deps: update icu to 77.1

PR-URL: https://github.com/nodejs/node/pull/57455
Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com>
Reviewed-By: Michaël Zasso <targos@protonmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Rafael Gonzaga <rafael.nunu@hotmail.com>
Reviewed-By: LiviaMedeiros <livia@cirno.name>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Steven R Loomis <srl295@gmail.com>
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
Reviewed-By: Richard Lau <rlau@redhat.com>

deps/icu-small/LICENSE

@@ -2,7 +2,7 @@ UNICODE LICENSE V3
 
 COPYRIGHT AND PERMISSION NOTICE
 
-Copyright © 2016-2024 Unicode, Inc.
+Copyright © 2016-2025 Unicode, Inc.
 
 NOTICE TO USER: Carefully read the following legal agreement. BY
 DOWNLOADING, INSTALLING, COPYING OR OTHERWISE USING DATA FILES, AND/OR

deps/icu-small/README-FULL-ICU.txt

@@ -1,8 +1,8 @@
 ICU sources - auto generated by shrink-icu-src.py
 
 This directory contains the ICU subset used by --with-intl=full-icu
-It is a strict subset of ICU 76 source files with the following exception(s):
+It is a strict subset of ICU 77 source files with the following exception(s):
-* deps/icu-small/source/data/in/icudt76l.dat.bz2 : compressed data file
+* deps/icu-small/source/data/in/icudt77l.dat.bz2 : compressed data file
 
 
 To rebuild this directory, see ../../tools/icu/README.md

deps/icu-small/source/common/brkiter.cpp

@@ -59,7 +59,7 @@ BreakIterator::buildInstance(const Locale& loc, const char *type, UErrorCode &st
 {
     char fnbuff[256];
     char ext[4]={'\0'};
-    CharString actualLocale;
+    CharString actual;
     int32_t size;
     const char16_t* brkfname = nullptr;
     UResourceBundle brkRulesStack;
@@ -94,7 +94,7 @@ BreakIterator::buildInstance(const Locale& loc, const char *type, UErrorCode &st
 
         // Use the string if we found it
         if (U_SUCCESS(status) && brkfname) {
-            actualLocale.append(ures_getLocaleInternal(brkName, &status), -1, status);
+            actual.append(ures_getLocaleInternal(brkName, &status), -1, status);
 
             char16_t* extStart=u_strchr(brkfname, 0x002e);
             int len = 0;
@@ -124,9 +124,8 @@ BreakIterator::buildInstance(const Locale& loc, const char *type, UErrorCode &st
         U_LOCALE_BASED(locBased, *(BreakIterator*)result);
 
         locBased.setLocaleIDs(ures_getLocaleByType(b, ULOC_VALID_LOCALE, &status),
-                              actualLocale.data());
+                              actual.data(), status);
-        uprv_strncpy(result->requestLocale, loc.getName(), ULOC_FULLNAME_CAPACITY);
+        LocaleBased::setLocaleID(loc.getName(), result->requestLocale, status);
-        result->requestLocale[ULOC_FULLNAME_CAPACITY-1] = 0; // always terminate
     }
 
     ures_close(b);
@@ -206,26 +205,32 @@ BreakIterator::getAvailableLocales(int32_t& count)
 
 BreakIterator::BreakIterator()
 {
-    *validLocale = *actualLocale = *requestLocale = 0;
 }
 
 BreakIterator::BreakIterator(const BreakIterator &other) : UObject(other) {
-    uprv_strncpy(actualLocale, other.actualLocale, sizeof(actualLocale));
+    UErrorCode status = U_ZERO_ERROR;
-    uprv_strncpy(validLocale, other.validLocale, sizeof(validLocale));
+    U_LOCALE_BASED(locBased, *this);
-    uprv_strncpy(requestLocale, other.requestLocale, sizeof(requestLocale));
+    locBased.setLocaleIDs(other.validLocale, other.actualLocale, status);
+    LocaleBased::setLocaleID(other.requestLocale, requestLocale, status);
+    U_ASSERT(U_SUCCESS(status));
 }
 
 BreakIterator &BreakIterator::operator =(const BreakIterator &other) {
     if (this != &other) {
-        uprv_strncpy(actualLocale, other.actualLocale, sizeof(actualLocale));
+        UErrorCode status = U_ZERO_ERROR;
-        uprv_strncpy(validLocale, other.validLocale, sizeof(validLocale));
+        U_LOCALE_BASED(locBased, *this);
-        uprv_strncpy(requestLocale, other.requestLocale, sizeof(requestLocale));
+        locBased.setLocaleIDs(other.validLocale, other.actualLocale, status);
+        LocaleBased::setLocaleID(other.requestLocale, requestLocale, status);
+        U_ASSERT(U_SUCCESS(status));
     }
     return *this;
 }
 
 BreakIterator::~BreakIterator()
 {
+    delete validLocale;
+    delete actualLocale;
+    delete requestLocale;
 }
 
 // ------------------------------------------
@@ -394,7 +399,7 @@ BreakIterator::createInstance(const Locale& loc, int32_t kind, UErrorCode& statu
         // revisit this in ICU 3.0 and clean it up/fix it/remove it.
         if (U_SUCCESS(status) && (result != nullptr) && *actualLoc.getName() != 0) {
             U_LOCALE_BASED(locBased, *result);
-            locBased.setLocaleIDs(actualLoc.getName(), actualLoc.getName());
+            locBased.setLocaleIDs(actualLoc.getName(), actualLoc.getName(), status);
         }
         return result;
     }
@@ -488,6 +493,7 @@ BreakIterator::makeInstance(const Locale& loc, int32_t kind, UErrorCode& status)
     }
 
     if (U_FAILURE(status)) {
+        delete result;
         return nullptr;
     }
 
@@ -496,20 +502,25 @@ BreakIterator::makeInstance(const Locale& loc, int32_t kind, UErrorCode& status)
 
 Locale
 BreakIterator::getLocale(ULocDataLocaleType type, UErrorCode& status) const {
-    if (type == ULOC_REQUESTED_LOCALE) {
+    if (U_FAILURE(status)) {
-        return {requestLocale};
+        return Locale::getRoot();
     }
-    U_LOCALE_BASED(locBased, *this);
+    if (type == ULOC_REQUESTED_LOCALE) {
-    return locBased.getLocale(type, status);
+        return requestLocale == nullptr ?
+            Locale::getRoot() : Locale(requestLocale->data());
+    }
+    return LocaleBased::getLocale(validLocale, actualLocale, type, status);
 }
 
 const char *
 BreakIterator::getLocaleID(ULocDataLocaleType type, UErrorCode& status) const {
-    if (type == ULOC_REQUESTED_LOCALE) {
+    if (U_FAILURE(status)) {
-        return requestLocale;
+        return nullptr;
     }
-    U_LOCALE_BASED(locBased, *this);
+    if (type == ULOC_REQUESTED_LOCALE) {
-    return locBased.getLocaleID(type, status);
+        return requestLocale == nullptr ?  "" : requestLocale->data();
+    }
+    return LocaleBased::getLocaleID(validLocale, actualLocale, type, status);
 }
 
 
@@ -536,8 +547,10 @@ int32_t BreakIterator::getRuleStatusVec(int32_t *fillInVec, int32_t capacity, UE
 }
 
 BreakIterator::BreakIterator (const Locale& valid, const Locale& actual) {
+  UErrorCode status = U_ZERO_ERROR;
   U_LOCALE_BASED(locBased, (*this));
-  locBased.setLocaleIDs(valid, actual);
+  locBased.setLocaleIDs(valid.getName(), actual.getName(), status);
+  U_ASSERT(U_SUCCESS(status));
 }
 
 U_NAMESPACE_END

deps/icu-small/source/common/charstr.cpp

@@ -70,6 +70,15 @@ CharString &CharString::copyFrom(const CharString &s, UErrorCode &errorCode) {
     return *this;
 }
 
+CharString &CharString::copyFrom(StringPiece s, UErrorCode &errorCode) {
+    if (U_FAILURE(errorCode)) {
+        return *this;
+    }
+    len = 0;
+    append(s, errorCode);
+    return *this;
+}
+
 int32_t CharString::lastIndexOf(char c) const {
     for(int32_t i=len; i>0;) {
         if(buffer[--i]==c) {
@@ -143,7 +152,7 @@ CharString &CharString::append(const char *s, int32_t sLength, UErrorCode &error
     return *this;
 }
 
-CharString &CharString::appendNumber(int32_t number, UErrorCode &status) {
+CharString &CharString::appendNumber(int64_t number, UErrorCode &status) {
     if (number < 0) {
         this->append('-', status);
         if (U_FAILURE(status)) {

deps/icu-small/source/common/charstr.h

@@ -74,6 +74,7 @@ public:
      * use a UErrorCode where memory allocations might be needed.
      */
     CharString &copyFrom(const CharString &other, UErrorCode &errorCode);
+    CharString &copyFrom(StringPiece s, UErrorCode &errorCode);
 
     UBool isEmpty() const { return len==0; }
     int32_t length() const { return len; }
@@ -135,7 +136,7 @@ public:
     }
     CharString &append(const char *s, int32_t sLength, UErrorCode &status);
 
-    CharString &appendNumber(int32_t number, UErrorCode &status);
+    CharString &appendNumber(int64_t number, UErrorCode &status);
 
     /**
      * Returns a writable buffer for appending and writes the buffer's capacity to

deps/icu-small/source/common/locbased.cpp

@@ -12,44 +12,84 @@
 */
 #include "locbased.h"
 #include "cstring.h"
+#include "charstr.h"
 
 U_NAMESPACE_BEGIN
 
-Locale LocaleBased::getLocale(ULocDataLocaleType type, UErrorCode& status) const {
+Locale LocaleBased::getLocale(const CharString* valid, const CharString* actual,
-    const char* id = getLocaleID(type, status);
+                              ULocDataLocaleType type, UErrorCode& status) {
+    const char* id = getLocaleID(valid, actual, type, status);
     return Locale(id != nullptr ? id : "");
 }
 
-const char* LocaleBased::getLocaleID(ULocDataLocaleType type, UErrorCode& status) const {
+const char* LocaleBased::getLocaleID(const CharString* valid, const CharString* actual,
+                                     ULocDataLocaleType type, UErrorCode& status) {
     if (U_FAILURE(status)) {
         return nullptr;
     }
 
     switch(type) {
     case ULOC_VALID_LOCALE:
-        return valid;
+        return valid == nullptr ? "" : valid->data();
     case ULOC_ACTUAL_LOCALE:
-        return actual;
+        return actual == nullptr ? "" : actual->data();
     default:
         status = U_ILLEGAL_ARGUMENT_ERROR;
         return nullptr;
     }
 }
 
-void LocaleBased::setLocaleIDs(const char* validID, const char* actualID) {
+void LocaleBased::setLocaleIDs(const CharString* validID, const CharString* actualID, UErrorCode& status) {
-    if (validID != nullptr) {
+    setValidLocaleID(validID, status);
-      uprv_strncpy(valid, validID, ULOC_FULLNAME_CAPACITY);
+    setActualLocaleID(actualID,status);
-      valid[ULOC_FULLNAME_CAPACITY-1] = 0; // always terminate
+}
+void LocaleBased::setLocaleIDs(const char* validID, const char* actualID, UErrorCode& status) {
+    setValidLocaleID(validID, status);
+    setActualLocaleID(actualID,status);
+}
+
+void LocaleBased::setLocaleID(const char* id, CharString*& dest, UErrorCode& status) {
+    if (U_FAILURE(status)) { return; }
+    if (id == nullptr || *id == 0) {
+        delete dest;
+        dest = nullptr;
+    } else {
+        if (dest == nullptr) {
+            dest = new CharString(id, status);
+            if (dest == nullptr) {
+                status = U_MEMORY_ALLOCATION_ERROR;
+                return;
+            }
+        } else {
+            dest->copyFrom(id, status);
         }
-    if (actualID != nullptr) {
-      uprv_strncpy(actual, actualID, ULOC_FULLNAME_CAPACITY);
-      actual[ULOC_FULLNAME_CAPACITY-1] = 0; // always terminate
     }
 }
 
-void LocaleBased::setLocaleIDs(const Locale& validID, const Locale& actualID) {
+void LocaleBased::setLocaleID(const CharString* id, CharString*& dest, UErrorCode& status) {
-  uprv_strcpy(valid, validID.getName());
+    if (U_FAILURE(status)) { return; }
-  uprv_strcpy(actual, actualID.getName());
+    if (id == nullptr || id->isEmpty()) {
+        delete dest;
+        dest = nullptr;
+    } else {
+        if (dest == nullptr) {
+            dest = new CharString(*id, status);
+            if (dest == nullptr) {
+                status = U_MEMORY_ALLOCATION_ERROR;
+                return;
+            }
+        } else {
+            dest->copyFrom(*id, status);
+        }
+    }
+}
+
+bool LocaleBased::equalIDs(const CharString* left, const CharString* right) {
+    // true if both are nullptr
+    if (left == nullptr && right == nullptr) return true;
+    // false if only one is nullptr
+    if (left == nullptr || right == nullptr) return false;
+    return *left == *right;
 }
 
 U_NAMESPACE_END

deps/icu-small/source/common/locbased.h

@@ -19,13 +19,14 @@
 /**
  * Macro to declare a locale LocaleBased wrapper object for the given
  * object, which must have two members named `validLocale' and
- * `actualLocale' of size ULOC_FULLNAME_CAPACITY
+ * `actualLocale' of which are pointers to the internal icu::CharString.
  */
 #define U_LOCALE_BASED(varname, objname) \
   LocaleBased varname((objname).validLocale, (objname).actualLocale)
 
 U_NAMESPACE_BEGIN
 
+class CharString;
 /**
  * A utility class that unifies the implementation of getLocale() by
  * various ICU services.  This class is likely to be removed in the
@@ -41,33 +42,35 @@ class U_COMMON_API LocaleBased : public UMemory {
      * Construct a LocaleBased wrapper around the two pointers.  These
      * will be aliased for the lifetime of this object.
      */
-    inline LocaleBased(char* validAlias, char* actualAlias);
+    inline LocaleBased(CharString*& validAlias, CharString*& actualAlias);
-
-    /**
-     * Construct a LocaleBased wrapper around the two const pointers.
-     * These will be aliased for the lifetime of this object.
-     */
-    inline LocaleBased(const char* validAlias, const char* actualAlias);
 
     /**
      * Return locale meta-data for the service object wrapped by this
      * object.  Either the valid or the actual locale may be
      * retrieved.
+     * @param valid The valid locale.
+     * @param actual The actual locale.
      * @param type either ULOC_VALID_LOCALE or ULOC_ACTUAL_LOCALE
      * @param status input-output error code
      * @return the indicated locale
      */
-    Locale getLocale(ULocDataLocaleType type, UErrorCode& status) const;
+    static Locale getLocale(
+        const CharString* valid, const CharString* actual,
+        ULocDataLocaleType type, UErrorCode& status);
 
     /**
      * Return the locale ID for the service object wrapped by this
      * object.  Either the valid or the actual locale may be
      * retrieved.
+     * @param valid The valid locale.
+     * @param actual The actual locale.
      * @param type either ULOC_VALID_LOCALE or ULOC_ACTUAL_LOCALE
      * @param status input-output error code
      * @return the indicated locale ID
      */
-    const char* getLocaleID(ULocDataLocaleType type, UErrorCode& status) const;
+    static const char* getLocaleID(
+        const CharString* valid, const CharString* actual,
+        ULocDataLocaleType type, UErrorCode& status);
 
     /**
      * Set the locale meta-data for the service object wrapped by this
@@ -75,31 +78,40 @@ class U_COMMON_API LocaleBased : public UMemory {
      * @param valid the ID of the valid locale
      * @param actual the ID of the actual locale
      */
-    void setLocaleIDs(const char* valid, const char* actual);
+    void setLocaleIDs(const char* valid, const char* actual, UErrorCode& status);
+    void setLocaleIDs(const CharString* valid, const CharString* actual, UErrorCode& status);
 
-    /**
+    static void setLocaleID(const char* id, CharString*& dest, UErrorCode& status);
-     * Set the locale meta-data for the service object wrapped by this
+    static void setLocaleID(const CharString* id, CharString*& dest, UErrorCode& status);
-     * object.
+
-     * @param valid the ID of the valid locale
+    static bool equalIDs(const CharString* left, const CharString* right);
-     * @param actual the ID of the actual locale
-     */
-    void setLocaleIDs(const Locale& valid, const Locale& actual);
 
  private:
 
-    char* valid;
+    void setValidLocaleID(const CharString* id, UErrorCode& status);
+    void setActualLocaleID(const CharString* id, UErrorCode& status);
+    void setValidLocaleID(const char* id, UErrorCode& status);
+    void setActualLocaleID(const char* id, UErrorCode& status);
 
-    char* actual;
+    CharString*& valid;
+    CharString*& actual;
 };
 
-inline LocaleBased::LocaleBased(char* validAlias, char* actualAlias) :
+inline LocaleBased::LocaleBased(CharString*& validAlias, CharString*& actualAlias) :
     valid(validAlias), actual(actualAlias) {
 }
 
-inline LocaleBased::LocaleBased(const char* validAlias,
+inline void LocaleBased::setValidLocaleID(const CharString* id, UErrorCode& status) {
-                                const char* actualAlias) :
+    setLocaleID(id, valid, status);
-    // ugh: cast away const
+}
-    valid(const_cast<char*>(validAlias)), actual(const_cast<char*>(actualAlias)) {
+inline void LocaleBased::setActualLocaleID(const CharString* id, UErrorCode& status) {
+    setLocaleID(id, actual, status);
+}
+inline void LocaleBased::setValidLocaleID(const char* id, UErrorCode& status) {
+    setLocaleID(id, valid, status);
+}
+inline void LocaleBased::setActualLocaleID(const char* id, UErrorCode& status) {
+    setLocaleID(id, actual, status);
 }
 
 U_NAMESPACE_END

deps/icu-small/source/common/locdispnames.cpp

@@ -19,6 +19,8 @@
 *   that then do not depend on resource bundle code and display name data.
 */
 
+#include <string_view>
+
 #include "unicode/utypes.h"
 #include "unicode/brkiter.h"
 #include "unicode/locid.h"
@@ -359,7 +361,7 @@ _getStringOrCopyKey(const char *path, const char *locale,
     return u_terminateUChars(dest, destCapacity, length, &errorCode);
 }
 
-using UDisplayNameGetter = icu::CharString(const char*, UErrorCode&);
+using UDisplayNameGetter = icu::CharString(std::string_view, UErrorCode&);
 
 int32_t
 _getDisplayNameForComponent(const char *locale,
@@ -377,6 +379,10 @@ _getDisplayNameForComponent(const char *locale,
         return 0;
     }
 
+    if (locale == nullptr) {
+        locale = uloc_getDefault();
+    }
+
     localStatus = U_ZERO_ERROR;
     icu::CharString localeBuffer = (*getter)(locale, localStatus);
     if (U_FAILURE(localStatus)) {

deps/icu-small/source/common/locid.cpp

@@ -1828,8 +1828,13 @@ ulocimp_isCanonicalizedLocaleForTest(const char* localeName)
 
 U_NAMESPACE_BEGIN
 
-/*This function initializes a Locale from a C locale ID*/
 Locale& Locale::init(const char* localeID, UBool canonicalize)
+{
+    return localeID == nullptr ? *this = getDefault() : init(StringPiece{localeID}, canonicalize);
+}
+
+/*This function initializes a Locale from a C locale ID*/
+Locale& Locale::init(StringPiece localeID, UBool canonicalize)
 {
     fIsBogus = false;
     /* Free our current storage */
@@ -1854,19 +1859,28 @@ Locale& Locale::init(const char* localeID, UBool canonicalize)
         int32_t length;
         UErrorCode err;
 
-        if(localeID == nullptr) {
-            // not an error, just set the default locale
-            return *this = getDefault();
-        }
-
         /* preset all fields to empty */
         language[0] = script[0] = country[0] = 0;
 
+        const auto parse = [canonicalize](std::string_view localeID,
+                                          char* name,
+                                          int32_t nameCapacity,
+                                          UErrorCode& status) {
+            return ByteSinkUtil::viaByteSinkToTerminatedChars(
+                name, nameCapacity,
+                [&](ByteSink& sink, UErrorCode& status) {
+                    if (canonicalize) {
+                        ulocimp_canonicalize(localeID, sink, status);
+                    } else {
+                        ulocimp_getName(localeID, sink, status);
+                    }
+                },
+                status);
+        };
+
         // "canonicalize" the locale ID to ICU/Java format
         err = U_ZERO_ERROR;
-        length = canonicalize ?
+        length = parse(localeID, fullName, sizeof fullNameBuffer, err);
-            uloc_canonicalize(localeID, fullName, sizeof(fullNameBuffer), &err) :
-            uloc_getName(localeID, fullName, sizeof(fullNameBuffer), &err);
 
         if (err == U_BUFFER_OVERFLOW_ERROR || length >= static_cast<int32_t>(sizeof(fullNameBuffer))) {
             U_ASSERT(baseName == nullptr);
@@ -1877,9 +1891,7 @@ Locale& Locale::init(const char* localeID, UBool canonicalize)
             }
             fullName = newFullName;
             err = U_ZERO_ERROR;
-            length = canonicalize ?
+            length = parse(localeID, fullName, length + 1, err);
-                uloc_canonicalize(localeID, fullName, length+1, &err) :
-                uloc_getName(localeID, fullName, length+1, &err);
         }
         if(U_FAILURE(err) || err == U_STRING_NOT_TERMINATED_WARNING) {
             /* should never occur */
@@ -2200,6 +2212,13 @@ Locale::createFromName (const char *name)
     }
 }
 
+Locale U_EXPORT2
+Locale::createFromName(StringPiece name) {
+    Locale loc("");
+    loc.init(name, false);
+    return loc;
+}
+
 Locale U_EXPORT2
 Locale::createCanonical(const char* name) {
     Locale loc("");

deps/icu-small/source/common/loclikely.cpp

@@ -300,6 +300,9 @@ ulocimp_addLikelySubtags(const char* localeID,
                          icu::ByteSink& sink,
                          UErrorCode& status) {
     if (U_FAILURE(status)) { return; }
+    if (localeID == nullptr) {
+        localeID = uloc_getDefault();
+    }
     icu::CharString localeBuffer = ulocimp_canonicalize(localeID, status);
     _uloc_addLikelySubtags(localeBuffer.data(), sink, status);
 }
@@ -334,6 +337,9 @@ ulocimp_minimizeSubtags(const char* localeID,
                         bool favorScript,
                         UErrorCode& status) {
     if (U_FAILURE(status)) { return; }
+    if (localeID == nullptr) {
+        localeID = uloc_getDefault();
+    }
     icu::CharString localeBuffer = ulocimp_canonicalize(localeID, status);
     _uloc_minimizeSubtags(localeBuffer.data(), sink, favorScript, status);
 }
@@ -349,7 +355,9 @@ uloc_isRightToLeft(const char *locale) {
     UErrorCode errorCode = U_ZERO_ERROR;
     icu::CharString lang;
     icu::CharString script;
-    ulocimp_getSubtags(locale, &lang, &script, nullptr, nullptr, nullptr, errorCode);
+    ulocimp_getSubtags(
+        locale == nullptr ? uloc_getDefault() : locale,
+        &lang, &script, nullptr, nullptr, nullptr, errorCode);
     if (U_FAILURE(errorCode) || script.isEmpty()) {
         // Fastpath: We know the likely scripts and their writing direction
         // for some common languages.
@@ -369,7 +377,7 @@ uloc_isRightToLeft(const char *locale) {
         if (U_FAILURE(errorCode)) {
             return false;
         }
-        ulocimp_getSubtags(likely.data(), nullptr, &script, nullptr, nullptr, nullptr, errorCode);
+        ulocimp_getSubtags(likely.toStringPiece(), nullptr, &script, nullptr, nullptr, nullptr, errorCode);
         if (U_FAILURE(errorCode) || script.isEmpty()) {
             return false;
         }
@@ -430,7 +438,7 @@ ulocimp_getRegionForSupplementalData(const char *localeID, bool inferRegion,
     icu::CharString rgBuf = GetRegionFromKey(localeID, "rg", status);
     if (U_SUCCESS(status) && rgBuf.isEmpty()) {
         // No valid rg keyword value, try for unicode_region_subtag
-        rgBuf = ulocimp_getRegion(localeID, status);
+        rgBuf = ulocimp_getRegion(localeID == nullptr ? uloc_getDefault() : localeID, status);
         if (U_SUCCESS(status) && rgBuf.isEmpty() && inferRegion) {
             // Second check for sd keyword value
             rgBuf = GetRegionFromKey(localeID, "sd", status);
@@ -439,7 +447,7 @@ ulocimp_getRegionForSupplementalData(const char *localeID, bool inferRegion,
                 UErrorCode rgStatus = U_ZERO_ERROR;
                 icu::CharString locBuf = ulocimp_addLikelySubtags(localeID, rgStatus);
                 if (U_SUCCESS(rgStatus)) {
-                    rgBuf = ulocimp_getRegion(locBuf.data(), status);
+                    rgBuf = ulocimp_getRegion(locBuf.toStringPiece(), status);
                 }
             }
         }

deps/icu-small/source/common/loclikelysubtags.cpp

@@ -527,7 +527,7 @@ LSR LikelySubtags::makeMaximizedLsrFrom(const Locale &locale,
         return {};
     }
     const char *name = locale.getName();
-    if (uprv_isAtSign(name[0]) && name[1] == 'x' && name[2] == '=') {  // name.startsWith("@x=")
+    if (!returnInputIfUnmatch && uprv_isAtSign(name[0]) && name[1] == 'x' && name[2] == '=') {  // name.startsWith("@x=")
         // Private use language tag x-subtag-subtag... which CLDR changes to
         // und-x-subtag-subtag...
         return LSR(name, "", "", LSR::EXPLICIT_LSR);

deps/icu-small/source/common/locresdata.cpp

@@ -161,6 +161,9 @@ _uloc_getOrientationHelper(const char* localeId,
 
     if (U_FAILURE(status)) { return result; }
 
+    if (localeId == nullptr) {
+        localeId = uloc_getDefault();
+    }
     icu::CharString localeBuffer = ulocimp_canonicalize(localeId, status);
 
     if (U_FAILURE(status)) { return result; }

deps/icu-small/source/common/punycode.cpp

@@ -193,7 +193,7 @@ u_strToPunycode(const char16_t *src, int32_t srcLength,
         return 0;
     }
 
-    if(src==nullptr || srcLength<-1 || (dest==nullptr && destCapacity!=0)) {
+    if(src==nullptr || srcLength<-1 || destCapacity<0 || (dest==nullptr && destCapacity!=0)) {
         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
         return 0;
     }

deps/icu-small/source/common/putil.cpp

@@ -76,7 +76,7 @@
 #include <float.h>
 
 #ifndef U_COMMON_IMPLEMENTATION
-#error U_COMMON_IMPLEMENTATION not set - must be set for all ICU source files in common/ - see https://unicode-org.github.io/icu/userguide/howtouseicu
+#error U_COMMON_IMPLEMENTATION not set - must be set for all ICU source files in common/ - see https://unicode-org.github.io/icu/userguide/icu/howtouseicu.html
 #endif
 
 

deps/icu-small/source/common/rbbinode.cpp

@@ -47,7 +47,10 @@ static int  gLastSerial = 0;
 //    Constructor.   Just set the fields to reasonable default values.
 //
 //-------------------------------------------------------------------------
-RBBINode::RBBINode(NodeType t) : UMemory() {
+RBBINode::RBBINode(NodeType t, UErrorCode& status) : UMemory() {
+    if (U_FAILURE(status)) {
+        return;
+    }
 #ifdef RBBI_DEBUG
     fSerialNum    = ++gLastSerial;
 #endif
@@ -65,10 +68,13 @@ RBBINode::RBBINode(NodeType t) : UMemory() {
     fVal          = 0;
     fPrecedence   = precZero;
 
-    UErrorCode     status = U_ZERO_ERROR;
+    fFirstPosSet  = new UVector(status);
-    fFirstPosSet  = new UVector(status);  // TODO - get a real status from somewhere
     fLastPosSet   = new UVector(status);
     fFollowPos    = new UVector(status);
+    if (U_SUCCESS(status) &&
+        (fFirstPosSet == nullptr || fLastPosSet == nullptr || fFollowPos == nullptr)) {
+        status =  U_MEMORY_ALLOCATION_ERROR;
+    }
     if      (t==opCat)    {fPrecedence = precOpCat;}
     else if (t==opOr)     {fPrecedence = precOpOr;}
     else if (t==opStart)  {fPrecedence = precStart;}
@@ -77,7 +83,10 @@ RBBINode::RBBINode(NodeType t) : UMemory() {
 }
 
 
-RBBINode::RBBINode(const RBBINode &other) : UMemory(other) {
+RBBINode::RBBINode(const RBBINode &other, UErrorCode& status) : UMemory(other) {
+    if (U_FAILURE(status)) {
+        return;
+    }
 #ifdef RBBI_DEBUG
     fSerialNum   = ++gLastSerial;
 #endif
@@ -94,10 +103,13 @@ RBBINode::RBBINode(const RBBINode &other) : UMemory(other) {
     fVal         = other.fVal;
     fRuleRoot    = false;
     fChainIn     = other.fChainIn;
-    UErrorCode     status = U_ZERO_ERROR;
     fFirstPosSet = new UVector(status);   // TODO - get a real status from somewhere
     fLastPosSet  = new UVector(status);
     fFollowPos   = new UVector(status);
+    if (U_SUCCESS(status) &&
+        (fFirstPosSet == nullptr || fLastPosSet == nullptr || fFollowPos == nullptr)) {
+        status =  U_MEMORY_ALLOCATION_ERROR;
+    }
 }
 
 
@@ -193,27 +205,54 @@ void RBBINode::NRDeleteNode(RBBINode *node) {
 //                  references in preparation for generating the DFA tables.
 //
 //-------------------------------------------------------------------------
-RBBINode *RBBINode::cloneTree() {
+constexpr int kRecursiveDepthLimit = 3500;
+RBBINode *RBBINode::cloneTree(UErrorCode &status, int depth) {
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+    // If the depth of the stack is too deep, we return U_INPUT_TOO_LONG_ERROR
+    // to avoid stack overflow crash.
+    if (depth > kRecursiveDepthLimit) {
+        status = U_INPUT_TOO_LONG_ERROR;
+        return nullptr;
+    }
     RBBINode    *n;
 
     if (fType == RBBINode::varRef) {
         // If the current node is a variable reference, skip over it
         //   and clone the definition of the variable instead.
-        n = fLeftChild->cloneTree();
+        n = fLeftChild->cloneTree(status, depth+1);
+        if (U_FAILURE(status)) {
+            return nullptr;
+        }
     } else if (fType == RBBINode::uset) {
         n = this;
     } else {
-        n = new RBBINode(*this);
+        n = new RBBINode(*this, status);
+        if (U_FAILURE(status)) {
+            delete n;
+            return nullptr;
+        }
         // Check for null pointer.
-        if (n != nullptr) {
+        if (n == nullptr) {
+            status =  U_MEMORY_ALLOCATION_ERROR;
+            return nullptr;
+        }
         if (fLeftChild != nullptr) {
-                n->fLeftChild          = fLeftChild->cloneTree();
+            n->fLeftChild          = fLeftChild->cloneTree(status, depth+1);
+            if (U_FAILURE(status)) {
+                delete n;
+                return nullptr;
+            }
             n->fLeftChild->fParent = n;
         }
         if (fRightChild != nullptr) {
-                n->fRightChild          = fRightChild->cloneTree();
+            n->fRightChild          = fRightChild->cloneTree(status, depth+1);
-                n->fRightChild->fParent = n;
+            if (U_FAILURE(status)) {
+                delete n;
+                return nullptr;
             }
+            n->fRightChild->fParent = n;
         }
     }
     return n;
@@ -239,7 +278,6 @@ RBBINode *RBBINode::cloneTree() {
 //                      nested references are handled by cloneTree(), not here.
 //
 //-------------------------------------------------------------------------
-constexpr int kRecursiveDepthLimit = 3500;
 RBBINode *RBBINode::flattenVariables(UErrorCode& status, int depth) {
     if (U_FAILURE(status)) {
         return this;
@@ -251,21 +289,34 @@ RBBINode *RBBINode::flattenVariables(UErrorCode& status, int depth) {
         return this;
     }
     if (fType == varRef) {
-        RBBINode *retNode  = fLeftChild->cloneTree();
+        RBBINode *retNode  = fLeftChild->cloneTree(status, depth+1);
-        if (retNode != nullptr) {
+        if (U_FAILURE(status)) {
+            return this;
+        }
         retNode->fRuleRoot = this->fRuleRoot;
         retNode->fChainIn  = this->fChainIn;
-        }
         delete this;   // TODO: undefined behavior. Fix.
         return retNode;
     }
 
     if (fLeftChild != nullptr) {
         fLeftChild = fLeftChild->flattenVariables(status, depth+1);
+        if (fLeftChild == nullptr) {
+            status = U_MEMORY_ALLOCATION_ERROR;
+        }
+        if (U_FAILURE(status)) {
+            return this;
+        }
         fLeftChild->fParent  = this;
     }
     if (fRightChild != nullptr) {
         fRightChild = fRightChild->flattenVariables(status, depth+1);
+        if (fRightChild == nullptr) {
+            status = U_MEMORY_ALLOCATION_ERROR;
+        }
+        if (U_FAILURE(status)) {
+            return this;
+        }
         fRightChild->fParent = this;
     }
     return this;
@@ -280,7 +331,16 @@ RBBINode *RBBINode::flattenVariables(UErrorCode& status, int depth) {
 //                 the left child of the uset node.
 //
 //-------------------------------------------------------------------------
-void RBBINode::flattenSets() {
+void RBBINode::flattenSets(UErrorCode &status, int depth) {
+    if (U_FAILURE(status)) {
+        return;
+    }
+    // If the depth of the stack is too deep, we return U_INPUT_TOO_LONG_ERROR
+    // to avoid stack overflow crash.
+    if (depth > kRecursiveDepthLimit) {
+        status = U_INPUT_TOO_LONG_ERROR;
+        return;
+    }
     U_ASSERT(fType != setRef);
 
     if (fLeftChild != nullptr) {
@@ -288,11 +348,15 @@ void RBBINode::flattenSets() {
             RBBINode *setRefNode = fLeftChild;
             RBBINode *usetNode   = setRefNode->fLeftChild;
             RBBINode *replTree   = usetNode->fLeftChild;
-            fLeftChild           = replTree->cloneTree();
+            fLeftChild           = replTree->cloneTree(status, depth+1);
+            if (U_FAILURE(status)) {
+                delete setRefNode;
+                return;
+            }
             fLeftChild->fParent  = this;
             delete setRefNode;
         } else {
-            fLeftChild->flattenSets();
+            fLeftChild->flattenSets(status, depth+1);
         }
     }
 
@@ -301,11 +365,15 @@ void RBBINode::flattenSets() {
             RBBINode *setRefNode = fRightChild;
             RBBINode *usetNode   = setRefNode->fLeftChild;
             RBBINode *replTree   = usetNode->fLeftChild;
-            fRightChild           = replTree->cloneTree();
+            fRightChild           = replTree->cloneTree(status, depth+1);
+            if (U_FAILURE(status)) {
+                delete setRefNode;
+                return;
+            }
             fRightChild->fParent  = this;
             delete setRefNode;
         } else {
-            fRightChild->flattenSets();
+            fRightChild->flattenSets(status, depth+1);
         }
     }
 }

deps/icu-small/source/common/rbbinode.h

@@ -91,14 +91,14 @@ class RBBINode : public UMemory {
         UVector       *fFollowPos;
 
 
-        RBBINode(NodeType t);
+        RBBINode(NodeType t, UErrorCode& status);
-        RBBINode(const RBBINode &other);
+        RBBINode(const RBBINode &other, UErrorCode& status);
         ~RBBINode();
         static void  NRDeleteNode(RBBINode *node);
         
-        RBBINode    *cloneTree();
+        RBBINode    *cloneTree(UErrorCode &status, int depth=0);
         RBBINode    *flattenVariables(UErrorCode &status, int depth=0);
-        void         flattenSets();
+        void         flattenSets(UErrorCode &status, int depth=0);
         void         findNodes(UVector *dest, RBBINode::NodeType kind, UErrorCode &status);
 
 #ifdef RBBI_DEBUG

deps/icu-small/source/common/rbbiscan.cpp

@@ -767,15 +767,24 @@ void RBBIRuleScanner::findSetFor(const UnicodeString &s, RBBINode *node, Unicode
             c = s.char32At(0);
             setToAdopt = new UnicodeSet(c, c);
         }
+        if (setToAdopt == nullptr) {
+            error(U_MEMORY_ALLOCATION_ERROR);
+            return;
+        }
     }
 
     //
     // Make a new uset node to refer to this UnicodeSet
     // This new uset node becomes the child of the caller's setReference node.
     //
-    RBBINode *usetNode    = new RBBINode(RBBINode::uset);
+    UErrorCode localStatus = U_ZERO_ERROR;
+    RBBINode *usetNode    = new RBBINode(RBBINode::uset, localStatus);
     if (usetNode == nullptr) {
-        error(U_MEMORY_ALLOCATION_ERROR);
+        localStatus = U_MEMORY_ALLOCATION_ERROR;
+    }
+    if (U_FAILURE(localStatus)) {
+        delete usetNode;
+        error(localStatus);
         delete setToAdopt;
         return;
     }
@@ -1191,7 +1200,7 @@ RBBINode  *RBBIRuleScanner::pushNewNode(RBBINode::NodeType  t) {
         return nullptr;
     }
     fNodeStackPtr++;
-    fNodeStack[fNodeStackPtr] = new RBBINode(t);
+    fNodeStack[fNodeStackPtr] = new RBBINode(t, *fRB->fStatus);
     if (fNodeStack[fNodeStackPtr] == nullptr) {
         *fRB->fStatus = U_MEMORY_ALLOCATION_ERROR;
     }

deps/icu-small/source/common/rbbisetb.cpp

@@ -375,7 +375,11 @@ void  RBBISetBuilder::addValToSets(UVector *sets, uint32_t val) {
 }
 
 void  RBBISetBuilder::addValToSet(RBBINode *usetNode, uint32_t val) {
-    RBBINode *leafNode = new RBBINode(RBBINode::leafChar);
+    RBBINode *leafNode = new RBBINode(RBBINode::leafChar, *fStatus);
+    if (U_FAILURE(*fStatus)) {
+        delete leafNode;
+        return;
+    }
     if (leafNode == nullptr) {
         *fStatus = U_MEMORY_ALLOCATION_ERROR;
         return;
@@ -388,9 +392,13 @@ void  RBBISetBuilder::addValToSet(RBBINode *usetNode, uint32_t val) {
         // There are already input symbols present for this set.
         // Set up an OR node, with the previous stuff as the left child
         //   and the new value as the right child.
-        RBBINode *orNode = new RBBINode(RBBINode::opOr);
+        RBBINode *orNode = new RBBINode(RBBINode::opOr, *fStatus);
         if (orNode == nullptr) {
             *fStatus = U_MEMORY_ALLOCATION_ERROR;
+        }
+        if (U_FAILURE(*fStatus)) {
+            delete orNode;
+            delete leafNode;
             return;
         }
         orNode->fLeftChild  = usetNode->fLeftChild;

deps/icu-small/source/common/rbbitblb.cpp

@@ -99,13 +99,22 @@ void  RBBITableBuilder::buildForwardTable() {
     //   {bof} fake character.
     // 
     if (fRB->fSetBuilder->sawBOF()) {
-        RBBINode *bofTop    = new RBBINode(RBBINode::opCat);
+        RBBINode *bofTop    = new RBBINode(RBBINode::opCat, *fStatus);
-        RBBINode *bofLeaf   = new RBBINode(RBBINode::leafChar);
+        if (bofTop == nullptr) {
-        // Delete and exit if memory allocation failed.
-        if (bofTop == nullptr || bofLeaf == nullptr) {
             *fStatus = U_MEMORY_ALLOCATION_ERROR;
+        }
+        if (U_FAILURE(*fStatus)) {
             delete bofTop;
+            return;
+        }
+        RBBINode *bofLeaf   = new RBBINode(RBBINode::leafChar, *fStatus);
+        // Delete and exit if memory allocation failed.
+        if (bofLeaf == nullptr) {
+            *fStatus = U_MEMORY_ALLOCATION_ERROR;
+        }
+        if (U_FAILURE(*fStatus)) {
             delete bofLeaf;
+            delete bofTop;
             return;
         }
         bofTop->fLeftChild  = bofLeaf;
@@ -120,18 +129,23 @@ void  RBBITableBuilder::buildForwardTable() {
     //   Appears as a cat-node, left child being the original tree,
     //   right child being the end marker.
     //
-    RBBINode *cn = new RBBINode(RBBINode::opCat);
+    RBBINode *cn = new RBBINode(RBBINode::opCat, *fStatus);
     // Exit if memory allocation failed.
     if (cn == nullptr) {
         *fStatus = U_MEMORY_ALLOCATION_ERROR;
+    }
+    if (U_FAILURE(*fStatus)) {
+        delete cn;
         return;
     }
     cn->fLeftChild = fTree;
     fTree->fParent = cn;
-    RBBINode *endMarkerNode = cn->fRightChild = new RBBINode(RBBINode::endMark);
+    RBBINode *endMarkerNode = cn->fRightChild = new RBBINode(RBBINode::endMark, *fStatus);
     // Delete and exit if memory allocation failed.
     if (cn->fRightChild == nullptr) {
         *fStatus = U_MEMORY_ALLOCATION_ERROR;
+    }
+    if (U_FAILURE(*fStatus)) {
         delete cn;
         return;
     }
@@ -142,7 +156,7 @@ void  RBBITableBuilder::buildForwardTable() {
     //  Replace all references to UnicodeSets with the tree for the equivalent
     //      expression.
     //
-    fTree->flattenSets();
+    fTree->flattenSets(*fStatus, 0);
 #ifdef RBBI_DEBUG
     if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "stree")) {
         RBBIDebugPuts("\nParse tree after flattening Unicode Set references.");

deps/icu-small/source/common/resbund.cpp

@@ -388,7 +388,7 @@ const Locale &ResourceBundle::getLocale() const {
     return ncThis->fLocale != nullptr ? *ncThis->fLocale : Locale::getDefault();
 }
 
-const Locale ResourceBundle::getLocale(ULocDataLocaleType type, UErrorCode &status) const
+Locale ResourceBundle::getLocale(ULocDataLocaleType type, UErrorCode &status) const
 {
   return ures_getLocaleByType(fResource, type, &status);
 }

deps/icu-small/source/common/ucnvmbcs.cpp

@@ -3146,11 +3146,8 @@ ucnv_MBCSGetNextUChar(UConverterToUnicodeArgs *pArgs,
     if(c<0) {
         if(U_SUCCESS(*pErrorCode) && source==sourceLimit && lastSource<source) {
             /* incomplete character byte sequence */
-            uint8_t *bytes=cnv->toUBytes;
             cnv->toULength = static_cast<int8_t>(source - lastSource);
-            do {
+            uprv_memcpy(cnv->toUBytes, lastSource, cnv->toULength);
-                *bytes++=*lastSource++;
-            } while(lastSource<source);
             *pErrorCode=U_TRUNCATED_CHAR_FOUND;
         } else if(U_FAILURE(*pErrorCode)) {
             /* callback(illegal) */

deps/icu-small/source/common/ucurr.cpp

@@ -372,12 +372,8 @@ struct CReg : public icu::UMemory {
     CReg(const char16_t* _iso, const char* _id)
         : next(nullptr)
     {
-        int32_t len = static_cast<int32_t>(uprv_strlen(_id));
+        uprv_strncpy(id, _id, sizeof(id)-1);
-        if (len > static_cast<int32_t>(sizeof(id) - 1)) {
+        id[sizeof(id)-1] = 0;
-            len = (sizeof(id)-1);
-        }
-        uprv_strncpy(id, _id, len);
-        id[len] = 0;
         u_memcpy(iso, _iso, ISO_CURRENCY_CODE_LENGTH);
         iso[ISO_CURRENCY_CODE_LENGTH] = 0;
     }
@@ -682,6 +678,9 @@ ucurr_getName(const char16_t* currency,
     // this function.
     UErrorCode ec2 = U_ZERO_ERROR;
 
+    if (locale == nullptr) {
+        locale = uloc_getDefault();
+    }
     CharString loc = ulocimp_getName(locale, ec2);
     if (U_FAILURE(ec2)) {
         *ec = U_ILLEGAL_ARGUMENT_ERROR;
@@ -780,6 +779,9 @@ ucurr_getPluralName(const char16_t* currency,
     // this function.
     UErrorCode ec2 = U_ZERO_ERROR;
 
