def main():
with open("./data/businesses_plus.csv") as csvfile:
reader = csv.reader(csvfile, delimiter=',')
i = 0
for data in reader:
i += 1
if i == 1:
continue
if data[0] in rep:
#print rep[data[0]], data
continue
rep[data[0]] = i
rec = models.Restaurant(id = int(data[0]),
name = preprocess(data[1]),
address = preprocess(data[2]),
postal_code = int(data[4]) if data[4].isdigit() else None,
latitude = float(data[5]) if isFloat(data[5]) else None,
longitude = float(data[6]) if isFloat(data[6]) else None,
phone_number = data[7][1:4]+'-'+data[7][4:7]+'-'+data[7][7:11] if data[7].isdigit() else None)
db.session.add(rec)
if i % 1000 == 0:
db.session.flush()
db.session.commit()
with open("./data/inspections_plus.csv") as csvfile:
reader = csv.reader(csvfile, delimiter=',')
i = 0
for data in reader:
i += 1
if i == 1:
continue
rec = models.Inspections(business_id = int(data[0]),
date = datetime.strptime(data[2], '%Y%m%d') if data[2] != '' else None,
score = int(data[1]) if data[1].isdigit() else None,
insp_type = data[3])
db.session.add(rec)
if i % 1000 == 0:
db.session.flush()
db.session.commit()
with open("./data/violations_plus.csv") as csvfile:
reader = csv.reader(csvfile, delimiter=',')
i = 0
for data in reader:
i += 1
if i == 1:
continue
rec = models.Violations(business_id = int(data[0]),
date = datetime.strptime(data[1], '%Y%m%d') if data[1] != '' else None,
typeID = int(data[2]) if data[2].isdigit() else -1,
risk_category = data[3],
description = data[4])
db.session.add(rec)
if i % 1000 == 0:
db.session.flush()
db.session.commit()
def main():
with open("./app/data/movies.csv") as csvfile:
reader = csv.reader(csvfile, delimiter=',')
i = 0
for data in reader:
i += 1
#print i
if i == 1:
continue
br = bracket(data[1])
mov = models.Movie(id = int(data[0]),
title = br[0][0: br[1]],
year = int(br[0][br[1]+1: br[1]+5]) if br[1] < len(br[0]) else 2011)
genres = data[2].split('|')
print mov.id, mov.year, mov.title
for genre in genres:
if genre != '(no genres listed)':
gre = models.Category(movie_id = int(data[0]), category = preprocess(genre))
print gre.movie_id, gre.category
db.session.add(gre)
db.session.add(mov)
if i % 1000 == 0:
db.session.flush()
db.session.commit()
with open("./app/data/links.csv") as csvfile:
reader = csv.reader(csvfile, delimiter=',')
i = 0
for data in reader:
i += 1
print i
if i == 1:
continue
link = models.Links(movie_id = int(data[0]),
imdb_link = int(data[1] if data[1].isdigit() else None),
moviedb_link = int(data[2]) if data[2].isdigit() else None)
#print link.movie_id. link.imdb_link, link.moviedb_link
db.session.add(link)
if i % 1000 == 0:
db.session.flush()
db.session.commit()
with open("./app/data/ratings.csv") as csvfile:
reader = csv.reader(csvfile, delimiter=',')
i = 0
for data in reader:
i += 1
#print i
if i == 1:
continue
rating = models.Rating(user_id = int(data[0]),
movie_id = int(data[1]),
score = float(data[2])*1.0,
time = datetime.fromtimestamp(int(data[3])))
print rating.user_id, rating.movie_id, rating.score, rating.time
db.session.add(rating)
if i % 1000 == 0:
db.session.flush()
db.session.commit()
def main():
with open("./data/businesses_plus.csv") as csvfile:
reader = csv.reader(csvfile, delimiter=',')
i = 0
for data in reader:
i += 1
if i == 1:
continue
if data[0] in rep:
#print rep[data[0]], data
continue
rep[data[0]] = i
rec = models.Restaurant(
id=int(data[0]),
name=preprocess(data[1]),
address=preprocess(data[2]),
postal_code=int(data[4]) if data[4].isdigit() else None,
latitude=float(data[5]) if isFloat(data[5]) else None,
longitude=float(data[6]) if isFloat(data[6]) else None,
phone_number=data[7][1:4] + '-' + data[7][4:7] + '-' +
data[7][7:11] if data[7].isdigit() else None)
db.session.add(rec)
if i % 1000 == 0:
db.session.flush()
db.session.commit()
with open("./data/inspections_plus.csv") as csvfile:
reader = csv.reader(csvfile, delimiter=',')
i = 0
for data in reader:
i += 1
if i == 1:
continue
rec = models.Inspections(
business_id=int(data[0]),
date=datetime.strptime(data[2], '%Y%m%d')
if data[2] != '' else None,
score=int(data[1]) if data[1].isdigit() else None,
insp_type=data[3])
db.session.add(rec)
if i % 1000 == 0:
db.session.flush()
db.session.commit()
with open("./data/violations_plus.csv") as csvfile:
reader = csv.reader(csvfile, delimiter=',')
i = 0
for data in reader:
i += 1
if i == 1:
continue
rec = models.Violations(
business_id=int(data[0]),
date=datetime.strptime(data[1], '%Y%m%d')
if data[1] != '' else None,
typeID=int(data[2]) if data[2].isdigit() else -1,
risk_category=data[3],
description=data[4])
db.session.add(rec)
if i % 1000 == 0:
db.session.flush()
db.session.commit()
def bracket(s):
s_new = preprocess(s)
for i in range(len(s_new)-1, -1, -1):
if s_new[i] == '(':
return (s_new, i)
return (s_new, len(s_new))
