9fa0f4299d
X-SVN-Rev: 34578
2400 lines
95 KiB
C++
2400 lines
95 KiB
C++
/*
|
|
********************************************************************************
|
|
* Copyright (C) 1997-2013, International Business Machines
|
|
* Corporation and others. All Rights Reserved.
|
|
********************************************************************************
|
|
*
|
|
* File DECIMFMT.H
|
|
*
|
|
* Modification History:
|
|
*
|
|
* Date Name Description
|
|
* 02/19/97 aliu Converted from java.
|
|
* 03/20/97 clhuang Updated per C++ implementation.
|
|
* 04/03/97 aliu Rewrote parsing and formatting completely, and
|
|
* cleaned up and debugged. Actually works now.
|
|
* 04/17/97 aliu Changed DigitCount to int per code review.
|
|
* 07/10/97 helena Made ParsePosition a class and get rid of the function
|
|
* hiding problems.
|
|
* 09/09/97 aliu Ported over support for exponential formats.
|
|
* 07/20/98 stephen Changed documentation
|
|
* 01/30/13 emmons Added Scaling methods
|
|
********************************************************************************
|
|
*/
|
|
|
|
#ifndef DECIMFMT_H
|
|
#define DECIMFMT_H
|
|
|
|
#include "unicode/utypes.h"
|
|
/**
|
|
* \file
|
|
* \brief C++ API: Formats decimal numbers.
|
|
*/
|
|
|
|
#if !UCONFIG_NO_FORMATTING
|
|
|
|
#include "unicode/dcfmtsym.h"
|
|
#include "unicode/numfmt.h"
|
|
#include "unicode/locid.h"
|
|
#include "unicode/fpositer.h"
|
|
#include "unicode/stringpiece.h"
|
|
#include "unicode/curramt.h"
|
|
#include "unicode/enumset.h"
|
|
|
|
/**
|
|
* \def UNUM_DECIMALFORMAT_INTERNAL_SIZE
|
|
* @internal
|
|
*/
|
|
#if UCONFIG_FORMAT_FASTPATHS_49
|
|
#define UNUM_DECIMALFORMAT_INTERNAL_SIZE 16
|
|
#endif
|
|
|
|
U_NAMESPACE_BEGIN
|
|
|
|
class DigitList;
|
|
class ChoiceFormat;
|
|
class CurrencyPluralInfo;
|
|
class Hashtable;
|
|
class UnicodeSet;
|
|
class FieldPositionHandler;
|
|
class DecimalFormatStaticSets;
|
|
class FixedDecimal;
|
|
|
|
// explicit template instantiation. see digitlst.h
|
|
#if defined (_MSC_VER)
|
|
template class U_I18N_API EnumSet<UNumberFormatAttribute,
|
|
UNUM_MAX_NONBOOLEAN_ATTRIBUTE+1,
|
|
UNUM_LIMIT_BOOLEAN_ATTRIBUTE>;
|
|
#endif
|
|
|
|
/**
|
|
* DecimalFormat is a concrete subclass of NumberFormat that formats decimal
|
|
* numbers. It has a variety of features designed to make it possible to parse
|
|
* and format numbers in any locale, including support for Western, Arabic, or
|
|
* Indic digits. It also supports different flavors of numbers, including
|
|
* integers ("123"), fixed-point numbers ("123.4"), scientific notation
|
|
* ("1.23E4"), percentages ("12%"), and currency amounts ("$123", "USD123",
|
|
* "123 US dollars"). All of these flavors can be easily localized.
|
|
*
|
|
* <p>To obtain a NumberFormat for a specific locale (including the default
|
|
* locale) call one of NumberFormat's factory methods such as
|
|
* createInstance(). Do not call the DecimalFormat constructors directly, unless
|
|
* you know what you are doing, since the NumberFormat factory methods may
|
|
* return subclasses other than DecimalFormat.
|
|
*
|
|
* <p><strong>Example Usage</strong>
|
|
*
|
|
* \code
|
|
* // Normally we would have a GUI with a menu for this
|
|
* int32_t locCount;
|
|
* const Locale* locales = NumberFormat::getAvailableLocales(locCount);
|
|
*
|
|
* double myNumber = -1234.56;
|
|
* UErrorCode success = U_ZERO_ERROR;
|
|
* NumberFormat* form;
|
|
*
|
|
* // Print out a number with the localized number, currency and percent
|
|
* // format for each locale.
|
|
* UnicodeString countryName;
|
|
* UnicodeString displayName;
|
|
* UnicodeString str;
|
|
* UnicodeString pattern;
|
|
* Formattable fmtable;
|
|
* for (int32_t j = 0; j < 3; ++j) {
|
|
* cout << endl << "FORMAT " << j << endl;
|
|
* for (int32_t i = 0; i < locCount; ++i) {
|
|
* if (locales[i].getCountry(countryName).size() == 0) {
|
|
* // skip language-only
|
|
* continue;
|
|
* }
|
|
* switch (j) {
|
|
* case 0:
|
|
* form = NumberFormat::createInstance(locales[i], success ); break;
|
|
* case 1:
|
|
* form = NumberFormat::createCurrencyInstance(locales[i], success ); break;
|
|
* default:
|
|
* form = NumberFormat::createPercentInstance(locales[i], success ); break;
|
|
* }
|
|
* if (form) {
|
|
* str.remove();
|
|
* pattern = ((DecimalFormat*)form)->toPattern(pattern);
|
|
* cout << locales[i].getDisplayName(displayName) << ": " << pattern;
|
|
* cout << " -> " << form->format(myNumber,str) << endl;
|
|
* form->parse(form->format(myNumber,str), fmtable, success);
|
|
* delete form;
|
|
* }
|
|
* }
|
|
* }
|
|
* \endcode
|
|
* <P>
|
|
* Another example use createInstance(style)
|
|
* <P>
|
|
* <pre>
|
|
* <strong>// Print out a number using the localized number, currency,
|
|
* // percent, scientific, integer, iso currency, and plural currency
|
|
* // format for each locale</strong>
|
|
* Locale* locale = new Locale("en", "US");
|
|
* double myNumber = 1234.56;
|
|
* UErrorCode success = U_ZERO_ERROR;
|
|
* UnicodeString str;
|
|
* Formattable fmtable;
|
|
* for (int j=NumberFormat::kNumberStyle;
|
|
* j<=NumberFormat::kPluralCurrencyStyle;
|
|
* ++j) {
|
|
* NumberFormat* format = NumberFormat::createInstance(locale, j, success);
|
|
* str.remove();
|
|
* cout << "format result " << form->format(myNumber, str) << endl;
|
|
* format->parse(form->format(myNumber, str), fmtable, success);
|
|
* }</pre>
|
|
*
|
|
*
|
|
* <p><strong>Patterns</strong>
|
|
*
|
|
* <p>A DecimalFormat consists of a <em>pattern</em> and a set of
|
|
* <em>symbols</em>. The pattern may be set directly using
|
|
* applyPattern(), or indirectly using other API methods which
|
|
* manipulate aspects of the pattern, such as the minimum number of integer
|
|
* digits. The symbols are stored in a DecimalFormatSymbols
|
|
* object. When using the NumberFormat factory methods, the
|
|
* pattern and symbols are read from ICU's locale data.
|
|
*
|
|
* <p><strong>Special Pattern Characters</strong>
|
|
*
|
|
* <p>Many characters in a pattern are taken literally; they are matched during
|
|
* parsing and output unchanged during formatting. Special characters, on the
|
|
* other hand, stand for other characters, strings, or classes of characters.
|
|
* For example, the '#' character is replaced by a localized digit. Often the
|
|
* replacement character is the same as the pattern character; in the U.S. locale,
|
|
* the ',' grouping character is replaced by ','. However, the replacement is
|
|
* still happening, and if the symbols are modified, the grouping character
|
|
* changes. Some special characters affect the behavior of the formatter by
|
|
* their presence; for example, if the percent character is seen, then the
|
|
* value is multiplied by 100 before being displayed.
|
|
*
|
|
* <p>To insert a special character in a pattern as a literal, that is, without
|
|
* any special meaning, the character must be quoted. There are some exceptions to
|
|
* this which are noted below.
|
|
*
|
|
* <p>The characters listed here are used in non-localized patterns. Localized
|
|
* patterns use the corresponding characters taken from this formatter's
|
|
* DecimalFormatSymbols object instead, and these characters lose
|
|
* their special status. Two exceptions are the currency sign and quote, which
|
|
* are not localized.
|
|
*
|
|
* <table border=0 cellspacing=3 cellpadding=0>
|
|
* <tr bgcolor="#ccccff">
|
|
* <td align=left><strong>Symbol</strong>
|
|
* <td align=left><strong>Location</strong>
|
|
* <td align=left><strong>Localized?</strong>
|
|
* <td align=left><strong>Meaning</strong>
|
|
* <tr valign=top>
|
|
* <td><code>0</code>
|
|
* <td>Number
|
|
* <td>Yes
|
|
* <td>Digit
|
|
* <tr valign=top bgcolor="#eeeeff">
|
|
* <td><code>1-9</code>
|
|
* <td>Number
|
|
* <td>Yes
|
|
* <td>'1' through '9' indicate rounding.
|
|
* <tr valign=top>
|
|
* <td><code>\htmlonly@\endhtmlonly</code> <!--doxygen doesn't like @-->
|
|
* <td>Number
|
|
* <td>No
|
|
* <td>Significant digit
|
|
* <tr valign=top bgcolor="#eeeeff">
|
|
* <td><code>#</code>
|
|
* <td>Number
|
|
* <td>Yes
|
|
* <td>Digit, zero shows as absent
|
|
* <tr valign=top>
|
|
* <td><code>.</code>
|
|
* <td>Number
|
|
* <td>Yes
|
|
* <td>Decimal separator or monetary decimal separator
|
|
* <tr valign=top bgcolor="#eeeeff">
|
|
* <td><code>-</code>
|
|
* <td>Number
|
|
* <td>Yes
|
|
* <td>Minus sign
|
|
* <tr valign=top>
|
|
* <td><code>,</code>
|
|
* <td>Number
|
|
* <td>Yes
|
|
* <td>Grouping separator
|
|
* <tr valign=top bgcolor="#eeeeff">
|
|
* <td><code>E</code>
|
|
* <td>Number
|
|
* <td>Yes
|
|
* <td>Separates mantissa and exponent in scientific notation.
|
|
* <em>Need not be quoted in prefix or suffix.</em>
|
|
* <tr valign=top>
|
|
* <td><code>+</code>
|
|
* <td>Exponent
|
|
* <td>Yes
|
|
* <td>Prefix positive exponents with localized plus sign.
|
|
* <em>Need not be quoted in prefix or suffix.</em>
|
|
* <tr valign=top bgcolor="#eeeeff">
|
|
* <td><code>;</code>
|
|
* <td>Subpattern boundary
|
|
* <td>Yes
|
|
* <td>Separates positive and negative subpatterns
|
|
* <tr valign=top>
|
|
* <td><code>\%</code>
|
|
* <td>Prefix or suffix
|
|
* <td>Yes
|
|
* <td>Multiply by 100 and show as percentage
|
|
* <tr valign=top bgcolor="#eeeeff">
|
|
* <td><code>\\u2030</code>
|
|
* <td>Prefix or suffix
|
|
* <td>Yes
|
|
* <td>Multiply by 1000 and show as per mille
|
|
* <tr valign=top>
|
|
* <td><code>\htmlonly¤\endhtmlonly</code> (<code>\\u00A4</code>)
|
|
* <td>Prefix or suffix
|
|
* <td>No
|
|
* <td>Currency sign, replaced by currency symbol. If
|
|
* doubled, replaced by international currency symbol.