+    if (locale == nullptr) {
+        locale = uloc_getDefault();
+    }
     CharString loc = ulocimp_getName(locale, ec2);
     if (U_FAILURE(ec2)) {
         *ec = U_ILLEGAL_ARGUMENT_ERROR;
@@ -973,6 +975,9 @@ collectCurrencyNames(const char* locale,
     // Look up the Currencies resource for the given locale.
     UErrorCode ec2 = U_ZERO_ERROR;
 
+    if (locale == nullptr) {
+        locale = uloc_getDefault();
+    }
     CharString loc = ulocimp_getName(locale, ec2);
     if (U_FAILURE(ec2)) {
         ec = U_ILLEGAL_ARGUMENT_ERROR;

deps/icu-small/source/common/uloc.cpp

@@ -482,8 +482,8 @@ constexpr CanonicalizationMap CANONICALIZE_MAP[] = {
 
 /* ### BCP47 Conversion *******************************************/
 /* Gets the size of the shortest subtag in the given localeID. */
-int32_t getShortestSubtagLength(const char *localeID) {
+int32_t getShortestSubtagLength(std::string_view localeID) {
-    int32_t localeIDLength = static_cast<int32_t>(uprv_strlen(localeID));
+    int32_t localeIDLength = static_cast<int32_t>(localeID.length());
     int32_t length = localeIDLength;
     int32_t tmpLength = 0;
     int32_t i;
@@ -507,8 +507,8 @@ int32_t getShortestSubtagLength(const char *localeID) {
     return length;
 }
 /* Test if the locale id has BCP47 u extension and does not have '@' */
-inline bool _hasBCP47Extension(const char *id) {
+inline bool _hasBCP47Extension(std::string_view id) {
-    return id != nullptr && uprv_strstr(id, "@") == nullptr && getShortestSubtagLength(id) == 1;
+    return id.find('@') == std::string_view::npos && getShortestSubtagLength(id) == 1;
 }
 
 /* ### Keywords **************************************************/
@@ -523,10 +523,9 @@ inline bool UPRV_OK_VALUE_PUNCTUATION(char c) { return c == '_' || c == '-' || c
 #define ULOC_MAX_NO_KEYWORDS 25
 
 U_CAPI const char * U_EXPORT2
-locale_getKeywordsStart(const char *localeID) {
+locale_getKeywordsStart(std::string_view localeID) {
-    const char *result = nullptr;
+    if (size_t pos = localeID.find('@'); pos != std::string_view::npos) {
-    if((result = uprv_strchr(localeID, '@')) != nullptr) {
+        return localeID.data() + pos;
-        return result;
     }
 #if (U_CHARSET_FAMILY == U_EBCDIC_FAMILY)
     else {
@@ -536,8 +535,8 @@ locale_getKeywordsStart(const char *localeID) {
         static const uint8_t ebcdicSigns[] = { 0x7C, 0x44, 0x66, 0x80, 0xAC, 0xAE, 0xAF, 0xB5, 0xEC, 0xEF, 0x00 };
         const uint8_t *charToFind = ebcdicSigns;
         while(*charToFind) {
-            if((result = uprv_strchr(localeID, *charToFind)) != nullptr) {
+            if (size_t pos = localeID.find(*charToFind); pos != std::string_view::npos) {
-                return result;
+                return localeID.data() + pos;
             }
             charToFind++;
         }
@@ -590,7 +589,7 @@ compareKeywordStructs(const void * /*context*/, const void *left, const void *ri
 }  // namespace
 
 U_EXPORT CharString
-ulocimp_getKeywords(const char* localeID,
+ulocimp_getKeywords(std::string_view localeID,
                     char prev,
                     bool valuesToo,
                     UErrorCode& status)
@@ -607,7 +606,7 @@ ulocimp_getKeywords(const char* localeID,
 }
 
 U_EXPORT void
-ulocimp_getKeywords(const char* localeID,
+ulocimp_getKeywords(std::string_view localeID,
                     char prev,
                     ByteSink& sink,
                     bool valuesToo,
@@ -619,9 +618,8 @@ ulocimp_getKeywords(const char* localeID,
 
     int32_t maxKeywords = ULOC_MAX_NO_KEYWORDS;
     int32_t numKeywords = 0;
-    const char* pos = localeID;
+    size_t equalSign = std::string_view::npos;
-    const char* equalSign = nullptr;
+    size_t semicolon = std::string_view::npos;
-    const char* semicolon = nullptr;
     int32_t i = 0, j, n;
 
     if(prev == '@') { /* start of keyword definition */
@@ -629,74 +627,72 @@ ulocimp_getKeywords(const char* localeID,
         do {
             bool duplicate = false;
             /* skip leading spaces */
-            while(*pos == ' ') {
+            while (localeID.front() == ' ') {
-                pos++;
+                localeID.remove_prefix(1);
             }
-            if (!*pos) { /* handle trailing "; " */
+            if (localeID.empty()) { /* handle trailing "; " */
                 break;
             }
             if(numKeywords == maxKeywords) {
                 status = U_INTERNAL_PROGRAM_ERROR;
                 return;
             }
-            equalSign = uprv_strchr(pos, '=');
+            equalSign = localeID.find('=');
-            semicolon = uprv_strchr(pos, ';');
+            semicolon = localeID.find(';');
             /* lack of '=' [foo@currency] is illegal */
             /* ';' before '=' [foo@currency;collation=pinyin] is illegal */
-            if(!equalSign || (semicolon && semicolon<equalSign)) {
+            if (equalSign == std::string_view::npos ||
+                (semicolon != std::string_view::npos && semicolon < equalSign)) {
+                status = U_INVALID_FORMAT_ERROR;
+                return;
+            }
+            /* zero-length keyword is an error. */
+            if (equalSign == 0) {
                 status = U_INVALID_FORMAT_ERROR;
                 return;
             }
             /* need to normalize both keyword and keyword name */
-            if(equalSign - pos >= ULOC_KEYWORD_BUFFER_LEN) {
+            if (equalSign >= ULOC_KEYWORD_BUFFER_LEN) {
                 /* keyword name too long for internal buffer */
                 status = U_INTERNAL_PROGRAM_ERROR;
                 return;
             }
-            for(i = 0, n = 0; i < equalSign - pos; ++i) {
+            for (i = 0, n = 0; static_cast<size_t>(i) < equalSign; ++i) {
-                if (pos[i] != ' ') {
+                if (localeID[i] != ' ') {
-                    keywordList[numKeywords].keyword[n++] = uprv_tolower(pos[i]);
+                    keywordList[numKeywords].keyword[n++] = uprv_tolower(localeID[i]);
                 }
             }
 
-            /* zero-length keyword is an error. */
-            if (n == 0) {
-                status = U_INVALID_FORMAT_ERROR;
-                return;
-            }
-
             keywordList[numKeywords].keyword[n] = 0;
             keywordList[numKeywords].keywordLen = n;
             /* now grab the value part. First we skip the '=' */
             equalSign++;
             /* then we leading spaces */
-            while(*equalSign == ' ') {
+            while (equalSign < localeID.length() && localeID[equalSign] == ' ') {
                 equalSign++;
             }
 
             /* Premature end or zero-length value */
-            if (!*equalSign || equalSign == semicolon) {
+            if (equalSign == localeID.length() || equalSign == semicolon) {
                 status = U_INVALID_FORMAT_ERROR;
                 return;
             }
 
-            keywordList[numKeywords].valueStart = equalSign;
+            keywordList[numKeywords].valueStart = localeID.data() + equalSign;
 
-            pos = semicolon;
+            std::string_view value = localeID;
-            i = 0;
+            if (semicolon != std::string_view::npos) {
-            if(pos) {
+                value.remove_suffix(value.length() - semicolon);
-                while(*(pos - i - 1) == ' ') {
+                localeID.remove_prefix(semicolon + 1);
-                    i++;
-                }
-                keywordList[numKeywords].valueLen = static_cast<int32_t>(pos - equalSign - i);
-                pos++;
             } else {
-                i = static_cast<int32_t>(uprv_strlen(equalSign));
+                localeID = {};
-                while(i && equalSign[i-1] == ' ') {
-                    i--;
             }
-                keywordList[numKeywords].valueLen = i;
+            value.remove_prefix(equalSign);
+            if (size_t last = value.find_last_not_of(' '); last != std::string_view::npos) {
+                value.remove_suffix(value.length() - last - 1);
             }
+            keywordList[numKeywords].valueLen = static_cast<int32_t>(value.length());
+
             /* If this is a duplicate keyword, then ignore it */
             for (j=0; j<numKeywords; ++j) {
                 if (uprv_strcmp(keywordList[j].keyword, keywordList[numKeywords].keyword) == 0) {
@@ -707,7 +703,7 @@ ulocimp_getKeywords(const char* localeID,
             if (!duplicate) {
                 ++numKeywords;
             }
-        } while(pos);
+        } while (!localeID.empty());
 
         /* now we have a list of keywords */
         /* we need to sort it */
@@ -784,7 +780,7 @@ ulocimp_getKeywordValue(const char* localeID,
       return;
     }
 
-    if (_hasBCP47Extension(localeID)) {
+    if (localeID != nullptr && _hasBCP47Extension(localeID)) {
         tempBuffer = ulocimp_forLanguageTag(localeID, -1, nullptr, status);
         tmpLocaleID = U_SUCCESS(status) && !tempBuffer.isEmpty() ? tempBuffer.data() : localeID;
     } else {
@@ -889,7 +885,8 @@ uloc_setKeywordValue(const char* keywordName,
         return 0;
     }
 
-    char* keywords = const_cast<char*>(locale_getKeywordsStart(buffer));
+    char* keywords = const_cast<char*>(
+        locale_getKeywordsStart({buffer, static_cast<std::string_view::size_type>(bufLen)}));
     int32_t baseLen = keywords == nullptr ? bufLen : keywords - buffer;
     // Remove -1 from the capacity so that this function can guarantee NUL termination.
     CheckedArrayByteSink sink(keywords == nullptr ? buffer + bufLen : keywords,
@@ -921,7 +918,7 @@ ulocimp_setKeywordValue(std::string_view keywordName,
 {
     if (U_FAILURE(status)) { return; }
     std::string_view keywords;
-    if (const char* start = locale_getKeywordsStart(localeID.data()); start != nullptr) {
+    if (const char* start = locale_getKeywordsStart(localeID.toStringPiece()); start != nullptr) {
         // This is safe because CharString::truncate() doesn't actually erase any
         // data, but simply sets the position for where new data will be written.
         int32_t size = start - localeID.data();
@@ -1138,15 +1135,18 @@ inline bool _isPrefixLetter(char a) { return a == 'x' || a == 'X' || a == 'i' ||
 
 /*returns true if one of the special prefixes is here (s=string)
   'x-' or 'i-' */
-inline bool _isIDPrefix(const char *s) { return _isPrefixLetter(s[0]) && _isIDSeparator(s[1]); }
+inline bool _isIDPrefix(std::string_view s) {
+    return s.size() >= 2 && _isPrefixLetter(s[0]) && _isIDSeparator(s[1]);
+}
 
 /* Dot terminates it because of POSIX form  where dot precedes the codepage
  * except for variant
  */
-inline bool _isTerminator(char a) { return a == 0 || a == '.' || a == '@'; }
+inline bool _isTerminator(char a) { return a == '.' || a == '@'; }
 
-inline bool _isBCP47Extension(const char* p) {
+inline bool _isBCP47Extension(std::string_view p) {
-    return p[0] == '-' &&
+    return p.size() >= 3 &&
+           p[0] == '-' &&
            (p[1] == 't' || p[1] == 'T' ||
             p[1] == 'u' || p[1] == 'U' ||
             p[1] == 'x' || p[1] == 'X') &&
@@ -1202,49 +1202,44 @@ namespace {
  * TODO try to use this in Locale
  */
 
-void
+size_t _getLanguage(std::string_view localeID, ByteSink* sink, UErrorCode& status) {
-_getLanguage(const char* localeID,
+    size_t skip = 0;
-             ByteSink* sink,
+    if (localeID.size() == 4 && uprv_strnicmp(localeID.data(), "root", 4) == 0) {
-             const char** pEnd,
+        skip = 4;
-             UErrorCode& status) {
+        localeID.remove_prefix(skip);
-    U_ASSERT(pEnd != nullptr);
+    } else if (localeID.size() >= 3 && uprv_strnicmp(localeID.data(), "und", 3) == 0 &&
-    *pEnd = localeID;
+               (localeID.size() == 3 ||
-
-    if (uprv_stricmp(localeID, "root") == 0) {
-        localeID += 4;
-    } else if (uprv_strnicmp(localeID, "und", 3) == 0 &&
-               (localeID[3] == '\0' ||
                 localeID[3] == '-' ||
                 localeID[3] == '_' ||
                 localeID[3] == '@')) {
-        localeID += 3;
+        skip = 3;
+        localeID.remove_prefix(skip);
     }
 
     constexpr int32_t MAXLEN = ULOC_LANG_CAPACITY - 1;  // Minus NUL.
 
     /* if it starts with i- or x- then copy that prefix */
-    int32_t len = _isIDPrefix(localeID) ? 2 : 0;
+    size_t len = _isIDPrefix(localeID) ? 2 : 0;
-    while (!_isTerminator(localeID[len]) && !_isIDSeparator(localeID[len])) {
+    while (len < localeID.size() && !_isTerminator(localeID[len]) && !_isIDSeparator(localeID[len])) {
         if (len == MAXLEN) {
             status = U_ILLEGAL_ARGUMENT_ERROR;
-            return;
+            return 0;
         }
         len++;
     }
 
-    *pEnd = localeID + len;
+    if (sink == nullptr || len == 0) { return skip + len; }
-    if (sink == nullptr || len == 0) { return; }
 
-    int32_t minCapacity = uprv_max(len, 4);  // Minimum 3 letters plus NUL.
+    int32_t minCapacity = uprv_max(static_cast<int32_t>(len), 4);  // Minimum 3 letters plus NUL.
     char scratch[MAXLEN];
     int32_t capacity = 0;
     char* buffer = sink->GetAppendBuffer(
             minCapacity, minCapacity, scratch, UPRV_LENGTHOF(scratch), &capacity);
 
-    for (int32_t i = 0; i < len; ++i) {
+    for (size_t i = 0; i < len; ++i) {
         buffer[i] = uprv_tolower(localeID[i]);
     }
-    if (_isIDSeparator(localeID[1])) {
+    if (localeID.size() >= 2 && _isIDSeparator(localeID[1])) {
         buffer[1] = '-';
     }
 
@@ -1256,32 +1251,26 @@ _getLanguage(const char* localeID,
         if (offset.has_value()) {
             const char* const alias = LANGUAGES[*offset];
             sink->Append(alias, static_cast<int32_t>(uprv_strlen(alias)));
-            return;
+            return skip + len;
         }
     }
 
-    sink->Append(buffer, len);
+    sink->Append(buffer, static_cast<int32_t>(len));
+    return skip + len;
 }
 
-void
+size_t _getScript(std::string_view localeID, ByteSink* sink) {
-_getScript(const char* localeID,
-           ByteSink* sink,
-           const char** pEnd) {
-    U_ASSERT(pEnd != nullptr);
-    *pEnd = localeID;
-
     constexpr int32_t LENGTH = 4;
 
-    int32_t len = 0;
+    size_t len = 0;
-    while (!_isTerminator(localeID[len]) && !_isIDSeparator(localeID[len]) &&
+    while (len < localeID.size() && !_isTerminator(localeID[len]) && !_isIDSeparator(localeID[len]) &&
             uprv_isASCIILetter(localeID[len])) {
-        if (len == LENGTH) { return; }
+        if (len == LENGTH) { return 0; }
         len++;
     }
-    if (len != LENGTH) { return; }
+    if (len != LENGTH) { return 0; }
 
-    *pEnd = localeID + LENGTH;
+    if (sink == nullptr) { return len; }
-    if (sink == nullptr) { return; }
 
     char scratch[LENGTH];
     int32_t capacity = 0;
@@ -1294,27 +1283,21 @@ _getScript(const char* localeID,
     }
 
     sink->Append(buffer, LENGTH);
+    return len;
 }
 
-void
+size_t _getRegion(std::string_view localeID, ByteSink* sink) {
-_getRegion(const char* localeID,
-           ByteSink* sink,
-           const char** pEnd) {
-    U_ASSERT(pEnd != nullptr);
-    *pEnd = localeID;
-
     constexpr int32_t MINLEN = 2;
     constexpr int32_t MAXLEN = ULOC_COUNTRY_CAPACITY - 1;  // Minus NUL.
 
-    int32_t len = 0;
+    size_t len = 0;
-    while (!_isTerminator(localeID[len]) && !_isIDSeparator(localeID[len])) {
+    while (len < localeID.size() && !_isTerminator(localeID[len]) && !_isIDSeparator(localeID[len])) {
-        if (len == MAXLEN) { return; }
+        if (len == MAXLEN) { return 0; }
         len++;
     }
-    if (len < MINLEN) { return; }
+    if (len < MINLEN) { return 0; }
 
-    *pEnd = localeID + len;
+    if (sink == nullptr) { return len; }
-    if (sink == nullptr) { return; }
 
     char scratch[ULOC_COUNTRY_CAPACITY];
     int32_t capacity = 0;
@@ -1325,7 +1308,7 @@ _getRegion(const char* localeID,
             UPRV_LENGTHOF(scratch),
             &capacity);
 
-    for (int32_t i = 0; i < len; ++i) {
+    for (size_t i = 0; i < len; ++i) {
         buffer[i] = uprv_toupper(localeID[i]);
     }
 
@@ -1337,26 +1320,25 @@ _getRegion(const char* localeID,
         if (offset.has_value()) {
             const char* const alias = COUNTRIES[*offset];
             sink->Append(alias, static_cast<int32_t>(uprv_strlen(alias)));
-            return;
+            return len;
         }
     }
 
-    sink->Append(buffer, len);
+    sink->Append(buffer, static_cast<int32_t>(len));
+    return len;
 }
 
 /**
  * @param needSeparator if true, then add leading '_' if any variants
  * are added to 'variant'
  */
-void
+size_t
-_getVariant(const char* localeID,
+_getVariant(std::string_view localeID,
             char prev,
             ByteSink* sink,
-            const char** pEnd,
             bool needSeparator,
             UErrorCode& status) {
-    if (U_FAILURE(status)) return;
+    if (U_FAILURE(status) || localeID.empty()) return 0;
-    if (pEnd != nullptr) { *pEnd = localeID; }
 
     // Reasonable upper limit for variants
     // There are no strict limitation of the syntax of variant in the legacy
@@ -1369,42 +1351,62 @@ _getVariant(const char* localeID,
     constexpr int32_t MAX_VARIANTS_LENGTH = 179;
 
     /* get one or more variant tags and separate them with '_' */
-    int32_t index = 0;
+    size_t index = 0;
     if (_isIDSeparator(prev)) {
         /* get a variant string after a '-' or '_' */
-        for (index=0; !_isTerminator(localeID[index]); index++) {
+        for (std::string_view sub = localeID;;) {
-            if (index >= MAX_VARIANTS_LENGTH) { // same as length > MAX_VARIANTS_LENGTH
+            size_t next = sub.find_first_of(".@_-");
+            // For historical reasons, a trailing separator is included in the variant.
+            bool finished = next == std::string_view::npos || next + 1 == sub.length();
+            size_t limit = finished ? sub.length() : next;
+            index += limit;
+            if (index > MAX_VARIANTS_LENGTH) {
                 status = U_ILLEGAL_ARGUMENT_ERROR;
-                return;
+                return 0;
-            }
-            if (needSeparator) {
-                if (sink != nullptr) {
-                    sink->Append("_", 1);
-                }
-                needSeparator = false;
-            }
-            if (sink != nullptr) {
-                char c = uprv_toupper(localeID[index]);
-                if (c == '-') c = '_';
-                sink->Append(&c, 1);
-            }
-        }
-        if (pEnd != nullptr) { *pEnd = localeID+index; }
             }
 
-    /* if there is no variant tag after a '-' or '_' then look for '@' */
+            if (sink != nullptr) {
-    if (index == 0) {
+                if (needSeparator) {
-        if (prev=='@') {
+                    sink->Append("_", 1);
-            /* keep localeID */
-        } else if((localeID=locale_getKeywordsStart(localeID))!=nullptr) {
-            ++localeID; /* point after the '@' */
                 } else {
-            return;
+                    needSeparator = true;
                 }
-        for(; !_isTerminator(localeID[index]); index++) {
+
+                int32_t length = static_cast<int32_t>(limit);
+                int32_t minCapacity = uprv_min(length, MAX_VARIANTS_LENGTH);
+                char scratch[MAX_VARIANTS_LENGTH];
+                int32_t capacity = 0;
+                char* buffer = sink->GetAppendBuffer(
+                        minCapacity, minCapacity, scratch, UPRV_LENGTHOF(scratch), &capacity);
+
+                for (size_t i = 0; i < limit; ++i) {
+                    buffer[i] = uprv_toupper(sub[i]);
+                }
+                sink->Append(buffer, length);
+            }
+
+            if (finished) { return index; }
+            sub.remove_prefix(next);
+            if (_isTerminator(sub.front()) || _isBCP47Extension(sub)) { return index; }
+            sub.remove_prefix(1);
+            index++;
+        }
+    }
+
+    size_t skip = 0;
+    /* if there is no variant tag after a '-' or '_' then look for '@' */
+    if (prev == '@') {
+        /* keep localeID */
+    } else if (const char* p = locale_getKeywordsStart(localeID); p != nullptr) {
+        skip = 1 + p - localeID.data(); /* point after the '@' */
+        localeID.remove_prefix(skip);
+    } else {
+        return 0;
+    }
+    for (; index < localeID.size() && !_isTerminator(localeID[index]); index++) {
         if (index >= MAX_VARIANTS_LENGTH) { // same as length > MAX_VARIANTS_LENGTH
             status = U_ILLEGAL_ARGUMENT_ERROR;
-                return;
+            return 0;
         }
         if (needSeparator) {
             if (sink != nullptr) {
@@ -1418,14 +1420,13 @@ _getVariant(const char* localeID,
             sink->Append(&c, 1);
         }
     }
-        if (pEnd != nullptr) { *pEnd = localeID + index; }
+    return skip + index;
-    }
 }
 
 }  // namespace
 
 U_EXPORT CharString
-ulocimp_getLanguage(const char* localeID, UErrorCode& status) {
+ulocimp_getLanguage(std::string_view localeID, UErrorCode& status) {
     return ByteSinkUtil::viaByteSinkToCharString(
         [&](ByteSink& sink, UErrorCode& status) {
             ulocimp_getSubtags(
@@ -1441,7 +1442,7 @@ ulocimp_getLanguage(const char* localeID, UErrorCode& status) {
 }
 
 U_EXPORT CharString
-ulocimp_getScript(const char* localeID, UErrorCode& status) {
+ulocimp_getScript(std::string_view localeID, UErrorCode& status) {
     return ByteSinkUtil::viaByteSinkToCharString(
         [&](ByteSink& sink, UErrorCode& status) {
             ulocimp_getSubtags(
@@ -1457,7 +1458,7 @@ ulocimp_getScript(const char* localeID, UErrorCode& status) {
 }
 
 U_EXPORT CharString
-ulocimp_getRegion(const char* localeID, UErrorCode& status) {
+ulocimp_getRegion(std::string_view localeID, UErrorCode& status) {
     return ByteSinkUtil::viaByteSinkToCharString(
         [&](ByteSink& sink, UErrorCode& status) {
             ulocimp_getSubtags(
@@ -1473,7 +1474,7 @@ ulocimp_getRegion(const char* localeID, UErrorCode& status) {
 }
 
 U_EXPORT CharString
-ulocimp_getVariant(const char* localeID, UErrorCode& status) {
+ulocimp_getVariant(std::string_view localeID, UErrorCode& status) {
     return ByteSinkUtil::viaByteSinkToCharString(
         [&](ByteSink& sink, UErrorCode& status) {
             ulocimp_getSubtags(
@@ -1490,7 +1491,7 @@ ulocimp_getVariant(const char* localeID, UErrorCode& status) {
 
 U_EXPORT void
 ulocimp_getSubtags(
-        const char* localeID,
+        std::string_view localeID,
         CharString* language,
         CharString* script,
         CharString* region,
@@ -1521,7 +1522,7 @@ ulocimp_getSubtags(
 
 U_EXPORT void
 ulocimp_getSubtags(
-        const char* localeID,
+        std::string_view localeID,
         ByteSink* language,
         ByteSink* script,
         ByteSink* region,
@@ -1531,7 +1532,7 @@ ulocimp_getSubtags(
     if (U_FAILURE(status)) { return; }
 
     if (pEnd != nullptr) {
-        *pEnd = localeID;
+        *pEnd = localeID.data();
     } else if (language == nullptr &&
                script == nullptr &&
                region == nullptr &&
@@ -1539,62 +1540,94 @@ ulocimp_getSubtags(
         return;
     }
 
+    if (localeID.empty()) { return; }
+
     bool hasRegion = false;
 
-    if (localeID == nullptr) {
+    {
-        localeID = uloc_getDefault();
+        size_t len = _getLanguage(localeID, language, status);
+        if (U_FAILURE(status)) { return; }
+        if (len > 0) {
+            localeID.remove_prefix(len);
+        }
     }
 
-    _getLanguage(localeID, language, &localeID, status);
-    if (U_FAILURE(status)) { return; }
-    U_ASSERT(localeID != nullptr);
-
     if (pEnd != nullptr) {
-        *pEnd = localeID;
+        *pEnd = localeID.data();
     } else if (script == nullptr &&
                region == nullptr &&
                variant == nullptr) {
         return;
     }
 
-    if (_isIDSeparator(*localeID)) {
+    if (localeID.empty()) { return; }
-        const char* begin = localeID + 1;
+
-        const char* end = nullptr;
+    if (_isIDSeparator(localeID.front())) {
-        _getScript(begin, script, &end);
+        std::string_view sub = localeID;
-        U_ASSERT(end != nullptr);
+        sub.remove_prefix(1);
-        if (end != begin) {
+        size_t len = _getScript(sub, script);
-            localeID = end;
+        if (len > 0) {
-            if (pEnd != nullptr) { *pEnd = localeID; }
+            localeID.remove_prefix(len + 1);
+            if (pEnd != nullptr) { *pEnd = localeID.data(); }
         }
     }
 
-    if (region == nullptr && variant == nullptr && pEnd == nullptr) { return; }
+    if ((region == nullptr && variant == nullptr && pEnd == nullptr) || localeID.empty()) { return; }
 
-    if (_isIDSeparator(*localeID)) {
+    if (_isIDSeparator(localeID.front())) {
-        const char* begin = localeID + 1;
+        std::string_view sub = localeID;
-        const char* end = nullptr;
+        sub.remove_prefix(1);
-        _getRegion(begin, region, &end);
+        size_t len = _getRegion(sub, region);
-        U_ASSERT(end != nullptr);
+        if (len > 0) {
-        if (end != begin) {
             hasRegion = true;
-            localeID = end;
+            localeID.remove_prefix(len + 1);
-            if (pEnd != nullptr) { *pEnd = localeID; }
+            if (pEnd != nullptr) { *pEnd = localeID.data(); }
         }
     }
 
-    if (variant == nullptr && pEnd == nullptr) { return; }
+    if ((variant == nullptr && pEnd == nullptr) || localeID.empty()) { return; }
 
-    if (_isIDSeparator(*localeID) && !_isBCP47Extension(localeID)) {
+    bool hasVariant = false;
+
+    if (_isIDSeparator(localeID.front()) && !_isBCP47Extension(localeID)) {
+        std::string_view sub = localeID;
         /* If there was no country ID, skip a possible extra IDSeparator */
-        if (!hasRegion && _isIDSeparator(localeID[1])) {
+        size_t skip = !hasRegion && localeID.size() > 1 && _isIDSeparator(localeID[1]) ? 2 : 1;
-            localeID++;
+        sub.remove_prefix(skip);
-        }
+        size_t len = _getVariant(sub, localeID[0], variant, false, status);
-        const char* begin = localeID + 1;
-        const char* end = nullptr;
-        _getVariant(begin, *localeID, variant, &end, false, status);
         if (U_FAILURE(status)) { return; }
-        U_ASSERT(end != nullptr);
+        if (len > 0) {
-        if (end != begin && pEnd != nullptr) { *pEnd = end; }
+            hasVariant = true;
+            localeID.remove_prefix(skip + len);
+            if (pEnd != nullptr) { *pEnd = localeID.data(); }
+        }
+    }
+
+    if ((variant == nullptr && pEnd == nullptr) || localeID.empty()) { return; }
+
+    if (_isBCP47Extension(localeID)) {
+        localeID.remove_prefix(2);
+        constexpr char vaposix[] = "-va-posix";
+        constexpr size_t length = sizeof vaposix - 1;
+        for (size_t next;; localeID.remove_prefix(next)) {
+            next = localeID.find('-', 1);
+            if (next == std::string_view::npos) { break; }
+            next = localeID.find('-', next + 1);
+            bool finished = next == std::string_view::npos;
+            std::string_view sub = localeID;
+            if (!finished) { sub.remove_suffix(sub.length() - next); }
+
+            if (sub.length() == length && uprv_strnicmp(sub.data(), vaposix, length) == 0) {
+                if (variant != nullptr) {
+                    if (hasVariant) { variant->Append("_", 1); }
+                    constexpr char posix[] = "POSIX";
+                    variant->Append(posix, sizeof posix - 1);
+                }
+                if (pEnd != nullptr) { *pEnd = localeID.data() + length; }
+            }
+
+            if (finished) { break; }
+        }
     }
 }
 
@@ -1700,7 +1733,7 @@ uloc_openKeywords(const char* localeID,
     CharString tempBuffer;
     const char* tmpLocaleID;
 
-    if (_hasBCP47Extension(localeID)) {
+    if (localeID != nullptr && _hasBCP47Extension(localeID)) {
         tempBuffer = ulocimp_forLanguageTag(localeID, -1, nullptr, *status);
         tmpLocaleID = U_SUCCESS(*status) && !tempBuffer.isEmpty() ? tempBuffer.data() : localeID;
     } else {
@@ -1753,7 +1786,7 @@ constexpr int32_t I_DEFAULT_LENGTH = UPRV_LENGTHOF(i_default);
  * This is the code underlying uloc_getName and uloc_canonicalize.
  */
 void
-_canonicalize(const char* localeID,
+_canonicalize(std::string_view localeID,
               ByteSink& sink,
               uint32_t options,
               UErrorCode& err) {
@@ -1764,33 +1797,30 @@ _canonicalize(const char* localeID,
     int32_t j, fieldCount=0;
     CharString tempBuffer;  // if localeID has a BCP47 extension, tmpLocaleID points to this
     CharString localeIDWithHyphens;  // if localeID has a BPC47 extension and have _, tmpLocaleID points to this
-    const char* origLocaleID;
+    std::string_view origLocaleID;
-    const char* tmpLocaleID;
+    std::string_view tmpLocaleID;
-    const char* keywordAssign = nullptr;
+    size_t keywordAssign = std::string_view::npos;
-    const char* separatorIndicator = nullptr;
+    size_t separatorIndicator = std::string_view::npos;
 
     if (_hasBCP47Extension(localeID)) {
-        const char* localeIDPtr = localeID;
+        std::string_view localeIDPtr = localeID;
 
         // convert all underbars to hyphens, unless the "BCP47 extension" comes at the beginning of the string
-        if (uprv_strchr(localeID, '_') != nullptr && localeID[1] != '-' && localeID[1] != '_') {
+        if (localeID.size() >= 2 && localeID.find('_') != std::string_view::npos && localeID[1] != '-' && localeID[1] != '_') {
-            localeIDWithHyphens.append(localeID, -1, err);
+            localeIDWithHyphens.append(localeID, err);
             if (U_SUCCESS(err)) {
                 for (char* p = localeIDWithHyphens.data(); *p != '\0'; ++p) {
                     if (*p == '_') {
                         *p = '-';
                     }
                 }
-                localeIDPtr = localeIDWithHyphens.data();
+                localeIDPtr = localeIDWithHyphens.toStringPiece();
             }
         }
 
-        tempBuffer = ulocimp_forLanguageTag(localeIDPtr, -1, nullptr, err);
+        tempBuffer = ulocimp_forLanguageTag(localeIDPtr.data(), static_cast<int32_t>(localeIDPtr.size()), nullptr, err);
-        tmpLocaleID = U_SUCCESS(err) && !tempBuffer.isEmpty() ? tempBuffer.data() : localeIDPtr;
+        tmpLocaleID = U_SUCCESS(err) && !tempBuffer.isEmpty() ? static_cast<std::string_view>(tempBuffer.toStringPiece()) : localeIDPtr;
     } else {
-        if (localeID==nullptr) {
-           localeID=uloc_getDefault();
-        }
         tmpLocaleID=localeID;
     }
 
@@ -1801,20 +1831,25 @@ _canonicalize(const char* localeID,
     CharString script;
     CharString country;
     CharString variant;
+    const char* end = nullptr;
     ulocimp_getSubtags(
             tmpLocaleID,
             &tag,
             &script,
             &country,
             &variant,
-            &tmpLocaleID,
+            &end,
             err);
     if (U_FAILURE(err)) {
         return;
     }
+    U_ASSERT(end != nullptr);
+    if (end > tmpLocaleID.data()) {
+        tmpLocaleID.remove_prefix(end - tmpLocaleID.data());
+    }
 
-    if (tag.length() == I_DEFAULT_LENGTH &&
+    if (tag.length() == I_DEFAULT_LENGTH && origLocaleID.length() >= I_DEFAULT_LENGTH &&
-            uprv_strncmp(origLocaleID, i_default, I_DEFAULT_LENGTH) == 0) {
+            uprv_strncmp(origLocaleID.data(), i_default, I_DEFAULT_LENGTH) == 0) {
         tag.clear();
         tag.append(uloc_getDefault(), err);
     } else {
@@ -1839,15 +1874,14 @@ _canonicalize(const char* localeID,
     }
 
     /* Copy POSIX-style charset specifier, if any [mr.utf8] */
-    if (!OPTION_SET(options, _ULOC_CANONICALIZE) && *tmpLocaleID == '.') {
+    if (!OPTION_SET(options, _ULOC_CANONICALIZE) && !tmpLocaleID.empty() && tmpLocaleID.front() == '.') {
         tag.append('.', err);
-        ++tmpLocaleID;
+        tmpLocaleID.remove_prefix(1);
-        const char *atPos = nullptr;
         size_t length;
-        if((atPos = uprv_strchr(tmpLocaleID, '@')) != nullptr) {
+        if (size_t atPos = tmpLocaleID.find('@'); atPos != std::string_view::npos) {
-            length = atPos - tmpLocaleID;
+            length = atPos;
         } else {
-            length = uprv_strlen(tmpLocaleID);
+            length = tmpLocaleID.length();
         }
         // The longest charset name we found in IANA charset registry
         // https://www.iana.org/assignments/character-sets/ is
@@ -1859,33 +1893,34 @@ _canonicalize(const char* localeID,
            err = U_ILLEGAL_ARGUMENT_ERROR; /* malformed keyword name */
            return;
         }
-        tag.append(tmpLocaleID, static_cast<int32_t>(length), err);
+        if (length > 0) {
-        tmpLocaleID += length;
+            tag.append(tmpLocaleID.data(), static_cast<int32_t>(length), err);
+            tmpLocaleID.remove_prefix(length);
+        }
     }
 
     /* Scan ahead to next '@' and determine if it is followed by '=' and/or ';'
-       After this, tmpLocaleID either points to '@' or is nullptr */
+       After this, tmpLocaleID either starts at '@' or is empty. */
-    if ((tmpLocaleID=locale_getKeywordsStart(tmpLocaleID))!=nullptr) {
+    if (const char* start = locale_getKeywordsStart(tmpLocaleID); start != nullptr) {
-        keywordAssign = uprv_strchr(tmpLocaleID, '=');
+        if (start > tmpLocaleID.data()) {
-        separatorIndicator = uprv_strchr(tmpLocaleID, ';');
+            tmpLocaleID.remove_prefix(start - tmpLocaleID.data());
+        }
+        keywordAssign = tmpLocaleID.find('=');
+        separatorIndicator = tmpLocaleID.find(';');
+    } else {
+        tmpLocaleID = {};
     }
 
     /* Copy POSIX-style variant, if any [mr@FOO] */
     if (!OPTION_SET(options, _ULOC_CANONICALIZE) &&
-        tmpLocaleID != nullptr && keywordAssign == nullptr) {
+        !tmpLocaleID.empty() && keywordAssign == std::string_view::npos) {
-        for (;;) {
+        tag.append(tmpLocaleID, err);
-            char c = *tmpLocaleID;
+        tmpLocaleID = {};
-            if (c == 0) {
-                break;
-            }
-            tag.append(c, err);
-            ++tmpLocaleID;
-        }
     }
 
     if (OPTION_SET(options, _ULOC_CANONICALIZE)) {
         /* Handle @FOO variant if @ is present and not followed by = */
-        if (tmpLocaleID!=nullptr && keywordAssign==nullptr) {
+        if (!tmpLocaleID.empty() && keywordAssign == std::string_view::npos) {
             /* Add missing '_' if needed */
             if (fieldCount < 2 || (fieldCount < 3 && !script.isEmpty())) {
                 do {
@@ -1895,7 +1930,9 @@ _canonicalize(const char* localeID,
             }
 
             CharStringByteSink s(&tag);
-            _getVariant(tmpLocaleID+1, '@', &s, nullptr, !variant.isEmpty(), err);
+            std::string_view sub = tmpLocaleID;
+            sub.remove_prefix(1);
+            _getVariant(sub, '@', &s, !variant.isEmpty(), err);
             if (U_FAILURE(err)) { return; }
         }
 
@@ -1903,7 +1940,7 @@ _canonicalize(const char* localeID,
         for (j=0; j<UPRV_LENGTHOF(CANONICALIZE_MAP); j++) {
             StringPiece id(CANONICALIZE_MAP[j].id);
             if (tag == id) {
-                if (id.empty() && tmpLocaleID != nullptr) {
+                if (id.empty() && !tmpLocaleID.empty()) {
                     break; /* Don't remap "" if keywords present */
                 }
                 tag.clear();
@@ -1916,11 +1953,12 @@ _canonicalize(const char* localeID,
     sink.Append(tag.data(), tag.length());
 
     if (!OPTION_SET(options, _ULOC_STRIP_KEYWORDS)) {
-        if (tmpLocaleID!=nullptr && keywordAssign!=nullptr &&
+        if (!tmpLocaleID.empty() && keywordAssign != std::string_view::npos &&
-            (!separatorIndicator || separatorIndicator > keywordAssign)) {
+            (separatorIndicator == std::string_view::npos || separatorIndicator > keywordAssign)) {
             sink.Append("@", 1);
             ++fieldCount;
-            ulocimp_getKeywords(tmpLocaleID+1, '@', sink, true, err);
+            tmpLocaleID.remove_prefix(1);
+            ulocimp_getKeywords(tmpLocaleID, '@', sink, true, err);
         }
     }
 }
@@ -1989,6 +2027,10 @@ uloc_getLanguage(const char*    localeID,
          int32_t languageCapacity,
          UErrorCode* err)
 {
+    if (localeID == nullptr) {
+        localeID = uloc_getDefault();
+    }
+
     /* uloc_getLanguage will return a 2 character iso-639 code if one exists. *CWB*/
     return ByteSinkUtil::viaByteSinkToTerminatedChars(
         language, languageCapacity,
@@ -2011,6 +2053,10 @@ uloc_getScript(const char*    localeID,
          int32_t scriptCapacity,
          UErrorCode* err)
 {
+    if (localeID == nullptr) {
+        localeID = uloc_getDefault();
+    }
+
     return ByteSinkUtil::viaByteSinkToTerminatedChars(
         script, scriptCapacity,
         [&](ByteSink& sink, UErrorCode& status) {
@@ -2032,6 +2078,10 @@ uloc_getCountry(const char* localeID,
             int32_t countryCapacity,
             UErrorCode* err)
 {
+    if (localeID == nullptr) {
+        localeID = uloc_getDefault();
+    }
+
     return ByteSinkUtil::viaByteSinkToTerminatedChars(
         country, countryCapacity,
         [&](ByteSink& sink, UErrorCode& status) {
@@ -2053,6 +2103,10 @@ uloc_getVariant(const char* localeID,
                 int32_t variantCapacity,
                 UErrorCode* err)
 {
+    if (localeID == nullptr) {
+        localeID = uloc_getDefault();
+    }
+
     return ByteSinkUtil::viaByteSinkToTerminatedChars(
         variant, variantCapacity,
         [&](ByteSink& sink, UErrorCode& status) {
@@ -2074,6 +2128,9 @@ uloc_getName(const char* localeID,
              int32_t nameCapacity,
              UErrorCode* err)
 {
+    if (localeID == nullptr) {
+        localeID = uloc_getDefault();
+    }
     return ByteSinkUtil::viaByteSinkToTerminatedChars(
         name, nameCapacity,
         [&](ByteSink& sink, UErrorCode& status) {
@@ -2083,7 +2140,7 @@ uloc_getName(const char* localeID,
 }
 
 U_EXPORT CharString
-ulocimp_getName(const char* localeID,
+ulocimp_getName(std::string_view localeID,
                 UErrorCode& err)
 {
     return ByteSinkUtil::viaByteSinkToCharString(
@@ -2094,7 +2151,7 @@ ulocimp_getName(const char* localeID,
 }
 
 U_EXPORT void
-ulocimp_getName(const char* localeID,
+ulocimp_getName(std::string_view localeID,
                 ByteSink& sink,
                 UErrorCode& err)
 {
@@ -2107,6 +2164,9 @@ uloc_getBaseName(const char* localeID,
                  int32_t nameCapacity,
                  UErrorCode* err)
 {
+    if (localeID == nullptr) {
+        localeID = uloc_getDefault();
+    }
     return ByteSinkUtil::viaByteSinkToTerminatedChars(
         name, nameCapacity,
         [&](ByteSink& sink, UErrorCode& status) {
@@ -2116,7 +2176,7 @@ uloc_getBaseName(const char* localeID,
 }
 
 U_EXPORT CharString
-ulocimp_getBaseName(const char* localeID,
+ulocimp_getBaseName(std::string_view localeID,
                     UErrorCode& err)
 {
     return ByteSinkUtil::viaByteSinkToCharString(
@@ -2127,7 +2187,7 @@ ulocimp_getBaseName(const char* localeID,
 }
 
 U_EXPORT void
-ulocimp_getBaseName(const char* localeID,
+ulocimp_getBaseName(std::string_view localeID,
                     ByteSink& sink,
                     UErrorCode& err)
 {
@@ -2140,6 +2200,9 @@ uloc_canonicalize(const char* localeID,
                   int32_t nameCapacity,
                   UErrorCode* err)
 {
+    if (localeID == nullptr) {
+        localeID = uloc_getDefault();
+    }
     return ByteSinkUtil::viaByteSinkToTerminatedChars(
         name, nameCapacity,
         [&](ByteSink& sink, UErrorCode& status) {
@@ -2149,7 +2212,7 @@ uloc_canonicalize(const char* localeID,
 }
 
 U_EXPORT CharString
-ulocimp_canonicalize(const char* localeID,
+ulocimp_canonicalize(std::string_view localeID,
                      UErrorCode& err)
 {
     return ByteSinkUtil::viaByteSinkToCharString(
@@ -2160,7 +2223,7 @@ ulocimp_canonicalize(const char* localeID,
 }
 
 U_EXPORT void
-ulocimp_canonicalize(const char* localeID,
+ulocimp_canonicalize(std::string_view localeID,
                      ByteSink& sink,
                      UErrorCode& err)
 {

deps/icu-small/source/common/uloc_tag.cpp

@@ -1043,7 +1043,7 @@ _initializeULanguageTag(ULanguageTag* langtag) {
 }
 
 void
-_appendLanguageToLanguageTag(const char* localeID, icu::ByteSink& sink, bool strict, UErrorCode& status) {
+_appendLanguageToLanguageTag(std::string_view localeID, icu::ByteSink& sink, bool strict, UErrorCode& status) {
     UErrorCode tmpStatus = U_ZERO_ERROR;
 
     if (U_FAILURE(status)) {
@@ -1088,7 +1088,7 @@ _appendLanguageToLanguageTag(const char* localeID, icu::ByteSink& sink, bool str
 }
 
 void
-_appendScriptToLanguageTag(const char* localeID, icu::ByteSink& sink, bool strict, UErrorCode& status) {
+_appendScriptToLanguageTag(std::string_view localeID, icu::ByteSink& sink, bool strict, UErrorCode& status) {
     UErrorCode tmpStatus = U_ZERO_ERROR;
 
     if (U_FAILURE(status)) {
@@ -1118,7 +1118,7 @@ _appendScriptToLanguageTag(const char* localeID, icu::ByteSink& sink, bool stric
 }
 
 void
-_appendRegionToLanguageTag(const char* localeID, icu::ByteSink& sink, bool strict, UErrorCode& status) {
+_appendRegionToLanguageTag(std::string_view localeID, icu::ByteSink& sink, bool strict, UErrorCode& status) {
     UErrorCode tmpStatus = U_ZERO_ERROR;
 
     if (U_FAILURE(status)) {
@@ -1169,7 +1169,7 @@ void _sortVariants(VariantListEntry* first) {
 }
 
 void
-_appendVariantsToLanguageTag(const char* localeID, icu::ByteSink& sink, bool strict, bool& hadPosix, UErrorCode& status) {
+_appendVariantsToLanguageTag(std::string_view localeID, icu::ByteSink& sink, bool strict, bool& hadPosix, UErrorCode& status) {
     if (U_FAILURE(status)) { return; }
 
