Extra Newline When Highlighting Specific Lines

[damned-me]

Hello!

I was trying out a new Jekyll feature — Marking Specific Lines, implemented here — and encountered some strange behavior.

The issue can be reproduced by rendering the following (I'm using the Minima v3 theme).

{% highlight ruby linenos mark_lines="2" %}
def foo
  puts 'foo'
end
{% endhighlight %}

{% highlight ruby linenos %}
def foo
  puts 'foo'
end
{% endhighlight %}

And it should look something like the following:

I would expect the first block to present the same content as the second one. However, this doesn't happen due to what I believe is a bug in the HTML generation. Specifically, a newline is appended to the last processed line (after <span class="k">end</span>).

Generated HTML:

<figure class="highlight"><pre><code class="language-ruby" data-lang="ruby"><table class="rouge-table"><tbody><tr><td class="gutter gl"><pre class="lineno">1
2
3
</pre></td><td class="code"><pre><span class="k">def</span> <span class="nf">foo</span>
<span class="hll">  <span class="nb">puts</span> <span class="s1">'foo'</span>
</span><span class="k">end</span>

</pre></td></tr></tbody></table></code></pre></figure>
<figure class="highlight"><pre><code class="language-ruby" data-lang="ruby"><table class="rouge-table"><tbody><tr><td class="gutter gl"><pre class="lineno">1
2
3
</pre></td><td class="code"><pre><span class="k">def</span> <span class="nf">foo</span>
  <span class="nb">puts</span> <span class="s1">'foo'</span>
<span class="k">end</span>
</pre></td></tr></tbody></table></code></pre></figure>

I think the root cause for this behavior is here, where the newline character is appended without checking whether the line is the last one being processed.

While I don’t usually code in Ruby — apologies if the implementation quality is lacking — the following rewrite of the stream function effectively fixes the issue on my setup:

def stream(tokens)
    last_line = nil
    token_lines(tokens).with_index(1) do |line_tokens, lineno|
        if last_line
            yield last_line + "\n"
        end
        
        line = @delegate.format(line_tokens)
        line = %(<span class="#{@highlight_line_class}">#{line}</span>) if @highlight_lines.include?(lineno)
        last_line = line
    end
    
    yield last_line if last_line
end

Resulting in the following generated HTML:

Correctly generated HTML

<figure class="highlight"><pre><code class="language-ruby" data-lang="ruby"><table class="rouge-table"><tbody><tr><td class="gutter gl"><pre class="lineno">1
2
3
</pre></td><td class="code"><pre><span class="k">def</span> <span class="nf">foo</span>
<span class="hll">  <span class="nb">puts</span> <span class="s1">'foo'</span></span>
<span class="k">end</span>
</pre></td></tr></tbody></table></code></pre></figure>
<figure class="highlight"><pre><code class="language-ruby" data-lang="ruby"><table class="rouge-table"><tbody><tr><td class="gutter gl"><pre class="lineno">1
2
3
</pre></td><td class="code"><pre><span class="k">def</span> <span class="nf">foo</span>
  <span class="nb">puts</span> <span class="s1">'foo'</span>
<span class="k">end</span>
</pre></td></tr></tbody></table></code></pre></figure>

System:

• OS: Linux 9fe7e33eef81 6.8.0-1020-raspi #24-Ubuntu SMP PREEMPT_DYNAMIC Sun Feb 23 08:39:32 UTC 2025 aarch64 GNU/Linux
• Ruby version : ruby 3.3.8 (2025-04-09 revision b200bad6cd) [aarch64-linux]
• Rouge version : Rouge 4.5.1

Let me know if I can provide any additional useful information.

[jneen]

Hi there! I would recommend to the Jekyll developers that they write their own formatter by subclassing Rouge::Formatters::HTML - the HTMLLineFormatter exists primarily for documentation purposes, and the needs of any specific application's line-handling are too varied for us to handle in every case.

[damned-me]

Hey thanks for the fast reply!
While I acknowledge that its simple to monkey-patch the class on Jekyll side, I think that the output of the formatter should be consistent with its input and not modify it by adding anything superfluous to it, even if its a single character. As a very wide and adopted project is building over this library it would be awesome if you guys could patch the behavior at its root.

[jneen]

No problem - I think there may be a misunderstanding here, though. I'm not suggesting anyone monkey-patch a class. Formatters::HTML is designed to be subclassed by users of this library, so they can use the span and token_lines helpers to format HTML any way they like in their own stream(...) method. This is not a monkey-patch, but a basic supported use of rouge - creating a new class in their own namespace. If Jekyll is using Formatters::HighlightLineFormatter directly then this is a problem that should be solved imo.

As for the extra newline at the end, it's likely that in a production setting you would not actually want to use <pre> for this, but rather have <div> or <tr> or similar wrapped around each line. But Rouge cannot know what kind of layout you are using on a given page, so in the builtin formatters we do something fairly generic.

[tancnle]

But Rouge cannot know what kind of layout you are using on a given page, so in the builtin formatters we do something fairly generic.

IMHO, I think the generic approach would be to not having the extra newline at the end. The change suggested by @damned-me looks quite reasonable to be integrated in HTMLLineHighlighter.