scuffed-code/icu4c/source/i18n/number_mapper.cpp
Victor Chang b62200061c
ICU-20058 Fix mimimum significant digits in engineering notation
- Follow the spec to calculate the mimimum significant digits in engineering notation
- The bug is regression since ICU 58. The new test still passes on
ICU58-based DecimalFormat
- Maximum significant digits is not changed
2018-09-27 14:27:40 -07:00

509 lines
21 KiB
C++

// © 2018 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
#include "unicode/utypes.h"
#if !UCONFIG_NO_FORMATTING
// Allow implicit conversion from char16_t* to UnicodeString for this file:
// Helpful in toString methods and elsewhere.
#define UNISTR_FROM_STRING_EXPLICIT
#include "number_mapper.h"
#include "number_patternstring.h"
#include "unicode/errorcode.h"
#include "number_utils.h"
#include "number_currencysymbols.h"
using namespace icu;
using namespace icu::number;
using namespace icu::number::impl;
UnlocalizedNumberFormatter NumberPropertyMapper::create(const DecimalFormatProperties& properties,
const DecimalFormatSymbols& symbols,
DecimalFormatWarehouse& warehouse,
UErrorCode& status) {
return NumberFormatter::with().macros(oldToNew(properties, symbols, warehouse, nullptr, status));
}
UnlocalizedNumberFormatter NumberPropertyMapper::create(const DecimalFormatProperties& properties,
const DecimalFormatSymbols& symbols,
DecimalFormatWarehouse& warehouse,
DecimalFormatProperties& exportedProperties,
UErrorCode& status) {
return NumberFormatter::with().macros(
oldToNew(
properties, symbols, warehouse, &exportedProperties, status));
}
MacroProps NumberPropertyMapper::oldToNew(const DecimalFormatProperties& properties,
const DecimalFormatSymbols& symbols,
DecimalFormatWarehouse& warehouse,
DecimalFormatProperties* exportedProperties,
UErrorCode& status) {
MacroProps macros;
Locale locale = symbols.getLocale();
/////////////
// SYMBOLS //
/////////////
macros.symbols.setTo(symbols);
//////////////////
// PLURAL RULES //
//////////////////
if (!properties.currencyPluralInfo.fPtr.isNull()) {
macros.rules = properties.currencyPluralInfo.fPtr->getPluralRules();
}
/////////////
// AFFIXES //
/////////////
AffixPatternProvider* affixProvider;
if (properties.currencyPluralInfo.fPtr.isNull()) {
warehouse.currencyPluralInfoAPP.setToBogus();
warehouse.propertiesAPP.setTo(properties, status);
affixProvider = &warehouse.propertiesAPP;
} else {
warehouse.currencyPluralInfoAPP.setTo(*properties.currencyPluralInfo.fPtr, properties, status);
warehouse.propertiesAPP.setToBogus();
affixProvider = &warehouse.currencyPluralInfoAPP;
}
macros.affixProvider = affixProvider;
///////////
// UNITS //
///////////
bool useCurrency = (
!properties.currency.isNull() || !properties.currencyPluralInfo.fPtr.isNull() ||
!properties.currencyUsage.isNull() || affixProvider->hasCurrencySign());
CurrencyUnit currency = resolveCurrency(properties, locale, status);
UCurrencyUsage currencyUsage = properties.currencyUsage.getOrDefault(UCURR_USAGE_STANDARD);
if (useCurrency) {
// NOTE: Slicing is OK.
macros.unit = currency; // NOLINT
}
warehouse.currencySymbols = {currency, locale, symbols, status};
macros.currencySymbols = &warehouse.currencySymbols;
///////////////////////
// ROUNDING STRATEGY //
///////////////////////
int32_t maxInt = properties.maximumIntegerDigits;
int32_t minInt = properties.minimumIntegerDigits;
int32_t maxFrac = properties.maximumFractionDigits;
int32_t minFrac = properties.minimumFractionDigits;
int32_t minSig = properties.minimumSignificantDigits;
int32_t maxSig = properties.maximumSignificantDigits;
double roundingIncrement = properties.roundingIncrement;
RoundingMode roundingMode = properties.roundingMode.getOrDefault(UNUM_ROUND_HALFEVEN);
bool explicitMinMaxFrac = minFrac != -1 || maxFrac != -1;
bool explicitMinMaxSig = minSig != -1 || maxSig != -1;
// Resolve min/max frac for currencies, required for the validation logic and for when minFrac or
// maxFrac was
// set (but not both) on a currency instance.
