Revert "merge charlock_holmes to master"

This reverts commit a25a85b9396cbc07bf2bfcd12d3c0a6685df66d6.

Revert "merge charlock_holmes to master"
This reverts commit a25a85b9396cbc07bf2bfcd12d3c0a6685df66d6.
Saito
1 parent 5719f02f
Showing 3 changed files with 11 additions and 9 deletions Show diff stats
Gemfile
Gemfile.lock
lib/utils.rb
@@ -22,9 +22,9 @@ gem &quot;acts_as_list&quot;
 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"
@@ -77,7 +77,6 @@ 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)
@@ -173,6 +172,7 @@ 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)
@@ -285,7 +285,6 @@ DEPENDENCIES
   awesome_print
   capybara
   carrierwave
-  charlock_holmes
   coffee-rails (~> 3.1.0)
   database_cleaner
   devise (= 1.5.0)
@@ -303,6 +302,7 @@ DEPENDENCIES
   pygments.rb (= 0.2.3)
   rails (= 3.1.1)
   rails-footnotes (~> 3.7.5)
+  rchardet19 (~> 1.3.5)
   rdiscount
   resque
   rspec-rails
@@ -17,13 +17,15 @@ module Utils
   end
  
   module CharEncode
-    def encode(content)
-      content ||= ''
-      detection = CharlockHolmes::EncodingDetector.detect(content)
-      if hash = detection
-        content = CharlockHolmes::Converter.convert(content, hash[:encoding], 'UTF-8') if hash[:encoding]
+    def encode(string)
+      return '' unless string
+      cd = CharDet.detect(string)
+      if cd.confidence > 0.6
+        string.force_encoding(cd.encoding)
       end
-      content
+      string.encode("utf-8", :undef => :replace, :replace => "?", :invalid => :replace)
+    rescue
+      "Invalid Encoding"
     end
   end
...	...	@@ -22,9 +22,9 @@ gem "acts_as_list"
22	22	gem "rdiscount"
23	23	gem "acts-as-taggable-on", "~> 2.1.0"
24	24	gem "drapper"
	25	+gem "rchardet19", "~> 1.3.5"
25	26	gem "resque"
26	27	gem "httparty"
27		-gem "charlock_holmes"
28	28
29	29	group :assets do
30	30	gem "sass-rails", "~> 3.1.0"
...	...
...	...	@@ -77,7 +77,6 @@ GEM
77	77	xpath (~> 0.1.4)
78	78	carrierwave (0.5.8)
79	79	activesupport (~> 3.0)
80		- charlock_holmes (0.6.8)
81	80	childprocess (0.2.2)
82	81	ffi (~> 1.0.6)
83	82	coffee-rails (3.1.1)
...	...	@@ -173,6 +172,7 @@ GEM
173	172	rdoc (~> 3.4)
174	173	thor (~> 0.14.6)
175	174	rake (0.9.2.2)
	175	+ rchardet19 (1.3.5)
176	176	rdiscount (1.6.8)
177	177	rdoc (3.11)
178	178	json (~> 1.4)
...	...	@@ -285,7 +285,6 @@ DEPENDENCIES
285	285	awesome_print
286	286	capybara
287	287	carrierwave
288		- charlock_holmes
289	288	coffee-rails (~> 3.1.0)
290	289	database_cleaner
291	290	devise (= 1.5.0)
...	...	@@ -303,6 +302,7 @@ DEPENDENCIES
303	302	pygments.rb (= 0.2.3)
304	303	rails (= 3.1.1)
305	304	rails-footnotes (~> 3.7.5)
	305	+ rchardet19 (~> 1.3.5)
306	306	rdiscount
307	307	resque
308	308	rspec-rails
...	...
...	...	@@ -17,13 +17,15 @@ module Utils
17	17	end
18	18
19	19	module CharEncode
20		- def encode(content)
21		- content \|\|= ''
22		- detection = CharlockHolmes::EncodingDetector.detect(content)
23		- if hash = detection
24		- content = CharlockHolmes::Converter.convert(content, hash[:encoding], 'UTF-8') if hash[:encoding]
	20	+ def encode(string)
	21	+ return '' unless string
	22	+ cd = CharDet.detect(string)
	23	+ if cd.confidence > 0.6
	24	+ string.force_encoding(cd.encoding)
25	25	end
26		- content
	26	+ string.encode("utf-8", :undef => :replace, :replace => "?", :invalid => :replace)
	27	+ rescue
	28	+ "Invalid Encoding"
27	29	end
28	30	end
29	31
...	...