scripts.mit.edu

Can I use scripts.mit.edu to generate an Apache-style log file for my web.mit.edu site?

Yes. The combination of Server Side Includes on web.mit.edu, a Perl script, and the SIPB web script service allows you to get a log file similar to the data you would be able to get if you were hosting the web pages on your own server.

How to set it up:

1. Sign up to use scripts.mit.edu.
2. Save the fake_log.pl script into your ~/web_scripts directory with a name that ends in .pl.
3. Make sure your web pages have the extension .shtml, and insert the following code into your web pages, replacing USERNAME with your username (and, if you called fake_log.pl something else, use the name that you chose):

<!-- Generate data that looks like an Apache log -->
<img src="https://USERNAME.scripts.mit.edu/~USERNAME/fake_log.pl?uri=
<!--#echo var="DOCUMENT_URI" -->&amp;ref=
<!--#echo var="HTTP_REFERER" -->&amp;met=
<!--#echo var="REQUEST_METHOD" -->&amp;pro=
<!--#echo var="SERVER_PROTOCOL" -->" 
width="1" height="1" alt="fake image for tracking"/>

4. The fake_log.pl script logs each hit to a file called fake.log in your ~/web_scripts directory. This log is designed to be compatible with Apache log files, so that you can use any of the myriad tools for analyzing and visualizing data from Apache log files.

Some questions and answers:

How does this work?

The fake_log.pl script works by taking the data returned from several Server Side Includes that work on web.mit.edu and using that and other data to write out an entry with the same format that Apache uses.

Since you’re just feeding information into a Perl script instead of getting the real server logs, couldn’t someone feed the Perl script garbage, or misleading information?

Yes. Invalid information will probably make the Perl script just plain fail. This method of logging data does not prevent people from deliberately feeding you bad data, but in most cases, no one will care to take the time to do so.

Will this log hits from text-only browsers?
Since this tracking method relies on the browser loading images, this will not log hits from text-only browsers.