Noosfero/noosfero

View on GitHub
app/services/feed_handler.rb

Summary

Maintainability
A
3 hrs
Test Coverage
require "feedparser"
require "open-uri"

# This class is responsible for processing feeds and pass the items to the
# respective container.
#
# The <tt>max_errors</tt> attribute controls how many times it will retry in
# case of failure. If a feed fails for <tt>max_errors+1</tt> times, it will be
# disabled and the last error message will be recorder in the container.
# The default value is *6*, if you need to change it you can do that in your
# config/local.rb file like this:
#
#   FeedHandler.max_errors = 10
#
# For the update interval, see FeedUpdater.
class FeedHandler
  # The maximum number
  cattr_accessor :max_errors
  cattr_accessor :disabled_period

  self.max_errors = 6
  self.disabled_period = 1.week

  def parse(content)
    raise FeedHandler::ParseError, "Content is nil" if content.nil?

    begin
      return FeedParser::Feed::new(content.force_encoding("utf-8"))
    rescue Exception => ex
      raise FeedHandler::ParseError, "Invalid feed format."
    end
  end

  def fetch(address, header = {})
    begin
      content = ""
      block = lambda { |s| content = s.read }
      content =
        if Rails.env == "test" && File.exists?(address)
          File.read(address)
        else
          if !valid_url?(address)
            raise InvalidUrl.new("\"%s\" is not a valid URL" % address)
          end

          header.merge!("User-Agent" => "Noosfero/#{Noosfero::VERSION}")
          open(address, header, &block)
        end
      return content
    rescue Exception => ex
      raise FeedHandler::FetchError, ex.message
    end
  end

  def fetch_through_proxy(address, environment)
    header = {}
    if address.starts_with?("https://")
      header.merge!(proxy: environment.https_feed_proxy) if environment.https_feed_proxy
    else
      header.merge!(proxy: environment.http_feed_proxy) if environment.http_feed_proxy
    end
    header.merge!(ssl_verify_mode: OpenSSL::SSL::VERIFY_NONE) if environment.disable_feed_ssl
    fetch(address, header)
  end

  def process(container)
    begin
      container.class.transaction do
        if failed_too_many_times(container) && enough_time_since_last_failure(container)
          container.enabled = true
          container.update_errors = 0
          container.save
        end
        next unless container.enabled

        actually_process_container(container)
        container.update_errors = 0
        container.finish_fetch
      end
    rescue Exception => exception
      Rails.logger.warn("Unknown error from %s ID %d\n%s" % [container.class.name, container.id, exception.to_s])
      Rails.logger.warn("Backtrace:\n%s" % exception.backtrace.join("\n"))
      container.reload
      container.update_errors += 1
      container.error_message = exception.to_s
      if container.update_errors > FeedHandler.max_errors
        container.fetched_at = Time.now
        container.enabled = false
      end
      begin
        container.finish_fetch
      rescue Exception => finish_fetch_exception
        Rails.logger.warn("Unable to finish fetch from %s ID %d\n%s" % [container.class.name, container.id, finish_fetch_exception.to_s])
        Rails.logger.warn("Backtrace:\n%s" % finish_fetch_exception.backtrace.join("\n"))
      end
    end
  end

  class InvalidUrl < Exception; end
  class ParseError < Exception; end
  class FetchError < Exception; end

  protected

    def actually_process_container(container)
      container.clear
      if container.environment.enable_feed_proxy
        content = fetch_through_proxy(container.address, container.environment)
      else
        content = fetch(container.address)
      end
      container.fetched_at = Time.now
      parsed_feed = parse(content)
      container.feed_title = parsed_feed.title
      parsed_feed.items[0..container.limit - 1].reverse.each do |item|
        container.add_item(item.title, item.link, item.date, item.content)
      end
    end

    def valid_url?(url)
      url =~ URI.regexp("http") || url =~ URI.regexp("https")
    end

    def failed_too_many_times(container)
      container.update_errors > FeedHandler.max_errors
    end

    def enough_time_since_last_failure(container)
      container.fetched_at.nil? || container.fetched_at < (Time.now - FeedHandler.disabled_period)
    end
end