def map_by_presence_of_coupons(self, _, line):
data = line
COUPONS = tools.read_coupons_file()
if any(word in data['text'].lower() for word in COUPONS):
yield "review_c", data["stars"]
else:
yield "review", data["stars"]
def separate_map_by_stars_and_categories(self, _, line):
data = line
if data.get("review_id"):
# business_id, ["review", user_id, stars]
COUPONS = tools.read_coupons_file()
if any(word in data['text'].lower() for word in COUPONS):
yield data["business_id"], ["review", data["review_id"], data["stars"] ]
elif data.get("business_id"):
# business_id, ["business", categories]
yield data["business_id"], ["categories", data["categories"]]
def separate_map_by_stars_and_categories(self, _, line):
data = line
if data.get("review_id"):
# business_id, ["review", user_id, stars]
has_coupon = False
COUPONS = tools.read_coupons_file()
for coupon in COUPONS:
if coupon in data["text"].lower():
has_coupon = True
yield data["business_id"], ["review_c", data["review_id"], data["stars"], coupon]
if has_coupon == False:
yield data["business_id"], ["review", data["review_id"], data["stars"]]
elif data.get("business_id"):
# business_id, ["business_info", name, categories]
yield data["business_id"], ["business_info", data["name"], data["categories"]]
def mapper_input(self, _, line):
COUPONS = tools.read_coupons_file()
if any(word in line['text'].lower() for word in COUPONS):
yield 'COUNTER', 1