picturehouse_uk 2.0.5 → 3.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: f3c596d190d8b0e0453014e3f73950904b75582f
-  data.tar.gz: f4b17b74683cd8f29840c4b4005b8132312005e9
+  metadata.gz: 400c5d45b2f84ec815b86ff05d36ba60864aef9e
+  data.tar.gz: 682bc113b527cf7f3ad011a65c048118658a58c5
 SHA512:
-  metadata.gz: 0099e2fa74f59ec702dea26248c800c2338112116965b563cf8c2652b67593f75c870edfd4bcf5fd545b16856f24727b9e0f27495f3565eb3893fe4314a640d5
-  data.tar.gz: 4e7ca06a8d4ef528d547e0bf19219cea236b36b525649fb202d18a7b6291013058cac323cb1833a2e7438da0fbf2ccc15f86ab131466cf7f5d2941bfaa399587
+  metadata.gz: 59b2fb923ff3edea3a4d837cd3615f08e995c0b63f06e3ba114684d32c691c630b16e1fdfc692dbe71047570c0f783b1d6ee11d639bdbf63af755cbdfe60f0fb
+  data.tar.gz: ad018b472fbc9680f36aafc708be483f61572e986a1e3477068a1f0d17d71847285863e49db6af0512a3e1c6988ebe1238f411ed5ce8704fd75931de3746b1ae

data/CHANGELOG.md

@@ -1,13 +1,18 @@
 # Change Log
 All notable changes to this project will be documented in this file.
 
-## v.2.0.5 -2015-01-03
+## [3.0.0] - 2015-02-15
+
+### Fixed
+- the whole damn parsing, we now look at picturehouses.com
+
+## [2.0.5] - 2015-01-03
 
 ### Added
 - parents and babies screening title
 - subtitled screening
 
-## v.2.0.4 -2015-01-01
+## [2.0.4] - 2015-01-01
 
 ### Added
 - remove imax from film title
@@ -15,14 +20,14 @@ All notable changes to this project will be documented in this file.
 - remove 'toddler time' from film title
 - remove 'singalong' from film title
 
-## v.2.0.3 -2014-10-19
+## [2.0.3] - 2014-10-19
 
 ### Added
 - Better Q&A/panel removal from titles
 - deal with various title edge cases
 - remove years from titles
 
-## v.2.0.2 -2014-10-19
+## [2.0.2] - 2014-10-19
 
 ### Added
 - No screenings in York for 'basement events'
@@ -31,7 +36,7 @@ All notable changes to this project will be documented in this file.
 - deal with ROH Encore
 - better spacing on 'event' titles
 
-## v.2.0.1 -2014-10-19
+## [2.0.1] - 2014-10-19
 
 ### Added
 - Better parsing of ampersands in film titles
@@ -41,7 +46,7 @@ All notable changes to this project will be documented in this file.
 - Better parsing of rogue screening types
 - Better parsing of free screening types
 
-## 2.0.0 - 2014-10-16
+## [2.0.0] - 2014-10-16
 
 ### Added
 - Live testing rake task

data/README.md

@@ -1,6 +1,6 @@
 # PicturehouseUk
 
-A simple gem to parse the [Picturehouse Cinemas UK website](http://picturehouses.co.uk) and spit out useful formatted info.
+A simple gem to parse the [Picturehouse Cinemas UK website](http://picturehouses.com) and spit out useful formatted info.
 
