def getworkday():
strt = time.strftime('%Y-%m-%d', time.localtime(time.time()))
strt = strt.split("-")
y = int(strt[0])
m = int(strt[1])
d = int(strt[2])
april_last = date(y, m, d)
print(april_last)
print(is_workday(april_last))
print("fuckshit", is_holiday(april_last))
print("dick", not is_holiday(april_last))
print("elecshit", is_workday(april_last) and (not is_holiday(april_last)))
return is_workday(april_last) and (not is_holiday(april_last))
def get_trade_days(start_date, end_date):
"""计算两个日期间的工作日"""
from chinese_calendar import is_holiday
# 字符串格式日期的处理
if type(start_date) == str:
start_date = datetime.strptime(start_date, "%Y-%m-%d").date()
if type(end_date) == str:
end_date = datetime.strptime(end_date, "%Y-%m-%d").date()
# 开始日期大，颠倒开始日期和结束日期
if start_date > end_date:
start_date, end_date = end_date, start_date
date_list = []
counts = 0
while True:
if start_date > end_date:
break
if is_holiday(start_date) or start_date.weekday() == 5 or start_date.weekday() == 6:
start_date += timedelta(days=1)
continue
date_list.append(str(start_date.strftime("%Y-%m-%d")))
counts += 1
start_date += timedelta(days=1)
return counts, date_list
def tradedays(start, end):
'''
计算两个日期间的工作日
start:开始时间
end:结束时间
'''
from datetime import datetime, timedelta
from chinese_calendar import is_holiday
# 字符串格式日期的处理
if type(start) == str:
start = datetime.strptime(start, '%Y-%m-%d').date()
if type(end) == str:
end = datetime.strptime(end, '%Y-%m-%d').date()
# 开始日期大，颠倒开始日期和结束日期
if start > end:
start, end = end, start
counts = 0
while True:
if start > end:
break
if is_holiday(start) or start.weekday() == 5 or start.weekday() == 6:
start += timedelta(days=1)
continue
counts += 1
start += timedelta(days=1)
return counts
def process_query(name):
print('process query:', name)
if name == 'train':
train = pd.read_csv(data_path + 'train_queries.csv')
else:
train = pd.read_csv(data_path + 'test_queries.csv')
train['o_lon'] = train.o.apply(lambda x: float(x.split(',')[0]))
train['o_lat'] = train.o.apply(lambda x: float(x.split(',')[1]))
train['d_lon'] = train.d.apply(lambda x: float(x.split(',')[0]))
train['d_lat'] = train.d.apply(lambda x: float(x.split(',')[1]))
train['o_i'] = (train['o_lon'] - lon_min) // precision
train['o_j'] = (train['o_lat'] - lat_min) // precision
train['d_i'] = (train['d_lon'] - lon_min) // precision
train['d_j'] = (train['d_lat'] - lat_min) // precision
num_i = np.ceil((lon_max - lon_min) / precision)
num_j = np.ceil((lat_max - lat_min) / precision)
train['o_grid'] = train['o_i'] * num_j + train['o_j']
train['d_grid'] = train['d_i'] * num_j + train['d_j']
train['req_time'] = pd.to_datetime(train['req_time'])
train['hod'] = train['req_time'].dt.hour
train['dow'] = train['req_time'].dt.dayofweek + 1
train['holiday'] = train.req_time.apply(lambda x: is_holiday(x) * 1)
train['holiday'][(train.dow == 6) | (train.dow == 7)] = 1
if name == 'train':
train.to_csv(save_path + 'train_od_date.csv', index=False)
else:
train.to_csv(save_path + 'test_od_date.csv', index=False)
train.drop(['o', 'd', 'req_time'], axis=1, inplace=True)
return train
def get_pro_features(self, ds_pt_list, hours, weather_dict):
holiday = is_holiday(ds_pt_list[0].time)*1
day = ds_pt_list[0].time.day
hour = {'hour': np.bincount(hours).max()} # find most frequent hours as hour of the trajectory
