Re: [xsl] Sorting complex objects

["G. Ken Holman" <gkholman@xxxxxxxxxxxxxxxxxxxx>]

At 2009-03-08 01:47 -0700, Mark Wilson wrote:
> I need to transform a file of random-order complex items into an 
> ordered file. I tried using 'for-each-group' nested, but failed to 
> get the results I wanted
>
> As I said, in the input file, the items are random and an item may 
> occur more than once and have <SomeData> elements with different 
> contents - both <SomeData> elements need to be preserved in the 
> output file. The <Items> are ordered by <Heading>, then sequentially 
> by the three <SubDivX> elements, and finally by <SomeData>. 
> Everything sorts as strings.
>
> I have been thinking about something like:

You were very close, all you were missing was the reconstitution of 
the <Item> element.

Also, where you simply concatenated your string components into your 
key, I introduced a delimiter between them.  Most times a simple 
concatenation won't do, because the concatenation puts nodes out of order.

Consider the two sequences in order of string values, left to right:

  ('ab','xy')
  ('abcd',ef')

Simple concatenation will put these out of order as:

  ('abcd','ef') <!--key "abcdef"-->
  ('ab','xy')   <!--key "abxy"-->

But, what to use for the delimiter?  Unicode reserves U+FDD0 as 
"non-characters", and as such, expects these not to be in any file 
interchanged by users.  But they are, effectively, available as 
application-internal private-use code points.  And they sort *after* 
characters used by users in XML documents, which makes them flexible 
in the use of concatenation.

But what about at the low end of the scale?  Any valid XML character, 
just based on the property of being a valid XML character, is a 
character that might very well have been input by the user.  So you 
have to use a character that is unlikely to be input by the 
user.  Because of end-of-line-sequence normalization, I think that 
the carriage return is a very unlikely character to find in the input 
by the user, and it sorts *before* characters used in XML documents.

Putting that together, if I wanted absent values to sort after 
present values, I would use the following with your data:

  group-by="concat(Heading,'&#xfdd0;',
                   SubDiv1,'&#xfdd0;',
                   SubDiv2,'&#xfdd0;',SubDiv3)">

... thereby absent values would be considered the COBOL "high 
values".  If I wanted absent values to sort before present values, as 
I do, I would use the following with your data to mimic COBOL "low 
values", keeping the U+FDD0 just because of the very (very!) small 
chance of finding a CR in the input:

  group-by="concat(Heading,'&#xd;&#xfdd0;',
                   SubDiv1,'&#xd;&#xfdd0;',
                   SubDiv2,'&#xd;&#xfdd0;',SubDiv3)">

And I get the solution I think you need.  But note that your proposed 
output seems a little out of order due to the trailing "s" on "Avenue".

I hope this helps.

. . . . . . . Ken

t:\ftemp>type mark.xml
<List>
<Item>
   <Heading>Prague</Heading>
   <SubDiv1>Avenues</SubDiv1>
   <SubDiv2>Crosswalks</SubDiv2>
   <SomeData>12</SomeData>
</Item>
<Item >
   <Heading>Prague</Heading>
   <SomeData>1</SomeData> <!-- Needs to me consolidated-->
</Item>
<Item>
   <Heading>Prague</Heading>
   <SubDiv1>Avenues</SubDiv1>
   <SomeData></SomeData>
</Item>
<Item>
<Heading>Prague</Heading>
<SubDiv1>Avenue</SubDiv1>
<SubDiv2>Crosswalks</SubDiv2>
<SubDiv3>Dangerous</SubDiv3>
<SomeData></SomeData>
</Item>
<Item>
   <Heading>Bonn</Heading>
   <SubDiv1>Avenue</SubDiv1>
   <SubDiv2>Crosswalks</SubDiv2>
   <SubDiv3>Dangerous</SubDiv3>
   <SomeData></SomeData>
</Item>
<Item>
   <Heading>Prague</Heading>
   <SubDiv1>Streets</SubDiv1>
   <SomeData></SomeData>
</Item>
<Item>
   <Heading>Washington</Heading>
   <SomeData></SomeData>
</Item>
<Item >
   <Heading>Prague</Heading>
   <SomeData>2</SomeData><!-- Needs to me consolidated-->
</Item>
</List>

t:\ftemp>call xslt2 mark.xml mark.xsl
<?xml version="1.0" encoding="UTF-8"?>
<Item>
   <Heading>Bonn</Heading>
   <SubDiv1>Avenue</SubDiv1>
   <SubDiv2>Crosswalks</SubDiv2>
   <SubDiv3>Dangerous</SubDiv3>
   <SomeData/>
</Item>
<Item>
   <Heading>Prague</Heading>
   <SomeData>1</SomeData>
   <SomeData>2</SomeData>
</Item>
<Item>
   <Heading>Prague</Heading>
   <SubDiv1>Avenue</SubDiv1>
   <SubDiv2>Crosswalks</SubDiv2>
   <SubDiv3>Dangerous</SubDiv3>
   <SomeData/>
</Item>
<Item>
   <Heading>Prague</Heading>
   <SubDiv1>Avenues</SubDiv1>
   <SomeData/>
</Item>
<Item>
   <Heading>Prague</Heading>
   <SubDiv1>Avenues</SubDiv1>
   <SubDiv2>Crosswalks</SubDiv2>
   <SomeData>12</SomeData>
</Item>
<Item>
   <Heading>Prague</Heading>
   <SubDiv1>Streets</SubDiv1>
   <SomeData/>
</Item>
<Item>
   <Heading>Washington</Heading>
   <SomeData/>
</Item>
t:\ftemp>type mark.xsl
<?xml version="1.0" encoding="US-ASCII"?>
<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform";
                version="2.0">

<xsl:output indent="yes"/>

<xsl:template match="List">
  <xsl:for-each-group select="Item"
                      group-by="concat(Heading,'&#xd;&#xfdd0;',
                                       SubDiv1,'&#xd;&#xfdd0;',
                                       SubDiv2,'&#xd;&#xfdd0;',SubDiv3)">
    <xsl:sort select="current-grouping-key()"/>
    <!--reconstitute the Item-->
    <Item>
      <xsl:copy-of select="Heading,SubDiv1,SubDiv2,SubDiv3"/>
      <!--consolidating the <SomeData> elements-->
      <xsl:for-each select="current-group()">
        <xsl:sort select="SomeData"/>
        <xsl:copy-of select="SomeData"/>
      </xsl:for-each>
    </Item>
  </xsl:for-each-group>
</xsl:template>

</xsl:stylesheet>

t:\ftemp>rem Done!



--
XQuery/XSLT training in Prague, CZ 2009-03 http://www.xmlprague.cz
XQuery/XSLT/XSL-FO training in Los Angeles/Anaheim - 2009-06-01/10
Training tools: Comprehensive interactive XSLT/XPath 1.0/2.0 video
Video lesson:    http://www.youtube.com/watch?v=PrNjJCh7Ppg&fmt=18
Video overview:  http://www.youtube.com/watch?v=VTiodiij6gE&fmt=18
G. Ken Holman                 mailto:gkholman@xxxxxxxxxxxxxxxxxxxx
Crane Softwrights Ltd.          http://www.CraneSoftwrights.com/s/
Male Cancer Awareness Nov'07  http://www.CraneSoftwrights.com/s/bc
Legal business disclaimers:  http://www.CraneSoftwrights.com/legal