def _get_business_from_caseone(self, caseone):
resp = {}
exist_keys = self._get_business_names()
setupId = caseone.get('SetupID', '')
processId = caseone.get("ProcessID", [])
teardownId = caseone.get('TeardownID', '')
keys = list(filter(None, list(set(processId + [setupId, teardownId]))))
import copy
keys_copy = copy.deepcopy(keys)
for key in keys_copy:
if key in exist_keys:
keys.remove(key)
for key in keys:
if key == setupId:
doc = caseone.get("SetupDoc", "")
param = ""
resp.update({
key: {
'businessID': key,
"Documentation": doc,
"Params": param
}
})
if key == teardownId:
doc = caseone.get("TeardownDoc", "")
param = ""
resp.update({
key: {
'businessID': key,
"Documentation": doc,
"Params": param
}
})
if key in processId:
index = processId.index(key)
doc = caseone.get("ProcessDoc", [])[index]
param = caseone.get("Params", [])[index]
resp.update({
key: {
'businessID': key,
"Documentation": doc,
"Params": param
}
})
return resp
'copy.deepcopy(x)' '''一般赋值中，是把变量指向某内存，当修改该变量时（除非是重新指向其他内存）， 会影响到所指向的内存中的变量，赋值时，用deepcopy即可避免内存变量的变动''' import copy a = [1,2,3] b = a b[0] = 'modified' print(a) a1 = [2,3,4] b1 = [] b1.append(a1) b1[-1][0] = 'modified' print(a1) a2 = [6, 6, 6] b2 = copy.deepcopy(a2) b2[0] = 'modified' print(a2) '# 结束----------------------------' 'pcolor(xx,yy,zz)' '''云图，xx,yy为二维数据，一般通过等间隔一维数组用Meshgrid产生，指示平面内每个点的横纵坐标， zz和xx,yy形式相同，表示每个坐标点上的z值''' x = numpy.arange(-2, 3, 1) y = numpy.arange(-2, 3, 1) xx, yy = numpy.meshgrid(x, y) zz = xx + yy qq=plt.pcolor(xx, yy,zz, cmap='jet') '# 结束----------------------------' 'plt.pie'
def vinicial(X1, X2, lv, y1, y2):
X11 = copy.deepcopy(X1)
X22 = copy.deepcopy(X2)
X11 = np.append(X11, np.zeros(lv - len(X11) % lv))
X22 = np.append(X22, np.zeros(lv - len(X22) % lv))
l = len(X11) / lv
rms1 = float(np.sqrt(np.mean(X11**2)))
rms2 = float(np.sqrt(np.mean(X22**2)))
X11 = X11 / rms1
X22 = X22 / rms2
X11 = X11[:l * lv]
X111 = copy.deepcopy(X11)
X22 = X22[:l * lv]
nl = np.floor(np.log2(len(X11) / lv)) #Numero de loops
cmax = np.zeros(2)
cpos = np.zeros(2)
Cposa = 0
X2pos = 0 # Posicion inicial del segmento final de X2
j = 0
while j < nl:
l2 = np.round(len(X111) / 2)
for i in range(2):
u1 = l2 * i
u2 = u1 + l2
X2C = X22[u1:u2] # Corta la senal modificada en partes igualles
Y = signal.correlate(
X111, X2C, 'full', 'fft'
) # Correlacion entre senal modificada cortada y de referencia
cmax[i] = float(
max(Y) /
float(len(Y))) # Vector de valores maximos de correlacion
s = len(Y) / 2
cpos[i] = np.argmax(
Y) - s # Vector de posicion de valores maximos de correlacion
Cposr = np.argmax(cmax) # Maxima posicion relativa
X2pos = Cposr * l2 + X2pos
Cposa = int(cpos[Cposr]) + Cposa # Maxima posicion absoluta
X22 = X22[l2 * Cposr:l2 * (Cposr + 1)]
# print Cposa
mp = len(
X11
) / 2 + Cposa - l2 / 2 # Posicion inicial del vector X111 con respecto a X11
if mp >= 0:
X111 = X11[mp:mp + l2]
else:
X111 = np.append(np.zeros(abs(mp)), X11[:l2 + mp])
# c = signal.correlate(X111,X22,'full','fft')
# print(np.argmax(c)-len(c)/2)
j += 1
n = mp - X2pos
if n < 0:
z = np.zeros(abs(n))
X1 = np.append(z, X1)
y1 = np.append(z, y1)
X2 = np.append(X2, z)
y2 = np.append(y2, z)
r = X2pos
elif n > 0:
z = np.zeros(abs(n))
X2 = np.append(z, X2)
y2 = np.append(z, y2)
X1 = np.append(X1, z)
y1 = np.append(y1, z)
r = mp
return X1, X2, r, y1, y2
def cor2(X1, X2, lv, p, g, mv, fs, CCM, u):
"""
Correlacion cruzada de audios.
lv es la longitud de la ventana temporal a utilizar.
p es el porcentaje de lv donde transiciona entre 0 y 1.
