yayimdbs 0.2.3 → 0.2.4

data/lib/yay_imdbs.rb

@@ -18,6 +18,13 @@ class YayImdbs
 
   STRIP_WHITESPACE = /(\s{2,}|\n|\||\302\240\302\273)/u
 
+  MORE_INFO_LINKS = ['See more',
+                     'Add/edit official sites',
+                     'See all certifications',
+                     'See full summary',
+                     'see all',
+                    ]
+
   DATE_PROPERTIES = [:release_date]
   LIST_PROPERTIES = [:genres, :plot_keywords, :country, :sound_mix, :language]
   INT_LIST_PROPERTIES = [:year, :season]
@@ -26,7 +33,8 @@ class YayImdbs
                      :year => :years, 
                      :season => :seasons,
                      :language => :languages,
-                     :motion_picture_rating_mpaa => :mpaa}
+                     :motion_picture_rating_mpaa => :mpaa,
+					 :official_sites => :official_site}
 
   class << self
 
@@ -95,7 +103,7 @@ class YayImdbs
       movie_properties(doc) do |key, value|
         found_info_divs = true
         info_hash["raw_#{key}"] = value
-        info_hash[key] = clean_movie_property(key, value)
+        info_hash[key] = clean_movie_property(key, value, imdb_id)
         info_hash[PROPERTY_ALIAS[key]] = info_hash[key] if PROPERTY_ALIAS[key]
       end
 
@@ -115,7 +123,7 @@ class YayImdbs
       return info_hash
     end
 
-    def clean_movie_property(key, value)
+    def clean_movie_property(key, value, imdb_id)
       if DATE_PROPERTIES.include?(key)
         value = Date.strptime(value, '%d %B %Y') rescue nil
       elsif key == :runtime
@@ -124,6 +132,8 @@ class YayImdbs
         else
           value = nil
         end
+      elsif key == :official_sites
+        value = get_official_site_url(value, imdb_id)
       elsif LIST_PROPERTIES.include?(key)
         value = value.split('|').collect { |l| l.gsub(/[^a-zA-Z0-9\-]/, '') }
       elsif INT_LIST_PROPERTIES.include?(key)
@@ -138,15 +148,21 @@ class YayImdbs
         raw_key = h4.inner_text
         key = raw_key.sub(':', '').strip.downcase
         value = div.inner_text[((div.inner_text =~ /#{Regexp.escape(raw_key)}/) + raw_key.length).. -1]
-        value = value.gsub(/\302\240\302\273/u, '').strip.gsub(/(See more)|(see all)|(See all certifications)|(Add\/edit official sites)|(See full summary)$/, '').strip
+        value = value.gsub(/\302\240\302\273/u, '').strip.gsub(/(#{MORE_INFO_LINKS.join(')|(')})$/i, '').strip
         symbol_key = key.downcase.gsub(/[^a-zA-Z0-9 ]/, '').gsub(/\s/, '_').to_sym
-		if symbol_key == :official_sites
-			value = div.inner_html.match(/href=\"(.*?)\"/)[1]
-		end
         yield symbol_key, value
       end
     end
 
+    # TODO capture all official sites, not all sites have an "Official site" link (e.g. Lost)
+    def get_official_site_url(value, imdb_id)
+        value = value.match(/<a href="(.*?)">Official site<\/a>/)
+        if value.nil?
+            value = get_official_sites_page(imdb_id).inner_html.match(/<a href="(.*?)">Official site<\/a>/)
+        end
+        return $1
+    end
+
     def scrap_images(doc, info_hash)
       #scrap poster image urls
       thumbnail_url = doc.at_css("td[id=img_primary] a img").try(:[], 'src')
@@ -174,9 +190,9 @@ class YayImdbs
 
         raw_date = e_div.at_css('strong').inner_text.strip
         episode['date'] = Date.parse(raw_date) rescue nil
-        if e_div.inner_text =~ /#{raw_date}/
-          episode['plot'] = $'.strip
-        end
+
+        # Seems that the day can sometimes be ???? which doesnt play will with regex
+        episode['plot'] = $'.strip if e_div.inner_text =~ /#{raw_date}/ rescue nil
 
         episodes << episode
       end
@@ -190,6 +206,10 @@ class YayImdbs
       def get_movie_page(imdb_id)
         Nokogiri::HTML(open(IMDB_MOVIE_URL + imdb_id))
       end
+	  
+      def get_official_sites_page(imdb_id)
+        Nokogiri::HTML(open(IMDB_MOVIE_URL + imdb_id + '/officialsites'	))
+      end
 
       def get_episodes_page(imdb_id)
         Nokogiri::HTML(open(IMDB_MOVIE_URL + imdb_id + '/episodes'))

metadata

@@ -5,8 +5,8 @@ version: !ruby/object:Gem::Version
   segments: 
   - 0
   - 2
-  - 3
-  version: 0.2.3
+  - 4
+  version: 0.2.4
 platform: ruby
 authors: 
 - Sam Cavenagh
@@ -14,7 +14,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-07-28 00:00:00 +10:00
+date: 2011-08-02 00:00:00 +10:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency