deal with just plain busted fec filings, e.g. 897015

[jsfenfen]

Something about how this filing broke
http://query.nictusa.com/cgi-bin/dcdev/forms/C/897015/
threw sand into the gears. The filing seems to be effectively removed from the site now.. need some type of check to make sure this doesn't happen again.

[jsfenfen]

Hmm, it seems like at the moment it loaded a 'page not found' http page, instead of an error. Maybe this is a secondary change related to the move? May be able to get this via filtering response headers, or just checking if it's an error page. I can't seem to reproduce this behaviour now; the download link:
query.nictusa.com/dcdev/posted/897015.fec
Just doesn't load.

[jsfenfen]

This may be a temporary condition. Anecdotal reports are that during high load times FEC servers can fail to retrieve filings that are later available. There's a potential that we're marking filings as missing when they are available later. This wasn't a problem when the FEC returned a 404 error, but it seems like it's now returning a 200-level response that says effectively the file can't be found. We really need to parse these responses and--if not downloadabe--mark them as not downloaded, and retry them. Some--maybe most of them--may never be downloadable.

I'm unclear if there are weird consequences to downloading a filing that's substantially outta order. There shouldn't be, but it's not something that's really been tested much.