suby 0.0.8 → 0.1.0

data/.gitignore

@@ -1 +1,3 @@
 *.gem
+spec/fixtures/*
+!spec/fixtures/.gitkeep

data/README.md

@@ -2,7 +2,7 @@
 
 Find and download subtitles
 
-suby is a little script to find and download subtitles for TV series
+`suby` is a little script to find and download subtitles for TV series
 
 ## Install
 
@@ -10,14 +10,18 @@ suby is a little script to find and download subtitles for TV series
 
 ## Synopsis
 
-    suby 'my show 1x01 - Pilot.avi' # => Downloads my show 1x01 - Pilot.srt
+    suby 'My Show 1x01 - Pilot.avi' # => Downloads 'My Show 1x01 - Pilot.srt'
 
-## Status
+## Features
 
-Under heavy development.
+* Parse filename to detect show, season and episode
+* Search and download appropriate subtitle, extracting it from the archive and renaming it
+* Accept a lang option (defaults to en)
+* Try multiple sites, falling back on the next one if it was not found on the current
+* Detailed error messages
 
-* Currently only accept one format for input.  
-  You could use [tvnamer](https://github.com/dbr/tvnamer) to rename the video files before.
+## TODO
 
-* Search only on tvsubtitles.net.  
-  Many series have subtitles there, but not all.
+* usual movies support (via opensubtitles.org)
+* multi-episodes support
+* choose wiser the right subtitle if many are available

data/lib/suby/downloader/addic7ed.rb

@@ -1,5 +1,7 @@
 module Suby
   class Downloader::Addic7ed < Downloader
+    Downloader.add(self)
+
     SITE = 'www.addic7ed.com'
     FORMAT = :file
     LANG_IDS = {
@@ -7,21 +9,45 @@ module Suby
       es: 5,
       fr: 8
     }
-    FILTER_IGNORED = "Couldn't find any subs with the specified language. Filter ignored"
+    FILTER_IGNORED = "Couldn't find any subs with the specified language. " +
+                     "Filter ignored"
+
+    def subtitles_url
+      "/serie/#{CGI.escape show}/#{season}/#{episode}/#{LANG_IDS[lang]}"
+    end
+
+    def subtitles_response
+      response = get(subtitles_url, {}, false)
+      unless Net::HTTPSuccess === response
+        raise NotFoundError, "show/season/episode not found"
+      end
+      response
+    end
+
+    def subtitles_body
+      body = subtitles_response.body
+      if body.include? FILTER_IGNORED
+        raise NotFoundError, "no subtitle available"
+      end
+      body
+    end
+
+    def redirected_url download_url
+      header = { 'Referer' => "http://#{SITE}#{subtitles_url}" }
+      location = get_redirection download_url, header # They check Referer
+      if location == '/downloadexceeded.php'
+        raise NotFoundError, "download exceeded"
+      end
+      URI.escape location
+    end
 
     def download_url
-      subtitles_url = "/serie/#{CGI.escape show}/#{season}/#{episode}/#{LANG_IDS[lang]}"
-      response = http.get(subtitles_url)
-      throw :downloader, "show/season/episode not found" unless Net::HTTPSuccess === response
-      body = response.body
-      throw :downloader, "no subtitle available" if body.include? FILTER_IGNORED
-      download_url = Nokogiri(body).css('a').find { |a|
+      download_url = Nokogiri(subtitles_body).css('a').find { |a|
         a[:href].start_with? '/original/' or
         a[:href].start_with? '/updated/'
       }[:href]
-      location = get_redirection download_url, 'Referer' => "http://#{SITE}#{subtitles_url}" # They check Referer
-      throw :downloader, "download exceeded" if location == '/downloadexceeded.php'
-      URI.escape location
+
+      redirected_url download_url
     end
   end
 end

data/lib/suby/downloader/tvsubtitles.rb

@@ -1,5 +1,7 @@
 module Suby
   class Downloader::TVSubtitles < Downloader
+    Downloader.add(self)
+
     SITE = 'www.tvsubtitles.net'
     FORMAT = :zip
     SEARCH_URL = '/search.php'
@@ -8,20 +10,22 @@ module Suby
     SHOW_URLS = {}
     SHOW_PAGES = {}
 
