texticle 1.0.3 → 1.0.4.20101004123327

data/CHANGELOG.rdoc

@@ -1,3 +1,16 @@
+=== 1.0.4 / 2010-08-19
+
+* 2 major enhancements
+
+  * use Rails.root instead of RAILS_ROOT
+  * refactored tasks to ease maintainance and patchability
+
+* 3 minor enhancements
+
+  * fix timestamp for migrationfile
+  * fixed deprecation warning for rails3 (dropping rails2-support)
+  * prevented warning about defined constant
+
 === 1.0.3 / 2010-07-07
 
 * 1 major enhancement

data/README.rdoc

@@ -18,7 +18,7 @@ named_scope methods making searching easy and fun!
 
 In the project's Gemfile add
 
-    gem 'texticle', '1.0.2', :git => 'git://github.com/tenderlove/texticle'
+    gem 'texticle', '1.0.4'
 
 === Rails 2 Configuration
 
@@ -54,6 +54,29 @@ Alternatively, you can create a migration that will build your indexes:
 
   $ rake texticle:migration
 
+== BETA: FUZZY SEARCH
+
+We're adding support for Postgres' optional trigram matching module, which allows for fuzzier text searches.
+
+Since it's an optional module, first step is to install it into the current database:
+	
+	$ rake texticle:install_trigram
+
+If you installed Postgres with its optional 'contrib' modules, you should be safe.
+
+Declare your indexes as you normally would:
+
+	class Book < ActiveRecord::Base
+		index { title }
+	end
+	
+Finally, say hello to #tsearch:
+
+	Book.create :title => "_why's Poignant Guide to Ruby"
+	Book.tsearch "_whys Poignnt Gd t Rb" # => [#<Book title: "_why's Poignant Guide to Ruby">]
+	
+Trigram searching is still in beta, so expect the API to change and/or improve.
+
 == REQUIREMENTS:
 
 * Texticle may be used outside rails, but works best with rails.

data/Rakefile

@@ -3,6 +3,7 @@
 require 'rubygems'
 require 'hoe'
 
+Hoe.plugin :gemspec
 Hoe.spec 'texticle' do
   developer('Aaron Patterson', 'aaronp@rubyforge.org')
   self.readme_file   = 'README.rdoc'

data/lib/texticle.rb

@@ -49,7 +49,7 @@ require 'texticle/railtie' if defined?(Rails) and Rails::VERSION::MAJOR > 2
 #   end
 module Texticle
   # The version of Texticle you are using.
-  VERSION = '1.0.3'
+  VERSION = '1.0.4' unless defined?(Texticle::VERSION)
 
   # A list of full text indexes
   attr_accessor :full_text_indexes
@@ -59,19 +59,53 @@ module Texticle
   def index name = nil, dictionary = 'english', &block
     search_name = ['search', name].compact.join('_')
 
