Sorting complex records

Anonymous Monk has asked for the wisdom of the Perl Monks concerning the following question:

Subject	Query	e_value	start	end
ID12	KBrH	2e-26	22400316	22400515
ID34	BrH0	2e-05	3876860	3876936
ID12	BrH0	2e-05	3876700	3873500

Replies are listed 'Best First'.
Re: Sorting complex records by Anonymous Monk on Sep 17, 2003 at 03:15 UTC
`@data = map { [split] } <>; @data = sort { $a->[0] cmp $b->[0] \|\| $a->[3] <=> $b->[3] } @data; print "@$_\n" for @data; __END__ ID12 KBrH 2e-26 22400316 22400515 ID34 BrH0 2e-05 3876860 3876936 ID12 BrH0 2e-05 3876700 3873500` [download]	[reply] [d/l]
Re: Sorting complex records by Roger (Parson) on Sep 17, 2003 at 04:16 UTC
Ok, if your data are stored in a file, say, in tab delimited format, you can sort the records without using perl at all: `sort -t "\t" -k 2 file.txt > file.out` [download] This will sort the file based on the 2nd field in the file, and output the sorted records to file.out. `sort -t "\t" -k 3 -n file.txt > file.out` [download] This will sort the file based on the numerical value of field 3. (Untested)	[reply] [d/l] [select]
Re: Sorting complex records by simonm (Vicar) on Sep 17, 2003 at 17:22 UTC
Generally, you can do this with multi-level sorts. A naive implementation might look like this: `# load data into a list of hash-refs my @data = ( { Subject => 'ID12', Query => 'KBrH', start => 22400316, ... }, ... ); my @sorted = sort { $a->{Subject} cmp $b->{Subject} or $a->{start} <=> $b->{start} } @data;` [download] I've written a CPAN module to make this kind of thing trivial: Data::Sorting. If you've loaded your data into a list of hash-refs: `use Data::Sorting 'sort_array'; sort_array( @data, 'Subject', 'start' );` [download] Alternately, if you've loaded your data into an array of arrays like this: `my $data = [ [ 'ID12', 'KBrH', '2e-26', 22400316, ... ], ... ]; use Data::Sorting 'sort_arrayref'; sort_arrayref( $data, 0, 3 );` [download] There are other functions that return a sorted copy of the array, if you don't want to change the order of the original. `use Data::Sorting 'sorted_array'; my @sorted = sorted_array( @data, 'Subject', 'start' );` [download] It'll run slower than a well-written inline sort statement, but it does some tricks under the covers to keep the performance up to acceptable levels (automatically picking a Schwartizian Transform or Guttman-Rossler strategy based on the arguments received), and you only need to write one line of code rather than a confusing block expression.	[reply] [d/l] [select]
Re: Sorting complex records by Not_a_Number (Prior) on Sep 17, 2003 at 19:21 UTC
If, as appears likely, the first field is an ID number (ie always the two letters 'ID' followed by a number), it would be nice to sort these numerically rather than ascibeticallly (ie with `ID100` coming after rather than before `ID20`...). Adapting the first answer given above, you could do this: `@data = sort { substr ($a->[0], 2) <=> substr ($b->[0], 2) \|\| $a->[3] <=> $b->[3] } @data;` [download] Meanwhile, I've downloaded the Data::Sorting module and have started playing with it, but haven't yet worked out how this would be done (although I'm sure it's simple :-). dave	[reply] [d/l] [select]
Re: Re: Sorting complex records by simonm (Vicar) on Sep 17, 2003 at 22:09 UTC
Data::Sorting allows you to pass in a subroutine reference that is responsible for extracting the value to sort on: `sort_arrayref( $data, -compare=>'numeric', sub { substr (shift)->[0] +, 2 }, 3 )` [download] There's also a (poorly-documented) interface to let you specify this in data rather than code: `sort_arrayref( $data, -compare=>'numeric', -extract=>'compound', [ index => 0, substr => 2 ], [ index => 3 ] )` [download]	[reply] [d/l] [select]