merge charlock_holmes to master

diff --git a/Gemfile b/Gemfile
index 3c7b1a4..e240fa6 100644 (file)
--- a/Gemfile
+++ b/Gemfile
@@ -22,9 +22,9 @@ gem "acts_as_list"
 gem "rdiscount"
 gem "acts-as-taggable-on", "~> 2.1.0"
 gem "drapper"
-gem "rchardet19", "~> 1.3.5"
 gem "resque"
 gem "httparty"
+gem "charlock_holmes"
 
 group :assets do
   gem "sass-rails",   "~> 3.1.0"

diff --git a/Gemfile.lock b/Gemfile.lock
index 09fecb8..86a6998 100644 (file)
--- a/Gemfile.lock
+++ b/Gemfile.lock
@@ -77,6 +77,7 @@ GEM
       xpath (~> 0.1.4)
     carrierwave (0.5.8)
       activesupport (~> 3.0)
+    charlock_holmes (0.6.8)
     childprocess (0.2.2)
       ffi (~> 1.0.6)
     coffee-rails (3.1.1)
@@ -172,7 +173,6 @@ GEM
       rdoc (~> 3.4)
       thor (~> 0.14.6)
     rake (0.9.2.2)
-    rchardet19 (1.3.5)
     rdiscount (1.6.8)
     rdoc (3.11)
       json (~> 1.4)
@@ -285,6 +285,7 @@ DEPENDENCIES
   awesome_print
   capybara
   carrierwave
+  charlock_holmes
   coffee-rails (~> 3.1.0)
   database_cleaner
   devise (= 1.5.0)
@@ -302,7 +303,6 @@ DEPENDENCIES
   pygments.rb (= 0.2.3)
   rails (= 3.1.1)
   rails-footnotes (~> 3.7.5)
-  rchardet19 (~> 1.3.5)
   rdiscount
   resque
   rspec-rails

diff --git a/lib/utils.rb b/lib/utils.rb
index 2de1476..8e5d469 100644 (file)
--- a/lib/utils.rb
+++ b/lib/utils.rb
@@ -17,15 +17,13 @@ module Utils
   end
 
   module CharEncode
-    def encode(string)
-      return '' unless string
-      cd = CharDet.detect(string)
-      if cd.confidence > 0.6
-        string.force_encoding(cd.encoding)
+    def encode(content)
+      content ||= ''
+      detection = CharlockHolmes::EncodingDetector.detect(content)
+      if hash = detection
+        content = CharlockHolmes::Converter.convert(content, hash[:encoding], 'UTF-8') if hash[:encoding]
       end
-      string.encode("utf-8", :undef => :replace, :replace => "?", :invalid => :replace)
-    rescue
-      "Invalid Encoding"
+      content
     end
   end