[XSL-LIST Mailing List Archive Home] [By Thread] [By Date]

Re: [xsl] Replace content of element, then transform it...


Subject: Re: [xsl] Replace content of element, then transform it...
From: Wendell Piez <wapiez@xxxxxxxxxxxxxxxx>
Date: Thu, 30 Aug 2012 12:04:35 -0400

Trond,

Indeed, your case illustrates exactly why disable-output-escaping is a trap.

One might think that if you have

<e>&lt;f>g&lt;/f></e>

and then

<xsl:template match="e">
  <xsl:copy>
    <xsl:value-of select="." disable-output-escaping="yes"/>
  </xsl:copy>
</xsl:template>

you then get <e><f>g</f></e>.

And this might even be true, in some cases -- but not in all. The reason? "<e><f>g</f></e>" is a sequence of characters, and XSLT does not generate sequences of characters, but nodes arranged in trees. You get a sequence of characters only when some subsequent process creates one from a tree generated by the transformation.

In this case, the tree would be shaped like this:

- element 'e'
  - text '<f>g</f>'

The process that turns this into an XML representation (tags and text: "<e><f>g</f></e>") is called "serialization".

In your case, however, having generated such a tree, your XSLT submits it to be processed with a set of templates. No markup is created at all until the *final* result of your processing pipeline is written in the form of characters.

There is no 'f' element to be matched and processed in this second step.

Among other consequences of working with trees (by and large beneficial ones), this means that the design pattern of "hiding" markup inside XML by escaping it is a poor one, which is bound to create problems. At best it works only if it is carefully managed.

How to manage it (assuming you can't change your input data)? You have two choices:

1. Serialize your intermediate results before processing it again. (Split your stylesheet into two, and run the second on the serialized output of the first.) This can work, but it doesn't scale well, as it requires both serializing and then parsing again in the middle of every transformation. Even with solid-state storage devices, this tends to be time- and resource-intensive. And it locks you down to certain kinds of transformation architecture (namely on a file system).

2. Find another way of parsing the pseudo-markup hidden in your data. For example, use an extension such as saxon:parse(), which will turn a string into a tree of nodes (assuming it's well-formed XML).

Cheers,
Wendell

On 8/30/2012 9:01 AM, trond.huso@xxxxxx wrote:
Right. I take that note.

I am also noticing that disable-output-escaping is deprecated in XSLT2.0, so I guess I shall try and figure out another way of doing this...

Trond


-----Original Message----- From: Michael Kay [mailto:mike@xxxxxxxxxxxx] Sent: 30. august 2012 14:37 To: xsl-list@xxxxxxxxxxxxxxxxxxxxxx Subject: Re: [xsl] Replace content of element, then transform it...

One of the main reasons that we've been telling people for 12 years not to use disable-output-escaping is that it couples the transformation too closely to the serialization, meaning it's difficult and inefficient to reuse your code as part of a pipeline. A lot depends on what this d-o-e stuff is really doing. Having said that, from the information supplied I don't know why you are getting the error you are.

Michael Kay
Saxonica


On 30/08/2012 12:08, trond.huso@xxxxxx wrote:
Hi,

I have the following XSLT.
<xsl:template match="body">
          <body>
              <xsl:value-of select="." disable-output-escaping="yes"/>
          </body>
      </xsl:template>

      <xsl:template match="leadtext">
          <leadtext>
              <xsl:value-of select="." disable-output-escaping="yes"/>
          </leadtext>
      </xsl:template>


<xsl:template match="node()|@*"> <xsl:copy> <xsl:apply-templates select="node()|@*"/> </xsl:copy> </xsl:template>

Which does what is intended. Just that I want to not output it, but start w= orking on it in a phase-two process.
So I tried this

<xsl:template match="body">
          <body>
              <xsl:value-of select="." disable-output-escaping="yes"/>
          </body>
      </xsl:template>

      <xsl:template match="leadtext">
          <leadtext>
              <xsl:value-of select="." disable-output-escaping="yes"/>
          </leadtext>
      </xsl:template>


<xsl:template match="node()|@*"> <xsl:variable name="foo"> <xsl:copy> <xsl:apply-templates select="node()|@*"/> </xsl:copy> </xsl:variable> <xsl:apply-templates select="$foo" mode="phase2" /> </xsl:template> <!-- Error message: Description: Cannot create an attribute node (id) whose parent is a documen= t node -->

After reading about how this works, I now understand why I am getting the e= rror. Is there another alternative to make this possible in one document, o= r do I have to send the output to a new document?

Best regards,

Trond Husx
System Developer
Mobile : +47 450 35 715
E-mail : trond.huso@xxxxxx
www.ntb.no

-- ====================================================================== Wendell Piez mailto:wapiez@xxxxxxxxxxxxxxxx Mulberry Technologies, Inc. http://www.mulberrytech.com 17 West Jefferson Street Direct Phone: 301/315-9635 Suite 207 Phone: 301/315-9631 Rockville, MD 20850 Fax: 301/315-8285 ---------------------------------------------------------------------- Mulberry Technologies: A Consultancy Specializing in SGML and XML ======================================================================


Current Thread
Keywords