From 7dd49c6293172b494c78918507242cdb55d35137 Mon Sep 17 00:00:00 2001
From: Joel Klinghed <the_jk@spawned.biz>
Date: Sun, 21 Jan 2024 12:31:30 +0100
Subject: WIP

---
 utf/src/utf16.cc |  4 ++--
 utf/src/utf32.cc |  4 ++--
 utf/src/utf8.cc  | 35 +++++++++++++++++++++++++++++++++--
 3 files changed, 37 insertions(+), 6 deletions(-)

(limited to 'utf/src')

diff --git a/utf/src/utf16.cc b/utf/src/utf16.cc
index 43595bf..623c1be 100644
--- a/utf/src/utf16.cc
+++ b/utf/src/utf16.cc
@@ -16,7 +16,7 @@ inline bool is_low_surrogate(uint16_t c) {
 
 }  // namespace
 
-uint32_t read16be(std::string_view data, std::size_t& offset) {
+uint32_t read16be(std::span<uint8_t const> data, std::size_t& offset) {
   if (offset > data.size() || data.size() - offset < 2)
     return NEED_MORE;
   uint16_t c = static_cast<uint16_t>(data[offset]) << 8
@@ -40,7 +40,7 @@ uint32_t read16be(std::string_view data, std::size_t& offset) {
   return c;
 }
 
-uint32_t read16le(std::string_view data, std::size_t& offset) {
+uint32_t read16le(std::span<uint8_t const> data, std::size_t& offset) {
   if (offset > data.size() || data.size() - offset < 2)
     return NEED_MORE;
   uint16_t c = static_cast<uint16_t>(data[offset + 1]) << 8
diff --git a/utf/src/utf32.cc b/utf/src/utf32.cc
index cfa29b6..e33b0b4 100644
--- a/utf/src/utf32.cc
+++ b/utf/src/utf32.cc
@@ -12,7 +12,7 @@ inline bool valid_codepoint(uint32_t c) {
 
 }  // namespace
 
-uint32_t read32be(std::string_view data, std::size_t& offset) {
+uint32_t read32be(std::span<uint8_t const> data, std::size_t& offset) {
   if (offset > data.size() || data.size() - offset < 4)
     return NEED_MORE;
   uint32_t c = static_cast<uint32_t>(data[offset]) << 24
@@ -26,7 +26,7 @@ uint32_t read32be(std::string_view data, std::size_t& offset) {
   return INVALID;
 }
 
-uint32_t read32le(std::string_view data, std::size_t& offset) {
+uint32_t read32le(std::span<uint8_t const> data, std::size_t& offset) {
   if (offset > data.size() || data.size() - offset < 4)
     return NEED_MORE;
   uint32_t c = static_cast<uint32_t>(data[offset + 3]) << 24
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
-- 
cgit v1.2.3-70-g09d2