     UErrorCode tmpStatus = U_ZERO_ERROR;
@@ -1872,7 +1872,7 @@ _appendKeywords(ULanguageTag* langtag, icu::ByteSink& sink, UErrorCode& status)
 }
 
 void
-_appendPrivateuseToLanguageTag(const char* localeID, icu::ByteSink& sink, bool strict, bool /*hadPosix*/, UErrorCode& status) {
+_appendPrivateuseToLanguageTag(std::string_view localeID, icu::ByteSink& sink, bool strict, bool /*hadPosix*/, UErrorCode& status) {
     if (U_FAILURE(status)) { return; }
 
     UErrorCode tmpStatus = U_ZERO_ERROR;
@@ -2596,6 +2596,9 @@ ulocimp_toLanguageTag(const char* localeID,
     bool hadPosix = false;
     const char* pKeywordStart;
 
+    if (localeID == nullptr) {
+        localeID = uloc_getDefault();
+    }
     /* Note: uloc_canonicalize returns "en_US_POSIX" for input locale ID "".  See #6835 */
     icu::CharString canonical = ulocimp_canonicalize(localeID, tmpStatus);
     if (U_FAILURE(tmpStatus)) {
@@ -2604,7 +2607,7 @@ ulocimp_toLanguageTag(const char* localeID,
     }
 
     /* For handling special case - private use only tag */
-    pKeywordStart = locale_getKeywordsStart(canonical.data());
+    pKeywordStart = locale_getKeywordsStart(canonical.toStringPiece());
     if (pKeywordStart == canonical.data()) {
         int kwdCnt = 0;
         bool done = false;
@@ -2642,12 +2645,12 @@ ulocimp_toLanguageTag(const char* localeID,
         }
     }
 
-    _appendLanguageToLanguageTag(canonical.data(), sink, strict, status);
+    _appendLanguageToLanguageTag(canonical.toStringPiece(), sink, strict, status);
-    _appendScriptToLanguageTag(canonical.data(), sink, strict, status);
+    _appendScriptToLanguageTag(canonical.toStringPiece(), sink, strict, status);
-    _appendRegionToLanguageTag(canonical.data(), sink, strict, status);
+    _appendRegionToLanguageTag(canonical.toStringPiece(), sink, strict, status);
-    _appendVariantsToLanguageTag(canonical.data(), sink, strict, hadPosix, status);
+    _appendVariantsToLanguageTag(canonical.toStringPiece(), sink, strict, hadPosix, status);
     _appendKeywordsToLanguageTag(canonical.data(), sink, strict, hadPosix, status);
-    _appendPrivateuseToLanguageTag(canonical.data(), sink, strict, hadPosix, status);
+    _appendPrivateuseToLanguageTag(canonical.toStringPiece(), sink, strict, hadPosix, status);
 }
 
 

deps/icu-small/source/common/ulocale.cpp

@@ -10,7 +10,6 @@
 #include "unicode/locid.h"
 
 #include "bytesinkutil.h"
-#include "charstr.h"
 #include "cmemory.h"
 
 U_NAMESPACE_USE
@@ -24,9 +23,7 @@ ulocale_openForLocaleID(const char* localeID, int32_t length, UErrorCode* err) {
     if (length < 0) {
         return EXTERNAL(icu::Locale::createFromName(localeID).clone());
     }
-    CharString str(localeID, length, *err);  // Make a NUL terminated copy.
+    return EXTERNAL(icu::Locale::createFromName(StringPiece{localeID, length}).clone());
-    if (U_FAILURE(*err)) { return nullptr; }
-    return EXTERNAL(icu::Locale::createFromName(str.data()).clone());
 }
 
 ULocale*

deps/icu-small/source/common/ulocimp.h

@@ -68,42 +68,42 @@ U_EXPORT std::optional<std::string_view>
 ulocimp_toLegacyTypeWithFallback(std::string_view keyword, std::string_view value);
 
 U_EXPORT icu::CharString
-ulocimp_getKeywords(const char* localeID,
+ulocimp_getKeywords(std::string_view localeID,
                     char prev,
                     bool valuesToo,
                     UErrorCode& status);
 
 U_EXPORT void
-ulocimp_getKeywords(const char* localeID,
+ulocimp_getKeywords(std::string_view localeID,
                     char prev,
                     icu::ByteSink& sink,
                     bool valuesToo,
                     UErrorCode& status);
 
 U_EXPORT icu::CharString
-ulocimp_getName(const char* localeID,
+ulocimp_getName(std::string_view localeID,
                 UErrorCode& err);
 
 U_EXPORT void
-ulocimp_getName(const char* localeID,
+ulocimp_getName(std::string_view localeID,
                 icu::ByteSink& sink,
                 UErrorCode& err);
 
 U_EXPORT icu::CharString
-ulocimp_getBaseName(const char* localeID,
+ulocimp_getBaseName(std::string_view localeID,
                     UErrorCode& err);
 
 U_EXPORT void
-ulocimp_getBaseName(const char* localeID,
+ulocimp_getBaseName(std::string_view localeID,
                     icu::ByteSink& sink,
                     UErrorCode& err);
 
 U_EXPORT icu::CharString
-ulocimp_canonicalize(const char* localeID,
+ulocimp_canonicalize(std::string_view localeID,
                      UErrorCode& err);
 
 U_EXPORT void
-ulocimp_canonicalize(const char* localeID,
+ulocimp_canonicalize(std::string_view localeID,
                      icu::ByteSink& sink,
                      UErrorCode& err);
 
@@ -119,16 +119,16 @@ ulocimp_getKeywordValue(const char* localeID,
                         UErrorCode& status);
 
 U_EXPORT icu::CharString
-ulocimp_getLanguage(const char* localeID, UErrorCode& status);
+ulocimp_getLanguage(std::string_view localeID, UErrorCode& status);
 
 U_EXPORT icu::CharString
-ulocimp_getScript(const char* localeID, UErrorCode& status);
+ulocimp_getScript(std::string_view localeID, UErrorCode& status);
 
 U_EXPORT icu::CharString
-ulocimp_getRegion(const char* localeID, UErrorCode& status);
+ulocimp_getRegion(std::string_view localeID, UErrorCode& status);
 
 U_EXPORT icu::CharString
-ulocimp_getVariant(const char* localeID, UErrorCode& status);
+ulocimp_getVariant(std::string_view localeID, UErrorCode& status);
 
 U_EXPORT void
 ulocimp_setKeywordValue(std::string_view keywordName,
@@ -145,7 +145,7 @@ ulocimp_setKeywordValue(std::string_view keywords,
 
 U_EXPORT void
 ulocimp_getSubtags(
-        const char* localeID,
+        std::string_view localeID,
         icu::CharString* language,
         icu::CharString* script,
         icu::CharString* region,
@@ -155,7 +155,7 @@ ulocimp_getSubtags(
 
 U_EXPORT void
 ulocimp_getSubtags(
-        const char* localeID,
+        std::string_view localeID,
         icu::ByteSink* language,
         icu::ByteSink* script,
         icu::ByteSink* region,
@@ -165,7 +165,7 @@ ulocimp_getSubtags(
 
 inline void
 ulocimp_getSubtags(
-        const char* localeID,
+        std::string_view localeID,
         std::nullptr_t,
         std::nullptr_t,
         std::nullptr_t,
@@ -364,7 +364,7 @@ ulocimp_minimizeSubtags(const char* localeID,
                         UErrorCode& err);
 
 U_CAPI const char * U_EXPORT2
-locale_getKeywordsStart(const char *localeID);
+locale_getKeywordsStart(std::string_view localeID);
 
 bool
 ultag_isExtensionSubtags(const char* s, int32_t len);

deps/icu-small/source/common/umapfile.cpp

@@ -237,8 +237,13 @@ typedef HANDLE MemoryMap;
         pData->map = (char *)data + length;
         pData->pHeader=(const DataHeader *)data;
         pData->mapAddr = data;
-#if U_PLATFORM == U_PF_IPHONE
+#if U_PLATFORM == U_PF_IPHONE || U_PLATFORM == U_PF_ANDROID
+    // Apparently supported from Android 23 and higher:
+    //   https://github.com/ggml-org/llama.cpp/pull/3631
+    // Checking for the flag itself is safer than checking for __ANDROID_API__.
+#   ifdef POSIX_MADV_RANDOM
         posix_madvise(data, length, POSIX_MADV_RANDOM);
+#   endif
 #endif
         return true;
     }

deps/icu-small/source/common/unicode/brkiter.h

@@ -58,6 +58,8 @@ U_NAMESPACE_END
 
 U_NAMESPACE_BEGIN
 
+class CharString;
+
 /**
  * The BreakIterator class implements methods for finding the location
  * of boundaries in text. BreakIterator is an abstract base class.
@@ -646,9 +648,9 @@ protected:
 private:
 
     /** @internal (private) */
-    char actualLocale[ULOC_FULLNAME_CAPACITY];
+    CharString* actualLocale = nullptr;
-    char validLocale[ULOC_FULLNAME_CAPACITY];
+    CharString* validLocale = nullptr;
-    char requestLocale[ULOC_FULLNAME_CAPACITY];
+    CharString* requestLocale = nullptr;
 };
 
 #ifndef U_HIDE_DEPRECATED_API

deps/icu-small/source/common/unicode/char16ptr.h

@@ -9,10 +9,13 @@
 
 #include "unicode/utypes.h"
 
-#if U_SHOW_CPLUSPLUS_API
+#if U_SHOW_CPLUSPLUS_API || U_SHOW_CPLUSPLUS_HEADER_API
 
 #include <cstddef>
 #include <string_view>
+#include <type_traits>
+
+#endif
 
 /**
  * \file
@@ -21,8 +24,6 @@
  *        Also conversion functions from char16_t * to UChar * and OldUChar *.
  */
 
-U_NAMESPACE_BEGIN
-
 /**
  * \def U_ALIASING_BARRIER
  * Barrier for pointer anti-aliasing optimizations even across function boundaries.
@@ -36,6 +37,11 @@ U_NAMESPACE_BEGIN
 #   define U_ALIASING_BARRIER(ptr)
 #endif
 
+// ICU DLL-exported
+#if U_SHOW_CPLUSPLUS_API
+
+U_NAMESPACE_BEGIN
+
 /**
  * char16_t * wrapper with implicit conversion from distinct but bit-compatible pointer types.
  * @stable ICU 59
@@ -251,6 +257,60 @@ const char16_t *ConstChar16Ptr::get() const { return u_.cp; }
 #endif
 /// \endcond
 
+U_NAMESPACE_END
+
+#endif  // U_SHOW_CPLUSPLUS_API
+
+// Usable in header-only definitions
+#if U_SHOW_CPLUSPLUS_API || U_SHOW_CPLUSPLUS_HEADER_API
+
+namespace U_ICU_NAMESPACE_OR_INTERNAL {
+
+#ifndef U_FORCE_HIDE_INTERNAL_API
+/** @internal */
+template<typename T, typename = std::enable_if_t<std::is_same_v<T, UChar>>>
+inline const char16_t *uprv_char16PtrFromUChar(const T *p) {
+    if constexpr (std::is_same_v<UChar, char16_t>) {
+        return p;
+    } else {
+#if U_SHOW_CPLUSPLUS_API
+        return ConstChar16Ptr(p).get();
+#else
+#ifdef U_ALIASING_BARRIER
+        U_ALIASING_BARRIER(p);
+#endif
+        return reinterpret_cast<const char16_t *>(p);
+#endif
+    }
+}
+#if !U_CHAR16_IS_TYPEDEF && (!defined(_LIBCPP_VERSION) || _LIBCPP_VERSION < 180000)
+/** @internal */
+inline const char16_t *uprv_char16PtrFromUint16(const uint16_t *p) {
+#if U_SHOW_CPLUSPLUS_API
+    return ConstChar16Ptr(p).get();
+#else
+#ifdef U_ALIASING_BARRIER
+    U_ALIASING_BARRIER(p);
+#endif
+    return reinterpret_cast<const char16_t *>(p);
+#endif
+}
+#endif
+#if U_SIZEOF_WCHAR_T==2
+/** @internal */
+inline const char16_t *uprv_char16PtrFromWchar(const wchar_t *p) {
+#if U_SHOW_CPLUSPLUS_API
+    return ConstChar16Ptr(p).get();
+#else
+#ifdef U_ALIASING_BARRIER
+    U_ALIASING_BARRIER(p);
+#endif
+    return reinterpret_cast<const char16_t *>(p);
+#endif
+}
+#endif
+#endif
+
 /**
  * Converts from const char16_t * to const UChar *.
  * Includes an aliasing barrier if available.
@@ -307,6 +367,15 @@ inline OldUChar *toOldUCharPtr(char16_t *p) {
     return reinterpret_cast<OldUChar *>(p);
 }
 
+}  // U_ICU_NAMESPACE_OR_INTERNAL
+
+#endif  // U_SHOW_CPLUSPLUS_API || U_SHOW_CPLUSPLUS_HEADER_API
+
+// ICU DLL-exported
+#if U_SHOW_CPLUSPLUS_API
+
+U_NAMESPACE_BEGIN
+
 #ifndef U_FORCE_HIDE_INTERNAL_API
 /**
  * Is T convertible to a std::u16string_view or some other 16-bit string view?
@@ -379,6 +448,6 @@ inline std::u16string_view toU16StringViewNullable(const T& text) {
 
 U_NAMESPACE_END
 
-#endif /* U_SHOW_CPLUSPLUS_API */
+#endif  // U_SHOW_CPLUSPLUS_API
 
 #endif  // __CHAR16PTR_H__

deps/icu-small/source/common/unicode/locid.h

@@ -449,6 +449,11 @@ public:
      */
     static Locale U_EXPORT2 createFromName(const char *name);
 
+#ifndef U_HIDE_INTERNAL_API
+    /** @internal */
+    static Locale U_EXPORT2 createFromName(StringPiece name);
+#endif  /* U_HIDE_INTERNAL_API */
+
     /**
      * Creates a locale from the given string after canonicalizing
      * the string according to CLDR by calling uloc_canonicalize().
@@ -1133,7 +1138,9 @@ private:
      * @param cLocaleID The new locale name.
      * @param canonicalize whether to call uloc_canonicalize on cLocaleID
      */
-    Locale& init(const char* cLocaleID, UBool canonicalize);
+    Locale& init(const char* localeID, UBool canonicalize);
+    /** @internal */
+    Locale& init(StringPiece localeID, UBool canonicalize);
 
     /*
      * Internal constructor to allow construction of a locale object with

deps/icu-small/source/common/unicode/resbund.h

@@ -450,7 +450,7 @@ public:
      * @return a Locale object
      * @stable ICU 2.8
      */
-    const Locale
+    Locale
       getLocale(ULocDataLocaleType type, UErrorCode &status) const;
 #ifndef U_HIDE_INTERNAL_API
     /**

deps/icu-small/source/common/unicode/uchar.h

@@ -675,14 +675,14 @@ typedef enum UProperty {
      * @stable ICU 63
      */
     UCHAR_VERTICAL_ORIENTATION=0x1018,
-#ifndef U_HIDE_DRAFT_API
     /**
      * Enumerated property Identifier_Status.
      * Used for UTS #39 General Security Profile for Identifiers
      * (https://www.unicode.org/reports/tr39/#General_Security_Profile).
-     * @draft ICU 75
+     * @stable ICU 75
      */
     UCHAR_IDENTIFIER_STATUS=0x1019,
+#ifndef U_HIDE_DRAFT_API
     /**
      * Enumerated property Indic_Conjunct_Break.
      * Used in the grapheme cluster break algorithm in UAX #29.
@@ -796,7 +796,6 @@ typedef enum UProperty {
     UCHAR_SCRIPT_EXTENSIONS=0x7000,
     /** First constant for Unicode properties with unusual value types. @stable ICU 4.6 */
     UCHAR_OTHER_PROPERTY_START=UCHAR_SCRIPT_EXTENSIONS,
-#ifndef U_HIDE_DRAFT_API
     /**
      * Miscellaneous property Identifier_Type.
      * Used for UTS #39 General Security Profile for Identifiers
@@ -808,10 +807,9 @@ typedef enum UProperty {
      *
      * @see u_hasIDType
      * @see u_getIDTypes
-     * @draft ICU 75
+     * @stable ICU 75
      */
     UCHAR_IDENTIFIER_TYPE=0x7001,
-#endif  // U_HIDE_DRAFT_API
 #ifndef U_HIDE_DEPRECATED_API
     /**
      * One more than the last constant for Unicode properties with unusual value types.
@@ -2791,13 +2789,12 @@ typedef enum UVerticalOrientation {
     U_VO_UPRIGHT,
 } UVerticalOrientation;
 
-#ifndef U_HIDE_DRAFT_API
 /**
  * Identifier Status constants.
  * See https://www.unicode.org/reports/tr39/#Identifier_Status_and_Type.
  *
  * @see UCHAR_IDENTIFIER_STATUS
- * @draft ICU 75
+ * @stable ICU 75
  */
 typedef enum UIdentifierStatus {
     /*
@@ -2806,9 +2803,9 @@ typedef enum UIdentifierStatus {
      *     U_ID_STATUS_<Unicode Identifier_Status value name>
      */
 
-    /** @draft ICU 75 */
+    /** @stable ICU 75 */
     U_ID_STATUS_RESTRICTED,
-    /** @draft ICU 75 */
+    /** @stable ICU 75 */
     U_ID_STATUS_ALLOWED,
 } UIdentifierStatus;
 
@@ -2817,7 +2814,7 @@ typedef enum UIdentifierStatus {
  * See https://www.unicode.org/reports/tr39/#Identifier_Status_and_Type.
  *
  * @see UCHAR_IDENTIFIER_TYPE
- * @draft ICU 75
+ * @stable ICU 75
  */
 typedef enum UIdentifierType {
     /*
@@ -2826,32 +2823,31 @@ typedef enum UIdentifierType {
      *     U_ID_TYPE_<Unicode Identifier_Type value name>
      */
 
-    /** @draft ICU 75 */
+    /** @stable ICU 75 */
     U_ID_TYPE_NOT_CHARACTER,
-    /** @draft ICU 75 */
+    /** @stable ICU 75 */
     U_ID_TYPE_DEPRECATED,
-    /** @draft ICU 75 */
+    /** @stable ICU 75 */
     U_ID_TYPE_DEFAULT_IGNORABLE,
-    /** @draft ICU 75 */
+    /** @stable ICU 75 */
     U_ID_TYPE_NOT_NFKC,
-    /** @draft ICU 75 */
+    /** @stable ICU 75 */
     U_ID_TYPE_NOT_XID,
-    /** @draft ICU 75 */
+    /** @stable ICU 75 */
     U_ID_TYPE_EXCLUSION,
-    /** @draft ICU 75 */
+    /** @stable ICU 75 */
     U_ID_TYPE_OBSOLETE,
-    /** @draft ICU 75 */
+    /** @stable ICU 75 */
     U_ID_TYPE_TECHNICAL,
-    /** @draft ICU 75 */
+    /** @stable ICU 75 */
     U_ID_TYPE_UNCOMMON_USE,
-    /** @draft ICU 75 */
+    /** @stable ICU 75 */
     U_ID_TYPE_LIMITED_USE,
-    /** @draft ICU 75 */
+    /** @stable ICU 75 */
     U_ID_TYPE_INCLUSION,
-    /** @draft ICU 75 */
+    /** @stable ICU 75 */
     U_ID_TYPE_RECOMMENDED,
 } UIdentifierType;
-#endif  // U_HIDE_DRAFT_API
 
 /**
  * Check a binary Unicode property for a code point.
@@ -4057,7 +4053,6 @@ u_isIDStart(UChar32 c);
 U_CAPI UBool U_EXPORT2
 u_isIDPart(UChar32 c);
 
-#ifndef U_HIDE_DRAFT_API
 /**
  * Does the set of Identifier_Type values code point c contain the given type?
  *
@@ -4069,7 +4064,7 @@ u_isIDPart(UChar32 c);
  * @param c code point
  * @param type Identifier_Type to check
  * @return true if type is in Identifier_Type(c)
- * @draft ICU 75
+ * @stable ICU 75
  */
 U_CAPI bool U_EXPORT2
 u_hasIDType(UChar32 c, UIdentifierType type);
@@ -4104,11 +4099,10 @@ u_hasIDType(UChar32 c, UIdentifierType type);
  *                   function chaining. (See User Guide for details.)
  * @return number of values in c's Identifier_Type,
  *         written to types unless U_BUFFER_OVERFLOW_ERROR indicates insufficient capacity
- * @draft ICU 75
+ * @stable ICU 75
  */
 U_CAPI int32_t U_EXPORT2
 u_getIDTypes(UChar32 c, UIdentifierType *types, int32_t capacity, UErrorCode *pErrorCode);
-#endif  // U_HIDE_DRAFT_API
 
 /**
  * Determines if the specified character should be regarded

deps/icu-small/source/common/unicode/uniset.h

@@ -1173,10 +1173,12 @@ public:
     inline U_HEADER_NESTED_NAMESPACE::USetStrings strings() const {
         return U_HEADER_NESTED_NAMESPACE::USetStrings(toUSet());
     }
+#endif  // U_HIDE_DRAFT_API
 
+#ifndef U_HIDE_DRAFT_API
     /**
      * Returns a C++ iterator for iterating over all of the elements of this set.
-     * Convenient all-in one iteration, but creates a UnicodeString for each
+     * Convenient all-in one iteration, but creates a std::u16string for each
      * code point or string.
      * (Similar to how Java UnicodeSet *is an* Iterable<String>.)
      *
@@ -1185,13 +1187,14 @@ public:
      * \code
      * UnicodeSet set(u"[abcçカ🚴{}{abc}{de}]", errorCode);
      * for (auto el : set) {
+     *     UnicodeString us(el);
      *     std::string u8;
-     *     printf("set.string length %ld \"%s\"\n", (long)el.length(), el.toUTF8String(u8).c_str());
+     *     printf("set.element length %ld \"%s\"\n", (long)us.length(), us.toUTF8String(u8).c_str());
      * }
      * \endcode
      *
      * @return an all-elements iterator.
-     * @draft ICU 76
+     * @draft ICU 77
      * @see end
      * @see codePoints
      * @see ranges
@@ -1203,7 +1206,7 @@ public:
 
     /**
      * @return an exclusive-end sentinel for iterating over all of the elements of this set.
-     * @draft ICU 76
+     * @draft ICU 77
      * @see begin
      * @see codePoints
      * @see ranges

deps/icu-small/source/common/unicode/uset.h

@@ -32,12 +32,13 @@
 #include "unicode/utypes.h"
 #include "unicode/uchar.h"
 
-#if U_SHOW_CPLUSPLUS_API
+#if U_SHOW_CPLUSPLUS_API || U_SHOW_CPLUSPLUS_HEADER_API
+#include <string>
 #include <string_view>
 #include "unicode/char16ptr.h"
 #include "unicode/localpointer.h"
-#include "unicode/unistr.h"
+#include "unicode/utf16.h"
-#endif   // U_SHOW_CPLUSPLUS_API
+#endif
 
 #ifndef USET_DEFINED
 
@@ -1392,8 +1393,8 @@ public:
 private:
     friend class USetCodePoints;
 
-    USetCodePointIterator(const USet *uset, int32_t rangeIndex, int32_t rangeCount)
+    USetCodePointIterator(const USet *pUset, int32_t nRangeIndex, int32_t nRangeCount)
-            : uset(uset), rangeIndex(rangeIndex), rangeCount(rangeCount),
+            : uset(pUset), rangeIndex(nRangeIndex), rangeCount(nRangeCount),
                 c(U_SENTINEL), end(U_SENTINEL) {
         // Fetch the first range.
         operator++();
@@ -1429,7 +1430,7 @@ public:
      * Constructs a C++ "range" object over the code points of the USet.
      * @draft ICU 76
      */
-    USetCodePoints(const USet *uset) : uset(uset), rangeCount(uset_getRangeCount(uset)) {}
+    USetCodePoints(const USet *pUset) : uset(pUset), rangeCount(uset_getRangeCount(pUset)) {}
 
     /** @draft ICU 76 */
     USetCodePoints(const USetCodePoints &other) = default;
@@ -1460,7 +1461,7 @@ struct CodePointRange {
     /** @draft ICU 76 */
     struct iterator {
         /** @draft ICU 76 */
-        iterator(UChar32 c) : c(c) {}
+        iterator(UChar32 aC) : c(aC) {}
 
         /** @draft ICU 76 */
         bool operator==(const iterator &other) const { return c == other.c; }
@@ -1573,8 +1574,8 @@ public:
 private:
     friend class USetRanges;
 
-    USetRangeIterator(const USet *uset, int32_t rangeIndex, int32_t rangeCount)
+    USetRangeIterator(const USet *pUset, int32_t nRangeIndex, int32_t nRangeCount)
-            : uset(uset), rangeIndex(rangeIndex), rangeCount(rangeCount) {}
+            : uset(pUset), rangeIndex(nRangeIndex), rangeCount(nRangeCount) {}
 
     const USet *uset;
     int32_t rangeIndex;
@@ -1610,7 +1611,7 @@ public:
      * Constructs a C++ "range" object over the code point ranges of the USet.
      * @draft ICU 76
      */
-    USetRanges(const USet *uset) : uset(uset), rangeCount(uset_getRangeCount(uset)) {}
+    USetRanges(const USet *pUset) : uset(pUset), rangeCount(uset_getRangeCount(pUset)) {}
 
     /** @draft ICU 76 */
     USetRanges(const USetRanges &other) = default;
@@ -1657,7 +1658,7 @@ public:
             int32_t length;
             const UChar *uchars = uset_getString(uset, index, &length);
             // assert uchars != nullptr;
-            return {ConstChar16Ptr(uchars), static_cast<uint32_t>(length)};
+            return {uprv_char16PtrFromUChar(uchars), static_cast<size_t>(length)};
         }
         return {};
     }
@@ -1684,8 +1685,8 @@ public:
 private:
     friend class USetStrings;
 
-    USetStringIterator(const USet *uset, int32_t index, int32_t count)
+    USetStringIterator(const USet *pUset, int32_t nIndex, int32_t nCount)
-            : uset(uset), index(index), count(count) {}
+            : uset(pUset), index(nIndex), count(nCount) {}
 
     const USet *uset;
     int32_t index;
@@ -1699,9 +1700,11 @@ private:
  * using U_HEADER_NESTED_NAMESPACE::USetStrings;
  * LocalUSetPointer uset(uset_openPattern(u"[abcçカ🚴{}{abc}{de}]", -1, &errorCode));
  * for (auto s : USetStrings(uset.getAlias())) {
- *     UnicodeString us(s);
+ *     int32_t len32 = s.length();
- *     std::string u8;
+ *     char utf8[200];
- *     printf("uset.string length %ld \"%s\"\n", (long)s.length(), us.toUTF8String(u8).c_str());
+ *     u_strToUTF8WithSub(utf8, int32_t{sizeof(utf8) - 1}, nullptr,
+ *                        s.data(), len32, 0xFFFD, nullptr, errorCode);
+ *     printf("uset.string length %ld \"%s\"\n", long{len32}, utf8);
  * }
  * \endcode
  *
@@ -1718,7 +1721,7 @@ public:
      * Constructs a C++ "range" object over the strings of the USet.
      * @draft ICU 76
      */
-    USetStrings(const USet *uset) : uset(uset), count(uset_getStringCount(uset)) {}
+    USetStrings(const USet *pUset) : uset(pUset), count(uset_getStringCount(pUset)) {}
 
     /** @draft ICU 76 */
     USetStrings(const USetStrings &other) = default;
@@ -1737,17 +1740,19 @@ private:
     const USet *uset;
     int32_t count;
 };
+#endif  // U_HIDE_DRAFT_API
 
+#ifndef U_HIDE_DRAFT_API
 /**
  * Iterator returned by USetElements.
- * @draft ICU 76
+ * @draft ICU 77
  */
 class USetElementIterator {
 public:
-    /** @draft ICU 76 */
+    /** @draft ICU 77 */
     USetElementIterator(const USetElementIterator &other) = default;
 
-    /** @draft ICU 76 */
+    /** @draft ICU 77 */
     bool operator==(const USetElementIterator &other) const {
         // No need to compare rangeCount & end given private constructor
         // and assuming we don't compare iterators across the set being modified.
@@ -1756,26 +1761,28 @@ public:
         return uset == other.uset && c == other.c && index == other.index;
     }
 
-    /** @draft ICU 76 */
+    /** @draft ICU 77 */
     bool operator!=(const USetElementIterator &other) const { return !operator==(other); }
 
-    /** @draft ICU 76 */
+    /** @draft ICU 77 */
-    UnicodeString operator*() const {
+    std::u16string operator*() const {
         if (c >= 0) {
-            return UnicodeString(c);
+            return c <= 0xffff ?
+                std::u16string({static_cast<char16_t>(c)}) :
+                std::u16string({U16_LEAD(c), U16_TRAIL(c)});
         } else if (index < totalCount) {
             int32_t length;
             const UChar *uchars = uset_getString(uset, index - rangeCount, &length);
             // assert uchars != nullptr;
-            return UnicodeString(uchars, length);
+            return {uprv_char16PtrFromUChar(uchars), static_cast<size_t>(length)};
         } else {
-            return UnicodeString();
+            return {};
         }
     }
 
     /**
      * Pre-increment.
-     * @draft ICU 76
+     * @draft ICU 77
      */
     USetElementIterator &operator++() {
         if (c < end) {
@@ -1800,7 +1807,7 @@ public:
 
     /**
      * Post-increment.
-     * @draft ICU 76
+     * @draft ICU 77
      */
     USetElementIterator operator++(int) {
         USetElementIterator result(*this);
@@ -1811,8 +1818,8 @@ public:
 private:
     friend class USetElements;
 
-    USetElementIterator(const USet *uset, int32_t index, int32_t rangeCount, int32_t totalCount)
+    USetElementIterator(const USet *pUset, int32_t nIndex, int32_t nRangeCount, int32_t nTotalCount)
-            : uset(uset), index(index), rangeCount(rangeCount), totalCount(totalCount),
+            : uset(pUset), index(nIndex), rangeCount(nRangeCount), totalCount(nTotalCount),
                 c(U_SENTINEL), end(U_SENTINEL) {
         if (index < rangeCount) {
             // Fetch the first range.
@@ -1840,7 +1847,7 @@ private:
 
 /**
  * A C++ "range" for iterating over all of the elements of a USet.
- * Convenient all-in one iteration, but creates a UnicodeString for each
+ * Convenient all-in one iteration, but creates a std::u16string for each
  * code point or string.
  *
  * Code points are returned first, then empty and multi-character strings.
@@ -1849,15 +1856,18 @@ private:
  * using U_HEADER_NESTED_NAMESPACE::USetElements;
  * LocalUSetPointer uset(uset_openPattern(u"[abcçカ🚴{}{abc}{de}]", -1, &errorCode));
  * for (auto el : USetElements(uset.getAlias())) {
- *     std::string u8;
+ *     int32_t len32 = el.length();
- *     printf("uset.string length %ld \"%s\"\n", (long)el.length(), el.toUTF8String(u8).c_str());
+ *     char utf8[200];
+ *     u_strToUTF8WithSub(utf8, int32_t{sizeof(utf8) - 1}, nullptr,
+ *                        el.data(), len32, 0xFFFD, nullptr, errorCode);
+ *     printf("uset.element length %ld \"%s\"\n", long{len32}, utf8);
  * }
  * \endcode
  *
  * C++ UnicodeSet has member functions for iteration, including begin() and end().
  *
  * @return an all-elements iterator.
- * @draft ICU 76
+ * @draft ICU 77
  * @see USetCodePoints
  * @see USetRanges
  * @see USetStrings
@@ -1866,21 +1876,21 @@ class USetElements {
 public:
     /**
      * Constructs a C++ "range" object over all of the elements of the USet.
-     * @draft ICU 76
+     * @draft ICU 77
      */
-    USetElements(const USet *uset)
+    USetElements(const USet *pUset)
-        : uset(uset), rangeCount(uset_getRangeCount(uset)),
+        : uset(pUset), rangeCount(uset_getRangeCount(pUset)),
-            stringCount(uset_getStringCount(uset)) {}
+            stringCount(uset_getStringCount(pUset)) {}
 
-    /** @draft ICU 76 */
+    /** @draft ICU 77 */
     USetElements(const USetElements &other) = default;
 
-    /** @draft ICU 76 */
+    /** @draft ICU 77 */
     USetElementIterator begin() const {
         return USetElementIterator(uset, 0, rangeCount, rangeCount + stringCount);
     }
 
-    /** @draft ICU 76 */
+    /** @draft ICU 77 */
     USetElementIterator end() const {
         return USetElementIterator(uset, rangeCount + stringCount, rangeCount, rangeCount + stringCount);
     }

deps/icu-small/source/common/unicode/utf8.h

@@ -124,7 +124,7 @@
  * @internal
  */
 U_CAPI UChar32 U_EXPORT2
-utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict);
+utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, int8_t strict);
 
 /**
  * Function for handling "append code point" with error-checking.
@@ -148,7 +148,7 @@ utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool
  * @internal
  */
 U_CAPI UChar32 U_EXPORT2
-utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict);
+utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, int8_t strict);
 
 /**
  * Function for handling "skip backward one code point" with error-checking.

deps/icu-small/source/common/unicode/utypes.h

@@ -598,12 +598,13 @@ typedef enum UErrorCode {
     U_MF_DUPLICATE_DECLARATION_ERROR, /**< The same variable is declared in more than one .local or .input declaration. @internal ICU 75 technology preview @deprecated This API is for technology preview only. */
     U_MF_OPERAND_MISMATCH_ERROR,     /**< An operand provided to a function does not have the required form for that function @internal ICU 75 technology preview @deprecated This API is for technology preview only. */
     U_MF_DUPLICATE_VARIANT_ERROR, /**< A message includes a variant with the same key list as another variant. @internal ICU 76 technology preview @deprecated This API is for technology preview only. */
+    U_MF_BAD_OPTION,             /**< An option value provided to a function does not have the required form for that option. @internal ICU 77 technology preview @deprecated This API is for technology preview only. */
 #ifndef U_HIDE_DEPRECATED_API
     /**
      * One more than the highest normal formatting API error code.
      * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
      */
-    U_FMT_PARSE_ERROR_LIMIT = 0x10120,
+    U_FMT_PARSE_ERROR_LIMIT = 0x10121,
 #endif  // U_HIDE_DEPRECATED_API
 
     /*

deps/icu-small/source/common/unicode/uvernum.h

@@ -53,7 +53,7 @@
  *  This value will change in the subsequent releases of ICU
  *  @stable ICU 2.4
  */
-#define U_ICU_VERSION_MAJOR_NUM 76
+#define U_ICU_VERSION_MAJOR_NUM 77
 
 /** The current ICU minor version as an integer.
  *  This value will change in the subsequent releases of ICU
@@ -79,7 +79,7 @@
  *  This value will change in the subsequent releases of ICU
  *  @stable ICU 2.6
  */
-#define U_ICU_VERSION_SUFFIX _76
+#define U_ICU_VERSION_SUFFIX _77
 
 /**
  * \def U_DEF2_ICU_ENTRY_POINT_RENAME
@@ -132,7 +132,7 @@
  *  This value will change in the subsequent releases of ICU
  *  @stable ICU 2.4
  */
-#define U_ICU_VERSION "76.1"
+#define U_ICU_VERSION "77.1"
 
 /**
  * The current ICU library major version number as a string, for library name suffixes.
@@ -145,13 +145,13 @@
  *
  * @stable ICU 2.6
  */
-#define U_ICU_VERSION_SHORT "76"
+#define U_ICU_VERSION_SHORT "77"
 
 #ifndef U_HIDE_INTERNAL_API
 /** Data version in ICU4C.
  * @internal ICU 4.4 Internal Use Only
  **/
-#define U_ICU_DATA_VERSION "76.1"
+#define U_ICU_DATA_VERSION "77.1"
 #endif  /* U_HIDE_INTERNAL_API */
 
 /*===========================================================================

deps/icu-small/source/common/unicode/uversion.h

@@ -125,7 +125,7 @@ typedef uint8_t UVersionInfo[U_MAX_VERSION_LENGTH];
         U_NAMESPACE_USE
 #   endif
 
-#ifndef U_HIDE_DRAFT_API
+#ifndef U_FORCE_HIDE_DRAFT_API
 /**
  * \def U_HEADER_NESTED_NAMESPACE
  * Nested namespace used inside U_ICU_NAMESPACE for header-only APIs.
@@ -150,22 +150,37 @@ typedef uint8_t UVersionInfo[U_MAX_VERSION_LENGTH];
  * @draft ICU 76
  */
 
+/**
+ * \def U_ICU_NAMESPACE_OR_INTERNAL
+ * Namespace used for header-only APIs that used to be regular C++ APIs.
+ * Different when used inside ICU to prevent public use of internal instantiations.
+ * Similar to U_HEADER_ONLY_NAMESPACE, but the public definition is the same as U_ICU_NAMESPACE.
+ * "U_ICU_NAMESPACE" or "U_ICU_NAMESPACE::internal".
+ *
+ * @draft ICU 77
+ */
+
 // The first test is the same as for defining U_EXPORT for Windows.
 #if defined(_MSC_VER) || (UPRV_HAS_DECLSPEC_ATTRIBUTE(__dllexport__) && \
                           UPRV_HAS_DECLSPEC_ATTRIBUTE(__dllimport__))
 #   define U_HEADER_NESTED_NAMESPACE header
+#   define U_ICU_NAMESPACE_OR_INTERNAL U_ICU_NAMESPACE
 #elif defined(U_COMBINED_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION) || \
         defined(U_I18N_IMPLEMENTATION) || defined(U_IO_IMPLEMENTATION) || \
         defined(U_LAYOUTEX_IMPLEMENTATION) || defined(U_TOOLUTIL_IMPLEMENTATION)
 #   define U_HEADER_NESTED_NAMESPACE internal
+#   define U_ICU_NAMESPACE_OR_INTERNAL U_ICU_NAMESPACE::internal
+    namespace U_ICU_NAMESPACE_OR_INTERNAL {}
+    using namespace U_ICU_NAMESPACE_OR_INTERNAL;
 #else
 #   define U_HEADER_NESTED_NAMESPACE header
+#   define U_ICU_NAMESPACE_OR_INTERNAL U_ICU_NAMESPACE
 #endif
 
 #define U_HEADER_ONLY_NAMESPACE U_ICU_NAMESPACE::U_HEADER_NESTED_NAMESPACE
 
 namespace U_HEADER_ONLY_NAMESPACE {}
-#endif  // U_HIDE_DRAFT_API
+#endif  // U_FORCE_HIDE_DRAFT_API
 
 #endif /* __cplusplus */
 

deps/icu-small/source/common/unistr.cpp

@@ -1945,6 +1945,13 @@ UnicodeString::cloneArrayIfNeeded(int32_t newCapacity,
       growCapacity = newCapacity;
     } else if(newCapacity <= US_STACKBUF_SIZE && growCapacity > US_STACKBUF_SIZE) {
       growCapacity = US_STACKBUF_SIZE;
+    } else if(newCapacity > growCapacity) {
+      setToBogus();
+      return false;  // bad inputs
+    }
+    if(growCapacity > kMaxCapacity) {
+      setToBogus();
+      return false;
     }
 
     // save old values

deps/icu-small/source/common/uresbund.cpp

@@ -2716,6 +2716,9 @@ ures_openWithType(UResourceBundle *r, const char* path, const char* localeID,
 
     UResourceDataEntry *entry;
     if(openType != URES_OPEN_DIRECT) {
+        if (localeID == nullptr) {
+            localeID = uloc_getDefault();
+        }
         /* first "canonicalize" the locale ID */
         CharString canonLocaleID = ulocimp_getBaseName(localeID, *status);
         if(U_FAILURE(*status)) {
@@ -3080,6 +3083,9 @@ ures_getFunctionalEquivalent(char *result, int32_t resultCapacity,
             kwVal.clear();
         }
     }
+    if (locid == nullptr) {
+        locid = uloc_getDefault();
+    }
     CharString base = ulocimp_getBaseName(locid, subStatus);
 #if defined(URES_TREE_DEBUG)
     fprintf(stderr, "getFunctionalEquivalent: \"%s\" [%s=%s] in %s - %s\n", 
@@ -3244,7 +3250,7 @@ ures_getFunctionalEquivalent(char *result, int32_t resultCapacity,
             const char *validLoc = ures_getLocaleByType(res, ULOC_VALID_LOCALE, &subStatus);
             if (U_SUCCESS(subStatus) && validLoc != nullptr && validLoc[0] != 0 && uprv_strcmp(validLoc, "root") != 0) {
                 CharString validLang = ulocimp_getLanguage(validLoc, subStatus);
-                CharString parentLang = ulocimp_getLanguage(parent.data(), subStatus);
+                CharString parentLang = ulocimp_getLanguage(parent.toStringPiece(), subStatus);
                 if (U_SUCCESS(subStatus) && validLang != parentLang) {
                     // validLoc is not root and has a different language than parent, use it instead
                     found.clear().append(validLoc, subStatus);

deps/icu-small/source/common/uscript.cpp

@@ -59,6 +59,9 @@ getCodesFromLocale(const char *locale,
     if (U_FAILURE(*err)) { return 0; }
     icu::CharString lang;
     icu::CharString script;
+    if (locale == nullptr) {
+        locale = uloc_getDefault();
+    }
     ulocimp_getSubtags(locale, &lang, &script, nullptr, nullptr, nullptr, *err);
     if (U_FAILURE(*err)) { return 0; }
     // Multi-script languages, equivalent to the LocaleScript data

deps/icu-small/source/common/ushape.cpp

@@ -28,6 +28,7 @@
 #include "ubidi_props.h"
 #include "uassert.h"
 
+#include <limits>
 /*
  * This implementation is designed for 16-bit Unicode strings.
  * The main assumption is that the Arabic characters and their
@@ -747,6 +748,10 @@ handleGeneratedSpaces(char16_t *dest, int32_t sourceLength,
         }
     }
 
+    if (static_cast<size_t>(sourceLength) + 1 > std::numeric_limits<size_t>::max() / U_SIZEOF_UCHAR) {
+        *pErrorCode = U_INDEX_OUTOFBOUNDS_ERROR;
+        return 0;
+    }
     tempbuffer = static_cast<char16_t*>(uprv_malloc((sourceLength + 1) * U_SIZEOF_UCHAR));
     /* Test for nullptr */
     if(tempbuffer == nullptr) {

deps/icu-small/source/common/usprep.cpp

@@ -126,7 +126,7 @@ compareEntries(const UHashTok p1, const UHashTok p2) {
     name2.pointer = b2->name;
     path1.pointer = b1->path;
     path2.pointer = b2->path;
-    return uhash_compareChars(name1, name2) & uhash_compareChars(path1, path2);
+    return uhash_compareChars(name1, name2) && uhash_compareChars(path1, path2);
 }
 
 static void 

deps/icu-small/source/common/utf_impl.cpp

@@ -124,11 +124,9 @@ errorValue(int32_t count, int8_t strict) {
  * >0  Obsolete "strict" behavior of UTF8_NEXT_CHAR_SAFE(..., true):
  *     Same as the obsolete "safe" behavior, but non-characters are also treated
  *     like illegal sequences.
- *
- * Note that a UBool is the same as an int8_t.
  */
 U_CAPI UChar32 U_EXPORT2
-utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, UBool strict) {
+utf8_nextCharSafeBody(const uint8_t *s, int32_t *pi, int32_t length, UChar32 c, int8_t strict) {
     // *pi is one after byte c.
     int32_t i=*pi;
     // length can be negative for NUL-terminated strings: Read and validate one byte at a time.
@@ -233,7 +231,7 @@ utf8_appendCharSafeBody(uint8_t *s, int32_t i, int32_t length, UChar32 c, UBool
 }
 
