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;

.gitignore

@@ -201,6 +201,40 @@ icu4c/source/samples/ustring/ustring.exe
 icu4c/source/stubdata/stubdatabuilt.txt
 icu4c/source/test/cintltst/cintltst
 icu4c/source/test/cintltst/cintltst.exe
+icu4c/source/test/fuzzer/break_iterator_fuzzer
+icu4c/source/test/fuzzer/calendar_fuzzer
+icu4c/source/test/fuzzer/collator_compare_fuzzer
+icu4c/source/test/fuzzer/collator_rulebased_fuzzer
+icu4c/source/test/fuzzer/converter_fuzzer
+icu4c/source/test/fuzzer/date_format_fuzzer
+icu4c/source/test/fuzzer/date_time_pattern_generator_fuzzer
+icu4c/source/test/fuzzer/decimal_format_symbols_fuzzer
+icu4c/source/test/fuzzer/dtfmtsym_fuzzer
+icu4c/source/test/fuzzer/dummytestcase
+icu4c/source/test/fuzzer/list_format_fuzzer
+icu4c/source/test/fuzzer/locale_fuzzer
+icu4c/source/test/fuzzer/locale_morph_fuzzer
+icu4c/source/test/fuzzer/message_formatter_fuzzer
+icu4c/source/test/fuzzer/normalizer2_fuzzer
+icu4c/source/test/fuzzer/number_format_fuzzer
+icu4c/source/test/fuzzer/number_formatter_fuzzer
+icu4c/source/test/fuzzer/plurrule_fuzzer
+icu4c/source/test/fuzzer/relative_date_time_formatter_fuzzer
+icu4c/source/test/fuzzer/rule_based_break_iterator_fuzzer
+icu4c/source/test/fuzzer/time_zone_names_fuzzer
+icu4c/source/test/fuzzer/timezone_create_fuzzer
+icu4c/source/test/fuzzer/ubidi_fuzzer
+icu4c/source/test/fuzzer/ucasemap_fuzzer
+icu4c/source/test/fuzzer/uloc_canonicalize_fuzzer
+icu4c/source/test/fuzzer/uloc_for_language_tag_fuzzer
+icu4c/source/test/fuzzer/uloc_get_name_fuzzer
+icu4c/source/test/fuzzer/uloc_is_right_to_left_fuzzer
+icu4c/source/test/fuzzer/uloc_open_keywords_fuzzer
+icu4c/source/test/fuzzer/unicode_string_codepage_create_fuzzer
+icu4c/source/test/fuzzer/unicodeset_fuzzer
+icu4c/source/test/fuzzer/uprop_fuzzer
+icu4c/source/test/fuzzer/uregex_match_fuzzer
+icu4c/source/test/fuzzer/uregex_open_fuzzer
 icu4c/source/test/intltest/intltest
 icu4c/source/test/intltest/intltest.exe
 icu4c/source/test/iotest/iotest

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[] = {
+static const UChar32 pairedChars[] = {
     0x0028, 0x0029, // ascii paired punctuation
     0x003c, 0x003e,
     0x005b, 0x005d,
@@ -43,11 +43,9 @@ 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;
+static constexpr int32_t pairedCharCount = UPRV_LENGTHOF(pairedChars);
 
-int8_t ScriptRun::highBit(int32_t value)
+static int8_t highBit(int32_t value)
 {
     if (value <= 0) {
         return -32;
@@ -83,10 +81,16 @@ int8_t ScriptRun::highBit(int32_t value)
     return bit;
 }
 
-int32_t ScriptRun::getPairIndex(UChar32 ch)
+static int32_t getPairedCharPower() {
+    static const int32_t pairedCharPower = 1 << highBit(pairedCharCount);
+    return pairedCharPower;
+}
+
+static int32_t getPairIndex(UChar32 ch)
 {
-    int32_t probe = pairedCharPower;
+    int32_t probe = getPairedCharPower();
     int32_t index = 0;
+    int32_t pairedCharExtra = pairedCharCount - probe;
 
     if (ch >= pairedChars[pairedCharExtra]) {
         index = pairedCharExtra;

icu4c/source/extra/scrptrun/scrptrun.h

@@ -86,14 +86,6 @@ 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 int32_t pairedCharCount;
-    static const int32_t pairedCharPower;
-    static const int32_t pairedCharExtra;
-
     /**
      * The address of this static class variable serves as this class's ID
      * for ICU "poor man's RTTI".

icu4c/source/i18n/dtptngen.cpp

@@ -369,6 +369,7 @@ DateTimePatternGenerator::DateTimePatternGenerator(const DateTimePatternGenerato
     UObject(),
     skipMatcher(nullptr),
     fAvailableFormatKeyHash(nullptr),
+    emptyString(u"", 0),
     fDefaultHourFormatChar(0),
     internalErrorCode(U_ZERO_ERROR)
 {
@@ -1406,7 +1407,6 @@ 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

@@ -1483,8 +1483,7 @@ 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 +1564,7 @@ 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.compare(UnicodeString(u"hebr", 4))==0 && 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 +3054,6 @@ 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 +3221,7 @@ 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.compare(UnicodeString(u"hebr", 4))==0 && 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 +3261,7 @@ 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.compare(UnicodeString(u"hebr", 4))==0 && 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() {
@@ -756,8 +757,7 @@ const UnicodeString& TransliteratorRegistry::getAvailableID(int32_t index) const
 
     // 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;
+    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
 };