|
|
* If tripled, replaced by currency plural names, for example,
|
|
* "US dollar" or "US dollars" for America.
|
|
* If present in a pattern, the monetary decimal separator
|
|
* is used instead of the decimal separator.
|
|
* <tr valign=top bgcolor="#eeeeff">
|
|
* <td><code>'</code>
|
|
* <td>Prefix or suffix
|
|
* <td>No
|
|
* <td>Used to quote special characters in a prefix or suffix,
|
|
* for example, <code>"'#'#"</code> formats 123 to
|
|
* <code>"#123"</code>. To create a single quote
|
|
* itself, use two in a row: <code>"# o''clock"</code>.
|
|
* <tr valign=top>
|
|
* <td><code>*</code>
|
|
* <td>Prefix or suffix boundary
|
|
* <td>Yes
|
|
* <td>Pad escape, precedes pad character
|
|
* </table>
|
|
*
|
|
* <p>A DecimalFormat pattern contains a postive and negative
|
|
* subpattern, for example, "#,##0.00;(#,##0.00)". Each subpattern has a
|
|
* prefix, a numeric part, and a suffix. If there is no explicit negative
|
|
* subpattern, the negative subpattern is the localized minus sign prefixed to the
|
|
* positive subpattern. That is, "0.00" alone is equivalent to "0.00;-0.00". If there
|
|
* is an explicit negative subpattern, it serves only to specify the negative
|
|
* prefix and suffix; the number of digits, minimal digits, and other
|
|
* characteristics are ignored in the negative subpattern. That means that
|
|
* "#,##0.0#;(#)" has precisely the same result as "#,##0.0#;(#,##0.0#)".
|
|
*
|
|
* <p>The prefixes, suffixes, and various symbols used for infinity, digits,
|
|
* thousands separators, decimal separators, etc. may be set to arbitrary
|
|
* values, and they will appear properly during formatting. However, care must
|
|
* be taken that the symbols and strings do not conflict, or parsing will be
|
|
* unreliable. For example, either the positive and negative prefixes or the
|
|
* suffixes must be distinct for parse() to be able
|
|
* to distinguish positive from negative values. Another example is that the
|
|
* decimal separator and thousands separator should be distinct characters, or
|
|
* parsing will be impossible.
|
|
*
|
|
* <p>The <em>grouping separator</em> is a character that separates clusters of
|
|
* integer digits to make large numbers more legible. It commonly used for
|
|
* thousands, but in some locales it separates ten-thousands. The <em>grouping
|
|
* size</em> is the number of digits between the grouping separators, such as 3
|
|
* for "100,000,000" or 4 for "1 0000 0000". There are actually two different
|
|
* grouping sizes: One used for the least significant integer digits, the
|
|
* <em>primary grouping size</em>, and one used for all others, the
|
|
* <em>secondary grouping size</em>. In most locales these are the same, but
|
|
* sometimes they are different. For example, if the primary grouping interval
|
|
* is 3, and the secondary is 2, then this corresponds to the pattern
|
|
* "#,##,##0", and the number 123456789 is formatted as "12,34,56,789". If a
|
|
* pattern contains multiple grouping separators, the interval between the last
|
|
* one and the end of the integer defines the primary grouping size, and the
|
|
* interval between the last two defines the secondary grouping size. All others
|
|
* are ignored, so "#,##,###,####" == "###,###,####" == "##,#,###,####".
|
|
*
|
|
* <p>Illegal patterns, such as "#.#.#" or "#.###,###", will cause
|
|
* DecimalFormat to set a failing UErrorCode.
|
|
*
|
|
* <p><strong>Pattern BNF</strong>
|
|
*
|
|
* <pre>
|
|
* pattern := subpattern (';' subpattern)?
|
|
* subpattern := prefix? number exponent? suffix?
|
|
* number := (integer ('.' fraction)?) | sigDigits
|
|
* prefix := '\\u0000'..'\\uFFFD' - specialCharacters
|
|
* suffix := '\\u0000'..'\\uFFFD' - specialCharacters
|
|
* integer := '#'* '0'* '0'
|
|
* fraction := '0'* '#'*
|
|
* sigDigits := '#'* '@' '@'* '#'*
|
|
* exponent := 'E' '+'? '0'* '0'
|
|
* padSpec := '*' padChar
|
|
* padChar := '\\u0000'..'\\uFFFD' - quote
|
|
*
|
|
* Notation:
|
|
* X* 0 or more instances of X
|
|
* X? 0 or 1 instances of X
|
|
* X|Y either X or Y
|
|
* C..D any character from C up to D, inclusive
|
|
* S-T characters in S, except those in T
|
|
* </pre>
|
|
* The first subpattern is for positive numbers. The second (optional)
|
|
* subpattern is for negative numbers.
|
|
*
|
|
* <p>Not indicated in the BNF syntax above:
|
|
*
|
|
* <ul><li>The grouping separator ',' can occur inside the integer and
|
|
* sigDigits elements, between any two pattern characters of that
|
|
* element, as long as the integer or sigDigits element is not
|
|
* followed by the exponent element.
|
|
*
|
|
* <li>Two grouping intervals are recognized: That between the
|
|
* decimal point and the first grouping symbol, and that
|
|
* between the first and second grouping symbols. These
|
|
* intervals are identical in most locales, but in some
|
|
* locales they differ. For example, the pattern
|
|
* "#,##,###" formats the number 123456789 as
|
|
* "12,34,56,789".</li>
|
|
*
|
|
* <li>The pad specifier <code>padSpec</code> may appear before the prefix,
|
|
* after the prefix, before the suffix, after the suffix, or not at all.
|
|
*
|
|
* <li>In place of '0', the digits '1' through '9' may be used to
|
|
* indicate a rounding increment.
|
|
* </ul>
|
|
*
|
|
* <p><strong>Parsing</strong>
|
|
*
|
|
* <p>DecimalFormat parses all Unicode characters that represent
|
|
* decimal digits, as defined by u_charDigitValue(). In addition,
|
|
* DecimalFormat also recognizes as digits the ten consecutive
|
|
* characters starting with the localized zero digit defined in the
|
|
* DecimalFormatSymbols object. During formatting, the
|
|
* DecimalFormatSymbols-based digits are output.
|
|
*
|
|
* <p>During parsing, grouping separators are ignored if in lenient mode;
|
|
* otherwise, if present, they must be in appropriate positions.
|
|
*
|
|
* <p>For currency parsing, the formatter is able to parse every currency
|
|
* style formats no matter which style the formatter is constructed with.
|
|
* For example, a formatter instance gotten from
|
|
* NumberFormat.getInstance(ULocale, NumberFormat.CURRENCYSTYLE) can parse
|
|
* formats such as "USD1.00" and "3.00 US dollars".
|
|
*
|
|
* <p>If parse(UnicodeString&,Formattable&,ParsePosition&)
|
|
* fails to parse a string, it leaves the parse position unchanged.
|
|
* The convenience method parse(UnicodeString&,Formattable&,UErrorCode&)
|
|
* indicates parse failure by setting a failing
|
|
* UErrorCode.
|
|
*
|
|
* <p><strong>Formatting</strong>
|
|
*
|
|
* <p>Formatting is guided by several parameters, all of which can be
|
|
* specified either using a pattern or using the API. The following
|
|
* description applies to formats that do not use <a href="#sci">scientific
|
|
* notation</a> or <a href="#sigdig">significant digits</a>.
|
|
*
|
|
* <ul><li>If the number of actual integer digits exceeds the
|
|
* <em>maximum integer digits</em>, then only the least significant
|
|
* digits are shown. For example, 1997 is formatted as "97" if the
|
|
* maximum integer digits is set to 2.
|
|
*
|
|
* <li>If the number of actual integer digits is less than the
|
|
* <em>minimum integer digits</em>, then leading zeros are added. For
|
|
* example, 1997 is formatted as "01997" if the minimum integer digits
|
|
* is set to 5.
|
|
*
|
|
* <li>If the number of actual fraction digits exceeds the <em>maximum
|
|
* fraction digits</em>, then rounding is performed to the
|
|
* maximum fraction digits. For example, 0.125 is formatted as "0.12"
|
|
* if the maximum fraction digits is 2. This behavior can be changed
|
|
* by specifying a rounding increment and/or a rounding mode.
|
|
*
|
|
* <li>If the number of actual fraction digits is less than the
|
|
* <em>minimum fraction digits</em>, then trailing zeros are added.
|
|
* For example, 0.125 is formatted as "0.1250" if the mimimum fraction
|
|
* digits is set to 4.
|
|
*
|
|
* <li>Trailing fractional zeros are not displayed if they occur
|
|
* <em>j</em> positions after the decimal, where <em>j</em> is less
|
|
* than the maximum fraction digits. For example, 0.10004 is
|
|
* formatted as "0.1" if the maximum fraction digits is four or less.
|
|
* </ul>
|
|
*
|
|
* <p><strong>Special Values</strong>
|
|
*
|
|
* <p><code>NaN</code> is represented as a single character, typically
|
|
* <code>\\uFFFD</code>. This character is determined by the
|
|
* DecimalFormatSymbols object. This is the only value for which
|
|
* the prefixes and suffixes are not used.
|
|
*
|
|
* <p>Infinity is represented as a single character, typically
|
|
* <code>\\u221E</code>, with the positive or negative prefixes and suffixes
|
|
* applied. The infinity character is determined by the
|
|
* DecimalFormatSymbols object.
|
|
*
|
|
* <a name="sci"><strong>Scientific Notation</strong></a>
|
|
*
|
|
* <p>Numbers in scientific notation are expressed as the product of a mantissa
|
|
* and a power of ten, for example, 1234 can be expressed as 1.234 x 10<sup>3</sup>. The
|
|
* mantissa is typically in the half-open interval [1.0, 10.0) or sometimes [0.0, 1.0),
|
|
* but it need not be. DecimalFormat supports arbitrary mantissas.
|
|
* DecimalFormat can be instructed to use scientific
|
|
* notation through the API or through the pattern. In a pattern, the exponent
|
|
* character immediately followed by one or more digit characters indicates
|
|
* scientific notation. Example: "0.###E0" formats the number 1234 as
|
|
* "1.234E3".
|
|
*
|
|
* <ul>
|
|
* <li>The number of digit characters after the exponent character gives the
|
|
* minimum exponent digit count. There is no maximum. Negative exponents are
|
|
* formatted using the localized minus sign, <em>not</em> the prefix and suffix
|
|
* from the pattern. This allows patterns such as "0.###E0 m/s". To prefix
|
|
* positive exponents with a localized plus sign, specify '+' between the
|
|
* exponent and the digits: "0.###E+0" will produce formats "1E+1", "1E+0",
|
|
* "1E-1", etc. (In localized patterns, use the localized plus sign rather than
|
|
* '+'.)
|
|
*
|
|
* <li>The minimum number of integer digits is achieved by adjusting the
|
|
* exponent. Example: 0.00123 formatted with "00.###E0" yields "12.3E-4". This
|
|
* only happens if there is no maximum number of integer digits. If there is a
|
|
* maximum, then the minimum number of integer digits is fixed at one.
|
|
*
|
|
* <li>The maximum number of integer digits, if present, specifies the exponent
|
|
* grouping. The most common use of this is to generate <em>engineering
|
|
* notation</em>, in which the exponent is a multiple of three, e.g.,
|
|
* "##0.###E0". The number 12345 is formatted using "##0.####E0" as "12.345E3".
