ratebeer 0.0.8 → 0.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 89147b9ea6840ec56edb38f6d24341f2d15a43dd
-  data.tar.gz: 5a3bbf6f0c78245d5c028f5c65a8dc2b0b5e5e99
+  metadata.gz: 45bf83b47740dbbdbed33401175d33db8917f672
+  data.tar.gz: f5eaa7e062b02b8391d094e8395c57265b3f92e2
 SHA512:
-  metadata.gz: 55bb181d53621951d0a04e7c235723eef68f4be2174dd922d2108339a2b8294b2bdbafdac6908775a9ac6a164be2b1e1be12d1499bd38456787621222536fc84
-  data.tar.gz: a27cacfa4adb464499e38911528af4bbdda4246cf3071237b1016f5edf0309f5052ff539613b9a489b96632a9e25617623ae3244858f8e752d1666f4f4e6449d
+  metadata.gz: fc34adee973d864596e83ee5570c5540a8ff0216c847fc2b9dd2a8b5e4121240ec13a443dfb346b69768be0e9ea0898b35eed18df8349879b0dbd221a3dda6f5
+  data.tar.gz: 3bcf0702e0cad27ef4b2a08da5000b8465043c30d11934d49191078e3df001c66900a09393826f9d42075e1d3381257f146a829bb6c9a3d1f21e5732f8878def

data/Gemfile.lock

@@ -1,7 +1,8 @@
 PATH
   remote: .
   specs:
-    ratebeer (0.0.6)
+    ratebeer (0.0.8)
+      i18n
       nokogiri
 
 GEM

data/lib/ratebeer.rb

@@ -1,19 +1,18 @@
+Dir[File.expand_path('../ratebeer/*.rb', __FILE__)].each { |f| require f }
+
 # RateBeer.com scraper
 #
-# Scrapes required information on beers, breweries, ratings, etc. from the 
+# Scrapes required information on beers, breweries, ratings, etc. from the
 # RateBeer.com beer database.
 #
-Dir[File.expand_path("../ratebeer/*.rb", __FILE__)].each { |f| require f }
-
 module RateBeer
-
   # Create new beer instance, using ID and name passed as arguments.
   #
   # @param [Integer, String] id ID# of beer to retrieve
   # @param [String] name Name of the beer to which ID# relates if known
   # @return [RateBeer::Beer] beer with passed ID#
   #
-  def beer(id, name=nil)
+  def beer(id, name = nil)
     Beer.new(id, name: name)
   end
 
@@ -23,7 +22,7 @@ module RateBeer
   # @param [String] name Name of the brewery to which ID# relates if known
   # @return [RateBeer::Brewery] brewery with passed ID#
   #
-  def brewery(id, name=nil)
+  def brewery(id, name = nil)
     Brewery.new(id, name: name)
   end
 
@@ -33,7 +32,7 @@ module RateBeer
   # @param [String] name Name of the style to which ID# relates if known
   # @return [RateBeer::Style] style with passed ID#
   #
-  def style(id, name=nil)
+  def style(id, name = nil)
     Style.new(id, name: name)
   end
 

data/lib/ratebeer/beer.rb

@@ -11,12 +11,12 @@ module RateBeer
     #
     def self.data_keys
       [:name,
-       :brewery, 
-       :style, 
-       :glassware, 
-       :availability, 
-       :abv, 
-       :calories, 
+       :brewery,
+       :style,
+       :glassware,
+       :availability,
+       :abv,
+       :calories,
        :description,
        :retired,
        :rating]
@@ -25,6 +25,12 @@ module RateBeer
     include RateBeer::Scraping
     include RateBeer::URLs
 
+    # CSS selector for the root element containing beer information.
+    ROOT_SELECTOR = '#container table'.freeze
+
+    # CSS selector for the beer information element.
+    INFO_SELECTOR = 'table'.freeze
+
     # Create RateBeer::Beer instance.
     #
     # Requires the RateBeer ID# for the beer in question.
@@ -37,6 +43,23 @@ module RateBeer
       super
     end
 
