I wanted to be able to have all my RSS feeds combined in the cool (looking but totally impractical) RSS screen saver but sadly it only allows you to select one source feed.

I tried two approaches to solve this:

1. Set up a personal planet planet instance that aggregates all my feeds and point the screen saver at that.
2. Create a combined RSS feed using AppleScript.

Item one I got working but didn’t like the fact that I was then downloading all my RSS feeds twice, so I scrapped the idea.  If however you ever want to create a planet instance that stays in sync with your RSS feeds, use something along the lines of the AppleScript below and execute it periodically.  I’ve snipped most of the paths involved to try to avoid making the lines too long.  A bunch of lines have also been wrapped – the compiler should find them.  It reads the configuration file from config.ini.tmpl, adds the RSS feeds to the end and writes it out to config.ini

set theFile to alias "Groovy:...:config.ini.tmpl"
set fileRef to open for access theFile
set configStart to read fileRef from 0
    until (get eof theFile) as string
close fileRef

set config to configStart
tell application "NetNewsWire"
  set subs to get the subscriptions
    repeat with sub in subs
      if the RSS URL of sub as string is not "" then
        set config to config & "\n" &
          "[" & the RSS URL of sub & "]" & "\n" &
          "name = " & the display name of sub & "\n"
      end if
    end repeat
end tell
set fileRef to open for access
   file "Groovy:...:config.ini" with write permission
set eof fileRef to 0
write config to fileRef as string
close access fileRef
do shell script
  "cd /.../planet-nightly/; python planet.py aj/config.ini"

The second approach seems to work much better – the AppleScript just creates the combined RSS feed directly.  This relies on the first subscription in NetNewsWire being the “All Feeds” item which combines all the feeds together and writes them out to a file.  It may be possible to get this to work directly as a CGI script from Apache and output the RSS instead of storing it in a file, but I couldn’t make it work so the script is just run by cron every so often and the output file is stored in /Library/WebServer/Documents/  You’ll probably need to tweak paths and again there’s a bunch of wrapped lines to fix up.

tell application "NetNewsWire"
  set sub to subscription 1
  set entries to "" as string
  set crlf to "\r\n"
  repeat with entry in the headlines of sub
    set entries to entries & "<item>" & crlf &
      "<title>" & the title of entry & "</title>"
      & crlf & "<guid>" & the guid of entry &
      "</guid>" & crlf & ¬
      "<link>" & the URL of entry &
      "</link>" & crlf & ¬
      "<description><![CDATA[" &
      the description of entry & "]]></description>"
      & crlf & ¬
      "<pubDate>" & the date published of entry &
      "</pubDate>" & crlf & ¬
      "<dc:creator>" & the creator of entry &
      "</dc:creator>" & crlf & ¬
      "</item>" & crlf
  end repeat
  set header to "<?xml version=" & quote &
    "1.0" & quote & "?>" & crlf & ¬
    "<rss version=" & quote & "2.0" & quote &
    " xmlns:dc='http://purl.org/dc/elements/1.1/'>" & crlf & ¬
    "<channel>" & crlf & ¬
    "<title>Planet AJ</title>" & crlf & ¬
    "<language>en</language>" & crlf & ¬
    "<description>Planet AJ</description>" & crlf
	
  set footer to "</channel></rss>"
  set rss to header & entries & footer
end tell
set fileRef to open for access file "Groovy:...:rss20.xml"
    with write permission
set eof fileRef to 0
write rss to fileRef as string starting at 0
close access fileRef

The resulting feed isn’t valid but it’s close enough (the dates are in the wrong format and various problems in the source feeds can cause other errors).  It seems to work well with the RSS screen saver though and that’s all I needed.