GSP
Quick Navigator

Search Site

Unix VPS
A - Starter
B - Basic
C - Preferred
D - Commercial
MPS - Dedicated
Previous VPSs
* Sign Up! *

Support
Contact Us
Online Help
Handbooks
Domain Status
Man Pages

FAQ
Virtual Servers
Pricing
Billing
Technical

Network
Facilities
Connectivity
Topology Map

Miscellaneous
Server Agreement
Year 2038
Credits
 

USA Flag

 

 

Man Pages
Gungho(3) User Contributed Perl Documentation Gungho(3)

Gungho - Yet Another High Performance Web Crawler Framework

  use Gungho;
  Gungho->run($config);

Gungho provides a complete out-of-the-box web crawler framework with high performance and great felxibility.

Please note that Gungho is in beta. It has been stable for some time, but its internals may still change, including the API.

Gungho comes with many features that solve recurring problems when building a spider:

Event-Based, Asynchronous Engine
Gungho uses event-based dispatch via POE, Danga::Socket, or IO::Async. Choose the best engine that fits your needs.
Asynchronous DNS lookups
HTTP connections are handled asynchronously, why not DNS lookups? Gungho doesn't block while hostnames are being resolved, so other jobs can continue.
Automatic robots.txt Handling
Every crawler needs to respect robots.txt. Gungho offers automatic handling of robots.txt. If you use it in conjunction with memcached, you can even do this in a distributed environment, where farms of Gungho crawler hosts are all fetching pages.
Robots META Directives
Robots META directives embedded in HTML text can also be parsed automatically. You can then access this resulting structure to decide if you can process the fetched URL.
Throttling
You don't want your crawl targets to go under just because you let loose a crawler against it and did a million fetches per hour. With Gungho's throttling component, you can throttle the amount of requests that are sent against a domain.
Private IP Blocking
Malicious sites may embed hostnames that resolve to internal IP address ranges such as 192.168.11.*, which may lead to a DoS attack to your private servers. Gungho has an automatic option to block such IP addresses via BlockPrivateIP component.
Caching
Whatever you want to cache, Gungho offers a generic cache interface a-la Catalyst via Gungho::Component::Cache
Web::Scraper Integration
(Note: This is not quite production ready) Gungho has Web::Scraper integration that allows you to easily call Web::Scraper sripts defined in your config files.
Request Logging
Requests can be automatically logged to a file, a database, to screen, via Gungho::Plugin::RequestLog, which gives you the full power of Log::Dispatch for your logging needs.

First there were a bunch of scripts that used scrape a bunch of RSS feeds. Then I got tired of writing scripts, so I decided a framework is the way to go, and Xango was born.

Xango was my first attempt at trying to harness the full power of event-based framework. It was fast. It wasn't fun to extend. It had a nightmare-ish way to deal with robots.txt.

Couple of more attempts later, more inspirations and lessons learned from Catalyst, Plagger, DBIx::Class, Gungho was born.

Since its inception, Gungho has been in successfully used as crawlers that fetch hundreds of thousands of urls to a few million urls per day.

Gungho is designed to so that it can handle massive amount of traffic. If you're careful enough with your Provider and Handler implementation, you can in fact hit millions of URL with this crawler.

So PLEASE DO NOT LET IT LOOSE. DO NOT OVERLOAD your crawl targets. You are STRONGLY advised to use Gungho::Component::Throttle to throttle your fetches.

Also PLEASE CHANGE THE USER AGENT NAME OF YOUR CRAWLER. If you hit your targets hard with the default name (Gungho/VERSION X.XXXX), it will look as though a service called Gungho is hitting their site, which really isn't the case. Whatever it is, please specify at least a simple user agent in your config

Gungho is comprised of three parts. A Provider, which provides Gungho with requests to process, a Handler, which handles the fetched page, and an Engine, which controls the entire process.

There are also "hooks". These hooks can be registered from anywhere by invoking the register_hook() method. They are run at particular points, which are specified when you call register_hook().

All components (engine, provider, handler) are overridable and switcheable. However, do note that if you plan on customizing stuff, you should be aware that Gungho uses Class::C3 extensively, and hence you may see warnings about the code you use.

One note about Gungho - Don't use it if you are planning on accessing a single url -- It's usually not worth it, so you might as well use LWP::UserAgent or an equivalent module.

Gungho's event driven engine works best when you are accessing hundreds, if not thousands of urls. It may in fact be slower than using LWP::UserAgent if you are accessing just a single url.

Of course, you may wish to utilize features other than speed that Gungho provides, so at that point, it's simply up to you.

Gungho has experimental support for running in distributed environments.

Strictly speaking, each crawler needs to have its own strategy to enable itself to to run in a distribued environment. What Gungho offers is a "good enough" solution that may work for your. If what Gungho offers isn't enough, at least what comes with it might help to show you what needs to be tweaked for your particular environment.

Roughly speaking, there are three components you need to worry about in order to make a well bahaved and distributed crawler. Check out the below list and documentation for each component.

Distributed Throttling
As of version 0.08010, Throttle::Domain and Throttle::Simple can be configured to use whatever Data::Throttler-based throttling object as its engine.

Download Data::Throttler::Memcached, and specify it as the engine behind your throttling for Gungho. Using Data::Throttler::Memcached will make Gungho store throttling information in a shared Memcached server, which will allow separate Gungho instances to share that information.

Distributed robots.txt Handling
As of version 0.08013, RobotRules can be configured to use a cache in the backend. You can specify your choice of distributed cache (e.g. Memcached) and use that as the storage for robots.txt data.

Of course, this means that robots.txt data isn't persitent, but you should be expiring robots.txt once in while to reflect new data, anyways.

Distributed Provider
This is actually the simplest aspect, as it's usually done by hooking the provider with a database. However, if you prefer, you may use some sort of Message Queue as your backend.

debug
   ---
   debug: 1
    

Setting debug to a non-zero value will trigger debug messages to be displayed.

Components add new functionality to Gungho. Components are loaded at startup time from the config file / hash given to Gungho constructor.

  Gungho->run({
    components => [
      'Throttle::Simple'
    ],
    throttle => {
      max_interval => ...,
    }
  });

Components modify Gungho's inheritance structure at run time to add extra functionality to Gungho, and therefore should only be loaded before starting the engine.

Please refer to each component's document for details

Gungho::Component::Authentication::Basic
Gungho::Component::BlockPrivateIP
Gungho::Component::Cache
Gungho::Component::RobotRules
Gungho::Component::RobotsMETA
Gungho::Component::Scraper
Gungho::Component::Throttle::Domain
Gungho::Component::Throttle::Simple

If you're looking into simple crawlers, you may want to look at Gungho::Inline,

  Gungho::Inline->run({
    provider => sub { ... },
    handler  => sub { ... }
  });

See the manual for Gungho::Inline for details.

Plugins are different from components in that, whereas components require the developer to explicitly call the methods, plugins are loaded and are not touched afterwards.

Please refer to the documentation of each plugin for details.

RequestLog
Statistics

Currently available hooks are:

Used for Class::C3::Componentised

You can obtain the current code base from

  http://gungho-crawler.googlecode.com/svn/trunk

Copyright (c) 2007 Daisuke Maki <daisuke@endeworks.jp>

Jeff Kim
Kazuho Oku
Keiichi Okabe

This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.

See http://www.perl.com/perl/misc/Artistic.html

2008-07-28 perl v5.32.1

Search for    or go to Top of page |  Section 3 |  Main Index

Powered by GSP Visit the GSP FreeBSD Man Page Interface.
Output converted with ManDoc.