picturehouse_uk 1.0.2 → 2.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 944f2d9ed771bfd8d95ceea29fda42f77346ef03
-  data.tar.gz: 87cd57db1cbe6ffc6ceb949332749ed94180620a
+  metadata.gz: 852cffd41d42e302152ccedd703e8f265b48015f
+  data.tar.gz: a2cfd6c6dbb93d841cc2d7c22423d9228363d342
 SHA512:
-  metadata.gz: 398e86805f4619f29d567599c775b7f5e101806809ccac053b6b3098ccadccf2a8f2872019e64d6bed8cfbca4a722cfd97b7038e05ec97e87205eb08ca65d954
-  data.tar.gz: b37051a1d76065d43c293c115abd37d0e1f77b32e932c139b8600ee23c698ac26235ce92b2b25ca276ea71db44fa3021197817c5de812a897ab631dcc0e0964c
+  metadata.gz: a5da4c31df5e98af0f65e4261eca2de02da13e6f5f80ab5a75171c7b393537e801bab7740cffdfee6ce39caa6994165c3ddc4c91e28ba3f26367590c31863de3
+  data.tar.gz: 36e769a6373c03914f241ca65babc5e878552902627d53b6123bdc99918a605009ba0dcc41815933139c918925d7d84be828496f2dfb0853c051e9efcde3f429

data/.travis.yml

@@ -1,4 +1,5 @@
 language: ruby
 rvm:
-  - 1.9.3
   - 2.0.0
+  - 2.1.3
+script: "rake && rake live"

data/CHANGELOG.md

@@ -0,0 +1,18 @@
+# Change Log
+All notable changes to this project will be documented in this file.
+
+## 2.0.0 - 2014-10-16
+
+### Added
+- Live testing rake task
+- Ruby 2.1.3 on Travis
+
+### Changed
+- Internal structure of parsers
+- Title sanitization
+- Broke up monolith of Cinema class
+- Fixture update
+- Rubocop
+
+### Removed
+- Only the badness

data/README.md

@@ -5,6 +5,7 @@ A simple gem to parse the [Picturehouse Cinemas UK website](http://picturehouses
 [![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)
 [![Build Status](https://travis-ci.org/andycroll/picturehouse_uk.png?branch=master)](https://travis-ci.org/andycroll/picturehouse_uk)
+[![Inline docs](http://inch-ci.org/github/andycroll/picturehouse_uk.png)](http://inch-ci.org/github/andycroll/picturehouse_uk)
 
 ## Installation
 
@@ -26,12 +27,13 @@ Or install it yourself as:
 
 ``` ruby
 PicturehouseUK::Cinema.all
-#=> [<PicturehouseUK::Cinema brand="Picturehouse" name="Duke's At Komedia" slug="dukes-at-komedia" chain_id="Dukes_At_Komedia" url="...">, #=> <PicturehouseUK::Cinema brand="Picturehouse" name="Duke o York's" slug="duke-of-yorks" chain_id="Duke_Of_Yorks" url="...">, ...]
+#=> [
+      <PicturehouseUK::Cinema brand="Picturehouse" name="Duke's at Komedia" slug="dukes-at-komedia" chain_id="Dukes_At_Komedia" url="...">,
+      <PicturehouseUK::Cinema brand="Picturehouse" name="Duke of York's" slug="duke-of-yorks" chain_id="Duke_Of_Yorks" url="...">,
+      ...
+    ]
 
-PicturehouseUK::Cinema.find_by_id('Duke_Of_Yorks')
-#=> <PicturehouseUK::Cinema brand="Picturehouse" name="Duke_Of_Yorks" slug="duke-of-yorks" address="..." chain_id="Duke_Of_Yorks" url="...">
-
-cinema = PicturehouseUK::Cinema.find_by_slug('duke-of-yorks')
+PicturehouseUK::Cinema.find('Duke_Of_Yorks')
 #=> <PicturehouseUK::Cinema brand="Picturehouse" name="Duke_Of_Yorks" slug="duke-of-yorks" address="..." chain_id="Duke_Of_Yorks" url="...">
 
 cinema.brand
@@ -49,11 +51,6 @@ cinema.films
 cinema.screenings
 #=> [<PicturehouseUK::Screening film="About Time" when="2013-09-09 11:00 UTC" varient="3d">, <PicturehouseUK::Screening film="Iron Man 3" when="2013-09-09 13:50 UTC" varient="kids">, <PicturehouseUK::Screening ..>, <PicturehouseUK::Screening ...>]
 
-cinema.screenings_of 'Iron Man 3'
-#=> [<PicturehouseUK::Screening film="Iron Man 3" when="2013-09-09 11:00 UTC" varient="3d">, <PicturehouseUK::Screening film="Iron Man 3" when="2013-09-09 13:50 UTC" varient="kids">]
-
-cinema.screenings_of <PicturehouseUK::Film name="Iron Man 3">
-#=> [<PicturehouseUK::Screening film="Iron Man 3" when="2013-09-09 11:00 UTC" varient="3d">, <PicturehouseUK::Screening film="Iron Man 3" when="2013-09-09 13:50 UTC" varient="kids">]
 ```
 
