Calling all UNIX gurus...

[cleo]

My web host runs Apache. I'm trying to set up a cgi script that basically gives a graphical analysis of my referrer logs. My host names the logs like this:

referrer_log.1050192000.gz
referrer_log.1050364800.gz
referrer_log.1050451200.gz
referrer_log.1050537600.gz
referrer_log.1050624000.gz
referrer_log.1050710400

The suffix is assigned server-wide, not just on my account, which is why the numbers skip around. So anyway, what I want to do is write a cron job to copy the most recent log file to another file called referrer_log. My problem lies in selecting the right file to copy. How do I tell the server to select only the most recent log, and only if it's not already gzip'd?

Any help would be greatly appreciated...

 

[wheezl]

Hmm, the question is a bit vague. a really ugly way to do this from the command line that should work is:

cp `ls -ut1 referrer_log.?????????? | head -n 1` referrer_log

What you should actually do is find your favorite scripting language and write a nice logic block that does exactly what you described in your post.

 

[zimv20]

Originally posted by wheezl

cp `ls -ut1 referrer_log.?????????? | head -n 1` referrer_log

good one. if apache doesn't guarantee using 10 digits, you can substitute:

referrer_log.*[0-9]

for

referrer_log.??????????

 

[wheezl]

good one. if apache doesn't guarantee using 10 digits, you can substitute:

referrer_log.*[0-9]

for

referrer_log.??????????

And it's prettier that way too

 

[Nipsy]

There are many freely available analysis tools which you can set up as a cron job to scan your log directory daily, and create html output.

Most of these also handle .gz logs regardless of naming convention by scanning the folder in question.