This file is indexed.

/usr/include/unicode/numfmt.h is in libicu-dev 60.2-3ubuntu3.

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
// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
********************************************************************************
* Copyright (C) 1997-2016, International Business Machines Corporation and others.
* All Rights Reserved.
********************************************************************************
*
* File NUMFMT.H
*
* Modification History:
*
*   Date        Name        Description
*   02/19/97    aliu        Converted from java.
*   03/18/97    clhuang     Updated per C++ implementation.
*   04/17/97    aliu        Changed DigitCount to int per code review.
*    07/20/98    stephen        JDK 1.2 sync up. Added scientific support.
*                            Changed naming conventions to match C++ guidelines
*                            Derecated Java style constants (eg, INTEGER_FIELD)
********************************************************************************
*/

#ifndef NUMFMT_H
#define NUMFMT_H


#include "unicode/utypes.h"

/**
 * \file
 * \brief C++ API: Abstract base class for all number formats.
 */

#if !UCONFIG_NO_FORMATTING

#include "unicode/unistr.h"
#include "unicode/format.h"
#include "unicode/unum.h" // UNumberFormatStyle
#include "unicode/locid.h"
#include "unicode/stringpiece.h"
#include "unicode/curramt.h"
#include "unicode/udisplaycontext.h"

class NumberFormatTest;

U_NAMESPACE_BEGIN

class SharedNumberFormat;

#if !UCONFIG_NO_SERVICE
class NumberFormatFactory;
class StringEnumeration;
#endif

/**
 *
 * Abstract base class for all number formats.  Provides interface for
 * formatting and parsing a number.  Also provides methods for
 * determining which locales have number formats, and what their names
 * are.
 *
 * <p><strong>NOTE:</strong> Starting in ICU 60, there is a new set of APIs for localized number
 * formatting that are designed to be an improvement over DecimalFormat.  New users are discouraged
 * from using DecimalFormat.  For more information, see numberformatter.h.
 *
 * \headerfile unicode/numfmt.h "unicode/numfmt.h"
 * <P>
 * NumberFormat helps you to format and parse numbers for any locale.
 * Your code can be completely independent of the locale conventions
 * for decimal points, thousands-separators, or even the particular
 * decimal digits used, or whether the number format is even decimal.
 * <P>
 * To format a number for the current Locale, use one of the static
 * factory methods:
 * \code
 *    #include <iostream>
 *    #include "unicode/numfmt.h"
 *    #include "unicode/unistr.h"
 *    #include "unicode/ustream.h"
 *    using namespace std;
 *    
 *    int main() {
 *        double myNumber = 7.0;
 *        UnicodeString myString;
 *        UErrorCode success = U_ZERO_ERROR;
 *        NumberFormat* nf = NumberFormat::createInstance(success);
 *        nf->format(myNumber, myString);
 *        cout << " Example 1: " << myString << endl;
 *    }
 * \endcode
 * Note that there are additional factory methods within subclasses of
 * NumberFormat.
 * <P>
 * If you are formatting multiple numbers, it is more efficient to get
 * the format and use it multiple times so that the system doesn't
 * have to fetch the information about the local language and country
 * conventions multiple times.
 * \code
 *     UnicodeString myString;
 *     UErrorCode success = U_ZERO_ERROR;
 *     NumberFormat *nf = NumberFormat::createInstance( success );
 *     for (int32_t number: {123, 3333, -1234567}) {
 *         nf->format(number, myString);
 *         myString += "; ";
 *     }
 *     cout << " Example 2: " << myString << endl;
 * \endcode
 * To format a number for a different Locale, specify it in the
 * call to \c createInstance().
 * \code
 *     nf = NumberFormat::createInstance(Locale::getFrench(), success);
 * \endcode
 * You can use a \c NumberFormat to parse also.
 * \code
 *    UErrorCode success;
 *    Formattable result(-999);  // initialized with error code
 *    nf->parse(myString, result, success);
 * \endcode
 * Use \c createInstance() to get the normal number format for a \c Locale.
 * There are other static factory methods available.  Use \c createCurrencyInstance()
 * to get the currency number format for that country.  Use \c createPercentInstance()
 * to get a format for displaying percentages. With this format, a
 * fraction from 0.53 is displayed as 53%.
 * <P>
 * The type of number formatting can be specified by passing a 'style' parameter to \c createInstance().
 * For example, use\n
 * \c createInstance(locale, UNUM_DECIMAL, errorCode) to get the normal number format,\n
 * \c createInstance(locale, UNUM_PERCENT, errorCode) to get a format for displaying percentage,\n
 * \c createInstance(locale, UNUM_SCIENTIFIC, errorCode) to get a format for displaying scientific number,\n
 * \c createInstance(locale, UNUM_CURRENCY, errorCode) to get the currency number format,
 * in which the currency is represented by its symbol, for example, "$3.00".\n
 * \c createInstance(locale, UNUM_CURRENCY_ISO, errorCode)  to get the currency number format,
 * in which the currency is represented by its ISO code, for example "USD3.00".\n
 * \c createInstance(locale, UNUM_CURRENCY_PLURAL, errorCode) to get the currency number format,
 * in which the currency is represented by its full name in plural format,
 * for example, "3.00 US dollars" or "1.00 US dollar".
 * <P>
 * You can also control the display of numbers with such methods as
 * \c getMinimumFractionDigits().  If you want even more control over the
 * format or parsing, or want to give your users more control, you can
 * try dynamic_casting the \c NumberFormat you get from the factory methods to a
 * \c DecimalFormat. This will work for the vast majority of
 * countries; just remember to test for NULL in case you
 * encounter an unusual one.
 * <P>
 * You can also use forms of the parse and format methods with
 * \c ParsePosition and \c FieldPosition to allow you to:
 * <ul type=round>
 *   <li>(a) progressively parse through pieces of a string.
 *   <li>(b) align the decimal point and other areas.
 * </ul>
 * For example, you can align numbers in two ways.
 * <P>
 * If you are using a monospaced font with spacing for alignment, you
 * can pass the \c FieldPosition in your format call, with field =
 * \c UNUM_INTEGER_FIELD. On output, \c getEndIndex will be set to the offset
 * between the last character of the integer and the decimal. Add
 * (desiredSpaceCount - getEndIndex) spaces at the front of the
 * string.
 * <P>
 * If you are using proportional fonts, instead of padding with
 * spaces, measure the width of the string in pixels from the start to
 * getEndIndex.  Then move the pen by (desiredPixelWidth -
 * widthToAlignmentPoint) before drawing the text.  It also works
 * where there is no decimal, but possibly additional characters at
 * the end, e.g. with parentheses in negative numbers: "(12)" for -12.
 * <p>
 * <em>User subclasses are not supported.</em> While clients may write
 * subclasses, such code will not necessarily work and will not be
 * guaranteed to work stably from release to release.
 *
 * @stable ICU 2.0
 */
class U_I18N_API NumberFormat : public Format {
public:
    /**
     * Rounding mode.
     *
     * <p>
     * For more detail on rounding modes, see:
     * http://userguide.icu-project.org/formatparse/numbers/rounding-modes
     *
     * @stable ICU 2.4
     */
    enum ERoundingMode {
        kRoundCeiling,  /**< Round towards positive infinity */
        kRoundFloor,    /**< Round towards negative infinity */
        kRoundDown,     /**< Round towards zero */
        kRoundUp,       /**< Round away from zero */
        kRoundHalfEven, /**< Round towards the nearest integer, or
                             towards the nearest even integer if equidistant */
        kRoundHalfDown, /**< Round towards the nearest integer, or
                             towards zero if equidistant */
        kRoundHalfUp,   /**< Round towards the nearest integer, or
                             away from zero if equidistant */
        /**
          *  Return U_FORMAT_INEXACT_ERROR if number does not format exactly.
          *  @stable ICU 4.8
          */
        kRoundUnnecessary
    };