g es el umbral del gate
mv es la maxima velocidad de desplazamiento entre microfonos
fs es la frecuencia de muestreo
CCM es el coeficiente de correlacion minima
"""
X11 = copy.deepcopy(X1)
X12 = copy.deepcopy(X2)
X11 = np.append(X11, np.zeros(lv - len(X11) % lv))
X12 = np.append(X12, np.zeros(lv - len(X12) % lv))
################## Ventana temporal #########################
w = np.blackman(round(lv * p)) # Genera la ventana blackman
w1 = np.append(w[:len(w) / 2], np.ones(int(round(lv * (1 - p)))))
w1 = np.append(w1,
w[len(w) / 2:]) # Crea la ventana temporal de longitud lv
################## Generacion de arrays del bucle for #######
l = int(len(X11) / float(lv)) # Cantidad de ventanas temporales/ cilos for
ta = np.zeros(l) # Vector de corrimentos absolutos
CC = np.zeros(l) # Vector de maximas correlaciones por ventana
M = np.ones(
l) # Vector de coeficientes de correccion por umbral de amplitud
N = np.ones(
l) # Vector de coeficientes de correccion por umbral de correlacion
################## Maximo Corrimiento Posible ###############
h = 2.0 # Multiplicador del maximo corrimiento
mcs = math.ceil(fs * mv /
343.0) # Maximo corrimiento en muestras por segundo
mcv = int(math.ceil(lv * mcs / fs) *
h) # Maximo corrimiento en muestras por ventana
mcvl = int(math.ceil(mcv / 2.0)) # Maximo corrimiento lateral
################## Parametros del bucle #####################
m = 3 # Coeficiente de correccion por umbral de amplitud
mm = 5 # Maximo valor de m
n = 10 # Coeficiente de correccion por umbral de correlacion
nm = 5 # Maximo valor de n
tt = 0 # Parametro de ajuste para la correlacion
rms1 = float(np.sqrt(np.mean(X11**2))) # Valor rms de la senal X11
rms2 = float(np.sqrt(np.mean(X12**2))) # Valor rms de la senal X12
################## Bucle de correlacion #####################
z = np.arange(u / lv - 1, -1, -1)
t0 = np.zeros(len(z)) # Vector de corrimientos relativos a i-1
for i in z: # Calculo de corrimientos a la izquierda de la ventana inicial
lv1 = int(lv * i) # Limite inferior
lv2 = int(lv * (i + 1)) # Limite superior
Y1 = X11[lv1:lv2] * w1 # Senal de referencia ventaneada en tiempo
RMS1 = float(np.sqrt(np.mean(
Y1**2))) # Valor RMS de la ventana de referencia
Y1 = Y1 / RMS1 # Senal de referencia normalizada a su valor RMS
if RMS1 / rms1 > g: # Gate de la senal de referencia
r = int(np.ceil(
mcvl * m *
n)) # Cantidad de muestras del corrimiento lateral por ventana
c = np.zeros(int(np.ceil(mcv * m * n) +
1)) # Vector de correlaciones por ciclo del bucle
for j in range(-r, r + 1): # Ventaneo de la senal desplazada
if i == 0 and j < 0: # Condicion inicial de corrimiento hacia la izquierda
Y2 = np.append(np.zeros(abs(j)),
X12[:int(lv2 + j)]) * w1 # Ventana Y2
RMS2 = float(np.sqrt(np.mean(Y2**2))) # Valor RMS de Y2
Y2 = Y2 / RMS2 # Y2 Normalizada a su valor RMS
elif i == 0 and j >= 0: # Condicion inicial de corrimiento hacia la derecha
Y2 = X12[j:int(lv2 + j)] * w1 # Ventana Y2
RMS2 = float(np.sqrt(np.mean(Y2**2))) # Valor RMS de Y2
Y2 = Y2 / RMS2 # Y2 Normalizada a su valor RMS
elif len(X12[int(lv1 - tt -
j):int(lv2 - tt -
j)]) == lv: # Condicion normal
Y2 = X12[int(lv1 - tt - j):int(lv2 - tt -
j)] * w1 # Ventana Y2
RMS2 = float(np.sqrt(np.mean(
Y2**2))) # Valor cuadratico medio de Y2
Y2 = Y2 / RMS2 # Y2 Normalizada a su valor RMS
else:
T = len(X12[int(lv1 - tt - j):int(lv2 - tt -
j)]) # Ultima ventana
TT = np.append(X12[int(lv1 - tt - j):int(lv2 - tt - j)],
np.zeros(lv - T)) # Ventana Y2
Y2 = TT * w1
RMS2 = float(np.sqrt(np.mean(
Y2**2))) # Valor cuadratico medio de Y2
Y2 = Y2 / float(RMS2) # Y2 Normalizada a su valor RMS
if RMS2 / rms2 > g: # Gate de la senal desplazada
m = 1 # Reset del coeficiente de corrimiento maximo por amplitud
c[j + r] = signal.correlate(
Y1, Y2, 'valid', 'fft') # Correlacion de senales
C = max(c) / float(lv) # Valor de correlacion maxima
if C >= CCM: # Condicion para valores de correlacion por debajo del umbral CCM
cm = np.argmax(
c) - r # Posicion del valor de maxima correlacion
t0[i] = cm
n = 1 # Reset del coeficiente de corrimiento maximo por correlacion
CC[i] = C # Vector de valores maximos de correlacion
else:
cm = 0
CC[i] = CCM
if n < nm:
n += 1
N[i] = n # Vector de coeficientes de correccion por umbral de correlacion
else:
t0[i] = 0
if m < mm:
m += 1
else:
t0[i] = 0
if m < mm:
m += 1
M[i] = m # Vector de coeficientes de correccion por umbral de amplitud
####### Vectores de corrimientos tt y ta ########
tt = sum(t0)
ta[i] = tt
t1 = np.zeros(2 + l - u / lv) # Vector de corrimientos relativos a i-1
m = 1 # Coeficiente de correccion por umbral de amplitud
n = 10 # Coeficiente de correccion por umbral de correlacion
tt = 0 # Parametro de ajuste para la correlacion
