def main(*args):
utf8_writer = codecs.getwriter('utf8')
single_categories = False
output_destination = utf8_writer(sys.stdout)
formatstring = "{}\n"
categories_only = False
for arg in pywikibot.handle_args(list(args)):
if arg == "-out:categories":
single_categories = True
elif arg == "-out:file":
output_destination = codecs.open("Denkmallistenliste.txt", "w", 'utf-8')
elif arg == "-fmt:wiki":
formatstring = "[[{}]]\n"
elif arg == "-fmt:url":
formatstring = "https://de.wikipedia.org/wiki/{}\n"
elif arg == "-list-cat-only":
categories_only = True
site = pywikibot.Site()
fetcher = CategoryFetcher(site)
if categories_only and not single_categories:
export_to_file(output_destination, fetcher.get_categories(False), formatstring)
return
if single_categories:
for category in fetcher.get_categories():
with codecs.open(category.title() + u".txt", "w", 'utf-8') as outfile:
export_to_file(outfile, category.articles(), formatstring)
else:
collection = ArticleCollection()
callbacks = ArticleIteratorCallbacks(article_callback=collection.cb_add_article)
article_iterator = ArticleIterator(callbacks, fetcher.get_categories())
article_iterator.iterate_categories()
export_to_file(output_destination, collection.article_list, formatstring)
def main(*args):
site = pywikibot.Site()
fetcher = CategoryFetcher(site)
checker = TemplateChecker()
checker.load_config("config/templates.json")
checker_bot = CheckerBot(checker, site)
all_categories = fetcher.get_categories()
callbacks = ArticleIteratorCallbacks(
category_callback=checker_bot.cb_store_category_result,
article_callback=checker_bot.cb_check_article,
logging_callback=pywikibot.log,
)
article_iterator = ArticleIterator(callbacks, categories=all_categories)
parser = ArticleIteratorArgumentParser(article_iterator, fetcher)
for argument in pywikibot.handle_args(list(args)):
if parser.check_argument(argument):
continue
elif argument.find("-outputpage:") == 0:
checker_bot.outputpage = argument[12:]
elif argument.find("-exclude-articles:") == 0:
page = pywikibot.Page(site, argument[18:])
article_iterator.excluded_articles = load_excluded_articles_from_wiki(page)
article_iterator.iterate_categories()
if article_iterator.categories != all_categories: # Don't update summary page if only single categories were crawled
return
summary = checker_bot.generate_summary_page()
if checker_bot.outputpage:
checker_bot.save_wikipage(summary, checker_bot.outputpage + u"/Zusammenfassung")
else:
pywikibot.output(u"Zusammenfassung")
pywikibot.output(u"===============")
pywikibot.output(summary)
pywikibot.output(checker_bot.generate_config_table())
def main():
utf8_writer = codecs.getwriter('utf8')
output_destination = utf8_writer(sys.stdout)
sample_size = 100
site = pywikibot.Site()
fetcher = CategoryFetcher(site)
counties = fetcher.get_categories()
for county in counties:
sample_county(county, sample_size, output_destination)
