in reply to Need help comparing 4 dates

Comparing more than two values--is sorting.

In theory, this should work(Update:see 2 posts down for fixed code): 

#! perl -slw
use warnings;
use strict;
use Data::Dumper;
use Date::Manip;

print scalar <DATA>;
while (<DATA>) {
    chomp;
    my @dates = split '~';
    my $label = shift @dates;
    my @order = map $_+1, sort{
        Date_Cmp( $dates[ $a ], $dates[ $b ] );
    } 0 .. $#dates;
    print join "\t", $label, @order;
}

__DATA__
EVENT   JIM     BOB     SAM     JACK
PTRED~09/29/10 03:23:05 PM ~09/28/10 02:21:09 PM ~09/26/10 11:00:03 AM
+ ~09/27/10 09:33:41 PM
RED~08/29/10 01:55:00 AM ~08/30/09 12:10:10 PM ~08/27/10 08:16:21 PM ~
+09/01/10 12:12:12 AM
INT~07/04/10 03:21:15 AM ~07/08/10 04:17:33 PM ~06/30/10 04:22:11 AM ~
+06/28/10 10:11:01 PM
PTRED~06/19/10 09:19:55 PM ~04/25/10 07:39:22 PM ~09/16/10 10:34:24 AM
+ ~07/22/10 06:19:38 PM
RED~04/29/10 12:10:59 AM ~04/20/10 02:13:33 AM ~07/17/10 01:00:05 PM ~
+09/01/10 11:10:15 PM
INT~05/23/10 11:11:11 PM ~01/08/10 10:45:12 PM ~05/15/09 03:29:37 AM ~
+05/18/09 12:59:59 PM

[download]

It populates an array with the dates; sorts the indices of the array, by the (ascending) values of the dates indexed; then adds one to the ordered indices to produce your "ranking value".

It really should work, but it doesn't! It produces this output which is clearly wrong, but for reasons I do not (yet) understand:

c:\test>junk51
EVENT   JIM     BOB     SAM     JACK

PTRED   3       4       2       1
RED     2       3       1       4
INT     4       3       1       2
PTRED   2       1       4       3
RED     2       1       3       4
INT     3       4       2       1

[download]

At first I thought that maybe Date_Cmp(), needed the dates to be pre-parsed, but that didn't fix it.

Then I thought that maybe D::M was getting confused by the 2 digit years, so I fixed that up, but no change either.

So then I thought that the (elusively documented) "flag" returned by Date_Cmp() might be incompatible with the expectations of sort, but that doesn't seem to be the case either.

This modified version:

#! perl -slw
use warnings;
use strict;
use Data::Dumper;
use Date::Manip;

print scalar <DATA>;
while (<DATA>) {
    chomp;
    my @dates = split '~';
    my $label = shift @dates;
    my @parsed = map{ s[/(\d\d) ][/20$1 ]; ParseDate( $_ ) } @dates;
    my @order = map $_+1, sort{
        my $res = Date_Cmp( $dates[ $a ], $dates[ $b ] );
        print "$dates[ $a ] <-> $dates[ $b ] := $res";
        $res;
    } 0 .. $#parsed;
    print join "\t", $label, @order;
}
__DATA__
EVENT   JIM     BOB     SAM     JACK
PTRED~09/29/10 03:23:05 PM ~09/28/10 02:21:09 PM ~09/26/10 11:00:03 AM
+ ~09/27/10 09:33:41 PM
RED~08/29/10 01:55:00 AM ~08/30/09 12:10:10 PM ~08/27/10 08:16:21 PM ~
+09/01/10 12:12:12 AM
INT~07/04/10 03:21:15 AM ~07/08/10 04:17:33 PM ~06/30/10 04:22:11 AM ~
+06/28/10 10:11:01 PM
PTRED~06/19/10 09:19:55 PM ~04/25/10 07:39:22 PM ~09/16/10 10:34:24 AM
+ ~07/22/10 06:19:38 PM
RED~04/29/10 12:10:59 AM ~04/20/10 02:13:33 AM ~07/17/10 01:00:05 PM ~
+09/01/10 11:10:15 PM
INT~05/23/10 11:11:11 PM ~01/08/10 10:45:12 PM ~05/15/09 03:29:37 AM ~
+05/18/09 12:59:59 PM

[download]

Produces this:

