ruby-oembed 0.7.0

data/lib/oembed.rb

@@ -0,0 +1,14 @@
+$:.unshift File.dirname(__FILE__)
+
+require 'net/http'
+
+require 'oembed/errors'
+require 'oembed/formatters'
+require 'oembed/provider'
+require 'oembed/provider_discovery'
+require 'oembed/providers'
+require 'oembed/response'
+require 'oembed/response/photo'
+require 'oembed/response/video'
+require 'oembed/response/link'
+require 'oembed/response/rich'

data/lib/oembed/embedly_urls.json

@@ -0,0 +1,143 @@
+[
+    "http://*youtube.com/watch*", 
+    "http://*.youtube.com/v/*", 
+    "http://youtu.be/*", 
+    "http://*.youtube.com/user/*#*", 
+    "http://*.youtube.com/*#*/*", 
+    "http://www.veoh.com/*/watch/*", 
+    "http://*justin.tv/*", 
+    "http://*justin.tv/*/b/*", 
+    "http://www.ustream.tv/recorded/*", 
+    "http://www.ustream.tv/channel/*", 
+    "http://qik.com/video/*", 
+    "http://qik.com/*", 
+    "http://*revision3.com/*", 
+    "http://*.dailymotion.com/video/*", 
+    "http://*.dailymotion.com/*/video/*", 
+    "http://www.collegehumor.com/video:*", 
+    "http://*twitvid.com/*", 
+    "http://www.break.com/*/*", 
+    "http://vids.myspace.com/index.cfm?fuseaction=vids.individual&videoid*", 
+    "http://www.myspace.com/index.cfm?fuseaction=*&videoid*", 
+    "http://www.metacafe.com/watch/*", 
+    "http://blip.tv/file/*", 
+    "http://*.blip.tv/file/*", 
+    "http://video.google.com/videoplay?*", 
+    "http://*revver.com/video/*", 
+    "http://video.yahoo.com/watch/*/*", 
+    "http://video.yahoo.com/network/*", 
+    "http://*viddler.com/explore/*/videos/*", 
+    "http://liveleak.com/view?*", 
+    "http://www.liveleak.com/view?*", 
+    "http://animoto.com/play/*", 
+    "http://dotsub.com/view/*", 
+    "http://www.overstream.net/view.php?oid=*", 
+    "http://*yfrog.*/*", 
+    "http://tweetphoto.com/*", 
+    "http://www.flickr.com/photos/*", 
+    "http://*twitpic.com/*", 
+    "http://*imgur.com/*", 
+    "http://*.posterous.com/*", 
+    "http://post.ly/*", 
+    "http://twitgoo.com/*", 
+    "http://i*.photobucket.com/albums/*", 
+    "http://gi*.photobucket.com/groups/*", 
+    "http://phodroid.com/*/*/*", 
+    "http://www.mobypicture.com/user/*/view/*", 
+    "http://moby.to/*", 
+    "http://xkcd.com/*", 
+    "http://www.xkcd.com/*", 
+    "http://www.asofterworld.com/index.php?id=*", 
+    "http://www.qwantz.com/index.php?comic=*", 
+    "http://23hq.com/*/photo/*", 
+    "http://www.23hq.com/*/photo/*", 
+    "http://*dribbble.com/shots/*", 
+    "http://drbl.in/*", 
+    "http://*.smugmug.com/*", 
+    "http://*.smugmug.com/*#*", 
+    "http://emberapp.com/*/images/*", 
+    "http://emberapp.com/*/images/*/sizes/*", 
+    "http://emberapp.com/*/collections/*/*", 
+    "http://emberapp.com/*/categories/*/*/*", 
+    "http://embr.it/*", 
+    "http://picasaweb.google.com*/*/*#*", 
+    "http://picasaweb.google.com*/lh/photo/*", 
+    "http://picasaweb.google.com*/*/*", 
+    "http://www.whitehouse.gov/photos-and-video/video/*", 
+    "http://www.whitehouse.gov/video/*", 
+    "http://wh.gov/photos-and-video/video/*", 
+    "http://wh.gov/video/*", 
+    "http://www.hulu.com/watch*", 
+    "http://www.hulu.com/w/*", 
+    "http://hulu.com/watch*", 
+    "http://hulu.com/w/*", 
+    "http://movieclips.com/watch/*/*/", 
+    "http://movieclips.com/watch/*/*/*/*", 
+    "http://*crackle.com/c/*", 
+    "http://www.fancast.com/*/videos", 