+    def doc
+      unless instance_variable_defined?("@doc")
+        @doc = noko_doc(URI.join(BASE_URL, beer_url(id)))
+        validate_beer
+        redirect_if_aliased
+      end
+      @doc
+    end
+
+    def root
+      @root ||= doc.at_css(ROOT_SELECTOR)
+    end
+
+    def info_root
+      @info_root ||= root.at_css(INFO_SELECTOR)
+    end
+
     # Return reviews of this beer.
     #
     def reviews(order: :most_recent, limit: 10)
@@ -45,99 +68,85 @@ module RateBeer
 
     private
 
-    # Retrieve details about this beer from the website.
+    # Redirects this beer to the "proper" beer page if it represents an alias
+    # of another beer.
     #
-    # This method stores the retrieved details in instance variables
-    # of the beer instance.
-    #
-    def retrieve_details
-      doc      = noko_doc(URI.join(BASE_URL, beer_url(id)))
-      root     = doc.css('#container table').first
-      info_tbl = root.css('table').first 
-
-      @name = doc.css("h1")
-                 .text
-                 .strip
-      @name = fix_characters(@name)
-      raise PageNotFoundError.new("Beer not found - #{id}") if name.empty?
-
-      # If this beer is an alias, change ID to that of "proper" beer and
+    # This method overwrites the value of @doc, so that this will scrape the
+    # details of the proper beer, and not the alias.
+    def redirect_if_aliased
       # retrieve details of the proper beer instead.
       alias_pattern = /Also known as(.|\n)*Proceed to the aliased beer\.{3}/
-      if root.css('tr')[1].css('div div').text =~ alias_pattern
-        alias_node = root.css('tr')[1]
-                         .css('div div')
-                         .css('a')
-                         .first
-        alias_name = alias_node.text
-        alias_id   = alias_node['href'].split('/').last.to_i
-        @id = alias_id
-        retrieve_details
-        return nil
+      local_root = doc.at_css(ROOT_SELECTOR)
+      if local_root.css('tr')[1].css('div div').text =~ alias_pattern
+        scrape_name # Set the name to the original, non-aliased beer.
+        alias_node = local_root.css('tr')[1]
+                               .css('div div')
+                               .css('a')
+                               .first
+        @alias_id = alias_node['href'].split('/').last.to_i
+        @doc = noko_doc(URI.join(BASE_URL, beer_url(@alias_id)))
+      end
+    end
+
+    def validate_beer
+      error_message = 'we didn\'t find this beer'
+      if name == error_message
+        raise PageNotFoundError.new("Beer not found - #{id}")
+      end
+    end
+
+    def scrape_name
+      @name ||= fix_characters(doc.css('h1').text.strip)
+    end
+
+    def scrape_brewery
+      brewery_element = doc.at_css("a[itemprop='brand']")
+      brewery_id = id_from_link(brewery_element)
+      brewery_name = fix_characters(brewery_element.text)
+      @brewery = Brewery.new(brewery_id, name: brewery_name)
+    end
+
+    def scrape_style
+      style_element = doc.at_css("a[href^='/beerstyles']")
+      style_id = id_from_link(style_element)
+      style_name = fix_characters(style_element.text)
+      @style = Style.new(style_id, name: style_name)
+    end
+
+    def scrape_glassware
+      glassware_elements = doc.css("a[href^='/ShowGlassware.asp']")
+      @glassware = glassware_elements.map do |el|
+        [:id, :name].zip([el['href'].split('GWID=').last.to_i, el.text]).to_h
       end
+    end
+
+    def scrape_availability
+      raw_info = info_root.css('td')[1]
+                          .css('table')
+                          .css('td')
+                          .children
+                          .children
+                          .map(&:text)
+                          .reject(&:empty?)
+                          .each_slice(2)
+                          .to_a
+                          .tap { |a| a.last.unshift('distribution') }
+                          .map do |(k, v)|
+        [k =~ /bottl/ ? :bottling : symbolize_text(k), v]
+      end
+      @availability = raw_info.to_h.merge(seasonal: scrape_misc[:seasonal])
+    end
+
+    def scrape_abv
+      @abv = scrape_misc[:abv]
+    end
+
+    def scrape_calories
+      @calories = scrape_misc[:est_calories]
+    end
 
