If you want to recognize language dependent expressions (such as complex ratio expressions -e.g. ``three out of four''- or currency expression -e.g. ``2,000 australian dollar''), you have to program a quantities_mylanguage class derived from abstract class quantities_module. Those classes are finite automata that recognize word sequences. An abstract class automat controls the sequence advance, so your derived class has little work to do apart from defining states and transitions for the automaton.
A good idea to start with this issue is having a look at the quantities_es and quantities_ca classes.
In the case your language is a roman language (or at least, has a similar structure for currency expressions) you can easily develop your currency expression detector by copying the quantities_es class, and modifying the CurrencyFile option to provide a file in which lexical items are adapted to your language. For instance: Catalan currency recognizer uses a copy of the quantities_es class, but a different CurrencyFile, since the syntactical structure for currency expression is the same in both languages, but lexical forms are different.
If your language has a very different structure for those expressions, you may require a different format for the CurrencyFile contents. Since that file will be used only for your language, feel free to readjust its format.
2008-01-24