Fix a big with html entities

Author: nono

lib/html_truncator.rb

@@ -72,18 +72,25 @@ def self_closing?
 
 class Nokogiri::XML::Text
   def truncate(max, opts)
-    words = to_xhtml.scan(/\s*\S+/)
     if opts[:length_in_chars]
-      count = to_xhtml.length
+      count = content.length
       return [to_xhtml, count, opts] if count <= max && max > 0
+      words = content.scan(/\s*\S+/)
       if words.size > 1
         words.inject('') do |string, word|
-          return [string, count, opts] if string.length + word.length > max
+          if string.length + word.length > max
+            txt = dup
+            txt.content = string
+            return [txt.to_xhtml, count, opts]
+          end
           string + word
         end
       end
-      [content.slice(0, max), count, opts]
+      txt = dup
+      txt.content = content.slice(0, max)
+      [txt.to_xhtml, count, opts]
     else
+      words = to_xhtml.scan(/\s*\S+/)
       count = words.length
       return [to_xhtml, count, opts] if count <= max && max > 0
       [words.slice(0, max).join, count, opts]

spec/html_truncator_spec.rb

@@ -163,4 +163,12 @@
   it "should remove punctuation before the ellipsis" do
     HTML_Truncator.truncate(short_text, 1).should == "<p>Foo…</p>"
   end
+
+  it "should behave correctly with html entities" do
+    HTML_Truncator.truncate("<br>&lt;br&gt;123456789012", 9, length_in_chars: true).should == "<br>&lt;br&gt;12345…"
+    HTML_Truncator.truncate("<p>&lt;br&gt; 1</p>2 3456789", 9, length_in_chars: true).should == "<p>&lt;br&gt; 1</p>2…"
+    HTML_Truncator.truncate("&lt;br&gt;", 10, length_in_chars: true).should == "&lt;br&gt;"
+    HTML_Truncator.truncate("<br>12345678901", 10, length_in_chars: true).should == "<br>1234567890…"
+    HTML_Truncator.truncate("<br>&lt;br&gt; 1 2 3", 2).should == "<br>&lt;br&gt; 1…"
+  end
 end