Creating a Robot (Perl Cookbook, 2nd Edition)

20.11.3. Discussion

To avoid marauding robots and web crawlers hammering their servers, sites are encouraged to create a file with access rules called robots.txt. If you're fetching only one document, this is no big deal, but if your script fetches many documents from the same server, you could easily exhaust that site's bandwidth.

When writing scripts to run around the Web, it's important to be a good net citizen: don't request documents from the same server too often, and heed the advisory access rules in their robots.txt file.

The easiest way to handle this is to use the LWP::RobotUA module instead of LWP::UserAgent to create agents. This agent automatically knows to fetch data slowly when calling the same server repeatedly. It also checks each site's robots.txt file to see whether you're trying to grab a file that is off-limits. If you do, you'll get a response like this:

403 (Forbidden) Forbidden by robots.txt

Here's an example robots.txt file, fetched using the GET program that comes with the LWP module suite:

% GET http://www.webtechniques.com/robots.txt 
User-agent: *
     Disallow: /stats
     Disallow: /db
     Disallow: /logs
     Disallow: /store
     Disallow: /forms
     Disallow: /gifs
     Disallow: /wais-src
     Disallow: /scripts
     Disallow: /config

A more interesting and extensive example is at http://www.cnn.com/robots.txt. This file is so big, they even keep it under RCS control!

% GET http://www.cnn.com/robots.txt | head
# robots, scram
# $I d : robots.txt,v 1.2 1998/03/10 18:27:01 mreed Exp $
User-agent: *
Disallow: /
User-agent:     Mozilla/3.01 (hotwired-test/0.1)
Disallow:   /cgi-bin
Disallow:   /TRANSCRIPTS
Disallow:   /development

20.11. Creating a Robot

20.11.1. Problem

20.11.2. Solution

20.11.3. Discussion

20.11.4. See Also