+    "http://www.funnyordie.com/videos/*", 
+    "http://www.vimeo.com/groups/*/videos/*", 
+    "http://www.vimeo.com/*", 
+    "http://vimeo.com/groups/*/videos/*", 
+    "http://vimeo.com/*", 
+    "http://www.ted.com/talks/*.html*", 
+    "http://www.ted.com/talks/lang/*/*.html*", 
+    "http://www.ted.com/index.php/talks/*.html*", 
+    "http://www.ted.com/index.php/talks/lang/*/*.html*", 
+    "http://*omnisio.com/*", 
+    "http://*nfb.ca/film/*", 
+    "http://www.thedailyshow.com/watch/*", 
+    "http://www.thedailyshow.com/full-episodes/*", 
+    "http://www.thedailyshow.com/collection/*/*/*", 
+    "http://movies.yahoo.com/movie/*/video/*", 
+    "http://movies.yahoo.com/movie/*/info", 
+    "http://movies.yahoo.com/movie/*/trailer", 
+    "http://www.colbertnation.com/the-colbert-report-collections/*", 
+    "http://www.colbertnation.com/full-episodes/*", 
+    "http://www.colbertnation.com/the-colbert-report-videos/*", 
+    "http://www.comedycentral.com/videos/index.jhtml?*", 
+    "http://www.theonion.com/video/*", 
+    "http://theonion.com/video/*", 
+    "http://wordpress.tv/*/*/*/*/", 
+    "http://www.traileraddict.com/trailer/*", 
+    "http://www.traileraddict.com/clip/*", 
+    "http://www.traileraddict.com/poster/*", 
+    "http://www.escapistmagazine.com/videos/*", 
+    "http://www.trailerspy.com/trailer/*/*", 
+    "http://www.trailerspy.com/trailer/*", 
+    "http://www.trailerspy.com/view_video.php*", 
+    "http://soundcloud.com/*", 
+    "http://soundcloud.com/*/*", 
+    "http://soundcloud.com/*/sets/*", 
+    "http://soundcloud.com/groups/*", 
+    "http://www.lala.com/#album/*", 
+    "http://www.lala.com/album/*", 
+    "http://www.lala.com/#song/*", 
+    "http://www.lala.com/song/*", 
+    "http://www.mixcloud.com/*/*/", 
+    "http://*amazon.*/gp/product/*", 
+    "http://*amazon.*/*/dp/*", 
+    "http://*amazon.*/dp/*", 
+    "http://*amazon.*/o/ASIN/*", 
+    "http://*amazon.*/gp/offer-listing/*", 
+    "http://*amazon.*/*/ASIN/*", 
+    "http://*amazon.*/gp/product/images/*", 
+    "http://www.amzn.com/*", 
+    "http://amzn.com/*", 
+    "http://twitter.com/*/status/*", 
+    "http://twitter.com/*/statuses/*", 
+    "http://www.slideshare.net/*/*", 
+    "http://*.scribd.com/doc/*", 
+    "http://screenr.com/*", 
+    "http://www.5min.com/Video/*", 
+    "http://www.howcast.com/videos/*", 
+    "http://www.screencast.com/*/media/*", 
+    "http://screencast.com/*/media/*", 
+    "http://www.screencast.com/t/*", 
+    "http://screencast.com/t/*", 
+    "http://www.clearspring.com/widgets/*", 
+    "http://my.opera.com/*/albums/show.dml?id=*", 
+    "http://my.opera.com/*/albums/showpic.dml?album=*&picture=*", 
+    "http://tumblr.com/*", 
+    "http://*.tumblr.com/post/*"
+]

data/lib/oembed/errors.rb

@@ -0,0 +1,28 @@
+module OEmbed
+  class Error < StandardError
+  end
+
+  class NotFound < OEmbed::Error
+    def to_s
+      "No embeddable content at '#{super}'"
+    end
+  end
+
+  class UnknownFormat < OEmbed::Error
+    def to_s
+      "The provider doesn't support the '#{super}' format"
+    end
+  end
+
+  class FormatNotSupported < OEmbed::Error
+    def to_s
+      "This server doesn't have the correct libraries installed to support the '#{super}' format"
+    end
+  end
+
+  class UnknownResponse < OEmbed::Error
+    def to_s
+      "Got unknown response (#{super}) from server"
+    end
+  end
+end

