COMMON: Convert endianity when needed in Encoding.

2 files changed, 77 insertions, 3 deletions

diff --git a/common/encoding.cpp b/common/encoding.cpp
index 719887d35e..aeae520dbb 100644
--- a/common/encoding.cpp
+++ b/common/encoding.cpp
@@ -24,6 +24,7 @@
 #include "common/textconsole.h"
 #include "common/system.h"
 #include "common/translation.h"
+#include "common/endian.h"
 #include <errno.h>
 
 namespace Common {
@@ -38,6 +39,29 @@ Encoding::~Encoding() {
 	deinitIconv(_iconvHandle);
 }
 
+char *Encoding::switchEndian(const char *string, int length, int bitCount) {
+	assert(bitCount % 8 == 0);
+	assert(length % (bitCount / 8) == 0);
+	char *newString = (char *) malloc(length);
+	if (!newString) {
+		warning("Could not allocate memory for string conversion");
+		return nullptr;
+	}
+	if (bitCount == 16) {
+		int characterCount = length / 2;
+		for(int i = 0; i < characterCount ; i++)
+			((uint16 *) newString)[i] = SWAP_BYTES_16(((const uint16 *) string)[i]);
+		return newString;
+	} else if (bitCount == 32) {
+		int characterCount = length / 4;
+		for(int i = 0; i < characterCount ; i++)
+			((uint32 *) newString)[i] = SWAP_BYTES_32(((const uint32 *) string)[i]);
+		return newString;
+	} else {
+		return nullptr;
+	}
+}
+
 String Encoding::addUtfEndianness(const String &str) {
 	if (str.equalsIgnoreCase("utf-16") || str.equalsIgnoreCase("utf-32")) {
 #ifdef SCUMM_BIG_ENDIAN
@@ -102,6 +126,21 @@ char *Encoding::convertWithTransliteration(iconv_t iconvHandle, const String &to
 		memcpy(result, string, length);
 		return result;
 	}
+	if ((addUtfEndianness(to).equalsIgnoreCase("utf-16be") &&
+		addUtfEndianness(from).equalsIgnoreCase("utf-16le")) ||
+		(addUtfEndianness(to).equalsIgnoreCase("utf-16le") &&
+		addUtfEndianness(from).equalsIgnoreCase("utf-16be")) ||
+		(addUtfEndianness(to).equalsIgnoreCase("utf-32be") &&
+		addUtfEndianness(from).equalsIgnoreCase("utf-32le")) ||
+		(addUtfEndianness(to).equalsIgnoreCase("utf-32le") &&
+		addUtfEndianness(from).equalsIgnoreCase("utf-32be")))
+	{
+		// The encoding is the same, we just need to switch the endianness
+		if (to.hasPrefixIgnoreCase("utf-16"))
+			return switchEndian(string, length, 16);
+		else
+			return switchEndian(string, length, 32);
+	}
 	char *newString = nullptr;
 	String newFrom = from;
 	size_t newLength = length;
@@ -265,20 +304,34 @@ char *Encoding::convertTransManMapping(const char *to, const char *from, const c
 		free(partialResult);
 		return finalResult;
 	} else if (currentCharset.equalsIgnoreCase(to) && String(from).hasPrefixIgnoreCase("utf-32")) {
-		// We accept only the machine endianness
+		bool swapEndian = false;
+		char *newString = nullptr;
+
 #ifdef SCUMM_BIG_ENDIAN
 		if (String(from).hasSuffixIgnoreCase("LE"))
-			return nullptr;
+			swapEndian = true;
 #else
 		if (String(from).hasSuffixIgnoreCase("BE"))
-			return nullptr;
+			swapEndian = true;
 #endif
+		if (swapEndian) {
+			if (String(from).hasPrefixIgnoreCase("utf-16"))
+				newString = switchEndian(string, length, 16);
+			if (String(from).hasPrefixIgnoreCase("utf-32"))
+				newString = switchEndian(string, length, 32);
+			if (newString != nullptr)
+				string = newString;
+			else
+				return nullptr;
+		}
 		// We can do reverse mapping
 		const uint32 *mapping = TransMan.getCharsetMapping();
 		const uint32 *src = (const uint32 *) string;
 		char *result = (char *) calloc(sizeof(char), (length + 4));
 		if (!result) {
 			warning("Couldn't allocate memory for encoding conversion");
+			if (newString != nullptr)
+				free(newString);
 			return nullptr;
 		}
 		for (unsigned i = 0; i < length; i++) {
@@ -289,6 +342,8 @@ char *Encoding::convertTransManMapping(const char *to, const char *from, const c
 				}
 			}
 		}
+		if (newString != nullptr)
+			free(newString);
 		return result;
 	} else
 		return nullptr;
diff --git a/common/encoding.h b/common/encoding.h
index 6018eaf6e4..c8f864d2ec 100644
--- a/common/encoding.h
+++ b/common/encoding.h
@@ -35,6 +35,11 @@ typedef void* iconv_t;
 
 #include "common/scummsys.h"
 #include "common/str.h"
+#include "common/system.h"
+
+#ifdef WIN32
+#include "backends/platform/sdl/win32/win32.h"
+#endif
 
 namespace Common {
 
@@ -44,6 +49,9 @@ namespace Common {
  * ScummVM is compiled with or without iconv.
  */
 class Encoding {
+#ifdef WIN32
+	friend char *OSystem_Win32::convertEncoding(const char*, const char *, const char *, size_t);
+#endif
 	public:
 		/**
 		 * Constructs everything needed for the conversion between 2 encodings
@@ -234,6 +242,17 @@ class Encoding {
 		 * the same string.
 		 */
 		static String addUtfEndianness(const String &str);
+
+		/**
+		 * Switches the endianity of a string.
+		 *
+		 * @param string Array containing the characters of a string.
+		 * @param length Length of the string in bytes
+		 * @param bitCount Number of bits used for each character.
+		 *
+		 * @return Array of characters with the opposite endianity
+		 */
+		static char *switchEndian(const char *string, int length, int bitCount);
 };
 
 }