+    # "Show (2009-2011)" => "Show"
+    def clean_show_name show_name
+      show_name.sub(/ \(\d{4}-\d{4}\)$/, '')
+    end
+
     def show_url
       SHOW_URLS[show] ||= begin
-        post = Net::HTTP::Post.new(SEARCH_URL)
-        post.form_data = { 'q' => show }
-        results = Nokogiri http.request(post).body
+        results = Nokogiri(post(SEARCH_URL, 'q' => show))
         a = results.css('ul li div a').find { |a|
-           # "Show (2009-2011)" => "Show"
-          a.text.sub(/ \(\d{4}-\d{4}\)$/, '').casecmp(show) == 0
+          clean_show_name(a.text).casecmp(show) == 0
         }
-        throw :downloader, "show not found" unless a
+        raise NotFoundError, "show not found" unless a
         url = a[:href]
 
-        raise 'could not find the show' unless /^\/tvshow-(\d+)\.html$/ =~ url
-        "/tvshow-#{$1}.html"
+        raise 'invalid show url' unless /^\/tvshow-\d+\.html$/ =~ url
+        url
       end
     end
 
@@ -29,37 +33,51 @@ module Suby
       show_url.sub(/\.html$/, "-#{season}.html")
     end
 
+    def has_season?
+      season_text = "Season #{season}"
+      bs = SHOW_PAGES[show].css('div.left_articles p.description b')
+      unless bs.find { |b| b.text == season_text }
+        raise NotFoundError, "season not found"
+      end
+    end
+
+    def find_episode_row
+      row = SHOW_PAGES[show].css('div.left_articles table tr').find { |tr|
+        tr.children.find { |td| td.name == 'td' and
+                                td.text =~ /\A#{season}x0?#{episode}\z/ }
+      }
+      raise NotFoundError, "episode not found" unless row
+      row
+    end
+
     def episode_url
       @episode_url ||= begin
-        SHOW_PAGES[show] ||= Nokogiri get season_url
-
-        season_text = /^Season #{season}$/
-        SHOW_PAGES[show].css('div.left_articles p.description b').find { |b|
-          b.text =~ season_text
-        } or throw :downloader, "season not found"
+        SHOW_PAGES[show] ||= Nokogiri(get(season_url))
+        has_season?
 
         url = nil
-        SHOW_PAGES[show].css('div.left_articles table tr').find { |tr|
-          tr.children.find { |td| td.name == 'td' && td.text =~ /\A#{season}x0?#{episode}\z/ }
-        }.tap { |tr|
-          throw :downloader, "episode not found" unless tr
-        }.children.find { |td|
+        find_episode_row.children.find { |td|
           td.children.find { |a|
-            a.name == 'a' && a[:href].start_with?('episode') && url = a[:href]
+            a.name == 'a' and a[:href].start_with?('episode') and url = a[:href]
           }
         }
-        raise "invalid episode url: #{episode_url}" unless url =~ /^episode-(\d+)\.html$/
+        unless url =~ /^episode-(\d+)\.html$/
+          raise "invalid episode url: #{episode_url}"
+        end
+
         "/episode-#{$1}-#{lang}.html"
       end
     end
 
     def subtitles_url
       @subtitles_url ||= begin
-        subtitles = Nokogiri get episode_url
+        subtitles = Nokogiri(get(episode_url))
 
         # TODO: choose 720p or most downloaded instead of first found
-        a = subtitles.css('div.left_articles a').find { |a| a.name == 'a' && a[:href].start_with?('/subtitle') }
-        throw :downloader, "no subtitle available" unless a
+        a = subtitles.css('div.left_articles a').find { |a|
+          a.name == 'a' and a[:href].start_with?('/subtitle')
+        }
+        raise NotFoundError, "no subtitle available" unless a
         url = a[:href]
         raise 'invalid subtitle url' unless url =~ /^\/subtitle-(\d+)\.html/
         url
@@ -67,7 +85,8 @@ module Suby
     end
 
     def download_url
-      @download_url ||= URI.escape '/' + get_redirection(subtitles_url.sub('subtitle', 'download'))
+      @download_url ||= URI.escape '/' +
+                      get_redirection(subtitles_url.sub('subtitle', 'download'))
     end
   end
 end

data/lib/suby/downloader.rb

@@ -6,8 +6,8 @@ require_relative 'filename_parser'
 module Suby
   class Downloader
     DOWNLOADERS = []
-    def self.inherited(subclass)
-      DOWNLOADERS << subclass
+    def self.add(downloader)
+      DOWNLOADERS << downloader
     end
 
     attr_reader :show, :season, :episode, :file, :lang
@@ -21,17 +21,36 @@ module Suby
       @http ||= Net::HTTP.new(self.class::SITE).start
     end
 
