Creates BreakIterator::GetWordBreakStatus.

base/i18n/break_iterator.cc

Index: base/i18n/break_iterator.cc
diff --git a/base/i18n/break_iterator.cc b/base/i18n/break_iterator.cc
index e2ed667572f52d8b033cb9899186d73852846755..acf5342a5e2664a0c09d2b9a33d79c5a15dd69b1 100644
--- a/base/i18n/break_iterator.cc
+++ b/base/i18n/break_iterator.cc
@@ -138,10 +138,16 @@ bool BreakIterator::SetText(const base::char16* text, const size_t length) {
 }
 
 bool BreakIterator::IsWord() const {
-  int32_t status = ubrk_getRuleStatus(static_cast<UBreakIterator*>(iter_));
+  return IsWordBreak() == IS_WORD_BREAK;
+}
+
+BreakIterator::WordBreakStatus BreakIterator::IsWordBreak() const {
   if (break_type_ != BREAK_WORD && break_type_ != RULE_BASED)
-    return false;
-  return status != UBRK_WORD_NONE;
+    return IS_NOT_WORD_BREAK;

[please use gerrit instead, 2015/08/07 20:53:10]

Call ubrk_getRuleStatus() before the if statement.

The original code called urbk_getRuleStatus() before checking break_type_. This
seems unusual, but there could be a good reason for it. Either way, the code
that's using IsWord() implicitly assumes that ubrk_getRuleStatus() is always
called inside of IsWord(). Let's not break that assumption.

[Julius, 2015/08/10 16:06:37]

Done.

+  return ubrk_getRuleStatus(static_cast<UBreakIterator*>(iter_)) ==
+                 UBRK_WORD_NONE
+             ? IS_SKIPPABLE_WORD
+             : IS_WORD_BREAK;
 }
 
 bool BreakIterator::IsEndOfWord(size_t position) const {