I'm trying to sort a file containing fields.
The two I use for sort contain numerical values, but they includes "+" or "-" signes.
Even using the -n flag of sort, final result is uncorrect, because it doesn't care about the signe.
There is a way to make that two-field sort, even not using the sort command? (I'm not expert of perl nor awk)
Thanks!

Remove the "+" sign(s) in the input file

Luke Chi

Thanks! Now it works... but unfortunately it appears that just first field it's actually used for sorting.
The command line is:
sort -n -k 1.44,1.48 -k 2.36,2.41 tobeclustered.txt
to sort using respectively 6th and 5th fields. Some sample lines are like this:
1 1 | -8.05 | 98 | -4.45 | 25 |#########################
2 3 | -6.16 | 61 | -3.90 | 18 |##################
3 5 | -5.55 | 1 | -3.11 | 11 |###########
4 4 | -6.02 | 63 | -3.35 | 9 |#########
5 2 | -6.56 | 68 | -3.10 | 5 |#####
6 7 | -3.77 | 9 | -2.76 | 4 |####
7 17 | 0.14 | 32 | 1.14 | 3 |###
8 11 | -2.03 | 24 | -0.05 | 3 |###
9 8 | -2.42 | 99 | -0.16 | 2 |##
10 6 | -4.01 | 22 | -2.47 | 2 |##
11 19 | 0.99 | 81 | 2.18 | 2 |##
12 18 | 0.52 | 25 | 1.94 | 2 |##
13 13 | -0.95 | 64 | 0.62 | 2 |##
14 10 | -2.35 | 97 | -2.00 | 2 |##
15 9 | -2.35 | 5 | -2.35 | 1 |#
16 24 | 29.75 | 10 | 29.75 | 1 |#
17 23 | 8.80 | 62 | 8.80 | 1 |#
18 22 | 5.29 | 70 | 5.29 | 1 |#
19 21 | 1.97 | 73 | 1.97 | 1 |#
20 20 | 1.16 | 17 | 1.16 | 1 |#
21 16 | -0.17 | 38 | -0.17 | 1 |#
22 15 | -0.69 | 77 | -0.69 | 1 |#
23 14 | -0.75 | 19 | -0.75 | 1 |#
24 12 | -1.44 | 6 | -1.44 | 1 |#

My idea is that sort uses the 6th field to sort first, then, values showing the same value for that field will be sorted further using the 5th field. It's this correct?
Thanks,

Stefano

I was imprecise when I sayd "first field": I meant the first field used (the 6th, or the 1.44,1.48 as seen by sort).
I would like sort will order lines in a way that lines
...
7 17 | 0.14 | 32 | 1.14 | 3 |###
8 11 | -2.03 | 24 | -0.05 | 3 |###
...
will be ranked as:
8 11 | -2.03 | 24 | -0.05 | 3 |###
7 17 | 0.14 | 32 | 1.14 | 3 |###
accordingly first to the 6th field then to the 5th field.
I hope that's more clear, now... :-P

S.

input file:

1 1 | -8.05 | 98 | -4.45 | 25 |#########################
10 6 | -4.01 | 22 | -2.47 | 2 |##
11 19 | 0.99 | 81 | 2.18 | 2 |##
12 18 | 0.52 | 25 | 1.94 | 2 |##
13 13 | -0.95 | 64 | 0.62 | 2 |##
14 10 | -2.35 | 97 | -2.00 | 2 |##
15 9 | -2.35 | 5 | -2.35 | 1 |#
16 24 | 29.75 | 10 | 29.75 | 1 |#
17 23 | 8.80 | 62 | 8.80 | 1 |#
18 22 | 5.29 | 70 | 5.29 | 1 |#
19 21 | 1.97 | 73 | 1.97 | 1 |#
2 3 | -6.16 | 61 | -3.90 | 18 |##################
20 20 | 1.16 | 17 | 1.16 | 1 |#
21 16 | -0.17 | 38 | -0.17 | 1 |#
22 15 | -0.69 | 77 | -0.69 | 1 |#
23 14 | -0.75 | 19 | -0.75 | 1 |#
24 12 | -1.44 | 6 | -1.44 | 1 |#
3 5 | -5.55 | 1 | -3.11 | 11 |###########
4 4 | -6.02 | 63 | -3.35 | 9 |#########
5 2 | -6.56 | 68 | -3.10 | 5 |#####
6 7 | -3.77 | 9 | -2.76 | 4 |####
7 17 | 0.14 | 32 | 1.14 | 3 |###
8 11 | -2.03 | 24 | -0.05 | 3 |###
9 8 | -2.42 | 99 | -0.16 | 2 |##

