Use of nmap.fetchfile in scripts

[Daniel Miller <bonsaiviking () gmail com>]

List,

Today on IRC, a user had a problem with dns-brute.nse. The 
dns-brute.hostlist script-arg would not accept a relative filename like 
so: `--script-args dns-brute.hostlist=./list.txt` I provided him a small 
patch (http://pastebin.com/QHMavbKa) that checked for a file at the 
relative or absolute path after first checking the locations that 
nmap.fetchfile returns.

dns-brute is not the only script that does this. Lots of scripts are 
using the pattern of passing script-args to nmap.fetchfile, when the 
function really makes better sense to retrieve a static filename 
relative to the Nmap search paths. On the other hand, the unpwdb library 
does what I would expect: If the userdb script-arg is provided, it uses 
the value literally as a path, otherwise it uses nmap.fetchfile to grab 
"nselib/data/usernames.lst"

My question is: Should this behavior be standardized? I have no problem 
doing the grunt work, but I'd like the community's consensus on these rules:

1. Filenames from script-args should not be passed to nmap.fetchfile, 
but instead directly to io.open.
2. When necessary, to preserve backwards-compatibility, the script-arg 
may be passed to nmap.fetchfile, but if the return value is nil, the 
script-arg should be used verbatim.

This means that new scripts should use this idiom:

filename = stdnse.get_script_args("file") or 
nmap.fetchfile("default.filename")

and older scripts can be brought into conformity by using this formula:

arg = stdnse.get_script_args("file")
filename = (arg and (nmap.fetchfile(arg) or arg) or 
nmap.fetchfile("default.filename")

Thoughts?

Dan
_______________________________________________
Sent through the dev mailing list
http://nmap.org/mailman/listinfo/dev
Archived at http://seclists.org/nmap-dev/