def __init__(self):
self.shop_extractor = ShopExtractor()
self.archiver = S3Archiver()
class Miner:
def __init__(self):
self.shop_extractor = ShopExtractor()
self.archiver = S3Archiver()
def construct_parser(self, source):
parser = None
if source == "dida":
parser = DidaParser()
elif source == "dianping":
parser = DianpingParser()
elif source == "ftuan":
parser = FtuanParser()
elif source == "lashou":
parser = LashouParser()
elif source == "manzuo":
parser = ManzuoParser()
elif source == "meituan":
parser = MeituanParser()
elif source == "nuomi":
parser = NuomiParser()
elif source == "wowo":
parser = WowoParser()
elif source == "wuba":
parser = WubaParser()
return parser
def mine(self, source, mine_shop=True, mine_deal=True):
price_not_match = 0
parser = self.construct_parser(source)
if not parser:
return None
deal_set = dict()
shop_set = dict()
filelist = self.shop_extractor.get_deal_file_list(source)
for filename in filelist:
filepath = os.path.join(source, filename)
print "processing", source, filename
try:
parser.parse(filepath)
except Exception, e:
print "Error: miner.py", source, filename
continue # get rid of this error file
# assume all deals in the same file belong to the same city
city_name = None
city_index = None
for deal in parser.deals:
if deal["deal_city"] in self.shop_extractor.cache:
city_name = deal["deal_city"]
city_index = self.shop_extractor.cache[city_name]
if city_index == "283": # if it is the whole country
continue
break
for deal in parser.deals:
sales_num = deal["sales_num"]
price = deal["price"]
current_city_index = None
if deal["deal_city"] in self.shop_extractor.cache:
current_city_index = self.shop_extractor.cache[deal["deal_city"]]
if (
current_city_index == "283"
): # although it is country-wise, it still sells inside the current city.
current_city_index = city_index
else:
current_city_index = city_index
deal_id = str(current_city_index) + "_" + deal["deal_id"] # add city information
# mine deal information
if mine_deal:
if deal_id not in deal_set:
deal_set[deal_id] = [price, sales_num]
else:
if price != deal_set[deal_id][0] or sales_num != deal_set[deal_id][1]:
print deal_id, "price or sales_num does not match!"
price_not_match = price_not_match + 1
# mine shop information
if mine_shop:
deal_sold_at = deal["deal_sold_at"]
for shop in deal_sold_at:
shop_entry = self.shop_extractor.make_shop_entry_name_as_key(shop)
if len(shop_entry) > 0:
shop_entry[0] = str(city_index) + "_" + shop_entry[0]
if shop_entry[0] not in shop_set:
shop_set[shop_entry[0]] = [
deal_id,
shop_entry[1],
shop_entry[2],
shop_entry[4],
shop_entry[5],
] # tel, addr, lat, lng
print "price or sales_num not match:", price_not_match
return [deal_set, shop_set]
