linkscape 0.2.6

data/lib/linkscape/constants/link-metrics.rb

@@ -0,0 +1,100 @@
+module Linkscape
+  module Constants
+    module LinkMetrics
+
+      RequestBits = {
+        :flags => {
+          :name => 'Flags',
+          :flag => 2,
+          :desc => %Q[A bit field indicating a variety of attributes which apply to this link.]
+        },
+        :text => {
+          :name => 'Anchor Text',
+          :flag => 4,
+          :desc => %Q[The anchor text of the link, including any markup (e.g. image tags with alt text).]
+        },
+        :mozrank => {
+          :name => 'mozRank Passed',
+          :flag => 16,
+          :desc => %Q[The amount of mozRank passed by the link.  Requesting this metric will provide both the pretty 10-point score and the raw score.]
+        }
+      }
+      RequestBits[:all] = {
+        :name => 'All columnts',
+        :flag => RequestBits.keys.inject(0) {|sum,k| sum + RequestBits[k][:flag]},
+        :desc => %Q[Requests all known columns from the API]
+      }
+
+      ResponseFlags = {
+        :no_follow => {
+          :name => 'No Follow',
+          :flag => 1,
+          :desc => %Q[The link in question bore a "rel=nofollow" directive indicating that no juice should flow over the link.]
+        },
+        :same_subdomain => {
+          :name => 'Same Subdomain',
+          :flag => 2,
+          :desc => %Q[The link is between two pages on the same domain.  This is an internal link.]
+        },
+        :meta_refresh => {
+          :name => 'Meta Refresh',
+          :flag => 4,
+          :desc => %Q[The link is actually a meta refresh from the source page to the target.]
+        },
+        :same_ip_address => {
+          :name => 'Same IP Address',
+          :flag => 8,
+          :desc => %Q[The link is between two pages hosted on the same IP address, strongly indicating a potential administrative relationship between the two.]
+        },
+        :same_c_block => {
+          :name => 'Same C-Block',
+          :flag => 16,
+          :desc => %Q[The link is between two pages hosted on the same C Block of IP addresses, indicating a potential administrative relationship between the two.]
+        },
+        :redirect301 => {
+          :name => '301',
+          :flag => 64,
+          :desc => %Q[The link is a 301 redirect.  The source page returned a 301 redirect to our crawler, indicating that the resource was available on the target.]
+        },
+        :redirect302 => {
+          :name => '302',
+          :flag => 128,
+          :desc => %Q[The link is a 302 redirect.  The source page returned a 302 redirect to our crawler, indicating that the resource was temporarily available on the target.]
+        },
+        :no_script => {
+          :name => 'No Script',
+          :flag => 256,
+          :desc => %Q[The link was located within a noscript html block.  This means the link may not have been visible to users using javascript.]
+        },
+        :off_screen => {
+          :name => 'Off Screen',
+          :flag => 512,
+          :desc => %Q[We determined that the link likely appears offscreen.  This means that the link may not have been visible to most users.]
+        },
+        :meta_no_follow => {
+          :name => 'Meta No Follow',
+          :flag => 2048,
+          :desc => %Q[The link appeared on a page using a page level (meta) no follow directive.  This link passes no juice.]
+        },
+        :same_root_domain => {
+          :name => 'Same Root Domain',
+          :flag => 4096,
+          :desc => %Q[The link is between two pages on the same root domain.  The link is not internal, but this strongly indicates an administrative relationship between the two pages.]
+        },
+        :feed_autodiscovery => {
+          :name => 'Feed Autodiscovery',
+          :flag => 16384,
+          :desc => %Q[The link indicates a syndication feed (e.g. rss or atom) for the source page.]
+        },
+        :rel_canonical => {
+          :name => 'Rel Canonical',
+          :flag => 32768,
+          :desc => %Q[The link indicates a canonical form of the page using the rel=canonical directive]
+        }
+      }
+      ResponseFlagMap = {}
+      ResponseFlags.each {|k,v| ResponseFlagMap[v[:flag]] = k }
+
+    end
+  end
+end

