Converting Markdown to PDF with PHP

Recently, I had to take some content in markdown, specifically markdown extra, and convert it to a series of PDFs styled with a specific branding. While some will argue that PDFs are dead and “long live the web”, many of us still need to produce PDFs for one reason or another.

In this case I had to take markdown extra, with some html sprinkled in, clean it up, and convert it to a styled PDF. What follows is how I did that using QueryPath for the cleanup and DOMPDF to make the conversion.

The Setup

At the root of this little app was a PHP script with the dependencies managed through composer. The composer.json file looked like:

{
    "name": "foo/bar",
    "description": "Convert markdown to PDF.",
    "type": "application",
    "require": {
        "php": ">=5.3.0",
        "michelf/php-markdown": "1.4.*",
        "dompdf/dompdf" : "0.6.*",
        "querypath/querypath": "3.*",
        "masterminds/html5": "1.*"
    }
}

Turning the Markdown into HTML

Within the script I started with a file we’ll call $file. Form here it was easy using the official markdown extra conversion utility.

$markdown = file_get_contents($file);
$markdownParser = new \Michelf\MarkdownExtra();
$html = $markdownParser->transform($markdown);

This produces the html needed to go inside the body of an html page. From here I wrapped it in a document because I could easily link to a CSS file for styling purposes. DOMPDF supports quite a bit of CSS 2.1.

$html = '<html>
    <head>
        <link rel="stylesheet" type="text/css" href="pdf.css">
    </head>
    <body>' . $html . '</body>
    </html>’;

pdf.css is where you can style the PDF. If you know how to style web pages using CSS you can manage to style a PDF document.

Cleaning Up The Content

There were a number of places html had been injected into the markdown that was either broken, unwanted in a PDF, or an edge case that DOMPDF didn’t support. To make these changes I used QueryPath. For example, I needed to take relative links, normally used in generation of a website, and add a domain name to them:

$dom = \HTML5::loadHTML($html);
$links = htmlqp($dom, 'a');
foreach ($links as $link) {
    $href = $link->attr('href');
    if (substr($href, 0, 1) == '/' && substr($href, 1, 1) != '/') {
        $link->attr('href', $domain_name . $href);
    }
}
$html = \HTML5::saveHTML($dom);

Note, I used the HTML5 parser and writer rather than the built-in one designed for xhtml and HTML 4. This is because DOMPDF attempts to work with HTML5 and I wanted to keep that consistent from the beginning.

Converting to PDF

There is a little setup before using DOMPDF. It has a built in autoloader which should be disabled and needs a config file. In my case I used the default config file and handled this with:

define('DOMPDF_ENABLE_AUTOLOAD', false);
require_once __DIR__ . '/vendor/dompdf/dompdf/dompdf_config.inc.php';

The conversion was fairly straight forward. I used a snippet like:

$dompdf = new DOMPDF();
$dompdf->load_html($html);
$dompdf->render();  
$output = $dompdf->output();
file_put_contents(‘path/to/file.pdf', $output);

DOMPDF has a lot of options and some quirks. It wasn’t exactly designed for composer. For example, if you want to work with custom fonts you need to get the project from git and install submodules.

Despite the quirks, needing to cleanup some of the html, and brand the documents, I was able to write a conversion script that handled dozens of documents quickly. Almost all of my time was on html cleanup and css styling.