From: Phrogz <phrogz@...>
Date: 2008-01-24T08:19:57+09:00
Subject: Re: Parsing poorly written XML

On Jan 23, 4:07 pm, Phrogz <phr...@mac.com> wrote:
> On Jan 23, 3:35 pm, Patrick Callahan <hoc...@youskate.com> wrote:
>
> > I've found an application which dumps all the registry settings for
> > installed applications into an .xml file for me (you can find it as
> > "myuninstall" via google).
>
> > I can't figure out the ruby code I need from REXML to do the following:
> > -find a particular "product_name" in the xml file
> > -if found, save off the "uninstall_string" so I can execute it on the
> > commandline and uninstall the app.
>
> require 'rexml/document'
> doc = REXML::Document.new( DATA.read )
>
> # Find 'item' elements at any level
> # that have a child 'name' element whose text value is 'foo'
> doc.each_element( "//item[name='foo']" ){ |item|
>   # Find every child element of the item element
>   # whose name is 'value', get the first one, and get its text
>   puts item.get_elements( "value" ).first.text}
>
> #=> 42
> #=> 17
>
> __END__
> <root>
>   <item>
>     <name>foo</name>
>     <value>42</value>
>   </item>
>   <item>
>     <name>bar</name>
>     <value>17</value>
>   </item>
>   <item>
>     <name>foo</name>
>     <value>17</value>
>   </item>
> </root>

Or, using XPath more liberally (with the same DATA section as quoted
above):

require 'rexml/document'
doc = REXML::Document.new( DATA.read )
p REXML::XPath.match( doc, "//item[name='foo']/value/text()" )
#=> ["42", "17"]