textrazor 0.0.8 → 1.0.0

data/lib/textrazor/sentence.rb

@@ -0,0 +1,24 @@
+module TextRazor
+
+  class Sentence
+
+    attr_reader :position, :words
+
+    def initialize(params)
+      @position = params[:position]
+      @words = params[:words].map do |word_hash|
+        Word.create_from_hash(word_hash)
+      end
+    end
+
+    def number_of_words
+      @words.size
+    end
+
+    def self.create_from_hash(params)
+      new(params)
+    end
+
+  end
+
+end

data/lib/textrazor/topic.rb

@@ -2,17 +2,19 @@ module TextRazor
 
   class Topic
 
+    extend Util
+
     attr_reader :id, :label, :wiki_link, :score
 
-    def initialize(id, label, wiki_link, score)
-      @id = id
-      @label = label
-      @wiki_link = wiki_link
-      @score = score
+    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)
-      new(params["id"], params["label"], params["wikiLink"], params["score"])
+      params = Hash[params.map {|k, v| [standardize(k), v] }]
+      new(params)
     end
 
   end

data/lib/textrazor/util.rb

@@ -1,7 +1,7 @@
 module TextRazor
   module Util
     def standardize(param)
-      param.gsub(/::/, '/').
+      param.to_s.gsub(/::/, '/').
       gsub(/([A-Z]+)([A-Z][a-z])/,'\1_\2').
       gsub(/([a-z\d])([A-Z])/,'\1_\2').
       tr("-", "_").

data/lib/textrazor/version.rb

@@ -1,3 +1,3 @@
 module TextRazor
-  VERSION = "0.0.8"
+  VERSION = "1.0.0"
 end

data/lib/textrazor/word.rb

@@ -4,7 +4,7 @@ module TextRazor
 
     extend Util
 
-    attr_reader :position, :starting_pos, :ending_pos, :stem, :lemma, 
+    attr_reader :position, :starting_pos, :ending_pos, :stem, :lemma,
                 :token, :part_of_speech, :parent_position
 
     def initialize(params = {})

data/spec/functional/service_spec.rb

@@ -0,0 +1,29 @@
+require 'spec_helper'
+
+describe 'Functional spec', functional: true do
+  let(:api_key) do
+    ENV["TEXTRAZOR_API_KEY"] or raise 'Please specify a TEXTRAZOR_API_KEY in your local environment to run this'
+  end
+
+  let(:client) do
+    TextRazor::Client.new(api_key)
+  end
+
+  let(:text) do
+    t = <<TEXT
+Barclays misled shareholders and the public about one of the biggest investments in the bank's history, a BBC Panorama investigation has found.
+The bank announced in 2008 that Manchester City owner Sheikh Mansour had agreed to invest more than £3bn.
+But the BBC found that the money, which helped Barclays avoid a bailout by British taxpayers, actually came from the Abu Dhabi government.
+Barclays said the mistake in its accounts was "a drafting error".
+Unlike RBS and Lloyds TSB, Barclays narrowly avoided having to request a government bailout late in 2008 after it was rescued by £7bn worth of new investment, most of which came from the gulf states of Qatar and Abu Dhabi.
+Half of the cash was supposed to be coming from Sheikh Mansour.
+But Barclays has admitted it was told the investor might change shortly before shareholders voted to approve the deal on 24 November 2008.
+But instead of telling shareholders, the bank remained silent until the change of investor was confirmed a few hours later.
+TEXT
+  end
+
+  it 'returns a response' do
+    expect(client.analyse(text)).to be_ok
+  end
+end
+

