The code involving imp is equivalent to "from ndiff import *", and ndiff
imports StringIO.StringIO as StringIO, which meant that while ndifftest
intended the name to be the module, it was really the class. Just moved
the import later to be sure the name was the one we intend.
http://seclists.org/nmap-dev/2014/q2/318
Essentially, we import the xml name, then override its search path,
stripping out the _xmlplus paths that PyXML uses. This leaves only the
Python 2 standard library path, which is what Zenmap was written for.
Issues fixed:
1 E111 indentation is not a multiple of four
1 E201 whitespace after '['
14 E251 no spaces around keyword / parameter equals
7 E301 expected 1 blank line, found 0
55 E302 expected 2 blank lines, found 1
69 E501 line too long (80 characters)
3 W291 trailing whitespace
4 W601 .has_key() is deprecated, use 'in'
if there was only one extraports state, would create a Port object for
each scanned port that was not listed explictly. So for example, if a
scan of 100 ports had 5 open ports and an extraports of "filtered",
Ndiff would create 95 port records in the "filtered" state in addition
to the 5 "open". If there was more than one extraports state this was
not possible.
This turned out to be a slow operation, out of all proportion to its
utility. Here are times for a diff of random-1.xml and random-2.xml,
before and after:
real 0m11.912s
user 0m10.970s
sys 0m0.249s
real 0m0.773s
user 0m0.726s
sys 0m0.046s
In most cases extraports are not shown in the output, so this was wasted
effort. I know of only one place where it affected the output, when an
extraports in the A scan because a non-extraports in a different state
in the B scan. Then the previous state would be included in the <a>
port, where now it will just be listed without a state.
was that if a host had multiple addresses of the same type, they should
be sorted numerically, not alphabetically. So 20.0.0.1 should come
before 100.0.0.1. But the way Nmap works, I don't think a host can have
more than one address of the same type, and it's not worth the code it
would take to make them sort properly if it did.
from
21/tcp is open, was filtered.
23/tcp is open, was filtered.
80/tcp is open, was filtered.
8701/tcp is filtered, was open.
to
-21/tcp filtered
+21/tcp open ftp Netgear broadband router ftpd 1.0
-23/tcp filtered
+23/tcp open telnet Netgear broadband router admin telnetd
-80/tcp filtered
+80/tcp open http Embedded Allegro RomPager webserver 4.07 UPnP/1.0 (ZyXEL ZyWALL 2)
-8701/tcp open unknown
+8701/tcp filtered
the DiffHunk class with a type tag. Now output is handled with polymorphism
rather than dispatching with if/else. It also better shows what members each
hunk type has.
happens when a host was scanned in the A scan but wasn't scanned in the B scan.
I previously had it ignore such changes using the logic that the diff should be
like scan aggregation: no new information means no state change. But I think
it's more useful to see those changes in which hosts were scanned.
This is analogous to r10263, which did the same thing for port state changes.
