Merge pull request #284 from pragdave/i279-csharp-code-blocks

Fixes: #279;

lib/earmark/line_scanner.ex

@@ -111,9 +111,9 @@ defmodule Earmark.LineScanner do
         [_, spaces, code] = match
         %Line.Indent{level: div(String.length(spaces), 4), content: code}
 
-      match = Regex.run(~r/^\s*(```|~~~)\s*([\w\-]*)\s*$/u, line) ->
+      match = Regex.run(~r/^\s*(```|~~~)\s*([^`\s]*)\s*$/u, line) ->
         [_, fence, language] = match
-        %Line.Fence{delimiter: fence, language: language}
+        %Line.Fence{delimiter: fence, language: _attribute_escape(language)}
 
       #   Although no block tags I still think they should close a preceding para as do many other
       #   implementations.
@@ -218,6 +218,13 @@ defmodule Earmark.LineScanner do
     end
   end
 
+
+  defp _attribute_escape(string), do:
+    string
+    |> String.replace("&", "&")
+    |> String.replace("<", "&lt;")
+
+
   @block_tags ~w< address article aside blockquote canvas dd div dl fieldset figcaption h1 h2 h3 h4 h5 h6 header hgroup li main nav noscript ol output p pre section table tfoot ul video>
               |> Enum.into(MapSet.new())
   defp block_tag?(tag), do: MapSet.member?(@block_tags, tag)

test/acceptance/html/fenced_code_blocks_test.exs

@@ -3,7 +3,7 @@ defmodule Acceptance.Html.FencedCodeBlocksTest do
 
   import Support.Helpers, only: [as_html: 1, as_html: 2]
 
-  # describe "Fenced code blocks" do
+  describe "Fenced code blocks" do
     test "no lang" do
       markdown = "```\n<\n >\n```\n"
       html     = "<pre><code class=\"\">&lt;\n &gt;</code></pre>\n"
@@ -43,8 +43,33 @@ defmodule Acceptance.Html.FencedCodeBlocksTest do
 
       assert as_html(markdown) == {:ok, html, messages}
     end
+  end
 
-  # end
+  describe "do not make too many assumptions about programming language names" do
+    test "at least the existing ones shall work" do
+      markdown = "```c#\nI do not know c# code\n```\n"
+      html     = "<pre><code class=\"c#\">I do not know c# code</code></pre>\n"
+      messages = []
+
+      assert as_html(markdown) == {:ok, html, messages}
+    end
+
+    test "and let us anticipate creative language designers too" do
+      markdown = "```42lang!\nassert x == 42\n```\n"
+      html     = "<pre><code class=\"42lang!\">assert x == 42</code></pre>\n"
+      messages = []
+
+      assert as_html(markdown) == {:ok, html, messages}
+    end
+
+    test "be careful about what can go into an HTML attribute though" do
+      markdown = "```a<b&\nassert x == 42\n```\n"
+      html     = "<pre><code class=\"a&lt;b&amp;\">assert x == 42</code></pre>\n"
+      messages = []
+
+      assert as_html(markdown) == {:ok, html, messages}
+    end
+  end
 end
 
 # SPDX-License-Identifier: Apache-2.0

test/acceptance/html1/fenced_code_blocks_test.exs

@@ -62,7 +62,38 @@ defmodule Acceptance.Html1.FencedCodeBlocksTest do
 
       assert to_html1(markdown) == {:ok, html, messages}
     end
+  end
+
+  describe "do not make too many assumptions about programming language names" do
+    test "at least the existing ones shall work" do
+      markdown = "```c#\nI do not know c# code\n```\n"
+      html     = construct([
+        :pre,
+        {:code, ~s{class="c#"}, "I do not know c# code"}])
+      messages = []
 
+      assert to_html1(markdown) == {:ok, html, messages}
+    end
+
+    test "and let us anticipate creative language designers too" do
+      markdown = "```42lang!\nassert x == 42\n```\n"
+      html     = construct([
+        :pre,
+        {:code, ~s{class="42lang!"}, "assert x == 42"}])
+      messages = []
+
+      assert to_html1(markdown) == {:ok, html, messages}
+    end
+
+    test "be careful about what can go into an HTML attribute though" do
+      markdown = "```a<b&\nassert x == 42\n```\n"
+      html     = construct([
+        :pre,
+        {:code, ~s{class="a&lt;b&amp;"}, "assert x == 42"}])
+      messages = []
+
+      assert to_html1(markdown) == {:ok, html, messages}
+    end
   end
 end