fuzzily 0.0.1 → 0.0.2

data/.gitignore

@@ -3,7 +3,6 @@
 .bundle
 .config
 .yardoc
-Gemfile.lock
 InstalledFiles
 _yardoc
 coverage

data/Gemfile.lock

@@ -0,0 +1,46 @@
+PATH
+  remote: .
+  specs:
+    fuzzily (0.0.2)
+      activerecord (~> 2.3)
+
+GEM
+  remote: https://rubygems.org/
+  specs:
+    activerecord (2.3.14)
+      activesupport (= 2.3.14)
+    activesupport (2.3.14)
+    appraisal (0.4.1)
+      bundler
+      rake
+    coderay (1.0.8)
+    diff-lcs (1.1.3)
+    method_source (0.8.1)
+    pry (0.9.10)
+      coderay (~> 1.0.5)
+      method_source (~> 0.8)
+      slop (~> 3.3.1)
+    pry-nav (0.2.2)
+      pry (~> 0.9.10)
+    rake (0.9.2.2)
+    rspec (2.11.0)
+      rspec-core (~> 2.11.0)
+      rspec-expectations (~> 2.11.0)
+      rspec-mocks (~> 2.11.0)
+    rspec-core (2.11.1)
+    rspec-expectations (2.11.3)
+      diff-lcs (~> 1.1.3)
+    rspec-mocks (2.11.3)
+    slop (3.3.3)
+    sqlite3 (1.3.6)
+
+PLATFORMS
+  ruby
+
+DEPENDENCIES
+  appraisal
+  fuzzily!
+  pry
+  pry-nav
+  rspec
+  sqlite3

data/lib/fuzzily/migration.rb

@@ -2,34 +2,37 @@ require 'active_record'
 
 module Fuzzily
   module Migration
+    def self.extended(by)
+      by.singleton_class.class_eval do
+        def trigrams_table_name=(custom_name)
+          @trigrams_table_name = custom_name
+        end
 
-    def trigrams_table_name=(custom_name)
-      @trigrams_table_name = custom_name
-    end
-
-    def trigrams_table_name
-      @trigrams_table_name ||= :trigrams
-    end
+        def trigrams_table_name
+          @trigrams_table_name ||= :trigrams
+        end
 
-    def up
-      create_table trigrams_table_name do |t|
-        t.string  :trigram, :limit => 3
-        t.integer :score
-        t.integer :owner_id
-        t.string  :owner_type
-        t.string  :fuzzy_field
-      end
+        def up
+          create_table trigrams_table_name do |t|
+            t.string  :trigram, :limit => 3
+            t.integer :score
+            t.integer :owner_id
+            t.string  :owner_type
+            t.string  :fuzzy_field
+          end
 
-      add_index trigrams_table_name,
-        [:owner_type, :fuzzy_field, :trigram, :owner_id, :score],
-        :name => :index_for_match
-      add_index trigrams_table_name,
-        [:owner_type, :owner_id],
-        :name => :index_by_owner
-    end
+          add_index trigrams_table_name,
+            [:owner_type, :fuzzy_field, :trigram, :owner_id, :score],
+            :name => :index_for_match
+          add_index trigrams_table_name,
+            [:owner_type, :owner_id],
+            :name => :index_by_owner
+        end
 
-    def down
-      drop_table trigrams_table_name
+        def down
+          drop_table trigrams_table_name
+        end
+      end
     end
   end
 end

data/lib/fuzzily/model.rb

@@ -39,9 +39,10 @@ module Fuzzily
       def matches_for(text, options = {})
         options[:limit] ||= 10
         self.
-          scoped(:select => 'owner_id, owner_type, SUM(score) AS score').
+          scoped(:select => 'owner_id, owner_type, SUM(score) AS total_score').
           scoped(:group => :owner_id).
-          scoped(:order => 'score DESC', :limit => options[:limit]).
+          scoped(:order => 'total_score DESC').
+          scoped(:limit => options[:limit]).
           with_trigram(text.extend(String).trigrams).
           map(&:owner)
       end

