webs-allocine 0.3.0

data/README.textile

@@ -0,0 +1,92 @@
+h2. Allocine
+
+Allocine consiste en trois classes.
+
+* @Allocine@ : permet de rechercher des films/séries,
+* @Allocine::Movie@ qui permet de récupérer toutes les informations sur un film,
+* @Allocine::Show@ qui permet de récupérer toutes les information sur les séries.
+
+Une documentation (et un code) plus propre arrive prochainement.
+
+Bug reports : http://webs.lighthouseapp.com/projects/18095-allocine
+
+Mis a disposition sous licence MIT. (c) 2008 Jordan Bracco <jordan@lifeisdead.net>
+
+h3. Installation
+
+<pre>
+  $ gem sources -a http://gems.github.com
+  $ sudo gem install webs-allocine
+</pre>
+
+h3. Recherche
+
+Les recherches s'effectuent avec @Allocine.find_movie@ et @Allocine.find_show@, et renvoie un Hash de l'ID Allocine et du titre exact.
+
+h4. Recherche d'un film
+
+La recherche s'effectue avec @Allocine.find_movie@ et renvoie un Hash de l'ID allocine et du titre exact.
+
+<pre>
+  >> Allocine.find_movie('Plan 9 From Outer Space')
+  => {"13031"=>"Plan 9", "136668"=>"Plan 9 from Outer Space"}
+</pre>
+
+h4. Recherche d'une série
+
+<pre>
+  >> Allocine.find_show('xfiles')
+  => {"182"=>"X-Files : Aux frontières du réel"}
+</pre>
+
+h3. Récupération des informations d'un film/série
+
+h4. Film
+
+<pre>
+  >> movie = Allocine::Movie.new('13031')
+  => #<AllocineMovie:0x11fb250 ....>
+  >> movie.title
+  => "Plan 9"
+  >> movie.original_title
+  => "Plan 9 from Outer Space"
+  >> movie.actors
+  => "Tom Keene, Tor Johnson, Vampira"
+  >> movie.directors
+  => "Ed Wood"
+  >> movie.production_date
+  => "1959"
+  >> movie.genres
+  => "Science fiction, Epouvante-horreur, Fantastique"
+  >> movie.nat
+  => "am\303\251ricain"
+  >> movie.image
+  => ""
+  >> movie.synopsis
+  => "Des extraterrestres appliquent le plan 9 destine a manoeuvrer les terriens. Plan diabolique, il consiste a la resurrection des morts en introduisant des electrodes a longue portee stimulant la glande pineale de cadavres recemment enterres."
+</pre>
+
+h4. Série
+
+<pre>
+  >> show = Allocine::Show.new('182')
+  => #<AllocineShow:0x1255908 ...>
+  >> show.title
+  => "X-Files : Aux fronti\303\250res du r\303\251el"
+  >> show.created_by
+  => "Chris Carter"
+  >> show.image
+  => "http://a69.g.akamai.net/n/69/10688/v1/img5.allocine.fr/acmedia/medias/nmedia/18/35/65/25/18600368.jpg"
+  >> show.actors
+  => "Robert Patrick, Annabeth Gish"
+  >> show.nat
+  => "Am\303\251ricaine, canadienne"
+  >> show.duree
+  => "42 mn"
+  >> show.genres
+  => "Fantastique"
+  >> show.original_title
+  => "The X Files"
+  >> show.synopsis
+  => "Deux agents du FBI sont charg\303\251s d'enqu\303\252ter sur les dossiers non r\303\251solus, appel\303\251s \"X-files\" la plupart du temps des affaires o\303\271 le paranormal entre en jeu. L'agent Fox Mulder, malgr\303\251 le scepticisme de sa co-\303\251quipi\303\250re Dana Scully, tente de prouver sa th\303\250se du complot gouvernement/extra-terrestres...\r\n\r\nFox Mulder n'aura de cesse de rechercher sa soeur disparue des ann\303\251es auparavant lorsqu'il avait 12 ans."
+</pre>

