C[omp]ute

From: "andrew cooke" <andrew@...>

Date: Mon, 10 Mar 2008 07:18:08 -0400 (CLT)

Over the weekend I made a new (sub-)site for my photos -
http://www.acooke.org/photography/

I generated the site using a similar approach to the Pancito site, but
instead of using Halipeto (the Haskell based templating system that uses
XML as a database) I used XSLT with a Python script.

There are two XML files - one describes the structure of the site and the
other describes each photograph.  There are also two XSLT files - one for
the front page and that is re-used for each photograph.  The Python script
looks after directory deletion/creation, copying of photos, and the
calling of the two XSLTs.

The site is completely static (from the server's POV) - each image has a
different page.  The front page randomly loads an image via Javascript on
the client.

The trickiest part was getting the previous/next links working from XSLT. 
The XSLT for a photo is invoked on the photo description XML, so it needs
to explicitly load the site description XML to find the neighbouring
photographs.  However, that turned out to be quite simple in practice:

  <xsl:variable name="site" select="document('../xml/site.xml')"/>
  <xsl:variable name="previous"
select="$site//site:image[@id=$image]/preceding-sibling::*[position()=1]"/>

where $image is the ID of the current photo (passed from Python).

Andrew