Merge pull request #498 from leo-souza/master

Use database's lower function for case-insensitive match

CHANGELOG.md

@@ -25,7 +25,7 @@ As such, a _Feature_ would map to either major or minor. A _bug fix_ to a patch.
   * [@rgould #417 Let '.count' work when tagged_with is accompanied by a group clause](https://github.com/mbleigh/acts-as-taggable-on/pull/417)
   * [@developer88 #461 Move 'Distinct' out of select string and use .uniq instead](https://github.com/mbleigh/acts-as-taggable-on/pull/461)
   * [@gerard-leijdekkers #473 Fixed down migration index name](https://github.com/mbleigh/acts-as-taggable-on/pull/473)
-  * [@leo-souza #492 Fix downcasing of unicode tag names](https://github.com/mbleigh/acts-as-taggable-on/pull/492)
+  * [@leo-souza #498 Use database's lower function for case-insensitive match](https://github.com/mbleigh/acts-as-taggable-on/pull/498)
 * Misc
   * [@billychan #463 Thread safe support](https://github.com/mbleigh/acts-as-taggable-on/pull/463)
   * [@billychan #386 Add parse:true instructions to README](https://github.com/mbleigh/acts-as-taggable-on/pull/386)

README.md

@@ -340,6 +340,8 @@ If you want to change the default delimiter (it defaults to ','). You can also p
 ActsAsTaggableOn.delimiter = ','
 ```
 
+*NOTE: SQLite by default can't upcase or downcase multibyte characters, resulting in unwanted behavior. Load the SQLite ICU extension for proper handle of such characters. [See docs](http://www.sqlite.org/src/artifact?ci=trunk&filename=ext/icu/README.txt)*
+
 ## Contributors
 
 We have a long list of valued contributors. [Check them all](https://github.com/mbleigh/acts-as-taggable-on/contributors)

lib/acts_as_taggable_on/tag.rb

@@ -24,9 +24,9 @@ def validates_name_uniqueness?
 
     def self.named(name)
       if ActsAsTaggableOn.strict_case_match
-        where(["name = #{binary}?", name])
+        where(["name = #{binary}?", as_8bit_ascii(name)])
       else
-        where(["lower(name) = ?", name.downcase])
+        where(["LOWER(name) = LOWER(?)", as_8bit_ascii(unicode_downcase(name))])
       end
     end
 
@@ -38,8 +38,7 @@ def self.named_any(list)
         where(clause)
       else
         clause = list.map { |tag|
-          lowercase_ascii_tag = as_8bit_ascii(tag, true)
-          sanitize_sql(["lower(name) = ?", lowercase_ascii_tag])
+          sanitize_sql(["LOWER(name) = LOWER(?)", as_8bit_ascii(unicode_downcase(tag))])
         }.join(" OR ")
         where(clause)
       end
@@ -101,23 +100,29 @@ class << self
 
       def comparable_name(str)
         if ActsAsTaggableOn.strict_case_match
-          as_8bit_ascii(str)
+          str
         else
-          as_8bit_ascii(str, true)
+          unicode_downcase(str.to_s)
         end
       end
 
       def binary
         using_mysql? ? "BINARY " : nil
       end
 
-      def as_8bit_ascii(string, downcase=false)
-        string = string.to_s.dup.mb_chars
-        string.downcase! if downcase
+      def unicode_downcase(string)
+        if ActiveSupport::Multibyte::Unicode.respond_to?(:downcase)
+          ActiveSupport::Multibyte::Unicode.downcase(string)
+        else
+          ActiveSupport::Multibyte::Chars.new(string).downcase.to_s  
+        end
+      end
+
+      def as_8bit_ascii(string)
         if defined?(Encoding)
-          string.to_s.force_encoding('BINARY')
+          string.to_s.dup.force_encoding('BINARY')
         else
-          string.to_s
+          string.to_s.mb_chars
         end
       end
     end

spec/acts_as_taggable_on/taggable_spec.rb

@@ -246,6 +246,7 @@
   unless ActsAsTaggableOn::Tag.using_sqlite?
     it "should not care about case for unicode names" do
       ActsAsTaggableOn.strict_case_match = false
+
       anya = TaggableModel.create(:name => "Anya", :tag_list => "ПРИВЕТ")
       igor = TaggableModel.create(:name => "Igor", :tag_list => "привет")
       katia = TaggableModel.create(:name => "Katia", :tag_list => "ПРИВЕТ")
@@ -255,6 +256,21 @@
     end
   end
 
+  it "should be able to create and find tags in languages without capitalization" do
+    ActsAsTaggableOn.strict_case_match = false
+    chihiro = TaggableModel.create(:name => "Chihiro", :tag_list => "日本の")
+    salim = TaggableModel.create(:name => "Salim", :tag_list => "עברית")
+    ieie = TaggableModel.create(:name => "Ieie", :tag_list => "中国的")
+    yasser = TaggableModel.create(:name => "Yasser", :tag_list => "العربية")
+    emo = TaggableModel.create(:name => "Emo", :tag_list => "✏")
+
+    TaggableModel.tagged_with("日本の").to_a.size.should == 1
+    TaggableModel.tagged_with("עברית").to_a.size.should == 1
+    TaggableModel.tagged_with("中国的").to_a.size.should == 1
+    TaggableModel.tagged_with("العربية").to_a.size.should == 1
+    TaggableModel.tagged_with("✏").to_a.size.should == 1
+  end 
+
   it "should be able to get tag counts on model as a whole" do
     bob = TaggableModel.create(:name => "Bob", :tag_list => "ruby, rails, css")
     frank = TaggableModel.create(:name => "Frank", :tag_list => "ruby, rails")