views:

194

answers:

3

Hi! I am in the process of refreshing my XST knowledge, and have decided to have a go at making an XSLT 1.0 stylesheet that convert the XMLHelp files from the C# compiler into a better formatted form.

There are a number of issues to get around, but currently I am simply trying to parse those annoying member name attributes, and create another XML document in which the tokens between ':' and '.' characters are extracted and turned into elements.

To start with, I want to turn something like:

<member name="T:PrimeNumbers.Properties.Resources">

into something like:

<member type="T">
  <Properties>
    <Resources />
  </Properties>
</member>

Now, I have gone as far as writing a stylesheet which looks as if it ought to output another XML document, but sadly when I process the data, XML Notepad 2007 crashes, and both IE7 and Firefox 3.5.5 give me an error saying that there is an infinite recursion going on in there.

I would be grateful if somebody could tell me what I have done wrong.

<xsl:transform version="1.0">

  <xsl:output 
    method="xml" version="4.0" encoding="iso-8859-1" 
    indent="yes" media-type="text/xml" 
  />

  <xsl:variable name="AssemblyName" />

  <xsl:template match="/">
    <xsl:apply-templates select="/doc/assembly" />
  </xsl:template>

  <xsl:template match="assembly/name">
    <xsl:variable name="AssemblyName" select="text()" />
    <assembly name="{$AssemblyName}">
      <xsl:apply-templates select="/doc" />
    </assembly>
  </xsl:template>

  <xsl:template match="/doc/members/member">
    <!-- This gives you a single letter 
         (T=Type P=Property M=Method F=Field) -->
    <member type="{substring-before(@name,':')}">
      <xsl:call-template name="RecurseName">
        <!-- This gives you the type name from the beginning 
             of the Namespace to the final local name. -->
        <xsl:with-param name="Path" select="
          substring-after(substring-after(@name,':'),'.')
        " />
      </xsl:call-template>
    </member>
  </xsl:template>

  <xsl:template name="RecurseName">
    <xsl:param name="Path" select="'default'" />
    <xsl:variable name="PathRemainder" select="substring-after($Path,'.')" />
    <xsl:value-of select="$PathRemainder" />
    <xsl:element name="{substring-before($Path,'.')}">
      <xsl:if test="$PathRemainder != ''">
        <xsl:call-template name="RecurseName">
          <xsl:with-param name="Path" select="$PathRemainder" />
        </xsl:call-template>
      </xsl:if>
    </xsl:element>
  </xsl:template>

</xsl:transform>

[Later on, I will transform this document so that if members have matching elements, they are merged together.]

Basically, it crashes XML Notepad 2007 if the RecurseName template even exists - with no elements. Note that <xsl:value-of select="$PathRemainder"/> is pure debuggery.

Any ideas?

Appendix A: Test XML

<?xml version="1.0" encoding="utf-8"?>
<?xml-stylesheet type="text/xsl" href="XmlDocTemplate3.xml" ?>
<doc>
  <assembly>
    <name>PrimeNumbers</name>
  </assembly>
  <members>
    <member name="T:PrimeNumbers.Properties.Resources">
      <summary>
              A strongly-typed resource class, for looking up localized strings, etc.
            </summary>
    </member>
    <member name="P:PrimeNumbers.Properties.Resources.ResourceManager">
      <summary>
              Returns the cached ResourceManager instance used by this class.
            </summary>
    </member>
    <member name="P:PrimeNumbers.Properties.Resources.Culture">
      <summary>
              Overrides the current thread's CurrentUICulture property for all
              resource lookups using this strongly typed resource class.
            </summary>
    </member>
    <member name="M:PrimeNumbers.Program.Main">
      <summary>
            The main entry point for the application.
            </summary>
    </member>
    <member name="M:PrimeNumbers.PrimeNumberForm.CalculatePrimeNumbers(System.Int32)">
      <summary>
            Calculates the prime numbers between 1 and the (count)th prime number.
            </summary>
      <param name="count">The number of prime numbers to return.</param>
      <returns>List of integers</returns>
      <exception cref="T:System.ArgumentOutOfRangeException">Thrown if <paramref name="count" /> is negative.</exception>
    </member>
    <member name="F:PrimeNumbers.PrimeNumberForm.components">
      <summary>
            Required designer variable.
            </summary>
    </member>
    <member name="M:PrimeNumbers.PrimeNumberForm.Dispose(System.Boolean)">
      <summary>
            Clean up any resources being used.
            </summary>
      <param name="disposing">true if managed resources should be disposed; otherwise, false.</param>
    </member>
    <member name="M:PrimeNumbers.PrimeNumberForm.InitializeComponent">
      <summary>
            Required method for Designer support - do not modify
            the contents of this method with the code editor.
            </summary>
    </member>
  </members>
</doc>
+2  A: 

You are encountering an error on the line:

<xsl:element name="{substring-before($Path,'.')}">

when the last portion of the namespace is reached which doesn't contain a dot '.'