data/lib/oembed/formatters.rb

@@ -0,0 +1,29 @@
+module OEmbed
+  class Formatters
+    FORMATS = Hash.new { |_, format| raise OEmbed::FormatNotSupported, format }
+
+    # Load XML
+    begin
+      require 'xmlsimple'
+      FORMATS[:xml] = proc { |r| XmlSimple.xml_in(r, 'ForceArray' => false)}
+    rescue LoadError
+    end
+
+    # Load JSON
+    begin
+      require 'json'
+      FORMATS[:json] = proc { |r| ::JSON.load(r) }
+    rescue LoadError
+    end
+
+    DEFAULT = FORMATS.keys.first
+
+    def self.verify?(type)
+      FORMATS[type] && type
+    end
+
+    def self.convert(type, value)
+      FORMATS[type].call(value)
+    end
+  end
+end

data/lib/oembed/provider.rb

@@ -0,0 +1,86 @@
+module OEmbed
+  class Provider
+    attr_accessor :format, :name, :url, :urls, :endpoint
+
+    def initialize(endpoint, format = OEmbed::Formatters::DEFAULT)
+      @endpoint = endpoint
+      @urls = []
+      # Try to use the best available format
+      @format = OEmbed::Formatters.verify?(format)
+    end
+
+    def <<(url)
+      if url.is_a? Regexp
+        @urls << url
+        return
+      end
+      full, scheme, domain, path = *url.match(%r{([^:]*)://?([^/?]*)(.*)})
+      domain = Regexp.escape(domain).gsub("\\*", "(.*?)").gsub("(.*?)\\.", "([^\\.]+\\.)?")
+      path = Regexp.escape(path).gsub("\\*", "(.*?)")
+      @urls << Regexp.new("^#{Regexp.escape(scheme)}://#{domain}#{path}")
+    end
+
+    def build(url, options = {})
+      raise OEmbed::NotFound, url unless include?(url)
+      query = options.merge({:url => url})
+      endpoint = @endpoint.clone
+
+      if format_in_url?
+        format = endpoint["{format}"] = (query[:format] || @format).to_s
+        query.delete(:format)
+      else
+        format = query[:format] ||= @format
+      end
+
+      query_string = "?" + query.inject("") do |memo, (key, value)|
+        "#{key}=#{value}&#{memo}"
+      end.chop
+
+      URI.parse(endpoint + query_string).instance_eval do
+        @format = format; def format; @format; end
+        self
+      end
+    end
+
+    def raw(url, options = {})
+      uri = build(url, options)
+      
+      found = false
+      max_redirects = 4
+      until found
+        host, port = uri.host, uri.port if uri.host && uri.port
+        res = Net::HTTP.start(uri.host, uri.port) {|http|  http.get(uri.request_uri) }
+        res.header['location'] ? uri = URI.parse(res.header['location']) : found = true
+        if max_redirects == 0
+            found = true
+        else
+            max_redirects = max_redirects - 1
+        end
+      end
+      
+      case res
+      when Net::HTTPNotImplemented
+        raise OEmbed::UnknownFormat, uri.format
+      when Net::HTTPNotFound
+        raise OEmbed::NotFound, url
+      when Net::HTTPOK
+        res.body
+      else
+        raise OEmbed::UnknownResponse, res.code
+      end
+    end
+
+    def get(url, options = {})
+      options[:format] ||= @format if @format
+      OEmbed::Response.create_for(raw(url, options), self, options[:format])
+    end
+
+    def format_in_url?
+      @endpoint.include?("{format}")
+    end
+
+    def include?(url)
+      @urls.empty? || !!@urls.detect{ |u| u =~ url }
+    end
+  end
+end

