Untitled

require 'hpricot'
require 'iconv'
require 'open-uri'

def parse_microfox(page)
  doc = Hpricot(Iconv.conv('utf-8', 'LATIN-9', "#{page.doc}"))

  @title = ""
  @description = ""
  @price = ""
  @page_url = ""

  # GET TITLE
  if @title = doc.search("//p.modello") then
    @title = @title.inner_html
  end

  # GET DESCRIPTION
  # rimuovi prezzo di mercato
  if @description = doc.search("//p.descrizione_generale") then
    @description = @description.inner_html
  end

  # GET PRICE
  if @price = doc.search("//p.prezzo_menoiva/span")[1] then
    @price = @price.inner_html.split[1]
  end

  @page_url = "#{page.url}"

  if p = Product.find_by_page_url(@page_url) then
    puts "Product already fetched."
  else
    p = Product.new
    p.name = @title

    # STRIP HTLM TAGS
    #remove format tags
    p.description = @description.gsub(/[\r\n\t]/, "").gsub(/^\s+|\s+$/,"").gsub(/<\/?[^>]*>/,  "")

    # CONVERT PRICE FORMAT
    if @price.nil? then
      p.price = 0
    else
      #convert price format
      p.price = @price.gsub(".","").gsub(",",".").to_f
    end

    p.page_url = @page_url
    p.shop_id = Shop.find_by_name("Microfox3000").id
  end

  if p.save then
    return p
  else
    return nil
  end
end