Flat XML to nested

IgorK · November 23rd, 2008, 10:35 AM

Hi!

I need to convert some flat XML to nested one. But I don't know if it's possible.

Can anyone help?

Here is my source structure:

Code:

<root>
    <Level_1>
        <Item_1></Item_1>
        <Item_2></Item_2>
    </Level_1>

    <Level_2>
        <Item_21></Item_21>
        <Item_22></Item_22>
    </Level_2>

    <Level_3>
        <Item_31></Item_31>
        <Item_32></Item_32>
    </Level_3>
</root>

This is the target:

Code:

<root>
 <Level_1>
        <Item_1></Item_1>
        <Item_2></Item_2>
     <Level_2>
          <Item_a></Item_a>
          <Item_b></Item_b>
       <Level_3>
           <Item_a1></Item_a1>
           <Item_a2></Item_a2>
       </Level_3>
     </Level_2>
 </Level_1>
</root>

Each of levels can have 1 to n occurrences.

I found a post with similar problem:
(as I am a new member I can't include URL in my post, but you can see it on page 5, post named "Group by" problem )
But there was some relation between nodes and I don't have any.

So I don't know if it's possible, and I don't have any idea how to do this :(

Thanks for any help.
Igor

Martin Honnen · November 23rd, 2008, 11:09 AM

Do you want to use XSLT 2.0 or 1.0?

--
Martin Honnen
Microsoft MVP - XML

mhkay · November 23rd, 2008, 11:10 AM

In XSLT 2.0 use

<xsl:template match="root">
  <xsl:for-each-group select="*" group-starting-with="Level_1">
    <Level_1>
      <xsl:for-each-group select="current-group()" group-starting-with="Level_2">
        <Level_2>
          <xsl:for-each-group select="current-group()" group-starting-with="Level_3">

etc.

In XSLT 1.0 it's rather more difficult. Search for "positional grouping" or "sibling recursion".

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

Martin Honnen · November 23rd, 2008, 11:52 AM

Also if you have

Code:

<root>
Ã Ã <Level_1>
        <Item_1></Item_1>
        <Item_2></Item_2>
Ã Ã </Level_1>
Ã Ã <Level_2>
        <Item_21></Item_21>
        <Item_22></Item_22>
Ã Ã </Level_2>
Ã Ã <Level_1>
        <Item_3></Item_3>
        <Item_3></Item_3>
Ã Ã </Level_1>
Ã Ã <Level_2>
        <Item_23></Item_23>
        <Item_24></Item_24>
Ã Ã </Level_2>
</root>

should that become

Code:

<root>
Ã Ã <Level_1>
        <Item_1></Item_1>
        <Item_2></Item_2>
        <Item_3></Item_3>
        <Item_3></Item_3>
        <Level_2>
        Ã Ã <Item_21></Item_21>
        Ã Ã <Item_22></Item_22>
        Ã Ã <Item_23></Item_23>
        Ã Ã <Item_24></Item_24>
        </Level_2>
Ã Ã </Level_1>
</root>

or something else?

--
Ã Ã Martin Honnen
Ã Ã Microsoft MVP - XML

IgorK · November 24th, 2008, 05:44 AM

Quote:

quote:Originally posted by Martin Honnen
Do you want to use XSLT 2.0 or 1.0?

--
Martin Honnen
Microsoft MVP - XML

Hi Martin!
I am using XSLT 2.0

IgorK · November 24th, 2008, 05:54 AM

Quote:

quote:Originally posted by mhkay
In XSLT 2.0 use

<xsl:template match="root">
<xsl:for-each-group select="*" group-starting-with="Level_1">
    <Level_1>
     <xsl:for-each-group select="current-group()" group-starting-with="Level_2">
        <Level_2>
         <xsl:for-each-group select="current-group()" group-starting-with="Level_3">

etc.

In XSLT 1.0 it's rather more difficult. Search for "positional grouping" or "sibling recursion".

Michael Kay

Author, XSLT 2.0 and XPath 2.0 Programmer's Reference

Hi Michael!

I tried to use your solution.
Here is my code:

Code:

    <xsl:template match="root">

        <xsl:for-each-group select="*" group-starting-with="Level_1">
            <Level_1>
                <Item_1>
                    <xsl:value-of select="Item_1"/>
                </Item_1>
                <Item_2>
                    <xsl:value-of select="Item_2"/>
                </Item_2>
                <xsl:for-each-group select="current-group()" group-starting-with="Level_2">
                    <Level_2>
                        <Item_21>
                            <xsl:value-of select="Item_21"/>
                        </Item_21>
                        <Item_22>
                            <xsl:value-of select="Item_22"/>
                        </Item_22>
                        <xsl:for-each-group select="current-group()" group-starting-with="Level_3">
                            <Level_3>
                                <Item_31>
                                    <xsl:value-of select="Item_31"/>
                                </Item_31>
                                <Item_32>
                                    <xsl:value-of select="Item_32"/>
                                </Item_32>
                            </Level_3>
                        </xsl:for-each-group>
                    </Level_2>
                </xsl:for-each-group>
            </Level_1>
        </xsl:for-each-group>
    </xsl:template>

It's partly working. Inserting blank nodes. For example:

Source:

Code:

<root>
    <Level_1>
        <Item_1>11</Item_1>
        <Item_2>22</Item_2>
    </Level_1>

    <Level_2>
        <Item_21>aa</Item_21>
        <Item_22>dd</Item_22>
    </Level_2>

    <Level_3>
        <Item_31>1234</Item_31>
        <Item_32>4321</Item_32>
    </Level_3>
    <Level_3>
        <Item_31>1235</Item_31>
        <Item_32>556</Item_32>
    </Level_3>
</root>