data/allocine.gemspec

@@ -0,0 +1,28 @@
+Gem::Specification.new do |s|
+  s.name     = "allocine"
+  s.version  = "0.3.0"
+  s.date     = "2009-02-08"
+  s.summary  = "Allocine.fr parser"
+  s.email    = "jordan@lifeisdead.net"
+  s.homepage = "http://github.com/webs/allocine"
+  s.description = "Allocine.fr parser"
+  s.has_rdoc = false
+  s.authors  = ["Jordan Bracco", "Sunny Ripert"]
+  s.files    = ["README.textile", 
+    "allocine.gemspec", 
+    "lib/allocine.rb",
+    "lib/allocine/allocine.rb",
+    "lib/allocine/movie.rb", 
+    "lib/allocine/show.rb"
+  ]
+  s.test_files = ["spec/allocine.rb", 
+    "spec/helper.rb",
+    "spec/movie.rb",
+    "spec/show.rb"
+  ]
+  #s.rdoc_options = ["--main", "README.txt"]
+  #s.extra_rdoc_files = ["History.txt", "Manifest.txt", "README.txt"]
+  #s.add_dependency("diff-lcs", ["> 0.0.0"])
+  #s.add_dependency("mime-types", ["> 0.0.0"])
+  #s.add_dependency("open4", ["> 0.0.0"])
+end

data/lib/allocine.rb

@@ -0,0 +1,11 @@
+$:.unshift File.dirname(__FILE__)
+require 'rubygems'
+require 'open-uri'
+require 'iconv'
+require 'allocine/allocine'
+require 'allocine/movie'
+require 'allocine/show'
+MOVIE_SEARCH_URL = "http://www.allocine.fr/recherche/?motcle=%s&rub=1"
+MOVIE_DETAIL_URL = "http://www.allocine.fr/film/fichefilm_gen_cfilm=%s.html"
+SHOW_SEARCH_URL = "http://www.allocine.fr/recherche/?motcle=%s&rub=6"
+SHOW_DETAIL_URL = "http://www.allocine.fr/series/ficheserie_gen_cserie=%s.html"

data/lib/allocine/allocine.rb

@@ -0,0 +1,21 @@
+module Allocine
+  
+  # Make a search on movies
+  def self.find_movie(search)
+    Allocine::Movie.find(search)
+  end
+  
+  # Make a search on shows
+  def self.find_show(search)
+    Allocine::Show.find(search)
+  end
+  
+  # Returns the first result
+  def self.lucky_movie(search)
+    Allocine::Movie.lucky_find(search)
+  end
+  
+  def self.lucky_show(search)
+    Allocine::Show.lucky_find(search)
+  end
+end

data/lib/allocine/movie.rb

@@ -0,0 +1,54 @@
+module Allocine
+class Movie
+  attr_accessor :title, :directors, :nat, :genres, :out_date, :duree, :production_date, :original_title, :actors, :synopsis, :image, :interdit
+  
+  def self.find(search)
+    search.gsub!(' ', '+')
+    str = open(MOVIE_SEARCH_URL % search).read.to_s
+    data = Iconv.conv('UTF-8', 'ISO-8859-1', str)
+    movies = {}
+    while data =~ /<a href="\/film\/fichefilm_gen_cfilm=(\d+).html" class="link(\d+)">(.*?)<\/a>/i
+      id, klass, name = $1, $2, $3
+      data.gsub!("<a href=\"/film/fichefilm_gen_cfilm=#{id}.html\" class=\"link#{klass}\">#{name}</a>", "")
+      name.gsub!(/<(.+?)>/,'')
+      movies[id] = name
+    end
+    movies
+  end
+  
+  def self.lucky_find(search)
+    results = find(search)
+    new(results.keys.first)
+  end
+  
+  def initialize(id, debug=false)
+    regexps = {
+      :title => '<h1 class="TitleFilm">(.*?)<\/h1>',
+      :directors => '<h3 class="SpProse">Réalisé par <a .*?>(.*?)<\/a><\/h3>',
+      :nat => '<h3 class="SpProse">Film (.*?).&nbsp;</h3>',
+      :genres => '<h3 class="SpProse">Genre : (.*?)</h3>',
+      :out_date => '<h3 class="SpProse">Date de sortie : <b>(.*?)</b>',
+      :duree => '<h3 class="SpProse">Dur\ée : (.*?).&nbsp;</h3>',
+      :production_date => '<h3 class="SpProse">Année de production : (.*?)</h3>',
+      :original_title => '<h3 class="SpProse">Titre original : <i>(.*?)</i></h3>',
+      :actors => '<h3 class="SpProse">Avec (.*?) &nbsp;&nbsp;',
+      :synopsis => '<td valign="top" style="padding:10 0 0 0"><div align="justify"><h4>(.*?)</h4>',
+      :image => '<td valign="top" width="120".*?img src="(.*?)" border="0" alt=".*?" class="affichette" />',
+      :interdit => '<h4 style="color: #D20000;">Interdit(.*?)</h4>'
+    }
+    str = open(MOVIE_DETAIL_URL % id).read.to_s
+    data = Iconv.conv('UTF-8', 'ISO-8859-1', str)
+    regexps.each do |reg|
+      print "#{reg[0]}: " if debug
+      r = data.scan Regexp.new(reg[1], Regexp::MULTILINE)
+      r = r.first.to_s.strip
+      r.gsub!(/<.*?>/, '')
+      if r[0..0] == " "
+        r = r.reverse.chop.reverse # that's a little bit ugly, but the only simple way i found to remove the first space in the out date
+      end
+      self.instance_variable_set("@#{reg[0]}", r)
+      print "#{r}\n" if debug
+    end
+  end
+end
+end

