textrazor 1.0.1 → 1.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: a5dd726ebc5946e424815cfad69070296b759d2e
-  data.tar.gz: fda4507e8aa64b77073789d97526fa661aa63c13
+  metadata.gz: 97bc6dcda89ce7ef88868906d97b92b0e46a9ef8
+  data.tar.gz: f4e7d689ddeb6b41130d0fabb0e020dff8d9eb24
 SHA512:
-  metadata.gz: aa872f1ddd89bb8b2273997a068fb026f90504cb610f6a319b32207964041d2b59052648fc77f6538f1263d08c9e07806686e80de30416cc2c099ca01bc97962
-  data.tar.gz: d3e63881541061cc0a3042e683d2f14cf7780f07563d852737e644ee3a4d99bb515d962c7492f6786df634ffa778e09800f96731327e6b6eeebd0f1ab5cf2671
+  metadata.gz: 506f8a61cda18fc018267cee10323dfc2a9bef7a52edacac46924e0cfcc2689e1348f834e67879dff892b8fb36b9e7ea6aa8b638ee9727579ccaafdf8e71b850
+  data.tar.gz: 013233c7cd51d4467aeaf5e5a2e5f3960a4acdc1e17724a1b0a2092cab28f732ec59605033cdd890d0481011d9191742f4d89e57788876444c1a3a0fc105e27d

data/lib/textrazor.rb

@@ -4,6 +4,7 @@ require "textrazor/util"
 require "textrazor/client"
 require "textrazor/request"
 require "textrazor/response"
+require "textrazor/category"
 require "textrazor/topic"
 require "textrazor/entailment"
 require "textrazor/entity"

data/lib/textrazor/category.rb

@@ -0,0 +1,15 @@
+module TextRazor
+
+  class Category
+
+    include Util
+
+    attr_reader :id, :category_id, :label, :score, :classifier_id
+
+    def initialize(params = {})
+      initialize_params params
+    end
+
+  end
+
+end

data/lib/textrazor/client.rb

@@ -17,7 +17,7 @@ module TextRazor
 
     REQUEST_OPTIONS = [:extractors, :rules, :cleanup_mode, :cleanup_return_cleaned, :cleanup_return_raw,
                        :language, :dictionaries, :filter_dbpedia_types, :filter_freebase_types, :allow_overlap,
-                       :enrichment_queries]
+                       :enrichment_queries, :classifiers]
 
     attr_reader :response, :api_key, :request_options
 
@@ -43,6 +43,12 @@ module TextRazor
         topics
     end
 
+    def self.categories(api_key, text, options = {})
+      new(api_key, options.merge(classifiers: ['textrazor_iab'])).
+        analyse(text).
+        categories
+    end
+
     def self.coarse_topics(api_key, text, options = {})
       new(api_key, options.merge(extractors: ['topics'])).
         analyse(text).

data/lib/textrazor/entailment.rb

@@ -2,21 +2,14 @@ module TextRazor
 
   class Entailment
 
-    extend Util
+    include Util
 
     attr_reader :id, :word_positions, :prior_score, :context_score,
                 :score, :entailed_tree, :entailed_words
 
     def initialize(params = {})
       @type = []
-      params.each do |k, v|
-        instance_variable_set(:"@#{k}", v) if v && self.respond_to?(:"#{k}")
-      end
-    end
-
-    def self.create_from_hash(params)
-      params = Hash[params.map {|k, v| [standardize(k), v] }]
-      new(params)
+      initialize_params params
     end
 
   end

data/lib/textrazor/entity.rb

@@ -2,22 +2,15 @@ module TextRazor
 
   class Entity
 
-    extend Util
+    include Util
 
     attr_reader :id, :type, :matching_tokens, :entity_id, :freebase_types, :confidence_score,
       :wiki_link, :matched_text, :freebase_id, :relevance_score, :entity_english_id,
-      :starting_pos, :ending_pos, :data
+      :starting_pos, :ending_pos, :data, :wikidata_id
 
     def initialize(params = {})
       @type = []
-      params.each do |k, v|
-        instance_variable_set(:"@#{k}", v) if v && self.respond_to?(:"#{k}")
-      end
-    end
-
-    def self.create_from_hash(params)
-      params = Hash[params.map {|k, v| [standardize(k), v] }]
-      new(params)
+      initialize_params params
     end
 
