feeds/app/models/site_feed.rb

class SiteFeed
  include Mongoid::Document
  include Mongoid::Timestamps
  ModuleAppSyncFields = ["feeds_model", "feeds_uid_field", "feeds_update_callback", "feeds_time_field", "feeds_finish_callback", "feeds_update_statuses_callback"]
  field :remote_site_url
  field :merge_with_category
  field :channel_name
  field :channel_title, :localize => true
  field :channel_key
  field :feed_name, localize: true
  field :disabled, type: Boolean, default: false
  field :feed_url
  field :feed_uid
  field :feeds_model
  field :feeds_uid_field
  field :feeds_update_callback
  field :feeds_time_field, type: String, default: "postdate"
  field :feeds_finish_callback
  field :feeds_update_statuses_callback
  field :enable_notify, type: Boolean, default: false
  field :is_rss, type: Boolean, default: false
  require 'feed_model/cache'
  require 'fileutils'
  include FeedModel::Cache
  Category.send(:include,FeedModel::Cache)
  after_create do
    if !self.is_rss
      self.add_notify
    end
    crontab_list = `crontab -l`.split("\n") rescue []
    site_root = Rails.root.to_s
    if crontab_list.select{|s| s.include?(site_root) && s.include?("feeds_module:make_cache")} == []
      `(crontab -l ; echo "*/5 * * * * /bin/bash -l -c 'cd #{site_root} && bundle exec rake feeds_module:make_cache > /dev/null'") | crontab`
    end
  end
  before_destroy do
    if !self.is_rss
      self.remove_notify
    end
    tmp = SiteFeedAnnc.where(feed_id: self.id)
    main_directory = File.join("#{Rails.root}","public","site_feeds")
    feed_directory = File.join(main_directory.to_s, self.id.to_s)
    FileUtils.rm_rf(feed_directory.to_s)
    if tmp.count!=0
      tmp.destroy
    end
  end
  after_save do
    if @skip_callback
      unless @skip_fix_data
        update_url = self.remote_site_url_changed?
        if self.channel_title_changed?
          self.update_channel_title(update_url)
        elsif update_url
          self.update_remote_site_url
        end
      end
    end
    @site_feed_annc = nil
  end
  scope :enabled, ->{where(:disabled => false)}
  def site_feed_annc
    @site_feed_annc ||= SiteFeedAnnc.where(:feed_id=>self.id).first
  end
  def update_channel_title(update_url=false) #update_url=true will also fix remote_site_url in data
    site_feed_annc.update_channel_title(update_url) if site_feed_annc
  end
  def update_remote_site_url
    site_feed_annc.update_remote_site_url if site_feed_annc
  end
  def sync_data_to_annc(site_feed_annc=nil)
    category_title = self.category[:title] rescue {}
    tmp_channel_title = self.channel_title_for_cache
    clone_fields =["channel_key", "merge_with_category", "remote_site_url", "feeds_model", "feeds_uid_field", "feeds_update_callback", "feeds_time_field", "feeds_update_statuses_callback"]
    (site_feed_annc ? [site_feed_annc] : SiteFeedAnnc.where(:feed_id=>self.id)).each do |tmp|
      clone_fields.each do |f|
        tmp.send("#{f}=", self.send(f))
      end
      tmp[:feed_name] = self[:feed_name]
      tmp.category_title = category_title
      tmp.channel_title = tmp_channel_title
      tmp.save
    end
  end
  def get_annc(force_refresh=false)
    def to_I18n(title, have_key=false, key="title_translations")
      value = I18n.available_locales.map{|v| [v,title]}.to_h
      if have_key
        {
          key => value
        }
      else
        value
      end
    end
    main_directory = File.join("#{Rails.root}","public","site_feeds")
    feed_directory = File.join(main_directory.to_s, self.id.to_s)
    if !force_refresh && File.exists?(feed_directory)
      anns = JSON.parse(File.read(File.join(feed_directory.to_s, self.feed_uid + ".json")))[self.channel_key.pluralize] rescue []
    else
      uri = URI(self.feed_url)
      res = get_response_body(uri) rescue ''
      if self.is_rss
        rss_data = Hash.from_xml(res)
        category_title = rss_data['rss']['channel']['title']
        category_title_trans = to_I18n(category_title, true)
        res_to_json = rss_data['rss']['channel']["item"].collect do |item|
          tag_titles = Array(item['類別']).collect{|title| to_I18n(title, true, "name_translations")}
          id = item['link'].split("&s=")[-1]
          title = Array(item['title'])[0]
          text = Array(item['description'])[0]
          files = []
          if text.include?('附件')
            files = Nokogiri::HTML(get_response_body(URI(item['link']))).search('#RelData1_liFile ol a').collect do |v|
              file_title = v.attr('title')
              {
                "description_translations" => to_I18n(file_title),
                "title_translations" => to_I18n(file_title.sub(/^\[(?:(?!\]).)*\]/,'').rpartition(".")[0]),
                "url" => v.attr('href')
              }
            end
