govkit-h 0.7.1.0

data/lib/generators/govkit/templates/mention.rb

@@ -0,0 +1,15 @@
+# A model to contain mentions of the :owner in the media
+class Mention < ActiveRecord::Base
+  belongs_to :owner, :polymorphic => true
+
+  scope :since, lambda { |d| where(["mentions.date > ?", d]) }
+
+  # Returns the mentions in JSON form
+  # 
+  # @params [Hash] A hash of options
+  # @return The mentions in JSON form
+  def as_json(opts = {})
+    default_opts = {:except => [:owner_id, :owner_type]}
+    super(default_opts.merge(opts))
+  end
+end

data/lib/gov_kit.rb

@@ -0,0 +1,45 @@
+$LOAD_PATH.unshift(File.expand_path(File.dirname(__FILE__))) unless $LOAD_PATH.include?(File.expand_path(File.dirname(__FILE__)))
+
+require 'digest/md5'
+require 'active_support'
+require 'nokogiri'
+require 'iconv'
+require 'httparty'
+require 'open-uri'
+require 'json'
+require 'gov_kit/configuration'
+require 'csv'
+
+if RUBY_VERSION[0,3] == "1.8"
+  require 'fastercsv'
+end
+
+module GovKit
+  autoload :Resource, 'gov_kit/resource'
+  autoload :OpenStates, 'gov_kit/open_states'
+  autoload :TransparencyData, 'gov_kit/transparency_data'
+  autoload :VoteSmart, 'gov_kit/vote_smart'
+  autoload :ActsAsNoteworthy, 'gov_kit/acts_as_noteworthy'
+  autoload :FollowTheMoney, 'gov_kit/follow_the_money'
+  autoload :OpenCongress, 'gov_kit/open_congress'
+  autoload :SearchEngines, 'gov_kit/search_engines'
+  
+  # Convenience class to represent a news story or blog post.
+  # Used by GovKit::SearchEngines classes.
+  class Mention
+    attr_accessor :url, :excerpt, :title, :source, :date, :weight, :search_source
+  end
+
+  class GovKitError < StandardError
+  end
+
+  class NotAuthorized < GovKitError; end
+
+  class InvalidRequest < GovKitError; end
+
+  class ResourceNotFound < GovKitError; end
+  
+  class ServerError < GovKitError; end
+  
+  class ClientError < GovKitError; end
+end

data/lib/gov_kit/acts_as_noteworthy.rb

@@ -0,0 +1,63 @@
+module GovKit::ActsAsNoteworthy
+
+  def self.included(base)
+    base.extend ActMethods
+  end
+
+  # Module to make a rails model act as a noteworthy object, linking it to govkit's mention model
+  module ActMethods
+    # Sets up the relationship between the model and the mention model
+    # 
+    # @param [Hash] opts a hash of options to be used by the relationship
+    def acts_as_noteworthy(options={})
+      class_inheritable_accessor :options
+      self.options = options
+
+      unless included_modules.include? InstanceMethods
+        instance_eval do
+          has_many :mentions, :as => :owner, :order => 'date desc'
+
+          with_options :as => :owner, :class_name => "Mention" do |c|
+            c.has_many :google_news_mentions, :conditions => {:search_source => "Google News"}, :order => 'date desc'
+            c.has_many :google_blog_mentions, :conditions => {:search_source => "Google Blogs"}, :order => 'date desc'
+#            c.has_many :technorati_mentions, :conditions => {:search_source => "Technorati"}, :order => 'date desc'
+            c.has_many :bing_mentions, :conditions => {:search_source => "Bing"}, :order => 'date desc'
+          end
+        end
+
+        extend ClassMethods
+        include InstanceMethods
+      end
+    end
+  end
+
+  module ClassMethods
+  end
+  
+  # A module that adds methods to individual instances of the noteworthy class.
+  module InstanceMethods
+    # Generates the raw mentions to be loaded into the Mention objects
+    # 
+    # @return [Hash] a hash of all the mentions found of the object in question.
+    def raw_mentions
+      opts = self.options.clone
+      attributes = opts.delete(:with)
+
+      if opts[:geo]
+        opts[:geo] = self.instance_eval("#{opts[:geo]}")
+      end
+
+      query = []
+      attributes.each do |attr|
+        query << self.instance_eval("#{attr}")
+      end
+
+      {
+        :google_news => GovKit::SearchEngines::GoogleNews.search(query, opts),
+        :google_blogs => GovKit::SearchEngines::GoogleBlog.search(query, opts),
+#        :technorati => GovKit::SearchEngines::Technorati.search(query),
+        :bing => GovKit::SearchEngines::Bing.search(query, opts)
+      }
+    end
+  end
+end

