metal_archives 0.4.0 → 0.5.0

data/lib/metal_archives/models/base_model.rb

@@ -1,8 +1,8 @@
 module MetalArchives
   ##
-  # Base model class all models are derived from
+  # Abstract model class
   #
-  class BaseModel # :nodoc:
+  class BaseModel
     ##
     # Generic shallow copy constructor
     #
@@ -21,21 +21,32 @@ module MetalArchives
       obj.instance_of? self.class and self.id == obj.id
     end
 
+    ##
+    # Fetch, parse and load the data
+    #
+    # [Raises]
+    # - rdoc-ref:MetalArchives::Errors::InvalidIDError when no id
+    # - rdoc-ref:MetalArchives::Errors::APIError when receiving a status code >= 400 (except 404)
+    #
+    def load!
+      raise Errors::InvalidIDError, 'no id present' unless !!id
+
+      # Use constructor to set attributes
+      initialize assemble
+    end
+
     protected
       ##
-      # Eagerly fetch the data
+      # Fetch the data and assemble the model
       #
-      # Raises rdoc-ref:MetalArchives::Errors::InvalidIDError when no id
-      # Raises rdoc-ref:MetalArchives::Errors::NotImplementedError when no :assemble method is implemented
-      # Raises rdoc-ref:MetalArchives::Errors::APIError when receiving a status code >= 400 (except 404)
+      # Override this method
       #
-      def fetch
-        raise Errors::InvalidIDError, 'no id present' unless !!id
-
-        raise Errors::NotImplementedError, 'no :assemble method in model' unless self.respond_to? :assemble, true
-
-        # Use constructor to set attributes
-        initialize assemble
+      # [Raises]
+      # - rdoc-ref:MetalArchives::Errors::InvalidIDError when no or invalid id
+      # - rdoc-ref:MetalArchives::Errors::APIError when receiving a status code >= 400 (except 404)
+      #
+      def assemble
+        raise Errors::NotImplementedError, 'method :assemble not implemented'
       end
 
     class << self
@@ -66,13 +77,13 @@ module MetalArchives
 
           # property
           define_method(name) do
-            self.fetch unless instance_variable_defined?("@#{name}") or name == :id
+            load! unless instance_variable_defined?("@#{name}") or name == :id
             instance_variable_get("@#{name}")
           end
 
           # property?
           define_method("#{name}?") do
-            self.fetch unless instance_variable_defined?("@#{name}") or name == :id
+            load! unless instance_variable_defined?("@#{name}") or name == :id
 
             property = instance_variable_get("@#{name}")
             property.respond_to?(:empty?) ? !property.empty? : !!property
@@ -116,13 +127,13 @@ module MetalArchives
 
           # property
           define_method(name) do
-            self.fetch unless instance_variable_defined?("@#{name}")
+            load! unless instance_variable_defined?("@#{name}")
             instance_variable_get("@#{name}")
           end
 
           # property?
           define_method("#{name}?") do
-            self.fetch unless instance_variable_defined?("@#{name}")
+            load! unless instance_variable_defined?("@#{name}")
 
             property = instance_variable_get("@#{name}")
             property.respond_to?(:empty?) ? !property.empty? : !!property
@@ -145,7 +156,7 @@ module MetalArchives
         end
 
         ##
-        # Defines a model boolean property. This method is an alias for +enum name, :values => [true, false]+
+        # Defines a model boolean property. This method is an alias for <tt>enum name, :values => [true, false]</tt>
         #
         # [+name+]
         #     Name of the property

data/lib/metal_archives/parsers/artist.rb

@@ -7,7 +7,7 @@ module Parsers
   ##
   # Artist parser
   #
-  class Artist # :nodoc:
+  class Artist < Parser # :nodoc:
     class << self
       ##
       # Map attributes to MA attributes
@@ -19,47 +19,52 @@ module Parsers
       #
       def map_params(query)
         params = {
-          :query => query[:name] || '',
-
-          :iDisplayStart => query[:iDisplayStart] || 0
+          :query => query[:name] || ''
         }
 
