old obfu tricks revealed

[Felonious]

I don't seem to have the time these days to do really involved obfus like I used to, but I thought it would be fun to pass along some of the neat things I've learned while writing them. Tonight's tricks illustrate some oddities you can take advantage of when naming subs, so here's a short illustration:

# a little tricky
print "\ntrick one\n";
sub ::0 { print "::0 called\n" }
%42->();

# a little more tricky, think about implicit return values
print "\ntrick two\n";
sub :: { print ":: called\n" }
sub ::1{ print "::1 called\n" }
::->();

# and think about evaluation order
print "\ntrick three\n";
sub _ {print "_ called\n"}
_ _ -> ();

# another trick that eliminates white space in sub declarations
print "\ntrick four\n";
sub'x{print "x called\n"}
x;

Take special note of the '::' function, it has an odd characteristic in that it doesn't get shown by Deparse in most perl versions (or any?). That makes it a great place to hide stuff that might become too obvious under Deparse. Now, a very simple Japh that demonstrates these all use toghther:

sub _ { print "perl "; ::->() }
sub :: { print "hacker\n"}
sub ::1 { _ print  "another "}
sub ::0 { print "just " }
1->(%1->());

Of course, you have to go to greater lengths in the sub bodies than I did in the above for the output to be suprising to anyone, but I think the '1->(%1->())' line makes it suffuciently interesting to figure out.

Finally, sometimes messing with the sysbol table in this manner exposes even deeper mysteries to exploit...
Contemplate why the below causes an endless loop just by re-ordering two of the subs...
Note: May not behave the same on all perl versions. Please post your version/platform if the below does not recurse?

sub :: { print "hacker\n"}
sub _ { print "perl "; ::->() }
sub ::1 { _ print  "another "}
sub ::0 { print "just " }
1->(%1->());

Try deparsing the two versions and observing the difference.

Well, I hope this has been interesting and possibly leads some others to investigate oddities in how the symbol table behaves. If there's interest, I may post a few other interesting tricks I picked up another time.

[Felonious]

Ok, tonight's trick simply illustrates the principle of replacing the standard '/' separator in substitutions with other characters. It's a simple and well known trick, but can be used pretty effectively in some instances. Here goes:

$_ = "Sello World\n";

# think about alternative separators
s'S'H's;
print;

# Deparse and compare the difference in the next two
s=>o=>O=>o;
print;

s,o,O,o;
print;

# and now, for the mystery...
sub s { ("x","y","z"); }
s()[0];
print;

The 'mystery' has nothing to do with mixing separators or that you can use 4 instead of three, that's pretty simple stuff... it's why it replaces anything at all, much less the character it does. Happy obfu-writing!

[jdalbec]

sub _ { print "perl "; ::->() }
sub :: { print "hacker\n"}
sub ::1 { _ print  "another "}
sub ::0 { print "just " }
1->(%1->());

<Reveal this spoiler or all spoilers in this node or all in this thread>

sub :: { print "hacker\n"}
sub _ { print "perl "; ::->() }
sub ::1 { _ print  "another "}
sub ::0 { print "just " }
1->(%1->());

<Reveal this spoiler>