QHash non-ANSI key problem

[roxton]

Hello! I've found a strange behavior of a hash lookup with a non-ANSI key. When I use Latin keys, hash works fine - all values are returned by the QHash.value function. But when I'm trying to use a Russian keys (in the QString unicode, of course), sometimes hash does not return a value. And vice versa - sometime I can't get a key but the value - even if key or value if a one-character strings.
Also I have to notice that all keys and values are presented in the hash table, as I can check it with keys() and values() functions.
If QHash table have such problems, what you can suggest me to use for key/value handling with a full Unicode support in keys and values?

[ChristianEhrlicher]

I don't think it's a qhash - problem but something on your side. Plz post some code to show us the problem.

[roxton]

OK, here is all code that related to my problem. It's a part of my Morse code converter.

//qstring_load - just a utility function for text file loading into QString
QString qstring_load (const QString &fileName)
{
  QFile file (fileName);
 
  QString s;
 
  if (! file.open(QFile::ReadOnly | QFile::Text))
       return s;
 
  QTextStream in(&file);
  in.setCodec ("UTF-8");
 
  s = in.readAll();
  return s;
}
 
//hash_load_keyval - loads a key/value pairs table into the hash:
QHash<QString, QString> hash_load_keyval (const QString &fname)
{
  QHash<QString, QString> result;
  if (! file_exists (fname))
     return result;
 
  QStringList l = qstring_load (fname).split ("\n");
 
  int c = l.size();
  for (int i = 0; i < c; i++)
      {
       QStringList sl = l[i].split("=");
       if (sl.size() == 2)
          result.insert (sl[0], sl[1]);
      }
 
  return result;
}
 
//morse_from_lang  - converts QString s into the Morse code using the key/value file that embeded into the resource. The QString lang is a locale id, for example "en" or "ru".
Please note that all data from the file loads OK.
 
QString morse_from_lang (const QString &s, const QString &lang)
{
 
  QHash<QString, QString> h = hash_load_keyval (":/text-data/morse-" + lang);
 
  QString result;
 
  int c = s.size();
  for (int i = 0; i < c; i++)
      {
       QString t = h.value (QString (s[i])); //here can be an empty value if key is non-Latin, and I don't know why
       if (! t.isNull() || ! t.isEmpty())
          result.append (t + " ");
      }
 
  return result;
}

//qstring_load - just a utility function for text file loading into QString
QString qstring_load (const QString &fileName)
{
  QFile file (fileName);

  QString s;

  if (! file.open(QFile::ReadOnly | QFile::Text))
       return s;

  QTextStream in(&file);
  in.setCodec ("UTF-8");

  s = in.readAll();
  return s;
}

//hash_load_keyval - loads a key/value pairs table into the hash:
QHash<QString, QString> hash_load_keyval (const QString &fname)
{
  QHash<QString, QString> result;
  if (! file_exists (fname))
     return result;

  QStringList l = qstring_load (fname).split ("\n");

  int c = l.size();
  for (int i = 0; i < c; i++)
      {
       QStringList sl = l[i].split("=");
       if (sl.size() == 2)
          result.insert (sl[0], sl[1]);
      }

  return result;
}

//morse_from_lang  - converts QString s into the Morse code using the key/value file that embeded into the resource. The QString lang is a locale id, for example "en" or "ru".
Please note that all data from the file loads OK.

QString morse_from_lang (const QString &s, const QString &lang)
{

  QHash<QString, QString> h = hash_load_keyval (":/text-data/morse-" + lang);

  QString result;

  int c = s.size();
  for (int i = 0; i < c; i++)
      {
       QString t = h.value (QString (s[i])); //here can be an empty value if key is non-Latin, and I don't know why
       if (! t.isNull() || ! t.isEmpty())
          result.append (t + " ");
      }

  return result;
}

To copy to clipboard, switch view to plain text mode 

[ChristianEhrlicher]

First: Please use the '[[code]]' - tags to make your code more readable

I would suggest adding some debug output to see what's going wrong. Maybe the utf-8 codec is not correct or something like this.
Add a debug output to see which keys you really put into the hash. Also use QHash<QChar,QString> because you only test for a char, not a string.
And then put a debug output into your other loop to see what s[i] actually is.

The debug-output should output the unicode numbers:

QChar c = sl[0][0];
qDebug() << "Adding " << c << " (= " << c.unicode() << ") to the hash";

QChar c = sl[0][0];
qDebug() << "Adding " << c << " (= " << c.unicode() << ") to the hash";

To copy to clipboard, switch view to plain text mode 

[roxton]

Thanks, all works fine now. It was my fault - I've copied a Morse table data from Wikipedia, and there are was some weird stuff with Russian letters (Latin "A" instead on a Russian "A", etc).