def __init__(self, id, buffer):
self.buffer = buffer
self.html = fromstring(buffer)
self.id = id
self.alternative_title = None
self.countries = list()
self.countries_to_save = list()
self.slogan = None
self.persons = list()
self.length = None
self.year = None
self.cast = list()
self.ratings = list()
self.genres = list()
self.rating_kinopoisk = None
self.rating_imdb = None
self.rating_critics = None
self.age_restriction = None
self.premieres = list()
self.world_premiere = None
self.dates = list()
self.boxes = list()
self.rating_mpaa = None
self.production_status = None
self.full_id = self.get_full_id()
logger.info('Full ID = %s' % self.full_id)
self.cache = FileCache(namespace='kinopoisk', path=os.environ.get('CACHE_PATH'))
self.net = NetworkManager()
self.parse()
def __init__(self):
parser = argparse.ArgumentParser(
description='Parse database export script')
parser.add_argument('--url',
help='Process only this URL',
type=str,
required=False)
self.args = parser.parse_args()
self.cache = FileCache(namespace='germany-cities',
path=os.environ.get('CACHE_PATH'))
self.net = NetworkManager()
def __init__(self, id):
self.id = id
self.name = None
self.alternative_name = None
self.birth_date = None
self.birth_place = None
self.death_date = None
self.death_place = None
self.growth = None
self.cache = FileCache(namespace='kinopoisk',
path=os.environ.get('CACHE_PATH'))
self.net = NetworkManager()
page = self.get_page('%s/name/%s/' % (self.base, id))
self.parse(page)
def __init__(self):
parser = argparse.ArgumentParser(description='kinopoisk.ru parser')
parser.add_argument('--year', type=int, help='Year to process')
parser.add_argument('--hostname',
type=str,
help='Hostname',
required=False,
default=gethostname())
parser.add_argument('--film-id', type=int, help='Film ID')
parser.add_argument('--sleep-time',
type=int,
help='Max sleep time between requests',
default=20)
parser.add_argument('--total',
required=False,
default=False,
action='store_true')
parser.add_argument('--read-only',
required=False,
default=False,
action='store_true')
parser.add_argument('--update',
required=False,
default=False,
action='store_true')
parser.add_argument('--start-page',
required=False,
default=1,
type=int)
parser.add_argument('--persons',
required=False,
default=False,
action='store_true')
parser.add_argument('--from-id', required=False, default=1, type=int)
parser.add_argument('--to-id', required=False, default=None)
self.args = parser.parse_args()
self.cache = FileCache(namespace='kinopoisk',
path=os.environ.get('CACHE_PATH'))
self.net = NetworkManager()
# Initialization of database connection
db.connect(config.dsn)
if self.args.year is not None:
self.set_year(self.args.year)
def __init__(self):
logging.basicConfig(
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
level=logging.INFO,
stream=sys.stdout)
logger.setLevel(logging.INFO)
logging.getLogger('cache').setLevel(logging.INFO)
super(App, self).__init__()
self.parser.add_argument('--post', type=str, help='Post to parse')
self.parser.add_argument('--from-year',
type=int,
help='Year to start parse from',
default=2006)
self.parser.add_argument('--from-month',
type=int,
help='Month to start parse from',
default=1)
self.parser.add_argument(
'--update',
action='store_true',
default=False,
help='Do not use cache to construct post list')
self.parser.add_argument('--db',
type=str,
help='Database DSN',
default=db)
self.parser.add_argument('--image',
type=str,
help='Process one image and exit')
self.args = self.parser.parse_args()
self.net = NetworkManager()
self.cache = FileCache(path=cache_path, namespace='varlamov.ru')
self.conn = psycopg2.connect(self.args.db)
self.conn.set_isolation_level(
psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
def __init__(self):
self.cache = FileCache(namespace='russian-cities',
path=os.environ.get('CACHE_PATH'))
self.net = NetworkManager()