number_longnames.cpp | Explore in Territory

// © 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

#include <cstdlib>

#include "unicode/simpleformatter.h"
#include "unicode/ures.h"
#include "ureslocs.h"
#include "charstr.h"
#include "uresimp.h"
#include "measunit_impl.h"
#include "number_longnames.h"
#include "number_microprops.h"
#include <algorithm>
#include "cstring.h"
#include "util.h"

usingnamespaceicu;
usingnamespaceicu::number;
usingnamespaceicu::number::impl;

namespace {

/**
 * Display Name (this format has no placeholder).
 *
 * Used as an index into the LongNameHandler::simpleFormats array. Units
 * resources cover the normal set of PluralRules keys, as well as `dnam` and
 * `per` forms.
 */
constexpr int32_t DNAM_INDEX = …;
/**
 * "per" form (e.g. "{0} per day" is day's "per" form).
 *
 * Used as an index into the LongNameHandler::simpleFormats array. Units
 * resources cover the normal set of PluralRules keys, as well as `dnam` and
 * `per` forms.
 */
constexpr int32_t PER_INDEX = …;
/**
 * Gender of the word, in languages with grammatical gender.
 */
constexpr int32_t GENDER_INDEX = …;
// Number of keys in the array populated by PluralTableSink.
constexpr int32_t ARRAY_LENGTH = …;

// TODO(icu-units#28): load this list from resources, after creating a "&set"
// function for use in ldml2icu rules.
const int32_t GENDER_COUNT = …;
const char *gGenders[GENDER_COUNT] = …;

// Converts a UnicodeString to a const char*, either pointing to a string in
// gGenders, or pointing to an empty string if an appropriate string was not
// found.
const char *getGenderString(UnicodeString uGender, UErrorCode status) { … }

// Returns the array index that corresponds to the given pluralKeyword.
static int32_t getIndex(const char* pluralKeyword, UErrorCode& status) { … }

// Selects a string out of the `strings` array which corresponds to the
// specified plural form, with fallback to the OTHER form.
//
// The `strings` array must have ARRAY_LENGTH items: one corresponding to each
// of the plural forms, plus a display name ("dnam") and a "per" form.
static UnicodeString getWithPlural(
        const UnicodeString* strings,
        StandardPlural::Form plural,
        UErrorCode& status) { … }

enum PlaceholderPosition { … };

/**
 * Returns three outputs extracted from pattern.
 *
 * @param coreUnit is extracted as per Extract(...) in the spec:
 *   https://unicode.org/reports/tr35/tr35-general.html#compound-units
 * @param PlaceholderPosition indicates where in the string the placeholder was
 *   found.
 * @param joinerChar Iff the placeholder was at the beginning or end, joinerChar
 *   contains the space character (if any) that separated the placeholder from
 *   the rest of the pattern. Otherwise, joinerChar is set to NUL. Only one
 *   space character is considered.
 */
void extractCorePattern(const UnicodeString &pattern,
                        UnicodeString &coreUnit,
                        PlaceholderPosition &placeholderPosition,
                        char16_t &joinerChar) { … }

//////////////////////////
/// BEGIN DATA LOADING ///
//////////////////////////

// Gets the gender of a built-in unit: unit must be a built-in. Returns an empty
// string both in case of unknown gender and in case of unknown unit.
UnicodeString
getGenderForBuiltin(const Locale &locale, const MeasureUnit &builtinUnit, UErrorCode &status) { … }

// Loads data from a resource tree with paths matching
// $key/$pluralForm/$gender/$case, with lateral inheritance for missing cases
// and genders.
//
// An InflectedPluralSink is configured to load data for a specific gender and
// case. It loads all plural forms, because selection between plural forms is
// dependent upon the value being formatted.
//
// See data/unit/de.txt and data/unit/fr.txt for examples - take a look at
// units/compound/power2: German has case, French has differences for gender,
// but no case.
//
// TODO(icu-units#138): Conceptually similar to PluralTableSink, however the
// tree structures are different. After homogenizing the structures, we may be
// able to unify the two classes.
//
// TODO: Spec violation: expects presence of "count" - does not fallback to an
// absent "count"! If this fallback were added, getCompoundValue could be
// superseded?
class InflectedPluralSink : public ResourceSink { … };

// Fetches localised formatting patterns for the given subKey. See documentation
// for InflectedPluralSink for details.
//
// Data is loaded for the appropriate unit width, with missing data filled in
// from unitsShort.
void getInflectedMeasureData(StringPiece subKey,
                             const Locale &locale,
                             const UNumberUnitWidth &width,
                             const char *gender,
                             const char *caseVariant,
                             UnicodeString *outArray,
                             UErrorCode &status) { … }

class PluralTableSink : public ResourceSink { … };

