf(
                        self.logfile,
                        "INFO: EXTERNSHEET needs %d bytes, have %d\n",
                        bytes_reqd, len(data),
                        )
                code2, length2, data2 = self.get_record_parts()
                if code2 != XL_CONTINUE:
                    raise XLRDError("Missing CONTINUE after EXTERNSHEET record")
                data += data2
            pos = 2
            for k in xrange(num_refs):
                info = unpack("<HHH", data[pos:pos+6])
                ref_recordx, ref_first_sheetx, ref_last_sheetx = info
                self._externsheet_info.append(info)
                pos += 6
                if blah2:
                    fprintf(
                        self.logfile,
                        "EXTERNSHEET(b8): k = %2d, record = %2d, first_sheet = %5d, last sheet = %5d\n",
                        k, ref_recordx, ref_first_sheetx, ref_last_sheetx,
                        )
        else:
            nc, ty = unpack("<BB", data[:2])
            if blah2:
                print("EXTERNSHEET(b7-):", file=self.logfile)
                hex_char_dump(data, 0, len(data), fout=self.logfile)
                msg = {
                    1: "Encoded URL",
                    2: "Current sheet!!",
                    3: "Specific sheet in own doc't",
                    4: "Nonspecific sheet in own doc't!!",
                    }.get(ty, "Not encoded")
                print("   %3d chars, type is %d (%s)" % (nc, ty, msg), file=self.logfile)
            if ty == 3:
                sheet_name = unicode(data[2:nc+2], self.encoding)
                self._extnsht_name_from_num[self._extnsht_count] = sheet_name
                if blah2: print(self._extnsht_name_from_num, file=self.logfile)
            if not (1 <= ty <= 4):
                ty = 0
            self._externsheet_type_b57.append(ty)

    def handle_filepass(self, data):
        if self.verbosity >= 2:
            logf = self.logfile
            fprintf(logf, "FILEPASS:\n")
            hex_char_dump(data, 0, len(data), base=0, fout=logf)
            if self.biff_version >= 80:
                kind1, = unpack('<H', data[:2])
                if kind1 == 0: # weak XOR encryption
                    key, hash_value = unpack('<HH', data[2:])
                    fprintf(logf,
                        'weak XOR: key=0x%04x hash=0x%04x\n',
                        key, hash_value)
                elif kind1 == 1:
                    kind2, = unpack('<H', data[4:6])
                    if kind2 == 1: # BIFF8 standard encryption
                        caption = "BIFF8 std"
                    elif kind2 == 2:
                        caption = "BIFF8 strong"
                    else:
                        caption = "** UNKNOWN ENCRYPTION METHOD **"
                    fprintf(logf, "%s\n", caption)
        raise XLRDError("Workbook is encrypted")

    def handle_name(self, data):
        blah = DEBUG or self.verbosity >= 2
        bv = self.biff_version
        if bv < 50:
            return
        self.derive_encoding()
        # print
        # hex_char_dump(data, 0, len(data), fout=self.logfile)
        (
        option_flags, kb_shortcut, name_len, fmla_len, extsht_index, sheet_index,
        menu_text_len, description_text_len, help_topic_text_len, status_bar_text_len,
        ) = unpack("<HBBHHH4B", data[0:14])
        nobj = Name()
        nobj.book = self ### CIRCULAR ###
        name_index = len(self.name_obj_list)
        nobj.name_index = name_index
        self.name_obj_list.append(nobj)
        nobj.option_flags = option_flags
        for attr, mask, nshift in (
            ('hidden', 1, 0),
            ('func', 2, 1),
            ('vbasic', 4, 2),
            ('macro', 8, 3),
            ('complex', 0x10, 4),
            ('builtin', 0x20, 5),
            ('funcgroup', 0xFC0, 6),
            ('binary', 0x1000, 12),
            ):
            setattr(nobj, attr, (option_flags & mask) >> nshift)

