shrug-blocket-scraper.rb

#!/usr/bin/env ruby
#
# SHRUG Blocket Scraper
#
# Author:       Peter Hellberg  <@peterhellberg>
# Microformat:  hListing        <http://microformats.org/wiki/hlisting-proposal>
# License:      WTFPL

require 'cgi'
require 'rubygems'
require 'open-uri'
require 'nokogiri'
require 'sinatra'
require 'haml'
require 'sass'

class ShrugBlocketScraper
  def self.scrape(q)
    return [] if q.nil?
    
    url = "http://www.blocket.se/stockholm?q=#{CGI.escape(q)}&cg=0&w=1&st=s&ca=11&md=li"
    
    # Development (Faking the response for fun and profit)
    if File.file?('powerbook.html')
      require 'fakeweb'
      FakeWeb.allow_net_connect = false
      FakeWeb.register_uri(:get, url, :body => IO.read('powerbook.html'))
    end
    
    doc = Nokogiri::HTML(open(url))
    ads = []
    
    doc.css('table.listing tr').each do |row|
      date_str = row.css('th.listing_date').inner_text.strip
      time_str = row.css('td.listing_time').inner_text.strip
      price_str = row.css('td.align_right').inner_text.strip
      
      price = (price_str.empty?) ? nil : price_str.gsub(' ', '').to_i
      
      subject_link = row.css('td.subject a').first
      
      title = subject_link.inner_text.strip
      link = subject_link['href']
      
      date = case date_str
        when 'Idag' then "#{Date.today} #{time_str}"
        when 'Igår' then "#{Date.today - 1} #{time_str}"
        else (Date.parse(date_str) > Date.today) ?
          "#{date_str} #{Date.today.year - 1} #{time_str}" :
          "#{date_str} #{time_str}"
      end
          
      ads << BlocketAd.new(title, price, link, Time.parse(date))
    end
    
    ads
  end
end

class BlocketAd
  attr_reader :title, :price, :link, :time
  
  def initialize(title, price, link, time)
    @title = title
    @price = price
    @link  = link
    @time  = time
  end
end

get '/' do
  @ads = ShrugBlocketScraper.scrape(params[:q])
  haml :index
end

get '/got_style.css' do
  sass :stylesheet
end

__END__

@@ layout
%html{ :xmlns => "http://www.w3.org/1999/xhtml", :lang => "en", "xml:lang" => "en"}
  %head
    %title= 'SHRUG Blocket Scraper'
    %meta{ "http-equiv" => "Content-Type", :content => "text/html; charset=utf-8"}
    %meta{ :name => "viewport", :content => "width=320; initial-scale=0.9; maximum-scale=2.0; user-scalable=false;"}
    %link{ :rel => :stylesheet, :type => 'text/css', :href => '/got_style.css' }
  %body
    = yield

@@ index
%h1.title SHRUG Blocket Scraper
%form{:action => '/', :method => :get}
  %input{:type => :text, :name => :q, :size => 20, :value => params[:q]}
  %input{:type => :submit, :value => 'Sök'}
%div.listings
  - @ads.each do |ad|
    %div.hlisting
      %span.item
        %a.url{:href => ad.link}
          %span.fn= ad.title
        
        %abbr.dtlisted{:title => ad.time.iso8601}= ad.time.strftime('%Y-%m-%d')
        - if ad.price
          %span.price= "#{ad.price}:-"

@@ stylesheet
body
  border-top: 8px solid #7CAF3C 
  margin: 0
  padding: 2px 20px
  background: #fff
  color: #333
  
  
h1
  color: #433C2A
  font-size: 100px
  letter-spacing: -11px
  line-height: 75px
  font-family: Helvetica Neue, cursive
  margin-top: 0.4em
  

form
  input
    padding: 0.3em

.hlisting
  margin-bottom: 1em
  background-color: #fff
  border: 1px solid #ccc
  padding: 0.7em
  max-width: 700px
  -moz-border-radius: 8px
  -webkit-border-radius: 8px
  
.price
  color: #7CAF3C
  font-size: 1.8em
  margin-top: -1em
  float: right
  
a
  color: #433C2A
  text-decoration: none
  
.fn
  max-width: 70%
  color: #433C2A
  font-size: 1.3em
  margin-bottom: 0.6em
  display: block

.dtlisted
  color: #ccc