Re^2: Delete the file with checking the value

Replies are listed 'Best First'.
Re^3: Delete the file with checking the value by almut (Canon) on Feb 28, 2010 at 09:45 UTC
As usual, there are several ways to do it. In the sample code I've taken care of it by tagging the time value onto the end of the date string, aligning it such that the whole string can simply be sorted asciibetically to yield the proper result. Note that space (ASCII 32) orders before digits (ASCII 48..57). This is done with the `sprintf("%s %6d",$date,$time)`. With the following sample input `__DATA__ 0.ids.xml 500 2004-10-01 1.ids.xml 2 2004-10-01 2.ids.xml 30 2004-10-01 3.ids.xml 600 2004-10-01 4.ids.xml 40 2004-10-01 5.ids.xml 7000 2004-10-01 6.ids.xml 8000 2004-10-01 7.ids.xml 1 2004-10-01 8.ids.xml 100000 2004-10-01 9.ids.xml 90000 2004-10-01` [download] and the comparison operation as shown — `$b->[1] cmp $a->[1]` (string sort, reversed) — this would order as `2004-10-01 100000 2004-10-01 90000 2004-10-01 8000 2004-10-01 7000 2004-10-01 600 2004-10-01 500 2004-10-01 40 2004-10-01 30 2004-10-01 2 2004-10-01 1` [download] i.e. you get the entry with the highest time value as the first element. Another way would be to store the date and time values separately `push @{$hash{$common}}, [ $input_file, $date, $time ];` [download] and then use a generic chained sort operation `@files = sort {$b->[1] cmp $a->[1] \|\| $b->[2] <=> $a->[2]} @{$hash +{$k}};` [download] This works because if the date value is equal, the first comparison (`$b->[1] cmp $a->[1]`) evaluates to zero, so the next comparison (`$b->[2] <=> $a->[2]`) after the logical or "`\|\|`" is tested to determine if the time differs (it kind of "falls through"). Note that in this case the time value must be compared numerically, i.e. with `<=>`, or else (with string comparison `cmp`) the `100000` would be ordered in between `1` and `2`. See sort.	[reply] [d/l] [select]

Replies are listed 'Best First'.

Re^3: Delete the file with checking the value
by almut (Canon) on Feb 28, 2010 at 09:45 UTC

As usual, there are several ways to do it. In the sample code I've taken care of it by tagging the time value onto the end of the date string, aligning it such that the whole string can simply be sorted asciibetically to yield the proper result. Note that space (ASCII 32) orders before digits (ASCII 48..57). This is done with the sprintf("%s %6d",$date,$time).

With the following sample input

__DATA__
0.ids.xml 500 2004-10-01
1.ids.xml 2 2004-10-01
2.ids.xml 30 2004-10-01
3.ids.xml 600 2004-10-01
4.ids.xml 40 2004-10-01
5.ids.xml 7000 2004-10-01
6.ids.xml 8000 2004-10-01
7.ids.xml 1 2004-10-01
8.ids.xml 100000 2004-10-01
9.ids.xml 90000 2004-10-01
[download]

and the comparison operation as shown — $b->[1] cmp $a->[1] (string sort, reversed) — this would order as

2004-10-01 100000
2004-10-01  90000
2004-10-01   8000
2004-10-01   7000
2004-10-01    600
2004-10-01    500
2004-10-01     40
2004-10-01     30
2004-10-01      2
2004-10-01      1
[download]

i.e. you get the entry with the highest time value as the first element.

Another way would be to store the date and time values separately

    push @{$hash{$common}}, [ $input_file, $date, $time ];
[download]

and then use a generic chained sort operation

    @files = sort {$b->[1] cmp $a->[1] || $b->[2] <=> $a->[2]} @{$hash
+{$k}};
[download]

This works because if the date value is equal, the first comparison ($b->[1] cmp $a->[1]) evaluates to zero, so the next comparison ($b->[2] <=> $a->[2]) after the logical or "||" is tested to determine if the time differs (it kind of "falls through"). Note that in this case the time value must be compared numerically, i.e. with <=>, or else (with string comparison cmp) the 100000 would be ordered in between 1 and 2. See sort.

[reply]
[d/l]
[select]