data/lib/allocine/show.rb

@@ -0,0 +1,49 @@
+module Allocine
+class Show
+  attr_accessor :title, :created_by, :producters, :nat, :genres, :duree, :original_title, :actors, :synopsis, :image
+  
+  def self.find(search)
+    search.gsub!(' ', '+')
+    str = open(SHOW_SEARCH_URL % search).read.to_s
+    data = Iconv.conv('UTF-8', 'ISO-8859-1', str)
+    shows = {}
+    while data =~ /<a href="\/series\/ficheserie_gen_cserie=(\d+).html" class="link(\d+)">(.*?)<\/a>/i
+      id, klass, name = $1, $2, $3
+      data.gsub!("<a href=\"/series/ficheserie_gen_cserie=#{id}.html\" class=\"link#{klass}\">#{name}</a>", "")
+      name.gsub!(/<(.+?)>/,'')
+      shows[id] = name
+    end
+    shows
+  end
+  
+  def self.lucky_find(search)
+    results = find(search)
+    new(results.keys.first)
+  end
+  
+  def initialize(id, debug=false)
+    regexps = {
+      :title => '<title>(.*?)<\/title>',
+      :created_by => '<h4>Série créée par <a .*?>(.*?)</a>', 
+      :producters => '<h4>Producteurs : (.*?)</h4>', 
+      :nat => '<span style=\'font-weight:bold\'>Nationalité</span> : (.*?)</h5>', 
+      :genres => '<span style=\'font-weight:bold\'>Genre</span> : (.*?)&nbsp;&nbsp;', 
+      :duree => '<span style=\'font-weight:bold\'>Format</span> : (.+?).&nbsp;', 
+      :original_title => '<h4><b>Titre original : </b></h4><h4 style="color:#D20000"><b>(.*?)</b></h4>',
+      :actors => '<h4>Avec : (.*?)&nbsp;&nbsp;', 
+      :synopsis => '<h5><span style=\'font-weight:bold\'>Synopsis</span>&nbsp;&nbsp;&nbsp;.*?<br />(.*?)</h5>',
+      :image => '<td><div id=\'divM\' .*?><img src=\'(.*?)\' style=\'border:1px solid black;.*?>',
+    }
+    str = open(SHOW_DETAIL_URL % id).read.to_s
+    data = Iconv.conv('UTF-8', 'ISO-8859-1', str)
+    regexps.each do |reg|
+      print "#{reg[0]}: " if debug
+      r = data.scan Regexp.new(reg[1], Regexp::MULTILINE)
+      r = r.first.to_s.strip
+      r.gsub!(/<.*?>/, '')
+      self.instance_variable_set("@#{reg[0]}", r)
+      print "#{r}\n" if debug
+    end
+  end
+end
+end

data/spec/allocine.rb

@@ -0,0 +1,17 @@
+describe "Allocine" do
+  it "find movie" do
+    Allocine.find_movie('Plan 9').should == {"13031"=>"Plan 9", "136668"=>"Plan 9 from Outer Space"}
+  end
+  
+  it "find show" do
+    Allocine.find_show('XFiles').should == {"182"=>"X-Files : Aux fronti\303\250res du r\303\251el"}
+  end
+  
+  it "lucky movie" do
+    Allocine.lucky_movie('Les pirates de la Silicon Valley').is_a?(Allocine::Movie).should == true
+  end
+  
+  it "lucky show" do
+    Allocine.lucky_show('Stargate').is_a?(Allocine::Show).should == true
+  end
+end

data/spec/helper.rb

@@ -0,0 +1,3 @@
+require File.dirname(__FILE__) + '/../lib/allocine.rb'
+require "rubygems"
+require "spec"

