Sorting number ranges from two arrays in bash

Question

I have two arrays with the same number of elements I want to manipulate. They are read from a file into two arrays (odd numbered lines go into array one, even numbered into array 2):

arr1=("1" "1" "3" "2" "4" "7" "7" "7" "1" "2" "3" "3" "3" "3" "7" "5")
arr2=("4" "1" "3" "5" "7" "1" "2" "3" "2" "9" "2" "6" "8" "9" "4" "6")

The data refers to season numbers and episode numbers in the same position across the two arrays. So array 1 (arr1) is seasons, array 2 (arr2) is episodes, and they align by element number. So ${arr1[0]} corresponds to ${arr2[0]}, etc.

What I am trying to do is sort them so that they are numerically ordered first by season, then by episode. So the original array (with annotations for which item is which element):

       1   2   3   4   5   6   7   8   9   10  11  12  13  14  15  16
arr1=("1" "1" "3" "2" "4" "7" "7" "7" "1" "2" "3" "3" "3" "3" "7" "5")
arr2=("4" "1" "3" "5" "7" "1" "2" "3" "2" "9" "2" "6" "8" "9" "4" "6")
       1   2   3   4   5   6   7   8   9   10  11  12  13  14  15  16

Becomes:

       2   9   1   4   10  12  3   12  13  14  5   16  6   7   8   15
arr1=("1" "1" "1" "2" "2" "3" "3" "3" "3" "3" "4" "5" "7" "7" "7" "7")
arr2=("1" "2" "4" "5" "9" "2" "3" "6" "8" "9" "7" "6" "1" "2" "3" "4")
       2   9   1   4   10  12  3   12  13  14  5   16  6   7   8   15

Possible ideas:

1. For each item i in ${arr1[@]}, write the corresponding element of ${arr2[n]} to a file. The file can then have sort run on it.

n="0"
for i in "${arr1[@]}"; do
    echo "${arr2[${n}]}" >> "${i}.txt"
    (( n++ ))
done

But I want to try avoid involving a disk write and sort if I don't have to.

2. Sort the data into some kind of an individual array? Each season gets its own array, and then the array can be sorted with something like sort -n "${season1Arr[@]}" -- But I don't know how this would be done.

3. Change the way the data is manipulated? I can't change the input file, but I can change the way it's processed. Perhaps instead of reading lines into two arrays based on even/odd line numbers, they could be managed some other way?

I'm trying to keep it as pure-bash as possible for compatibility, but I understand that external programs very may well need to be used. Appreciate any ideas.

Answer 1

Here's an example of what I was talking about in my comments, using perl and an associative array (hash), but using a version sort rather than the natural sort I mentioned (a simple numeric or alpha-numeric sort would put, e.g. 1.10 before 1.2, which is clearly wrong for "series.episode"):

#!/usr/bin/perl

use strict;

use Sort::Versions;

my %data;
my ($key, $series, $episode);

while (<>) {
  chomp; # remove trailing newline from input

  if ($. % 2 == 1) {
    $series = $_;
  } else {
    $episode = $_;
    $key = "$series.$episode";
    $data{$key} = 1;
  };
}

print join(", ", sort { versioncmp($a, $b) } keys %data), "\n";

This could be shortened a bit - the only variables really needed are the %data hash and $series. I've used $key and $episode to make it obvious how the input values are being used. BTW, $_ is the default input/value/iterator. It has lots of uses in perl and many functions & syntax elements use it if another variable isn't provided - in this script, it's the value of the current line as read by the while (<>) loop. See man perlvar and search for "General Variables".

#!/usr/bin/perl

use strict;
use Sort::Versions;

my (%data, $series);

while (<>) {
  chomp; 

  if ($. % 2 == 1) {
    $series = $_;
  } else {
    $data{"$series.$_"} = 1;
  };
}

print join(", ", sort { versioncmp($a, $b) } keys %data), "\n";

Sample run (both versions produce the same output):

$ ./sort-series-episode.pl input.txt  
1.1, 1.2, 1.4, 1.15, 2.5, 2.9, 3.2, 3.3, 3.6, 3.8, 3.9, 4.7, 5.6, 7.1, 7.2, 7.3, 7.4

Note: the Sort::Versions module is not a core perl module and needs to be installed separately, either via cpan or with a distro package - e.g. in Debian, apt-get install libsort-versions-perl

Answer 2

While perl is much better for such tasks, it is possible to do it in almost pure bash (safe for external sort):

#!/bin/bash
arr1=( "1" "1" "3" "2" "4" "7" "7" "7" "1" "2" "3" "3" "3" "3" "7" "5" )
arr2=( "4" "1" "3" "5" "7" "1" "2" "3" "2" "9" "2" "6" "8" "9" "4" "6" )

# make an array with strings from both initial arrays:
arr_length=${#arr1[@]}
for ((i=0; i< $arr_length; i++)); do
   arr_combined[$i]="${arr1[$i]}!${arr2[$i]}"
done

# sort the combined strings
arr_sorted=( $(printf "%s\n" "${arr_combined[@]}" | sort) )

# split the elements of sorted array back into two arrays
for pair in ${arr_sorted[@]} ; do
    arr1n+="${pair%!*} "
    arr2n+="${pair#*!} "
done

# print the results
printf "%s\n" "${arr1n[@]}"
printf "%s\n" "${arr2n[@]}"

Answer 3

You can make life much easier if you sort the numbers in the original file first and then populate your arrays. I assume your original file contained something like this:

$ cat file1
1 1 3 2 4 7 7 7 1 2 3 3 3 3 7 5
4 1 3 5 7 1 2 3 2 9 2 6 8 9 4 6

So this command:

transpose file1 | sort | transpose

would produce what you want:

1 1 1 2 2 3 3 3 3 3 4 5 7 7 7 7
1 2 4 5 9 2 3 6 8 9 7 6 1 2 3 4

The transpose function is very simple. You can write your own. Here is one example:

transpose () 
{ 
    awk '{ for (i=1; i<=NF; i++) a[i]= (i in a?a[i] OFS :"") $i; } 
    END{ for (i=1; i<=NF; i++) print a[i] }' $1
}

Answer 4

From your description, the following may be the original input data:

1
4
1
1
3
3
2
5
4
7
7
1
7
2
7
3
1
2
2
9
3
2
3
6
3
8
3
9
7
4
5
6

To process this in the way you seem to want:

paste - - <file | sort -k 1,1n -k 2,2n

This first transforms the data into two tab-delimited columns, with the season in the first column and the episode in the second.

The lines are sorted numerically on the first column, and any lines with identical first columns are sorted numerically on the second.

With the input shown above, this yields:

1       1
1       2
1       4
2       5
2       9
3       2
3       3
3       6
3       8
3       9
4       7
5       6
7       1
7       2
7       3
7       4