data/lib/linkscape/constants/url-metrics.rb

@@ -0,0 +1,183 @@
+module Linkscape
+  module Constants
+    module URLMetrics
+
+      RequestBits = {
+        :title => {
+          :name => 'Title',
+          :flag => 1,
+          :desc => %Q[The title of the page if available. For example: "Request-Response Format"]
+        },
+        :url => {
+          :name => 'URL',
+          :flag => 4,
+          :desc => %Q[The url of the page.  For example: "apiwiki.seomoz.org/Request-Response+Format"]
+        },
+        :fq_domain => {
+          :name => 'Subdomain',
+          :flag => 8,
+          :desc => %Q[The subdomain of the url.  For example: "apiwiki.seomoz.org"]
+        },
+        :pl_domain => {
+          :name => 'Root Domain',
+          :flag => 16,
+          :desc => %Q[The root domain of the url.  For example: "seomoz.org"]
+        },
+        :external_links => {
+          :name => 'External Links',
+          :flag => 32,
+          :desc => %Q[The number of juice-passing external links to the url.]
+        },
+        :fq_domain_external_links => {
+          :name => 'Subdomain External Links',
+          :flag => 64,
+          :desc => %Q[The number of juice-passing external links to the subdomain of the url.]
+        },
+        :pl_domain_external_links => {
+          :name => 'Root Domain External Links',
+          :flag => 128,
+          :desc => %Q[The number of juice-passing external links to the root domain of the url.]
+        },
+        :juice_links => {
+          :name => 'Juice-Passing Links',
+          :flag => 256,
+          :desc => %Q[The number of juice-passing links (internal or external) to the url.]
+        },
+        :fq_domains_linking => {
+          :name => 'Subdomains Linking',
+          :flag => 512,
+          :desc => %Q[The number of subdomains with any pages linking to the url.]
+        },
+        :pl_domains_linking => {
+          :name => 'Root Domains Linking',
+          :flag => 1024,
+          :desc => %Q[The number of root domains with any pages linking to the url.]
+        },
+        :links => {
+          :name => 'Links',
+          :flag => 2048,
+          :desc => %Q[The number of links (juice-passing or not, internal or external) to the url.]
+        },
+        :fq_domain_links => {
+          :name => 'Subdomain Links',
+          :flag => 4294967296,
+          :desc => %Q[The number of links to any page on the subdomain of the url.]
+        },
+        :fq_domain_fq_domains_linking => {
+          :name => 'Subdomain Subdomains Linking',
+          :flag => 4096,
+          :desc => %Q[The number of subdomains with any pages linking to the subdomain of the url.]
+        },
+        :fq_domain_pl_domains_linking => {
+          :name => 'Subdomain Root Domains Linking',
+          :flag => 17179869184,
+          :desc => %Q[The number of domains with any pages linking to the subdomain of the url.]
+        },
+        :pl_domain_links => {
+          :name => 'Root Domain Links',
+          :flag => 8589934592,
+          :desc => %Q[The number of links to any page on the root domain of the url.]
+        },
+        :pl_domain_pl_domains_linking => {
+          :name => 'Root Domain Root Domains Linking',
+          :flag => 8192,
+          :desc => %Q[The number of root domains with any pages linking to the root domain of the url.]
+        },
+        :mozrank => {
+          :name => 'mozRank',
+          :flag => 16384,
+          :desc => %Q[The mozRank of the url.  Requesting this metric will provide both the pretty 10-point score and the raw score.]
+        },
+        :fq_domain_mozrank => {
+          :name => 'Subdomain mozRank',
+          :flag => 32768,
+          :desc => %Q[The mozRank of the subdomain of the url.  Requesting this metric will provide both the pretty 10-point score and the raw score.]
+        },
+        :pl_domain_mozrank => {
+          :name => 'Root Domain mozRank',
+          :flag => 65536,
+          :desc => %Q[The mozRank of the Root Domain of the url.  Requesting this metric will provide both the pretty 10-point score and the raw score.]
