xsl-list
[Top] [All Lists]

RE: [xsl] Slow XSLT

2008-03-02 06:38:00
Hi Michael,

Michael wrote:

Its easy to see why your code is slow. But I'm
afraid I can't suggest any
way to speed it up, because I simply can't work out
what it is trying to do.

Please find below my previous posts and replies I got.
I hope it becomes clearer what I am trying to do. If
not, please let me know so that I can try and come up
with a better explanation.

Cheers

C
++++++++++
I need to build a nested heading Rows/Columns (<TR>
and <TD>). The output should look like this for the
2003 heading:

                        2003
Quarter 1        Quarter 2        Quarter 3
    Quarter 4

 <ColGrp heading="Quarter">
   <ColGrp heading="2003">
     <Col heading="Quarter 1" />
     <Col heading="Quarter 2" />
     <Col heading="Quarter 3" />
     <Col heading="Quarter 4" />
   </ColGrp>
  </ColGrp>


I use the xpath below to get me to the inner most
ColGrp (parent of Col elements).

Normally $depth = 1. This means I do not want to
output the outer most ColGrp - <ColGrp
heading="Quarter">

//ColGrp[count(ancestor::ColGrp)=$depth]

//ColGrp[count(ancestor::ColGrp)=1]

So, I go to the ancestor:ColGrp whose depth = 1
This will produce the first <TD>s (

2003      2004

Then all the Col elements will produce the second
row with all the <TD>s

Quarter 1   Quarter 2      Quarter 3       Quarter
4

And lastly, for each Col element I have to go back
to the Measure elements to produce the final Row which
will be the same for all the Col elements

P.S. The last row and <td>s come from the Measure
elements and I use the position() function to output
them like that:
 
Total Pages Cost Total Pages Cost Total Pages Cost
Total Pages Cost

Therefore, this is the output I get with my XSLT
but it is too slow for big xml docs.

                          2003
Quarter 1     Quarter 2    Quarter 3        Quarter
4
Total Pages Cost Total Pages Cost Total Pages Cost


<Measures>
  <Measure idx="1" heading="Total Pages" />
  <Measure idx="2" heading="Cost" />
</Measures>
<Columns>
 <ColGrp heading="Quarter">
   <ColGrp heading="2003">
     <Col heading="Quarter 1" />
     <Col heading="Quarter 2" />
     <Col heading="Quarter 3" />
     <Col heading="Quarter 4" />
   </ColGrp>
  </ColGrp>
</Columns>


++++++++++++++++++

Hi Manfred,

Thanks for your reply.

I think I can now understand what David poited out
i.e. that when using // can be very costly. 

However, I thought that by using // followed by a
predicate ( //ColGrp[count(ancestor::ColGrp)=$depth]
), I would go straight to only those inner most ColGrp
elements parent of the Col elements.

There is one problem with your approach though. You
are assuming that there will always be only two ColGrp
like this:

<xsl:for-each
select="/Reports/Report/Columns/ColGrp/ColGrp[count(ancestor::ColGrp)=$depth]">

<ColGrp heading="Quarter">
   <ColGrp heading="2003">
     <Col heading="Quarter 1" />
     <Col heading="Quarter 2" />
     <Col heading="Quarter 3" />
     <Col heading="Quarter 4" />
   </ColGrp>
</ColGrp>

But That is not the case. The xml can have as many
nested ColGrp as necessary. For instance:

<ColGrp heading="Quarter">
   <ColGrp heading="2003">
     <ColGrp heading="Name">
       <Col heading="Quarter 1" />
       <Col heading="Quarter 2" />
       <Col heading="Quarter 3" />
       <Col heading="Quarter 4" />
    </ColGrp>
   </ColGrp>
</ColGrp>

or

<ColGrp heading="Quarter">
   <ColGrp heading="2003">
     <ColGrp heading="Name">
       <ColGrp heading="Charity Report IO">
          <Col heading="Quarter 1" />
          <Col heading="Quarter 2" />
          <Col heading="Quarter 3" />
          <Col heading="Quarter 4" />
       </ColGrp>
    </ColGrp>
   </ColGrp>
</ColGrp>

As you can see, there can be many ColGroup levels
nested inside each other. Therefore, we need an
approach that is dynamic i.e. no matter how many
nested ColGrp there are, it will produce the right
output.

As David has pointed out the // operator is very
costly. So I have
replaced them by explicit xpath expressions and
would be interested
about the performance now. However I could not
resist to replace
  count(.//Col)*$msrs
by
  count(Col)*$msrs
Most of the rest I have l left unchanged. I also
noticed the
depreciated <nobr>, which should be replaced by css
white-space:
nowrap.

