anstaendig 0.1.0

data/README.rdoc

@@ -0,0 +1,7 @@
+= anstaendig
+
+== Installation
+
+gem build anstaendig.gemspec
+sudo gem install anstaendig-*.gem
+

data/bin/anstaendig

@@ -0,0 +1,114 @@
+#!/usr/bin/env ruby
+#
+# Script to download videos from various streaming sites.
+#
+# Author:: henning mueller (mailto:henning@orgizm.net)
+# Copyright:: Copyright (c) 2010 henning mueller
+# License:: Beerware
+
+require( 'uri' )
+require( 'getoptlong' )
+
+# Add source lib directory to include paths for require to work if anstaendig
+# not installed as a gem.
+$:.unshift( File.dirname( __FILE__ ) + '/../lib' )
+require( 'anstaendig' )
+include( Anstaendig )
+
+# Shows a summary of the command line options.
+def usageMessage
+	$stderr << "#{File.basename( $0 )} [options] <url>
+Download videos from video streaming sites.
+
+  -h           This help.
+  -l           List available agents.
+  -m           Keep mess (do not clean up on Interrupt).
+  -o name      File to save the video to.
+  -q quality   Quality of the video.
+  -r           Recode to Theora/Vorbis.
+  -s           Only show stream URL.
+
+"
+	exit( 1 )
+end
+
+# Prints available download agents.
+def listAgents
+	puts( AGENTS.keys.sort )
+	exit( 0 )
+end
+
+# Returns a new filename, if the desired one is taken.
+def nextFilename( filename )
+	if( File.exists?( filename ) )
+		new = filename.insert( filename.index( /\./, -5 ), '.new' )
+		nextFilename( new )
+	else
+		return( filename )
+	end
+end
+
+options = GetoptLong.new(
+	[ '-h', GetoptLong::NO_ARGUMENT ],
+	[ '-l', GetoptLong::NO_ARGUMENT ],
+	[ '-m', GetoptLong::NO_ARGUMENT ],
+	[ '-o', GetoptLong::REQUIRED_ARGUMENT ],
+	[ '-q', GetoptLong::REQUIRED_ARGUMENT ],
+	[ '-r', GetoptLong::NO_ARGUMENT ],
+	[ '-s', GetoptLong::NO_ARGUMENT ]
+)
+
+saveAs = ""
+quality = :default
+cleanUp = true
+recodeOGG, showStreamURL = false
+
+options.each do |option, argument|
+	case( option )
+		when '-h'
+			usageMessage
+		when '-l'
+			listAgents
+		when '-m'
+			cleanUp = false
+		when '-o'
+			saveAs = argument
+		when '-q'
+			quality = argument.to_sym
+		when '-r'
+			recodeOGG = true
+		when '-s'
+			showStreamURL = true
+	end
+end
+
+begin
+	uri = URI.parse( ARGV[0] )
+rescue URI::InvalidURIError
+	usageMessage
+end
+
+agent = AGENTS[ uri.host ].new( uri, quality )
+
+if( showStreamURL )
+	puts( agent.url )
+	exit( 0 )
+end
+
+if( saveAs.empty? )
+	agent.filename = nextFilename( agent.filename )
+else
+	agent.filename = saveAs
+end
+
+begin
+	agent.run
+rescue Interrupt
+	if( cleanUp and File.exists?( agent.filename ) )
+		File.unlink( agent.filename )
+	end
+end
+
+if( recodeOGG )
+	agent.recode
+end

data/lib/anstaendig/abstractagent.rb

@@ -0,0 +1,48 @@
+module Anstaendig
+	AGENTS = Hash.new
+
+	# This interface describes download agents.
+	class AbstractAgent
+		attr_accessor :filename, :url
+
+		# Exception should be thrown, if an non-existent quality is demanded
+		# by the user.
+		class NoSuchQualityException
+		end
+
+		# If an agent does not set @filename, it would be the current time
+		# and date.
+		def initialize( url, quality = :high )
+			@filename = Time.now.to_i
+		end
+
+		# Calls the ripping process after substitution of certain characters
+		# in the filename generated by the agent.
+		def run
+			[ '/', '"', "'" ].each do |char|
+				@filename.gsub!( char, '' )
+			end
+
+			rip
+		end
+
+		# Contains the subshell command to rip the video stream. mplayer by
+		# default. Meant to be overloaded by the agent if necessary.
+		def rip
+			`mplayer -user-agent "#{userAgent}" -dumpstream -dumpfile "#{@filename}" "#{@url}"`
+		end
+
+		# Contains the subshell command to recode the video to a Vorbis/Theora
+		# encoded video in an OGG container. Meant to be overloaded by the
+		# agent if necessary.
+		def recode
+			`ffmpeg -i "#{@filename}" -acodec libvorbis -vcodec libtheora "#{@filename}.ogg"`
+		end
+
+		# Return the user agent which is sent on stream retrieval. Meant to be
+		# overloaded by the agent if necessary.
+		def userAgent
+			"Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.2.2) Gecko/20100323 Namoroka/3.6.2"
+		end
+	end
+end