 [![Gem Version](https://badge.fury.io/rb/picturehouse_uk.png)](http://badge.fury.io/rb/picturehouse_uk)
 [![Code Climate](https://codeclimate.com/github/andycroll/picturehouse_uk.png)](https://codeclimate.com/github/andycroll/picturehouse_uk)
@@ -53,6 +53,14 @@ cinema.screenings
 
 ```
 
+## Running Tests
+
+To redownload the test URLs run
+
+```
+ruby test/fixture_updater.rb
+```
+
 ## Contributing
 
 1. Fork it

data/lib/picturehouse_uk.rb

@@ -4,9 +4,8 @@ require 'tzinfo/data'
 
 require_relative './picturehouse_uk/version'
 
-require_relative './picturehouse_uk/internal/address_parser'
-require_relative './picturehouse_uk/internal/cinema_page'
-require_relative './picturehouse_uk/internal/film_with_screenings_parser'
+require_relative './picturehouse_uk/internal/parser/address'
+require_relative './picturehouse_uk/internal/parser/screenings'
 require_relative './picturehouse_uk/internal/title_sanitizer'
 require_relative './picturehouse_uk/internal/website'
 

data/lib/picturehouse_uk/cinema.rb

@@ -2,9 +2,9 @@ module PicturehouseUk
   # The object representing a cinema on the Picturehouse UK website
   class Cinema
     # address css
-    ADDRESS_CSS = '.box6 .txt6'
+    ADDRESS_CSS = '.static-content #contact-us + p:first'
     # cinema link css
-    CINEMA_LINKS_CSS = '#cinemalisthome .cinemas a'
+    CINEMA_LINKS_CSS = '.footer .col-sm-3 option + option'
 
     # @return [String] the brand of the cinema
     attr_reader :brand
@@ -25,7 +25,7 @@ module PicturehouseUk
       @name  = options[:name]
       @slug  = @name.downcase.gsub(/[^0-9a-z ]/, '').gsub(/\s+/, '-')
       @url   = if options[:url][0] == '/'
-                 "http://www.picturehouses.co.uk#{options[:url]}"
+                 "http://www.picturehouses.com#{options[:url]}"
                else
                  options[:url]
                end
@@ -59,6 +59,7 @@ module PicturehouseUk
     #         street_address:   '44-47 Gardner Street',
     #         extended_address: 'North Laine',
     #         locality:         'Brighton',
+    #         region:           'East Sussex',
     #         postal_code:      'BN1 1UN',
     #         country_name: 'United Kingdom'
     #       }
@@ -121,6 +122,17 @@ module PicturehouseUk
       address[:postal_code]
     end
 
+    # The region (county) of the cinema
+    # @return [String]
+    # @example
+    #   cinema = PicturehouseUk::Cinema.find('Dukes_At_Komedia')
+    #   cinema.region
+    #   #=> 'East Sussex'
+    # @note Uses the standard method naming as at http://microformats.org/wiki/adr
+    def region
+      address[:region]
+    end
+
     # All planned screenings
     # @return [Array<PicturehouseUk::Screening>]
     # @example
@@ -157,20 +169,20 @@ module PicturehouseUk
     end
 
     def self.new_from_link(link)
-      url = link.get_attribute('href')
-      new(
-        id: url.match(%r{/cinema/(.+)/$})[1],
-        name: link.css('span:nth-child(2)').text,
-        url:  url
-      )
+      url =  link.get_attribute('data-href')
+      name = link.children.first.to_s.split(' — ')[1]
+
+      new(id:   url.match(%r{/cinema/(.+)$})[1],
+          name: name,
+          url:  url)
     end
 
     def address_node
-      @address_node ||= contact_us_doc.css(ADDRESS_CSS)
+      @address_node ||= info_doc.css(ADDRESS_CSS)
     end
 
-    def contact_us_doc
-      @contact_us_doc ||= Nokogiri::HTML(self.class.website.contact_us(id))
+    def info_doc
+      @info_doc ||= Nokogiri::HTML(self.class.website.info(id))
     end
   end
 end

data/lib/picturehouse_uk/film.rb

@@ -19,9 +19,7 @@ module PicturehouseUk
     # @param [String] cinema_id the id of the cinema
     # @return [Array<PicturehouseUk::Film>]
     def self.at(cinema_id)
-      cinema_page(cinema_id).film_html.map do |html|
-        new(Internal::FilmWithScreeningsParser.new(html).film_name)
