Difficulties converting XML to XML using XSLT

Reznik · May 31st, 2008, 11:13 AM

Hello,

i have this XML structure

Code:

<group>
        <name>person</name>
        <group>
            <name>caucasian</name>
            <group>
                <field>
                    <name>age</name>
                    <value>3</value>
                    <children>0</children>
                </field>
                <field>
                    <name>age</name>
                    <value>24</value>
                    <children>0</children>
                </field>
            </group>
       </group>
       <group>
            <name>asian</name>
            <group>
                <field>
                    <name>age</name>
                    <age>36</age>
                    <children>2</children>
                </field>
(...)

and i want to transform it to the following format using xsl transformations

Code:

<person>
        <caucasian>
            <age>
                 <value>3</value>
                 <children>0</children>
            </age>
            <age>
                 <value>24</value>
                 <children>0</children>
            </age>
        </caucasian>
       <asian>
            <age>
                 <value>36</value>
                 <children>2</children>
            </age>
    </asian>
(...)

I already can split to the values i want using this code

Code:

<xsl:stylesheet version="1.0"
xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
    <xsl:output method="xml"/>
    <xsl:template match="/">
        <xsl:apply-templates/> 
    </xsl:template>

    <xsl:template match="main">
        <xsl:apply-templates select="field"/>
    </xsl:template>

    <xsl:template match="field">
        <xsl:if test="child::*[1][self::name]">
                <xsl:value-of select="."/>
        </xsl:if>

        <xsl:if test="child::*[1][self::size]">
            <xsl:copy-of select="child::*[1]"/>
        </xsl:if>

        <xsl:if test="child::*[1][self::allow]">
            <xsl:copy-of select="child::*[1]"/>
        </xsl:if>
    </xsl:template>
</xsl:stylesheet>

however i can't make it close the <name> tag from the original in the right place.
Right now i can only get to this

Code:

(...)
<caucasian></caucasian>
            <age></age>
                 <value>3</value>
                 <children>0</children>

            <age></age>
                 <value>24</value>
                 <children>0</children>
(...)

Any help? I'm killing myself with this

Martin Honnen · May 31st, 2008, 11:42 AM

Your XML sample is inconsistent, the first two 'field' elements have child elements 'value' and 'children' while the third has 'age' and 'children' child elements.
Assuming the third 'field' element also has 'value' instead of 'age' then it is easy:

Code:

<xsl:stylesheet
  xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
  version="1.0">

  <xsl:output method="xml" indent="yes"/>

  <xsl:strip-space elements="*"/>

  <xsl:template match="/group">
    <person>
      <xsl:apply-templates select="group"/>
    </person>
  </xsl:template>

  <xsl:template match="/group/group">
    <xsl:element name="{name}">
      <xsl:apply-templates select="group/field"/>
    </xsl:element>
  </xsl:template>

  <xsl:template match="group/field">
    <age>
      <xsl:copy-of select="value | children"/>
    </age>
  </xsl:template>

</xsl:stylesheet>

--
Martin Honnen
Microsoft MVP - XML

Reznik · May 31st, 2008, 01:34 PM

Yes, it it value and not age.

I'm going to try what you wrote. Thank you!

mhkay · June 1st, 2008, 03:12 AM

This logic looks very clumsy

<xsl:template match="field">
        <xsl:if test="child::*[1][self::name]">
                <xsl:value-of select="."/>
        </xsl:if>

Can't be definitive because you haven't said what all the input possibilities are (e.g. you mention element <allow>), but to handle your sample data (as corrected) I might write

<xsl:template match="field">
  <xsl:element name="{name}">
    <xsl:copy-of select="*[not(self::name)]"/>
  </xsl:element>
</xsl:template>

>however i can't make it close the <name> tag from the original in the right place.

Do try to think of XSLT as producing a result tree of nodes, not producing lexical XML with tags. Changing your thought process here is the first step to understanding what's going on. In the above template, the xsl:element instruction generates an element node in the result tree, and the xsl:copy-of instruction generates its children in the result tree. If you get the tree of nodes right, the tags will look after themselves.

Michael Kay
http://www.saxonica.com/
Author, XSLT Programmer's Reference and XPath 2.0 Programmer's Reference

Reznik · June 2nd, 2008, 01:50 PM

First of all thank you for your help.

I'm close to what i want to obtain as a resulting XML, however there are a few bugs that i can't get rid of...

