Generate list of revision URLs from a mediawiki dump

gen_rev_list.py

import sys

with open(sys.argv[1]) as fin:
  lines = fin.readlines()

INDEX_URL = sys.argv[2]

TITLE_PREFIX = '    <title>'
TITLE_SUFFIX = '</title>\n'
REV_PREFIX   = '      <id>'
REV_SUFFIX   = '</id>\n'

for line in lines:
  if line.startswith(TITLE_PREFIX):
    cur_title = line[len(TITLE_PREFIX):-len(TITLE_SUFFIX)].replace(' ', '_')
  if line.startswith(REV_PREFIX):
    rev=line[len(REV_PREFIX):-len(REV_SUFFIX)]
    print(INDEX_URL + '?title=' + cur_title + '&oldid=' + rev)
    print(INDEX_URL + '?title=' + cur_title + '&action=edit&oldid=' + rev)