sax parser and nested tags

Thread view:

grasp06110 - 12 Sep 2006 20:23 GMT

Hi Everybody!

I'm doing some xml stuff and have gone down the road of writting a sax
parser to get data out of the file. I've over-written the startElement
and characters methods and I am getting the data I need... kind of. I
am not seeing an easy way to detect where I am in a nested structure
(e.g. a house has a room that has a cabinet that has a name v. a person
has a name). The stuff I've been reading talks about fully qualified
names. I would guess that that would give some indication of the
nesting of an element (e.g. house.room.cabenet.name v. person.name). I
am not seeing this behavior. In the house example and in the person
example I get "name" as the qualified name.

Any help would be appreciated.

Thanks,
John

Reply to this Message

grasp06110 - 12 Sep 2006 21:35 GMT

Hi everybody,

Roll-your-own seem to work prety well. Are there any potential pit
falls I'm missing with this solution?

This goes in startNode:

if (this.nodeName_ == null
|| this.nodeName_.trim().length() == 0) {
this.nodeName_ = name;
} else {
this.nodeName_ += "." + name;
}

This goes in endNode:

if (this.nodeName_ != null) {
int start = 0;
int end = nodeName_.lastIndexOf(".");
if (end == -1) {
nodeName_ = "";
} else {
nodeName_ = nodeName_.substring(start, end);
}
}

Reply to this Message

Tor Iver Wilhelmsen - 13 Sep 2006 15:54 GMT

> I am not seeing an easy way to detect where I am in a nested
> structure (e.g. a house has a room that has a cabinet that has a
> name v. a person has a name).

Push the elements onto a stack as you go. Peek at the stack to see the
context you're in. Pop when you reach the closing tag.

Reply to this Message

Lew - 17 Sep 2006 16:23 GMT

>> I am not seeing an easy way to detect where I am in a nested
>> structure (e.g. a house has a room that has a cabinet that has a
>> name v. a person has a name).
>
> Push the elements onto a stack as you go. Peek at the stack to see the
> context you're in. Pop when you reach the closing tag.

You might also explore the jaxb parser library.

Often it is enough just to refer the enclosing element (or its name).

class ElementRepresentation
{
    ElementRepresentation encloser;
    String tag;
    ...
}

Then if you want to distinguish, say, person.name from organization.name you
can put code in the "name" handler like:

ElementRepresentation name;
...
if ( name.getEncloser().getTag().equals( "person" ) )
{ ... }
else if ( name.getEncloser().getTag().equals( "organization" ) )
{ ... }
...

Crude, but effective. If necessary you can follow the encloser chain through
more than one layer -
if ( name.getEncloser().getEncloser().getTag().equals( "supervisor" )) ...

I've implemented a few projects using SAX parsing and this type of strategy,
where DOM parsing or XSL seemed like overkill. SAX parsing can process an XML
document in a single pass, and can be amazingly fast. The downside is that it
hard-codes schema structure into the logic, not always a Bad Thing but a
limitation nonetheless.

- Lew

Reply to this Message

sax parser and nested tags

Free Magazines