sort child elements based on its attribute value which is expected to alphanumeric - XSLT

[Sharmu Dev]

Hi,

I need to sort child elements in a xml file based on its attribute value which is expected to alphanumeric using XSLT.
Please give some idea.

In the below xml file I need to sort H2 using its @PNO

Input file:

<WINDOW>
<H2 PNO="10,888a">...</H2>
<H2 PNO="10,888">...</H2>
<H2 PNO="10,887a">...</H2>
<H2 PNO="10,887">...</H2>
....
<H2 PNO="1000">...</H2>
</WINDOW>

<WINDOW> transforms to <asselmbly> and <H2> transforms to <include>

Expected output should be

<assembly>
<include cite="1000"/>
....
<include cite="10,887"/>
<include cite="10,887a"/>
<include cite="10,888"/>
<include cite="10,888a"/>
</assembly>

[g tsuji]

You should have read in every tutorial discussing sorting in xslt 1.0 that xsl:sort can appear in two kinds of structure. This is what they want to say.

[1] In xsl:apply-templates

<xsl:template match="WINDOW">
    <assembly>
        <xsl:apply-templates select="H2">
            <xsl:sort select="@PNO" order="ascending" data-type="text" />
        </xsl:apply-templates>
    </assembly>
</xsl:template>

Test you skill in making the template for H2.

[2] In xsl:for-each

<xsl:template match="WINDOW">
    <assembly>
        <xsl:for-each select="H2">
            <xsl:sort select="@PNO" order="ascending" data-type="text" />
            <!-- again test you skill in transforming the H2 here -->
        </xsl:for-each>
    </assembly>
</xsl:template>

[Sharmu Dev]

Hi g tsuji,

Thanks for your response. I tried the way you gave <xsl:sort select="@PNO" order="ascending" data-type="text" /> but little bit modified as <xsl:sort select="if(contains(@PNO,',')) then replace(@PNO,',','') else @PNO" order="ascending" data-type="text" /> it worked for the scenario which i showed in my post with H2/@PNO values (10,888a, 10,888, 10,887a, 10,887, 1000) and when I added another H2 with @PNO="2000" it didnt work.

My code as of now:

<xsl:template match="WINDOW">
<assembly>
<xsl:apply-templates>
<xsl:sort select="if(contains(@PNO,',')) then replace(@PNO,',','') else @PNO" order="ascending" data-type="text" />
</xsl:apply-templates>
</assembly>
</xsl:template>

<xsl:template match="H2">
<include area="h2area" num="{@PNO}"/>
</xsl:template>

Input:

<?xml version="1.0" encoding="UTF-8"?>
<WINDOW>
<H2 PNO="10,888a">...</H2>
<H2 PNO="10,888">...</H2>
<H2 PNO="10,887a">...</H2>
<H2 PNO="10,887">...</H2>
<H2 PNO="1000">...</H2>
<H2 PNO="2000">...</H2>
</WINDOW>

Output:

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

<assembly>
<include area="h2area" num="1000"/>
<include area="h2area" num="10,887"/>
<include area="h2area" num="10,887a"/>
<include area="h2area" num="10,888"/>
<include area="h2area" num="10,888a"/>
<include area="h2area" num="2000"/>
</assembly>

[Sharmu Dev]

Hi g tsuji,

I further tried on sorting the attribute value by separating its numeric and alphabets as below it worked

My code:
<xsl:template match="WINDOW"> <assembly> <xsl:apply-templates select="H2"> <xsl:sort select="if(contains(functx:substring-before-match(@PNO,'[A-Za-z]'),',')) then replace(functx:substring-before-match(@PNO,'[A-Za-z]'),',','') else functx:substring-before-match(@PNO,'[A-Za-z]')" order="ascending" data-type="number" /> <xsl:sort select="functx:substring-after-match(replace(@PNO,',',''),'[0-9]+')" order="ascending" data-type="text"/> </xsl:apply-templates> </assembly> </xsl:template> <xsl:template match="H2"> <include area="h2area" num="{@PNO}"/> </xsl:template> <xsl:function name="functx:substring-before-match" as="xs:string"> <xsl:param name="arg" as="xs:string?"/> <xsl:param name="regex" as="xs:string"/> <xsl:sequence select="tokenize($arg,$regex)[1]"/> </xsl:function> <xsl:function name="functx:substring-after-match" as="xs:string?"> <xsl:param name="arg" as="xs:string?"/> <xsl:param name="regex" as="xs:string"/> <xsl:sequence select="replace($arg,concat('^.*?',$regex),'')"/> </xsl:function>

Input:

<?xml version="1.0" encoding="UTF-8"?><?dtd name="cptj"?> <WINDOW> <H2 PNO="2000">...</H2> <H2 PNO="10888az">...</H2> <H2 PNO="10888">...</H2> <H2 PNO="10888a">...</H2> <H2 PNO="3000">...</H2> <H2 PNO="10887">...</H2> <H2 PNO="10887b">...</H2> <H2 PNO="1000">...</H2> </WINDOW>

Output:
<?xml version="1.0" encoding="utf-8"?> <assembly> <include area="h2area" num="1000"/> :D <include area="h2area" num="2000"/> :D <include area="h2area" num="3000"/> :D <include area="h2area" num="10887"/> :D <include area="h2area" num="10887b"/> :D <include area="h2area" num="10888"/> :D <include area="h2area" num="10888a"/> :D <include area="h2area" num="10888az"/> :D </assembly>


Thanks please provide comments in order to improvement my implementation.

[Paul Clapham]

I put your XML into XML code tags, so it would be easier for people to read.

As for improvements: it seems to me that
if(contains(functx:substring-before-match(@PNO,'[A-Za-z]'),',')) then replace(functx:substring-before-match(@PNO,'[A-Za-z]'),',','') else functx:substring-before-match(@PNO,'[A-Za-z]')
is equivalent to
replace(functx:substring-before-match(@PNO,'[A-Za-z]'),',','')
is it not? This change would make the code easier to read as well.