-      @brewery = info_tbl.css('td')[1]
-                         .css('div')
-                         .first
-                         .css('a')
-                         .map { |a| [:id, 
-                                     :name].zip([a['href'].split('/')
-                                                          .last
-                                                          .to_i, a.text]).to_h }.first
-      @brewery = Brewery.new(@brewery[:id], name: fix_characters(@brewery[:name]))
-      @style = info_tbl.css('td')[1]
-                       .css('div')
-                       .first
-                       .css('a')
-                       .select { |a| a['href'] =~ /beerstyles/ }
-                       .map { |a| [:id, 
-                                   :name].zip([a['href'].split('/')
-                                                        .last
-                                                        .to_i, a.text]).to_h }.first
-      @style = Style.new(@style[:id], name: fix_characters(@style[:name]))
-      @glassware = info_tbl.css('td')[1]
-                          .css('div')[1]
-                          .css('a')
-                          .map { |a| [:id, 
-                                      :name].zip([a['href'].split('GWID=')
-                                                           .last
-                                                           .to_i, a.text]).to_h }.first
-      misc = info_tbl.next_element
-                     .first_element_child
-                     .children 
-                     .map(&:text)
-                     .flat_map { |x| x.gsub(nbsp, ' ').strip.split(':') }
-                     .map(&:strip)
-                     .reject(&:empty?)
-                     .each_slice(2)
-                     .map { |(k, v)| [symbolize_text(k),
-                                      v.to_f.zero? ? v : v.to_f] }
-                     .to_h
-      @abv      = misc[:abv]
-      @calories = misc[:est_calories]
-      @rating   = [:overall, 
-                   :style].zip(info_tbl.css('div')
-                          .select { |d| d['title'] =~ /This figure/ }
-                          .map    { |d| d['title'].split(':').first.to_f }).to_h
-      @rating.merge!({ ratings:       misc[:ratings], 
-                       weighted_avg:  misc[:weighted_avg],
-                       mean:          misc[:mean] })
-      @availability  = info_tbl.css('td')[1]
-                               .css('table')
-                               .css('td')
-                               .children
-                               .children
-                               .map(&:text)
-                               .reject(&:empty?)
-                               .each_slice(2)
-                               .to_a
-                               .tap { |a| a.last.unshift('distribution') }
-                               .map { |(k, v)| [k =~ /bottl/ ?  
-                                                :bottling : 
-                                                symbolize_text(k), v] }
-                               .to_h
-      @availability.merge!({ seasonal: misc[:seasonal] })
-      @description  = info_tbl.next_element
+    def scrape_description
+      @description = info_root.next_element
                               .next_element
                               .children
                               .children
@@ -147,10 +156,38 @@ module RateBeer
                               .reject(&:empty?)
                               .join("\n")
       @description = fix_characters(@description)
+    end
+
+    def scrape_retired
       @retired = !(root.css('span.beertitle2') && 
-                   root.css('span.beertitle2').text =~ /RETIRED/).nil?
+                   root.css('span.beertitle2').text =~ /RETIRED/).nil?  
+    end
+
+    def scrape_rating
+      raw_rating = [:overall,
+                    :style].zip(info_root.css('div')
+                           .select { |d| d['title'] =~ /This figure/ }
+                           .map { |d| d['title'].split(':').first.to_f }).to_h
+      @rating = raw_rating.merge(ratings:       scrape_misc[:ratings],
+                                 weighted_avg:  scrape_misc[:weighted_avg],
+                                 mean:          scrape_misc[:mean])
+    end
 
-      nil
+    # Scrapes the miscellaneous information contained on the beer page.
+    #
+    # This information relates to various other specific types of information.
+    # As such, other scrapers rely on this method for information.
+    def scrape_misc
+      info_root.next_element
+               .first_element_child
+               .children
+               .map(&:text)
+               .flat_map { |x| x.gsub(nbsp, ' ').strip.split(':') }
+               .map(&:strip)
+               .reject(&:empty?)
+               .each_slice(2)
+               .map { |(k, v)| [symbolize_text(k), v.to_f.zero? ? v : v.to_f] }
+               .to_h
     end
   end
 end

data/lib/ratebeer/brewery.rb

@@ -1,7 +1,11 @@
-require_relative "scraping"
-require_relative "urls"
+# frozen_string_literal: true
+
+require_relative 'scraping'
+require_relative 'urls'
 
 module RateBeer
+  # The brewery class represents one brewery found in RateBeer, with methods
+  # for accessing information found about the brewery on the site.
   class Brewery
     # Each key represents an item of data accessible for each beer, and defines
     # dynamically a series of methods for accessing this data.
@@ -19,11 +23,14 @@ module RateBeer
 
     attr_reader :established, :location
 
+    # CSS selector for the brewery information element.
+    INFO_SELECTOR = "div[itemtype='http://schema.org/LocalBusiness']".freeze
+
     # Create RateBeer::Brewery instance.
     #
     # Requires the RateBeer ID# for the brewery in question. Optionally accepts
     # a name parameter where the name is already known.
-    # 
+    #
     # @param [Integer, String] id ID# for the brewery
     # @param [String] name The name of the specified brewery
     # @param [hash] options Options hash for entity created
@@ -38,87 +45,70 @@ module RateBeer
       end
     end
 