|
|
*
|
|
* <li>When using scientific notation, the formatter controls the
|
|
* digit counts using significant digits logic. The maximum number of
|
|
* significant digits limits the total number of integer and fraction
|
|
* digits that will be shown in the mantissa; it does not affect
|
|
* parsing. For example, 12345 formatted with "##0.##E0" is "12.3E3".
|
|
* See the section on significant digits for more details.
|
|
*
|
|
* <li>The number of significant digits shown is determined as
|
|
* follows: If areSignificantDigitsUsed() returns false, then the
|
|
* minimum number of significant digits shown is one, and the maximum
|
|
* number of significant digits shown is the sum of the <em>minimum
|
|
* integer</em> and <em>maximum fraction</em> digits, and is
|
|
* unaffected by the maximum integer digits. If this sum is zero,
|
|
* then all significant digits are shown. If
|
|
* areSignificantDigitsUsed() returns true, then the significant digit
|
|
* counts are specified by getMinimumSignificantDigits() and
|
|
* getMaximumSignificantDigits(). In this case, the number of
|
|
* integer digits is fixed at one, and there is no exponent grouping.
|
|
*
|
|
* <li>Exponential patterns may not contain grouping separators.
|
|
* </ul>
|
|
*
|
|
* <a name="sigdig"><strong>Significant Digits</strong></a>
|
|
*
|
|
* <code>DecimalFormat</code> has two ways of controlling how many
|
|
* digits are shows: (a) significant digits counts, or (b) integer and
|
|
* fraction digit counts. Integer and fraction digit counts are
|
|
* described above. When a formatter is using significant digits
|
|
* counts, the number of integer and fraction digits is not specified
|
|
* directly, and the formatter settings for these counts are ignored.
|
|
* Instead, the formatter uses however many integer and fraction
|
|
* digits are required to display the specified number of significant
|
|
* digits. Examples:
|
|
*
|
|
* <table border=0 cellspacing=3 cellpadding=0>
|
|
* <tr bgcolor="#ccccff">
|
|
* <td align=left>Pattern
|
|
* <td align=left>Minimum significant digits
|
|
* <td align=left>Maximum significant digits
|
|
* <td align=left>Number
|
|
* <td align=left>Output of format()
|
|
* <tr valign=top>
|
|
* <td><code>\@\@\@</code>
|
|
* <td>3
|
|
* <td>3
|
|
* <td>12345
|
|
* <td><code>12300</code>
|
|
* <tr valign=top bgcolor="#eeeeff">
|
|
* <td><code>\@\@\@</code>
|
|
* <td>3
|
|
* <td>3
|
|
* <td>0.12345
|
|
* <td><code>0.123</code>
|
|
* <tr valign=top>
|
|
* <td><code>\@\@##</code>
|
|
* <td>2
|
|
* <td>4
|
|
* <td>3.14159
|
|
* <td><code>3.142</code>
|
|
* <tr valign=top bgcolor="#eeeeff">
|
|
* <td><code>\@\@##</code>
|
|
* <td>2
|
|
* <td>4
|
|
* <td>1.23004
|
|
* <td><code>1.23</code>
|
|
* </table>
|
|
*
|
|
* <ul>
|
|
* <li>Significant digit counts may be expressed using patterns that
|
|
* specify a minimum and maximum number of significant digits. These
|
|
* are indicated by the <code>'@'</code> and <code>'#'</code>
|
|
* characters. The minimum number of significant digits is the number
|
|
* of <code>'@'</code> characters. The maximum number of significant
|
|
* digits is the number of <code>'@'</code> characters plus the number
|
|
* of <code>'#'</code> characters following on the right. For
|
|
* example, the pattern <code>"@@@"</code> indicates exactly 3
|
|
* significant digits. The pattern <code>"@##"</code> indicates from
|
|
* 1 to 3 significant digits. Trailing zero digits to the right of
|
|
* the decimal separator are suppressed after the minimum number of
|
|
* significant digits have been shown. For example, the pattern
|
|
* <code>"@##"</code> formats the number 0.1203 as
|
|
* <code>"0.12"</code>.
|
|
*
|
|
* <li>If a pattern uses significant digits, it may not contain a
|
|
* decimal separator, nor the <code>'0'</code> pattern character.
|
|
* Patterns such as <code>"@00"</code> or <code>"@.###"</code> are
|
|
* disallowed.
|
|
*
|
|
* <li>Any number of <code>'#'</code> characters may be prepended to
|
|
* the left of the leftmost <code>'@'</code> character. These have no
|
|
* effect on the minimum and maximum significant digits counts, but
|
|
* may be used to position grouping separators. For example,
|
|
* <code>"#,#@#"</code> indicates a minimum of one significant digits,
|
|
* a maximum of two significant digits, and a grouping size of three.
|
|
*
|
|
* <li>In order to enable significant digits formatting, use a pattern
|
|
* containing the <code>'@'</code> pattern character. Alternatively,
|
|
* call setSignificantDigitsUsed(TRUE).
|
|
*
|
|
* <li>In order to disable significant digits formatting, use a
|
|
* pattern that does not contain the <code>'@'</code> pattern
|
|
* character. Alternatively, call setSignificantDigitsUsed(FALSE).
|
|
*
|
|
* <li>The number of significant digits has no effect on parsing.
|
|
*
|
|
* <li>Significant digits may be used together with exponential notation. Such
|
|
* patterns are equivalent to a normal exponential pattern with a minimum and
|
|
* maximum integer digit count of one, a minimum fraction digit count of
|
|
* <code>getMinimumSignificantDigits() - 1</code>, and a maximum fraction digit
|
|
* count of <code>getMaximumSignificantDigits() - 1</code>. For example, the
|
|
* pattern <code>"@@###E0"</code> is equivalent to <code>"0.0###E0"</code>.
|
|
*
|
|
* <li>If signficant digits are in use, then the integer and fraction
|
|
* digit counts, as set via the API, are ignored. If significant
|
|
* digits are not in use, then the signficant digit counts, as set via
|
|
* the API, are ignored.
|
|
*
|
|
* </ul>
|
|
*
|
|
* <p><strong>Padding</strong>
|
|
*
|
|
* <p>DecimalFormat supports padding the result of
|
|
* format() to a specific width. Padding may be specified either
|
|
* through the API or through the pattern syntax. In a pattern the pad escape
|
|
* character, followed by a single pad character, causes padding to be parsed
|
|
* and formatted. The pad escape character is '*' in unlocalized patterns, and
|
|
* can be localized using DecimalFormatSymbols::setSymbol() with a
|
|
* DecimalFormatSymbols::kPadEscapeSymbol
|
|
* selector. For example, <code>"$*x#,##0.00"</code> formats 123 to
|
|
* <code>"$xx123.00"</code>, and 1234 to <code>"$1,234.00"</code>.
|
|
*
|
|
* <ul>
|
|
* <li>When padding is in effect, the width of the positive subpattern,
|
|
* including prefix and suffix, determines the format width. For example, in
|
|
* the pattern <code>"* #0 o''clock"</code>, the format width is 10.
|
|
*
|
|
* <li>The width is counted in 16-bit code units (UChars).
|
|
*
|
|
* <li>Some parameters which usually do not matter have meaning when padding is
|
|
* used, because the pattern width is significant with padding. In the pattern
|
|
* "* ##,##,#,##0.##", the format width is 14. The initial characters "##,##,"
|
|
* do not affect the grouping size or maximum integer digits, but they do affect
|
|
* the format width.
|
|
*
|
|
* <li>Padding may be inserted at one of four locations: before the prefix,
|
|
* after the prefix, before the suffix, or after the suffix. If padding is
|
|
* specified in any other location, applyPattern()
|
|
* sets a failing UErrorCode. If there is no prefix,
|
|
* before the prefix and after the prefix are equivalent, likewise for the
|
|
* suffix.
|
|
*
|
|
* <li>When specified in a pattern, the 32-bit code point immediately
|
|
* following the pad escape is the pad character. This may be any character,
|
|
* including a special pattern character. That is, the pad escape
|
|
* <em>escapes</em> the following character. If there is no character after
|
|
* the pad escape, then the pattern is illegal.
|
|
*
|
|
* </ul>
|
|
*
|
|
* <p><strong>Rounding</strong>
|
|
*
|
|
* <p>DecimalFormat supports rounding to a specific increment. For
|
|
* example, 1230 rounded to the nearest 50 is 1250. 1.234 rounded to the
|
|
* nearest 0.65 is 1.3. The rounding increment may be specified through the API
|
|
* or in a pattern. To specify a rounding increment in a pattern, include the
|
|
* increment in the pattern itself. "#,#50" specifies a rounding increment of
|
|
* 50. "#,##0.05" specifies a rounding increment of 0.05.
|
|
*
|
|
* <p>In the absense of an explicit rounding increment numbers are
|
|
* rounded to their formatted width.
|
|
*
|
|
* <ul>
|
|
* <li>Rounding only affects the string produced by formatting. It does
|
|
* not affect parsing or change any numerical values.
|
|
*
|
|
* <li>A <em>rounding mode</em> determines how values are rounded; see
|
|
* DecimalFormat::ERoundingMode. The default rounding mode is
|
|
* DecimalFormat::kRoundHalfEven. The rounding mode can only be set
|
|
* through the API; it can not be set with a pattern.
|
|
*
|
|
* <li>Some locales use rounding in their currency formats to reflect the
|
|
* smallest currency denomination.
|
|
*
|
|
* <li>In a pattern, digits '1' through '9' specify rounding, but otherwise
|
|
* behave identically to digit '0'.
|
|
* </ul>
|
|
*
|
|
* <p><strong>Synchronization</strong>
|
|
*
|
|
* <p>DecimalFormat objects are not synchronized. Multiple
|
|
* threads should not access one formatter concurrently.
|
|
*
|
|
* <p><strong>Subclassing</strong>
|
|
*
|
|
* <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.
|
|
*/
|
|
class U_I18N_API DecimalFormat: public NumberFormat {
|
|
public:
|
|
/**
|
|
* Rounding mode.
|
|
* @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
|
|
};
|
|
|
|
/**
|
|
* Pad position.
|
|
* @stable ICU 2.4
|
|
*/
|
|
enum EPadPosition {
|
|
kPadBeforePrefix,
|
|
kPadAfterPrefix,
|
|
kPadBeforeSuffix,
|
|
kPadAfterSuffix
|
|
};
|
|
|
|
/**
|
|
* Create a DecimalFormat using the default pattern and symbols
|
|
* for the default locale. This is a convenient way to obtain a
|
|
* DecimalFormat when internationalization is not the main concern.
|
|
* <P>
|
|
* To obtain standard formats for a given locale, use the factory methods
|
|
* on NumberFormat such as createInstance. These factories will
|
|
* return the most appropriate sub-class of NumberFormat for a given
|
|
* locale.
|
|
* @param status Output param set to success/failure code. If the
|
|
* pattern is invalid this will be set to a failure code.
|
|
* @stable ICU 2.0
|
|
*/
|
|
DecimalFormat(UErrorCode& status);
|
|
|
|
/**
|
|
* Create a DecimalFormat from the given pattern and the symbols
|
|
* for the default locale. This is a convenient way to obtain a
|
|
* DecimalFormat when internationalization is not the main concern.
|
|
* <P>
|
|
* To obtain standard formats for a given locale, use the factory methods
|
|
* on NumberFormat such as createInstance. These factories will
|
|
* return the most appropriate sub-class of NumberFormat for a given
|
|
* locale.
|
|
* @param pattern A non-localized pattern string.