data/lib/gov_kit/configuration.rb

@@ -0,0 +1,58 @@
+module GovKit
+  class Configuration
+    attr_accessor :sunlight_apikey, :openstates_base_url, :transparency_data_base_url, :transparency_data_categories_url
+    attr_accessor :votesmart_apikey, :votesmart_base_url
+    attr_accessor :openstates_apikey, :ftm_apikey, :ftm_base_url
+    attr_accessor :opencongress_apikey, :opencongress_base_url
+    attr_accessor :technorati_apikey, :technorati_base_url
+    attr_accessor :google_blog_base_url, :google_news_base_url
+    attr_accessor :wikipedia_base_url
+    attr_accessor :bing_appid, :bing_base_url
+
+    def initialize
+      @sunlight_apikey = @openstates_apikey = @votesmart_apikey = @ftm_apikey = ''
+      @openstates_base_url = 'openstates.sunlightlabs.com/api/v1/'
+      @transparency_data_base_url = 'transparencydata.com/api/1.0/'
+      @votesmart_base_url = 'api.votesmart.org/'
+      @ftm_base_url = 'api.followthemoney.org/'
+      @opencongress_base_url = 'www.opencongress.org/'
+      @technorati_base_url = 'api.technorati.com'
+      @bing_base_url = 'api.search.live.net'
+      @google_blog_base_url = 'blogsearch.google.com'
+      @google_news_base_url = 'news.google.com'
+      @wikipedia_base_url = 'en.wikipedia.org'
+
+      # Permant home for contribution category mappings
+      @transparency_data_categories_url = 'http://assets.transparencydata.org.s3.amazonaws.com/docs/catcodes.csv'
+    end
+    
+    def opencongress_apikey= key
+      warn "[DEPRECATION] OpenCongress no longer requires an API Key. Ability to set it will be removed in future versions"
+      @opencongress_apikey = key
+    end
+
+    def openstates_apikey= key
+      warn "[DEPRECATION] Use sunlight_apikey instead of openstates_apikey. Ability to set it will be removed in future versions"
+      @sunlight_apikey = key
+    end
+  end
+
+  class << self
+    attr_accessor :configuration
+    
+    def configuration
+      @configuration = Configuration.new if @configuration.nil?
+      @configuration
+    end
+  end
+
+  # Configure GovKit in config/initializers/govkit.rb
+  #
+  # @example
+  #   GovKit.configure do |config|
+  #     config.sunlight_apikey = ''
+  #   end
+  def self.configure
+    yield(configuration)
+  end
+end

