Red <funkandlove@y...> wrote in
<1178711269.587341.163300@n...>:
> We have certain reports that contain grouping and sums in
> the output. The existing XSL just reads out all rows to
> html tables. I would like to make these easier to read by
> eliminating all duplicates within a row.  At the moment, I
> dont have the access to change the XML output, just the
> XSL.
> 
> <?xml version="1.0"?>
>  <rs:data>

Namespace declarations seem to be missing.

> Birmingham  1001001  Open    5380.04
>             1001002  Open    1281.12
> London      1001003  Closed  1015.32
>             1001004  Open    9866.53
>             1001005  Open    1659.55
> Glasgow     1001006  Open    6944.21

Generalising a bit, it's a trivial grouping problem. The
following is a working example of how it's done:

<xsl:stylesheet version="1.0"
  xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
  xmlns:rs="http://example.org/rs"
  xmlns:z="http://example.org/z">
  <xsl:output method="html"/>
  <xsl:key name="branch" match="z:row" use="@BRANCH"/>
  <xsl:key name="branches" match="z:row"
    use="count(.|key('branch',@BRANCH)[1])=1"/>
  <xsl:template match="rs:data">
    <data>
      <xsl:apply-templates
        select="key('branches',true())" mode="branch"/>
    </data>
  </xsl:template>
  <xsl:template match="z:row" mode="branch">
    <xsl:apply-templates select="key('branch',@BRANCH)"/>
  </xsl:template>
  <xsl:template
    match="z:row[count(.|key('branch',@BRANCH)[1])=1]">
    <row>
      <cell><xsl:apply-templates select="@BRANCH"/></cell>
      <xsl:call-template name="acct-stat-bal"/>
    </row>
  </xsl:template>
  <xsl:template match="z:row">
    <row>
      <cell></cell>
      <xsl:call-template name="acct-stat-bal"/>
    </row>
  </xsl:template>
  <xsl:template name="acct-stat-bal">
    <cell>
      <xsl:apply-templates select="@ACCOUNT_NO"/>
    </cell>
    <cell>
      <xsl:apply-templates select="@STATUS"/>
    </cell>
    <cell>
      <xsl:apply-templates select="@BALANCE"/>
    </cell>
  </xsl:template>
</xsl:stylesheet>

It gets a bit hairier if you need to combine grouping and
sorting. On the other hand, if you can use an
XSLT2-compliant processor, everything suddenly becomes very
easy as long as overusing sequences does not lead to
performance issues. 

<irony intensity="0.75">

Note that for-each is evil and employing it where unneeded
may darn you to heck. If that training next month doesn't
mention it, I advise accusing your mentors of being
blasphemous for-each-worshippers, then preaching the
virtues of template-based approach. Who knows, they might
not be beyond redemption yet.

</irony>

-- 
Pavel Lepin