map{push@L,@C+0;push@C,lc=~/./g}<>;sub p{pop@n}sub w{push@n,@_}while($
+n<@C){$_=
$C[$n++];if(/h/){$_=p;w$n[-1-abs];$_>0&&splice@n,-2-$_,1}if(/n/){w 0;/
+-/||w$_+7
*p while($_=index"htaoinse",$C[$n++])<7}$0=p,$_=p,w int$_/$0,$_%$0if/e
+/;if(/t/)
{$_=p;$_?$n=$L[$_-1]:last if p}/a/?w 1+grep$n>$_,@L:/o/?print chr p:/s
+/?w-(p)+p
:/i/&&w($_=getc)?ord:-1}
We can s/last if p/p&&last/ and that brings it own to 338 bytes:
map{push@L,@C+0;push@C,lc=~/./g}<>;sub p{pop@n}sub w{push@n,@_}while($
+n<@C){$_=
$C[$n++];if(/h/){$_=p;w$n[-1-abs];$_>0&&splice@n,-2-$_,1}if(/n/){w 0;/
+-/||w$_+7
*p while($_=index"htaoinse",$C[$n++])<7}$0=p,$_=p,w int$_/$0,$_%$0if/e
+/;if(/t/)
{$_=p;$_?$n=$L[$_-1]:p&&last}/a/?w 1+grep$n>$_,@L:/o/?print chr p:/s/?
+w-(p)+p:
/i/&&w($_=getc)?ord:-1}
But wait, there's more! change $0=p,$_=p,w int$_/$0,$_%$0 to w int($_=p)/($0=p),$_%$0}
Which gives 337 (I think) bytes:
map{push@L,@C+0;push@C,lc=~/./g}<>;sub p{pop@n}sub w{push@n,@_}while($
+n<@C){$_=
$C[$n++];if(/h/){$_=p;w$n[-1-abs];$_>0&&splice@n,-2-$_,1}if(/n/){w 0;/
+-/||w$_+7
*p while($_=index"htaoinse",$C[$n++])<7}w int($_=p)/($0=p),$_%$0if/e/;
+if(/t/)
{$_=p;$_?$n=$L[$_-1]:p&&last}/a/?w 1+grep$n>$_,@L:/o/?print chr p:/s/?
+w-(p)+p:
/i/&&w($_=getc)?ord:-1}
A ternary operator can be replaced by a && and the plus in w+($_=getc)?ord:-1 trick is not needed, it works without (or, at least, hello.eta still outputs correctly).
I have an idea it must be possible to replace the w$_+7*p while($_=index"htaoinse",$C[$n++])<7 section with a redo and/or a pos, but it is so
difficult to understand what the program does I am
beginning to wonder whether it wouldn't be easier to look
at the instruction set and come up with another angle of attack. For instance, it would be nice to be able to do away with the @L array as well. Replace @C and @L by a hash that is keyed on byte offset and the value is the byte.
print@_{sort keys %_},$/if%_=split//,'= & *a?b:e\f/h^h!j+n,o@o;r$s-t%t#u' |