+        },
+        :moztrust => {
+          :name => 'mozTrust',
+          :flag => 131072,
+          :desc => %Q[The mozTrust of the url.    Requesting this metric will provide both the pretty 10-point score and the raw score.]
+        },
+        :fq_domain_moztrust => {
+          :name => 'Subdomain mozTrust',
+          :flag => 262144,
+          :desc => %Q[The mozTrust of the subdomain of the url.  Requesting this metric will provide both the pretty 10-point score and the raw score.]
+        },
+        :pl_domain_moztrust => {
+          :name => 'Root Domain mozTrust',
+          :flag => 524288,
+          :desc => %Q[The mozTrust of the root domain of the url.  Requesting this metric will provide both the pretty 10-point score and the raw score.]
+        },
+        :external_mozrank => {
+          :name => 'External mozRank',
+          :flag => 1048576,
+          :desc => %Q[The portion of the url's mozRank coming from external links.  Requesting this metric will provide both the pretty 10-point score and the raw score.]
+        },
+        :fq_domain_external_mozrank_sum_raw => {
+          :name => 'Subdomain External Domain Juice',
+          :flag => 2097152,
+          :desc => %Q[The portion of the mozRank of all pages on the subdomain coming from external links.  Requesting this metric will provide both the pretty 10-point score and the raw score.]
+        },
+        :pl_domain_external_mozrank_sum_raw => {
+          :name => 'Root Domain External Domain Juice',
+          :flag => 4194304,
+          :desc => %Q[The portion of the mozRank of all pages on the root domain coming from external links.  Requesting this metric will provide both the pretty 10-point score and the raw score.]
+          #     source.External mozRank sum of all PL Domain Pages (raw)                       - 9.8334596959365e-11
+        },
+        :fq_domain_mozrank_sum_raw => {
+          :name => 'Subdomain Domain Juice',
+          :flag => 8388608,
+          :desc => %Q[The mozRank of all pages on the subdomain combined.  Requesting this metric will provide both the pretty 10-point score and the raw score.]
+        },
+        :pl_domain_mozrank_sum_raw => {
+          :name => 'Root Domain Domain Juice',
+          :flag => 16777216,
+          :desc => %Q[The mozRank of all pages on the root domain combined.  Requesting this metric will provide both the pretty 10-point score and the raw score.]
+        },
+        :canonical_url => {
+          :name => 'Canonical URL',
+          :flag => 268435456,
+          :desc => %Q[If the url canaonicalizes to a different form, that canonical form will be available in this field]
+        },
+        :status => {
+          :name => 'HTTP Status Code',
+          :flag => 536870912,
+          :desc => %Q[The HTTP status code recorded by Linkscape for this URL (if available)]
+        },
+
+        :page_authority => {
+          :name => 'Page Authority',
+          :flag => 34359738368,
+          :desc => %Q[The page authority of this URL. This will return the pretty 100-point score.]
+        },
+        :domain_authority => {
+          :name => 'Domain Authority',
+          :flag => 68719476736,
+          :desc => %Q[The page authority of all pages on the root domain. This will return the pretty 100-point score.]
+        },
+        :page_authority_raw => {
+          :name => 'Raw Page Authority',
+          :flag => 137438953472,
+          :desc => %Q[The page authority of this URL. This will return the raw score.]
+        },
+        :domain_authority_raw => {
+          :name => 'Raw Domain Authority',
+          :flag => 274877906944,
+          :desc => %Q[The page authority of all pages on the root domain. This will return the raw score.]
+        },
+
+      }
+      RequestBits[:all] = {
+        :name => 'All columnts',
+        :flag => RequestBits.keys.inject(0) {|sum,k| sum + RequestBits[k][:flag]},
+        :desc => %Q[Requests all known columns from the API]
+      }
+
+    end
+  end
+end

