Simple Transformation: Problems with encoding conversion (angle brackets)

[patricksteffens]

I use ST to create an XML file from ABAP. The transformation works finde except for one tiny, yet crucial part.

Here's what the XML should look like in the end:

<?xml version="1.0" encoding="utf-8"?>
<Paket>
    <Metadaten>
        <DOKNR>N1-QCPRT-000231</DOKNR><DOKAR>DOC</DOKAR><DOKVR>-</DOKVR><DOKTL>000</DOKTL><WFRelevant>J</WFRelevant>
    </Metadaten>
</Paket>

The tags in line 4 have dynamic names, depending on the metadata parameters read by the program. Since ST is not built for handling such dynamic stuff, I create this line manually in ABAP by writing it into a string variable (PAKET.METADATEN) and use it in my transformation:

<?sap.transform simple?>
<tt:transform xmlns:tt="http://www.sap.com/transformation-templates" xmlns:ddic="http://www.sap.com/abapxml/types/dictionary" xmlns:def="http://www.sap.com/abapxml/types/defined">
  <tt:root name="ROOT" type="?"/>
  <tt:root name="PAKET" type="ddic:ZPD_S_XML_PAKET_ROOT"/>
  <tt:template>
    <Paket>
      <tt:serialize>
        <Metadaten><tt:value ref=".PAKET.METADATEN"/></Metadaten>
      </tt:serialize>
    </Paket>
  </tt:template>
</tt:transform>

I transform it writing the output into an xstring variable (to get an UTF-8 XML). The transformation now interprets the angle brackets in the string as '& lt;' for '<' and '& gt;' for '>' (ignore the whitespace ;-)) which leads to the following output XML (in bytes, since this editor here is able to interpret it correctly:

3c3f786d6c2076657273696f6e3d22312e302220656e636f64696e673d227574662d38223f3e0a3c50616b65743e0a093c4d657461646174656e3e266c743b444f4b4e522667743b4e312d51435052542d303030323331266c743b2f444f4b4e522667743b266c743b444f4b41522667743b444f43266c743b2f444f4b41522667743b266c743b444f4b56522667743b2d266c743b2f444f4b56522667743b266c743b444f4b544c2667743b303030266c743b2f444f4b544c2667743b3c2f4d657461646174656e3e0a3c2f50616b65743e<br>

I'm working on a unicode system and thus the string value is represented in unicode. The output must be a UTF-8 xml file.

So the basic question is: How can I either realize such dynamic tag names without filling a string variable or have ST interpret the string properly, so that the angle brackets are shown in the output?

[Tomas_Buryanek]

Check out XML Fragments.

Demo program for ST = DEMO_ST_XSDANY - method prepare_fragment is creating XML string which is later inserted in transformation to result XML.

By the way XML Fragments works in XSL Transformation too (not only in Simple Transformations).

[patricksteffens]

Hi Tomas,

thanks a lot for this hint. This solved the initial problem of broken tags 🙂 However, < and > within such nodes (e.g. arrow symbols in description texts) are still translated into & lt ; and & gt ; (without whitespace) . Do you know why this happens?

Best
Patrick

[Tomas_Buryanek]

patricksteffens It looks like you need to escape the values in your XML string (fragment) before transformation.

For example you have a value "two > one" in tag <value>. After escaping the complete string would look like - EDIT: without spaces after ampresand, sap.com forum is unescaping escaped value > 😞

<value>two & gt ; one</value>

[patricksteffens]

I replaced all < by & lt ; and all > by & gt ; (without whitespace) before the id transformation.

In the output XML, they are not translated back into < and >.Even worse: The & is interpreted as & amp ; which makes somewhat sense to me.

Or did I get you wrong?

[Sandra_Rossi]

Patrick Steffens Why is it a problem that a text with character > is seritalized into & gt; ? (by the way it's not related to the question which was about including an XML fragment in an XML element) That's how XML works: the deserialization will interpret & gt; as the character >. Same thing for & which is serialized into & amp;. Do you have an example to describe your problem? Or is it just because you don't "like" it?

[patricksteffens]

sandra.rossi : We encountered problems where description texts are not translated consistently. E.g. the text "ä<" would be translated into "ä& lt ;" (without whitespace) which is a problem for our software parsing the XML (it fails reading the 'ä' symbol). Thus I would like to have either all or no symbol transformed. So either I would expect all symbols to be translated or none.

Is there an explanation for this observation?

[patricksteffens]

By the way, I tried to escape the tag containing these symbols via CDATA but this did not have an effect.