        macro_flag = " M"[nobj.macro]
        if bv < 80:
            internal_name, pos = unpack_string_update_pos(data, 14, self.encoding, known_len=name_len)
        else:
            internal_name, pos = unpack_unicode_update_pos(data, 14, known_len=name_len)
        nobj.extn_sheet_num = extsht_index
        nobj.excel_sheet_index = sheet_index
        nobj.scope = None # patched up in the names_epilogue() method
        if blah:
            fprintf(
                self.logfile,
                "NAME[%d]:%s oflags=%d, name_len=%d, fmla_len=%d, extsht_index=%d, sheet_index=%d, name=%r\n",
                name_index, macro_flag, option_flags, name_len,
                fmla_len, extsht_index, sheet_index, internal_name)
        name = internal_name
        if nobj.builtin:
            name = builtin_name_from_code.get(name, "??Unknown??")
            if blah: print("    builtin: %s" % name, file=self.logfile)
        nobj.name = name
        nobj.raw_formula = data[pos:]
        nobj.basic_formula_len = fmla_len
        nobj.evaluated = 0
        if blah:
            nobj.dump(
                self.logfile,
                header="--- handle_name: name[%d] ---" % name_index,
                footer="-------------------",
                )

    def names_epilogue(self):
        blah = self.verbosity >= 2
        f = self.logfile
        if blah:
            print("+++++ names_epilogue +++++", file=f)
            print("_all_sheets_map", REPR(self._all_sheets_map), file=f)
            print("_extnsht_name_from_num", REPR(self._extnsht_name_from_num), file=f)
            print("_sheet_num_from_name", REPR(self._sheet_num_from_name), file=f)
        num_names = len(self.name_obj_list)
        for namex in range(num_names):
            nobj = self.name_obj_list[namex]
            # Convert from excel_sheet_index to scope.
            # This is done here because in BIFF7 and earlier, the
            # BOUNDSHEET records (from which _all_sheets_map is derived)
            # come after the NAME records.
            if self.biff_version >= 80:
                sheet_index = nobj.excel_sheet_index
                if sheet_index == 0:
                    intl_sheet_index = -1 # global
                elif 1 <= sheet_index <= len(self._all_sheets_map):
                    intl_sheet_index = self._all_sheets_map[sheet_index-1]
                    if intl_sheet_index == -1: # maps to a macro or VBA sheet
                        intl_sheet_index = -2 # valid sheet reference but not useful
                else:
                    # huh?
                    intl_sheet_index = -3 # invalid
            elif 50 <= self.biff_version <= 70:
                sheet_index = nobj.extn_sheet_num
                if sheet_index == 0:
                    intl_sheet_index = -1 # global
                else:
                    sheet_name = self._extnsht_name_from_num[sheet_index]
                    intl_sheet_index = self._sheet_num_from_name.get(sheet_name, -2)
            nobj.scope = intl_sheet_index

        for namex in range(num_names):
            nobj = self.name_obj_list[namex]
            # Parse the formula ...
            if nobj.macro or nobj.binary: continue
            if nobj.evaluated: continue
            evaluate_name_formula(self, nobj, namex, blah=blah)

        if self.verbosity >= 2:
            print("---------- name object dump ----------", file=f)
            for namex in range(num_names):
                nobj = self.name_obj_list[namex]
                nobj.dump(f, header="--- name[%d] ---" % namex)
            print("--------------------------------------", file=f)
        #
        # Build some dicts for access to the name objects
        #
        name_and_scope_map = {} # (name.lower(), scope): Name_object
        name_map = {}           # name.lower() : list of Name_objects (sorted in scope order)
        for namex in range(num_names):
            nobj = self.name_obj_list[namex]
            name_lcase = nobj.name.lower()
            key = (name_lcase, nobj.scope)
            if key in name_and_scope_map and self.verbosity:
                fprintf(f, 'Duplicate entry %r in name_and_scope_map\n', key)
            name_and_scope_map[key] = nobj
            sort_data = (nobj.scope, namex, nobj)
            # namex (a temp unique ID) ensures the Name objects will not
            # be compared (fatal in py3)
            if name_lcase in name_map:
                name_map[name_lcase].append(sort_data)
            else:
                name_map[name_lcase] = [sort_data]
        for key in name_map.keys():
            alist = name_map[key]
            alist.sort()
            name_map[key] = [x[2] for x in alist]
        self.name_and_scope_map = name_and_scope_map
        self.name_map = name_map