    /**
     * Alignment Field constants used to construct a FieldPosition object.
     * Signifies that the position of the integer part or fraction part of
     * a formatted number should be returned.
     *
     * Note: as of ICU 4.4, the values in this enum have been extended to
     * support identification of all number format fields, not just those
     * pertaining to alignment.
     *
     * These constants are provided for backwards compatibility only.
     * Please use the C style constants defined in the header file unum.h.
     *
     * @see FieldPosition
     * @stable ICU 2.0
     */
    enum EAlignmentFields {
        /** @stable ICU 2.0 */
        kIntegerField = UNUM_INTEGER_FIELD,
        /** @stable ICU 2.0 */
        kFractionField = UNUM_FRACTION_FIELD,
        /** @stable ICU 2.0 */
        kDecimalSeparatorField = UNUM_DECIMAL_SEPARATOR_FIELD,
        /** @stable ICU 2.0 */
        kExponentSymbolField = UNUM_EXPONENT_SYMBOL_FIELD,
        /** @stable ICU 2.0 */
        kExponentSignField = UNUM_EXPONENT_SIGN_FIELD,
        /** @stable ICU 2.0 */
        kExponentField = UNUM_EXPONENT_FIELD,
        /** @stable ICU 2.0 */
        kGroupingSeparatorField = UNUM_GROUPING_SEPARATOR_FIELD,
        /** @stable ICU 2.0 */
        kCurrencyField = UNUM_CURRENCY_FIELD,
        /** @stable ICU 2.0 */
        kPercentField = UNUM_PERCENT_FIELD,
        /** @stable ICU 2.0 */
        kPermillField = UNUM_PERMILL_FIELD,
        /** @stable ICU 2.0 */
        kSignField = UNUM_SIGN_FIELD,

    /**
     * These constants are provided for backwards compatibility only.
     * Please use the constants defined in the header file unum.h.
     */
        /** @stable ICU 2.0 */
        INTEGER_FIELD        = UNUM_INTEGER_FIELD,
        /** @stable ICU 2.0 */
        FRACTION_FIELD       = UNUM_FRACTION_FIELD
    };

    /**
     * Destructor.
     * @stable ICU 2.0
     */
    virtual ~NumberFormat();

    /**
     * Return true if the given Format objects are semantically equal.
     * Objects of different subclasses are considered unequal.
     * @return    true if the given Format objects are semantically equal.
     * @stable ICU 2.0
     */
    virtual UBool operator==(const Format& other) const;


    using Format::format;

