Handle code blocks with and without language spec

This commit includes the following:

 * Improve `pretty_codeblocks` function to work with Hoedown and Pandoc
 * Add more tests for `pretty_codeblocks`
 * Delete duplicate code in `ExDoc.Markdown.to_html`
 * Avoid to run too many Regex.replace expressions
 * Add logic for pretty blocks in Hoedown and Pandoc

Author: milmazz

lib/ex_doc/formatter/html.ex

@@ -75,11 +75,17 @@ defmodule ExDoc.Formatter.HTML do
   end
 
   @doc false
-  # Helper to handle plain code blocks (```...```) without
+  # Helper to handle plain code blocks (```...```) with and without
   # language specification and indentation code blocks
   def pretty_codeblocks(bin) do
-    Regex.replace(~r/<pre><code\s*(class=\"\")?>/,
-                  bin, "<pre class=\"codeblock\">")
+    bin = Regex.replace(~r/<pre><code(\s+class=\"\")?>\s*iex&gt;/,
+                        # Add "elixir" class for now, until we have support for
+                        # "iex" in highlight.js
+                        bin, "<pre><code class=\"iex elixir\">iex&gt;")
+    bin = Regex.replace(~r/<pre><code(\s+class=\"\")?>/,
+                        bin, "<pre><code class=\"elixir\">")
+
+    bin
   end
 
   @doc false

lib/ex_doc/markdown.ex

@@ -7,16 +7,11 @@ defmodule ExDoc.Markdown do
 
   @markdown_processor_key :markdown_processor
 
-  def to_html(text) when is_binary(text) do
-    text = get_markdown_processor().to_html(text)
+  import ExDoc.Formatter.HTML, only: [pretty_codeblocks: 1]
 
-    # handle code blocks (```...```) with language specification
-    text = Regex.replace(~r/<pre><code\s*(class=\"\")?>\s*iex&gt;/,
-                  #add "elixir" class for now, until we have support for iex in highlight.js
-                  text, "<pre><code class=\"iex elixir\">iex&gt;")
+  def to_html(text) when is_binary(text) do
+    text = get_markdown_processor().to_html(text) |> pretty_codeblocks
 
-    text = Regex.replace(~r/<pre><code(\s+class=\"\")?>/,
-                  text, "<pre><code class=\"elixir\">")
     text
   end
 

lib/ex_doc/markdown/hoedown.ex

@@ -16,5 +16,17 @@ defmodule ExDoc.Markdown.Hoedown do
       autolink: Keyword.get(opts, :autolink, true),
       fenced_code: Keyword.get(opts, :fenced_code, true),
       tables: Keyword.get(opts, :tables, true))
+    |> pretty_codeblocks
+  end
+
+  @doc false
+  # Helper to handle fenced code blocks (```...```) with
+  # language specification
+  defp pretty_codeblocks(bin) do
+    # Hoedown parser puts the prefix "language-" as part of the class value
+    bin = Regex.replace(~r/<pre><code\s+class=\"language-([^\"]+)\">/,
+                        bin, "<pre><code class=\"\\1\">")
+
+    bin
   end
 end

lib/ex_doc/markdown/pandoc.ex

@@ -15,6 +15,7 @@ defmodule ExDoc.Markdown.Pandoc do
     |> text_to_file()
     |> open_port(opts)
     |> process_port()
+    |> pretty_codeblocks()
   end
 
   defp text_to_file(text) do
@@ -53,4 +54,16 @@ defmodule ExDoc.Markdown.Pandoc do
         {status, List.to_string(data)}
     end
   end
+
+  @doc false
+  # Helper to handle fenced code blocks (```...```) with
+  # language specification
+  defp pretty_codeblocks(bin) do
+    # Pandoc parser puts the class attribute inside the `pre` tag
+    # Move the class attribute to the code element to keep consistency.
+    bin = Regex.replace(~r/<pre\s+class=\"([^\"]+)\"><code>/,
+                        bin, "<pre><code class=\"\\1\">")
+
+    bin
+  end
 end

test/ex_doc/formatter/html_test.exs

@@ -92,10 +92,16 @@ defmodule ExDoc.Formatter.HTMLTest do
   test "make markdown codeblocks pretty" do
     with_empty_class = "<pre><code class=\"\">mix run --no-halt path/to/file.exs"
     without_class = "<pre><code>mix run --no-halt path/to/file.exs"
+    iex_detected_with_empty_class = "<pre><code class=\"\">iex&gt; max(4, 5)"
+    iex_detected_without_class = "<pre><code>iex&gt; max(4, 5)"
 
     assert HTML.pretty_codeblocks(with_empty_class) ==
-           "<pre class=\"codeblock\">mix run --no-halt path/to/file.exs"
+           "<pre><code class=\"elixir\">mix run --no-halt path/to/file.exs"
     assert HTML.pretty_codeblocks(without_class) ==
-           "<pre class=\"codeblock\">mix run --no-halt path/to/file.exs"
+           "<pre><code class=\"elixir\">mix run --no-halt path/to/file.exs"
+    assert HTML.pretty_codeblocks(iex_detected_with_empty_class) ==
+          "<pre><code class=\"iex elixir\">iex&gt; max(4, 5)"
+    assert HTML.pretty_codeblocks(iex_detected_without_class) ==
+          "<pre><code class=\"iex elixir\">iex&gt; max(4, 5)"
   end
 end

test/ex_doc/markdown/hoedown_test.exs

@@ -23,4 +23,10 @@ defmodule ExDoc.Markdown.HoedownTest do
   test "to_html handles empty input" do
     assert Markdown.to_html("") == ""
   end
+
+  test "pretty Markdown fenced code blocks for Hoedown" do
+    hoedown_with_language_specified = "```elixir\nmix run --no-halt path/to/file.exs\n```"
+    expected = "<pre><code class=\"elixir\">mix run --no-halt path/to/file.exs\n</code></pre>\n"
+    assert Markdown.to_html(hoedown_with_language_specified) == expected
+  end
 end

test/ex_doc/markdown/pandoc_test.exs

@@ -30,4 +30,10 @@ defmodule MarkdownTest.PandocTest do
   test "to_html converts to rst" do
     assert Markdown.to_html("`hello`", format: "rst") == "``hello``\n"
   end
+
+  test "pretty Markdown fenced code blocks for Pandoc" do
+    pandoc_with_language_specified = "```elixir\nmix run --no-halt path/to/file.exs\n```"
+    expected = "<pre><code class=\"elixir\">mix run --no-halt path/to/file.exs</code></pre>\n"
+    assert Markdown.to_html(pandoc_with_language_specified) == expected
+  end
 end