moin2txt

moin2txt.py

#!/usr/bin/env python2
# coding=utf-8

# convert MoinMoin wiki to text
# for moinmoin 1.9.x
# author: alswl
# update at: 2012-07-22

import sys
import os
import argparse
import binascii
import re
import shutil

IS_DECODE_PATH = False

def convert(root, dir, target):
    name = name_decode(dir)
    if not name:
        return
    dst = os.path.join(target, name.replace('/', '-') + '.txt')
    try:
        version = open(os.path.join(root, dir, 'current'),
                       'r').readline().strip()
        src = os.path.join(root, dir, 'revisions', version)
        shutil.copyfile(src, dst)
    except IOError, e:
        if IS_DECODE_PATH:
            dir = dir.replace('(', r'\(').replace(')', r'\)')
        sys.stderr.write('File %s, Name: %s, Message: %s\n'
                         %(dir, name, str(e)))

def name_decode(name):
    raw = ''
    lastpos = 0
    ENCODE_RE = re.compile(r'\(([\w\d]+)\)')
    match =ENCODE_RE.search(name)
    while(match):
        raw += name[lastpos : lastpos + match.start()]
        raw += binascii.unhexlify(match.groups()[0])
        lastpos += match.end()
        match = ENCODE_RE.search(name[lastpos:])
    return raw

def walk(path, target):
    for dir in os.listdir(path):
        convert(path, dir, target)

def main():
    parser = argparse.ArgumentParser(
        description='Convert moin wiki to text archieves'
        )
    parser.add_argument('--input', '-i',
                        help='the path of moinmoin/data/pages',
                        type=str,
                        required=True)
    parser.add_argument('--output', '-o',
                        help='the path os target',
                        type=str,
                        required=True)
    args = parser.parse_args()

    walk(args.input, args.output)

if __name__ == '__main__':
    main()