 ## Contributing

data/Rakefile

@@ -1,20 +1,33 @@
 #!/usr/bin/env rake
-require "bundler/gem_tasks"
+require 'bundler/gem_tasks'
 
 require 'rake/testtask'
 
 Rake::TestTask.new do |t|
   t.libs << 'lib/picturehouse_uk'
-  t.test_files = FileList['test/lib/picturehouse_uk/*_test.rb', 'test/lib/picturehouse_uk/internal/*_test.rb']
+  t.test_files = FileList[
+    'test/lib/picturehouse_uk/*_test.rb',
+    'test/lib/picturehouse_uk/internal/*_test.rb'
+  ]
   t.verbose = true
 end
 
-task :build do
-  system "gem build picturehouse_uk.gemspec"
+Rake::TestTask.new do |t|
+  t.libs << 'lib/picturehouse_uk'
+  t.name = :live
+  t.test_files = FileList[
+    'test/live/*_test.rb'
+  ]
+  t.verbose = true
 end
 
-task :release => :build do
-  system "gem push picturehouse_uk-#{PicturehouseUk::VERSION}"
+# http://erniemiller.org/2014/02/05/7-lines-every-gems-rakefile-should-have/
+task :console do
+  require 'irb'
+  require 'irb/completion'
+  require 'picturehouse_uk'
+  ARGV.clear
+  IRB.start
 end
 
-task :default => :test
+task default: :test

data/lib/picturehouse_uk.rb

@@ -1,11 +1,14 @@
-require 'httparty'
 require 'nokogiri'
 require 'tzinfo'
 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/title_sanitizer'
+require_relative './picturehouse_uk/internal/website'
 
 require_relative './picturehouse_uk/cinema'
 require_relative './picturehouse_uk/film'

data/lib/picturehouse_uk/cinema.rb

@@ -1,7 +1,10 @@
 module PicturehouseUk
-
   # The object representing a cinema on the Picturehouse UK website
   class Cinema
+    # address css
+    ADDRESS_CSS = '.box6 .txt6'
+    # cinema link css
+    CINEMA_LINKS_CSS = '#cinemalisthome .cinemas a'
 
     # @return [String] the brand of the cinema
     attr_reader :brand
@@ -14,27 +17,27 @@ module PicturehouseUk
     # @return [String] the url of the cinema on the Picturehouse website
     attr_reader :url
 
-    # @param [String] id cinema id from the site
-    # @param [String] name cinema name
-    # @param [String] url url on Picturehouse website
+    # @param [Hash] options id, name and url of the cinemas
     # @return [PicturehouseUk::Cinema]
-    def initialize(id, name, url)
+    def initialize(options)
       @brand = 'Picturehouse'
-      @id    = id
-      @name  = name
-      @slug  = name.downcase.gsub(/[^0-9a-z ]/,'').gsub(/\s+/, '-')
-      @url   = (url[0] == '/') ? "http://www.picturehouses.co.uk#{url}" : url
+      @id    = options[:id]
+      @name  = options[:name]
+      @slug  = @name.downcase.gsub(/[^0-9a-z ]/, '').gsub(/\s+/, '-')
+      @url   = if options[:url][0] == '/'
+                 "http://www.picturehouses.co.uk#{options[:url]}"
+               else
+                 options[:url]
+               end
     end
 