    /**
     * Format an object to produce a string.  This method handles
     * Formattable objects with numeric types. If the Formattable
     * object type is not a numeric type, then it returns a failing
     * UErrorCode.
     *
     * @param obj       The object to format.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @param pos       On input: an alignment field, if desired.
     *                  On output: the offsets of the alignment field.
     * @param status    Output param filled with success/failure status.
     * @return          Reference to 'appendTo' parameter.
     * @stable ICU 2.0
     */
    virtual UnicodeString& format(const Formattable& obj,
                                  UnicodeString& appendTo,
                                  FieldPosition& pos,
                                  UErrorCode& status) const;

    /**
     * Format an object to produce a string.  This method handles
     * Formattable objects with numeric types. If the Formattable
     * object type is not a numeric type, then it returns a failing
     * UErrorCode.
     *
     * @param obj       The object to format.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @param posIter   On return, can be used to iterate over positions
     *                  of fields generated by this format call.  Can be
     *                  NULL.
     * @param status    Output param filled with success/failure status.
     * @return          Reference to 'appendTo' parameter.
     * @stable ICU 4.4
     */
    virtual UnicodeString& format(const Formattable& obj,
                                  UnicodeString& appendTo,
                                  FieldPositionIterator* posIter,
                                  UErrorCode& status) const;

    /**
     * Parse a string to produce an object.  This methods handles
     * parsing of numeric strings into Formattable objects with numeric
     * types.
     * <P>
     * Before calling, set parse_pos.index to the offset you want to
     * start parsing at in the source. After calling, parse_pos.index
     * indicates the position after the successfully parsed text.  If
     * an error occurs, parse_pos.index is unchanged.
     * <P>
     * When parsing, leading whitespace is discarded (with successful
     * parse), while trailing whitespace is left as is.
     * <P>
     * See Format::parseObject() for more.
     *
     * @param source    The string to be parsed into an object.
     * @param result    Formattable to be set to the parse result.
     *                  If parse fails, return contents are undefined.
     * @param parse_pos The position to start parsing at. Upon return
     *                  this param is set to the position after the
     *                  last character successfully parsed. If the
     *                  source is not parsed successfully, this param
     *                  will remain unchanged.
     * @return          A newly created Formattable* object, or NULL
     *                  on failure.  The caller owns this and should
     *                  delete it when done.
     * @stable ICU 2.0
     */
    virtual void parseObject(const UnicodeString& source,
                             Formattable& result,
                             ParsePosition& parse_pos) const;

    /**
     * Format a double number. These methods call the NumberFormat
     * pure virtual format() methods with the default FieldPosition.
     *
     * @param number    The value to be formatted.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @return          Reference to 'appendTo' parameter.
     * @stable ICU 2.0
     */
    UnicodeString& format(  double number,
                            UnicodeString& appendTo) const;

    /**
     * Format a long number. These methods call the NumberFormat
     * pure virtual format() methods with the default FieldPosition.
     *
     * @param number    The value to be formatted.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @return          Reference to 'appendTo' parameter.
     * @stable ICU 2.0
     */
    UnicodeString& format(  int32_t number,
                            UnicodeString& appendTo) const;

    /**
     * Format an int64 number. These methods call the NumberFormat
     * pure virtual format() methods with the default FieldPosition.
     *
     * @param number    The value to be formatted.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @return          Reference to 'appendTo' parameter.
     * @stable ICU 2.8
     */
    UnicodeString& format(  int64_t number,
                            UnicodeString& appendTo) const;

    /**
     * Format a double number. Concrete subclasses must implement
     * these pure virtual methods.
     *
     * @param number    The value to be formatted.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @param pos       On input: an alignment field, if desired.
     *                  On output: the offsets of the alignment field.
     * @return          Reference to 'appendTo' parameter.
     * @stable ICU 2.0
     */
    virtual UnicodeString& format(double number,
                                  UnicodeString& appendTo,
                                  FieldPosition& pos) const = 0;
    /**
     * Format a double number. By default, the parent function simply
     * calls the base class and does not return an error status.
     * Therefore, the status may be ignored in some subclasses.
     *
     * @param number    The value to be formatted.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @param pos       On input: an alignment field, if desired.
     *                  On output: the offsets of the alignment field.
     * @param status    error status
     * @return          Reference to 'appendTo' parameter.
     * @internal
     */
    virtual UnicodeString& format(double number,
                                  UnicodeString& appendTo,
                                  FieldPosition& pos,
                                  UErrorCode &status) const;
    /**
     * Format a double number. Subclasses must implement
     * this method.
     *
     * @param number    The value to be formatted.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @param posIter   On return, can be used to iterate over positions
     *                  of fields generated by this format call.
     *                  Can be NULL.
     * @param status    Output param filled with success/failure status.
     * @return          Reference to 'appendTo' parameter.
     * @stable ICU 4.4
     */
    virtual UnicodeString& format(double number,
                                  UnicodeString& appendTo,
                                  FieldPositionIterator* posIter,
                                  UErrorCode& status) const;
    /**
     * Format a long number. Concrete subclasses must implement
     * these pure virtual methods.
     *
     * @param number    The value to be formatted.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @param pos       On input: an alignment field, if desired.
     *                  On output: the offsets of the alignment field.
     * @return          Reference to 'appendTo' parameter.
     * @stable ICU 2.0
    */
    virtual UnicodeString& format(int32_t number,
                                  UnicodeString& appendTo,
                                  FieldPosition& pos) const = 0;

