picturehouse_uk 1.0.2 → 2.0.0

data/lib/picturehouse_uk/internal/cinema_page.rb

@@ -0,0 +1,35 @@
+module PicturehouseUk
+  # @api private
+  module Internal
+    # Parses a chunk of HTML to derive movie showing data
+    class CinemaPage
+      # css for a film + screenings
+      FILM_CSS = '#events .largelist .item'
+
+      # @param [Integer] cinema_id cineworld cinema id
+      def initialize(cinema_id)
+        @cinema_id = cinema_id
+      end
+
+      # break up the page into individual chunks for each film
+      # @return [Array<String>] html chunks for a film and it's screenings
+      def film_html
+        film_nodes.map { |n| n.to_s.gsub(/^\s+/, '') }
+      end
+
+      private
+
+      def cinema
+        @cinema ||= PicturehouseUk::Internal::Website.new.cinema(@cinema_id)
+      end
+
+      def cinema_doc
+        @cinema_doc ||= Nokogiri::HTML(cinema)
+      end
+
+      def film_nodes
+        cinema_doc.css(FILM_CSS)
+      end
+    end
+  end
+end

data/lib/picturehouse_uk/internal/film_with_screenings_parser.rb

@@ -1,59 +1,109 @@
 module PicturehouseUk
-
-  # Internal utility classes: Do not use
   # @api private
   module Internal
-
     # Parses a chunk of HTML to derive movie showing data
     class FilmWithScreeningsParser
+      # film name css
+      FILM_NAME_CSS = '.movielink'
+      # showings css
+      SCREENING_CSS = 'a[epoch]'
 
-      # @param [String] film_html a chunk of html
-      def initialize(film_html)
-        @nokogiri_html = Nokogiri::HTML(film_html)
+      # @param [Nokogiri::HTML] film_html a chunk of html
+      def initialize(html)
+        @html = html
       end
 
       # The film name
       # @return [String]
       def film_name
-        name = @nokogiri_html.css('.movielink').children.first.to_s
+        TitleSanitizer.new(raw_film_name).sanitized
+      end
+
+      # Showings hashes
+      # @return [Array<Hash>]
+      def to_a
+        return [] unless screenings?
+        screening_nodes.map do |node|
+          {
+            film_name: film_name,
+            dimension: dimension
+          }.merge(ScreeningParser.new(node).to_hash)
+        end
+      end
+
+      private
+
+      def dimension
+        raw_film_name.match(/3d/i) ? '3d' : '2d'
+      end
+
+      def doc
+        @doc ||= Nokogiri::HTML(@html)
+      end
+
+      def raw_film_name
+        @raw_film_name ||= doc.css(FILM_NAME_CSS).children.first.to_s
+      end
 
-        # screening types
-        name = name.gsub /\s\[(AS LIVE: )?[ACPGU1258]+\]/, '' # remove certificate
-        name = name.gsub /\s\[NO CERT\]/, '' # remove no certificate
-        name = name.gsub /\s\[\]/, '' # remove no certificate
-        name = name.gsub /\s+[23][dD]/, '' # remove 2d or 3d from title
+      def screening_nodes
+        @screening_nodes ||= doc.css(SCREENING_CSS)
+      end
 
-        # special screenings
-        name = name.gsub 'ROH. Live:', 'Royal Opera House:' # fill out Royal Opera House
-        name = name.gsub 'Met. Encore:', 'Met Opera:' # fill out Met Opera
-        name = name.gsub 'NT Encore:', 'National Theatre:' # National theatre
-        name = name.gsub 'RSC Live:', 'Royal Shakespeare Company:' # RSC
-        name = name.gsub 'RSC Encore:', 'Royal Shakespeare Company:' # RSC
+      def screenings?
+        !!screening_nodes
+      end
+    end
 
-        name = name.squeeze(' ') # spaces compressed
-        name = name.gsub /\A\s+/, '' # remove leading spaces
-        name = name.gsub /\s+\z/, '' # remove trailing spaces
+    # parse an individual screening node
+    class ScreeningParser
+      # @param [Nokogiri::HTML] node a node with a film screening
+      def initialize(node)
+        @node = node
       end
 
-      # Showings
+      # is the screening bookable?
+      # @return [Boolean]
+      def bookable?
+        !!booking_url
+      end
+
+      # the attributes of a single screening
       # @return [Hash]
       # @example
-      #   {
-      #     "2D" => [Time.utc, Time.utc]
-      #   }
-      def showings
-        @nokogiri_html.css('a[epoch]').inject({}) do |result, link|
-          key = case link['class']
-            when /big_scream/ then 'baby'
-            when /kids_club|toddler_time/ then 'kids'
-            when /silver_screen/ then 'silver'
-            when /subtitled_cinema/ then 'subtitled'
-            else '2d'
-          end
-          # this is a hack because Time.at() only uses local time
-          time = Time.utc(1970)+link['epoch'].to_i
+      #   Cineworld::Internal::ScreeningParser.new(html).to_hash
+      #   => {
+      #        booking_url: 'http://...',
+      #        time:        <Time>,
+      #        variant:     ['imax']
+      #      }
+      def to_hash
+        {
+          booking_url: "http://www.picturehouses.co.uk#{booking_url}",
+          time:        time,
+          variant:     variant
+        }
+      end
 