-    class_eval do
-      named_scope search_name.to_sym, lambda { |term|
-        # Let's extract the individual terms to allow for quoted terms.
-        term = term.scan(/"([^"]+)"|(\S+)/).flatten.compact.map {|lex| "'#{lex}'"}.join(' & ')
-        {
-          :select => "#{table_name}.*, ts_rank_cd((#{full_text_indexes.first.to_s}),
-            to_tsquery(#{connection.quote(term)})) as rank",
-          :conditions =>
-            ["#{full_text_indexes.first.to_s} @@ to_tsquery(?)", term],
-          :order => 'rank DESC'
-        }
+    scope_lamba = lambda { |term|
+      # Let's extract the individual terms to allow for quoted and wildcard terms.
+      term = term.scan(/"([^"]+)"|(\S+)/).flatten.compact.map do |lex|
+        lex =~ /(.+)\*\s*$/ ? "'#{$1}':*" : "'#{lex}'"
+      end.join(' & ')
+
+      {
+        :select => "#{table_name}.*, ts_rank_cd((#{full_text_indexes.first.to_s}),
+          to_tsquery(#{connection.quote(term)})) as rank",
+        :conditions =>
+          ["#{full_text_indexes.first.to_s} @@ to_tsquery(?)", term],
+        :order => 'rank DESC'
+      }
+    }
+    
+    # tsearch, i.e. trigram search
+    trigram_scope_lambda = lambda { |term|
+      term = "'#{term.gsub("'", "''")}'"
+
+      similarities = full_text_indexes.first.index_columns.values.flatten.inject([]) do |array, index|
+        array << "similarity(#{index}, #{term})"
+      end.join(" + ")
+      
+      conditions = full_text_indexes.first.index_columns.values.flatten.inject([]) do |array, index|
+        array << "(#{index} % #{term})"
+      end.join(" OR ")
+      
+      {
+        :select => "#{table_name}.*, #{similarities} as rank",
+        :conditions => conditions,
+        :order => 'rank DESC'
       }
+    }
+
+    class_eval do
+      # Trying to avoid the deprecation warning when using :named_scope
+      # that Rails 3 emits. Can't use #respond_to?(:scope) since scope
+      # is a protected method in Rails 2, and thus still returns true.
+      if self.respond_to?(:scope) and not protected_methods.include?('scope')
+        scope search_name.to_sym, scope_lamba
+        scope ('t' + search_name).to_sym, trigram_scope_lambda
+      elsif self.respond_to? :named_scope
+        named_scope search_name.to_sym, scope_lamba
+        named_scope ('t' + search_name).to_sym, trigram_scope_lambda
+      end
     end
+
     index_name = [table_name, name, 'fts_idx'].compact.join('_')
     (self.full_text_indexes ||= []) <<
       FullTextIndex.new(index_name, dictionary, self, &block)

data/lib/texticle/full_text_index.rb

@@ -11,6 +11,10 @@ module Texticle
       instance_eval(&block)
     end
 
+    def self.find_constant_of(filename)
+      File.basename(filename, '.rb').pluralize.classify.constantize
+    end
+
     def create
       @model_class.connection.execute create_sql
     end
@@ -20,10 +24,10 @@ module Texticle
     end
 
     def create_sql
-      <<-eosql
-        CREATE index #{@name}
-        ON #{@model_class.table_name}
-        USING gin((#{to_s}))
+      <<-eosql.chomp
+CREATE index #{@name}
+      ON #{@model_class.table_name}
+      USING gin((#{to_s}))
       eosql
     end
 
@@ -35,7 +39,7 @@ module Texticle
       return @string if @string
       vectors = []
       @index_columns.sort_by { |k,v| k }.each do |weight, columns|
-        c = columns.map { |x| "coalesce(#{@model_class.table_name}.#{x}, '')" }
+        c = columns.map { |x| "coalesce(\"#{@model_class.table_name}\".\"#{x}\", '')" }
         if weight == 'none'
           vectors << "to_tsvector('#{@dictionary}', #{c.join(" || ' ' || ")})"
         else

data/lib/texticle/tasks.rb

@@ -5,34 +5,39 @@ namespace :texticle do
   desc "Create full text index migration"
   task :migration => :environment do
     now = Time.now.utc
-    filename = "#{now.strftime('%Y%m%d%H%m%S')}_full_text_search_#{now.to_i}.rb"
-    File.open(File.join(RAILS_ROOT, 'db', 'migrate', filename), 'wb') { |fh|
-      fh.puts "class FullTextSearch#{now.to_i} < ActiveRecord::Migration"
-      fh.puts "  def self.up"
-      Dir[File.join(RAILS_ROOT, 'app', 'models', '*.rb')].each do |f|
-        klass = File.basename(f, '.rb').pluralize.classify.constantize
+    filename = "#{now.strftime('%Y%m%d%H%M%S')}_full_text_search_#{now.to_i}.rb"
+    File.open(Rails.root + 'db' + 'migrate' + filename, 'wb') do |fh|
+      up_sql_statements = []
+      dn_sql_statements = []
+
+      Dir[Rails.root + 'app' + 'models' + '*.rb'].each do |f|
+        klass = Texticle::FullTextIndex.find_constant_of(f)
         if klass.respond_to?(:full_text_indexes)
           (klass.full_text_indexes || []).each do |fti|
-            fh.puts <<-eostmt
-      ActiveRecord::Base.connection.execute(<<-'eosql')
-        #{fti.destroy_sql}
-      eosql
-      ActiveRecord::Base.connection.execute(<<-'eosql')
-        #{fti.create_sql}
-      eosql
-            eostmt
+            up_sql_statements << fti.destroy_sql
+            up_sql_statements << fti.create_sql
+            dn_sql_statements << fti.destroy_sql
           end
         end
       end
+
+      fh.puts "class FullTextSearch#{now.to_i} < ActiveRecord::Migration"
+      fh.puts "  def self.up"
+      insert_sql_statements_into_migration_file(up_sql_statements, fh)
+      fh.puts "  end"
+      fh.puts ""
+
+      fh.puts "  def self.down"
+      insert_sql_statements_into_migration_file(dn_sql_statements, fh)
       fh.puts "  end"
       fh.puts "end"
-    }
+    end
   end
 