    /**
     * Format a long number. Concrete subclasses may override
     * this function to provide status return.
     *
     * @param number    The value to be formatted.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @param pos       On input: an alignment field, if desired.
     *                  On output: the offsets of the alignment field.
     * @param status the output status.
     * @return          Reference to 'appendTo' parameter.
     * @internal
    */
    virtual UnicodeString& format(int32_t number,
                                  UnicodeString& appendTo,
                                  FieldPosition& pos,
                                  UErrorCode &status) const;

    /**
     * Format an int32 number. Subclasses must implement
     * this method.
     *
     * @param number    The value to be formatted.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @param posIter   On return, can be used to iterate over positions
     *                  of fields generated by this format call.
     *                  Can be NULL.
     * @param status    Output param filled with success/failure status.
     * @return          Reference to 'appendTo' parameter.
     * @stable ICU 4.4
     */
    virtual UnicodeString& format(int32_t number,
                                  UnicodeString& appendTo,
                                  FieldPositionIterator* posIter,
                                  UErrorCode& status) const;
    /**
     * Format an int64 number. (Not abstract to retain compatibility
     * with earlier releases, however subclasses should override this
     * method as it just delegates to format(int32_t number...);
     *
     * @param number    The value to be formatted.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @param pos       On input: an alignment field, if desired.
     *                  On output: the offsets of the alignment field.
     * @return          Reference to 'appendTo' parameter.
     * @stable ICU 2.8
    */
    virtual UnicodeString& format(int64_t number,
                                  UnicodeString& appendTo,
                                  FieldPosition& pos) const;

    /**
     * Format an int64 number. (Not abstract to retain compatibility
     * with earlier releases, however subclasses should override this
     * method as it just delegates to format(int32_t number...);
     *
     * @param number    The value to be formatted.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @param pos       On input: an alignment field, if desired.
     *                  On output: the offsets of the alignment field.
     * @param status    Output param filled with success/failure status.
     * @return          Reference to 'appendTo' parameter.
     * @internal
    */
    virtual UnicodeString& format(int64_t number,
                                  UnicodeString& appendTo,
                                  FieldPosition& pos,
                                  UErrorCode& status) const;
    /**
     * Format an int64 number. Subclasses must implement
     * this method.
     *
     * @param number    The value to be formatted.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @param posIter   On return, can be used to iterate over positions
     *                  of fields generated by this format call.
     *                  Can be NULL.
     * @param status    Output param filled with success/failure status.
     * @return          Reference to 'appendTo' parameter.
     * @stable ICU 4.4
     */
    virtual UnicodeString& format(int64_t number,
                                  UnicodeString& appendTo,
                                  FieldPositionIterator* posIter,
                                  UErrorCode& status) const;

    /**
     * Format a decimal number. Subclasses must implement
     * this method.  The syntax of the unformatted number is a "numeric string"
     * as defined in the Decimal Arithmetic Specification, available at
     * http://speleotrove.com/decimal
     *
     * @param number    The unformatted number, as a string, to be formatted.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @param posIter   On return, can be used to iterate over positions
     *                  of fields generated by this format call.
     *                  Can be NULL.
     * @param status    Output param filled with success/failure status.
     * @return          Reference to 'appendTo' parameter.
     * @stable ICU 4.4
     */
    virtual UnicodeString& format(StringPiece number,
                                  UnicodeString& appendTo,
                                  FieldPositionIterator* posIter,
                                  UErrorCode& status) const;
public:
    /**
     * Format a decimal number.
     * The number is a DigitList wrapper onto a floating point decimal number.
     * The default implementation in NumberFormat converts the decimal number
     * to a double and formats that.  Subclasses of NumberFormat that want
     * to specifically handle big decimal numbers must override this method.
     * class DecimalFormat does so.
     *
     * @param number    The number, a DigitList format Decimal Floating Point.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @param posIter   On return, can be used to iterate over positions
     *                  of fields generated by this format call.
     * @param status    Output param filled with success/failure status.
     * @return          Reference to 'appendTo' parameter.
     * @internal
     */
    virtual UnicodeString& format(const DigitList &number,
                                  UnicodeString& appendTo,
                                  FieldPositionIterator* posIter,
                                  UErrorCode& status) const;

    /**
     * Format a decimal number.
     * The number is a DigitList wrapper onto a floating point decimal number.
     * The default implementation in NumberFormat converts the decimal number
     * to a double and formats that.  Subclasses of NumberFormat that want
     * to specifically handle big decimal numbers must override this method.
     * class DecimalFormat does so.
     *
     * @param number    The number, a DigitList format Decimal Floating Point.
     * @param appendTo  Output parameter to receive result.
     *                  Result is appended to existing contents.
     * @param pos       On input: an alignment field, if desired.
     *                  On output: the offsets of the alignment field.
     * @param status    Output param filled with success/failure status.
     * @return          Reference to 'appendTo' parameter.
     * @internal
     */
    virtual UnicodeString& format(const DigitList &number,
                                  UnicodeString& appendTo,
                                  FieldPosition& pos,
                                  UErrorCode& status) const;

public:

