IF statement to control logging to specific files based upon the browser name. Apache 1.3x

Goal: Log to specific files based upon the browser name.

I noticed that there's a <IfDefine parameter> statement for the
httpd.conf file. But it seems to be used when defining parameters
on the command-line... unless I'm misreading the documentation.

The documentation states...

$ httpd -DReverseProxy ...

I'm not sure what the $ symbol denotes in the line above, and
the name of the Apache executable is NOT httpd.

Can I get any suggestions on what to read or perhaps an example
of how to accomplish the above stated goals?

Thanks.

--
Jim Carlock
Post replies to the group.

On 2007-01-12, Jim Carlock asked...
> Can I get any suggestions on what to read or perhaps an example
> of how to accomplish the above stated goals?


"Davide Bianchi" posted...
: First idea: pipe the log into an external program and apply your
: manipulation in the aforementioned program, second idea: just
: dump everything into the log file and apply the separation at the
: rotation of the log (so once a day) or when you analyze the logs.
:
: Exactly why you want separated log files for clients?

Thinking it would be easy to grab the date off the file to get the
last mod date. And I wondered if Apache provided an IF
statement whereby I could log to different files based upon the
client browser name. And I started thinking that an IF statement
could be quite handy inside an .htaccess file or the conf files.

The only IFs I found though involved <IfModule ...> and
<IfDefine ...>. There's only sample listed under the <IfDefine ...>
involved what appeared to be a constant defined at the command
line. The documentation could be more helpful by providing more
examples.

Do you know of a website that provides extra help for Apache
that might be good for searching? I'm getting some garbage sites
when searching for things. And I don't have a list of good Apache
sites.Well, I take that back, the only one currently in the list is...

http://httpd.apache.org/

I recently converted all my logs to get recorded in a comma
delimited, quote encased manner. Although I forsee a problem
if a URL contains an embedded quote. I don't know IF I'll run
into that problem.

Going a little off topic, I ran into a problem with Time Zone
formatting. I couldn't get the abbreviated Time Zone to appear.
So I'm wondering about that issue as well.

How do I define a constant inside of the httpd.conf file?

Do you use Apache to do log rotations? I just noticed rotatelogs
and logresolve in the bin folder. So I'm currently looking over
what these two apps perform.

--
Jim Carlock
Post replies to the group.

Jim Carlock wrote:
> Goal: Log to specific files based upon the browser name.
>
> I noticed that there's a <IfDefine parameter> statement for the
> httpd.conf file. But it seems to be used when defining parameters
> on the command-line... unless I'm misreading the documentation.
>
> The documentation states...
>
> $ httpd -DReverseProxy ...
>
> I'm not sure what the $ symbol denotes in the line above, and
> the name of the Apache executable is NOT httpd.
>
> Can I get any suggestions on what to read or perhaps an example
> of how to accomplish the above stated goals?
>
> Thanks.
>
> --
> Jim Carlock
> Post replies to the group.


I'm with the post-parsing of log files idea from Davide.

Here's how you would detect a user-agent and set an apache
environmental variable. You could then use the presense of this env var
to trigger a script that logs and serves the requested_uri

you could have reg exp like
SetEnvIf User-Agent 2.0.0.1$ firefox2_0_0_1_on
Header add Firefox2_0_0_1 "%D %t" env=firefox2_0_0_1_on

or just a string
SetEnvIf User-Agent "MSIE 6.0" ie6_on
Header add ie6 "%D %t" env=ie6_on

the request would then be rewritten to a script on the basis of whether
this user-agent was present -> which would set the env var for that
request, you would append the QS for this rewrite.
The php/cgi/perl/ruby script would use the requested URI and
Query_String as args to build the data to serve.

It would log the user agent, and serve the data.

However this requires an overhead, I would hope though that your server
would move to a post-parsing of log files, which is more scalable.

"Davide Bianchi" <davideyeahsure@onlyforfun.net> wrote:
: Still, you haven't explained why you want to do such thing.
: What's the problem you are trying to solve?

My first thoughts involved separating the search engine logs
from humans. The bots go through quite a few pages. And
I'm looking for a ways to reduce the size of the logs. I'm not
sure that I want to completely drop the search engine logs
completely, so I wondered about putting those items into
separate files.

But now, that I've got the logs going into a .csv format,
perhaps building an external application to handle zipping
the files up and moving them...

I probably should have spent the time today by starting my
own app instead of trying to compile someone elses app
(Cronolog).

--
Jim Carlock
Post replies to the group.