data/spec/lib/textrazor/client_spec.rb

@@ -4,55 +4,103 @@ module TextRazor
 
   describe Client do
 
-    let(:api_key) { 'api_key' }
-    let(:client) { custom_options_client }
-    let(:nil_api_key_client) { Client.new(nil) }
-    let(:empty_api_key_client) { Client.new('') }
-    let(:custom_options_client) { Client.new(api_key, {extractors: %w(entities topics words),
-                                    cleanup_html: true, filter_dbpedia_types: %w(type1),
-                                    language: 'fre',
-                                    filter_freebase_types: %w(type2)}) }
-    let(:default_options_client) { Client.new(api_key) }
-
-    context "#initialize" do
-
-      context "valid parameters" do
-
-        it "should assign correct api_key, text and default options" do
-          expect(default_options_client.api_key).to eq(api_key)
-          expect(default_options_client.request_options).
-            to eq({extractors: %w(entities topics words phrases dependency-trees relations entailments senses)})
+    let(:api_key) do
+     'api_key'
+    end
+
+    let(:client) do
+      custom_options_client
+    end
+
+    let(:nil_api_key_client) do
+      Client.new(nil)
+    end
+
+    let(:empty_api_key_client) do
+      Client.new('')
+    end
+
+    let(:custom_options_client) do
+      Client.new(api_key, {
+        extractors: %w(entities topics words), cleanup_mode: 'raw',
+        cleanup_return_cleaned: true, cleanup_return_raw: true,
+        filter_dbpedia_types: %w(type1), language: 'fre',
+        filter_freebase_types: %w(type2), allow_overlap: false
+      })
+    end
+
+    let(:default_options_client) do
+      Client.new(api_key)
+    end
+
+    describe "#initialize" do
+
+      context 'with valid api key' do
+
+        context "and default request options" do
+
+          it 'assigns correctly' do
+            expect(default_options_client.api_key).to eq(api_key)
+            expect(default_options_client.request_options).
+              to eq({extractors: %w(entities topics words phrases dependency-trees
+                     relations entailments senses), cleanup_mode: 'raw'})
+          end
+
         end
 
-        it "should assign correct api_key, text and passed in options" do
-          expect(custom_options_client.api_key).to eq(api_key)
-          expect(custom_options_client.request_options).
-            to eq({extractors: %w(entities topics words), cleanup_html: true, language: 'fre',
-                   filter_dbpedia_types: %w(type1), filter_freebase_types: %w(type2)})
+        context 'and custom request options' do
+
+          it "assigns correctly" do
+            expect(custom_options_client.api_key).to eq(api_key)
+            expect(custom_options_client.request_options).
+              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})
+          end
+
         end
 
       end
 
-      context "invalid parameters" do
+      context "with invalid api key" do
+
+        context "when nil" do
+
+          it "raises an exception" do
+            expect { nil_api_key_client }.
+              to raise_error(Client::EmptyApiKey)
+          end
+
+        end
+
+        context "when empty" do
+
+          it "raises an exception" do
+            expect { empty_api_key_client }.
+              to raise_error(Client::EmptyApiKey)
+          end
+
+        end
 
-        context "api_key" do
+      end
 
-          context "is nil" do
+      context 'with invalid request options' do
 
-            it "should raise an exception" do
-              expect { nil_api_key_client }.
-                to raise_error(Client::EmptyApiKey)
-            end
+        context 'when an invalid extractor value is supplied' do
 
+          it 'raises an exception' do
+            expect { Client.new(api_key, {extractors: ['invalid-extractor', 'topics']}) }.
+              to raise_error(Client::UnsupportedExtractor)
           end
 
-          context "is empty" do
+        end
 
-            it "should raise an exception" do
-              expect { empty_api_key_client }.
-                to raise_error(Client::EmptyApiKey)
-            end
+        context 'when an invalid cleanup_mode value is supplied' do
 
+          it 'raises an exception' do
+            expect { Client.new(api_key, {cleanup_mode: 'invalid-cleanup-mode'}) }.
+              to raise_error(Client::UnsupportedCleanupMode)
           end
 
         end
@@ -63,16 +111,20 @@ module TextRazor
 
     context "#analyse" do
 
-      let(:very_long_text) { "L" * 201 * 1024 }
+      let(:very_long_text) do
+        "L" * 201 * 1024
+      end
 
