gchar *
tracker_encoding_guess_meegotouch (const gchar *buffer,
gsize size)
{
/* Initialize detector */
MCharsetDetector detector ((const char *)buffer, (int)size);
gchar *locale;
gchar *encoding = NULL;
if (detector.hasError ()) {
g_warning ("Charset detector error when creating: %s",
detector.errorString ().toUtf8 (). data ());
return NULL;
}
locale = tracker_locale_get (TRACKER_LOCALE_LANGUAGE);
detector.setDeclaredLocale (locale);
MCharsetMatch bestMatch = detector.detect ();
if (detector.hasError ()) {
g_warning ("Charset detector error when detecting: %s",
detector.errorString ().toUtf8 (). data ());
g_free (locale);
return NULL;
}
if (bestMatch.confidence () > 30) {
encoding = g_strdup (bestMatch.name ().toUtf8 ().data ());
#if 0
QList<MCharsetMatch> mCharsetMatchList = detector.detectAll();
if (detector.hasError ()) {
g_warning ("Charset detector error when detecting all: %s",
detector.errorString ().toUtf8 (). data ());
}
g_debug ("Detecting all charsets...");
for (gint i = 0; i < mCharsetMatchList.size (); ++i) {
g_debug (" Charset '%s' with %d%% confidence...",
mCharsetMatchList[i].name (). toUtf8 ().data (),
mCharsetMatchList[i].confidence ());
}
#endif
g_debug ("Guessing charset as '%s' with %d%% confidence",
encoding, bestMatch.confidence ());
} else {
g_debug ("Ignoring charset as '%s' with %d%% (< 30%%) confidence",
bestMatch.name ().toUtf8 ().data (),
bestMatch.confidence ());
}
g_free (locale);
return encoding;
}
QString MCharsetDetector::text(const MCharsetMatch &charsetMatch)
{
Q_D(MCharsetDetector);
clearError();
QTextCodec *codec
= QTextCodec::codecForName(charsetMatch.name().toLatin1());
if (codec == NULL) { // there is no codec matching the name
d->_status = U_ILLEGAL_ARGUMENT_ERROR;
qWarning() << __PRETTY_FUNCTION__
<< "no codec for the name" << charsetMatch.name()
<< errorString();
// return empty string to indicate that no conversion is possible:
return QString();
}
else {
QTextCodec::ConverterState state;
QString text =
codec->toUnicode(d->_ba.constData(), d->_ba.size(), &state);
if (state.invalidChars > 0)
d->_status = U_INVALID_CHAR_FOUND;
return text;
}
}
