1 files changed, 33 insertions, 2 deletions
diff --git a/utf/src/utf8.cc b/utf/src/utf8.cc
index 54b0296..0e444ae 100644
--- a/utf/src/utf8.cc
+++ b/utf/src/utf8.cc
@@ -12,12 +12,12 @@ inline bool valid_codepoint(uint32_t c) {
 
 }  // namespace
 
-uint32_t read8(std::string_view data, std::size_t& offset) {
+uint32_t read8(std::span<uint8_t const> data, std::size_t& offset) {
   if (offset >= data.size())
     return NEED_MORE;
   uint32_t ret;
   uint8_t size;
-  switch (static_cast<uint8_t>(data[offset]) >> 4) {
+  switch (data[offset] >> 4) {
     case 15:
       if (data[offset] & 0x08)
         return INVALID;
@@ -65,4 +65,35 @@ uint32_t read8(std::string_view data, std::size_t& offset) {
   return ret;
 }
 
+bool write8(uint32_t codepoint, std::span<uint8_t> data, std::size_t& offset) {
+  if (offset >= data.size()) UNLIKELY {
+    return false;
+  }
+  if (codepoint < 0x80) {
+    data[offset++] = codepoint;
+  } else if (codepoint < 0x800) {
+    if (data.size() - offset < 2) UNLIKELY {
+      return false;
+    }
+    data[offset++] = 0xc0 | (codepoint >> 6);
+    data[offset++] = 0x80 | (codepoint & 0x3f);
+  } else if (codepoint < 0x10000) {
+    if (data.size() - offset < 3) UNLIKELY {
+      return false;
+    }
+    data[offset++] = 0xe0 | (codepoint >> 12);
+    data[offset++] = 0x80 | ((codepoint >> 6) & 0x3f);
+    data[offset++] = 0x80 | (codepoint & 0x3f);
+  } else {
+    if (data.size() - offset < 4) UNLIKELY {
+      return false;
+    }
+    data[offset++] = 0xf0 | (codepoint >> 18);
+    data[offset++] = 0x80 | ((codepoint >> 12) & 0x3f);
+    data[offset++] = 0x80 | ((codepoint >> 6) & 0x3f);
+    data[offset++] = 0x80 | (codepoint & 0x3f);
+  }
+  return true;
+}
+
 }  // namespace utf