          end
          {
            "show_url" => item['link'],
            "org_is_top" => 0,
            "id" => id,
            "title_translations" => to_I18n(title),
            "subtitle_translations" => {},
            "text_translations" => to_I18n(text),
            "postdate" => DateTime.parse(item['pubDate']).to_s,
            "image_description_translations" => {},
            "image" => {},
            "display_img" => false,
            "tags" => tag_titles,
            "category" => category_title_trans,
            "author" => item['author'],
            "params" => "#{title}-#{id}",
            "bulletin_links" => [],
            "bulletin_files" => files,
            "bulletin_carousel_images" => [],
            "external_link" => nil
          }
        end
        res = JSON.dump({"announcements" => res_to_json})
      end
      FileUtils.mkdir_p(feed_directory) if !File.exists?(feed_directory)
      File.open(File.join(feed_directory.to_s,self.feed_uid + ".json"),"w") do |file|
        res.force_encoding("utf-8")
        file.write(res)
      end
      anns = JSON.parse(res)[self.channel_key.pluralize] rescue []
    end
    anns
  end
  def get_response_body(uri)
    res = Net::HTTP.start(uri.host, uri.port,
                          :use_ssl => uri.scheme == 'https',
                          open_timeout: 60,read_timeout: 60,
                          verify_mode: OpenSSL::SSL::VERIFY_NONE) do |http|
      req = Net::HTTP::Get.new(uri)
      req['User-Agent'] = 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/105.0.0.0 Safari/537.36'
      http.request(req)
    end
    if res.code == "302" ||  res.code == "301"
      location = res['Location']
      cookie = res['Set-Cookie']  
      if location[0] == "/"
          uri = URI.parse("#{uri.scheme}://#{uri.host}#{location}")
      else
          uri = URI.parse(location)
      end
      begin
        res = Net::HTTP.start(uri.host, uri.port,
                              :use_ssl => uri.scheme == 'https',
                              open_timeout: 60,read_timeout: 60,
                              verify_mode: OpenSSL::SSL::VERIFY_NONE) do |http|
          req = Net::HTTP::Get.new(uri)
          req['Cookie'] = cookie
          http.request(req)
        end
        res_body = res.body
      rescue => e
        res_body = ''
        puts e
        puts e.backtrace
      end
      return res_body
    else
      return res.body
    end
  end
  def category
    Category.find(self.merge_with_category) rescue nil
  end
  def channel_title_for_cache #empty? must for hash
    !self[:channel_title].blank? ? self[:channel_title] : {}#I18n.available_locales.collect{|v| [v,I18n.t("feed.source")]}.to_h 
  end
  def http_request(http, request)
    response = http.request(request)
    if response.code.to_i == 301 || response.code.to_i == 302
      location = response["location"]
      new_uri = URI(location)
      http = Net::HTTP.new(new_uri.host, new_uri.port)
      if location.include?('https')
        http.use_ssl = true
        http.verify_mode = OpenSSL::SSL::VERIFY_NONE
      end
      request.instance_variable_set(:@path, new_uri.path)
      response = http_request(http, request)
    end
    response
  end
  def add_notify
    unless self.enable_notify
      root_url = Site.first.root_url rescue ""
      if root_url.present?
        uri = URI(self.remote_site_url)
        http_req = Net::HTTP.new(uri.host, uri.port)
        if self.remote_site_url.include?('https')
          http_req.use_ssl = true
          http_req.verify_mode = OpenSSL::SSL::VERIFY_NONE
        end
        http_req.open_timeout = 10
        request = Net::HTTP::Post.new("/xhr/#{self.channel_key.pluralize}/feed_add_remote/#{self.feed_uid}", 'Content-Type' => 'application/json')
        request.body = {"url"=>root_url}.to_json
        response = http_request( http_req , request )
        if response.code.to_i == 200
          self.update(:enable_notify=>true)
        end
      end
    end
  end
  def remove_notify
    if self.enable_notify
      root_url = Site.first.root_url rescue ""
      if root_url.present?
        uri = URI(self.remote_site_url)
        http_req = Net::HTTP.new(uri.host, uri.port)
        if self.remote_site_url.include?('https')
          http_req.use_ssl = true
          http_req.verify_mode = OpenSSL::SSL::VERIFY_NONE
        end
        http_req.open_timeout = 10
        request = Net::HTTP::Post.new("/xhr/#{self.channel_key.pluralize}/feed_remove_remote/#{self.feed_uid}", 'Content-Type' => 'application/json')
        request.body = {"url"=>root_url}.to_json
        response = http_request( http_req , request )
        if response.code.to_i == 200
          self.update(:enable_notify=>false)
        end
      end
    end
  end
end