Pagination

lib/smart_search.rb

@@ -17,29 +17,40 @@ def self.included(base)
   module ClassMethods
     # Enable SmartSearch for the current ActiveRecord model.
     def smart_search(options = {:on => [], :split => false})
-      if table_exists?
-        # Check if search_tags exists
-        if !is_smart_search?
-
-          cattr_accessor :condition_default, :group_default, :tags, :order_default, :enable_similarity, :default_template_path
-
-          # BETA!
-          cattr_accessor :split_searchable_fields
-          self.split_searchable_fields = options[:split]
-
-          send :include, InstanceMethods
-          self.send(:after_commit, :create_search_tags, :if => :update_search_tags?) unless options[:auto] == false
-          self.send(:after_destroy, :clear_search_tags)
-          self.enable_similarity ||= true
+      begin
+        if table_exists?
+          if SmartSearch::Config.search_models.index(self.name).nil?
+            SmartSearch::Config.search_models << self.name
+          end
+          # Check if search_tags exists
+          if !is_smart_search?
+
+            cattr_accessor :condition_default, :group_default, :tags, :order_default, :enable_similarity, :default_template_path
+
+            # BETA!
+            cattr_accessor :split_searchable_fields
+            self.split_searchable_fields = options[:split]
+
+            send :include, InstanceMethods
+            self.send(:after_commit, :create_search_tags, :if => :update_search_tags?) unless options[:auto] == false
+            self.send(:after_destroy, :clear_search_tags)
+            if options[:disable_similarity].present?
+              self.enable_similarity = false
+            else
+              self.enable_similarity = true
+            end
 
-          attr_accessor :query_score, :dont_update_search_tags
+            attr_accessor :query_score, :dont_update_search_tags
 
-          self.tags = options[:on] || []
-        elsif is_smart_search? && Rails.env.production?
-          # Allow re-adding attributes for search
-          logger.info("Re-Adding search data on #{self.name}: #{options[:on].inspect}".yellow)
-          self.tags += options[:on]
+            self.tags = options[:on] || []
+          elsif is_smart_search?
+            # Allow re-adding attributes for search
+            logger.info("Re-Adding search data on #{self.name}: #{options[:on].inspect}".yellow)
+            self.tags += options[:on]
+          end
         end
+      rescue ActiveRecord::NoDatabaseError => e
+        puts "Could not set up smart_search due to missing database"
       end
     end
 
@@ -49,40 +60,54 @@ def is_smart_search?
     end
 
     # Serach database for given search tags
-    def find_by_tags(tags = "", options = {})
+    def find_by_tags(tags = "", options = {}, &block)
 
       tags = store_history_and_get_sanitized_search_tags(tags)
       tags = map_similarity_tags(tags)
 
-      # Load ranking from Search tags
-      result_ids = []
-      result_scores = {}
+      base_select = "#{self.quoted_table_name}.* #{SmartSearch::Config.order_by_score ? ', SUM(boost) AS score' : ''}"
 
-      results = SmartSearchTag.select("entry_id, sum(boost) as score").group(:entry_id).where(table_name: self.table_name)
+      join_query = "LEFT JOIN #{SmartSearchTag.quoted_table_name}
+        ON #{SmartSearchTag.quoted_table_name}.#{ActiveRecord::Base.connection.quote_column_name('table_name')} = '#{self.table_name}'
+        AND #{SmartSearchTag.quoted_table_name}.#{ActiveRecord::Base.connection.quote_column_name('entry_id')} = #{self.quoted_table_name}.#{self.primary_key}"
 
-      query = case ActiveRecord::Base.connection.adapter_name
-          when 'PostgreSQL'
-            "select entry_id, sum(boost) as score
-                  from smart_search_tags where #{ActiveRecord::Base.connection.quote_column_name('table_name')}= '#{self.table_name}'
-                  group by entry_id
-                  HAVING (#{tags.join(' AND ')})
-                  order by score DESC"
-          else
-            "select entry_id, sum(boost) as score, #{adapater_based_group_method}(search_tags) as grouped_tags
-                  from smart_search_tags where #{ActiveRecord::Base.connection.quote_column_name('table_name')}= '#{self.table_name}' and
-                  (#{tags.join(' AND ')}) group by entry_id order by score DESC"
-          end
+      group_clause = "#{self.quoted_table_name}.#{self.primary_key}, #{SmartSearchTag.column_names.map {|c| SmartSearchTag.quoted_table_name + '.' + c }.join(", ") }"
+
+      order_clause =  if self.order_default
+        "#{self.order_default}"
+      elsif SmartSearch::Config.order_by_score
+        "score DESC"
+      end
+
+      results = case ActiveRecord::Base.connection.adapter_name
+      when 'PostgreSQL'
 