-    def get(path, initheader = {})
+    def get(path, initheader = {}, parse_response = true)
       response = http.get(path, initheader)
-      raise "Invalid response for #{path}: #{response}" unless Net::HTTPSuccess === response
+      if parse_response
+        unless Net::HTTPSuccess === response
+          raise DownloaderError, "Invalid response for #{path}: #{response}"
+        end
+        response.body
+      else
+        response
+      end
+    end
+
+    def post(path, data = {}, initheader = {})
+      post = Net::HTTP::Post.new(path, initheader)
+      post.form_data = data
+      response = http.request(post)
+      unless Net::HTTPSuccess === response
+        raise DownloaderError, "Invalid response for #{path}(#{data}): " +
+                               response.inspect
+      end
       response.body
     end
 
     def get_redirection(path, initheader = {})
       response = http.get(path, initheader)
       location = response['Location']
-      unless (Net::HTTPFound === response or Net::HTTPSuccess === response) and location
-        raise "Invalid response for #{path}: #{response}: location: #{location.inspect}"
+      unless (Net::HTTPFound === response or
+              Net::HTTPSuccess === response) and location
+        raise DownloaderError, "Invalid response for #{path}: " +
+              "#{response}: location: #{location.inspect}, #{response.body}"
       end
       location
     end
@@ -44,17 +63,23 @@ module Suby
       contents = get(url)
       http.finish
       format = self.class::FORMAT
-      if format == :file
+      case format
+      when :file
         open(sub_name(url), 'wb') { |f| f.write contents }
-      else
+      when :zip
         open(TEMP_ARCHIVE_NAME, 'wb') { |f| f.write contents }
-        sub = Suby.extract_sub_from_archive(TEMP_ARCHIVE_NAME, format)
-        File.rename sub, sub_name(sub)
+        Suby.extract_sub_from_archive(TEMP_ARCHIVE_NAME, format, basename)
+      else
+        raise "unknown subtitles format: #{format}"
       end
     end
 
+    def basename
+      File.basename(file, File.extname(file))
+    end
+
     def sub_name(sub)
-      File.basename(file, File.extname(file)) + File.extname(sub)
+      basename + File.extname(sub)
     end
   end
 end

data/lib/suby/downloader_error.rb

@@ -0,0 +1,4 @@
+module Suby
+  class DownloaderError < StandardError
+  end
+end

data/lib/suby/filename_parser.rb

@@ -79,9 +79,10 @@ module Suby
 
     def parse(file)
       filename = File.basename(file)
-      FILENAME_PATTERNS.find { |pattern|
+      found = FILENAME_PATTERNS.find { |pattern|
         pattern =~ filename
-      } or raise "wrong file format (#{file})"
+      }
+      raise "wrong file format (#{file})" unless found
       [clean_show_name($~[:show]), $~[:season].to_i, $~[:episode].to_i]
     end
 

data/lib/suby/not_found_error.rb

@@ -0,0 +1,4 @@
+module Suby
+  class NotFoundError < StandardError
+  end
+end

data/lib/suby.rb

@@ -1,4 +1,7 @@
+require_relative 'suby/downloader_error'
+require_relative 'suby/not_found_error'
 require_relative 'suby/downloader'
+require 'zip/zip'
 
 module Suby
   extend self
@@ -8,46 +11,61 @@ module Suby
 
   def download_subtitles(files, options = {})
     files.each { |file|
-      next if SUB_EXTENSIONS.include? File.extname(file)
       next puts "Skipping: #{file}" if SUB_EXTENSIONS.any? { |ext|
-        File.exist? File.basename(file, File.extname(file)) + ".#{ext}" }
+        File.exist? File.basename(file, File.extname(file)) + ".#{ext}"
+      }
+      download_subtitles_for_file(file, options)
+    }
+  end
 
-      begin
-        success = Downloader::DOWNLOADERS.find do |downloader|
-          error = catch :downloader do
-            downloader.new(file, options[:lang]).download
-            :success
-          end
-          if error == :success
-            puts "#{downloader} found subtitles for #{file}"
-          else
-            puts "#{downloader} did not find subtitles for #{file} (#{error})"
-          end
-          error == :success
-        end
-        STDERR.puts "No downloader could find subtitles for #{file}" unless success
-      rescue
-        puts "  The download of the subtitles failed for #{file}:"
-        puts "  #{$!.class}: #{$!.message}"
-        puts $!.backtrace.map { |line| line.prepend ' '*4 }
+  def download_subtitles_for_file(file, options)
+    begin
+      success = Downloader::DOWNLOADERS.find { |downloader_class|
+        try_downloader(downloader_class.new(file, options[:lang]))
+      }
+      unless success
+        STDERR.puts "No downloader could find subtitles for #{file}"
       end
-    }
+    rescue
+      puts "  The download of the subtitles failed for #{file}:"
+      puts "  #{$!.class}: #{$!.message}"
+      puts $!.backtrace.map { |line| line.prepend ' '*4 }
+    end
   end
 