         params
       end
 
+      ##
+      # Parse main HTML page
+      #
+      # Returns +Hash+
+      #
+      # [Raises]
+      # - rdoc-ref:MetalArchives::Errors::ParserError when parsing failed. Please report this error.
+      #
       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
+            content = sanitize(dt.next_element.content)
+
+            next if content == 'N/A'
+
+            case sanitize(dt.content)
             when 'Real/full name:'
-              props[:name] = dt.next_element.content.strip
+              props[:name] = content
             when 'Age:'
-              break if dt.next_element.content == 'N/A'
-              date = dt.next_element.content.gsub(/ [0-9]* \(born ([^\)]*)\)/, '\1')
+              date = 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
+              props[:date_of_death] = Date.parse content
             when 'Died of:'
-              break if dt.next_element.content = 'N/A'
-              props[:cause_of_death] = dt.next_element.content
+              props[:cause_of_death] = 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)
+              props[:country] = ISO3166::Country.find_country_by_name(sanitize(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
+              case content
               when 'Male'
                 props[:gender] = :male
               when 'Female'
                 props[:gender] = :female
               else
-                raise Errors::ParserError, "Unknown gender: #{dt.next_element.content}"
+                raise Errors::ParserError, "Unknown gender: #{content}"
               end
             else
               raise Errors::ParserError, "Unknown token: #{dt.content}"
@@ -68,14 +73,23 @@ module Parsers
         end
 
         props[:aliases] = []
-        alt = doc.css('.band_member_name').first.content
+        alt = sanitize doc.css('.band_member_name').first.content
         props[:aliases] << alt unless props[:name] == alt
 
         props
       rescue => e
-        raise ParserError, e
+        e.backtrace.each { |b| MetalArchives::config.logger.error b }
+        raise Errors::ParserError, e
       end
 
+      ##
+      # Parse links HTML page
+      #
+      # Returns +Hash+
+      #
+      # [Raises]
+      # - rdoc-ref:MetalArchives::Errors::ParserError when parsing failed. Please report this error.
+      #
       def parse_links_html(response)
         links = []
 
@@ -89,6 +103,10 @@ module Parsers
             type = row['id'].gsub(/^header_/, '').downcase.to_sym
           else
             a = row.css('td a').first
+
+            # No links have been added yet
+            next unless a
+
             links << {
                 :url => a['href'],
                 :type => type,
@@ -98,6 +116,9 @@ module Parsers
         end
 
         links
+      rescue => e
+        e.backtrace.each { |b| MetalArchives::config.logger.error b }
+        raise Errors::ParserError, e
       end
     end
   end

data/lib/metal_archives/parsers/band.rb

@@ -7,7 +7,7 @@ module Parsers
   ##
   # Band parser
   #
-  class Band # :nodoc:
+  class Band < Parser # :nodoc:
     class << self
       ##
       # Map attributes to MA attributes
@@ -29,9 +29,7 @@ module Parsers
           :themes => query[:lyrical_themes] || '',
           :location => query[:location] || '',
           :bandLabelName => query[:label] || '',
-          :indieLabelBand => (!!query[:independent] ? 1 : 0),
-
-          :iDisplayStart => query[:iDisplayStart] || 0
+          :indieLabelBand => (!!query[:independent] ? 1 : 0)
         }
 
         params[:country] = []
@@ -43,11 +41,19 @@ module Parsers
         params
       end
 
+      ##
+      # Parse main HTML page
+      #
+      # Returns +Hash+
+      #
+      # [Raises]
+      # - rdoc-ref:MetalArchives::Errors::ParserError when parsing failed. Please report this error.
+      #
       def parse_html(response)
         props = {}
         doc = Nokogiri::HTML response
 