   /**
    * Return a long if possible (e.g. within range LONG_MAX,
    * LONG_MAX], and with no decimals), otherwise a double.  If
    * IntegerOnly is set, will stop at a decimal point (or equivalent;
    * e.g. for rational numbers "1 2/3", will stop after the 1).
    * <P>
    * If no object can be parsed, index is unchanged, and NULL is
    * returned.
    * <P>
    * This is a pure virtual which concrete subclasses must implement.
    *
    * @param text           The text to be parsed.
    * @param result         Formattable to be set to the parse result.
    *                       If parse fails, return contents are undefined.
    * @param parsePosition  The position to start parsing at on input.
    *                       On output, moved to after the last successfully
    *                       parse character. On parse failure, does not change.
    * @stable ICU 2.0
    */
    virtual void parse(const UnicodeString& text,
                       Formattable& result,
                       ParsePosition& parsePosition) const = 0;

    /**
     * Parse a string as a numeric value, and return a Formattable
     * numeric object. This method parses integers only if IntegerOnly
     * is set.
     *
     * @param text          The text to be parsed.
     * @param result        Formattable to be set to the parse result.
     *                      If parse fails, return contents are undefined.
     * @param status        Output parameter set to a failure error code
     *                      when a failure occurs.
     * @see                 NumberFormat::isParseIntegerOnly
     * @stable ICU 2.0
     */
    virtual void parse(const UnicodeString& text,
                       Formattable& result,
                       UErrorCode& status) const;

    /**
     * Parses text from the given string as a currency amount.  Unlike
     * the parse() method, this method will attempt to parse a generic
     * currency name, searching for a match of this object's locale's
     * currency display names, or for a 3-letter ISO currency code.
     * This method will fail if this format is not a currency format,
     * that is, if it does not contain the currency pattern symbol
     * (U+00A4) in its prefix or suffix.
     *
     * @param text the string to parse
     * @param pos  input-output position; on input, the position within text
     *             to match; must have 0 <= pos.getIndex() < text.length();
     *             on output, the position after the last matched character.
     *             If the parse fails, the position in unchanged upon output.
     * @return     if parse succeeds, a pointer to a newly-created CurrencyAmount
     *             object (owned by the caller) containing information about
     *             the parsed currency; if parse fails, this is NULL.
     * @stable ICU 49
     */
    virtual CurrencyAmount* parseCurrency(const UnicodeString& text,
                                          ParsePosition& pos) const;

    /**
     * Return true if this format will parse numbers as integers
     * only.  For example in the English locale, with ParseIntegerOnly
     * true, the string "1234." would be parsed as the integer value
     * 1234 and parsing would stop at the "." character.  Of course,
     * the exact format accepted by the parse operation is locale
     * dependant and determined by sub-classes of NumberFormat.
     * @return    true if this format will parse numbers as integers
     *            only.
     * @stable ICU 2.0
     */
    UBool isParseIntegerOnly(void) const;

    /**
     * Sets whether or not numbers should be parsed as integers only.
     * @param value    set True, this format will parse numbers as integers
     *                 only.
     * @see isParseIntegerOnly
     * @stable ICU 2.0
     */
    virtual void setParseIntegerOnly(UBool value);

    /**
     * Sets whether lenient parsing should be enabled (it is off by default).
     *
     * @param enable \c TRUE if lenient parsing should be used,
     *               \c FALSE otherwise.
     * @stable ICU 4.8
     */
    virtual void setLenient(UBool enable);

    /**
     * Returns whether lenient parsing is enabled (it is off by default).
     *
     * @return \c TRUE if lenient parsing is enabled,
     *         \c FALSE otherwise.
     * @see #setLenient
     * @stable ICU 4.8
     */
    virtual UBool isLenient(void) const;

    /**
     * Create a default style NumberFormat for the current default locale.
     * The default formatting style is locale dependent.
     * @stable ICU 2.0
     */
    static NumberFormat* U_EXPORT2 createInstance(UErrorCode&);

    /**
     * Create a default style NumberFormat for the specified locale.
     * The default formatting style is locale dependent.
     * @param inLocale    the given locale.
     * @stable ICU 2.0
     */
    static NumberFormat* U_EXPORT2 createInstance(const Locale& inLocale,
                                        UErrorCode&);

    /**
     * Create a specific style NumberFormat for the specified locale.
     * @param desiredLocale    the given locale.
     * @param style            the given style.
     * @param errorCode        Output param filled with success/failure status.
     * @return                 A new NumberFormat instance.
     * @stable ICU 4.8
     */
    static NumberFormat* U_EXPORT2 createInstance(const Locale& desiredLocale,
                                                  UNumberFormatStyle style,
                                                  UErrorCode& errorCode);

#ifndef U_HIDE_INTERNAL_API

    /**
     * ICU use only.
     * Creates NumberFormat instance without using the cache.
     * @internal
     */
    static NumberFormat* internalCreateInstance(
            const Locale& desiredLocale,
            UNumberFormatStyle style,
            UErrorCode& errorCode);

    /**
     * ICU use only.
     * Returns handle to the shared, cached NumberFormat instance for given
     * locale. On success, caller must call removeRef() on returned value
     * once it is done with the shared instance.
     * @internal
     */
    static const SharedNumberFormat* U_EXPORT2 createSharedInstance(
            const Locale& inLocale, UNumberFormatStyle style, UErrorCode& status);

