Support Sitemap Crawler in Nutch 2.x Final Report

Title :

GSOC 2015 Final Report

Reporting Date :

23th August 2015

Issue :

NUTCH-1741 - Support Sitemap Crawler in Nutch 2.x

Student :

Cihad Güzel - cguzelg@gmail.com

Mentors :

Lewis John McGibbney, Talat Uyarer

Development Codebase: :

Github Repo Url

Abstract

The url’s can be got from only pages that were scanned before in nutch crawler system. This method is expensive. But, it is possible to find the whole of urls in a up-to-date sitemap file. For this reason, sitemap files in website should be crawled. Nutch project will have that support of sitemap crawler thanks to this development.

Introduction

Sitemap is a file guiding to crawl website in a better way and it has different file formats (such as simple text format, xml format, rss 2.0, atom 0.3 & 1.0). It is possible to find the whole of urls in a up-to-date sitemap file. Websites can be crawled faster by means of sitemap crawler that will be developed. In addition, some knowledge can be detected such as “change frequance”, “last update time” and “the priority” of the pages. Shortly, a better url list will be got easily and fast from sitemap file thanks to this software. It is another advantage that this process is under the control of the user. Finally, when the project concluded;

Project Details

It is aimed to power nutch project by sitemap crawler support. The main target is to detect the sitemap having correct urls and to be crawled. It is easy and fast to find correct ursl by sitemap crawler. The software will make following features possible.

  1. sitemap detection: Sitemap files will be detected automatically, if available.
  2. sitemap list injection: Sitemap urls will be injected by using Nutch injection
  3. “Change frequence” mechanism must be supported by the crawler.
  4. Supporting multi-sitemap.
  5. Sitemap constraint: The maximum sitemap size can not be greater than 10 MB and the maximum urls can not be greater than 50,000 in a sitemap file.
  6. Sitemaps must have only inlink. Outlinks must be ignored.
  7. Sitemap crawler is the part of Nutch Life Cycle [3]. Sitemap crawler is designed according to these cases:
    • Sitemap urls can be injected from seedlist.
    • Sitemap files can be detected automatically from sites crawled.
    • It can be wanted to crawl only sitemaps.
    • It can be wanted to crawl urls except sitemap.
    • A sitemap file can give reference another sitemap file.
    • Sitemap file can be in zip format.
    • Sitemap file may be larger than 50mb. In case of this some limits must be defined.
    • Sitemaps file may include more url than 50,000. In case of this some limits must be defined.

The advatages of the process of developing project

  1. The new features that will be developed can be entegrated easily thanks to the nutch pluginer design and nutch life cycle.
  2. The current nutch plugins can be used.
  3. There are some studies about sitemap crawler in Nutch project (NUTCH-1741 [1], NUTCH-1465 [2]). The process improves by taking hand the weak and strong sides of the project

How Nutch 2.x processes Sitemap?

There are two use cases supported in Nutch's Sitemap processing:

  1. Sitemaps are considered as "remote seed lists". Crawl administrators can prepare a list of sitemap links and get only those sitemap pages. This suits well for targeted crawl of specific hosts. The sitemap urls are directly injected, fetched and parsed, if pass parameter as “-sitemap”. Nutch uses Crawler Commons Project for parsing sitemaps.
  2. For open web crawl, it is not possible to track each host and get the sitemap links manually. Nutch would automatically get the sitemaps for all the hosts seen in the crawls and inject the urls from sitemap to the crawldb, if pass parameter as “-stmDetect” when fetch . we need a list of all hosts see throughout the duration of nutch crawl. Nutch's HostDb stores all the hosts that were seen in the long crawl. Link to the robots.txt of these hosts is generated by pre-pending "http://" or "https://" schemes to the hostname. Crawler Commons is used for robots.txt parsing and thus get the sitemap links. These sitemap links are then processed same as #1.

You can see sitemap crawler life cycle schema as follow: https://issues.apache.org/jira/secure/attachment/12707721/SitemapCrawlerLifeCycle.pdf

GoogleSummerOfCode/SitemapCrawler/finalreport (last edited 2015-08-23 12:01:30 by CihadGuzel)