-        props[:name] = doc.css('#band_info .band_name a').first.content
+        props[:name] = sanitize 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?
@@ -55,43 +61,42 @@ module Parsers
 
         doc.css('#band_stats dl').each do |dl|
           dl.search('dt').each do |dt|
+            content = sanitize(dt.next_element.content)
+
+            next if content == 'N/A'
+
             case dt.content
             when 'Country of origin:'
-              props[:country] = ISO3166::Country.find_country_by_name dt.next_element.css('a').first.content
+              props[:country] = ISO3166::Country.find_country_by_name sanitize(dt.next_element.css('a').first.content)
             when 'Location:'
-              break if dt.next_element.content == 'N/A'
-              props[:location] = dt.next_element.content
+              props[:location] = content
             when 'Status:'
-              props[:status] = dt.next_element.content.downcase.gsub(/ /, '_').to_sym
+              props[:status] = 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
+              props[:date_formed] = Date.new content.to_i
             when 'Genre:'
-              break if dt.next_element.content == 'N/A'
-              props[:genres] = ParserHelper.parse_genre dt.next_element.content
+              props[:genres] = parse_genre content
             when 'Lyrical themes:'
               props[:lyrical_themes] = []
-              break if dt.next_element.content == 'N/A'
-              dt.next_element.content.split(',').each do |theme|
+              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')
+              props[:independent] = (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|
+              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)
+                props[:date_active] << MetalArchives::Range.new(date_start, date_end)
               end
             else
               raise MetalArchives::Errors::ParserError, "Unknown token: #{dt.content}"
@@ -101,9 +106,18 @@ module Parsers
 
         props
       rescue => e
-        raise ParserError, e
+        e.backtrace.each { |b| MetalArchives::config.logger.error b }
+        raise Errors::ParserError, e
       end
 
+      ##
+      # Parse similar bands HTML page
+      #
+      # Returns +Hash+
+      #
+      # [Raises]
+      # - rdoc-ref:MetalArchives::Errors::ParserError when parsing failed. Please report this error.
+      #
       def parse_similar_bands_html(response)
         similar = []
 
@@ -116,8 +130,19 @@ module Parsers
         end
 
         similar
+      rescue => e
+        e.backtrace.each { |b| MetalArchives::config.logger.error b }
+        raise Errors::ParserError, e
       end
 
+      ##
+      # Parse related links HTML page
+      #
+      # Returns +Hash+
+      #
+      # [Raises]
+      # - rdoc-ref:MetalArchives::Errors::ParserError when parsing failed. Please report this error.
+      #
       def parse_related_links_html(response)
         links = []
 
@@ -138,9 +163,20 @@ module Parsers
         end
 
         links
+      rescue => e
+        e.backtrace.each { |b| MetalArchives::config.logger.error b }
+        raise Errors::ParserError, e
       end
 
       private
+        ##
+        # Map MA band status
+        #
+        # Returns +Symbol+
+        #
+        # [Raises]
+        # - rdoc-ref:MetalArchives::Errors::ParserError when parsing failed. Please report this error.
+        #
         def map_status(status)
           s = {
             nil => '',

data/lib/metal_archives/parsers/{parser_helper.rb → parser.rb}

@@ -7,9 +7,9 @@ module MetalArchives
 #
 module Parsers # :nodoc:
   ##
-  # Parsing utilities
+  # Parser base class
   #
-  class ParserHelper # :nodoc:
+  class Parser
     class << self
       ##
       # Parse a country
@@ -20,6 +20,15 @@ module Parsers # :nodoc:
         ISO3166::Country.find_country_by_name (input)
       end
 
+      ##
+      # Sanitize a string
+      #
+      # Return +String+
+      #
+      def sanitize(input)
+        input.gsub(/^"/, '').gsub(/"$/, '').strip
+      end
+
       ##
       # Opinionated parsing of genres
       #
@@ -32,7 +41,7 @@ module Parsers # :nodoc:
       #
       # All genres are capitalized.
       #
-      # For examples on how genres are parsed, refer to +ParserHelperTest::test_parse_genre+
+      # For examples on how genres are parsed, refer to +ParserTest#test_parse_genre+
       #
       def parse_genre(input)
         genres = []

data/lib/metal_archives/version.rb

@@ -2,5 +2,5 @@ module MetalArchives
   ##
   # MetalArchives API version
   #
-  VERSION = '0.4.0'
+  VERSION = '0.5.0'
 end

data/metal_archives.gemspec

@@ -14,15 +14,15 @@ Gem::Specification.new do |gem|
   gem.version       = MetalArchives::VERSION
   gem.license       = "MIT"
 
