...

1 files changed, 89 insertions, 26 deletions

diff --git a/include/cru/common/StringUtil.hpp b/include/cru/common/StringUtil.hpp
index 714f1d49..b8edc302 100644
--- a/include/cru/common/StringUtil.hpp
+++ b/include/cru/common/StringUtil.hpp
@@ -10,51 +10,114 @@ class TextEncodeException : public std::runtime_error {
   using runtime_error::runtime_error;
 };
 
-inline bool IsSurrogatePair(char16_t c) { return c >= 0xD800 && c <= 0xDFFF; }
+inline bool IsUtf16SurrogatePairCodeUnit(char16_t c) {
+  return c >= 0xD800 && c <= 0xDFFF;
+}
 
-inline bool IsSurrogatePairLeading(char16_t c) {
+inline bool IsUtf16SurrogatePairLeading(char16_t c) {
   return c >= 0xD800 && c <= 0xDBFF;
 }
 
-inline bool IsSurrogatePairTrailing(char16_t c) {
+inline bool IsUtf16SurrogatePairTrailing(char16_t c) {
   return c >= 0xDC00 && c <= 0xDFFF;
 }
 
-class Utf16Iterator : public Object {
+CodePoint Utf8NextCodePoint(std::string_view str, Index current,
+                            Index* next_position);
+
+CodePoint Utf16NextCodePoint(std::u16string_view str, Index current,
+                             Index* next_position);
+CodePoint Utf16PreviousCodePoint(std::u16string_view str, Index current,
+                                 Index* previous_position);
+
+template <typename StringType>
+using NextCodePointFunctionType = CodePoint (*)(StringType, Index, Index*);
+
+template <typename StringType,
+          NextCodePointFunctionType<StringType> NextCodePointFunction>
+class CodePointIterator {
+ public:
+  using difference_type = Index;
+  using value_type = CodePoint;
+  using pointer = void;
+  using reference = value_type;
+  using iterator_category = std::forward_iterator_tag;
+
  public:
-  explicit Utf16Iterator(std::u16string_view string)
-      : string_(std::move(string)) {}
-  Utf16Iterator(std::u16string_view string, Index position)
-      : string_(std::move(string)), position_(position) {}
+  struct past_end_tag_t {};
 
-  CRU_DEFAULT_COPY(Utf16Iterator)
-  CRU_DEFAULT_MOVE(Utf16Iterator)
+  explicit CodePointIterator(StringType string)
+      : string_(std::move(string)), position_(0) {}
+  explicit CodePointIterator(StringType string, past_end_tag_t)
+      : string_(std::move(string)), position_(string_.size()) {}
 
-  ~Utf16Iterator() = default;
+  CRU_DEFAULT_COPY(CodePointIterator)
+  CRU_DEFAULT_MOVE(CodePointIterator)
+
+  ~CodePointIterator() = default;
 
  public:
-  void SetPositionToHead() { position_ = 0; }
-  void SetPosition(Index position) { position_ = position; }
+  StringType GetString() const { return string_; }
+  Index GetPosition() const { return position_; }
+
+  bool IsPastEnd() const {
+    return position_ == static_cast<Index>(string_.size());
+  }
 
-  // Backward current position and get previous code point. Return
-  // k_invalid_code_point if reach head. Throw TextEncodeException if encounter
-  // encoding problem.
-  CodePoint Previous();
+ public:
+  CodePointIterator begin() const { return *this; }
+  CodePointIterator end() const {
+    return CodePointIterator{string_, past_end_tag_t{}};
+  }
 
-  // Advance current position and get next code point. Return
-  // k_invalid_code_point if reach tail. Throw TextEncodeException if encounter
-  // encoding problem.
-  CodePoint Next();
+ public:
+  bool operator==(const CodePointIterator& other) const {
+    // You should compare iterator that iterate on the same string.
+    Expects(this->string_.data() == other.string_.data() &&
+            this->string_.size() == other.string_.size());
+    return this->position_ == other.position_;
+  }
+  bool operator!=(const CodePointIterator& other) const {
+    return !this->operator==(other);
+  }
+
+  CodePointIterator& operator++() {
+    Expects(!IsPastEnd());
+    Forward();
+    return *this;
+  }
+
+  CodePointIterator operator++(int) {
+    Expects(!IsPastEnd());
+    CodePointIterator old = *this;
+    Forward();
+    return old;
+  }
+
+  CodePoint operator*() const {
+    return NextCodePointFunction(string_, position_, &next_position_cache_);
+  }
 
-  Index CurrentPosition() const { return this->position_; }
+ private:
+  void Forward() {
+    if (next_position_cache_ > position_) {
+      position_ = next_position_cache_;
+    } else {
+      NextCodePointFunction(string_, position_, &position_);
+    }
+  }
 
  private:
-  std::u16string_view string_;
-  Index position_ = 0;
+  StringType string_;
+  Index position_;
+  mutable Index next_position_cache_;
 };
 
-Index PreviousIndex(std::u16string_view string, Index current);
-Index NextIndex(std::u16string_view string, Index current);
+using Utf8CodePointIterator =
+    CodePointIterator<std::string_view, &Utf8NextCodePoint>;
+
+using Utf16CodePointIterator =
+    CodePointIterator<std::u16string_view, &Utf16NextCodePoint>;
 
 std::string ToUtf8(const std::u16string& s);
 inline std::string ToUtf8(std::u16string_view s) {