osdb 0.0.10 → 0.1.0

data/bin/getsub

@@ -1,197 +1,159 @@
 #!/usr/bin/env ruby
-# TODO: getsub is becomming a bit to complexe, rewrite it OO or move it in another gem
 require 'optparse'
 require 'uri'
 require 'rubygems'
 require File.expand_path(File.join(File.dirname(__FILE__), '..', 'lib', 'osdb'))
 
-def env_lang
-  OSDb::Language.from_locale(ENV['LANG'])
-end
 
-@options = {:language => env_lang.to_iso639_2b, :force => false, :dir => nil }
-@parser ||= OptionParser.new do |opts|
-  opts.banner = "Automatically download subs for your video files using opensubtitle.org"
-  opts.separator ""
-  opts.separator "Usage: getsub [options] DIRECTORY | VIDEO_FILE [VIDEO_FILE ...]"
-  opts.separator ""
-  opts.separator "Main options:"
-
-  opts.on("-l", "--language LANGUAGE", "Sub language ISO 639-2 code like fre or eng. Default: env $LANG (#{env_lang.to_iso639_2b})") do |language|
-    if language.to_s.length != 3
-      STDERR.puts "Language should specified as ISO 639-2 (ie, 3 letters, like 'eng' or 'fre')"
-      exit 1
-    end
-    @options[:language] = language.to_s
-  end
+class GetSub
 
-  opts.on("-d", "--directory DIRECTORY", "Specify a directory to search recursively for movies") do |dir|
-    @options[:dir] = dir
-  end
-  
-  opts.on("-f", "--force", "Download sub even if video already has one") { @options[:force] = true }
+  def initialize
+    @options = {:language => env_lang.to_iso639_2b, :force => false, :dir => nil, :methods => 'h'}
+    @parser ||= OptionParser.new do |opts|
+      opts.banner = "Automatically download subs for your video files using opensubtitle.org"
+      opts.separator ""
+      opts.separator "Usage: getsub [options] DIRECTORY | VIDEO_FILE [VIDEO_FILE ...]"
+      opts.separator ""
+      opts.separator "Main options:"
 
-  opts.on("-t", "--type FORMATS", "Select only subtitles in specified formats. e.g -t srt,sub") { |formats| @options[:formats] = formats.to_s.split(',') }
+      opts.on("-a", "--auto", "Do not ask user to resolve hash conflicts.") { @options[:auto] = true }
 
-end
-@parser.parse!
+      opts.on("-l", "--language LANGUAGE", "Sub language ISO 639-2 code like fre or eng. Default: env $LANG (#{env_lang.to_iso639_2b})") do |language|
+        if language.to_s.length != 3
+          STDERR.puts "Invalid argument: Language should specified as ISO 639-2 (ie, 3 letters, like 'eng' or 'fre')"
+          exit 1
+        end
+        @options[:language] = language.to_s
+      end
 
-def curl_available?
-  %x{ curl --version 2> /dev/null > /dev/null }
-  $?.success?
-end
+      opts.on("-f", "--force", "Download sub even if video already has one") { @options[:force] = true }
 
-def wget_available?
-  %x{ wget --version 2> /dev/null > /dev/null }
-  $?.success?
-end
+      opts.on("-t", "--type FORMATS", "Select only subtitles in specified formats. e.g -t srt,sub") { |formats| @options[:formats] = formats.to_s.split(',') }
 
-def download!(url, local_path)
-  puts "* download #{url} to #{local_path}"
-  if curl_available?
-    %x{ curl -s '#{url}' | gunzip > "#{local_path}" }
-  elsif wget_available?
-    %x{ wget -O - -q '#{url}' | gunzip > "#{local_path}"}
-  else
-    puts "Can't found any curl or wget please install one of them or manualy download your sub"
-    puts url
-  end
-end
+      methods_help = "Ordered list of search methods. h: by movie hash, i: by name on IMDB, n: by name on OSDb, p: by filename on OSDb. e.g -s hi . Default: h"
+      opts.on("-s", "--search-by METHODS", methods_help) do |methods|
+        unless methods =~ /^[hinp]+$/
+          STDERR.puts "Invalid argument: Available search methods are: h, i, n and p."
+          exit 1
+        end
+        @options[:methods] = methods
+      end
 