   end

data/lib/textrazor/property.rb

@@ -2,19 +2,12 @@ module TextRazor
 
   class Property
 
-    extend Util
+    include Util
 
     attr_reader :id, :word_positions, :property_positions
 
     def initialize(params = {})
-      params.each do |k, v|
-        instance_variable_set(:"@#{k}", v) if v && self.respond_to?(:"#{k}")
-      end
-    end
-
-    def self.create_from_hash(params)
-      params = Hash[params.map {|k, v| [standardize(k), v] }]
-      new(params)
+      initialize_params params
     end
 
   end

data/lib/textrazor/request.rb

@@ -17,7 +17,8 @@ module TextRazor
       filter_dbpedia_types: 'entities.filterDbpediaTypes',
       filter_freebase_types: 'entities.filterFreebaseTypes',
       allow_overlap: 'entities.allowOverlap',
-      enrichment_queries: 'entities.enrichmentQueries'
+      enrichment_queries: 'entities.enrichmentQueries',
+      classifiers: 'classifiers'
     }
 
     def self.post(text, options)

data/lib/textrazor/response.rb

@@ -84,6 +84,10 @@ module TextRazor
        @sentences ||= parse_sentences
     end
 
+    def categories
+      @categories ||= parse_categories
+    end
+
     def language
       raw_response[:language]
     end
@@ -122,6 +126,10 @@ module TextRazor
       parse(:topic, raw_response[:topics])
     end
 
+    def parse_categories
+      parse(:category, raw_response[:categories])
+    end
+
     def parse_phrases
       raw_phrases = raw_response[:nounPhrases]
       return if raw_phrases.nil?

data/lib/textrazor/topic.rb

@@ -2,19 +2,12 @@ module TextRazor
 
   class Topic
 
-    extend Util
+    include Util
 
-    attr_reader :id, :label, :wiki_link, :score
+    attr_reader :id, :label, :wiki_link, :score, :wikidata_id
 
     def initialize(params = {})
-      params.each do |k, v|
-        instance_variable_set(:"@#{k}", v) if v && self.respond_to?(:"#{k}")
-      end
-    end
-
-    def self.create_from_hash(params)
-      params = Hash[params.map {|k, v| [standardize(k), v] }]
-      new(params)
+      initialize_params params
     end
 
   end

data/lib/textrazor/util.rb

@@ -1,11 +1,30 @@
 module TextRazor
   module Util
-    def standardize(param)
-      param.to_s.gsub(/::/, '/').
-      gsub(/([A-Z]+)([A-Z][a-z])/,'\1_\2').
-      gsub(/([a-z\d])([A-Z])/,'\1_\2').
-      tr("-", "_").
-      downcase
+    def self.included(base)
+      base.extend ClassMethods
+
+      base.class_eval do
+        def initialize_params(params)
+          params.each do |k, v|
+            instance_variable_set(:"@#{k}", v) if self.respond_to?(:"#{k}") && v && (!v.is_a?(String) || !v.empty?)
+          end
+        end
+      end
+    end
+
+    module ClassMethods
+      def standardize(param)
+        param.to_s.gsub(/::/, '/').
+        gsub(/([A-Z]+)([A-Z][a-z])/,'\1_\2').
+        gsub(/([a-z\d])([A-Z])/,'\1_\2').
+        tr("-", "_").
+        downcase
+      end
+
+      def create_from_hash(params)
+        params = Hash[params.map {|k, v| [standardize(k), v] }]
+        new(params)
+      end
     end
   end
 end

data/lib/textrazor/version.rb

@@ -1,3 +1,3 @@
 module TextRazor
-  VERSION = "1.0.1"
+  VERSION = "1.1"
 end

data/lib/textrazor/word.rb

@@ -1,22 +1,14 @@
 module TextRazor
 
   class Word
-
-    extend Util
+    include Util
 
     attr_reader :position, :starting_pos, :ending_pos, :stem, :lemma,
                 :token, :part_of_speech, :parent_position
 
     def initialize(params = {})
       @type = []
-      params.each do |k, v|
-        instance_variable_set(:"@#{k}", v) if v && self.respond_to?(:"#{k}")
-      end
-    end
-
-    def self.create_from_hash(params)
-      params = Hash[params.map {|k, v| [standardize(k), v] }]
-      new(params)
+      initialize_params params
     end
 