 U_CAPI UChar32 U_EXPORT2
-utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, UBool strict) {
+utf8_prevCharSafeBody(const uint8_t *s, int32_t start, int32_t *pi, UChar32 c, int8_t strict) {
     // *pi is the index of byte c.
     int32_t i=*pi;
     if(U8_IS_TRAIL(c) && i>start) {

deps/icu-small/source/common/utypes.cpp

@@ -140,7 +140,8 @@ _uFmtErrorName[U_FMT_PARSE_ERROR_LIMIT - U_FMT_PARSE_ERROR_START] = {
     "U_MF_MISSING_SELECTOR_ANNOTATION_ERROR",
     "U_MF_DUPLICATE_DECLARATION_ERROR",
     "U_MF_OPERAND_MISMATCH_ERROR",
-    "U_MF_DUPLICATE_VARIANT_ERROR"
+    "U_MF_DUPLICATE_VARIANT_ERROR",
+    "U_MF_BAD_OPTION"
 };
 
 static const char * const

deps/icu-small/source/i18n/basictz.cpp

@@ -160,12 +160,13 @@ BasicTimeZone::getSimpleRulesNear(UDate date, InitialTimeZoneRule*& initial,
               || (tr.getFrom()->getDSTSavings() != 0 && tr.getTo()->getDSTSavings() == 0))
             && (date + MILLIS_PER_YEAR > nextTransitionTime)) {
  
-            int32_t year, month, dom, dow, doy, mid;
+            int32_t year, mid;
+            int8_t month, dom, dow;
             UDate d;
 
             // Get local wall time for the next transition time
             Grego::timeToFields(nextTransitionTime + initialRaw + initialDst,
-                year, month, dom, dow, doy, mid, status);
+                year, month, dom, dow, mid, status);
             if (U_FAILURE(status)) return;
             int32_t weekInMonth = Grego::dayOfWeekInMonth(year, month, dom);
             // Create DOW rule
@@ -193,7 +194,7 @@ BasicTimeZone::getSimpleRulesNear(UDate date, InitialTimeZoneRule*& initial,
 
                         // Get local wall time for the next transition time
                         Grego::timeToFields(tr.getTime() + tr.getFrom()->getRawOffset() + tr.getFrom()->getDSTSavings(),
-                            year, month, dom, dow, doy, mid, status);
+                            year, month, dom, dow, mid, status);
                         if (U_FAILURE(status)) return;
                         weekInMonth = Grego::dayOfWeekInMonth(year, month, dom);
                         // Generate another DOW rule
@@ -225,7 +226,7 @@ BasicTimeZone::getSimpleRulesNear(UDate date, InitialTimeZoneRule*& initial,
 
                         // Generate another DOW rule
                         Grego::timeToFields(tr.getTime() + tr.getFrom()->getRawOffset() + tr.getFrom()->getDSTSavings(),
-                            year, month, dom, dow, doy, mid, status);
+                            year, month, dom, dow, mid, status);
                         if (U_FAILURE(status)) return;
                         weekInMonth = Grego::dayOfWeekInMonth(year, month, dom);
                         dtr = new DateTimeRule(month, weekInMonth, dow, mid, DateTimeRule::WALL_TIME);
@@ -486,8 +487,7 @@ BasicTimeZone::getTimeZoneRulesAfter(UDate start, InitialTimeZoneRule*& initial,
                 }
             } else {
                 // Calculate the transition year
-                int32_t year, month, dom, dow, doy, mid;
+                int32_t year = Grego::timeToYear(tzt.getTime(), status);
-                Grego::timeToFields(tzt.getTime(), year, month, dom, dow, doy, mid, status);
                 if (U_FAILURE(status)) {
                     return;
                 }

deps/icu-small/source/i18n/buddhcal.cpp

@@ -36,7 +36,6 @@ static const int32_t kGregorianEpoch = 1970;    // used as the default value of
 BuddhistCalendar::BuddhistCalendar(const Locale& aLocale, UErrorCode& success)
 :   GregorianCalendar(aLocale, success)
 {
-    setTimeInMillis(getNow(), success); // Call this again now that the vtable is set up properly.
 }
 
 BuddhistCalendar::~BuddhistCalendar()
@@ -48,12 +47,6 @@ BuddhistCalendar::BuddhistCalendar(const BuddhistCalendar& source)
 {
 }
 
-BuddhistCalendar& BuddhistCalendar::operator= ( const BuddhistCalendar& right)
-{
-    GregorianCalendar::operator=(right);
-    return *this;
-}
-
 BuddhistCalendar* BuddhistCalendar::clone() const
 {
     return new BuddhistCalendar(*this);

deps/icu-small/source/i18n/buddhcal.h

@@ -82,13 +82,6 @@ public:
      */
     BuddhistCalendar(const BuddhistCalendar& source);
 
-    /**
-     * Default assignment operator
-     * @param right    the object to be copied.
-     * @internal
-     */
-    BuddhistCalendar& operator=(const BuddhistCalendar& right);
-
     /**
      * Create and return a polymorphic copy of this calendar.
      * @return    return a polymorphic copy of this calendar.

deps/icu-small/source/i18n/calendar.cpp

@@ -156,7 +156,7 @@ U_CFUNC void ucal_dump(UCalendar* cal) {
 #endif
 
 /* Max value for stamp allowable before recalculation */
-#define STAMP_MAX 10000
+#define STAMP_MAX 127
 
 static const char * const gCalTypes[] = {
     "gregorian",
@@ -700,15 +700,10 @@ fIsTimeSet(false),
 fAreFieldsSet(false),
 fAreAllFieldsSet(false),
 fAreFieldsVirtuallySet(false),
-fNextStamp(static_cast<int32_t>(kMinimumUserStamp)),
-fTime(0),
 fLenient(true),
-fZone(nullptr),
 fRepeatedWallTime(UCAL_WALLTIME_LAST),
 fSkippedWallTime(UCAL_WALLTIME_LAST)
 {
-    validLocale[0] = 0;
-    actualLocale[0] = 0;
     clear();
     if (U_FAILURE(success)) {
         return;
@@ -722,26 +717,21 @@ fSkippedWallTime(UCAL_WALLTIME_LAST)
 
 // -------------------------------------
 
-Calendar::Calendar(TimeZone* zone, const Locale& aLocale, UErrorCode& success)
+Calendar::Calendar(TimeZone* adoptZone, const Locale& aLocale, UErrorCode& success)
 :   UObject(),
 fIsTimeSet(false),
 fAreFieldsSet(false),
 fAreAllFieldsSet(false),
 fAreFieldsVirtuallySet(false),
-fNextStamp(static_cast<int32_t>(kMinimumUserStamp)),
-fTime(0),
 fLenient(true),
-fZone(nullptr),
 fRepeatedWallTime(UCAL_WALLTIME_LAST),
 fSkippedWallTime(UCAL_WALLTIME_LAST)
 {
-    validLocale[0] = 0;
+    LocalPointer<TimeZone> zone(adoptZone, success);
-    actualLocale[0] = 0;
     if (U_FAILURE(success)) {
-        delete zone;
         return;
     }
-    if (zone == nullptr) {
+    if (zone.isNull()) {
 #if defined (U_DEBUG_CAL)
         fprintf(stderr, "%s:%d: ILLEGAL ARG because timezone cannot be 0\n",
             __FILE__, __LINE__);
@@ -751,7 +741,7 @@ fSkippedWallTime(UCAL_WALLTIME_LAST)
     }
 
     clear();
-    fZone = zone;
+    fZone = zone.orphan();
     setWeekData(aLocale, nullptr, success);
 }
 
@@ -763,15 +753,10 @@ fIsTimeSet(false),
 fAreFieldsSet(false),
 fAreAllFieldsSet(false),
 fAreFieldsVirtuallySet(false),
-fNextStamp(static_cast<int32_t>(kMinimumUserStamp)),
-fTime(0),
 fLenient(true),
-fZone(nullptr),
 fRepeatedWallTime(UCAL_WALLTIME_LAST),
 fSkippedWallTime(UCAL_WALLTIME_LAST)
 {
-    validLocale[0] = 0;
-    actualLocale[0] = 0;
     if (U_FAILURE(success)) {
         return;
     }
@@ -779,6 +764,7 @@ fSkippedWallTime(UCAL_WALLTIME_LAST)
     fZone = zone.clone();
     if (fZone == nullptr) {
         success = U_MEMORY_ALLOCATION_ERROR;
+        return;
     }
     setWeekData(aLocale, nullptr, success);
 }
@@ -788,6 +774,8 @@ fSkippedWallTime(UCAL_WALLTIME_LAST)
 Calendar::~Calendar()
 {
     delete fZone;
+    delete actualLocale;
+    delete validLocale;
 }
 
 // -------------------------------------
@@ -795,7 +783,6 @@ Calendar::~Calendar()
 Calendar::Calendar(const Calendar &source)
 :   UObject(source)
 {
-    fZone = nullptr;
     *this = source;
 }
 
@@ -806,7 +793,6 @@ Calendar::operator=(const Calendar &right)
 {
     if (this != &right) {
         uprv_arrayCopy(right.fFields, fFields, UCAL_FIELD_COUNT);
-        uprv_arrayCopy(right.fIsSet, fIsSet, UCAL_FIELD_COUNT);
         uprv_arrayCopy(right.fStamp, fStamp, UCAL_FIELD_COUNT);
         fTime                    = right.fTime;
         fIsTimeSet               = right.fIsTimeSet;
@@ -828,10 +814,10 @@ Calendar::operator=(const Calendar &right)
         fWeekendCease            = right.fWeekendCease;
         fWeekendCeaseMillis      = right.fWeekendCeaseMillis;
         fNextStamp               = right.fNextStamp;
-        uprv_strncpy(validLocale, right.validLocale, sizeof(validLocale));
+        UErrorCode status = U_ZERO_ERROR;
-        uprv_strncpy(actualLocale, right.actualLocale, sizeof(actualLocale));
+        U_LOCALE_BASED(locBased, *this);
-        validLocale[sizeof(validLocale)-1] = 0;
+        locBased.setLocaleIDs(right.validLocale, right.actualLocale, status);
-        actualLocale[sizeof(validLocale)-1] = 0;
+        U_ASSERT(U_SUCCESS(status));
     }
 
     return *this;
@@ -1167,13 +1153,9 @@ Calendar::setTimeInMillis( double millis, UErrorCode& status ) {
     fAreFieldsSet = fAreAllFieldsSet = false;
     fIsTimeSet = fAreFieldsVirtuallySet = true;
 
-    for (int32_t i=0; i<UCAL_FIELD_COUNT; ++i) {
+    uprv_memset(fFields, 0, sizeof(fFields));
-        fFields[i]     = 0;
+    uprv_memset(fStamp, kUnset, sizeof(fStamp));
-        fStamp[i]     = kUnset;
+    fNextStamp = kMinimumUserStamp;
-        fIsSet[i]     = false;
-    }
-
-
 }
 
 // -------------------------------------
@@ -1208,12 +1190,11 @@ Calendar::set(UCalendarDateFields field, int32_t value)
         computeFields(ec);
     }
     fFields[field]     = value;
-    /* Ensure that the fNextStamp value doesn't go pass max value for int32_t */
+    /* Ensure that the fNextStamp value doesn't go pass max value for int8_t */
     if (fNextStamp == STAMP_MAX) {
         recalculateStamp();
     }
     fStamp[field]     = fNextStamp++;
-    fIsSet[field]     = true; // Remove later
     fIsTimeSet = fAreFieldsSet = fAreFieldsVirtuallySet = false;
 }
 
@@ -1270,11 +1251,9 @@ void Calendar::setRelatedYear(int32_t year)
 void
 Calendar::clear()
 {
-    for (int32_t i=0; i<UCAL_FIELD_COUNT; ++i) {
+    uprv_memset(fFields, 0, sizeof(fFields));
-        fFields[i]     = 0; // Must do this; other code depends on it
+    uprv_memset(fStamp, kUnset, sizeof(fStamp));
-        fStamp[i]     = kUnset;
+    fNextStamp = kMinimumUserStamp;
-        fIsSet[i]     = false; // Remove later
-    }
     fIsTimeSet = fAreFieldsSet = fAreAllFieldsSet = fAreFieldsVirtuallySet = false;
     // fTime is not 'cleared' - may be used if no fields are set.
 }
@@ -1296,12 +1275,10 @@ Calendar::clear(UCalendarDateFields field)
     if (field == UCAL_MONTH) {
         fFields[UCAL_ORDINAL_MONTH]         = 0;
         fStamp[UCAL_ORDINAL_MONTH]         = kUnset;
-        fIsSet[UCAL_ORDINAL_MONTH]         = false; // Remove later
     }
     if (field == UCAL_ORDINAL_MONTH) {
         fFields[UCAL_MONTH]         = 0;
         fStamp[UCAL_MONTH]         = kUnset;
-        fIsSet[UCAL_MONTH]         = false; // Remove later
     }
     fIsTimeSet = fAreFieldsSet = fAreAllFieldsSet = fAreFieldsVirtuallySet = false;
 }
@@ -1434,10 +1411,8 @@ void Calendar::computeFields(UErrorCode &ec)
     for (int32_t i=0; i<UCAL_FIELD_COUNT; ++i) {
         if ((mask & 1) == 0) {
             fStamp[i] = kInternallySet;
-            fIsSet[i] = true; // Remove later
         } else {
             fStamp[i] = kUnset;
-            fIsSet[i] = false; // Remove later
         }
         mask >>= 1;
     }
@@ -1467,7 +1442,7 @@ void Calendar::computeFields(UErrorCode &ec)
     //__FILE__, __LINE__, fFields[UCAL_JULIAN_DAY], localMillis);
 #endif
 
-    computeGregorianAndDOWFields(fFields[UCAL_JULIAN_DAY], ec);
+    computeGregorianFields(fFields[UCAL_JULIAN_DAY], ec);
 
     // Call framework method to have subclass compute its fields.
     // These must include, at a minimum, MONTH, DAY_OF_MONTH,
@@ -1538,32 +1513,6 @@ uint8_t Calendar::julianDayToDayOfWeek(int32_t julian)
     return result;
 }
 
-/**
-* Compute the Gregorian calendar year, month, and day of month from
-* the given Julian day.  These values are not stored in fields, but in
-* member variables gregorianXxx.  Also compute the DAY_OF_WEEK and
-* DOW_LOCAL fields.
-*/
-void Calendar::computeGregorianAndDOWFields(int32_t julianDay, UErrorCode &ec)
-{
-    computeGregorianFields(julianDay, ec);
-    if (U_FAILURE(ec)) {
-        return;
-    }
-
-    // Compute day of week: JD 0 = Monday
-    int32_t dow = julianDayToDayOfWeek(julianDay);
-    internalSet(UCAL_DAY_OF_WEEK,dow);
-
-    // Calculate 1-based localized day of week
-    int32_t dowLocal = dow - getFirstDayOfWeek() + 1;
-    if (dowLocal < 1) {
-        dowLocal += 7;
-    }
-    internalSet(UCAL_DOW_LOCAL,dowLocal);
-    fFields[UCAL_DOW_LOCAL] = dowLocal;
-}
-
 /**
 * Compute the Gregorian calendar year, month, and day of month from the
 * Julian day.  These values are not stored in fields, but in member
@@ -1575,14 +1524,13 @@ void Calendar::computeGregorianFields(int32_t julianDay, UErrorCode& ec) {
     if (U_FAILURE(ec)) {
         return;
     }
-    int32_t gregorianDayOfWeekUnused;
     if (uprv_add32_overflow(
             julianDay, -kEpochStartAsJulianDay, &julianDay)) {
         ec = U_ILLEGAL_ARGUMENT_ERROR;
         return;
     }
     Grego::dayToFields(julianDay, fGregorianYear, fGregorianMonth,
-                       fGregorianDayOfMonth, gregorianDayOfWeekUnused,
+                       fGregorianDayOfMonth,
                        fGregorianDayOfYear, ec);
 }
 
@@ -1610,8 +1558,19 @@ void Calendar::computeWeekFields(UErrorCode &ec) {
     if(U_FAILURE(ec)) {
         return;
     }
+
+    // Compute day of week: JD 0 = Monday
+    int32_t dayOfWeek = julianDayToDayOfWeek(fFields[UCAL_JULIAN_DAY]);
+    internalSet(UCAL_DAY_OF_WEEK, dayOfWeek);
+    int32_t firstDayOfWeek = getFirstDayOfWeek();
+    // Calculate 1-based localized day of week
+    int32_t dowLocal = dayOfWeek - firstDayOfWeek + 1;
+    if (dowLocal < 1) {
+        dowLocal += 7;
+    }
+    internalSet(UCAL_DOW_LOCAL,dowLocal);
+
     int32_t eyear = fFields[UCAL_EXTENDED_YEAR];
-    int32_t dayOfWeek = fFields[UCAL_DAY_OF_WEEK];
     int32_t dayOfYear = fFields[UCAL_DAY_OF_YEAR];
 
     // WEEK_OF_YEAR start
@@ -1624,10 +1583,11 @@ void Calendar::computeWeekFields(UErrorCode &ec) {
     // first week of the next year.  ASSUME that the year length is less than
     // 7000 days.
     int32_t yearOfWeekOfYear = eyear;
-    int32_t relDow = (dayOfWeek + 7 - getFirstDayOfWeek()) % 7; // 0..6
+    int32_t relDow = (dayOfWeek + 7 - firstDayOfWeek) % 7; // 0..6
-    int32_t relDowJan1 = (dayOfWeek - dayOfYear + 7001 - getFirstDayOfWeek()) % 7; // 0..6
+    int32_t relDowJan1 = (dayOfWeek - dayOfYear + 7001 - firstDayOfWeek) % 7; // 0..6
     int32_t woy = (dayOfYear - 1 + relDowJan1) / 7; // 0..53
-    if ((7 - relDowJan1) >= getMinimalDaysInFirstWeek()) {
+    int32_t minimalDaysInFirstWeek = getMinimalDaysInFirstWeek();
+    if ((7 - relDowJan1) >= minimalDaysInFirstWeek) {
         ++woy;
     }
 
@@ -1639,11 +1599,13 @@ void Calendar::computeWeekFields(UErrorCode &ec) {
         // to handle the case in which we are the first week of the
         // next year.
 
-        int32_t prevDoy = dayOfYear + handleGetYearLength(eyear - 1);
+        int32_t prevDoy = dayOfYear + handleGetYearLength(eyear - 1, ec);
+        if(U_FAILURE(ec)) return;
         woy = weekNumber(prevDoy, dayOfWeek);
         yearOfWeekOfYear--;
     } else {
-        int32_t lastDoy = handleGetYearLength(eyear);
+        int32_t lastDoy = handleGetYearLength(eyear, ec);
+        if(U_FAILURE(ec)) return;
         // Fast check: For it to be week 1 of the next year, the DOY
         // must be on or after L-5, where L is yearLength(), then it
         // cannot possibly be week 1 of the next year:
@@ -1655,7 +1617,7 @@ void Calendar::computeWeekFields(UErrorCode &ec) {
             if (lastRelDow < 0) {
                 lastRelDow += 7;
             }
-            if (((6 - lastRelDow) >= getMinimalDaysInFirstWeek()) &&
+            if (((6 - lastRelDow) >= minimalDaysInFirstWeek) &&
                 ((dayOfYear + 7 - relDow) > lastDoy)) {
                     woy = 1;
                     yearOfWeekOfYear++;
@@ -2946,7 +2908,7 @@ void Calendar::validateField(UCalendarDateFields field, UErrorCode &status) {
         if (U_FAILURE(status)) {
            return;
         }
-        validateField(field, 1, handleGetYearLength(y), status);
+        validateField(field, 1, handleGetYearLength(y, status), status);
         break;
     case UCAL_DAY_OF_WEEK_IN_MONTH:
         if (internalGet(field) == 0) {
@@ -3607,9 +3569,19 @@ int32_t Calendar::handleComputeJulianDay(UCalendarDateFields bestField, UErrorCo
                     fprintf(stderr, "%s:%d - y=%d, y-1=%d doy%d, njd%d (C.F. %d)\n",
                         __FILE__, __LINE__, year, year-1, testDate, julianDay+testDate, nextJulianDay);
 #endif
-                    if(julianDay+testDate > nextJulianDay) { // is it past Dec 31?  (nextJulianDay is day BEFORE year+1's  Jan 1)
+                    if (uprv_add32_overflow(julianDay, testDate, &testDate)) {
+                        status = U_ILLEGAL_ARGUMENT_ERROR;
+                        return 0;
+                    }
+
+                    if(testDate > nextJulianDay) { // is it past Dec 31?  (nextJulianDay is day BEFORE year+1's  Jan 1)
                         // Fire up the calculating engines.. retry YWOY = (year-1)
-                        julianDay = handleComputeMonthStart(year-1, 0, false, status); // jd before Jan 1 of previous year
+                        int32_t prevYear;
+                        if (uprv_add32_overflow(year, -1, &prevYear)) {
+                            status = U_ILLEGAL_ARGUMENT_ERROR;
+                            return 0;
+                        }
+                        julianDay = handleComputeMonthStart(prevYear, 0, false, status); // jd before Jan 1 of previous year
                         if (U_FAILURE(status)) {
                             return 0;
                         }
@@ -3834,16 +3806,20 @@ int32_t Calendar::handleGetExtendedYearFromWeekFields(int32_t yearWoy, int32_t w
 
 int32_t Calendar::handleGetMonthLength(int32_t extendedYear, int32_t month, UErrorCode& status) const
 {
-    return handleComputeMonthStart(extendedYear, month+1, true, status) -
+    int32_t nextMonth;
+    if (uprv_add32_overflow(month, 1, &nextMonth)) {
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+    return handleComputeMonthStart(extendedYear, nextMonth, true, status) -
         handleComputeMonthStart(extendedYear, month, true, status);
 }
 
-int32_t Calendar::handleGetYearLength(int32_t eyear) const
+int32_t Calendar::handleGetYearLength(int32_t eyear, UErrorCode& status) const
 {
-    UErrorCode status = U_ZERO_ERROR;
     int32_t result = handleComputeMonthStart(eyear+1, 0, false, status) -
         handleComputeMonthStart(eyear, 0, false, status);
-    U_ASSERT(U_SUCCESS(status));
+    if (U_FAILURE(status)) return 0;
     return result;
 }
 
@@ -3882,7 +3858,7 @@ Calendar::getActualMaximum(UCalendarDateFields field, UErrorCode& status) const
             }
             cal->setLenient(true);
             cal->prepareGetActual(field,false,status);
-            result = handleGetYearLength(cal->get(UCAL_EXTENDED_YEAR, status));
+            result = handleGetYearLength(cal->get(UCAL_EXTENDED_YEAR, status), status);
             delete cal;
         }
         break;
@@ -4141,7 +4117,7 @@ Calendar::setWeekData(const Locale& desiredLocale, const char *type, UErrorCode&
     if (U_SUCCESS(status)) {
         U_LOCALE_BASED(locBased,*this);
         locBased.setLocaleIDs(ures_getLocaleByType(monthNames.getAlias(), ULOC_VALID_LOCALE, &status),
-                              ures_getLocaleByType(monthNames.getAlias(), ULOC_ACTUAL_LOCALE, &status));
+                              ures_getLocaleByType(monthNames.getAlias(), ULOC_ACTUAL_LOCALE, &status), status);
     } else {
         status = U_USING_FALLBACK_WARNING;
         return;
@@ -4229,14 +4205,12 @@ Calendar::updateTime(UErrorCode& status)
 
 Locale
 Calendar::getLocale(ULocDataLocaleType type, UErrorCode& status) const {
-    U_LOCALE_BASED(locBased, *this);
+    return LocaleBased::getLocale(validLocale, actualLocale, type, status);
-    return locBased.getLocale(type, status);
 }
 
 const char *
 Calendar::getLocaleID(ULocDataLocaleType type, UErrorCode& status) const {
-    U_LOCALE_BASED(locBased, *this);
+    return LocaleBased::getLocaleID(validLocale, actualLocale, type, status);
-    return locBased.getLocaleID(type, status);
 }
 
 void
@@ -4245,7 +4219,7 @@ Calendar::recalculateStamp() {
     int32_t currentValue;
     int32_t j, i;
 
-    fNextStamp = 1;
+    fNextStamp = kInternallySet;
 
     for (j = 0; j < UCAL_FIELD_COUNT; j++) {
         currentValue = STAMP_MAX;

deps/icu-small/source/i18n/cecal.cpp

@@ -53,7 +53,6 @@ static const int32_t LIMITS[UCAL_FIELD_COUNT][4] = {
 CECalendar::CECalendar(const Locale& aLocale, UErrorCode& success)
 :   Calendar(TimeZone::forLocaleOrDefault(aLocale), aLocale, success)
 {
-    setTimeInMillis(getNow(), success);
 }
 
 CECalendar::CECalendar (const CECalendar& other) 
@@ -65,13 +64,6 @@ CECalendar::~CECalendar()
 {
 }
 
-CECalendar&
-CECalendar::operator=(const CECalendar& right)
-{
-    Calendar::operator=(right);
-    return *this;
-}
-
 //-------------------------------------------------------------------------
 // Calendar framework
 //-------------------------------------------------------------------------

deps/icu-small/source/i18n/cecal.h

@@ -82,13 +82,6 @@ protected:
      */
     virtual ~CECalendar();
 
-    /**
-     * Default assignment operator
-     * @param right    Calendar object to be copied
-     * @internal
-     */
-    CECalendar& operator=(const CECalendar& right);
-
 protected:
     //-------------------------------------------------------------------------
     // Calendar framework

deps/icu-small/source/i18n/chnsecal.cpp

@@ -130,7 +130,6 @@ ChineseCalendar::ChineseCalendar(const Locale& aLocale, UErrorCode& success)
 :   Calendar(TimeZone::forLocaleOrDefault(aLocale), aLocale, success),
     hasLeapMonthBetweenWinterSolstices(false)
 {
-    setTimeInMillis(getNow(), success); // Call this again now that the vtable is set up properly.
 }
 
 ChineseCalendar::ChineseCalendar(const ChineseCalendar& other) : Calendar(other) {
@@ -219,7 +218,9 @@ int32_t ChineseCalendar::handleGetExtendedYear(UErrorCode& status) {
     }
 
     int32_t year;
-    if (newestStamp(UCAL_ERA, UCAL_YEAR, kUnset) <= fStamp[UCAL_EXTENDED_YEAR]) {
+    // if UCAL_EXTENDED_YEAR is not older than UCAL_ERA nor UCAL_YEAR
+    if (newerField(UCAL_EXTENDED_YEAR, newerField(UCAL_ERA, UCAL_YEAR)) ==
+        UCAL_EXTENDED_YEAR) {
         year = internalGet(UCAL_EXTENDED_YEAR, 1); // Default to year 1
     } else {
         // adjust to the instance specific epoch
@@ -252,11 +253,16 @@ int32_t ChineseCalendar::handleGetExtendedYear(UErrorCode& status) {
  * @stable ICU 2.8
  */
 int32_t ChineseCalendar::handleGetMonthLength(int32_t extendedYear, int32_t month, UErrorCode& status) const {
+    bool isLeapMonth = internalGet(UCAL_IS_LEAP_MONTH) == 1;
+    return handleGetMonthLengthWithLeap(extendedYear, month, isLeapMonth, status);
+}
+
+int32_t ChineseCalendar::handleGetMonthLengthWithLeap(int32_t extendedYear, int32_t month, bool leap, UErrorCode& status) const {
     const Setting setting = getSetting(status);
     if (U_FAILURE(status)) {
         return 0;
     }
-    int32_t thisStart = handleComputeMonthStart(extendedYear, month, true, status);
+    int32_t thisStart = handleComputeMonthStartWithLeap(extendedYear, month, leap, status);
     if (U_FAILURE(status)) {
         return 0;
     }
@@ -332,18 +338,24 @@ struct MonthInfo computeMonthInfo(
  * @stable ICU 2.8
  */
 int64_t ChineseCalendar::handleComputeMonthStart(int32_t eyear, int32_t month, UBool useMonth, UErrorCode& status) const {
+    bool isLeapMonth = false;
+    if (useMonth) {
+        isLeapMonth = internalGet(UCAL_IS_LEAP_MONTH) != 0;
+    }
+    return handleComputeMonthStartWithLeap(eyear, month, isLeapMonth, status);
+}
+
+int64_t ChineseCalendar::handleComputeMonthStartWithLeap(int32_t eyear, int32_t month, bool isLeapMonth, UErrorCode& status) const {
     if (U_FAILURE(status)) {
        return 0;
     }
     // If the month is out of range, adjust it into range, and
     // modify the extended year value accordingly.
     if (month < 0 || month > 11) {
-        double m = month;
+        if (uprv_add32_overflow(eyear, ClockMath::floorDivide(month, 12, &month), &eyear)) {
-        if (uprv_add32_overflow(eyear, ClockMath::floorDivide(m, 12.0, &m), &eyear)) {
             status = U_ILLEGAL_ARGUMENT_ERROR;
             return 0;
         }
-        month = static_cast<int32_t>(m);
     }
 
     const Setting setting = getSetting(status);
@@ -362,19 +374,9 @@ int64_t ChineseCalendar::handleComputeMonthStart(int32_t eyear, int32_t month, U
        return 0;
     }
 
-    // Ignore IS_LEAP_MONTH field if useMonth is false
+    int32_t newMonthYear = Grego::dayToYear(newMoon, status);
-    bool isLeapMonth = false;
-    if (useMonth) {
-        isLeapMonth = internalGet(UCAL_IS_LEAP_MONTH) != 0;
-    }
 
-    int32_t unusedMonth;
+    struct MonthInfo monthInfo = computeMonthInfo(setting, newMonthYear, newMoon, status);
-    int32_t unusedDayOfWeek;
-    int32_t unusedDayOfMonth;
-    int32_t unusedDayOfYear;
-    Grego::dayToFields(newMoon, gyear, unusedMonth, unusedDayOfWeek, unusedDayOfMonth, unusedDayOfYear, status);
-
-    struct MonthInfo monthInfo = computeMonthInfo(setting, gyear, newMoon, status);
     if (U_FAILURE(status)) {
        return 0;
     }
@@ -794,6 +796,9 @@ struct MonthInfo computeMonthInfo(
         solsticeBefore = solsticeAfter;
         solsticeAfter = winterSolstice(setting, gyear + 1, status);
     }
+    if (!(solsticeBefore <= days && days < solsticeAfter)) {
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+    }
     if (U_FAILURE(status)) {
         return output;
     }
@@ -1043,7 +1048,12 @@ void ChineseCalendar::offsetMonth(int32_t newMoon, int32_t dayOfMonth, int32_t d
     }
 
     // Find the target dayOfMonth
-    int32_t jd = newMoon + kEpochStartAsJulianDay - 1 + dayOfMonth;
+    int32_t jd;
+    if (uprv_add32_overflow(newMoon, kEpochStartAsJulianDay - 1, &jd) ||
+        uprv_add32_overflow(jd, dayOfMonth, &jd)) {
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
 
     // Pin the dayOfMonth.  In this calendar all months are 29 or 30 days
     // so pinning just means handling dayOfMonth 30.
@@ -1182,6 +1192,27 @@ ChineseCalendar::Setting ChineseCalendar::getSetting(UErrorCode&) const {
   };
 }
 
+int32_t
+ChineseCalendar::getActualMaximum(UCalendarDateFields field, UErrorCode& status) const
+{
+    if (U_FAILURE(status)) {
+       return 0;
+    }
+    if (field == UCAL_DATE) {
+        LocalPointer<ChineseCalendar> cal(clone(), status);
+        if(U_FAILURE(status)) {
+            return 0;
+        }
+        cal->setLenient(true);
+        cal->prepareGetActual(field,false,status);
+        int32_t year = cal->get(UCAL_EXTENDED_YEAR, status);
+        int32_t month = cal->get(UCAL_MONTH, status);
+        bool leap = cal->get(UCAL_IS_LEAP_MONTH, status) != 0;
+        return handleGetMonthLengthWithLeap(year, month, leap, status);
+    }
+    return Calendar::getActualMaximum(field, status);
+}
+
 U_NAMESPACE_END
 
 #endif

deps/icu-small/source/i18n/chnsecal.h

@@ -194,6 +194,10 @@ class U_I18N_API ChineseCalendar : public Calendar {
   virtual void handleComputeFields(int32_t julianDay, UErrorCode &status) override;
   virtual const UFieldResolutionTable* getFieldResolutionTable() const override;
 
+ private:
+  int32_t handleGetMonthLengthWithLeap(int32_t extendedYear, int32_t month, bool isLeap, UErrorCode& status) const;
+  int64_t handleComputeMonthStartWithLeap(int32_t eyear, int32_t month, bool isLeap, UErrorCode& status) const;
+
  public:
   virtual void add(UCalendarDateFields field, int32_t amount, UErrorCode &status) override;
   virtual void add(EDateFields field, int32_t amount, UErrorCode &status) override;
@@ -254,6 +258,8 @@ class U_I18N_API ChineseCalendar : public Calendar {
    */
   virtual const char * getType() const override;
 
+  virtual int32_t getActualMaximum(UCalendarDateFields field, UErrorCode& status) const override;
+
   struct Setting {
       int32_t epochYear;
       const TimeZone* zoneAstroCalc;

deps/icu-small/source/i18n/collationruleparser.cpp

@@ -613,18 +613,24 @@ CollationRuleParser::parseSetting(UErrorCode &errorCode) {
                 return;
             }
             // localeID minus all keywords
-            char baseID[ULOC_FULLNAME_CAPACITY];
+            CharString baseID = ulocimp_getBaseName(localeID.toStringPiece(), errorCode);
-            int32_t length = uloc_getBaseName(localeID.data(), baseID, ULOC_FULLNAME_CAPACITY, &errorCode);
+            if (U_FAILURE(errorCode)) {
-            if(U_FAILURE(errorCode) || length >= ULOC_KEYWORDS_CAPACITY) {
                 errorCode = U_ZERO_ERROR;
                 setParseError("expected language tag in [import langTag]", errorCode);
                 return;
             }
-            if(length == 0) {
+            if (baseID.isEmpty()) {
-                uprv_strcpy(baseID, "root");
+                baseID.copyFrom("root", errorCode);
-            } else if(*baseID == '_') {
+            } else if (baseID[0] == '_') {
-                uprv_memmove(baseID + 3, baseID, length + 1);
+                // CharString doesn't have any insert() method, only append().
-                uprv_memcpy(baseID, "und", 3);
+                constexpr char und[] = "und";
+                constexpr int32_t length = sizeof und - 1;
+                int32_t dummy;
+                char* tail = baseID.getAppendBuffer(length, length, dummy, errorCode);
+                char* head = baseID.data();
+                uprv_memmove(head + length, head, baseID.length());
+                uprv_memcpy(head, und, length);
+                baseID.append(tail, length, errorCode);
             }
             // @collation=type, or length=0 if not specified
             CharString collationType = ulocimp_getKeywordValue(localeID.data(), "collation", errorCode);
@@ -637,7 +643,7 @@ CollationRuleParser::parseSetting(UErrorCode &errorCode) {
                 setParseError("[import langTag] is not supported", errorCode);
             } else {
                 UnicodeString importedRules;
-                importer->getRules(baseID,
+                importer->getRules(baseID.data(),
                                    !collationType.isEmpty() ? collationType.data() : "standard",
                                    importedRules, errorReason, errorCode);
                 if(U_FAILURE(errorCode)) {

deps/icu-small/source/i18n/datefmt.cpp

@@ -40,6 +40,7 @@
 #if defined( U_DEBUG_CALSVC ) || defined (U_DEBUG_CAL)
 #include <stdio.h>
 #endif
+#include <typeinfo>
 
 // *****************************************************************************
 // class DateFormat
@@ -279,9 +280,8 @@ UnicodeString&
 DateFormat::format(UDate date, UnicodeString& appendTo, FieldPosition& fieldPosition) const {
     if (fCalendar != nullptr) {
         UErrorCode ec = U_ZERO_ERROR;
-        const auto* calType = fCalendar->getType();
         // Avoid a heap allocation and corresponding free for the common case
-        if (uprv_strcmp(calType, "gregorian") == 0) {
+        if (typeid(*fCalendar) == typeid(GregorianCalendar)) {
             GregorianCalendar cal(*static_cast<GregorianCalendar*>(fCalendar));
             cal.setTime(date, ec);
             if (U_SUCCESS(ec)) {
@@ -309,9 +309,8 @@ DateFormat::format(UDate date, UnicodeString& appendTo, FieldPositionIterator* p
                    UErrorCode& status) const {
     if (fCalendar != nullptr) {
         UErrorCode ec = U_ZERO_ERROR;
-        const auto* calType = fCalendar->getType();
         // Avoid a heap allocation and corresponding free for the common case
-        if (uprv_strcmp(calType, "gregorian") == 0) {
+        if (typeid(*fCalendar) == typeid(GregorianCalendar)) {
             GregorianCalendar cal(*static_cast<GregorianCalendar*>(fCalendar));
             cal.setTime(date, ec);
             if (U_SUCCESS(ec)) {

deps/icu-small/source/i18n/dcfmtsym.cpp

@@ -118,7 +118,6 @@ DecimalFormatSymbols::DecimalFormatSymbols(const Locale& loc, const NumberingSys
 
 DecimalFormatSymbols::DecimalFormatSymbols()
         : UObject(), locale(Locale::getRoot()) {
-    *validLocale = *actualLocale = 0;
     initialize();
 }
 
@@ -136,6 +135,8 @@ DecimalFormatSymbols::createWithLastResortData(UErrorCode& status) {
 
 DecimalFormatSymbols::~DecimalFormatSymbols()
 {
+    delete actualLocale;
+    delete validLocale;
 }
 
 // -------------------------------------
@@ -163,8 +164,12 @@ DecimalFormatSymbols::operator=(const DecimalFormatSymbols& rhs)
             currencySpcAfterSym[i].fastCopyFrom(rhs.currencySpcAfterSym[i]);
         }
         locale = rhs.locale;
-        uprv_strcpy(validLocale, rhs.validLocale);
+
-        uprv_strcpy(actualLocale, rhs.actualLocale);
+        UErrorCode status = U_ZERO_ERROR;
+        U_LOCALE_BASED(locBased, *this);
+        locBased.setLocaleIDs(rhs.validLocale, rhs.actualLocale, status);
+        U_ASSERT(U_SUCCESS(status));
+
         fIsCustomCurrencySymbol = rhs.fIsCustomCurrencySymbol; 
         fIsCustomIntlCurrencySymbol = rhs.fIsCustomIntlCurrencySymbol; 
         fCodePointZero = rhs.fCodePointZero;
@@ -203,8 +208,8 @@ DecimalFormatSymbols::operator==(const DecimalFormatSymbols& that) const
     }
     // No need to check fCodePointZero since it is based on fSymbols
     return locale == that.locale &&
-        uprv_strcmp(validLocale, that.validLocale) == 0 &&
+        LocaleBased::equalIDs(actualLocale, that.actualLocale) &&
-        uprv_strcmp(actualLocale, that.actualLocale) == 0;
+        LocaleBased::equalIDs(validLocale, that.validLocale);
 }
 
 // -------------------------------------
@@ -353,7 +358,6 @@ DecimalFormatSymbols::initialize(const Locale& loc, UErrorCode& status,
     UBool useLastResortData, const NumberingSystem* ns)
 {
     if (U_FAILURE(status)) { return; }
-    *validLocale = *actualLocale = 0;
 
     // First initialize all the symbols to the fallbacks for anything we can't find
     initialize();
@@ -409,7 +413,8 @@ DecimalFormatSymbols::initialize(const Locale& loc, UErrorCode& status,
             ULOC_VALID_LOCALE, &status),
         ures_getLocaleByType(
             numberElementsRes.getAlias(),
-            ULOC_ACTUAL_LOCALE, &status));
+            ULOC_ACTUAL_LOCALE, &status),
+        status);
 
     // Now load the rest of the data from the data sink.
     // Start with loading this nsName if it is not Latin.
@@ -568,8 +573,7 @@ void DecimalFormatSymbols::setCurrency(const char16_t* currency, UErrorCode& sta
 
 Locale
 DecimalFormatSymbols::getLocale(ULocDataLocaleType type, UErrorCode& status) const {
-    U_LOCALE_BASED(locBased, *this);
+    return LocaleBased::getLocale(validLocale, actualLocale, type, status);
-    return locBased.getLocale(type, status);
 }
 
 const UnicodeString&

deps/icu-small/source/i18n/dtfmtsym.cpp

@@ -402,9 +402,8 @@ void
 DateFormatSymbols::copyData(const DateFormatSymbols& other) {
     UErrorCode status = U_ZERO_ERROR;
     U_LOCALE_BASED(locBased, *this);
-    locBased.setLocaleIDs(
+    locBased.setLocaleIDs(other.validLocale, other.actualLocale, status);
-        other.getLocale(ULOC_VALID_LOCALE, status),
+    U_ASSERT(U_SUCCESS(status));
-        other.getLocale(ULOC_ACTUAL_LOCALE, status));
     assignArray(fEras, fErasCount, other.fEras, other.fErasCount);
     assignArray(fEraNames, fEraNamesCount, other.fEraNames, other.fEraNamesCount);
     assignArray(fNarrowEras, fNarrowErasCount, other.fNarrowEras, other.fNarrowErasCount);
@@ -497,6 +496,8 @@ DateFormatSymbols& DateFormatSymbols::operator=(const DateFormatSymbols& other)
 DateFormatSymbols::~DateFormatSymbols()
 {
     dispose();
+    delete actualLocale;
+    delete validLocale;
 }
 
 void DateFormatSymbols::dispose()
@@ -536,6 +537,10 @@ void DateFormatSymbols::dispose()
     delete[] fStandaloneWideDayPeriods;
     delete[] fStandaloneNarrowDayPeriods;
 
+    delete actualLocale;
+    actualLocale = nullptr;
+    delete validLocale;
+    validLocale = nullptr;
     disposeZoneStrings();
 }
 
@@ -2302,7 +2307,7 @@ DateFormatSymbols::initializeData(const Locale& locale, const char *type, UError
     // of it that we need except for the time-zone and localized-pattern data, which
     // are stored in a separate file
     locBased.setLocaleIDs(ures_getLocaleByType(cb.getAlias(), ULOC_VALID_LOCALE, &status),
-                          ures_getLocaleByType(cb.getAlias(), ULOC_ACTUAL_LOCALE, &status));
+                          ures_getLocaleByType(cb.getAlias(), ULOC_ACTUAL_LOCALE, &status), status);
 
     // Load eras
     initField(&fEras, fErasCount, calendarSink, buildResourcePath(path, gErasTag, gNamesAbbrTag, status), status);
@@ -2528,8 +2533,7 @@ DateFormatSymbols::initializeData(const Locale& locale, const char *type, UError
 
 Locale
 DateFormatSymbols::getLocale(ULocDataLocaleType type, UErrorCode& status) const {
-    U_LOCALE_BASED(locBased, *this);
+    return LocaleBased::getLocale(validLocale, actualLocale, type, status);
-    return locBased.getLocale(type, status);
 }
 
 U_NAMESPACE_END

deps/icu-small/source/i18n/erarules.cpp

@@ -305,8 +305,9 @@ void EraRules::initCurrentEra() {
         localMillis += (rawOffset + dstOffset);
     }
 
-    int year, month0, dom, dow, doy, mid;
+    int32_t year, mid;
-    Grego::timeToFields(localMillis, year, month0, dom, dow, doy, mid, ec);
+    int8_t  month0, dom;
+    Grego::timeToFields(localMillis, year, month0, dom, mid, ec);
     if (U_FAILURE(ec)) return;
     int currentEncodedDate = encodeDate(year, month0 + 1 /* changes to 1-base */, dom);
     int eraIdx = numEras - 1;

deps/icu-small/source/i18n/format.cpp

@@ -24,6 +24,7 @@
 #include "utypeinfo.h"  // for 'typeid' to work
 
 #include "unicode/utypes.h"
+#include "charstr.h"
 
 #ifndef U_I18N_IMPLEMENTATION
 #error U_I18N_IMPLEMENTATION not set - must be set for all ICU source files in i18n/ - see https://unicode-org.github.io/icu/userguide/howtouseicu
@@ -72,13 +73,14 @@ FieldPosition::clone() const {
 Format::Format()
     : UObject()
 {
-    *validLocale = *actualLocale = 0;
 }
 
 // -------------------------------------
 
 Format::~Format()
 {
+    delete actualLocale;
+    delete validLocale;
 }
 
 // -------------------------------------
@@ -97,8 +99,10 @@ Format&
 Format::operator=(const Format& that)
 {
     if (this != &that) {
-        uprv_strcpy(validLocale, that.validLocale);
+        UErrorCode status = U_ZERO_ERROR;
-        uprv_strcpy(actualLocale, that.actualLocale);
+        U_LOCALE_BASED(locBased, *this);
+        locBased.setLocaleIDs(that.validLocale, that.actualLocale, status);
+        U_ASSERT(U_SUCCESS(status));
     }
     return *this;
 }
@@ -196,20 +200,20 @@ void Format::syntaxError(const UnicodeString& pattern,
 
