libtxt: Clone an ICU line break iterator for each Paragraph/WordBreaker (#22594)

third_party/txt/benchmarks/paragraph_benchmarks.cc

@@ -405,7 +405,7 @@ BENCHMARK_DEFINE_F(ParagraphFixture, AddStyleRun)(benchmark::State& state) {
   paint.wordSpacing = text_style.word_spacing;
 
   minikin::LineBreaker breaker;
-  breaker.setLocale(icu::Locale(), nullptr);
+  breaker.setLocale();
   breaker.resize(text.size());
   memcpy(breaker.buffer(), text.data(), text.size() * sizeof(text[0]));
   breaker.setText();

third_party/txt/src/minikin/LineBreaker.cpp

@@ -68,10 +68,10 @@ const float SHRINKABILITY = 1.0 / 3.0;
 // not precisely match the postBreak width.
 const float LIBTXT_WIDTH_ADJUST = 0.00001;
 
-void LineBreaker::setLocale(const icu::Locale& locale, Hyphenator* hyphenator) {
-  mWordBreaker.setLocale(locale);
-  mLocale = locale;
-  mHyphenator = hyphenator;
+void LineBreaker::setLocale() {
+  mWordBreaker.setLocale();
+  mLocale = icu::Locale();
+  mHyphenator = nullptr;
 }
 
 void LineBreaker::setText() {

third_party/txt/src/minikin/LineBreaker.h

@@ -97,7 +97,10 @@ class LineBreaker {
   // could be here but it's better for performance that it's upstream because of
   // the cost of constructing and comparing the ICU Locale object.
   // Note: caller is responsible for managing lifetime of hyphenator
-  void setLocale(const icu::Locale& locale, Hyphenator* hyphenator);
+  //
+  // libtxt extension: always use the default locale so that a cached instance
+  // of the ICU break iterator can be reused.
+  void setLocale();
 
   void resize(size_t size) {
     mTextBuf.resize(size);

third_party/txt/src/minikin/WordBreaker.cpp

@@ -31,9 +31,19 @@ namespace minikin {
 const uint32_t CHAR_SOFT_HYPHEN = 0x00AD;
 const uint32_t CHAR_ZWJ = 0x200D;
 
-void WordBreaker::setLocale(const icu::Locale& locale) {
+// libtxt extension: avoid the cost of initializing new ICU break iterators
+// by constructing a global iterator using the default locale and then
+// creating a clone for each WordBreaker instance.
+static std::once_flag gLibtxtBreakIteratorInitFlag;
+static icu::BreakIterator* gLibtxtDefaultBreakIterator = nullptr;
+
+void WordBreaker::setLocale() {
   UErrorCode status = U_ZERO_ERROR;
-  mBreakIterator.reset(icu::BreakIterator::createLineInstance(locale, status));
+  std::call_once(gLibtxtBreakIteratorInitFlag, [&status] {
+    gLibtxtDefaultBreakIterator =
+        icu::BreakIterator::createLineInstance(icu::Locale(), status);
+  });
+  mBreakIterator.reset(gLibtxtDefaultBreakIterator->clone());
   // TODO: handle failure status
   if (mText != nullptr) {
     mBreakIterator->setText(&mUText, status);

third_party/txt/src/minikin/WordBreaker.h

@@ -33,7 +33,9 @@ class WordBreaker {
  public:
   ~WordBreaker() { finish(); }
 
-  void setLocale(const icu::Locale& locale);
+  // libtxt extension: always use the default locale so that a cached instance
+  // of the ICU break iterator can be reused.
+  void setLocale();
 
   void setText(const uint16_t* data, size_t size);
 

third_party/txt/src/txt/paragraph_txt.cc

@@ -229,7 +229,7 @@ void ParagraphTxt::CodeUnitRun::Shift(double delta) {
 }
 
 ParagraphTxt::ParagraphTxt() {
-  breaker_.setLocale(icu::Locale(), nullptr);
+  breaker_.setLocale();
 }
 
 ParagraphTxt::~ParagraphTxt() = default;