+        self.select(base_select)
+          .joins(join_query)
+          .group(group_clause)
+          .having("#{tags.join(' AND ')}")
+      else
+        self.select(base_select)
+          .joins(join_query)
+          .group(group_clause)
+          .having("#{tags.join(' AND ')}")
+      end
+
+      if options[:distinct] == true
+        results = results.distinct
+      end
 
-      SmartSearchTag.connection.select_all(query).each do |r|
-        result_ids << r["entry_id"].to_i
-        result_scores[r["entry_id"].to_i] = r['score'].to_f
+      if block_given?
+        results = yield(results)
       end
 
-      results     =  self.where(self.primary_key => result_ids)
+      results = results.offset(options[:offset]) if options[:offset]
+      results = results.limit(options[:per_page]) if options[:per_page]
+
+      return results
     end
 
-    def find_by_splitted_tags(search_fields = {})
+    def find_by_splitted_tags(search_fields = {}, &block)
       sanitized_search_fields = {}
       search_fields.each do |field, tags|
         next if tags.blank?
@@ -100,14 +125,18 @@ def find_by_splitted_tags(search_fields = {})
       end
 
       result_ids = eval(result_list.map(&:to_s).join(" & "))
-      self.where(self.primary_key => result_ids)
+      results = self.where(self.primary_key => result_ids)
+
+      if block_given?
+        results = yield(results)
+      end
     end
 
     # Private Query Helper Methods
     private
     def store_history_and_get_sanitized_search_tags(orig_tags)
       orig_tags = orig_tags.join(" ") if orig_tags.is_a?(Array)
-      sanitized_tags = orig_tags.gsub(/[\(\)\[\]\'\"\*\%\|\&]/, '').split(/[\ -]/).select {|t| !t.blank?}
+      sanitized_tags = orig_tags.gsub(/[\(\)\[\]\'\"\*\%\|\&\+\.\$\?]/, ' ').split(/[\ -]/).select {|t| !t.blank?}
 
       # Save Data for similarity analysis
       if sanitized_tags.join(' ').size > 3
@@ -131,7 +160,7 @@ def map_similarity_tags(tags)
           when 'PostgreSQL'
             "string_agg(search_tags, ' ') ~* '#{similars}'"
           else
-            "search_tags REGEXP '#{similars}'"
+             "search_tags REGEXP '#{similars}'"
           end
         end
 
@@ -187,23 +216,33 @@ def update_search_tags?
     end
 
     # create search tags for this very record based on the attributes defined in ':on' option passed to the 'Class.smart_search' method
-    def create_search_tags
+    def create_search_tags(options = {})
       # storing tags must never fail the systems
       begin
 
         self.get_calculated_tags_list!
         self.clear_search_tags
 
+        data = []
+
         (self.class.split_searchable_fields ? @calculated_tags_list : get_merged_calculated_tags).each do |t|
           if !t[:search_tags].blank? && t[:search_tags].size > 1
-            begin
-              SmartSearchTag.create(t.merge!(:table_name => self.class.table_name, :entry_id => self.id, :search_tags => t[:search_tags].strip.split(" ").uniq.join(" ")))
-            rescue Exception => e
-
+            if options[:export]
+              data << t.merge!(:table_name => self.class.table_name, :entry_id => self.id, :search_tags => t[:search_tags].strip.split(" ").uniq.join(" "))
+            else
+              begin
+                SmartSearchTag.create(t.merge!(:table_name => self.class.table_name, :entry_id => self.id, :search_tags => t[:search_tags].strip.split(" ").uniq.join(" ")))
+              rescue Exception => e
+
+              end
             end
           end
         end
 
