sentimental 1.1.1 → 1.2.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 63ca1cc3a2e131c766fda7b6ffef711958b77af4
-  data.tar.gz: da25b0c71603372b554e3624e4113d667155d6dc
+  metadata.gz: 3589f3fdd19386fdd901d673bd905f2703102aa7
+  data.tar.gz: 355ec5e97a133254c82839b70db0e2a12fe0462a
 SHA512:
-  metadata.gz: 33f62d43418c9d02206ec0897572a9efb47ae5dac4189c8fa6c07e889608f23c92fa3478990611d2b697d630fba4652dba5d501483f547a6d278ebfde5930ee6
-  data.tar.gz: 3c45a95fafdcf12949dd67ea1c0b075d0628951f07168de5c803935db456d07ba4cbc478acf5715fcd830b67808c04dcceaf97f130f7cc99277afd2b83251aaf
+  metadata.gz: ec5d616c70d55299d58a520c5482e1dc0ff451b6405337d02e6ba0a9c5b77ca7553839a12dbb40c23f65e071e1022fc3b8dfeed0d022e96abde51d7532345935
+  data.tar.gz: d935a09a091dce7796717563bd9251cf4575ae50eae1bbfe7e0ed4a2b09398f4976a09d076bb4a3d445a1aec2e2f7aa01b4316b1079438862a74ec76c206df25

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    sentimental (1.0.4)
+    sentimental (1.2.1)
 
 GEM
   remote: https://rubygems.org/

data/README.md

@@ -71,6 +71,19 @@ scores and tokens, e.g.:
     0.0 Meh
     -1.0 Horrible
 
+## N-grams
+
+You can parse n-grams of words by specifying their max size in the initializer:
+```
+  Sentimental.new(ngrams: 4)
+```
+
+The dictionary must have this format:
+
+    1.0 very happy
+    -2.0 no
+    0.0 meh
+
 ## Installation
 
     gem install sentimental

data/lib/sentimental.rb

@@ -1,8 +1,14 @@
 class Sentimental
-  attr_accessor :threshold, :word_scores, :neutral_regexps
+  attr_accessor :threshold, :word_scores, :neutral_regexps, :ngrams
 
-  def initialize(threshold: 0, word_scores: nil, neutral_regexps: [])
-    @word_scores = Hash.new(0.0) || word_scores
+  def initialize(threshold: 0, word_scores: nil, neutral_regexps: [], ngrams: 1)
+    if ngrams >= 1
+      @ngrams = ngrams.to_i
+    else
+      @ngrams = 1
+    end
+    @word_scores = word_scores || {}
+    @word_scores.default = 0.0
     @threshold = threshold
     @neutral_regexps = neutral_regexps
   end
@@ -10,7 +16,7 @@ class Sentimental
   def score(string)
     return 0 if neutral_regexps.any? {|regexp| string =~ regexp}
 
-    extract_words(string).inject(0) do |score, token| 
+    extract_words_with_n_grams(string).inject(0) do |score, token| 
       score += word_scores[token]
     end
   end
@@ -27,26 +33,47 @@ class Sentimental
     end
   end
 
+  def classify(string)
+    sentiment(string) == :positive
+  end
+
   def load_defaults
     ['sentiwords', 'sentislang'].each do |filename|
-      load_senti_file(File.dirname(__FILE__) + "/../data/#{filename}.txt")
+      load_from(File.dirname(__FILE__) + "/../data/#{filename}.txt")
     end
   end
 
-  def load_senti_file(filename)
+  def load_from(filename)
     File.open(filename) do |file|
       file.each_line do |line|
-        parsed_line = line.chomp.split(/\s+/)
-        sentiscore = parsed_line[0]
-        text = parsed_line[1]
-        word_scores[text] = sentiscore.to_f
+        if parsed_line = (line.chomp.scan(/^([^\s]+)\s+(.+)/).first)
+          sentiscore = parsed_line[0]
+          text = parsed_line[1]
+          word_scores[text] = sentiscore.to_f
+        end
       end
     end
   end
+  
+  alias_method :load_senti_file, :load_from
 
   private
 
   def extract_words(string)
     string.to_s.downcase.scan(/([\w']+|\S{2,})/).flatten
   end
+
+  def extract_words_with_n_grams(string)
+    words = extract_words(string)
+    (1..ngrams).to_a.map do |ngram_size|
+      ngramify(words, ngram_size)
+    end.flatten
+  end
+
+  def ngramify(words, max_size)
+    return [words.join(" ")] if words.size <= max_size
+    tail = words.last(words.size - 1)
+    
+    [words.first(max_size).join(" ")] + ngramify(tail, max_size)
+  end
 end

