This isn't still true for latter version of FPC like 2.4.1, right?

http://wiki.freepascal.org/XML_Tutorial#Printing_the_names_of_nodes

"A quick note on navigating the DOM tree: When you need to access nodes in sequence, it is best to use FirstChild and NextSibling properties (to iterate forward), or LastChild and PreviousSibling (to iterate backward). For random access it is possible to use ChildNodes or GetElementsByTagName methods, but these will create a TDOMNodeList object which eventually must be freed. This differs from other DOM implementations like MSXML, because FCL implementation is object-based, not interface-based. "

I think I read that had been "fixed".

Thanks,

I am glad to see someone with an interest in FPCs XML. Do you have some experience with XPath usage in PFC/Lazarus?

Frank Church wrote

I am glad to see someone with an interest in FPCs XML. Do you have some experience with XPath usage in PFC/Lazarus?

No, other than some cursory google searches. I figured I'd be asking about that myself soon.

Frank Church wrote

I am glad to see someone with an interest in FPCs XML. Do you have some experience with XPath usage in PFC/Lazarus?

In this program:
http://www.silvioprog.com.br/downloads/Cliente_SOAP_SAH.zip?attredirects=0&d=1 you can find a simple usage of XPath + WST

Luiz

Thanks for the link. What is WST? My searches turn up something related to eclipse.

Frank Church wrote

Thanks for the link. What is WST? My searches turn up something related to eclipse.

Web Services Toolkit: http://wiki.lazarus.freepascal.org/Web_Service_Toolkit

Luiz

In our previous episode, Lee Jenkins said:

Most, if not all trunk XML fixes have been merged to 2.4.1. And there were a lot of them, thanks to Sergei.

The funny thing is I have already read that wiki page, but did not recognize it in your description.
When I added 'pascal' to the Google search, then it turned up on the first page..

FreePascal users ought to blog some more. ;-)

Luiz,

In Unit1.pas:

XPathResult := EvaluateXPathExpression('//Descricao', XmlTree.DocumentElement);
   if (XPathResult  nil) then
   begin
     for i := 0 to XPathResult.AsNodeSet.Count - 1 do
     begin
       Node := TDOMNode(XPathResult.AsNodeSet[i]);
       if Node.FirstChild  nil then
         Memo1.Lines.Add(UTF8Encode(Node.FirstChild.NodeValue));
     end;
     XPathResult.Destroy;
   end;
   XmlTree.Free;

Why the call to XPathResult.Destroy ? Is it for reference counting? Just seems something else to remember (use destroy instead of free)...

Thanks

No special reason.

Free is redundant in this case since i already knows that XPathResult nil

Luiz

I am having some problems with XPath.

EvaluateXPathExpression raises exceptions with some search strings and I am not sure if it is my limited knowledge of XPath, or a buggy XPath implementation.

The attached demo uses the example from
http://www.w3schools.com/xpath/xpath_examples.asp and I am running 0.9.28.2 beta on Windows

I'm trying to put together a simple test application for xpath now and I've noticed, there doesn't seem to be a way in fpc xml implementation to simply save to a string, an xml document or node.

ReadXML()
ReadXMLFile()

All have no implementation to write a value to a string, only Text, Stream and File. I have to create an intermediary stream, save it to that and then convert that to string or use TStringStream in order to convert them to a string. Odd.

Also the wiki page:
http://wiki.freepascal.org/XML_Tutorial#Create_a_TXMLDocument_from_a_string

Is incorrect. The example will error out because the string passed in is interpreted as a file path, not as a string representation of an xml document.

The XPath implementation works flawlessly with the examples given on w3schools.com site's xpath tutorial.

See the attached simple lcl project. Compile it and then copy and paste examples from the w3schools site and you'll see fpc xpath implementation will make you smile.

I have run it with some success.

What I find missing now is the ability to give the correct output for a an index value, such as //bookstore/book[1]/title. It works on the w3schools version but not in the FPC one.

Are we on the same versions of Lazarus and XPath?

Try changing to only using one front slash instead of two:

/bookstore/book[price>35.00]/title

^--- One slash

If I'm not mistaken, // means from current node, search. / means the entire document. Not sure what you're intent is with the example so that may be exactly what you're trying to do.

When I try it the /bookstore/book[price>35.00]/title I get the error

Conversion from node set to number not possible. The error is the same whether I use a // or a / .

Are we using the same version? Marco's notes refer to updates made in FPC 2.4.1 and I am using 2.2.4 with Lazarus 0.9.28

0.9.29 rev 23574 with FPC 2.4.1

May be this can help you:

procedure Str2XmlFile(var XmlTree: TXMLDocument; const S: String);
var
  Stream: TStringStream;
begin
  Stream := TStringStream.Create(S);
  try
    ReadXmlFile(XmlTree, Stream);
  finally
    Stream.Destroy;
  end;
end;

Luiz

Frank Church wrote

Are we using the same version? Marco's notes refer to updates made in FPC 2.4.1 and I am using 2.2.4 with Lazarus 0.9.28

2.2.4 misses the bulk of Sergei's fixes. If you are serious about fcl-xml, upgrade, preferably to todays 2.4.1 or 2.5.1 from SVN.