+        if options[:export]
+          return data
+        end
+
       rescue Exception => e
         Rails.logger.error "SMART SEARCH FAILED TO TO STORE SEARCH TAGS #{self.class.name} #{self.id}"
         Rails.logger.error e.message
@@ -268,9 +307,11 @@ class Config
 
     cattr_accessor  :search_models
     cattr_accessor  :public_models
+    cattr_accessor  :order_by_score
 
     self.search_models = []
     self.public_models = []
+    self.order_by_score = true
 
     def self.get_search_models
       self.search_models.map {|m| m.constantize}
@@ -279,6 +320,30 @@ def self.get_search_models
     def self.get_public_models
       self.public_models.map {|m| m.constantize}
     end
+
+    def self.rebuild_index
+      puts "Rebuilding Search index..."
+
+      SmartSearch::Config.search_models.each do |name|
+        puts "... #{name}"
+      end
+
+      ActiveRecord::Base.logger = nil
+
+      model_bar = ProgressBar.create(:title => "Building models", :total => SmartSearch::Config.search_models.size, format: "%t: (%c/%C) |%W| %f")
+
+      SmartSearch::Config.get_search_models.each do |model|
+        puts model.tags.join("\n")
+        entry_bar = ProgressBar.create(:title => model.name, :total => model.all.size, format: "%t: (%c/%C) |%W| %f")
+        model.all.each do |entry|
+          entry.create_search_tags
+          entry_bar.increment
+        end
+
+        model_bar.increment
+        puts "\n\n"
+      end
+    end
   end
 
 

lib/smart_search/engine.rb

@@ -2,10 +2,9 @@ module SmartSearch
   class Engine < Rails::Engine
 
     engine_name 'smart_search'
-    
+
     isolate_namespace SmartSearch
     require "friendly_extensions"
-    require "amatch"
 
   end
 end

lib/smart_similarity.rb

@@ -17,12 +17,6 @@ class SmartSimilarity < ActiveRecord::Base
       #== Konstanten
           # Defines the min. result of word simililarity check
           SIMILARITY_FACTOR = 0.81
-          # Defines first simililarity check method
-          SIMILARITY_METHOD_1 = :jarowinkler
-          # Defines first simililarity check method
-          SIMILARITY_METHOD_2 = :levenshtein
-
-          # An average of both results will generated and compered with 'SIMILARITY_FACTOR'
 
           # Limit Number of similar words
           SIMILARITY_LIMIT  = 5
@@ -100,8 +94,8 @@ def self.similars(word, options = {})
 
     # Return match score for two words bases und the two defined similarity methods
     def self.match_words(word1, word2)
-      x1 = word1.downcase.send("#{SIMILARITY_METHOD_1}_similar", word2.downcase)
-      x2 = word1.downcase.send("#{SIMILARITY_METHOD_2}_similar", word2.downcase)
+      x1 = DidYouMean::JaroWinkler.distance(word1.downcase, word2.downcase)
+      x2 = DidYouMean::Levenshtein.distance(word1.downcase, word2.downcase)
       return (x1+x2)/2.0
     end
 

lib/tasks/smart_search.rake

@@ -4,6 +4,11 @@ namespace :smart_search do
     require File.expand_path("../../smart_similarity", __FILE__)
     SmartSimilarity.load_from_query_history
   end
+
+  desc "Rebuild search index for all avialble models"
+  task :rebuild_index => :environment  do
+    SmartSearch::Config.rebuild_index
+  end
 end
 
 

smart_search.gemspec

@@ -1,6 +1,6 @@
 Gem::Specification.new do |s|
   s.name        = 'smart_search'
-  s.version     = '1.0.2'
+  s.version     = '1.2.2'
   s.summary     = "Simple, easy to use search MySQL based search for ActiveRecord"
   s.description = "SmartSearch adds full-text search functions to ActiveRecord running with MySQL, including search for similiar words. Its fast, simple, and works with almost zero-config!"
   s.authors     = ["Florian Eck"]
@@ -9,10 +9,6 @@ Gem::Specification.new do |s|
   s.test_files  =   Dir.glob("test/**/*")
   s.homepage    = 'https://github.com/florianeck/smart_search'
 
-  s.add_dependency "rails", ">= 4.0.4"
-  s.add_dependency "amatch"
   s.add_dependency "friendly_extensions"
   s.add_dependency "unicode-emoji"
-  s.add_dependency "ruby-progressbar"
-  s.add_dependency  "parallel"
 end