def create_vk_spider(name, module, boards=None, owner_id=None, url=None,
access_token=''):
if not owner_id and not url:
raise exc.SpiderException("Either owner_id or url must be specified!")
if owner_id and url:
raise exc.SpiderException("Both owner_id and url given, choose one")
if url:
raise exc.SpiderException("Url passing not supported yet")
# XXX call to utils.get_access_token left only for convenient
# scrapy crawl spider-name calls.
# FIXME change to calls from control.py one day
access_token = access_token or utils.get_access_token()
generated = gen_vk_spider_class(
name=name, owner_id=owner_id, boards=boards, access_token=access_token)
# a nasty hack to make generated class discoverable by scrapy
generated.__module__ = module
return generated
def crawl_all(token=None):
if not token:
LOG.warn("No token passed, "
"acquiring one using login data from settings")
token = utils.get_access_token()
LOG.info("Access token: %s" % token)
runner = crawler.CrawlerRunner(project.get_project_settings())
dispatcher.connect(on_close, signal=signals.spider_closed)
for spider_cls in spider_utils.find_spiders():
# FIXME incapsulation vialation
# inject access_token to a VK spider
spider_cls.access_token = token
RUNNING_CRAWLERS.append(spider_cls)
runner.crawl(spider_cls)
d = runner.join()
d.addBoth(lambda _: send_mail())
internet.reactor.run()