data/lib/anstaendig/agents/3sat.rb

@@ -0,0 +1,31 @@
+module Anstaendig
+	# Agent for anstaendig to download videos from the 3sat media center.
+	#
+	# Author:: henning mueller (mailto:henning@orgizm.net)
+	# Copyright:: Copyright (c) 2010 henning mueller
+	# License:: Beerware
+
+	# 3sat agent class inherits from agent interface.
+	class DreiSatAgent < AbstractAgent
+		AGENTS[ 'www.3sat.de' ] = self
+
+		def initialize( uri, quality )
+			require( 'nokogiri' )
+			require( 'open-uri' )
+			require( 'iconv' )
+
+			uri.query += '&mode=play'
+
+			doc = Nokogiri::HTML( open( uri.to_s ) )
+
+			title = doc.xpath( '//title' ).inner_html
+			title = title.gsub!( '3sat.Mediathek - Video: ', '' ) + '.wmv'
+
+			asx = doc.xpath( '//embed[@id="Player2"]' ).attr( 'data' ) 
+			asx = Nokogiri::HTML( open( asx ) )
+
+			@filename = Iconv.conv( 'utf8', 'latin1', title )
+			@url = asx.xpath( '//ref' ).attr( 'href' )
+		end
+	end
+end

data/lib/anstaendig/agents/ard.rb

@@ -0,0 +1,49 @@
+module Anstaendig
+	# Agent for anstaendig to download videos from the ARD media center.
+	#
+	# Author:: henning mueller (mailto:henning@orgizm.net)
+	# Copyright:: Copyright (c) 2010 henning mueller
+	# License:: Beerware
+
+	# ARD agent class inherits from agent interface.
+	class ArdAgent < AbstractAgent
+		AGENTS[ 'www.ardmediathek.de' ] = self
+
+		def initialize( uri, quality )
+			require( 'nokogiri' )
+			require( 'open-uri' )
+
+			doc = Nokogiri::HTML( open( uri.to_s ) )
+			raw = doc.content.split( "\r\n" ).grep( /rtmp/ )
+
+			@url = case( quality )
+				when :high, :default
+					conjure( raw.last )
+				when :medium
+					begin
+						conjure( raw[1] )
+					rescue( NoMethodError )
+						raise( NoSuchQualityException )
+					end
+				when :low
+					conjure( raw.first )
+				else
+					raise( NoSuchQualityException )
+			end
+
+			title = doc.xpath( '//title' ).inner_text
+			title = title.gsub( 'ARD Mediathek: ', '' )
+			n = title.index( /\ -\ [A-Za-z]*,/ ) - 1
+
+			@filename = title[0..n] + '.flv'
+		end
+
+		def conjure( string )
+			a = string.index( '"' ) + 1
+			b = string.index( '", "' ) - 1
+			c = string.index( '")' ) - 1
+
+			string.slice( a..b ) + string.slice( b+5..c )
+		end
+	end
+end

data/lib/anstaendig/agents/arte.rb

@@ -0,0 +1,49 @@
+module Anstaendig
+	# Agent for anstaendig to download videos from the arte media center.
+	#
+	# Author:: henning mueller (mailto:henning@orgizm.net)
+	# Copyright:: Copyright (c) 2010 henning mueller
+	# License:: Beerware
+
+	#
+	# TODO
+	# This extracts the correct filename and url, but the stream ripping dies
+	# somehow after a few minutes.
+	#
+
+	# arte agent class inherits from agent interface.
+	class ArteAgent < AbstractAgent
+		AGENTS[ 'videos.arte.tv' ] = self
+
+		def initialize( uri, quality )
+			require( 'nokogiri' )
+			require( 'open-uri' )
+
+			lang = ENV['LANG'].slice(0..1)
+
+			doc = open( uri.to_s ).readlines
+
+			doc = Nokogiri::XML( open(
+				doc.grep( /vars_player\.videorefFileUrl/ )
+					.first.match( /(http.*xml)/ )[1]
+			))
+
+			doc.xpath( '//videos/video' ).each do |item|
+				if( item.attr( 'lang' ) == lang )
+					doc = Nokogiri::XML( open( item.attr( 'ref' ) ) )
+				end
+			end
+
+			@filename = doc.xpath( '//name' ).first.text + '.wmv'
+
+			case( quality )
+				when :hd, :high, :default
+					@url = doc.xpath( '//urls/url[@quality="hd"]' ).text
+				when :sd, :medium, :low
+					@url = doc.xpath( '//urls/url[@quality="sd"]' ).text
+				else
+					raise( NoSuchQualityException )
+			end
+		end
+	end
+end