Manfred

<?xml version="1.0" encoding="UTF-8" ?>
<xsl:stylesheet version="2.0"
        xmlns:xsl="http://www.w3.org/1999/XSL/Transform";>
<xsl:param name="axisHeads" select="'false'" />
<!--
java -jar D:\Programme\Saxon6\saxon.jar slow.xml
slow.xsl
-->
<xsl:output indent="yes"/>

 <xsl:variable name="msrs">
   <xsl:choose>
     <xsl:when
test="/Reports/Report/Measures/Measure">
       <xsl:value-of
select="count(/Reports/Report/Measures/Measure)"/>
     </xsl:when>
     <xsl:otherwise>
       <xsl:value-of select="1"/>
     </xsl:otherwise>
   </xsl:choose>
 </xsl:variable>

 <xsl:template match="Reports">
 <html>
 <body>
   <xsl:apply-templates/>
 </body>
 </html>
 </xsl:template>

 <xsl:template match="Report" >
     <!-- Top -->
     <div id="g1" style="position: absolute; top: 0px;
left: 0px;
width: 400px; height: 12px">
       <table class="grdTop" border="0"
cellspacing="1"
 cellpadding="0">
         <tbody>
           <xsl:apply-templates select="Columns" />
         </tbody>
       </table>
     </div>
  </xsl:template>

 <xsl:template match="Columns">
   <xsl:apply-templates select="ColGrp[1]"
mode="Header">
     <!-- 0 for top level heading, 1 to cut it out -->
     <xsl:with-param name="depth">
       <xsl:choose>
         <xsl:when test="$axisHeads='true'">
           <xsl:value-of select="0"/>
         </xsl:when>
         <xsl:otherwise>
           <xsl:value-of select="1"/>
         </xsl:otherwise>
       </xsl:choose>
     </xsl:with-param>
   </xsl:apply-templates>
 </xsl:template>

 <xsl:template match="ColGrp" mode="Header">
   <xsl:param name="depth" />
   <tr>
     <!-- the very first row needs a padding cell -->
     <xsl:for-each
select="/Reports/Report/Columns/ColGrp/ColGrp[count(ancestor::ColGrp)=$depth]">
         <td colspan="{count(Col)*$msrs}"
align="center"
 style="overflow:none">
         <nobr>
           <div>
             <xsl:value-of select="@heading"/>
           </div>
         </nobr>
       </td>
     </xsl:for-each>
   </tr>
   <xsl:choose>
     <xsl:when test="ColGrp">
       <xsl:apply-templates select="ColGrp[1]"
mode="Header">
         <xsl:with-param name="depth"
select="$depth+1" />
       </xsl:apply-templates>
     </xsl:when>
     <xsl:otherwise>
       <xsl:apply-templates select="Col[1]"
mode="colHead" />
     </xsl:otherwise>
   </xsl:choose>
 </xsl:template>

 <xsl:template match="Col" mode="colHead">
   <tr>
     <xsl:for-each
select="ancestor::Columns/ColGrp/ColGrp/Col">
       <td colspan="{$msrs}" valign="top"
align="center"
 style="overflow:none">
         <nobr>
           <!-- 2003/2004 -->
           <div>
             <xsl:value-of select="@heading"/>
           </div>
         </nobr>
       </td>
     </xsl:for-each>
   </tr>
   <tr valign="bottom">
     <xsl:for-each
select="/Reports/Report/Columns/ColGrp/ColGrp/Col">
       <xsl:apply-templates
select="/Reports/Report/Measures">
         <xsl:with-param name="pos"
select="position()" />
       </xsl:apply-templates>
     </xsl:for-each>
   </tr>
 </xsl:template>

 <xsl:template match="Measures">
   <xsl:param name="pos" />
   <xsl:for-each select="Measure">
     <xsl:variable name="mPos">
       <xsl:value-of select="position()" />
     </xsl:variable>
     <td align="center">
       <nobr>
         <div class="action" style="width:90px;
overflow:none"
onclick="sortFullGrid({$mPos}, {$pos}, '{(_at_)class}')">
           <xsl:value-of select="@heading"/>
         </div>
       </nobr>
     </td>
   </xsl:for-each>
 </xsl:template>

</xsl:stylesheet>

+++++++++++++++++++++





 

--- Michael Kay <mike(_at_)saxonica(_dot_)com> wrote:


I have written the XSLT below and it works fine.
However, when the XML has lots of RowGrp elements
then it 
becomes quite slow to transform.

I see no RowGrp elements, I assume you mean
ColGrp...

Its easy to see why your code is slow. But I'm
afraid I can't suggest any
way to speed it up, because I simply can't work out
what it is trying to do.
You haven't given any clues, and reverse engineering
the requirements from
some rather strangely-written code is not easy.

Michael Kay
http://www.saxonica.com/



The difficulty that I am having is to find a way
to speed it 
up. I think there could be two problems with my
XSLT:

1 - Recursion in the ColGrp Template mode="header"
 

<xsl:for-each
select="//ColGrp[count(ancestor::ColGrp)=$depth]">

2 - Also, when I am in the Col Template I have to
go back to 
the measures element for each Col

<xsl:for-each select="//Col">
  <xsl:apply-templates select="//Measures">
    <xsl:with-param name="pos" select="position()"
/>
  </xsl:apply-templates>
</xsl:for-each>

I would appreciate if someone could sched some
light.
I have read Jeni's XSLT pages (key and grouping)
but have not 
found any posts or example similar to mine.

