Stop combining text runs which are connected by 'COMMON' blocks.

ui/gfx/text_utils.cc

Index: ui/gfx/text_utils.cc
diff --git a/ui/gfx/text_utils.cc b/ui/gfx/text_utils.cc
index a31ef3d3cc999cd844fb4042b17628b9db976bcf..c62b88b122ccefc2a3f6ee73cd822a408652650c 100644
--- a/ui/gfx/text_utils.cc
+++ b/ui/gfx/text_utils.cc
@@ -5,9 +5,33 @@
 #include "ui/gfx/text_utils.h"
 
 #include "base/i18n/char_iterator.h"
+#include "base/logging.h"
+#include "base/numerics/safe_conversions.h"
+#include "third_party/icu/source/common/unicode/uchar.h"
+#include "third_party/icu/source/common/unicode/utf16.h"
 
 namespace gfx {
 
+namespace {
+
+// Returns true if the code point |c| is a combining mark character in Unicode.
+bool CharIsMark(UChar32 c) {
+  int8_t char_type = u_charType(c);
+  return char_type == U_NON_SPACING_MARK || char_type == U_ENCLOSING_MARK ||
+         char_type == U_COMBINING_SPACING_MARK;
+}
+
+// Gets the code point of |str| at the given code unit position |index|. If
+// |index| is a surrogate code unit, returns the whole code point (unless the
+// code unit is unpaired, in which case it just returns the surrogate value).
+UChar32 GetCodePointAt(const base::string16& str, size_t index) {
+  UChar32 c;
+  U16_GET(str.data(), 0, index, str.size(), c);
+  return c;
+}
+
+}  // namespace
+
 base::string16 RemoveAcceleratorChar(const base::string16& s,
                                      base::char16 accelerator_char,
                                      int* accelerated_char_pos,
@@ -46,4 +70,40 @@ base::string16 RemoveAcceleratorChar(const base::string16& s,
   return accelerator_removed;
 }
 
+size_t FindValidBoundaryBefore(const base::string16& text, size_t index) {
+  size_t length = text.length();
+  DCHECK_LE(index, length);
+  if (index == length)
+    return index;
+
+  // If |index| straddles a combining character sequence, go back until we find
+  // a base character.
+  while (index > 0 && CharIsMark(GetCodePointAt(text, index)))
+    --index;
+
+  // If |index| straddles a UTF-16 surrogate pair, go back.
+  U16_SET_CP_START(text.data(), 0, index);
+  return index;
+}
+
+size_t FindValidBoundaryAfter(const base::string16& text, size_t index) {
+  DCHECK_LE(index, text.length());
+  if (index == text.length())
+    return index;
+
+  int32_t text_index = base::checked_cast<int32_t>(index);
+  int32_t text_length = base::checked_cast<int32_t>(text.length());
+
+  // If |index| straddles a combining character sequence, go forward until we
+  // find a base character.
+  while (text_index < text_length &&
+         CharIsMark(GetCodePointAt(text, text_index))) {
+    ++text_index;
+  }
+
+  // If |index| straddles a UTF-16 surrogate pair, go forward.
+  U16_SET_CP_LIMIT(text.data(), 0, text_index, text_length);
+  return static_cast<size_t>(text_index);
+}
+
 }  // namespace gfx