weather = {'weather': weather_dict[(day, hour['hour'])]}
features = self.one_hot(hour) + self.one_hot(weather) + [holiday]
return features
def gen_feas(all_data_df):
# 添加特征
all_data_df["year"] = all_data_df["date_t"].dt.year
all_data_df["month"] = all_data_df["date_t"].dt.month
all_data_df["day"] = all_data_df["date_t"].dt.day
# mon -> 0, sun->6
all_data_df["dow"] = all_data_df["date_t"].dt.dayofweek
all_data_df["doy"] = all_data_df["date_t"].dt.dayofyear
all_data_df["is_sa"] = all_data_df["dow"] == 5
all_data_df["is_su"] = all_data_df["dow"] == 6
BC.log("Add more features : year, month, day, dow, doy, is_sa, is_su")
# 处理节假日
all_data_df["is_holiday"] = [
1 if cc.is_holiday(x.date()) else 0 for x in all_data_df.date_t
]
all_data_df["is_holiday"] = (all_data_df["is_sa"] != 1) & (
all_data_df["is_su"] != 1) & (all_data_df["is_holiday"] == 1)
all_data_df["is_holiday"] = [1 if x else 0 for x in all_data_df.is_holiday]
all_data_df["is_WDA"] = [
1 if cc.is_workday(x.date()) else 0 for x in all_data_df.date_t
]
all_data_df["is_WDA"] = (
(all_data_df["is_sa"] == 1) |
(all_data_df["is_su"] == 1)) & (all_data_df["is_WDA"] == 1)
BC.log("Add more features : is_holiday, is_WDA")
# 对比了下，单独划分没有明显效果，根据数据特征分组
all_data_df["is_B12367"] = all_data_df["brand"].isin([1, 2, 3, 6, 7])
all_data_df["is_B410"] = all_data_df["brand"].isin([4, 10])
all_data_df["is_B5"] = all_data_df["brand"].isin([5])
all_data_df["is_B8"] = all_data_df["brand"].isin([8])
all_data_df["is_B9"] = all_data_df["brand"].isin([9])
BC.log("Add more features : is_B12367, is_B410, is_B5, is_B8，is_B9")
# 外部数据
# 增加车牌量
# 处理最后一个月的值
filename = "../Input/car_sale_volume.csv"
car_sales_df = pd.read_csv(filename)
car_sales_df = car_sales_df[["sale_quantity", "year", "month"]]
BC.log("Read cas sales : " + filename)
# 最后一个月的数据处理
df_ = pd.Series([car_sales_df[-3:]["sale_quantity"].mean(), 2017, 11],
index=["sale_quantity", "year", "month"]).to_frame()
car_sales_df = pd.concat([car_sales_df, df_.T])
# car_sales_df.reset_index(inplace=True)
all_data_df = pd.merge(all_data_df, car_sales_df, on=["year", "month"])
all_data_df.drop("day_of_week", inplace=True, axis=1)
all_data_df["is_Jan"] = all_data_df["month"] == 1
all_data_df["is_Feb"] = all_data_df["month"] == 2
BC.log("Add more features : is_Jan, is_Feb")
filename = BC.outputFolder + "all_data.csv"
all_data_df.to_csv(filename)
BC.log("Write all_data_df to " + filename)
return all_data_df
def test_chinesecalendar():
"""
测试判断chinese_calendar这个库CC
:return:
"""
test_day = datetime.date(2019, 10, 1)
print("是工作日：%s" % is_workday(test_day))
print("是假日：%s" % is_holiday(test_day))
def __get_datetime_range(cls, current, offset):
'''
获得时间的上下限，目标是找到一个日期，该日期是有休息日变成工作日的界限
offset为负数时，则反过来
:param current:当前日期
:param offset:偏移量
:return:
'''
while True:
another_day = cls.__get_offset_date(current, offset)
current_is_holiday = cc.is_holiday(current)
another_day_is_holiday = cc.is_holiday(another_day)
if not another_day_is_holiday and current_is_holiday:
return current
else:
current = another_day
before_day = cls.__get_offset_date(current, offset)
def is_holidays(date):
'''
判断是否为节假日
'''
Y = date.year
M = date.month
D = date.day
april_last = datetime.date(Y, M, D)
return is_holiday(april_last)
def send_set_time():
now = datetime.datetime.now()
set_time = [
