Re: sorting an array with decimal points

[Laurent_R]

Hi levW,

this is a typical case for using the Schwartzian Transform (see https://en.wikipedia.org/wiki/Schwartzian_transform):

use strict;
use warnings;
use feature "say";

my @array = qw(Patch_11.4 Patch_1.0 Patch_2.0 Patch_3.1 Patch_5.0 Patc
+h_4.2 Patch_6.0
               Patch_11.0 Patch_7.0 Patch_8.0 Patch_9.3 Patch_10.2 Pat
+ch_11.2);

@array = map { $_->[0] } 
    sort { $a->[1] <=> $b->[1] or $a->[2] <=> $b->[2] } 
    map { /Patch_(\d+)\.(\d+)/; [$_, $1, $2] } @array;

say for @array;

The map on the last line creates an anonymous array in which the items are array refs containing the whole string, the first number and the second number (for example: [Patch_11.4, 11, 4]). The items are then sorted according to the first number and then according to the second number. At the end, the map on the first line retrieves the original strings from the sorted array refs.

Output:

$ perl sort_versions.pl
Patch_1.0
Patch_2.0
Patch_3.1
Patch_4.2
Patch_5.0
Patch_6.0
Patch_7.0
Patch_8.0
Patch_9.3
Patch_10.2
Patch_11.0
Patch_11.2
Patch_11.4

[kcott]

G'day Laurent,

An alternative to this could be a Guttman-Rosler Transform. In the following, I've kept the same data, in the same order; and, I retained the same basic code layout for the transform.

#!/usr/bin/env perl -l

use strict;
use warnings;

my @array = qw{
    Patch_11.4 Patch_1.0 Patch_2.0 Patch_3.1 Patch_5.0
    Patch_4.2 Patch_6.0 Patch_11.0 Patch_7.0 Patch_8.0
    Patch_9.3 Patch_10.2 Patch_11.2
};

@array = map substr($_, 2),
         sort
         map pack("C2", /^Patch_(\d+)\.(\d+)/) . $_, @array;

print for @array;

The output is identical to what you show.

Just as a side note, given the list context provided by "[ ... ]", regex captures will be evaluated in that context, and your second map would only need a single statement. Here's a quick one-liner to explain:

$ perl -E 'my @x = qw{X1.2 X3.4}; say "@$_" for map { [ $_, /X(\d+)\.(
+\d+)/ ] } @x'
X1.2 1 2
X3.4 3 4

[Laurent_R]

Hi Ken,

Thank you for your comment. In fact, I had briefly thought about the GRT, but did not find a nice implementation of it in that specific case, so I stuck to the more traditional ST.

You're right about that fact that the initial map in the ST solution can be made in only one statement. Given that the ST is a bit complicated to understand for a beginner not knowing about it, I thought it would be a little clearer with two separate statements in the map, I'm no longer sure that this is really the case.

[levW]

thanks,will need a deeper look into this code...:)