Start of (still disabled) ID3v1 encoding detector.

Comes with const_cast hackiness \o/
Updates issue #254

src/CMakeLists.txt

@@ -390,6 +390,7 @@ target_link_libraries(clementine_lib
     qtsingleapplication
     qtiocompressor
     lastfm
+    chardet
     ${GOBJECT_LIBRARIES}
     ${GLIB_LIBRARIES}
     ${TAGLIB_LIBRARIES}

src/song.cpp

@@ -44,6 +44,7 @@
 #include <QFileInfo>
 #include <QTime>
 #include <QSqlQuery>
+#include <QTextCodec>
 #include <QVariant>
 
 #include <boost/scoped_ptr.hpp>
@@ -87,6 +88,46 @@ static TagLib::String QStringToTaglibString(const QString& s);
 
 TagLibFileRefFactory Song::kDefaultFactory;
 
+UniversalEncodingHandler::UniversalEncodingHandler()
+  : nsUniversalDetector(NS_FILTER_ALL),
+    current_codec_(NULL) {
+}
+
+TagLib::String UniversalEncodingHandler::parse(const TagLib::ByteVector& data) const {
+  const_cast<UniversalEncodingHandler*>(this)->Reset();
+  const_cast<UniversalEncodingHandler*>(this)->HandleData(data.data(), data.size());
+  const_cast<UniversalEncodingHandler*>(this)->DataEnd();
+
+  if (!current_codec_) {
+    return TagLib::String(data);  // Latin-1
+  } else {
+    // Detected codec -> QString (UTF-16) -> UTF8 -> UTF16-BE (TagLib::String)
+    // That's probably expensive.
+    QString unicode = current_codec_->toUnicode(data.data(), data.size());
+    qDebug() << "Decoded to:" << unicode;
+    return TagLib::String(unicode.toUtf8().constData(), TagLib::String::UTF8);
+  }
+}
+
+TagLib::ByteVector UniversalEncodingHandler::render(const TagLib::String& s) const {
+  // TODO: what should we do here?
+  // 1. Coerce to ASCII
+  // 2. Just write UTF8
+  // 3. Write what we read
+  // 4. Nothing and rewrite the tag as ID3v2 & UTF8
+  return TagLib::ByteVector();
+}
+
+void UniversalEncodingHandler::Report(const char* charset) {
+  qDebug() << "Detected as" << charset;
+  QTextCodec* codec = QTextCodec::codecForName(charset);
+  if (!codec) {
+    qWarning() << "Could not identify encoding in ID3v1 tag. Assuming ASCII.";
+  }
+  current_codec_ = codec;
+}
+
+
 Song::Private::Private()
   : valid_(false),
     id_(-1),

src/song.h

@@ -27,6 +27,9 @@
 
 #include "engines/engine_fwd.h"
 
+#include <taglib/id3v1tag.h>
+#include "nsUniversalDetector.h"
+
 namespace lastfm {
   class Track;
 }
@@ -50,6 +53,22 @@ class TagLibFileRefFactory : public FileRefFactory {
   virtual TagLib::FileRef* GetFileRef(const QString& filename);
 };
 
+class UniversalEncodingHandler : public TagLib::ID3v1::StringHandler,
+                                 nsUniversalDetector {
+ public:
+  UniversalEncodingHandler();
+
+  // TagLib::ID3v1::StringHandler
+  virtual TagLib::String parse(const TagLib::ByteVector& data) const;
+  virtual TagLib::ByteVector render(const TagLib::String& s) const;
+
+ private:
+  // nsUniversalDetector
+  virtual void Report(const char* charset);
+
+  QTextCodec* current_codec_;
+};
+
 class Song {
  public:
   Song();