char16_t formatting support

hecl/extern/athena

@@ -1 +1 @@
-Subproject commit 483870850c24956d9549bad5149a97e80de42f2a
+Subproject commit dd8b60f779a97281ebf9f7778c4c0365fdc54457

hecl/include/hecl/hecl.hpp

@@ -52,7 +52,9 @@ extern logvisor::Module LogModule;
 #endif
 
 std::string WideToUTF8(const std::wstring& src);
+std::string Char16ToUTF8(const std::u16string& src);
 std::wstring UTF8ToWide(const std::string& src);
+std::u16string UTF8ToChar16(const std::string& src);
 
 /* humanize_number port from FreeBSD's libutil */
 enum class HNFlags
@@ -424,6 +426,7 @@ std::string Format(const char* format, ...);
 
 std::wstring WideFormat(const wchar_t* format, ...);
 
+std::u16string Char16Format(const wchar_t* format, ...);
 
 static inline bool CheckFreeSpace(const SystemChar* path, size_t reqSz)
 {

hecl/lib/WideStringConvert.cpp

@@ -22,6 +22,24 @@ std::string WideToUTF8(const std::wstring& src)
     return retval;
 }
 
+std::string Char16ToUTF8(const std::u16string& src)
+{
+    std::string retval;
+    retval.reserve(src.length());
+    for (char16_t ch : src)
+    {
+        utf8proc_uint8_t mb[4];
+        utf8proc_ssize_t c = utf8proc_encode_char(utf8proc_int32_t(ch), mb);
+        if (c < 0)
+        {
+            LogModule.report(logvisor::Warning, "invalid UTF-8 character while encoding");
+            return retval;
+        }
+        retval.append(reinterpret_cast<char*>(mb), c);
+    }
+    return retval;
+}
+
 std::wstring UTF8ToWide(const std::string& src)
 {
     std::wstring retval;
@@ -42,4 +60,24 @@ std::wstring UTF8ToWide(const std::string& src)
     return retval;
 }
 
+std::u16string UTF8ToChar16(const std::string& src)
+{
+    std::u16string retval;
+    retval.reserve(src.length());
+    const utf8proc_uint8_t* buf = reinterpret_cast<const utf8proc_uint8_t*>(src.c_str());
+    while (*buf)
+    {
+        utf8proc_int32_t wc;
+        utf8proc_ssize_t len = utf8proc_iterate(buf, -1, &wc);
+        if (len < 0)
+        {
+            LogModule.report(logvisor::Warning, "invalid UTF-8 character while decoding");
+            return retval;
+        }
+        buf += len;
+        retval += char16_t(wc);
+    }
+    return retval;
+}
+
 }

hecl/lib/hecl.cpp

@@ -65,6 +65,20 @@ std::wstring WideFormat(const wchar_t* format, ...)
     return std::wstring(resultBuf, printSz);
 }
 
+std::u16string Char16Format(const wchar_t* format, ...)
+{
+    wchar_t resultBuf[FORMAT_BUF_SZ];
+    va_list va;
+    va_start(va, format);
+    int printSz = vswprintf(resultBuf, FORMAT_BUF_SZ, format, va);
+    va_end(va);
+    std::u16string res;
+    res.reserve(printSz);
+    for (size_t i=0 ; i<printSz ; ++i)
+        res.push_back(resultBuf[i]);
+    return res;
+}
+
 void SanitizePath(std::string& path)
 {
     if (path.empty())