-    def name
-      @name ||= retrieve_brewery_info
+    def doc
+      @doc ||= noko_doc(URI.join(BASE_URL, brewery_url(id)))
+      validate_brewery
+      @doc
     end
 
-    def beers
-      @beers ||= retrieve_brewery_beers
+    def info_root
+      @info_root ||= doc.at_css(INFO_SELECTOR)
     end
 
     private
 
-    # Retrieve details about this brewery from the website.
-    #
-    # This method stores the retrieved details in instance variables
-    # of the brewery instance.
+    # Validates whether the brewery with the given ID exists.
     #
-    def retrieve_details
-      @doc  = noko_doc(URI.join(BASE_URL, brewery_url(id)))
-
-      brewery_info = retrieve_brewery_info
-
-      @beers = []
-      if pagination?(@doc)
-        (1..page_count(@doc)).flat_map do |page_no|
-          @doc = noko_doc(URI.join(BASE_URL, brewery_url(id), "0/", "#{page_no}/"))
-          retrieve_brewery_beers
-        end
-      else
-        retrieve_brewery_beers
+    # Throws an exception if the brewery does not exist.
+    def validate_brewery
+      error_message = "This brewer, ID##{id}, is no longer in the database. "\
+                      'RateBeer Home'
+      if @doc.at_css('body p').text == error_message
+        raise PageNotFoundError.new("Brewery not found - #{id}")
       end
-      nil
     end
 
-    # Scrape brewery info from Nokogiri Doc for brewery page
-    #
-    def retrieve_brewery_info
-      root = @doc.css('#container table').first
-      contact_node = root.css('td').first
-
-      @name = fix_characters(root.css('h1').first.text)
-      raise PageNotFoundError.new("Brewery not found - #{id}") if @name.empty?
-
-      @type = root.css('span.beerfoot')
-                  .select { |x| x.text =~ /Type: .*/ }
-                  .first
-                  .text
-                  .strip
-                  .split("Type: ")
-                  .last
-                  .split(/\s{2,}/)
-                  .first
-      @address = root.css('div[itemprop="address"] b span')
-                     .map { |elem| key = case elem.attributes['itemprop'].value
-                                         when 'streetAddress' then :street
-                                         when 'addressLocality' then :city
-                                         when 'addressRegion' then :state
-                                         when 'addressCountry' then :country
-                                         when 'postalCode' then :postcode
-                                         else raise "unrecognised attribute"
-                                         end
-                                   [key, elem.text.strip] }
-                     .to_h
-
-      @telephone = root.css('span[itemprop="telephone"]').first && 
-                   root.css('span[itemprop="telephone"]').first.text
-
-    end
-
-    # Scrape beer details from Nokogiri Doc for brewery page
-    #
-    def retrieve_brewery_beers
-      location, brewer = nil  # Variables used in the map below
-      root = @doc.css('table.maintable.nohover').first
-      @beers += root.css('tr').drop(1).map do |row|
-        if row.text =~ /^Brewed at (?<location>.+?)(?: by\/for (?<brewer>.+))?$/
-          location = Regexp.last_match['location']
-          brewer   = Regexp.last_match['brewer']
-          nil
-        else
-          process_beer_row(row, location, brewer)
-        end
-      end.reject(&:nil?)
+    # Scrapes the brewery's name.
+    def scrape_name
+      @name = fix_characters(info_root.css('h1').first.text)
+    end
+
+    # Scrapes the brewery's address.
+    def scrape_address
+      address_root = info_root.css('div[itemprop="address"] b span')
+      address_details = address_root.map { |e| extract_address_element(e) }
+      @address = address_details.to_h
+    end
+
+    # Extracts one element of address details from a node contained within the
+    # address div.
+    def extract_address_element(node)
+      key = case node.attributes['itemprop'].value
+            when 'streetAddress'   then :street
+            when 'addressLocality' then :city
+            when 'addressRegion'   then :state
+            when 'addressCountry'  then :country
+            when 'postalCode'      then :postcode
+            else raise 'unrecognised attribute'
+            end
+      [key, node.text.strip]
+    end
+
+    # Scrapes the telephone number of the brewery.
+    def scrape_telephone
+      @telephone = info_root.at_css('span[itemprop="telephone"]')
+    end
+
+    # Scrapes the type of brewery.
+    def scrape_type
+      @type = info_root.css('div')[1]
+    end
+
+    # Scrapes beers list for brewery.
+    def scrape_beers
+      beers_doc = noko_doc(URI.join(BASE_URL, brewery_beers_url(id)))
+      rows = beers_doc.css('table#brewer-beer-table tbody tr')
+      @beers = rows.map { |row| process_beer_row(row) }.reject(&:nil?)
     end
 
