Working XPath and mapping examples

*UPDATED 9 JULY 2010*

These examples were provided by our institution's resident XPath/XML expert (not me). Hopefully I have the details correct. Note that I'm using the feeds_xmlparser code from http://github.com/chrisirhc/feeds_xmlparser which allows for namespaces.

Example Feed: http://rss.oucs.ox.ac.uk/engfac/notshakespeare-audio/rss20.xml

XPath: /rss/channel/item or //item

/rss/channel/item will only look for items that nested within /rss/channel e.g.:

<rss>
    <channel> 
        <item>blah</item>
        <item>more blah</item>
        <item>another blah</item>
    </channel>
<rss>

//item will look for items anywhere in the XML, regardless of nesting. For this particular RSS feed, either XPath setting is suitable. With my settings, 6 nodes would be created from this feed, one for each item.

Mapping Sources:

1. title - maps the item title e.g. "The Spanish Tragedy: Thomas Kyd"
2. link - maps the URL e.g. http://media.podcasts.ox.ac.uk/engfac/fhs/01_spanish_tragedy.mp3
3. category - maps *all* instances of category. So if I mapped 'category' to a Taxonomy field, I would get the following comma-separated values for the first node:

english, language, jacobean, elizabethan, theatre, renaissance, Q323, 1, ukoer, 106104

In my feed, Q323, 1, ukoer, 106104 are auto-generated from certain category choices in the feed source, and I wish to separate them from the user-entered freetext keywords english, language etc. So instead I use:

4. category[not(@domain)] - this excludes any category element that has 'domain' in it. Fortunately that is all the ones I don't want and so this gives me:

english, language, jacobean, elizabethan, theatre, renaissance

Now, I may want to map Q323, 1, ukoer, 106104 to individual taxonomy or CCK fields. Here is how I do that:

5. category[@domain='http://rss.oucs.ox.ac.uk/transcripts_available'] - maps 1
6. category[@domain='http://rss.oucs.ox.ac.uk/jacs_codes'] - maps Q323
7. category[@domain='http://www.jisc.ac.uk/oer/'] - maps ukoer
8. category[@domain='http://www.itunesu.com/feed'] - maps 106104

Perhaps I might want to combine some of these into a single field, for example ukoer and Q323, as two comma-separated values in a single taxonomy field.

9. category[@domain='http://www.jisc.ac.uk/oer/' or @domain='http://rss.oucs.ox.ac.uk/jacs_codes']

The namespaces were initially a little trickier to figure out.

10. *[local-name()='author'] - maps itunes:author e.g. Emma Smith. Strictly, this is mapping all instances of ***:author which would cause problems if I had an additional element called, say, podcast:author. Fortunately I don't, but let's say I did:

11. *[starts-with(name(),'itunes:') and local-name()='author'] - maps the element that begins with the namespace 'itunes:' and has the local name or 'author'. But we can simplify this:

12. *[name()='itunes:author'] - maps itunes:author - this is the one we would use

I'll continue to post more detailed examples from this feed once I've had a chance to test them all. Please post your own examples below including a link to your feed.