data/lib/fuzzily/searchable.rb

@@ -24,7 +24,8 @@ module Fuzzily
         :class_name => trigram_class_name,
         :as => :owner,
         :conditions => { :fuzzy_field => field.to_s },
-        :dependent => :destroy
+        :dependent => :destroy,
+        :autosave => true
 
       singleton_class.send(:define_method,"find_by_fuzzy_#{field}".to_sym) do |*args|
         case args.size
@@ -32,13 +33,41 @@ module Fuzzily
           when 2 then pattern, options = args
           else        raise 'Wrong # of arguments'
         end
-        Trigram.scoped(options).for_model(self.name).for_field(field).matches(pattern)
+
+        trigram_class_name.constantize.
+          scoped(options).
+          for_model(self.name).
+          for_field(field.to_s).
+          matches_for(pattern)
+      end
+
+      singleton_class.send(:define_method,"bulk_update_fuzzy_#{field}".to_sym) do
+        trigram_class = trigram_class_name.constantize
+
+        self.scoped(:include => trigram_association).find_in_batches(:batch_size => 100) do |batch|
+          inserts = []
+          batch.each do |record|
+            record.send(field).extend(String).trigrams.each do |trigram|
+              inserts << sanitize_sql_array(['(?,?,?,?,?)', self.name, record.id, field.to_s, 1, trigram])
+            end
+          end
+
+          trigram_class.transaction do
+            batch.each { |record| record.send(trigram_association).delete_all }
+            trigram_class.connection.insert(%Q{
+              INSERT INTO `#{trigram_class.table_name}`
+              (`owner_type`, `owner_id`, `fuzzy_field`, `score`, `trigram`)
+              VALUES
+              #{inserts.join(", ")}
+            })
+          end
+        end
       end
 
       define_method update_trigrams_method do
-        self.send(trigram_association).destroy_all
+        self.send(trigram_association).delete_all
         self.send(field).extend(String).trigrams.each do |trigram|
-          self.send(trigram_association).create!(:score => 1, :trigram => trigram)
+          self.send(trigram_association).create!(:score => 1, :trigram => trigram, :owner_type => self.class.name)
         end
       end
 

data/lib/fuzzily/trigram.rb

@@ -1,25 +1,23 @@
-require 'iconv'
+require 'active_support/core_ext/string/multibyte'
 
 module Fuzzily
   module String
     def trigrams
-      normalized_words.map do |word|
-        (0..(word.length - 3)).map { |index| word[index,3] }
-      end.flatten.uniq
+      normalized = self.normalize
+      (0..(normalized.length - 3)).map { |index| normalized[index,3] }.uniq
     end
 
-    private
+    protected
 
     # Remove accents, downcase, replace spaces and word start with '*',
     # return list of normalized words
-    def normalized_words
-      self.split(/\s+/).map { |word|
-        Iconv.iconv('ascii//translit//ignore', 'utf-8', word).first.downcase.gsub(/\W/,'')
-      }.
-      delete_if(&:empty?).
-      map { |word|
-        "**#{word}"
-      }
+    def normalize
+      # Iconv.iconv('ascii//translit//ignore', 'utf-8', self).first.
+      ActiveSupport::Multibyte::Chars.new(self).
+        mb_chars.normalize(:kd).gsub(/[^\x00-\x7F]/,'').downcase.to_s.
+        gsub(/\W/,' ').
+        gsub(/\s+/,'*').
+        gsub(/^/,'**')
     end
   end
 end

data/lib/fuzzily/version.rb

@@ -1,3 +1,3 @@
 module Fuzzily
-  VERSION = "0.0.1"
+  VERSION = "0.0.2"
 end

data/spec/fuzzily/searchable_spec.rb

@@ -10,10 +10,11 @@ describe Fuzzily::Searchable do
   before(:each) { prepare_owners_table   }
 