data/spec/movie.rb

@@ -0,0 +1,49 @@
+describe "Allocine Movie (Star Trek - 4854)" do
+  
+  before(:each) { @movie = Allocine::Movie.new('4854') }
+  
+  it 'should have a title' do
+    @movie.title.should == "Star Trek : Premier contact"
+  end
+  
+  it 'should have a synopsis' do
+    @movie.synopsis.should == "De m\303\251chants extraterrestres, les Borgs, complotent contre les habitants de la Terre. Ils mettent au point une machination diabolique pour d\303\251truire l'humanit\303\251."
+  end
+  
+  it 'should have an original title' do
+    @movie.original_title.should == "Star Trek: First Contact"
+  end
+  
+  it 'should have a nationality' do
+    @movie.nat.should == "am\303\251ricain"
+  end
+  
+  it 'should have a picture' do
+    @movie.image.should == "http://a69.g.akamai.net/n/69/10688/v1/img5.allocine.fr/acmedia/medias/nmedia/18/66/82/23/18957530.jpg"
+  end
+  
+  it 'should have a genre' do
+    @movie.genres.should == "Science fiction"
+  end
+  
+  it 'should have a director' do
+    @movie.directors.should == "Jonathan Frakes"
+  end
+  
+  it 'should have a duree (lolwtffrench)' do
+    @movie.duree.should == "1h 52min"
+  end
+  
+  it 'should have actors' do
+    @movie.actors.should == "Jonathan Frakes, Patrick Stewart, Brent Spiner"
+  end
+  
+  it 'should have an out date' do
+    @movie.out_date.should == "05 Mars 1997"
+  end
+  
+  it 'should have a production date' do
+    @movie.production_date.should == "1996"
+  end
+  
+end

data/spec/show.rb

@@ -0,0 +1,40 @@
+describe "Allocine Show (Enterprise - 109)" do
+  
+  before(:each) { @show = Allocine::Show.new('109') }
+  
+  it 'should have a title' do
+    @show.title.should == "Enterprise"
+  end
+  
+  it 'should have producters' do
+    @show.producters.should == "J.P. Farrell, Antoinette Stella"
+  end
+  
+  it 'should have a genre' do
+    @show.genres.should == "Fantastique"
+  end
+  
+  it 'should have an image' do
+    @show.image.should == "http://a69.g.akamai.net/n/69/10688/v1/img5.allocine.fr/acmedia/medias/nmedia/18/35/64/59/18415355.jpg"
+  end
+  
+  it 'should have a creator' do
+    @show.created_by.should == "Rick Berman"
+  end
+  
+  it 'should have actors' do
+    @show.actors.should == "Dominic Keating, Scott Bakula"
+  end
+  
+  it 'should have a nationality' do
+    @show.nat.should == "Am\303\251ricaine"
+  end
+  
+  it 'should have a synopsis' do
+    @show.synopsis.should == "Le premier vaisseau terrien Enterprise, dirig\303\251 par le capitaine Jonathan Archer, part \303\240 la d\303\251couverte de mondes nouveaux, de nouvelles formes de vie et de civilisations diff\303\251rentes.Note : L'histoire se d\303\251roule 100 ans avant les aventures du capitaine Kirk dans la s\303\251rie originale Star Trek."
+  end
+  
+  it 'should have a duree' do
+    @show.duree.should == "42 mn" # \o/
+  end
+end

metadata

@@ -0,0 +1,62 @@
+--- !ruby/object:Gem::Specification 
+name: webs-allocine
+version: !ruby/object:Gem::Version 
+  version: 0.3.0
+platform: ruby
+authors: 
+- Jordan Bracco
+- Sunny Ripert
+autorequire: 
+bindir: bin
+cert_chain: []
+
+date: 2009-02-08 00:00:00 -08:00
+default_executable: 
+dependencies: []
+
+description: Allocine.fr parser
+email: jordan@lifeisdead.net
+executables: []
+
+extensions: []
+
+extra_rdoc_files: []
+
+files: 
+- README.textile
+- allocine.gemspec
+- lib/allocine.rb
+- lib/allocine/allocine.rb
+- lib/allocine/movie.rb
+- lib/allocine/show.rb
+has_rdoc: false
+homepage: http://github.com/webs/allocine
+post_install_message: 
+rdoc_options: []
+
+require_paths: 
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement 
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      version: "0"
+  version: 
+required_rubygems_version: !ruby/object:Gem::Requirement 
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      version: "0"
+  version: 
+requirements: []
+
+rubyforge_project: 
+rubygems_version: 1.2.0
+signing_key: 
+specification_version: 2
+summary: Allocine.fr parser
+test_files: 
+- spec/allocine.rb
+- spec/helper.rb
+- spec/movie.rb
+- spec/show.rb