How to use xpointer with Xinclude to reference elements

Question

I want to merge 2 XML files with the same structure to make one. For example;

Test1.xml

<?xml version="1.0" encoding="UTF-8"?>

<ns:Root
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xmlns:ns="urn:TestNamespace"
    xsi:schemaLocation="urn:Test.Namespace Test1.xsd"
    >
    <ns:element1 id="001">
       <ns:element2 id="001.1" order="1">
           <ns:element3 id="001.1.1" />
       </ns:element2>
       <ns:element2 id="001.2" order="2">
           <ns:element3 id="001.1.2" />
       </ns:element2>
    </ns:element1>
</ns:Root>

and Test2.xml

<?xml version="1.0" encoding="UTF-8"?>

<ns:Root
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xmlns:ns="urn:TestNamespace"
    xsi:schemaLocation="urn:Test.Namespace Test1.xsd"
    >
    <ns:element1 id="999">
        <ns:element2 id="999.1" order="1">
            <ns:element3 id="999.1.1" />
        </ns:element2>
    </ns:element1>
</ns:Root>

To create

TestOutput.xml

<?xml version="1.0" encoding="UTF-8"?>

<ns:Root
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xmlns:ns="urn:TestNamespace"
    xsi:schemaLocation="urn:Test.Namespace Test1.xsd"
    >
    <ns:element1 id="001">
       <ns:element2 id="001.1" order="1">
           <ns:element3 id="001.1.1" />
       </ns:element2>
       <ns:element2 id="001.2" order="2">
           <ns:element3 id="001.1.2" />
       </ns:element2>
    </ns:element1>
    <ns:element1 id="999">
        <ns:element2 id="999.1" order="1">
            <ns:element3 id="999.1.1" />
        </ns:element2>
    </ns:element1>
</ns:Root>

ie one XML file with all the elements from each included.

I found a useful question on StackOverflow, and came up with this;

Merge.xml

<?xml version="1.0"?>

<ns:Root xmlns:xi="http://www.w3.org/2003/XInclude"
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xmlns:ns="urn:TestNamespace">

    <xi:include href="Test1.xml" parse="xml" xpointer="element(//ns:Root/ns:element1)" />  

    <xi:include href="Test2.xml" parse="xml" xpointer="element(//ns:Root/ns:element1)" />

</ns:Root>

Which I run by doing this (I need to use xmllint for reasons to involved to go into)

xmllint -xinclude Merge.xml

But this does not work, it complains about various thiongs, which seem to relate to xpointer.

parser error : warning: ChildSeq not starting by /1
Merge.xml:7: element include: XInclude error : XPointer evaluation failed: #element(//ns:Root/ns:element1)
Merge.xml:7: element include: XInclude error : could not load Test1.xml, and no fallback was found
parser error : warning: ChildSeq not starting by /1
Merge.xml:9: element include: XInclude error : XPointer evaluation failed: #element(//ns:Root/ns:element1)
Merge.xml:9: element include: XInclude error : could not load Test2.xml, and no fallback was found
<?xml version="1.0"?>
<ns:Root xmlns:xi="http://www.w3.org/2003/XInclude" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:ns="urn:TestNamespace">

    <xi:include href="Test1.xml" parse="xml" xpointer="element(//ns:Root/ns:element1)"/>

    <xi:include href="Test2.xml" parse="xml" xpointer="element(//ns:Root/ns:element1)"/>

</ns:Root>

If I omit the xpointer attributes in Merge.xml then I get some sensible output, but it has done more than include the elements I want of course.

Can someone offer some advice as to what I am doing wrong with xpointer please?

Thanks in antcipation.

If I remove the namespaces, the above works, so this just looks to be an issue with XPointer and how I am dealing with the namespaces — Nerdio, May 15 '13 at 13:35
The `element()` scheme does not support qualified names (see https://www.w3.org/TR/xptr-element/). A name specified with `element()` must be a NCName and refers to a _single_ element identified with an xs:ID of that name. That's obviously not what you want. — Adrian W, Nov 12 '16 at 18:49

score 5 · Answer 1 · answered May 16 '13 at 07:47

I have dabbled with this a bit more, and found plenty of examples on the web that suggest what I am doing is correct.This is now a working version...