You can use this snippet for the RecurseName template instead:

<xsl:template name="RecurseName">
  <xsl:param name="Path" select="'default'"/>
  <xsl:choose>
    <xsl:when test="contains($Path, '.')">
      <xsl:element name="{substring-before($Path,'.')}">
        <xsl:call-template name="RecurseName">
          <xsl:with-param name="Path" select="substring-after($Path,'.')"/>
        </xsl:call-template>
      </xsl:element>
    </xsl:when>
    <xsl:otherwise>
      <xsl:element name="{$Path}" />
    </xsl:otherwise>
  </xsl:choose>
</xsl:template>

I have tested it. Output generated:

<member type="T">
  <Properties>
    <Resources />
  </Properties>
</member>
Rashmi Pandit
An alternative would be to use: `substring-before(concat($Path, '.'),'.')`. This way the tested string always end with a separator, all you need to do is `<xsl:if test="$stringpart != ''">`.
Tomalak
Sadly, this code also seems to have the recursion error. As for the change in code, whilst I would agree that it is better because it more explicit, mine would be more compact. Am I making a fundamental error in assuming that the string functions don't return empty strings ('') and I can't test versus this?Rashmi - just for interest, what XML are you transforming with this XSLT? I am going to post my file, just to be sure.
Mark Bertenshaw
Mark I transformed the member element only, since you had not posted the entire xml. I didn't get any recursion error for that snippet though. Glad, that you found the solution :)
Rashmi Pandit
+1  A: 

I think the infinite recursion error in question is actually being caused by this template here

<xsl:template match="assembly/name">
  <xsl:variable name="AssemblyName" select="text()"/>
  <assembly name="{$AssemblyName}">
    <xsl:apply-templates select="/doc"/>
  </assembly>
</xsl:template>

Where you do <xsl:apply-templates select="/doc" /> within this template, this will match the template itself, because the assembly element is beneath the doc element. Try selecting on /doc/members instead

<xsl:template match="assembly/name">
  <xsl:variable name="AssemblyName" select="text()"/>
  <assembly name="{$AssemblyName}">
    <xsl:apply-templates select="/doc/members"/>
  </assembly>
</xsl:template>

This should stop the infinite recursion errors at least.

Tim C
Tim -You are absolutely right. I'm kicking myself for not spotting this. However, even if this is fixed, there are a few errors which occur. For a start, I haven't taken account of what happens with methods that have parameters - the round brackets are illegal in element names. These have to be stripped off in the template that matches the original member element.Anyway, thanks a lot. This is only stage one. I have a lot more to learn.Cheers,--Mark
Mark Bertenshaw
A: 

Tim C got the right answer, however there are a few other mistakes, which I am correcting purely for posterity. Round brackets in the method names will break my XSLT as well. So before we start recursing, I need to preserve everything between '(' and ')' as a element, and send the name string before the first '(' to the recursive template.

<xsl:transform version="1.0">
 <xsl:output method="xml" version="1.0" encoding="iso-8859-1" indent="yes" media-type="text/xml"/>
 <xsl:variable name="AssemblyName"/>
<xsl:template match="/">
 <xsl:apply-templates select="/doc/assembly"/>
 </xsl:template>
<xsl:template match="assembly/name">
 <xsl:variable name="AssemblyName" select="text()"/>
<assembly name="{$AssemblyName}">
 <xsl:apply-templates select="/doc/members"/>
 </assembly>
 </xsl:template>
<xsl:template match="member">
<member type="{substring-before(@name,':')}">
<xsl:choose>
<xsl:when test="contains(@name,'(')">
<params>
 <xsl:value-of select="substring-before(substring-after(@name,'('),')')"/>
 </params>
<xsl:call-template name="RecurseName">
 <xsl:with-param name="Path" select="substring-after(substring-before(substring-after(@name,':'),'('),'.')"/>
 </xsl:call-template>
 </xsl:when>
<xsl:otherwise>
<xsl:call-template name="RecurseName">
 <xsl:with-param name="Path" select="substring-after(substring-after(@name,':'),'.')"/>
 </xsl:call-template>
 </xsl:otherwise>
 </xsl:choose>
 </member>
 </xsl:template>
<xsl:template name="RecurseName">
 <xsl:param name="Path" select="'default'"/>
<xsl:choose>
<xsl:when test="contains($Path,'.')">
<xsl:element name="{substring-before($Path,'.')}">
<xsl:call-template name="RecurseName">
 <xsl:with-param name="Path" select="substring-after($Path,'.')"/>
 </xsl:call-template>
 </xsl:element>
 </xsl:when>
<xsl:otherwise>
 <xsl:element name="{$Path}"/>
 </xsl:otherwise>
 </xsl:choose>
 </xsl:template>
 </xsl:transform>
Mark Bertenshaw