If you want to trundle through an arbitrarily deep reference tree with grep-like syntax, it's actually fairly easy to implement:

    use 5.016; # For __SUB__, or pre-declare $grep and use $grep->() s
+yntax

    sub ref_grep(&$) {
        my ($wanted, $ref) = @_;
        my $seen = {}; # Refs we've already seen (avoid circular refs)
        my @found;     # Result list
        
        my $grep = sub {
            my ($ref, $path) = @_;
            return if ref $ref and $seen->{$ref}++;
            { $_ = $ref; push @found, $path if $wanted->() }
            for (ref $ref) {
                when ('HASH') {
                    __SUB__->($ref->{$_}, $path."->{$_}") for sort key
+s $ref;
                }
                when ('ARRAY') {
                    __SUB__->($ref->[$_], $path."->[$_]") for sort key
+s $ref;
                }
            }
        };
        $grep->($ref, '');
        return @found;
    }
[download]

Invocation is just like grep:

It's not limited to scalars, either; the BLOCK can perform any test it likes.

Of course, this is not without limitations. Support for nested objects, tests based on the path itself, pruning, etc., would take more work, and I'd certainly not try much harder to re-invent this wheel. What I'm saying is, take this as a starting point.

Thanks for that! I was kinda stuck on grep, and grep needing a list as an argument, but I certainly get the idea.

I wonder if this is an XY Problem?

Can you explain what you are going to do with this *location* once you've got it?

That's a fair point. I'm doing a text search through a book. I've parsed the book into chapters and paragraph numbers, so that when I find the location I can say "text found at paragraph 10 in chapter 2".

Then you should probably invert your data structure. Instead of

my $hash = { base => { key => 'value', key2 => 'value2'}};
[download]

consider

my $hash = {
  "value"   => [
    { 'chapter' => 2, 'paragraph' => 10 },
    { 'chapter' => 2, 'paragraph' => 11 },
   ...
  ],
  "value2" => [ 
    { 'chapter' => 3, 'paragraph' => 1 }
  ]
};
[download]

-derby

I agree with derby. Your data structure should explicitly model the whole book, including the text, the chapter numbers, and the paragraph numbers. It's better not to infer chapter numbers and paragraph numbers implicitly from their positions within a Perl nested data structure. A well-designed data structure will be extensible. If you decide later you want to model another attribute of the book—say, sentence numbers—you can do so without having to re-implement the data structure or create a new one. The data structure will then also map neatly to an external, reusable representation such as XML.

But what I really want is like the location of $VAR1=>{'base'}{'key2'}. How can I get that (elegantly)?

That really does not make much sense to me. Location? With your data, you could presumably say that the location of 'value2' is $hash{'base'}{'key2'}, because 'value2' is stored there. But the location %VAR1 being {'base'}{'key2'}, this I don't understand.

Sorry, like I said I wasn't even sure how to phrase the question. The first thing you said is what I'm after.