Re: [Aramorph-users] XML tables

[Pierrick Brihaye]

Hi,

Ahmed El-dawy a écrit :

>     1) remember that implementing classes should be singletons. The
>     interface may so have a GetInstance() method. 
>   
> I don't think so. The getInstance() method will not be called from the 
> interface because it is static. It will be called from the implementing 
> class directly.

OK.

>     2) I would like an init() method to pass, for example, a JDBCConnection.
>  
> Do you mean something like constructor? 

No since we shouldn't have access to the constructor.

> It will be constructed in the 
> getInstance() method, so this will be useless. Also I don't know which 
> parameter it may receive.
> If it depends on the implementation, it cannot 
> be added to the interface.

It should be generic enough. See the design of :
http://excalibur.apache.org/apidocs/org/apache/avalon/excalibur/component/ExcaliburComponentManager.html

... which implements Configurable, initializable...

Welcome to COP (component oriented programming).

>     3) we must have a LoadFromFile() method and maybe even a LoadFromXML.
>  
> Is this different from init()?

IMHO, yes. init() would be used to initialize a connection to a 
database. LoadFromFile would actually send its content. Well... as you 
like but we need at least one method :-)

> I think this is implementation specific. 
> LoadFromXML, for example, will not be used with JDBCDictionaryHandler.

On the first time, it will !

>     4) rather then "translitered" arguments, consider Unicode ones. We may
>     even have methods for both. 
>
> Right! And I think we will have to change the InMemoryDictionaryHandler 
> to handle this.

Yes.

>     Parsing a comment : ; conjunctions
>     Parsing unvocalized : w
>     Parsing vocalized : wa
>     Parsing morphological category : Pref-Wa
>     Parsing gloss and grammatical category : and <pos>wa/CONJ+</pos>
>
>     etc, etc. It will be useful for the bloody GlossPOS ;-)
>
>  
> Where will this go?

In the code.

> By the way, I have set the transformer to write XML comments in place of 
> dictionary comments.

Good idea.
>     Oh, one more important thing I would like you to do. To remain ASCII
>     compliant, use the Java \uXXXXsyntax for arabic characters. Aramorph
>     must comple even on computers that don't support arabic.
>  
> OK! I will do it.

OK.


>     Could you please provide an example ? Since I've planned to write a
>     LemmaIDAnalyzer, this is bad news :-) 
>
> The code you have writes error message when finding one. Here's an example:
>
>         Found an element without lemma at line: 5811
>
>         Found an element without lemma at line: 5812
>         Found an element without lemma at line: 5813
>         Found an element without lemma at line: 5814 

You're right :

> ;--- Amd
> > md      >amad        N       term;period;extent;range;duration
> Amd     >amad        N       term;period;extent;range;duration
> |mAd    |mAd    N       terms;periods
> AmAd    |mAd    N       terms;periods

In such a case, I would create a lemmaID from the root.

> Another question. For which XML dictionary format do I start writing the 
> parser?

Still have to think about this. Stay tuned !

p.b.