From 8d8e1eca1e6fcce2b4cf22c7f06cbc1facb834c4 Mon Sep 17 00:00:00 2001
From: Alexey Kopytko <alexey@kopytko.ru>
Date: Wed, 21 Dec 2016 11:43:01 +0900
Subject: [PATCH] rndr_header_anchor: do not remove character if nothing was
 added

---
 ext/redcarpet/html.c     | 7 ++++++-
 test/html_render_test.rb | 7 +++++++
 2 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/ext/redcarpet/html.c b/ext/redcarpet/html.c
index 1e3a9f61..2f717ba6 100644
--- a/ext/redcarpet/html.c
+++ b/ext/redcarpet/html.c
@@ -281,16 +281,20 @@ rndr_header_anchor(struct buf *out, const struct buf *anchor)
 	int stripped = 0, inserted = 0;
 
 	for (; i < size; ++i) {
+		// skip html tags
 		if (a[i] == '<') {
 			while (i < size && a[i] != '>')
 				i++;
+		// skip html entities
 		} else if (a[i] == '&') {
 			while (i < size && a[i] != ';')
 				i++;
 		}
+		// replace non-ascii or invalid characters with dashes
 		else if (!isascii(a[i]) || strchr(STRIPPED, a[i])) {
 			if (inserted && !stripped)
 				bufputc(out, '-');
+			// and do it only once
 			stripped = 1;
 		}
 		else {
@@ -300,7 +304,8 @@ rndr_header_anchor(struct buf *out, const struct buf *anchor)
 		}
 	}
 
-	if (stripped)
+	// replace the last dash if there was anything added
+	if (stripped && inserted)
 		out->size--;
 }
 
diff --git a/test/html_render_test.rb b/test/html_render_test.rb
index 42a7eb01..9bc3da95 100644
--- a/test/html_render_test.rb
+++ b/test/html_render_test.rb
@@ -252,6 +252,13 @@ def test_non_ascii_removal_in_header_anchors
     assert_equal html, render(markdown, with: [:with_toc_data])
   end
 
+  def test_utf8_only_header_anchors
+    markdown = "# 見出し"
+    html = "<h1 id=\"\">見出し</h1>"
+
+    assert_equal html, render(markdown, with: [:with_toc_data])
+  end
+
   def test_escape_entities_removal_from_anchor
     output = render("# Foo's & Bar's", with: [:with_toc_data])
     result = %(<h1 id="foos-bars">Foo&#39;s &amp; Bar&#39;s</h1>)