Upgrading the Markdown Parser (Because Regex is a Bad Idea)

Ten minutes after publishing my first post, I realized my markdown parser was terrible.

The Problem

Built a custom markdown converter using regex. Worked for basic stuff. Broke on anything complex.

Here's what I had:

private function convertMarkdown($text)
{
    $text = preg_replace('/^### (.+)$/m', '<h3>$1</h3>', $text);
    $text = preg_replace('/^## (.+)$/m', '<h2>$1</h2>', $text);
    $text = preg_replace('/\*\*(.+?)\*\*/s', '<strong>$1</strong>', $text);
    // ... more regex hell
}

Problems:

• Didn't handle nested markdown
• Code blocks broke everything
• Lists didn't work properly
• Tables? Forget it
• Edge cases everywhere

Regex for parsing structured text is a known bad idea. Did it anyway because I wanted to ship fast. Paid for it immediately.

The Fix

Switched to Parsedown. Proper markdown parser. Handles everything.

Installation

composer require erusev/parsedown

Implementation

Replaced entire convertMarkdown() method with:

public function getContentHtmlAttribute()
{
    $parsedown = new Parsedown();
    $parsedown->setSafeMode(true); // Prevents XSS
    return $parsedown->text($this->content);
}

Three lines. Done.

What This Gives Me

Better markdown support:

• All heading levels (not just three)
• Proper code blocks with ``` syntax
• Nested lists work correctly
• Blockquotes
• Tables
• Links and images handled properly
• Strikethrough text
• Horizontal rules

Security:

• setSafeMode(true) prevents XSS attacks
• No raw HTML execution
• Safe by default

Maintainability:

• Not maintaining custom regex
• Battle-tested library
• Edge cases already handled

The Deploy

Changed locally. Pushed. Pulled on server.

Got this error:

Class "Parsedown" not found

Right. Forgot to install on the server.

ssh liam@46.225.13.75
cd ~/sites/hizhub
composer require erusev/parsedown

Refreshed. Worked.

Testing the New Features

Let me test some markdown features that didn't work before:

Code Blocks

function toggleTheme() {
    const html = document.documentElement;
    html.classList.toggle('dark');
}

Works now. Didn't before.

Tables

Feature	Regex Parser	Parsedown
Speed	Fast	Fast enough
Reliability	Breaks	Works
Maintenance	Pain	None

Tables work. Nice.

Nested Lists

1. First item
   • Sub-item one
   • Sub-item two
2. Second item
   • Another sub-item
     • Even deeper

All renders correctly now.

Blockquotes

"Premature optimization is the root of all evil."

• Donald Knuth

Handles quotes properly.

---

Horizontal rules work too.

What I Learned

Don't reinvent the wheel:

• Used regex because I thought it'd be faster
• Took longer to debug than installing a library would have
• Classic mistake

Ship fast, but not that fast:

• Could have installed Parsedown from the start
• Would have saved time overall
• Sometimes the "quick" solution costs more

Test before you ship:

• Should have tried more complex markdown before publishing
• Would have caught the limitations immediately
• Live and learn

Use battle-tested libraries:

• Parsedown has been around since 2013
• Millions of downloads
• Edge cases already handled
• Just use it

The Result

Blog now handles proper markdown. Code blocks render. Tables work. Lists nest correctly.

Took 10 minutes to fix. Should have done it right the first time.

But hey - shipped, broke, fixed. That's the process.

---

Lesson: Don't parse structured text with regex. Use a proper parser. Your future self will thank you.