'10:00', '10:30', '11:00', '11:30', '13:30', '14:00', '14:30', '15:00'
]
time1_str = datetime.datetime.strftime(now, '%Y-%m-%d %H:%M:%S')[-8:-3]
if time1_str in set_time and is_holiday(now) == False:
return True
else:
return False
def test_workdays(self):
dates = [
datetime.date(year=2017, month=5, day=27),
datetime.date(year=2017, month=6, day=5),
datetime.date(year=2017, month=10, day=31),
datetime.date(year=2018, month=5, day=10),
]
for date in dates:
self.assertFalse(is_holiday(date))
self.assertTrue(is_workday(date))
def add_holiday_info(cls, vol: pd.DataFrame):
'''
为数据添加是否为节假日的信息
:param vol: 原始的数据
:return: 添加了'whether_holiday_filter'列，1表示这一天是节假日，为0则表示不是节假日
'''
vol = copy.deepcopy(vol)
vol[Columns.Is_holiday] = pd.to_datetime(
vol[Columns.Date]).apply(lambda t: {
True: 1,
False: 0
}[cc.is_holiday(t)])
return vol
def alarm_clock():
date = datetime.date.today()
localtime = time.localtime(time.time())
if is_holiday(date):
text = holiday_msg(localtime)
elif is_workday(date):
text = workday_msg(localtime)
if text is not None:
msg_body = {"msgtype": "text",
"text": {"content": text},
"at": { "isAtAll": True}
}
ret = requests.post(ding_url, json = msg_body)
def deal_TestInput(testFilePath):
data = pd.read_csv(testFilePath)
deal_data = data.copy(deep=True)
deal_data['date'] = pd.to_datetime(deal_data['date'])
deal_data['week'] = deal_data['date'].dt.dayofweek
deal_data['month'] = deal_data['date'].dt.month
deal_data['year'] = deal_data['date'].dt.year
deal_data['is_workday'] = deal_data['date'].apply(
lambda x: 1 if is_workday(pd.to_datetime(x)) else 0)
deal_data['is_holiday'] = deal_data['date'].apply(
lambda x: 1 if is_holiday(pd.to_datetime(x)) else 0)
deal_data.drop(['date'], axis=1, inplace=True)
deal_data.to_csv('data/dealed_testInput.csv', index=False)
def test_same_code_as_readme(self):
import datetime
# Check if 2018-04-30 is holiday in China
from chinese_calendar import is_workday, is_holiday
april_last = datetime.date(2018, 4, 30)
self.assertFalse(is_workday(april_last))
self.assertTrue(is_holiday(april_last))
# or check and get the holiday name
import chinese_calendar as calendar # with different import style
on_holiday, holiday_name = calendar.get_holiday_detail(april_last)
self.assertTrue(on_holiday)
self.assertEqual(calendar.Holiday.labour_day.value, holiday_name)
def holiday_and_dev(self, line):
"""
判断是否为假期和开发服务器，同时为True才返回holiday。
line为传进来的每一行日志。
:return:
"""
now = datetime.now()
now2 = now + timedelta(hours=15) # 需要加15小时时差才能获取到服务器上最新log的日期
dev = ["devel", "beta"]
for hostname in dev:
try:
# 判断是假期的同时满足是开发服务器
if is_holiday(now2) and hostname in line:
return "holiday"
except NotImplementedError:
return "NotImplementedError"
def isHolidayOrWeekend(day):
"""
判断是否是星期六 0-周一，1-周二.....5-周六，6-周日, 返回值是<int>类型
:param day: %Y%m%d
:return:
"""
april_last = datetime.date(int(day[:4]), int(day[4:6]), int(day[6:]))
if is_holiday(april_last):
return True
day_date = datetime.datetime.strptime(day, '%Y%m%d')
day_week = day_date.weekday()
# print('new_time:', new_time, ' day_week: ', day_week)
if day_week == 5 or day_week == 6:
return True
return False
def getweekday(DataSet):
dataset_test = DataSet
dataset_test['is_weekend'] = dataset_test['day_of_week'].map(
lambda x: 1 if x == 5 or x == 6 else 0)
weekday_dummies = pd.get_dummies(dataset_test['day_of_week'])
