[xsl] accumulators and continuous numbering

["Graydon graydon@xxxxxxxxx" <xsl-list-service@xxxxxxxxxxxxxxxxxxxxxx>]

So I'm converting a bunch of OOXML into DITA using XSLT 3 in oXygen with Saxon
9.8.0.12 PE.

There's a requirement to use short (4 digit) unique identifiers for the
resuling files and for these to be unique across the content set.  There are
many resulting DITA files per source OOXML document.

"A chance to try accumulators!" I thought; I can pre-process the whole thing
with file numbers in attributes.  This isn't an attempt to use streaming; I've
got a few tens of MB of source.

The accumulator works, BUT the number sequence restarts if I try to use the
accumuator across a sequence of document nodes or a sequence of element nodes.

To produce a single continuous sequence of monotonically increasing integers
with the accumulator, I have to get the whole content set into the same tree
before creating the numbering attributes.

Is that the expected behaviour?

What I have defines the accumulator,

<xsl:accumulator as="xs:integer" initial-value="0" name="documentNumber">
  <xsl:accumulator-rule
    match="w:p[w:pPr/w:pStyle/@w:val = $listOfFileStyles]"
    select="$value + 1">
  </xsl:accumulator-rule>
</xsl:accumulator>

pulls the content set into a variable using the file and arch extensions:

<xsl:variable as="element(w:document)+" name="contentSet">
  <xsl:for-each select="$OOXMLsrc"> <!-- a sequence of path strings -->
    <xsl:sequence
      select="(file:path-to-uri(.) => file:read-binary() => arch:extract-text('word/document.xml') => parse-xml())/*" />
  </xsl:for-each>
</xsl:variable>

then applies the numbering template

<xsl:variable as="element(w:document)+" name="numberedSrc">
 <xsl:apply-templates mode="fileNumber" select="$contentSet" />
</xsl:variable>

<xsl:mode name="fileNumber" on-no-match="shallow-copy" />

<xsl:template
		match="w:p[w:pPr/w:pStyle/@w:val = $listOfFileStyles]"
		mode="fileNumber" >
  <xsl:copy>
	<xsl:apply-templates mode="fileNumber" select="@*" />
	<xsl:attribute name="fileNumber" select="accumulator-before('documentNumber') => format-number('0000')" />
	<xsl:apply-templates mode="fileNumber" select="node()" />
  </xsl:copy>
</xsl:template>

It works; the correct paragraphs are numbered.  But the sequence restarts for
each new member of the sequence.

If instead I use:

<xsl:variable as="element(bucket)+" name="contentSet">
  <bucket>
    <xsl:for-each select="$OOXMLsrc"> <!-- a sequence of path strings -->
      <xsl:sequence
        select="(file:path-to-uri(.) => file:read-binary() => arch:extract-text('word/document.xml') => parse-xml())/*"
 	  />
    </xsl:for-each>
  </bucket>
</xsl:variable>

<xsl:variable as="element(bucket)" name="numberedSrc">
  <xsl:apply-templates mode="fileNumber" select="$contentSet" />
</xsl:variable>

I get one continous sequence of numbers.

Is there a better way to get the single continuous sequence of numbers?

Thanks!

-- Graydon