[From nobody Fri Aug  6 17:04:56 2004
Delivered-To: jhunter@collab.net
Return-Path: sweiss@aamc.org
Received: (qmail 910 invoked from network); 5 Jul 2000 16:03:38 -0000
Received: from wpo.aamc.org (143.220.4.36)
	by laswell.collab.net with SMTP; 5 Jul 2000 16:03:38 -0000
Received: from aamc.org (U024223.aamc.org [143.220.24.223])
	by wpo.aamc.org; Wed, 05 Jul 2000 12:01:48 -0500
Sender: steve
Message-ID: <39635C04.2AC314FF@aamc.org>
Date: Wed, 05 Jul 2000 12:02:12 -0400
From: Steve Weiss <sweiss@aamc.org>
Organization: Association of American Medical Colleges
X-Mailer: Mozilla 4.73 [en] (X11; U; Linux 2.2.14-12 i686)
X-Accept-Language: en, de, fr
MIME-Version: 1.0
To: Gabor Greif <gabor@no.netopia.com>
CC: jdom-interest@jdom.org, Jason Hunter <jhunter@collab.net>
Subject: Re: [jdom-interest] Entities with inner structure (proposed fix)
References: <B5879490-C72A6@192.109.102.124>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mozilla-Status2: 00000000

I'm not sure if using an abstract base class would be necessary
(although it might be a nice design approach), given Gabor's solution. I
did a quick hack on SAXHandler.java (I downloaded the latest from CVS
and startElement() has moved there from SAXBuilder.java) and Gabor's
code seemed to fix the problem. I still don't have things quite working,
but that may be a problem with the way I'm using entities. I'm an XML
newbie so I may not be doing things quite right.

Anyway, I'll keep fooling around with things- I'd be curious to know if
anyone else has been able to use entities though.

-Steve

Gabor Greif wrote:
> 
> I would underline Steve's diagnosis and add one more twist.
> The inEntity instance variable has  to be updated to mirror the class of
> the object on the top of the stack.
> 
> In SAXBuilder.startElement:
> 
>         if (atRoot) {
>             document.setRootElement(element);
>             stack.push(element);
>             atRoot = false;
>         } else {
>                   Object peeked = stack.peek();
>             if (inEntity)
>             {
>                 ((Entity)peeked).addChild(element);
>                 inEntity = false;
>             }
>             else
>                 ((Element)peeked).addChild(element);
> 
>             stack.push(element);
>         }
> 
> In SAXBuilder.endElement:
> 
>         stack.pop();
> 
>         inEntity = !stack.empty() && stack.peek() instanceof Entity;
> 
> Of course a common (abstract) baseclass to Element and Entity factoring out
> their intersection would be the best solution (as suggested by Steve).
> 
>         Gabor
> 
> On Wed, Jun 21, 2000 20:49 Uhr, Steve Weiss <mailto:sweiss@aamc.org> wrote:
> >I've had trouble trying to use Entities also, and it looks to me like
> >there is a problem in SAXBuilder.java.  In startElement(), at line 561
> >(beta4 release) there is this:
> >
> >    if (atRoot) {
> >        document.setRootElement(element);
> >        stack.push(element);
> >        atRoot = false;
> >    } else {
> >        ((Element)stack.peek()).addChild(element); <<---- throws
> >ClassCastException
> >        stack.push(element);
> >    }
> >
> >It looks like the ClassCastException is being thrown when an Entity
> >comes off the stack rather than an Element. Looking at these two
> >classes, it seems that both could extend some TBD AbstractXmlThing
> >class, which might solve this problem (i.e, cast to AbstractXmlThing
> >instead of Element). Of course, I could be way off on this.
> >
> >-Steve
> >
> >Gabor Greif wrote:
> >>
> >> I get an exception when parsing something like
> >>
> >> <tag>
> >>         &CancelButton;
> >> </tag>
> >>
> >> where CancelButton is an entity that is not string valued but consists
> of
> >> tag(s):
> >>
> >> <!ENTITY CancelButton "<pane type='Button'><ostype
> >> name='id'>&CancelButtonID;</ostype><string
> >> name="label">Cancel</string></pane>" >
> >>
> >> I am using JDOM b4.
> >>
> >> What is the preferred way to use this feature?
> >>
> >> I would propose that a subclass of Entity would be created in the "tag"
> >> Element, that would have a method for parsing the internal structure of
> the
> >> entity and giving back its constituents.
> >>
> >> Any ideas?
> >>
> >> Thanks,
> >>
> >>         Gabor

-- 
Steve Weiss      Association of American Medical Colleges
(202)828-0428    mailto:sweiss@aamc.org    http://www.aamc.org

]