From 6b705d56525372cfaaca0b10d8208020660f435e Mon Sep 17 00:00:00 2001
From: Martchus <martchus@gmx.net>
Date: Wed, 31 May 2017 00:53:42 +0200
Subject: [PATCH] Mind encoding when converting TagValue to pos

---
 id3/id3v2frame.cpp | 22 +++-------------------
 tagvalue.cpp       | 26 ++++++++++++++++++++++++--
 tagvalue.h         |  6 +++++-
 3 files changed, 32 insertions(+), 22 deletions(-)

diff --git a/id3/id3v2frame.cpp b/id3/id3v2frame.cpp
index 199e2c1..2b0af7f 100644
--- a/id3/id3v2frame.cpp
+++ b/id3/id3v2frame.cpp
@@ -65,13 +65,10 @@ Id3v2Frame::Id3v2Frame(const identifierType &id, const TagValue &value, const by
  * \returns Returns the genre index or -1 if the specified string does not denote a genre index.
  */
 template<class stringtype>
-int parseGenreIndex(const stringtype &denotation, bool isBigEndian = false)
+int parseGenreIndex(const stringtype &denotation)
 {
     int index = -1;
     for(auto c : denotation) {
-        if(sizeof(typename stringtype::value_type) == 2 && isBigEndian != CONVERSION_UTILITIES_IS_BYTE_ORDER_BIG_ENDIAN) {
-            c = swapOrder(static_cast<uint16>(c));
-        }
         if(index == -1) {
             switch(c) {
             case ' ':
@@ -272,7 +269,7 @@ void Id3v2Frame::parse(BinaryReader &reader, const uint32 version, const uint32
             int genreIndex;
             if(characterSize(dataEncoding) > 1) {
                 auto genreDenotation = parseWideString(buffer.get() + 1, m_dataSize - 1, dataEncoding);
-                genreIndex = parseGenreIndex(genreDenotation, dataEncoding == TagTextEncoding::Utf16BigEndian);
+                genreIndex = parseGenreIndex(genreDenotation);
             } else {
                 auto genreDenotation = parseString(buffer.get() + 1, m_dataSize - 1, dataEncoding);
                 genreIndex = parseGenreIndex(genreDenotation);
@@ -675,20 +672,7 @@ u16string Id3v2Frame::parseWideString(const char *buffer, size_t dataSize, TagTe
 {
     auto substr = parseSubstring(buffer, dataSize, encoding, addWarnings);
     u16string res(reinterpret_cast<u16string::const_pointer>(get<0>(substr)), get<1>(substr) / 2);
-    if(encoding !=
-        #if defined(CONVERSION_UTILITIES_BYTE_ORDER_LITTLE_ENDIAN)
-            TagTextEncoding::Utf16LittleEndian
-        #elif defined(CONVERSION_UTILITIES_BYTE_ORDER_BIG_ENDIAN)
-            TagTextEncoding::Utf16BigEndian
-        #else
-        # error "Host byte order not supported"
-        #endif
-            ) {
-        // ensure byte order matches host byte order
-        for(auto &c : res) {
-            c = ((c >> 8) & 0x00FF) | ((c << 8) & 0xFF00);
-        }
-    }
+    TagValue::ensureHostByteOrder(res, encoding);
     return res;
 }
 
diff --git a/tagvalue.cpp b/tagvalue.cpp
index d09f5fa..fcbf5db 100644
--- a/tagvalue.cpp
+++ b/tagvalue.cpp
@@ -249,8 +249,9 @@ PositionInSet TagValue::toPositionInSet() const
                 return PositionInSet(string(m_ptr.get(), m_size));
             case TagTextEncoding::Utf16LittleEndian:
             case TagTextEncoding::Utf16BigEndian:
-                // FIXME: Ensure endianness is correctly
-                return PositionInSet(u16string(reinterpret_cast<char16_t *>(m_ptr.get()), m_size / 2));
+                u16string u16str(reinterpret_cast<char16_t *>(m_ptr.get()), m_size / 2);
+                ensureHostByteOrder(u16str, m_encoding);
+                return PositionInSet(u16str);
             }
         case TagDataType::Integer:
         case TagDataType::PositionInSet:
@@ -705,6 +706,27 @@ void TagValue::stripBom(const char *&text, size_t &length, TagTextEncoding encod
     }
 }
 
+/*!
+ * \brief Ensures the byte-order of the specified UTF-16 string matches the byte-order of the machine.
+ * \remarks Does nothing if \a currentEncoding already matches the byte-order of the machine.
+ */
+void TagValue::ensureHostByteOrder(u16string &u16str, TagTextEncoding currentEncoding)
+{
+    if(currentEncoding !=
+        #if defined(CONVERSION_UTILITIES_BYTE_ORDER_LITTLE_ENDIAN)
+            TagTextEncoding::Utf16LittleEndian
+        #elif defined(CONVERSION_UTILITIES_BYTE_ORDER_BIG_ENDIAN)
+            TagTextEncoding::Utf16BigEndian
+        #else
+        # error "Host byte order not supported"
+        #endif
+            ) {
+        for(auto &c : u16str) {
+            c = swapOrder(static_cast<uint16>(c));
+        }
+    }
+}
+
 /*!
  * \brief Returns an empty TagValue.
  */
diff --git a/tagvalue.h b/tagvalue.h
index 12d8b58..7e2c539 100644
--- a/tagvalue.h
+++ b/tagvalue.h
@@ -14,6 +14,7 @@
 namespace Media {
 
 class Tag;
+class Id3v2Frame;
 
 /*!
  * \brief Specifies the text encoding.
@@ -62,6 +63,8 @@ enum class TagDataType : unsigned int
 
 class TAG_PARSER_EXPORT TagValue
 {
+    friend class Id3v2Frame; // FIXME: make ensureHostByteOrder() public in next minor release
+
 public:
     // constructor, destructor
     TagValue();
@@ -124,7 +127,8 @@ public:
 
 
 private:
-    void stripBom(const char *&text, size_t &length, TagTextEncoding encoding);
+    static void stripBom(const char *&text, size_t &length, TagTextEncoding encoding);
+    static void ensureHostByteOrder(std::u16string &u16str, TagTextEncoding currentEncoding);
 
     std::unique_ptr<char[]> m_ptr;
     std::string::size_type m_size;