Re: generate a rss.xml from a bunch of HTML files

[davidson <davidson@freevolt.org>]

On Mon, 10 May 2021 Emanuel Berg wrote:
> > > ...and this somewhat more complex-looking one...
> > >
> > >  "W3C RSS 1.0 News Feed Creation How-To"
> > >  https://www.w3.org/2001/10/glance/doc/howto
> >
> > Great, but stops on <figure> and <figurecaption>,

Elsewhere in the thread you seem to have moved on from XSLT to more
promising options, but I'll make a few comments here anyways.

I suspect that those specific tags are not a primary cause of
difficulty.

I could be mistaken, of course. But I am unable to replicate this
without more information about the input document.

I wonder whether the input document is XML. (No unclosed tags, etc.)

> > these are HTML5 tags:
> >
> >   http://html5doctor.com/the-figure-figcaption-elements/
> >
> > so either we must change the XSLT rules to make use of for
> > example the caption at least, _or_ we must either make
> > a rule or tell the tool to ignore them, if such an option
> > exists...
>
> Here is the Makefile [last] only one problem, the XSLT file or
> xsltproc tool (?) doesn't seem to transform the HTML into RSS,
> really, output is basically a text file with no markup whatsoever
> except for the first line which is
>
>  <?xml version="1.0" encoding="utf-8"?>

TLDR: What you describe will happen when none of an XSLT stylesheet's
template rules match anything in the input.

XSLT is template-based, a little bit like sed or awk, but instead of
processing records (lines) in a text file it processes the nodes of an
XML tree.

When a node matches no template in your stylesheet, then *built-in*
template rules are applied:

  * The built-in template rule for the "document node" (at top of the
  tree), is to apply templates to that node's children.

  * The built-in template rule for any element is the same as for the
    document node -- apply templates to the children of that element.

  * And, when we reach the leaves of the tree, the built-in template
  rule for text nodes is to copy the text to the result tree.

As a consequence, applying an XSLT stylesheet to a document that
matches none of the templates in the stylesheet results in output that
looks identical to the output you would get by applying a trivial
stylesheet containing no template rules at all!

It's a little like how the output of

 $ sed '' somefile

is indistinguishable from

 $ cat somefile

> Maybe I do something wrong?

I lack fluency in make/Makefile, and I have not dug into the weeds of
that stylesheet at https://www.w3.org/2001/10/glance/doc/howto .

However, when you call xsltproc it looks to me like you not are
supplying any of the four parameters that the stylesheet html2rss.xsl
expects:

 <xsl:param name = "Base" />
 <xsl:param name = "Channel" />
 <xsl:param name = "xmlfile" />
 <xsl:param name = "xslfile" />
 <xsl:param name = "Page" />

You might supply them like so

 $ xsltproc -o Overview.rss
            --stringparam xmldata "$webpage" \
            --stringparam xlsfile html2rss.xsl \
            --stringparam Base "$(dirname "$webpage")" \
            --stringparam Page "$(dirname "$webpage")" \
            --stringparam Channel Overview.rss \
            html2rss.xsl \
            "$webpage"

> name = tree-house
>
> src = ${name}.html
>
> srcpp = ${name}-pp.html
>
> trans = html2rss.xsl
>
> dst = ${name}.rss
>
> opts = --html
>
> all: ${dst}
>
> ${srcpp}: ${src}
> 	sed -e 's/<\/*fig\(ure\|caption\)>//g' $< > $@
>
> ${dst}: ${srcpp}
> 	xsltproc -o $@ ${opts} ${trans} $<

--
Ce qui est important est rarement urgent
et ce qui est urgent est rarement important
-- Dwight David Eisenhower