324 lines
13 KiB
C++
324 lines
13 KiB
C++
// © 2017 and later: Unicode, Inc. and others.
|
|
// License & terms of use: http://www.unicode.org/copyright.html
|
|
|
|
#include "unicode/utypes.h"
|
|
|
|
#if !UCONFIG_NO_FORMATTING && !UPRV_INCOMPLETE_CPP11_SUPPORT
|
|
|
|
#include "resource.h"
|
|
#include "number_compact.h"
|
|
#include "unicode/ustring.h"
|
|
#include "unicode/ures.h"
|
|
#include "cstring.h"
|
|
#include "charstr.h"
|
|
#include "uresimp.h"
|
|
|
|
using namespace icu::number::impl;
|
|
using namespace icu;
|
|
|
|
namespace {
|
|
|
|
// A dummy object used when a "0" compact decimal entry is encountered. This is necessary
|
|
// in order to prevent falling back to root. Object equality ("==") is intended.
|
|
const UChar *USE_FALLBACK = u"<USE FALLBACK>";
|
|
|
|
/** Produces a string like "NumberElements/latn/patternsShort/decimalFormat". */
|
|
void getResourceBundleKey(const char *nsName, CompactStyle compactStyle, CompactType compactType,
|
|
CharString &sb, UErrorCode &status) {
|
|
sb.clear();
|
|
sb.append("NumberElements/", status);
|
|
sb.append(nsName, status);
|
|
sb.append(compactStyle == CompactStyle::UNUM_SHORT ? "/patternsShort" : "/patternsLong", status);
|
|
sb.append(compactType == CompactType::TYPE_DECIMAL ? "/decimalFormat" : "/currencyFormat", status);
|
|
}
|
|
|
|
int32_t getIndex(int32_t magnitude, StandardPlural::Form plural) {
|
|
return magnitude * StandardPlural::COUNT + plural;
|
|
}
|
|
|
|
int32_t countZeros(const UChar *patternString, int32_t patternLength) {
|
|
// NOTE: This strategy for computing the number of zeros is a hack for efficiency.
|
|
// It could break if there are any 0s that aren't part of the main pattern.
|
|
int32_t numZeros = 0;
|
|
for (int32_t i = 0; i < patternLength; i++) {
|
|
if (patternString[i] == u'0') {
|
|
numZeros++;
|
|
} else if (numZeros > 0) {
|
|
break; // zeros should always be contiguous
|
|
}
|
|
}
|
|
return numZeros;
|
|
}
|
|
|
|
} // namespace
|
|
|
|
// NOTE: patterns and multipliers both get zero-initialized.
|
|
CompactData::CompactData() : patterns(), multipliers(), largestMagnitude(0), isEmpty(TRUE) {
|
|
}
|
|
|
|
void CompactData::populate(const Locale &locale, const char *nsName, CompactStyle compactStyle,
|
|
CompactType compactType, UErrorCode &status) {
|
|
CompactDataSink sink(*this);
|
|
LocalUResourceBundlePointer rb(ures_open(nullptr, locale.getName(), &status));
|
|
|
|
bool nsIsLatn = strcmp(nsName, "latn") == 0;
|
|
bool compactIsShort = compactStyle == CompactStyle::UNUM_SHORT;
|
|
|
|
// Fall back to latn numbering system and/or short compact style.
