correctly truncate text that contains MB chars, fixes #4135

TextHelper#truncate_html doesn't appear to have this issue, because it
always splits and truncates on word boundaries (using String#split()).

Change-Id: I8bf301055818d7080d8e12859707458ba41e89f4
Reviewed-on: https://gerrit.instructure.com/2851
Reviewed-by: Zach Wily <zach@instructure.com>
Tested-by: Hudson <hudson@instructure.com>

lib/text_helper.rb

@@ -119,14 +119,14 @@ module TextHelper
   def truncate_text(text, options={})
     max_length = options[:max_length] || 30
     ellipsis = options[:ellipsis] || "..."
-    words = options[:words] || false
     ellipsis_length = ellipsis.length
-    content_length = text.length
     actual_length = max_length - ellipsis_length
-    if content_length > max_length
+
-      truncated = text[0, actual_length] + ellipsis
+    truncated = (text || "")[/.{0,#{actual_length}}/mu]
+    if truncated.length < text.length
+      truncated + ellipsis
     else
-      text
+      truncated
     end
   end
   

spec/lib/text_helper_spec.rb

@@ -96,4 +96,16 @@ describe TextHelper do
     end
   end
 
+  context "truncate_text" do
+    it "should split on multi-byte character boundaries" do
+      str = "This\ntext\nhere\n获\nis\nutf-8"
+      th.truncate_text(str, :max_length => 9).should ==  "This\nt..."
+      th.truncate_text(str, :max_length => 21).should == "This\ntext\nhere\n获\ni..."
+      th.truncate_text(str, :max_length => 18).should == "This\ntext\nhere\n..."
+      th.truncate_text(str, :max_length => 19).should == "This\ntext\nhere\n获..."
+      th.truncate_text(str, :max_length => 20).should == "This\ntext\nhere\n获\n..."
+      th.truncate_text(str, :max_length => 80).should == str
+    end
+  end
+
 end