def get_gacha_record_to_sql(num=5):
db = pymysql.connect(host,username,password,database)
hash_list = []
print("一共进行{}次爬取".format(str(num)))
for i in range(num):
valid_times = 0
time1 = time.time()
print("正在进行第{}次爬取..".format(str(i+1)))
sample_gacha_record = get_sample_gacha_record()
record_hash = get_md5(json.dumps(sample_gacha_record))
if record_hash in hash_list:
continue
else:
hash_list.append(record_hash)
# 应判断item是否已经存在于数据库中，如何判断是否同一次出货？
for item in sample_gacha_record:
# print(item['uid'])
if is_valid_data(db,item):
valid_times += 1
insert_data(db,item)
db.commit()
print("第{}次爬取成功,本次共爬取有效数据{}条,用时{}秒".format(str(i+1),str(valid_times),str((time.time()-time1))))
next_time = int(60 * ((-valid_times/40)+5))
print("下次爬取时间为{}秒后".format(next_time))
time.sleep(next_time)
def do_request(url, oauth, carcel, payload, geocode):
if payload != None:
r = requests.get(url=url, auth=oauth, params=payload)
data = r.json()
else:
r = requests.get(url=url, auth=oauth)
data = r.json()
try:
next_results = data['search_metadata']['next_results']
print next_results
except:
print "There are not next_results"
next_results = "None"
for status in data['statuses']:
obj = {}
obj['carcel'] = carcel
status_id = status['id']
obj['status_id'] = status_id
text = status['text']
obj['text'] = text
screen_name = status['user']['screen_name']
obj['screen_name'] = screen_name
profile_img_url = status['user']['profile_image_url_https']
download_profile_image(profile_img_url, screen_name)
utc_offset = status['user']['utc_offset']
obj['utc_offset'] = utc_offset
user_id = status['user']['id']
obj['user_id'] = user_id
created_at = status['created_at']
obj['created_at'] = created_at
if 'geo' in status and status['geo'] != None:
latitude = status['geo']['coordinates'][0]
longitude = status['geo']['coordinates'][1]
obj['latitude'] = latitude
obj['longitude'] = longitude
obj['retuited'] = "no"
obj['in_jail'] = "no"
if calc_distance(latitude, longitude, float(geocode.split(",")[0]),
float(geocode.split(",")[1])) < 1.1:
print obj['status_id'], calc_distance(
latitude, longitude, float(geocode.split(",")[0]),
float(geocode.split(",")[1]))
insert_data(obj)
return next_results
def do_request(url, oauth, carcel, payload, geocode):
if payload != None:
r = requests.get(url=url, auth=oauth, params=payload)
data = r.json()
else:
r = requests.get(url=url, auth=oauth)
data = r.json()
try:
next_results = data['search_metadata']['next_results']
print next_results
except:
print "There are not next_results"
next_results = "None"
for status in data['statuses']:
obj = {}
obj['carcel'] = carcel
status_id = status['id']
obj['status_id'] = status_id
text = status['text']
obj['text'] = text
screen_name = status['user']['screen_name']
obj['screen_name'] = screen_name
profile_img_url = status['user']['profile_image_url_https']
download_profile_image(profile_img_url, screen_name)
utc_offset = status['user']['utc_offset']
obj['utc_offset'] = utc_offset
user_id = status['user']['id']
obj['user_id'] = user_id
created_at = status['created_at']
obj['created_at'] = created_at
if 'geo' in status and status['geo'] != None:
latitude = status['geo']['coordinates'][0]
longitude = status['geo']['coordinates'][1]
obj['latitude'] = latitude
obj['longitude'] = longitude
obj['retuited'] = "no"
if calc_distance(latitude, longitude, float(geocode.split(",")[0]), float(geocode.split(",")[1])) < 1.1:
print obj['status_id'], calc_distance(latitude, longitude, float(geocode.split(",")[0]), float(geocode.split(",")[1]))
insert_data(obj)
return next_results
def run(username, flow):
# data = lib.get_mixedtables(username, flow)
# return jsonify(
# data=data,
# username=username,
# flow=flow)
data = [(u'news:5', 1), (u'norris:6', 2)]
print(data)
simpleList = []
for line in data:
action_data = lib.run_action(line)
simpleList.append(action_data)
return jsonify(data=simpleList,
username=username,
flow=flow,
time=datetime.datetime.now())
newslist = []
for line in news:
list_data = lib.insert_news(line)
newslist.append[news_data]
norrislist = []
for line in norris:
list_data = lib.insert_data(line)
norrislist.append[norris_data]
import lib
if len(sys.argv) < 3:
print "This script inserts 1 tuit into our database using the status_id value and a string for the jail"
print "Usage python insert_tuit.py 123208309281908 'Penal Diroes'"
sys.exit()
oauth = api.get_oauth()
# get tuit data
url = "https://api.twitter.com/1.1/statuses/show.json"
payload = {'id': sys.argv[1].strip()}
r = requests.get(url=url, auth=oauth, params=payload)
data = r.json()
obj = {}
obj['carcel'] = unicode(sys.argv[2].strip(), "utf-8")
obj['created_at'] = data['created_at']
obj['screen_name'] = data['user']['screen_name']
obj['status_id'] = data['id']
obj['text'] = data['text']
obj['user_id'] = data['user']['id']
obj['utc_offset'] = data['user']['utc_offset']
coords = data['geo']['coordinates']
obj['latitude'] = coords[0]
obj['longitude'] = coords[1]
lib.insert_data(obj)