turkish_stemmer 0.1.4 → 0.1.5

data/spec/spec_helper.rb

@@ -0,0 +1,19 @@
+# This file was generated by the `rspec --init` command. Conventionally, all
+# specs live under a `spec` directory, which RSpec adds to the `$LOAD_PATH`.
+# Require this file using `require "spec_helper"` to ensure that it is only
+# loaded once.
+#
+# See http://rubydoc.info/gems/rspec-core/RSpec/Core/Configuration
+require 'turkish_stemmer'
+
+RSpec.configure do |config|
+  config.treat_symbols_as_metadata_keys_with_true_values = true
+  config.run_all_when_everything_filtered = true
+  config.filter_run :focus
+
+  # Run specs in random order to surface order dependencies. If you find an
+  # order dependency and want to debug it, you can fix the order by providing
+  # the seed, which is printed after each run.
+  #     --seed 1234
+  config.order = 'random'
+end

data/spec/support/fixtures.csv

@@ -0,0 +1,101 @@
+evimden,ev,from my house,ev-(i)m-den,
+göz,göz,eye,--,
+güzelmişsin,güzel,you were beautiful,güzel-miş-sin,rumor
+etkilerden,etki,from the effects,etki-ler-den,
+çocukmuş,çocuk,it was child,çocuk-miş,rumor
+kediymiş,kedi,it was cat,kedi-(y)miş,rumor
+balığım,balık,my fish,balık-(i)m,
+doktoruymuşsunuz,doktor,you were his/her/its doctor,doktor-i-(y)miş-siniz,rumor
+kalelerimizdekilerden,kale,the ones that are from our castle,kale-ler-(i)miz-de-ki-ler-den,
+çocuğuymuşumçasına,çocuk,as if i was his/her child ,çocuk-i-(y)miş-im-cesine,
+kedileriyle,kedi,with his/her/its cats,kedi-ler-i-(y)le,kedileri+ile
+çocuklarımmış,çocuk,they were my children,çocuk-ler-(i)m-miş,rumor
+kitabımızdı,kitap,it was our book,kitap-(i)miz-di,
+kelimelerin,kelime,"""your"" -or- ""of"" words",kelime-ler-(i)n -or- kelime-ler-in,both ways
+kayısısı,kayısı,his/her/its apricot,kayısı-(s)ı,
+eriğinin,erik,"of ""your"" -or- ""his/her/its"" plum",erik-(i)n-in -or- erik-i-(n)in,both ways
+eriğindeki,erik,"the one that is at ""your"" -or- ""his/her/its"" plum ",erik-(i)n-de-ki - or- erik-i-(n)de-ki,both ways
+eriğinden,erik,"from ""your"" -or- ""his/her/its"" plum",erik-(i)n-den -or- erik-i-(n)den,both ways
+eriğine,erik,"to ""your"" -or- ""his/her/its"" plum",erik-(i)n-e -or- erik-i-(n)e,both ways
+eriğinde,erik,"at ""your"" -or- ""his/her/its"" plum",erik-(i)n-de -or- erik-i-(n)de,both ways
+kayısısına,kayısı,his/her/its apricot,kayısı-(s)ı-(n)a,
+kayısısında,kayısı,at his/her/its apricot,kayısı-(s)ı-(n)da,
+saatlerimiz,saat,our watches/hours,saat-ler-(i)miz,
+kalemimin,kalem,of my pencil,kalem-(i)m-in,
+ucu,uç,nib of...,uç-i,
+kalelerdekilerden,kale,from the ones that are at (the) castle,kale-ler-de-ki-ler-den,
+kalelerdekilerin,kale,of the ones that are at (the) castle,kale-ler-de-ki-ler-in,
+kalelerimizdekilerde,kale,at the ones that are at (the) castle,kale-ler-(i)miz-de-ki-ler-de,
+kaleninkinin,kale,of the one that belongs to (the) castle,kale-(n)in-ki-nin,
+kalemizinkinin,kale,of the one that belongs to our castle,kale-miz-(i)n-ki-(n)in,
+kalelerindeki,kale,"the one that is at ""their castle"" -or- ""his/her/its castles""",kale-leri-(n)de-ki -or- kale-ler-i-(n)de-ki,both ways