<?xml version="1.0"?>

<Root xmlns:xi="http://www.w3.org/2003/XInclude"
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xmlns:ns="http://testurl.com/now">

    <xi:include href="Test1.xml" xpointer="xmlns(ns=http://testurl.com/now)xpointer(/ns:Root/ns:element1)" parse="xml" />
    <xi:include href="Test2.xml" xpointer="xpointer(//Root/element1)" parse="xml" />

</Root>

This example uses a version of Test1.xml which has namespaces, and Test2.xml which does not.

The output now looks like this....

<?xml version="1.0"?>
<Root xmlns:xi="http://www.w3.org/2003/XInclude" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:ns="http://testurl.com/now">

    <ns:element1 xmlns:ns="http://testurl.com/now" id="001">
        <ns:element2 id="001.1" order="1">
            <ns:element3 id="001.1.1"/>
        </ns:element2>
        <ns:element2 id="001.2" order="2">
            <ns:element3 id="001.1.2"/>
        </ns:element2>
    </ns:element1><ns:element1 xmlns:ns="http://testurl.com/now" id="003">
        <ns:element2 id="007.0" order="1">
            <ns:element3 id="007.1.1"/>
        </ns:element2>
    </ns:element1><ns:element1 xmlns:ns="http://testurl.com/now" id="002">
        <ns:element2 id="002.1" order="3">
            <ns:element3 id="002.1.1"/>
        </ns:element2>
        <ns:element2 id="002.2" order="4">
            <ns:element3 id="002.1.2"/>
        </ns:element2>
    </ns:element1>
    <element1 id="999">
        <element2 id="999.1" order="1">
            <element3 id="999.1.1"/>
        </element2>
    </element1>

</Root>

This is of course acceptable, it would be nice if the line breaks between the open and close of element1 were still there

The line breaks are not part of the elements, so that can't be cought with referencing the elements. Try adding `--pretty 1` to the xmllint command line: `xmllint --pretty 1 -xinclude Merge.xml`. That still will not reproduce the original spacing, but looks a bit nicer. — Adrian W, Nov 12 '16 at 19:03

score 2 · Answer 2 · answered Nov 12 '16 at 19:08

This works with and without namespaces:

<?xml version="1.0"?>
<ns:Root xmlns:xi="http://www.w3.org/2003/XInclude"
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xmlns:ns="urn:TestNamespace">

    <xi:include href="Test1.xml" xpointer="xpointer(*/*)" />  
    <xi:include href="Test2.xml" xpointer="xpointer(*/*)" />

</ns:Root>

Also parse="xml" is default. You don't need to specify it.

score 0 · Answer 3 · answered Sep 16 '21 at 21:06

For those using Xerces in Java: it only supports xpointer="element(...)" pointers. This is defined at https://www.w3.org/TR/2003/REC-xptr-element-20030325/

It has an example:

For example, the following pointer part identifies the element with an ID (as defined in XPointer Framework) of "intro"

but I failed to understand XPointer Framework determined ID is from https://www.w3.org/TR/2003/REC-xptr-framework-20030325/#shorthand

Scanning through https://www.ibiblio.org/xml/books/bible3/chapters/ch18.html
and reading https://xerces.apache.org/xerces2-j/faq-xinclude.html
I realized that this is possible to achieve what you asked for:

<?xml version="1.0"?>
<ns:Root xmlns:xi="http://www.w3.org/2003/XInclude"
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xmlns:ns="urn:TestNamespace">

    <xi:include href="Test1.xml" xpointer="element(/1)" />  
    <xi:include href="Test2.xml" xpointer="element(/1)" />

</ns:Root>

The good thing about this is that I guess element() schema is supported in more places than the full xpointer() schema.

Note: this addressing scheme can be nested, so for example /1/2 means root's (/) 1st element that has a child (/) at position 2. So it would select 001.2 from Test1.xml.

How to use xpointer with Xinclude to reference elements

3 Answers3

Linked