Spidey

Spidey provides a bare-bones framework for crawling and scraping web sites. Its goal is to keep boilerplate scraping logic out of your code.

Example

This example spider crawls an eBay page, follows links to category pages, continues to auction detail pages, and finally records a few scraped item details as a result.

class EbayPetSuppliesSpider < Spidey::AbstractSpider
  handle "http://pet-supplies.shop.ebay.com", :process_home

  def process_home(page, default_data = {})
    page.search("#AllCats a[role=menuitem]").each do |a|
      handle resolve_url(a.attr('href'), page), :process_category, category: a.text.strip
    end
  end

  def process_category(page, default_data = {})
    page.search("#ResultSetItems table.li td.dtl a").each do |a|
      handle resolve_url(a.attr('href'), page), :process_auction, default_data.merge(title: a.text.strip)
    end
  end

  def process_auction(page, default_data = {})
    image_el = page.search('div.vi-ipic1 img').first
    price_el = page.search('span[itemprop=price]').first
    record default_data.merge(
      image_url: (image_el.attr('src') if image_el),
      price: price_el.text.strip
    )
  end

end

spider = EbayPetSuppliesSpider.new verbose: true
spider.crawl max_urls: 100

spider.results  # => [{category: "Aquarium & Fish", title: "5 Gal. Fish Tank"...

Implement a spider class extending Spidey::AbstractSpider for each target site. The class can declare starting URLs by calling handle at the class level. Spidey invokes each of the methods specified in those calls, passing in the resulting page (a Mechanize Page object) and, optionally, some scraped data. The methods can do whatever processing of the page is necessary, calling handle with additional URLs to crawl and/or record with scraped results.

Storage Strategies

By default, the lists of URLs being crawled, results scraped, and errors encountered are stored as simple arrays in the spider (i.e., in memory):

spider.urls     # => ["http://www.ebay.com", "http://www.ebay.com/...", ...]
spider.results  # => [{auction_title: "...", sale_price: "..."}, ...]
spider.errors   # => [{url: "...", handler: :process_home, error: FooException}, ...]

Add the spidey-mongo gem and include Spidey::Strategies::Mongo in your spider to instead use MongoDB to persist these data. See the docs for more information. Or, you can implement your own strategy by overriding the appropriate methods from AbstractSpider.

Logging

You may set Spidey.logger to a logger of your choosing. When used in a Rails environment, the logger defaults to the Rails logger. Otherwise, it's directed to STDOUT.

Contributing

Spidey is very much a work in progress. Pull requests welcome.

To Do