-      end.uniq
+      screenings(cinema_id).map { |hash| new hash[:film_name] }.uniq
     end
 
     # Allows sort on objects
@@ -54,8 +52,8 @@ module PicturehouseUk
 
     private
 
-    def self.cinema_page(cinema_id)
-      @cinema_page ||= PicturehouseUk::Internal::CinemaPage.new(cinema_id)
+    def self.screenings(cinema_id)
+      PicturehouseUk::Internal::Parser::Screenings.new(cinema_id).to_a
     end
   end
 end

data/lib/picturehouse_uk/internal/parser/address.rb

@@ -0,0 +1,37 @@
+module PicturehouseUk
+  # @api private
+  module Internal
+    # Parses a chunk of HTML to derive address
+    class AddressParser
+      # @param [String] node the HTML to parse into an address
+      # @return [PicturehouseUk::Internal::AddressParser]
+      def initialize(html)
+        @html = html
+      end
+
+      # @return [Hash] contains :street_address, :extended_address, :locality,
+      # :postal_code, :country
+      # @note Uses the address naming from http://microformats.org/wiki/adr
+      def address
+        {
+          street_address: array[1],
+          extended_address: array.length > 5 ? array[2] : nil,
+          locality: town,
+          region: array[-2] == town ? nil : array[-2],
+          postal_code: array[-1],
+          country: 'United Kingdom'
+        }
+      end
+
+      private
+
+      def town
+        @town ||= array[0].split(', ')[-1]
+      end
+
+      def array
+        @array ||= @html.gsub(/\<.?p.?\>/, '').split('<br>')
+      end
+    end
+  end
+end

data/lib/picturehouse_uk/internal/parser/screenings.rb

@@ -0,0 +1,117 @@
+module PicturehouseUk
+  # @api private
+  module Internal
+    # @api private
+    module Parser
+      # Parses screenings page into an array of hashes for an individual cinema
+      class Screenings < Struct.new(:cinema_id)
+        # css for a day of films & screenings
+        LISTINGS = '#this-week .listings > li, #further-ahead .listings > li'
+
+        # parse the cinema page into an array of screenings attributes
+        # @return [Array<Hash>] 
+        def to_a
+          date = nil
+          doc.css(LISTINGS).each_with_object([]) do |node, result|
+            if node.attribute('title') != nil
+              date = Date.parse(node.attribute('title').value)
+            else
+              result << FilmWithShowtimes.new(node, date).to_a
+            end
+          end.flatten
+        end
+
+        private
+
+        def doc
+          @doc ||= Nokogiri::HTML(page)
+        end
+
+        def page
+          @page ||= PicturehouseUk::Internal::Website.new.cinema(cinema_id)
+        end
+      end
+    end
+
+    class FilmWithShowtimes < Struct.new(:node, :date)
+      # film name css
+      NAME = '.top-mg-sm a'
+      # variants css
+      VARIANTS = '.film-times .col-xs-10'
+
+      # The film name
+      # @return [String]
+      def name
+        TitleSanitizer.new(raw_name).sanitized
+      end
+
+      # Showings hashes
+      # @return [Array<Hash>]
+      def to_a
+        Array(node.css(VARIANTS)).flat_map do |variant|
+          Variant.new(variant, date).to_a.map do |hash|
+            {
+              film_name: name,
+              dimension: dimension
+            }.merge(hash)
+          end
+        end
+      end
+
+      private
+
+      def dimension
+        raw_name.match(/3d/i) ? '3d' : '2d'
+      end
+
+      def raw_name
+        @raw_name ||= node.css(NAME).children.first.to_s
+      end
+    end
+
+    class Variant < Struct.new(:node, :date)
+      SHOWTIMES = 'a'
+      VARIENT   = '.film-type-desc'
+
+      def to_a
+        node.css(SHOWTIMES).map do |node|
+          { variant: variant }.merge(Showtime.new(node, date).to_hash)
+        end
+      end
+
+      private
+
+      def variant
+        @variant ||= nil
+      end
+    end
+
