From e68e0d9a5130e8bc0b634572b7fd44b9bfc0f8ef Mon Sep 17 00:00:00 2001
From: crupest <crupest@outlook.com>
Date: Sat, 30 Oct 2021 21:08:43 +0800
Subject: ...

---
 include/cru/common/String.hpp     |  2 +
 include/cru/common/StringUtil.hpp | 78 ++++++++++++++++++++++++++++++++++++++-
 2 files changed, 78 insertions(+), 2 deletions(-)

(limited to 'include/cru/common')

diff --git a/include/cru/common/String.hpp b/include/cru/common/String.hpp
index c05ab6e0..544d24a4 100644
--- a/include/cru/common/String.hpp
+++ b/include/cru/common/String.hpp
@@ -171,6 +171,8 @@ class CRU_BASE_API String {
   }
 
  public:
+  void AppendCodePoint(CodePoint code_point);
+
   Utf16CodePointIterator CodePointIterator() const {
     return Utf16CodePointIterator(
         std::u16string_view(reinterpret_cast<char16_t*>(buffer_), size_));
diff --git a/include/cru/common/StringUtil.hpp b/include/cru/common/StringUtil.hpp
index 4291a0da..cd2f4e16 100644
--- a/include/cru/common/StringUtil.hpp
+++ b/include/cru/common/StringUtil.hpp
@@ -2,6 +2,7 @@
 #include "Base.hpp"
 
 #include <functional>
+#include <stdexcept>
 #include <string>
 #include <string_view>
 
@@ -121,8 +122,81 @@ using Utf16CodePointIterator =
 
 void CRU_BASE_API Utf8EncodeCodePointAppend(CodePoint code_point,
                                             std::string& str);
-void CRU_BASE_API Utf16EncodeCodePointAppend(CodePoint code_point,
-                                             std::u16string& str);
+
+namespace details {
+template <typename UInt, int number_of_bit, typename ReturnType>
+inline std::enable_if_t<std::is_unsigned_v<UInt>, ReturnType> ExtractBits(
+    UInt n) {
+  return static_cast<ReturnType>(n & ((1u << number_of_bit) - 1));
+}
+}  // namespace details
+
+template <typename TAppend>
+bool Utf8EncodeCodePointAppendWithFunc(CodePoint code_point, TAppend&& append) {
+  auto write_continue_byte = [&append](std::uint8_t byte6) {
+    append((1u << 7) + (((1u << 6) - 1) & byte6));
+  };
+
+  if (code_point >= 0 && code_point <= 0x007F) {
+    append(static_cast<char>(code_point));
+    return true;
+  } else if (code_point >= 0x0080 && code_point <= 0x07FF) {
+    std::uint32_t unsigned_code_point = code_point;
+    append(
+        static_cast<char>(details::ExtractBits<std::uint32_t, 5, std::uint8_t>(
+                              (unsigned_code_point >> 6)) +
+                          0b11000000));
+    write_continue_byte(details::ExtractBits<std::uint32_t, 6, std::uint8_t>(
+        unsigned_code_point));
+    return true;
+  } else if (code_point >= 0x0800 && code_point <= 0xFFFF) {
+    std::uint32_t unsigned_code_point = code_point;
+    append(
+        static_cast<char>(details::ExtractBits<std::uint32_t, 4, std::uint8_t>(
+                              (unsigned_code_point >> (6 * 2))) +
+                          0b11100000));
+    write_continue_byte(details::ExtractBits<std::uint32_t, 6, std::uint8_t>(
+        unsigned_code_point >> 6));
+    write_continue_byte(details::ExtractBits<std::uint32_t, 6, std::uint8_t>(
+        unsigned_code_point));
+    return true;
+  } else if (code_point >= 0x10000 && code_point <= 0x10FFFF) {
+    std::uint32_t unsigned_code_point = code_point;
+    append(
+        static_cast<char>(details::ExtractBits<std::uint32_t, 3, std::uint8_t>(
+                              (unsigned_code_point >> (6 * 3))) +
+                          0b11110000));
+    write_continue_byte(details::ExtractBits<std::uint32_t, 6, std::uint8_t>(
+        unsigned_code_point >> (6 * 2)));
+    write_continue_byte(details::ExtractBits<std::uint32_t, 6, std::uint8_t>(
+        unsigned_code_point >> 6));
+    write_continue_byte(details::ExtractBits<std::uint32_t, 6, std::uint8_t>(
+        unsigned_code_point));
+    return true;
+  } else {
+    return false;
+  }
+}
+
+template <typename TAppend>
+bool Utf16EncodeCodePointAppendWithFunc(CodePoint code_point,
+                                        TAppend&& append) {
+  if ((code_point >= 0 && code_point <= 0xD7FF) ||
+      (code_point >= 0xE000 && code_point <= 0xFFFF)) {
+    append(static_cast<char16_t>(code_point));
+    return true;
+  } else if (code_point >= 0x10000 && code_point <= 0x10FFFF) {
+    std::uint32_t u = code_point - 0x10000;
+    append(static_cast<char16_t>(
+        details::ExtractBits<std::uint32_t, 10, std::uint32_t>(u >> 10) +
+        0xD800u));
+    append(static_cast<char16_t>(
+        details::ExtractBits<std::uint32_t, 10, std::uint32_t>(u) + 0xDC00u));
+    return true;
+  } else {
+    return false;
+  }
+}
 
 std::string CRU_BASE_API ToUtf8(std::u16string_view s);
 std::u16string CRU_BASE_API ToUtf16(std::string_view s);
-- 
cgit v1.2.3