#endif  /* U_HIDE_INTERNAL_API */

    /**
     * Returns a currency format for the current default locale.
     * @stable ICU 2.0
     */
    static NumberFormat* U_EXPORT2 createCurrencyInstance(UErrorCode&);

    /**
     * Returns a currency format for the specified locale.
     * @param inLocale    the given locale.
     * @stable ICU 2.0
     */
    static NumberFormat* U_EXPORT2 createCurrencyInstance(const Locale& inLocale,
                                                UErrorCode&);

    /**
     * Returns a percentage format for the current default locale.
     * @stable ICU 2.0
     */
    static NumberFormat* U_EXPORT2 createPercentInstance(UErrorCode&);

    /**
     * Returns a percentage format for the specified locale.
     * @param inLocale    the given locale.
     * @stable ICU 2.0
     */
    static NumberFormat* U_EXPORT2 createPercentInstance(const Locale& inLocale,
                                               UErrorCode&);

    /**
     * Returns a scientific format for the current default locale.
     * @stable ICU 2.0
     */
    static NumberFormat* U_EXPORT2 createScientificInstance(UErrorCode&);

    /**
     * Returns a scientific format for the specified locale.
     * @param inLocale    the given locale.
     * @stable ICU 2.0
     */
    static NumberFormat* U_EXPORT2 createScientificInstance(const Locale& inLocale,
                                                UErrorCode&);

    /**
     * Get the set of Locales for which NumberFormats are installed.
     * @param count    Output param to receive the size of the locales
     * @stable ICU 2.0
     */
    static const Locale* U_EXPORT2 getAvailableLocales(int32_t& count);

#if !UCONFIG_NO_SERVICE
    /**
     * Register a new NumberFormatFactory.  The factory will be adopted.
     * Because ICU may choose to cache NumberFormat objects internally,
     * this must be called at application startup, prior to any calls to
     * NumberFormat::createInstance to avoid undefined behavior.
     * @param toAdopt the NumberFormatFactory instance to be adopted
     * @param status the in/out status code, no special meanings are assigned
     * @return a registry key that can be used to unregister this factory
     * @stable ICU 2.6
     */
    static URegistryKey U_EXPORT2 registerFactory(NumberFormatFactory* toAdopt, UErrorCode& status);

    /**
     * Unregister a previously-registered NumberFormatFactory using the key returned from the
     * register call.  Key becomes invalid after a successful call and should not be used again.
     * The NumberFormatFactory corresponding to the key will be deleted.
     * Because ICU may choose to cache NumberFormat objects internally,
     * this should be called during application shutdown, after all calls to
     * NumberFormat::createInstance to avoid undefined behavior.
     * @param key the registry key returned by a previous call to registerFactory
     * @param status the in/out status code, no special meanings are assigned
     * @return TRUE if the factory for the key was successfully unregistered
     * @stable ICU 2.6
     */
    static UBool U_EXPORT2 unregister(URegistryKey key, UErrorCode& status);

    /**
     * Return a StringEnumeration over the locales available at the time of the call,
     * including registered locales.
     * @return a StringEnumeration over the locales available at the time of the call
     * @stable ICU 2.6
     */
    static StringEnumeration* U_EXPORT2 getAvailableLocales(void);
#endif /* UCONFIG_NO_SERVICE */

    /**
     * Returns true if grouping is used in this format. For example,
     * in the English locale, with grouping on, the number 1234567
     * might be formatted as "1,234,567". The grouping separator as
     * well as the size of each group is locale dependent and is
     * determined by sub-classes of NumberFormat.
     * @see setGroupingUsed
     * @stable ICU 2.0
     */
    UBool isGroupingUsed(void) const;

    /**
     * Set whether or not grouping will be used in this format.
     * @param newValue    True, grouping will be used in this format.
     * @see getGroupingUsed
     * @stable ICU 2.0
     */
    virtual void setGroupingUsed(UBool newValue);

    /**
     * Returns the maximum number of digits allowed in the integer portion of a
     * number.
     * @return     the maximum number of digits allowed in the integer portion of a
     *             number.
     * @see setMaximumIntegerDigits
     * @stable ICU 2.0
     */
    int32_t getMaximumIntegerDigits(void) const;

    /**
     * Sets the maximum number of digits allowed in the integer portion of a
     * number. maximumIntegerDigits must be >= minimumIntegerDigits.  If the
     * new value for maximumIntegerDigits is less than the current value
     * of minimumIntegerDigits, then minimumIntegerDigits will also be set to
     * the new value.
     *
     * @param newValue    the new value for the maximum number of digits
     *                    allowed in the integer portion of a number.
     * @see getMaximumIntegerDigits
     * @stable ICU 2.0
     */
    virtual void setMaximumIntegerDigits(int32_t newValue);

    /**
     * Returns the minimum number of digits allowed in the integer portion of a
     * number.
     * @return    the minimum number of digits allowed in the integer portion of a
     *            number.
     * @see setMinimumIntegerDigits
     * @stable ICU 2.0
     */
    int32_t getMinimumIntegerDigits(void) const;