     # Return basic cinema information for all cinemas
     # @return [Array<PicturehouseUk::Cinema>]
     # @example
     #   PicturehouseUk::Cinema.all
-    #   # => [<PicturehouseUK::Cinema brand="Picturehouse" name="Duke's At Komedia" slug="dukes-at-komedia" id="Dukes_At_Komedia" url="...">, #=> <PicturehouseUK::Cinema brand="Picturehouse" name="Duke o York's" slug="duke-of-yorks" id="Duke_Of_Yorks" url="...">, ...]
+    #   # => [<PicturehouseUK::Cinema ...>, <PicturehouseUK::Cinema ...>, ...]
     def self.all
-      cinema_links.map do |link|
-        new_from_link link
-      end
+      cinema_links.map { |link| new_from_link(link) }
     end
 
     # Find a single cinema
@@ -42,9 +45,9 @@ module PicturehouseUk
     # @return [PicturehouseUk::Cinema, nil]
     # @example
     #   PicturehouseUk::Cinema.find('Dukes_At_Komedia')
-    #   # => <PicturehouseUK::Cinema brand="Picturehouse" name="Duke's At Komedia" slug="dukes-at-komedia" id="Dukes_At_Komedia" url="...">
+    #   # => <PicturehouseUK::Cinema ...>
     def self.find(id)
-      all.select { |cinema| cinema.id == id }[0]
+      all.find { |cinema| cinema.id == id }
     end
 
     # Address of the cinema
@@ -52,16 +55,16 @@ module PicturehouseUk
     # @example
     #   cinema = PicturehouseUk::Cinema.find('Dukes_At_Komedia')
     #   cinema.adr
-    #   #=> { street_address: '44-47 Gardner Street', extended_address: 'North Laine', locality: 'Brighton', postal_code: 'BN1 1UN', country_name: 'United Kingdom' }
-    # @note Uses the standard method naming as at http://microformats.org/wiki/adr
+    #   #=> {
+    #         street_address:   '44-47 Gardner Street',
+    #         extended_address: 'North Laine',
+    #         locality:         'Brighton',
+    #         postal_code:      'BN1 1UN',
+    #         country_name: 'United Kingdom'
+    #       }
+    # @note Uses method naming as at http://microformats.org/wiki/adr
     def adr
-      {
-        street_address: street_address,
-        extended_address: extended_address,
-        locality: locality,
-        postal_code: postal_code,
-        country: 'United Kingdom'
-      }
+      PicturehouseUk::Internal::AddressParser.new(address_node.to_s).address
     end
     alias_method :address, :adr
 
@@ -71,9 +74,9 @@ module PicturehouseUk
     #   cinema = PicturehouseUk::Cinema.find('Dukes_At_Komedia')
     #   cinema.extended_address
     #   #=> 'North Laine'
-    # @note Uses the standard method naming as at http://microformats.org/wiki/adr
+    # @note Uses method naming as at http://microformats.org/wiki/adr
     def extended_address
-      address_strings.length > 3 ? address_strings[1] : nil
+      address[:extended_address]
     end
 
     # Films with showings scheduled at this cinema
@@ -81,12 +84,9 @@ module PicturehouseUk
     # @example
     #   cinema = PicturehouseUk::Cinema.find('Dukes_At_Komedia')
     #   cinema.films
-    #   # => [<PicturehouseUk::Film name="Iron Man 3">, <PicturehouseUk::Film name="Star Trek Into Darkness">]
+    #   # => [<PicturehouseUk::Film ...>, <PicturehouseUk::Film ...>, ...]
     def films
-      film_nodes.map do |node|
-        parser = PicturehouseUk::Internal::FilmWithScreeningsParser.new node.to_s
-        PicturehouseUk::Film.new parser.film_name
-      end.uniq
+      PicturehouseUk::Film.at(@id)
     end
 
     # The name of the cinema (might include brand)
@@ -107,7 +107,7 @@ module PicturehouseUk
     #   #=> 'Brighton'
     # @note Uses the standard method naming as at http://microformats.org/wiki/adr
     def locality
-      address_strings[-2]
+      address[:locality]
     end
 
