Workaround for navigating by word in the omnibox on Windows.

ui/accessibility/ax_text_utils.cc

 // Copyright (c) 2011 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "ui/accessibility/ax_text_utils.h"
 
+#include "base/i18n/break_iterator.h"
 #include "base/logging.h"
 #include "base/strings/string_util.h"
 
 namespace ui {
 
 // line_breaks is a Misnomer. Blink provides the start offsets of each line
 // not the line breaks.
 // TODO(nektar): Rename line_breaks a11y attribute and variable references.
 size_t FindAccessibleTextBoundary(const base::string16& text,
                                   const std::vector<int>& line_breaks,
                                   TextBoundaryType boundary,
                                   size_t start_offset,
                                   TextBoundaryDirection direction) {
   size_t text_size = text.size();
   DCHECK_LE(start_offset, text_size);
 
   if (boundary == CHAR_BOUNDARY) {
     if (direction == FORWARDS_DIRECTION && start_offset < text_size)
       return start_offset + 1;
     else
       return start_offset;
-  } else if (boundary == LINE_BOUNDARY) {
+  }
+
+  base::i18n::BreakIterator word_iter(text,
+                                      base::i18n::BreakIterator::BREAK_WORD);
+  if (boundary == WORD_BOUNDARY) {
+    if (!word_iter.Init())
+      return start_offset;
+  }
+
+  if (boundary == LINE_BOUNDARY) {
     if (direction == FORWARDS_DIRECTION) {
       for (size_t j = 0; j < line_breaks.size(); ++j) {
           size_t line_break = line_breaks[j] >= 0 ? line_breaks[j] : 0;
         if (line_break > start_offset)
           return line_break;
       }
       return text_size;
     } else {
       for (size_t j = line_breaks.size(); j != 0; --j) {
         size_t line_break = line_breaks[j - 1] >= 0 ? line_breaks[j - 1] : 0;
@@ skipping 16 matching lines @@
         return 0;
       pos = result - 1;
     }
 
     switch (boundary) {
       case CHAR_BOUNDARY:
       case LINE_BOUNDARY:
         NOTREACHED();  // These are handled above.
         break;
       case WORD_BOUNDARY:
-        if (base::IsUnicodeWhitespace(text[pos]))
-          return result;
+        if (word_iter.IsStartOfWord(result)) {
+          // If we are searching forward and we are still at the start offset,
+          // we need to find the next word.
+          if (direction == BACKWARDS_DIRECTION || result != start_offset)
+            return result;
+        }
         break;
       case PARAGRAPH_BOUNDARY:
         if (text[pos] == '\n')
           return result;
         break;
       case SENTENCE_BOUNDARY:
         if ((text[pos] == '.' || text[pos] == '!' || text[pos] == '?') &&
             (pos == text_size - 1 ||
              base::IsUnicodeWhitespace(text[pos + 1]))) {
           return result;
         }
         break;
       case ALL_BOUNDARY:
       default:
         break;
     }
 
     if (direction == FORWARDS_DIRECTION) {
       result++;
     } else {
       result--;
     }
   }
 }
 
-}  // Namespace ui
+}  // namespace ui