// NOTE: Increments are handled in "Precision.constructCurrency()".
if (useCurrency && (minFrac == -1 || maxFrac == -1)) {
int32_t digits = ucurr_getDefaultFractionDigitsForUsage(
currency.getISOCurrency(), currencyUsage, &status);
if (minFrac == -1 && maxFrac == -1) {
minFrac = digits;
maxFrac = digits;
} else if (minFrac == -1) {
minFrac = std::min(maxFrac, digits);
} else /* if (maxFrac == -1) */ {
maxFrac = std::max(minFrac, digits);
}
}
// Validate min/max int/frac.
// For backwards compatibility, minimum overrides maximum if the two conflict.
// The following logic ensures that there is always a minimum of at least one digit.
if (minInt == 0 && maxFrac != 0) {
// Force a digit after the decimal point.
minFrac = minFrac <= 0 ? 1 : minFrac;
maxFrac = maxFrac < 0 ? -1 : maxFrac < minFrac ? minFrac : maxFrac;
minInt = 0;
maxInt = maxInt < 0 ? -1 : maxInt > kMaxIntFracSig ? -1 : maxInt;
} else {
// Force a digit before the decimal point.
minFrac = minFrac < 0 ? 0 : minFrac;
maxFrac = maxFrac < 0 ? -1 : maxFrac < minFrac ? minFrac : maxFrac;
minInt = minInt <= 0 ? 1 : minInt > kMaxIntFracSig ? 1 : minInt;
maxInt = maxInt < 0 ? -1 : maxInt < minInt ? minInt : maxInt > kMaxIntFracSig ? -1 : maxInt;
}
Precision precision;
if (!properties.currencyUsage.isNull()) {
precision = Precision::constructCurrency(currencyUsage).withCurrency(currency);
} else if (roundingIncrement != 0.0) {
precision = Precision::constructIncrement(roundingIncrement, minFrac);
} else if (explicitMinMaxSig) {
minSig = minSig < 1 ? 1 : minSig > kMaxIntFracSig ? kMaxIntFracSig : minSig;
maxSig = maxSig < 0 ? kMaxIntFracSig : maxSig < minSig ? minSig : maxSig > kMaxIntFracSig
? kMaxIntFracSig : maxSig;
precision = Precision::constructSignificant(minSig, maxSig);
} else if (explicitMinMaxFrac) {
precision = Precision::constructFraction(minFrac, maxFrac);
} else if (useCurrency) {
precision = Precision::constructCurrency(currencyUsage);
}
if (!precision.isBogus()) {
precision = precision.withMode(roundingMode);
macros.precision = precision;
}
///////////////////
// INTEGER WIDTH //
///////////////////
macros.integerWidth = IntegerWidth(
static_cast<digits_t>(minInt),
static_cast<digits_t>(maxInt),
properties.formatFailIfMoreThanMaxDigits);
///////////////////////
// GROUPING STRATEGY //
///////////////////////
macros.grouper = Grouper::forProperties(properties);
/////////////
// PADDING //
/////////////
if (properties.formatWidth != -1) {
macros.padder = Padder::forProperties(properties);
}
///////////////////////////////
// DECIMAL MARK ALWAYS SHOWN //
///////////////////////////////
macros.decimal = properties.decimalSeparatorAlwaysShown ? UNUM_DECIMAL_SEPARATOR_ALWAYS
: UNUM_DECIMAL_SEPARATOR_AUTO;
///////////////////////
// SIGN ALWAYS SHOWN //
///////////////////////
macros.sign = properties.signAlwaysShown ? UNUM_SIGN_ALWAYS : UNUM_SIGN_AUTO;
/////////////////////////
// SCIENTIFIC NOTATION //
/////////////////////////
if (properties.minimumExponentDigits != -1) {
// Scientific notation is required.
// This whole section feels like a hack, but it is needed for regression tests.
// The mapping from property bag to scientific notation is nontrivial due to LDML rules.
if (maxInt > 8) {
// But #13110: The maximum of 8 digits has unknown origins and is not in the spec.