-  gem.add_development_dependency 'byebug', '~> 9.0.6', '>= 9.0.6'
-  gem.add_development_dependency 'rake', '~> 11.3.0', '>= 11.3.0'
-  gem.add_development_dependency 'rdoc', '~> 5.0.0', '>= 5.0.0'
-  gem.add_development_dependency 'test-unit', '~> 3.2.2', '>= 3.2.2'
-  gem.add_development_dependency 'activesupport', '~> 5.0.0.1', '>= 5.0.0'
+  gem.add_development_dependency 'byebug', '~> 9.0'
+  gem.add_development_dependency 'rake', '~> 11.0'
+  gem.add_development_dependency 'rdoc', '~> 5.0'
+  gem.add_development_dependency 'test-unit', '~> 3.0'
+  gem.add_development_dependency 'activesupport', '~> 5.0'
 
-  gem.add_dependency 'faraday', '~> 0.10.0', '>= 0.10.0'
-  gem.add_dependency 'faraday-http-cache', '~> 1.3.1', '>= 1.3'
-  gem.add_dependency 'faraday_throttler', '~> 0.0.3', '>= 0.0.3'
-  gem.add_dependency 'nokogiri', '~> 1.6.8.1', '>= 1.6.8'
-  gem.add_dependency 'countries', '~> 1.2.5', '>= 1.2.5'
+  gem.add_dependency 'faraday', '~> 0.9'
+  gem.add_dependency 'faraday-http-cache', '~> 2.0'
+  gem.add_dependency 'faraday_throttler', '~> 0.0.3'
+  gem.add_dependency 'nokogiri', '~> 1.6.8'
+  gem.add_dependency 'countries', '~> 1.2.5'
 end

data/test/parser_test.rb

@@ -0,0 +1,37 @@
+require 'test_helper'
+
+require 'date'
+require 'countries'
+
+require 'metal_archives/parsers/parser.rb'
+
+class ParserTest < Test::Unit::TestCase
+  def test_parse_country
+    assert_equal ISO3166::Country['US'], MetalArchives::Parsers::Parser.parse_country('United States')
+    assert_equal ISO3166::Country['DE'], MetalArchives::Parsers::Parser.parse_country('Germany')
+    assert_equal ISO3166::Country['BE'], MetalArchives::Parsers::Parser.parse_country('Belgium')
+  end
+
+  def test_parse_genre
+    assert_equal ['Black', 'Death', 'Power'].sort,
+        MetalArchives::Parsers::Parser.parse_genre('Death, Power, Black').sort
+
+    assert_equal ['Black', 'Death', 'Power'].sort,
+        MetalArchives::Parsers::Parser.parse_genre('Death, Power, Black').sort
+
+    assert_equal ['Black', 'Death', 'Heavy', 'Power'].sort,
+        MetalArchives::Parsers::Parser.parse_genre('Death (early), Heavy/Power Metal, Black (later)').sort
+
+    assert_equal ['Death', 'Power'].sort,
+        MetalArchives::Parsers::Parser.parse_genre('        Death       ,      Power   Metal, Power, Power').sort
+
+    assert_equal ['Heavy Power', 'Speed Power'].sort,
+        MetalArchives::Parsers::Parser.parse_genre('Heavy/Speed Power Metal').sort
+
+    assert_equal ['Traditional Heavy', 'Traditional Power'].sort,
+        MetalArchives::Parsers::Parser.parse_genre('Traditional Heavy/Power Metal').sort
+
+    assert_equal ['Traditional Heavy', 'Traditional Power', 'Classical Heavy', 'Classical Power'].sort,
+        MetalArchives::Parsers::Parser.parse_genre('Traditional/Classical Heavy/Power Metal').sort
+  end
+end