Hello! I've finished the script and let it run through all the OPML files that have been sent to me. All feeds seem to work great with the current file naming scheme, except for two: http://feeds.themerlinshow.com/TheMerlinShow This feed always has "episode.mp4" as file name. The episode URLs are redirects to the real files, which resolve to something like this: > $ HEAD -S > http://feeds.themerlinshow.com/~r/TheMerlinShow/~5/125754107/redirect.mp4 > HEAD > http://feeds.themerlinshow.com/~r/TheMerlinShow/~5/125754107/redirect.mp4 --> > 302 Moved Temporarily > HEAD > http://www.podtrac.com/pts/redirect.mp4?http://media.libsyn.com/media/themerlinshow/tms_020_Jesse_Thorn_02_ipod.mp4 > --> 302 Found > HEAD > http://media.libsyn.com/media/themerlinshow/tms_020_Jesse_Thorn_02_ipod.mp4 > --> 302 Found > HEAD > http://cdn2.libsyn.com/themerlinshow/tms_020_Jesse_Thorn_02_ipod.mp4 > --> 200 OK This means that for this podcast with "redirect.mp4" as base name for every episode, we can use a HTTP head request to determine a valid file name. This is a suboptimal solution, so I'm thinking about how we can avoid such situations. http://americanpublicmedia.publicradio.org/podcasts/xml/prairie_home_companion/news_from_lake_wobegon.xml This one unfortunately constructs its URLs so that episodes from different months but same day-of-month collide, e.g. http://download.publicradio.org/podcast/phc/2008/01/26_nflw_64.mp3 and http://download.publicradio.org/podcast/phc/2008/04/26_nflw_64.mp3 I have to think about what we can do in such a situation. I am thinking about adding a dictionary of filename => url pairs that can be used to check if a downloaded file has been downloaded from a specific URL or not. If so, we know that the filename corresponds to the URL. If not, we have to rename the file and (after downloading it) save the file name in the dictionary. The only problem is that we also have to look in our dictionary if there is a generated filename for every URL for which a downloaded file doesn't exist in the default location. Another problem is that we want to be downwards-compatible with previous versions (i.e. downloads from previous versions should be detected and used). Any ideas? Thanks, Thomas