z = np.arange(u / lv, l)
for i in z:
lv1 = int(lv * i) # Limite inferior
lv2 = int(lv * (i + 1)) # Limite superior
Y1 = X11[lv1:lv2] * w1 # Senal de referencia ventaneada en tiempo
RMS1 = float(np.sqrt(np.mean(
Y1**2))) # Valor RMS de la ventana de referencia
Y1 = Y1 / RMS1 # Senal de referencia normalizada a su valor RMS
if RMS1 / rms1 > g: # Gate de la senal de referencia
r = int(np.ceil(
mcvl * m *
n)) # Cantidad de muestras del corrimiento lateral por ventana
c = np.zeros(int(np.ceil(mcv * m * n) +
1)) # Vector de correlaciones por ciclo del bucle
for j in range(-r, r + 1): # Ventaneo de la senal desplazada
if i == 0 and j < 0: # Condicion inicial de corrimiento hacia la izquierda
Y2 = np.append(np.zeros(abs(j)),
X12[:int(lv2 + j)]) * w1 # Ventana Y2
RMS2 = float(np.sqrt(np.mean(Y2**2))) # Valor RMS de Y2
Y2 = Y2 / RMS2 # Y2 Normalizada a su valor RMS
elif i == 0 and j >= 0: # Condicion inicial de corrimiento hacia la derecha
Y2 = X12[j:int(lv2 + j)] * w1 # Ventana Y2
RMS2 = float(np.sqrt(np.mean(Y2**2))) # Valor RMS de Y2
Y2 = Y2 / RMS2 # Y2 Normalizada a su valor RMS
elif len(X12[int(lv1 - tt -
j):int(lv2 - tt -
j)]) == lv: # Condicion normal
Y2 = X12[int(lv1 - tt - j):int(lv2 - tt -
j)] * w1 # Ventana Y2
RMS2 = float(np.sqrt(np.mean(
Y2**2))) # Valor cuadratico medio de Y2
Y2 = Y2 / RMS2 # Y2 Normalizada a su valor RMS
else:
T = len(X12[int(lv1 - tt - j):int(lv2 - tt -
j)]) # Ultima ventana
TT = np.append(X12[int(lv1 - tt - j):int(lv2 - tt - j)],
np.zeros(lv - T)) # Ventana Y2
Y2 = TT * w1
RMS2 = float(np.sqrt(np.mean(
Y2**2))) # Valor cuadratico medio de Y2
Y2 = Y2 / float(RMS2) # Y2 Normalizada a su valor RMS
if RMS2 / rms2 > g: # Gate de la senal desplazada
m = 1 # Reset del coeficiente de corrimiento maximo por amplitud
c[j + r] = signal.correlate(
Y1, Y2, 'valid', 'fft') # Correlacion de senales
C = max(c) / float(lv) # Valor de correlacion maxima
if C >= CCM: # Condicion para valores de correlacion por debajo del umbral CCM
cm = np.argmax(
c) - r # Posicion del valor de maxima correlacion
t1[i - u / lv] = cm
n = 1 # Reset del coeficiente de corrimiento maximo por correlacion
CC[i] = C # Vector de valores maximos de correlacion
else:
cm = 0
CC[i] = CCM
if n < nm:
n += 1
N[i] = n # Vector de coeficientes de correccion por umbral de correlacion
else:
t1[i - u / lv] = 0
if m < mm:
m += 1
else:
t1[i - u / lv] = 0
if m < mm:
m += 1
M[i] = m # Vector de coeficientes de correccion por umbral de amplitud
####### Vectores de corrimientos tt y ta ########
tt = sum(t1)
ta[i] = tt
return t0, t1, ta, l, CC, N, M
def merge(self, to_excel, from_excel_list, from_keys):
#statistic
contract_num = {}
contract_amount = {}
contract_detail = {}
for idx, key in enumerate(from_keys):
from_excel = from_excel_list[idx]
contract_num[key] = {
"this_month": {},
"last_month": {},
"smly": {}
}
contract_amount[key] = {
"this_month": {},
"last_month": {},
"smly": {}
}
contract_detail[key] = {}
smly_num = contract_num[key]["smly"]
smly_num["sum"] = from_excel.get_value(2, "B7", -1, 0, int)
smly_num["carry"] = from_excel.get_value(2, "C7", -1, 0, int)
smly_num["new"] = from_excel.get_value(2, "E7", -1, 0, int)
smly_num["accum"] = from_excel.get_value(2, "D7", -1, 0, int)
thism_num = contract_num[key]["this_month"]
thism_num["sum"] = from_excel.get_value(2, "F7", -1, 0, int)
thism_num["carry"] = from_excel.get_value(2, "G7", -1, 0, int)
thism_num["accum"] = from_excel.get_value(2, "H7", -1, 0, int)
thism_num["new"] = from_excel.get_value(2, "I7", -1, 0, int)
smly_amount = contract_amount[key]["smly"]
smly_amount["sum"] = from_excel.get_value(2, "J7", -1, 0.0, float)
smly_amount["carry"] = from_excel.get_value(
2, "K7", -1, 0.0, float)
smly_amount["accum"] = from_excel.get_value(
2, "L7", -1, 0.0, float)
smly_amount["new"] = from_excel.get_value(2, "M7", -1, 0.0, float)
thism_amount = contract_amount[key]["this_month"]
thism_amount["sum"] = from_excel.get_value(2, "N7", -1, 0.0, float)
thism_amount["carry"] = from_excel.get_value(
2, "O7", -1, 0.0, float)
thism_amount["accum"] = from_excel.get_value(
2, "P7", -1, 0.0, float)
thism_amount["new"] = from_excel.get_value(2, "Q7", -1, 0.0, float)
#details
details = contract_detail[key]
for d_idx in range(8):
lin_name = d_idx + 8
busi_name = from_excel.get_value(2, "A%d" % (lin_name), -1)
details[busi_name] = {
"num": {
"this_month": {},
"last_month": {},
"smly": {}
},
"amount": {
"this_month": {},
"last_month": {},
"smly": {}
}
}
num = details[busi_name]["num"]
num["smly"]["sum"] = from_excel.get_value(
2, "B%d" % lin_name, -1, 0.0, float)
num["smly"]["carry"] = from_excel.get_value(
2, "C%d" % lin_name, -1, 0.0, float)
