myimdb 0.4.0 → 0.4.1

data/VERSION

@@ -1 +1 @@
-0.4.0
+0.4.1

data/bin/myimdb

@@ -21,6 +21,10 @@ OptionParser.new do |opts|
     options[:metacritic] = true
   end
   
+  opts.on("-b", "--freebase", "Generates data from Freebase") do
+    options[:freebase] = true
+  end
+  
   opts.on("-i", "--imdb", "Generates data from Imdb") do
     options[:imdb] = true
   end
@@ -46,8 +50,7 @@ name = ARGV.join(' ')
 
 def details(klass_name, name)
   search_scope  = "#{klass_name.downcase}.com"
-  
-  search_result = Myimdb::Search::Google.search_text(name, :restrict_to=> search_scope)[0]
+  search_result = Myimdb::Search::Base.search(name, :restrict_to=> search_scope)[0]
   site          = eval("Myimdb::Scraper::#{klass_name}").new(search_result[:url])
   puts "===================================================="
   puts "#{klass_name} details for: #{name}"
@@ -57,6 +60,6 @@ rescue Exception=> ex
   puts "Unable to fetch #{klass_name} details for: #{name} because: #{ex.message}"
 end
 
-['Imdb', 'RottenTomatoes', 'Metacritic'].each do |site|
+['Imdb', 'RottenTomatoes', 'Metacritic', 'Freebase'].each do |site|
   details(site, name) if options[site.downcase.to_sym]
 end

data/bin/myimdb-catalogue

@@ -147,7 +147,7 @@ end
 def generate_metadata(path, name)
   search_scope  = "imdb.com"
 
-  search_result = Myimdb::Search::Google.search_text(name, :restrict_to=> search_scope)[0]
+  search_result = Myimdb::Search::Base.search(name, :restrict_to=> search_scope)[0]
   imdb          = Myimdb::Scraper::Imdb.new(search_result[:url])
   new_name      = name.gsub(/\[\S+\]/, "").strip
   new_name      << " [#{imdb.year}] [#{imdb.rating},#{imdb.votes}] [#{imdb.directors.join(',')}]"

data/lib/myimdb/scraper/base.rb

@@ -1,4 +1,5 @@
 class UnformattedHtml < Exception; end
+class DocumentNotFound < Exception; end
 
 module HandleExceptions
   def self.included(base)
@@ -30,7 +31,8 @@ module Myimdb
     class Base
       include HandleExceptions
       include Myimdb::Scraper::StringExtensions
-
+      
+      def name; end
       def directors; end
       def directors_with_url; end
       def writers; end

data/lib/myimdb/scraper/freebase.rb

@@ -0,0 +1,74 @@
+module Myimdb
+  module Scraper
+    class Freebase < Scraper::Base
+      include HTTParty
+      format  :json
+      headers 'Content-Type' => 'application/json'
+
+      def initialize(url)
+        @key = url.gsub(/.*?\/view/, '')
+        @url = "http://www.freebase.com/experimental/topic/standard?id=#{@key}"
+      end
+
+      def directors
+        parse_text '/film/film/directed_by'
+      end
+
+      def directors_with_url
+        parse_text_and_url '/film/film/directed_by'
+      end
+
+      def writers
+        parse_text '/film/film/written_by'
+      end
+
+      def writers_with_url
+        parse_text_and_url '/film/film/written_by'
+      end
+
+      def genres
+        parse_text '/film/film/genre'
+      end
+
+      def tagline
+        parse_text('/film/film/tagline')[0]
+      end
+
+      def plot
+        document['description']
+      end
+
+      def year
+        release_date.year
+      end
+
+      def release_date
+        Date.parse(parse_text('/film/film/initial_release_date')[0])
+      end
+
+      def image
+      end
+      
+      private
+        def document
+          @document ||= begin
+            result = self.class.get( @url )[@key]
+            unless result['code'].index("ok")
+              raise DocumentNotFound.new("Unable to locate freebase article")
+            end
+            result['result']
+          end
+        end
+
+        def parse_text(path)
+          document['properties'][path]['values'].collect{ |obj| obj['text'] }
+        end
+
+        def parse_text_and_url(path)
+          document['properties'][path]['values'].collect{ |obj| { :name=> obj['text'], :url=> obj['url'] } }
+        end
+        
+        handle_exceptions_for :directors, :directors_with_url, :writers, :writers_with_url, :genres, :tagline, :plot, :year, :image
+    end
+  end
+end

data/lib/myimdb/scraper.rb

@@ -2,4 +2,5 @@ require 'myimdb/scraper/string_extensions'
 require 'myimdb/scraper/base'
 require 'myimdb/scraper/imdb'
 require 'myimdb/scraper/metacritic'
-require 'myimdb/scraper/rotten_tomatoes'
+require 'myimdb/scraper/rotten_tomatoes'
+require 'myimdb/scraper/freebase'

data/lib/myimdb/search/base.rb

@@ -8,6 +8,20 @@ module Myimdb
             search_result[:title].gsub(/ \(.*$/, "")
           end
         end
+
+        def search( text, options )
+          engines = [Myimdb::Search::Google, Myimdb::Search::Bing]
+
+          def search(engine, text, options)
+            engine.search_text(text, options)
+          end
+
+          engines.each do |engine|
+            puts "Retrying using #{engine}" unless engines.index(engine) == 0
+            result = search(engine, text, options)
+            return result unless result.nil? or result.empty?
+          end
+        end
       end
     end
   end

data/lib/myimdb.rb

@@ -8,7 +8,7 @@ require 'myimdb/scraper'
 
 class ImdbMovie
   def self.search(movie_name)
-    search_result = Myimdb::Search::Google.search_text(movie_name, :restrict_to=> 'imdb.com')[0]
+    search_result = Myimdb::Search::Base.search(movie_name, :restrict_to=> 'imdb.com')[0]
     Myimdb::Scraper::Imdb.new(search_result[:url]) if search_result
   end
 end

data/myimdb.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = %q{myimdb}
-  s.version = "0.4.0"
+  s.version = "0.4.1"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Gaurav"]
-  s.date = %q{2010-02-22}
+  s.date = %q{2010-03-23}
   s.email = %q{gaurav@vinsol.com}
   s.executables = ["myimdb", "myimdb-catalogue"]
   s.extra_rdoc_files = [
@@ -26,6 +26,7 @@ Gem::Specification.new do |s|
      "lib/myimdb.rb",
      "lib/myimdb/scraper.rb",
      "lib/myimdb/scraper/base.rb",
+     "lib/myimdb/scraper/freebase.rb",
      "lib/myimdb/scraper/imdb.rb",
      "lib/myimdb/scraper/metacritic.rb",
      "lib/myimdb/scraper/rotten_tomatoes.rb",

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: myimdb
 version: !ruby/object:Gem::Version 
-  version: 0.4.0
+  version: 0.4.1
 platform: ruby
 authors: 
 - Gaurav
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-02-22 00:00:00 +05:30
+date: 2010-03-23 00:00:00 +05:30
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -52,6 +52,7 @@ files:
 - lib/myimdb.rb
 - lib/myimdb/scraper.rb
 - lib/myimdb/scraper/base.rb
+- lib/myimdb/scraper/freebase.rb
 - lib/myimdb/scraper/imdb.rb
 - lib/myimdb/scraper/metacritic.rb
 - lib/myimdb/scraper/rotten_tomatoes.rb