Hi Vairoj,
This structure does present a problem, because the element you want to
use to represent an EquipmentLineItem is determined by the actual type
of the Equipment reference. I don't see any clean way to handle this
directly in the binding for marshalling. If you were only binding for
unmarshalling I think you could do this (at least with the CVS code)
using a collection of the form:
<collection name="equipment" ordered="false" name="equipment"
ordered="false" ...
<structure name="phone" usage="optional">
<structure field="equipment">
<value name="phone-name" field="name"/>
</structure>
<value name="phone-quantity" field="quantity"/>
</structure>
<structure name="pda" usage="optional">
<structure field="equipment">
<value name="pda-name" field="name"/>
</structure>
<value name="pda-quantity" field="quantity"/>
</structure>
</collection>
since for unmarshalling only the element names of the child components
of a <collection> need to be unique. If you also need marshalling you
could probably add test-methods to the EquipmentLineItem to distinguish
between the different variations, then in theory use a binding like:
<collection name="equipment" ordered="false"
item-type="EquipmentLineItem" ...
<structure name="phone" test-method="isPhone" usage="optional">
<structure field="equipment">
<value name="phone-name" field="name"/>
</structure>
<value name="phone-quantity" field="quantity"/>
</structure>
<structure name="pda" test-method="isPda" usage="optional">
<structure field="equipment">
<value name="pda-name" field="name"/>
</structure>
<value name="pda-quantity" field="quantity"/>
</structure>
</collection>
I say "in theory" because I don't know that this code generation path
(structures distinguished by test-methods as items in a collection) has
ever been tested. <mapping> definitions (abstract or not) really
wouldn't gain you much here, since you're using distinct names even for
the shared values.
I don't really understand the logic behind the current representation,
so I'm not sure what to suggest in the way of changes to make it map
more cleanly. The easiest way to handle things would be to distinguish
at the line item level between the different types of devices, so that
you'd have PhoneLineItem and PdaLineItem classes. These could derive
from EquipmentLineItem, which would have quantity and name fields. Aside
from that, the best I can suggest is a custom marshaller/unmarshaller
for handling EquipmentLineItem. For the data you've shown this custom
mar/umar could just get the class name of the Equipment instance and use
in the XML representation as both the element name and the prefix on the
child element names.
The other alternative is to change the XML representation, assuming you
have control over that. I'd suggest something along the lines of:
<equipment>
<item>
<phone>Siemens S45</phone>
<quantity>5</quantity>
</item>
this could be done using <mapping> definitions for the different
subtypes of equipment.
- Dennis
Vairoj A. wrote:
Hi,
This question combines structure mapping with abstract mapping, so it
makes me confuse. However, I still not sure on how to separate the
issue so maybe you guys could help me here.
This is a sample XML data:
<person>
<name>Jack</name>
<equipment>
<phone>
<phone-name>Siemens S45</name>
<phone-quantity>5</quantity>
</phone>
<pda>
<pda-name>O2 XDA mini</name>
<pda-quantity>2</quantity>
</pda>
</equipment>
</person>
A person has a set of equipment line items, equipments can be
categorized into 2 categories - phone and pda. Each line item is
composed of name of the equipment and quantity. Please note that the
tag that describe each line item is different by the type of equipment
it contains.
The following are the classes which shows only parts that are relevant
to the mapping:
class Person {
private Set<EquipmentLineItem> lineItems;
}
class EquipmentLineItem {
private Equipment equipment;
private int quantity;
}
abstract class Equipment {
private String name;
}
class Phone extends Equipment {
}
class Pda extends Equipment {
}
I have also attached an image file which is the class diagram for the
preceding classes.
The question is how to map this class structure to the above XML
format. There is several issue (for me).
- the structure is not symmetric. While in java class it is a
collection of equipments, in XML the tag is up to the specific
equipment type.
- In java class quantity attribute resides in EquipmentLineItem,
whereas in XML they are all in the same level.
- sub element in each equipment has different element name depends on
equipment type
Is it possible to do this type of mapping? Please suggest me. If it is
not possible, I could adjust the class structure to which can be map
better.
Regards,
Vairoj
------------------------------------------------------------------------
-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. Download
it for free - -and be entered to win a 42" plasma tv or your very own
Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php
_______________________________________________
jibx-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/jibx-users
