Re: What does !$saw{$

Let's split it up a bit:

You have a hash, %saw. Individual elements of a hash are accessed via $hash{key}. If you append the postfix ++ operator to it, it looks like $saw{key}++, which increments the hash element by one.

Now, what does that do?

The grep iterates over the whole array @in, sets $_ to each element and then executes the code, in our case the expression !$saw{$_}++. If the expression returns a true value, grep keeps the array element in its result, otherwise it's discarded.

Now, when a key in %saw does not already exists, the code sets $saw{key} to 1 (incrementing by one from undef), and then returns the negation of the previous hash value (undef is false, so it returns true). So, if the hash key did not yet exist in the hash, the array element is put into @out.

The other case is that the hash key already exists in the hash. Then $saw{$_} returns a number greater than zero, which is interpreted as true, and the negation of that is false, so the (duplicate) array element in @in is discarded.

This method is a nice and easy way (once you understand it) to get a unique list of elements in an array while retaining the order. There are other methods, like using the keys of %saw:

undef %saw;
$saw{$_}++
  for @in;
@out = keys %saw;
[download]

This code puts the same elements into @out, but you lose the order.

I don't remember if it was quicker, but there also is the non-looped version:

undef %saw;
@saw{@in} = (1) x @in;
@out = keys %saw;
[download]

Comment on Re: What does !$saw{$_}++ means Select or Download Code

Replies are listed 'Best First'.
Re^2: What does !$saw{$_}++ means by ysth (Canon) on Jan 26, 2005 at 09:45 UTC
returns the negation of the previous hash value (undef is false, so it returns true) Nit: post-increment returns 0 if the value was undef, not undef. This is occasionally criticised, defended, pointed out as inconsistent with post-decrement (which does return undef), and then dropped. A comment in pp.c refers one to http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2003-03/msg00536.html for further info (where you can read the criticism, defense, etc.).	[reply]
Re^2: What does !$saw{$_}++ means by Eimi Metamorphoumai (Deacon) on Jan 26, 2005 at 13:36 UTC
I don't remember if it was quicker, but there also is the non-looped version: `undef %saw; @saw{@in} = (1) x @in; @out = keys %saw;` [download] I haven't benchmarked it, but I've frequently heard it asserted that even faster is `undef %saw; @saw{@in} = (); @out = keys %saw;` [download] That is, using undef as your values (so you don't need the increment or to create the list of ones).	[reply] [d/l] [select]
Re^2: What does !$saw{$_}++ means by ikegami (Patriarch) on Jan 26, 2005 at 17:06 UTC
but there also is the non-looped version: Except that `x` is a loop operator, just like `map`. In fact, it not only loops for the number of duplications, it also implicitely loops over each element of the list on the LHS. `@saw{@in}` and `keys(%saw)` are also loops, but they are implicit unlike `x`. In fact, you can remove two of the four loops of your "non-looped version": `undef %saw; undef(@saw{@in}); @out = keys(%saw);` [download] Update: `undef %saw` is also an implicit loop, on calls other than the first.	[reply] [d/l] [select]


Do you know where your variables are?
	PerlMonks

Re: What does !$saw{$_}++ means