 Locale
 Format::getLocale(ULocDataLocaleType type, UErrorCode& status) const {
-    U_LOCALE_BASED(locBased, *this);
+    return LocaleBased::getLocale(validLocale, actualLocale, type, status);
-    return locBased.getLocale(type, status);
 }
 
 const char *
 Format::getLocaleID(ULocDataLocaleType type, UErrorCode& status) const {
-    U_LOCALE_BASED(locBased, *this);
+    return LocaleBased::getLocaleID(validLocale,actualLocale, type, status);
-    return locBased.getLocaleID(type, status);
 }
 
 void
 Format::setLocaleIDs(const char* valid, const char* actual) {
     U_LOCALE_BASED(locBased, *this);
-    locBased.setLocaleIDs(valid, actual);
+    UErrorCode status = U_ZERO_ERROR;
+    locBased.setLocaleIDs(valid, actual, status);
+    U_ASSERT(U_SUCCESS(status));
 }
 
 U_NAMESPACE_END

deps/icu-small/source/i18n/formattedvalue.cpp

@@ -193,6 +193,11 @@ ucfpos_close(UConstrainedFieldPosition* ptr) {
 }
 
 
+// -Wreturn-local-addr first found in https://gcc.gnu.org/onlinedocs/gcc-4.8.5/gcc/Warning-Options.html#Warning-Options
+#if U_GCC_MAJOR_MINOR >= 409
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wreturn-local-addr"
+#endif
 U_CAPI const char16_t* U_EXPORT2
 ufmtval_getString(
         const UFormattedValue* ufmtval,
@@ -213,6 +218,9 @@ ufmtval_getString(
     // defined to return memory owned by the ufmtval argument.
     return readOnlyAlias.getBuffer();
 }
+#if U_GCC_MAJOR_MINOR >= 409
+#pragma GCC diagnostic pop
+#endif
 
 
 U_CAPI UBool U_EXPORT2

deps/icu-small/source/i18n/gregocal.cpp

@@ -147,6 +147,7 @@ UOBJECT_DEFINE_RTTI_IMPLEMENTATION(GregorianCalendar)
 // in Java, -12219292800000L
 //const UDate GregorianCalendar::kPapalCutover = -12219292800000L;
 static const uint32_t kCutoverJulianDay = 2299161;
+static const int32_t kDefaultCutoverYear = 1582;
 static const UDate kPapalCutover = (2299161.0 - kEpochStartAsJulianDay) * U_MILLIS_PER_DAY;
 //static const UDate kPapalCutoverJulian = (2299161.0 - kEpochStartAsJulianDay);
 
@@ -155,7 +156,7 @@ static const UDate kPapalCutover = (2299161.0 - kEpochStartAsJulianDay) * U_MILL
 GregorianCalendar::GregorianCalendar(UErrorCode& status)
 :   Calendar(status),
 fGregorianCutover(kPapalCutover),
-fCutoverJulianDay(kCutoverJulianDay), fNormalizedGregorianCutover(fGregorianCutover), fGregorianCutoverYear(1582),
+fCutoverJulianDay(kCutoverJulianDay), fNormalizedGregorianCutover(fGregorianCutover), fGregorianCutoverYear(kDefaultCutoverYear),
 fIsGregorian(true), fInvertGregorian(false)
 {
     setTimeInMillis(getNow(), status);
@@ -164,34 +165,22 @@ fIsGregorian(true), fInvertGregorian(false)
 // -------------------------------------
 
 GregorianCalendar::GregorianCalendar(TimeZone* zone, UErrorCode& status)
-:   Calendar(zone, Locale::getDefault(), status),
+:   GregorianCalendar(zone, Locale::getDefault(), status)
-fGregorianCutover(kPapalCutover),
-fCutoverJulianDay(kCutoverJulianDay), fNormalizedGregorianCutover(fGregorianCutover), fGregorianCutoverYear(1582),
-fIsGregorian(true), fInvertGregorian(false)
 {
-    setTimeInMillis(getNow(), status);
 }
 
 // -------------------------------------
 
 GregorianCalendar::GregorianCalendar(const TimeZone& zone, UErrorCode& status)
-:   Calendar(zone, Locale::getDefault(), status),
+:   GregorianCalendar(zone, Locale::getDefault(), status)
-fGregorianCutover(kPapalCutover),
-fCutoverJulianDay(kCutoverJulianDay), fNormalizedGregorianCutover(fGregorianCutover), fGregorianCutoverYear(1582),
-fIsGregorian(true), fInvertGregorian(false)
 {
-    setTimeInMillis(getNow(), status);
 }
 
 // -------------------------------------
 
 GregorianCalendar::GregorianCalendar(const Locale& aLocale, UErrorCode& status)
-:   Calendar(TimeZone::forLocaleOrDefault(aLocale), aLocale, status),
+:   GregorianCalendar(TimeZone::forLocaleOrDefault(aLocale), aLocale, status)
-fGregorianCutover(kPapalCutover),
-fCutoverJulianDay(kCutoverJulianDay), fNormalizedGregorianCutover(fGregorianCutover), fGregorianCutoverYear(1582),
-fIsGregorian(true), fInvertGregorian(false)
 {
-    setTimeInMillis(getNow(), status);
 }
 
 // -------------------------------------
@@ -200,7 +189,7 @@ GregorianCalendar::GregorianCalendar(TimeZone* zone, const Locale& aLocale,
                                      UErrorCode& status)
                                      :   Calendar(zone, aLocale, status),
                                      fGregorianCutover(kPapalCutover),
-                                     fCutoverJulianDay(kCutoverJulianDay), fNormalizedGregorianCutover(fGregorianCutover), fGregorianCutoverYear(1582),
+                                     fCutoverJulianDay(kCutoverJulianDay), fNormalizedGregorianCutover(fGregorianCutover), fGregorianCutoverYear(kDefaultCutoverYear),
                                      fIsGregorian(true), fInvertGregorian(false)
 {
     setTimeInMillis(getNow(), status);
@@ -212,7 +201,7 @@ GregorianCalendar::GregorianCalendar(const TimeZone& zone, const Locale& aLocale
                                      UErrorCode& status)
                                      :   Calendar(zone, aLocale, status),
                                      fGregorianCutover(kPapalCutover),
-                                     fCutoverJulianDay(kCutoverJulianDay), fNormalizedGregorianCutover(fGregorianCutover), fGregorianCutoverYear(1582),
+                                     fCutoverJulianDay(kCutoverJulianDay), fNormalizedGregorianCutover(fGregorianCutover), fGregorianCutoverYear(kDefaultCutoverYear),
                                      fIsGregorian(true), fInvertGregorian(false)
 {
     setTimeInMillis(getNow(), status);
@@ -224,7 +213,7 @@ GregorianCalendar::GregorianCalendar(int32_t year, int32_t month, int32_t date,
                                      UErrorCode& status)
                                      :   Calendar(TimeZone::createDefault(), Locale::getDefault(), status),
                                      fGregorianCutover(kPapalCutover),
-                                     fCutoverJulianDay(kCutoverJulianDay), fNormalizedGregorianCutover(fGregorianCutover), fGregorianCutoverYear(1582),
+                                     fCutoverJulianDay(kCutoverJulianDay), fNormalizedGregorianCutover(fGregorianCutover), fGregorianCutoverYear(kDefaultCutoverYear),
                                      fIsGregorian(true), fInvertGregorian(false)
 {
     set(UCAL_ERA, AD);
@@ -237,15 +226,8 @@ GregorianCalendar::GregorianCalendar(int32_t year, int32_t month, int32_t date,
 
 GregorianCalendar::GregorianCalendar(int32_t year, int32_t month, int32_t date,
                                      int32_t hour, int32_t minute, UErrorCode& status)
-                                     :   Calendar(TimeZone::createDefault(), Locale::getDefault(), status),
+                                     :   GregorianCalendar(year, month, date, status)
-                                     fGregorianCutover(kPapalCutover),
-                                     fCutoverJulianDay(kCutoverJulianDay), fNormalizedGregorianCutover(fGregorianCutover), fGregorianCutoverYear(1582),
-                                     fIsGregorian(true), fInvertGregorian(false)
 {
-    set(UCAL_ERA, AD);
-    set(UCAL_YEAR, year);
-    set(UCAL_MONTH, month);
-    set(UCAL_DATE, date);
     set(UCAL_HOUR_OF_DAY, hour);
     set(UCAL_MINUTE, minute);
 }
@@ -255,17 +237,8 @@ GregorianCalendar::GregorianCalendar(int32_t year, int32_t month, int32_t date,
 GregorianCalendar::GregorianCalendar(int32_t year, int32_t month, int32_t date,
                                      int32_t hour, int32_t minute, int32_t second,
                                      UErrorCode& status)
-                                     :   Calendar(TimeZone::createDefault(), Locale::getDefault(), status),
+                                     :   GregorianCalendar(year, month, date, hour, minute, status)
-                                     fGregorianCutover(kPapalCutover),
-                                     fCutoverJulianDay(kCutoverJulianDay), fNormalizedGregorianCutover(fGregorianCutover), fGregorianCutoverYear(1582),
-                                     fIsGregorian(true), fInvertGregorian(false)
 {
-    set(UCAL_ERA, AD);
-    set(UCAL_YEAR, year);
-    set(UCAL_MONTH, month);
-    set(UCAL_DATE, date);
-    set(UCAL_HOUR_OF_DAY, hour);
-    set(UCAL_MINUTE, minute);
     set(UCAL_SECOND, second);
 }
 
@@ -601,7 +574,8 @@ int32_t GregorianCalendar::handleGetMonthLength(int32_t extendedYear, int32_t mo
     return isLeapYear(extendedYear) ? kLeapMonthLength[month] : kMonthLength[month];
 }
 
-int32_t GregorianCalendar::handleGetYearLength(int32_t eyear) const {
+int32_t GregorianCalendar::handleGetYearLength(int32_t eyear, UErrorCode& status) const {
+    if (U_FAILURE(status)) return 0;
     return isLeapYear(eyear) ? 366 : 365;
 }
 
@@ -871,13 +845,14 @@ GregorianCalendar::roll(UCalendarDateFields field, int32_t amount, UErrorCode& s
         }
         if (month == UCAL_JANUARY) {
             if (woy >= 52) {
-                isoDoy += handleGetYearLength(isoYear);
+                isoDoy += handleGetYearLength(isoYear, status);
             }
         } else {
             if (woy == 1) {
-                isoDoy -= handleGetYearLength(isoYear - 1);
+                isoDoy -= handleGetYearLength(isoYear - 1, status);
             }
         }
+        if (U_FAILURE(status)) return;
         if (uprv_add32_overflow(woy, amount, &woy)) {
             status = U_ILLEGAL_ARGUMENT_ERROR;
             return;
@@ -890,7 +865,8 @@ GregorianCalendar::roll(UCalendarDateFields field, int32_t amount, UErrorCode& s
             // days at the end of the year are going to fall into
             // week 1 of the next year, we drop the last week by
             // subtracting 7 from the last day of the year.
-            int32_t lastDoy = handleGetYearLength(isoYear);
+            int32_t lastDoy = handleGetYearLength(isoYear, status);
+            if (U_FAILURE(status)) return;
             int32_t lastRelDow = (lastDoy - isoDoy + internalGet(UCAL_DAY_OF_WEEK) -
                 getFirstDayOfWeek()) % 7;
             if (lastRelDow < 0) lastRelDow += 7;
@@ -1186,14 +1162,10 @@ int32_t GregorianCalendar::handleGetExtendedYear(UErrorCode& status) {
     int32_t year = kEpochYear;
 
     // year field to use
-    int32_t yearField = UCAL_EXTENDED_YEAR;
-
     // There are three separate fields which could be used to
     // derive the proper year.  Use the one most recently set.
-    if (fStamp[yearField] < fStamp[UCAL_YEAR])
+    UCalendarDateFields yearField = newerField(
-        yearField = UCAL_YEAR;
+        newerField(UCAL_EXTENDED_YEAR, UCAL_YEAR), UCAL_YEAR_WOY);
-    if (fStamp[yearField] < fStamp[UCAL_YEAR_WOY])
-        yearField = UCAL_YEAR_WOY;
 
     // based on the "best" year field, get the year
     switch(yearField) {

deps/icu-small/source/i18n/gregoimp.cpp

@@ -117,57 +117,110 @@ int64_t Grego::fieldsToDay(int32_t year, int32_t month, int32_t dom) {
     return julian - JULIAN_1970_CE; // JD => epoch day
 }
 
-void Grego::dayToFields(int32_t day, int32_t& year, int32_t& month,
+void Grego::dayToFields(int32_t day, int32_t& year, int8_t& month,
-                        int32_t& dom, int32_t& dow, int32_t& doy, UErrorCode& status) {
+                        int8_t& dom, int8_t& dow, int16_t& doy, UErrorCode& status) {
-
+    year = dayToYear(day, doy, status); // one-based doy
     if (U_FAILURE(status)) return;
+
     // Convert from 1970 CE epoch to 1 CE epoch (Gregorian calendar)
     if (uprv_add32_overflow(day, JULIAN_1970_CE - JULIAN_1_CE, &day)) {
         status = U_ILLEGAL_ARGUMENT_ERROR;
         return;
     }
 
-    // Convert from the day number to the multiple radix
-    // representation.  We use 400-year, 100-year, and 4-year cycles.
-    // For example, the 4-year cycle has 4 years + 1 leap day; giving
-    // 1461 == 365*4 + 1 days.
-    int32_t n400 = ClockMath::floorDivide(day, 146097, &doy); // 400-year cycle length
-    int32_t n100 = ClockMath::floorDivide(doy, 36524, &doy); // 100-year cycle length
-    int32_t n4   = ClockMath::floorDivide(doy, 1461, &doy); // 4-year cycle length
-    int32_t n1   = ClockMath::floorDivide(doy, 365, &doy);
-    year = 400*n400 + 100*n100 + 4*n4 + n1;
-    if (n100 == 4 || n1 == 4) {
-        doy = 365; // Dec 31 at end of 4- or 400-year cycle
-    } else {
-        ++year;
-    }
-    
-    UBool isLeap = isLeapYear(year);
-    
     // Gregorian day zero is a Monday.
     dow = (day + 1) % 7;
     dow += (dow < 0) ? (UCAL_SUNDAY + 7) : UCAL_SUNDAY;
 
     // Common Julian/Gregorian calculation
     int32_t correction = 0;
+    bool isLeap = isLeapYear(year);
     int32_t march1 = isLeap ? 60 : 59; // zero-based DOY for March 1
-    if (doy >= march1) {
+    if (doy > march1) {
         correction = isLeap ? 1 : 2;
     }
-    month = (12 * (doy + correction) + 6) / 367; // zero-based month
+    month = (12 * (doy - 1 + correction) + 6) / 367; // zero-based month
-    dom = doy - DAYS_BEFORE[month + (isLeap ? 12 : 0)] + 1; // one-based DOM
+    dom = doy - DAYS_BEFORE[month + (isLeap ? 12 : 0)]; // one-based DOM
-    doy++; // one-based doy
 }
 
-void Grego::timeToFields(UDate time, int32_t& year, int32_t& month,
+int32_t Grego::dayToYear(int32_t day, UErrorCode& status) {
-                        int32_t& dom, int32_t& dow, int32_t& doy, int32_t& mid, UErrorCode& status) {
+    int16_t unusedDOY;
+    return dayToYear(day, unusedDOY, status);
+}
+
+int32_t Grego::dayToYear(int32_t day, int16_t& doy, UErrorCode& status) {
+    if (U_FAILURE(status)) return 0;
+    // Convert from 1970 CE epoch to 1 CE epoch (Gregorian calendar)
+    if (uprv_add32_overflow(day, JULIAN_1970_CE - JULIAN_1_CE, &day)) {
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    // Convert from the day number to the multiple radix
+    // representation.  We use 400-year, 100-year, and 4-year cycles.
+    // For example, the 4-year cycle has 4 years + 1 leap day; giving
+    // 1461 == 365*4 + 1 days.
+    int32_t doy32;
+    int32_t n400 = ClockMath::floorDivide(day, 146097, &doy32); // 400-year cycle length
+    int32_t n100 = ClockMath::floorDivide(doy32, 36524, &doy32); // 100-year cycle length
+    int32_t n4   = ClockMath::floorDivide(doy32, 1461, &doy32); // 4-year cycle length
+    int32_t n1   = ClockMath::floorDivide(doy32, 365, &doy32);
+    int32_t year = 400*n400 + 100*n100 + 4*n4 + n1;
+    if (n100 == 4 || n1 == 4) {
+        doy = 365; // Dec 31 at end of 4- or 400-year cycle
+    } else {
+        doy = doy32;
+        ++year;
+    }
+    doy++; // one-based doy
+    return year;
+}
+
+void Grego::dayToFields(int32_t day, int32_t& year, int8_t& month,
+                        int8_t& dom, int8_t& dow, UErrorCode& status) {
+    int16_t unusedDOY;
+    dayToFields(day, year, month, dom, dow, unusedDOY, status);
+}
+
+void Grego::dayToFields(int32_t day, int32_t& year, int8_t& month,
+                        int8_t& dom, int16_t& doy, UErrorCode& status) {
+    int8_t unusedDOW;
+    dayToFields(day, year, month, dom, unusedDOW, doy, status);
+}
+
+void Grego::timeToFields(UDate time, int32_t& year, int8_t& month,
+                        int8_t& dom, int32_t& mid, UErrorCode& status) {
+    int8_t unusedDOW;
+    timeToFields(time, year, month, dom, unusedDOW, mid, status);
+}
+
+void Grego::timeToFields(UDate time, int32_t& year, int8_t& month,
+                        int8_t& dom, int8_t& dow, int32_t& mid, UErrorCode& status) {
+    int16_t unusedDOY;
+    timeToFields(time, year, month, dom, dow, unusedDOY, mid, status);
+}
+
+void Grego::timeToFields(UDate time, int32_t& year, int8_t& month,
+                        int8_t& dom, int8_t& dow, int16_t& doy, int32_t& mid, UErrorCode& status) {
     if (U_FAILURE(status)) return;
-    double millisInDay;
+    double day = ClockMath::floorDivide(time, U_MILLIS_PER_DAY, &mid);
-    double day = ClockMath::floorDivide(static_cast<double>(time), static_cast<double>(U_MILLIS_PER_DAY), &millisInDay);
+    if (day > INT32_MAX || day < INT32_MIN) {
-    mid = static_cast<int32_t>(millisInDay);
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }
     dayToFields(day, year, month, dom, dow, doy, status);
 }
 
+int32_t Grego::timeToYear(UDate time, UErrorCode& status) {
+    if (U_FAILURE(status)) return 0;
+    double day = ClockMath::floorDivide(time, double(U_MILLIS_PER_DAY));
+    if (day > INT32_MAX || day < INT32_MIN) {
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+    return Grego::dayToYear(day, status);
+}
+
 int32_t Grego::dayOfWeek(int32_t day) {
     int32_t dow;
     ClockMath::floorDivide(day + int{UCAL_THURSDAY}, 7, &dow);

deps/icu-small/source/i18n/gregoimp.h

@@ -210,8 +210,21 @@ class Grego {
      * @param doy output parameter to receive day-of-year (1-based)
      * @param status error code.
      */
-    static void dayToFields(int32_t day, int32_t& year, int32_t& month,
+    static void dayToFields(int32_t day, int32_t& year, int8_t& month,
-                            int32_t& dom, int32_t& dow, int32_t& doy, UErrorCode& status);
+                            int8_t& dom, int8_t& dow, int16_t& doy, UErrorCode& status);
+
+    /**
+     * Convert a 1970-epoch day number to proleptic Gregorian year,
+     * month, day-of-month, and day-of-week.
+     * @param day 1970-epoch day
+     * @param year output parameter to receive year
+     * @param month output parameter to receive month (0-based, 0==Jan)
+     * @param dom output parameter to receive day-of-month (1-based)
+     * @param doy output parameter to receive day-of-year (1-based)
+     * @param status error code.
+     */
+    static void dayToFields(int32_t day, int32_t& year, int8_t& month,
+                            int8_t& dom, int16_t& doy, UErrorCode& status);
 
     /**
      * Convert a 1970-epoch day number to proleptic Gregorian year,
@@ -223,8 +236,24 @@ class Grego {
      * @param dow output parameter to receive day-of-week (1-based, 1==Sun)
      * @param status error code.
      */
-    static inline void dayToFields(int32_t day, int32_t& year, int32_t& month,
+    static void dayToFields(int32_t day, int32_t& year, int8_t& month,
-                                   int32_t& dom, int32_t& dow, UErrorCode& status);
+                            int8_t& dom, int8_t& dow, UErrorCode& status);
+
+    /**
+     * Convert a 1970-epoch day number to proleptic Gregorian year.
+     * @param day 1970-epoch day
+     * @param status error code.
+     * @return year.
+     */
+    static int32_t dayToYear(int32_t day, UErrorCode& status);
+    /**
+     * Convert a 1970-epoch day number to proleptic Gregorian year.
+     * @param day 1970-epoch day
+     * @param doy output parameter to receive day-of-year (1-based)
+     * @param status error code.
+     * @return year.
+     */
+    static int32_t dayToYear(int32_t day, int16_t& doy, UErrorCode& status);
 
     /**
      * Convert a 1970-epoch milliseconds to proleptic Gregorian year,
@@ -238,8 +267,43 @@ class Grego {
      * @param mid output parameter to receive millis-in-day
      * @param status error code.
      */
-    static void timeToFields(UDate time, int32_t& year, int32_t& month,
+    static void timeToFields(UDate time, int32_t& year, int8_t& month,
-                            int32_t& dom, int32_t& dow, int32_t& doy, int32_t& mid, UErrorCode& status);
+                            int8_t& dom, int8_t& dow, int16_t& doy, int32_t& mid, UErrorCode& status);
+
+    /**
+     * Convert a 1970-epoch milliseconds to proleptic Gregorian year,
+     * month, day-of-month, and day-of-week, day of year and millis-in-day.
+     * @param time 1970-epoch milliseconds
+     * @param year output parameter to receive year
+     * @param month output parameter to receive month (0-based, 0==Jan)
+     * @param dom output parameter to receive day-of-month (1-based)
+     * @param dow output parameter to receive day-of-week (1-based, 1==Sun)
+     * @param mid output parameter to receive millis-in-day
+     * @param status error code.
+     */
+    static void timeToFields(UDate time, int32_t& year, int8_t& month,
+                            int8_t& dom, int8_t& dow, int32_t& mid, UErrorCode& status);
+
+    /**
+     * Convert a 1970-epoch milliseconds to proleptic Gregorian year,
+     * month, day-of-month, and day-of-week, day of year and millis-in-day.
+     * @param time 1970-epoch milliseconds
+     * @param year output parameter to receive year
+     * @param month output parameter to receive month (0-based, 0==Jan)
+     * @param dom output parameter to receive day-of-month (1-based)
+     * @param mid output parameter to receive millis-in-day
+     * @param status error code.
+     */
+    static void timeToFields(UDate time, int32_t& year, int8_t& month,
+                            int8_t& dom, int32_t& mid, UErrorCode& status);
+
+    /**
+     * Convert a 1970-epoch milliseconds to proleptic Gregorian year.
+     * @param time 1970-epoch milliseconds
+     * @param status error code.
+     * @return year.
+     */
+    static int32_t timeToYear(UDate time, UErrorCode& status);
 
     /**
      * Return the day of week on the 1970-epoch day
@@ -305,12 +369,6 @@ Grego::previousMonthLength(int y, int m) {
   return (m > 0) ? monthLength(y, m-1) : 31;
 }
 
-inline void Grego::dayToFields(int32_t day, int32_t& year, int32_t& month,
-                               int32_t& dom, int32_t& dow, UErrorCode& status) {
-  int32_t doy_unused;
-  dayToFields(day,year,month,dom,dow,doy_unused, status);
-}
-
 inline double Grego::julianDayToMillis(int32_t julian)
 {
   return (static_cast<double>(julian) - kEpochStartAsJulianDay) * kOneDay;

deps/icu-small/source/i18n/hebrwcal.cpp

@@ -164,7 +164,6 @@ HebrewCalendar::HebrewCalendar(const Locale& aLocale, UErrorCode& success)
 :   Calendar(TimeZone::forLocaleOrDefault(aLocale), aLocale, success)
 
 {
-    setTimeInMillis(getNow(), success); // Call this again now that the vtable is set up properly.
 }
 
 
@@ -591,13 +590,8 @@ int32_t HebrewCalendar::handleGetMonthLength(int32_t extendedYear, int32_t month
 * Returns the number of days in the given Hebrew year
 * @internal
 */
-int32_t HebrewCalendar::handleGetYearLength(int32_t eyear) const {
+int32_t HebrewCalendar::handleGetYearLength(int32_t eyear, UErrorCode& status) const {
-    UErrorCode status = U_ZERO_ERROR;
+    return daysInYear(eyear, status);
-    int32_t len = daysInYear(eyear, status);
-    if (U_FAILURE(status)) {
-        return 12;
-    }
-    return len;
 }
 
 void HebrewCalendar::validateField(UCalendarDateFields field, UErrorCode &status) {

deps/icu-small/source/i18n/hebrwcal.h

@@ -326,9 +326,9 @@ public:
      * calendar system.  Subclasses should override this method if they can
      * provide a more correct or more efficient implementation than the
      * default implementation in Calendar.
-     * @stable ICU 2.0
+     * @internal
      */
-    virtual int32_t handleGetYearLength(int32_t eyear) const override;
+    virtual int32_t handleGetYearLength(int32_t eyear, UErrorCode& status) const override;
 
     /**
      * Subclasses may override this method to compute several fields

deps/icu-small/source/i18n/indiancal.cpp

@@ -41,7 +41,6 @@ IndianCalendar* IndianCalendar::clone() const {
 IndianCalendar::IndianCalendar(const Locale& aLocale, UErrorCode& success)
   :   Calendar(TimeZone::forLocaleOrDefault(aLocale), aLocale, success)
 {
-  setTimeInMillis(getNow(), success); // Call this again now that the vtable is set up properly.
 }
 
 IndianCalendar::IndianCalendar(const IndianCalendar& other) : Calendar(other) {
@@ -129,7 +128,8 @@ int32_t IndianCalendar::handleGetMonthLength(int32_t eyear, int32_t month, UErro
  *
  * @param eyear The year in Saka Era.
  */
-int32_t IndianCalendar::handleGetYearLength(int32_t eyear) const {
+int32_t IndianCalendar::handleGetYearLength(int32_t eyear, UErrorCode& status) const {
+    if (U_FAILURE(status)) return 0;
     return isGregorianLeap(eyear + INDIAN_ERA_START) ? 366 : 365;
 }
 /*
@@ -143,18 +143,6 @@ static double gregorianToJD(int32_t year, int32_t month, int32_t date) {
    return Grego::fieldsToDay(year, month, date) + kEpochStartAsJulianDay - 0.5;
 }
 
-/*
- * Returns the Gregorian Date corresponding to a given Julian Day
- * Month is 0 based.
- * @param jd The Julian Day
- */
-static int32_t* jdToGregorian(double jd, int32_t gregorianDate[3], UErrorCode& status) {
-   int32_t gdow;
-   Grego::dayToFields(jd - kEpochStartAsJulianDay,
-                      gregorianDate[0], gregorianDate[1], gregorianDate[2], gdow, status);
-   return gregorianDate;
-}
-
    
 //-------------------------------------------------------------------------
 // Functions for converting from field values to milliseconds....
@@ -266,10 +254,9 @@ int32_t IndianCalendar::handleGetExtendedYear(UErrorCode& status) {
 void IndianCalendar::handleComputeFields(int32_t julianDay, UErrorCode&  status) {
     double jdAtStartOfGregYear;
     int32_t leapMonth, IndianYear, yday, IndianMonth, IndianDayOfMonth, mday;
-    int32_t gregorianYear;      // Stores gregorian date corresponding to Julian day;
+    // Stores gregorian date corresponding to Julian day;
-    int32_t gd[3];
+    int32_t gregorianYear = Grego::dayToYear(julianDay - kEpochStartAsJulianDay, status);
 
-    gregorianYear = jdToGregorian(julianDay, gd, status)[0];          // Gregorian date for Julian day
     if (U_FAILURE(status)) return;
     IndianYear = gregorianYear - INDIAN_ERA_START;            // Year in Saka era
     jdAtStartOfGregYear = gregorianToJD(gregorianYear, 0, 1); // JD at start of Gregorian year

deps/icu-small/source/i18n/indiancal.h

@@ -215,7 +215,7 @@ public:
    * Return the number of days in the given Indian year
    * @internal
    */
-  virtual int32_t handleGetYearLength(int32_t extendedYear) const override;
+  virtual int32_t handleGetYearLength(int32_t extendedYear, UErrorCode& status) const override;
 
   //-------------------------------------------------------------------------
   // Functions for converting from field values to milliseconds....

deps/icu-small/source/i18n/islamcal.cpp

@@ -202,7 +202,6 @@ IslamicCalendar* IslamicCalendar::clone() const {
 IslamicCalendar::IslamicCalendar(const Locale& aLocale, UErrorCode& success)
 :   Calendar(TimeZone::forLocaleOrDefault(aLocale), aLocale, success)
 {
-    setTimeInMillis(getNow(), success); // Call this again now that the vtable is set up properly.
 }
 
 IslamicCalendar::~IslamicCalendar()
@@ -444,15 +443,8 @@ int32_t yearLength(int32_t extendedYear, UErrorCode& status) {
 * Return the number of days in the given Islamic year
 * @draft ICU 2.4
 */
-int32_t IslamicCalendar::handleGetYearLength(int32_t extendedYear) const {
+int32_t IslamicCalendar::handleGetYearLength(int32_t extendedYear, UErrorCode& status) const {
-    UErrorCode status = U_ZERO_ERROR;
+    return yearLength(extendedYear, status);
-    int32_t length = yearLength(extendedYear, status);
-    if (U_FAILURE(status)) {
-        // fallback to normal Islamic calendar length 355 day a year if we
-        // encounter error and cannot propagate.
-        return 355;
-    }
-    return length;
 }
 
 //-------------------------------------------------------------------------
@@ -706,7 +698,8 @@ int32_t IslamicCivilCalendar::handleGetMonthLength(int32_t extendedYear, int32_t
 * Return the number of days in the given Islamic year
 * @draft ICU 2.4
 */
-int32_t IslamicCivilCalendar::handleGetYearLength(int32_t extendedYear) const {
+int32_t IslamicCivilCalendar::handleGetYearLength(int32_t extendedYear, UErrorCode& status) const {
+    if (U_FAILURE(status)) return 0;
     return 354 + (civilLeapYear(extendedYear) ? 1 : 0);
 }
 
@@ -872,7 +865,7 @@ int32_t IslamicUmalquraCalendar::handleGetMonthLength(int32_t extendedYear, int3
 
 int32_t IslamicUmalquraCalendar::yearLength(int32_t extendedYear, UErrorCode& status) const {
     if (extendedYear<UMALQURA_YEAR_START || extendedYear>UMALQURA_YEAR_END) {
-        return IslamicCivilCalendar::handleGetYearLength(extendedYear);
+        return IslamicCivilCalendar::handleGetYearLength(extendedYear, status);
     }
     int length = 0;
     for(int i=0; i<12; i++) {
@@ -888,15 +881,8 @@ int32_t IslamicUmalquraCalendar::yearLength(int32_t extendedYear, UErrorCode& st
 * Return the number of days in the given Islamic year
 * @draft ICU 2.4
 */
-int32_t IslamicUmalquraCalendar::handleGetYearLength(int32_t extendedYear) const {
+int32_t IslamicUmalquraCalendar::handleGetYearLength(int32_t extendedYear, UErrorCode& status) const {
-    UErrorCode status = U_ZERO_ERROR;
+    return yearLength(extendedYear, status);
-    int32_t length = yearLength(extendedYear, status);
-    if (U_FAILURE(status)) {
-        // fallback to normal Islamic calendar length 355 day a year if we
-        // encounter error and cannot propagate.
-        return 355;
-    }
-    return length;
 }
 
 /**

deps/icu-small/source/i18n/islamcal.h

@@ -235,7 +235,7 @@ class U_I18N_API IslamicCalendar : public Calendar {
    * Return the number of days in the given Islamic year
    * @internal
    */
-  virtual int32_t handleGetYearLength(int32_t extendedYear) const override;
+  virtual int32_t handleGetYearLength(int32_t extendedYear, UErrorCode& status) const override;
     
   //-------------------------------------------------------------------------
   // Functions for converting from field values to milliseconds....
@@ -438,7 +438,7 @@ class U_I18N_API IslamicCivilCalendar : public IslamicCalendar {
    * Return the number of days in the given Islamic year
    * @internal
    */
-  virtual int32_t handleGetYearLength(int32_t extendedYear) const override;
+  virtual int32_t handleGetYearLength(int32_t extendedYear, UErrorCode& status) const override;
 
   /**
    * Override Calendar to compute several fields specific to the Islamic
@@ -621,7 +621,7 @@ class U_I18N_API IslamicUmalquraCalendar : public IslamicCivilCalendar {
    * Return the number of days in the given Islamic year
    * @internal
    */
-  virtual int32_t handleGetYearLength(int32_t extendedYear) const override;
+  virtual int32_t handleGetYearLength(int32_t extendedYear, UErrorCode& status) const override;
 
   /**
    * Override Calendar to compute several fields specific to the Islamic

deps/icu-small/source/i18n/japancal.cpp

@@ -115,7 +115,6 @@ JapaneseCalendar::JapaneseCalendar(const Locale& aLocale, UErrorCode& success)
 :   GregorianCalendar(aLocale, success)
 {
     init(success);
-    setTimeInMillis(getNow(), success); // Call this again now that the vtable is set up properly.
 }
 
 JapaneseCalendar::~JapaneseCalendar()
@@ -130,12 +129,6 @@ JapaneseCalendar::JapaneseCalendar(const JapaneseCalendar& source)
     U_ASSERT(U_SUCCESS(status));
 }
 
-JapaneseCalendar& JapaneseCalendar::operator= ( const JapaneseCalendar& right)
-{
-    GregorianCalendar::operator=(right);
-    return *this;
-}
-
 JapaneseCalendar* JapaneseCalendar::clone() const
 {
     return new JapaneseCalendar(*this);

deps/icu-small/source/i18n/japancal.h

@@ -104,13 +104,6 @@ public:
      */
     JapaneseCalendar(const JapaneseCalendar& source);
 
-    /**
-     * Default assignment operator
-     * @param right    the object to be copied.
-     * @internal
-     */
-    JapaneseCalendar& operator=(const JapaneseCalendar& right);
-
     /**
      * Create and return a polymorphic copy of this calendar.
      * @return    return a polymorphic copy of this calendar.

deps/icu-small/source/i18n/measunit.cpp

@@ -41,26 +41,26 @@ static const int32_t gOffsets[] = {
     2,
     7,
     17,
-    27,
+    28,
-    31,
+    32,
-    333,
+    334,
-    344,
+    345,
-    362,
+    363,
-    366,
+    367,
-    375,
+    376,
-    378,
+    379,
-    382,
+    383,
-    390,
+    391,
-    412,
+    413,
-    416,
+    417,
-    431,
     432,
-    438,
+    433,
-    449,
+    439,
-    455,
+    450,
-    459,
+    456,
-    461,
+    460,
-    495
+    462,
+    496
 };
 
 static const int32_t kCurrencyOffset = 5;
@@ -121,6 +121,7 @@ static const char * const gSubTypes[] = {
     "permille",
     "permillion",
     "permyriad",
+    "portion-per-1e9",
     "liter-per-100-kilometer",
     "liter-per-kilometer",
     "mile-per-gallon",
@@ -811,6 +812,14 @@ MeasureUnit MeasureUnit::getPermyriad() {
     return MeasureUnit(3, 9);
 }
 
+MeasureUnit *MeasureUnit::createPortionPer1E9(UErrorCode &status) {
+    return MeasureUnit::create(3, 10, status);
+}
+
+MeasureUnit MeasureUnit::getPortionPer1E9() {
+    return MeasureUnit(3, 10);
+}
+
 MeasureUnit *MeasureUnit::createLiterPer100Kilometers(UErrorCode &status) {
     return MeasureUnit::create(4, 0, status);
 }
@@ -2400,6 +2409,7 @@ MeasureUnitImpl MeasureUnitImpl::copy(UErrorCode &status) const {
     MeasureUnitImpl result;
     result.complexity = complexity;
     result.identifier.append(identifier, status);
+    result.constantDenominator = constantDenominator;
     for (int32_t i = 0; i < singleUnits.length(); i++) {
         SingleUnitImpl *item = result.singleUnits.emplaceBack(*singleUnits[i]);
         if (!item) {

deps/icu-small/source/i18n/measunit_extra.cpp

@@ -15,6 +15,7 @@
 #include "charstr.h"
 #include "cmemory.h"
 #include "cstring.h"
+#include "double-conversion-string-to-double.h"
 #include "measunit_impl.h"
 #include "resource.h"
 #include "uarrsort.h"
@@ -30,13 +31,15 @@
 #include "unicode/ustringtrie.h"
 #include "uresimp.h"
 #include "util.h"
+#include <limits.h>
 #include <cstdlib>
-
 U_NAMESPACE_BEGIN
 
 
 namespace {
 
+using icu::double_conversion::StringToDoubleConverter;
+
 // TODO: Propose a new error code for this?
 constexpr UErrorCode kUnitIdentifierSyntaxError = U_ILLEGAL_ARGUMENT_ERROR;
 
@@ -467,7 +470,30 @@ void U_CALLCONV initUnitExtras(UErrorCode& status) {
 
 class Token {
 public:
-    Token(int32_t match) : fMatch(match) {}
+  Token(int64_t match) : fMatch(match) {
+      if (fMatch < kCompoundPartOffset) {
+          this->fType = TYPE_PREFIX;
+      } else if (fMatch < kInitialCompoundPartOffset) {
+          this->fType = TYPE_COMPOUND_PART;
+      } else if (fMatch < kPowerPartOffset) {
+          this->fType = TYPE_INITIAL_COMPOUND_PART;
+      } else if (fMatch < kSimpleUnitOffset) {
+          this->fType = TYPE_POWER_PART;
+      } else {
+          this->fType = TYPE_SIMPLE_UNIT;
+      }
+  }
+
+  static Token constantToken(StringPiece str, UErrorCode &status) {
+      Token result;
+      auto value = Token::parseStringToLong(str, status);
+      if (U_FAILURE(status)) {
+          return result;
+      }
+      result.fMatch = value;
+      result.fType = TYPE_CONSTANT_DENOMINATOR;
+      return result;
+  }
 
   enum Type {
       TYPE_UNDEFINED,
@@ -478,25 +504,20 @@ public:
       TYPE_INITIAL_COMPOUND_PART,
       TYPE_POWER_PART,
       TYPE_SIMPLE_UNIT,
+      TYPE_CONSTANT_DENOMINATOR,
   };
 
   // Calling getType() is invalid, resulting in an assertion failure, if Token
   // value isn't positive.
   Type getType() const {
-        U_ASSERT(fMatch > 0);
+      U_ASSERT(fMatch >= 0);
-        if (fMatch < kCompoundPartOffset) {
+      return this->fType;
-            return TYPE_PREFIX;
   }
-        if (fMatch < kInitialCompoundPartOffset) {
+
-            return TYPE_COMPOUND_PART;
+  // Retrieve the value of the constant denominator if the token is of type TYPE_CONSTANT_DENOMINATOR.
-        }
+  uint64_t getConstantDenominator() const {
-        if (fMatch < kPowerPartOffset) {
+      U_ASSERT(getType() == TYPE_CONSTANT_DENOMINATOR);
-            return TYPE_INITIAL_COMPOUND_PART;
+      return static_cast<uint64_t>(fMatch);
-        }
-        if (fMatch < kSimpleUnitOffset) {
-            return TYPE_POWER_PART;
-        }
-        return TYPE_SIMPLE_UNIT;
   }
 
     UMeasurePrefix getUnitPrefix() const {
@@ -530,8 +551,41 @@ public:
         return fMatch - kSimpleUnitOffset;
     }
 
+    // TODO: Consider moving this to a separate utility class.
+    // Utility function to parse a string into an unsigned long value.
+    // The value must be a positive integer within the range [1, INT64_MAX].
+    // The input can be in integer or scientific notation.
+    static uint64_t parseStringToLong(const StringPiece strNum, UErrorCode &status) {
+        // We are processing well-formed input, so we don't need any special options to
+        // StringToDoubleConverter.
+        StringToDoubleConverter converter(0, 0, 0, "", "");
+        int32_t count;
+        double double_result = converter.StringToDouble(strNum.data(), strNum.length(), &count);
+        if (count != strNum.length()) {
+            status = kUnitIdentifierSyntaxError;
+            return 0;
+        }
+
+        if (U_FAILURE(status) || double_result < 1.0 || double_result > static_cast<double>(INT64_MAX)) {
+            status = kUnitIdentifierSyntaxError;
+            return 0;
+        }
+
+        // Check if the value is integer.
+        uint64_t int_result = static_cast<uint64_t>(double_result);
+        const double kTolerance = 1e-9;
+        if (abs(double_result - int_result) > kTolerance) {
+            status = kUnitIdentifierSyntaxError;
+            return 0;
+        }
+
+        return int_result;
+    }
+
 private:
-    int32_t fMatch;
+  Token() = default;
+  int64_t fMatch;
+  Type fType = TYPE_UNDEFINED;
 };
 
 class Parser {
@@ -555,6 +609,50 @@ public:
         return {source};
     }
 
+    /**
+     * A single unit or a constant denominator.
+     */
+    struct SingleUnitOrConstant {
+        enum ValueType {
+            kSingleUnit,
+            kConstantDenominator,
+        };
+
+        ValueType type = kSingleUnit;
+        SingleUnitImpl singleUnit;
+        uint64_t constantDenominator;
+
+        static SingleUnitOrConstant singleUnitValue(SingleUnitImpl singleUnit) {
+            SingleUnitOrConstant result;
+            result.type = kSingleUnit;
+            result.singleUnit = singleUnit;
+            result.constantDenominator = 0;
+            return result;
+        }
+
+        static SingleUnitOrConstant constantDenominatorValue(uint64_t constant) {
+            SingleUnitOrConstant result;
+            result.type = kConstantDenominator;
+            result.singleUnit = {};
+            result.constantDenominator = constant;
+            return result;
+        }
+
+        uint64_t getConstantDenominator() const {
+            U_ASSERT(type == kConstantDenominator);
+            return constantDenominator;
+        }
+
+        SingleUnitImpl getSingleUnit() const {
+            U_ASSERT(type == kSingleUnit);
+            return singleUnit;
+        }
+
+        bool isSingleUnit() const { return type == kSingleUnit; }
+
+        bool isConstantDenominator() const { return type == kConstantDenominator; }
+    };
+
     MeasureUnitImpl parse(UErrorCode& status) {
         MeasureUnitImpl result;
 
@@ -569,12 +667,19 @@ public:
         while (hasNext()) {
             bool sawAnd = false;
 
-            SingleUnitImpl singleUnit = nextSingleUnit(sawAnd, status);
+            auto singleUnitOrConstant = nextSingleUnitOrConstant(sawAnd, status);
             if (U_FAILURE(status)) {
                 return result;
             }
 
-            bool added = result.appendSingleUnit(singleUnit, status);
+            if (singleUnitOrConstant.isConstantDenominator()) {
+                result.constantDenominator = singleUnitOrConstant.getConstantDenominator();
+                result.complexity = UMEASURE_UNIT_COMPOUND;
+                continue;
+            }
+
+            U_ASSERT(singleUnitOrConstant.isSingleUnit());
+            bool added = result.appendSingleUnit(singleUnitOrConstant.getSingleUnit(), status);
             if (U_FAILURE(status)) {
                 return result;
             }
@@ -604,6 +709,12 @@ public:
             }
         }
 
+        if (result.singleUnits.length() == 0) {
+            // The identifier was empty or only had a constant denominator.
+            status = kUnitIdentifierSyntaxError;
+            return result; // add it for code consistency.
+        }
+
         return result;
     }
 
@@ -622,6 +733,10 @@ private:
     // identifier is invalid pending TODO(CLDR-13701).
     bool fAfterPer = false;
 
+    // Set to true when we've just seen a "per-". This is used to determine if
+    // the next token can be a constant denominator token.
+    bool fJustSawPer = false;
+
     Parser() : fSource(""), fTrie(u"") {}
 
     Parser(StringPiece source)
@@ -640,6 +755,10 @@ private:
         // Saves the position in the fSource string for the end of the most
         // recent matching token.
         int32_t previ = -1;
+
+        // Saves the position in the fSource string for later use in case of unit constant found.
+        int32_t currentFIndex = fIndex;
+
         // Find the longest token that matches a value in the trie:
         while (fIndex < fSource.length()) {
             auto result = fTrie.next(fSource.data()[fIndex++]);
@@ -658,14 +777,27 @@ private:
             // continue;
         }
 
-        if (match < 0) {
+        if (match >= 0) {
-            status = kUnitIdentifierSyntaxError;
-        } else {
             fIndex = previ;
-        }
             return {match};
         }
 