data/lib/oembed/provider_discovery.rb

@@ -0,0 +1,55 @@
+module OEmbed
+  class ProviderDiscovery
+    class << self
+    def raw(url, options = {})
+      provider = discover_provider(url, options)
+      provider.raw(url, options)
+    end
+
+    def get(url, options = {})
+      provider = discover_provider(url, options)
+      provider.get(url, options)
+    end
+
+    def discover_provider(url, options = {})
+      uri = URI.parse(url)
+
+      res = Net::HTTP.start(uri.host, uri.port) do |http|
+        http.get(uri.request_uri)
+      end
+
+      case res
+      when Net::HTTPNotFound
+        raise OEmbed::NotFound, url
+      when Net::HTTPOK
+        format = options[:format]
+
+        if format.nil? || format == :json
+          provider_endpoint ||= /<link.*href=['"]*([^\s'"]+)['"]*.*application\/json\+oembed.*>/.match(res.body)
+          provider_endpoint ||= /<link.*application\/json\+oembed.*href=['"]*([^\s'"]+)['"]*.*>/.match(res.body)
+          format ||= :json if provider_endpoint
+        end
+        if format.nil? || format == :xml
+          provider_endpoint ||= /<link.*href=['"]*([^\s'"]+)['"]*.*application\/xml\+oembed.*>/.match(res.body)
+          provider_endpoint ||= /<link.*application\/xml\+oembed.*href=['"]*([^\s'"]+)['"]*.*>/.match(res.body)
+          format ||= :xml if provider_endpoint
+        end
+
+        begin
+          provider_endpoint = URI.parse(provider_endpoint && provider_endpoint[1])
+          provider_endpoint.query = nil
+          provider_endpoint = provider_endpoint.to_s
+        rescue URI::Error
+          raise OEmbed::NotFound, url
+        end
+
+
+        Provider.new(provider_endpoint, format || OEmbed::Formatters::DEFAULT)
+      else
+        raise OEmbed::UnknownResponse, res.code
+      end
+    end
+
+    end
+  end
+end