data/lib/linkscape/errors.rb

@@ -0,0 +1,4 @@
+class RecursionError < StandardError; end
+class AuthenticationError < StandardError; end
+class InvalidArgument < StandardError; end
+class MissingArgument < StandardError; end

data/lib/linkscape/request.rb

@@ -0,0 +1,80 @@
+module Linkscape
+  class Request
+    require 'net/http'
+    require 'uri'
+    require 'cgi'
+    # require 'base64'
+    # require 'rubygems'
+    # require 'hmac-sha1'
+    
+    attr_accessor :requestURL
+    
+    URL_TEMPLATE = %Q[http://:apiHost:/:apiRoot:/:api:/:url:?AccessID=:accessID:&Expires=:expiration:&Signature=:signature:]
+
+    def self.run(options)
+      self.new(options).run
+    end
+    
+    def initialize(options)
+      
+      case options[:url]
+      when String
+        new_vals = {:url => CGI::escape(options[:url].sub(/^https?:\/\//, '')) }
+      when Array
+        @body = options[:url].collect{ |u| u.sub(/^https?:\/\//, '') }
+        new_vals = {:url => ""}
+      else
+        raise "URL most be a String or an Array"
+      end
+      
+      @requestURL = URL_TEMPLATE.template(signRequest(options.merge(new_vals)))
+      @requestURL += "&" + options[:query].collect{|k,v| "#{CGI::escape(k.to_s)}=#{CGI::escape(v.to_s)}"}.join('&') if options[:query] && Hash === options[:query]
+      @requestURL += "&" + options[:query] if options[:query] && String === options[:query]
+
+      options[:offset] = 0 if options[:offset] && options[:offset] < 0
+      @requestURL += "&Offset=#{options[:offset]}" if options[:offset]
+
+      options[:limit] = 1000 if options[:limit] && options[:limit] > 1000
+      @requestURL += "&Limit=#{options[:limit]}" if options[:limit]
+    end
+    
+    def run
+      res = fetch(URI.parse(@requestURL))
+      # res = fetch(URI.parse('http://martian.at/other/ose.json'))
+      return Response.new(self, res)
+    end
+
+    def inspect
+      #<Linkscape::Request:0x1016228a0 @requestURL="http://lsapi.seomoz.com/linkscape/mozrank/www.martian.at%2F?AccessID=ose&Expires=1258772331&Signature=Hfwssn0ZbWMe9MEf6%2FWoHOGFHzQ%3D">
+      %Q[#<#{self.class}="#{@requestURL}">]
+    end
+
+    private
+    def signRequest options
+      Linkscape::Signer.signParams(options)
+      options
+    end
+
+    def fetch(uri, limit = 10)
+      # You should choose better exception.
+      raise RecursionError, 'HTTP redirect too deep' if limit == 0
+      
+      # Fetch with a POST of thers is a body
+      response = if @body
+        http = Net::HTTP.new(uri.host, uri.port)
+        request = Net::HTTP::Post.new(uri.request_uri)
+        request.body = @body.to_json
+        http.request(request)
+      else
+        Net::HTTP.get_response(uri)
+      end
+      
+      
+
+      return fetch(response['location'], limit - 1) if Net::HTTPSuccess == response
+
+      response
+    end
+
+  end
+end