   subject do 
-    Stuff.clone.class_eval do
-      def self.name ; 'Stuff' ; end
-      self
+    silence_warnings do
+      Stuff = Class.new(ActiveRecord::Base)
     end
+    def Stuff.name ; 'Stuff' ; end
+    Stuff
   end
 
   describe '.fuzzily_searchable' do
@@ -39,14 +40,14 @@ describe Fuzzily::Searchable do
   end
 
   describe '(callbacks)' do
+    before { subject.fuzzily_searchable :name }
+
     it 'generates trigram records on creation' do
-      subject.fuzzily_searchable :name
       subject.create(:name => 'Paris')
       subject.last.trigrams_for_name.should_not be_empty
     end
 
     it 'generates the correct trigrams' do
-      subject.fuzzily_searchable :name
       record = subject.create(:name => 'FOO')
       Trigram.first.trigram.should    == '**f'
       Trigram.first.owner_id.should   == record.id
@@ -54,19 +55,35 @@ describe Fuzzily::Searchable do
     end
 
     it 'updates all trigram records on save' do
-      subject.fuzzily_searchable :name
       subject.create(:name => 'Paris')
       subject.first.update_attribute :name, 'Rome'
       Trigram.all.map(&:trigram).should =~ %w(**r *ro rom ome)
     end
   end
 
-  describe '#find_by_fuzzy_<field>' do
-    it 'works'
+  describe '#update_fuzzy_<field>!' do
+    it 're-creates trigrams' do
+      subject.fuzzily_searchable :name
+      subject.create(:name => 'Paris')
+      old_ids = Trigram.all.map(&:id)
+      subject.last.update_fuzzy_name!
+      (old_ids & Trigram.all.map(&:id)).should be_empty
+    end
   end
 
-  describe '#update_fuzzy_<field>!' do
-    it 'works'
+  context '(integrationg test)' do
+    describe '#find_by_fuzzy_<field>' do
+      it 'returns records' do
+        subject.fuzzily_searchable :name
+        @paris =   subject.create(:name => 'Paris')
+        @palma =   subject.create(:name => 'Palma de Majorca')
+        @palmyre = subject.create(:name => 'La Palmyre')
+
+        subject.find_by_fuzzy_name('Piris').should_not be_empty
+        subject.find_by_fuzzy_name('Piris').should =~ [@paris, @palma]
+        subject.find_by_fuzzy_name('Paradise').should =~ [@paris, @palma, @palmyre]
+      end
+    end
   end
 
 end

data/spec/fuzzily/trigram_spec.rb

@@ -1,8 +1,23 @@
 require 'spec_helper'
 
 describe Fuzzily::String do
-  it 'splits strings into trigrams'
-  it 'removes accents'
-  it 'removes symbols'
-  it 'handles multi word strings'
+  def result(string)
+    string.extend(described_class).trigrams
+  end
+  
+  it 'splits strings into trigrams' do
+    result('Paris').should == %w(**p *pa par ari ris)
+  end
+
+  it 'removes accents' do
+    result('Montélimar').should == %w(**m *mo mon ont nte tel eli lim ima mar)
+  end
+
+  it 'handles multi word strings' do
+    result('Le Mans').should == %w(**l *le le* e*m *ma man ans)
+  end
+
+  it 'removes symbols' do
+    result('Besse-en-Chandesse').should == %w(**b *be bes ess sse se* e*e *en en* n*c *ch cha han and nde des)
+  end
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: fuzzily
 version: !ruby/object:Gem::Version 
-  hash: 29
+  hash: 27
   prerelease: 
   segments: 
   - 0
   - 0
-  - 1
-  version: 0.0.1
+  - 2
+  version: 0.0.2
 platform: ruby
 authors: 
 - Julien Letessier
@@ -116,6 +116,7 @@ files:
 - .gitignore
 - .rspec
 - Gemfile
+- Gemfile.lock
 - LICENSE.txt
 - README.md
 - Rakefile