+erikleri,erik,"""their plum"" -or- ""his/her/its plums""",erik-leri -or- erik-ler-i,both ways
+erikler,erik,(the) plums,erik-ler,
+eriğim,erik,my plum,erik-(i)m,
+eriğimiz,erik,our plum,erik-(i)miz,
+eriğin,erik,your plum,erik-(i)n,
+eriğiniz,erik,your plum,erik-(i)niz,2nd person in plural
+eriği,erik,his/her/its plum,erik-i,
+eriğini,erik,"""your"" -or- ""his/her/its"" plum",erik-(i)n-i -or- erik-i-(n)i,both ways
+eriğinin,erik,"of ""your"" -or- ""his/her/its"" plum",erik-(i)n-in -or- erik-i-(n)in,both ways
+eriğe,erik,to (the) plum,erik-e,
+eriğine,erik,"to ""your"" -or- ""his/her/its"" plum",erik-(i)n-e -or- erik-i-(n)e,both ways
+eriklerine,erik,"to ""their plum"" -or- ""his/her/its plums""",erik-leri-(n)e -or- erik-ler-i-(n)e,both ways
+erikte,erik,at (the) plum,erik-de,
+eriğinde,erik,"at ""your"" -or- ""his/her/its"" plum",erik-(i)n-de -or- erik-i-(n)de,both ways
+erikten,erik,from (the) plum,erik-den,
+eriğinden,erik,"from ""your"" -or- ""his/her/its"" plum",erik-(i)n-den -or- erik-i-(n)den,both ways
+eriğindeki,erik,"the one that is at ""your"" -or- ""his/her/its"" plum",erik-(i)n-de-ki - or- erik-i-(n)de-ki,both ways
+eriğiyle,erik,with his/her/its plum,erik-i-(y)le,
+eriğinin,erik,"of ""your"" -or- ""his/her/its"" plum",erik-(i)n-in -or- erik-i-(n)in,both ways
+eriğindeki,erik,"the one that is at ""your"" -or- ""his/her/its"" plum",erik-(i)n-de-ki - or- erik-i-(n)de-ki,both ways
+eriğince,erik,"after ""your"" -or- ""his/her/its"" plum",erik-(i)n-ce -or- erik-i-(n)ce,both ways
+gülüm,gül,my rose,gül-(i)m,
+erikteki,erik,the one that is at (the) plum,erik-de-ki,
+eriktekilerden,erik,the ones that are from (the) plum,erik-de-ki-ler-den,
+eriklerdeki,erik,the ones that are at (the) plum,erik-ler-de-ki,
+kitabı,kitap,(the) book,kitap-i,
+ağacı,ağaç,(the) tree,ağaç-i,
+eriğim,erik,my plum / i am plum,erik-(i)m / erik-im,
+kayısıyım,kayısı,i am apricot,kayısı-(y)ım,
+eriksem,erik,if i am plum,erik-se-m,
+eriksen,erik,if you are plum,erik-se-n,
+erikse,erik,if he/she/it is plum,erik-se,
+erikseniz,erik,if you are plum,erik-se-niz,2nd person in plural
+erikseler,erik,if they are plum,erik-se-ler,
+erikti,erik,he/she/it was plum,erik-di,
+eriktiniz,erik,you were plum,erik-di-niz,2nd person in plural
+eriktiler,erik,they were plum,erik-di-ler,
+erikmiş,erik,it was plum,erik-miş,rumor
+erikmişçesine,erik,as if it was plum,erik-miş-cesine,
+erikmiştir,erik,it was plum,erik-miş-dir,rumor
+erikmişim,erik,i was plum,erik-miş-(i)m,rumor
+erikmişsin,erik,you were plum,erik-miş-sin,rumor
+erikmişsindir,erik,you happened to be plum,erik-miş-sin-dir,rumor
+erikmişimdir,erik,i happened to be plum,erik-miş-im-dir,rumor
+erikmişiz,erik,we were plum,erik-miş-iz,rumor
+erikmişizdir,erik,we happened to be plum,erik-miş-iz-dir,rumor
+erikmişsiniz,erik,you were plum,erik-miş-siniz,2nd person in plural + rumor
+erikmişsinizdir,erik,you happened to be plum,erik-miş-siniz-dir,2nd person in plural + rumor
+erikmişler,erik,they were plum,erik-miş-ler,rumor
+erikmişlerdir,erik,they happened to be plum,erik-miş-ler-dir,rumor
+erikmişimcesine,erik,as if i was plum,erik-miş-im-cesine,
