def __init__( self, start_url, message_handler=None, max_trys=10, reader=None, rss_proxy_url=None, rss_proxy_token=None, dns_service=DNS_SERVICE, ): start_url = coerce_url(start_url) try: start_url = validate_url(start_url) except Invalid: raise ValueError(f"invalid start_url {start_url!r}") self._set_start_url(start_url) self.message_handler = message_handler self.max_trys = max_trys if reader is None: reader = FeedReader( dns_service=dns_service, rss_proxy_url=rss_proxy_url, rss_proxy_token=rss_proxy_token, ) self._close_reader = True else: self._close_reader = False self.reader = reader self._links = {start_url: ScoredLink(start_url, 1.0)} self._visited = set() self._guessed = False
def __init__( self, start_url, message_handler=None, max_trys=10, reader=None, allow_private_address=False, rss_proxy_url=None, rss_proxy_token=None, ): start_url = coerce_url(start_url) self._set_start_url(start_url) self.message_handler = message_handler self.max_trys = max_trys if reader is None: reader = FeedReader( allow_private_address=allow_private_address, rss_proxy_url=rss_proxy_url, rss_proxy_token=rss_proxy_token, ) self._close_reader = True else: self._close_reader = False self.reader = reader self._links = {start_url: ScoredLink(start_url, 1.0)} self._visited = set() self._guessed = False
def _import_one_line_text(text): text = text.strip() parts = text.split(maxsplit=2) if len(parts) != 1: return None url = coerce_url(parts[0]) try: _validate_url(url) except Invalid: return None return url
def __init__(self, start_url, message_handler=None, max_trys=10, reader=None, validate=True): start_url = coerce_url(start_url) self._set_start_url(start_url) self.message_handler = message_handler self.max_trys = max_trys if reader is None: reader = FeedReader() self._close_reader = True else: self._close_reader = False self.reader = reader self.validate = validate self._links = {start_url: ScoredLink(start_url, 1.0)} self._visited = set()