api for funcheap - fetch events from funcheap

funcheap_events.rake

require 'chronic'
require 'nokogiri'
require 'open-uri'
require 'date'
require 'json'
require 'paperclip'

namespace :evee do
  desc "Fetch Events from Funcheap"
  task pull_funcheap_events: :environment do
    
    def duplicate_event(uniqkey)
       event = Event.where(uniqkey: uniqkey)
       return event.present? ? true : false
    end

  	def event_parse(url)
      return nil unless url
      doc = Nokogiri::XML(open(url))
      event_info =  doc.css('span.left span script')
      if event_info[5]
        maps_raw    = event_info[5].text
        maps_json   = maps_raw[maps_raw.index('{'), maps_raw.rindex('}]}') - maps_raw.index('{') + 3]
        loc_info    = JSON.parse(maps_json)
        image       = doc.css('img.alignright')
        coordinates = loc_info["pois"][0]["point"]
        address     = loc_info["pois"][0]["address"]
        return [coordinates, address]
      end
    end
   
   required_date     =  Date.new(2015,10,1)
   month_start_date  =  required_date.beginning_of_month  #|| Date.today.beginning_of_month
   month_end_date    =  required_date.end_of_month        # || Date.today.end_of_month
     
      (month_start_date).upto(month_end_date) do |d| 
       begin
        puts eventlist_url = "http://sf.funcheap.com/#{d.strftime("%Y/%m/%d")}/"
        
        page = Nokogiri::HTML(open(eventlist_url))
        title_parse = page.css('div.tanbox.left')
        
         title_parse.each do |event|
          if event.css('span.title').css('a')[0] != nil
            coordinates, address, image   =   event_parse(event.css('span.title').css('a')[0]["href"])
            events_array                  =   event.css('span.title').css('a').text.split(" | ")
            date_time                     =   event.css('div.meta').text
            misc                          =   date_time[0, date_time.index("|")]
            time_extract                  =   Chronic.parse(date_time[0, date_time.index("|")])
            descrip_text                  =   (event.css('p').text)
            d_string                      =   event.children.last.text.gsub("\r", "").gsub("\n", "").gsub("\t", "").strip
            d_blurb                       =   if descrip_text == "" then d_string else descrip_text end
            event_name                    =   events_array[0]
            event_description             =   d_blurb.gsub("\n", " ")
            event_url                     =   event.css('a')[0]["href"]
            event_image_url               =   event.css('img')[0]["src"]
            uniqkey                       =   event_name.split.join + "-" + d.strftime("%Y%m%d")
                 unless duplicate_event(uniqkey)
                  event = Event.new(:title 			     =>  event_name,
                                    :description     =>  event_description.present? ? event_description : "No Description Available",
                                    :url             =>  event_url.present? ? event_url : "http://sf.funcheap.com",
                                    :event_image     =>  event_image_url.present? ? URI.parse(event_image_url).open : URI.parse("http://dnh7crmrr8pwi.cloudfront.net/images/events/574.large.png?1441967456").open,
                                    :venue           =>  address.present? ? address : "All Over San Francisco",
                                    :address         =>  "San Francisco, CA",
                                    :latitude        =>  coordinates.present? ? coordinates["lat"].to_f : 37.773972,
                                    :longitude       =>  coordinates.present? ? coordinates["lng"].to_f : -122.431297,
                                    :source          =>  "Funcheap",
                                    :price           =>  0,
                                    :city            =>  "San Francisco",
                                    :moods           =>  ["social"],
                                    :startdate       =>  d.strftime("%Y/%m/%d").gsub('/','-'),
                                    :enddate         =>  d.strftime("%Y/%m/%d").gsub('/','-'),
                                    :uniqkey         =>  uniqkey
                                   )
                 event.save
                 puts "objectId: #{event.id} -  #{event.title} - #{event.event_image.url}"
                else
                 puts "Event already exist"
                end #unless close
            end # if close
        end #title-parse do close
    
      rescue OpenURI::HTTPError => e
        if e.message == '500 Internal Server Error'
          next
        else
          raise e
        end
      end #begin close

    end #upto loop close
    
  end #end task
end #end namespace