-          result.merge(key => (result[key] || []) << time)
+      private
+
+      def booking_url
+        @booking_url ||= @node['html']
+      end
+
+      def time
+        @time ||= Time.utc(1970) + @node['epoch'].to_i
+      end
+
+      def variant
+        @variant ||= begin
+          case @node['class']
+          when /big_scream/ then ['baby']
+          when /kids_club|toddler_time/ then ['kids']
+          when /silver_screen/ then ['silver']
+          when /subtitled_cinema/ then ['subtitled']
+          else
+            []
+          end
         end
       end
     end

data/lib/picturehouse_uk/internal/title_sanitizer.rb

@@ -0,0 +1,49 @@
+module PicturehouseUk
+  # @api private
+  module Internal
+    # Sanitize and standardize film titles
+    class TitleSanitizer
+      # strings and regex to be removed
+      REMOVE = [
+        /\s\[(AS LIVE: )?[ACPGU1258]+\]/, # regular certificate
+        /\s+[23][dD]/,                    # 2d or 3d from title
+        /\s\[NO CERT\]/,                  # no certificate
+        /\s\[\]/,                         # blank certificate
+        /ourscreen\: /,                   # ourscreen
+        /\s\(Re(\: \d{0,4})?\)/i,         # Re-release
+        /\s\[CERT TBC\]/,                 # certificate TBC
+      ]
+
+      # regexes and their replacements
+      REPLACE = {
+        /Met\.? Encore: (.*)/ => 'Met Opera:',
+        /Met\.? Opera: (.*)/  => 'Met Opera: ',
+        /NT Encore: (.*)/     => 'National Theatre:',
+        /NT Live: (.*)/       => 'National Theatre:',
+        /ROH\.? Live: (.*)/   => 'Royal Opera House:',
+        /RSC\.? Live: (.*)/   => 'Royal Shakespeare Company:',
+        /RSC\.? Encore: (.*)/ => 'Royal Shakespeare Company:'
+      }
+
+      # @param [String] title a film title
+      def initialize(title)
+        @title = title
+      end
+
+      # sanitized and standardized title
+      # @return [String] title
+      def sanitized
+        @sanitzed ||= begin
+          sanitized = @title
+          REMOVE.each do |pattern|
+            sanitized.gsub! pattern, ''
+          end
+          REPLACE.each do |pattern, prefix|
+            sanitized.gsub!(pattern) { |_| prefix + $1 }
+          end
+          sanitized.squeeze(' ').strip
+        end
+      end
+    end
+  end
+end

data/lib/picturehouse_uk/internal/website.rb

@@ -0,0 +1,39 @@
+require 'open-uri'
+
+module PicturehouseUk
+  # @api private
+  module Internal
+    # fetches pages from the picturehouse.co.uk website
+    class Website
+      # get the cinema page with showings for passed id
+      # @return [String]
+      def 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/")
+      end
+
+      # get the home page
+      # @return [String]
+      def home
+        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
+      end
+    end
+  end
+end

data/lib/picturehouse_uk/screening.rb

@@ -1,36 +1,80 @@
 module PicturehouseUk
-
   # A single screening of a film on the Picturehouse UK website
   class Screening
-
+    # @return [String] the booking URL on the cinema website
+    attr_reader :booking_url
     # @return [String] the cinema name
     attr_reader :cinema_name
+    # @return [String] 2d or 3d
+    attr_reader :dimension
     # @return [String] the film name
     attr_reader :film_name
-    # @return [Time] the UTC time of the screening
-    attr_reader :when
-    # @return [String] the type of screening (2D, 3D, IMAX...)
-    attr_reader :variant
 
-    # @param [String] film_name the film name
-    # @param [String] cinema_name the cinema name
-    # @param [Time] time datetime of the screening (UTC preferred)
-    # @param [String] variant the type of showing (e.g. 3d/baby/live)
-    def initialize(film_name, cinema_name, time, variant=nil)
-      @cinema_name, @film_name, @variant = cinema_name, film_name, variant
-      @when = time.utc? ? time : TZInfo::Timezone.get('Europe/London').local_to_utc(time)
+    # @param [Hash] options
+    def initialize(options)
+      @booking_url = options.fetch(:booking_url, nil)
+      @cinema_name = options.fetch(:cinema_name)
+      @cinema_id   = options.fetch(:cinema_id)
+      @dimension   = options.fetch(:dimension, '2d')
+      @film_name   = options.fetch(:film_name)
+      @time        = options.fetch(:time)
+      @variant     = options.fetch(:variant, [])
+    end
+
+    # Screenings at a single cinema
+    # @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
+    end
+
+    # The UTC time of the screening
+    # @return [Time]
+    def showing_at
+      @showing_at ||= begin
+        if @time.utc?
+          @time
+        else
+          TZInfo::Timezone.get('Europe/London').local_to_utc(@time)
+        end
+      end
     end
 