+erikmişsincesine,erik,as if you were plum,erik-miş-sin-cesine,
+erikmişizcesine,erik,as if we were plum,erik-miş-iz-cesine,
+erikmişsinizcesine,erik,as if you were plum,erik-miş-siniz-cesine,2nd person in plural
+erikmişlercesine,erik,as if they were plum,erik-miş-ler-cesine,
+erikler,erik,plums,erik-ler,
+eriğim,erik,my plum / i am plum,erik-(i)m / erik-im,
+eriksin,erik,you are plum,erik-sin,
+erik,erik,plum / he/she/it is plum,erik / erik,
+eriğiz,erik,we are plum,erik-iz,
+eriksiniz,erik,you are plum,erik-siniz,2nd person in plural
+erikler,erik,they are plum,erik-ler,
+eriktir,erik,it is plum,erik-dir,assumption
+eriktirler,erik,they are plum,erik-dir-ler,assumption
+erikken,erik,while he/she/it was plum,erik-(i)ken,
+kötüymüş,kötü,he/she/it is bad,kötü-(y)miş,rumor
+yüz,yüz,face / hundred,yüz,double meaning
+muş,muş,--,--,this is a suffix
+ad,ad,name,ad,
+soyad,soyad,surname,soyad,soy+ad = lineage+name

data/spec/turkish_stemmer_spec.rb

@@ -0,0 +1,522 @@
+# coding: utf-8
+require "spec_helper"
+require "pry"
+require "csv"
+
+describe TurkishStemmer do
+
+  describe ".count_syllables" do
+    it "counts syllables correctly" do
+      expect(described_class.count_syllables("erikler")).to eq 3
+      expect(described_class.count_syllables("çocuklarımmış")).to eq 5
+    end
+  end
+
+  describe ".vowels" do
+    it "returns all vowels of a word" do
+      expect(described_class.vowels("kötüymüş")).to eq(%w(ö ü ü))
+    end
+  end
+
+  describe ".has_roundness?" do
+    context "when vowel is empty" do
+      it "has roundness" do
+        expect(described_class).to have_roundness(nil, "a")
+      end
+    end
+
+    context "when candidate is empty" do
+      it "has roundness" do
+        expect(described_class).to have_roundness("a", nil)
+      end
+    end
+
+    context "when an unrounded vowel is passed" do
+      let(:vowel) { described_class::UNROUNDED_VOWELS.chars.to_a.sample }
+
+      context "and candidate is an unrounded vowel too" do
+        let(:candidate) { described_class::UNROUNDED_VOWELS.chars.to_a.sample }
+
+        it "has roundness" do
+          expect(described_class).to have_roundness(vowel, candidate)
+        end
+      end
+
+      context "and candidate is not an unrounded vowel" do
+        let(:candidate) { described_class::ROUNDED_VOWELS.chars.to_a.sample }
+
+        it "does not have roundness" do
+          expect(described_class).not_to have_roundness(vowel, candidate)
+        end
+      end
+    end
+
+    context "when a rounded vowel is passed" do
+      let(:vowel) { described_class::ROUNDED_VOWELS.chars.to_a.sample }
+
+      context "and one of 'a', 'e', 'u' or 'ü' is a candidate" do
+        let(:candidate) { described_class::FOLLOWING_ROUNDED_VOWELS.chars.to_a.sample }
+
+        it "has roundness" do
+          expect(described_class).to have_roundness(vowel, candidate)
+        end
+      end
+
+      context "and candidate is 'o'" do
+        let(:candidate) { 'o' }
+
+        it "does not have roundness" do
+          expect(described_class).not_to have_roundness(vowel, candidate)
+        end
+      end
+    end
+  end
+
