in reply to Re: Parse a list of path strings into a nested hash
in thread Parse a list of path strings into a nested hash

Can you please briefly explain paths2TreeGF to us, especially the tricky bit? It looks slick. I just can't get my head around it.

$scan = $scan->{shift @parts} ||= {} while @parts;

Thanks.

Replies are listed 'Best First'.
Re^3: Parse a list of path strings into a nested hash
by GrandFather (Saint) on Sep 06, 2010 at 02:32 UTC

    Sure. For each path split the path into parts:

    my @parts = split /\\/, $path;

    Then for each part in the path starting at the 'root' (drive specifier for example) add it to the parent hash. %pTree is the parent hash for the whole tree so we start out by setting $scan to reference that:

    my $scan = \%pTree;

    then the 'tricky' bit adds the parts. shift @parts removes the left most part of the path. $scan->{shift @parts} ||= {} will create a new hash element if there isn't one already. In any case it returns the hash ref for the current part of the path and that becomes the new parent $scan = ... while @parts;. Taken all together the 'tricky' bit walks down the tree generating entries as required:

    $scan = $scan->{shift @parts} ||= {} while @parts;

    A few important tricks:

    1. $scan->{shift @parts} ||= {} autovivifies (creates) an entry in the hash if it doesn't exist yet.
    2. ||= {} only assigns a new (empty) hash if a new entry has been created (or in the nasty special case that the directory name is 0).
    3. assignment operators (=, ||=, +=, ...) 'return' the value assigned which is how $scan gets updated.

    Note that with Perl 5.10 and later you can use //= in place of ||= and it fixes the 0 issue mentioned in item 2.

    True laziness is hard work