Get the list of OpenStack-approved vendors from the OpenStack Marketplace

openstack_approved_vendors.rb

#!/usr/bin/env ruby
require 'rubygems'
require 'nokogiri'
require 'open-uri'
require 'uri'

class String
  def titleize
    humanize.split(' ').map(&:capitalize).join(' ')
  end
  
  def humanize
    tr('-', ' ')
  end
end

class OpenStackMarketplacePage
  attr_accessor :category, :url, :html

  BASE_URL = 'http://www.openstack.org'

  def initialize(category, url)
    @category = category
    @url = url
  end
  
  def html
    @html ||= Nokogiri.HTML(open(URI.join(BASE_URL, url))) if url
  end
  
  def urls(filter=nil)
    urls = html.css('a').map {|i| i.attr('href')}
    filter ? urls.select {|url| url =~ filter} : urls
  end

  def vendors
    result = unless category == 'drivers' # there's always one ...
      vendor_elem = category == 'distros' ? 4 : 3
      urls(/^#{url}/).map {|url| url.split('/')[vendor_elem]}.compact
    else
      rows = html.css('table#releaseTable tr')
      rows[1..-1].map {|row| row.css('td')[1]}.map(&:text)
    end
    
    result.compact.sort.uniq
  end
  
  def to_s
    "%s\n%s\n%s" % [category.titleize, category_underbar, vendors.map(&:titleize).join("\n")]
  end
  
  private
  def category_underbar
    '-' * category.length
  end
end

def item_id_and_url(item)
  [item['id'], item.css('a').attribute('href').value]
end

MARKETPLACE_URL = "http://www.openstack.org/marketplace"

# Get category names and URLs from marketplace nav

marketplace = Nokogiri.HTML(open(MARKETPLACE_URL))
nav_items = marketplace.css('ul.marketplace-nav > li')
categories = nav_items.map {|item| item_id_and_url(item)}.to_h

# Walk each category page for links into category, pull vendor out of that
categories.each do |category, url|
  puts ''
  page = OpenStackMarketplacePage.new(category, url)
  puts page.to_s
end

Yes, it's probably a sacrilege to write something related to OpenStack in Ruby. I'm okay with that.