metal_archives 0.2.0

data/lib/metal_archives/models/range.rb

@@ -0,0 +1,58 @@
+module MetalArchives
+  ##
+  # Range which can start and/or end with +nil+
+  #
+  class Range
+    include Comparable
+
+    ##
+    # Begin- and endpoint of range
+    #
+    attr_accessor :begin, :end
+
+    ##
+    # Create a new range
+    #
+    # [+_begin+]
+    #     Start of range
+    #
+    #     Default: +nil+
+    #
+    # [+_end+]
+    #     End of range
+    #
+    #     Default: +nil+
+    #
+    def initialize(_begin = nil, _end = nil)
+      @begin = _begin
+      @end = _end
+    end
+
+    ##
+    # Whether start of range is present
+    #
+    def begin?
+    !!@begin
+    end
+
+    ##
+    # Whether end of range is present
+    #
+    def end?
+    !!@end
+    end
+
+    ##
+    # Comparison operator
+    #
+    def <=>(other)
+      comparison = self.begin <=> other.begin
+
+      if comparison == 0
+        return self.end <=> other.end
+      else
+        return comparison
+      end
+    end
+  end
+end

data/lib/metal_archives/parsers/artist.rb

@@ -0,0 +1,103 @@
+require 'json'
+require 'date'
+require 'countries'
+
+module MetalArchives
+module Parsers
+  ##
+  # Artist parser
+  #
+  class Artist # :nodoc:
+    class << self
+      ##
+      # Map attributes to MA attributes
+      #
+      # Returns +Hash+
+      #
+      # [+params+]
+      #     +Hash+
+      #
+      def map_params(query)
+        params = {
+          :query => query[:name] || '',
+
+          :iDisplayStart => query[:iDisplayStart] || 0
+        }
+
+        params
+      end
+
+      def parse_html(response)
+        props = {}
+        doc = Nokogiri::HTML response
+
+        doc.css('#member_info dl').each do |dl|
+          dl.css('dt').each do |dt|
+            case dt.content.strip
+            when 'Real/full name:'
+              props[:name] = dt.next_element.content.strip
+            when 'Age:'
+              break if dt.next_element.content == 'N/A'
+              date = dt.next_element.content.gsub(/ [0-9]* \(born ([^\)]*)\)/, '\1')
+              props[:date_of_birth] = Date.parse date
+            when 'R.I.P.:'
+              break if dt.next_element.content == 'N/A'
+              props[:date_of_death] = Date.parse dt.next_element.content
+            when 'Died of:'
+              break if dt.next_element.content = 'N/A'
+              props[:cause_of_death] = dt.next_element.content
+            when 'Place of origin:'
+              break if dt.next_element.content == 'N/A'
+              props[:country] = ISO3166::Country.find_country_by_name(dt.next_element.css('a').first.content)
+              location = dt.next_element.xpath('text()').map { |x| x.content }.join('').strip.gsub(/[()]/, '')
+              props[:location] = location unless location.empty?
+            when 'Gender:'
+              break if dt.next_element.content == 'N/A'
+              case dt.next_element.content
+              when 'Male'
+                props[:gender] = :male
+              when 'Female'
+                props[:gender] = :female
+              else
+                raise Errors::ParserError, "Unknown gender: #{dt.next_element.content}"
+              end
+            else
+              raise Errors::ParserError, "Unknown token: #{dt.content}"
+            end
+          end
+        end
+
+        props[:aliases] = []
+        alt = doc.css('.band_member_name').first.content
+        props[:aliases] << alt unless props[:name] == alt
+
+        props
+      end
+
+      def parse_links_html(response)
+        links = []
+
+        doc = Nokogiri::HTML response
+
+        # Default to official links
+        type = :official
+
+        doc.css('#linksTablemain tr').each do |row|
+          if row['id'] =~ /^header_/
+            type = row['id'].gsub(/^header_/, '').downcase.to_sym
+          else
+            a = row.css('td a').first
+            links << {
+                :url => a['href'],
+                :type => type,
+                :title => a.content
+            }
+          end
+        end
+
+        links
+      end
+    end
+  end
+end
+end

