jakal 0.1.92 → 0.1.93

data/lib/jkl/text_client.rb

@@ -2,8 +2,8 @@ module Jkl
   module Text
     class << self
 
-      def sanitize(text)
-        remove_short_lines(strip_all_tags(remove_script_tags(text)))
+      def sanitize(text, number = 5)
+        remove_short_lines(strip_all_tags(remove_script_tags(text)), number)
       end
       alias :clean :sanitize
 
@@ -24,12 +24,12 @@ module Jkl
         text.gsub(/((<[\s\/]*script\b[^>]*>)([^>]*)(<\/script>))/i, "")
       end
 
-      def remove_short_lines(text)
+      def remove_short_lines(text, number = 5)
         text = text.gsub(/\s\s/, "\n")
         str = ""
         # remove short lines - ususally just navigation
         text.split("\n").each do |l|
-          str << l unless l.count(" ") < 5
+          str << l unless l.count(" ") < number
         end
         str
       end

data/test/unit/text_cleaning_test.rb

@@ -16,6 +16,16 @@ HTML
       assert result == "the cat sat on the mat"
     end
     
+    should "Remove shorter lines" do
+      input = <<-HTML
+the cat sat on the mat
+the cat sat on the slightly fluffy, yet worn and homely mat
+a short line
+HTML
+      result = Jkl::Text::remove_short_lines(input, 8)
+      assert result == "the cat sat on the slightly fluffy, yet worn and homely mat"
+    end
+    
     should "Remove script tags" do
       input = <<-HTML
 the cat sat on the mat

metadata

@@ -5,8 +5,8 @@ version: !ruby/object:Gem::Version
   segments: 
   - 0
   - 1
-  - 92
-  version: 0.1.92
+  - 93
+  version: 0.1.93
 platform: ruby
 authors: 
 - sshingler