No such thing as a small change PerlMonks

### Re: Golf: Tree searching

by japhy (Canon)
 on Apr 19, 2001 at 22:17 UTC ( #73917=note: print w/replies, xml ) Need Help??

in reply to Golf: Tree searching

Here's code in your vein (72 chars not counting 'sub '):
```sub f{(\$s,\$t)=@_;while(\$t->{d}ne\$s){\$t=\$t->{(r,l)[\$s lt\$t->{d}]}||retu
+rn}\$t}
Here's 63 chars:
```sub f{(\$^,*_)=@_;*_=\$_{(r,l)[\$^lt\$_{d}]}||return while\$_{d}ne\$^;*_}

japhy -- Perl and Regex Hacker

Replies are listed 'Best First'.
Re: Re: Golf: Tree searching
by twerq (Deacon) on Apr 19, 2001 at 22:48 UTC
Here's 58 chars with a completely different approach. .

sub f{(\$s,\$t)=@_;\$\$t{d}eq\$s?\$t:f(\$s,\$\$t{l})||f(\$s,\$\$t{r})}

--twerq
As koolade pointed out, not all of these work. Yours in particular goes into deep recursion. Works in theory is not good enough, in golf you are going to learn a lot about the assumptions you cannot make. And yours has at least 2 separate (and serious) mistakes.

BTW the counts as I list them are just for the body of the sub, so if yours did work it would be 51 characters.

Playing independently I came in with several solutions that are small, and the following which is similar to yours really does work at 58:

```sub f {
my(\$s,\$t)=@_;\$t?\$\$t{d}eq\$s?\$t:f(\$s,\$\$t{l})||f(\$s,\$\$t{r}):0
}
And here, thanks to koolade, is the test that I use:
```\$t = {
d => 'd',
l => {
d => 'b',
l => { d => 'a', l => 0, r => 0, },
r => { d => 'c', l => 0, r => 0, },
},
r => {
d => 'f',
l => { d => 'e', l => 0, r => 0, },
r => { d => 'g', l => 0, r => 0, },
}
};

sub test {
my \$val = f(@_);
print \$val ? "\$val->{d}:\$val\n" : "\$val\n";
}

test('e',\$t);
test('O',\$t);
Both tilly and indigo have exactly the same result, which is a fix of twerq's including 1) my, and 2) the test for \$t which prevents infinite looping.

Here is a solution which is short, but not 'strict' compliant: sub f{for((\$s,\$t)=@_;\$_=\$\$t{(0,l,r)[\$\$t{d}cmp\$s]};\$t=\$_){}\$t}
Alas, although 4 characters lighter, it doesn't return 0 on misses properly. This does, but is admittedly 3 characters heavy: sub f{for((\$s,\$t)=@_;\$_=\$t&&(0,l,r)[\$\$t{d}cmp\$s];){\$t=\$\$t{\$_}||0}\$t}
Or, a tie, provided the tree has 0-value stubs as it does in other examples: sub f{for((\$s,\$t)=@_;\$_=\$t&&(0,l,r)[\$\$t{d}cmp\$s];){\$t=\$\$t{\$_}}\$t}
Fun test code below for a tree:
```    \$|++;
\$table = {
d => 'h',
l => {
d => 'd',
l => { d => 'b', l => { d => 'a' }, r => { d => 'c' } },
r => { d => 'f', l => { d => 'e' }, r => { d => 'g' } },
},
r => {
d => 'l',
l => { d => 'j', l => { d => 'i' }, r => { d => 'k' } },
r => { d => 'm', l => { d => 'l' }, r => { d => 'n' } },
}
};

foreach \$y (qw[ a b c d e f g h i j k l m n o p q ])
{
\$x = f(\$y,\$table);
print "Answer for \$y = ";
print \$x," ", \$\$x{d},"\n";
}
value entry in the hash.

Create A New User
Node Status?
node history
Node Type: note [id://73917]
help
Chatterbox?
 [LanX]: hm just read an article on heise.de claiming that RPerl is 250 times faster. ... it's that even possible? [marto]: under certain circumstances perhaps [RonW]: as I understand it, RPerl "compiles" a subset of Perl to C, so I would expect the final executable to run a lot faster

How do I use this? | Other CB clients
Other Users?
Others chanting in the Monastery: (11)
As of 2017-05-22 20:13 GMT
Sections?
Information?
Find Nodes?
Leftovers?
Voting Booth?
My favorite model of computation is ...

Results (174 votes). Check out past polls.