Reading from a xml file and writing to another

billy_bob_the_3rd · April 28th, 2005, 04:15 PM

Hi there,

Using ASP.NET my goal is to modify a xml document using XmlTextReader and XmlTextWriter (performance reasons). While reading a xml document line by line I want to write selective lines to another file.

My problem lies in the fact that XmlTextReader does not allow you to return a single line of xml, only the InnerXml or OuterXml of a node.

For example, consider the following xml

Code:

<animals>
  <fish selected="true">
    <type>Water</type>
  </fish>
  <monkey selected="false">
    <type>Land</type>
  </monkey>
</animals>

Say I want to create a new xml file that contains all of the selected animals (i.e. selected="true") from the above xml. I would write the xml line by line if the elements do not contain attribute selected="false".

Any suggestions on how I can accomplish this?

Thanks!

mhkay · April 28th, 2005, 04:29 PM

Lines have no significance in XML. If you know how the XML is formatted and want to exploit this fact, then treat is as text and don't use XML software for the job. If you're parsing it as XML, then the significant units are elements, not lines.

Michael Kay
http://www.saxonica.com/
Author, XSLT Programmer's Reference and XPath 2.0 Programmer's Reference

billy_bob_the_3rd · April 29th, 2005, 08:57 AM

Thanks for the reply. I've come up with another way to accomplish my task and wanted your opinion on which method is less process intensive and more optimal.

The first is your suggestion. I would store the xml as a string and split it on a line break. For each index I would check for "selected='true'" and write as needed.

The second method is to re-create the xml using the XmlTextWriter methods (WriteStartElement, WriteAttributeString, etc). Using XmlTextReader, if I find an element that has attribute selected="true" then create a start element with the same name and create all the same attribute value pairs.

Thanks again :-)

billy_bob_the_3rd · April 29th, 2005, 09:12 AM

I just realized that your method won't be as simple as it seems. The concept of depth is gone when using an array of strings vs XmlTextReader. That is, if I have nested elements with the same name then I won't know what the corresponding end element is without keeping track (i.e. creating a stack -> more overhead).

mhkay · April 29th, 2005, 09:14 AM

I wouldn't do either of these, I would do it with XSLT. I don't have time for low-level coding...

Michael Kay
http://www.saxonica.com/
Author, XSLT Programmer's Reference and XPath 2.0 Programmer's Reference

billy_bob_the_3rd · April 29th, 2005, 01:40 PM

Using XSLT involves loading the xml document into memory, which I want to avoid since I'm dealing with very large files.