data/lib/oembed/providers.rb

@@ -0,0 +1,156 @@
+require 'rubygems'
+require 'json'
+
+module OEmbed
+  class Providers
+    class << self
+      @@urls = {}
+      @@fallback = []
+
+      def urls
+        @@urls
+      end
+
+      def register(*providers)
+        providers.each do |provider|
+          provider.urls.each do |url|
+            @@urls[url] = provider
+          end
+        end
+      end
+
+      def unregister(*providers)
+        providers.each do |provider|
+          provider.urls.each do |url|
+            @@urls.delete(url)
+          end
+        end
+      end
+
+      def register_all
+        register(Youtube, Flickr, Viddler, Qik, Pownce, Revision3, Hulu, Vimeo, PollEverywhere, MyOpera, ClearspringWidgets, MovieClips, TwentyThree)
+      end
+
+      # Takes an array of OEmbed::Provider instances or OEmbed::ProviderDiscovery
+      # Use this method to register fallback providers.
+      # When the raw or get methods are called, if the URL doesn't match
+      # any of the registerd url patters the fallback providers
+      # will be called (in order) with the URL.
+      #
+      # A common example:
+      #  OEmbed::Providers.register_fallback(OEmbed::ProviderDiscovery, OEmbed::Providers::OohEmbed)
+      def register_fallback(*providers)
+        @@fallback += providers
+      end
+
+      # Returns an array of all registerd fallback providers
+      def fallback
+        @@fallback
+      end
+
+      def find(url)
+        @@urls[@@urls.keys.detect { |u| u =~ url }] || false
+      end
+
+      def raw(url, options = {})
+        provider = find(url)
+        if provider
+          provider.raw(url, options)
+        else
+          fallback.each do |p|
+            return p.raw(url, options) rescue OEmbed::Error
+          end
+          raise(OEmbed::NotFound)
+        end
+      end
+
+      def get(url, options = {})
+        provider = find(url)
+        if provider
+          provider.get(url, options)
+        else
+          fallback.each do |p|
+            return p.get(url, options) rescue OEmbed::Error
+          end
+          raise(OEmbed::NotFound)
+        end
+      end
+    end
+
+    # Custom providers:
+    Youtube = OEmbed::Provider.new("http://www.youtube.com/oembed/")
+    Youtube << "http://*.youtube.com/*"
+
+    Flickr = OEmbed::Provider.new("http://www.flickr.com/services/oembed/")
+    Flickr << "http://*.flickr.com/*"
+
+    Viddler = OEmbed::Provider.new("http://lab.viddler.com/services/oembed/")
+    Viddler << "http://*.viddler.com/*"
+
+    Qik = OEmbed::Provider.new("http://qik.com/api/oembed.{format}")
+    Qik << "http://qik.com/*"
+    Qik << "http://qik.com/video/*"
+
+    Revision3 = OEmbed::Provider.new("http://revision3.com/api/oembed/")
+    Revision3 << "http://*.revision3.com/*"
+
+    Hulu = OEmbed::Provider.new("http://www.hulu.com/api/oembed.{format}")
+    Hulu << "http://www.hulu.com/watch/*"
+
+    Vimeo = OEmbed::Provider.new("http://www.vimeo.com/api/oembed.{format}")
+    Vimeo << "http://*.vimeo.com/*"
+    Vimeo << "http://*.vimeo.com/groups/*/videos/*"
+
+    Pownce = OEmbed::Provider.new("http://api.pownce.com/2.1/oembed.{format}")
+    Pownce << "http://*.pownce.com/*"
+
+    # A general end point, which then calls other APIs and returns OEmbed info
+    OohEmbed = OEmbed::Provider.new("http://oohembed.com/oohembed/")
+    #OohEmbed << %r{http://yfrog.(com|ru|com.tr|it|fr|co.il|co.uk|com.pl|pl|eu|us)/(.*?)} # image & video hosting
+    #OohEmbed << "http://*.xkcd.com/*" # A hilarious stick figure comic
+    OohEmbed << "http://*.wordpress.com/*/*/*/*" # Blogging Engine & community
+    OohEmbed << "http://*.wikipedia.org/wiki/*" # Online encyclopedia
+    #OohEmbed << "http://*.twitpic.com/*" # Picture hosting for Twitter
+    #OohEmbed << "http://twitter.com/*/statuses/*" # Mirco-blogging network
+    #OohEmbed << "http://*.slideshare.net/*" # Share presentations online
+    #OohEmbed << "http://*.phodroid.com/*/*/*" # Photo host
+    #OohEmbed << "http://*.metacafe.com/watch/*" # Video host
+    #OohEmbed << "http://video.google.com/videoplay?*" # Video hosting
+    #OohEmbed << "http://*.funnyordie.com/videos/*" # Comedy video host
+    #OohEmbed << "http://*.thedailyshow.com/video/*" # Syndicated show
+    #OohEmbed << "http://*.collegehumor.com/video:*" # Comedic & original videos
+    #OohEmbed << %r{http://(.*?).amazon.(com|co.uk|de|ca|jp)/(.*?)/(gp/product|o/ASIN|obidos/ASIN|dp)/(.*?)} # Online product shopping
+    #OohEmbed << "http://*.5min.com/Video/*" # micro-video host
+
+    # A general end point, which then calls other APIs and returns OEmbed info
+    Embedly = OEmbed::Provider.new("http://api.embed.ly/v1/api/oembed")
+    urls = JSON.parse(File.open(File.dirname(__FILE__) + "/embedly_urls.json", "r").read)
+    urls.each do |url|
+        Embedly << url
+    end
+    
+    PollEverywhere = OEmbed::Provider.new("http://www.polleverywhere.com/services/oembed/")
+    PollEverywhere << "http://www.polleverywhere.com/polls/*"
+    PollEverywhere << "http://www.polleverywhere.com/multiple_choice_polls/*"
+    PollEverywhere << "http://www.polleverywhere.com/free_text_polls/*"
+
+    MyOpera = OEmbed::Provider.new("http://my.opera.com/service/oembed", :json)
+    MyOpera << "http://my.opera.com/*"
+
+    ClearspringWidgets = OEmbed::Provider.new("http://widgets.clearspring.com/widget/v1/oembed/")
+    ClearspringWidgets << "http://www.clearspring.com/widgets/*"
+
+    NFBCanada = OEmbed::Provider.new("http://www.nfb.ca/remote/services/oembed/")
+    NFBCanada << "http://*.nfb.ca/film/*"
+
+    Scribd = OEmbed::Provider.new("http://www.scribd.com/services/oembed")
+    Scribd << "http://*.scribd.com/*"
+
+    MovieClips = OEmbed::Provider.new("http://movieclips.com/services/oembed/")
+    MovieClips << "http://movieclips.com/watch/*/*/"
+
+    TwentyThree = OEmbed::Provider.new("http://www.23hq.com/23/oembed")
+    TwentyThree << "http://www.23hq.com/*"
+
+  end
+end