+  describe ".has_frontness?" do
+    context "when vowel is empty" do
+      it "has frontness" do
+        expect(described_class).to have_frontness(nil, "a")
+      end
+    end
+
+    context "when candidate is empty" do
+      it "has frontness" do
+        expect(described_class).to have_frontness("a", nil)
+      end
+    end
+
+    context "when a front vowel is passed" do
+      let(:vowel) { described_class::FRONT_VOWELS.chars.to_a.sample }
+
+      context "and candidate is a front vowel" do
+        let(:candidate) { described_class::FRONT_VOWELS.chars.to_a.sample }
+
+        it "has frontness" do
+          expect(described_class).to have_frontness(vowel, candidate)
+        end
+      end
+
+      context "and candidate is a back vowel" do
+        let(:candidate) { described_class::BACK_VOWELS.chars.to_a.sample }
+
+        it "does not have frontness" do
+          expect(described_class).not_to have_frontness(vowel, candidate)
+        end
+      end
+    end
+
+    context "when a back vowel is passed" do
+      let(:vowel) { described_class::BACK_VOWELS.chars.to_a.sample }
+
+      context "and candidate is a front vowel" do
+        let(:candidate) { described_class::FRONT_VOWELS.chars.to_a.sample }
+
+        it "does not have frontness" do
+          expect(described_class).not_to have_frontness(vowel, candidate)
+        end
+      end
+
+      context "and candidate is a back vowel" do
+        let(:candidate) { described_class::BACK_VOWELS.chars.to_a.sample }
+
+        it "has frontness" do
+          expect(described_class).to have_frontness(vowel, candidate)
+        end
+      end
+    end
+  end
+
+  describe ".has_vowel_harmony?" do
+    it "has vowel harmony for valid Turkish words" do
+      expect(described_class).to have_vowel_harmony("Türkiyedir")
+      expect(described_class).to have_vowel_harmony("kapıdır")
+      expect(described_class).to have_vowel_harmony("gündür")
+      expect(described_class).to have_vowel_harmony("paltodur")
+    end
+
+    it "does not have vowel harmony for loanwords" do
+      expect(described_class).not_to have_vowel_harmony("kürdan")
+    end
+
+    it "does not have vowel harmony for exceptions" do
+      expect(described_class).not_to have_vowel_harmony("anne")
+      expect(described_class).not_to have_vowel_harmony("kardeş")
+    end
+  end
+
+  describe ".affix_morphological_stripper" do
+    context "when states are empty" do
+      it "returns the word" do
+        expect(
+          described_class.
+            affix_morphological_stripper("kapıdır", suffixes: :test)).
+        to eq(["kapıdır"])
+      end
+    end
+
+    context "when suffixes are empty" do
+      it "return the word" do
+        expect(
+          described_class.
+            affix_morphological_stripper("kapıdır", states: :test)).
+        to eq(["kapıdır"])
+      end
+    end
+
+    context "when there exist states and suffixes"  do
+      let(:states) {
+        described_class.
+          load_states_or_suffixes("spec/fixtures/simple_state.yml")
+      }
+
+      let(:suffixes) {
+        described_class.
