ICU-22781 Support Arbitrary Constant Unit Formatting (C++)

See unicode-org#3381
younies · Feb 10, 2025 · d132924 · d132924
1 parent f495d10
commit d132924
Show file tree

Hide file tree

Showing 4 changed files with 172 additions and 75 deletions.
diff --git a/icu4c/source/i18n/measunit_extra.cpp b/icu4c/source/i18n/measunit_extra.cpp
@@ -1322,7 +1322,7 @@ void MeasureUnitImpl::serialize(UErrorCode &status) {
         return;
     }
 
-    if (this->singleUnits.length() == 0) {
+    if (this->singleUnits.length() == 0 && this->constantDenominator == 0) {
         // Dimensionless, constructed by the default constructor.
         return;
     }
@@ -1497,9 +1497,20 @@ MeasureUnit MeasureUnit::product(const MeasureUnit& other, UErrorCode& status) c
     for (int32_t i = 0; i < otherImpl.singleUnits.length(); i++) {
         impl.appendSingleUnit(*otherImpl.singleUnits[i], status);
     }
-    if (impl.singleUnits.length() > 1) {
+
+    if (this->getConstantDenominator(status) != 0 && other.getConstantDenominator(status) != 0) {
+        // Cannot multiply units that both of them have a constant denominator
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return {};
+    }
+
+    impl.constantDenominator =
+        this->getConstantDenominator(status) + other.getConstantDenominator(status);
+
+    if (impl.singleUnits.length() > 1 || impl.constantDenominator > 0) {
         impl.complexity = UMEASURE_UNIT_COMPOUND;
     }
+
     return std::move(impl).build(status);
 }
 

diff --git a/icu4c/source/i18n/number_longnames.cpp b/icu4c/source/i18n/number_longnames.cpp
@@ -48,8 +48,12 @@ constexpr int32_t PER_INDEX = StandardPlural::Form::COUNT + 1;
  * Gender of the word, in languages with grammatical gender.
  */
 constexpr int32_t GENDER_INDEX = StandardPlural::Form::COUNT + 2;
+/**
+ *  Denominator constant of the unit.
+ */
+constexpr int32_t CONSTANT_DENOMINATOR_INDEX = StandardPlural::Form::COUNT + 3;
 // Number of keys in the array populated by PluralTableSink.
-constexpr int32_t ARRAY_LENGTH = StandardPlural::Form::COUNT + 3;
+constexpr int32_t ARRAY_LENGTH = StandardPlural::Form::COUNT + 4;
 
 // TODO(icu-units#28): load this list from resources, after creating a "&set"
 // function for use in ldml2icu rules.
@@ -1010,6 +1014,11 @@ void LongNameHandler::forArbitraryUnit(const Locale &loc,
     //    denominator (the part after the "-per-). If both are empty, fail
     MeasureUnitImpl unit;
     MeasureUnitImpl perUnit;
+
+    if (unitRef.getConstantDenominator(status) != 0) {
+        perUnit.constantDenominator = unitRef.getConstantDenominator(status);
+    }
+
     {
         MeasureUnitImpl fullUnit = MeasureUnitImpl::forMeasureUnitMaybeCopy(unitRef, status);
         if (U_FAILURE(status)) {
@@ -1196,6 +1205,12 @@ void LongNameHandler::processPatternTimes(MeasureUnitImpl &&productUnit,
     DerivedComponents derivedTimesCases(loc, "case", "times");
     DerivedComponents derivedPowerCases(loc, "case", "power");
 
+    if (productUnit.constantDenominator != 0) {
+        CharString constantString;
+        constantString.appendNumber(productUnit.constantDenominator, status);
+        outArray[CONSTANT_DENOMINATOR_INDEX] = UnicodeString::fromUTF8(constantString.toStringPiece());
+    }
+
     // 4. For each single_unit in product_unit
     for (int32_t singleUnitIndex = 0; singleUnitIndex < productUnit.singleUnits.length();
          singleUnitIndex++) {
@@ -1454,6 +1469,39 @@ void LongNameHandler::processPatternTimes(MeasureUnitImpl &&productUnit,
             }
         }
     }
+
+    // 5. Handling constant denominator if it exists.
+    if (productUnit.constantDenominator != 0) {
+        int32_t pluralIndex = -1;
+        for (int32_t index = 0; index < StandardPlural::Form::COUNT; index++) {
+            if (!outArray[index].isBogus()) {
+                pluralIndex = index;
+                break;
+            }
+        }
+
+        U_ASSERT(pluralIndex >= 0); // "No plural form found for constant denominator"
+
+        // TODO(ICU-23039):
+        // Improve the handling of constant_denominator representation.
+        // For instance, a constant_denominator of 1000000 should be adaptable to
+        // formats like
+        // 1,000,000, 1e6, or 1 million.
+        // Furthermore, ensure consistent pluralization rules for units. For example,
+        // "meter per 100 seconds" should be evaluated for correct singular/plural
+        // usage: "second" or "seconds"?
+        // Similarly, "kilogram per 1000 meters" should be checked for "meter" or
+        // "meters"?
+        if (outArray[pluralIndex].length() == 0) {
+            outArray[pluralIndex] = outArray[CONSTANT_DENOMINATOR_INDEX];
+        } else {
+            UnicodeString tmp;
+            timesPatternFormatter.format(outArray[CONSTANT_DENOMINATOR_INDEX], outArray[pluralIndex],
+                                         tmp, status);
+            outArray[pluralIndex] = tmp;
+        }
+    }
+
     for (int32_t pluralIndex = 0; pluralIndex < StandardPlural::Form::COUNT; pluralIndex++) {
         if (globalPlaceholder[pluralIndex] == PH_BEGINNING) {
             UnicodeString tmp;

diff --git a/icu4c/source/test/intltest/numbertest.h b/icu4c/source/test/intltest/numbertest.h
@@ -103,6 +103,7 @@ class NumberFormatterApiTest : public IntlTestWithFieldPosition {
     void toDecimalNumber();
     void microPropsInternals();
     void formatUnitsAliases();
+    void formatArbitraryConstant();
     void testIssue22378();
 
     void runIndexedTest(int32_t index, UBool exec, const char*& name, char* par = nullptr) override;