num["smly"]["accum"] = from_excel.get_value(
2, "D%d" % lin_name, -1, 0.0, float)
num["smly"]["new"] = from_excel.get_value(
2, "E%d" % lin_name, -1, 0.0, float)
num["this_month"]["sum"] = from_excel.get_value(
2, "F%d" % lin_name, -1, 0.0, float)
num["this_month"]["carry"] = from_excel.get_value(
2, "G%d" % lin_name, -1, 0.0, float)
num["this_month"]["accum"] = from_excel.get_value(
2, "H%d" % lin_name, -1, 0.0, float)
num["this_month"]["new"] = from_excel.get_value(
2, "I%d" % lin_name, -1, 0.0, float)
amount = details[busi_name]["amount"]
amount["smly"]["sum"] = from_excel.get_value(
2, "J%d" % lin_name, -1, 0.0, float)
amount["smly"]["carry"] = from_excel.get_value(
2, "K%d" % lin_name, -1, 0.0, float)
amount["smly"]["accum"] = from_excel.get_value(
2, "L%d" % lin_name, -1, 0.0, float)
amount["smly"]["new"] = from_excel.get_value(
2, "M%d" % lin_name, -1, 0, float)
amount["this_month"]["sum"] = from_excel.get_value(
2, "N%d" % lin_name, -1, 0.0, float)
amount["this_month"]["carry"] = from_excel.get_value(
2, "O%d" % lin_name, -1, 0.0, float)
amount["this_month"]["accum"] = from_excel.get_value(
2, "P%d" % lin_name, -1, 0.0, float)
amount["this_month"]["new"] = from_excel.get_value(
2, "Q%d" % lin_name, -1, 0.0, float)
out_ws = to_excel.get_new_sheet(self.title)
out_ws['A1'].font = pxl.styles.Font(name=u'宋体', size=16, bold=True)
out_ws['A1'] = self.title + u'(单位：万元)'
out_ws.merge_cells('A1:K1')
titles = [
u'单位名称', u'本月新签合同额', u'本月累计新签合同额', u'本月合同总额', u'去年同期新签合同额',
u'去年同期累计新签合同额', u'去年同期合同总额', u'本月累计合同份数', u'本月新签合同份数', u'本月合同总份数',
u'去年同期累计合同份数', u'去年同期新签合同份数', u'去年同期合同总份数'
]
cell_name = 'A2'
for i, t in enumerate(titles):
out_ws[cell_name] = t
out_ws.merge_cells(
'%s:%s' % (cell_name, get_next_cell_name(cell_name, True)))
cell_name = get_next_cell_name(cell_name)
start_cell_name = 'A4'
format_start = 'A1'
new_sum_amount = 0
for key in from_keys:
new_sum_amount += contract_amount[key]['this_month']['new']
for key in from_keys:
out_ws[start_cell_name] = key
cur_cell = get_next_cell_name(start_cell_name)
out_ws[cur_cell] = contract_amount[key]['this_month']['new']
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = contract_amount[key]['this_month']['accum']
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = contract_amount[key]['this_month']['sum']
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = contract_amount[key]['smly']['new']
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = contract_amount[key]['smly']['accum']
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = contract_amount[key]['smly']['sum']
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = contract_num[key]['this_month']['accum']
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = contract_num[key]['this_month']['new']
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = contract_num[key]['this_month']['sum']
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = contract_num[key]['smly']['accum']
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = contract_num[key]['smly']['new']
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = contract_num[key]['smly']['sum']
format_end = cur_cell
start_cell_name = get_next_cell_name(start_cell_name, True)
bd = pxl.styles.Side(style='thin', color='000000')
to_excel.style_range(cell_range="%s:%s" % (format_start, format_end), \
border=pxl.styles.Border(left=bd, top=bd, right=bd, bottom=bd), \
fill=None, \
alignment=pxl.styles.Alignment(horizontal='center', vertical='center'))
start_cell_name = get_next_cell_name(start_cell_name, True)
start_cell_name = get_next_cell_name(start_cell_name, True)
out_ws[start_cell_name] = u'本月各公司各类项目合同额'
out_ws[start_cell_name].font = pxl.styles.Font(name=u'宋体',
size=16,
bold=True)
format_start = start_cell_name
busi_names = [
u'测绘地理信息', u'管线工程', u'应用地球物理工程', u'新兴业务航空遥感', u'新兴业务智慧城市',
u'其他新兴业务\n（测绘地理信息类）', u'其他新兴业务\n（管线工程类）', u'其他新兴业务\n（应用地球物理工程类）'
]
titles = copy.deepcopy(busi_names)
titles.insert(0, u'各单位名称')
titles.extend([u'新兴业务新签合同', u'备注'])
out_ws.merge_cells(
'%s:%s' % (start_cell_name,
get_next_cell_name(start_cell_name, False,
len(titles) + len(busi_names) - 1)))
start_cell_name = get_next_cell_name(start_cell_name, True)
cell_name = start_cell_name
format_end = start_cell_name
for i, t in enumerate(titles):
out_ws[cell_name] = t
cell_end = cell_name
if t in busi_names:
cell_end = get_next_cell_name(get_next_cell_name(cell_name))
else:
cell_end = get_next_cell_name(cell_name, True)
out_ws.merge_cells('%s:%s' % (cell_name, cell_end))
if t in busi_names:
first_cell = get_next_cell_name(cell_name, True)
second_cell = get_next_cell_name(
get_next_cell_name(cell_name, True))
third_cell = get_next_cell_name(
