From e76698ae59da551df9af58bbb0c8ae04430e8466 Mon Sep 17 00:00:00 2001
From: Vicki Pfau <vi@endrift.com>
Date: Sun, 15 May 2022 21:27:37 -0700
Subject: [PATCH] Util: Add latin1ToUtf8

---
 include/mgba-util/string.h |  1 +
 src/util/string.c          | 48 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 49 insertions(+)

diff --git a/include/mgba-util/string.h b/include/mgba-util/string.h
index cc7169bf2..de7de640f 100644
--- a/include/mgba-util/string.h
+++ b/include/mgba-util/string.h
@@ -33,6 +33,7 @@ int utfcmp(const uint16_t* utf16, const char* utf8, size_t utf16Length, size_t u
 char* utf16to8(const uint16_t* utf16, size_t length);
 uint32_t utf8Char(const char** unicode, size_t* length);
 uint32_t utf16Char(const uint16_t** unicode, size_t* length);
+char* latin1ToUtf8(const char* latin1, size_t length);
 char* gbkToUtf8(const char* gbk, size_t length);
 size_t utf8strlen(const char* string);
 
diff --git a/src/util/string.c b/src/util/string.c
index a561a5fb1..fdb5305d4 100644
--- a/src/util/string.c
+++ b/src/util/string.c
@@ -285,6 +285,54 @@ char* utf16to8(const uint16_t* utf16, size_t length) {
 	return newUTF8;
 }
 
+char* latin1ToUtf8(const char* latin1, size_t length) {
+	char* utf8 = NULL;
+	char* utf8Offset = NULL;
+	size_t offset;
+	char buffer[4];
+	size_t utf8TotalBytes = 0;
+	size_t utf8Length = 0;
+	for (offset = 0; offset < length; ++offset) {
+		if (length == 0) {
+			break;
+		}
+		uint8_t unichar = latin1[offset];
+		size_t bytes = toUtf8(unichar, buffer);
+		utf8Length += bytes;
+		if (!utf8) {
+			utf8 = malloc(length);
+			if (!utf8) {
+				return NULL;
+			}
+			utf8TotalBytes = length;
+			memcpy(utf8, buffer, bytes);
+			utf8Offset = utf8 + bytes;
+		} else if (utf8Length < utf8TotalBytes) {
+			memcpy(utf8Offset, buffer, bytes);
+			utf8Offset += bytes;
+		} else if (utf8Length >= utf8TotalBytes) {
+			ptrdiff_t o = utf8Offset - utf8;
+			char* newUTF8 = realloc(utf8, utf8TotalBytes * 2);
+			utf8Offset = o + newUTF8;
+			if (!newUTF8) {
+				free(utf8);
+				return 0;
+			}
+			utf8 = newUTF8;
+			memcpy(utf8Offset, buffer, bytes);
+			utf8Offset += bytes;
+		}
+	}
+
+	char* newUTF8 = realloc(utf8, utf8Length + 1);
+	if (!newUTF8) {
+		free(utf8);
+		return 0;
+	}
+	newUTF8[utf8Length] = '\0';
+	return newUTF8;
+}
+
 extern const uint16_t gbkUnicodeTable[];
 
 char* gbkToUtf8(const char* gbk, size_t length) {