ICU-23254 Remove C++ static initialization

Author: grhoten

.github/workflows/icu4c.yml

@@ -373,12 +373,12 @@ jobs:
 
   # MacOS with clang
   macos-clang:
-    runs-on: macos-14  # Updated in BRS
+    runs-on: macos-15  # Updated in BRS
     steps:
       - uses: actions/checkout@ff7abcd0c3c05ccf6adc123a8cd1fd4fb30fb493 # v4.1.1
       - name: ICU4C with clang on MacOS
         env:
-          CPPFLAGS: '-Wall -Wextra -Wextra-semi -Wundef -Wnon-virtual-dtor -Wctad-maybe-unsupported -Werror'
+          CPPFLAGS: '-Wall -Wextra -Wextra-semi -Wundef -Wnon-virtual-dtor -Wctad-maybe-unsupported -Wglobal-constructors -Wexit-time-destructors -Werror'
         run: |
           cd icu4c/source;
           PYTHON=python3 ./runConfigureICU macOS;

icu4c/source/common/loclikely.cpp

@@ -418,7 +418,7 @@ GetRegionFromKey(const char* localeID, std::string_view key, UErrorCode& status)
     if (U_SUCCESS(status) && len >= 3 && len <= 6 &&
         uprv_isASCIILetter(kw[0]) && uprv_isASCIILetter(kw[1])) {
         // Additional Check
-        static icu::RegionValidateMap valid;
+        icu::RegionValidateMap valid;
         const char region[] = {kw[0], kw[1], '\0'};
         if (valid.isSet(region)) {
             result.append(uprv_toupper(kw[0]), status);

icu4c/source/extra/scrptrun/scrptrun.cpp

@@ -23,7 +23,7 @@ U_NAMESPACE_BEGIN
 
 const char ScriptRun::fgClassID=0;
 
-UChar32 ScriptRun::pairedChars[] = {
+const UChar32 ScriptRun::pairedChars[] = {
     0x0028, 0x0029, // ascii paired punctuation
     0x003c, 0x003e,
     0x005b, 0x005d,
@@ -43,11 +43,7 @@ UChar32 ScriptRun::pairedChars[] = {
     0x301a, 0x301b
 };
 
-const int32_t ScriptRun::pairedCharCount = UPRV_LENGTHOF(pairedChars);
-const int32_t ScriptRun::pairedCharPower = 1 << highBit(pairedCharCount);
-const int32_t ScriptRun::pairedCharExtra = pairedCharCount - pairedCharPower;
-
-int8_t ScriptRun::highBit(int32_t value)
+static constexpr int8_t highBit(int32_t value)
 {
     if (value <= 0) {
         return -32;
@@ -83,6 +79,10 @@ int8_t ScriptRun::highBit(int32_t value)
     return bit;
 }
 
+const int32_t ScriptRun::pairedCharCount = UPRV_LENGTHOF(pairedChars);
+const int32_t ScriptRun::pairedCharPower = 1 << highBit(pairedCharCount);
+const int32_t ScriptRun::pairedCharExtra = pairedCharCount - pairedCharPower;
+
 int32_t ScriptRun::getPairIndex(UChar32 ch)
 {
     int32_t probe = pairedCharPower;

icu4c/source/extra/scrptrun/scrptrun.h

@@ -86,10 +86,9 @@ class ScriptRun : public UObject {
     ParenStackEntry parenStack[128];
     int32_t parenSP;
 
-    static int8_t highBit(int32_t value);
     static int32_t getPairIndex(UChar32 ch);
 
-    static UChar32 pairedChars[];
+    static const UChar32 pairedChars[];
     static const int32_t pairedCharCount;
     static const int32_t pairedCharPower;
     static const int32_t pairedCharExtra;

icu4c/source/i18n/dtptngen.cpp

@@ -333,11 +333,9 @@ DateTimePatternGenerator::createEmptyInstance(UErrorCode& status) {
 }
 
 DateTimePatternGenerator::DateTimePatternGenerator(UErrorCode &status) :
-    skipMatcher(nullptr),
-    fAvailableFormatKeyHash(nullptr),
-    fDefaultHourFormatChar(0),
-    internalErrorCode(U_ZERO_ERROR)
+    UObject()
 {
+    emptyString.getTerminatedBuffer();
     fp = new FormatParser();
     dtMatcher = new DateTimeMatcher();
     distanceInfo = new DistanceInfo();
@@ -348,30 +346,15 @@ DateTimePatternGenerator::DateTimePatternGenerator(UErrorCode &status) :
 }
 
