def __init__(self):
AExternalEntitySource.__init__(self, "NYTimesBestSellers", self.TYPES, 512)
self.base = 'http://www.nytimes.com'
self.seen = set()
self.details_re = re.compile('.*___by ([^(]*)\. \(([^)]*)\) (.*)')
self.date_re = re.compile('.*/(\d\d\d\d)-(\d\d)-(\d\d)/.*')
def __init__(self):
AExternalEntitySource.__init__(self, "TheTVDBCrawler", self.TYPES, 512)
self.base = 'http://thetvdb.com'
self._thetvdb = TheTVDB()
self._id_re = re.compile('.*&id=([0-9]+).*')
self._actor_re = re.compile('.*___([^_]+)___as ([^_]+)___.*')
self._season_re = re.compile('([0-9]+) - ([0-9]+)')
self._date_re = re.compile('([0-9]+)-([0-9]+)-([0-9]+)')
def __init__(self):
AExternalEntitySource.__init__(self, "Yelp", self.TYPES, 512)
self.base = 'http://www.yelp.com'
self.title_re = re.compile('[0-9]*\. (.*)')
self.address_re = re.compile('([^_]*)___*([^_]*)___*([^_]*)_*')
self.rating_reviews_re = re.compile('([0-9.]*) .*')
self.start_re = re.compile('.*start=([0-9]*).*')
self.category_re = re.compile('refine_category_.*')
self.results_per_page = 40
self.seen = set()
def __init__(self):
AExternalEntitySource.__init__(self, "BostonMag", self.TYPES, 512)
self._seen = set()
def __init__(self):
AExternalEntitySource.__init__(self, "LATimes", self.TYPES, 512)
self._count = {}
self._seen = set()
def __init__(self):
AExternalEntitySource.__init__(self, "TimeOutLA", self.TYPES, 512)
self._seen = set()
def __init__(self):
AExternalEntitySource.__init__(self, "SeattleTimes", self.TYPES, 512)
self.base = 'http://www.seattletimes.com'
self.seen = set()
self.page_re = re.compile('.*&page=([0-9]*)')
def __init__(self):
AExternalEntitySource.__init__(self, "NYMag", self.TYPES, 512)
def __init__(self):
AExternalEntitySource.__init__(self, "Urbanspoon", self.TYPES, 512)
self.base = 'http://www.urbanspoon.com'
def __init__(self):
AExternalEntitySource.__init__(self, "AwardAnnals", self.TYPES, 512)
self.base = 'http://www.awardannals.com'
self.seen = set()
self.page_re = re.compile('Page 1 of ([0-9]*)')
def __init__(self):
AExternalEntitySource.__init__(self, "AmazonBestSellerBookFeeds", self.TYPES, 512)
self.base = 'http://www.amazon.com'
self.seen = set()
self.max_depth = 2
def __init__(self):
AExternalEntitySource.__init__(self, "SFWeekly", self.TYPES, 512)
self._seen = set()
def __init__(self):
AExternalEntitySource.__init__(self, "Zagat", self.TYPES, 512)
self.base = 'http://www.zagat.com'