+          load_states_or_suffixes("spec/fixtures/simple_suffix.yml")
+      }
+
+      it "generates pendings for the initial state" do
+        described_class.should_receive(:generate_pendings).with(:a,
+          "word", states, suffixes).and_call_original
+
+        described_class.affix_morphological_stripper("word",
+          states: states, suffixes: suffixes)
+      end
+    end
+
+    context "when a transition is valid"  do
+      let(:states) {
+        described_class.
+          load_states_or_suffixes("spec/fixtures/simple_state.yml")
+      }
+
+      let(:suffixes) {
+        described_class.
+          load_states_or_suffixes("spec/fixtures/simple_suffix.yml")
+      }
+
+      context "and the transit state is a final state" do
+        it "removes similar pending transitions" do
+          described_class.should_receive(:mark_stem).with(
+            "guzelim", suffixes[:s1]).and_call_original
+
+          described_class.affix_morphological_stripper(
+            "guzelim", states: states, suffixes: suffixes)
+        end
+
+        context "with no other transitions" do
+          it "stems the word" do
+            expect(
+              described_class.
+                affix_morphological_stripper("guzelim",
+                  states: states, suffixes: suffixes)).
+            to eq ["guzel"]
+          end
+        end
+
+        context "with other transitions"  do
+          let(:states) {
+            described_class.load_states_or_suffixes("spec/fixtures/simple_state_02.yml")
+          }
+
+          it "adds more pendings to check" do
+            described_class.should_receive(:mark_stem).with("guzelim",
+              suffixes[:s1]).and_call_original
+
+            described_class.affix_morphological_stripper("guzelim",
+              states: states, suffixes: suffixes)
+          end
+        end
+      end
+    end
+
+    context "when one suffix matches correctly with a given word" do
+      it "does not compare other suffixes in the same transition" do
+        described_class.
+          should_receive(:mark_stem).
+          with(anything, anything).
+          # only for suffixes [sUnUz, nUz]
+          exactly(2).times.
+          and_call_original
+
+        puts described_class.
+          affix_morphological_stripper("taksicisiniz",
+            states: described_class::NOMINAL_VERB_STATES,
+            suffixes: described_class::NOMINAL_VERB_SUFFIXES)
+      end
+    end
+  end
+
+  describe ".stem"  do
+    context "when input is single syllable" do
+      it "returns the input as is" do
+        expect(described_class.stem("ev")).to eq "ev"
+      end
+    end
+
+    context "when input has zero syllables - one consonant" do
+      it "returns the input as is" do
+        expect(described_class.stem("p")).to eq "p"
+      end
+    end
+  end
+
+  describe ".last_consonant!"  do
+    context "when last consonant is among 'b', 'c', 'd' or 'ğ'" do
+      it "is replaced by 'p', 'ç', 't' or 'k'" do
+        expect(described_class.last_consonant!('kebab')).to eq('kebap')
+        expect(described_class.last_consonant!('kebac')).to eq('kebaç')
+        expect(described_class.last_consonant!('kebad')).to eq('kebat')
+        expect(described_class.last_consonant!('kebağ')).to eq('kebak')
+      end
+    end
+
+    context "when word belongs to protected words" do
+      it "does not replace last consonant" do
+        expect(described_class.last_consonant!('ad')).to eq('ad')
+      end
+    end
+  end
+
+  describe ".mark_stem" do
+    let(:suffix) do
+      {
+        "name" => "-dir",
+        "regex" => "dir",
+        "optional_letter" => false,
+        "check_harmony" => true
+      }
+    end
+
+    context "when suffix has harmony check on" do
+      before do
+        suffix["regex"] = "dan"
+      end
+
+      context "and word does not obey harmony rules" do
+        it "does not stem a word that does not obey harmony rules" do
+          expect(described_class.mark_stem("kürdan", suffix)).to eq(
+            { stem: false, word: "kürdan", suffix_applied: nil })
+        end
+
+        context "and word belongs to exceptions" do
+          before do
+            suffix["regex"] = "ler"
+          end
+          it "stems the word" do
+            expect(described_class.mark_stem("saatler", suffix)).to eq(
+              { stem: true, word: "saat", suffix_applied: "ler" })
+          end
+        end
+      end
+
+    end
+
+    context "when suffix has harmony check off" do
+      before do
+        suffix["regex"] = "dan"
+        suffix["check_harmony"] = false
+      end
+
+      it "stems a word that does not obey harmony rules" do
+        expect(
+          described_class.
