Functions of particular interest to you will be split and sort. Here's a code example to get you on your way.

#! /usr/bin/perl

use strict ;
use warnings ;
$|++ ;

# Read the data into a 2d array
chomp( my @lines = <DATA> ) ;
my @data = () ;
foreach ( @lines )
{
    my @cells = split ;
    push @data, \@cells ;
}

# Sort the list by user.
my @data_by_name = sort { $a->[0] cmp $b->[0] } @data ;

# Sort the list by date & time.
my @data_by_date = sort {    $a->[3] <=> $b->[3] 
                          || $a->[4] <=> $b->[4] } @data ;

__DATA__
u12345 x10 qwerty  20020725 1421
u12357 x11 asdf;;  20020727 1524
u12245 x12 perl    20020722 1941
u12145 x13 python  20020725 1825
u12945 x14 /bin/sh 20020724 1331
u12545 x15 grep    20020721 1921
[download]

foreach ( @lines )
{
    my @cells = split ;
    push @data, \@cells ;
}
[download]

I wouldn't exactly be showing this to a newbie. As the original poster pointed out, he has "never even programmed", I have been using perl for about two years, and I still have some problems with references..

maybe building a hash with original lines as keys and specific column as values would be easier to understand (but maybe harder on the memory, depends on how big is the list.) don't you agree?

---

He who asks will be a fool for five minutes, but he who doesn't ask will remain a fool for life.

Chady | http://chady.net/

++, though I tend to disagree regarding references. Your suggestion was to build a hash with original lines as keys, and a specific column as values. To me, that seems inefficient, and no easier to understand. The data is already in a tabular format. The operations we want to do (sorting by a specific column) are easily realized with tabular data. Why not, then, continue to treat the data as a table once it's loaded into memory? The only thing I might do differently here would be to use an AoH (array of hashes) instead of an AoA (array of arrays), thus giving the columns names instead of numbers.

If I am wrong, and the code I presented is actually way over the original poster's head, forgive me. Here are some resources that will help:

• perlreftut, perlref
• perldsc, perldata
• perllol

_______________
D a m n D i r t y A p e
Home Node | Email

 
#!/usr/bin/perl -w 
use strict;   #always use strict

my %sortKey;

open (INFILE, "filename");   #open the file
while <INFILE> { #iterate through the file line by line
   my @tmp = split /,/;   #split the file on the comma
   
   #for the sake of the example we will say that the first 
   #element of the file is what you want to sort on
   #so we will make it the key to our hash
   $sortKey{$tmp[0]}=$_;
}

open (OUTFILE ">outfile");

#this will take the keys, shove them in an array, sort 
#them and set $key equal to the current sorted key
foreach my $key (sort(keys(%sortKey))) {
   print OUTFILE " %sortKey{$key};
}
[download]

It may well be that the file sizes are small enough that you can safely sort within Perl: it won't take too much memory, and it won't take too long.

But there is a threshold to be aware of: when the file reaches a certain size with respect to free RAM available on your computer it is more efficient to use the sort utility that comes with the operating system. (Unless you happen to be stuck on Windows, although Cygwin can can help you out there).

A sufficiently full-featured sort utility will be able to sort your file on userID and by time within userID (ascending or descending) in a single run, and it will probably be faster than Perl could do it by an order of magnitude or two. Once it is sorted it will be a snap to write a simple Perl script to walk down the file and split it out into new files when the userID changes, and those new files will already be sorted.

Back in the '60s, someone (Knuth? Hoare? Dijkstra?) observed that 50% of CPU time is spent sorting. In this age of GUIs, that proportion has no doubt decreased, but you can be sure that the sort utility that comes with your OS has had an awful lot of time spent on it making sure it runs as fast as possible (especially when the files exceed the amount of available RAM). Know when to use it.

<update>

Given a datafile as follows (I'm assuming your data really are separated by dashes):

u213-alpha-r-2002/03/19-00:09
u213-alpha-q-2002/03/19-00:08
u213-alpha-j-2002/03/19-00:01
u214-bravo-k-2002/03/19-00:02
u214-bravo-l-2002/03/19-00:03
u214-bravo-o-2002/03/19-00:06
u214-bravo-n-2002/03/19-00:05
u214-bravo-t-2002/03/19-00:11
u214-bravo-u-2002/03/19-00:12
u212-charlie-m-2002/03/19-00:04
u212-charlie-v-2002/03/19-00:13
u212-charlie-p-2002/03/19-00:07
u212-charlie-w-2002/03/19-00:14
u213-delta-s-2002/03/19-00:10

You can sort this using - as a delimiter, on the first column and then on the 4th column (descending) with the following command:

Hope this helps.

</update>

In this case my big is about 590,000 entries. I have tried the above two scripts and both have generated errors and I have no idea how to correct them. I am using Red Hat 7.3. I'm not sure what sort utilities it has that I can use. I am a little familiar with the bash shell so if there's anything I can do using that it might help. Thanks.

Your differing potential sort requirements make this sound like a problem that could be usefully approached by means of a database. Perl has lots of nifty ways to access information stored in databases, and even to treat simple files as databases. I know it sounds like learning yet another technology, but databases are pretty straightforward and easy to talk to. You could look at the DBI modules, for starters.

BCE
--Your punctuation skills are insufficient!

I'm working on the same project with Infinity. How would we do this sort command in *nix? I think we may have already tried this.

man sort

T I M T O W T D I