-      context "valid parameters" do
+      context "valid value of 'text'" do
 
-        it "should make correct calls" do
-          request = Object.new
+        it "makes correct calls" do
+          request = BasicObject.new
 
           expect(Request).to receive(:post).
-            with('text', {api_key: 'api_key', extractors: %w(entities topics words), cleanup_html: true,
-                          language: 'fre', filter_dbpedia_types: %w(type1), filter_freebase_types: %w(type2)}).
+            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}).
             and_return(request)
 
           expect(Response).to receive(:new).with(request)
@@ -82,35 +134,31 @@ module TextRazor
 
       end
 
-      context "invalid parameters" do
+      context "invalid value of 'text'" do
 
-        context "text" do
-
-          context "is nil" do
-
-            it "should raise an exception" do
-              expect { client.analyse(nil) }.
-                to raise_error(Client::EmptyText)
-            end
+        context "when nil" do
 
+          it "raises an exception" do
+            expect { client.analyse(nil) }.
+              to raise_error(Client::EmptyText)
           end
 
-          context "is empty" do
+        end
 
-            it "should raise an exception" do
-              expect { client.analyse('') }.
-                to raise_error(Client::EmptyText)
-            end
+        context "when empty" do
 
+          it "raises an exception" do
+            expect { client.analyse('') }.
+              to raise_error(Client::EmptyText)
           end
 
-          context "size is > 200kb" do
+        end
 
-            it "should raise an exception" do
-              expect { client.analyse(very_long_text) }.
-                to raise_error(Client::TextTooLong)
-            end
+        context "when size is > 200kb" do
 
+          it "raises an exception" do
+            expect { client.analyse(very_long_text) }.
+              to raise_error(Client::TextTooLong)
           end
 
         end
@@ -121,7 +169,7 @@ module TextRazor
 
     context ".topics" do
 
-      it "should make correct calls" do
+      it "makes correct calls" do
         client = OpenStruct.new
         response = OpenStruct.new topics: ['topic1'], coarseTopics: ['topic1']
 
@@ -140,7 +188,7 @@ module TextRazor
 
     context ".coarse_topics" do
 
-      it "should make correct calls" do
+      it "makes correct calls" do
         client = OpenStruct.new
         response = OpenStruct.new topics: ['topic1'], coarseTopics: ['topic1']
 
@@ -159,7 +207,7 @@ module TextRazor
 
     context ".entities" do
 
-      it "should make correct calls" do
+      it "makes correct calls" do
         client = OpenStruct.new
         response = OpenStruct.new entities: ['Entity1']
 
@@ -178,7 +226,7 @@ module TextRazor
 
     context ".words" do
 
-      it "should make correct calls" do
+      it "makes correct calls" do
         client = OpenStruct.new
         response = OpenStruct.new words: ['Word1']
 
@@ -197,7 +245,7 @@ module TextRazor
 
     context ".phrases" do
 
-      it "should make correct calls" do
+      it "makes correct calls" do
         client = OpenStruct.new
         response = OpenStruct.new phrases: ['Phrase1']
 

data/spec/lib/textrazor/entailment_spec.rb

@@ -0,0 +1,36 @@
+require "spec_helper"
+
+module TextRazor
+
+  describe Entailment do
+
+    context "#create_from_hash" do
+
+      let(:entailment_hash) {
+        {
+          :id=>2, :wordPositions=>[1], :entailedWords=>["misrepresentation"],
+          :entailedTree=>{
+            :word=>"misrepresentation", :wordId=>0, :parentRelation=>-1
+          },
+          :priorScore=>0.00132419, :contextScore=>0.0694058, :score=>0.154246
+        }
+      }
+
+      it "creates a new instance" do
+        entailment = Entailment.create_from_hash(entailment_hash)
+
+        expect(entailment.id).to eq(2)
+        expect(entailment.entailed_tree).to eq({
+          :word=>"misrepresentation", :wordId=>0, :parentRelation=>-1
+        })
+        expect(entailment.word_positions).to eq([1])
+        expect(entailment.prior_score).to eq(0.00132419)
+        expect(entailment.context_score).to eq(0.0694058)
+        expect(entailment.score).to eq(0.154246)
+      end
+
+    end
+
+  end
+
+end