data/lib/linkscape/response.rb

@@ -0,0 +1,138 @@
+require 'forwardable'
+module Linkscape
+  class Response
+    extend Forwardable
+    
+    class ResponseData
+      include Enumerable
+      extend Forwardable
+      
+      attr_reader :type, :subjects
+
+      class Flags
+        def initialize(bitfield, type)
+          @value = bitfield
+          @flags = Linkscape::Constants::LinkMetrics::ResponseFlags.to_a.collect{|k,vv| k if (@value & vv[:flag]) == vv[:flag]}.compact if type == :link
+          @flags = Linkscape::Constants::AnchorMetrics::ResponseFlags.to_a.collect{|k,vv| k if (@value & vv[:flag]) == vv[:flag]}.compact if type == :anchor
+        end
+        def [](key); @flags.include? key.to_sym; end
+        def to_a; @flags; end
+        def to_hash; @flags.inject({}){|h,f|h[f]=true;h}; end
+        def to_s
+          %Q[#{@value}=#{self.to_a.inspect}]
+        end
+      end
+
+      def initialize(data, type=nil)
+        @data = data
+        @type = if type
+          type.to_sym
+        elsif Hash === @data
+          :hash
+        elsif Array === @data
+          :array
+        end
+        @data.symbolize_keys! if Hash === @data
+        @data = @data.collect{|d|ResponseData.new(d)} if Array === @data
+        
+        if @type == :hash
+          subdatas = {}
+          @data.each do |k,v|
+            if field = Linkscape::Constants::ResponseFields[k]
+              if subject = field[:subject]
+                subdatas[subject] ||= {}
+                v = ResponseData::Flags.new(v, field[:bitfield]) if field[:bitfield]
+                subdatas[subject][field[:key]] = v
+              end
+            end
+          end
+          @subjects = []
+          subdatas.each do |k,v|
+            @data[k] = ResponseData.new(v, k)
+            @subjects.push k
+          end
+        end
+        
+      end
+
+      def_delegators :@data, :length, :each, :each_index, :map, :collect, :select, :keys
+      def [](*args)
+        if Array === @data
+          @data[*args]
+        else
+          k = args.first.to_sym
+          if @subjects && @subjects.length == 1 && @data[@subjects.first][k]
+            @data[@subjects.first][k]
+          else
+            @data[args.first.to_sym]
+          end
+        end
+      rescue
+        nil
+      end
+      
+      def to_s(indent="")
+        printer = Proc.new do |h,prefix|
+          o = ""
+          h.sort{|l,r|l[0].to_s<=>r[0].to_s}.each do |k,v|
+            field = Linkscape::Constants::ResponseFields[k]
+            v = v.to_s
+            # v = ((field && field[:bitfield]) ? v.to_a.inspect : v).to_s
+            #desc = field ? field[:name] : '*'+k.inspect
+            #o += %Q[%s%-#{Linkscape::Constants::LongestNameLength+15}.#{Linkscape::Constants::LongestNameLength+15}s - %s\n] % [prefix, desc, v]
+            o += %Q[%s%-#{Linkscape::Constants::LongestKeyLength+5}.#{Linkscape::Constants::LongestKeyLength+5}s - %s\n] % [prefix, k, v]
+          end
+          o
+        end
+        if type == :array
+          o = ""
+          @data.each_with_index do |d,idx|
+            o += %Q[#{indent}[#{idx}]\n] + d.to_s("#{indent}  ") + "\n"
+          end
+          o
+        elsif @subjects
+          o = ""
+          @subjects.each do |s|
+            o += %Q[#{indent}#{s}\n]
+            o += printer.call(@data[s], "#{indent}  #{s}.")
+          end
+          o
+        else
+          printer.call(@data, indent)
+        end
+      end
+
+      def inspect
+        #<Linkscape::Response:0x10161d8a0 @response=#<Net::HTTPUnauthorized 401 Unauthorized readbody=true>>
+        %Q[#<#{self.class} @type=#{@type.inspect}] + (@subjects ? %Q[ @subjects=#{@subjects.inspect}] : "") + %Q[>]
+      end
+
+    end
+    
+    require 'rubygems'
+    require 'json'
+    
+    attr_accessor :request, :response, :data, :valid
+    
+    def initialize(request, response)
+      @valid = false
+      @request = request
+      @response = response
+      if Net::HTTPSuccess === response
+        @data = ResponseData.new(JSON.parse(response.body))
+        @valid = true
+      end
+    end
+    
+    def_delegators :@data, :[], :length, :each, :each_index, :map, :collect, :select
+    
+    def valid?; valid; end
+    
+    def inspect
+      #<Linkscape::Response:0x10161d8a0 @response=#<Net::HTTPUnauthorized 401 Unauthorized readbody=true>>
+      %Q[#<#{self.class} @response=#{@response.class.inspect} @request="#{@request.requestURL}">]
+    end
+    
+  end
+  
+end