OFX download problems

[Jon Hamkins]

I have been putting together a set of perl scripts, using
WWW::Mechanize, to negotiate financial institution web sites to download
OFX files and monthly statements.  WWW::Mechanize is great at automating
web-browsing, but unfortunately, it doesn't do javascript, which means I
have had to painfully manually reimplement any javascript these sites
use (setting cookies, redirections, hidden fields, etc.).

Happily, I happened upon the ofx.py and ofx-ba.py scripts at
http://www.jongsma.org/gc/ which connect directly to the OFX servers,
along with the list of OFX servers.  Thanks!  In principle, this should
be enough for me to give up my own scripts.

Except, I can't get ofx.py to work.  

1. For Discover, even after changing the APPID and APPVER with

            config["appid"] = "Money"
            config["appver"] = "0900"

the OFX file, which arrives with no error messages in it, has all
boilerplate and no actual transactions.  Any ideas?  Unfortunately, I
don't grok python.  The script like it is requesting 31 days worth of
transactions.

2.  For Bank of America, there is no OFX server site listed in 5043.xml.
I haven't managed to get that site working yet with my perl scripts
either.  Has anyone automated OFX download from B of A?

Any help is appreciated!

     ----Jon