Retrieves the full source of a title from Wikisource

gimmesrc.py

#!/usr/bin/env python
# Example: python gimmesrc.py De_Cive > txt

import sys, urllib, urllib2

URL = 'http://en.wikisource.org/w/index.php?action=raw&title='

if __name__ == '__main__':
    title = sys.argv[1]
    title_parts = []
    data = urllib2.urlopen(URL+title).read()
    in_contents = False
    for line in data.split('\n'):
        if line.strip() == '===Contents===':
            in_contents = True
            continue
        if in_contents:
            if line == '':
                in_contents = False
            else:
                line = line.split('|')
                title_parts.append(line[0].strip('*['))
    for part in title_parts:
        if not part.startswith('/'):
            continue
        print part
        part = urllib.quote_plus(part.replace(' ', '_'))
        data = urllib2.urlopen(URL+title+part).read()
        in_contents = False
        for line in data.split('\n'):
            if line.strip() == '<div class="prose">':
                in_contents = True
                continue
            if line.startswith('[['):
                in_contents = False
            if in_contents:
                print line