“tree” command output with “pure” (7-bit) ASCII output

Only for tree 1.6 and above

You might want to look at:

man tree

--du For each directory report its size as the accumulation of sizes of all its files and sub-directories (and their files, and so on). The total amount of used space is also given in the final report (like the 'du -c' command.) This option requires tree to read the entire directory tree before emitting it, see BUGS AND NOTES below. Implies -s.

So you should use:

tree --du -h

Attempt 1

A solution using just perl, returning a simple hash of hashes structure. Before the OP clarified data format of JSON.

#! /usr/bin/perl

use File::Find;
use JSON;

use strict;
use warnings;

my $dirs={};
my $encoder = JSON->new->ascii->pretty;

find({wanted => \&process_dir, no_chdir => 1 }, ".");
print $encoder->encode($dirs);

sub process_dir {
    return if !-d $File::Find::name;
    my $ref=\%$dirs;
    for(split(/\//, $File::Find::name)) {
        $ref->{$_} = {} if(!exists $ref->{$_});
        $ref = $ref->{$_};
    }
}

File::Find module works in a similar way to the unix find command. The JSON module takes perl variables and converts them into JSON.

find({wanted => \&process_dir, no_chdir => 1 }, ".");

Will iterate down the file structure from the present working directory calling the subroutine process_dir for each file/directory under ".", and the no_chdir tell perl not to issue a chdir() for each directory it finds.

process_dir returns if the present examined file is not a directory:

return if !-d $File::Find::name;

We then grab a reference of the existing hash %$dirs into $ref, split the file path around / and loop with for adding a new hash key for each path.

Making a directory structure like slm did:

mkdir -p dir{1..5}/dir{A,B}/subdir{1..3}

The output is:

{
   "." : {
      "dir3" : {
         "dirA" : {
            "subdir2" : {},
            "subdir3" : {},
            "subdir1" : {}
         },
         "dirB" : {
            "subdir2" : {},
            "subdir3" : {},
            "subdir1" : {}
         }
      },
      "dir2" : {
         "dirA" : {
            "subdir2" : {},
            "subdir3" : {},
            "subdir1" : {}
         },
         "dirB" : {
            "subdir2" : {},
            "subdir3" : {},
            "subdir1" : {}
         }
      },
      "dir5" : {
         "dirA" : {
            "subdir2" : {},
            "subdir3" : {},
            "subdir1" : {}
         },
         "dirB" : {
            "subdir2" : {},
            "subdir3" : {},
            "subdir1" : {}
         }
      },
      "dir1" : {
         "dirA" : {
            "subdir2" : {},
            "subdir3" : {},
            "subdir1" : {}
         },
         "dirB" : {
            "subdir2" : {},
            "subdir3" : {},
            "subdir1" : {}
         }
      },
      "dir4" : {
         "dirA" : {
            "subdir2" : {},
            "subdir3" : {},
            "subdir1" : {}
         },
         "dirB" : {
            "subdir2" : {},
            "subdir3" : {},
            "subdir1" : {}
         }
      }
   }
}

Attempt 2

Okay now with different data structure...

#! /usr/bin/perl

use warnings;
use strict;
use JSON;

my $encoder = JSON->new->ascii->pretty;   # ascii character set, pretty format
my $dirs;                                 # used to build the data structure

my $path=$ARGV[0] || '.';                 # use the command line arg or working dir

# Open the directory, read in the file list, grep out directories and skip '.' and '..'
# and assign to @dirs
opendir(my $dh, $path) or die "can't opendir $path: $!";
my @dirs = grep { ! /^[.]{1,2}/ && -d "$path/$_" } readdir($dh);
closedir($dh);

# recurse the top level sub directories with the parse_dir subroutine, returning
# a hash reference.
%$dirs = map { $_ => parse_dir("$path/$_") } @dirs;

# print out the JSON encoding of this data structure
print $encoder->encode($dirs);

sub parse_dir {
    my $path = shift;    # the dir we're working on

    # get all sub directories (similar to above opendir/readdir calls)
    opendir(my $dh, $path) or die "can't opendir $path: $!";
    my @dirs = grep { ! /^[.]{1,2}/ && -d "$path/$_" } readdir($dh);
    closedir($dh);

    return undef if !scalar @dirs; # nothing to do here, directory empty

    my $vals = [];                            # set our result to an empty array
    foreach my $dir (@dirs) {                 # loop the sub directories         
        my $res = parse_dir("$path/$dir");    # recurse down each path and get results

        # does the returned value have a result, and is that result an array of at 
        # least one element, then add these results to our $vals anonymous array 
        # wrapped in a anonymous hash
        # ELSE
        # push just the name of that directory our $vals anonymous array
        push(@$vals, (defined $res and scalar @$res) ? { $dir => $res } : $dir);
    }

    return $vals;  # return the recursed result
}

And then running the script on the proposed directory structure...

./tree2json2.pl .
{
   "dir2" : [
      "dirB",
      "dirA"
   ],
   "dir1" : [
      "dirB",
      {
         "dirA" : [
            "dirBB",
            "dirAA"
         ]
      }
   ]
}

I found this pretty damn tricky to get right (especially given the "hash if sub directories, array if not, OH UNLESS top level, then just hashes anyway" logic). So I'd be surprised if this was something you could do with sed / awk ... but then Stephane hasn't looked at this yet I bet :)