   end

data/spec/lib/textrazor/category_spec.rb

@@ -0,0 +1,34 @@
+require "spec_helper"
+
+module TextRazor
+
+  describe Category do
+
+    context "#create_from_hash" do
+      let(:category_hash) do
+        {
+          "id":0,
+          "classifierId":"textrazor_iab",
+          "categoryId":"IAB11",
+          "label":"Law, Gov’t & Politics",
+          "score":0.809611
+        }
+      end
+
+      let(:category) do
+        Category.create_from_hash(category_hash)
+      end
+
+      it "should create a new instance" do
+        expect(category.id).to eq(0)
+        expect(category.classifier_id).to eq("textrazor_iab")
+        expect(category.category_id).to eq("IAB11")
+        expect(category.label).to eq("Law, Gov’t & Politics")
+        expect(category.score).to eq(0.809611)
+      end
+
+    end
+
+  end
+
+end

data/spec/lib/textrazor/client_spec.rb

@@ -23,6 +23,7 @@ module TextRazor
     let(:custom_options_client) do
       Client.new(api_key, {
         extractors: %w(entities topics words), cleanup_mode: 'raw',
+        classifiers: 'textrazor_newscodes',
         cleanup_return_cleaned: true, cleanup_return_raw: true,
         filter_dbpedia_types: %w(type1), language: 'fre',
         filter_freebase_types: %w(type2), allow_overlap: false,
@@ -57,7 +58,8 @@ module TextRazor
               to eq({extractors: %w(entities topics words), cleanup_mode: 'raw', language: 'fre',
                      cleanup_return_cleaned: true, cleanup_return_raw: true,
                      filter_dbpedia_types: %w(type1), filter_freebase_types: %w(type2),
-                     allow_overlap: false, dictionaries: %w(test)})
+                     allow_overlap: false, dictionaries: %w(test),
+                     classifiers: 'textrazor_newscodes'})
           end
 
         end
@@ -125,7 +127,7 @@ module TextRazor
             with('text', {api_key: 'api_key', extractors: %w(entities topics words), cleanup_mode: 'raw',
                           cleanup_return_cleaned: true, cleanup_return_raw: true, language: 'fre',
                           filter_dbpedia_types: %w(type1), filter_freebase_types: %w(type2),
-                          allow_overlap: false, dictionaries: %w(test)}).
+                          allow_overlap: false, dictionaries: %w(test), classifiers: 'textrazor_newscodes'}).
             and_return(request)
 
           expect(Response).to receive(:new).with(request)
@@ -263,6 +265,25 @@ module TextRazor
 
     end
 
+    context ".categories" do
+
+      it "makes correct calls" do
+        client = OpenStruct.new
+        response = OpenStruct.new categories: ['Category1']
+
+        expect(Client).to receive(:new).
+          with(api_key, {classifiers: ['textrazor_iab']}).
+          and_return(client)
+
+        expect(client).to receive(:analyse).
+          with("text").
+          and_return(response)
+
+        Client.categories(api_key, 'text', {})
+      end
+
+    end
+
   end
 
 end

data/spec/lib/textrazor/entity_spec.rb

@@ -28,7 +28,8 @@ module TextRazor
             "endingPos" => 20,
             "data" => {
               "type" => ['person', 'company']
-            }
+            },
+            "wikidataId" => 'Q7330070'
           }
         end
 
@@ -47,6 +48,7 @@ module TextRazor
           expect(entity.starting_pos).to eq(3)
           expect(entity.ending_pos).to eq(20)
           expect(entity.data['type']).to match_array(['person', 'company'])
+          expect(entity.wikidata_id).to eq('Q7330070')
         end
       end
 

data/spec/lib/textrazor/request_spec.rb

@@ -60,13 +60,13 @@ module TextRazor
           options = {api_key: 'api_key', extractors: %w(entities topics words), cleanup_mode: 'raw',
                      cleanup_return_cleaned: true, cleanup_return_raw: true, language: 'fre',
                      filter_dbpedia_types: %w(type1), filter_freebase_types: %w(type2), allow_overlap: false,
-                     enrichment_queries: 'queries'}
+                     enrichment_queries: 'queries', classifiers: 'textrazor_iab'}
 
           expect(::RestClient).to receive(:post).
             with("https://api.textrazor.com/", { "text" => 'text', "apiKey" => 'api_key', "extractors" => "entities,topics,words",
             "cleanup.mode" => "raw", "cleanup.returnCleaned" => true, "cleanup.returnRaw" => true, "languageOverride" => 'fre',
             "entities.filterDbpediaTypes" => "type1", "entities.filterFreebaseTypes" => "type2" , "entities.allowOverlap" => false,
-            "entities.enrichmentQueries" => "queries"},
+            "entities.enrichmentQueries" => "queries", "classifiers" => 'textrazor_iab'},
              accept_encoding: 'gzip')
 