weekday_dummies.columns = [
'weekday' + str(i) for i in range(weekday_dummies.shape[1])
]
dataset_test = pd.concat([dataset_test, weekday_dummies], axis=1)
#新加的特征
dataset_test['is_holiday'] = dataset_test['date_received'].map(
lambda x: 1 if is_holiday(x) else 0)
dataset_test['is_workday'] = dataset_test['date_received'].map(
lambda x: 1 if is_workday(x) else 0)
return dataset_test
def __supplementary_days(cls, data: pd.DataFrame):
'''补充时间，使得数据中的时间以完整的假期开始，以完整的假期结束'''
# 对数据按照时间进行排序
df = copy.deepcopy(data)
df.index = pd.to_datetime(df[Columns.Date])
del df[Columns.Date]
df.sort_index(inplace=True)
df.reset_index(inplace=True)
upday = cls.__get_datetime_range(df.iloc[0][Columns.Date], -1)
lowerday = cls.__get_datetime_range(df.iloc[-1][Columns.Date], 1)
newdf = pd.DataFrame({Columns.Date: pd.date_range(upday, lowerday)})
result = pd.merge(newdf, df, how='left', on=[Columns.Date])
result[Columns.Is_holiday] = pd.to_datetime(result[Columns.Date])\
.apply(lambda t: {True:int(1),False:int(0)}[cc.is_holiday(t)])
if Columns.OutpatientVol in result.columns:
result[Columns.OutpatientVol].fillna(0, inplace=True)
return result
def test_holiday_and_dev(line):
"""
判断是否为假期和开发服务器，同时为True才返回True。
:return: 布尔值
"""
now = datetime.datetime.now()
now2 = now + datetime.timedelta(hours=15) # 需要加15小时时差才能获取到服务器上最新log的日期
test_day = datetime.date(
2019, 10, 2) # 测试的时候修改这里, 分别测2019，10，2；2019，10，8；2020，10，2之类
dev = ["devel", "beta"]
for hostname in dev:
try:
if is_holiday(test_day) and hostname in line:
# 测试的时候使用 if is_holiday(test_day) and hostname in line:
return "holiday"
except NotImplementedError:
return "NotImplementedError"
def clockIn(userId, clockTime):
res = Record.query.filter(
and_(
Record.time.like('{}%'.format(
clockTime.isoformat(' ').split()[0])),
Record.user_id == userId)).order_by(Record.time).all()
print('a1', res)
# 判断当天是否已打卡两次
if len(res) > 1:
return False
data = Record(userId, clockTime)
db.session.add(data)
# 法定假日
if is_holiday(clockTime):
if len(res) == 1:
res = Record.query.filter(
and_(
Record.time.like('{}%'.format(
clockTime.isoformat(' ').split()[0])),
Record.user_id == userId)).order_by(Record.time).all()
dura = (res[1].time - res[0].time).total_seconds()
dura = format(dura / 3600., '.2f')
data = Summary(res[0].user_id, res[0].time.year, res[0].time.month,
res[0].time.day, dura)
db.session.add(data)
db.session.commit()
return True
else: # 工作日
# if clockoff, compute overtime duration
hms = clockTime.time()
if hms > datetime.time.fromisoformat('17:00:00'):
assert len(res) == 1
res = Record.query.filter(
and_(
Record.time.like('{}%'.format(
clockTime.isoformat(' ').split()[0])),
Record.user_id == userId)).order_by(Record.time).all()
print("dura: ", res)
dura = computeDurationPerDay(res[0].time, res[1].time)
data = Summary(res[0].user_id, res[1].time.year, res[1].time.month,
res[1].time.day, dura)
db.session.add(data)
db.session.commit()
return True
def add_overtime(e_id, time_len, ot_reason, ot_date):
'''
添加加班记录
Args:
e_id : 员工编号
time_len：加班时长
ot_reason:加班理由
ot_date:加班日期
Returns:
void
'''
sign = is_holiday(ot_date)
e = User.objects.get(e_id=e_id)
add_ot = Overtime(ot_date=ot_date,
time_length=time_len,
o_reason=ot_reason,
is_holiday=sign,
o_emp=e)
add_ot.save()
def build_date_fea_defult(df, date_col):
""" 构建日期相关特征.
