# Static site generator in 90 lines of Python code

Created: Apr 16, 2025

[На русском](/ru/blog/2025/01-sitegen)

A long time ago, I made a small business card website. It consisted of three 
simple HTML pages, one CSS file (which I generated from SCSS), several fonts 
and images. That was more than enough to get a link to my website featured 
in a resume or social media profile.


![Picture of the old website](/images/01-oldsite.png "Picture of the old website")

I recently decided to continue working <a href="https://github.com/blankhex/bhlib" target="_blank">on my pet-project</a> 
and would like to publish all sorts of notes and articles on this topic on my 
website. I didn't want to manually mess with HTML files, so I decided to look 
for an alternative in the form of some kind of static website generator. 
Ideally, I would like it to be:

- Small and simple
- Able to work with Markdown
- Able syntax-highlight blocks of code

Unfortunately, I couldn't find any suitable solutions for myself, so I decided
to build my own using Python, <a href="https://mistune.lepture.com/en/latest/" target="_blank">mistune</a> 
Markdown parser, <a href="https://jinja.palletsprojects.com/en/stable/" target="_blank">Jinja2</a> 
template engine, and <a href="https://pygments.org" target="_blank">Pygments</a>.
The whole generation process boils down to the following:

1. For every file in the input directory check whether it is Markdown
   - If yes - convert it to HTML (with highlighting) and write to output directory
   - If no - copy as is to output directory
2. Compress content of the output directory

This generation process has a rather major drawback - due to the fact that 
there is no post-processing of HTML, any links to other Markdown pages must 
end with a `.html` extension[^1].

[^1]: This can be mitigated by special web-server configuration, that replaces
  `.md` extension with `.html` or by omitting `.md` extension entirely and 
  using something like `try_files $uri $uri.html`

Here is the code:

```python
import re, jinja2, mistune, shutil, os, pathlib, tarfile
from pygments.lexers import get_lexer_by_name
from pygments.formatters import HtmlFormatter
from pygments import highlight


class PygmentsHTMLRenderer(mistune.HTMLRenderer):
    def block_code(self, code: str, info = None):
        if not info:
            return '\n<pre><code>%s</code></pre>\n' % mistune.escape(code)
        lexer = get_lexer_by_name(info, stripall=True)
        formatter = HtmlFormatter(lineseparator='<br>')
        return highlight(code, lexer, formatter)


def convert_markdown(page: str):
    plugins = ['footnotes', 'table', 'strikethrough', 'url']
    renderer = PygmentsHTMLRenderer(escape=False)
    return mistune.create_markdown(plugins=plugins, renderer=renderer)(page)


def extract_title(page: str):
    matches = re.match('<h1>(.*?)</h1>', page)
    if matches:
        return matches.group(1)
    return 'BlankHex'


def handle_file(path: str, input_dir: str, output_dir: str, template_name: str):
    # Calculate input and output paths
    relpath = os.path.relpath(path, input_dir)
    input_path = path
    output_path = os.path.join(output_dir, relpath)
    if input_path.endswith('.md'):
        output_path = output_path.replace('.md', '.html')

    # Don't convert if output path exists
    if os.path.exists(output_path):
        return

    # Run conversion
    pathlib.Path(os.path.dirname(output_path)).mkdir(parents=True, exist_ok=True)
    if input_path.endswith('.md'):
        # Read Markdown document
        with open(input_path, 'r') as handle:
            markdown_page = handle.read()

        # Get Pygments styles for light and dark themes
        light_style = HtmlFormatter(style='default').get_style_defs()
        dark_style = HtmlFormatter(style='monokai').get_style_defs()

        # Convert Markdown document to HTML document
        html_page = convert_markdown(markdown_page)
        html_header = extract_title(html_page)
        environment = jinja2.Environment(loader=jinja2.FileSystemLoader('template/'))
        template = environment.get_template(template_name)
        output_page = template.render(title=html_header,
                                      body=html_page,
                                      light_style=light_style,
                                      dark_style=dark_style)

        # Write HTML document
        with open(output_path, 'w') as handle:
            handle.write(output_page)
    else:
        # Copy file as is
        shutil.copy(path, output_path)


def convert_dir(input_dir: str, output_dir: str, template_name: str):
    # Convert or copy every file from the input directory to the output directory
    for subdir, dirs, files in os.walk(input_dir):
        for file in files:
            handle_file(os.path.join(subdir, file), input_dir, output_dir, template_name)


# Remove output from previous run
if os.path.isdir('public'):
    shutil.rmtree('public')
if os.path.isfile('public.tgz'):
    os.remove('public.tgz')

# Run conversion
convert_dir('content', 'public', 'template.html')
with tarfile.open('public.tgz', 'w:gz') as tar:
    for file in os.listdir('public'):
        tar.add(os.path.join('public', file), file)
```