+        // If no match was found, we check if the token is a constant denominator.
+        // 1. We find the index of the start of the next token or the end of the string.
+        int32_t endOfConstantIndex = fSource.find("-", currentFIndex);
+        endOfConstantIndex = (endOfConstantIndex == -1) ? fSource.length() : endOfConstantIndex;
+        if (endOfConstantIndex <= currentFIndex) {
+            status = kUnitIdentifierSyntaxError;
+            return {match};
+        }
+
+        // 2. We extract the substring from the start of the constant to the end of the constant.
+        StringPiece constantDenominatorStr =
+            fSource.substr(currentFIndex, endOfConstantIndex - currentFIndex);
+        fIndex = endOfConstantIndex;
+        return Token::constantToken(constantDenominatorStr, status);
+    }
+
     /**
      * Returns the next "single unit" via result.
      *
@@ -680,10 +812,10 @@ private:
      * unit", sawAnd is set to true. If not, it is left as is.
      * @param status ICU error code.
      */
-    SingleUnitImpl nextSingleUnit(bool &sawAnd, UErrorCode &status) {
+    SingleUnitOrConstant nextSingleUnitOrConstant(bool &sawAnd, UErrorCode &status) {
-        SingleUnitImpl result;
+        SingleUnitImpl singleUnitResult;
         if (U_FAILURE(status)) {
-            return result;
+            return {};
         }
 
         // state:
@@ -695,19 +827,22 @@ private:
         bool atStart = fIndex == 0;
         Token token = nextToken(status);
         if (U_FAILURE(status)) {
-            return result;
+            return {};
         }
 
+        fJustSawPer = false;
+
         if (atStart) {
             // Identifiers optionally start with "per-".
             if (token.getType() == Token::TYPE_INITIAL_COMPOUND_PART) {
                 U_ASSERT(token.getInitialCompoundPart() == INITIAL_COMPOUND_PART_PER);
                 fAfterPer = true;
-                result.dimensionality = -1;
+                fJustSawPer = true;
+                singleUnitResult.dimensionality = -1;
 
                 token = nextToken(status);
                 if (U_FAILURE(status)) {
-                    return result;
+                    return {};
                 }
             }
         } else {
@@ -715,7 +850,7 @@ private:
             // via a compound part:
             if (token.getType() != Token::TYPE_COMPOUND_PART) {
                 status = kUnitIdentifierSyntaxError;
-                return result;
+                return {};
             }
 
             switch (token.getMatch()) {
@@ -724,15 +859,16 @@ private:
                     // Mixed compound units not yet supported,
                     // TODO(CLDR-13701).
                     status = kUnitIdentifierSyntaxError;
-                    return result;
+                    return {};
                 }
                 fAfterPer = true;
-                result.dimensionality = -1;
+                fJustSawPer = true;
+                singleUnitResult.dimensionality = -1;
                 break;
 
             case COMPOUND_PART_TIMES:
                 if (fAfterPer) {
-                    result.dimensionality = -1;
+                    singleUnitResult.dimensionality = -1;
                 }
                 break;
 
@@ -741,7 +877,7 @@ private:
                     // Can't start with "-and-", and mixed compound units
                     // not yet supported, TODO(CLDR-13701).
                     status = kUnitIdentifierSyntaxError;
-                    return result;
+                    return {};
                 }
                 sawAnd = true;
                 break;
@@ -749,52 +885,65 @@ private:
 
             token = nextToken(status);
             if (U_FAILURE(status)) {
-                return result;
+                return {};
             }
         }
 
+        if (token.getType() == Token::TYPE_CONSTANT_DENOMINATOR) {
+            if (!fJustSawPer) {
+                status = kUnitIdentifierSyntaxError;
+                return {};
+            }
+
+            return SingleUnitOrConstant::constantDenominatorValue(token.getConstantDenominator());
+        }
+
         // Read tokens until we have a complete SingleUnit or we reach the end.
         while (true) {
             switch (token.getType()) {
             case Token::TYPE_POWER_PART:
                 if (state > 0) {
                     status = kUnitIdentifierSyntaxError;
-                        return result;
+                    return {};
                 }
-                    result.dimensionality *= token.getPower();
+                singleUnitResult.dimensionality *= token.getPower();
                 state = 1;
                 break;
 
             case Token::TYPE_PREFIX:
                 if (state > 1) {
                     status = kUnitIdentifierSyntaxError;
-                        return result;
+                    return {};
                 }
-                    result.unitPrefix = token.getUnitPrefix();
+                singleUnitResult.unitPrefix = token.getUnitPrefix();
                 state = 2;
                 break;
 
             case Token::TYPE_SIMPLE_UNIT:
-                    result.index = token.getSimpleUnitIndex();
+                singleUnitResult.index = token.getSimpleUnitIndex();
-                    return result;
+                break;
 
             default:
                 status = kUnitIdentifierSyntaxError;
-                    return result;
+                return {};
+            }
+
+            if (token.getType() == Token::TYPE_SIMPLE_UNIT) {
+                break;
             }
 
             if (!hasNext()) {
                 // We ran out of tokens before finding a complete single unit.
                 status = kUnitIdentifierSyntaxError;
-                return result;
+                return {};
             }
             token = nextToken(status);
             if (U_FAILURE(status)) {
-                return result;
+                return {};
             }
         }
 
-        return result;
+        return SingleUnitOrConstant::singleUnitValue(singleUnitResult);
     }
 };
 
@@ -1120,6 +1269,51 @@ MeasureUnitImpl::extractIndividualUnitsWithIndices(UErrorCode &status) const {
     return result;
 }
 
+int32_t countCharacter(const CharString &str, char c) {
+    int32_t count = 0;
+    for (int32_t i = 0, n = str.length(); i < n; i++) {
+        if (str[i] == c) {
+            count++;
+        }
+    }
+    return count;
+}
+
+/**
+ * Internal function that returns a string of the constants in the correct
+ * format.
+ *
+ * Example:
+ * 1000 --> "-per-1000"
+ * 1000000 --> "-per-1e6"
+ *
+ * NOTE: this function is only used when the constant denominator is greater
+ * than 0.
+ */
+CharString getConstantsString(uint64_t constantDenominator, UErrorCode &status) {
+    U_ASSERT(constantDenominator > 0 && constantDenominator <= LLONG_MAX);
+
+    CharString result;
+    result.appendNumber(constantDenominator, status);
+    if (U_FAILURE(status)) {
+        return result;
+    }
+
+    if (constantDenominator <= 1000) {
+        return result;
+    }
+
+    // Check if the constant is a power of 10.
+    int32_t zeros = countCharacter(result, '0');
+    if (zeros == result.length() - 1 && result[0] == '1') {
+        result.clear();
+        result.append(StringPiece("1e"), status);
+        result.appendNumber(zeros, status);
+    }
+
+    return result;
+}
+
 /**
  * Normalize a MeasureUnitImpl and generate the identifier string in place.
  */
@@ -1128,7 +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;
     }
@@ -1145,6 +1339,7 @@ void MeasureUnitImpl::serialize(UErrorCode &status) {
     CharString result;
     bool beforePer = true;
     bool firstTimeNegativeDimension = false;
+    bool constantDenominatorAppended = false;
     for (int32_t i = 0; i < this->singleUnits.length(); i++) {
         if (beforePer && (*this->singleUnits[i]).dimensionality < 0) {
             beforePer = false;
@@ -1168,43 +1363,103 @@ void MeasureUnitImpl::serialize(UErrorCode &status) {
                 } else {
                     result.append(StringPiece("-per-"), status);
                 }
-            } else {
+
-                if (result.length() != 0) {
+                if (this->constantDenominator > 0) {
+                    result.append(getConstantsString(this->constantDenominator, status), status);
                     result.append(StringPiece("-"), status);
+                    constantDenominatorAppended = true;
                 }
+
+            } else if (result.length() != 0) {
+                result.append(StringPiece("-"), status);
             }
         }
 
         this->singleUnits[i]->appendNeutralIdentifier(result, status);
     }
 
+    if (!constantDenominatorAppended && this->constantDenominator > 0) {
+        result.append(StringPiece("-per-"), status);
+        result.append(getConstantsString(this->constantDenominator, status), status);
+    }
+
+    if (U_FAILURE(status)) {
+        return;
+    }
     this->identifier = CharString(result, status);
 }
 
-MeasureUnit MeasureUnitImpl::build(UErrorCode& status) && {
+MeasureUnit MeasureUnitImpl::build(UErrorCode &status) && {
     this->serialize(status);
     return MeasureUnit(std::move(*this));
 }
 
-MeasureUnit MeasureUnit::forIdentifier(StringPiece identifier, UErrorCode& status) {
+MeasureUnit MeasureUnit::forIdentifier(StringPiece identifier, UErrorCode &status) {
     return Parser::from(identifier, status).parse(status).build(status);
 }
 
-UMeasureUnitComplexity MeasureUnit::getComplexity(UErrorCode& status) const {
+UMeasureUnitComplexity MeasureUnit::getComplexity(UErrorCode &status) const {
     MeasureUnitImpl temp;
     return MeasureUnitImpl::forMeasureUnit(*this, temp, status).complexity;
 }
 
-UMeasurePrefix MeasureUnit::getPrefix(UErrorCode& status) const {
+UMeasurePrefix MeasureUnit::getPrefix(UErrorCode &status) const {
     return SingleUnitImpl::forMeasureUnit(*this, status).unitPrefix;
 }
 
-MeasureUnit MeasureUnit::withPrefix(UMeasurePrefix prefix, UErrorCode& status) const UPRV_NO_SANITIZE_UNDEFINED {
+MeasureUnit MeasureUnit::withPrefix(UMeasurePrefix prefix,
+                                    UErrorCode &status) const UPRV_NO_SANITIZE_UNDEFINED {
     SingleUnitImpl singleUnit = SingleUnitImpl::forMeasureUnit(*this, status);
     singleUnit.unitPrefix = prefix;
     return singleUnit.build(status);
 }
 
+uint64_t MeasureUnit::getConstantDenominator(UErrorCode &status) const {
+    auto complexity = this->getComplexity(status);
+    if (U_FAILURE(status)) {
+        return 0;
+    }
+
+    if (complexity != UMEASURE_UNIT_SINGLE && complexity != UMEASURE_UNIT_COMPOUND) {
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return 0;
+    }
+
+    if (this->fImpl == nullptr) {
+        return 0;
+    }
+
+    return this->fImpl->constantDenominator;
+}
+
+MeasureUnit MeasureUnit::withConstantDenominator(uint64_t denominator, UErrorCode &status) const {
+    // To match the behavior of the Java API, we do not allow a constant denominator
+    // bigger than LONG_MAX.
+    if (denominator > LONG_MAX) {
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return {};
+    }
+
+    auto complexity = this->getComplexity(status);
+    if (U_FAILURE(status)) {
+        return {};
+    }
+    if (complexity != UMEASURE_UNIT_SINGLE && complexity != UMEASURE_UNIT_COMPOUND) {
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return {};
+    }
+
+    MeasureUnitImpl impl = MeasureUnitImpl::forMeasureUnitMaybeCopy(*this, status);
+    if (U_FAILURE(status)) {
+        return {};
+    }
+
+    impl.constantDenominator = denominator;
+    impl.complexity = (impl.singleUnits.length() < 2 && denominator == 0) ? UMEASURE_UNIT_SINGLE
+                                                                          : UMEASURE_UNIT_COMPOUND;
+    return std::move(impl).build(status);
+}
+
 int32_t MeasureUnit::getDimensionality(UErrorCode& status) const {
     SingleUnitImpl singleUnit = SingleUnitImpl::forMeasureUnit(*this, status);
     if (U_FAILURE(status)) { return 0; }
@@ -1222,6 +1477,11 @@ MeasureUnit MeasureUnit::withDimensionality(int32_t dimensionality, UErrorCode&
 
 MeasureUnit MeasureUnit::reciprocal(UErrorCode& status) const {
     MeasureUnitImpl impl = MeasureUnitImpl::forMeasureUnitMaybeCopy(*this, status);
+    // The reciprocal of a unit that has a constant denominator is not allowed.
+    if (impl.constantDenominator != 0) {
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return {};
+    }
     impl.takeReciprocal(status);
     return std::move(impl).build(status);
 }
@@ -1237,9 +1497,25 @@ 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) {
+
+    uint64_t currentConstatDenominator = this->getConstantDenominator(status);
+    uint64_t otherConstantDenominator = other.getConstantDenominator(status);
+
+    // TODO: we can also multiply the constant denominators instead of returning an error.
+    if (currentConstatDenominator != 0 && otherConstantDenominator != 0) {
+        // There is only `one` constant denominator in a compound unit.
+        // Therefore, we Cannot multiply units that both of them have a constant denominator
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return {};
+    }
+
+    // Because either one of the constant denominators is zero, we can use the maximum of them.
+    impl.constantDenominator = uprv_max(currentConstatDenominator, otherConstantDenominator);
+
+    if (impl.singleUnits.length() > 1 || impl.constantDenominator > 0) {
         impl.complexity = UMEASURE_UNIT_COMPOUND;
     }
+
     return std::move(impl).build(status);
 }
 

deps/icu-small/source/i18n/measunit_impl.h

@@ -328,6 +328,14 @@ class U_I18N_API MeasureUnitImpl : public UMemory {
      */
     CharString identifier;
 
+    /**
+     * Represents the unit constant denominator.
+     *
+     * NOTE:
+     *   if set to 0, it means that the constant is not set.
+     */
+    uint64_t constantDenominator = 0;
+
     // For calling serialize
     // TODO(icu-units#147): revisit serialization
     friend class number::impl::LongNameHandler;

deps/icu-small/source/i18n/messageformat2.cpp

@@ -3,6 +3,8 @@
 
 #include "unicode/utypes.h"
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
@@ -11,8 +13,10 @@
 #include "unicode/messageformat2_data_model.h"
 #include "unicode/messageformat2_formattable.h"
 #include "unicode/messageformat2.h"
+#include "unicode/normalizer2.h"
 #include "unicode/unistr.h"
 #include "messageformat2_allocation.h"
+#include "messageformat2_checker.h"
 #include "messageformat2_evaluation.h"
 #include "messageformat2_macros.h"
 
@@ -37,7 +41,7 @@ static Formattable evalLiteral(const Literal& lit) {
         // The fallback for a variable name is itself.
         UnicodeString str(DOLLAR);
         str += var;
-        const Formattable* val = context.getGlobal(var, errorCode);
+        const Formattable* val = context.getGlobal(*this, var, errorCode);
         if (U_SUCCESS(errorCode)) {
             return (FormattedPlaceholder(*val, str));
         }
@@ -51,7 +55,7 @@ static Formattable evalLiteral(const Literal& lit) {
     return FormattedPlaceholder(evalLiteral(lit), lit.quoted());
 }
 
-[[nodiscard]] FormattedPlaceholder MessageFormatter::formatOperand(const Environment& env,
+[[nodiscard]] InternalValue* MessageFormatter::formatOperand(const Environment& env,
                                                             const Operand& rand,
                                                             MessageContext& context,
                                                             UErrorCode &status) const {
@@ -60,7 +64,7 @@ static Formattable evalLiteral(const Literal& lit) {
     }
 
     if (rand.isNull()) {
-        return FormattedPlaceholder();
+        return create<InternalValue>(InternalValue(FormattedPlaceholder()), status);
     }
     if (rand.isVariable()) {
         // Check if it's local or global
@@ -71,15 +75,19 @@ static Formattable evalLiteral(const Literal& lit) {
         // Eager vs. lazy evaluation is an open issue:
         // see https://github.com/unicode-org/message-format-wg/issues/299
 
+        // NFC-normalize the variable name. See
+        // https://github.com/unicode-org/message-format-wg/blob/main/spec/syntax.md#names-and-identifiers
+        const VariableName normalized = normalizeNFC(var);
+
         // Look up the variable in the environment
-        if (env.has(var)) {
+        if (env.has(normalized)) {
           // `var` is a local -- look it up
-          const Closure& rhs = env.lookup(var);
+          const Closure& rhs = env.lookup(normalized);
           // Format the expression using the environment from the closure
           return formatExpression(rhs.getEnv(), rhs.getExpr(), context, status);
         }
         // Variable wasn't found in locals -- check if it's global
-        FormattedPlaceholder result = evalArgument(var, context, status);
+        FormattedPlaceholder result = evalArgument(normalized, context, status);
         if (status == U_ILLEGAL_ARGUMENT_ERROR) {
             status = U_ZERO_ERROR;
             // Unbound variable -- set a resolution error
@@ -88,12 +96,12 @@ static Formattable evalLiteral(const Literal& lit) {
             // https://github.com/unicode-org/message-format-wg/blob/main/spec/formatting.md#fallback-resolution
             UnicodeString str(DOLLAR);
             str += var;
-            return FormattedPlaceholder(str);
+            return create<InternalValue>(InternalValue(FormattedPlaceholder(str)), status);
         }
-        return result;
+        return create<InternalValue>(InternalValue(std::move(result)), status);
     } else {
         U_ASSERT(rand.isLiteral());
-        return formatLiteral(rand.asLiteral());
+        return create<InternalValue>(InternalValue(formatLiteral(rand.asLiteral())), status);
     }
 }
 
@@ -114,28 +122,32 @@ FunctionOptions MessageFormatter::resolveOptions(const Environment& env, const O
 
         // Options are fully evaluated before calling the function
         // Format the operand
-        FormattedPlaceholder rhsVal = formatOperand(env, v, context, status);
+        LocalPointer<InternalValue> rhsVal(formatOperand(env, v, context, status));
         if (U_FAILURE(status)) {
             return {};
         }
-        if (!rhsVal.isFallback()) {
+        // Note: this means option values are "eagerly" evaluated.
-            resolvedOpt.adoptInstead(create<ResolvedFunctionOption>(ResolvedFunctionOption(k, rhsVal.asFormattable()), status));
+        // Currently, options don't have options. This will be addressed by the
+        // full FormattedPlaceholder redesign.
+        FormattedPlaceholder optValue = rhsVal->forceFormatting(context.getErrors(), status);
+        resolvedOpt.adoptInstead(create<ResolvedFunctionOption>
+                                 (ResolvedFunctionOption(k,
+                                                         optValue.asFormattable()),
+                                  status));
         if (U_FAILURE(status)) {
             return {};
         }
         optionsVector->adoptElement(resolvedOpt.orphan(), status);
     }
-    }
-
     return FunctionOptions(std::move(*optionsVector), status);
 }
 
 // Overload that dispatches on argument type. Syntax doesn't provide for options in this case.
-[[nodiscard]] FormattedPlaceholder MessageFormatter::evalFormatterCall(FormattedPlaceholder&& argument,
+[[nodiscard]] InternalValue* MessageFormatter::evalFunctionCall(FormattedPlaceholder&& argument,
                                                                 MessageContext& context,
                                                                 UErrorCode& status) const {
     if (U_FAILURE(status)) {
-        return {};
+        return nullptr;
     }
 
     // These cases should have been checked for already
@@ -153,8 +165,8 @@ FunctionOptions MessageFormatter::resolveOptions(const Environment& env, const O
             // No formatter for this type -- follow default behavior
             break;
         }
-        return evalFormatterCall(functionName,
+        return evalFunctionCall(functionName,
-                                 std::move(argument),
+                                create<InternalValue>(std::move(argument), status),
                                 FunctionOptions(),
                                 context,
                                 status);
@@ -167,12 +179,13 @@ FunctionOptions MessageFormatter::resolveOptions(const Environment& env, const O
     }
     // No formatter for this type, or it's a primitive type (which will be formatted later)
     // -- just return the argument itself
-    return std::move(argument);
+    return create<InternalValue>(std::move(argument), status);
 }
 
 // Overload that dispatches on function name
-[[nodiscard]] FormattedPlaceholder MessageFormatter::evalFormatterCall(const FunctionName& functionName,
+// Adopts `arg`
-                                                                 FormattedPlaceholder&& argument,
+[[nodiscard]] InternalValue* MessageFormatter::evalFunctionCall(const FunctionName& functionName,
+                                                                InternalValue* arg_,
                                                                 FunctionOptions&& options,
                                                                 MessageContext& context,
                                                                 UErrorCode& status) const {
@@ -180,69 +193,41 @@ FunctionOptions MessageFormatter::resolveOptions(const Environment& env, const O
         return {};
     }
 
-    DynamicErrors& errs = context.getErrors();
+    LocalPointer<InternalValue> arg(arg_);
 
+    // Look up the formatter or selector
+    LocalPointer<Formatter> formatterImpl(nullptr);
+    LocalPointer<Selector> selectorImpl(nullptr);
+    if (isFormatter(functionName)) {
+        formatterImpl.adoptInstead(getFormatter(functionName, status));
+        U_ASSERT(U_SUCCESS(status));
+    }
+    if (isSelector(functionName)) {
+        selectorImpl.adoptInstead(getSelector(context, functionName, status));
+        U_ASSERT(U_SUCCESS(status));
+    }
+    if (formatterImpl == nullptr && selectorImpl == nullptr) {
+        // Unknown function error
+        context.getErrors().setUnknownFunction(functionName, status);
+
+        if (arg->hasNullOperand()) {
+            // Non-selector used as selector; an error would have been recorded earlier
             UnicodeString fallback(COLON);
             fallback += functionName;
-    if (!argument.isNullOperand()) {
+            return new InternalValue(FormattedPlaceholder(fallback));
-        fallback = argument.fallback;
-    }
-
-    if (isFormatter(functionName)) {
-        LocalPointer<Formatter> formatterImpl(getFormatter(functionName, status));
-        if (U_FAILURE(status)) {
-            if (status == U_MF_FORMATTING_ERROR) {
-                errs.setFormattingError(functionName, status);
-                status = U_ZERO_ERROR;
-                return {};
-            }
-            if (status == U_MF_UNKNOWN_FUNCTION_ERROR) {
-                errs.setUnknownFunction(functionName, status);
-                status = U_ZERO_ERROR;
-                return {};
-            }
-            // Other errors are non-recoverable
-            return {};
-        }
-        U_ASSERT(formatterImpl != nullptr);
-
-        UErrorCode savedStatus = status;
-        FormattedPlaceholder result = formatterImpl->format(std::move(argument), std::move(options), status);
-        // Update errors
-        if (savedStatus != status) {
-            if (U_FAILURE(status)) {
-                if (status == U_MF_OPERAND_MISMATCH_ERROR) {
-                    status = U_ZERO_ERROR;
-                    errs.setOperandMismatchError(functionName, status);
         } else {
-                    status = U_ZERO_ERROR;
+            return new InternalValue(FormattedPlaceholder(arg->getFallback()));
-                    // Convey any error generated by the formatter
-                    // as a formatting error, except for operand mismatch errors
-                    errs.setFormattingError(functionName, status);
-                }
-                return FormattedPlaceholder(fallback);
-            } else {
-                // Ignore warnings
-                status = savedStatus;
         }
     }
-        // Ignore the output if any errors occurred
+    return new InternalValue(arg.orphan(),
-        if (errs.hasFormattingError()) {
+                             std::move(options),
-            return FormattedPlaceholder(fallback);
+                             functionName,
-        }
+                             formatterImpl.isValid() ? formatterImpl.orphan() : nullptr,
-        return result;
+                             selectorImpl.isValid() ? selectorImpl.orphan() : nullptr);
-    }
-    // No formatter with this name -- set error
-    if (isSelector(functionName)) {
-        errs.setFormattingError(functionName, status);
-    } else {
-        errs.setUnknownFunction(functionName, status);
-    }
-    return FormattedPlaceholder(fallback);
 }
 
 // Formats an expression using `globalEnv` for the values of variables
-[[nodiscard]] FormattedPlaceholder MessageFormatter::formatExpression(const Environment& globalEnv,
+[[nodiscard]] InternalValue* MessageFormatter::formatExpression(const Environment& globalEnv,
                                                                const Expression& expr,
                                                                MessageContext& context,
                                                                UErrorCode &status) const {
@@ -252,19 +237,18 @@ FunctionOptions MessageFormatter::resolveOptions(const Environment& env, const O
 
     const Operand& rand = expr.getOperand();
     // Format the operand (formatOperand handles the case of a null operand)
-    FormattedPlaceholder randVal = formatOperand(globalEnv, rand, context, status);
+    LocalPointer<InternalValue> randVal(formatOperand(globalEnv, rand, context, status));
 
-    // Don't call the function on error values
+    FormattedPlaceholder maybeRand = randVal->takeArgument(status);
-    if (randVal.isFallback()) {
-        return randVal;
-    }
 
-    if (!expr.isFunctionCall()) {
+    if (!expr.isFunctionCall() && U_SUCCESS(status)) {
         // Dispatch based on type of `randVal`
-        return evalFormatterCall(std::move(randVal),
+         if (maybeRand.isFallback()) {
-                                 context,
+            return randVal.orphan();
-                                 status);
+        }
-    } else {
+        return evalFunctionCall(std::move(maybeRand), context, status);
+    } else if (expr.isFunctionCall()) {
+        status = U_ZERO_ERROR;
         const Operator* rator = expr.getOperator(status);
         U_ASSERT(U_SUCCESS(status));
         const FunctionName& functionName = rator->getFunctionName();
@@ -273,19 +257,14 @@ FunctionOptions MessageFormatter::resolveOptions(const Environment& env, const O
         FunctionOptions resolvedOptions = resolveOptions(globalEnv, options, context, status);
 
         // Call the formatter function
-        // The fallback for a nullary function call is the function name
+        return evalFunctionCall(functionName,
-        UnicodeString fallback;
+                                randVal.orphan(),
-        if (rand.isNull()) {
-            fallback = UnicodeString(COLON);
-            fallback += functionName;
-        } else {
-            fallback = randVal.fallback;
-        }
-        return evalFormatterCall(functionName,
-                                 std::move(randVal),
                                 std::move(resolvedOptions),
                                 context,
                                 status);
+    } else {
+        status = U_ZERO_ERROR;
+        return randVal.orphan();
     }
 }
 
@@ -301,11 +280,13 @@ void MessageFormatter::formatPattern(MessageContext& context, const Environment&
             // Markup is ignored
         } else {
 	      // Format the expression
-	      FormattedPlaceholder partVal = formatExpression(globalEnv, part.contents(), context, status);
+              LocalPointer<InternalValue> partVal(
+                  formatExpression(globalEnv, part.contents(), context, status));
+              FormattedPlaceholder partResult = partVal->forceFormatting(context.getErrors(),
+                                                                         status);
               // Force full evaluation, e.g. applying default formatters to
 	      // unformatted input (or formatting numbers as strings)
-              UnicodeString partResult = partVal.formatToString(locale, status);
+              result += partResult.formatToString(locale, status);
-              result += partResult;
               // Handle formatting errors. `formatToString()` can't take a context and thus can't
               // register an error directly
               if (status == U_MF_FORMATTING_ERROR) {
@@ -328,14 +309,14 @@ void MessageFormatter::resolveSelectors(MessageContext& context, const Environme
     CHECK_ERROR(status);
     U_ASSERT(!dataModel.hasPattern());
 
-    const Expression* selectors = dataModel.getSelectorsInternal();
+    const VariableName* selectors = dataModel.getSelectorsInternal();
     // 1. Let res be a new empty list of resolved values that support selection.
     // (Implicit, since `res` is an out-parameter)
     // 2. For each expression exp of the message's selectors
     for (int32_t i = 0; i < dataModel.numSelectors(); i++) {
         // 2i. Let rv be the resolved value of exp.
-        ResolvedSelector rv = formatSelectorExpression(env, selectors[i], context, status);
+        LocalPointer<InternalValue> rv(formatOperand(env, Operand(selectors[i]), context, status));
-        if (rv.hasSelector()) {
+        if (rv->canSelect()) {
             // 2ii. If selection is supported for rv:
             // (True if this code has been reached)
         } else {
@@ -344,17 +325,17 @@ void MessageFormatter::resolveSelectors(MessageContext& context, const Environme
             // Append nomatch as the last element of the list res.
             // Emit a Selection Error.
             // (Note: in this case, rv, being a fallback, serves as `nomatch`)
-            #if U_DEBUG
+            DynamicErrors& err = context.getErrors();
-            const DynamicErrors& err = context.getErrors();
+            err.setSelectorError(rv->getFunctionName(), status);
-            U_ASSERT(err.hasError());
+            rv.adoptInstead(new InternalValue(FormattedPlaceholder(rv->getFallback())));
-            U_ASSERT(rv.argument().isFallback());
+            if (!rv.isValid()) {
-            #endif
+                status = U_MEMORY_ALLOCATION_ERROR;
+                return;
+            }
         }
         // 2ii(a). Append rv as the last element of the list res.
         // (Also fulfills 2iii)
-        LocalPointer<ResolvedSelector> v(create<ResolvedSelector>(std::move(rv), status));
+        res.adoptElement(rv.orphan(), status);
-        CHECK_ERROR(status);
-        res.adoptElement(v.orphan(), status);
     }
 }
 
@@ -362,18 +343,17 @@ void MessageFormatter::resolveSelectors(MessageContext& context, const Environme
 // `keys` and `matches` are vectors of strings
 void MessageFormatter::matchSelectorKeys(const UVector& keys,
                                          MessageContext& context,
-					 ResolvedSelector&& rv,
+					 InternalValue* rv, // Does not adopt `rv`
 					 UVector& keysOut,
 					 UErrorCode& status) const {
     CHECK_ERROR(status);
 
-    if (!rv.hasSelector()) {
+    if (U_FAILURE(status)) {
         // Return an empty list of matches
+        status = U_ZERO_ERROR;
         return;
     }
 
-    auto selectorImpl = rv.getSelector();
-    U_ASSERT(selectorImpl != nullptr);
     UErrorCode savedStatus = status;
 
     // Convert `keys` to an array
@@ -400,15 +380,17 @@ void MessageFormatter::matchSelectorKeys(const UVector& keys,
     int32_t prefsLen = 0;
 
     // Call the selector
-    selectorImpl->selectKey(rv.takeArgument(), rv.takeOptions(),
+    FunctionName name = rv->getFunctionName();
-                            adoptedKeys.getAlias(), keysLen, adoptedPrefs.getAlias(), prefsLen,
+    rv->forceSelection(context.getErrors(),
+                       adoptedKeys.getAlias(), keysLen,
+                       adoptedPrefs.getAlias(), prefsLen,
                        status);
 
     // Update errors
     if (savedStatus != status) {
         if (U_FAILURE(status)) {
             status = U_ZERO_ERROR;
-            context.getErrors().setSelectorError(rv.getSelectorName(), status);
+            context.getErrors().setSelectorError(name, status);
         } else {
             // Ignore warnings
             status = savedStatus;
@@ -461,8 +443,8 @@ void MessageFormatter::resolvePreferences(MessageContext& context, UVector& res,
             if (!key.isWildcard()) {
                 // 2ii(b)(a) Assert that key is a literal.
                 // (Not needed)
-                // 2ii(b)(b) Let `ks` be the resolved value of `key`.
+                // 2ii(b)(b) Let `ks` be the resolved value of `key` in Unicode Normalization Form C.
-                ks = key.asLiteral().unquoted();
+                ks = normalizeNFC(key.asLiteral().unquoted());
                 // 2ii(b)(c) Append `ks` as the last element of the list `keys`.
                 ksP.adoptInstead(create<UnicodeString>(std::move(ks), status));
                 CHECK_ERROR(status);
@@ -471,7 +453,7 @@ void MessageFormatter::resolvePreferences(MessageContext& context, UVector& res,
         }
         // 2iii. Let `rv` be the resolved value at index `i` of `res`.
         U_ASSERT(i < res.size());
-        ResolvedSelector rv = std::move(*(static_cast<ResolvedSelector*>(res[i])));
+        InternalValue* rv = static_cast<InternalValue*>(res[i]);
         // 2iv. Let matches be the result of calling the method MatchSelectorKeys(rv, keys)
         LocalPointer<UVector> matches(createUVector(status));
         matchSelectorKeys(*keys, context, std::move(rv), *matches, status);
@@ -523,7 +505,7 @@ void MessageFormatter::filterVariants(const UVector& pref, UVector& vars, UError
             // 2i(c). Assert that `key` is a literal.
             // (Not needed)
             // 2i(d). Let `ks` be the resolved value of `key`.
-            UnicodeString ks = key.asLiteral().unquoted();
+            UnicodeString ks = normalizeNFC(key.asLiteral().unquoted());
             // 2i(e). Let `matches` be the list of strings at index `i` of `pref`.
             const UVector& matches = *(static_cast<UVector*>(pref[i])); // `matches` is a vector of strings
             // 2i(f). If `matches` includes `ks`
@@ -585,7 +567,7 @@ void MessageFormatter::sortVariants(const UVector& pref, UVector& vars, UErrorCo
                 // 5iii(c)(a). Assert that `key` is a literal.
                 // (Not needed)
                 // 5iii(c)(b). Let `ks` be the resolved value of `key`.
-                UnicodeString ks = key.asLiteral().unquoted();
+                UnicodeString ks = normalizeNFC(key.asLiteral().unquoted());
                 // 5iii(c)(c) Let matchpref be the integer position of ks in `matches`.
                 matchpref = vectorFind(matches, ks);
                 U_ASSERT(matchpref >= 0);
@@ -604,123 +586,13 @@ void MessageFormatter::sortVariants(const UVector& pref, UVector& vars, UErrorCo
     // 7. Select the pattern of `var`
 }
 
-
-// Evaluate the operand
-ResolvedSelector MessageFormatter::resolveVariables(const Environment& env, const Operand& rand, MessageContext& context, UErrorCode &status) const {
-    if (U_FAILURE(status)) {
-        return {};
-    }
-
-    if (rand.isNull()) {
-        return ResolvedSelector(FormattedPlaceholder());
-    }
-
-    if (rand.isLiteral()) {
-        return ResolvedSelector(formatLiteral(rand.asLiteral()));
-    }
-
-    // Must be variable
-    const VariableName& var = rand.asVariable();
-    // Resolve the variable
-    if (env.has(var)) {
-        const Closure& referent = env.lookup(var);
-        // Resolve the referent
-        return resolveVariables(referent.getEnv(), referent.getExpr(), context, status);
-    }
-    // Either this is a global var or an unbound var --
-    // either way, it can't be bound to a function call.
-    // Check globals
-    FormattedPlaceholder val = evalArgument(var, context, status);
-    if (status == U_ILLEGAL_ARGUMENT_ERROR) {
-        status = U_ZERO_ERROR;
-        // Unresolved variable -- could be a previous warning. Nothing to resolve
-        U_ASSERT(context.getErrors().hasUnresolvedVariableError());
-        return ResolvedSelector(FormattedPlaceholder(var));
-    }
-    // Pass through other errors
-    return ResolvedSelector(std::move(val));
-}
-
-// Evaluate the expression except for not performing the top-level function call
-// (which is expected to be a selector, but may not be, in error cases)
-ResolvedSelector MessageFormatter::resolveVariables(const Environment& env,
-                                                    const Expression& expr,
-                                                    MessageContext& context,
-                                                    UErrorCode &status) const {
-    if (U_FAILURE(status)) {
-        return {};
-    }
-
-    // Function call -- resolve the operand and options
-    if (expr.isFunctionCall()) {
-        const Operator* rator = expr.getOperator(status);
-        U_ASSERT(U_SUCCESS(status));
-        // Already checked that rator is non-reserved
-        const FunctionName& selectorName = rator->getFunctionName();
-        if (isSelector(selectorName)) {
-            auto selector = getSelector(context, selectorName, status);
-            if (U_SUCCESS(status)) {
-                FunctionOptions resolvedOptions = resolveOptions(env, rator->getOptionsInternal(), context, status);
-                // Operand may be the null argument, but resolveVariables() handles that
-                FormattedPlaceholder argument = formatOperand(env, expr.getOperand(), context, status);
-                return ResolvedSelector(selectorName, selector, std::move(resolvedOptions), std::move(argument));
-            }
-        } else if (isFormatter(selectorName)) {
-            context.getErrors().setSelectorError(selectorName, status);
-        } else {
-            context.getErrors().setUnknownFunction(selectorName, status);
-        }
-        // Non-selector used as selector; an error would have been recorded earlier
-        UnicodeString fallback(COLON);
-        fallback += selectorName;
-        if (!expr.getOperand().isNull()) {
-            fallback = formatOperand(env, expr.getOperand(), context, status).fallback;
-        }
-        return ResolvedSelector(FormattedPlaceholder(fallback));
-    } else {
-        // Might be a variable reference, so expand one more level of variable
-        return resolveVariables(env, expr.getOperand(), context, status);
-    }
-}
-
-ResolvedSelector MessageFormatter::formatSelectorExpression(const Environment& globalEnv, const Expression& expr, MessageContext& context, UErrorCode &status) const {
-    if (U_FAILURE(status)) {
-        return {};
-    }
-
-    // Resolve expression to determine if it's a function call
-    ResolvedSelector exprResult = resolveVariables(globalEnv, expr, context, status);
-
-    DynamicErrors& err = context.getErrors();
-
-    // If there is a selector, then `resolveVariables()` recorded it in the context
-    if (exprResult.hasSelector()) {
-        // Check if there was an error
-        if (exprResult.argument().isFallback()) {
-            // Use a null expression if it's a syntax or data model warning;
-            // create a valid (non-fallback) formatted placeholder from the
-            // fallback string otherwise
-            if (err.hasSyntaxError() || err.hasDataModelError()) {
-                return ResolvedSelector(FormattedPlaceholder()); // Null operand
-            } else {
-                return ResolvedSelector(exprResult.takeArgument());
-            }
-        }
-        return exprResult;
-    }
-
-    // No selector was found; error should already have been set
-    U_ASSERT(err.hasMissingSelectorAnnotationError() || err.hasUnknownFunctionError() || err.hasSelectorError());
-    return ResolvedSelector(FormattedPlaceholder(exprResult.argument().fallback));
-}
-
 void MessageFormatter::formatSelectors(MessageContext& context, const Environment& env, UErrorCode &status, UnicodeString& result) const {
     CHECK_ERROR(status);
 
     // See https://github.com/unicode-org/message-format-wg/blob/main/spec/formatting.md#pattern-selection
 
     // Resolve Selectors
-    // res is a vector of FormattedPlaceholders
+    // res is a vector of InternalValues
     LocalPointer<UVector> res(createUVector(status));
     CHECK_ERROR(status);
     resolveSelectors(context, env, status, *res);
@@ -761,16 +633,21 @@ void MessageFormatter::formatSelectors(MessageContext& context, const Environmen
 UnicodeString MessageFormatter::formatToString(const MessageArguments& arguments, UErrorCode &status) {
     EMPTY_ON_ERROR(status);
 
-    // Create a new environment that will store closures for all local variables
-    Environment* env = Environment::create(status);
     // Create a new context with the given arguments and the `errors` structure
     MessageContext context(arguments, *errors, status);
+    UnicodeString result;
 
+    if (!(errors->hasSyntaxError() || errors->hasDataModelError())) {
+        // Create a new environment that will store closures for all local variables
         // Check for unresolved variable errors
+        // checkDeclarations needs a reference to the pointer to the environment
+        // since it uses its `env` argument as an out-parameter. So it needs to be
+        // temporarily not a LocalPointer...
+        Environment* env(Environment::create(status));
         checkDeclarations(context, env, status);
+        // ...and then it's adopted to avoid leaks
         LocalPointer<Environment> globalEnv(env);
 
-    UnicodeString result;
         if (dataModel.hasPattern()) {
             formatPattern(context, *globalEnv, dataModel.getPattern(), status, result);
         } else {
@@ -783,6 +660,8 @@ UnicodeString MessageFormatter::formatToString(const MessageArguments& arguments
                 formatSelectors(context, *globalEnv, status, result);
             }
         }
+    }
+
     // Update status according to all errors seen while formatting
     if (signalErrors) {
         context.checkErrors(status);
@@ -813,12 +692,14 @@ void MessageFormatter::check(MessageContext& context, const Environment& localEn
 
     // Check that variable is in scope
     const VariableName& var = rand.asVariable();
+    UnicodeString normalized = normalizeNFC(var);
+
     // Check local scope
-    if (localEnv.has(var)) {
+    if (localEnv.has(normalized)) {
         return;
     }
     // Check global scope
-    context.getGlobal(var, status);
+    context.getGlobal(*this, normalized, status);
     if (status == U_ILLEGAL_ARGUMENT_ERROR) {
         status = U_ZERO_ERROR;
         context.getErrors().setUnresolvedVariable(var, status);
@@ -855,7 +736,10 @@ void MessageFormatter::checkDeclarations(MessageContext& context, Environment*&
         // memoizing the value of localEnv up to this point
 
         // Add the LHS to the environment for checking the next declaration
-        env = Environment::create(decl.getVariable(), Closure(rhs, *env), env, status);
+        env = Environment::create(normalizeNFC(decl.getVariable()),
+                                  Closure(rhs, *env),
+                                  env,
+                                  status);
         CHECK_ERROR(status);
     }
 }
@@ -866,3 +750,5 @@ U_NAMESPACE_END
 #endif /* #if !UCONFIG_NO_MF2 */
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
+
+#endif /* #if !UCONFIG_NO_NORMALIZATION */

deps/icu-small/source/i18n/messageformat2_allocation.h

@@ -10,6 +10,8 @@
 
 #if U_SHOW_CPLUSPLUS_API
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
@@ -139,6 +141,8 @@ U_NAMESPACE_END
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
 
+#endif /* #if !UCONFIG_NO_NORMALIZATION */
+
 #endif /* U_SHOW_CPLUSPLUS_API */
 
 #endif // MESSAGEFORMAT2_UTILS_H

deps/icu-small/source/i18n/messageformat2_arguments.cpp

@@ -3,12 +3,16 @@
 
 #include "unicode/utypes.h"
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
 
+#include "unicode/messageformat2.h"
 #include "unicode/messageformat2_arguments.h"
 #include "unicode/messageformat2_data_model_names.h"
+#include "messageformat2_evaluation.h"
 #include "uvector.h" // U_ASSERT
 
 U_NAMESPACE_BEGIN
@@ -22,11 +26,15 @@ namespace message2 {
 
     using Arguments = MessageArguments;
 
-    const Formattable* Arguments::getArgument(const VariableName& arg, UErrorCode& errorCode) const {
+    const Formattable* Arguments::getArgument(const MessageFormatter& context,
+                                              const VariableName& arg,
+                                              UErrorCode& errorCode) const {
         if (U_SUCCESS(errorCode)) {
             U_ASSERT(argsLen == 0 || arguments.isValid());
             for (int32_t i = 0; i < argsLen; i++) {
-                if (argumentNames[i] == arg) {
+                UnicodeString normalized = context.normalizeNFC(argumentNames[i]);
+                // arg already assumed to be normalized
+                if (normalized == arg) {
                     return &arguments[i];
                 }
             }
@@ -57,3 +65,5 @@ U_NAMESPACE_END
 #endif /* #if !UCONFIG_NO_MF2 */
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
+
+#endif /* #if !UCONFIG_NO_NORMALIZATION */

deps/icu-small/source/i18n/messageformat2_checker.cpp

@@ -3,12 +3,16 @@
 
 #include "unicode/utypes.h"
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
 
+#include "unicode/messageformat2.h"
 #include "messageformat2_allocation.h"
 #include "messageformat2_checker.h"
+#include "messageformat2_evaluation.h"
 #include "messageformat2_macros.h"
 #include "uvector.h" // U_ASSERT
 
@@ -104,6 +108,14 @@ TypeEnvironment::~TypeEnvironment() {}
 
 // ---------------------
 
+Key Checker::normalizeNFC(const Key& k) const {
+    if (k.isWildcard()) {
+        return k;
+    }
+    return Key(Literal(k.asLiteral().isQuoted(),
+                       context.normalizeNFC(k.asLiteral().unquoted())));
+}
+
 static bool areDefaultKeys(const Key* keys, int32_t len) {
     U_ASSERT(len > 0);
     for (int32_t i = 0; i < len; i++) {
@@ -185,7 +197,7 @@ void Checker::checkVariants(UErrorCode& status) {
                 // This variant was already checked,
                 // so we know keys1.len == len
                 for (int32_t kk = 0; kk < len; kk++) {
-                    if (!(keys[kk] == keys1[kk])) {
+                    if (!(normalizeNFC(keys[kk]) == normalizeNFC(keys1[kk]))) {
                         allEqual = false;
                         break;
                     }
@@ -205,18 +217,14 @@ void Checker::checkVariants(UErrorCode& status) {
     }
 }
 
-void Checker::requireAnnotated(const TypeEnvironment& t, const Expression& selectorExpr, UErrorCode& status) {
+void Checker::requireAnnotated(const TypeEnvironment& t,
+                               const VariableName& selectorVar,
+                               UErrorCode& status) {
     CHECK_ERROR(status);
 