|
|
* @param status Output param set to success/failure code. If the
|
|
* pattern is invalid this will be set to a failure code.
|
|
* @stable ICU 2.0
|
|
*/
|
|
DecimalFormat(const UnicodeString& pattern,
|
|
UErrorCode& status);
|
|
|
|
/**
|
|
* Create a DecimalFormat from the given pattern and symbols.
|
|
* Use this constructor when you need to completely customize the
|
|
* behavior of the format.
|
|
* <P>
|
|
* To obtain standard formats for a given
|
|
* locale, use the factory methods on NumberFormat such as
|
|
* createInstance or createCurrencyInstance. If you need only minor adjustments
|
|
* to a standard format, you can modify the format returned by
|
|
* a NumberFormat factory method.
|
|
*
|
|
* @param pattern a non-localized pattern string
|
|
* @param symbolsToAdopt the set of symbols to be used. The caller should not
|
|
* delete this object after making this call.
|
|
* @param status Output param set to success/failure code. If the
|
|
* pattern is invalid this will be set to a failure code.
|
|
* @stable ICU 2.0
|
|
*/
|
|
DecimalFormat( const UnicodeString& pattern,
|
|
DecimalFormatSymbols* symbolsToAdopt,
|
|
UErrorCode& status);
|
|
|
|
#ifndef U_HIDE_INTERNAL_API
|
|
/**
|
|
* This API is for ICU use only.
|
|
* Create a DecimalFormat from the given pattern, symbols, and style.
|
|
*
|
|
* @param pattern a non-localized pattern string
|
|
* @param symbolsToAdopt the set of symbols to be used. The caller should not
|
|
* delete this object after making this call.
|
|
* @param style style of decimal format
|
|
* @param status Output param set to success/failure code. If the
|
|
* pattern is invalid this will be set to a failure code.
|
|
* @internal
|
|
*/
|
|
DecimalFormat( const UnicodeString& pattern,
|
|
DecimalFormatSymbols* symbolsToAdopt,
|
|
UNumberFormatStyle style,
|
|
UErrorCode& status);
|
|
|
|
#if UCONFIG_HAVE_PARSEALLINPUT
|
|
/**
|
|
* @internal
|
|
*/
|
|
void setParseAllInput(UNumberFormatAttributeValue value);
|
|
#endif
|
|
|
|
#endif /* U_HIDE_INTERNAL_API */
|
|
|
|
|
|
/**
|
|
* Set an integer attribute on this DecimalFormat.
|
|
* May return U_UNSUPPORTED_ERROR if this instance does not support
|
|
* the specified attribute.
|
|
* @param attr the attribute to set
|
|
* @param newvalue new value
|
|
* @param status the error type
|
|
* @return *this - for chaining (example: format.setAttribute(...).setAttribute(...) )
|
|
* @draft ICU 51
|
|
*/
|
|
virtual DecimalFormat& setAttribute( UNumberFormatAttribute attr,
|
|
int32_t newvalue,
|
|
UErrorCode &status);
|
|
|
|
/**
|
|
* Get an integer
|
|
* May return U_UNSUPPORTED_ERROR if this instance does not support
|
|
* the specified attribute.
|
|
* @param attr the attribute to set
|
|
* @param status the error type
|
|
* @return the attribute value. Undefined if there is an error.
|
|
* @draft ICU 51
|
|
*/
|
|
virtual int32_t getAttribute( UNumberFormatAttribute attr,
|
|
UErrorCode &status) const;
|
|
|
|
|
|
|
|
/**
|
|
* Create a DecimalFormat from the given pattern and symbols.
|
|
* Use this constructor when you need to completely customize the
|
|
* behavior of the format.
|
|
* <P>
|
|
* To obtain standard formats for a given
|
|
* locale, use the factory methods on NumberFormat such as
|
|
* createInstance or createCurrencyInstance. If you need only minor adjustments
|
|
* to a standard format, you can modify the format returned by
|
|
* a NumberFormat factory method.
|
|
*
|
|
* @param pattern a non-localized pattern string
|
|
* @param symbolsToAdopt the set of symbols to be used. The caller should not
|
|
* delete this object after making this call.
|
|
* @param parseError Output param to receive errors occured during parsing
|
|
* @param status Output param set to success/failure code. If the
|
|
* pattern is invalid this will be set to a failure code.
|
|
* @stable ICU 2.0
|
|
*/
|
|
DecimalFormat( const UnicodeString& pattern,
|
|
DecimalFormatSymbols* symbolsToAdopt,
|
|
UParseError& parseError,
|
|
UErrorCode& status);
|
|
/**
|
|
* Create a DecimalFormat from the given pattern and symbols.
|
|
* Use this constructor when you need to completely customize the
|
|
* behavior of the format.
|
|
* <P>
|
|
* To obtain standard formats for a given
|
|
* locale, use the factory methods on NumberFormat such as
|
|
* createInstance or createCurrencyInstance. If you need only minor adjustments
|
|
* to a standard format, you can modify the format returned by
|
|
* a NumberFormat factory method.
|
|
*
|
|
* @param pattern a non-localized pattern string
|
|
* @param symbols the set of symbols to be used
|
|
* @param status Output param set to success/failure code. If the
|
|
* pattern is invalid this will be set to a failure code.
|
|
* @stable ICU 2.0
|
|
*/
|
|
DecimalFormat( const UnicodeString& pattern,
|
|
const DecimalFormatSymbols& symbols,
|
|
UErrorCode& status);
|
|
|
|
/**
|
|
* Copy constructor.
|
|
*
|
|
* @param source the DecimalFormat object to be copied from.
|
|
* @stable ICU 2.0
|
|
*/
|
|
DecimalFormat(const DecimalFormat& source);
|
|
|
|
/**
|
|
* Assignment operator.
|
|
*
|
|
* @param rhs the DecimalFormat object to be copied.
|
|
* @stable ICU 2.0
|
|
*/
|
|
DecimalFormat& operator=(const DecimalFormat& rhs);
|
|
|
|
/**
|
|
* Destructor.
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual ~DecimalFormat();
|
|
|
|
/**
|
|
* Clone this Format object polymorphically. The caller owns the
|
|
* result and should delete it when done.
|
|
*
|
|
* @return a polymorphic copy of this DecimalFormat.
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual Format* clone(void) const;
|
|
|
|
/**
|
|
* Return true if the given Format objects are semantically equal.
|
|
* Objects of different subclasses are considered unequal.
|
|
*
|
|
* @param other the object to be compared with.
|
|
* @return true if the given Format objects are semantically equal.
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual UBool operator==(const Format& other) const;
|
|
|
|
|
|
using NumberFormat::format;
|
|
|
|
/**
|
|
* Format a double or long number using base-10 representation.
|
|
*
|
|
* @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;
|
|
|
|
|
|
/**
|
|
* Format a double or long number using base-10 representation.
|
|
*
|
|
* @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
|
|
* @return Reference to 'appendTo' parameter.
|
|
* @internal
|
|
*/
|
|
virtual UnicodeString& format(double number,
|
|
UnicodeString& appendTo,
|
|
FieldPosition& pos,
|
|
UErrorCode &status) const;
|
|
|
|
/**
|
|
* Format a double or long number using base-10 representation.
|
|
*
|
|
* @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 4.4
|
|
*/
|
|
virtual UnicodeString& format(double number,
|
|
UnicodeString& appendTo,
|
|
FieldPositionIterator* posIter,
|
|
UErrorCode& status) const;
|
|
|
|
/**
|
|
* Format a long number using base-10 representation.
|
|
*
|
|
* @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;
|
|
|
|
/**
|
|
* Format a long number using base-10 representation.
|
|
*
|
|
* @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.
|
|
* @internal
|
|
*/
|
|
virtual UnicodeString& format(int32_t number,
|
|
UnicodeString& appendTo,
|
|
FieldPosition& pos,
|
|
UErrorCode &status) const;
|
|
|
|
/**
|
|
* Format a long number using base-10 representation.
|
|
*
|
|
* @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 4.4
|
|
*/
|
|
virtual UnicodeString& format(int32_t number,
|
|
UnicodeString& appendTo,
|
|
FieldPositionIterator* posIter,
|
|
UErrorCode& status) const;
|
|
|
|
/**
|
|
* Format an int64 number using base-10 representation.
|
|
*
|
|
* @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 using base-10 representation.
|
|
*
|
|
* @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.
|
|
* @internal
|
|
*/
|
|
virtual UnicodeString& format(int64_t number,
|
|
UnicodeString& appendTo,
|
|
FieldPosition& pos,
|
|
UErrorCode &status) const;
|
|
|
|
/**
|
|
* Format an int64 number using base-10 representation.
|
|
*
|
|
* @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 4.4
|
|
*/
|
|
virtual UnicodeString& format(int64_t number,
|
|
UnicodeString& appendTo,
|
|
FieldPositionIterator* posIter,
|
|
UErrorCode& status) const;
|
|
|
|
/**
|
|
* Format a decimal number.
|
|
* 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.
|
|
* @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 4.4
|
|
*/
|
|
virtual UnicodeString& format(const StringPiece &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.
|
|
*
|
|
* @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.
|
|
*
|
|
* @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;
|
|
|
|
using NumberFormat::parse;
|
|
|
|
/**
|
|
* Parse the given string using this object's choices. The method
|
|
* does string comparisons to try to find an optimal match.
|
|
* If no object can be parsed, index is unchanged, and NULL is
|
|
* returned. The result is returned as the most parsimonious
|
|
* type of Formattable that will accomodate all of the
|
|
* necessary precision. For example, if the result is exactly 12,
|
|
* it will be returned as a long. However, if it is 1.5, it will
|
|
* be returned as a double.
|
|
*
|
|
* @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.
|
|
* @see Formattable
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void parse(const UnicodeString& text,
|
|
Formattable& result,
|
|
ParsePosition& parsePosition) 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;
|
|
|
|
/**
|
|
* Returns the decimal format symbols, which is generally not changed
|
|
* by the programmer or user.
|
|
* @return desired DecimalFormatSymbols
|
|
* @see DecimalFormatSymbols
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual const DecimalFormatSymbols* getDecimalFormatSymbols(void) const;
|
|
|
|
/**
|
|
* Sets the decimal format symbols, which is generally not changed
|
|
* by the programmer or user.
|
|
* @param symbolsToAdopt DecimalFormatSymbols to be adopted.
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void adoptDecimalFormatSymbols(DecimalFormatSymbols* symbolsToAdopt);
|
|
|
|
/**
|
|
* Sets the decimal format symbols, which is generally not changed
|
|
* by the programmer or user.
|
|
* @param symbols DecimalFormatSymbols.
