XSLT subtree extraction

dextermagnific · June 29th, 2006, 08:22 AM

Hi everybody,

i have troubles extracting subtrees from an xml document.

i'd like to extract everything before a <if> tag and everything after it.

e.g:

Code:

<instruction> ... </instruction>
<instruction> ... </instruction>
<if> .... </if>
<instruction> ... </instruction>
<instruction> ... </instruction>

<instruction> could be any tag. i'd like to extract all the content before and after <if>

<xsl:value-of select="if/preceding-sibling::*" /> doesn't work at all. it extract also the contents of the <if> tag

thanks

joefawcett · June 29th, 2006, 12:28 PM

If I start with this:

Code:

<data>
  <element1>1</element1>
  <element2>2</element2>
  <element3>3</element3>
  <if>IF</if>
  <element4>4</element4>
  <element5>5</element5>
  <element6>6</element6>
</data>

and apply this transform:

Code:

<xsl:stylesheet version="1.0"
    xmlns:xsl="http://www.w3.org/1999/XSL/Transform">

<xsl:template match="/">
  <root>
    <xsl:apply-templates select="data/if/preceding-sibling::*"/>
    <xsl:apply-templates select="data/if/following-sibling::*"/>
  </root>
</xsl:template>

  <xsl:template match="*">
    <xsl:copy-of select="."/>
  </xsl:template>
</xsl:stylesheet>

I get this:

Code:

<root>
  <element1>1</element1>
  <element2>2</element2>
  <element3>3</element3>
  <element4>4</element4>
  <element5>5</element5>
  <element6>6</element6>
</root>

So you must be doing something else...

--

Joe (Microsoft MVP - XML)

mhkay · June 29th, 2006, 06:09 PM

You haven't specified the problem very precisely, for example, is there always exactly one <if> element?

But: in XSLT 2.0, do

<xsl:copy-of select="* except if"/>

In 1.0, do

<xsl:copy-of select="*[not(name() = 'if')]"/>

Michael Kay
http://www.saxonica.com/
Author, XSLT Programmer's Reference and XPath 2.0 Programmer's Reference

dextermagnific · June 30th, 2006, 02:02 AM

Hi,

There are many <if> tags. I have a C program in a XML form. the aim is : 1. extract all <if> tags (the 'control') and put them in a particular place in the dest file. 2. extract all but <if> tags (the 'dataflow') and put them elsewhere. Of course, we must extract them by sub-blocks.

e.g :

Code:

// block 1
inst1;
inst2;
...

// control1
if (c) {
  // block 2
  inst3;
  inst4;

  //control2
  if (c2) {
    // block 3
    ...
  } else {
    // block 4
    ...
  }

  // block 5
  inst....
}

block 6
....

the XML C file is the exact translation of the sourcecode.

i have this template

Code:

  <xsl:template name="dataflow-extract-process-flows">
    <xsl:for-each select="block">

      <xsl:value-of select="if/preceding-sibling::*" />

    </xsl:for-each>
  </xsl:template>

What i missed, is if[position()=1]/preceding-sibling::*, now i can call recursively the template with the rest of the code after the first <if>

Hope this will fix my problem.

Thank you for your help.

mhkay · June 30th, 2006, 04:11 AM

Thanks for the more complete explanation. It's always a good idea to post a complete specimen of the input and desired output.

You have a classic "positional grouping" problem (as it happens, I'm talking at the XIME-P conference in Chicago about positional grouping in XQuery later this morning).

Since you're generating the XML, it would be better to generate the proper hierarchical structure in the first place, rather than trying to infer it later.

In XSLT 2.0, the problem is easily tackled using <xsl:for-each-group group-starting-with="if">.

In 1.0, try a search for "XSLT positional grouping". There are two basic approaches:

(a) sibling recursion: apply-templates to all the <if> children. From that template, apply-templates to following-sibling::*[1][not(self::if)], say in mode="next-sibling". From that template, again apply-templates to following-sibling::*[1][not(self::if)], in mode="next-sibling". The recursion terminates automatically when there's no next sibling or when the next sibling is an <if> element.

(b) reduce it to a value-based grouping problem, which you can tackle using the Muenchian grouping method, in which the grouping key is the generate-id() of "." in the case of an <if> element, or of preceding-sibling::if[1] in the case of any other element.

Michael Kay
http://www.saxonica.com/
Author, XSLT Programmer's Reference and XPath 2.0 Programmer's Reference

dextermagnific · June 30th, 2006, 05:00 AM

YES ! this seems a very good solution for my problem ! Thanks a lot.
I'm not aware of XSLT 2.0 functions. I practised XSLT 1.0 in the past and i'm having lot of troubles in grouping items. I hope this will work.

Thanks