data/lib/gov_kit/follow_the_money.rb

@@ -0,0 +1,176 @@
+module GovKit
+
+  # Subclass of {Resource} for FollowTheMoney data. This 
+  # is subclassed further for each of the different types of record
+  # returned by FollowTheMoney.
+  #
+  # For the details on the FollowTheMoney queries, see {http://www.followthemoney.org/services/methods.phtml the FollowTheMoney API documentation}.
+  class FollowTheMoneyResource < Resource
+    default_params :key => GovKit::configuration.ftm_apikey
+    base_uri GovKit::configuration.ftm_base_url
+
+    # Common method used by subclasses to get data from the service.
+    #
+    # @return [Nokogiri::XML::Document] a {http://nokogiri.org/Nokogiri/HTML/Document.html Nokogiri XML::Document} object
+    # @param [String] path query path that specifies the required data
+    # @param [Hash]   options query options
+    #
+    # @example
+    #   doc = get_xml("/base_level.industries.list.php", :query => {:page => page_num})
+    #
+    def self.get_xml(path, options)
+      doc = Nokogiri::XML(get(path, options))
+
+      e = doc.search("//error")
+
+      # Deal with whatever error comes back
+      if e.size > 0
+        raise case e.first.attributes['code'].value
+        when "100"
+          GovKit::NotAuthorized
+        when "300"
+          GovKit::InvalidRequest
+        when "200"
+          GovKit::ResourceNotFound
+        else
+          GovKit::InvalidRequest
+        end, e.first.attributes['text'].value
+      end
+
+      doc
+    end
+    
+  # Convert the hash array returned by Nokogiri, which has Nokogiri::XML::Attr objects as values,
+  # to an array of hashes with string values.
+  #
+  # @param [Array] result array of hashes, with object values.
+  # @return [Array] array of hashes, with string values.
+    def self.stringify_values_of(result)
+      result.collect! { |r| r.inject({}) {|h, (k, v)| h[k] = v.to_s; h } }
+    end
+  end
+
+  # Provides classes to wrap {http://www.followthemoney.org/index.phtml FollowTheMoney} data. 
+  #
+  # For the details on the FollowTheMoney queries, see {http://www.followthemoney.org/services/methods.phtml the FollowTheMoney API documentation}.
+  module FollowTheMoney
+
+    # Wrap {http://www.followthemoney.org/services/method_doc.phtml?a=11 Industry data}
+    class Business < FollowTheMoneyResource
+
+      # Return a list of all business, industry, and sector categories. See the {http://www.followthemoney.org/services/method_doc.phtml?a=11 FollowTheMoney API}.
+      #
+      # @return [Business] A list of Business objects.
+      def self.list
+        next_page, result, page_num = "yes", [], 0
+
+        until next_page != "yes"
+          # puts "Getting batch number #{page_num}"
+
+          doc = get_xml("/base_level.industries.list.php", :query => {:page => page_num})
+
+          next_page = doc.children.first.attributes['next_page'].value
+
+          page_num += 1
+
+          result += doc.search('//business_detail').collect { |x| x.attributes }
+        end
+
+        stringify_values_of(result)
+        parse(result)
+      end
+    end
+
+    # Wrap contributions to a candidate. See the {http://www.followthemoney.org/services/method_doc.phtml?a=32 FollowTheMoney API}.
+    class Contribution < FollowTheMoneyResource
+
+      # Return contributions to a candidate. See the {http://www.followthemoney.org/services/method_doc.phtml?a=32 FollowTheMoney API}.
+      #
+      # @param [Integer] nimsp_id the candidate id.
+      #
+      # @return [Contribution] a Contribution object, or array of Contribution objects, representing the contributions.
+      def self.find(nimsp_id)
+        next_page, result, page_num = "yes", [], 0
+
+        until next_page != "yes"
+          doc = get_xml("/candidates.contributions.php", :query => {"imsp_candidate_id" => nimsp_id, :page => page_num})
+
+          next_page = doc.children.first.attributes['next_page'].value
+
+          page_num += 1
+
+          result += doc.search('//contribution').collect { |x| x.attributes }
+        end
+
+        stringify_values_of(result)
+        parse(result)
+      end
+
+      # Return a list of the top contributors to a candidate. See the {http://www.followthemoney.org/services/method_doc.phtml?a=20 FollowTheMoney API}.
+      #
+      # @param [Integer] nimsp_id the candidate id.
+      #
+      # @return [[Contribution]] an array of Contribution objects.
+      def self.top(nimsp_id)
+        doc = get_xml("/candidates.top_contributor.php", :query => {"imsp_candidate_id" => nimsp_id})
+        result = doc.search('//top_contributor').collect { |x| x.attributes }
+
+        stringify_values_of(result)
+        parse(result)
+      end
+    end
+
+    # Wrap contributions by industry to a candidate. See the {http://www.followthemoney.org/services/method_doc.phtml?a=24 FollowTheMoney API}.
+    #
+    class IndustryContribution < Contribution
+      # Return contributions by industry.
+      #
+      # @param [Integer] nimsp_id the candidate id.
+      #
+      # @return [[Contribution]] an array of Contribution objects.
+      def self.find(nimsp_id)
+        doc = get_xml("/candidates.industries.php", :query => {"imsp_candidate_id" => nimsp_id})
+        result = doc.search('//candidate_industry').collect { |x| x.attributes }
+
+        stringify_values_of(result)
+        parse(result)
+      end
+    end
+
+    # Wrap contributions by sector to a candidate. See the {http://www.followthemoney.org/services/method_doc.phtml?a=23 FollowTheMoney API}.
+    #
+    class SectorContribution < Contribution
+      # Return conributions by sector.
+      #
+      # @param [Integer] nimsp_id the candidate id.
+      #
+      # @return [[Contribution]] an array of Contribution objects.
+      def self.find(nimsp_id)
+        doc = get_xml("/candidates.sectors.php", :query => {"imsp_candidate_id" => nimsp_id})
+
+        result = doc.search('//candidate_sector').collect { |x| x.attributes }
+
+        stringify_values_of(result)
+        parse(result)
+      end
+    end
+
+    # Wrap contributions by business to a candidate. See the {http://www.followthemoney.org/services/method_doc.phtml?a=25 FollowTheMoney API}.
+    #
+    class BusinessContribution < Contribution
+      # Return contributions by business.
+      #
+      # @param [Integer] nimsp_id the candidate id.
+      #
+      # @return [[Contribution]] an array of Contribution objects.
+      def self.find(nimsp_id)
+        doc = get_xml("/candidates.businesses.php", :query => {"imsp_candidate_id" => nimsp_id})
+
+        result = doc.search('//candidate_business').collect { |x| x.attributes }
+
+        stringify_values_of(result)
+        parse(result)
+      end
+    end
+  end
+end