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setDecimalFormatSymbols(const DecimalFormatSymbols& symbols);
|
|
|
|
|
|
/**
|
|
* Returns the currency plural format information,
|
|
* which is generally not changed by the programmer or user.
|
|
* @return desired CurrencyPluralInfo
|
|
* @stable ICU 4.2
|
|
*/
|
|
virtual const CurrencyPluralInfo* getCurrencyPluralInfo(void) const;
|
|
|
|
/**
|
|
* Sets the currency plural format information,
|
|
* which is generally not changed by the programmer or user.
|
|
* @param toAdopt CurrencyPluralInfo to be adopted.
|
|
* @stable ICU 4.2
|
|
*/
|
|
virtual void adoptCurrencyPluralInfo(CurrencyPluralInfo* toAdopt);
|
|
|
|
/**
|
|
* Sets the currency plural format information,
|
|
* which is generally not changed by the programmer or user.
|
|
* @param info Currency Plural Info.
|
|
* @stable ICU 4.2
|
|
*/
|
|
virtual void setCurrencyPluralInfo(const CurrencyPluralInfo& info);
|
|
|
|
|
|
/**
|
|
* Get the positive prefix.
|
|
*
|
|
* @param result Output param which will receive the positive prefix.
|
|
* @return A reference to 'result'.
|
|
* Examples: +123, $123, sFr123
|
|
* @stable ICU 2.0
|
|
*/
|
|
UnicodeString& getPositivePrefix(UnicodeString& result) const;
|
|
|
|
/**
|
|
* Set the positive prefix.
|
|
*
|
|
* @param newValue the new value of the the positive prefix to be set.
|
|
* Examples: +123, $123, sFr123
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setPositivePrefix(const UnicodeString& newValue);
|
|
|
|
/**
|
|
* Get the negative prefix.
|
|
*
|
|
* @param result Output param which will receive the negative prefix.
|
|
* @return A reference to 'result'.
|
|
* Examples: -123, ($123) (with negative suffix), sFr-123
|
|
* @stable ICU 2.0
|
|
*/
|
|
UnicodeString& getNegativePrefix(UnicodeString& result) const;
|
|
|
|
/**
|
|
* Set the negative prefix.
|
|
*
|
|
* @param newValue the new value of the the negative prefix to be set.
|
|
* Examples: -123, ($123) (with negative suffix), sFr-123
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setNegativePrefix(const UnicodeString& newValue);
|
|
|
|
/**
|
|
* Get the positive suffix.
|
|
*
|
|
* @param result Output param which will receive the positive suffix.
|
|
* @return A reference to 'result'.
|
|
* Example: 123%
|
|
* @stable ICU 2.0
|
|
*/
|
|
UnicodeString& getPositiveSuffix(UnicodeString& result) const;
|
|
|
|
/**
|
|
* Set the positive suffix.
|
|
*
|
|
* @param newValue the new value of the positive suffix to be set.
|
|
* Example: 123%
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setPositiveSuffix(const UnicodeString& newValue);
|
|
|
|
/**
|
|
* Get the negative suffix.
|
|
*
|
|
* @param result Output param which will receive the negative suffix.
|
|
* @return A reference to 'result'.
|
|
* Examples: -123%, ($123) (with positive suffixes)
|
|
* @stable ICU 2.0
|
|
*/
|
|
UnicodeString& getNegativeSuffix(UnicodeString& result) const;
|
|
|
|
/**
|
|
* Set the negative suffix.
|
|
*
|
|
* @param newValue the new value of the negative suffix to be set.
|
|
* Examples: 123%
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setNegativeSuffix(const UnicodeString& newValue);
|
|
|
|
/**
|
|
* Get the multiplier for use in percent, permill, etc.
|
|
* For a percentage, set the suffixes to have "%" and the multiplier to be 100.
|
|
* (For Arabic, use arabic percent symbol).
|
|
* For a permill, set the suffixes to have "\\u2031" and the multiplier to be 1000.
|
|
*
|
|
* @return the multiplier for use in percent, permill, etc.
|
|
* Examples: with 100, 1.23 -> "123", and "123" -> 1.23
|
|
* @stable ICU 2.0
|
|
*/
|
|
int32_t getMultiplier(void) const;
|
|
|
|
/**
|
|
* Set the multiplier for use in percent, permill, etc.
|
|
* For a percentage, set the suffixes to have "%" and the multiplier to be 100.
|
|
* (For Arabic, use arabic percent symbol).
|
|
* For a permill, set the suffixes to have "\\u2031" and the multiplier to be 1000.
|
|
*
|
|
* @param newValue the new value of the multiplier for use in percent, permill, etc.
|
|
* Examples: with 100, 1.23 -> "123", and "123" -> 1.23
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setMultiplier(int32_t newValue);
|
|
|
|
/**
|
|
* Get the rounding increment.
|
|
* @return A positive rounding increment, or 0.0 if a custom rounding
|
|
* increment is not in effect.
|
|
* @see #setRoundingIncrement
|
|
* @see #getRoundingMode
|
|
* @see #setRoundingMode
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual double getRoundingIncrement(void) const;
|
|
|
|
/**
|
|
* Set the rounding increment. In the absence of a rounding increment,
|
|
* numbers will be rounded to the number of digits displayed.
|
|
* @param newValue A positive rounding increment, or 0.0 to
|
|
* use the default rounding increment.
|
|
* Negative increments are equivalent to 0.0.
|
|
* @see #getRoundingIncrement
|
|
* @see #getRoundingMode
|
|
* @see #setRoundingMode
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setRoundingIncrement(double newValue);
|
|
|
|
/**
|
|
* Get the rounding mode.
|
|
* @return A rounding mode
|
|
* @see #setRoundingIncrement
|
|
* @see #getRoundingIncrement
|
|
* @see #setRoundingMode
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual ERoundingMode getRoundingMode(void) const;
|
|
|
|
/**
|
|
* Set the rounding mode.
|
|
* @param roundingMode A rounding mode
|
|
* @see #setRoundingIncrement
|
|
* @see #getRoundingIncrement
|
|
* @see #getRoundingMode
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setRoundingMode(ERoundingMode roundingMode);
|
|
|
|
/**
|
|
* Get the width to which the output of format() is padded.
|
|
* The width is counted in 16-bit code units.
|
|
* @return the format width, or zero if no padding is in effect
|
|
* @see #setFormatWidth
|
|
* @see #getPadCharacterString
|
|
* @see #setPadCharacter
|
|
* @see #getPadPosition
|
|
* @see #setPadPosition
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual int32_t getFormatWidth(void) const;
|
|
|
|
/**
|
|
* Set the width to which the output of format() is padded.
|
|
* The width is counted in 16-bit code units.
|
|
* This method also controls whether padding is enabled.
|
|
* @param width the width to which to pad the result of
|
|
* format(), or zero to disable padding. A negative
|
|
* width is equivalent to 0.
|
|
* @see #getFormatWidth
|
|
* @see #getPadCharacterString
|
|
* @see #setPadCharacter
|
|
* @see #getPadPosition
|
|
* @see #setPadPosition
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setFormatWidth(int32_t width);
|
|
|
|
/**
|
|
* Get the pad character used to pad to the format width. The
|
|
* default is ' '.
|
|
* @return a string containing the pad character. This will always
|
|
* have a length of one 32-bit code point.
|
|
* @see #setFormatWidth
|
|
* @see #getFormatWidth
|
|
* @see #setPadCharacter
|
|
* @see #getPadPosition
|
|
* @see #setPadPosition
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual UnicodeString getPadCharacterString() const;
|
|
|
|
/**
|
|
* Set the character used to pad to the format width. If padding
|
|
* is not enabled, then this will take effect if padding is later
|
|
* enabled.
|
|
* @param padChar a string containing the pad charcter. If the string
|
|
* has length 0, then the pad characer is set to ' '. Otherwise
|
|
* padChar.char32At(0) will be used as the pad character.
|
|
* @see #setFormatWidth
|
|
* @see #getFormatWidth
|
|
* @see #getPadCharacterString
|
|
* @see #getPadPosition
|
|
* @see #setPadPosition
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setPadCharacter(const UnicodeString &padChar);
|
|
|
|
/**
|
|
* Get the position at which padding will take place. This is the location
|
|
* at which padding will be inserted if the result of format()
|
|
* is shorter than the format width.
|
|
* @return the pad position, one of kPadBeforePrefix,
|
|
* kPadAfterPrefix, kPadBeforeSuffix, or
|
|
* kPadAfterSuffix.
|
|
* @see #setFormatWidth
|
|
* @see #getFormatWidth
|
|
* @see #setPadCharacter
|
|
* @see #getPadCharacterString
|
|
* @see #setPadPosition
|
|
* @see #EPadPosition
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual EPadPosition getPadPosition(void) const;
|
|
|
|
/**
|
|
* Set the position at which padding will take place. This is the location
|
|
* at which padding will be inserted if the result of format()
|
|
* is shorter than the format width. This has no effect unless padding is
|
|
* enabled.
|
|
* @param padPos the pad position, one of kPadBeforePrefix,
|
|
* kPadAfterPrefix, kPadBeforeSuffix, or
|
|
* kPadAfterSuffix.
|
|
* @see #setFormatWidth
|
|
* @see #getFormatWidth
|
|
* @see #setPadCharacter
|
|
* @see #getPadCharacterString
|
|
* @see #getPadPosition
|
|
* @see #EPadPosition
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setPadPosition(EPadPosition padPos);
|
|
|
|
/**
|
|
* Return whether or not scientific notation is used.
|
|
* @return TRUE if this object formats and parses scientific notation
|
|
* @see #setScientificNotation
|
|
* @see #getMinimumExponentDigits
|
|
* @see #setMinimumExponentDigits
|
|
* @see #isExponentSignAlwaysShown
|
|
* @see #setExponentSignAlwaysShown
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual UBool isScientificNotation(void) const;
|
|
|
|
/**
|
|
* Set whether or not scientific notation is used. When scientific notation
|
|
* is used, the effective maximum number of integer digits is <= 8. If the
|
|
* maximum number of integer digits is set to more than 8, the effective
|
|
* maximum will be 1. This allows this call to generate a 'default' scientific
|
|
* number format without additional changes.
|
|
* @param useScientific TRUE if this object formats and parses scientific
|
|
* notation
|
|
* @see #isScientificNotation
|
|
* @see #getMinimumExponentDigits
|
|
* @see #setMinimumExponentDigits
|
|
* @see #isExponentSignAlwaysShown
|
|
* @see #setExponentSignAlwaysShown
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setScientificNotation(UBool useScientific);
|
|
|
|
/**
|
|
* Return the minimum exponent digits that will be shown.
|
|
* @return the minimum exponent digits that will be shown
|
|
* @see #setScientificNotation
|
|
* @see #isScientificNotation
|
|
* @see #setMinimumExponentDigits
|
|
* @see #isExponentSignAlwaysShown
|
|
* @see #setExponentSignAlwaysShown
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual int8_t getMinimumExponentDigits(void) const;
|
|
|
|
/**
|
|
* Set the minimum exponent digits that will be shown. This has no
|
|
* effect unless scientific notation is in use.
|
|
* @param minExpDig a value >= 1 indicating the fewest exponent digits
|
|
* that will be shown. Values less than 1 will be treated as 1.
|
|
* @see #setScientificNotation
|
|
* @see #isScientificNotation
|
|
* @see #getMinimumExponentDigits
|
|
* @see #isExponentSignAlwaysShown
|
|
* @see #setExponentSignAlwaysShown
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setMinimumExponentDigits(int8_t minExpDig);
|
|
|
|
/**
|
|
* Return whether the exponent sign is always shown.
|
|
* @return TRUE if the exponent is always prefixed with either the
|
|
* localized minus sign or the localized plus sign, false if only negative
|
|
* exponents are prefixed with the localized minus sign.
|
|
* @see #setScientificNotation
|
|
* @see #isScientificNotation
|
|
* @see #setMinimumExponentDigits
|
|
* @see #getMinimumExponentDigits
|
|
* @see #setExponentSignAlwaysShown
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual UBool isExponentSignAlwaysShown(void) const;
|
|
|
|
/**
|
|
* Set whether the exponent sign is always shown. This has no effect
|
|
* unless scientific notation is in use.
|
|
* @param expSignAlways TRUE if the exponent is always prefixed with either
|
|
* the localized minus sign or the localized plus sign, false if only
|
|
* negative exponents are prefixed with the localized minus sign.
