RE: [xml-dev] question about xml:base value generated by xi:include

Hi Micheal,
Thank you for quick answer.
Just to better understand: I found the same issue on Saxon and BaseX.
BaseX developers told me that they are not responsible for this issue, since their application just wraps around SAX xml library (and they thus recommended me to raise the question to xml-dev list).
Therefore, my questions are:
1) is Saxon using the same xml parser of baseX?
2) is this issue going to be fixed in Saxon, or in an external xml parser library (SAX?)

Thank you,
Marco Randazzo
=================================
Marco Randazzo, PhD
iCub Facility
Istituto Italiano di Tecnologia (IIT)
Via Morego 30
16163 Genova, Italy
www.iit.it
=================================

From: Michael Kay <mike@saxonica.com>
Sent: Tuesday, July 31, 2018 4:37 PM
To: Marco Randazzo <Marco.Randazzo@iit.it>
Cc: xml-dev@lists.xml.org
Subject: Re: [xml-dev] question about xml:base value generated by xi:include

Note that this has been discussed on the Saxon forum at

https://saxonica.plan.io/issues/3864

and

https://saxonica.plan.io/issues/3530

and I previously raised questions on xml-dev in the thread at

https://markmail.org/thread/ka6uzz6ppali5amn

The current status in Saxon (I believe) is that I found a fix for the problem which I implemented on the development branch, but which I felt was too risky for the 9.8 (or earlier) branch. Part of the problem is that it's the kind of thing that should really be tested against multiple XML parsers, and we only get around to that quite rarely (in fact, it hasn't been done for a while, given how stable the XML parser scene is).

Michael Kay

Saxonica

On 31 Jul 2018, at 13:59, Marco Randazzo <Marco.Randazzo@iit.it> wrote:

Dear all,
I have a question about the value of attribute xml:base generated when processing an xi:include. Let me explain it with an example.

My example employs xi:include elements to include other xml files. It is constituted by a main.xml file which includes another file (include1.xml) stored in a subfolder, called include1.

include1.xml, in turn, includes another xml file (include2.xml) stored in a subfolder of include1 which is called include2.

Overall, the example consists of four folder levels and only the last file, include4.xml, contains data.

This is the contents of the files:

./main.xml

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

<main xmlns:xi="http://www.w3.org/2001/XInclude">

    <xi:include href="" />

</main >

/include1/include1.xml

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

<level1 xmlns:xi="http://www.w3.org/2001/XInclude">

               <xi:include href="" />

</level1 >

./include1/include2/include2.xml

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

<level2 xmlns:xi="http://www.w3.org/2001/XInclude">

               <xi:include href="" />

</level2 >

./include1/include2/include3/include3.xml

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

<level3 xmlns:xi="http://www.w3.org/2001/XInclude">

               <xi:include href="" />

</level3 >

./include1/include2/include3/include4/include4.xml

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

<level4 xmlns:xi="http://www.w3.org/2001/XInclude">

    <data> 0 </data>

    <data> 1 </data>

</level4 >

When I process main.xml, I obtain the following result:

<main xmlns:xi="http://www.w3.org/2001/XInclude">

<level1 xml:base="./include1/include1.xml">

    <level2 xml:base="./include1/include2/include2.xml">

      <level3 xml:base="./include2/include3/include3.xml">

        <level4 xml:base="./include3/include4/include4.xml">

          <data>0</data>

          <data>1</data>

        </level4>

      </level3>

    </level2>

</level1>

</main>

The xml tree is correct, but I have doubts about the value of the attribute xml:base.

In particular:

- level1: I think that the value of xml:base ( i.e. ./include1/include1.xml) is correct.

- level2: xml:base value is correct if its meaning is a path relative to the root file. Otherwise, if xml:base is supposed to store a path relative to the current file, then it should be ./include2/include2.xml.

- level3: This seems to be not correct, neither as an absolute path, nor as a relative path. I think it should be ./include1/include2/include3/include3.xml as a path relative to the root file or ./include3/include3.xml as a path relative to the current file.

- level4: Same as level3.

I also tried to execute a Xquery command http://www.xqueryfunctions.com/xq/fn_base-uri.html on the output.

The result returned by base-uri(//level3) is:

file:///C:/software/test_include/include1/include1/include2/include2/include3/include3.xml

which I think is wrong, I think it should be:

file:///C:/software/test_include/include1/include2/include3/include3.xml

I thus suspect that either the implementation of how xi:include generates the value of xml:base or the implementation of base-uri function has a problem.

My suspect is on xi:include. What do you think?
I currently experienced the same strange behavior on two different XML/XQuery processors, i.e.
- baseX http://basex.org (wrapping SAX xml library)
- Saxon EE 9.7 https://www.saxonica.com/download/java.xml