-def group_by_movie_name(subs)
-  subs.inject({}) do |hash, sub| 
-    hash[sub.movie_name] ||= []
-    hash[sub.movie_name] << sub
-    hash
+    end
   end
-end
 
-def ask_user_to_identify_movie(movies)
-  movies.keys.each_with_index do |name, index|
-    puts " #{index} - #{name}"
+  def env_lang
+    OSDb::Language.from_locale(ENV['LANG'])
   end
-  print 'id: '
-  str = STDIN.gets # TODO: rule #1, never trust user input
-  puts
-  movies[movies.keys[str.to_i]] || []
-end
-
-def normalize_name(name)
-  name.downcase.gsub(/[\s\.\-\_]+/, ' ')
-end
 
-def select_movie(movies)
-  return movies.values.first || [] if movies.length <= 1
-
-  puts "D'oh! You stumbled upon a hash conflict, please resolve it:"
-  puts
-  ask_user_to_identify_movie(movies)
-end
-
-def select_format(subs)
-  return subs unless @options[:formats]
-  subs.select{ |s| @options[:formats].include?(s.format) }
-end
-
-def select_sub(subs)
-  subs = select_format(subs)
-  movies = group_by_movie_name(subs)
-  subs = select_movie(movies)
-  subs.max_by(&:score)
-end
-
-def download_sub!(sub, movie)
-  sub_path = movie.sub_path(sub.format)
-  download!(sub.url, sub_path)
-end
+  def run!(files)
+    @parser.parse!
+
+    movie_files = files.map{ |path| OSDb::MovieFile.new(path) }
+
+    movie_files.each do |movie_file|
+      begin
+        puts "* Search subtitles for #{movie_file.name}"
+        if movie_file.has_sub? && !@options[:force]
+          puts "* Sub already there. To override it use --force"
+          puts
+          next
+        end
+
+        if sub = subtitle_finder.find_sub_for(movie_file, @options[:language])
+          download_sub!(sub, movie_file)
+        else
+          puts "* No sub found"
+        end
+        puts
+      rescue Exception => e
+        report_exception(e)
+      end
+    end
+  end
 
-def arg_files
-  return ARGV unless @options[:dir]
-  Dir.glob(File.join(@options[:dir], '**', "*.{#{OSDb::Movie::EXTENSIONS.join(',')}}"))
-end
+  def report_exception(exception)
+    puts "Something crashed."
+    puts "Feel free to report the error here: https://github.com/byroot/ruby-osdb/issues"
+    puts "With the following debug informations:"
+    puts
+    puts "#{exception.class.name}: #{exception.message}:"
+    puts exception.backtrace
+    puts
+  end
 
-movies = arg_files.map{ |path| OSDb::Movie.new(path) }
-movies.reject!(&:has_sub?) unless @options[:force]
-
-server = OSDb::Server.new(
-  :host => 'api.opensubtitles.org', 
-  :path => '/xml-rpc', 
-  :timeout => 90, 
-  :useragent => "SubDownloader 2.0.10" # register useragent ? WTF ? too boring.
-) 
-STDOUT.sync = true
-
-if movies.empty?
-  puts "No file provided"
-  puts @parser.help
-  exit 1
-end
+  def subtitle_finder
+    @subtitle_finder ||= OSDb::SubtitleFinder.new(search_engines, finders, selectors)
+  end
 
-def search_by_hash(server, movie)
-  server.search_subtitles(:moviehash => movie.hash, :moviebytesize => movie.size.to_s, :sublanguageid => @options[:language])
-end
+  SEARCH_ENGINES = {
+    'h' => OSDb::Search::MovieHash,
+    'i' => OSDb::Search::IMDB,
+    'n' => OSDb::Search::Name,
+    'p' => OSDb::Search::Path
+  }
 
-def search_by_path(server, movie)
-  server.search_subtitles(:sublanguageid => @options[:language], :tag => movie.path)
-end
+  def search_engines
+    @options[:methods].to_s.each_char.to_a.uniq.map do |char|
+      SEARCH_ENGINES[char.to_s].new(server)
+    end
+  end
 