   desc "Create full text indexes"
   task :create_indexes => ['texticle:destroy_indexes'] do
-    Dir[File.join(RAILS_ROOT, 'app', 'models', '*.rb')].each do |f|
-      klass = File.basename(f, '.rb').pluralize.classify.constantize
+    Dir[Rails.root + 'app' + 'models' + '*.rb'].each do |f|
+      klass = Texticle::FullTextIndex.find_constant_of(f)
       if klass.respond_to?(:full_text_indexes)
         (klass.full_text_indexes || []).each do |fti|
           begin
@@ -47,8 +52,8 @@ namespace :texticle do
 
   desc "Destroy full text indexes"
   task :destroy_indexes => [:environment] do
-    Dir[File.join(RAILS_ROOT, 'app', 'models', '*.rb')].each do |f|
-      klass = File.basename(f, '.rb').pluralize.classify.constantize
+    Dir[Rails.root + 'app' + 'models' + '*.rb'].each do |f|
+      klass = Texticle::FullTextIndex.find_constant_of(f)
       if klass.respond_to?(:full_text_indexes)
         (klass.full_text_indexes || []).each do |fti|
           fti.destroy
@@ -56,4 +61,30 @@ namespace :texticle do
       end
     end
   end
+
+  desc "Install trigram text search module"
+  task :install_trigram => [:environment] do
+    share_dir = `pg_config --sharedir`.chomp
+    raise RuntimeError, 'cannot find Postgres\' shared directory' unless $?.exitstatus.zero?
+    trigram = "#{share_dir}/contrib/pg_trgm.sql"
+    unless system("ls #{trigram}")
+      raise RuntimeError, 'cannot find trigram module; was it compiled and installed?'
+    end
+    db_name = ActiveRecord::Base.connection.current_database
+    unless system("psql -d #{db_name} -f #{trigram}")
+      raise RuntimeError, "`psql -d #{db_name} -f #{trigram}` cannot complete successfully"
+    end
+    puts "Trigram text search module successfully installed into '#{db_name}' database."
+  end
+
+  def insert_sql_statements_into_migration_file statements, fh
+    statements.each do |statement|
+      fh.puts <<-eostmt
+    execute(<<-'eosql'.strip)
+      #{statement}
+    eosql
+      eostmt
+    end
+  end
 end
+

data/test/helper.rb

@@ -33,6 +33,7 @@ class TexticleTestCase < Test::Unit::TestCase
         def named_scope *args
           @named_scopes << args
         end
+        alias :scope :named_scope
 
         def quote thing
           "'#{thing}'"

data/test/test_full_text_index.rb

@@ -45,14 +45,14 @@ class TestFullTextIndex < TexticleTestCase
     fti = Texticle::FullTextIndex.new('ft_index', 'english', @fm) do
       name
     end
-    assert_equal "to_tsvector('english', coalesce(#{@fm.table_name}.name, ''))", fti.to_s
+    assert_equal "to_tsvector('english', coalesce(\"#{@fm.table_name}\".\"name\", ''))", fti.to_s
   end
 
   def test_to_s_A_weight
     fti = Texticle::FullTextIndex.new('ft_index', 'english', @fm) do
       name 'A'
     end
-    assert_equal "setweight(to_tsvector('english', coalesce(#{@fm.table_name}.name, '')), 'A')", fti.to_s
+    assert_equal "setweight(to_tsvector('english', coalesce(\"#{@fm.table_name}\".\"name\", '')), 'A')", fti.to_s
   end
 
