[JXPath] Why array range check exists in isCollectionElement() ?

[Adrian Perez Jorge <ap...@ignos.com>]

Hi,

  In 
org.apache.commons.jxpath.ri.axes.SimplePathInterpreter.isCollectionElement(), 
why is there a range check?  Why is it necessary?

  My problem is this.  Suppose I have developed a Bean wich has an 
indexed property, and the real storage is, let say, an ArrayList:

  public class MyBean {

    ArrayList storage = new ArrayList();

    public String[] getStorage() {
       return storage.toArray(new String[] {} );
    }

    public String getStorage(int i) {
       this.storage.get(i);
    }

    public void setStorage(String[] storage) {

        this.storage.clear();
        for (int i = 0; i < storage.length; ++i) {
           this.storage.add(storage[i]);
        }
    }

    public void setStorage(int i, String str) {

        this.storage.set(i, str);
    }
  }

 When I call JXPathContext.getValue("storage[3]") it will do range 
check, so the ArrayList must be converted to an array each time I access 
to any of the elements.

 Is it possible to avoid range check and just catch and manage the 
ArrayIndexOutOfBoundsException?

 BTW, why range check is done for (xpath) indexes > 1?  If I do 
JXPathContext.getValue("storage[1]") it does not do range check, but for 
JXPathContext.getValue("storage[2]") it does not.  Why 1 is different?  
Think about an empty array.

  Thanks in advance,

Adrian P.J.


---------------------------------------------------------------------
To unsubscribe, e-mail: commons-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: commons-user-help@jakarta.apache.org

Re: [JXPath] Why array range check exists in isCollectionElement() ?

[Adrian Perez Jorge <ap...@ignos.com>]

> BTW, why range check is done for (xpath) indexes > 1?  If I do 
> JXPathContext.getValue("storage[1]") it does not do range check, but 
> for JXPathContext.getValue("storage[2]") it does not.  Why 1 is 
> different?  Think about an empty array.
>
BTW, why range check is done for (xpath) indexes > 1?  If I do 
JXPathContext.getValue("storage[1]") it does not do range check, but for 
JXPathContext.getValue("storage[2]") it _does_. <snip/>

Excuse me for the typo.


---------------------------------------------------------------------
To unsubscribe, e-mail: commons-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: commons-user-help@jakarta.apache.org

Re: [JXPath] Why array range check exists in isCollectionElement() ?

[Dmitri Plotnikov <dm...@apache.org>]

Adrian,
 
I wouldn't really want to rely on the ArrayIndexOutOfBoundsException.  There is no guarantee that the bean will properly throw this exception and keep the state of the bean intact.  There is no way of knowing that passing an invalid index won't leave some kind of horrible side-effect.
 
However, you don't need to convert the collection to an array. JXPath can handle all kinds of Collection objects.
 
As to your second question, the explanation has to do with the fact that XPath is designed to work with XML, which is missing an explicit notion of a collection.
 
Let's start with an XML file that looks like this:
 
<a>
  <b>first</b>
  <b>second</b>
</a>
 
The xpath /a/b[1] will return "first" and /a/b[2] will return "second", just as expected.
 
Now, let's say the file contains only one <b> element:
 
<a>
  <b>bee</b>
</a>
 
The xpath /a/b[1] will return "bee", but so will "/a/b".
 
As you can see, in XML a collection containing exactly one element can be treated either as a collection or a scalar.
 
In order to conform to the XPath standard at least in some sense, JXPath makes the same to be true for java objects: a collection of exactly one element can be treated either as a collection or as a scalar.
 
I hope this helps.
 
- Dmitri

Adrian Perez Jorge <ap...@ignos.com> wrote:
Hi,

In 
org.apache.commons.jxpath.ri.axes.SimplePathInterpreter.isCollectionElement(), 
why is there a range check? Why is it necessary?

My problem is this. Suppose I have developed a Bean wich has an 
indexed property, and the real storage is, let say, an ArrayList:

public class MyBean {

ArrayList storage = new ArrayList();

public String[] getStorage() {
return storage.toArray(new String[] {} );
}

public String getStorage(int i) {
this.storage.get(i);
}

public void setStorage(String[] storage) {

this.storage.clear();
for (int i = 0; i < storage.length; ++i) {
this.storage.add(storage[i]);
}
}

public void setStorage(int i, String str) {

this.storage.set(i, str);
}
}

When I call JXPathContext.getValue("storage[3]") it will do range 
check, so the ArrayList must be converted to an array each time I access 
to any of the elements.

Is it possible to avoid range check and just catch and manage the 
ArrayIndexOutOfBoundsException?

BTW, why range check is done for (xpath) indexes > 1? If I do 
JXPathContext.getValue("storage[1]") it does not do range check, but for 
JXPathContext.getValue("storage[2]") it does not. Why 1 is different? 
Think about an empty array.

Thanks in advance,

Adrian P.J.


---------------------------------------------------------------------
To unsubscribe, e-mail: commons-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: commons-user-help@jakarta.apache.org