data/lib/gov_kit/open_congress.rb

@@ -0,0 +1,125 @@
+require 'nokogiri'
+require 'open-uri'
+require 'json'
+require 'CGI'
+
+module GovKit::OpenCongress
+  autoload :Bill,               'gov_kit/open_congress/bill'
+  autoload :BlogPost,           'gov_kit/open_congress/blog_post'
+  autoload :NewsPost,           'gov_kit/open_congress/news_post'
+  autoload :VotingComparison,   'gov_kit/open_congress/voting_comparison'
+  autoload :RollCallComparison, 'gov_kit/open_congress/roll_call_comparison'
+  autoload :Person,             'gov_kit/open_congress/person'
+  autoload :PersonStat,         'gov_kit/open_congress/person_stat'
+
+  # Parent class for classes that wrap {http://www.opencongress.org/api OpenCongress data}.
+  #
+  # Unlike the wrapper classes for data from {FollowTheMoneyResource FollowTheMoney}, 
+  # {OpenStatesResource OpenStates}, {TransparencyDataResource TransparencyData}, 
+  # and {VoteSmartResource VoteSmart}, OpenCongressObject does not inherit 
+  # from {GovKit::Resource}
+  #
+  class OpenCongressObject
+    
+    def initialize(obj, params)
+      params.each do |key, value|
+        key = key.to_sym if RUBY_VERSION[0,3] == "1.9"
+        instance_variable_set("@#{key}", value) if obj.instance_methods.include? key
+      end      
+    end
+
+    # Create a query url, by adding the method path and query parameters to the
+    # base url of {http://www.opencongress.org/api}.
+    def self.construct_url(api_method, params)
+      url = nil
+      getkey = GovKit::configuration.opencongress_apikey.nil? ? "" : "&key=#{GovKit::configuration.opencongress_apikey}"
+      url = "http://#{GovKit::configuration.opencongress_base_url}api/#{api_method}?format=json#{hash2get(params)}#{getkey}"
+      return url
+    end
+
+    # Convert a hash to a string of query parameters.
+    #
+    # @param [Hash] h a hash.
+    # @return [String] a string of query parameters.
+    def self.hash2get(h)
+      get_string = ""
+    
+      h.each_pair do |key, value|
+        get_string += "&#{key.to_s}=#{CGI::escape(value.to_s)}"
+      end
+
+      get_string
+    end
+
+    # Iterates through the array returned by {make_call}, 
+    # converting each hash to an OpenCongressObject subclass.
+    #
+    # @param [Hash] results the array returned by make_call.
+    # @return a hash of arrays of OpenCongressObject objects, with these keys: 
+    #   * :also_supporting_bills 
+    #   * :also_opposing_bills
+    #   * :also_disapproved_senators
+    #   * :also_disapproved_representatives
+    #   * :also_approved_senators
+    #   * :also_approved_representatives
+    def self.parse_supporting_results(result)
+      working = result["opencongress_users_tracking"]
+
+      also_supporting_bills = []
+      working["also_supporting_bills"]["bill"].each do |bill|
+        also_supporting_bills << Bill.new(bill)
+      end
+
+      also_opposing_bills = []
+      working["also_opposing_bills"]["bill"].each do |bill|
+        also_opposing_bills << Bill.new(bill)
+      end
+
+      also_disapproved_senators = []
+      working["also_disapproved_senators"]["person"].each do |person|
+        also_disapproved_senators << Person.new(person)
+      end
+
+      also_disapproved_representatives = []
+      working["also_disapproved_representatives"]["person"].each do |person|
+        also_disapproved_representatives << Person.new(person)
+      end
+
+      also_approved_senators = []
+      working["also_approved_senators"]["person"].each do |person|
+        also_approved_senators << Person.new(person)
+      end
+
+      also_approved_representatives = []
+      working["also_approved_representatives"]["person"].each do |person|
+        also_approved_representatives << Person.new(person)
+      end
+
+      return {:also_supporting_bills => also_supporting_bills,
+              :also_opposing_bills => also_opposing_bills,
+              :also_disapproved_senators => also_disapproved_senators,
+              :also_disapproved_representatives => also_disapproved_representatives,
+              :also_approved_senators => also_approved_senators,
+              :also_approved_representatives => also_approved_representatives}
+    
+    end
+  
+    # Get the data from {http://www.opencongress.org/api OpenCongress data}. Called by subclasses.
+    #
+    # Parses the data using {http://flori.github.com/json/doc/index.html JSON.parse}, which
+    # returns an array of hashes.
+    #
+    # @return the returned data, as an array of hashes.
+    def self.make_call(this_url)
+      result = nil
+      begin
+        result = JSON.parse(open(this_url).read)
+      rescue => e
+        puts e
+      end
+    
+      return result
+    
+    end
+  end
+end