command: (-t is the field delimiter)

sort -n -t" " -k 10,10 -k 8,8 input.txt

(For eample: in line of
15 9 | -2.35 | 5 | -2.35 | 1 |#
-2.35 is the 8th field
1 is the 10th field
)

output:

15 9 | -2.35 | 5 | -2.35 | 1 |#
24 12 | -1.44 | 6 | -1.44 | 1 |#
23 14 | -0.75 | 19 | -0.75 | 1 |#
22 15 | -0.69 | 77 | -0.69 | 1 |#
21 16 | -0.17 | 38 | -0.17 | 1 |#
20 20 | 1.16 | 17 | 1.16 | 1 |#
19 21 | 1.97 | 73 | 1.97 | 1 |#
18 22 | 5.29 | 70 | 5.29 | 1 |#
17 23 | 8.80 | 62 | 8.80 | 1 |#
16 24 | 29.75 | 10 | 29.75 | 1 |#
10 6 | -4.01 | 22 | -2.47 | 2 |##
14 10 | -2.35 | 97 | -2.00 | 2 |##
9 8 | -2.42 | 99 | -0.16 | 2 |##
13 13 | -0.95 | 64 | 0.62 | 2 |##
12 18 | 0.52 | 25 | 1.94 | 2 |##
11 19 | 0.99 | 81 | 2.18 | 2 |##
8 11 | -2.03 | 24 | -0.05 | 3 |###
7 17 | 0.14 | 32 | 1.14 | 3 |###
6 7 | -3.77 | 9 | -2.76 | 4 |####
5 2 | -6.56 | 68 | -3.10 | 5 |#####
4 4 | -6.02 | 63 | -3.35 | 9 |#########
3 5 | -5.55 | 1 | -3.11 | 11 |###########
2 3 | -6.16 | 61 | -3.90 | 18 |##################
1 1 | -8.05 | 98 | -4.45 | 25 |#########################

Luke Chi

Well, it's not exactly what I'm looking for, but probably it's my fault because it's a bit complicated for me too. To be honest, the first field was added by me, accordingly to a trick that (maybe) your self suggest in a prevoius post (a piping to a "| cat -n" after a first sort).
A simpler example, four fake lines:
INPUT:
X | XX | XX | XX | -4.9 | 4
X | XX | XX | XX | 1.9 | 4
X | XX | XX | XX | -8.9 | 10
X | XX | XX | XX | -3.9 | 20

There are two criteria I would use when sorting:
1. the 6th field, decreasing (higher values=better)
2. the 5th field, increasing (lower values=better)

That's the *real* problem.

It should be damn'easy, using the fields 1st and 6st:

FINAL OUTPUT (DESIRED)
X | XX | XX | XX | -3.9 | 20
X | XX | XX | XX | -8.9 | 10
X | XX | XX | XX | -4.9 | 4
X | XX | XX | XX | 1.9 | 4

...but how to manage the two sort orders (decreasing and increasing) in the same time?

Anyway, my best thanks for your patience up to now.

Stefano

Once -r, decending, in sort comand is specified, ALL columns in the input file will be sorted in decending order.

If -r is not specified, ALL coulmns in the input file will be sorted in ascending order.

sort command in Unix is not smart enough.

Luke Chi