Hi,

I have an XML file with the folowing structure:

<?xml version="1.0" encoding="utf-8"?>
<AppName>
<Info>
<Application>Versie 3.0.1</Application>
<CopyRight>(C) FooBar</CopyRight>
<Warning>A message in a bottle</Warning>
</Info>
<Persons Count=99>
<Person Index="000">
<Name>Jan</Name>
<City>Bergentheim</City>
<Email></Email>
</Person">
<Person Index="001">
<Name>Mary</Name>
<City>Amsterdam</City>
<Email></Email>
</Person>
....
</Persons>
</AppName>

What would be the best/most simple way to:
- sequentially read "Count" persons
- find person with Index=X and read Name, City etc?

Currently I iterate through all nodes from the TXMlDocument I read using ReadXMLFile() and parsing all Text-node values. It goes something like this:

    iNode := XMLDoc.FirstChild;
    while iNode  nil do
    begin
      ProcessNode(iNode,  0); // Recursive
      iNode := iNode.NextSibling;
    end;

In ProcessNode I check the nodetype, construct the path, check if the path is valid, determine if it is a name, city, some other field, or invalid and store the value in the appropriate recordfield.

This is rather cumbersome.
Mosty, because I need (want) to check that e.g. a City node must be a child of a Person node, which must be child of Persons node, which must be child the root node.

E.g. if someone text-edits the file (that's the downside of making it human-readable) and makes it like so:

...
<Persons Count=99>
<Name>Klaas</Name>
<Person Index="000">
<Name>Jan</Name>
<City>Bergentheim</City>
<Email></Email>
</Person">

The <Name>Klaas</Name> (whilst being legal xml) should be (silently) disregarded.

So, I'm thinking I'm not using the most optimal method reading the XML-file, certainly there must be suitable classes to do such a thing?

So, basically what I am looking for is, given the TXMlDocument object:
- find value of Persons/Count
- find Persons/Person with Index=X/Name

Preferrably I want a solution that does not drag in databse components (the program is supposed to be stand-alone, no external libraries should be needed).

Bart

Have you considered xpath? It's a core feature of all xml engines (msxml, libxml2, whatever else).

Syntax overview:

http://msdn.microsoft.com/en-us/library/ms256086(v=vs.110).aspx

In my wrapper would look like this:

Person := Document.FindNode(' = "0001"]');

Other examples

// All persons in a city that start with the letter A
  Document.FindNodes('//Person[starts-with(City, "A")]');
// Or to processing save time, select the Persons node
  People := Document.FindNode('//Persons');
// And now xpath is evaluated relative to that
  People.FindNode('Person[Name = "Jan"]');

And so on
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20150111/e4819232/attachment.html>

Free Pascal comes with the XPath unit, which implements the XPath spec. It does not drag in anything except the dom unit.

Michael.

Michael Van Canneyt wrote

Free Pascal comes with the XPath unit, which implements the XPath spec. It does not drag in anything except the dom unit.

I'll have a