Hi thanos,

To clarify, the structures I am trying to parse are generated by TreeDumper with $Data::TreeDumper::Displayaddress = 0.

Here is a quick and dirty function I have whipped up which suffices for my possible inputs, but is not quite correct for all possible inputs. Specifically, it will interpret arrays as hashes with keys 0 .. n.

#!/usr/bin/env perl

use strict;
use warnings FATAL => 'all';
use feature qw(say);

use Test::Deep qw(eq_deeply);

sub parse_treedump {
    my $lines = shift;
    my @lines = @$lines;

    my %obj;
    my @stack;

    my $parse_next_line = sub {
        my $line = shift @lines;
        my $key_indicator = $line =~ /\|-/ ? '|-' : '`-';

        my $key_idx = index( $line, $key_indicator );

        my $level = @stack;
        my $key_level = $key_idx / 3;

        if ($key_level < $level) {
            pop @stack until $key_level == ($level = @stack);
        }

        my $current_obj = \%obj;
        $current_obj = $current_obj->{$_} foreach @stack;

        if ($line =~ /(\w+)\s=\s(.*)$/ && $key_level == $level) {
            $current_obj->{$1} = $2 eq 'undef' ? undef : $2;
        }

        elsif ($line =~ /(\w+)\s\(no\selements\)/) {
            $current_obj->{$1} = [ ];
        }

        elsif ($line =~ /(\w+)\s=$/) {
            $current_obj->{$1} = '';
        }

        elsif ($line =~ /(\w+)/ && $key_level == $level) {
            push @stack, $1;
            $current_obj->{$1} = { };
        }
    };

    $parse_next_line->() while @lines;

    return \%obj
}

my $expected_hash = {
    one      => 'two',
    three    => {
        four   => 'five',
        six    => 'seven',
        eight  => {
            nine => 'ten'
        },
        eleven => [ ],
        twelve => undef,
    },
    thirteen => 'fourteen'
};

my $lines = [];
while (<DATA>) {
    chomp;
    push @$lines, $_;
}
close DATA;
my $output_hash = parse_treedump($lines);

if (eq_deeply($output_hash, $expected_hash)) {
    say "OK";
}

1;

__DATA__
|- one = two
|- three
|  |- four = five
|  |- six = seven
|  |- eight
|  |  `- nine = ten
|  |- eleven (no elements)
|  `- twelve = undef
`- thirteen = fourteen
[download]

As you can see, the TreeDump output in the __DATA__ section gets converted to the expected hash. I would prefer it to work for all possible inputs...

Best,
Jim

Here's a crazy idea. Apply some simple transforms to the the TreeDump output:

'|- ' => '',
'|  ' => '',
'`-'  => '',
'='   => ':',
'undef' => 'null',
[download]

You may need some more, but after that the text will be in YAML format, so you can use one of the YAML parsers to convert it to a hash.

    my $parse_next_line = sub {
        my $line = shift @lines;
        ...
    };
    $parse_next_line->() while @lines;
[download]

This is very silly. Just write

    foreach my $line (@lines) {
        ...
    }
[download]

Good point. I started off trying to build an iterator, but eventually settled into this pseudo foreach loop as you have pointed out. Thanks for the input!

I kind of like this solution, but YMMV. The approach isn't much different from yours.

my @stack = ({});
while (<DATA>) {
   s/\s+$//; # remove trailing space and newline
   my $depth = 0;
   $depth++ while $depth < @stack && s/^[|` ][- ] //;
   die 'malformed input' if $depth < 1;
   splice @stack, $depth;
   my $obj = $stack[-1];
   if (/^(.*?) = ?(.*)/) {
      $obj->{$1} = $2 eq 'undef' ? undef : $2;
   }
   elsif (s/ \(no elements\)$//) {
      $obj->{$_} = [];
   }
   else {
      push @stack, $obj->{$_} = {};
   }
}
my $output_hash = $stack[0];
[download]