charles 0.0.5 → 0.0.6

data/bin/charles

@@ -16,7 +16,7 @@ puts({
   :content => document.content,
   :interesting_content => document.interesting_content,
   :title => document.title,
-  :filtered_images => document.filtered_images.collect{|image| image[:url]}
+  :filtered_images => document.filtered_images
 }.to_yaml)
 
 

data/lib/charles/images.rb

@@ -43,6 +43,9 @@ module Charles
     end
     
     def filtered_images
+      filtered_images_extra.collect{|image| image[:url]}
+    end
+    def filtered_images_extra
       _max_proportion = 2.5
       _min_area = 88*88
       _filtered_images = []

data/lib/charles/internal_attributes.rb

@@ -33,8 +33,8 @@ module Charles
           _title_words[token]||=0; _title_words[token]+=1
         }
       }
-      _threshold = (0.9 * @options[:sample_titles].size).ceil
-      _words_to_filter = _title_words.select{|k,v| v >= _threshold}.collect{|k,v| k} #select words used in more than 90% of the titles
+      _threshold = (0.8 * @options[:sample_titles].size).ceil
+      _words_to_filter = _title_words.select{|k,v| v >= _threshold}.collect{|k,v| k} #select words used in more than 80% of the titles
     end
   end
 end

data/lib/charles/version.rb

@@ -1,3 +1,3 @@
 module Charles
-  VERSION = "0.0.5"
+  VERSION = "0.0.6"
 end

data/test/test_charles.rb

@@ -60,10 +60,11 @@ class CharlesTest < Test::Unit::TestCase
     article = TEST_ARTICLES.detect{|article| article[:url] == 'http://online.wsj.com/article/SB10001424052702303674004577433160886451978.html'}
     input = File.read("test/articles/#{article[:file]}.html")
     document = Charles::Document.new(input, :url => article[:url])
-    assert document.filtered_images.size > 3
-    assert document.filtered_images.last[:data].size > 1000
-    assert document.filtered_images.last[:width] > 100
-    assert document.filtered_images.last[:height] > 100
+    filtered_images_extra = document.filtered_images_extra
+    assert filtered_images_extra.size > 3
+    assert filtered_images_extra.last[:data].size > 1000
+    assert filtered_images_extra.last[:width] > 100
+    assert filtered_images_extra.last[:height] > 100
   end
   
   

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: charles
 version: !ruby/object:Gem::Version 
-  hash: 21
+  hash: 19
   prerelease: 
   segments: 
   - 0
   - 0
-  - 5
-  version: 0.0.5
+  - 6
+  version: 0.0.6
 platform: ruby
 authors: 
 - Jason Ling Xiaowei