[lxml] Re: Accessing variables from within an XSL file

Hi Adrian/Stefan, Apologies for dredging up a 2-year-old thread but I've hit the same issue and found that although the posted solution works for variables declared with simple string values, it fails when considering xsl:param values or "calculated" values. Consider: http://www.w3.org/1999/XSL/Transform" xmlns:abc="bar" extension-element-prefixes="abc"> default param value <baz></baz> Now let's say I want to receive the value of "$var1" in my extension-element foo(): #!/usr/local/bin/python3 # -*- coding: utf-8 -*- import sys from lxml import etree from copy import deepcopy class FooExtensionElement( etree.XSLTExtension ): def execute( self, context, self_node, input_node, output_parent ): print( "attr1: {}".format( self_node.get( "attr1" ))) parser = etree.XMLParser( load_dtd=True ) extensions = { ( 'bar', 'foo' ) : FooExtensionElement() } transformer = etree.XSLT( etree.parse( sys.stdin, parser ), extensions = extensions ) xml_doc = etree.XML( '<dummy/>' ) result = transformer( xml_doc, param1=etree.XSLT.strparam( "passed param value" )) print( result ) When run, the output is: attr1: {$var1} <?xml version="1.0"?> <baz bif="http://passed param value">XSLT ouput: http://passed param value</baz> This shows that: * self_node.get( "attr1" ) returns the literal string "{$var1}" in execute(), i.e. the transformer has not evaluated {$var1} when passing it to abc:foo/@attr1 * baz/@bif gets the evaluated value of {$var1} as expected If I were to perform the solution suggested below and lookup //xsl:variable[ @name = 'var1']/@select from within execute(), it would return the string "concat( 'http://', $param1 )" I could attempt to evaluate this in self_node.xpath(), but I don't have access to the value of $param1, right? So this would result in: lxml.etree.XPathEvalError: Undefined variable What if I first looked up the value of $param1 in the same way as suggested below for xsl:variable? //xsl:param[ @name = 'param1']/text() would return "default param value", NOT the actual value ("passed param value") which was passed in: AFAIK, I have *no* way to access that value from execute(). This problem goes away if evaluation of {$xyz} -syntax is honored for extension-elements. As demonstrated above, this is already working perfectly for elements in the XSL which have no namespace (the "baz" element in my example). Is there a fundamental reason why extension-elements are excluded from these evaluations? If this contravenes XSLT convention, is it possible to place a map of variable/param names and their values within the 'context' parameter passed to the execute() function so that extension elements can perform their own lookup? Many thanks! Mike Adrian Bool wrote:

Hi Stefan, Many thanks for your thoughts. I hadn't considered just executing an XPath query directly against the XSL as simple XML – but that works! I combined this idea with your xpath expression and now have a function that will return the correct value: def get_variable_from_xsl(variable_name, xsl): xpath_results = xsl.xpath("//xsl:variable[@name=$name]", name=variable_name, namespaces={'xsl':'http://www.w3.org/1999/XSL/Transform' }) for xpath_result in xpath_results: if xpath_result.get('name') == variable_name: for xpath_child in xpath_result.getchildren(): if 'select' in xpath_child.keys(): return xpath_child.get('select') return None destination_url = get_variable_from_xsl('destination_url', xsl) print(f"destination_url = {destination_url}") Which returns: destination_url = 'https://my.domain.example.com/' Thanks again! aid On Wed, 10 Apr 2019 at 18:25, Stefan Behnel stefan_ml@behnel.de wrote:

...

Adrian Bool schrieb am 10.04.19 um 10:49: I'm trying to extract from data from XSL files on an IBM DataPower system. I can't change the XSLT files - and therefore need to work around their current structure to get the info I need. Well, XSL is XML, so you can always post-process the XSL document after parsing it as XML, before you pass it into the XSLT() constructor. I'm doing this from within Python 3.7 using etree.__version__ 4.3.3. Here are my imports, a dummy XML object and a cutdown version of the XSL file from the IBM system: from lxml import etree from textwrap import dedent

xml = etree.XML("""<root></root>""")

xsl = etree.XML(dedent( """<?xml version="1.0"?> http://www.w3.org/1999/XSL/Transform" xmlns:dp="http://www.datapower.com/extensions" extension-element-prefixes="dp">

xsl:copy

https://my.domain.example.com/'"

/>

<!--Set the destination URL in the context variable-->

value="$destination_url" /> I don't think I've ever tried this myself in lxml. I guess I would have expected value="{$destination_url}" rather than the plain variable reference here. Can't say how XSLT is meant to deal with this. """))

My goal is to use lxml to process this file and extract the contents of the DataPower variable 'var://context/mpgw/Destination' - i.e. the contents of the xsl:variable $destination_url - in the above case ' https://my.domain.example.com/'. I've created an XSLT extension for the "dp:set-variable" element to try to capture this information: class MyExtElement(etree.XSLTExtension): def execute(self, context, self_node, input_node, output_parent): if self_node.tag == '{

http://www.datapower.com/extensions%7Dset-variable': print(f"Setting variable of name {self_node.get('name')} to value {self_node.get('value')}") extensions = { ('http://www.datapower.com/extensions',

'set-variable') : MyExtElement() } result = etree.XSLT(xsl, extensions=extensions)(xml) print(f"Result: {result}") When this is being called; instead of the value of destination_url I'm getting the static "$destination_url" string instead - with the output of the above script being: Setting variable of name 'var://context/mpgw/Destination' to value

$destination_url Result: <?xml version="1.0"?> <root/> I am rather suprised that the $destination_url xsl:variable was not being evaluated - does anyone know why this is the case? Given the above, I've been trying to find a way to access the $destination_url xsl:variable from within my XSLTExtension - but unfortunately with no luck. Did you try looking it up in the XSL document via XPath? Something like .xpath("//xsl:variable[@name=$name]", name="destination_url") (plus namespaces) Stefan _________________________________________________________________ Mailing list for the lxml Python XML toolkit - http://lxml.de/ lxml@lxml.de https://mailman-mail5.webfaction.com/listinfo/lxml