-def search_by_name(server, movie)
-  subs = server.search_subtitles(:sublanguageid => @options[:language], :query => movie.name)
-  normalized_movie_name = normalize_name(movie.name)
-  subs.select! do |sub|
-    normalize_name(sub.filename).index(normalized_movie_name) # MAYBE: Levenshtein ?
+  def finders
+    [OSDb::Finder::Score.new]
   end
-  subs if subs.any?
-end
 
-def search_by_imdb(server, movie)
-  imdb_results = server.search_imdb(:query => movie.name)
-  if imdb_results.any?
-    if imdb_results.length == 1
-      imdb_result = imdb_results.first
-      puts "* found on IMDB with ID: #{imdb_result.imdbid}"
+  def selectors
+    movie_finder = if @options[:auto]
+      OSDb::Finder::First.new # TODO: try to match subtitle movie name with filename
     else
-      movies = Hash[imdb_results.map{ |r| [r.title, r] }]
-      imdb_result = ask_user_to_identify_movie(movies)
+      OSDb::Finder::Interactive.new
     end
-    server.search_subtitles(:sublanguageid => @options[:language], :imdbid => imdb_result.imdbid)
+    
+    selectors = [
+      OSDb::Selector::Movie.new(movie_finder)
+    ]
+    selectors << OSDb::Selector::Format.new(@options[:formats]) if @options[:formats]
+    selectors
   end
-end
 
-movies.each do |movie|
-  begin
-    puts "* search subs by hash for: #{movie.path}"
-    sub = select_sub search_by_hash(server, movie)
+  def server
+    @server ||= OSDb::Server.new(
+      :host => 'api.opensubtitles.org',
+      :path => '/xml-rpc', 
+      :timeout => 90, 
+      :useragent => "SubDownloader 2.0.10" # register useragent ? WTF ? too boring.
+    ) 
+  end
 
-    unless sub
-      puts "* could not find sub by hash, trying IMDB"
-      sub = select_sub search_by_imdb(server, movie)
-    end
+  def download_sub!(sub, movie_file)
+    sub_path = movie_file.sub_path(sub.format)
+    download!(sub.url, sub_path)
+  end
 
-    unless sub
-      puts "* no matches"
-      puts "* search subs by path"
-      sub = select_sub search_by_path(server, movie)
-    end
+  def curl_available?
+    %x{ curl --version 2> /dev/null > /dev/null }
+    $?.success?
+  end
 
-    unless sub
-      puts "* still no matches"
-      puts "* search subs by filename"
-      sub = select_sub search_by_name(server, movie)
-    end
+  def wget_available?
+    %x{ wget --version 2> /dev/null > /dev/null }
+    $?.success?
+  end
 
-    if sub
-      download_sub!(sub, movie)
+  def download!(url, local_path)
+    # TODO: use Net::HTTP and ZLib to avoid shell dependency
+    puts "* download #{url} to #{local_path}"
+    if curl_available?
+      %x{ curl -s '#{url}' | gunzip > "#{local_path}" }
+    elsif wget_available?
+      %x{ wget -O - -q '#{url}' | gunzip > "#{local_path}"}
     else
-      puts "Nothing worked, you are very unlucky :'("
+      puts "Can't found any curl or wget please install one of them or manualy download your sub"
+      puts url
     end
-    puts
-  rescue Exception => e
-    puts "Something crashed."
-    puts "Feel free to report the error here: https://github.com/byroot/ruby-osdb/issues"
-    puts "With the following debug informations:"
-    puts
-    puts "#{e.class.name}: #{e.message}:"
-    puts e.backtrace
   end
+
 end
+
+GetSub.new.run!(ARGV)

data/lib/osdb.rb

@@ -2,8 +2,13 @@ require 'xmlrpc/client'
 