// If maxInt is greater than 8, it is set to minInt, even if minInt is greater than 8.
maxInt = minInt;
macros.integerWidth = IntegerWidth::zeroFillTo(minInt).truncateAt(maxInt);
} else if (maxInt > minInt && minInt > 1) {
// Bug #13289: if maxInt > minInt > 1, then minInt should be 1.
minInt = 1;
macros.integerWidth = IntegerWidth::zeroFillTo(minInt).truncateAt(maxInt);
}
int engineering = maxInt < 0 ? -1 : maxInt;
macros.notation = ScientificNotation(
// Engineering interval:
static_cast<int8_t>(engineering),
// Enforce minimum integer digits (for patterns like "000.00E0"):
(engineering == minInt),
// Minimum exponent digits:
static_cast<digits_t>(properties.minimumExponentDigits),
// Exponent sign always shown:
properties.exponentSignAlwaysShown ? UNUM_SIGN_ALWAYS : UNUM_SIGN_AUTO);
// Scientific notation also involves overriding the rounding mode.
// TODO: Overriding here is a bit of a hack. Should this logic go earlier?
if (macros.precision.fType == Precision::PrecisionType::RND_FRACTION) {
// For the purposes of rounding, get the original min/max int/frac, since the local
// variables have been manipulated for display purposes.
int maxInt_ = properties.maximumIntegerDigits;
int minInt_ = properties.minimumIntegerDigits;
int minFrac_ = properties.minimumFractionDigits;
int maxFrac_ = properties.maximumFractionDigits;
if (minInt_ == 0 && maxFrac_ == 0) {
// Patterns like "#E0" and "##E0", which mean no rounding!
macros.precision = Precision::unlimited().withMode(roundingMode);
} else if (minInt_ == 0 && minFrac_ == 0) {
// Patterns like "#.##E0" (no zeros in the mantissa), which mean round to maxFrac+1
macros.precision = Precision::constructSignificant(1, maxFrac_ + 1).withMode(roundingMode);
} else {
int maxSig_ = minInt_ + maxFrac_;
// Bug #20058: if maxInt_ > minInt_ > 1, then minInt_ should be 1.
if (maxInt_ > minInt_ && minInt_ > 1) {
minInt_ = 1;
}
int minSig_ = minInt_ + minFrac_;
// To avoid regression, maxSig is not reset when minInt_ set to 1.
// TODO: Reset maxSig_ = 1 + minFrac_ to follow the spec.
macros.precision = Precision::constructSignificant(minSig_, maxSig_).withMode(roundingMode);
}
}
}
//////////////////////
// COMPACT NOTATION //
//////////////////////
if (!properties.compactStyle.isNull()) {
if (properties.compactStyle.getNoError() == UNumberCompactStyle::UNUM_LONG) {
macros.notation = Notation::compactLong();
} else {
macros.notation = Notation::compactShort();
}
// Do not forward the affix provider.
macros.affixProvider = nullptr;
}
/////////////////
// MULTIPLIERS //
/////////////////
macros.scale = scaleFromProperties(properties);
//////////////////////
// PROPERTY EXPORTS //
//////////////////////
if (exportedProperties != nullptr) {
exportedProperties->currency = currency;
exportedProperties->roundingMode = roundingMode;
exportedProperties->minimumIntegerDigits = minInt;
exportedProperties->maximumIntegerDigits = maxInt == -1 ? INT32_MAX : maxInt;
Precision rounding_;
if (precision.fType == Precision::PrecisionType::RND_CURRENCY) {
rounding_ = precision.withCurrency(currency, status);
} else {
rounding_ = precision;
}
int minFrac_ = minFrac;
int maxFrac_ = maxFrac;
int minSig_ = minSig;
int maxSig_ = maxSig;
double increment_ = 0.0;
if (rounding_.fType == Precision::PrecisionType::RND_FRACTION) {
minFrac_ = rounding_.fUnion.fracSig.fMinFrac;
maxFrac_ = rounding_.fUnion.fracSig.fMaxFrac;
} else if (rounding_.fType == Precision::PrecisionType::RND_INCREMENT) {
increment_ = rounding_.fUnion.increment.fIncrement;
minFrac_ = rounding_.fUnion.increment.fMinFrac;
maxFrac_ = rounding_.fUnion.increment.fMinFrac;
} else if (rounding_.fType == Precision::PrecisionType::RND_SIGNIFICANT) {
minSig_ = rounding_.fUnion.fracSig.fMinSig;
maxSig_ = rounding_.fUnion.fracSig.fMaxSig;
}
exportedProperties->minimumFractionDigits = minFrac_;
exportedProperties->maximumFractionDigits = maxFrac_;
exportedProperties->minimumSignificantDigits = minSig_;
exportedProperties->maximumSignificantDigits = maxSig_;
exportedProperties->roundingIncrement = increment_;
}
return macros;
}
void PropertiesAffixPatternProvider::setTo(const DecimalFormatProperties& properties, UErrorCode&) {
fBogus = false;
// There are two ways to set affixes in DecimalFormat: via the pattern string (applyPattern), and via the
// explicit setters (setPositivePrefix and friends). The way to resolve the settings is as follows:
//
// 1) If the explicit setting is present for the field, use it.