|
|
* @see #setScientificNotation
|
|
* @see #isScientificNotation
|
|
* @see #setMinimumExponentDigits
|
|
* @see #getMinimumExponentDigits
|
|
* @see #isExponentSignAlwaysShown
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setExponentSignAlwaysShown(UBool expSignAlways);
|
|
|
|
/**
|
|
* Return the grouping size. Grouping size is the number of digits between
|
|
* grouping separators in the integer portion of a number. For example,
|
|
* in the number "123,456.78", the grouping size is 3.
|
|
*
|
|
* @return the grouping size.
|
|
* @see setGroupingSize
|
|
* @see NumberFormat::isGroupingUsed
|
|
* @see DecimalFormatSymbols::getGroupingSeparator
|
|
* @stable ICU 2.0
|
|
*/
|
|
int32_t getGroupingSize(void) const;
|
|
|
|
/**
|
|
* Set the grouping size. Grouping size is the number of digits between
|
|
* grouping separators in the integer portion of a number. For example,
|
|
* in the number "123,456.78", the grouping size is 3.
|
|
*
|
|
* @param newValue the new value of the grouping size.
|
|
* @see getGroupingSize
|
|
* @see NumberFormat::setGroupingUsed
|
|
* @see DecimalFormatSymbols::setGroupingSeparator
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setGroupingSize(int32_t newValue);
|
|
|
|
/**
|
|
* Return the secondary grouping size. In some locales one
|
|
* grouping interval is used for the least significant integer
|
|
* digits (the primary grouping size), and another is used for all
|
|
* others (the secondary grouping size). A formatter supporting a
|
|
* secondary grouping size will return a positive integer unequal
|
|
* to the primary grouping size returned by
|
|
* getGroupingSize(). For example, if the primary
|
|
* grouping size is 4, and the secondary grouping size is 2, then
|
|
* the number 123456789 formats as "1,23,45,6789", and the pattern
|
|
* appears as "#,##,###0".
|
|
* @return the secondary grouping size, or a value less than
|
|
* one if there is none
|
|
* @see setSecondaryGroupingSize
|
|
* @see NumberFormat::isGroupingUsed
|
|
* @see DecimalFormatSymbols::getGroupingSeparator
|
|
* @stable ICU 2.4
|
|
*/
|
|
int32_t getSecondaryGroupingSize(void) const;
|
|
|
|
/**
|
|
* Set the secondary grouping size. If set to a value less than 1,
|
|
* then secondary grouping is turned off, and the primary grouping
|
|
* size is used for all intervals, not just the least significant.
|
|
*
|
|
* @param newValue the new value of the secondary grouping size.
|
|
* @see getSecondaryGroupingSize
|
|
* @see NumberFormat#setGroupingUsed
|
|
* @see DecimalFormatSymbols::setGroupingSeparator
|
|
* @stable ICU 2.4
|
|
*/
|
|
virtual void setSecondaryGroupingSize(int32_t newValue);
|
|
|
|
/**
|
|
* Allows you to get the behavior of the decimal separator with integers.
|
|
* (The decimal separator will always appear with decimals.)
|
|
*
|
|
* @return TRUE if the decimal separator always appear with decimals.
|
|
* Example: Decimal ON: 12345 -> 12345.; OFF: 12345 -> 12345
|
|
* @stable ICU 2.0
|
|
*/
|
|
UBool isDecimalSeparatorAlwaysShown(void) const;
|
|
|
|
/**
|
|
* Allows you to set the behavior of the decimal separator with integers.
|
|
* (The decimal separator will always appear with decimals.)
|
|
*
|
|
* @param newValue set TRUE if the decimal separator will always appear with decimals.
|
|
* Example: Decimal ON: 12345 -> 12345.; OFF: 12345 -> 12345
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setDecimalSeparatorAlwaysShown(UBool newValue);
|
|
|
|
/**
|
|
* Synthesizes a pattern string that represents the current state
|
|
* of this Format object.
|
|
*
|
|
* @param result Output param which will receive the pattern.
|
|
* Previous contents are deleted.
|
|
* @return A reference to 'result'.
|
|
* @see applyPattern
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual UnicodeString& toPattern(UnicodeString& result) const;
|
|
|
|
/**
|
|
* Synthesizes a localized pattern string that represents the current
|
|
* state of this Format object.
|
|
*
|
|
* @param result Output param which will receive the localized pattern.
|
|
* Previous contents are deleted.
|
|
* @return A reference to 'result'.
|
|
* @see applyPattern
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual UnicodeString& toLocalizedPattern(UnicodeString& result) const;
|
|
|
|
/**
|
|
* Apply the given pattern to this Format object. A pattern is a
|
|
* short-hand specification for the various formatting properties.
|
|
* These properties can also be changed individually through the
|
|
* various setter methods.
|
|
* <P>
|
|
* There is no limit to integer digits are set
|
|
* by this routine, since that is the typical end-user desire;
|
|
* use setMaximumInteger if you want to set a real value.
|
|
* For negative numbers, use a second pattern, separated by a semicolon
|
|
* <pre>
|
|
* . Example "#,#00.0#" -> 1,234.56
|
|
* </pre>
|
|
* This means a minimum of 2 integer digits, 1 fraction digit, and
|
|
* a maximum of 2 fraction digits.
|
|
* <pre>
|
|
* . Example: "#,#00.0#;(#,#00.0#)" for negatives in parantheses.
|
|
* </pre>
|
|
* In negative patterns, the minimum and maximum counts are ignored;
|
|
* these are presumed to be set in the positive pattern.
|
|
*
|
|
* @param pattern The pattern to be applied.
|
|
* @param parseError Struct to recieve information on position
|
|
* of error if an error is encountered
|
|
* @param status Output param set to success/failure code on
|
|
* exit. If the pattern is invalid, this will be
|
|
* set to a failure result.
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void applyPattern(const UnicodeString& pattern,
|
|
UParseError& parseError,
|
|
UErrorCode& status);
|
|
/**
|
|
* Sets the pattern.
|
|
* @param pattern The pattern to be applied.
|
|
* @param status Output param set to success/failure code on
|
|
* exit. If the pattern is invalid, this will be
|
|
* set to a failure result.
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void applyPattern(const UnicodeString& pattern,
|
|
UErrorCode& status);
|
|
|
|
/**
|
|
* Apply the given pattern to this Format object. The pattern
|
|
* is assumed to be in a localized notation. A pattern is a
|
|
* short-hand specification for the various formatting properties.
|
|
* These properties can also be changed individually through the
|
|
* various setter methods.
|
|
* <P>
|
|
* There is no limit to integer digits are set
|
|
* by this routine, since that is the typical end-user desire;
|
|
* use setMaximumInteger if you want to set a real value.
|
|
* For negative numbers, use a second pattern, separated by a semicolon
|
|
* <pre>
|
|
* . Example "#,#00.0#" -> 1,234.56
|
|
* </pre>
|
|
* This means a minimum of 2 integer digits, 1 fraction digit, and
|
|
* a maximum of 2 fraction digits.
|
|
*
|
|
* Example: "#,#00.0#;(#,#00.0#)" for negatives in parantheses.
|
|
*
|
|
* In negative patterns, the minimum and maximum counts are ignored;
|
|
* these are presumed to be set in the positive pattern.
|
|
*
|
|
* @param pattern The localized pattern to be applied.
|
|
* @param parseError Struct to recieve information on position
|
|
* of error if an error is encountered
|
|
* @param status Output param set to success/failure code on
|
|
* exit. If the pattern is invalid, this will be
|
|
* set to a failure result.
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void applyLocalizedPattern(const UnicodeString& pattern,
|
|
UParseError& parseError,
|
|
UErrorCode& status);
|
|
|
|
/**
|
|
* Apply the given pattern to this Format object.
|
|
*
|
|
* @param pattern The localized pattern to be applied.
|
|
* @param status Output param set to success/failure code on
|
|
* exit. If the pattern is invalid, this will be
|
|
* set to a failure result.
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void applyLocalizedPattern(const UnicodeString& pattern,
|
|
UErrorCode& status);
|
|
|
|
|
|
/**
|
|
* Sets the maximum number of digits allowed in the integer portion of a
|
|
* number. This override limits the integer digit count to 309.
|
|
*
|
|
* @param newValue the new value of the maximum number of digits
|
|
* allowed in the integer portion of a number.
|
|
* @see NumberFormat#setMaximumIntegerDigits
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setMaximumIntegerDigits(int32_t newValue);
|
|
|
|
/**
|
|
* Sets the minimum number of digits allowed in the integer portion of a
|
|
* number. This override limits the integer digit count to 309.
|
|
*
|
|
* @param newValue the new value of the minimum number of digits
|
|
* allowed in the integer portion of a number.
|
|
* @see NumberFormat#setMinimumIntegerDigits
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setMinimumIntegerDigits(int32_t newValue);
|
|
|
|
/**
|
|
* Sets the maximum number of digits allowed in the fraction portion of a
|
|
* number. This override limits the fraction digit count to 340.
|
|
*
|
|
* @param newValue the new value of the maximum number of digits
|
|
* allowed in the fraction portion of a number.
|
|
* @see NumberFormat#setMaximumFractionDigits
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setMaximumFractionDigits(int32_t newValue);
|
|
|
|
/**
|
|
* Sets the minimum number of digits allowed in the fraction portion of a
|
|
* number. This override limits the fraction digit count to 340.
|
|
*
|
|
* @param newValue the new value of the minimum number of digits
|
|
* allowed in the fraction portion of a number.
|
|
* @see NumberFormat#setMinimumFractionDigits
|
|
* @stable ICU 2.0
|
|
*/
|
|
virtual void setMinimumFractionDigits(int32_t newValue);
|
|
|
|
/**
|
|
* Returns the minimum number of significant digits that will be
|
|
* displayed. This value has no effect unless areSignificantDigitsUsed()
|
|
* returns true.
|
|
* @return the fewest significant digits that will be shown
|
|
* @stable ICU 3.0
|
|
*/
|
|
int32_t getMinimumSignificantDigits() const;
|
|
|
|
/**
|
|
* Returns the maximum number of significant digits that will be
|
|
* displayed. This value has no effect unless areSignificantDigitsUsed()
|
|
* returns true.
|
|
* @return the most significant digits that will be shown
|
|
* @stable ICU 3.0
|
|
*/
|
|
int32_t getMaximumSignificantDigits() const;
|
|
|
|
/**
|
|
* Sets the minimum number of significant digits that will be
|
|
* displayed. If <code>min</code> is less than one then it is set
|
|
* to one. If the maximum significant digits count is less than
|
|
* <code>min</code>, then it is set to <code>min</code>.
|
|
* This function also enables the use of significant digits
|
|
* by this formatter - areSignificantDigitsUsed() will return TRUE.
|
|
* @see #areSignificantDigitsUsed
|
|
* @param min the fewest significant digits to be shown
|
|
* @stable ICU 3.0
|
|
*/
|
|
void setMinimumSignificantDigits(int32_t min);
|
|
|
|
/**
|
|
* Sets the maximum number of significant digits that will be
|
|
* displayed. If <code>max</code> is less than one then it is set
|
|
* to one. If the minimum significant digits count is greater
|
|
* than <code>max</code>, then it is set to <code>max</code>.
|
|
* This function also enables the use of significant digits
|
|
* by this formatter - areSignificantDigitsUsed() will return TRUE.
|
|
* @see #areSignificantDigitsUsed
|
|
* @param max the most significant digits to be shown
|
|
* @stable ICU 3.0
|
|
*/
|
|
void setMaximumSignificantDigits(int32_t max);
|
|
|
|
/**
|
|
* Returns true if significant digits are in use, or false if
|
|
* integer and fraction digit counts are in use.