get_next_cell_name(get_next_cell_name(cell_name, True)))
out_ws[first_cell] = u'新增合同额'
out_ws[second_cell] = u'今年累计合同额'
out_ws[third_cell] = u'去年累计合同额'
format_end = second_cell
cell_name = get_next_cell_name(
get_next_cell_name(get_next_cell_name(cell_name)))
else:
format_end = get_next_cell_name(cell_name, True)
cell_name = get_next_cell_name(cell_name)
bd = pxl.styles.Side(style='thin', color='000000')
to_excel.style_range("%s:%s" % (start_cell_name, format_end), \
border=pxl.styles.Border(left=bd, top=bd, right=bd, bottom=bd), \
fill=pxl.styles.PatternFill('solid', fgColor='ffff00'), \
font=pxl.styles.Font(name=u'宋体', size=14, bold=True))
start_cell_name = get_next_cell_name(start_cell_name, True)
start_cell_name = get_next_cell_name(start_cell_name, True)
new_busi_info = {}
for key in from_keys:
out_ws[start_cell_name] = key
cur_cell = get_next_cell_name(start_cell_name)
new_busi_sum = 0.0
for busi in busi_names:
out_ws[cur_cell] = contract_detail[key][busi]['amount'][
'this_month']['new']
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = contract_detail[key][busi]['amount'][
'this_month']['accum']
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = contract_detail[key][busi]['amount'][
'smly']['accum']
cur_cell = get_next_cell_name(cur_cell)
if busi.find(u'新兴') >= 0:
new_busi_sum += contract_detail[key][busi]['amount'][
'this_month']['new']
if contract_detail[key][busi]['amount']['this_month'][
'new'] > 0.0:
if key not in new_busi_info:
new_busi_info[key] = []
new_busi_info[key].append(busi)
out_ws[cur_cell] = new_busi_sum
cur_cell = get_next_cell_name(cur_cell)
if key in new_busi_info:
out_ws[cur_cell] = u"、".join(new_busi_info[key])
format_end = cur_cell
start_cell_name = get_next_cell_name(start_cell_name, True)
to_excel.style_range(cell_range="%s:%s" % (format_start, format_end),
alignment=pxl.styles.Alignment(
horizontal='center', vertical='center'))
#every business
busi_new_amount = [0.0] * len(busi_names)
busi_accum_amount = [0.0] * len(busi_names)
all_busi_accum_amount = 0.0
for i, busi in enumerate(busi_names):
for key in from_keys:
busi_new_amount[i] += contract_detail[key][busi]['amount'][
'this_month']['new']
busi_accum_amount[i] += contract_detail[key][busi]['amount'][
'this_month']['accum']
all_busi_accum_amount += contract_detail[key][busi]['amount'][
'this_month']['accum']
start_cell_name = get_next_cell_name(start_cell_name, True)
single_tname = [u'新增合同额', u'今年累计合同额', u'去年累计合同额']
single_key = [
'amount:this_month:new', 'amount:this_month:accum',
'amount:smly:accum'
]
for idx, tname in enumerate(single_tname):
format_start = start_cell_name
out_ws[start_cell_name] = u'按版块各数据单表:%s' % tname
titles = [
u'各单位名称',
]
titles.extend(busi_names)
out_ws.merge_cells('%s:%s' %
(start_cell_name,
get_next_cell_name(start_cell_name, False,
len(titles) - 1)))
start_cell_name = get_next_cell_name(start_cell_name, True)
cell_name = start_cell_name
for t in titles:
out_ws[cell_name] = t
cell_name = get_next_cell_name(cell_name)
start_cell_name = get_next_cell_name(start_cell_name, True)
for key in from_keys:
out_ws[start_cell_name] = key
cur_cell = get_next_cell_name(start_cell_name)
for busi in busi_names:
out_ws[cur_cell] = get_keys(contract_detail[key][busi],
single_key[idx])
format_end = cur_cell
cur_cell = get_next_cell_name(cur_cell)
start_cell_name = get_next_cell_name(start_cell_name, True)
to_excel.style_range(cell_range='%s:%s' % (format_start, format_end), \
border=pxl.styles.Border(left=bd, top=bd, right=bd, bottom=bd), \
fill=None, \
font=pxl.styles.Font(name=u'宋体', size=12), \
alignment=pxl.styles.Alignment(horizontal='center', vertical='center'))
start_cell_name = get_next_cell_name(start_cell_name, True)
format_start = start_cell_name
out_ws[start_cell_name] = u'各类项目本月合同额统计'
out_ws[start_cell_name].font = pxl.styles.Font(name=u'宋体',
size=16,
bold=True)
out_ws[start_cell_name].alignment = pxl.styles.Alignment(
horizontal='center')
titles = [u'各单位名称', u'新签合同额', u'新签合同额占比', u'累计合同额', u'累计合同额占比']
out_ws.merge_cells('%s:%s' %
(start_cell_name,
get_next_cell_name(start_cell_name, False,
len(titles) - 1)))
start_cell_name = get_next_cell_name(start_cell_name, True)
cur_cell = start_cell_name
title_start = start_cell_name
for t in titles:
out_ws[cur_cell] = t
cur_cell = get_next_cell_name(cur_cell)
start_cell_name = get_next_cell_name(start_cell_name, True)
to_excel.style_range("%s:%s" % (title_start, get_next_cell_name(title_start, False, len(titles) - 1)), \
border=pxl.styles.Border(left=bd, top=bd, right=bd, bottom=bd), \
fill=pxl.styles.PatternFill('solid', fgColor='ffff00'), \
font=pxl.styles.Font(name=u'宋体', size=14, bold=True))