|
|
CharString resourceKey;
|
|
getResourceBundleKey(nsName, compactStyle, compactType, resourceKey, status);
|
|
UErrorCode localStatus = U_ZERO_ERROR;
|
|
ures_getAllItemsWithFallback(rb.getAlias(), resourceKey.data(), sink, localStatus);
|
|
if (isEmpty && !nsIsLatn) {
|
|
getResourceBundleKey("latn", compactStyle, compactType, resourceKey, status);
|
|
localStatus = U_ZERO_ERROR;
|
|
ures_getAllItemsWithFallback(rb.getAlias(), resourceKey.data(), sink, localStatus);
|
|
}
|
|
if (isEmpty && !compactIsShort) {
|
|
getResourceBundleKey(nsName, CompactStyle::UNUM_SHORT, compactType, resourceKey, status);
|
|
localStatus = U_ZERO_ERROR;
|
|
ures_getAllItemsWithFallback(rb.getAlias(), resourceKey.data(), sink, localStatus);
|
|
}
|
|
if (isEmpty && !nsIsLatn && !compactIsShort) {
|
|
getResourceBundleKey("latn", CompactStyle::UNUM_SHORT, compactType, resourceKey, status);
|
|
localStatus = U_ZERO_ERROR;
|
|
ures_getAllItemsWithFallback(rb.getAlias(), resourceKey.data(), sink, localStatus);
|
|
}
|
|
|
|
// The last fallback should be guaranteed to return data.
|
|
if (isEmpty) {
|
|
status = U_INTERNAL_PROGRAM_ERROR;
|
|
}
|
|
}
|
|
|
|
int32_t CompactData::getMultiplier(int32_t magnitude) const {
|
|
if (magnitude < 0) {
|
|
return 0;
|
|
}
|
|
if (magnitude > largestMagnitude) {
|
|
magnitude = largestMagnitude;
|
|
}
|
|
return multipliers[magnitude];
|
|
}
|
|
|
|
const UChar *CompactData::getPattern(int32_t magnitude, StandardPlural::Form plural) const {
|
|
if (magnitude < 0) {
|
|
return nullptr;
|
|
}
|
|
if (magnitude > largestMagnitude) {
|
|
magnitude = largestMagnitude;
|
|
}
|
|
const UChar *patternString = patterns[getIndex(magnitude, plural)];
|
|
if (patternString == nullptr && plural != StandardPlural::OTHER) {
|
|
// Fall back to "other" plural variant
|
|
patternString = patterns[getIndex(magnitude, StandardPlural::OTHER)];
|
|
}
|
|
if (patternString == USE_FALLBACK) { // == is intended
|
|
// Return null if USE_FALLBACK is present
|
|
patternString = nullptr;
|
|
}
|
|
return patternString;
|
|
}
|
|
|
|
void CompactData::getUniquePatterns(UVector &output, UErrorCode &status) const {
|
|
U_ASSERT(output.isEmpty());
|
|
// NOTE: In C++, this is done more manually with a UVector.
|
|
// In Java, we can take advantage of JDK HashSet.
|
|
for (auto pattern : patterns) {
|
|
if (pattern == nullptr || pattern == USE_FALLBACK) {
|
|
continue;
|
|
}
|
|
|
|
// Insert pattern into the UVector if the UVector does not already contain the pattern.
|
|
// Search the UVector from the end since identical patterns are likely to be adjacent.
|
|
for (int32_t i = output.size() - 1; i >= 0; i--) {
|
|
if (u_strcmp(pattern, static_cast<const UChar *>(output[i])) == 0) {
|
|
goto continue_outer;
|
|
}
|
|
}
|
|
|
|
// The string was not found; add it to the UVector.
|
|
// ANDY: This requires a const_cast. Why?
|
|
output.addElement(const_cast<UChar *>(pattern), status);
|
|
|
|
continue_outer:
|
|
continue;
|
|
}
|
|
}
|
|
|
|
void CompactData::CompactDataSink::put(const char *key, ResourceValue &value, UBool /*noFallback*/,
|
|
UErrorCode &status) {
|
|
// traverse into the table of powers of ten
|
|
ResourceTable powersOfTenTable = value.getTable(status);
|
|
if (U_FAILURE(status)) { return; }
|
|
for (int i3 = 0; powersOfTenTable.getKeyAndValue(i3, key, value); ++i3) {
|
|
|
|
// Assumes that the keys are always of the form "10000" where the magnitude is the
|
|
// length of the key minus one. We expect magnitudes to be less than MAX_DIGITS.