data/lib/metal_archives/parsers/band.rb

@@ -0,0 +1,160 @@
+require 'json'
+require 'date'
+require 'countries'
+
+module MetalArchives
+module Parsers
+  ##
+  # Band parser
+  #
+  class Band # :nodoc:
+    class << self
+      ##
+      # Map attributes to MA attributes
+      #
+      # Returns +Hash+
+      #
+      # [+params+]
+      #     +Hash+
+      #
+      def map_params(query)
+        params = {
+          :bandName => query[:name] || '',
+          :exactBandMatch => (!!query[:exact] ? 1 : 0),
+          :genre => query[:genre] || '',
+          :yearCreationFrom => (query[:year] and query[:year].begin ? query[:year].begin.year : '') || '',
+          :yearCreationTo => (query[:year] and query[:year].end ? query[:year].end.year : '') || '',
+          :bandNotes => query[:comment] || '',
+          :status => map_status(query[:status]),
+          :themes => query[:lyrical_themes] || '',
+          :location => query[:location] || '',
+          :bandLabelName => query[:label] || '',
+          :indieLabelBand => (!!query[:independent] ? 1 : 0),
+
+          :iDisplayStart => query[:iDisplayStart] || 0
+        }
+
+        params[:country] = []
+        Array(query[:country]).each do |country|
+          params[:country] << (country.is_a?(ISO3166::Country) ? country.alpha2 : (country || ''))
+        end
+        params[:country] = params[:country].first if (params[:country].size == 1)
+
+        params
+      end
+
+      def parse_html(response)
+        props = {}
+        doc = Nokogiri::HTML response
+
+        props[:name] = doc.css('#band_info .band_name a').first.content
+
+        props[:aliases] = []
+        props[:logo] = doc.css('.band_name_img img').first.attr('src') unless doc.css('.band_name_img').empty?
+        props[:photo] = doc.css('.band_img img').first.attr('src') unless doc.css('.band_img').empty?
+
+        doc.css('#band_stats dl').each do |dl|
+          dl.search('dt').each do |dt|
+            case dt.content
+            when 'Country of origin:'
+              props[:country] = ISO3166::Country.find_country_by_name dt.next_element.css('a').first.content
+            when 'Location:'
+              break if dt.next_element.content == 'N/A'
+              props[:location] = dt.next_element.content
+            when 'Status:'
+              props[:status] = dt.next_element.content.downcase.gsub(/ /, '_').to_sym
+            when 'Formed in:'
+              break if dt.next_element.content == 'N/A'
+              props[:date_formed] = Date.new dt.next_element.content.to_i
+            when 'Genre:'
+              break if dt.next_element.content == 'N/A'
+              props[:genres] = ParserHelper.parse_genre dt.next_element.content
+            when 'Lyrical themes:'
+              props[:lyrical_themes] = []
+              break if dt.next_element.content == 'N/A'
+              dt.next_element.content.split(',').each do |theme|
+                t = theme.split.map(&:capitalize)
+                t.delete '(early)'
+                t.delete '(later)'
+                props[:lyrical_themes] << t.join(' ')
+              end
+            when /(Current|Last) label:/
+              props[:independent] = (dt.next_element.content == 'Unsigned/independent')
+              # TODO
+            when 'Years active:'
+              break if dt.next_element.content == 'N/A'
+              props[:date_active] = []
+              dt.next_element.content.split(',').each do |range|
+                # Aliases
+                range.scan(/\(as ([^)]*)\)/).each { |name| props[:aliases] << name.first }
+                # Ranges
+                r = range.gsub(/ *\(as ([^)]*)\) */, '').strip.split('-')
+                date_start = (r.first == '?' ? nil : Date.new(r.first.to_i))
+                date_end = (r.last == '?' or r.last == 'present' ? nil : Date.new(r.first.to_i))
+                props[:date_active] << Range.new(date_start, date_end)
+              end
+            else
+              raise MetalArchives::Errors::ParserError, "Unknown token: #{dt.content}"
+            end
+          end
+        end
+
+        props
+      end
+
+      def parse_similar_bands_html(response)
+        similar = []
+
+        doc = Nokogiri::HTML response
+        doc.css('#artist_list tbody tr').each do |row|
+          similar << {
+              :band => MetalArchives::Band.new(:id => row.css('td a').first['href'].split('/').last.to_i),
+              :score => row.css('td').last.content.strip
+          }
+        end
+
+        similar
+      end
+
+      def parse_related_links_html(response)
+        links = []
+
+        doc = Nokogiri::HTML response
+        doc.css('#linksTableOfficial td a').each do |a|
+          links << {
+            :url => a['href'],
+            :type => :official,
+            :title => a.content
+          }
+        end
+        doc.css('#linksTableOfficial_merchandise td a').each do |a|
+          links << {
+            :url => a['href'],
+            :type => :merchandise,
+            :title => a.content
+          }
+        end
+
+        links
+      end
+
+      private
+        def map_status(status)
+          s = {
+            nil => '',
+            :active => 'Active',
+            :split_up => 'Split-up',
+            :on_hold => 'On hold',
+            :unknown => 'Unknown',
+            :changed_name => 'Changed name',
+            :disputed => 'Disputed'
+          }
+
+          raise MetalArchives::Errors::ParserError, "Unknown status: #{status}" unless s[status]
+
+          s[status]
+        end
+    end
+  end
+end
+end