 DateTimePatternGenerator::DateTimePatternGenerator(const Locale& locale, UErrorCode &status, UBool skipStdPatterns) :
-    skipMatcher(nullptr),
-    fAvailableFormatKeyHash(nullptr),
-    fDefaultHourFormatChar(0),
-    internalErrorCode(U_ZERO_ERROR)
+    DateTimePatternGenerator(status)
 {
-    fp = new FormatParser();
-    dtMatcher = new DateTimeMatcher();
-    distanceInfo = new DistanceInfo();
-    patternMap = new PatternMap();
-    if (fp == nullptr || dtMatcher == nullptr || distanceInfo == nullptr || patternMap == nullptr) {
-        internalErrorCode = status = U_MEMORY_ALLOCATION_ERROR;
-    }
-    else {
-        initData(locale, status, skipStdPatterns);
-    }
+    initData(locale, status, skipStdPatterns);
 }
 
 DateTimePatternGenerator::DateTimePatternGenerator(const DateTimePatternGenerator& other) :
-    UObject(),
-    skipMatcher(nullptr),
-    fAvailableFormatKeyHash(nullptr),
-    fDefaultHourFormatChar(0),
-    internalErrorCode(U_ZERO_ERROR)
+    UObject()
 {
+    emptyString.getTerminatedBuffer();
     fp = new FormatParser();
     dtMatcher = new DateTimeMatcher();
     distanceInfo = new DistanceInfo();
@@ -1406,7 +1389,6 @@ DateTimePatternGenerator::setDateTimeFormat(UDateFormatStyle style, const Unicod
 
 const UnicodeString&
 DateTimePatternGenerator::getDateTimeFormat(UDateFormatStyle style, UErrorCode& status) const {
-    static const UnicodeString emptyString = UNICODE_STRING_SIMPLE("");
     if (U_FAILURE(status)) {
         return emptyString;
     }

icu4c/source/i18n/smpdtfmt.cpp

@@ -236,6 +236,7 @@ static const int32_t gFieldRangeBias[] = {
 // offset the years within the current millennium down to 1-999
 static const int32_t HEBREW_CAL_CUR_MILLENIUM_START_YEAR = 5000;
 static const int32_t HEBREW_CAL_CUR_MILLENIUM_END_YEAR = 6000;
+static constexpr const char16_t HEBREW_CALENDAR_VALUE[] = u"hebr";
 
 /**
  * Maximum range for detecting daylight offset of a time zone when parsed time zone
@@ -1483,8 +1484,7 @@ SimpleDateFormat::subFormat(UnicodeString &appendTo,
                             Calendar& cal,
                             UErrorCode& status) const
 {
-    static const int32_t maxIntCount = 10;
-    static const UnicodeString hebr(u"hebr");
+    static constexpr int32_t maxIntCount = 10;
 
     if (U_FAILURE(status)) {
         return;
@@ -1565,7 +1565,7 @@ SimpleDateFormat::subFormat(UnicodeString &appendTo,
 //AD 12345 12345     45   12345    12345     12345
     case UDAT_YEAR_FIELD:
     case UDAT_YEAR_WOY_FIELD:
-        if (fDateOverride.compare(hebr)==0 && value>HEBREW_CAL_CUR_MILLENIUM_START_YEAR && value<HEBREW_CAL_CUR_MILLENIUM_END_YEAR) {
+        if (fDateOverride == HEBREW_CALENDAR_VALUE && value>HEBREW_CAL_CUR_MILLENIUM_START_YEAR && value<HEBREW_CAL_CUR_MILLENIUM_END_YEAR) {
             value-=HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
         }
         if(count == 2)
@@ -3055,7 +3055,6 @@ int32_t SimpleDateFormat::subParse(const UnicodeString& text, int32_t& start, ch
         return -start;
     }
     UCalendarDateFields field = fgPatternIndexToCalendarField[patternCharIndex]; // UCAL_FIELD_COUNT if irrelevant
-    UnicodeString hebr("hebr", 4, US_INV);
 