for i, busi in enumerate(busi_names):
cur_cell = start_cell_name
out_ws[cur_cell] = busi
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = busi_new_amount[i]
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = division(busi_new_amount[i], new_sum_amount)
out_ws[cur_cell].number_format = '0.00%'
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = busi_accum_amount[i]
cur_cell = get_next_cell_name(cur_cell)
out_ws[cur_cell] = division(busi_accum_amount[i],
all_busi_accum_amount)
out_ws[cur_cell].number_format = '0.00%'
format_end = cur_cell
start_cell_name = get_next_cell_name(start_cell_name, True)
to_excel.style_range(cell_range="%s:%s" % (format_start, format_end),
alignment=pxl.styles.Alignment(
horizontal='center', vertical='center'))
to_excel.fit_width(2)
def get_DNN_info_general(sess, first_jpeg_image, n_images=50):
#def get_DNN_info_general(sess, first_jpeg_image, n_images = 3):
graph_def = sess.graph.as_graph_def(add_shapes=True)
all_tensors = [
tensor for op in tf.get_default_graph().get_operations()
for tensor in op.values()
]
all_layers = []
input_tensor_list = []
conv_tensor_list = []
padding_tensor_list = []
sw_tensor_list = []
sh_tensor_list = []
first_input_tensor = []
k_tensor_list = []
kw_tensor_list = []
kh_tensor_list = []
# loop on all nodes in the graph
for nid, n in enumerate(graph_def.node):
try:
if nid == 0:
first_input_tensor.append(n.name + ':0')
#print(n.name)
if (
'Conv2D' in n.name or 'convolution' in n.name
) and '_bn_' not in n.name and 'Logits' not in n.name and 'logits' not in n.name:
output_tensor_name = n.name + ':0'
input_tensor_name = n.input[0] + ':0'
input_tensor = sess.graph.get_tensor_by_name(input_tensor_name)
input_tensor_list.append(input_tensor_name)
conv_tensor_list.append(
sess.graph.get_tensor_by_name(output_tensor_name))
if 'padding' in n.attr.keys():
padding_type = n.attr['padding'].s.decode(encoding='utf-8')
padding_tensor_list.append(padding_type)
if 'strides' in n.attr.keys():
art_tensor_name = n.name + ':0'
strides_list = [int(a) for a in n.attr['strides'].list.i]
Sh = strides_list[1]
Sw = strides_list[2]
sh_tensor_list.append(Sh)
sw_tensor_list.append(Sw)
conv_tensor_params_name = n.input[1] + ':0'
conv_tensor_params = sess.graph.get_tensor_by_name(
conv_tensor_params_name)
filter_shape = conv_tensor_params.shape
Kh = filter_shape[0]
Kw = filter_shape[1]
K = filter_shape[3]
k_tensor_list.append(K)
kh_tensor_list.append(Kh)
kw_tensor_list.append(Kw)
except ValueError:
print('%s is an Op.' % n.name)
# ensure that created lists have the same length
assert (len(sw_tensor_list) == len(sh_tensor_list) ==
len(padding_tensor_list) == len(conv_tensor_list) ==
len(input_tensor_list))
assert (len(k_tensor_list) == len(kh_tensor_list) == len(kw_tensor_list) ==
len(sw_tensor_list))
# Write the header in text file
txt_file = FLAGS.gen_dir + FLAGS.model_name + '.info'
info_file = open(txt_file, 'w')
current_string = (
'====input_tensor_name\toutput_tensor_name\tIh\tIw\tOh\tOw\tKh\tKw\tSh\tSw\tIc\tK\tru\tlowering_density====\n'
)
info_file.write(current_string)
print('Fetched Primary information about DNN...')
# Loop through convolutions to get the conv dimensions
file_list = FLAGS.gen_dir + "file_list"
conv_shape = FLAGS.gen_dir + "conv_shape"
file_list_file = open(file_list, 'w')
conv_shape_file = open(conv_shape, 'w')
for ic, c in enumerate(conv_tensor_list):
# Get image data
image_data = get_image_data(first_jpeg_image)
# Run the first image
current_feature_map, input_to_feature_map = run_DNN_for_analysis(
sess, ic, c, input_tensor_list, first_input_tensor, image_data)
_, Oh, Ow, _ = current_feature_map.shape
In, Ih, Iw, Ic = input_to_feature_map.shape
K, Kh, Kw, = k_tensor_list[ic], kh_tensor_list[ic], kw_tensor_list[ic]
Sh, Sw = sh_tensor_list[ic], sw_tensor_list[ic]
padding_type = padding_tensor_list[ic]
input_tensor_name = input_tensor_list[ic]
output_tensor_name = c.name
# Create the conv_layer
conv_layer = Conv_Layer(input_tensor_name,
output_tensor_name,
K,
Kh,
Kw,
Sh,
Sw,
Oh,
Ow,
Ih,
Iw,
Ic,
In,
padding=padding_type)
conv_layer.padding_cal()
# print(input_tensor_name, output_tensor_name)
# Calculate the densities
input_to_feature_map = np.squeeze(input_to_feature_map)
####################################### CUDA data ########################################################
# Data for preformance paper
# File name the have the conv layer intput
input_tensor_FeatureMap_fileName = (FLAGS.model_name +
"_Conv_%d_ImgID_%d") % (ic, 1)
file_list_file.write(input_tensor_FeatureMap_fileName + "\n")
current_string = ("%d") % (Ih)
conv_shape_file.write(current_string + "\n")
conv_fileName = FLAGS.conv + input_tensor_FeatureMap_fileName
