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

Dear brethren,
I am fairly new to perl and am having difficulty parsing a logfile that does NOT have a common delimiter. A sample from the logfile looks like this:

10/1/2003 2:06:32 AM|1|Checkout Started for US-02-14@comany.com (requested by aad6870)
10/2/2003 2:07:17 AM|1|Checkout Processed for US-02-14@company.com (requested by aad6870)
10/3/2003 2:09:37 AM|1|Checkin Processed for DN-US-02-14@company.com (requested by aad6870)
10/4/2003 9:37:53 AM|1|Checkout Started for DN-US-02-14@company.com (requested by heavis6608)
10/5/2003 9:38:29 PM|1|Checkout Processed for US-02-14@company.com (requested by heavis6608)
10/6/2003 10:10:21 AM|1|Checkout Started for US-02-17@company.com (requested by vm_karthik3521)

I need to parse out the date stamp, the activity 'Checkout Started', the machine 'US-02-14@company.com', and the username 'aad6870'. What I've started on is the following: 
while (<LOGFILE1>) 
{
    
    # parse the date field with the '|' delimiter
        @dateField = split(/\|/);
    
        # parse the activity field by matching '|1|'word space word
        @activityField = split(/\|1\|\w\s\w/);

    # Start populating the fields as appropriate
    $date = $dateField[0];
    $activity = $activityField[0];
        
    ## Write the cleaned up data to the data file
    print DATAFILE "$date , ";
    print DATAFILE "$activity ,";
    print DATAFILE "\n";

}

[download]

My problem is that the parsing of the activity field returns me the entire row. Can anyone tell me what is wrong with my regular expression for the activifity field?
Am I heading in the right direction or have the I been possessed by Perl gremlins...
Any help or suggestions greatly appreciated....

Replies are listed 'Best First'.
Re: Regular Expression help
by Roger (Parson) on Nov 11, 2003 at 23:39 UTC
    Perhaps you are looking for something like this instead? :)
    use strict;
use Data::Dumper;

while (<DATA>)
{
    chomp;                  # remove trailing \n, optional
    my @rec = split /\|/;   # split records

    my $date = $rec[0];
    my ($activity, $machine, $requester) =
            $rec[2] =~ /(.*)\sfor\s(.*)\s\(requested by (.*)\)/;

    print "$date, $activity, $machine, $requester\n";
}

__DATA__
10/1/2003 2:06:32 AM|1|Checkout Started for US-02-14@comany.com (reque
+sted by aad6870)
10/2/2003 2:07:17 AM|1|Checkout Processed for US-02-14@company.com (re
+quested by aad6870)
10/3/2003 2:09:37 AM|1|Checkin Processed for DN-US-02-14@company.com (
+requested by aad6870)
10/4/2003 9:37:53 AM|1|Checkout Started for DN-US-02-14@company.com (r
+equested by heavis6608)
10/5/2003 9:38:29 PM|1|Checkout Processed for US-02-14@company.com (re
+quested by heavis6608)
10/6/2003 10:10:21 AM|1|Checkout Started for US-02-17@company.com (req
+uested by vm_karthik3521)

    [download]
    And the output is -
    10/1/2003 2:06:32 AM, Checkout Started, US-02-14@comany.com, aad6870
10/2/2003 2:07:17 AM, Checkout Processed, US-02-14@company.com, aad687
+0
10/3/2003 2:09:37 AM, Checkin Processed, DN-US-02-14@company.com, aad6
+870
10/4/2003 9:37:53 AM, Checkout Started, DN-US-02-14@company.com, heavi
+s6608
10/5/2003 9:38:29 PM, Checkout Processed, US-02-14@company.com, heavis
+6608
10/6/2003 10:10:21 AM, Checkout Started, US-02-17@company.com, vm_kart
+hik3521

    [download]
    I think the only trick here is with the my ($var) = $str =~ /(.*)/; idiom. Which is a handy one to master. tachyon had a Perl Meditation not long ago on this topic... 291543

    Also you don't need to print the elements one line at a time, you can print them all at once.

[reply]
[d/l]
[select]
      Hi Roger, Thanks for the help, this really deepens my understanding of regular expressions. I wonder could you tell me how I could parse it without stripping the date out first with the /\|/ split. In other words is there a way to get a record to parse without splitting?? Once again thanks for the help.
[reply]
        Ok, you can change my previous code to - 
        my ($activity, $machine, $requester) = /\|1\|(.*)\sfor\s(.*)\s\(reques
+ted by (.*)\)/;

        [download]
        I have omitted the implicit $_ =~ part in the idiom. What the new code does is to look for the |1| pattern followed by the stuff you are looking for. Note that at this point, $_ holds the entire line.

[reply]
[d/l]
[select]
Re: Regular Expression help
by davido (Cardinal) on Nov 12, 2003 at 05:44 UTC
    This may seem a little funky, and it is somewhat dependant on what you want to allow as machine and user names. But this should give you something to work with. 

    my ( $date, $activity, $machine, $company ) = $_ =~
    m/^(.+(?:AM|PM))
       \|\d\|
       (.+)
       \sfor\s
       ([\w\d-]+@[\w\d.-])
       \s.+\bby\s
       ([\w\d])
       [^\w\d]+$
     /x;

    [download]

    I used the /x modifier to allow whitespace within the RE, so that I could group it in segments that each accomplish a different portion of the match. You probably ought to also have a look at perlre and perlretut, as well as perlfaq6. They will go a long way toward giving you a good comfort level with RE's.


    Dave


    "If I had my life to live over again, I'd be a plumber." -- Albert Einstein
[reply]
[d/l]
Re: Regular Expression help
by ysth (Canon) on Nov 11, 2003 at 23:34 UTC
    \w matches a word character, not a word. Repeat it like \w+
[reply]
[d/l]
[select]

Back to Seekers of Perl Wisdom