dondebatch/guiaoleo.rb

   1 require 'net/http'
   2 require 'uri'
   3 require 'iconv'
   4
   5 require 'util'
   6
   7 class Guiaoleo
   8   @@url = 'http://www.guiaoleo.com.ar/list.php'
   9   @@iconv = Iconv.new('UTF-8','ISO-8859-1')
  10   attr_reader :pages
  11
  12   def fetch_page(page_num)
  13     puts 'Fetching page ' + page_num.to_s
  14     res = Net::HTTP.get(URI.parse(@@url+Util.build_uri_params({
  15       'Page' => page_num,
  16       'navby' => 'multiply',
  17       # For now, hardcode search to Capital only, since addresses returned in list don't include city
  18       'zona' => '10,15,31,5,30,19,2,22,32,13,18,7,20,3,23,55,6,1,24,25,4,8,17,12,16,9'
  19     })))
  20     res = @@iconv.iconv(res)
  21
  22     # Parse the list of restaurants. Currently parsed elements:
  23     # - name, address, detail URL
  24     restaurants = []
  25     res.scan(/detail.php.*?\/tr/).each do |row|
  26       restaurant = {}
  27       row =~ /ID=(\d+)/
  28       restaurant['url'] = 'http://www.guiaoleo.com.ar/detail.php?ID=' + $~[1]
  29       row =~ />(.*?)</
  30       restaurant['name'] = $~[1]
  31       cols = row.scan(/<td>.*?<\/td>/)
  32       cols[2]  =~ /2>(.+?)</
  33       if $~.nil?
  34          puts "Failed parsing. COL = [" + cols[2] + "] from ROW = [" + row + "]"
  35       else
  36         restaurant['address'] = $~[1] + ", capital federal, argentina"
  37       end
  38       restaurants.push(restaurant)
  39     end
  40
  41     # Parse the total number of pages
  42     @pages=res.scan(/Page=(\d+)/).inject(0) { |max,n| n[0].to_i>max ? n[0].to_i : max }
  43
  44     return restaurants
  45   end
  46 end