Re: [xsl] Grouping to Separate Out Block Level Elements from Mixed Content

Subject: Re: [xsl] Grouping to Separate Out Block Level Elements from Mixed Content
From: "Chris Papademetrious christopher.papademetrious@xxxxxxxxxxxx" <xsl-list-service@xxxxxxxxxxxxxxxxxxxxxx>
Date: Sat, 5 Feb 2022 15:50:53 -0000
Hi Don,

I recently had to solve the opposite problem b inline content in <li> *not*
being wrapped in <p>. But itbs a similar problem b wrap some stuff but not
other stuff. Ibm attaching the code I used.

The basic approach is to use helper functions to identify elements as
block/inline, then <for-each-group group-adjacent=...> to wrap inline element
sequences in the current element but not block element sequences.

Note the following:

  *   My code has some heuristics to wrap <indexterm> elements only when
theybre along with real inline content; you might not need that.
  *   My code uses <xsl:next-match/> because itbs extracted from a larger
XSLT file that does other stuff; you might not need that.
  *   My helper functions use a series of [self::A or self::B or b&] that is
a bit inelegant. I should rewrite it to use the template-based approach
suggested in the recent bUsing node-set variables in predicatesb
discussion on this list, as I now prefer that style for element-class helper
  *   Because youbre solving the reverse problem (breaking content out of an
existing context rather than wrapping it in a new context), be sure to copy
any <p @props/@audience/@product/...> attributes to block elements you push
out of the <p> context.

My template matches like this:

<xsl:template match="*[mine:disallow-inline(.)][node()[mine:is-inline(.)]]">



which says, bmatch any element that contains inline content (is-inline()),
but *I* donbt want it to (disallow-inline()). The important distinction is,
disallow-inline() means *I* donbt want it to contain inline content; it has
nothing to do with what DITA itself allows.

So youbll probably need something more like this:

<xsl:template match="*[mine:disallow-block(.)][node()[mine:is-block(.)]]">



As for what to include in the lists, I used the script at

to dump out our content models, then did some analysis based on which elements
are permitted in <body>, which elements are permitted in <p>, and so on. That
was a manual effort; maybe some day Ibll script it up so I can regenerate
the element lists as adjust our specializations.

  *   Chris

[demime 1.01d removed an attachment of type text/xml which had a name of wrap-inline-content-in-p.xsl]

Current Thread