+    # parse an individual screening node
+    class Showtime < Struct.new(:node, :date)
+      # the attributes of a single screening
+      # @return [Hash]
+      # @example
+      #   PicturehouseUk::Internal::ScreeningParser.new(html).to_hash
+      #   => {
+      #        booking_url: 'http://...',
+      #        time:        <Time>,
+      #        variant:     ['imax']
+      #      }
+      def to_hash
+        {
+          booking_url: node['href'],
+          time:        time
+        }
+      end
+
+      private
+
+      def time
+        @time ||= begin
+          hour, min = node.text.split('.').map(&:to_i)
+          date.to_time + (hour * 60 + min) * 60
+        end
+      end
+    end
+  end
+end

data/lib/picturehouse_uk/internal/title_sanitizer.rb

@@ -9,6 +9,7 @@ module PicturehouseUk
         /\s+\(?[23][dD]\)?/,                # 2d or 3d from title
         /\bIMAX\b/i,                        # imax from title
         /\s\[NO CERT\]/,                    # no certificate
+        /\s\[TBC\]/,                        # tbc
         /\s\[\]/,                           # blank certificate
         /ourscreen\: /,                     # ourscreen
         /\s\(Re(\: \d{0,4})?\)/i,           # Re-release
@@ -32,6 +33,7 @@ module PicturehouseUk
         /\bsingalong\b/i,                   # singalong
         /\s+\-?\s*Parents \& Babies/i,      # parents and babies
         /Subtitled\:*\s*/i,                 # subtitled
+        /\(?live\)?\z/i,                    # live
       ]
 
       # regexes and their replacements

data/lib/picturehouse_uk/internal/website.rb

@@ -8,13 +8,13 @@ module PicturehouseUk
       # get the cinema page with showings for passed id
       # @return [String]
       def cinema(id)
-        get("cinema/#{id}/")
+        get("cinema/#{id}")
       end
 
       # get the cinema contact information page for passed id
       # @return [String]
-      def contact_us(id)
-        get("cinema/#{id}/Hires_Info/Contact_Us/")
+      def info(id)
+        get("cinema/info/#{id}")
       end
 
       # get the home page
@@ -23,16 +23,10 @@ module PicturehouseUk
         get(nil)
       end
 
-      # get the cinema page containing all upcoming films and screenings
-      # @return [String]
-      def whatson(id)
-        get("whatson?cinema=#{id}")
-      end
-
       private
 
       def get(path)
-        open("http://www.picturehouses.co.uk/#{path}").read
+        open("http://www.picturehouses.com/#{path}").read
       end
     end
   end

data/lib/picturehouse_uk/screening.rb

@@ -25,9 +25,9 @@ module PicturehouseUk
     # @param [String] cinema_id the id of the cinema
     # @return [Array<PicturehouseUk::Screening>]
     def self.at(cinema_id)
-      cinema_page(cinema_id).film_html.map do |html|
-        create_for_single_film(html, cinema_id)
-      end.flatten
+      screenings(cinema_id).map do |attributes|
+        new cinema_hash(cinema_id).merge(attributes)
+      end.uniq
     end
 
     # The UTC time of the screening
@@ -63,18 +63,8 @@ module PicturehouseUk
       }
     end
 
-    def self.cinema_page(cinema_id)
-      PicturehouseUk::Internal::CinemaPage.new(cinema_id)
-    end
-
-    def self.create_for_single_film(html, cinema_id)
-      screenings_parser(html).to_a.map do |attributes|
-        new cinema_hash(cinema_id).merge(attributes)
-      end
-    end
-
-    def self.screenings_parser(html)
-      PicturehouseUk::Internal::FilmWithScreeningsParser.new(html)
+    def self.screenings(cinema_id)
+      PicturehouseUk::Internal::Parser::Screenings.new(cinema_id).to_a
     end
   end
 end

data/lib/picturehouse_uk/version.rb

@@ -1,6 +1,6 @@
 # Ruby interface for http://www.picturehouses.co.uk
-# @version 2.0.5
+# @version 3.0.0
 module PicturehouseUk
   # Gem version
-  VERSION = '2.0.5'
+  VERSION = '3.0.0'
 end