     # Post code of the cinema
@@ -118,7 +118,7 @@ module PicturehouseUk
     #   #=> 'BN1 1UN'
     # @note Uses the standard method naming as at http://microformats.org/wiki/adr
     def postal_code
-      address_strings[-1]
+      address[:postal_code]
     end
 
     # All planned screenings
@@ -126,17 +126,9 @@ module PicturehouseUk
     # @example
     #   cinema = PicturehouseUk::Cinema.find('Dukes_At_Komedia')
     #   cinema.screenings
-    #   # => [<PicturehouseUk::Screening film_name="Iron Man 3" cinema_name="Duke's At Komedia" when="..." variant="...">, <PicturehouseUk::Screening ...>]
+    #   # => [<PicturehouseUk::Screening ...>, <PicturehouseUk::Screening ...>]
     def screenings
-      film_nodes.map do |node|
-        parser = PicturehouseUk::Internal::FilmWithScreeningsParser.new node.to_s
-        parser.showings.map do |screening_type, times|
-          times.map do |time|
-            variant = screening_type == '2d' ? nil : screening_type
-            PicturehouseUk::Screening.new parser.film_name, self.name, time, variant
-          end
-        end
-      end.flatten
+      PicturehouseUk::Screening.at(@id)
     end
 
     # The street adress of the cinema
@@ -147,79 +139,38 @@ module PicturehouseUk
     #   #=> '44-47 Gardner Street'
     # @note Uses the standard method naming as at http://microformats.org/wiki/adr
     def street_address
-      address_strings[0]
+      address[:street_address]
     end
 
     private
 
     def self.cinema_links
-      parsed_homepage.css('#cinemalisthome .cinemas a')
+      home_doc.css(CINEMA_LINKS_CSS)
     end
 
-    def self.homepage_response
-      @homepage_response ||= HTTParty.get('http://www.picturehouses.co.uk/')
-    end
-
-    def self.new_from_link(link)
-      url  = link.get_attribute('href')
-      id   = url.match(/\/cinema\/(.+)\/$/)[1]
-      name = link.css('span:nth-child(2)').text
-      new id, name, url
+    def self.home_doc
+      @home_doc ||= Nokogiri::HTML(website.home)
     end
 
-    def self.parsed_homepage
-      Nokogiri::HTML(homepage_response)
+    def self.website
+      @website ||= PicturehouseUk::Internal::Website.new
     end
 
-    def address_parts
-      if pure_address_parts.length > 0 && pure_address_parts[0].match(/\d+\Z/)
-        ["#{pure_address_parts[0]} #{pure_address_parts[1]}"] + pure_address_parts[2..-1]
-      else
-        pure_address_parts
-      end
-    end
-
-    def address_strings
-      if address_parts && address_parts.length > 0
-        address_parts[0..post_code_index]
-      else
-        # this is a horrendous hack for Hackney Picturehouse
-        address_node.css('p').to_s.split('Box Office')[0].split('<br> ')[1..-1]
-      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
+      )
     end
 
     def address_node
-      parsed_contact_us.css('.box6 .txt6')
-    end
-
-    def contact_us_response
-      @contact_us_response ||= HTTParty.get("#{@url}Hires_Info/Contact_Us/")
-    end
-
-    def cinema_response
-      @cinema_response ||= HTTParty.get(@url)
-    end
-
-    def film_nodes
-      parsed_cinema.css('.box8_content .largelist .item')
-    end
-
-    def parsed_cinema
-      Nokogiri::HTML(cinema_response)
-    end
-
-    def parsed_contact_us
-      Nokogiri::HTML(contact_us_response)
-    end
-
-    def post_code_index
-      address_parts.index { |e| e.match /[A-Z]{1,2}\d{1,2}[A-Z]?\s\d{1,2}[A-Z]{1,2}/ }
+      @address_node ||= contact_us_doc.css(ADDRESS_CSS)
     end
 
