reformat.py

import os, subprocess

# Regex to detect [//]-style comments:
# \[\/\/\]: (?:# )?(?:\((.*)\)|([^(].*[^)]))

def process_pandoc(in_path, out_path):
    with open(in_path, 'r') as f:
        lines = f.readlines()

    if not lines:
        with open(out_path, 'w') as f:
            pass
        return

    state = 'before header'
    for header_end, line in enumerate(lines):
        if state == 'before header' and line.strip() == '---':
            state = 'in header'
        elif state == 'in header' and line.strip() == '---':
            state = 'after header'
        elif state == 'after header' and line.strip():
            break
    header = lines[:header_end]

    enabled_extensions = [
        'yaml_metadata_block',
        # 'grid_tables',  # not supported by kramdown
    ]

    disabled_extensions = [
        'hard_line_breaks',
        # 'raw_html',
        # 'simple_tables',
        # 'pipe_tables',
        # 'multiline_tables',
    ]

    extensions = (
        ''.join('+' + ext for ext in enabled_extensions) +
        ''.join('-' + ext for ext in disabled_extensions)
    )

    subprocess.run([
        'pandoc',
        '--from=markdown',
        '--to=markdown_github' + extensions,
        '--columns=80',
        '--atx-headers', # makes headers with ### instead of underlining
        '--normalize',   # removes redundant spaces between words
        # '--smart',       # adds fancy punctuation (—, …, etc)
        # fancy punctuation includes fancy quotes, which I personally don't like
        # '--standalone',  # adds YAML metadata header
        # pandoc does not preserve order in the header, so not adding it back
        in_path,
        '-o',
        out_path
    ])

    with open(out_path, 'r') as f:
        lines = f.readlines()

    fixed_lines = []
    for line, next_line in zip(lines[:-1], lines[1:]):
        fixed_lines.append(line)
        if line.strip().endswith('-->') and next_line.strip():
            fixed_lines.append('\n')
    fixed_lines.append(lines[-1])

    lines = header + fixed_lines

    with open(out_path, 'w') as f:
        f.writelines(lines)


def process(docs_in, docs_out, in_root, in_fname):
    in_path = os.path.join(in_root, in_fname)
    print(in_path)

    assert in_path.startswith(docs_in)
    out_root = docs_out + in_root[len(docs_in):]
    out_path = os.path.join(out_root, in_fname)
    if not os.path.exists(out_root):
        os.makedirs(out_root)

    process_pandoc(in_path, out_path)


def main():
    docs_in = '_docs'
    docs_out = '_formatted_docs'
    for root, dirs, files in os.walk(docs_in):
        for fname in files:
            if fname.endswith('.md'):
                process(docs_in, docs_out, root, fname)

if __name__ == '__main__':
    main()