def _build(self, config):
try:
self.name = config['name']
Logger.register(self.name)
Logger.load()
self.logger = Logger(self.name)
self.debug = config.get('debug', True)
self.logger.info(self.name, 'Start building...')
self.scraper = Scraper.get(config['scraper']['name'])(self)
if 'args' in config['scraper']:
self.scraper.setargs(config['scraper']['args'])
self.frontier = Frontier.get(config['frontier']['name'])(self)
if 'args' in config['frontier']:
self.frontier.setargs(config['frontier']['args'])
for each in config['handlers']:
handler = Handler.get(each['name'])(self)
if 'args' in each:
handler.setargs(each['args'])
self.handlers.append(handler)
self.logger.info(self.name, 'Build successful!')
except KeyError as e:
raise PyCrawlerException('Key \''+e.args[0]+'\' missing in config dict')
def test_getScraper(self):
self.assertRaises(ScraperException, Scraper.get, 'None')
self.assertEqual(DefaultScraper, Scraper.get('DefaultScraper'))
self.assertIsInstance(Scraper.get('DefaultScraper')(SpiderTest('testspider')), DefaultScraper)