-    if (selectorExpr.isFunctionCall()) {
+    if (t.get(selectorVar) == TypeEnvironment::Type::Annotated) {
         return; // No error
     }
-    const Operand& rand = selectorExpr.getOperand();
-    if (rand.isVariable()) {
-        if (t.get(rand.asVariable()) == TypeEnvironment::Type::Annotated) {
-            return; // No error
-        }
-    }
     // If this code is reached, an error was detected
     errors.addError(StaticErrorType::MissingSelectorAnnotation, status);
 }
@@ -226,7 +234,7 @@ void Checker::checkSelectors(const TypeEnvironment& t, UErrorCode& status) {
 
     // Check each selector; if it's not annotated, emit a
     // "missing selector annotation" error
-    const Expression* selectors = dataModel.getSelectorsInternal();
+    const VariableName* selectors = dataModel.getSelectorsInternal();
     for (int32_t i = 0; i < dataModel.numSelectors(); i++) {
         requireAnnotated(t, selectors[i], status);
     }
@@ -312,3 +320,5 @@ U_NAMESPACE_END
 #endif /* #if !UCONFIG_NO_MF2 */
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
+
+#endif /* #if !UCONFIG_NO_NORMALIZATION */

deps/icu-small/source/i18n/messageformat2_checker.h

@@ -10,6 +10,8 @@
 
 #if U_SHOW_CPLUSPLUS_API
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
@@ -56,15 +58,20 @@ namespace message2 {
                                         // an explicit declaration
     }; // class TypeEnvironment
 
+    class MessageFormatter;
+
     // Checks a data model for semantic errors
     // (Errors are defined in https://github.com/unicode-org/message-format-wg/blob/main/spec/formatting.md       )
     class Checker {
     public:
         void check(UErrorCode&);
-        Checker(const MFDataModel& m, StaticErrors& e) : dataModel(m), errors(e) {}
+        Checker(const MFDataModel& d, StaticErrors& e, const MessageFormatter& mf)
+            : dataModel(d), errors(e), context(mf) {}
     private:
 
-        void requireAnnotated(const TypeEnvironment&, const Expression&, UErrorCode&);
+        Key normalizeNFC(const Key&) const;
+
+        void requireAnnotated(const TypeEnvironment&, const VariableName&, UErrorCode&);
         void addFreeVars(TypeEnvironment& t, const Operand&, UErrorCode&);
         void addFreeVars(TypeEnvironment& t, const Operator&, UErrorCode&);
         void addFreeVars(TypeEnvironment& t, const OptionMap&, UErrorCode&);
@@ -78,6 +85,9 @@ namespace message2 {
         void check(const Pattern&);
         const MFDataModel& dataModel;
         StaticErrors& errors;
+
+        // Used for NFC normalization
+        const MessageFormatter& context;
     }; // class Checker
 
 } // namespace message2
@@ -88,6 +98,8 @@ U_NAMESPACE_END
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
 
+#endif /* #if !UCONFIG_NO_NORMALIZATION */
+
 #endif /* U_SHOW_CPLUSPLUS_API */
 
 #endif // MESSAGEFORMAT_CHECKER_H

deps/icu-small/source/i18n/messageformat2_data_model.cpp

@@ -3,6 +3,8 @@
 
 #include "unicode/utypes.h"
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
@@ -691,7 +693,7 @@ Matcher::Matcher(const Matcher& other) {
     numSelectors = other.numSelectors;
     numVariants = other.numVariants;
     UErrorCode localErrorCode = U_ZERO_ERROR;
-    selectors.adoptInstead(copyArray<Expression>(other.selectors.getAlias(),
+    selectors.adoptInstead(copyArray<VariableName>(other.selectors.getAlias(),
                                                    numSelectors,
                                                    localErrorCode));
     variants.adoptInstead(copyArray<Variant>(other.variants.getAlias(),
@@ -702,7 +704,7 @@ Matcher::Matcher(const Matcher& other) {
     }
 }
 
-Matcher::Matcher(Expression* ss, int32_t ns, Variant* vs, int32_t nv)
+Matcher::Matcher(VariableName* ss, int32_t ns, Variant* vs, int32_t nv)
     : selectors(ss), numSelectors(ns), variants(vs), numVariants(nv) {}
 
 Matcher::~Matcher() {}
@@ -724,7 +726,7 @@ const Binding* MFDataModel::getLocalVariablesInternal() const {
     return bindings.getAlias();
 }
 
-const Expression* MFDataModel::getSelectorsInternal() const {
+const VariableName* MFDataModel::getSelectorsInternal() const {
     U_ASSERT(!bogus);
     U_ASSERT(!hasPattern());
     return std::get_if<Matcher>(&body)->selectors.getAlias();
@@ -786,15 +788,13 @@ MFDataModel::Builder& MFDataModel::Builder::addBinding(Binding&& b, UErrorCode&
     return *this;
 }
 
-/*
+MFDataModel::Builder& MFDataModel::Builder::addSelector(VariableName&& selector,
-  selector must be non-null
+                                                        UErrorCode& status) {
-*/
-MFDataModel::Builder& MFDataModel::Builder::addSelector(Expression&& selector, UErrorCode& status) noexcept {
     THIS_ON_ERROR(status);
 
     buildSelectorsMessage(status);
     U_ASSERT(selectors != nullptr);
-    selectors->adoptElement(create<Expression>(std::move(selector), status), status);
+    selectors->adoptElement(create<VariableName>(std::move(selector), status), status);
 
     return *this;
 }
@@ -830,11 +830,11 @@ MFDataModel::MFDataModel(const MFDataModel& other) : body(Pattern()) {
     if (other.hasPattern()) {
         body = *std::get_if<Pattern>(&other.body);
     } else {
-        const Expression* otherSelectors = other.getSelectorsInternal();
+        const VariableName* otherSelectors = other.getSelectorsInternal();
         const Variant* otherVariants = other.getVariantsInternal();
         int32_t numSelectors = other.numSelectors();
         int32_t numVariants = other.numVariants();
-        Expression* copiedSelectors = copyArray(otherSelectors, numSelectors, localErrorCode);
+        VariableName* copiedSelectors = copyArray(otherSelectors, numSelectors, localErrorCode);
         Variant* copiedVariants = copyArray(otherVariants, numVariants, localErrorCode);
         if (U_FAILURE(localErrorCode)) {
             bogus = true;
@@ -863,7 +863,9 @@ MFDataModel::MFDataModel(const MFDataModel::Builder& builder, UErrorCode& errorC
         int32_t numVariants = builder.variants->size();
         int32_t numSelectors = builder.selectors->size();
         LocalArray<Variant> variants(copyVectorToArray<Variant>(*builder.variants, errorCode), errorCode);
-        LocalArray<Expression> selectors(copyVectorToArray<Expression>(*builder.selectors, errorCode), errorCode);
+        LocalArray<VariableName> selectors(copyVectorToArray<VariableName>(*builder.selectors,
+                                                                           errorCode),
+                                           errorCode);
         if (U_FAILURE(errorCode)) {
             bogus = true;
             return;
@@ -918,3 +920,5 @@ U_NAMESPACE_END
 #endif /* #if !UCONFIG_NO_MF2 */
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
+
+#endif /* #if !UCONFIG_NO_NORMALIZATION */

deps/icu-small/source/i18n/messageformat2_errors.cpp

@@ -3,6 +3,8 @@
 
 #include "unicode/utypes.h"
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
@@ -290,3 +292,5 @@ U_NAMESPACE_END
 #endif /* #if !UCONFIG_NO_MF2 */
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
+
+#endif /* #if !UCONFIG_NO_NORMALIZATION */

deps/icu-small/source/i18n/messageformat2_errors.h

@@ -15,6 +15,8 @@
  * \brief C++ API: Formats messages using the draft MessageFormat 2.0.
  */
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
@@ -151,6 +153,8 @@ U_NAMESPACE_END
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
 
+#endif /* #if !UCONFIG_NO_NORMALIZATION */
+
 #endif /* U_SHOW_CPLUSPLUS_API */
 
 #endif // MESSAGEFORMAT2_ERRORS_H

deps/icu-small/source/i18n/messageformat2_evaluation.cpp

@@ -3,6 +3,8 @@
 
 #include "unicode/utypes.h"
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
@@ -89,35 +91,54 @@ FunctionOptions::FunctionOptions(FunctionOptions&& other) {
 FunctionOptions::~FunctionOptions() {
     if (options != nullptr) {
         delete[] options;
+        options = nullptr;
     }
 }
-// ResolvedSelector
-// ----------------
 
-ResolvedSelector::ResolvedSelector(const FunctionName& fn,
+static bool containsOption(const UVector& opts, const ResolvedFunctionOption& opt) {
-                                   Selector* sel,
+    for (int32_t i = 0; i < opts.size(); i++) {
-                                   FunctionOptions&& opts,
+        if (static_cast<ResolvedFunctionOption*>(opts[i])->getName()
-                                   FormattedPlaceholder&& val)
+            == opt.getName()) {
-    : selectorName(fn), selector(sel), options(std::move(opts)), value(std::move(val))  {
+            return true;
-    U_ASSERT(sel != nullptr);
+        }
+    }
+    return false;
 }
 
-ResolvedSelector::ResolvedSelector(FormattedPlaceholder&& val) : value(std::move(val)) {}
+// Options in `this` take precedence
+// `this` can't be used after mergeOptions is called
+FunctionOptions FunctionOptions::mergeOptions(FunctionOptions&& other,
+                                              UErrorCode& status) {
+    UVector mergedOptions(status);
+    mergedOptions.setDeleter(uprv_deleteUObject);
 
-ResolvedSelector& ResolvedSelector::operator=(ResolvedSelector&& other) noexcept {
+    if (U_FAILURE(status)) {
-    selectorName = std::move(other.selectorName);
+        return {};
-    selector.adoptInstead(other.selector.orphan());
+    }
-    options = std::move(other.options);
+
-    value = std::move(other.value);
+    // Create a new vector consisting of the options from this `FunctionOptions`
-    return *this;
+    for (int32_t i = 0; i < functionOptionsLen; i++) {
+        mergedOptions.adoptElement(create<ResolvedFunctionOption>(std::move(options[i]), status),
+                                 status);
+    }
+
+    // Add each option from `other` that doesn't appear in this `FunctionOptions`
+    for (int i = 0; i < other.functionOptionsLen; i++) {
+        // Note: this is quadratic in the length of `options`
+        if (!containsOption(mergedOptions, other.options[i])) {
+            mergedOptions.adoptElement(create<ResolvedFunctionOption>(std::move(other.options[i]),
+                                                                    status),
+                                     status);
+        }
+    }
+
+    delete[] options;
+    options = nullptr;
+    functionOptionsLen = 0;
+
+    return FunctionOptions(std::move(mergedOptions), status);
 }
 
-ResolvedSelector::ResolvedSelector(ResolvedSelector&& other) {
-    *this = std::move(other);
-}
-
-ResolvedSelector::~ResolvedSelector() {}
-
 // PrioritizedVariant
 // ------------------
 
@@ -190,18 +211,210 @@ PrioritizedVariant::~PrioritizedVariant() {}
         errors.checkErrors(status);
     }
 
-    const Formattable* MessageContext::getGlobal(const VariableName& v, UErrorCode& errorCode) const {
+    const Formattable* MessageContext::getGlobal(const MessageFormatter& context,
-       return arguments.getArgument(v, errorCode);
+                                                 const VariableName& v,
+                                                 UErrorCode& errorCode) const {
+       return arguments.getArgument(context, v, errorCode);
     }
 
     MessageContext::MessageContext(const MessageArguments& args,
                                    const StaticErrors& e,
                                    UErrorCode& status) : arguments(args), errors(e, status) {}
+
     MessageContext::~MessageContext() {}
 
+    // InternalValue
+    // -------------
+
+    bool InternalValue::isFallback() const {
+        return std::holds_alternative<FormattedPlaceholder>(argument)
+            && std::get_if<FormattedPlaceholder>(&argument)->isFallback();
+    }
+
+    bool InternalValue::hasNullOperand() const {
+        return std::holds_alternative<FormattedPlaceholder>(argument)
+            && std::get_if<FormattedPlaceholder>(&argument)->isNullOperand();
+    }
+
+    FormattedPlaceholder InternalValue::takeArgument(UErrorCode& errorCode) {
+        if (U_FAILURE(errorCode)) {
+            return {};
+        }
+
+        if (std::holds_alternative<FormattedPlaceholder>(argument)) {
+            return std::move(*std::get_if<FormattedPlaceholder>(&argument));
+        }
+        errorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        return {};
+    }
+
+    const UnicodeString& InternalValue::getFallback() const {
+        if (std::holds_alternative<FormattedPlaceholder>(argument)) {
+            return std::get_if<FormattedPlaceholder>(&argument)->getFallback();
+        }
+        return (*std::get_if<InternalValue*>(&argument))->getFallback();
+    }
+
+    const Selector* InternalValue::getSelector(UErrorCode& errorCode) const {
+        if (U_FAILURE(errorCode)) {
+            return nullptr;
+        }
+
+        if (selector == nullptr) {
+            errorCode = U_ILLEGAL_ARGUMENT_ERROR;
+        }
+        return selector;
+    }
+
+    InternalValue::InternalValue(FormattedPlaceholder&& arg) {
+        argument = std::move(arg);
+        selector = nullptr;
+        formatter = nullptr;
+    }
+
+    InternalValue::InternalValue(InternalValue* operand,
+                                 FunctionOptions&& opts,
+                                 const FunctionName& functionName,
+                                 const Formatter* f,
+                                 const Selector* s) {
+        argument = operand;
+        options = std::move(opts);
+        name = functionName;
+        selector = s;
+        formatter = f;
+        U_ASSERT(selector != nullptr || formatter != nullptr);
+    }
+
+    // `this` cannot be used after calling this method
+    void InternalValue::forceSelection(DynamicErrors& errs,
+                                       const UnicodeString* keys,
+                                       int32_t keysLen,
+                                       UnicodeString* prefs,
+                                       int32_t& prefsLen,
+                                       UErrorCode& errorCode) {
+        if (U_FAILURE(errorCode)) {
+            return;
+        }
+
+        if (!canSelect()) {
+            errorCode = U_ILLEGAL_ARGUMENT_ERROR;
+            return;
+        }
+        // Find the argument and complete set of options by traversing `argument`
+        FunctionOptions opts;
+        InternalValue* p = this;
+        FunctionName selectorName = name;
+        while (std::holds_alternative<InternalValue*>(p->argument)) {
+            if (p->name != selectorName) {
+                // Can only compose calls to the same selector
+                errorCode = U_ILLEGAL_ARGUMENT_ERROR;
+                return;
+            }
+            // First argument to mergeOptions takes precedence
+            opts = opts.mergeOptions(std::move(p->options), errorCode);
+            if (U_FAILURE(errorCode)) {
+                return;
+            }
+            InternalValue* next = *std::get_if<InternalValue*>(&p->argument);
+            p = next;
+        }
+        FormattedPlaceholder arg = std::move(*std::get_if<FormattedPlaceholder>(&p->argument));
+
+        selector->selectKey(std::move(arg), std::move(opts),
+                            keys, keysLen,
+                            prefs, prefsLen, errorCode);
+        if (U_FAILURE(errorCode)) {
+            errorCode = U_ZERO_ERROR;
+            errs.setSelectorError(selectorName, errorCode);
+        }
+    }
+
+    FormattedPlaceholder InternalValue::forceFormatting(DynamicErrors& errs, UErrorCode& errorCode) {
+        if (U_FAILURE(errorCode)) {
+            return {};
+        }
+
+        if (formatter == nullptr && selector == nullptr) {
+            U_ASSERT(std::holds_alternative<FormattedPlaceholder>(argument));
+            return std::move(*std::get_if<FormattedPlaceholder>(&argument));
+        }
+        if (formatter == nullptr) {
+            errorCode = U_ILLEGAL_ARGUMENT_ERROR;
+            return {};
+        }
+
+        FormattedPlaceholder arg;
+
+        if (std::holds_alternative<FormattedPlaceholder>(argument)) {
+            arg = std::move(*std::get_if<FormattedPlaceholder>(&argument));
+        } else {
+            arg = (*std::get_if<InternalValue*>(&argument))->forceFormatting(errs,
+                                                                             errorCode);
+        }
+
+        if (U_FAILURE(errorCode)) {
+            return {};
+        }
+
+        // The fallback for a nullary function call is the function name
+        UnicodeString fallback;
+        if (arg.isNullOperand()) {
+            fallback = u":";
+            fallback += name;
+        } else {
+            fallback = arg.getFallback();
+        }
+
+        // Call the function with the argument
+        FormattedPlaceholder result = formatter->format(std::move(arg), std::move(options), errorCode);
+        if (U_FAILURE(errorCode)) {
+            if (errorCode == U_MF_OPERAND_MISMATCH_ERROR) {
+                errorCode = U_ZERO_ERROR;
+                errs.setOperandMismatchError(name, errorCode);
+            } else {
+                errorCode = U_ZERO_ERROR;
+                // Convey any error generated by the formatter
+                // as a formatting error, except for operand mismatch errors
+                errs.setFormattingError(name, errorCode);
+            }
+        }
+        // Ignore the output if any error occurred
+        if (errs.hasFormattingError()) {
+            return FormattedPlaceholder(fallback);
+        }
+
+        return result;
+    }
+
+    InternalValue& InternalValue::operator=(InternalValue&& other) noexcept {
+        argument = std::move(other.argument);
+        other.argument = nullptr;
+        options = std::move(other.options);
+        name = other.name;
+        selector = other.selector;
+        formatter = other.formatter;
+        other.selector = nullptr;
+        other.formatter = nullptr;
+
+        return *this;
+    }
+
+    InternalValue::~InternalValue() {
+        delete selector;
+        selector = nullptr;
+        delete formatter;
+        formatter = nullptr;
+        if (std::holds_alternative<InternalValue*>(argument)) {
+            delete *std::get_if<InternalValue*>(&argument);
+            argument = nullptr;
+        }
+    }
+
 } // namespace message2
 U_NAMESPACE_END
 
 #endif /* #if !UCONFIG_NO_MF2 */
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
+
+#endif /* #if !UCONFIG_NO_NORMALIZATION */

deps/icu-small/source/i18n/messageformat2_evaluation.h

@@ -14,6 +14,7 @@
  * \file
  * \brief C++ API: Formats messages using the draft MessageFormat 2.0.
  */
+#if !UCONFIG_NO_NORMALIZATION
 
 #if !UCONFIG_NO_FORMATTING
 
@@ -63,38 +64,6 @@ namespace message2 {
         return 1;
     }
 
-    // Encapsulates a value to be scrutinized by a `match` with its resolved
-    // options and the name of the selector
-    class ResolvedSelector : public UObject {
-    public:
-        ResolvedSelector() {}
-        ResolvedSelector(const FunctionName& fn,
-                         Selector* selector,
-                         FunctionOptions&& options,
-                         FormattedPlaceholder&& value);
-        // Used either for errors, or when selector isn't yet known
-        explicit ResolvedSelector(FormattedPlaceholder&& value);
-        bool hasSelector() const { return selector.isValid(); }
-        const FormattedPlaceholder& argument() const { return value; }
-        FormattedPlaceholder&& takeArgument() { return std::move(value); }
-        const Selector* getSelector() {
-            U_ASSERT(selector.isValid());
-            return selector.getAlias();
-        }
-        FunctionOptions&& takeOptions() {
-            return std::move(options);
-        }
-        const FunctionName& getSelectorName() const { return selectorName; }
-        virtual ~ResolvedSelector();
-        ResolvedSelector& operator=(ResolvedSelector&&) noexcept;
-        ResolvedSelector(ResolvedSelector&&);
-    private:
-        FunctionName selectorName; // For error reporting
-        LocalPointer<Selector> selector;
-        FunctionOptions options;
-        FormattedPlaceholder value;
-    }; // class ResolvedSelector
-
     // Closures and environments
     // -------------------------
 
@@ -174,11 +143,15 @@ namespace message2 {
     // The context contains all the information needed to process
     // an entire message: arguments, formatter cache, and error list
 
+    class MessageFormatter;
+
     class MessageContext : public UMemory {
     public:
         MessageContext(const MessageArguments&, const StaticErrors&, UErrorCode&);
 
-        const Formattable* getGlobal(const VariableName&, UErrorCode&) const;
+        const Formattable* getGlobal(const MessageFormatter&,
+                                     const VariableName&,
+                                     UErrorCode&) const;
 
         // If any errors were set, update `status` accordingly
         void checkErrors(UErrorCode& status) const;
@@ -191,8 +164,47 @@ namespace message2 {
         const MessageArguments& arguments; // External message arguments
         // Errors accumulated during parsing/formatting
         DynamicErrors errors;
+
     }; // class MessageContext
 
+    // InternalValue
+    // ----------------
+
+    class InternalValue : public UObject {
+    public:
+        const FunctionName& getFunctionName() const { return name; }
+        bool canSelect() const { return selector != nullptr; }
+        const Selector* getSelector(UErrorCode&) const;
+        FormattedPlaceholder forceFormatting(DynamicErrors& errs,
+                                             UErrorCode& errorCode);
+        void forceSelection(DynamicErrors& errs,
+                            const UnicodeString* keys,
+                            int32_t keysLen,
+                            UnicodeString* prefs,
+                            int32_t& prefsLen,
+                            UErrorCode& errorCode);
+        // Needs to be deep-copyable and movable
+        virtual ~InternalValue();
+        InternalValue(FormattedPlaceholder&&);
+        // Formatter and selector may be null
+        InternalValue(InternalValue*, FunctionOptions&&, const FunctionName&, const Formatter*,
+                      const Selector*);
+        const UnicodeString& getFallback() const;
+        bool isFallback() const;
+        bool hasNullOperand() const;
+        // Can't be used anymore after calling this
+        FormattedPlaceholder takeArgument(UErrorCode& errorCode);
+        InternalValue(InternalValue&& other) { *this = std::move(other); }
+        InternalValue& operator=(InternalValue&& other) noexcept;
+    private:
+        // InternalValue is owned (if present)
+        std::variant<InternalValue*, FormattedPlaceholder> argument;
+        FunctionOptions options;
+        FunctionName name;
+        const Selector* selector; // May be null
+        const Formatter* formatter; // May be null, but one or the other should be non-null unless argument is a FormattedPlaceholder
+    }; // class InternalValue
+
 } // namespace message2
 
 U_NAMESPACE_END
@@ -201,6 +213,8 @@ U_NAMESPACE_END
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
 
+#endif /* #if !UCONFIG_NO_NORMALIZATION */
+
 #endif /* U_SHOW_CPLUSPLUS_API */
 
 #endif // MESSAGEFORMAT2_EVALUATION_H

deps/icu-small/source/i18n/messageformat2_formattable.cpp

@@ -3,6 +3,8 @@
 
 #include "unicode/utypes.h"
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
@@ -336,3 +338,5 @@ U_NAMESPACE_END
 #endif /* #if !UCONFIG_NO_MF2 */
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
+
+#endif /* #if !UCONFIG_NO_NORMALIZATION */

deps/icu-small/source/i18n/messageformat2_formatter.cpp

@@ -3,6 +3,8 @@
 
 #include "unicode/utypes.h"
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
@@ -43,7 +45,8 @@ namespace message2 {
 
         // Parse the pattern
         MFDataModel::Builder tree(errorCode);
-        Parser(pat, tree, *errors, normalizedInput).parse(parseError, errorCode);
+        Parser(pat, tree, *errors, normalizedInput, errorCode)
+            .parse(parseError, errorCode);
 
         // Fail on syntax errors
         if (errors->hasSyntaxError()) {
@@ -116,6 +119,24 @@ namespace message2 {
 
     // MessageFormatter
 
+    // Returns the NFC-normalized version of s, returning s itself
+    // if it's already normalized.
+    UnicodeString MessageFormatter::normalizeNFC(const UnicodeString& s) const {
+        UErrorCode status = U_ZERO_ERROR;
+        // Check if string is already normalized
+        UNormalizationCheckResult result = nfcNormalizer->quickCheck(s, status);
+        // If so, return it
+        if (U_SUCCESS(status) && result == UNORM_YES) {
+            return s;
+        }
+        // Otherwise, normalize it
+        UnicodeString normalized = nfcNormalizer->normalize(s, status);
+        if (U_FAILURE(status)) {
+            return {};
+        }
+        return normalized;
+    }
+
     MessageFormatter::MessageFormatter(const MessageFormatter::Builder& builder, UErrorCode &success) : locale(builder.locale), customMFFunctionRegistry(builder.customMFFunctionRegistry) {
         CHECK_ERROR(success);
 
@@ -132,9 +153,13 @@ namespace message2 {
             .adoptFormatter(FunctionName(UnicodeString("time")), time, success)
             .adoptFormatter(FunctionName(UnicodeString("number")), number, success)
             .adoptFormatter(FunctionName(UnicodeString("integer")), integer, success)
+            .adoptFormatter(FunctionName(UnicodeString("test:function")), new StandardFunctions::TestFormatFactory(), success)
+            .adoptFormatter(FunctionName(UnicodeString("test:format")), new StandardFunctions::TestFormatFactory(), success)
             .adoptSelector(FunctionName(UnicodeString("number")), new StandardFunctions::PluralFactory(UPLURAL_TYPE_CARDINAL), success)
             .adoptSelector(FunctionName(UnicodeString("integer")), new StandardFunctions::PluralFactory(StandardFunctions::PluralFactory::integer()), success)
-            .adoptSelector(FunctionName(UnicodeString("string")), new StandardFunctions::TextFactory(), success);
+            .adoptSelector(FunctionName(UnicodeString("string")), new StandardFunctions::TextFactory(), success)
+            .adoptSelector(FunctionName(UnicodeString("test:function")), new StandardFunctions::TestSelectFactory(), success)
+            .adoptSelector(FunctionName(UnicodeString("test:select")), new StandardFunctions::TestSelectFactory(), success);
         CHECK_ERROR(success);
         standardMFFunctionRegistry = standardFunctionsBuilder.build();
         CHECK_ERROR(success);
@@ -163,6 +188,8 @@ namespace message2 {
             errors = errorsNew.orphan();
         }
 
+        nfcNormalizer = Normalizer2::getNFCInstance(success);
+
         // Note: we currently evaluate variables lazily,
         // without memoization. This call is still necessary
         // to check out-of-scope uses of local variables in
@@ -170,7 +197,7 @@ namespace message2 {
         // only be checked when arguments are known)
 
         // Check for resolution errors
-        Checker(dataModel, *errors).check(success);
+        Checker(dataModel, *errors, *this).check(success);
     }
 
     void MessageFormatter::cleanup() noexcept {
@@ -191,6 +218,7 @@ namespace message2 {
         signalErrors = other.signalErrors;
         errors = other.errors;
         other.errors = nullptr;
+        nfcNormalizer = other.nfcNormalizer;
         return *this;
     }
 
@@ -256,8 +284,11 @@ namespace message2 {
         return formatter;
     }
 
-    bool MessageFormatter::getDefaultFormatterNameByType(const UnicodeString& type, FunctionName& name) const {
+    bool MessageFormatter::getDefaultFormatterNameByType(const UnicodeString& type,
-        U_ASSERT(hasCustomMFFunctionRegistry());
+                                                         FunctionName& name) const {
+        if (!hasCustomMFFunctionRegistry()) {
+            return false;
+        }
         const MFFunctionRegistry& reg = getCustomMFFunctionRegistry();
         return reg.getDefaultFormatterNameByType(type, name);
     }
@@ -352,3 +383,5 @@ U_NAMESPACE_END
 #endif /* #if !UCONFIG_NO_MF2 */
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
+
+#endif /* #if !UCONFIG_NO_NORMALIZATION */

deps/icu-small/source/i18n/messageformat2_function_registry.cpp

@@ -3,6 +3,8 @@
 
 #include "unicode/utypes.h"
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
@@ -85,10 +87,11 @@ MFFunctionRegistry::Builder::Builder(UErrorCode& errorCode) {
     formattersByType = new Hashtable();
     if (!(formatters != nullptr && selectors != nullptr && formattersByType != nullptr)) {
         errorCode = U_MEMORY_ALLOCATION_ERROR;
-    }
+    } else {
         formatters->setValueDeleter(uprv_deleteUObject);
         selectors->setValueDeleter(uprv_deleteUObject);
         formattersByType->setValueDeleter(uprv_deleteUObject);
+    }
 }
 
 MFFunctionRegistry::Builder::~Builder() {
@@ -158,9 +161,13 @@ void MFFunctionRegistry::checkStandard() const {
     checkFormatter("time");
     checkFormatter("number");
     checkFormatter("integer");
+    checkFormatter("test:function");
+    checkFormatter("test:format");
     checkSelector("number");
     checkSelector("integer");
     checkSelector("string");
+    checkSelector("test:function");
+    checkSelector("test:select");
 }
 
 // Formatter/selector helpers
@@ -424,14 +431,14 @@ static FormattedPlaceholder notANumber(const FormattedPlaceholder& input) {
     return FormattedPlaceholder(input, FormattedValue(UnicodeString("NaN")));
 }
 
-static double parseNumberLiteral(const FormattedPlaceholder& input, UErrorCode& errorCode) {
+static double parseNumberLiteral(const Formattable& input, UErrorCode& errorCode) {
     if (U_FAILURE(errorCode)) {
         return {};
     }
 
     // Copying string to avoid GCC dangling-reference warning
     // (although the reference is safe)
-    UnicodeString inputStr = input.asFormattable().getString(errorCode);
+    UnicodeString inputStr = input.getString(errorCode);
     // Precondition: `input`'s source Formattable has type string
     if (U_FAILURE(errorCode)) {
         return {};
@@ -463,8 +470,42 @@ static double parseNumberLiteral(const FormattedPlaceholder& input, UErrorCode&
     return result;
 }
 
+static UChar32 digitToChar(int32_t val, UErrorCode errorCode) {
+    if (U_FAILURE(errorCode)) {
+        return '0';
+    }
+    if (val < 0 || val > 9) {
+        errorCode = U_ILLEGAL_ARGUMENT_ERROR;
+    }
+    switch(val) {
+        case 0:
+            return '0';
+        case 1:
+            return '1';
+        case 2:
+            return '2';
+        case 3:
+            return '3';
+        case 4:
+            return '4';
+        case 5:
+            return '5';
+        case 6:
+            return '6';
+        case 7:
+            return '7';
+        case 8:
+            return '8';
+        case 9:
+            return '9';
+        default:
+            errorCode = U_ILLEGAL_ARGUMENT_ERROR;
+            return '0';
+    }
+}
+
 static FormattedPlaceholder tryParsingNumberLiteral(const number::LocalizedNumberFormatter& nf, const FormattedPlaceholder& input, UErrorCode& errorCode) {
-    double numberValue = parseNumberLiteral(input, errorCode);
+    double numberValue = parseNumberLiteral(input.asFormattable(), errorCode);
     if (U_FAILURE(errorCode)) {
         return notANumber(input);
     }
@@ -1235,6 +1276,273 @@ void StandardFunctions::TextSelector::selectKey(FormattedPlaceholder&& toFormat,
 StandardFunctions::TextFactory::~TextFactory() {}
 StandardFunctions::TextSelector::~TextSelector() {}
 
+// ------------ TestFormatFactory
+
+Formatter* StandardFunctions::TestFormatFactory::createFormatter(const Locale& locale, UErrorCode& errorCode) {
+    NULL_ON_ERROR(errorCode);
+
+    // Results are not locale-dependent
+    (void) locale;
+
+    Formatter* result = new TestFormat();
+    if (result == nullptr) {
+        errorCode = U_MEMORY_ALLOCATION_ERROR;
+    }
+    return result;
+}
+
+StandardFunctions::TestFormatFactory::~TestFormatFactory() {}
+StandardFunctions::TestFormat::~TestFormat() {}
+
+// Extract numeric value from a Formattable or, if it's a string,
+// parse it as a number according to the MF2 `number-literal` grammar production
+double formattableToNumber(const Formattable& arg, UErrorCode& status) {
+    if (U_FAILURE(status)) {
+        return 0;
+    }
+
+    double result = 0;
+
+    switch (arg.getType()) {
+        case UFMT_DOUBLE: {
+            result = arg.getDouble(status);
+            U_ASSERT(U_SUCCESS(status));
+            break;
+        }
+        case UFMT_LONG: {
+            result = (double) arg.getLong(status);
+            U_ASSERT(U_SUCCESS(status));
+            break;
+        }
+        case UFMT_INT64: {
+            result = (double) arg.getInt64(status);
+            U_ASSERT(U_SUCCESS(status));
+            break;
+        }
+        case UFMT_STRING: {
+            // Try to parse the string as a number
+            result = parseNumberLiteral(arg, status);
+            if (U_FAILURE(status)) {
+                status = U_MF_OPERAND_MISMATCH_ERROR;
+            }
+            break;
+        }
+        default: {
+            // Other types can't be parsed as a number
+            status = U_MF_OPERAND_MISMATCH_ERROR;
+            break;
+        }
+        }
+    return result;
+}
+
+
+/* static */ void StandardFunctions::TestFormat::testFunctionParameters(const FormattedPlaceholder& arg,
+                                                                        const FunctionOptions& options,
+                                                                        int32_t& decimalPlaces,
+                                                                        bool& failsFormat,
+                                                                        bool& failsSelect,
+                                                                        double& input,
+                                                                        UErrorCode& status) {
+    CHECK_ERROR(status);
+
+    // 1. Let DecimalPlaces be 0.
+    decimalPlaces = 0;
+
+    // 2. Let FailsFormat be false.
+    failsFormat = false;
+
+    // 3. Let FailsSelect be false.
+    failsSelect = false;
+
+    // 4. Let arg be the resolved value of the expression operand.
+    // (already true)
+
+    // Step 5 omitted because composition isn't fully implemented yet
+    // 6. Else if arg is a numerical value or a string matching the number-literal production, then
+    input = formattableToNumber(arg.asFormattable(), status);
+    if (U_FAILURE(status)) {
+        // 7. Else,
+        // 7i. Emit "bad-input" Resolution Error.
+        status = U_MF_OPERAND_MISMATCH_ERROR;
+        // 7ii. Use a fallback value as the resolved value of the expression.
+        // Further steps of this algorithm are not followed.
+    }
+    // 8. If the decimalPlaces option is set, then
+    Formattable opt;
+    if (options.getFunctionOption(UnicodeString("decimalPlaces"), opt)) {
+        // 8i. If its value resolves to a numerical integer value 0 or 1
+        // or their corresponding string representations '0' or '1', then
+        double decimalPlacesInput = formattableToNumber(opt, status);
+        if (U_SUCCESS(status)) {
+            if (decimalPlacesInput == 0 || decimalPlacesInput == 1) {
+                // 8ia. Set DecimalPlaces to be the numerical value of the option.
+                decimalPlaces = decimalPlacesInput;
+            }
+        }
+        // 8ii. Else if its value is not an unresolved value set by option resolution,
+        else {
+            // 8iia. Emit "bad-option" Resolution Error.
+            status = U_MF_BAD_OPTION;
+            // 8iib. Use a fallback value as the resolved value of the expression.
+        }
+    }
+    // 9. If the fails option is set, then
+    Formattable failsOpt;
+    if (options.getFunctionOption(UnicodeString("fails"), failsOpt)) {
+        UnicodeString failsString = failsOpt.getString(status);
+        if (U_SUCCESS(status)) {
+            // 9i. If its value resolves to the string 'always', then
+            if (failsString == u"always") {
+                // 9ia. Set FailsFormat to be true
+                failsFormat = true;
+                // 9ib. Set FailsSelect to be true.
+                failsSelect = true;
+            }
+            // 9ii. Else if its value resolves to the string "format", then
+            else if (failsString == u"format") {
+                // 9ia. Set FailsFormat to be true
+                failsFormat = true;
+            }
+            // 9iii. Else if its value resolves to the string "select", then
+            else if (failsString == u"select") {
+                // 9iiia. Set FailsSelect to be true.
+                failsSelect = true;
+            }
+            // 9iv. Else if its value does not resolve to the string "never", then
+            else if (failsString != u"never") {
+                // 9iv(a). Emit "bad-option" Resolution Error.
+                status = U_MF_BAD_OPTION;
+            }
+        } else {
+            // 9iv. again
+            status = U_MF_BAD_OPTION;
+        }
+    }
+}
+
+FormattedPlaceholder StandardFunctions::TestFormat::format(FormattedPlaceholder&& arg,
+                                                           FunctionOptions&& options,
+                                                           UErrorCode& status) const{
+
+    int32_t decimalPlaces;
+    bool failsFormat;
+    bool failsSelect;
+    double input;
+
+    testFunctionParameters(arg, options, decimalPlaces,
+                           failsFormat, failsSelect, input, status);
+    if (U_FAILURE(status)) {
+        return FormattedPlaceholder(arg.getFallback());
+    }
+
+    // If FailsFormat is true, attempting to format the placeholder to any
+    // formatting target will fail.
+    if (failsFormat) {
+        status = U_MF_FORMATTING_ERROR;
+        return FormattedPlaceholder(arg.getFallback());
+    }
+    UnicodeString result;
+    // When :test:function is used as a formatter, a placeholder resolving to a value
+    // with a :test:function expression is formatted as a concatenation of the following parts:
+    // 1. If Input is less than 0, the character - U+002D Hyphen-Minus.
+    if (input < 0) {
+        result += HYPHEN;
+    }
+    // 2. The truncated absolute integer value of Input, i.e. floor(abs(Input)), formatted as a
+    // sequence of decimal digit characters (U+0030...U+0039).
+    char buffer[256];
+    bool ignore;
+    int ignoreLen;
+    int ignorePoint;
+    double_conversion::DoubleToStringConverter::DoubleToAscii(floor(abs(input)),
+                                                              double_conversion::DoubleToStringConverter::DtoaMode::SHORTEST,
+                                                              0,
+                                                              buffer,
+                                                              256,
+                                                              &ignore,
+                                                              &ignoreLen,
+                                                              &ignorePoint);
+    result += UnicodeString(buffer);
+    // 3. If DecimalPlaces is 1, then
+    if (decimalPlaces == 1) {
+        // 3i. The character . U+002E Full Stop.
+        result += u".";
+        // 3ii. The single decimal digit character representing the value
+        // floor((abs(Input) - floor(abs(Input))) * 10)
+        int32_t val = floor((abs(input) - floor(abs(input)) * 10));
+        result += digitToChar(val, status);
+        U_ASSERT(U_SUCCESS(status));
+    }
+    return FormattedPlaceholder(result);
+}
+
+// ------------ TestSelectFactory
+
+StandardFunctions::TestSelectFactory::~TestSelectFactory() {}
+StandardFunctions::TestSelect::~TestSelect() {}
+
+Selector* StandardFunctions::TestSelectFactory::createSelector(const Locale& locale,
+                                                               UErrorCode& errorCode) const {
+    NULL_ON_ERROR(errorCode);
+
+    // Results are not locale-dependent
+    (void) locale;
+
+    Selector* result = new TestSelect();
+    if (result == nullptr) {
+        errorCode = U_MEMORY_ALLOCATION_ERROR;
+    }
+    return result;
+}
+
+void StandardFunctions::TestSelect::selectKey(FormattedPlaceholder&& val,
+                                              FunctionOptions&& options,
+                                              const UnicodeString* keys,
+                                              int32_t keysLen,
+                                              UnicodeString* prefs,
+                                              int32_t& prefsLen,
+                                              UErrorCode& status) const {
+    int32_t decimalPlaces;
+    bool failsFormat;
+    bool failsSelect;
+    double input;
+
+    TestFormat::testFunctionParameters(val, options, decimalPlaces,
+                                       failsFormat, failsSelect, input, status);
+
+    if (U_FAILURE(status)) {
+        return;
+    }
+
+    if (failsSelect) {
+        status = U_MF_SELECTOR_ERROR;
+        return;
+    }
+
+    // If the Input is 1 and DecimalPlaces is 1, the method will return some slice
+    // of the list « '1.0', '1' », depending on whether those values are included in keys.
+    bool include1point0 = false;
+    bool include1 = false;
+    if (input == 1 && decimalPlaces == 1) {
+        include1point0 = true;
+        include1 = true;
+    } else if (input == 1 && decimalPlaces == 0) {
+        include1 = true;
+    }
+
+    // If the Input is 1 and DecimalPlaces is 0, the method will return the list « '1' » if
+    // keys includes '1', or an empty list otherwise.
+    // If the Input is any other value, the method will return an empty list.
+    for (int32_t i = 0; i < keysLen; i++) {
+        if ((keys[i] == u"1" && include1)
+            || (keys[i] == u"1.0" && include1point0)) {
+            prefs[prefsLen] = keys[i];
+            prefsLen++;
+        }
+    }
+}
+
 } // namespace message2
 U_NAMESPACE_END
 
@@ -1242,3 +1550,4 @@ U_NAMESPACE_END
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
 
+#endif /* #if !UCONFIG_NO_NORMALIZATION */

deps/icu-small/source/i18n/messageformat2_function_registry_internal.h

@@ -10,6 +10,8 @@
 
 #if U_SHOW_CPLUSPLUS_API
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
@@ -209,6 +211,60 @@ namespace message2 {
 
             TextSelector(const Locale& l) : locale(l) {}
         };
+
+        // See https://github.com/unicode-org/message-format-wg/blob/main/test/README.md
+        class TestFormatFactory : public FormatterFactory {
+        public:
+            Formatter* createFormatter(const Locale& locale, UErrorCode& status) override;
+            TestFormatFactory() {}
+            virtual ~TestFormatFactory();
+        };
+
+        class TestSelect;
+
+        class TestFormat : public Formatter {
+        public:
+            FormattedPlaceholder format(FormattedPlaceholder&& toFormat, FunctionOptions&& options, UErrorCode& status) const override;
+            virtual ~TestFormat();
+
+        private:
+            friend class TestFormatFactory;
+            friend class TestSelect;
+            TestFormat() {}
+            static void testFunctionParameters(const FormattedPlaceholder& arg,
+                                               const FunctionOptions& options,
+                                               int32_t& decimalPlaces,
+                                               bool& failsFormat,
+                                               bool& failsSelect,
+                                               double& input,
+                                               UErrorCode& status);
+
+        };
+
+        // See https://github.com/unicode-org/message-format-wg/blob/main/test/README.md
+        class TestSelectFactory : public SelectorFactory {
+        public:
+            Selector* createSelector(const Locale& locale, UErrorCode& status) const override;
+            TestSelectFactory() {}
+            virtual ~TestSelectFactory();
+        };
+
+        class TestSelect : public Selector {
+        public:
+            void selectKey(FormattedPlaceholder&& val,
+                           FunctionOptions&& options,
+                           const UnicodeString* keys,
+                           int32_t keysLen,
+                           UnicodeString* prefs,
+                           int32_t& prefsLen,
+                           UErrorCode& status) const override;
+            virtual ~TestSelect();
+
+        private:
+            friend class TestSelectFactory;
+            TestSelect() {}
+        };
+
     };
 
     extern void formatDateWithDefaults(const Locale& locale, UDate date, UnicodeString&, UErrorCode& errorCode);
@@ -226,6 +282,8 @@ U_NAMESPACE_END
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
 
+#endif /* #if !UCONFIG_NO_NORMALIZATION */
+
 #endif /* U_SHOW_CPLUSPLUS_API */
 
 #endif // MESSAGEFORMAT2_FUNCTION_REGISTRY_INTERNAL_H

deps/icu-small/source/i18n/messageformat2_macros.h

@@ -10,6 +10,8 @@
 
 #if U_SHOW_CPLUSPLUS_API
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
@@ -97,6 +99,8 @@ U_NAMESPACE_END
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
 
+#endif /* #if !UCONFIG_NO_NORMALIZATION */
+
 #endif /* U_SHOW_CPLUSPLUS_API */
 
 #endif // MESSAGEFORMAT2_MACROS_H

deps/icu-small/source/i18n/messageformat2_parser.cpp

@@ -3,13 +3,18 @@
 
 #include "unicode/utypes.h"
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
 
+#include "unicode/uniset.h"
 #include "messageformat2_errors.h"
 #include "messageformat2_macros.h"
 #include "messageformat2_parser.h"
+#include "ucln_in.h"
+#include "umutex.h"
 #include "uvector.h" // U_ASSERT
 
 U_NAMESPACE_BEGIN
@@ -91,14 +96,282 @@ static void copyContext(const UChar in[U_PARSE_CONTEXT_LEN], UChar out[U_PARSE_C
 }
 
 // -------------------------------------
-// Predicates
+// Initialization of UnicodeSets
 
-// Returns true if `c` is in the interval [`first`, `last`]
+namespace unisets {
-static bool inRange(UChar32 c, UChar32 first, UChar32 last) {
+
-    U_ASSERT(first < last);
+UnicodeSet* gUnicodeSets[unisets::UNISETS_KEY_COUNT] = {};
-    return c >= first && c <= last;
+
+inline UnicodeSet* getImpl(Key key) {
+    return gUnicodeSets[key];
 }
 
