Question

I have run into a dilemma. In a particular application, I’m receiving XML results from a SOAP request that look like this:

<env:Envelope xmlns:env='http://schemas.xmlsoap.org/soap/envelope/'>
  <env:Header />
  <env:Body>
    <ns1:searchResponse xmlns:ns1='http://url.to.namespace' xmlns:xsi='http://www.w3.org/2001/XMLSchema-instance'>
      <ns1:result>&lt;?xml version="1.0"?&gt;&lt;results count="201" returned="201" code="200" msg="successful"&gt;&lt;result order="0"&gt;&lt;dirkey&gt;DK886shn3525&lt;/dirkey&gt;&lt;eid&gt;smith&lt;/eid&gt;&lt;email&gt;[e-mail]&lt;/email&gt;&lt;fn&gt;Smith&lt;/fn&gt;&lt;ln&gt;Bob&lt;/ln&gt;&lt;wid&gt;859589157&lt;/wid&gt;&lt;score&gt;70&lt;/score&gt;&lt;/result&gt;&lt;result order="1"&gt;&lt;dirkey&gt;DK547fjx6702&lt;/dirkey&gt;&lt;eid&gt;james31&lt;/eid&gt;&lt;email&gt;[e-mail]&lt;/email&gt;&lt;fn&gt;Tim&lt;/fn&gt;&lt;ln&gt;Allen&lt;/ln&gt;&lt;stu&gt;&lt;lvl&gt;Senior&lt;/lvl&gt;&lt;plans&gt;&lt;plan&gt;Technology Management-B&lt;/plan&gt;&lt;/plans&gt;&lt;contacts&gt;&lt;contact type="permanent"&gt;&lt;city&gt;Salina&lt;/city&gt;&lt;phone&gt;[phone]&lt;/phone&gt;&lt;postal&gt;67401&lt;/postal&gt;&lt;street1&gt;1111 Main Ln&lt;/street1&gt;&lt;state&gt;KS&lt;/state&gt;&lt;/contact&gt;&lt;/contacts&gt;&lt;/stu&gt;&lt;wid&gt;2222222222&lt;/wid&gt;&lt;score&gt;20&lt;/score&gt;&lt;/result&gt;</ns1:result>
    </ns1:searchResponse>
  </env:Body>
</env:Envelope>

I am most interested in the data contained within the <ns1:result> element. While this might make sense in an HTML world, I need the <ns1:result> text as XML. Intrigued by the possibility of doing this via XSL, I constructed the following stylesheet:

<?xml version="1.0"?>
<xsl:stylesheet version="1.0"
  xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
  xmlns:env="http://schemas.xmlsoap.org/soap/envelope/"
  xmlns:ns1="http://url.to.namespace"
  exclude-result-prefixes="env ns1">

  <xsl:output omit-xml-declaration="yes" indent="yes" method="text" />
  <xsl:strip-space elements="*"/>

  <!-- Template #1 - Identity Transform -->
  <xsl:template match="node()|@*">
    <xsl:copy>
      <xsl:apply-templates select="node()|@*"/>
    </xsl:copy>
  </xsl:template>

  <!-- Template #2 - for all text() nodes, disable output escaping -->
  <xsl:template match="text()">
    <xsl:copy-of select="." disable-output-escaping="yes" />
  </xsl:template>

</xsl:stylesheet>

…which technically does produce what I want:

<?xml version="1.0"?>
<results count="201" returned="201" code="200" msg="successful">
  <result order="0">
    <dirkey>DK886shn3525</dirkey>
    <eid>smith</eid>
    <email>[e-mail]</email>
    <fn>Bob</fn>
    <ln>Smith</ln>
    <wid>859589157</wid>
    <score>70</score>
  </result>
  <result order="1">
    <dirkey>DK547fjx6702</dirkey>
    <eid>ta</eid>
    <email>[e-mail]</email>
    <fn>Tim</fn>
    <ln>Allen</ln>
    <stu>
      <lvl>Senior</lvl>
      <plans>
        <plan>Technology Management-B</plan>
      </plans>
      <contacts>
        <contact type="permanent">
          <city>Salina</city>
          <phone>[phone]</phone>
          <postal>67401</postal>
          <street1>1111 Main Ln</street1>
          <state>KS</state>
        </contact>
      </contacts>
    </stu>
    <wid>2222222222</wid>
    <score>20</score>
  </result>
</results>

However, I’ve heard it said that DOE is the sign of a desperate individual. Indeed, when I try to run this XSLT through an application of ours (one that is designed to transform XML before passing it on to a templating engine), it doesn’t work. I’m guessing that DOE is not implemented in our particular XSL parser…

So, here’s the ultimate question: is there a way in XSLT 1.0 to unescape these entities without using a parser-specific tactic like DOE? My one thought is constructing a method that translates certain escaped characters (e.g., >) into their literal counterparts (>)…but I’m not entirely sure how I’d go about that.

As always, I appreciate your assistance.

P.S. Please, don’t bother telling me how disgusting this output is or how they’ve mangled their document structure; we’ve already tried to get them to change it and that’s not an option. 🙁

Answer 1

So, here’s the ultimate question: is there a way in XSLT 1.0 to
unescape these entities without using a parser-specific tactic like
DOE? My one thought is constructing a method that translates certain
escaped characters (e.g., >) into their literal counterparts
(>)…but I’m not entirely sure how I’d go about that.

There isn’t a pure XSLT way to reconstruct destroyed markup — until XSLT 3.0 (still a W3C working draft) that will xave a standard function parse-xml()

Until you have XSLT 3.0 available, the safe way to reconstruct destroyed markup is to call an extension function with a similar signature that you have to write yourself.

This extension function will try to parse its string argument into an instance of XmlDocument and if successful, return back the result.