+            mark_stem("kürdan", suffix)).
+        to eq({ stem: true, word: "kür", suffix_applied: "dan" })
+      end
+    end
+
+    context "when word matches suffix" do
+      it "partially stems a word" do
+        expect(
+          described_class.
+            mark_stem("Türkiyedir", suffix)).
+        to eq({ stem: true, word: "Türkiye", suffix_applied: "dir" })
+      end
+
+
+      context "when suffix has (y) as optional letter" do
+        before do
+          suffix["optional_letter"] = "y|y"
+          suffix["regex"] = "um"
+        end
+
+        context "and new word has valid last 'y' symbol" do
+          it "stems correctly and increases the suffix" do
+            expect(
+              described_class.
+                mark_stem("loyum", suffix)).
+            to eq({ stem: true, word: "lo", suffix_applied: "yum" })
+          end
+        end
+
+        context "and new word does not have valid last 'y' symbol" do
+          it "does not stem the word" do
+            expect(
+              described_class.
+                mark_stem("lotyum", suffix)).
+            to eq({ stem: false, word: "lotyum", suffix_applied: nil })
+          end
+        end
+      end
+    end
+  end
+
+  describe ".generate_pendings" do
+    let(:states) { described_class::NOMINAL_VERB_STATES }
+    let(:suffixes) { described_class::NOMINAL_VERB_SUFFIXES }
+
+    it "raises an error if state does not exist" do
+      expect {
+        described_class.
+          generate_pendings(1, "satıyorsunuz", states, suffixes)
+      }.to raise_error(ArgumentError, "State #{1} does not exist")
+    end
+
+    context "when state key does not have transitions" do
+      it "returns an empty array" do
+        expect(
+          described_class.
+            # :f state does not have transitions
+            generate_pendings(:f, "satıyorsunuz", states, suffixes)).
+        to eq []
+      end
+    end
+
+    context "when state key has transitions" do
+      it "returns an array of hashes for each transition" do
+        expect(
+          described_class.
+            generate_pendings(:a, "satıyorsunuz", states, suffixes).first.keys).
+        to eq [:suffix, :to_state, :from_state, :word, :mark]
+      end
+
+      it "sets :from_state key to current key state" do
+        expect(
+          described_class.
+            generate_pendings(:a, "satıyorsunuz", states, suffixes).first[:from_state]).
+        to eq :a
+      end
+    end
+  end
+
+  describe ".valid_optional_letter?" do
+    context "when last letter of the word is not equal to candidate" do
+      it "responds with [true,nil] - indicating that there was not match" do
+        expect(
+          described_class.valid_optional_letter?("test", "r")).
+        to eq([true, nil])
+      end
+    end
+
+    context "when there is a vowel match" do
+      context "and the previous char is a vowel" do
+        it "responds with false" do
+          expect(
+            described_class.
+              valid_optional_letter?("takcicii", "i")).
+          to eq([false, "i"])
+        end
+      end
+
+      context "and the previous char is a consonant" do
+        it "responds with true" do
+          expect(
+            described_class.
+              valid_optional_letter?("okula", "a")).
+          to eq([true, "a"])
+        end
+      end
+    end
+
+    context "when there is a consonant match" do
+      context "and the previous char is a vowel" do
+        it "responds with true" do
+          expect(
+            described_class.
+              valid_optional_letter?("litiy", "y")).
+          to eq([true, "y"])
+        end
+      end
+
+      context "and the previous char is a consonant" do
+        it "responds with true" do
+          expect(
+            described_class.
+              valid_optional_letter?("lity", "y")).
