def test_catch_infinite_loop(self):
offset = offsets.DateOffset(minute=5)
# blow up, don't loop forever
msg = "Offset <DateOffset: minute=5> did not increment date"
with pytest.raises(ValueError, match=msg):
date_range(datetime(2011, 11, 11), datetime(2011, 11, 12),
freq=offset)
def test_catch_infinite_loop(self):
offset = offsets.DateOffset(minute=5)
# blow up, don't loop forever
pytest.raises(Exception,
date_range,
datetime(2011, 11, 11),
datetime(2011, 11, 12),
freq=offset)
offsets.BMonthEnd(),
offsets.BMonthBegin(),
offsets.CustomBusinessDay(),
offsets.CustomBusinessDay(calendar=hcal),
offsets.CustomBusinessMonthBegin(calendar=hcal),
offsets.CustomBusinessMonthEnd(calendar=hcal),
offsets.CustomBusinessMonthEnd(calendar=hcal),
]
other_offsets = [
offsets.YearEnd(),
offsets.YearBegin(),
offsets.QuarterEnd(),
offsets.QuarterBegin(),
offsets.MonthEnd(),
offsets.MonthBegin(),
offsets.DateOffset(months=2, days=2),
offsets.BusinessDay(),
offsets.SemiMonthEnd(),
offsets.SemiMonthBegin(),
]
offset_objs = non_apply + other_offsets
class OnOffset:
params = offset_objs
param_names = ["offset"]
def setup(self, offset):
self.dates = [
datetime(2016, m, d) for m in [10, 11, 12]
def __getitem__(self, index):
current_frame_time = pd.to_datetime(
self.data_list[index].split(" ")[2] + " " +
self.data_list[index].split(" ")[3])
#print(current_frame_time)
time_index = int(current_frame_time.microsecond / 50000)
start_time = pd.to_datetime(
str(current_frame_time)[:-7] + "." +
str(time_index * 50000).zfill(6))
#print(start_time)
next_frame_time = start_time + offsets.DateOffset(microseconds=50000)
#print(next_frame_time)
# start_time = str(start_time.hour).zfill(2) + ":" + str(start_time.minute).zfill(2) + ":" + str(start_time.second).zfill(2) + "." + str(start_time.microsecond).zfill(6)
# end_time = str(next_frame_time.hour).zfill(2) + ":" + str(next_frame_time.minute).zfill(2) + ":" + str(next_frame_time.second).zfill(2) + "." + str(next_frame_time.microsecond).zfill(6)
# print("'start_time', {}, 'end_time', {}, 'len', {}".format(start_time, end_time, len(self.current_csi_dat[start_time:end_time])))
#print(self.csi_dat[self.data_list[index].split(" ")[0][:-6]])
#print(self.csi_dat)
while (len(
np.array(self.csi_dat[self.data_list[index].split(" ")[0][:-6]]
[str(start_time):str(next_frame_time)])) < 5):
next_frame_time = next_frame_time + offsets.DateOffset(
microseconds=50000)
csi_temp = self.csi_dat[self.data_list[index].split(
" ")[0][:-6]][str(start_time):str(next_frame_time)][:5]
# print(csi_temp[0])
# csi_temp = np.array(csi_temp)
# print(np.concatenate(np.array(csi_temp), axis=2).shape)
# csi_temp = np.absolute(np.concatenate(np.array(csi_temp)))
# print(csi_temp.shape)
# csi_temp = csi_temp.transpose(0 ,2, 1)
# print(csi_temp.shape)
# # sample_csi = np.absolute(csi_temp)
# sample_csi = np.resize(csi_temp, (150, 3, 3))
#print(np.concatenate(csi_temp, axis=2))
sample_csi = np.concatenate(csi_temp, axis=2).transpose(2, 0, 1)
#print(sample_csi.shape)
#print(self.root_dir)
#print(self.data_list[index].split(" ")[:1][0])
#print(type(self.data_list[index].split(" ")[:1][0]))
sample_SM = sio.loadmat(
join(
self.root_dir, "mask_resize", self.data_list[index].split(" ")
[:1][0] + "_" + str(index + 2)) + ".mat")['masks']
sample_SM = self.pose_transform(sample_SM)
#转变float64 torch.size(1,46,82)
sample_SM = sample_SM.double()
#print(sample_SM.shape)
json_dir = join(self.root_dir, 'res', 'alphapose_results.json')
with open(json_dir) as f:
# print(f)
sample = json.load(f)
JHM = sample[index]["keypoints"]
JHM = np.array(JHM)
sample_JHMs = get_heatmap((1280, 720), JHM, (82, 46))[:, :, :-1]
sample_PAFs = get_vectormap((1280, 720), JHM, (82, 46))
'''
x = sample_PAFs.transpose((2,0,1))
print(x[0])
cv.namedWindow('input_image', cv.WINDOW_AUTOSIZE)
mask = 255*x[10]
mask = mask.astype(np.uint8)
cv.imshow('input_image', mask)
cv.waitKey(0)
cv.destroyAllWindows()
'''
sample_PAFs = self.pose_transform(sample_PAFs)
sample_JHMs = self.pose_transform(sample_JHMs)
sample_PAFs = sample_PAFs.double()
sample_JHMs = sample_JHMs.double()
#print(sample_JHMs.shape)
#print(sample_PAFs.shape)
# 转变float64 sample_JHMs torch.size(17,46,82) sample_PAFs torch.size(36,46,82)
video_name = self.data_list[index].split(" ")[0]
frame_number = self.data_list[index].split(" ")[1]
sample = {
'csi': sample_csi,
'JHMs': sample_JHMs,
'SM': sample_SM,
'video': video_name,
'PAFs': sample_PAFs,
'frame': frame_number
}
if self.transforms:
sample = self.transforms(sample)
return sample
else:
print('Values match')
print('\n')
#---------------------------------------------------------------------------------------------------
# profiling
#---------------------------------------------------------------------------------------------------
from pandas import concat
from pandas import offsets
import timeit
from pandas import to_datetime
df_big = read_csv(r'.\inputs\PD 2022 Wk 1 Input - Input.csv', parse_dates=['Date of Birth'],
usecols=['id', 'pupil first name', 'pupil last name', 'Date of Birth'])
df_big = concat([df_big]*1000)
# using replace
%timeit -n 1 -r 100 df_big['This Year\'s Birthday'] = df_big['Date of Birth'].apply(lambda x: x.replace(year=datetime.now().year))
# using offsets
%timeit -n 1 -r 100 df_big['This Year\'s Birthday'] = df_big['Date of Birth'] + offsets.DateOffset(year=2022)
# using string formatting
%timeit -n 1 -r 100 df_big['This Year\'s Birthday'] = to_datetime('2022-' + df_big['Date of Birth'].dt.strftime('%m-%d'))