     if (numericLeapMonthFormatter != nullptr) {
         numericLeapMonthFormatter->setFormats(reinterpret_cast<const Format**>(&currentNumberFormat), 1);
@@ -3223,7 +3222,7 @@ int32_t SimpleDateFormat::subParse(const UnicodeString& text, int32_t& start, ch
         // we made adjustments to place the 2-digit year in the proper
         // century, for parsed strings from "00" to "99".  Any other string
         // is treated literally:  "2250", "-1", "1", "002".
-        if (fDateOverride.compare(hebr)==0 && value < 1000) {
+        if (fDateOverride == HEBREW_CALENDAR_VALUE && value < 1000) {
             value += HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
         } else if (text.moveIndex32(start, 2) == pos.getIndex() && !isChineseCalendar
             && u_isdigit(text.char32At(start))
@@ -3263,7 +3262,7 @@ int32_t SimpleDateFormat::subParse(const UnicodeString& text, int32_t& start, ch
 
     case UDAT_YEAR_WOY_FIELD:
         // Comment is the same as for UDAT_Year_FIELDs - look above
-        if (fDateOverride.compare(hebr)==0 && value < 1000) {
+        if (fDateOverride == HEBREW_CALENDAR_VALUE && value < 1000) {
             value += HEBREW_CAL_CUR_MILLENIUM_START_YEAR;
         } else if (text.moveIndex32(start, 2) == pos.getIndex()
             && u_isdigit(text.char32At(start))

icu4c/source/i18n/transreg.cpp

@@ -542,6 +542,7 @@ TransliteratorRegistry::TransliteratorRegistry(UErrorCode& status) :
         variantList.adoptElement(emptyString, status);
     }
     specDAG.setValueDeleter(uhash_deleteHashtable);
+    fBogus.setToBogus();
 }
 
 TransliteratorRegistry::~TransliteratorRegistry() {
@@ -754,10 +755,11 @@ const UnicodeString& TransliteratorRegistry::getAvailableID(int32_t index) const
         return *static_cast<UnicodeString*>(e->key.pointer);
     }
 
-    // If the code reaches here, the hash table was likely modified during iteration.
-    // Return an statically initialized empty string due to reference return type.
-    static UnicodeString empty;
-    return empty;
+    // If the code reaches here, the hash table was likely modified during iteration,
+    // or the requested index is out of bounds.
+    // Return a bogus string due to reference return type,
+    // and inability to return a UErrorCode from this API.
+    return fBogus;
 }
 
 StringEnumeration* TransliteratorRegistry::getAvailableIDs() const {

icu4c/source/i18n/transreg.h

@@ -455,6 +455,8 @@ class TransliteratorRegistry : public UMemory {
      */
     Hashtable availableIDs;
 
+    UnicodeString fBogus;
+
     TransliteratorRegistry(const TransliteratorRegistry &other); // forbid copying of this class
     TransliteratorRegistry &operator=(const TransliteratorRegistry &other); // forbid copying of this class
 };

icu4c/source/i18n/udatpg.cpp

@@ -248,7 +248,7 @@ udatpg_getDateTimeFormatForStyle(const UDateTimePatternGenerator *udtpg,
     }
     // Note: The UnicodeString for the dateTimeFormat string in the DateTimePatternGenerator
     // was NUL-terminated what it was set, to avoid doing it here which could re-allocate
-    // the buffe and affect and cont references to the string or its buffer.
+    // the buffer and affect const references to the string or its buffer.
     return result.getBuffer();
  }
 

icu4c/source/i18n/unicode/dtptngen.h

@@ -606,17 +606,17 @@ class U_I18N_API_CLASS DateTimePatternGenerator : public UObject {
     UnicodeString fieldDisplayNames[UDATPG_FIELD_COUNT][UDATPG_WIDTH_COUNT];
     UnicodeString dateTimeFormat[4];
     UnicodeString decimal;
-    DateTimeMatcher *skipMatcher;
-    Hashtable *fAvailableFormatKeyHash;
+    DateTimeMatcher *skipMatcher {  };
+    Hashtable *fAvailableFormatKeyHash {  };
     UnicodeString emptyString;
-    char16_t fDefaultHourFormatChar;
+    char16_t fDefaultHourFormatChar {  };
 
     int32_t fAllowedHourFormats[7];  // Actually an array of AllowedHourFormat enum type, ending with UNKNOWN.
 
     // Internal error code used for recording/reporting errors that occur during methods that do not
     // have a UErrorCode parameter. For example: the Copy Constructor, or the ::clone() method.
     // When this is set to an error the object is in an invalid state.
-    UErrorCode internalErrorCode;
+    UErrorCode internalErrorCode { U_ZERO_ERROR };
 
     /* internal flags masks for adjustFieldTypes etc. */
     enum {