     # Process a row of data representing one beer brewed by/at a brewery.
@@ -129,47 +119,70 @@ module RateBeer
     #   where this location differs from the brewery's regular brewsite/venue
     # @param [String] brewer the client for whom this brewery brewed the beer,
     #   where the brewery is brewing for a different company/brewery
-    # @return [RateBeer::Beer] a beer object representing the scraped beer, 
+    # @return [RateBeer::Beer] a beer object representing the scraped beer,
     #   containing scraped attributes
     #
-    def process_beer_row(row, location=nil, brewer=nil)
-      # Attributes stored in each table row, with indices representing their
-      # position in each row
-      attributes = { name:            0,
-                     abv:             2,
-                     avg_rating:      3,
-                     overall_rating:  4,
-                     style_rating:    5,
-                     num_ratings:     6 }
-
-      beer = attributes.reduce({}) do |beer_hash, (attr, i)| 
-        val = row.css('td')[i].text.gsub(nbsp, ' ').strip rescue nil
-        case attr
-        when :name
-          fix_characters(val)
-        when :abv, :avg_rating
-          val = val.to_f
-        when :overall_rating, :style_rating, :num_ratings
-          val = val.to_i
-        end
-        beer_hash[attr] = val
-        beer_hash
-      end
-      beer[:url] = row.css('td').first.css('a').first['href']
-      id = beer[:url].split('/').last.to_i
+    def process_beer_row(row)
+      beer = process_beer_name_cell(row.css('td').first)
+      beer[:abv] = row.css('td')[1].text.to_f
+      beer[:date_added] = Date.strptime(row.css('td')[2].text, '%m/%d/%Y')
+      Beer.new(id, beer.merge(process_rating_info(row)))
+    end
+
+    # Processes the cell containing the beer's name and other information.
+    #
+    # This cell contains information on the beer's name, its style, whether it
+    # is retired, and who is was brewed for or by.
+    def process_beer_name_cell(node)
+      beer_link = node.at_css('strong a')
+      name = fix_characters(beer_link.text)
+      id = id_from_link(beer_link)
+      info = node.at_css('em.real-small')
+      brewed_at_for = process_brewed_at_for(node)
+      style = process_style_info(node)
+      { id:     id,
+        name:   name,
+        style:  style,
+        retired:  info && info.text =~ /retired/ || false }.merge(brewed_at_for)
+    end
 
-      # Apply additional location and brewer information if scraped
-      beer[:brewed_at]     = location unless location.nil?
-      beer[:brewed_by_for] = brewer unless brewer.nil?
+    # Processes information on who the beer was brewed for or by, or at.
+    def process_brewed_at_for(node)
+      brewed_at_for_node = node.at_css('div.small em')
+      return {} if brewed_at_for_node.nil?
+      node_text = brewed_at_for_node.children.first.text
+      key = if node_text.include?('Brewed at')
+              :brewed_at
+            elsif node_text.include?('Brewed by/for')
+              :brewed_by_for
+            end
+      other_brewer_node = brewed_at_for_node.at_css('a')
+      { key => Brewery.new(id_from_link(other_brewer_node),
+                           name: other_brewer_node.text) }
+    end
 
-      # Transform ratings into correct format
-      beer[:rating] = { overall: beer[:overall_rating], 
-                        style: beer[:style_rating],
-                        ratings: beer[:num_ratings],
-                        weighted_avg: beer[:avg_rating] }
+    # Processes the style information contained within a beer name cell.
+    def process_style_info(node)
+      style_node = node.css('a').find do |n|
+        n.children.any? { |c| c.name == 'span' }
+      end
+      name = style_node.text
+      id = id_from_link(style_node)
+      Style.new(id, name: name)
+    end
 
-      # Create beer instance from scraped data
-      Beer.new(id, beer)
+    # Processes rating information from a beer row.
+    def process_rating_info(row)
+      cell_indices = { avg_rating:      4,
+                       overall_rating:  5,
+                       style_rating:    6,
+                       num_ratings:     7 }
+      rating = cell_indices.map do |attr, i|
+        val = row.css('td')[i].text.gsub(nbsp, ' ').strip
+        conversion = attr == :avg_rating ? :to_f : :to_i
+        [attr, val.send(conversion)]
+      end
+      rating.to_h
     end
   end
 end