     # The date of the screening
     # @return [Date]
-    def date
-      @when.to_date
+    def showing_on
+      showing_at.to_date
+    end
+
+    # The kinds of screening
+    # @return <Array[String]>
+    def variant
+      @variant.map(&:downcase).sort
+    end
+
+    private
+
+    def self.cinema_hash(cinema_id)
+      {
+        cinema_id: cinema_id,
+        cinema_name: PicturehouseUk::Cinema.find(cinema_id).name
+      }
+    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
 
-    # @deprecated Please use {#variant} instead, I can't spell
-    def varient
-      warn "Please use #variant instead, I can't spell"
-      variant
+    def self.screenings_parser(html)
+      PicturehouseUk::Internal::FilmWithScreeningsParser.new(html)
     end
   end
 end

data/lib/picturehouse_uk/version.rb

@@ -1,6 +1,6 @@
 # Ruby interface for http://www.picturehouses.co.uk
-# @version 1.0.2
+# @version 2.0.0
 module PicturehouseUk
   # Gem version
-  VERSION = "1.0.2"
+  VERSION = '2.0.0'
 end

data/picturehouse_uk.gemspec

@@ -4,25 +4,25 @@ $LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
 require 'picturehouse_uk/version'
 
 Gem::Specification.new do |spec|
-  spec.name          = "picturehouse_uk"
+  spec.name          = 'picturehouse_uk'
   spec.version       = PicturehouseUk::VERSION
-  spec.authors       = ["Andy Croll"]
-  spec.email         = ["andy@goodscary.com"]
-  spec.description   = %q{An API to pull movie information from the picturehouse.co.uk website}
-  spec.summary       = %q{It's a scraper, but a nice one}
-  spec.homepage      = ""
-  spec.license       = "MIT"
+  spec.authors       = ['Andy Croll']
+  spec.email         = ['andy@goodscary.com']
+  spec.description   = 'An API to pull movie information from the picturehouse.co.uk website'
+  spec.summary       = "It's a scraper, but a nice one"
+  spec.homepage      = ''
+  spec.license       = 'MIT'
 
   spec.files         = `git ls-files`.split($/)
   spec.executables   = spec.files.grep(%r{^bin/}) { |f| File.basename(f) }
   spec.test_files    = spec.files.grep(%r{^(test|spec|features)/})
-  spec.require_paths = ["lib"]
+  spec.require_paths = ['lib']
 
-  spec.add_development_dependency "bundler", "~> 1.3"
-  spec.add_development_dependency "rake"
+  spec.add_development_dependency 'bundler', '~> 1.3'
+  spec.add_development_dependency 'minitest-reporters'
+  spec.add_development_dependency 'rake'
   spec.add_development_dependency 'webmock'
 
-  spec.add_runtime_dependency 'httparty'
   spec.add_runtime_dependency 'nokogiri'
   spec.add_runtime_dependency 'tzinfo'
   spec.add_runtime_dependency 'tzinfo-data'

data/test/fixture_updater.rb

@@ -0,0 +1,43 @@
+require File.expand_path('../../lib/picturehouse_uk.rb', __FILE__)
+
+def fixture(name)
+  File.expand_path("../fixtures/#{name}.html", __FILE__)
+end
+
+File.open(fixture('home'), 'w') do |file|
+  puts '* Homepage'
+  file.write PicturehouseUk::Internal::Website.new.home
+end
+
+# DUKE OF YORKS
+
+File.open(fixture('cinema/Duke_Of_Yorks'), 'w') do |file|
+  puts '* Duke of Yorks'
+  file.write PicturehouseUk::Internal::Website.new.cinema('Duke_Of_Yorks')
+end
+
+File.open(fixture('contact_us/Duke_Of_Yorks'), 'w') do |file|
+  puts '* Duke of Yorks Information'
+  file.write PicturehouseUk::Internal::Website.new.contact_us('Duke_Of_Yorks')
+end
+
+# KOMEDIA
+
+File.open(fixture('contact_us/Dukes_At_Komedia'), 'w') do |file|
+  puts '* Dukes at Komedia Information'
+  file.write PicturehouseUk::Internal::Website.new.contact_us('Dukes_At_Komedia')
+end
+
+# FILMS
+
+page = PicturehouseUk::Internal::CinemaPage.new('Duke_Of_Yorks')
+
+File.open(fixture('cinema/Duke_Of_Yorks/film_second'), 'w') do |file|
+  puts '* Duke of Yorks Second Film'
+  file.write page.film_html[1]
+end
+
+File.open(fixture('cinema/Duke_Of_Yorks/film_last'), 'w') do |file|
+  puts '* Duke of Yorks Last Film'
+  file.write page.film_html[-1]
+end