-    def pure_address_parts
-      @pure_address_parts = address_node.css('.cinemaListBox').map do |e|
-        e.children[0].to_s
-      end.select { |e| e != '' }
+    def contact_us_doc
+      @contact_us_doc ||= Nokogiri::HTML(self.class.website.contact_us(id))
     end
   end
 end

data/lib/picturehouse_uk/film.rb

@@ -1,5 +1,4 @@
 module PicturehouseUk
-
   # A film on the Picturehouse UK website
   class Film
     include Comparable
@@ -13,14 +12,23 @@ module PicturehouseUk
     # @return [PicturehouseUk::Film]
     def initialize(name)
       @name = name
-      @slug = name.downcase.gsub(/[^0-9a-z ]/,'').gsub(/\s+/, '-')
+      @slug = name.downcase.gsub(/[^0-9a-z ]/, '').gsub(/\s+/, '-')
+    end
+
+    # Films at a single cinema
+    # @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
     end
 
     # Allows sort on objects
     # @param [PicturehouseUk::Film] other another film object
     # @return [Integer] -1, 0 or 1
-    def <=> other
-      self.slug <=> other.slug
+    def <=>(other)
+      slug <=> other.slug
     end
 
     # Check an object is the same as another object.
@@ -28,20 +36,26 @@ module PicturehouseUk
     # @return [Boolean] True if both objects are the same exact object, or if
     #   they are of the same type and share an equal slug
     # @note Guided by http://woss.name/2011/01/20/equality-comparison-and-ordering-in-ruby/
-    def eql? other
+    def eql?(other)
       self.class == other.class && self == other
     end
 
     # Generates hash of slug in order to allow two records of the same type and
     # id to work with something like:
     #
-    #   [ Film.new('ABC'), Film.new('DEF') ] & [ Film.new('DEF'), Film.new('GHI') ]
-    #   #=> [ Film.new('DEF') ]
+    #   [ Film.new('AB'), Film.new('EF') ] & [ Film.new('EF'), Film.new('GH') ]
+    #   #=> [ Film.new('EF') ]
     #
     # @return [Integer] hash of slug
     # @note Guided by http://woss.name/2011/01/20/equality-comparison-and-ordering-in-ruby/
     def hash
-      self.slug.hash
+      slug.hash
+    end
+
+    private
+
+    def self.cinema_page(cinema_id)
+      @cinema_page ||= PicturehouseUk::Internal::CinemaPage.new(cinema_id)
     end
   end
 end

data/lib/picturehouse_uk/internal/address_parser.rb

@@ -0,0 +1,72 @@
+module PicturehouseUk
+  # @api private
+  module Internal
+    # Parses a chunk of HTML to derive address
+    class AddressParser
+      # CSS for address lines
+      ADDRESS_LINES_CSS = '.cinemaListBox'
+
+      # regular expression for postal code
+      POSTCODE_REGEX = /[A-Z]{1,2}\d{1,2}[A-Z]?\s\d{1,2}[A-Z]{1,2}/
+
+      # @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[0],
+          extended_address: array.length > 3 ? array[1] : nil,
+          locality: array[-2],
+          postal_code: array[-1],
+          country: 'United Kingdom'
+        }
+      end
+
+      private
+
+      def array
+        @array ||= begin
+          if standard?
+            lines[0..postal_code_index(lines)]
+          else
+            # this is a horrendous hack for Hackney Picturehouse
+            doc.css('p').to_s.split('Box Office')[0].split('<br> ')[1..-1]
+          end
+        end
+      end
+
+      def doc
+        @doc ||= Nokogiri::HTML(@html)
+      end
+
+      def matched_lines
+        @matched_lines ||= begin
+          matched = doc.css(ADDRESS_LINES_CSS).map { |n| n.children[0].to_s }
+          matched.reject { |e| e == '' }
+        end
+      end
+
+      def postal_code_index(array)
+        array.index { |element| element.match(POSTCODE_REGEX) }
+      end
+
+      def lines
+        if matched_lines.length > 0 && matched_lines[0].match(/\d+\Z/) # komedia
+          ["#{matched_lines[0]} #{matched_lines[1]}"] + matched_lines[2..-1]
+        else
+          matched_lines
+        end
+      end
+
+      def standard?
+        lines && lines.length > 0
+      end
+    end
+  end
+end