More UTF-16 stuff

luprex/cpp/drv/drvutil.cpp

@@ -109,6 +109,34 @@ static int buffer_codepoint_utf8(char32_t scp, char *buffer) {
     }
 }
 
+static int32_t read_codepoint_utf16(std::u16string_view &source) {
+    if (source.empty()) return -1;
+    
+    int32_t word0 = ((const uint16_t *)source.data())[0];
+    source.remove_prefix(1);
+
+    if (word0 < 0xD800) {
+        return word0;
+    } else if (word0 < 0xDC00) {
+        if (source.empty()) {
+            return -2;
+        }
+        int32_t word1 = ((const uint16_t *)source.data())[0];
+        if ((word1 < 0xDC00)||(word1 > 0xDFFF)) {
+            return -2;
+        }
+        int32_t part1 = word0 & 0x3FF;
+        int32_t part2 = word1 & 0x3FF;
+        int32_t result = ((part1 << 10) | part2) + 0x10000;
+        source.remove_prefix(1);
+        return result;
+    } else if (word0 < 0xE000) {
+        return -2;
+    } else {
+        return word0;
+    }
+}
+
 static int32_t read_codepoint_utf8(std::string_view &source) {
     size_t size = source.size();
     if (size == 0) return -1;
@@ -218,6 +246,18 @@ std::u16string utf8_to_ucs2(std::string_view s, int *consumed) {
     return result.substr(0, len);
 }
 
+std::string utf16_to_utf8(std::u16string_view s) {
+    std::string result(s.size() * 4, 0);
+    int len = 0;
+    while (true) {
+        int codepoint = read_codepoint_utf16(s);
+        if (codepoint == -1) break;
+        if (codepoint < 0) continue;
+        len += buffer_codepoint_utf8(codepoint, &result[len]);
+    }
+    return result.substr(0, len);
+}
+
 static std::vector<std::string> parse_control_lst(std::string_view ctrl) {
     std::vector<std::string> result;
     while (!ctrl.empty()) {

luprex/cpp/drv/drvutil.hpp

@@ -55,7 +55,7 @@ bool is_single_wchar_t(char32_t c);
 //
 std::string utf32_to_utf8(const std::u32string &cps);
 
-// Convert a UTF8 string to a codepoint string.
+// Convert a UTF8 string to a UTF32 string.
 // 
 // If the UTF8 string contains invalid sequences, they're silently dropped.
 // Some of the bytes may not be consumed, if the source ends with an unfinished
@@ -73,6 +73,12 @@ std::u32string utf8_to_utf32(std::string_view source, int *consumed);
 //
 std::u16string utf8_to_ucs2(std::string_view source, int *consumed);
 
+// Convert a UTF16 string to a UTF8 string.
+//
+// If the UTF16 string contains invalid sequences, they're silently dropped.
+//
+std::string utf16_to_utf8(std::u16string_view source);
+
 // Get a system error message, in an OS-independent manner.
 //
 // These versions of strerror is thread-safe, and it never fails