    def handle_obj(self, data):
        # Not doing much handling at all.
        # Worrying about embedded (BOF ... EOF) substreams is done elsewhere.
        # DEBUG = 1
        obj_type, obj_id = unpack('<HI', data[4:10])
        # if DEBUG: print "---> handle_obj type=%d id=0x%08x" % (obj_type, obj_id)

    def handle_supbook(self, data):
        # aka EXTERNALBOOK in OOo docs
        self._supbook_types.append(None)
        blah = DEBUG or self.verbosity >= 2
        if blah:
            print("SUPBOOK:", file=self.logfile)
            hex_char_dump(data, 0, len(data), fout=self.logfile)
        num_sheets = unpack("<H", data[0:2])[0]
        if blah: print("num_sheets = %d" % num_sheets, file=self.logfile)
        sbn = self._supbook_count
        self._supbook_count += 1
        if data[2:4] == b"\x01\x04":
            self._supbook_types[-1] = SUPBOOK_INTERNAL
            self._supbook_locals_inx = self._supbook_count - 1
            if blah:
                print("SUPBOOK[%d]: internal 3D refs; %d sheets" % (sbn, num_sheets), file=self.logfile)
                print("    _all_sheets_map", self._all_sheets_map, file=self.logfile)
            return
        if data[0:4] == b"\x01\x00\x01\x3A":
            self._supbook_types[-1] = SUPBOOK_ADDIN
            self._supbook_addins_inx = self._supbook_count - 1
            if blah: print("SUPBOOK[%d]: add-in functions" % sbn, file=self.logfile)
            return
        url, pos = unpack_unicode_update_pos(data, 2, lenlen=2)
        if num_sheets == 0:
            self._supbook_types[-1] = SUPBOOK_DDEOLE
            if blah: fprintf(self.logfile, "SUPBOOK[%d]: DDE/OLE document = %r\n", sbn, url)
            return
        self._supbook_types[-1] = SUPBOOK_EXTERNAL
        if blah: fprintf(self.logfile, "SUPBOOK[%d]: url = %r\n", sbn, url)
        sheet_names = []
        for x in range(num_sheets):
            try:
                shname, pos = unpack_unicode_update_pos(data, pos, lenlen=2)
            except struct.error:
                # #### FIX ME ####
                # Should implement handling of CONTINUE record(s) ...
                if self.verbosity:
                    print((
                        "*** WARNING: unpack failure in sheet %d of %d in SUPBOOK record for file %r" 
                        % (x, num_sheets, url)
                        ), file=self.logfile)
                break
            sheet_names.append(shname)
            if blah: fprintf(self.logfile, "  sheetx=%d namelen=%d name=%r (next pos=%d)\n", x, len(shname), shname, pos)

    def handle_sheethdr(self, data):
        # This a BIFF 4W special.
        # The SHEETHDR record is followed by a (BOF ... EOF) substream containing
        # a worksheet.
        # DEBUG = 1
        self.derive_encoding()
        sheet_len = unpack('<i', data[:4])[0]
        sheet_name = unpack_string(data, 4, self.encoding, lenlen=1)
        sheetno = self._sheethdr_count
        assert sheet_name == self._sheet_names[sheetno]
        self._sheethdr_count += 1
        BOF_posn = self._position
        posn = BOF_posn - 4 - len(data)
        if DEBUG: fprintf(self.logfile, 'SHEETHDR %d at posn %d: len=%d name=%r\n', sheetno, posn, sheet_len, sheet_name)
        self.initialise_format_info()
        if DEBUG: print('SHEETHDR: xf epilogue flag is %d' % self._xf_epilogue_done, file=self.logfile)
        self._sheet_list.append(None) # get_sheet updates _sheet_list but needs a None beforehand
        self.get_sheet(sheetno, update_pos=False)
        if DEBUG: print('SHEETHDR: posn after get_sheet() =', self._position, file=self.logfile)
        self._position = BOF_posn + sheet_len