为一个DataFrame的日期列构造日期相关特征,
构建的特征有:
['year', 'month', 'day','dayofyear',
'dayofweek', 'daysinmonth', 'is_leap_year',
'is_in_lieu', 'is_holiday', 'is_workday'].
Args:
df: 输入的DataFrame.
date_col: 日期列名,为一个字符串,需存在于df中,对应列可以是字符串格式或者日期格式.
Returns:
df: 构建完成的新的DataFrame.
"""
year, month, day, dayofyear, \
dayofweek, daysinmonth, is_leap_year, \
is_in_lieu, is_holiday, is_workday = create_empty_list(10)
for t in df[date_col]:
t = pd.to_datetime(t)
year.append(t.year)
month.append(t.month)
day.append(t.day)
dayofyear.append(t.dayofyear)
dayofweek.append(t.dayofweek)
daysinmonth.append(t.daysinmonth)
is_leap_year.append(int(t.is_leap_year))
is_in_lieu.append(int(cc.is_in_lieu(t)))
is_holiday.append(int(cc.is_holiday(t)))
is_workday.append(int(cc.is_workday(t)))
df['year'] = year
df['month'] = month
df['day'] = day
df['dayofyear'] = dayofyear
df['dayofweek'] = dayofweek
df['daysinmonth'] = daysinmonth
df['is_leap_year'] = is_leap_year
df['is_in_lieu'] = is_in_lieu
df['is_holiday'] = is_holiday
df['is_workday'] = is_workday
return df
def Get_external_dataset(start_time, stop_time):
time_slots = []
time_slots_between = []
index_to_delete = [27] # 10min：91 20min：43 30min：27
num_days = (stop_time - start_time).days
while start_time <= stop_time:
time_slots.append(start_time)
start_time = start_time + datetime.timedelta(
minutes=30) # 10min 20min 30min
for time in time_slots:
hour = time.hour
minute = time.minute
if is_time_between(hour, minute):
time_slots_between.append(time)
for i in range(1, num_days + 1):
index = 27 + i * 33 #10min：91 97 20min：43 49 30min：27 33
index_to_delete.append(index)
time_slots_between = [
time_slots_between[i] for i in range(0, len(time_slots_between), 1)
if i not in index_to_delete
]
external_feature = []
for time in time_slots_between:
day = time.day
cut_time = '2015-1-' + str(day)
temp_f = f[cut_time]
diff = temp_f.index - time
temp_f['diff'] = abs(diff)
idx = temp_f['diff'].idxmin()
row = temp_f.loc[idx] # 该时段的天气特征（数值 已min-max归一化） 6维 + 时间差（不用）
dim_0 = int(is_workday(time))
dim_1 = int(is_holiday(time)) # 该时间段的节假日特征（one-hot编码） 2维 [是否工作日，是否节假日]
feature = np.array(
(dim_0, dim_1, row[0], row[1], row[2], row[3], row[4], row[5]))
feature = tile(feature, (76, 1)) # 76个站点， 同一个时间external feature一样
external_feature.append(feature)
external_feature = np.array(external_feature)
return external_feature
def create_casual(casual_start, casual_end, casual_reason):
'''
经理创建临时加班
Args:
casual_start:开始时间
casual_end:结束时间
casual_reason:加班理由
Returns:
void
'''
c_date = casual_start.date()
sign = is_holiday(c_date)
cas = Casual(c_reason=casual_reason,
c_begin=casual_start,
c_end=casual_end,
is_holiday=sign)
cas.save()
emps = User.objects.all()
for e in emps:
emp_cas = Emp_Casual(casual=cas, emp=e)
emp_cas.save()
def get_tradingday(start_day,end_day):
start_week = start_day.isoweekday()
days = []
end = False
global number
for day in range(6-start_week):
t = start_day+timedelta(days=day)
if t == end_day:
end = True
day = 6 - start_week - 1
if t.isoweekday() < 6:
if not is_holiday(t):
days.append(t)
if day == 6 - start_week - 1:
if days:
result[str(number)] = days
number = number + 1
if end:
break
next_weekday = t + timedelta(days=3)