data/lib/anstaendig/agents/zdf.rb

@@ -0,0 +1,47 @@
+module Anstaendig
+	# Agent for anstaendig to download videos from the ZDF media center.
+	#
+	# Author:: henning mueller (mailto:henning@orgizm.net)
+	# Copyright:: Copyright (c) 2010 henning mueller
+	# License:: Beerware
+
+	# ZDF agent class inherits from agent interface.
+	class ZdfAgent < AbstractAgent
+		AGENTS[ 'www.zdf.de' ] = self
+
+		def initialize( uri, quality )
+			require( 'nokogiri' )
+			require( 'open-uri' )
+
+			uri = uri.to_s
+
+			if( uri =~ /hauptnavigation\/startseite\/#/ )
+				uri.gsub!( 'hauptnavigation/startseite/#/', '' )
+			end
+
+			doc = Nokogiri::HTML( open( uri ) )
+
+			quality = case( quality )
+				when :high, :default
+					'/veryhigh/'
+				when :medium, :low
+					'/300/'
+				else
+					raise( NoSuchQualityException )
+			end
+
+			doc.xpath( '//a[@class="play"]' ).each do |temp|
+				temp = temp.attribute( 'href' ).to_s
+				if( temp.include?( quality ) and temp.include?( '.asx' ) )
+					uri = temp
+				end
+			end
+
+			@filename = doc.xpath( '//h1[@class="beitragHeadline"]' ).inner_text + '.wmv'
+
+			doc = Nokogiri::HTML( open( uri ) )
+	#		@url = doc.xpath( '//ref' ).attr( 'href' )
+			@url = doc.to_s.match( /(mms.*wmv)/ )[0]
+		end
+	end
+end

data/lib/anstaendig.rb

@@ -0,0 +1,11 @@
+$:.unshift( File.dirname( __FILE__ ) )
+
+# The Anstaendig module includes an abstract agent class and all agents
+# implementing this.
+module Anstaendig
+	VERSION = '0.1.0'
+end
+
+Dir.glob( $:.first + '/**/*.rb' ).sort.each do |file|
+	require( file )
+end

metadata

@@ -0,0 +1,74 @@
+--- !ruby/object:Gem::Specification 
+name: anstaendig
+version: !ruby/object:Gem::Version 
+  hash: 27
+  prerelease: false
+  segments: 
+  - 0
+  - 1
+  - 0
+  version: 0.1.0
+platform: ruby
+authors: 
+- henning mueller
+autorequire: 
+bindir: bin
+cert_chain: []
+
+date: 2010-05-31 00:00:00 +02:00
+default_executable: 
+dependencies: []
+
+description: anstaendig is a extendable script to rip movie clips from TV station media center sites.
+email: henning@orgizm.net
+executables: 
+- anstaendig
+extensions: []
+
+extra_rdoc_files: []
+
+files: 
+- bin/anstaendig
+- lib/anstaendig.rb
+- lib/anstaendig/agents/zdf.rb
+- lib/anstaendig/agents/ard.rb
+- lib/anstaendig/agents/3sat.rb
+- lib/anstaendig/agents/arte.rb
+- lib/anstaendig/abstractagent.rb
+- README.rdoc
+has_rdoc: true
+homepage: 
+licenses: []
+
+post_install_message: 
+rdoc_options: []
+
+require_paths: 
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement 
+  none: false
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      hash: 3
+      segments: 
+      - 0
+      version: "0"
+required_rubygems_version: !ruby/object:Gem::Requirement 
+  none: false
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      hash: 3
+      segments: 
+      - 0
+      version: "0"
+requirements: []
+
+rubyforge_project: 
+rubygems_version: 1.3.7
+signing_key: 
+specification_version: 3
+summary: Rip clips from TV station media center sites.
+test_files: []
+