#11 medline import garbles url: two bugs?

[Jan Warnking]

This look to me like two independent bugs, actually.
In a medline research (Brain Res[ta] AND 830[vi] AND
1[ip] AND 43[pg]), I came across the following URL:

URLF-
http://www.elsevier.com:80/cgi-bin/.../cas_sub/browse/brow
se.cgi?year=1999&volume=830&issue=1&aid=15476

The linebreak is really present in the medline output.
I left out ("...") parts I believe to be irrelevant, to
fit this into the limited line width of sourceforge's
message system. This is what I got in my database:

http://www.elsevier.com:80/cgi-bin/.../cas\_sub/browse/brow
se.cgi?year=1999\&volume=830\&issue=1\&aid=15476

Actually, this isn't how this is displayed in sixpack.
Just after importing, there is a blank at the point of
the linebreak. After saving to bibtex and reloading,
there is a funny sort of "J" with a bar over it. In the
bibtex database file, there is a linebreak at this
point.

Note also the "_", which gets converted to a "\_".

None of the other URLs I came across (and which work
fine), contain underscores or linebreaks. I assume
these two problems to be independent.

This problem occurred indifferently when importing the
citation via wget or when saving the reference to a
medline format text file (.fcgi) from the web page and
importing this file into sixpack.

This occurred using:

sparc / solaris 2.7
bib: CVS 2001/07/02
perl: 5.005_3
perl-tk: 800.023

[Michael Lachmann]

Logged In: YES
user_id=120828

Hmmm.. this is somewhat complicated. The bug is in the bp library.
bp needs to know not to try to convert the characters of the url field into
tex chars. And it needs to handle urls that continue over one line better.
Currently, the medline returns also with '&' instead of '&', so that needs
to be treated, also. I did a quick fix, but the problem isn't solved yet.