           Request.post('text', options)

data/spec/lib/textrazor/response_spec.rb

@@ -441,6 +441,73 @@ module TextRazor
 
     end
 
+    describe "#categories" do
+
+      let(:http_response) do
+        ::OpenStruct.new(code: 200, body: body)
+      end
+
+      let(:response) do
+        Response.new(http_response)
+      end
+
+      context "if there are categories returned from api" do
+
+        let(:body) do
+          {
+            "time" => "0.013219",
+            "response" => {
+              "language" => "eng",
+              "languageIsReliable" => true,
+              "categories" => [
+                {
+                  "id" => 0,
+                  "classifierId" => "textrazor_iab",
+                  "categoryId" => "IAB11",
+                  "label" => "Law, Gov’t & Politics",
+                  "score" => 0.809611
+                },
+                {
+                  "id" => 1,
+                  "classifierId" => "textrazor_iab",
+                  "categoryId" => "IAB11-2",
+                  "label" => "Law, Gov’t & Politics>Legal Issues",
+                  "score" => 0.61239
+                }
+              ]
+            }
+          }.to_json
+        end
+
+        it "returns categories" do
+          categories = response.categories
+
+          expect(categories).to_not be_nil
+          expect(categories.size).to eq(2)
+        end
+
+      end
+
+      context "if there are no categories returned from api" do
+
+        let(:body) do
+          {
+            "time" => "0.013219",
+            "response" => {
+              "language" => "eng",
+              "languageIsReliable" => true
+            }
+          }.to_json
+        end
+
+        it "returns nil" do
+          expect(response.categories).to be_nil
+        end
+
+      end
+
+    end
+
     describe "#words" do
 
       let(:http_response) do

data/spec/lib/textrazor/topic_spec.rb

@@ -10,7 +10,8 @@ module TextRazor
           id: 1,
           label: "Sports",
           wikiLink: "link_to_wiki",
-          score: 1.03589
+          score: 1.03589,
+          wikidataId: "Q042"
         }
       end
 
@@ -23,6 +24,7 @@ module TextRazor
         expect(topic.label).to eq("Sports")
         expect(topic.wiki_link).to eq("link_to_wiki")
         expect(topic.score).to eq(1.03589)
+        expect(topic.wikidata_id).to eq("Q042")
       end
 
     end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: textrazor
 version: !ruby/object:Gem::Version
-  version: 1.0.1
+  version: '1.1'
 platform: ruby
 authors:
 - Anuj Dutta
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-02-15 00:00:00.000000000 Z
+date: 2016-05-26 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rest-client
@@ -111,6 +111,7 @@ files:
 - README.md
 - Rakefile
 - lib/textrazor.rb
+- lib/textrazor/category.rb
 - lib/textrazor/client.rb
 - lib/textrazor/configuration.rb
 - lib/textrazor/entailment.rb
@@ -127,6 +128,7 @@ files:
 - lib/textrazor/version.rb
 - lib/textrazor/word.rb
 - spec/functional/service_spec.rb
+- spec/lib/textrazor/category_spec.rb
 - spec/lib/textrazor/client_spec.rb
 - spec/lib/textrazor/configuration_spec.rb
 - spec/lib/textrazor/entailment_spec.rb
@@ -163,12 +165,13 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.2.5
+rubygems_version: 2.4.5.1
 signing_key: 
 specification_version: 4
 summary: An api wrapper for text razor in ruby
 test_files:
 - spec/functional/service_spec.rb
+- spec/lib/textrazor/category_spec.rb
 - spec/lib/textrazor/client_spec.rb
 - spec/lib/textrazor/configuration_spec.rb
 - spec/lib/textrazor/entailment_spec.rb