Commit a25a85b9396cbc07bf2bfcd12d3c0a6685df66d6
1 parent
6e5caa19
Exists in
master
and in
4 other branches
merge charlock_holmes to master
Showing
3 changed files
with
9 additions
and
11 deletions
Show diff stats
Gemfile
| @@ -22,9 +22,9 @@ gem "acts_as_list" | @@ -22,9 +22,9 @@ gem "acts_as_list" | ||
| 22 | gem "rdiscount" | 22 | gem "rdiscount" |
| 23 | gem "acts-as-taggable-on", "~> 2.1.0" | 23 | gem "acts-as-taggable-on", "~> 2.1.0" |
| 24 | gem "drapper" | 24 | gem "drapper" |
| 25 | -gem "rchardet19", "~> 1.3.5" | ||
| 26 | gem "resque" | 25 | gem "resque" |
| 27 | gem "httparty" | 26 | gem "httparty" |
| 27 | +gem "charlock_holmes" | ||
| 28 | 28 | ||
| 29 | group :assets do | 29 | group :assets do |
| 30 | gem "sass-rails", "~> 3.1.0" | 30 | gem "sass-rails", "~> 3.1.0" |
Gemfile.lock
| @@ -77,6 +77,7 @@ GEM | @@ -77,6 +77,7 @@ GEM | ||
| 77 | xpath (~> 0.1.4) | 77 | xpath (~> 0.1.4) |
| 78 | carrierwave (0.5.8) | 78 | carrierwave (0.5.8) |
| 79 | activesupport (~> 3.0) | 79 | activesupport (~> 3.0) |
| 80 | + charlock_holmes (0.6.8) | ||
| 80 | childprocess (0.2.2) | 81 | childprocess (0.2.2) |
| 81 | ffi (~> 1.0.6) | 82 | ffi (~> 1.0.6) |
| 82 | coffee-rails (3.1.1) | 83 | coffee-rails (3.1.1) |
| @@ -172,7 +173,6 @@ GEM | @@ -172,7 +173,6 @@ GEM | ||
| 172 | rdoc (~> 3.4) | 173 | rdoc (~> 3.4) |
| 173 | thor (~> 0.14.6) | 174 | thor (~> 0.14.6) |
| 174 | rake (0.9.2.2) | 175 | rake (0.9.2.2) |
| 175 | - rchardet19 (1.3.5) | ||
| 176 | rdiscount (1.6.8) | 176 | rdiscount (1.6.8) |
| 177 | rdoc (3.11) | 177 | rdoc (3.11) |
| 178 | json (~> 1.4) | 178 | json (~> 1.4) |
| @@ -285,6 +285,7 @@ DEPENDENCIES | @@ -285,6 +285,7 @@ DEPENDENCIES | ||
| 285 | awesome_print | 285 | awesome_print |
| 286 | capybara | 286 | capybara |
| 287 | carrierwave | 287 | carrierwave |
| 288 | + charlock_holmes | ||
| 288 | coffee-rails (~> 3.1.0) | 289 | coffee-rails (~> 3.1.0) |
| 289 | database_cleaner | 290 | database_cleaner |
| 290 | devise (= 1.5.0) | 291 | devise (= 1.5.0) |
| @@ -302,7 +303,6 @@ DEPENDENCIES | @@ -302,7 +303,6 @@ DEPENDENCIES | ||
| 302 | pygments.rb (= 0.2.3) | 303 | pygments.rb (= 0.2.3) |
| 303 | rails (= 3.1.1) | 304 | rails (= 3.1.1) |
| 304 | rails-footnotes (~> 3.7.5) | 305 | rails-footnotes (~> 3.7.5) |
| 305 | - rchardet19 (~> 1.3.5) | ||
| 306 | rdiscount | 306 | rdiscount |
| 307 | resque | 307 | resque |
| 308 | rspec-rails | 308 | rspec-rails |
lib/utils.rb
| @@ -17,15 +17,13 @@ module Utils | @@ -17,15 +17,13 @@ module Utils | ||
| 17 | end | 17 | end |
| 18 | 18 | ||
| 19 | module CharEncode | 19 | module CharEncode |
| 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) | 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] | ||
| 25 | end | 25 | end |
| 26 | - string.encode("utf-8", :undef => :replace, :replace => "?", :invalid => :replace) | ||
| 27 | - rescue | ||
| 28 | - "Invalid Encoding" | 26 | + content |
| 29 | end | 27 | end |
| 30 | end | 28 | end |
| 31 | 29 |