c:\test>junk51
EVENT   JIM     BOB     SAM     JACK

09/29/2010 03:23:05 PM  <-> 09/28/2010 02:21:09 PM  := 1
09/26/2010 11:00:03 AM  <-> 09/27/2010 09:33:41 PM := -1
09/28/2010 02:21:09 PM  <-> 09/26/2010 11:00:03 AM  := 1
09/28/2010 02:21:09 PM  <-> 09/27/2010 09:33:41 PM := 1
PTRED   3       4       2       1
08/29/2010 01:55:00 AM  <-> 08/30/2009 12:10:10 PM  := 1
08/27/2010 08:16:21 PM  <-> 09/01/2010 12:12:12 AM := -1
08/30/2009 12:10:10 PM  <-> 08/27/2010 08:16:21 PM  := -1
08/27/2010 08:16:21 PM  <-> 08/29/2010 01:55:00 AM  := -1
08/29/2010 01:55:00 AM  <-> 09/01/2010 12:12:12 AM := -1
RED     2       3       1       4
07/04/2010 03:21:15 AM  <-> 07/08/2010 04:17:33 PM  := -1
06/30/2010 04:22:11 AM  <-> 06/28/2010 10:11:01 PM := 1
07/04/2010 03:21:15 AM  <-> 06/28/2010 10:11:01 PM := 1
07/04/2010 03:21:15 AM  <-> 06/30/2010 04:22:11 AM  := 1
INT     4       3       1       2
06/19/2010 09:19:55 PM  <-> 04/25/2010 07:39:22 PM  := 1
09/16/2010 10:34:24 AM  <-> 07/22/2010 06:19:38 PM := 1
04/25/2010 07:39:22 PM  <-> 07/22/2010 06:19:38 PM := -1
07/22/2010 06:19:38 PM <-> 06/19/2010 09:19:55 PM  := 1
PTRED   2       1       4       3
04/29/2010 12:10:59 AM  <-> 04/20/2010 02:13:33 AM  := 1
07/17/2010 01:00:05 PM  <-> 09/01/2010 11:10:15 PM := -1
04/20/2010 02:13:33 AM  <-> 07/17/2010 01:00:05 PM  := -1
07/17/2010 01:00:05 PM  <-> 04/29/2010 12:10:59 AM  := 1
RED     2       1       3       4
05/23/2010 11:11:11 PM  <-> 01/08/2010 10:45:12 PM  := 1
05/15/2009 03:29:37 AM  <-> 05/18/2009 12:59:59 PM := -1
01/08/2010 10:45:12 PM  <-> 05/15/2009 03:29:37 AM  := 1
01/08/2010 10:45:12 PM  <-> 05/18/2009 12:59:59 PM := 1
INT     3       4       2       1

[download]

Each individual comparison produces the expected result, but still the ordering is wrong. For all teh world it looks like sort is making mistakes :)

At this point I ran out of ideas, so I throw it over to the monks to point out my mis deliberate error?

Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
"Science is about questioning the status quo. Questioning authority".
In the absence of evidence, opinion is indistinguishable from prejudice.
RIP an inspiration; A true Folk's Guy

Replies are listed 'Best First'.
Re^2: Need help comparing 4 dates (sort errors?)
by AnomalousMonk (Archbishop) on Oct 03, 2010 at 08:53 UTC
    It populates an array with the dates; sorts the indices of the array, by the (ascending) values of the dates indexed; then adds one to the ordered indices to produce your "ranking value".

    But the 'ranking value' is really the date itself: the earliest date is 'first place', the latest date is 'fourth place' (in the OPed example data). It's necessary to convert each date, at some point, into its 'place' from 1 to n.

    Once the ordered dates have been converted to place values, they have to be restored to the original order of the 'players' Jim .. Jack for printing.

[reply]

      Duh! Of course. Thanks.

      Once I've sorted the indices by the dates, in order to rank them; I need to sort the indices by that ranking to get them back into the original order: 

      #! perl -slw
use warnings;
use strict;
use Data::Dumper;
use Date::Manip;

print scalar <DATA>;
while (<DATA>) {
    chomp;
    my @dates = split '~';
    my $label = shift @dates;
    my @order = sort{
        Date_Cmp( $dates[ $a ], $dates[ $b ] );
    } 0 .. $#dates;

    print join "\t", $label, map $_+1, sort{
        $order[ $a ] <=> $order[ $b ]
    } 0..$#order;
}

__DATA__
EVENT   JIM     BOB     SAM     JACK
PTRED~09/29/10 03:23:05 PM ~09/28/10 02:21:09 PM ~09/26/10 11:00:03 AM
+ ~09/27/10 09:33:41 PM
RED~08/29/10 01:55:00 AM ~08/30/09 12:10:10 PM ~08/27/10 08:16:21 PM ~
+09/01/10 12:12:12 AM
INT~07/04/10 03:21:15 AM ~07/08/10 04:17:33 PM ~06/30/10 04:22:11 AM ~
+06/28/10 10:11:01 PM
PTRED~06/19/10 09:19:55 PM ~04/25/10 07:39:22 PM ~09/16/10 10:34:24 AM
+ ~07/22/10 06:19:38 PM
RED~04/29/10 12:10:59 AM ~04/20/10 02:13:33 AM ~07/17/10 01:00:05 PM ~
+09/01/10 11:10:15 PM
INT~05/23/10 11:11:11 PM ~01/08/10 10:45:12 PM ~05/15/09 03:29:37 AM ~
+05/18/09 12:59:59 PM

      [download]

      Produces:

      c:\test>junk51
EVENT   JIM     BOB     SAM     JACK

PTRED   4       3       1       2
RED     3       1       2       4
INT     3       4       2       1
PTRED   2       1       4       3
RED     2       1       3       4
INT     4       3       1       2

      [download]
      Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
      "Science is about questioning the status quo. Questioning authority".
      In the absence of evidence, opinion is indistinguishable from prejudice.
[reply]
[d/l]
[select]

        Thank you so much! This is spot on!

[reply]

        Hi All, this does work beautifully, except for one thing...i keep getting the following warning message:

        Use of uninitialized value in length at /u/capsftp/lib/perl5/site_perl/5.8.0/Date/Manip.pm line 247, <DATA> line 2

        I was curious if you were getting it also? and if so, any idea what is causing it? With Data::Dumper i can see the values all seem to be defined when passed to the Date_Cmp function so not sure why it's happening

[reply]

In Section Seekers of Perl Wisdom