    def handle_sheetsoffset(self, data):
        # DEBUG = 0
        posn = unpack('<i', data)[0]
        if DEBUG: print('SHEETSOFFSET:', posn, file=self.logfile)
        self._sheetsoffset = posn

    def handle_sst(self, data):
        # DEBUG = 1
        if DEBUG:
            print("SST Processing", file=self.logfile)
            t0 = time.time()
        nbt = len(data)
        strlist = [data]
        uniquestrings = unpack('<i', data[4:8])[0]
        if DEBUG  or self.verbosity >= 2:
            fprintf(self.logfile, "SST: unique strings: %d\n", uniquestrings)
        while 1:
            code, nb, data = self.get_record_parts_conditional(XL_CONTINUE)
            if code is None:
                break
            nbt += nb
            if DEBUG >= 2:
                fprintf(self.logfile, "CONTINUE: adding %d bytes to SST -> %d\n", nb, nbt)
            strlist.append(data)
        self._sharedstrings, rt_runlist = unpack_SST_table(strlist, uniquestrings)
        if self.formatting_info:
            self._rich_text_runlist_map = rt_runlist        
        if DEBUG:
            t1 = time.time()
            print("SST processing took %.2f seconds" % (t1 - t0, ), file=self.logfile)

    def handle_writeaccess(self, data):
        DEBUG = 0
        if self.biff_version < 80:
            if not self.encoding:
                self.raw_user_name = True
                self.user_name = data
                return
            strg = unpack_string(data, 0, self.encoding, lenlen=1)
        else:
            strg = unpack_unicode(data, 0, lenlen=2)
        if DEBUG: fprintf(self.logfile, "WRITEACCESS: %d bytes; raw=%s %r\n", len(data), self.raw_user_name, strg)
        strg = strg.rstrip()
        self.user_name = strg

    def parse_globals(self):
        # DEBUG = 0
        # no need to position, just start reading (after the BOF)
        formatting.initialise_book(self)
        while 1:
            rc, length, data = self.get_record_parts()
            if DEBUG: print("parse_globals: record code is 0x%04x" % rc, file=self.logfile)
            if rc == XL_SST:
                self.handle_sst(data)
            elif rc == XL_FONT or rc == XL_FONT_B3B4:
                self.handle_font(data)
            elif rc == XL_FORMAT: # XL_FORMAT2 is BIFF <= 3.0, can't appear in globals
                self.handle_format(data)
            elif rc == XL_XF:
                self.handle_xf(data)
            elif rc ==  XL_BOUNDSHEET:
                self.handle_boundsheet(data)
            elif rc == XL_DATEMODE:
                self.handle_datemode(data)
            elif rc == XL_CODEPAGE:
                self.handle_codepage(data)
            elif rc == XL_COUNTRY:
                self.handle_country(data)
            elif rc == XL_EXTERNNAME:
                self.handle_externname(data)
            elif rc == XL_EXTERNSHEET:
                self.handle_externsheet(data)
            elif rc == XL_FILEPASS:
                self.handle_filepass(data)
            elif rc == XL_WRITEACCESS:
                self.handle_writeaccess(data)
            elif rc == XL_SHEETSOFFSET:
                self.handle_sheetsoffset(data)
            elif rc == XL_SHEETHDR:
                self.handle_sheethdr(data)
            elif rc == XL_SUPBOOK:
                self.handle_supbook(data)
            elif rc == XL_NAME:
                self.handle_name(data)
            elif rc == XL_PALETTE:
                self.handle_palette(data)
            elif rc == XL_STYLE:
                self.handle_style(data)
            elif rc & 0xff == 9 and self.verbosity:
                fprintf(self.logfile, "*** Unexpected BOF at posn %d: 0x%04x len=%d data=%r\n",
                    self._position - length - 4, rc, length, data)
            elif rc ==  XL_EOF:
                self.xf_epilogue()
                self.names_epilogue()
                self.palette_epilogue()
                if not self.encoding:
                    self.derive_encoding()
                if self.biff_version == 45:
                    # DEBUG = 0
                    if DEBUG: print("global EOF: position", self._position, file=self.logfile)
                    # if DEBUG:
                    #     pos = self._position - 4
                    #     print repr(self.mem[pos:pos+40])
                return
            else:
                # if DEBUG:
                #     print >> self.logfile, "parse_globals: ignoring record code 0x%04x" % rc
                pass