// 2) Otherwise, follows UTS 35 rules based on the pattern string.
//
// Importantly, the explicit setters affect only the one field they override. If you set the positive
// prefix, that should not affect the negative prefix. Since it is impossible for the user of this class
// to know whether the origin for a string was the override or the pattern, we have to say that we always
// have a negative subpattern and perform all resolution logic here.
// Convenience: Extract the properties into local variables.
// Variables are named with three chars: [p/n][p/s][o/p]
// [p/n] => p for positive, n for negative
// [p/s] => p for prefix, s for suffix
// [o/p] => o for escaped custom override string, p for pattern string
UnicodeString ppo = AffixUtils::escape(properties.positivePrefix);
UnicodeString pso = AffixUtils::escape(properties.positiveSuffix);
UnicodeString npo = AffixUtils::escape(properties.negativePrefix);
UnicodeString nso = AffixUtils::escape(properties.negativeSuffix);
const UnicodeString& ppp = properties.positivePrefixPattern;
const UnicodeString& psp = properties.positiveSuffixPattern;
const UnicodeString& npp = properties.negativePrefixPattern;
const UnicodeString& nsp = properties.negativeSuffixPattern;
if (!properties.positivePrefix.isBogus()) {
posPrefix = ppo;
} else if (!ppp.isBogus()) {
posPrefix = ppp;
} else {
// UTS 35: Default positive prefix is empty string.
posPrefix = u"";
}
if (!properties.positiveSuffix.isBogus()) {
posSuffix = pso;
} else if (!psp.isBogus()) {
posSuffix = psp;
} else {
// UTS 35: Default positive suffix is empty string.
posSuffix = u"";
}
if (!properties.negativePrefix.isBogus()) {
negPrefix = npo;
} else if (!npp.isBogus()) {
negPrefix = npp;
} else {
// UTS 35: Default negative prefix is "-" with positive prefix.
// Important: We prepend the "-" to the pattern, not the override!
negPrefix = ppp.isBogus() ? u"-" : u"-" + ppp;
}
if (!properties.negativeSuffix.isBogus()) {
negSuffix = nso;
} else if (!nsp.isBogus()) {
negSuffix = nsp;
} else {
// UTS 35: Default negative prefix is the positive prefix.
negSuffix = psp.isBogus() ? u"" : psp;
}
}
char16_t PropertiesAffixPatternProvider::charAt(int flags, int i) const {
return getStringInternal(flags).charAt(i);
}
int PropertiesAffixPatternProvider::length(int flags) const {
return getStringInternal(flags).length();
}
UnicodeString PropertiesAffixPatternProvider::getString(int32_t flags) const {
return getStringInternal(flags);
}
const UnicodeString& PropertiesAffixPatternProvider::getStringInternal(int32_t flags) const {
bool prefix = (flags & AFFIX_PREFIX) != 0;
bool negative = (flags & AFFIX_NEGATIVE_SUBPATTERN) != 0;
if (prefix && negative) {
return negPrefix;
} else if (prefix) {
return posPrefix;
} else if (negative) {
return negSuffix;
} else {
return posSuffix;
}
}
bool PropertiesAffixPatternProvider::positiveHasPlusSign() const {
// TODO: Change the internal APIs to propagate out the error?