/**
 * Populates outArray with `locale`-specific values for `unit` through use of
 * PluralTableSink. Only the set of basic units are supported!
 *
 * Reading from resources *unitsNarrow* and *unitsShort* (for width
 * UNUM_UNIT_WIDTH_NARROW), or just *unitsShort* (for width
 * UNUM_UNIT_WIDTH_SHORT). For other widths, it reads just "units".
 *
 * @param unit must be a built-in unit, i.e. must have a type and subtype,
 *     listed in gTypes and gSubTypes in measunit.cpp.
 * @param unitDisplayCase the empty string and "nominative" are treated the
 *     same. For other cases, strings for the requested case are used if found.
 *     (For any missing case-specific data, we fall back to nominative.)
 * @param outArray must be of fixed length ARRAY_LENGTH.
 */
void getMeasureData(const Locale &locale,
                    const MeasureUnit &unit,
                    const UNumberUnitWidth &width,
                    const char *unitDisplayCase,
                    UnicodeString *outArray,
                    UErrorCode &status) { … }

// NOTE: outArray MUST have a length of at least ARRAY_LENGTH.
void getCurrencyLongNameData(const Locale &locale, const CurrencyUnit &currency, UnicodeString *outArray,
                             UErrorCode &status) { … }

UnicodeString getCompoundValue(StringPiece compoundKey,
                               const Locale &locale,
                               const UNumberUnitWidth &width,
                               UErrorCode &status) { … }

/**
 * Loads and applies deriveComponent rules from CLDR's grammaticalFeatures.xml.
 *
 * Consider a deriveComponent rule that looks like this:
 *
 *     <deriveComponent feature="case" structure="per" value0="compound" value1="nominative"/>
 *
 * Instantiating an instance as follows:
 *
 *     DerivedComponents d(loc, "case", "per");
 *
 * Applying the rule in the XML element above, `d.value0("foo")` will be "foo",
 * and `d.value1("foo")` will be "nominative".
 *
 * The values returned by value0(...) and value1(...) are valid only while the
 * instance exists. In case of any kind of failure, value0(...) and value1(...)
 * will return "".
 */
class DerivedComponents { … };

// TODO(icu-units#28): test somehow? Associate with an ICU ticket for adding
// testsuite support for testing with synthetic data?
/**
 * Loads and returns the value in rules that look like these:
 *
 * <deriveCompound feature="gender" structure="per" value="0"/>
 * <deriveCompound feature="gender" structure="times" value="1"/>
 *
 * Currently a fake example, but spec compliant:
 * <deriveCompound feature="gender" structure="power" value="feminine"/>
 *
 * NOTE: If U_FAILURE(status), returns an empty string.
 */ 
UnicodeString
getDeriveCompoundRule(Locale locale, const char *feature, const char *structure, UErrorCode &status) { … }

// Returns the gender string for structures following these rules:
//
// <deriveCompound feature="gender" structure="per" value="0"/>
// <deriveCompound feature="gender" structure="times" value="1"/>
//
// Fake example:
// <deriveCompound feature="gender" structure="power" value="feminine"/>
//
// data0 and data1 should be pattern arrays (UnicodeString[ARRAY_SIZE]) that
// correspond to value="0" and value="1".
//
// Pass a nullptr to data1 if the structure has no concept of value="1" (e.g.
// "prefix" doesn't).
UnicodeString getDerivedGender(Locale locale,
                               const char *structure,
                               UnicodeString *data0,
                               UnicodeString *data1,
                               UErrorCode &status) { … }

////////////////////////
/// END DATA LOADING ///
////////////////////////

// TODO: promote this somewhere? It's based on patternprops.cpp' trimWhitespace
const char16_t *trimSpaceChars(const char16_t *s, int32_t &length) { … }

/**
 * Calculates the gender of an arbitrary unit: this is the *second*
 * implementation of an algorithm to do this:
 *
 * Gender is also calculated in "processPatternTimes": that code path is "bottom
 * up", loading the gender for every component of a compound unit (at the same
 * time as loading the Long Names formatting patterns), even if the gender is
 * unneeded, then combining the single units' genders into the compound unit's
 * gender, according to the rules. This algorithm does a lazier "top-down"
 * evaluation, starting with the compound unit, calculating which single unit's
 * gender is needed by breaking it down according to the rules, and then loading
 * only the gender of the one single unit who's gender is needed.
 *
 * For future refactorings:
 * 1. we could drop processPatternTimes' gender calculation and just call this
 *    function: for UNUM_UNIT_WIDTH_FULL_NAME, the unit gender is in the very
 *    same table as the formatting patterns, so loading it then may be
 *    efficient. For other unit widths however, it needs to be explicitly looked
 *    up anyway.
 * 2. alternatively, if CLDR is providing all the genders we need such that we
 *    don't need to calculate them in ICU anymore, we could drop this function
 *    and keep only processPatternTimes' calculation. (And optimise it a bit?)
 *
 * @param locale The desired locale.
 * @param unit The measure unit to calculate the gender for.
 * @return The gender string for the unit, or an empty string if unknown or
 *     ungendered.
 */
UnicodeString calculateGenderForUnit(const Locale &locale, const MeasureUnit &unit, UErrorCode &status) { … }