data/lib/metal_archives/parsers/label.rb

@@ -0,0 +1,68 @@
+require 'date'
+require 'nokogiri'
+
+module MetalArchives
+module Parsers
+  ##
+  # Label parser
+  #
+  class Label # :nodoc:
+    class << self
+      def find_endpoint(params)
+        "http://www.metal-archives.com/labels/#{params[:name]}/#{params[:id]}"
+      end
+
+      def parse(response)
+        props = {}
+        doc = Nokogiri::HTML(response)
+
+        props[:name] = doc.css('#label_info .label_name').first.content
+
+        props[:contact] = []
+        doc.css('#label_contact a').each do |contact|
+          props[:contact] << {
+            :title => contact.content,
+            :content => contact.attr(:href)
+          }
+        end
+
+        doc.css('#label_info dl').each do |dl|
+          dl.search('dt').each do |dt|
+            case dt.content
+            when 'Address:'
+              break if dt.next_element.content == 'N/A'
+              props[:address] = dt.next_element.content
+            when 'Country:'
+              break if dt.next_element.content == 'N/A'
+              props[:country] = ParserHelper.parse_country dt.next_element.css('a').first.content
+            when 'Phone number:'
+              break if dt.next_element.content == 'N/A'
+              props[:phone] = dt.next_element.content
+            when 'Status:'
+              props[:status] = dt.next_element.content.downcase.gsub(/ /, '_').to_sym
+            when 'Specialised in:'
+              break if dt.next_element.content == 'N/A'
+              props[:specializations] = ParserHelper.parse_genre dt.next_element.content
+            when 'Founding date :'
+              break if dt.next_element.content == 'N/A'
+              props[:date_founded] = Date.new dt.next_element.content.to_i
+            when 'Sub-labels:'
+              # TODO
+            when 'Online shopping:'
+              if dt.next_element.content == 'Yes'
+                props[:online_shopping] = true
+              elsif dt.next_element.content == 'No'
+                props[:online_shopping] = false
+              end
+            else
+              raise "Unknown token: #{dt.content}"
+            end
+          end
+        end
+
+        props
+      end
+    end
+  end
+end
+end

data/lib/metal_archives/parsers/parser_helper.rb

@@ -0,0 +1,79 @@
+require 'date'
+require 'countries'
+
+module MetalArchives
+##
+# Mapping layer from and to MA Web Service
+#
+module Parsers # :nodoc:
+  ##
+  # Parsing utilities
+  #
+  class ParserHelper # :nodoc:
+    class << self
+      ##
+      # Parse a country
+      #
+      # Returns +ISO3166::Country+
+      #
+      def parse_country(input)
+        ISO3166::Country.find_country_by_name (input)
+      end
+
+      ##
+      # Opinionated parsing of genres
+      #
+      # Returns an +Array+ of +String+
+      #
+      # The following components are omitted:
+      # - Metal
+      # - (early)
+      # - (later)
+      #
+      # All genres are capitalized.
+      #
+      # For examples on how genres are parsed, refer to +ParserHelperTest::test_parse_genre+
+      #
+      def parse_genre(input)
+        genres = []
+        # Split fields
+        input.split(',').each do |genre|
+          ##
+          # Start with a single empty genre string. Split the genre by spaces
+          # and process each component. If a component does not have a slash,
+          # concatenate it to all genre strings present in +temp+. If it does
+          # have a slash present, duplicate all genre strings, and concatenate
+          # the first component (before the slash) to the first half, and the
+          # last component to the last half. +temp+ now has an array of genre
+          # combinations.
+          #
+          # 'Traditional Heavy/Power Metal' => ['Traditional Heavy', 'Traditional Power']
+          # 'Traditional/Classical Heavy/Power Metal' => [
+          #                                       'Traditional Heavy', 'Traditional Power',
+          #                                       'Classical Heavy', 'Classical Power']
+          #
+          temp = ['']
+          genre.downcase.split.reject { |g| ['(early)', '(later)', 'metal'].include? g }.each do |g|
+            unless g.include? '/'
+              temp.map! { |t| t.empty? ? g.capitalize : "#{t.capitalize} #{g.capitalize}" }
+            else
+              # Duplicate all WIP genres
+              temp2 = temp.dup
+
+              # Assign first and last components to temp and temp2 respectively
+              split = g.split '/'
+              temp.map! { |t| t.empty? ? split.first.capitalize : "#{t.capitalize} #{split.first.capitalize}" }
+              temp2.map! { |t| t.empty? ? split.last.capitalize : "#{t.capitalize} #{split.last.capitalize}" }
+
+              # Add both genre trees
+              temp += temp2
+            end
+          end
+          genres += temp
+        end
+        genres.uniq
+      end
+    end
+  end
+end
+end