/vendor/tools/parser/lib/parser/globo_esporte.rb
https://bitbucket.org/tbueno/planetafutebol · Ruby · 44 lines · 31 code · 13 blank · 0 comment · 1 complexity · 304360ed5e8f41f26b487b3ddd12aaf8 MD5 · raw file
- require 'open-uri'
- require 'action_view'
- module Parser
- class GloboEsporteEntry < Feedzirra::Parser::RSSEntry
- include ActionView::Helpers::SanitizeHelper
- include SAXMachine
- include Feedzirra::FeedEntryUtilities
- def content
- return @content if @content
- page = Nokogiri::HTML(open(url)).css('div#materia-letra').first
- @content = page.inner_html ? sanitize(page.inner_html, :attributes => %w(alt src title href)) : ''
- end
- def summary
- sanitize @summary, :tags => %w(p)
- end
- def author
- 'globoesporte.com'
- end
- end
- class GloboEsporte
- include SAXMachine
- include Feedzirra::FeedUtilities
- element :title
- element :link, :as => :url
- elements :item , :as => :entries, :class => GloboEsporteEntry
- attr_accessor :feed_url
- def self.able_to_parse?(xml)
- (/globoesporte.globo.com/ =~ xml)
- end
- end
- end