conv_file = open(conv_fileName, 'w')
for i_h in range(0, Ih):
for i_w in range(0, Iw):
conv_file.write(str(input_to_feature_map[i_h, i_w, 0]) + '\t')
conv_file.close()
####################################### END CUDA data ####################################################
# lowering_matrix = conv_layer.preprocessing_layer(np.squeeze(input_to_feature_map))
lowering_matrix = conv_layer.preprocessing_layer(input_to_feature_map)
#print('********* PADDING TYPE ', conv_layer.padding, ' PADDINGS: ' , conv_layer.paddings, ' Input: ', input_tensor_name, ' out: ', output_tensor_name,
# ' Ih: ', conv_layer.Ih)
# Get all layers
all_layers.append(conv_layer)
print('[%s] Analyzed Conv Node %d' % (FLAGS.model_name, ic))
print('Reset the session')
config = tf.ConfigProto(device_count={'GPU': 0})
sess = tf.Session(config=config)
create_graph()
# Loop through images to get the average density for images:
for imgID in range(2, FLAGS.END + 1):
current_jpeg_image = org_image_dir + '/shard-' + str(0) + '/' + str(
1) + '/' + 'ILSVRC2012_val_' + str(imgID).zfill(8) + '.JPEG'
image_data = get_image_data(current_jpeg_image)
print('[%s] %s' % (FLAGS.model_name, current_jpeg_image))
# Loop through all convs in the model to update their densities:
for ic, c in enumerate(conv_tensor_list):
# Run the DNN
current_feature_map, input_to_feature_map = run_DNN_for_analysis(
sess, ic, c, input_tensor_list, first_input_tensor, image_data)
# Get the old_conv_layer in a new deep copy
old_conv_layer = copy.deepcopy(all_layers[ic])
# Calculate the densities of old_conv_layer
input_to_feature_map = np.squeeze(input_to_feature_map)
####################################### CUDA data ########################################################
# Data for preformance paper
# File name the have the conv layer intput
input_tensor_FeatureMap_fileName = (
FLAGS.model_name + "_Conv_%d_ImgID_%d") % (ic, imgID)
file_list_file.write(input_tensor_FeatureMap_fileName + "\n")
current_string = ("%d") % (Ih)
conv_shape_file.write(current_string + "\n")
conv_fileName = FLAGS.conv + input_tensor_FeatureMap_fileName
conv_file = open(conv_fileName, 'w')
for i_h in range(0, Ih):
for i_w in range(0, Iw):
conv_file.write(
str(input_to_feature_map[i_h, i_w, 0]) + '\t')
conv_file.close()
####################################### END CUDA data ####################################################
lowering_matrix = old_conv_layer.preprocessing_layer(
input_to_feature_map)
# lowering_matrix = old_conv_layer.preprocessing_layer(np.squeeze(input_to_feature_map))
# Update the densities to sum
all_layers[ic].ru += old_conv_layer.ru
all_layers[ic].lowering_density += old_conv_layer.lowering_density
if not ic % 10:
print('[%s] Analyzed Conv Node %d' % (FLAGS.model_name, ic))
if ic == int(0.25 * len(conv_tensor_list)) or ic == int(
0.5 * len(conv_tensor_list)) or ic == int(
0.75 * len(conv_tensor_list)):
# reset the session:
config = tf.ConfigProto(device_count={'GPU': 0})
sess = tf.Session(config=config)
create_graph()
#print('i-', ic, ' ', input_tensor_list[ic],
# ' ru: ' , old_conv_layer.ru, ' ru2: ', old_conv_layer.lowering_density, ' sum1: ', all_layers[ic].ru, ' sum2: ', all_layers[ic].lowering_density)
# Reset the session
config = tf.ConfigProto(device_count={'GPU': 0})
sess = tf.Session(config=config)
create_graph()
file_list_file.close()
conv_shape_file.close()
# Get the average density and save it in *.info file:
txt_file = FLAGS.gen_dir + FLAGS.model_name + '.info'
info_file = open(txt_file, 'a')
for ilayer, layer in enumerate(all_layers):
all_layers[ilayer].ru = all_layers[ilayer].ru / n_images
all_layers[ilayer].lowering_density = all_layers[
ilayer].lowering_density / n_images
# current_string = ('%s\t%s\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%f\t%f\n' %
# (layer.input_tensor_name, layer.output_tensor_name, layer.Ih, layer.Iw, layer.Oh, layer.Ow,
# layer.Kh, layer.Kw,
# layer.Sh, layer.Sw,
# layer.Ic, layer.K,
# layer.ru, layer.lowering_density))
current_string = (
'%d & %d & %d & %d & %d & %d & %d & %d & %d & %d & %d & %.2f & %.2f \\\\\n'
% (ilayer, layer.Ih, layer.Iw, layer.Oh, layer.Ow, layer.Kh,
layer.Kw, layer.Sh, layer.Sw, layer.Ic, layer.K, layer.ru,
layer.lowering_density))
info_file.write(current_string)
#print(current_string)
#info_file.write('Extracted info for these %d layers... No Risk No Fun :) ' % len(all_layers))
info_file.close()
print('Extracted info for these %d layers... No Risk No Fun :) ' %
len(all_layers))
print('This code should work without exit.... Done!')
exit(0)
return all_layers
def readfiles(self, filepath): #存入数据
self.progressBar.setVisible(True)
self.statusBar().showMessage("正在进行数据转换...")