+icu::UInitOnce gMF2ParseUniSetsInitOnce {};
+}
+
+UnicodeSet* initContentChars(UErrorCode& status) {
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+
+    UnicodeSet* result = new UnicodeSet(0x0001, 0x0008); // Omit NULL, HTAB and LF
+    if (result == nullptr) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+        return nullptr;
+    }
+    result->add(0x000B, 0x000C); // Omit CR
+    result->add(0x000E, 0x001F); // Omit SP
+    result->add(0x0021, 0x002D); // Omit '.'
+    result->add(0x002F, 0x003F); // Omit '@'
+    result->add(0x0041, 0x005B); // Omit '\'
+    result->add(0x005D, 0x007A); // Omit { | }
+    result->add(0x007E, 0x2FFF); // Omit IDEOGRAPHIC_SPACE
+    result->add(0x3001, 0x10FFFF); // Allowing surrogates is intentional
+    result->freeze();
+    return result;
+}
+
+UnicodeSet* initWhitespace(UErrorCode& status) {
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+
+    UnicodeSet* result = new UnicodeSet();
+    if (result == nullptr) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+        return nullptr;
+    }
+    result->add(SPACE);
+    result->add(HTAB);
+    result->add(CR);
+    result->add(LF);
+    result->add(IDEOGRAPHIC_SPACE);
+    result->freeze();
+    return result;
+}
+
+UnicodeSet* initBidiControls(UErrorCode& status) {
+    UnicodeSet* result = new UnicodeSet(UnicodeString("[\\u061C]"), status);
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+    result->add(0x200E, 0x200F);
+    result->add(0x2066, 0x2069);
+    result->freeze();
+    return result;
+}
+
+UnicodeSet* initAlpha(UErrorCode& status) {
+    UnicodeSet* result = new UnicodeSet(UnicodeString("[:letter:]"), status);
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+    result->freeze();
+    return result;
+}
+
+UnicodeSet* initDigits(UErrorCode& status) {
+    UnicodeSet* result = new UnicodeSet(UnicodeString("[:number:]"), status);
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+    result->freeze();
+    return result;
+}
+
+UnicodeSet* initNameStartChars(UErrorCode& status) {
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+
+    UnicodeSet* isAlpha = unisets::gUnicodeSets[unisets::ALPHA] = initAlpha(status);
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+    UnicodeSet* result = new UnicodeSet(*isAlpha);
+    if (result == nullptr) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+        return nullptr;
+    };
+    result->add(UNDERSCORE);
+    result->add(0x00C0, 0x00D6);
+    result->add(0x00D8, 0x00F6);
+    result->add(0x00F8, 0x02FF);
+    result->add(0x0370, 0x037D);
+    result->add(0x037F, 0x061B);
+    result->add(0x061D, 0x1FFF);
+    result->add(0x200C, 0x200D);
+    result->add(0x2070, 0x218F);
+    result->add(0x2C00, 0x2FEF);
+    result->add(0x3001, 0xD7FF);
+    result->add(0xF900, 0xFDCF);
+    result->add(0xFDF0, 0xFFFD);
+    result->add(0x100000, 0xEFFFF);
+    result->freeze();
+    return result;
+}
+
+UnicodeSet* initNameChars(UErrorCode& status) {
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+
+    UnicodeSet* nameStart = unisets::gUnicodeSets[unisets::NAME_START] = initNameStartChars(status);
+    UnicodeSet* digit = unisets::gUnicodeSets[unisets::DIGIT] = initDigits(status);
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+    UnicodeSet* result = new UnicodeSet();
+    if (result == nullptr) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+        return nullptr;
+    };
+    result->addAll(*nameStart);
+    result->addAll(*digit);
+    result->add(HYPHEN);
+    result->add(PERIOD);
+    result->add(0x00B7);
+    result->add(0x0300, 0x036F);
+    result->add(0x203F, 0x2040);
+    result->freeze();
+    return result;
+}
+
+UnicodeSet* initTextChars(UErrorCode& status) {
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+
+    UnicodeSet* content = unisets::gUnicodeSets[unisets::CONTENT] = initContentChars(status);
+    UnicodeSet* whitespace = unisets::gUnicodeSets[unisets::WHITESPACE] = initWhitespace(status);
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+    UnicodeSet* result = new UnicodeSet();
+    if (result == nullptr) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+        return nullptr;
+    };
+    result->addAll(*content);
+    result->addAll(*whitespace);
+    result->add(PERIOD);
+    result->add(AT);
+    result->add(PIPE);
+    result->freeze();
+    return result;
+}
+
+UnicodeSet* initQuotedChars(UErrorCode& status) {
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+
+    unisets::gUnicodeSets[unisets::TEXT] = initTextChars(status);
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+    UnicodeSet* result = new UnicodeSet();
+    if (result == nullptr) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+        return nullptr;
+    };
+    // content and whitespace were initialized by `initTextChars()`
+    UnicodeSet* content = unisets::getImpl(unisets::CONTENT);
+    if (content == nullptr) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+        return nullptr;
+    }
+    result->addAll(*content);
+    UnicodeSet* whitespace = unisets::getImpl(unisets::WHITESPACE);
+    if (whitespace == nullptr) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+        return nullptr;
+    }
+    result->addAll(*whitespace);
+    result->add(PERIOD);
+    result->add(AT);
+    result->add(LEFT_CURLY_BRACE);
+    result->add(RIGHT_CURLY_BRACE);
+    result->freeze();
+    return result;
+}
+
+UnicodeSet* initEscapableChars(UErrorCode& status) {
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+
+    UnicodeSet* result = new UnicodeSet();
+    if (result == nullptr) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+        return nullptr;
+    }
+    result->add(PIPE);
+    result->add(BACKSLASH);
+    result->add(LEFT_CURLY_BRACE);
+    result->add(RIGHT_CURLY_BRACE);
+    result->freeze();
+    return result;
+}
+
+namespace unisets {
+
+UBool U_CALLCONV cleanupMF2ParseUniSets() {
+    for (int32_t i = 0; i < UNISETS_KEY_COUNT; i++) {
+        delete gUnicodeSets[i];
+        gUnicodeSets[i] = nullptr;
+    }
+    gMF2ParseUniSetsInitOnce.reset();
+    return true;
+}
+
+void U_CALLCONV initMF2ParseUniSets(UErrorCode& status) {
+    ucln_i18n_registerCleanup(UCLN_I18N_MF2_UNISETS, cleanupMF2ParseUniSets);
+    /*
+      Each of the init functions initializes the UnicodeSets
+      that it depends on.
+
+      initBidiControls (no dependencies)
+
+      initEscapableChars (no dependencies)
+
+      initNameChars depends on
+         initDigits
+         initNameStartChars depends on
+           initAlpha
+
+      initQuotedChars depends on
+         initTextChars depends on
+            initContentChars
+            initWhitespace
+     */
+    gUnicodeSets[unisets::BIDI] = initBidiControls(status);
+    gUnicodeSets[unisets::NAME_CHAR] = initNameChars(status);
+    gUnicodeSets[unisets::QUOTED] = initQuotedChars(status);
+    gUnicodeSets[unisets::ESCAPABLE] = initEscapableChars(status);
+
+    if (U_FAILURE(status)) {
+        cleanupMF2ParseUniSets();
+    }
+}
+
+const UnicodeSet* get(Key key, UErrorCode& status) {
+    umtx_initOnce(gMF2ParseUniSetsInitOnce, &initMF2ParseUniSets, status);
+    if (U_FAILURE(status)) {
+        return nullptr;
+    }
+    UnicodeSet* result = getImpl(key);
+    if (result == nullptr) {
+        status = U_MEMORY_ALLOCATION_ERROR;
+    }
+    return result;
+}
+
+}
+
+// -------------------------------------
+// Predicates
+
 /*
   The following helper predicates should exactly match nonterminals in the MessageFormat 2 grammar:
 
@@ -113,76 +386,50 @@ static bool inRange(UChar32 c, UChar32 first, UChar32 last) {
   `isWhitespace()`    : `s`
 */
 
-static bool isContentChar(UChar32 c) {
+bool Parser::isContentChar(UChar32 c) const {
-    return inRange(c, 0x0001, 0x0008)    // Omit NULL, HTAB and LF
+    return contentChars->contains(c);
-           || inRange(c, 0x000B, 0x000C) // Omit CR
-           || inRange(c, 0x000E, 0x001F) // Omit SP
-           || inRange(c, 0x0021, 0x002D) // Omit '.'
-           || inRange(c, 0x002F, 0x003F) // Omit '@'
-           || inRange(c, 0x0041, 0x005B) // Omit '\'
-           || inRange(c, 0x005D, 0x007A) // Omit { | }
-           || inRange(c, 0x007E, 0xD7FF) // Omit surrogates
-           || inRange(c, 0xE000, 0x10FFFF);
 }
 
-// See `s` in the MessageFormat 2 grammar
+// See `bidi` in the MF2 grammar
-inline bool isWhitespace(UChar32 c) {
+bool Parser::isBidiControl(UChar32 c) const {
-    switch (c) {
+    return bidiControlChars->contains(c);
-    case SPACE:
-    case HTAB:
-    case CR:
-    case LF:
-    case IDEOGRAPHIC_SPACE:
-        return true;
-    default:
-        return false;
-    }
 }
 
-static bool isTextChar(UChar32 c) {
+// See `ws` in the MessageFormat 2 grammar
-    return isContentChar(c)
+bool Parser::isWhitespace(UChar32 c) const {
-        || isWhitespace(c)
+    return whitespaceChars->contains(c);
-        || c == PERIOD
-        || c == AT
-        || c == PIPE;
 }
 
-static bool isAlpha(UChar32 c) { return inRange(c, 0x0041, 0x005A) || inRange(c, 0x0061, 0x007A); }
+bool Parser::isTextChar(UChar32 c) const {
-
+    return textChars->contains(c);
-static bool isDigit(UChar32 c) { return inRange(c, 0x0030, 0x0039); }
-
-static bool isNameStart(UChar32 c) {
-    return isAlpha(c) || c == UNDERSCORE || inRange(c, 0x00C0, 0x00D6) || inRange(c, 0x00D8, 0x00F6) ||
-           inRange(c, 0x00F8, 0x02FF) || inRange(c, 0x0370, 0x037D) || inRange(c, 0x037F, 0x1FFF) ||
-           inRange(c, 0x200C, 0x200D) || inRange(c, 0x2070, 0x218F) || inRange(c, 0x2C00, 0x2FEF) ||
-           inRange(c, 0x3001, 0xD7FF) || inRange(c, 0xF900, 0xFDCF) || inRange(c, 0xFDF0, 0xFFFD) ||
-           inRange(c, 0x10000, 0xEFFFF);
 }
 
-static bool isNameChar(UChar32 c) {
+bool Parser::isAlpha(UChar32 c) const {
-    return isNameStart(c) || isDigit(c) || c == HYPHEN || c == PERIOD || c == 0x00B7 ||
+    return alphaChars->contains(c);
-           inRange(c, 0x0300, 0x036F) || inRange(c, 0x203F, 0x2040);
 }
 
-static bool isUnquotedStart(UChar32 c) {
+bool Parser::isDigit(UChar32 c) const {
-    return isNameStart(c) || isDigit(c) || c == HYPHEN || c == PERIOD || c == 0x00B7 ||
+    return digitChars->contains(c);
-           inRange(c, 0x0300, 0x036F) || inRange(c, 0x203F, 0x2040);
 }
 
-static bool isQuotedChar(UChar32 c) {
+bool Parser::isNameStart(UChar32 c) const {
-    return isContentChar(c)
+    return nameStartChars->contains(c);
-        || isWhitespace(c)
-        || c == PERIOD
-        || c == AT
-        || c == LEFT_CURLY_BRACE
-        || c == RIGHT_CURLY_BRACE;
 }
 
-static bool isEscapableChar(UChar32 c) {
+bool Parser::isNameChar(UChar32 c) const {
-    return c == PIPE
+    return nameChars->contains(c);
-        || c == BACKSLASH
+}
-        || c == LEFT_CURLY_BRACE
+
-        || c == RIGHT_CURLY_BRACE;
+bool Parser::isUnquotedStart(UChar32 c) const {
+    return isNameChar(c);
+}
+
+bool Parser::isQuotedChar(UChar32 c) const {
+    return quotedChars->contains(c);
+}
+
+bool Parser::isEscapableChar(UChar32 c) const {
+    return escapableChars->contains(c);
 }
 
 // Returns true iff `c` can begin a `function` nonterminal
@@ -203,12 +450,12 @@ static bool isAnnotationStart(UChar32 c) {
 }
 
 // Returns true iff `c` can begin a `literal` nonterminal
-static bool isLiteralStart(UChar32 c) {
+bool Parser::isLiteralStart(UChar32 c) const {
     return (c == PIPE || isNameStart(c) || c == HYPHEN || isDigit(c));
 }
 
 // Returns true iff `c` can begin a `key` nonterminal
-static bool isKeyStart(UChar32 c) {
+bool Parser::isKeyStart(UChar32 c) const {
     return (c == ASTERISK || isLiteralStart(c));
 }
 
@@ -347,7 +594,7 @@ option, or the optional space before an attribute.
   No pre, no post.
   A message may end with whitespace, so `index` may equal `len()` on exit.
 */
-void Parser::parseWhitespaceMaybeRequired(bool required, UErrorCode& errorCode) {
+void Parser::parseRequiredWS(UErrorCode& errorCode) {
     bool sawWhitespace = false;
 
     // The loop exits either when we consume all the input,
@@ -358,7 +605,7 @@ void Parser::parseWhitespaceMaybeRequired(bool required, UErrorCode& errorCode)
             // If whitespace isn't required -- or if we saw it already --
             // then the caller is responsible for checking this case and
             // setting an error if necessary.
-            if (!required || sawWhitespace) {
+            if (sawWhitespace) {
                 // Not an error.
                 return;
             }
@@ -380,24 +627,51 @@ void Parser::parseWhitespaceMaybeRequired(bool required, UErrorCode& errorCode)
         }
     }
 
-    if (!sawWhitespace && required) {
+    if (!sawWhitespace) {
         ERROR(errorCode);
     }
 }
 
+void Parser::parseOptionalBidi() {
+    while (true) {
+        if (!inBounds()) {
+            return;
+        }
+        if (isBidiControl(peek())) {
+            next();
+        } else {
+            break;
+        }
+    }
+}
+
 /*
-  No pre, no post, for the same reason as `parseWhitespaceMaybeRequired()`.
+  No pre, no post, because a message may end with whitespace
+  Matches `s` in the MF2 grammar
 */
 void Parser::parseRequiredWhitespace(UErrorCode& errorCode) {
-    parseWhitespaceMaybeRequired(true, errorCode);
+    parseOptionalBidi();
+    parseRequiredWS(errorCode);
+    parseOptionalWhitespace();
     normalizedInput += SPACE;
 }
 
 /*
   No pre, no post, for the same reason as `parseWhitespaceMaybeRequired()`.
 */
-void Parser::parseOptionalWhitespace(UErrorCode& errorCode) {
+void Parser::parseOptionalWhitespace() {
-    parseWhitespaceMaybeRequired(false, errorCode);
+    while (true) {
+        if (!inBounds()) {
+            return;
+        }
+        auto cp = peek();
+        if (isWhitespace(cp) || isBidiControl(cp)) {
+            maybeAdvanceLine();
+            next();
+        } else {
+            break;
+        }
+    }
 }
 
 // Consumes a single character, signaling an error if `peek()` != `c`
@@ -442,11 +716,11 @@ void Parser::parseToken(const std::u16string_view& token, UErrorCode& errorCode)
 */
 void Parser::parseTokenWithWhitespace(const std::u16string_view& token, UErrorCode& errorCode) {
     // No need for error check or bounds check before parseOptionalWhitespace
-    parseOptionalWhitespace(errorCode);
+    parseOptionalWhitespace();
     // Establish precondition
     CHECK_BOUNDS(errorCode);
     parseToken(token, errorCode);
-    parseOptionalWhitespace(errorCode);
+    parseOptionalWhitespace();
     // Guarantee postcondition
     CHECK_BOUNDS(errorCode);
 }
@@ -458,12 +732,12 @@ void Parser::parseTokenWithWhitespace(const std::u16string_view& token, UErrorCo
    then consumes optional whitespace again
 */
 void Parser::parseTokenWithWhitespace(UChar32 c, UErrorCode& errorCode) {
-    // No need for error check or bounds check before parseOptionalWhitespace(errorCode)
+    // No need for error check or bounds check before parseOptionalWhitespace()
-    parseOptionalWhitespace(errorCode);
+    parseOptionalWhitespace();
     // Establish precondition
     CHECK_BOUNDS(errorCode);
     parseToken(c, errorCode);
-    parseOptionalWhitespace(errorCode);
+    parseOptionalWhitespace();
     // Guarantee postcondition
     CHECK_BOUNDS(errorCode);
 }
@@ -482,11 +756,17 @@ UnicodeString Parser::parseName(UErrorCode& errorCode) {
 
     U_ASSERT(inBounds());
 
-    if (!isNameStart(peek())) {
+    if (!(isNameStart(peek()) || isBidiControl(peek()))) {
         ERROR(errorCode);
         return name;
     }
 
+    // name       = [bidi] name-start *name-char [bidi]
+
+    // [bidi]
+    parseOptionalBidi();
+
+    // name-start *name-char
     while (isNameChar(peek())) {
         UChar32 c = peek();
         name += c;
@@ -497,6 +777,10 @@ UnicodeString Parser::parseName(UErrorCode& errorCode) {
             break;
         }
     }
+
+    // [bidi]
+    parseOptionalBidi();
+
     return name;
 }
 
@@ -510,21 +794,13 @@ VariableName Parser::parseVariableName(UErrorCode& errorCode) {
     VariableName result;
 
     U_ASSERT(inBounds());
-    // If the '$' is missing, we don't want a binding
+
-    // for this variable to be created.
-    bool valid = peek() == DOLLAR;
     parseToken(DOLLAR, errorCode);
     if (!inBounds()) {
         ERROR(errorCode);
         return result;
     }
-    UnicodeString varName = parseName(errorCode);
+    return VariableName(parseName(errorCode));
-    // Set the name to "" if the variable wasn't
-    // declared correctly
-    if (!valid) {
-        varName.remove();
-    }
-    return VariableName(varName);
 }
 
 /*
@@ -853,7 +1129,7 @@ void Parser::parseAttribute(AttributeAdder<T>& attrAdder, UErrorCode& errorCode)
     // about whether whitespace precedes another
     // attribute, or the '=' sign
     int32_t savedIndex = index;
-    parseOptionalWhitespace(errorCode);
+    parseOptionalWhitespace();
 
     Operand rand;
     if (peek() == EQUALS) {
@@ -861,19 +1137,9 @@ void Parser::parseAttribute(AttributeAdder<T>& attrAdder, UErrorCode& errorCode)
         parseTokenWithWhitespace(EQUALS, errorCode);
 
         UnicodeString rhsStr;
-        // Parse RHS, which is either a literal or variable
+        // Parse RHS, which must be a literal
-        switch (peek()) {
+        // attribute = "@" identifier [o "=" o literal]
-        case DOLLAR: {
-            rand = Operand(parseVariableName(errorCode));
-            break;
-        }
-        default: {
-            // Must be a literal
         rand = Operand(parseLiteral(errorCode));
-            break;
-        }
-        }
-        U_ASSERT(!rand.isNull());
     } else {
         // attribute -> "@" identifier [[s] "=" [s]]
         // Use null operand, which `rand` is already set to
@@ -881,7 +1147,7 @@ void Parser::parseAttribute(AttributeAdder<T>& attrAdder, UErrorCode& errorCode)
         index = savedIndex;
     }
 
-    attrAdder.addAttribute(lhs, std::move(rand), errorCode);
+    attrAdder.addAttribute(lhs, std::move(Operand(rand)), errorCode);
 }
 
 /*
@@ -1149,7 +1415,7 @@ the comment in `parseOptions()` for details.
       // (the character is either the required space before an annotation, or optional
       // trailing space after the literal or variable). It's still ambiguous which
       // one does apply.
-      parseOptionalWhitespace(status);
+      parseOptionalWhitespace();
       // Restore precondition
       CHECK_BOUNDS(status);
 
@@ -1220,7 +1486,7 @@ Expression Parser::parseExpression(UErrorCode& status) {
     // Parse opening brace
     parseToken(LEFT_CURLY_BRACE, status);
     // Optional whitespace after opening brace
-    parseOptionalWhitespace(status);
+    parseOptionalWhitespace();
 
     Expression::Builder exprBuilder(status);
     // Restore precondition
@@ -1263,7 +1529,7 @@ Expression Parser::parseExpression(UErrorCode& status) {
 
     // Parse optional space
     // (the last [s] in e.g. "{" [s] literal [s annotation] *(s attribute) [s] "}")
-    parseOptionalWhitespace(status);
+    parseOptionalWhitespace();
 
     // Either an operand or operator (or both) must have been set already,
     // so there can't be an error
@@ -1339,7 +1605,7 @@ void Parser::parseInputDeclaration(UErrorCode& status) {
     CHECK_BOUNDS(status);
 
     parseToken(ID_INPUT, status);
-    parseOptionalWhitespace(status);
+    parseOptionalWhitespace();
 
     // Restore precondition before calling parseExpression()
     CHECK_BOUNDS(status);
@@ -1400,7 +1666,7 @@ void Parser::parseDeclarations(UErrorCode& status) {
         // Avoid looping infinitely
         CHECK_ERROR(status);
 
-        parseOptionalWhitespace(status);
+        parseOptionalWhitespace();
         // Restore precondition
         CHECK_BOUNDS(status);
     }
@@ -1510,8 +1776,8 @@ This is addressed using "backtracking" (similarly to `parseOptions()`).
 
     // We've seen at least one whitespace-key pair, so now we can parse
     // *(s key) [s]
-    while (peek() != LEFT_CURLY_BRACE || isWhitespace(peek())) { // Try to recover from errors
+    while (peek() != LEFT_CURLY_BRACE || isWhitespace(peek()) || isBidiControl(peek())) {
-        bool wasWhitespace = isWhitespace(peek());
+        bool wasWhitespace = isWhitespace(peek()) || isBidiControl(peek());
         parseRequiredWhitespace(status);
         if (!wasWhitespace) {
             // Avoid infinite loop when parsing something like:
@@ -1569,7 +1835,7 @@ Markup Parser::parseMarkup(UErrorCode& status) {
     // Consume the '{'
     next();
     normalizedInput += LEFT_CURLY_BRACE;
-    parseOptionalWhitespace(status);
+    parseOptionalWhitespace();
     bool closing = false;
     switch (peek()) {
     case NUMBER_SIGN: {
@@ -1596,19 +1862,19 @@ Markup Parser::parseMarkup(UErrorCode& status) {
 
     // Parse the options, which must begin with a ' '
     // if present
-    if (inBounds() && isWhitespace(peek())) {
+    if (inBounds() && (isWhitespace(peek()) || isBidiControl(peek()))) {
         OptionAdder<Markup::Builder> optionAdder(builder);
         parseOptions(optionAdder, status);
     }
 
     // Parse the attributes, which also must begin
     // with a ' '
-    if (inBounds() && isWhitespace(peek())) {
+    if (inBounds() && (isWhitespace(peek()) || isBidiControl(peek()))) {
         AttributeAdder<Markup::Builder> attrAdder(builder);
         parseAttributes(attrAdder, status);
     }
 
-    parseOptionalWhitespace(status);
+    parseOptionalWhitespace();
 
     bool standalone = false;
     // Check if this is a standalone or not
@@ -1656,7 +1922,7 @@ std::variant<Expression, Markup> Parser::parsePlaceholder(UErrorCode& status) {
             isMarkup = true;
             break;
         }
-        if (!isWhitespace(c)){
+        if (!(isWhitespace(c) || isBidiControl(c))) {
             break;
         }
         tempIndex++;
@@ -1712,7 +1978,7 @@ Pattern Parser::parseSimpleMessage(UErrorCode& status) {
                 break;
             }
             // Don't loop infinitely
-            if (errors.hasSyntaxError()) {
+            if (errors.hasSyntaxError() || U_FAILURE(status)) {
                 break;
             }
         }
@@ -1720,6 +1986,22 @@ Pattern Parser::parseSimpleMessage(UErrorCode& status) {
     return result.build(status);
 }
 
+void Parser::parseVariant(UErrorCode& status) {
+    CHECK_ERROR(status);
+
+    // At least one key is required
+    SelectorKeys keyList(parseNonEmptyKeys(status));
+
+    // parseNonEmptyKeys() consumes any trailing whitespace,
+    // so the pattern can be consumed next.
+
+    // Restore precondition before calling parsePattern()
+    // (which must return a non-null value)
+    CHECK_BOUNDS(status);
+    Pattern rhs = parseQuotedPattern(status);
+
+    dataModel.addVariant(std::move(keyList), std::move(rhs), status);
+}
 
 /*
   Consume a `selectors` (matching the nonterminal in the grammar),
@@ -1739,22 +2021,25 @@ void Parser::parseSelectors(UErrorCode& status) {
     // Parse selectors
     // "Backtracking" is required here. It's not clear if whitespace is
     // (`[s]` selector) or (`[s]` variant)
-    while (isWhitespace(peek()) || peek() == LEFT_CURLY_BRACE) {
+    while (isWhitespace(peek()) || peek() == DOLLAR) {
-        parseOptionalWhitespace(status);
+        int32_t whitespaceStart = index;
+        parseRequiredWhitespace(status);
         // Restore precondition
         CHECK_BOUNDS(status);
-        if (peek() != LEFT_CURLY_BRACE) {
+        if (peek() != DOLLAR) {
             // This is not necessarily an error, but rather,
             // means the whitespace we parsed was the optional
             // whitespace preceding the first variant, not the
-            // optional whitespace preceding a subsequent expression.
+            // required whitespace preceding a subsequent variable.
+            // In that case, "push back" the whitespace.
+            normalizedInput.truncate(normalizedInput.length() - 1);
+            index = whitespaceStart;
             break;
         }
-        Expression expression;
+        VariableName var = parseVariableName(status);
-        expression = parseExpression(status);
         empty = false;
 
-        dataModel.addSelector(std::move(expression), status);
+        dataModel.addSelector(std::move(var), status);
         CHECK_ERROR(status);
     }
 
@@ -1770,27 +2055,29 @@ void Parser::parseSelectors(UErrorCode& status) {
         }                                          \
 
     // Parse variants
-    while (isWhitespace(peek()) || isKeyStart(peek())) {
+    // matcher = match-statement s variant *(o variant)
-        // Trailing whitespace is allowed
+
-        parseOptionalWhitespace(status);
+    // Parse first variant
+    parseRequiredWhitespace(status);
+    if (!inBounds()) {
+        ERROR(status);
+        return;
+    }
+    parseVariant(status);
+    if (!inBounds()) {
+        // Not an error; there might be only one variant
+        return;
+    }
+
+    while (isWhitespace(peek()) || isBidiControl(peek()) || isKeyStart(peek())) {
+        parseOptionalWhitespace();
+        // Restore the precondition.
+        // Trailing whitespace is allowed.
         if (!inBounds()) {
             return;
         }
 
-        // At least one key is required
+        parseVariant(status);
-        SelectorKeys keyList(parseNonEmptyKeys(status));
-
-        CHECK_ERROR(status);
-
-        // parseNonEmptyKeys() consumes any trailing whitespace,
-        // so the pattern can be consumed next.
-
-        // Restore precondition before calling parsePattern()
-        // (which must return a non-null value)
-        CHECK_BOUNDS(status);
-        Pattern rhs = parseQuotedPattern(status);
-
-        dataModel.addVariant(std::move(keyList), std::move(rhs), status);
 
         // Restore the precondition, *without* erroring out if we've
         // reached the end of input. That's because it's valid for the
@@ -1799,6 +2086,10 @@ void Parser::parseSelectors(UErrorCode& status) {
         // Because if we don't check it here, the `isWhitespace()` call in
         // the loop head will read off the end of the input string.
         CHECK_END_OF_INPUT
+
+        if (errors.hasSyntaxError() || U_FAILURE(status)) {
+            break;
+        }
     }
 }
 
@@ -1871,7 +2162,7 @@ void Parser::parse(UParseError &parseErrorResult, UErrorCode& status) {
     bool complex = false;
     // First, "look ahead" to determine if this is a simple or complex
     // message. To do that, check the first non-whitespace character.
-    while (inBounds(index) && isWhitespace(peek())) {
+    while (inBounds(index) && (isWhitespace(peek()) || isBidiControl(peek()))) {
         next();
     }
 
@@ -1891,10 +2182,10 @@ void Parser::parse(UParseError &parseErrorResult, UErrorCode& status) {
     // Message can be empty, so we need to only look ahead
     // if we know it's non-empty
     if (complex) {
-        parseOptionalWhitespace(status);
+        parseOptionalWhitespace();
         parseDeclarations(status);
         parseBody(status);
-        parseOptionalWhitespace(status);
+        parseOptionalWhitespace();
     } else {
         // Simple message
         // For normalization, quote the pattern
@@ -1926,3 +2217,4 @@ U_NAMESPACE_END
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
 
+#endif /* #if !UCONFIG_NO_NORMALIZATION */

deps/icu-small/source/i18n/messageformat2_parser.h

@@ -10,12 +10,15 @@
 
 #include "unicode/messageformat2_data_model.h"
 #include "unicode/parseerr.h"
+#include "unicode/uniset.h"
 
 #include "messageformat2_allocation.h"
 #include "messageformat2_errors.h"
 
 #if U_SHOW_CPLUSPLUS_API
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
@@ -54,6 +57,26 @@ namespace message2 {
             }
     };
 
+
+    // Initialization of UnicodeSets
+    namespace unisets {
+        enum Key {
+            CONTENT,
+            WHITESPACE,
+            BIDI,
+            ALPHA,
+            DIGIT,
+            NAME_START,
+            NAME_CHAR,
+            TEXT,
+            QUOTED,
+            ESCAPABLE,
+            UNISETS_KEY_COUNT
+        };
+
+    U_I18N_API const UnicodeSet* get(Key key, UErrorCode& status);
+    }
+
     // Parser class (private)
     class Parser : public UMemory {
     public:
@@ -82,8 +105,23 @@ namespace message2 {
 	    UChar   postContext[U_PARSE_CONTEXT_LEN];
 	} MessageParseError;
 
-	Parser(const UnicodeString &input, MFDataModel::Builder& dataModelBuilder, StaticErrors& e, UnicodeString& normalizedInputRef)
+	Parser(const UnicodeString &input,
-	  : source(input), index(0), errors(e), normalizedInput(normalizedInputRef), dataModel(dataModelBuilder) {
+               MFDataModel::Builder& dataModelBuilder,
+               StaticErrors& e,
+               UnicodeString& normalizedInputRef,
+               UErrorCode& status)
+            : contentChars(unisets::get(unisets::CONTENT, status)),
+              whitespaceChars(unisets::get(unisets::WHITESPACE, status)),
+              bidiControlChars(unisets::get(unisets::BIDI, status)),
+              alphaChars(unisets::get(unisets::ALPHA, status)),
+              digitChars(unisets::get(unisets::DIGIT, status)),
+              nameStartChars(unisets::get(unisets::NAME_START, status)),
+              nameChars(unisets::get(unisets::NAME_CHAR, status)),
+              textChars(unisets::get(unisets::TEXT, status)),
+              quotedChars(unisets::get(unisets::QUOTED, status)),
+              escapableChars(unisets::get(unisets::ESCAPABLE, status)),
+            source(input), index(0), errors(e), normalizedInput(normalizedInputRef), dataModel(dataModelBuilder) {
+            (void) status;
 	  parseError.line = 0;
 	  parseError.offset = 0;
 	  parseError.lengthBeforeCurrentLine = 0;
@@ -91,6 +129,20 @@ namespace message2 {
 	  parseError.postContext[0] = '\0';
 	}
 
+        bool isContentChar(UChar32) const;
+        bool isBidiControl(UChar32) const;
+        bool isWhitespace(UChar32) const;
+        bool isTextChar(UChar32) const;
+        bool isQuotedChar(UChar32) const;
+        bool isEscapableChar(UChar32) const;
+        bool isAlpha(UChar32) const;
+        bool isDigit(UChar32) const;
+        bool isNameStart(UChar32) const;
+        bool isNameChar(UChar32) const;
+        bool isUnquotedStart(UChar32) const;
+        bool isLiteralStart(UChar32) const;
+        bool isKeyStart(UChar32) const;
+
 	static void translateParseError(const MessageParseError&, UParseError&);
 	static void setParseError(MessageParseError&, uint32_t);
 	void maybeAdvanceLine();
@@ -101,10 +153,12 @@ namespace message2 {
         void parseLocalDeclaration(UErrorCode&);
         void parseInputDeclaration(UErrorCode&);
         void parseSelectors(UErrorCode&);
+        void parseVariant(UErrorCode&);
 
-	void parseWhitespaceMaybeRequired(bool, UErrorCode&);
+	void parseRequiredWS(UErrorCode&);
 	void parseRequiredWhitespace(UErrorCode&);
-	void parseOptionalWhitespace(UErrorCode&);
+	void parseOptionalBidi();
+	void parseOptionalWhitespace();
 	void parseToken(UChar32, UErrorCode&);
 	void parseTokenWithWhitespace(UChar32, UErrorCode&);
 	void parseToken(const std::u16string_view&, UErrorCode&);
@@ -149,6 +203,18 @@ namespace message2 {
         bool inBounds(uint32_t i) const { return source.moveIndex32(index, i) < source.length(); }
         bool allConsumed() const { return (int32_t) index == source.length(); }
 
+        // UnicodeSets for checking character ranges
+        const UnicodeSet* contentChars;
+        const UnicodeSet* whitespaceChars;
+        const UnicodeSet* bidiControlChars;
+        const UnicodeSet* alphaChars;
+        const UnicodeSet* digitChars;
+        const UnicodeSet* nameStartChars;
+        const UnicodeSet* nameChars;
+        const UnicodeSet* textChars;
+        const UnicodeSet* quotedChars;
+        const UnicodeSet* escapableChars;
+
 	// The input string
 	const UnicodeString &source;
 	// The current position within the input string -- counting in UChar32
@@ -165,8 +231,8 @@ namespace message2 {
 
 	// The parent builder
 	MFDataModel::Builder &dataModel;
-    }; // class Parser
 
+    }; // class Parser
 } // namespace message2
 
 U_NAMESPACE_END
@@ -175,6 +241,8 @@ U_NAMESPACE_END
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
 
+#endif /* #if !UCONFIG_NO_NORMALIZATION */
+
 #endif /* U_SHOW_CPLUSPLUS_API */
 
 #endif // MESSAGEFORMAT_PARSER_H

deps/icu-small/source/i18n/messageformat2_serializer.cpp

@@ -3,6 +3,8 @@
 
 #include "unicode/utypes.h"
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
@@ -244,11 +246,12 @@ void Serializer::serializeDeclarations() {
 
 void Serializer::serializeSelectors() {
     U_ASSERT(!dataModel.hasPattern());
-    const Expression* selectors = dataModel.getSelectorsInternal();
+    const VariableName* selectors = dataModel.getSelectorsInternal();
 
     emit(ID_MATCH);
     for (int32_t i = 0; i < dataModel.numSelectors(); i++) {
-        // No whitespace needed here -- see `selectors` in the grammar
+        whitespace();
+        emit(DOLLAR);
         emit(selectors[i]);
     }
 }
@@ -256,6 +259,7 @@ void Serializer::serializeSelectors() {
 void Serializer::serializeVariants() {
     U_ASSERT(!dataModel.hasPattern());
     const Variant* variants = dataModel.getVariantsInternal();
+    whitespace();
     for (int32_t i = 0; i < dataModel.numVariants(); i++) {
         const Variant& v = variants[i];
         emit(v.getKeys());
@@ -285,3 +289,4 @@ U_NAMESPACE_END
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
 
+#endif /* #if !UCONFIG_NO_NORMALIZATION */

deps/icu-small/source/i18n/messageformat2_serializer.h

@@ -10,6 +10,8 @@
 
 #if U_SHOW_CPLUSPLUS_API
 
+#if !UCONFIG_NO_NORMALIZATION
+
 #if !UCONFIG_NO_FORMATTING
 
 #if !UCONFIG_NO_MF2
@@ -63,6 +65,8 @@ U_NAMESPACE_END
 
 #endif /* #if !UCONFIG_NO_FORMATTING */
 
+#endif /* #if !UCONFIG_NO_NORMALIZATION */
+
 #endif /* U_SHOW_CPLUSPLUS_API */
 
 #endif // MESSAGEFORMAT_SERIALIZER_H

deps/icu-small/source/i18n/nfrs.cpp

@@ -152,7 +152,7 @@ NFRuleSet::NFRuleSet(RuleBasedNumberFormat *_owner, UnicodeString* descriptions,
 
     UnicodeString& description = descriptions[index]; // !!! make sure index is valid
 
-    if (description.length() == 0) {
+    if (description.isEmpty()) {
         // throw new IllegalArgumentException("Empty rule set description");
         status = U_PARSE_ERROR;
         return;
@@ -177,16 +177,16 @@ NFRuleSet::NFRuleSet(RuleBasedNumberFormat *_owner, UnicodeString* descriptions,
         name.setTo(UNICODE_STRING_SIMPLE("%default"));
     }
 
-    if (description.length() == 0) {
+    if (description.isEmpty()) {
         // throw new IllegalArgumentException("Empty rule set description");
         status = U_PARSE_ERROR;
     }
 
     fIsPublic = name.indexOf(gPercentPercent, 2, 0) != 0;
 
-    if ( name.endsWith(gNoparse,8) ) {
+    if (name.endsWith(gNoparse, 8)) {
         fIsParseable = false;
-        name.truncate(name.length()-8); // remove the @noparse from the name
+        name.truncate(name.length() - 8); // remove the @noparse from the name
     }
 
     // all of the other members of NFRuleSet are initialized

deps/icu-small/source/i18n/nfrule.cpp

@@ -19,7 +19,6 @@
 
 #if U_HAVE_RBNF
 
-#include <limits>
 #include "unicode/localpointer.h"
 #include "unicode/rbnf.h"
 #include "unicode/tblcoll.h"
@@ -65,6 +64,7 @@ NFRule::~NFRule()
 
 static const char16_t gLeftBracket = 0x005b;
 static const char16_t gRightBracket = 0x005d;
+static const char16_t gVerticalLine = 0x007C;
 static const char16_t gColon = 0x003a;
 static const char16_t gZero = 0x0030;
 static const char16_t gNine = 0x0039;
@@ -147,6 +147,7 @@ NFRule::makeRules(UnicodeString& description,
         // then it's really shorthand for two rules (with one exception)
         LocalPointer<NFRule> rule2;
         UnicodeString sbuf;
+        int32_t orElseOp = description.indexOf(gVerticalLine);
 
         // we'll actually only split the rule into two rules if its
         // base value is an even multiple of its divisor (or it's one
@@ -194,9 +195,13 @@ NFRule::makeRules(UnicodeString& description,
             rule2->radix = rule1->radix;
             rule2->exponent = rule1->exponent;
 
-            // rule2's rule text omits the stuff in brackets: initialize
+            // By default, rule2's rule text omits the stuff in brackets,
-            // its rule text and substitutions accordingly
+            // unless it contains a | between the brackets.
+            // Initialize its rule text and substitutions accordingly.
             sbuf.append(description, 0, brack1);
+            if (orElseOp >= 0) {
+                sbuf.append(description, orElseOp + 1, brack2 - orElseOp - 1);
+            }
             if (brack2 + 1 < description.length()) {
                 sbuf.append(description, brack2 + 1, description.length() - brack2 - 1);
             }
@@ -207,7 +212,12 @@ NFRule::makeRules(UnicodeString& description,
         // the brackets themselves: initialize _its_ rule text and
         // substitutions accordingly
         sbuf.setTo(description, 0, brack1);
+        if (orElseOp >= 0) {
+            sbuf.append(description, brack1 + 1, orElseOp - brack1 - 1);
+        }
+        else {
             sbuf.append(description, brack1 + 1, brack2 - brack1 - 1);
+        }
         if (brack2 + 1 < description.length()) {
             sbuf.append(description, brack2 + 1, description.length() - brack2 - 1);
         }
@@ -286,18 +296,17 @@ NFRule::parseRuleDescriptor(UnicodeString& description, UErrorCode& status)
             // into "tempValue", skip periods, commas, and spaces,
             // stop on a slash or > sign (or at the end of the string),
             // and throw an exception on any other character
-            int64_t ll_10 = 10;
             while (p < descriptorLength) {
                 c = descriptor.charAt(p);
                 if (c >= gZero && c <= gNine) {
-                    int32_t single_digit = static_cast<int32_t>(c - gZero);
+                    int64_t digit = static_cast<int64_t>(c - gZero);
-                    if ((val > 0 && val > (std::numeric_limits<int64_t>::max() - single_digit) / 10) ||
+                    if ((val > 0 && val > (INT64_MAX - digit) / 10) ||
-                        (val < 0 && val < (std::numeric_limits<int64_t>::min() - single_digit) / 10)) {
+                        (val < 0 && val < (INT64_MIN - digit) / 10)) {
                         // out of int64_t range
                         status = U_PARSE_ERROR;
                         return;
                     }
-                    val = val * ll_10 + single_digit;
+                    val = val * 10 + digit;
                 }
                 else if (c == gSlash || c == gGreaterThan) {
                     break;
@@ -322,11 +331,17 @@ NFRule::parseRuleDescriptor(UnicodeString& description, UErrorCode& status)
             if (c == gSlash) {
                 val = 0;
                 ++p;
-                ll_10 = 10;
                 while (p < descriptorLength) {
                     c = descriptor.charAt(p);
                     if (c >= gZero && c <= gNine) {
-                        val = val * ll_10 + static_cast<int32_t>(c - gZero);
+                        int64_t digit = static_cast<int64_t>(c - gZero);
+                        if ((val > 0 && val > (INT64_MAX - digit) / 10) ||
+                            (val < 0 && val < (INT64_MIN - digit) / 10)) {
+                            // out of int64_t range
+                            status = U_PARSE_ERROR;
+                            return;
+                        }
+                        val = val * 10 + digit;
                     }
                     else if (c == gGreaterThan) {
                         break;
@@ -400,7 +415,7 @@ NFRule::parseRuleDescriptor(UnicodeString& description, UErrorCode& status)
     // finally, if the rule body begins with an apostrophe, strip it off
     // (this is generally used to put whitespace at the beginning of
     // a rule's rule text)
-    if (description.length() > 0 && description.charAt(0) == gTick) {
+    if (!description.isEmpty() && description.charAt(0) == gTick) {
         description.removeBetween(0, 1);
     }
 

deps/icu-small/source/i18n/number_decimalquantity.cpp

@@ -1133,7 +1133,7 @@ void DecimalQuantity::setDigitPos(int32_t position, int8_t value) {
 }
 
 void DecimalQuantity::shiftLeft(int32_t numDigits) {
-    if (!usingBytes && precision + numDigits > 16) {
+    if (!usingBytes && precision + numDigits >= 16) {
         switchStorage();
     }
     if (usingBytes) {

deps/icu-small/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;

deps/icu-small/source/i18n/number_mapper.cpp

@@ -74,9 +74,11 @@ MacroProps NumberPropertyMapper::oldToNew(const DecimalFormatProperties& propert
             !properties.currencyPluralInfo.fPtr.isNull() ||
             !properties.currencyUsage.isNull() ||
             warehouse.affixProvider.get().hasCurrencySign());
-    CurrencyUnit currency = resolveCurrency(properties, locale, status);
+    CurrencyUnit currency;
-    UCurrencyUsage currencyUsage = properties.currencyUsage.getOrDefault(UCURR_USAGE_STANDARD);
+    UCurrencyUsage currencyUsage;
     if (useCurrency) {
+        currency = resolveCurrency(properties, locale, status);
+        currencyUsage = properties.currencyUsage.getOrDefault(UCURR_USAGE_STANDARD);
         // NOTE: Slicing is OK.
         macros.unit = currency; // NOLINT
     }
@@ -129,6 +131,7 @@ MacroProps NumberPropertyMapper::oldToNew(const DecimalFormatProperties& propert
     }
     Precision precision;
     if (!properties.currencyUsage.isNull()) {
+        U_ASSERT(useCurrency);
         precision = Precision::constructCurrency(currencyUsage).withCurrency(currency);
     } else if (roundingIncrement != 0.0) {
         if (PatternStringUtils::ignoreRoundingIncrement(roundingIncrement, maxFrac)) {
@@ -276,7 +279,7 @@ MacroProps NumberPropertyMapper::oldToNew(const DecimalFormatProperties& propert
         exportedProperties->maximumIntegerDigits = maxInt == -1 ? INT32_MAX : maxInt;
 
         Precision rounding_;
-        if (precision.fType == Precision::PrecisionType::RND_CURRENCY) {
+        if (useCurrency && precision.fType == Precision::PrecisionType::RND_CURRENCY) {
             rounding_ = precision.withCurrency(currency, status);
         } else {
             rounding_ = precision;