   def test_to_s_multi_weight
@@ -61,7 +61,7 @@ class TestFullTextIndex < TexticleTestCase
       value 'A'
       description 'B'
     end
-    assert_equal "setweight(to_tsvector('english', coalesce(#{@fm.table_name}.name, '') || ' ' || coalesce(#{@fm.table_name}.value, '')), 'A') || ' ' || setweight(to_tsvector('english', coalesce(#{@fm.table_name}.description, '')), 'B')", fti.to_s
+    assert_equal "setweight(to_tsvector('english', coalesce(\"#{@fm.table_name}\".\"name\", '') || ' ' || coalesce(\"#{@fm.table_name}\".\"value\", '')), 'A') || ' ' || setweight(to_tsvector('english', coalesce(\"#{@fm.table_name}\".\"description\", '')), 'B')", fti.to_s
   end
 
   def test_mixed_weight
@@ -69,6 +69,6 @@ class TestFullTextIndex < TexticleTestCase
       name
       value 'A'
     end
-    assert_equal "setweight(to_tsvector('english', coalesce(#{@fm.table_name}.value, '')), 'A') || ' ' || to_tsvector('english', coalesce(#{@fm.table_name}.name, ''))", fti.to_s
+    assert_equal "setweight(to_tsvector('english', coalesce(\"#{@fm.table_name}\".\"value\", '')), 'A') || ' ' || to_tsvector('english', coalesce(\"#{@fm.table_name}\".\"name\", ''))", fti.to_s
   end
 end

data/test/test_texticle.rb

@@ -10,11 +10,13 @@ class TestTexticle < TexticleTestCase
       end
     end
     assert_equal 1, x.full_text_indexes.length
-    assert_equal 1, x.named_scopes.length
+    # One named_scope for search, another for trigram search
+    assert_equal 2, x.named_scopes.length
 
     x.full_text_indexes.first.create
     assert_match "#{x.table_name}_fts_idx", x.executed.first
     assert_equal :search, x.named_scopes.first.first
+    assert_equal :tsearch, x.named_scopes[1].first
   end
 
   def test_named_index
@@ -26,11 +28,12 @@ class TestTexticle < TexticleTestCase
       end
     end
     assert_equal 1, x.full_text_indexes.length
-    assert_equal 1, x.named_scopes.length
+    assert_equal 2, x.named_scopes.length
 
     x.full_text_indexes.first.create
     assert_match "#{x.table_name}_awesome_fts_idx", x.executed.first
     assert_equal :search_awesome, x.named_scopes.first.first
+    assert_equal :tsearch_awesome, x.named_scopes[1].first
   end
 
   def test_named_scope_select
@@ -57,4 +60,17 @@ class TestTexticle < TexticleTestCase
     ns = x.named_scopes.first[1].call('foo bar "foo bar"')
     assert_match(/'foo' & 'bar' & 'foo bar'/, ns[:select])
   end
+  
+  def test_wildcard_queries
+    x = fake_model
+    x.class_eval do
+      extend Texticle
+      index('awesome') do
+        name
+      end
+    end
+    
+    ns = x.named_scopes.first[1].call('foo bar*')
+    assert_match(/'foo' & 'bar:*'/, ns[:select])
+  end
 end

metadata

@@ -1,13 +1,14 @@
 --- !ruby/object:Gem::Specification 
 name: texticle
 version: !ruby/object:Gem::Version 
-  hash: 17
+  hash: 40202008246577
   prerelease: false
   segments: 
   - 1
   - 0
-  - 3
-  version: 1.0.3
+  - 4
+  - 20101004123327
+  version: 1.0.4.20101004123327
 platform: ruby
 authors: 
 - Aaron Patterson
@@ -15,7 +16,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-07-08 00:00:00 -07:00
+date: 2010-10-04 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -42,12 +43,12 @@ dependencies:
     requirements: 
     - - ">="
       - !ruby/object:Gem::Version 
-        hash: 23
+        hash: 19
         segments: 
         - 2
         - 6
-        - 0
-        version: 2.6.0
+        - 2
+        version: 2.6.2
   type: :development
   version_requirements: *id002
 description: |-