Target:

Code:

<Level_1>
    <Item_1>11</Item_1>
    <Item_2>22</Item_2>
    <Level_2>
        <Item_21></Item_21>
        <Item_22></Item_22>
        <Level_3>
            <Item_31></Item_31>
            <Item_32></Item_32>
        </Level_3>
    </Level_2>
    <Level_2>
        <Item_21>aa</Item_21>
        <Item_22>dd</Item_22>
        <Level_3>
            <Item_31></Item_31>
            <Item_32></Item_32>
        </Level_3>
        <Level_3>
            <Item_31>1234</Item_31>
            <Item_32>4321</Item_32>
        </Level_3>
        <Level_3>
            <Item_31>1235</Item_31>
            <Item_32>556</Item_32>
        </Level_3>
    </Level_2>
</Level_1>

IgorK · November 24th, 2008, 05:56 AM

Quote:

quote:Originally posted by Martin Honnen
Also if you have

Code:

<root>
Ã Ã <Level_1>
        <Item_1></Item_1>
        <Item_2></Item_2>
Ã Ã </Level_1>
Ã Ã <Level_2>
        <Item_21></Item_21>
        <Item_22></Item_22>
Ã Ã </Level_2>
Ã Ã <Level_1>
        <Item_3></Item_3>
        <Item_3></Item_3>
Ã Ã </Level_1>
Ã Ã <Level_2>
        <Item_23></Item_23>
        <Item_24></Item_24>
Ã Ã </Level_2>
</root>

should that become

Code:

<root>
Ã Ã <Level_1>
        <Item_1></Item_1>
        <Item_2></Item_2>
        <Item_3></Item_3>
        <Item_3></Item_3>
        <Level_2>
        Ã Ã <Item_21></Item_21>
        Ã Ã <Item_22></Item_22>
        Ã Ã <Item_23></Item_23>
        Ã Ã <Item_24></Item_24>
        </Level_2>
Ã Ã </Level_1>
</root>

or something else?

--
Ã Ã Martin Honnen
Ã Ã Microsoft MVP - XML

That should become:

Code:

<root>
<Level_1>
        <Item_1></Item_1>
        <Item_2></Item_2>
                <Level_2>
                        <Item_21></Item_21>
                        <Item_22></Item_22>
                </Level_2>
</Level_1>

<Level_1>
        <Item_3></Item_3>
        <Item_4></Item_4>
                <Level_2>
                        <Item_23></Item_23>
                        <Item_24></Item_24>
                </Level_2>
</Level_1>

</root>

mhkay · November 24th, 2008, 07:43 AM

This

<Item_31>
<xsl:value-of select="Item_31"/>
</Item_31>

creates an empty element in the output if there is no element in the input. To avoid that, just replace it with

<xsl:copy-of select="Item_31"/>

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

IgorK · November 24th, 2008, 08:47 AM

Quote:

quote:Originally posted by mhkay
This

<Item_31>
<xsl:value-of select="Item_31"/>
</Item_31>

creates an empty element in the output if there is no element in the input. To avoid that, just replace it with

<xsl:copy-of select="Item_31"/>

Michael Kay

Author, XSLT 2.0 and XPath 2.0 Programmer's Reference

Thanks for your help. Now my code is this:

Code:

<xsl:template match="root">

        <xsl:for-each-group select="*" group-starting-with="Level_1">
            <Level_1>

                <xsl:copy-of select="Item_1"/>
                <xsl:copy-of select="Item_2"/>

                <xsl:for-each-group select="current-group()" group-starting-with="Level_2">
                    <Level_2>

                        <xsl:copy-of select="Item_21"/>
                        <xsl:copy-of select="Item_22"/>

                        <xsl:for-each-group select="current-group()" group-starting-with="Level_3">
                            <Level_3>

                                <xsl:copy-of select="Item_31"/>
                                <xsl:copy-of select="Item_32"/>

                            </Level_3>
                        </xsl:for-each-group>
                    </Level_2>
                </xsl:for-each-group>
            </Level_1>
        </xsl:for-each-group>
    </xsl:template>
</xsl:stylesheet>

Now it's doing the job well, but there is only one strange thing.
In each node there is one empty next node. For example:

Source

Code:

<root>
    <Level_1>
        <Item_1>11</Item_1>
        <Item_2>22</Item_2>
    </Level_1>

    <Level_2>
        <Item_21>aa</Item_21>
        <Item_22>dd</Item_22>
    </Level_2>

    <Level_3>
        <Item_31>1234</Item_31>
        <Item_32>4321</Item_32>
    </Level_3>
    <Level_3>
        <Item_31>1235</Item_31>
        <Item_32>556</Item_32>
    </Level_3>

</root>

Target

Code:

<Level_1>
    <Item_1>11</Item_1>
    <Item_2>22</Item_2>
    <Level_2>
        <Level_3/>
    </Level_2>
    <Level_2>
        <Item_21>aa</Item_21>
        <Item_22>dd</Item_22>
        <Level_3/>
        <Level_3>
            <Item_31>1234</Item_31>
            <Item_32>4321</Item_32>
        </Level_3>
        <Level_3>
            <Item_31>1235</Item_31>
            <Item_32>556</Item_32>
        </Level_3>
    </Level_2>
</Level_1>

mhkay · November 24th, 2008, 09:34 AM

I hope you've tried to understand my code and made an attempt to see where it was wrong - it's a bad idea to just use code without trying to understand it. The bug is where it does select="current-group()" - you only want to process all items in the current group after the first. You can do that using subsequence(current-group(),2).

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