void maybeCalculateGender(const Locale &locale,
                          const MeasureUnit &unitRef,
                          UnicodeString *outArray,
                          UErrorCode &status) { … }

} // namespace

void LongNameHandler::forMeasureUnit(const Locale &loc,
                                     const MeasureUnit &unitRef,
                                     const UNumberUnitWidth &width,
                                     const char *unitDisplayCase,
                                     const PluralRules *rules,
                                     const MicroPropsGenerator *parent,
                                     LongNameHandler *fillIn,
                                     UErrorCode &status) { … }

void LongNameHandler::forArbitraryUnit(const Locale &loc,
                                       const MeasureUnit &unitRef,
                                       const UNumberUnitWidth &width,
                                       const char *unitDisplayCase,
                                       LongNameHandler *fillIn,
                                       UErrorCode &status) { … }

void LongNameHandler::processPatternTimes(MeasureUnitImpl &&productUnit,
                                          Locale loc,
                                          const UNumberUnitWidth &width,
                                          const char *caseVariant,
                                          UnicodeString *outArray,
                                          UErrorCode &status) { … }

UnicodeString LongNameHandler::getUnitDisplayName(
        const Locale& loc,
        const MeasureUnit& unit,
        UNumberUnitWidth width,
        UErrorCode& status) { … }

UnicodeString LongNameHandler::getUnitPattern(
        const Locale& loc,
        const MeasureUnit& unit,
        UNumberUnitWidth width,
        StandardPlural::Form pluralForm,
        UErrorCode& status) { … }

LongNameHandler* LongNameHandler::forCurrencyLongNames(const Locale &loc, const CurrencyUnit &currency,
                                                      const PluralRules *rules,
                                                      const MicroPropsGenerator *parent,
                                                      UErrorCode &status) { … }

void LongNameHandler::simpleFormatsToModifiers(const UnicodeString *simpleFormats, Field field,
                                               UErrorCode &status) { … }

void LongNameHandler::multiSimpleFormatsToModifiers(const UnicodeString *leadFormats, UnicodeString trailFormat,
                                                    Field field, UErrorCode &status) { … }

void LongNameHandler::processQuantity(DecimalQuantity &quantity, MicroProps &micros,
                                      UErrorCode &status) const { … }

const Modifier* LongNameHandler::getModifier(Signum /*signum*/, StandardPlural::Form plural) const { … }

void MixedUnitLongNameHandler::forMeasureUnit(const Locale &loc,
                                              const MeasureUnit &mixedUnit,
                                              const UNumberUnitWidth &width,
                                              const char *unitDisplayCase,
                                              const PluralRules *rules,
                                              const MicroPropsGenerator *parent,
                                              MixedUnitLongNameHandler *fillIn,
                                              UErrorCode &status) { … }

void MixedUnitLongNameHandler::processQuantity(DecimalQuantity &quantity, MicroProps &micros,
                                               UErrorCode &status) const { … }

const Modifier *MixedUnitLongNameHandler::getMixedUnitModifier(DecimalQuantity &quantity,
                                                               MicroProps &micros,
                                                               UErrorCode &status) const { … }

const Modifier *MixedUnitLongNameHandler::getModifier(Signum /*signum*/,
                                                      StandardPlural::Form /*plural*/) const { … }

LongNameMultiplexer *LongNameMultiplexer::forMeasureUnits(const Locale &loc,
                                                          const MaybeStackVector<MeasureUnit> &units,
                                                          const UNumberUnitWidth &width,
                                                          const char *unitDisplayCase,
                                                          const PluralRules *rules,
                                                          const MicroPropsGenerator *parent,
                                                          UErrorCode &status) { … }

void LongNameMultiplexer::processQuantity(DecimalQuantity &quantity, MicroProps &micros,
                                          UErrorCode &status) const { … }

#endif /* #if !UCONFIG_NO_FORMATTING */
chromium/third_party/icu/source/i18n/number_longnames.cpp