    /**
     * Sets the minimum number of digits allowed in the integer portion of a
     * number. minimumIntegerDigits must be &lt;= maximumIntegerDigits.  If the
     * new value for minimumIntegerDigits exceeds the current value
     * of maximumIntegerDigits, then maximumIntegerDigits will also be set to
     * the new value.
     * @param newValue    the new value to be set.
     * @see getMinimumIntegerDigits
     * @stable ICU 2.0
     */
    virtual void setMinimumIntegerDigits(int32_t newValue);

    /**
     * Returns the maximum number of digits allowed in the fraction portion of a
     * number.
     * @return    the maximum number of digits allowed in the fraction portion of a
     *            number.
     * @see setMaximumFractionDigits
     * @stable ICU 2.0
     */
    int32_t getMaximumFractionDigits(void) const;

    /**
     * Sets the maximum number of digits allowed in the fraction portion of a
     * number. maximumFractionDigits must be >= minimumFractionDigits.  If the
     * new value for maximumFractionDigits is less than the current value
     * of minimumFractionDigits, then minimumFractionDigits will also be set to
     * the new value.
     * @param newValue    the new value to be set.
     * @see getMaximumFractionDigits
     * @stable ICU 2.0
     */
    virtual void setMaximumFractionDigits(int32_t newValue);

    /**
     * Returns the minimum number of digits allowed in the fraction portion of a
     * number.
     * @return    the minimum number of digits allowed in the fraction portion of a
     *            number.
     * @see setMinimumFractionDigits
     * @stable ICU 2.0
     */
    int32_t getMinimumFractionDigits(void) const;

    /**
     * Sets the minimum number of digits allowed in the fraction portion of a
     * number. minimumFractionDigits must be &lt;= maximumFractionDigits.   If the
     * new value for minimumFractionDigits exceeds the current value
     * of maximumFractionDigits, then maximumIntegerDigits will also be set to
     * the new value
     * @param newValue    the new value to be set.
     * @see getMinimumFractionDigits
     * @stable ICU 2.0
     */
    virtual void setMinimumFractionDigits(int32_t newValue);

    /**
     * Sets the currency used to display currency
     * amounts.  This takes effect immediately, if this format is a
     * currency format.  If this format is not a currency format, then
     * the currency is used if and when this object becomes a
     * currency format.
     * @param theCurrency a 3-letter ISO code indicating new currency
     * to use.  It need not be null-terminated.  May be the empty
     * string or NULL to indicate no currency.
     * @param ec input-output error code
     * @stable ICU 3.0
     */
    virtual void setCurrency(const char16_t* theCurrency, UErrorCode& ec);

    /**
     * Gets the currency used to display currency
     * amounts.  This may be an empty string for some subclasses.
     * @return a 3-letter null-terminated ISO code indicating
     * the currency in use, or a pointer to the empty string.
     * @stable ICU 2.6
     */
    const char16_t* getCurrency() const;
	
    /**
     * Set a particular UDisplayContext value in the formatter, such as
     * UDISPCTX_CAPITALIZATION_FOR_STANDALONE.
     * @param value The UDisplayContext value to set.
     * @param status Input/output status. If at entry this indicates a failure
     *               status, the function will do nothing; otherwise this will be
     *               updated with any new status from the function.
     * @stable ICU 53
     */
    virtual void setContext(UDisplayContext value, UErrorCode& status);

    /**
     * Get the formatter's UDisplayContext value for the specified UDisplayContextType,
     * such as UDISPCTX_TYPE_CAPITALIZATION.
     * @param type The UDisplayContextType whose value to return
     * @param status Input/output status. If at entry this indicates a failure
     *               status, the function will do nothing; otherwise this will be
     *               updated with any new status from the function.
     * @return The UDisplayContextValue for the specified type.
     * @stable ICU 53
     */
    virtual UDisplayContext getContext(UDisplayContextType type, UErrorCode& status) const;

    /**
     * Get the rounding mode. This will always return NumberFormat::ERoundingMode::kRoundUnnecessary
     * if the subclass does not support rounding. 
     * @return A rounding mode
     * @draft ICU 60
     */
    virtual ERoundingMode getRoundingMode(void) const;

    /**
     * Set the rounding mode. If a subclass does not support rounding, this will do nothing.
     * @param roundingMode A rounding mode
     * @draft ICU 60
     */
    virtual void setRoundingMode(ERoundingMode roundingMode);

public:

    /**
     * Return the class ID for this class.  This is useful for
     * comparing to a return value from getDynamicClassID(). Note that,
     * because NumberFormat is an abstract base class, no fully constructed object
     * will have the class ID returned by NumberFormat::getStaticClassID().
     * @return The class ID for all objects of this class.
     * @stable ICU 2.0
     */
    static UClassID U_EXPORT2 getStaticClassID(void);

    /**
     * Returns a unique class ID POLYMORPHICALLY.  Pure virtual override.
     * This method is to implement a simple version of RTTI, since not all
     * C++ compilers support genuine RTTI.  Polymorphic operator==() and
     * clone() methods call this method.
     * <P>
     * @return The class ID for this object. All objects of a
     * given class have the same class ID.  Objects of
     * other classes have different class IDs.
     * @stable ICU 2.0
     */
    virtual UClassID getDynamicClassID(void) const = 0;

protected:

    /**
     * Default constructor for subclass use only.
     * @stable ICU 2.0
     */
    NumberFormat();

    /**
     * Copy constructor.
     * @stable ICU 2.0
     */
    NumberFormat(const NumberFormat&);

    /**
     * Assignment operator.
     * @stable ICU 2.0
     */
    NumberFormat& operator=(const NumberFormat&);

    /**
     * Returns the currency in effect for this formatter.  Subclasses
     * should override this method as needed.  Unlike getCurrency(),
     * this method should never return "".
     * @result output parameter for null-terminated result, which must
     * have a capacity of at least 4
     * @internal
     */
    virtual void getEffectiveCurrency(char16_t* result, UErrorCode& ec) const;

#ifndef U_HIDE_INTERNAL_API
    /**
     * Creates the specified number format style of the desired locale.
     * If mustBeDecimalFormat is TRUE, then the returned pointer is
     * either a DecimalFormat or it is NULL.
     * @internal
     */
    static NumberFormat* makeInstance(const Locale& desiredLocale,
                                      UNumberFormatStyle style,
                                      UBool mustBeDecimalFormat,
                                      UErrorCode& errorCode);
#endif  /* U_HIDE_INTERNAL_API */

private:

    static UBool isStyleSupported(UNumberFormatStyle style);

    /**
     * Creates the specified decimal format style of the desired locale.
     * @param desiredLocale    the given locale.
     * @param style            the given style.
     * @param errorCode        Output param filled with success/failure status.
     * @return                 A new NumberFormat instance.
     */
    static NumberFormat* makeInstance(const Locale& desiredLocale,
                                      UNumberFormatStyle style,
                                      UErrorCode& errorCode);

    UBool       fGroupingUsed;
    int32_t     fMaxIntegerDigits;
    int32_t     fMinIntegerDigits;
    int32_t     fMaxFractionDigits;
    int32_t     fMinFractionDigits;

  protected:
    /** \internal */
    static const int32_t gDefaultMaxIntegerDigits;
    /** \internal */
    static const int32_t gDefaultMinIntegerDigits;

  private:
    UBool      fParseIntegerOnly;
    UBool      fLenient; // TRUE => lenient parse is enabled

    // ISO currency code
    char16_t      fCurrency[4];

    UDisplayContext fCapitalizationContext;

    friend class ICUNumberFormatFactory; // access to makeInstance
    friend class ICUNumberFormatService;
    friend class ::NumberFormatTest;  // access to isStyleSupported()
};

#if !UCONFIG_NO_SERVICE
/**
 * A NumberFormatFactory is used to register new number formats.  The factory
 * should be able to create any of the predefined formats for each locale it
 * supports.  When registered, the locales it supports extend or override the
 * locale already supported by ICU.
 *
 * @stable ICU 2.6
 */
class U_I18N_API NumberFormatFactory : public UObject {
public:

    /**
     * Destructor
     * @stable ICU 3.0
     */
    virtual ~NumberFormatFactory();

    /**
     * Return true if this factory will be visible.  Default is true.
     * If not visible, the locales supported by this factory will not
     * be listed by getAvailableLocales.
     * @stable ICU 2.6
     */
    virtual UBool visible(void) const = 0;

    /**
     * Return the locale names directly supported by this factory.  The number of names
     * is returned in count;
     * @stable ICU 2.6
     */
    virtual const UnicodeString * getSupportedIDs(int32_t &count, UErrorCode& status) const = 0;

    /**
     * Return a number format of the appropriate type.  If the locale
     * is not supported, return null.  If the locale is supported, but
     * the type is not provided by this service, return null.  Otherwise
     * return an appropriate instance of NumberFormat.
     * @stable ICU 2.6
     */
    virtual NumberFormat* createFormat(const Locale& loc, UNumberFormatStyle formatType) = 0;
};

/**
 * A NumberFormatFactory that supports a single locale.  It can be visible or invisible.
 * @stable ICU 2.6
 */
class U_I18N_API SimpleNumberFormatFactory : public NumberFormatFactory {
protected:
    /**
     * True if the locale supported by this factory is visible.
     * @stable ICU 2.6
     */
    const UBool _visible;

    /**
     * The locale supported by this factory, as a UnicodeString.
     * @stable ICU 2.6
     */
    UnicodeString _id;

public:
    /**
     * @stable ICU 2.6
     */
    SimpleNumberFormatFactory(const Locale& locale, UBool visible = TRUE);

    /**
     * @stable ICU 3.0
     */
    virtual ~SimpleNumberFormatFactory();

    /**
     * @stable ICU 2.6
     */
    virtual UBool visible(void) const;

    /**
     * @stable ICU 2.6
     */
    virtual const UnicodeString * getSupportedIDs(int32_t &count, UErrorCode& status) const;
};
#endif /* #if !UCONFIG_NO_SERVICE */

// -------------------------------------

inline UBool
NumberFormat::isParseIntegerOnly() const
{
    return fParseIntegerOnly;
}

inline UBool
NumberFormat::isLenient() const
{
    return fLenient;
}

U_NAMESPACE_END

#endif /* #if !UCONFIG_NO_FORMATTING */

#endif // _NUMFMT
//eof