|
|
auto magnitude = static_cast<int8_t> (strlen(key) - 1);
|
|
int8_t multiplier = data.multipliers[magnitude];
|
|
U_ASSERT(magnitude < COMPACT_MAX_DIGITS);
|
|
|
|
// Iterate over the plural variants ("one", "other", etc)
|
|
ResourceTable pluralVariantsTable = value.getTable(status);
|
|
if (U_FAILURE(status)) { return; }
|
|
for (int i4 = 0; pluralVariantsTable.getKeyAndValue(i4, key, value); ++i4) {
|
|
|
|
// Skip this magnitude/plural if we already have it from a child locale.
|
|
// Note: This also skips USE_FALLBACK entries.
|
|
StandardPlural::Form plural = StandardPlural::fromString(key, status);
|
|
if (U_FAILURE(status)) { return; }
|
|
if (data.patterns[getIndex(magnitude, plural)] != nullptr) {
|
|
continue;
|
|
}
|
|
|
|
// The value "0" means that we need to use the default pattern and not fall back
|
|
// to parent locales. Example locale where this is relevant: 'it'.
|
|
int32_t patternLength;
|
|
const UChar *patternString = value.getString(patternLength, status);
|
|
if (U_FAILURE(status)) { return; }
|
|
if (u_strcmp(patternString, u"0") == 0) {
|
|
patternString = USE_FALLBACK;
|
|
patternLength = 0;
|
|
}
|
|
|
|
// Save the pattern string. We will parse it lazily.
|
|
data.patterns[getIndex(magnitude, plural)] = patternString;
|
|
|
|
// If necessary, compute the multiplier: the difference between the magnitude
|
|
// and the number of zeros in the pattern.
|
|
if (multiplier == 0) {
|
|
int32_t numZeros = countZeros(patternString, patternLength);
|
|
if (numZeros > 0) { // numZeros==0 in certain cases, like Somali "Kun"
|
|
multiplier = static_cast<int8_t> (numZeros - magnitude - 1);
|
|
}
|
|
}
|
|
}
|
|
|
|
// Save the multiplier.
|
|
if (data.multipliers[magnitude] == 0) {
|
|
data.multipliers[magnitude] = multiplier;
|
|
if (magnitude > data.largestMagnitude) {
|
|
data.largestMagnitude = magnitude;
|
|
}
|
|
data.isEmpty = false;
|
|
} else {
|
|
U_ASSERT(data.multipliers[magnitude] == multiplier);
|
|
}
|
|
}
|
|
}
|
|
|
|
///////////////////////////////////////////////////////////
|
|
/// END OF CompactData.java; BEGIN CompactNotation.java ///
|
|
///////////////////////////////////////////////////////////
|
|
|
|
CompactHandler::CompactHandler(CompactStyle compactStyle, const Locale &locale, const char *nsName,
|
|
CompactType compactType, const PluralRules *rules,
|
|
MutablePatternModifier *buildReference, const MicroPropsGenerator *parent,
|
|
UErrorCode &status)
|
|
: rules(rules), parent(parent) {
|
|
data.populate(locale, nsName, compactStyle, compactType, status);
|
|
if (buildReference != nullptr) {
|
|
// Safe code path
|
|
precomputeAllModifiers(*buildReference, status);
|
|
safe = TRUE;
|
|
} else {
|
|
// Unsafe code path
|
|
safe = FALSE;
|
|
}
|
|
}
|
|
|
|
CompactHandler::~CompactHandler() {
|
|
for (int32_t i = 0; i < precomputedModsLength; i++) {
|
|
delete precomputedMods[i].mod;
|
|
}
|
|
}
|
|
|
|
void CompactHandler::precomputeAllModifiers(MutablePatternModifier &buildReference, UErrorCode &status) {
|
|
if (U_FAILURE(status)) { return; }
|
|
|
|
// Initial capacity of 12 for 0K, 00K, 000K, ...M, ...B, and ...T
|
|
UVector allPatterns(12, status);
|
|
if (U_FAILURE(status)) { return; }
|
|
data.getUniquePatterns(allPatterns, status);
|
|
if (U_FAILURE(status)) { return; }
|
|
|
|
// C++ only: ensure that precomputedMods has room.