ErrorCode localStatus;
return AffixUtils::containsType(posPrefix, TYPE_PLUS_SIGN, localStatus) ||
AffixUtils::containsType(posSuffix, TYPE_PLUS_SIGN, localStatus);
}
bool PropertiesAffixPatternProvider::hasNegativeSubpattern() const {
// See comments in the constructor for more information on why this is always true.
return true;
}
bool PropertiesAffixPatternProvider::negativeHasMinusSign() const {
ErrorCode localStatus;
return AffixUtils::containsType(negPrefix, TYPE_MINUS_SIGN, localStatus) ||
AffixUtils::containsType(negSuffix, TYPE_MINUS_SIGN, localStatus);
}
bool PropertiesAffixPatternProvider::hasCurrencySign() const {
ErrorCode localStatus;
return AffixUtils::hasCurrencySymbols(posPrefix, localStatus) ||
AffixUtils::hasCurrencySymbols(posSuffix, localStatus) ||
AffixUtils::hasCurrencySymbols(negPrefix, localStatus) ||
AffixUtils::hasCurrencySymbols(negSuffix, localStatus);
}
bool PropertiesAffixPatternProvider::containsSymbolType(AffixPatternType type, UErrorCode& status) const {
return AffixUtils::containsType(posPrefix, type, status) ||
AffixUtils::containsType(posSuffix, type, status) ||
AffixUtils::containsType(negPrefix, type, status) ||
AffixUtils::containsType(negSuffix, type, status);
}
bool PropertiesAffixPatternProvider::hasBody() const {
return true;
}
void CurrencyPluralInfoAffixProvider::setTo(const CurrencyPluralInfo& cpi,
const DecimalFormatProperties& properties,
UErrorCode& status) {
// We need to use a PropertiesAffixPatternProvider, not the simpler version ParsedPatternInfo,
// because user-specified affix overrides still need to work.
fBogus = false;
DecimalFormatProperties pluralProperties(properties);
for (int32_t plural = 0; plural < StandardPlural::COUNT; plural++) {
const char* keyword = StandardPlural::getKeyword(static_cast<StandardPlural::Form>(plural));
UnicodeString patternString;
patternString = cpi.getCurrencyPluralPattern(keyword, patternString);
PatternParser::parseToExistingProperties(
patternString,
pluralProperties,
IGNORE_ROUNDING_NEVER,
status);
affixesByPlural[plural].setTo(pluralProperties, status);
}
}
char16_t CurrencyPluralInfoAffixProvider::charAt(int32_t flags, int32_t i) const {
int32_t pluralOrdinal = (flags & AFFIX_PLURAL_MASK);
return affixesByPlural[pluralOrdinal].charAt(flags, i);
}
int32_t CurrencyPluralInfoAffixProvider::length(int32_t flags) const {
int32_t pluralOrdinal = (flags & AFFIX_PLURAL_MASK);
return affixesByPlural[pluralOrdinal].length(flags);
}
UnicodeString CurrencyPluralInfoAffixProvider::getString(int32_t flags) const {
int32_t pluralOrdinal = (flags & AFFIX_PLURAL_MASK);
return affixesByPlural[pluralOrdinal].getString(flags);
}
bool CurrencyPluralInfoAffixProvider::positiveHasPlusSign() const {
return affixesByPlural[StandardPlural::OTHER].positiveHasPlusSign();
}
bool CurrencyPluralInfoAffixProvider::hasNegativeSubpattern() const {
return affixesByPlural[StandardPlural::OTHER].hasNegativeSubpattern();
}
bool CurrencyPluralInfoAffixProvider::negativeHasMinusSign() const {
return affixesByPlural[StandardPlural::OTHER].negativeHasMinusSign();
}
bool CurrencyPluralInfoAffixProvider::hasCurrencySign() const {
return affixesByPlural[StandardPlural::OTHER].hasCurrencySign();
}
bool CurrencyPluralInfoAffixProvider::containsSymbolType(AffixPatternType type, UErrorCode& status) const {
return affixesByPlural[StandardPlural::OTHER].containsSymbolType(type, status);
}
bool CurrencyPluralInfoAffixProvider::hasBody() const {
return affixesByPlural[StandardPlural::OTHER].hasBody();
}
#endif /* #if !UCONFIG_NO_FORMATTING */