md2dk.py

#!/usr/bin/env python3
# md2dk.py
# Created by dturba at 13/03/2020
import re
from argparse import ArgumentParser, Namespace
from typing import List, Optional

_link_pattern = re.compile(r'\[(?P<name>[^\]]+)(\])(\()(?P<link>[^)]+)\)',
                           flags=0)


def title_formatting(*, line: str, **_) -> str:
    if line.startswith('#' * 6):
        line = f"{line.replace('#' * 6, '=')} ="
    elif line.startswith('#' * 5):
        line = f"{line.replace('#' * 5, '=' * 2)} {'=' * 2}"
    elif line.startswith('#' * 4):
        line = f"{line.replace('#' * 4, '=' * 3)} {'=' * 3}"
    elif line.startswith('#' * 3):
        line = f"{line.replace('#' * 3, '=' * 4)} {'=' * 4}"
    elif line.startswith('#' * 2):
        line = f"{line.replace('#' * 2, '=' * 5)} {'=' * 5}"
    elif line.startswith('#'):
        line = f"{line.replace('#', '=' * 6)} {'=' * 6}"
    return line


def code_block_formatting(
        *, line: str, lines: List[str], current_i: int, **_) -> str:
    if '```' in line:
        line = line.replace('```', '')
        for i, tmp_line in enumerate(lines[current_i + 1:], 1):
            if '```' in tmp_line:
                lines[current_i + i] = lines[current_i + i].replace('```', '')
                break
            else:
                lines[current_i + i] = ''.join(('  ', lines[current_i + i]))
    return line


def list_formatting(*, line: str, **_) -> str:
    if line.startswith('- '):
        line = line.replace('- ', '  * ')
    return line


def monospace_formatting(*, line: str, **_) -> str:
    new_line = line
    while new_line.find('`') != -1:
        new_line = new_line.replace('`', "''%%", 1)
        new_line = new_line.replace('`', "%%''", 1)
    return new_line


def link_formatting(*, line: str, **_) -> str:
    new_line = line
    link_match = _link_pattern.search(new_line)
    while link_match:
        new_line = f"{new_line[:link_match.start()]}[[" \
                   f"{link_match.group('link')}|" \
                   f"{link_match.group('name')}]]" \
                   f"{new_line[link_match.end():]}"
        link_match = _link_pattern.search(new_line)
    return new_line


def formatting(text: str) -> str:
    lines = text.splitlines()
    format_function_li = [
        title_formatting,
        code_block_formatting,
        list_formatting,
        monospace_formatting,
        link_formatting,
    ]
    final_lines = list()

    for i, line in enumerate(lines):
        for format_function in format_function_li:
            line = format_function(line=line, lines=lines, current_i=i)
        final_lines.append(line)
    final_text = '\n'.join(final_lines)
    return final_text


def convert_md_to_mw(input_filename: str, output_filename: Optional[str],
                     open_mode: str) -> None:
    with open(input_filename, 'r') as input_file:
        final_text = formatting(input_file.read())
    if not output_filename:
        return print(final_text)
    with open(output_filename, open_mode) as output_file:
        output_file.write(final_text)
    return print(f'{output_filename} written')


if __name__ == '__main__':
    parser = ArgumentParser(prog='md2dk',
                            usage='%(prog)s [-h] [-o O] input_file [--w]',
                            description='Convert a file from markdown to '
                                        'dokuwiki. If no output file is given,'
                                        ' the standard output will be used.')
    parser.add_argument('input_file', type=str,
                        help='the file you need to convert')
    parser.add_argument('-o',
                        help='name of output file in dokuwiki format')
    parser.add_argument('--w', default='x', const='w', action='store_const',
                        help='if you want to overwrite the output file')
    args: Namespace = parser.parse_args()
    convert_md_to_mw(args.input_file, args.o, args.w)