On 05/12/2011 13:47, franck.eyr...@jrc.ec.europa.eu wrote:
(I could try to contribute, if you think it would require few changes,
but I didn't work with c/c++ for a long time)
I anyway had a curious look at the source code; I noticed at line 349 of
MirrorJob.cc the following code :
if(target_is_local && !script_only)
{
if(lstat(target_name,&st)!=-1)
{
if(S_ISDIR(st.st_mode))
{
chmod(target_name,st.st_mode|0700);
create_target_dir=false;
}
else
{
Report(_("Removing old local file `%s'"),
dir_file(target_relative_dir,file->name));
if(remove(target_name)==-1)
{
eprintf("mirror: remove(%s):
%s\n",target_name,strerror(errno));
goto skip;
}
create_target_dir=true;
}
}
}
The S_ISDIR at line 353 test doesn't take into acount the case where the
target file is a link and the links points to a directory, which would
suits more my case. In fact, since the whole code above is executed only
when !script_only, I realized that running mirror --script=script.lft
and then lftp -c script.lftp, it mirrors correctly (but I'm loosing the
possibility of parallel downloads). So lftp is able to follow the links,
it is just that is decides a priori to delete them.
However, I'm not quite sure yet how to handle the "target is link and
link points to a directory" test in C++, and don't have time now.
How that helps,
Franck
_______________________________________________
lftp mailing list
lftp@uniyar.ac.ru
http://univ.uniyar.ac.ru/mailman/listinfo/lftp
