Re: Suggestion: Using script output - Human-readable and machine-parseable

[Daniel Miller <bonsaiviking () gmail com>]

On Mon, Apr 18, 2011 at 11:36 PM, David Fifield <david () bamsoftware com>wrote:

> Do we really need to link with libyaml? It seems like emitting YAML, as
> opposed to parsing it, would be easy enough to implement in stdnse.
>
> David Fifield

No, we really don't, especially if we want to have an intermediate form (C++
data structure) to hand off to output.cc for XML output. I've been working
on this angle (scrapped the YAML for now), and I have a decent first go at
it. I've been using stdnse.format_output as a model, since script authors
should be used to that, and it means the least changes to existing scripts.
Here's example output:

Normal:
Pre-scan script results:
| test:
|
|     Apple pie
|   DOMAINS
|         Domain 1
|         Domain 2
|   NAMES (WARNING: Not all names could be determined!)
|_        Name 1

XML:
<prescript><script id="test"><container name=""><elem>Apple
pie</elem><container name="DOMAINS"><elem>Domain 1</elem><elem>Domain
2</elem></container><container name="NAMES" warning="Not all names could be
determined!"><elem>Name
1</elem></container></container></script></prescript>

I'm currently in debug mode, and having a rough time. I can't get that blank
first line to go away. I know exactly which function is generating it, but
if I change the function to return a blank string instead of "\n", it
mangles the rest of the text output in a way that doesn't make any sense.
I'm also fighting with a function that returns different values based on if
there is a comment in it.

I'm attaching a patch with my current progress. I'd appreciate any input you
have at this point, but I realize it's not robust code yet.

Dan

Attachment: script-output.diff
Description:

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