From fc4547b412e28164af1bf8981234c6af959ccc0b Mon Sep 17 00:00:00 2001
From: Joel Klinghed <the_jk@spawned.biz>
Date: Tue, 13 Jun 2023 10:07:16 +0200
Subject: WIP

---
 utf/src/utf16.cc | 67 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 67 insertions(+)
 create mode 100644 utf/src/utf16.cc

(limited to 'utf/src/utf16.cc')

diff --git a/utf/src/utf16.cc b/utf/src/utf16.cc
new file mode 100644
index 0000000..43595bf
--- /dev/null
+++ b/utf/src/utf16.cc
@@ -0,0 +1,67 @@
+#include "utf16.hh"
+
+#include "utf_error.hh"
+
+namespace utf {
+
+namespace {
+
+inline bool is_high_surrogate(uint16_t c) {
+  return c >= 0xd800 && c <= 0xdbff;
+}
+
+inline bool is_low_surrogate(uint16_t c) {
+  return c >= 0xdc00 && c <= 0xdfff;
+}
+
+}  // namespace
+
+uint32_t read16be(std::string_view 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
+               | static_cast<uint16_t>(data[offset + 1] & 0xff);
+  if (is_high_surrogate(c)) {
+    if (data.size() - offset < 4)
+      return NEED_MORE;
+    uint16_t d = static_cast<uint16_t>(data[offset + 2]) << 8
+                 | static_cast<uint16_t>(data[offset + 3] & 0xff);
+    if (is_low_surrogate(d)) {
+      offset += 4;
+      return 0x10000
+          + (static_cast<uint32_t>(c & 0x3ff) << 10
+             | (d & 0x3ff));
+    }
+    return INVALID;
+  } else if (is_low_surrogate(c)) {
+    return INVALID;
+  }
+  offset += 2;
+  return c;
+}
+
+uint32_t read16le(std::string_view 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
+               | static_cast<uint16_t>(data[offset] & 0xff);
+  if (is_high_surrogate(c)) {
+    if (data.size() - offset < 4)
+      return NEED_MORE;
+    uint16_t d = static_cast<uint16_t>(data[offset + 3]) << 8
+                 | static_cast<uint16_t>(data[offset + 2] & 0xff);
+    if (is_low_surrogate(d)) {
+      offset += 4;
+      return 0x10000
+          + (static_cast<uint32_t>(c & 0x3ff) << 10
+             | (d & 0x3ff));
+    }
+    return INVALID;
+  } else if (is_low_surrogate(c)) {
+    return INVALID;
+  }
+  offset += 2;
+  return c;
+}
+
+}  // namespace utf
-- 
cgit v1.2.3-70-g09d2