def contents_first_line(): lines = [] for letter in ['#'] + russian_letters(): link = letter if letter != u'#' else u'0' lines.append(u'[[Участник:Vitalik7/Русский язык/%s|%s]]' % (link, letter)) return u'<small>Первая буква слова:</small><br/>\n' + \ u'{{Участник:Vitalik7/Первая буква слова}}\n\n'
def download_pages(): for letter in russian_letters(): print "=" * 10 print u"Буква: %s" % letter title = u"Индекс:Русский_язык/%s" % letter filename = join(path, letter) wiki = get_and_save_wiki(title, filename) lines = wiki.split("\n") for line in lines: if line.find(u"см.") != -1: m = re.match(u"^см. \[\[(/[а-яё])\|([а-яё]{2})\]\]$", line, re.UNICODE) if not m: print line continue m1 = m.group(1) m2 = m.group(2) if m1[1] != m2[1] or m2[0] != letter.lower(): print line continue title2 = title + m1 print m2 filename = join(path, letter + m1[1]) get_and_save_wiki(title2, filename) time.sleep(0.1)