+          to eq([false, "y"])
+        end
+      end
+    end
+  end
+
+  describe ".stem_post_process"  do
+    context "when input stream has words with last consonant replacements" do
+      it "replaces last consonant" do
+        expect(described_class.stem_post_process(["kebab"], "word")).to eq("kebap")
+      end
+    end
+
+    it "flattens and uniq results" do
+      expect(described_class.stem_post_process(["kitap",["kitap"]], "word")).to eq("kitap")
+    end
+
+    it "removes no syllables words" do
+      expect(described_class.stem_post_process(["kitap", "k"], "word")).to eq("kitap")
+    end
+
+    context "when multiple stem candidates exist" do
+      it "returns the shortest" do
+        pending("fix this")
+        expect(described_class.stem_post_process(["kitap", "kita", "kit"], "word")).to eq "kit"
+      end
+
+      context "and word belongs to selection list" do
+        it "returns this word" do
+          expect(described_class.stem_post_process(
+            ["su", "suy", "suyu"], "suyu")).to eq "su"
+        end
+      end
+    end
+  end
+
+  describe ".proceed_to_stem?"  do
+    context "when word has 1 or less syllables" do
+      it "returns false" do
+        expect(described_class.proceed_to_stem?("kit")).not_to be
+      end
+    end
+
+    context "when word is nil" do
+      it "returns false" do
+        expect(described_class.proceed_to_stem?(nil)).not_to be
+      end
+    end
+
+    context "when word is empty" do
+      it "returns false" do
+        expect(described_class.proceed_to_stem?("")).not_to be
+      end
+    end
+
+    context "when word is among protected words" do
+      it "returns false" do
+        expect(described_class.proceed_to_stem?("soyad")).not_to be
+      end
+    end
+
+    context "when word contains non Turkish letters" do
+      it "returns false" do
+        expect(described_class.proceed_to_stem?("τελειο")).not_to be
+        expect(described_class.proceed_to_stem?("&aa")).not_to be
+      end
+    end
+  end
+
+  context "1:1 testing with paper" do
+    CSV.read("spec/support/fixtures.csv").each do |row|
+      it "stems #{row[0]} correct" do
+        expect(described_class.stem(row[0].downcase)).to eq row[1].downcase
+      end
+    end
+  end
+end

data/turkish_stemmer.gemspec

@@ -0,0 +1,36 @@
+# coding: utf-8
+lib = File.expand_path('../lib', __FILE__)
+$LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
+require 'turkish_stemmer/version'
+
+Gem::Specification.new do |spec|
+  spec.name          = "turkish_stemmer"
+  spec.version       = TurkishStemmer::VERSION
+  spec.authors       = ["Tasos Stathopoulos", "Giorgos Tsiftsis"]
+  spec.email         = ["stathopa@skroutz.gr", "giorgos.tsiftsis@skroutz.gr"]
+  spec.summary       = %q{A simple Turkish stemmer}
+  spec.description   = %q{A simple Turkish stemmer}
+  spec.homepage      = "https://github.com/skroutz/turkish_stemmer"
+  spec.license       = "MIT"
+
+  spec.files         = `git ls-files`.split($/)
+  #spec.files.reject! { Dir['benchmarks'] }
+  spec.executables   = spec.files.grep(%r{^bin/}) { |f| File.basename(f) }
+  spec.test_files    = spec.files.grep(%r{^(test|spec|features)/})
+  spec.require_paths = ["lib"]
+
+  spec.add_dependency "activesupport", [">= 3.0.0"]
+
+  spec.add_development_dependency "bundler", "~> 1.5"
+  spec.add_development_dependency "rake"
+  spec.add_development_dependency "rspec"
+  spec.add_development_dependency "ruby-stemmer"
+
+  if RUBY_ENGINE == "ruby"
+    if RUBY_VERSION >= "2.0.0"
+      spec.add_development_dependency "pry-byebug"
+    else
+      spec.add_development_dependency "pry"
+    end
+  end
+end