Add function to find a UTF-16 position in a UTF-8 string.

2 files changed, 13 insertions, 0 deletions

diff --git a/src/UniConversion.cxx b/src/UniConversion.cxx
index 1287aa612..3b7472638 100644
--- a/src/UniConversion.cxx
+++ b/src/UniConversion.cxx
@@ -37,6 +37,18 @@ size_t UTF8Length(std::wstring_view wsv) {
 	return len;
 }
 
+size_t UTF8PositionFromUTF16Position(std::string_view u8Text, size_t positionUTF16) noexcept {
+	size_t positionUTF8 = 0;
+	for (size_t lengthUTF16 = 0; (positionUTF8 < u8Text.length()) && (lengthUTF16 < positionUTF16);) {
+		const unsigned char uch = u8Text[positionUTF8];
+		const unsigned int byteCount = UTF8BytesOfLead[uch];
+		lengthUTF16 += UTF16LengthFromUTF8ByteCount(byteCount);
+		positionUTF8 += byteCount;
+	}
+
+	return positionUTF8;
+}
+
 void UTF8FromUTF16(std::wstring_view wsv, char *putf, size_t len) {
 	size_t k = 0;
 	for (size_t i = 0; i < wsv.length() && wsv[i];) {
diff --git a/src/UniConversion.h b/src/UniConversion.h
index 1b84b8f81..e2be40776 100644
--- a/src/UniConversion.h
+++ b/src/UniConversion.h
@@ -15,6 +15,7 @@ const int UTF8MaxBytes = 4;
 const int unicodeReplacementChar = 0xFFFD;
 
 size_t UTF8Length(std::wstring_view wsv);
+size_t UTF8PositionFromUTF16Position(std::string_view u8Text, size_t positionUTF16) noexcept;
 void UTF8FromUTF16(std::wstring_view wsv, char *putf, size_t len);
 void UTF8FromUTF32Character(int uch, char *putf);
 size_t UTF16Length(std::string_view sv);