Simple log parser

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

Hi Monks,

I have a log of working hours which I'd like to parse. The idea is that I calculate the net hours I've done from this file and insert these values into a MySQL db. The log file has the following format (one entry per line):

dd-mm-yy hh:mm am-dd-mm-yy hh:mm pm

I've been using the regex below to pull out all the values (which works fine) and work out the hours. The one small hitch is that I need to convert pm hours into 24h format for the calculation to work (i.e. 3pm should become 15h). I've tried to do this with a conditional but cannot get it to evaluate properly - all times, regardless if am or pm, get converted to 24h format.

@lines = <LOGFILE>;

foreach $line (@lines) {
if ($line != m/\s/) {
# regex to turn dates into epoch
($dd, $mm, $yy, $hh, $mt, $tt, $tdd, $tmm, $tyy, $thh, $tmt, $ttt) = $
+line =~ /(\d+)-(\d+)-(\d+)\s(\d+):(\d+)\s(am|pm)-(\d+)-(\d+)-(\d+)\s(
+\d+):(\d+)\s(am|pm)/;

if ($tt == 'pm') {
$hh = $hh + 12;
}

if ($ttt == 'pm') {
$thh = $thh + 12;
}

$t_from = timelocal(0,$mt,$hh,$dd,$mm,$yy);
$t_to = timelocal(0,$tmt,$thh,$tdd,$tmm,$tyy);
$s_elapsed = $t_to - $t_from;

...
[download]

I'm guessing that this is one of those extremely simple problems - I just can't see where I'm going wrong. Any help would be greatly appreciated!

Mark

Comment on Simple log parser Download Code

Replies are listed 'Best First'.
Re: Simple log parser by seeker (Curate) on Aug 15, 2002 at 14:16 UTC
Shouldn't your string compares (i.e. ($tt == 'pm')) use the eq operator ($tt eq 'pm')? Try it with that change. Cheers;	[reply]
Re: Simple log parser by dimmesdale (Friar) on Aug 15, 2002 at 14:20 UTC
`@lines = <LOGFILE>; foreach $line (@lines) { if ($line != m/\s/) { # regex to turn dates into epoch ($dd, $mm, $yy, $hh, $mt, $tt, $tdd, $tmm, $tyy, $thh, $tmt, $ttt) = $ +line =~ /(\d+)-(\d+)-(\d+)\s(\d+):(\d+)\s(am\|pm)-(\d+)-(\d+)-(\d+)\s( +\d+):(\d+)\s(am\|pm)/;` [download] You say dd-mm... but then in the regex you have \d+. If you want to have just two digits, then something like this would work: `$line =~ /(\d\d)-(\d\d)\s+(\d\d):(\d\d)\s+([ap]m)-(\d\d)-(\d\d)-(\d\d) +\s+(\d\d):(\d\d)\s+[ap]m/` [download] Notice I changes `(am\|pm)` to `[ap]m`. Character class is faster than alternation, and since m is repeating in both you don't need to have that included in the alternative. Have you looked at the various Date:: modules? `if ($tt == 'pm') { $hh = $hh + 12; } if ($ttt == 'pm') { $thh = $thh + 12; }` [download] Uh-ohh.... someone didn't have use strict/use warnings on ... did they? `if ($tt eq 'pm') { $hh += 12; } if ($ttt eq 'pm') { $thh += 12; }` [download] You'll notice I used the perlop 'eq' here (== is numeric, we don't want numeric; use strict/use warnings issues a warning for this I believe). I also changed the $a = $a + $b form to $a += $b. Not necessary, but it looks a little better (to me :).	[reply] [d/l] [select]
Re: Simple log parser by grummerX (Pilgrim) on Aug 15, 2002 at 14:13 UTC
Take a look at the Delta_DHMS function in Date::Calc. -- grummerX	[reply]
Re: Simple log parser by Anonymous Monk on Aug 15, 2002 at 15:21 UTC
Wow! Thank you all kindly for your quick responses. The script works now. BTW, I didn't look into any of the Date:: modules because I wanted to try and stay with the standard Perl stuff. Anyway, thanks again!	[reply]