-  def extract_sub_from_archive(archive, format)
+  def try_downloader(downloader)
+    begin
+      downloader.download
+    rescue Suby::NotFoundError => error
+      puts "#{downloader.class} did not find subtitles for " +
+           "#{downloader.file} (#{error.message})"
+      false
+    rescue Suby::DownloaderError => error
+      puts "#{downloader.class} had a problem finding subtitles for " +
+           "#{downloader.file} (#{error.message})"
+      false
+    else
+      puts "#{downloader.class} found subtitles for #{downloader.file}"
+      true
+    end
+  end
+
+  def extract_sub_from_archive(archive, format, basename)
     case format
     when :zip
-      sub = `unzip -qql #{archive}`.scan(/\d{2}:\d{2}   (.+?(?:#{SUB_EXTENSIONS.join '|'}))$/).map(&:first).first
-      raise "no subtitles in #{archive}" unless sub
-      sub_for_unzip = sub.gsub(/(\[|\])/) { "\\#{$1}" }
-      system 'unzip', archive, sub_for_unzip, 1 => :close
-      puts "found subtitle: #{sub}" if $VERBOSE
+      Zip::ZipFile.open(archive) { |zip|
+        sub = zip.entries.find { |entry|
+          entry.to_s =~ /\.#{Regexp.union SUB_EXTENSIONS}$/
+        }
+        raise "no subtitles in #{archive}" unless sub
+        name = basename + File.extname(sub.to_s)
+        sub.extract(name)
+      }
     else
       raise "unknown archive type (#{archive})"
     end
-
+  ensure
     # Cleaning
-    File.unlink archive
-    sub
+    File.unlink archive if File.exist? archive
   end
 end

data/suby.gemspec

@@ -3,12 +3,15 @@ Gem::Specification.new do |s|
   s.summary = "Subtitles' downloader"
   s.description = "Find and download subtitles"
   s.author = 'eregon'
+  s.email = 'eregontp@gmail.com'
+  s.homepage = 'https://github.com/eregon/suby'
 
   s.files = Dir['bin/*', 'lib/**/*.rb', '.gitignore', 'README.md', 'suby.gemspec']
   s.executables = ['suby']
 
   s.required_ruby_version = '>= 1.9.2'
   s.add_dependency 'nokogiri'
+  s.add_dependency 'rubyzip'
 
-  s.version = '0.0.8'
+  s.version = '0.1.0'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: suby
 version: !ruby/object:Gem::Version
-  version: 0.0.8
+  version: 0.1.0
   prerelease: 
 platform: ruby
 authors:
@@ -9,11 +9,11 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2011-07-27 00:00:00.000000000 Z
+date: 2011-07-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: nokogiri
-  requirement: &2157439960 !ruby/object:Gem::Requirement
+  requirement: &2153546620 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -21,9 +21,20 @@ dependencies:
         version: '0'
   type: :runtime
   prerelease: false
-  version_requirements: *2157439960
+  version_requirements: *2153546620
+- !ruby/object:Gem::Dependency
+  name: rubyzip
+  requirement: &2153546120 !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :runtime
+  prerelease: false
+  version_requirements: *2153546120
 description: Find and download subtitles
-email: 
+email: eregontp@gmail.com
 executables:
 - suby
 extensions: []
@@ -33,12 +44,14 @@ files:
 - lib/suby/downloader/addic7ed.rb
 - lib/suby/downloader/tvsubtitles.rb
 - lib/suby/downloader.rb
+- lib/suby/downloader_error.rb
 - lib/suby/filename_parser.rb
+- lib/suby/not_found_error.rb
 - lib/suby.rb
 - .gitignore
 - README.md
 - suby.gemspec
-homepage: 
+homepage: https://github.com/eregon/suby
 licenses: []
 post_install_message: 
 rdoc_options: []
@@ -58,7 +71,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 1.8.5.1
+rubygems_version: 1.8.6
 signing_key: 
 specification_version: 3
 summary: Subtitles' downloader