 module OSDb
   base_path = File.expand_path(File.dirname(__FILE__) + '/osdb')
-  autoload :Language,  "#{base_path}/language"
-  autoload :Movie,  "#{base_path}/movie"
-  autoload :Server, "#{base_path}/server"
-  autoload :Sub,    "#{base_path}/sub"
-end
+  autoload :Finder,         "#{base_path}/finder"
+  autoload :Language,       "#{base_path}/language"
+  autoload :Movie,          "#{base_path}/movie"
+  autoload :MovieFile,      "#{base_path}/movie_file"
+  autoload :Search,         "#{base_path}/search"
+  autoload :Selector,       "#{base_path}/selector"
+  autoload :Server,         "#{base_path}/server"
+  autoload :Sub,            "#{base_path}/sub"
+  autoload :SubtitleFinder, "#{base_path}/subtitle_finder"
+end

data/lib/osdb/finder.rb

@@ -0,0 +1,8 @@
+module OSDb
+  module Finder
+    base_path = File.expand_path(File.dirname(__FILE__) + '/finder')
+    autoload :First,       "#{base_path}/first"
+    autoload :Interactive, "#{base_path}/interactive"
+    autoload :Score,       "#{base_path}/score"
+  end
+end

data/lib/osdb/finder/first.rb

@@ -0,0 +1,13 @@
+module OSDb
+  module Finder
+
+    class First
+
+      def chose(items)
+        items.first
+      end
+
+    end
+
+  end
+end

data/lib/osdb/finder/interactive.rb

@@ -0,0 +1,21 @@
+module OSDb
+  module Finder
+
+    class Interactive
+
+      def chose(items)
+        puts "D'oh! You stumbled upon a hash conflict, please resolve it:"
+        puts
+        items.each_with_index do |name, index|
+          puts " #{index} - #{name}"
+        end
+        print 'id: '
+        str = STDIN.gets # TODO: rule #1, never trust user input
+        puts
+        items[str.to_i]
+      end
+
+    end
+
+  end
+end

data/lib/osdb/finder/score.rb

@@ -0,0 +1,13 @@
+module OSDb
+  module Finder
+
+    class Score
+
+      def chose(items)
+        items.max_by(&:score)
+      end
+
+    end
+
+  end
+end

data/lib/osdb/movie.rb

@@ -1,61 +1,16 @@
 module OSDb
   class Movie
-    
-    EXTENSIONS = %w(avi mpg m4v mkv mov ogv mp4)
-    
-    attr_reader :path
-    
-    def initialize(path)
-      @path = path
-    end
 
-    def has_sub?
-      exist = false
-      %w(.srt .sub).each{ |ext| exist ||= File.exist?(path.gsub(File.extname(path), ext)) }
-      exist
-    end
+    attr_reader :id, :title, :year, :cover, :rating, :raw_data
 
-    def sub_path(format)
-      path.gsub(File.extname(path), ".#{format}")
-    end
-    
-    def hash
-      @hash ||= self.class.compute_hash(path)
-    end
-    
-    def size
-      @size ||= File.size(path)
+    def initialize(data)
+      @id = data['id']
+      @title = data['title']
+      @year = data['year'] && data['year'].to_i
+      @cover = data['cover']
+      @rating = data['rating'] && data['rating'].to_f
+      @raw_data = data
     end
 
-    def name
-      @name ||= File.basename(path, File.extname(path))
-    end
-    
-    CHUNK_SIZE = 64 * 1024 # in bytes
-    
-    # from http://trac.opensubtitles.org/projects/opensubtitles/wiki/HashSourceCodes
-    def self.compute_hash(path)
-      filesize = File.size(path)
-      hash = filesize
-      
-      # Read 64 kbytes, divide up into 64 bits and add each
-      # to hash. Do for beginning and end of file.
-      File.open(path, 'rb') do |f|    
-        # Q = unsigned long long = 64 bit
-        f.read(CHUNK_SIZE).unpack("Q*").each do |n|
-          hash = hash + n & 0xffffffffffffffff # to remain as 64 bit number
-        end
-        
-        f.seek([0, filesize - CHUNK_SIZE].max, IO::SEEK_SET)
-        
-        # And again for the end of the file
-        f.read(CHUNK_SIZE).unpack("Q*").each do |n|
-          hash = hash + n & 0xffffffffffffffff
-        end
-      end
-      
-      sprintf("%016x", hash)
-    end
-    
   end
 end