|
|
* @return true if significant digits are in use
|
|
* @stable ICU 3.0
|
|
*/
|
|
UBool areSignificantDigitsUsed() const;
|
|
|
|
/**
|
|
* Sets whether significant digits are in use, or integer and
|
|
* fraction digit counts are in use.
|
|
* @param useSignificantDigits true to use significant digits, or
|
|
* false to use integer and fraction digit counts
|
|
* @stable ICU 3.0
|
|
*/
|
|
void setSignificantDigitsUsed(UBool useSignificantDigits);
|
|
|
|
public:
|
|
/**
|
|
* 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 through the application of a new pattern.
|
|
* @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 UChar* theCurrency, UErrorCode& ec);
|
|
|
|
/**
|
|
* Sets the currency used to display currency amounts. See
|
|
* setCurrency(const UChar*, UErrorCode&).
|
|
* @deprecated ICU 3.0. Use setCurrency(const UChar*, UErrorCode&).
|
|
*/
|
|
virtual void setCurrency(const UChar* theCurrency);
|
|
|
|
/**
|
|
* The resource tags we use to retrieve decimal format data from
|
|
* locale resource bundles.
|
|
* @deprecated ICU 3.4. This string has no public purpose. Please don't use it.
|
|
*/
|
|
static const char fgNumberPatterns[];
|
|
|
|
#ifndef U_HIDE_INTERNAL_API
|
|
/**
|
|
* Get a FixedDecimal corresponding to a double as it would be
|
|
* formatted by this DecimalFormat.
|
|
* Internal, not intended for public use.
|
|
* @internal
|
|
*/
|
|
FixedDecimal getFixedDecimal(double number, UErrorCode &status) const;
|
|
|
|
/**
|
|
* Get a FixedDecimal corresponding to a formattable as it would be
|
|
* formatted by this DecimalFormat.
|
|
* Internal, not intended for public use.
|
|
* @internal
|
|
*/
|
|
FixedDecimal getFixedDecimal(const Formattable &number, UErrorCode &status) const;
|
|
|
|
/**
|
|
* Get a FixedDecimal corresponding to a DigitList as it would be
|
|
* formatted by this DecimalFormat. Note: the DigitList may be modified.
|
|
* Internal, not intended for public use.
|
|
* @internal
|
|
*/
|
|
FixedDecimal getFixedDecimal(DigitList &number, UErrorCode &status) const;
|
|
#endif /* U_HIDE_INTERNAL_API */
|
|
|
|
public:
|
|
|
|
/**
|
|
* Return the class ID for this class. This is useful only for
|
|
* comparing to a return value from getDynamicClassID(). For example:
|
|
* <pre>
|
|
* . Base* polymorphic_pointer = createPolymorphicObject();
|
|
* . if (polymorphic_pointer->getDynamicClassID() ==
|
|
* . Derived::getStaticClassID()) ...
|
|
* </pre>
|
|
* @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.
|
|
*
|
|
* @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;
|
|
|
|
private:
|
|
|
|
DecimalFormat(); // default constructor not implemented
|
|
|
|
int32_t precision() const;
|
|
|
|
/**
|
|
* Initialize all fields of a new DecimalFormatter to a safe default value.
|
|
* Common code for use by constructors.
|
|
*/
|
|
void init();
|
|
|
|
/**
|
|
* Do real work of constructing a new DecimalFormat.
|
|
*/
|
|
void construct(UErrorCode& status,
|
|
UParseError& parseErr,
|
|
const UnicodeString* pattern = 0,
|
|
DecimalFormatSymbols* symbolsToAdopt = 0
|
|
);
|
|
|
|
/**
|
|
* Does the real work of generating a pattern.
|
|
*
|
|
* @param result Output param which will receive the pattern.
|
|
* Previous contents are deleted.
|
|
* @param localized TRUE return localized pattern.
|
|
* @return A reference to 'result'.
|
|
*/
|
|
UnicodeString& toPattern(UnicodeString& result, UBool localized) const;
|
|
|
|
/**
|
|
* Does the real work of applying a pattern.
|
|
* @param pattern The pattern to be applied.
|
|
* @param localized If true, the pattern is localized; else false.
|
|
* @param parseError Struct to recieve information on position
|
|
* of error if an error is encountered
|
|
* @param status Output param set to success/failure code on
|
|
* exit. If the pattern is invalid, this will be
|
|
* set to a failure result.
|
|
*/
|
|
void applyPattern(const UnicodeString& pattern,
|
|
UBool localized,
|
|
UParseError& parseError,
|
|
UErrorCode& status);
|
|
|
|
/*
|
|
* similar to applyPattern, but without re-gen affix for currency
|
|
*/
|
|
void applyPatternInternally(const UnicodeString& pluralCount,
|
|
const UnicodeString& pattern,
|
|
UBool localized,
|
|
UParseError& parseError,
|
|
UErrorCode& status);
|
|
|
|
/*
|
|
* only apply pattern without expand affixes
|
|
*/
|
|
void applyPatternWithoutExpandAffix(const UnicodeString& pattern,
|
|
UBool localized,
|
|
UParseError& parseError,
|
|
UErrorCode& status);
|
|
|
|
|
|
/*
|
|
* expand affixes (after apply patter) and re-compute fFormatWidth
|
|
*/
|
|
void expandAffixAdjustWidth(const UnicodeString* pluralCount);
|
|
|
|
|
|
/**
|
|
* Do the work of formatting a number, either a double or a long.
|
|
*
|
|
* @param appendTo Output parameter to receive result.
|
|
* Result is appended to existing contents.
|
|
* @param handler Records information about field positions.
|
|
* @param digits the digits to be formatted.
|
|
* @param isInteger if TRUE format the digits as Integer.
|
|
* @return Reference to 'appendTo' parameter.
|
|
*/
|
|
UnicodeString& subformat(UnicodeString& appendTo,
|
|
FieldPositionHandler& handler,
|
|
DigitList& digits,
|
|
UBool isInteger,
|
|
UErrorCode &status) const;
|
|
|
|
|
|
void parse(const UnicodeString& text,
|
|
Formattable& result,
|
|
ParsePosition& pos,
|
|
UChar* currency) const;
|
|
|
|
enum {
|
|
fgStatusInfinite,
|
|
fgStatusLength // Leave last in list.
|
|
} StatusFlags;
|
|
|
|
UBool subparse(const UnicodeString& text,
|
|
const UnicodeString* negPrefix,
|
|
const UnicodeString* negSuffix,
|
|
const UnicodeString* posPrefix,
|
|
const UnicodeString* posSuffix,
|
|
UBool complexCurrencyParsing,
|
|
int8_t type,
|
|
ParsePosition& parsePosition,
|
|
DigitList& digits, UBool* status,
|
|
UChar* currency) const;
|
|
|
|
// Mixed style parsing for currency.
|
|
// It parses against the current currency pattern
|
|
// using complex affix comparison
|
|
// parses against the currency plural patterns using complex affix comparison,
|
|
// and parses against the current pattern using simple affix comparison.
|
|
UBool parseForCurrency(const UnicodeString& text,
|
|
ParsePosition& parsePosition,
|
|
DigitList& digits,
|
|
UBool* status,
|
|
UChar* currency) const;
|
|
|
|
int32_t skipPadding(const UnicodeString& text, int32_t position) const;
|
|
|
|
int32_t compareAffix(const UnicodeString& input,
|
|
int32_t pos,
|
|
UBool isNegative,
|
|
UBool isPrefix,
|
|
const UnicodeString* affixPat,
|
|
UBool complexCurrencyParsing,
|
|
int8_t type,
|
|
UChar* currency) const;
|
|
|
|
static UnicodeString& trimMarksFromAffix(const UnicodeString& affix, UnicodeString& trimmedAffix);
|
|
|
|
UBool equalWithSignCompatibility(UChar32 lhs, UChar32 rhs) const;
|
|
|
|
int32_t compareSimpleAffix(const UnicodeString& affix,
|
|
const UnicodeString& input,
|
|
int32_t pos,
|
|
UBool lenient) const;
|
|
|
|
static int32_t skipPatternWhiteSpace(const UnicodeString& text, int32_t pos);
|
|
|
|
static int32_t skipUWhiteSpace(const UnicodeString& text, int32_t pos);
|
|
|
|
static int32_t skipUWhiteSpaceAndMarks(const UnicodeString& text, int32_t pos);
|
|
|
|
static int32_t skipBidiMarks(const UnicodeString& text, int32_t pos);
|
|
|
|
int32_t compareComplexAffix(const UnicodeString& affixPat,
|
|
const UnicodeString& input,
|
|
int32_t pos,
|
|
int8_t type,
|
|
UChar* currency) const;
|
|
|
|
static int32_t match(const UnicodeString& text, int32_t pos, UChar32 ch);
|
|
|
|
static int32_t match(const UnicodeString& text, int32_t pos, const UnicodeString& str);
|
|
|
|
static UBool matchSymbol(const UnicodeString &text, int32_t position, int32_t length, const UnicodeString &symbol,
|
|
UnicodeSet *sset, UChar32 schar);
|
|
|
|
static UBool matchDecimal(UChar32 symbolChar,
|
|
UBool sawDecimal, UChar32 sawDecimalChar,
|
|
const UnicodeSet *sset, UChar32 schar);
|
|
|
|
static UBool matchGrouping(UChar32 groupingChar,
|
|
UBool sawGrouping, UChar32 sawGroupingChar,
|
|
const UnicodeSet *sset,
|
|
UChar32 decimalChar, const UnicodeSet *decimalSet,
|
|
UChar32 schar);
|
|
|
|
/**
|
|
* Get a decimal format symbol.
|
|
* Returns a const reference to the symbol string.
|
|
* @internal
|
|
*/
|
|
inline const UnicodeString &getConstSymbol(DecimalFormatSymbols::ENumberFormatSymbol symbol) const;
|
|
|
|
int32_t appendAffix(UnicodeString& buf,
|
|
double number,
|
|
FieldPositionHandler& handler,
|
|
UBool isNegative,
|
|
UBool isPrefix) const;
|
|
|
|
/**
|
|
* Append an affix to the given UnicodeString, using quotes if
|
|
* there are special characters. Single quotes themselves must be
|
|
* escaped in either case.
|
|
*/
|
|
void appendAffixPattern(UnicodeString& appendTo, const UnicodeString& affix,
|
|
UBool localized) const;
|
|
|
|
void appendAffixPattern(UnicodeString& appendTo,
|
|
const UnicodeString* affixPattern,
|
|
const UnicodeString& expAffix, UBool localized) const;
|
|
|
|
void expandAffix(const UnicodeString& pattern,
|
|
UnicodeString& affix,
|
|
double number,
|
|
FieldPositionHandler& handler,
|
|
UBool doFormat,
|
|
const UnicodeString* pluralCount) const;
|
|
|
|
void expandAffixes(const UnicodeString* pluralCount);
|
|
|
|
void addPadding(UnicodeString& appendTo,
|
|
FieldPositionHandler& handler,
|
|
int32_t prefixLen, int32_t suffixLen) const;
|
|
|
|
UBool isGroupingPosition(int32_t pos) const;
|
|
|
|
void setCurrencyForSymbols();
|
|
|
|
// similar to setCurrency without re-compute the affixes for currency.
|
|
// If currency changes, the affix pattern for currency is not changed,
|
|
// but the affix will be changed. So, affixes need to be
|
|
// re-computed in setCurrency(), but not in setCurrencyInternally().