get_tradingday(next_weekday,end_day)
def test_same_code_as_readme_english(self):
import datetime
# Check if 2018-04-30 is holiday in China
from chinese_calendar import is_holiday, is_workday
april_last = datetime.date(2018, 4, 30)
assert is_workday(april_last) is False
assert is_holiday(april_last) is True
# or check and get the holiday name
import chinese_calendar as calendar # 也可以这样 import
on_holiday, holiday_name = calendar.get_holiday_detail(april_last)
assert on_holiday is True
assert holiday_name == calendar.Holiday.labour_day.value
# even check if a holiday is in lieu
import chinese_calendar
assert chinese_calendar.is_in_lieu(datetime.date(2006, 2, 1)) is False
assert chinese_calendar.is_in_lieu(datetime.date(2006, 2, 2)) is True
def daily_check():
if is_holiday(
datetime.datetime.now(
datetime.timezone(datetime.timedelta(hours=8))).date()):
logging.info('Today is holiday')
else:
run_after = random.randint(0, 540)
logging.info('Auto Check in/out start in ' + str(run_after) + 'secs')
time.sleep(run_after)
try:
uid, token = login()
check(uid, token)
except LoginException as e:
logging.exception(e)
emergence_trigger()
except HTTPError as e:
logging.exception(e)
emergence_trigger()
except CheckException as e:
logging.exception(e)
emergence_trigger()
def start_date_match(start_date, end_date, **kw):
# 生成初始日期序列
rst_date = pd.DataFrame({"date": [], "count": []})
last_count = 0
for tryDate in pd.date_range(start=startDatein, end=endDatein):
# print(x.dayofweek) #0-6 -> 1-7
train_by_date_zeromatch_df = kw["train_by_date_df"]
if tryDate.dayofweek == 2:
train_by_date_zeromatch_df[
"date_t"] = utils.Briefcase.gen_date_by_dows(
tryDate, train_by_date_zeromatch_df["dow_diff"])
train_by_date_zeromatch_df["is_holiday"] = [
1 if cc.is_holiday(x.date()) else 0
for x in train_by_date_zeromatch_df["date_t"]
]
zeromatch_idx = train_by_date_zeromatch_df.loc[:,
"day_of_week"].isin(
[6, 7])
for x in range(len(zeromatch_idx)):
if zeromatch_idx[x] == True:
train_by_date_zeromatch_df.drop(x, inplace=True)
curr_count = train_by_date_zeromatch_df["is_holiday"].sum()
if last_count == 0:
last_count = curr_count
min_count = curr_count
min_count = min(min_count, curr_count)
last_count = curr_count
print(tryDate.date(), last_count, min_count)
df = pd.DataFrame({
"date": [tryDate],
"count": [train_by_date_zeromatch_df["is_holiday"].sum()]
})
rst_date = pd.concat([rst_date, df], ignore_index=True)
rst_date.describe()
return rst_date[rst_date["count"] == rst_date["count"].min()]
def test_same_code_as_readme_chinese(self):
import datetime
# 判断 2018年4月30号 是不是节假日
from chinese_calendar import is_holiday, is_workday
april_last = datetime.date(2018, 4, 30)
assert is_workday(april_last) is False
assert is_holiday(april_last) is True
# 或者在判断的同时，获取节日名
import chinese_calendar as calendar # 也可以这样 import
on_holiday, holiday_name = calendar.get_holiday_detail(april_last)
assert on_holiday is True
assert holiday_name == calendar.Holiday.labour_day.value
# 还能判断法定节假日是不是调休
import chinese_calendar
assert chinese_calendar.is_in_lieu(datetime.date(2006, 2, 1)) is False
assert chinese_calendar.is_in_lieu(datetime.date(2006, 2, 2)) is True
