Script for converting Python 2 to Python 3 in Markdown files

md2to3.py

#!/usr/bin/env python3

import sys
import tempfile
import logging
import re
from pathlib import Path
from lib2to3 import refactor
from lib2to3.pgen2.parse import ParseError


def main(infile, outfile, force=False):
    fixers = refactor.get_fixers_from_package('lib2to3.fixes')
    rt = refactor.RefactoringTool(fixers)

    made_changes = False
    parse_error = False
    with open(infile) as fin:
        with tempfile.NamedTemporaryFile('w', delete=False) as fout:
            ispystart = False
            ispy = False
            chunk_no = 0
            for line in fin:
                # Capture start of code block, but do not process this line
                if re.match('```\s*python', line.strip()):
                    ispystart = True
                    chunk = []
                    chunk_no += 1
                # Capture subsequent lines:
                elif ispystart:
                    ispystart = False
                    ispy = True
                # Stop capturing and process the chunk
                if ispy and line.strip() == '```':
                    ispy = False
                    code_str = ''.join(chunk)
                    try:
                        out_tree = rt.refactor_string(code_str, f'chunk{chunk_no}')
                    except ParseError:
                        logging.debug(f'Error parsing chunk #{chunk_no}!')
                        fout.write('# ParseError: Could not check this chunk!\n')
                        parse_error = True
                        out_str = code_str
                    else:
                        out_str = str(out_tree)
                    if code_str != out_str:
                        logging.debug(f'Updated chunk #{chunk_no}.')
                        made_changes = True
                        code_str = out_str
                    else:
                        logging.debug(f'No changes in chunk #{chunk_no}.')
                    fout.write(code_str)
                # Capture code lines
                if ispy:
                    chunk.append(line)
                # Write out all other lines unchanged
                else:
                    fout.write(line)
    if made_changes:
        logging.info('Changes required.')
        outpath = Path(outfile)
        if outpath.exists() and not force:
            logging.error(f'Not overwriting existing file {outfile}!')
        else:
            Path(fout.name).rename(outpath)
    elif parse_error:
        logging.info('Could not determine required changes.')
    else:
        logging.info('No changes necessary.')


if __name__ == '__main__':
    import argparse
    parser = argparse.ArgumentParser()
    group = parser.add_mutually_exclusive_group(required=True)
    group.add_argument('-o', '--outfile')
    group.add_argument('-w', '--writeback', action='store_true')
    parser.add_argument('-f', '--force', action='store_true')
    parser.add_argument('-v', '--verbose', action='store_true')
    parser.add_argument('infile')

    args = parser.parse_args()
    if args.writeback:
        args.outfile = args.infile
        args.force = True

    if args.verbose:
        level = logging.DEBUG
    else:
        level = logging.INFO
    logging.basicConfig(level=level)

    main(args.infile, outfile=args.infile, force=args.force)