    def read(self, pos, length):
        data = self.mem[pos:pos+length]
        self._position = pos + len(data)
        return data

    def getbof(self, rqd_stream):
        # DEBUG = 1
        # if DEBUG: print >> self.logfile, "getbof(): position", self._position
        if DEBUG: print("reqd: 0x%04x" % rqd_stream, file=self.logfile)
        def bof_error(msg):
            raise XLRDError('Unsupported format, or corrupt file: ' + msg)
        savpos = self._position
        opcode = self.get2bytes()
        if opcode == MY_EOF:
            bof_error('Expected BOF record; met end of file')
        if opcode not in bofcodes:
            bof_error('Expected BOF record; found %r' % self.mem[savpos:savpos+8])
        length = self.get2bytes()
        if length == MY_EOF:
            bof_error('Incomplete BOF record[1]; met end of file')
        if not (4 <= length <= 20):
            bof_error(
                'Invalid length (%d) for BOF record type 0x%04x'
                % (length, opcode))
        padding = b'\0' * max(0, boflen[opcode] - length)
        data = self.read(self._position, length);
        if DEBUG: fprintf(self.logfile, "\ngetbof(): data=%r\n", data)
        if len(data) < length:
            bof_error('Incomplete BOF record[2]; met end of file')
        data += padding
        version1 = opcode >> 8
        version2, streamtype = unpack('<HH', data[0:4])
        if DEBUG:
            print("getbof(): op=0x%04x version2=0x%04x streamtype=0x%04x" \
                % (opcode, version2, streamtype), file=self.logfile)
        bof_offset = self._position - 4 - length
        if DEBUG:
            print("getbof(): BOF found at offset %d; savpos=%d" \
                % (bof_offset, savpos), file=self.logfile)
        version = build = year = 0
        if version1 == 0x08:
            build, year = unpack('<HH', data[4:8])
            if version2 == 0x0600:
                version = 80
            elif version2 == 0x0500:
                if year < 1994 or build in (2412, 3218, 3321):
                    version = 50
                else:
                    version = 70
            else:
                # dodgy one, created by a 3rd-party tool
                version = {
                    0x0000: 21,
                    0x0007: 21,
                    0x0200: 21,
                    0x0300: 30,
                    0x0400: 40,
                    }.get(version2, 0)
        elif version1 in (0x04, 0x02, 0x00):
            version = {0x04: 40, 0x02: 30, 0x00: 21}[version1]

        if version == 40 and streamtype == XL_WORKBOOK_GLOBALS_4W:
            version = 45 # i.e. 4W

        if DEBUG or self.verbosity >= 2:
            print("BOF: op=0x%04x vers=0x%04x stream=0x%04x buildid=%d buildyr=%d -> BIFF%d" \
                % (opcode, version2, streamtype, build, year, version), file=self.logfile)
        got_globals = streamtype == XL_WORKBOOK_GLOBALS or (
            version == 45 and streamtype == XL_WORKBOOK_GLOBALS_4W)
        if (rqd_stream == XL_WORKBOOK_GLOBALS and got_globals) or streamtype == rqd_stream:
            return version
        if version < 50 and streamtype == XL_WORKSHEET:
            return version
        if version >= 50 and streamtype == 0x0100:
            bof_error("Workspace file -- no spreadsheet data")
        bof_error(
            'BOF not workbook/worksheet: op=0x%04x vers=0x%04x strm=0x%04x build=%d year=%d -> BIFF%d' \
            % (opcode, version2, streamtype, build, year, version)
            )