|
|
virtual void setCurrencyInternally(const UChar* theCurrency, UErrorCode& ec);
|
|
|
|
// set up currency affix patterns for mix parsing.
|
|
// The patterns saved here are the affix patterns of default currency
|
|
// pattern and the unique affix patterns of the plural currency patterns.
|
|
// Those patterns are used by parseForCurrency().
|
|
void setupCurrencyAffixPatterns(UErrorCode& status);
|
|
|
|
// set up the currency affixes used in currency plural formatting.
|
|
// It sets up both fAffixesForCurrency for currency pattern if the current
|
|
// pattern contains 3 currency signs,
|
|
// and it sets up fPluralAffixesForCurrency for currency plural patterns.
|
|
void setupCurrencyAffixes(const UnicodeString& pattern,
|
|
UBool setupForCurrentPattern,
|
|
UBool setupForPluralPattern,
|
|
UErrorCode& status);
|
|
|
|
// hashtable operations
|
|
Hashtable* initHashForAffixPattern(UErrorCode& status);
|
|
Hashtable* initHashForAffix(UErrorCode& status);
|
|
|
|
void deleteHashForAffixPattern();
|
|
void deleteHashForAffix(Hashtable*& table);
|
|
|
|
void copyHashForAffixPattern(const Hashtable* source,
|
|
Hashtable* target, UErrorCode& status);
|
|
void copyHashForAffix(const Hashtable* source,
|
|
Hashtable* target, UErrorCode& status);
|
|
|
|
UnicodeString& _format(int64_t number,
|
|
UnicodeString& appendTo,
|
|
FieldPositionHandler& handler,
|
|
UErrorCode &status) const;
|
|
UnicodeString& _format(double number,
|
|
UnicodeString& appendTo,
|
|
FieldPositionHandler& handler,
|
|
UErrorCode &status) const;
|
|
UnicodeString& _format(const DigitList &number,
|
|
UnicodeString& appendTo,
|
|
FieldPositionHandler& handler,
|
|
UErrorCode &status) const;
|
|
|
|
/**
|
|
* Constants.
|
|
*/
|
|
|
|
UnicodeString fPositivePrefix;
|
|
UnicodeString fPositiveSuffix;
|
|
UnicodeString fNegativePrefix;
|
|
UnicodeString fNegativeSuffix;
|
|
UnicodeString* fPosPrefixPattern;
|
|
UnicodeString* fPosSuffixPattern;
|
|
UnicodeString* fNegPrefixPattern;
|
|
UnicodeString* fNegSuffixPattern;
|
|
|
|
/**
|
|
* Formatter for ChoiceFormat-based currency names. If this field
|
|
* is not null, then delegate to it to format currency symbols.
|
|
* @since ICU 2.6
|
|
*/
|
|
ChoiceFormat* fCurrencyChoice;
|
|
|
|
DigitList * fMultiplier; // NULL for multiplier of one
|
|
int32_t fScale;
|
|
int32_t fGroupingSize;
|
|
int32_t fGroupingSize2;
|
|
UBool fDecimalSeparatorAlwaysShown;
|
|
DecimalFormatSymbols* fSymbols;
|
|
|
|
UBool fUseSignificantDigits;
|
|
int32_t fMinSignificantDigits;
|
|
int32_t fMaxSignificantDigits;
|
|
|
|
UBool fUseExponentialNotation;
|
|
int8_t fMinExponentDigits;
|
|
UBool fExponentSignAlwaysShown;
|
|
|
|
EnumSet<UNumberFormatAttribute,
|
|
UNUM_MAX_NONBOOLEAN_ATTRIBUTE+1,
|
|
UNUM_LIMIT_BOOLEAN_ATTRIBUTE>
|
|
fBoolFlags;
|
|
|
|
DigitList* fRoundingIncrement; // NULL if no rounding increment specified.
|
|
ERoundingMode fRoundingMode;
|
|
|
|
UChar32 fPad;
|
|
int32_t fFormatWidth;
|
|
EPadPosition fPadPosition;
|
|
|
|
/*
|
|
* Following are used for currency format
|
|
*/
|
|
// pattern used in this formatter
|
|
UnicodeString fFormatPattern;
|
|
// style is only valid when decimal formatter is constructed by
|
|
// DecimalFormat(pattern, decimalFormatSymbol, style)
|
|
int fStyle;
|
|
/*
|
|
* Represents whether this is a currency format, and which
|
|
* currency format style.
|
|
* 0: not currency format type;
|
|
* 1: currency style -- symbol name, such as "$" for US dollar.
|
|
* 2: currency style -- ISO name, such as USD for US dollar.
|
|
* 3: currency style -- plural long name, such as "US Dollar" for
|
|
* "1.00 US Dollar", or "US Dollars" for
|
|
* "3.00 US Dollars".
|
|
*/
|
|
int fCurrencySignCount;
|
|
|
|
|
|
/* For currency parsing purose,
|
|
* Need to remember all prefix patterns and suffix patterns of
|
|
* every currency format pattern,
|
|
* including the pattern of default currecny style
|
|
* and plural currency style. And the patterns are set through applyPattern.
|
|
*/
|
|
// TODO: innerclass?
|
|
/* This is not needed in the class declaration, so it is moved into decimfmp.cpp
|
|
struct AffixPatternsForCurrency : public UMemory {
|
|
// negative prefix pattern
|
|
UnicodeString negPrefixPatternForCurrency;
|
|
// negative suffix pattern
|
|
UnicodeString negSuffixPatternForCurrency;
|
|
// positive prefix pattern
|
|
UnicodeString posPrefixPatternForCurrency;
|
|
// positive suffix pattern
|
|
UnicodeString posSuffixPatternForCurrency;
|
|
int8_t patternType;
|
|
|
|
AffixPatternsForCurrency(const UnicodeString& negPrefix,
|
|
const UnicodeString& negSuffix,
|
|
const UnicodeString& posPrefix,
|
|
const UnicodeString& posSuffix,
|
|
int8_t type) {
|
|
negPrefixPatternForCurrency = negPrefix;
|
|
negSuffixPatternForCurrency = negSuffix;
|
|
posPrefixPatternForCurrency = posPrefix;
|
|
posSuffixPatternForCurrency = posSuffix;
|
|
patternType = type;
|
|
}
|
|
};
|
|
*/
|
|
|
|
/* affix for currency formatting when the currency sign in the pattern
|
|
* equals to 3, such as the pattern contains 3 currency sign or
|
|
* the formatter style is currency plural format style.
|
|
*/
|
|
/* This is not needed in the class declaration, so it is moved into decimfmp.cpp
|
|
struct AffixesForCurrency : public UMemory {
|
|
// negative prefix
|
|
UnicodeString negPrefixForCurrency;
|
|
// negative suffix
|
|
UnicodeString negSuffixForCurrency;
|
|
// positive prefix
|
|
UnicodeString posPrefixForCurrency;
|
|
// positive suffix
|
|
UnicodeString posSuffixForCurrency;
|
|
|
|
int32_t formatWidth;
|
|
|
|
AffixesForCurrency(const UnicodeString& negPrefix,
|
|
const UnicodeString& negSuffix,
|
|
const UnicodeString& posPrefix,
|
|
const UnicodeString& posSuffix) {
|
|
negPrefixForCurrency = negPrefix;
|
|
negSuffixForCurrency = negSuffix;
|
|
posPrefixForCurrency = posPrefix;
|
|
posSuffixForCurrency = posSuffix;
|
|
}
|
|
};
|
|
*/
|
|
|
|
// Affix pattern set for currency.
|
|
// It is a set of AffixPatternsForCurrency,
|
|
// each element of the set saves the negative prefix pattern,
|
|
// negative suffix pattern, positive prefix pattern,
|
|
// and positive suffix pattern of a pattern.
|
|
// It is used for currency mixed style parsing.
|
|
// It is actually is a set.
|
|
// The set contains the default currency pattern from the locale,
|
|
// and the currency plural patterns.
|
|
// Since it is a set, it does not contain duplicated items.
|
|
// For example, if 2 currency plural patterns are the same, only one pattern
|
|
// is included in the set. When parsing, we do not check whether the plural
|
|
// count match or not.
|
|
Hashtable* fAffixPatternsForCurrency;
|
|
|
|
// Following 2 are affixes for currency.
|
|
// It is a hash map from plural count to AffixesForCurrency.
|
|
// AffixesForCurrency saves the negative prefix,
|
|
// negative suffix, positive prefix, and positive suffix of a pattern.
|
|
// It is used during currency formatting only when the currency sign count
|
|
// is 3. In which case, the affixes are getting from here, not
|
|
// from the fNegativePrefix etc.
|
|
Hashtable* fAffixesForCurrency; // for current pattern
|
|
Hashtable* fPluralAffixesForCurrency; // for plural pattern
|
|
|
|
// Information needed for DecimalFormat to format/parse currency plural.
|
|
CurrencyPluralInfo* fCurrencyPluralInfo;
|
|
|
|
#if UCONFIG_HAVE_PARSEALLINPUT
|
|
UNumberFormatAttributeValue fParseAllInput;
|
|
#endif
|
|
|
|
// Decimal Format Static Sets singleton.
|
|
const DecimalFormatStaticSets *fStaticSets;
|
|
|
|
|
|
protected:
|
|
|
|
#ifndef U_HIDE_INTERNAL_API
|
|
/**
|
|
* Rounds a value according to the rules of this object.
|
|
* @internal
|
|
*/
|
|
DigitList& _round(const DigitList& number, DigitList& adjustedNum, UBool& isNegative, UErrorCode& status) const;
|
|
#endif /* U_HIDE_INTERNAL_API */
|
|
|
|
/**
|
|
* 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(UChar* result, UErrorCode& ec) const;
|
|
|
|
/** number of integer digits
|
|
* @stable ICU 2.4
|
|
*/
|
|
static const int32_t kDoubleIntegerDigits;
|
|
/** number of fraction digits
|
|
* @stable ICU 2.4
|
|
*/
|
|
static const int32_t kDoubleFractionDigits;
|
|
|
|
/**
|
|
* When someone turns on scientific mode, we assume that more than this
|
|
* number of digits is due to flipping from some other mode that didn't
|
|
* restrict the maximum, and so we force 1 integer digit. We don't bother
|
|
* to track and see if someone is using exponential notation with more than
|
|
* this number, it wouldn't make sense anyway, and this is just to make sure
|
|
* that someone turning on scientific mode with default settings doesn't
|
|
* end up with lots of zeroes.
|
|
* @stable ICU 2.8
|
|
*/
|
|
static const int32_t kMaxScientificIntegerDigits;
|
|
|
|
#if UCONFIG_FORMAT_FASTPATHS_49
|
|
private:
|
|
/**
|
|
* Internal state.
|
|
* @internal
|
|
*/
|
|
uint8_t fReserved[UNUM_DECIMALFORMAT_INTERNAL_SIZE];
|
|
|
|
|
|
/**
|
|
* Called whenever any state changes. Recomputes whether fastpath is OK to use.
|
|
*/
|
|
void handleChanged();
|
|
#endif
|
|
};
|
|
|
|
inline const UnicodeString &
|
|
DecimalFormat::getConstSymbol(DecimalFormatSymbols::ENumberFormatSymbol symbol) const {
|
|
return fSymbols->getConstSymbol(symbol);
|
|
}
|
|
|
|
U_NAMESPACE_END
|
|
|
|
#endif /* #if !UCONFIG_NO_FORMATTING */
|
|
|
|
#endif // _DECIMFMT
|
|
//eof
|