Programming Languages on Power

The XML string inside the <requestXml> tags can't have any of the XML-syntax characters like < > & etc. They have to be encoded as &lt; &gt; &amp; etc.

If I code my program like this, where the "embedded" tag is an XML document, I get RNX0353 saying that the XML document doesn't match the RPG variable. The <a> tag is seen as a child of my <embedded> tag. 

dcl-s xmlstring varchar(100); 
dcl-s embedded varchar(20); 

xmlstring = '<embedded><a>hello</a></embedded>'; 
xml-into embedded %xml(xmlstring);
return;  

This version of my program works, because it doesn't see "<a>", it sees "&lt;a&gt;"

dcl-s xmlstring varchar(100); 
dcl-s embedded varchar(20); 

xmlstring = '<embedded>&lt;a&gt;hello&lt;/a&gt;</embedded>'; 
xml-into embedded %xml(xmlstring);
return;  

Hi Paul
Scan and substitute is where I ended up too. But it felt wrong that something so basic could not be done with the standard tools and I wondered if I had missed something obvious.
Thanks for your help, Colin

But in the context of XML-INTO's design, it is not "something so basic".  The op-code is intended to break XML into its individual elements - it does it very well. You are effectively expecting a toaster to fry an egg for you.

There are tools available that could do this - although I doubt they would be any simpler than a couple of scans and a substring.

Are you passing this portion of the document to another process which will decompose it? Trying to understand why you want top extract this piece.

Hi Colin,

what you plan to do is not the purpose of a XML parser. You would be able to do this with SAX, but I think, its a PITA.

I would use embedded SQL and a regular expression to extract the part of the XML that I need:

exec sql values regexp_substr(
    '<client>...</client><authentication>....</authentication> <request><requestType>....</requestType><requestXML>.....</requestXML></request>',
    '<request>(.+?)<\/request>',
    1,
    1,
    'ni',
    1
) into :varCharVariable;

This should do what you want. Of course you can replace the source string also with a :HostVariable.

HTH and kind regards,

Daniel

Daniel - I don't think SAX would work for the same reason that XML-INTO won't it is designed to handle individual elements.

Your SQL is basically doing substring operations and given that that is built-in to RPG I'm sure it would be slower and, to me at least, is less obvious.

If this is a process (with differering data sources) then I might be tempted to write a general purpose parser that could be used with DATA-INTO. But straight forward substring extraction is about as easy as it gets,

Jon,

well - last time I programmed with SAX was in Java and its "many moons" ago - so I'm not sure, but I thought we were able to access the raw XML stream in the SAX "event handler" - but as I said, it's long ago, and my memory might not serve me right. 

The pattern matching with regular expressions is quite normal at any other platform - and it's also quite handy, because you're even able to check for contents inside the "substring" - and all that without having to code long %scan and %subst expressions. 

You can also test and develop your regular expressions on several websites (e.g. regex101.com) and your SQL interactively - not so easy with a complex %scan and %subst expression.

Regular expressions might be somehow unfamiliar for us (the seasoned RPG programmers) but it's mainstream for nearly all others. So I think of it as some sort of "future proofing" an application. 

And finally to speed/performance - we all know what Hoare said: Premature optimization is the root of all evil. So I didn't make any assumptions about performance as I don't know if it's in fact a factor. 

But if performance might be a problem, it's possible to compile the regular expression and apply it multiple times using the highly optimized C library - from an RPG programmers POV this a huge PITA but it's very fast at runtime. 

Regards,

Daniel