# === helper functions

def expand_cell_address(inrow, incol):
    # Ref : OOo docs, "4.3.4 Cell Addresses in BIFF8"
    outrow = inrow
    if incol & 0x8000:
        if outrow >= 32768:
            outrow -= 65536
        relrow = 1
    else:
        relrow = 0
    outcol = incol & 0xFF
    if incol & 0x4000:
        if outcol >= 128:
            outcol -= 256
        relcol = 1
    else:
        relcol = 0
    return outrow, outcol, relrow, relcol

def colname(colx, _A2Z="ABCDEFGHIJKLMNOPQRSTUVWXYZ"):
    assert colx >= 0
    name = UNICODE_LITERAL('')
    while 1:
        quot, rem = divmod(colx, 26)
        name = _A2Z[rem] + name
        if not quot:
            return name
        colx = quot - 1

def display_cell_address(rowx, colx, relrow, relcol):
    if relrow:
        rowpart = "(*%s%d)" % ("+-"[rowx < 0], abs(rowx))
    else:
        rowpart = "$%d" % (rowx+1,)
    if relcol:
        colpart = "(*%s%d)" % ("+-"[colx < 0], abs(colx))
    else:
        colpart = "$" + colname(colx)
    return colpart + rowpart

def unpack_SST_table(datatab, nstrings):
    "Return list of strings"
    datainx = 0
    ndatas = len(datatab)
    data = datatab[0]
    datalen = len(data)
    pos = 8
    strings = []
    strappend = strings.append
    richtext_runs = {}
    local_unpack = unpack
    local_min = min
    local_BYTES_ORD = BYTES_ORD
    latin_1 = "latin_1"
    for _unused_i in xrange(nstrings):
        nchars = local_unpack('<H', data[pos:pos+2])[0]
        pos += 2
        options = local_BYTES_ORD(data[pos])
        pos += 1
        rtcount = 0
        phosz = 0
        if options & 0x08: # richtext
            rtcount = local_unpack('<H', data[pos:pos+2])[0]
            pos += 2
        if options & 0x04: # phonetic
            phosz = local_unpack('<i', data[pos:pos+4])[0]
            pos += 4
        accstrg = UNICODE_LITERAL('')
        charsgot = 0
        while 1:
            charsneed = nchars - charsgot
            if options & 0x01:
                # Uncompressed UTF-16
                charsavail = local_min((datalen - pos) >> 1, charsneed)
                rawstrg = data[pos:pos+2*charsavail]
                # if DEBUG: print "SST U16: nchars=%d pos=%d rawstrg=%r" % (nchars, pos, rawstrg)
                try:
                    accstrg += unicode(rawstrg, "utf_16_le")
                except:
                    # print "SST U16: nchars=%d pos=%d rawstrg=%r" % (nchars, pos, rawstrg)
                    # Probable cause: dodgy data e.g. unfinished surrogate pair.
                    # E.g. file unicode2.xls in pyExcelerator's examples has cells containing
                    # unichr(i) for i in range(0x100000)
                    # so this will include 0xD800 etc
                    raise
                pos += 2*charsavail
            else:
                # Note: this is COMPRESSED (not ASCII!) encoding!!!
                charsavail = local_min(datalen - pos, charsneed)
                rawstrg = data[pos:pos+charsavail]
                # if DEBUG: print "SST CMPRSD: nchars=%d pos=%d rawstrg=%r" % (nchars, pos, rawstrg)
                accstrg += unicode(rawstrg, latin_1)
                pos += charsavail
            charsgot += charsavail
            if charsgot == nchars:
                break
            datainx += 1
            data = datatab[datainx]
            datalen = len(data)
            options = local_BYTES_ORD(data[0])
            pos = 1
        
        if rtcount:
            runs = []
            for runindex in xrange(rtcount):
                if pos == datalen:
                    pos = 0
                    datainx += 1
                    data = datatab[datainx]
                    datalen = len(data)
                runs.append(local_unpack("<HH", data[pos:pos+4]))
                pos += 4
            richtext_runs[len(strings)] = runs
                
        pos += phosz # size of the phonetic stuff to skip
        if pos >= datalen:
            # adjust to correct position in next record
            pos = pos - datalen
            datainx += 1
            if datainx < ndatas:
                data = datatab[datainx]
                datalen = len(data)
            else:
                assert _unused_i == nstrings - 1
        strappend(accstrg)
    return strings, richtext_runs