This is a piece of my XML code (with some edits):

Code:

<?xml version="1.0" encoding="ISO-8859-1"?>
<packet>
    <group>
        <name>ph</name>
        <offset>0</offset>
        <group>
            <name>pi</name>
            <offset>0</offset>
            <group>
                <field>
                    <name>vn</name>
                    <size>3</size>
                    <allow>0</allow>
                </field>
                <field>
                    <name>t</name>
                    <size>1</size>
                    <reserved>0</reserved>
                </field>
                <field>
                    <name>data</name>
                    <size>1</size>
                </field>
                <group>
                    <name>aid</name>
                    <offset>0</offset>
                    <group>
                        <field>
                            <name>pd</name>
                            <size>7</size>
                        </field>
                        <field>
                            <name>pt</name>
                            <size>4</size>
                        </field>
                    </group>
                </group>
            </group>
        </group>
(...)

this is my xsl

Code:

 <xsl:template match="/">
    <xsl:text disable-output-escaping="yes"><packet></xsl:text>
    <xsl:apply-templates/>
    <xsl:text disable-output-escaping="yes"></packet></xsl:text>
  </xsl:template>


  <xsl:template match="//group">
    <xsl:if test="child::node()[name()='name']">
        <xsl:text disable-output-escaping="yes"><</xsl:text>
        <xsl:value-of select="name"/>
        <xsl:text disable-output-escaping="yes">></xsl:text>
    </xsl:if>
    <xsl:apply-templates/>   
    <xsl:if test="child::node()[name()='name']">
        <xsl:text disable-output-escaping="yes"></</xsl:text>
        <xsl:value-of select="name"/>
        <xsl:text disable-output-escaping="yes">></xsl:text>
    </xsl:if>
  </xsl:template>

  <xsl:template match="group">

  </xsl:template>

  <xsl:template match="offset">
    <xsl:copy-of select="."/>
  </xsl:template>

  <xsl:template match="field">
    <xsl:element name="{name}">
      <xsl:copy-of select="size | allow | reserved"/>
    </xsl:element>
  </xsl:template>

i'm getting this result with some text in the middle... that shouldn't be there... also i'm open to suggestions for replacement of those if clauses in the beginning :)

Code:

<packet>

    <ph>
ph
<offset>0</offset>

    <pi>
pi
<offset>0</offset>

    <vn>
<size>3</size>
<allow>0</allow>
</vn>

    <t>
<size>1</size>
<reserved>0</reserved>
</t>

    <data>
<size>1</size>
</data>

    <aid>
aid
<offset>0</offset>

    <pd>
<size>7</size>
</pd>

    <pt>
<size>4</size>
</pt>
</aid>
</pi>

basically the text repeats after a "/group/name" from the original XML

Thank you

mhkay · June 2nd, 2008, 02:16 PM

I can't actually see how you're getting anything: your stylesheet isn't well-formed XML. Perhaps you really wrote <packet> rather than <packet>?

Please don't use disable-output-escaping until you're expert enough to know that there is no other solution. It doesn't work in all environments, and that's because it creates a nasty hidden dependency on the interface between the transformer and the serializer, which should be kept independent of each other.

I said this before: "Do try to think of XSLT as producing a result tree of nodes, not producing lexical XML with tags. " and you clearly haven't taken it on board.

In this example there's no excuse:

<xsl:template match="/">
    <xsl:text disable-output-escaping="yes"><packet></xsl:text>
    <xsl:apply-templates/>
    <xsl:text disable-output-escaping="yes"></packet></xsl:text>
  </xsl:template>

that should be

<xsl:template match="/">
    <packet>
    <xsl:apply-templates/>
    </packet>
  </xsl:template>

and this

<xsl:template match="//group">
    <xsl:if test="child::node()[name()='name']">
        <xsl:text disable-output-escaping="yes"><</xsl:text>
        <xsl:value-of select="name"/>
        <xsl:text disable-output-escaping="yes">></xsl:text>
    </xsl:if>
    <xsl:apply-templates/>
    <xsl:if test="child::node()[name()='name']">
        <xsl:text disable-output-escaping="yes"></</xsl:text>
        <xsl:value-of select="name"/>
        <xsl:text disable-output-escaping="yes">></xsl:text>
    </xsl:if>
  </xsl:template>

should be two separate rules:

<xsl:template match="group[name]">
  <xsl:element name="{name}">
    <xsl:apply-templates/>
  </xsl:element>
</xsl:template>

<xsl:template match="group[not(name)]">
    <xsl:apply-templates/>
</xsl:template>

>i'm getting this result with some text in the middle... that shouldn't be there...

That's caused by processing the <name> element twice: once explicitly using value-of, and once implicitly when you do apply-templates to all children. Simplest solution is

<xsl:template match="name"/>

Michael Kay
http://www.saxonica.com/
Author, XSLT Programmer's Reference and XPath 2.0 Programmer's Reference

Reznik · June 3rd, 2008, 05:39 AM

Thank you. Now it works :)

I think i will never work with XSL again. I rather prefer to code in assembly or something that i can put logic into it... i couldn't find logic in XSL... but this is just me

thanks again!

mhkay · June 3rd, 2008, 05:45 AM

Yes, you're clearly someone who isn't used to coding at a high level of abstraction, and prefers to be closer to the machine. Do persist though - high level languages (and thinking in terms of higher-level abstractions) will greatly increase your effectiveness as a programmer if you can learn to get to grips with them.

Michael Kay
http://www.saxonica.com/
Author, XSLT Programmer's Reference and XPath 2.0 Programmer's Reference