|
|
precomputedModsLength = allPatterns.size();
|
|
if (precomputedMods.getCapacity() < precomputedModsLength) {
|
|
precomputedMods.resize(allPatterns.size(), status);
|
|
if (U_FAILURE(status)) { return; }
|
|
}
|
|
|
|
for (int32_t i = 0; i < precomputedModsLength; i++) {
|
|
auto patternString = static_cast<const UChar *>(allPatterns[i]);
|
|
UnicodeString hello(patternString);
|
|
CompactModInfo &info = precomputedMods[i];
|
|
ParsedPatternInfo patternInfo;
|
|
PatternParser::parseToPatternInfo(UnicodeString(patternString), patternInfo, status);
|
|
if (U_FAILURE(status)) { return; }
|
|
buildReference.setPatternInfo(&patternInfo);
|
|
info.mod = buildReference.createImmutable(status);
|
|
if (U_FAILURE(status)) { return; }
|
|
info.numDigits = patternInfo.positive.integerTotal;
|
|
info.patternString = patternString;
|
|
}
|
|
}
|
|
|
|
void CompactHandler::processQuantity(DecimalQuantity &quantity, MicroProps µs,
|
|
UErrorCode &status) const {
|
|
parent->processQuantity(quantity, micros, status);
|
|
if (U_FAILURE(status)) { return; }
|
|
|
|
// Treat zero as if it had magnitude 0
|
|
int magnitude;
|
|
if (quantity.isZero()) {
|
|
magnitude = 0;
|
|
micros.rounding.apply(quantity, status);
|
|
} else {
|
|
// TODO: Revisit chooseMultiplierAndApply
|
|
int multiplier = micros.rounding.chooseMultiplierAndApply(quantity, data, status);
|
|
magnitude = quantity.isZero() ? 0 : quantity.getMagnitude();
|
|
magnitude -= multiplier;
|
|
}
|
|
|
|
StandardPlural::Form plural = quantity.getStandardPlural(rules);
|
|
const UChar *patternString = data.getPattern(magnitude, plural);
|
|
int numDigits = -1;
|
|
if (patternString == nullptr) {
|
|
// Use the default (non-compact) modifier.
|
|
// No need to take any action.
|
|
} else if (safe) {
|
|
// Safe code path.
|
|
// Java uses a hash set here for O(1) lookup. C++ uses a linear search.
|
|
// TODO: Benchmark this and maybe change to a binary search or hash table.
|
|
int32_t i = 0;
|
|
for (; i < precomputedModsLength; i++) {
|
|
const CompactModInfo &info = precomputedMods[i];
|
|
if (u_strcmp(patternString, info.patternString) == 0) {
|
|
info.mod->applyToMicros(micros, quantity);
|
|
numDigits = info.numDigits;
|
|
break;
|
|
}
|
|
}
|
|
// It should be guaranteed that we found the entry.
|
|
U_ASSERT(i < precomputedModsLength);
|
|
} else {
|
|
// Unsafe code path.
|
|
// Overwrite the PatternInfo in the existing modMiddle.
|
|
// C++ Note: Use unsafePatternInfo for proper lifecycle.
|
|
ParsedPatternInfo &patternInfo = const_cast<CompactHandler *>(this)->unsafePatternInfo;
|
|
PatternParser::parseToPatternInfo(UnicodeString(patternString), patternInfo, status);
|
|
static_cast<MutablePatternModifier*>(const_cast<Modifier*>(micros.modMiddle))
|
|
->setPatternInfo(&patternInfo);
|
|
numDigits = patternInfo.positive.integerTotal;
|
|
}
|
|
|
|
// FIXME: Deal with numDigits == 0 (Awaiting a test case)
|
|
|
|
// We already performed rounding. Do not perform it again.
|
|
micros.rounding = Rounder::constructPassThrough();
|
|
}
|
|
|
|
#endif /* #if !UCONFIG_NO_FORMATTING */
|