Code Blocks

Code is the one construct a Markdown parser must be able to keep its hands off. Everything inside a code span or a code fence is literal: no emphasis, no links, no entity interpretation. Getting that isolation right is what lets people paste Markdown examples into Markdown documents at all.

Fenced blocks: what exactly opens and closes one?

```python
print("hi")
```

Four rules do all the work, and each one has a failure mode:

The info string is the language tag. It becomes class="language-python" on the <code> element, which is all a syntax highlighter needs.

<pre><code class="language-python">print(&quot;hi&quot;)
</code></pre>

Note the two details in that output: the content is HTML-escaped even though it is code, and it ends with a newline before </code>. Both are what the tests compare against.

Inline code: how many backticks?

A code span opens with a run of N backticks and closes with a run of exactly N. That is how you put a backtick inside code — surround it with two.

There is one more normalisation: if the content begins and ends with a space, and is not entirely spaces, one space is removed from each end. That rule exists so you can write code that itself starts with a backtick.

Row three is the whole lesson. The naive pattern closes on the first backtick it finds, even though that backtick is part of a longer run — so a code span whose content is two backticks collapses to a code span containing a space. The lookarounds (?<!)and(?!) enforce "this run, and no more", which is precisely the spec's definition of a backtick string.

Indented code blocks

Four spaces of indentation also makes code, with two differences worth knowing: there is no language tag, and an indented block cannot interrupt a paragraph — indented lines under prose are just continuation lines of that paragraph. Fenced blocks have neither restriction, which is why they won.

Your exercise

Convert fenced blocks and inline code spans; leave everything else alone.

The mistake the grader catches is including the closing fence line in the content. The expected output for the plain-fence test is exactly <pre><code>plain code then a newline then </code></pre> — three lines. If your loop appends every line until it sees the closer and then also appends the closer, you emit a stray ``` inside the block and every fenced test fails at once.

The second trap is the trailing newline: the content ends with a newline before </code> even when the block holds a single line. Join the collected body lines with a newline, then append one more.