data/spec/lib/textrazor/entity_spec.rb

@@ -6,36 +6,60 @@ module TextRazor
 
     context "#create_from_hash" do
 
-      it "should create a new instance" do
-        entity_hash = {"id" => 1, "type" => ['Person'], "matchingTokens" => [1, 2], "entityId" => "Foreign minister",
-                       "freebaseTypes" => ["government/government_office_or_title"],
-                       "confidenceScore" => 0.897858, "wikiLink" => "http://en.wikipedia.org/wiki/Foreign_minister",
-                       "matchedText" => "foreign ministers", "freebaseId" => "/m/01t_55", "relevanceScore" => 0.311479,
-                       "entityEnglishId" => "Foreign minister", "startingPos" => 3, "endingPos" => 20}
-
-        entity = Entity.create_from_hash(entity_hash)
-
-        expect(entity.id).to eq(1)
-        expect(entity.type).to eq(['Person'])
-        expect(entity.matching_tokens).to eq([1,2])
-        expect(entity.entity_id).to eq("Foreign minister")
-        expect(entity.freebase_types).to eq(["government/government_office_or_title"])
-        expect(entity.confidence_score).to eq(0.897858)
-        expect(entity.wiki_link).to eq("http://en.wikipedia.org/wiki/Foreign_minister")
-        expect(entity.matched_text).to eq("foreign ministers")
-        expect(entity.freebase_id).to eq("/m/01t_55")
-        expect(entity.relevance_score).to eq(0.311479)
-        expect(entity.entity_english_id).to eq("Foreign minister")
-        expect(entity.starting_pos).to eq(3)
-        expect(entity.ending_pos).to eq(20)
+      let(:entity) do
+        Entity.create_from_hash(entity_hash)
       end
 
-      it "should use sensible defaults" do
-        entity_hash = {"id" => 1, "startingPos" => 3, "endingPos" => 20}
+      context "with defined values" do
+        let(:entity_hash) do
+          {
+            "id" => 1,
+            "type" => ['Person'],
+            "matchingTokens" => [1, 2],
+            "entityId" => "Foreign minister",
+            "freebaseTypes" => ["government/government_office_or_title"],
+            "confidenceScore" => 0.897858,
+            "wikiLink" => "http://en.wikipedia.org/wiki/Foreign_minister",
+            "matchedText" => "foreign ministers",
+            "freebaseId" => "/m/01t_55",
+            "relevanceScore" => 0.311479,
+            "entityEnglishId" => "Foreign minister",
+            "startingPos" => 3,
+            "endingPos" => 20
+          }
+        end
 
-        entity = Entity.create_from_hash(entity_hash)
+        it "should create a new instance" do
+          expect(entity.id).to eq(1)
+          expect(entity.type).to eq(['Person'])
+          expect(entity.matching_tokens).to eq([1,2])
+          expect(entity.entity_id).to eq("Foreign minister")
+          expect(entity.freebase_types).to eq(["government/government_office_or_title"])
+          expect(entity.confidence_score).to eq(0.897858)
+          expect(entity.wiki_link).to eq("http://en.wikipedia.org/wiki/Foreign_minister")
+          expect(entity.matched_text).to eq("foreign ministers")
+          expect(entity.freebase_id).to eq("/m/01t_55")
+          expect(entity.relevance_score).to eq(0.311479)
+          expect(entity.entity_english_id).to eq("Foreign minister")
+          expect(entity.starting_pos).to eq(3)
+          expect(entity.ending_pos).to eq(20)
+        end
+      end
+
+      context "with empty values" do
+
+        let(:entity_hash) do
+          {
+            "id" => 1,
+            "startingPos" => 3,
+            "endingPos" => 20
+          }
+        end
+
+        it "should use sensible defaults" do
+          expect(entity.type).to eq([])
+        end
 
-        expect(entity.type).to eq([])
       end
 
     end