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
}