porras-imdb 0.0.6 → 0.0.7

data/README

@@ -1,6 +1,4 @@
 
-ImdbMovie
-
 ImdbMovie Indiana Jones and the Last Crusade
 - should query IMDB url
 - should get director
@@ -8,6 +6,7 @@ ImdbMovie Indiana Jones and the Last Crusade
 - should get the rating
 - should get cast members
 - should get the writers
+- should get the year
 - should get the release date
 - should get the genres
 - should get the plot
@@ -40,24 +39,25 @@ ImdbMovie Han robado una estrella
 - should get the company
 - should not get any photos
 
-ImdbSearch
-
-ImdbSearch Indiana Jones
+ImdbSearch search that returns multiple movies
 - should query IMDB url
+- should not allow to change the query
 
-ImdbSearch Indiana Jones movies
+ImdbSearch search that returns multiple movies movies
 - should be a collection of ImdbMovie instances
 - should include 'Indiana Jones and the Last Crusade'
 - should have titles
 - should not have titles with HTML tags
+- should not have duplicate movies
 
-ImdbSearch searches with potential encoding issues
+ImdbSearch search that redirects to the lone matching movie movies
+- should be a collection containing a single ImdbMovie instance
+- should have the correct ID
+- should have the correct title
 
 ImdbSearch searches with potential encoding issues movies
 - should include 'Misión en Marbella'
 
-String
-
 String unescape_html
 - should convert & to &
 - should convert ó to ó
@@ -65,6 +65,6 @@ String unescape_html
 String strip_tags
 - should strip HTML tags
 
-Finished in 3.044303 seconds
+Finished in 0.829689 seconds
 
-42 examples, 0 failures
+48 examples, 0 failures

data/lib/imdb/imdb_movie.rb

@@ -27,6 +27,10 @@ class ImdbMovie
   def writers
     document.search("h5[text()^='Writers'] ~ a").map { |link| link.innerHTML.strip.unescape_html }.reject { |w| w == 'more' }.uniq rescue []
   end
+
+  def year
+    document.search('a[@href^="/Sections/Years/"]').innerHTML.to_i
+  end
   
   def release_date
     date = document.search("//h5[text()^='Release Date']/..").innerHTML[/^\d{1,2} \w+ \d{4}/]

data/lib/imdb/imdb_search.rb

@@ -1,23 +1,37 @@
 class ImdbSearch
 
-  attr_accessor :query
-
   def initialize(query)
-    self.query = query
+    @query = query
   end
 
   def movies
-    @movies ||= document.search('a[@href^="/title/tt"]').reject do |element|
-      element.innerHTML.strip_tags.empty?
-    end.map do |element|
-      ImdbMovie.new(element['href'][/\d+/], element.innerHTML.strip_tags.unescape_html)
-    end
+    @movies ||= (exact_match? ? parse_movie : parse_movies)
   end
 
   private
   
   def document
-    @document ||= Hpricot(open("http://www.imdb.com/find?q=#{CGI::escape(query)};s=tt").read)
+    @document ||= Hpricot(open("http://www.imdb.com/find?q=#{CGI::escape(@query)};s=tt").read)
+  end
+  
+  def parse_movies
+    document.search('a[@href^="/title/tt"]').reject do |element|
+      element.innerHTML.strip_tags.empty?
+    end.map do |element|
+      [element['href'][/\d+/], element.innerHTML.strip_tags.unescape_html]
+    end.uniq.map do |values|
+      ImdbMovie.new(*values)
+    end
+  end
+  
+  def parse_movie
+    id = document.at("a[@name='poster']")['href'][/\d+$/]
+    title = document.at("h1").innerHTML.split('<span').first.strip.unescape_html
+    [ImdbMovie.new(id, title)]
+  end
+  
+  def exact_match?
+    document.at("title[text()='IMDb Search']").nil?
   end
 
 end

data/spec/imdb_movie_spec.rb

@@ -40,6 +40,10 @@ describe ImdbMovie do
       @imdb_movie.writers.should include('Philip Kaufman')
       @imdb_movie.writers.should_not include('more')
     end
+
+    it "should get the year" do
+      @imdb_movie.year.should == 1989
+    end
   
     it "should get the release date" do
       @imdb_movie.release_date.should be_an_instance_of(Date)

data/spec/imdb_search_spec.rb

@@ -2,7 +2,7 @@ require File.dirname(__FILE__) + '/spec_helper'
 
 describe ImdbSearch do
 
-  describe 'Indiana Jones' do
+  describe 'search that returns multiple movies' do
 
     before(:each) do
       @imdb_search = ImdbSearch.new('indiana jones')
@@ -13,6 +13,10 @@ describe ImdbSearch do
       @imdb_search.should_receive(:open).with("http://www.imdb.com/find?q=indiana+jones;s=tt").and_return(open("#{$samples_dir}/sample_search.html"))
       @imdb_search.send(:document)
     end
+    
+    it "should not allow to change the query" do
+      lambda { @imdb_search.query = 'wadus' }.should raise_error(NoMethodError)
+    end
 
     describe "movies" do
 
@@ -39,7 +43,37 @@ describe ImdbSearch do
           movie.title.should_not match(/<.+>/)
         end
       end
+      
+      it "should not have duplicate movies" do
+        all_movie_ids = @imdb_search.movies.collect {|m| m.id}
+        unique_movie_ids = all_movie_ids.uniq
+        all_movie_ids.should == unique_movie_ids
+      end
+    end
 
+  end
+
+  describe 'search that redirects to the lone matching movie' do
+
+    before(:each) do
+      @imdb_search = ImdbSearch.new('some extremely specific search for indiana jones')
+      @imdb_search.stub!(:open).and_return(open("#{$samples_dir}/sample_movie.html"))
+    end
+
+    describe "movies" do
+
+      it "should be a collection containing a single ImdbMovie instance" do
+        @imdb_search.movies.size.should == 1
+        @imdb_search.movies.first.should be_an_instance_of(ImdbMovie)
+      end
+
+      it "should have the correct ID" do
+        @imdb_search.movies.first.id.should == '0097576'
+      end
+
+      it "should have the correct title" do
+        @imdb_search.movies.first.title.should == 'Indiana Jones and the Last Crusade'
+      end
     end
 
   end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: porras-imdb
 version: !ruby/object:Gem::Version 
-  version: 0.0.6
+  version: 0.0.7
 platform: ruby
 authors: 
 - Sergio Gil