data/sentimental.gemspec

@@ -1,6 +1,6 @@
 Gem::Specification.new do |spec|
   spec.name         = 'sentimental'
-  spec.version      = '1.1.1'
+  spec.version      = '1.2.1'
   spec.summary      = 'Simple sentiment analysis'
   spec.description  = 'A simple sentiment analysis gem'
   spec.authors      = ['Jeff Emminger', 'Christopher MacLellan', 'Denis Pasin']

data/spec/sentimental_spec.rb

@@ -8,7 +8,93 @@ describe Sentimental do
 
   let(:analyzer) { Sentimental.new(threshold: 0.1) }
 
-  describe "#score" do
+  describe "#sentiment" do
+    it "returns :positive when the score > threshold" do
+      expect(analyzer.sentiment("I love ruby <3")).to be :positive
+    end
+
+    it "returns :negative when the score < -threshold" do
+      expect(analyzer.sentiment("I hate javascript")).to be :negative
+    end
+
+    it "returns :positive when -threshold < score < threshold" do
+      expect(analyzer.sentiment("je en sais pas")).to be :neutral
+    end
+  end
+
+  describe "#classify" do
+    it "is true when in the class" do
+      expect(analyzer.classify("I love ruby")).to be_truthy
+    end
+
+    it "is false otherwise" do
+      expect(analyzer.classify("je ne sais pas")).to be_falsy
+      expect(analyzer.classify("i hate java")).to be_falsy
+    end
+  end
+
+  describe "initialization" do
+    subject do
+      Sentimental.new(
+        threshold: 0.2,
+        word_scores: {"non" => -1.0},
+        neutral_regexps: [/.*/],
+      )
+    end
+
+    it "takes multiple init params" do
+      expect(subject.threshold).to eq 0.2
+      expect(subject.word_scores["non"]).to eq -1.0
+      expect(subject.neutral_regexps).to include /.*/
+    end
+  end
+
+  describe "neutral regexp" do
+    context "when there is some neutral regexp" do
+      let(:text_neutral) {"Do you love ruby?"}
+      let(:text) {"I love ruby"}
+
+      before do
+        analyzer.neutral_regexps << /\?\s*$/
+      end
+
+      it "scores it to 0" do
+        expect(analyzer.score(text_neutral)).to eq 0
+        expect(analyzer.score(text)).not_to eq 0
+      end
+    end
+  end
+
+  describe "n-grams" do
+    let(:word_scores) { nil }
+    subject do
+      Sentimental.new(word_scores: word_scores, ngrams: 3)
+    end
+
+    it "is initialized by ngrams param" do
+      expect(subject.ngrams).to eq 3
+    end
+    
+    context "there is n-grams in the dictionary" do
+      let(:word_scores) {{"happy hour" => 1.0, "not happy hour" => -5.0}}
+      let(:text) { "why not happy hour, but happy so hour?" }
+
+      it "update scores regarding to n-grams" do
+        expect(subject.score(text)).to eq -4
+      end
+    end
+
+    context "there's n-grams longer than specified in dictionary" do
+      let(:word_scores) {{"happy hour" => 1.0, "not so happy hour" => -5.0}}
+      let(:text) { "why not so happy hour ?" }
+
+      it "ignores the lines" do
+        expect(subject.score(text)).to eq 1
+      end
+    end
+  end
+
+  describe "scoring in a normal context" do
     subject do
       analyzer.score(text)
     end
@@ -55,18 +141,5 @@ describe Sentimental do
       end
     end
 
-    context "when there is some neutral regexp" do
-      let(:text_neutral) {"Do you love ruby?"}
-      let(:text) {"I love ruby"}
-
-      before do
-        analyzer.neutral_regexps << /\?\s*$/
-      end
-
-      it "scores it to 0" do
-        expect(analyzer.score(text_neutral)).to eq 0
-        expect(analyzer.score(text)).not_to eq 0
-      end
-    end
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: sentimental
 version: !ruby/object:Gem::Version
-  version: 1.1.1
+  version: 1.2.1
 platform: ruby
 authors:
 - Jeff Emminger
@@ -10,7 +10,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-03-17 00:00:00.000000000 Z
+date: 2016-04-22 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -93,7 +93,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.4.6
+rubygems_version: 2.4.8
 signing_key: 
 specification_version: 4
 summary: Simple sentiment analysis