From 7b9d4b1f1b448cc5f6653c191beda8615b35fd7c Mon Sep 17 00:00:00 2001
From: Timothy Clem <timothy.clem@gmail.com>
Date: Thu, 16 Feb 2017 08:29:49 -0800
Subject: [PATCH] Add test for toc summaries of files with unicode chars

---
 test/TOCSpec.hs                   | 5 +++++
 test/corpus/toc/ruby/unicode.A.rb | 4 ++++
 test/corpus/toc/ruby/unicode.B.rb | 7 +++++++
 3 files changed, 16 insertions(+)
 create mode 100644 test/corpus/toc/ruby/unicode.A.rb
 create mode 100644 test/corpus/toc/ruby/unicode.B.rb

diff --git a/test/TOCSpec.hs b/test/TOCSpec.hs
index a0d414812..8be608155 100644
--- a/test/TOCSpec.hs
+++ b/test/TOCSpec.hs
@@ -46,6 +46,11 @@ spec = parallel $ do
       diff <- testDiff sourceBlobs
       diffTOC sourceBlobs diff `shouldBe` [ JSONSummary $ Summarizable C.Function "performHealthCheck" (sourceSpanBetween (8, 1) (29, 2)) "modified" ]
 
+    it "handles unicode characters in file" $ do
+      sourceBlobs <- blobsForPaths (both "ruby/unicode.A.rb" "ruby/unicode.B.rb")
+      diff <- testDiff sourceBlobs
+      diffTOC sourceBlobs diff `shouldBe` [ JSONSummary $ Summarizable C.Method "foo" (sourceSpanBetween (6, 1) (7, 4)) "added" ]
+
     prop "only methods and functions are summarized" $
       \iden body ->
         let
diff --git a/test/corpus/toc/ruby/unicode.A.rb b/test/corpus/toc/ruby/unicode.A.rb
new file mode 100644
index 000000000..890628027
--- /dev/null
+++ b/test/corpus/toc/ruby/unicode.A.rb
@@ -0,0 +1,4 @@
+# ’
+# RIGHT SINGLE QUOTATION MARK
+# Unicode: U+2019, UTF-8: E2 80 99
+"’"
diff --git a/test/corpus/toc/ruby/unicode.B.rb b/test/corpus/toc/ruby/unicode.B.rb
new file mode 100644
index 000000000..3d59ce4af
--- /dev/null
+++ b/test/corpus/toc/ruby/unicode.B.rb
@@ -0,0 +1,7 @@
+# ’
+# RIGHT SINGLE QUOTATION MARK
+# Unicode: U+2019, UTF-8: E2 80 99
+"’"
+
+def foo(a, b, c)
+end