From 6f03cfc67c0c389437b2935ad9d4ed92f900de4d Mon Sep 17 00:00:00 2001
From: Joshua Peek <josh@joshpeek.com>
Date: Thu, 23 Jun 2011 16:51:03 -0500
Subject: [PATCH] Assume utf8 when serving raw text blobs

---
 lib/linguist/pathname.rb |  2 +-
 test/test_blob.rb        | 16 ++++++++--------
 test/test_pathname.rb    |  4 ++--
 3 files changed, 11 insertions(+), 11 deletions(-)

diff --git a/lib/linguist/pathname.rb b/lib/linguist/pathname.rb
index 00a10f66..d9e6059d 100644
--- a/lib/linguist/pathname.rb
+++ b/lib/linguist/pathname.rb
@@ -87,7 +87,7 @@ module Linguist
     #
     # Returns a content type String.
     def content_type
-      @content_type ||= binary? ? mime_type : 'text/plain'
+      @content_type ||= binary? ? mime_type : 'text/plain; charset=utf8'
     end
 
     # Public: Is the path binary?
diff --git a/test/test_blob.rb b/test/test_blob.rb
index 6cc8511f..19eb69a9 100644
--- a/test/test_blob.rb
+++ b/test/test_blob.rb
@@ -35,14 +35,14 @@ class TestBlob < Test::Unit::TestCase
     assert_equal "application/octet-stream", blob("dog.o").content_type
     assert_equal "application/pdf", blob("foo.pdf").content_type
     assert_equal "image/png", blob("foo.png").content_type
-    assert_equal "text/plain", blob("README").content_type
-    assert_equal "text/plain", blob("foo.html").content_type
-    assert_equal "text/plain", blob("foo.pl").content_type
-    assert_equal "text/plain", blob("foo.py").content_type
-    assert_equal "text/plain", blob("foo.rb").content_type
-    assert_equal "text/plain", blob("foo.sh").content_type
-    assert_equal "text/plain", blob("foo.xhtml").content_type
-    assert_equal "text/plain", blob("foo.xml").content_type
+    assert_equal "text/plain; charset=utf8", blob("README").content_type
+    assert_equal "text/plain; charset=utf8", blob("foo.html").content_type
+    assert_equal "text/plain; charset=utf8", blob("foo.pl").content_type
+    assert_equal "text/plain; charset=utf8", blob("foo.py").content_type
+    assert_equal "text/plain; charset=utf8", blob("foo.rb").content_type
+    assert_equal "text/plain; charset=utf8", blob("foo.sh").content_type
+    assert_equal "text/plain; charset=utf8", blob("foo.xhtml").content_type
+    assert_equal "text/plain; charset=utf8", blob("foo.xml").content_type
   end
 
   def test_disposition
diff --git a/test/test_pathname.rb b/test/test_pathname.rb
index 520a6d38..8bc65fc2 100644
--- a/test/test_pathname.rb
+++ b/test/test_pathname.rb
@@ -58,8 +58,8 @@ class TestPathname < Test::Unit::TestCase
   end
 
   def test_content_type
-    assert_equal 'text/plain', Pathname.new("file.txt").content_type
-    assert_equal 'text/plain', Pathname.new("file.rb").content_type
+    assert_equal 'text/plain; charset=utf8', Pathname.new("file.txt").content_type
+    assert_equal 'text/plain; charset=utf8', Pathname.new("file.rb").content_type
     assert_equal 'image/png', Pathname.new("octocat.png").content_type
   end
 end