Re: [NSE] Microsoft SQL Server (MSSQL) library and scripts

[Fyodor <fyodor () insecure org>]

On Sun, Mar 28, 2010 at 07:54:59PM -0500, Kris Katterjohn wrote:
> OK, I guess I'll be the first to actually argue :)

Well I think I agree with your "argument" :).  I'm not suggesting we
go wholesale combining scripts with different purposes just because
they talk to the same service.  My suggestion relates to scripts which:
 o Talk to the same service
 o Their whole function is to query that service for information and
   report it.
 o Belong to (or can be made to properly belong to) the same
   categories

> I have
> manually removed scripts which shared categories with previous ones in the
> list (do correct my list if I made a mistake!).

In so doing, you have removed all of the scripts I would want to
combine :).

> I'm not sure how to go about sorting through this mess to combine the scripts
> into one.

Yeah, I don't want to combine any of those.  The groups I think are
POSSIBLE CANDIDATES for combining are:

citrix-enum-apps
citrix-enum-servers

citrix-enum-apps-xml
citrix-enum-servers-xml

couchdb-databases
couchdb-stats

db2-das-info (side note: it would be great if we could get a
db2-info      @usage and @output for the NSEdoc for these two)

dns-random-srcport
dns-random-txid

mysql-databases
mysql-info
mysql-users
mysql-variables

nfs-acls
nfs-dirlist
nfs-showmount
nfs-statfs

smb-enum-domains
smb-enum-groups
smb-enum-processes
smb-enum-sessions
smb-enum-shares
smb-enum-users
smb-os-discovery
smb-security-mode
smb-server-stats
smb-system-info

snmp-interfaces
snmp-netstat
snmp-processes
snmp-sysdescr
snmp-win32-services
snmp-win32-shares
snmp-win32-software
snmp-win32-users

Again, I'm not suggesting that each of these groups should be combined
into one script.  There are probably good reasons for keeping some of
them separate.  But I do think that some of these groups could be
reduced.  For an example of a script which does numerous queries
itself rather than splitting into one script per query type, take a
look at irc-info.

I see the benefits of the combination as:

o Easier to maintain one script than a bunch of very similar
  ones. Even with libraries, similar scripts can share a lot of code for
  output, control flow, etc.

o Easier for users to specify "--script smb-info" or even "--script
  smb-info --script-args smb-info=all" than the likes of:

  --script smb-enum-domains,smb-enum-groups,smb-enum-processes,\
    smb-enum-sessions,smb-enum-shares,smb-enum-users,\
    smb-os-discovery,smb-security-mode,smb-server-stats,\
    smb-system-info

  Even if you use wildcards the above will be pretty long. (smb-* may
  get much more than you want).

o A single -info script can be augmented with new stuff later, while
  adding a new query script will require the users to learn about it and
  add it to their command-line (though categories and wildcards can
  help here).

o Output can be combined more elegantly and intelligently if it is all
  in one script. If you break them up, the output has to be all
  separated by script.

o Easier for users to browse the lists of scripts (e.g. in the book or
  the directory or http://nmap.org/nsedoc/) if we don't have a bunch of
  similar scripts for each service.

But I also see advantages to individual scripts, particularly in cases
where:

o Users typically just want one particular piece of information. So
  snmp-sysdescr.nse might be best off on its own (even if an snmp-info
  might duplicate that functionality).

o Scripts which should be in different categories for one reason or
  another.  For example, we currently have mysql-info in default and safe
  because it just parses the banner it gets back, while mysql-databases
  must authenticate (even if using a blank password) and thus we
  consider it intrusive and non-default.

o Scripts with very different implementation

o Scripts which retrieve information which is so different that there
  isn't likely a big correlation between users wanting the output from
  one script and that from another.

I hope that makes sense.

> Only combine them if the categories are similar?  I don't know how
> you can have a file with mixed categories, or how you could possibly call it
> from --script.

Right.  Scripts which are different enough to require different
categories are probably different enough to remain separate.

> Please note that I'm not saying don't combine any scripts!

And I'm not saying we should combine all the scripts :).

Cheers,
Fyodor
_______________________________________________
Sent through the nmap-dev mailing list
http://cgi.insecure.org/mailman/listinfo/nmap-dev
Archived at http://seclists.org/nmap-dev/