Cheers

C
===========================
XML
<Reports>
<Report xmlns="">
  <Measures>
    <Measure idx="1" heading="Total Pages" />
    <Measure idx="2" heading="Cost" />
  </Measures>
  <Columns>
    <ColGrp heading="Quarter">
      <ColGrp heading="2003">
        <Col heading="Quarter 1" />
        <Col heading="Quarter 2" />
        <Col heading="Quarter 3" />
        <Col heading="Quarter 4" />
      </ColGrp>
      <ColGrp heading="2004">
        <Col heading="Quarter 1" />
        <Col heading="Quarter 2" />
        <Col heading="Quarter 3" />
        <Col heading="Quarter 4" />
      </ColGrp>
    </ColGrp>
    <!--===TOTALS====-->
    <ColGrp>
      <ColGrp>
        <Col heading="Total" />
      </ColGrp>
    </ColGrp>
  </Columns>
</Report>
</Reports>

==============================

XSLT

<xsl:param name="axisHeads" select="'false'" />

  <xsl:variable name="msrs">
    <xsl:choose>
      <xsl:when test="//Measure">
        <xsl:value-of select="count(//Measure)"/>
      </xsl:when>
      <xsl:otherwise>
        <xsl:value-of select="1"/>
      </xsl:otherwise>
    </xsl:choose>
  </xsl:variable>

 <xsl:template match="Reports">
    <xsl:apply-templates
select="Report"></xsl:apply-templates>
  </xsl:template>
  
  <xsl:template match="Report" >
      <!-- Top -->
      <div id="g1" style="position: absolute; top:
0px; left: 0px; width: 400px; height: 12px">
        <table class="grdTop" border="0"
cellspacing="1" cellpadding="0">
          <tbody>
            <xsl:apply-templates select="Columns"
/>
          </tbody>
        </table>
      </div>
   </xsl:template>

  <xsl:template match="Columns">
    <xsl:apply-templates select="ColGrp[1]"
mode="Header">
      <!-- 0 for top level heading, 1 to cut it
out
-->
      <xsl:with-param name="depth">
        <xsl:choose>
          <xsl:when test="$axisHeads='true'">
            <xsl:value-of select="0"/>
          </xsl:when>
          <xsl:otherwise>
            <xsl:value-of select="1"/>
          </xsl:otherwise>
        </xsl:choose>
      </xsl:with-param>
    </xsl:apply-templates>
  </xsl:template>
 
  <xsl:template match="ColGrp" mode="Header">
    <xsl:param name="depth" />
    <tr>
      <!-- the very first row needs a padding cell
-->
      <xsl:for-each
select="//ColGrp[count(ancestor::ColGrp)=$depth]">
          <td colspan="{count(.//Col)*$msrs}"
align="center" style="overflow:none">
          <nobr>
            <div>
              <xsl:value-of select="@heading"/>
            </div>
          </nobr>
        </td>
      </xsl:for-each>
    </tr>
    <xsl:choose>
      <xsl:when test="ColGrp"> 
        <xsl:apply-templates select="ColGrp[1]"
mode="Header">
          <xsl:with-param name="depth"
select="$depth+1" />
        </xsl:apply-templates>
      </xsl:when>
      <xsl:otherwise>
        <xsl:apply-templates select="Col[1]"
mode="colHead" />
      </xsl:otherwise>
    </xsl:choose>
  </xsl:template>

  <xsl:template match="Col" mode="colHead">
    <tr>
      <xsl:for-each
select="ancestor::Columns//Col">
        <td colspan="{$msrs}" valign="top"
align="center" style="overflow:none">
          <nobr>
            <!-- 2003/2004 -->
            <div>
              <xsl:value-of select="@heading"/>
            </div>
          </nobr>
        </td>
      </xsl:for-each>
    </tr>
    <tr valign="bottom">
      <xsl:for-each select="//Col">
        <xsl:apply-templates select="//Measures">
          <xsl:with-param name="pos"
select="position()" />
        </xsl:apply-templates>
      </xsl:for-each>
    </tr>
  </xsl:template>

  <xsl:template match="//Measures">
    <xsl:param name="pos" />
    <xsl:for-each select="Measure">

=== message truncated ===



      ___________________________________________________________ 
Rise to the challenge for Sport Relief with Yahoo! For Good  

http://uk.promotions.yahoo.com/forgood/

--~------------------------------------------------------------------
XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list
To unsubscribe, go to: http://lists.mulberrytech.com/xsl-list/
or e-mail: <mailto:xsl-list-unsubscribe(_at_)lists(_dot_)mulberrytech(_dot_)com>
--~--