self.progressBar.setValue(0)
files = os.listdir(filepath)
#排除隐藏文件和文件夹
for f in files:
if (os.path.isdir(filepath + '/' + f)):
# 排除隐藏文件夹。因为隐藏文件夹过多
if (f[0] == '.'):
pass
else:
# 添加非隐藏文件夹
self.dirList.append(f)
if (os.path.isfile(filepath + '/' + f)):
# 添加文件
if (os.path.splitext(f)[1] == ".txt"):
self.filenames.append(f)
p = 1
print(self.filenames)
for f in self.filenames:
self.statusBar().showMessage("正在转换 " + str(p) + "/" +
str(len(self.filenames)))
data = dataclass()
data.filepath = filepath
datarow = open(filepath + '/' + f) #读取的整个原始文件数据
datarowlines = datarow.readlines() #读取的整个原始文件的数据，按行分割
datapar = [] #真正的每行数据数组
for line in datarowlines:
linenew = line.strip()
if (linenew != ""):
datapar.append(linenew)
# self.ACQ_Time=re.search("(\d{4}-\d{1,2}-\d{1,2}\s*\d{1,2}:\d{1,2}:\d{1,2}:\s*\d{1,3})",datapar[1])
temptime = re.search(
"\d{4}-\s*\d{1,2}-\s*\d{1,2}\s*\d{1,2}:\s*\d{1,2}:\s*\d{1,2}:\s*\d{1,3}",
datapar[1]).group(0)
timelist = re.split('[- :]\s*', temptime)
# print(timelist)
timestr = timelist[0] + "-" + timelist[1] + "-" + timelist[
2] + " " + timelist[3] + ":" + timelist[4] + ":" + timelist[
5] + "." + timelist[6]
data.ACQ_Time = datetime.strptime(timestr, '%Y-%m-%d %H:%M:%S.%f')
data.Project = datapar[2].strip(
datapar[2].split(": ")[0]).strip(": ")
data.Name = datapar[3].strip(datapar[3].split(": ")[0]).strip(": ")
data.part = datapar[4].strip(datapar[4].split(": ")[0]).strip(": ")
data.Operator = datapar[5].strip(
datapar[5].split(": ")[0]).strip(": ")
data.Desc = datapar[6].strip(datapar[6].split(": ")[0]).strip(": ")
data.Excited_Peroid = int(datapar[9].strip(
datapar[9].split(": ")[0]).strip(": "))
data.Excited_Time = int(datapar[10].strip(
datapar[10].split(": ")[0]).strip(": "))
data.Acq_Delay_Time = int(datapar[11].strip(
datapar[11].split(": ")[0]).strip(": "))
data.Gate_Time = int(datapar[12].strip(
datapar[12].split(": ")[0]).strip(": "))
data.Count_Num_per_gate = int(datapar[13].strip(
datapar[13].split(": ")[0]).strip(": "))
data.Repeat_Times = int(datapar[14].strip(
datapar[14].split(": ")[0]).strip(": "))
data.Acq_Gate_Times = int(datapar[15].strip(
datapar[15].split(": ")[0]).strip(": "))
data.Interval_per_Gate = int(datapar[16].strip(
datapar[16].split(": ")[0]).strip(": "))
data.Channel_Number = int(datapar[17].strip(
datapar[17].split(": ")[0]).strip(": "))
# print(data.ACQ_Time)
#计算时间，以分钟为单位
# minutecount=0
# timelist=re.split('[- :]\s*',data.ACQ_Time)
# data1 =timelist[0] + "-" + timelist[1] + "-" + timelist[2] + " " + timelist[3] + ":" + timelist[4] + ":" + timelist[5]
#
# year=(float(timelist[0])-2019)*364*24*60
# mouth=(float(timelist[1])-1)*30.5*24*60
# day= (float(timelist[2])-1)*24*
# print(data)
datanum = int(data.Repeat_Times * data.Count_Num_per_gate)
data.Pro_Data1 = np.zeros(data.Acq_Gate_Times *
data.Count_Num_per_gate).tolist()
# [0 for i in range(data.Acq_Gate_Times*data.Count_Num_per_gate)]
# print(data.Gate_Time)
# print("len(data.Raw_Data1)",len(data.Pro_Data1))
# print("data.Count_Num_per_gate*data.Gate_Time",data.Count_Num_per_gate*data.Gate_Time)
for i in range(19, 19 + datanum):
data.Raw_Data1.append(int(datapar[i]))
if (data.Channel_Number == "2"):
for i in range(19 + datanum + 1, 19 + datanum + 1 + datanum):
data.Raw_Data2.append(int(datapar[i]))
# print(len(data.Raw_Data1))
# print(len(data.Raw_Data2))
dScale = data.Count_Num_per_gate * 1000 / (data.Gate_Time *
data.Repeat_Times)
for i in range(int(data.Acq_Gate_Times)):
for j in range(int(data.Count_Num_per_gate)):
ncps = 0
for k in range(int(data.Repeat_Times)):
ncps += data.Raw_Data1[k * data.Count_Num_per_gate +
j + i * datanum]
data.Pro_Data1[j * data.Acq_Gate_Times + i] = ncps * dScale
data.Interval = float(data.Gate_Time) / len(data.Pro_Data1)
for i in range(len(data.Pro_Data1)):
data.Pro_Data1_X.append(i * data.Interval)
# print(data.Pro_Data1)
data.Max = np.max(data.Pro_Data1)
data.Min = np.min(data.Pro_Data1)
data.cutendnum1 = len(data.Pro_Data1) - 1
#复制原始数据到Cut
data.Cut_Data1 = copy.deepcopy(data.Pro_Data1)
data.Cut_Data1_X = copy.deepcopy(data.Pro_Data1_X)
if (len(data.Cut_Data1) > self.maxCol):
self.maxCol = len(data.Cut_Data1)
# print("data.Cut_Data1",data.Cut_Data1)
self.filelist[f] = data
# print(self.filelist[f].Pro_Data1)
self.progressBar.setValue(p / len(self.filenames) * 100)
p += 1
self.statusBar().showMessage("完成！")
self.progressBar.setVisible(False)
self.statusBar().showMessage("数据转换成功！")