def post(self):
try:
jsonByte = self.request.body
jsonStr = jsonByte.decode("utf-8")
naiveBayesVo = json.loads(jsonStr, object_hook=dictToNaiveBayesVo)
base_path = naiveBayesVo.bayesDir
x_file_name = naiveBayesVo.trainxFileName
y_file_name = naiveBayesVo.trainyFileName
# 入参训练数据读取
(x_train, x_test) = get_data(base_path + x_file_name)
# 出参训练数据读取
y_train_dict = get_y_train_data(base_path + y_file_name)
clf = GaussianNB()
predictDataList = []
# 创建预测值文件对象
for (station, y_train) in y_train_dict.items():
clf.fit(x_train, y_train)
y_predict = clf.predict(x_test.reshape(1, -1))
# 写预测值
siteData = com.SiteData(str(station), str(y_predict[0]))
predictDataList.append(siteData.__dict__)
commonResponse = com.CommonResponse(
com.successCode, com.successMsg,
dict([("list", predictDataList)]))
except BaseException as e:
print("执行朴素贝叶斯算法失败：{}".format(e))
commonResponse = com.CommonResponse(com.errorCode, "{}".format(e),
dict([("list", [])]))
finally:
# 返回数据
self.write(commonResponse.__dict__)
def post(self):
try:
jsonByte = self.request.body
jsonStr = jsonByte.decode("utf-8")
rhythmVo = json.loads(jsonStr, object_hook=dictToRhythmVo)
datas = load_data(rhythmVo.sampleDataPath)
length = len(datas) - 1
observeData = datas[length]
index = 0
maxSimilarity = Similarity(np.array(observeData), np.array(datas[0]))
if pd.isnull(maxSimilarity):
maxSimilarity = 0
for i in range(1,length):
s = Similarity(np.array(observeData), np.array(datas[i]))
if pd.isnull(s):
s = 0
if s > maxSimilarity:
maxSimilarity = s
index = i
commonResponse = com.CommonResponse(com.successCode, com.successMsg, dict([("obj", index)]))
except BaseException as e:
print("执行韵律算法失败：{}".format(e))
commonResponse = com.CommonResponse(com.errorCode, "{}".format(e), dict([("obj", com.missingValues)]))
finally:
# 返回数据
self.write(commonResponse.__dict__)
def post(self):
jsonByte = self.request.body
jsonStr = jsonByte.decode("utf-8")
mgfVo = json.loads(jsonStr, object_hook=dictToMgfVo)
dataFilePath = mgfVo.sampleDataPath
file = open(dataFilePath, 'r')
predictDataList = []
for line in file.readlines():
line_arr = line.split(",")
station_id = line_arr.pop(0)
X = load_data(line_arr) #数据转float数组
N, = X.shape # shape返回数组维度，资料年份长度
NN = 1 # 需要预测的年数
N1 = N + NN
IN = 10 # 循环次数
Y = np.zeros(shape=N1)
try:
ACTA(X, Y, N, NN, N1, IN)
predict_value = Y[Y.size - 1]
siteData = com.SiteData(str(station_id), str(predict_value))
predictDataList.append(siteData.__dict__)
except TypeError as e:
print(station_id, "预测失败：", e)
if len(predictDataList) == 0:
commonResponse = com.CommonResponse(
com.errorCode, "每个站点都预测失败", dict([("list", predictDataList)]))
else:
commonResponse = com.CommonResponse(
com.successCode, com.successMsg,
dict([("list", predictDataList)]))
self.write(commonResponse.__dict__)
def post(self):
try:
commonResponse = com.CommonResponse(com.errorCode, "无同聚类数据",
dict([("list", [])]))
jsonByte = self.request.body
jsonStr = jsonByte.decode("utf-8")
kmeansVo = json.loads(jsonStr, object_hook=dictToKmeansVo)
k = 8 # 聚类中心的个数
data = load_data(kmeansVo.dataPath)
while k >= 2:
years = get_index_means(data, k)
if len(years) != 0:
print("聚类成功，k={}".format(k))
commonResponse = com.CommonResponse(
com.successCode, com.successMsg,
dict([("list", years)]))
break
k = k - 1
except BaseException as e:
print("执行kmeans失败：{}".format(e))
commonResponse = com.CommonResponse(com.errorCode, "{}".format(e),
dict([("list", [])]))
finally:
# 返回数据
self.write(commonResponse.__dict__)
def post(self):
try:
jsonByte = self.request.body
jsonStr = jsonByte.decode("utf-8")
svmVo = json.loads(jsonStr, object_hook=svm.dictToSvmVo)
file_name = svmVo.trainDataPath
model = svmVo.modelPath
# 1、导入训练数据
dataSet, labels = load_data_libsvm(file_name)
# 2、训练SVM模型
C = 0.6
toler = 0.001
maxIter = 500
svm_model = svm.SVM_training(dataSet, labels, C, toler, maxIter)
# 3、计算训练的准确性
svm.cal_accuracy(svm_model, dataSet, labels)
# 4、保存最终的SVM模型
svm.save_svm_model(svm_model, model)
commonResponse = com.CommonResponse(com.successCode,
com.successMsg,
dict([("obj", 1)]))
except BaseException as e:
print("执行支持向量机-生成模型失败：{}".format(e))
commonResponse = com.CommonResponse(com.errorCode, "{}".format(e),
dict([("obj", 1)]))
finally:
# 返回数据
self.write(commonResponse.__dict__)
def post(self):
try:
# 1、导入训练数据
jsonByte = self.request.body
jsonStr = jsonByte.decode("utf-8")
decisionTreeVo = json.loads(jsonStr,
object_hook=dictToDecisionTreeVo)
data_train = load_data(decisionTreeVo.trainDataPath)
# 2、创建CART分类树
tree = build_tree(data_train)
# 3、获取样本数据
test_data = load_data(decisionTreeVo.testDataPath)
# 4、对样本进行预测
result = list(predict(test_data[0], tree).keys())[0]
commonResponse = com.CommonResponse(com.successCode,
com.successMsg,
dict([("obj", result)]))
except BaseException as e:
print("执行决策树算法失败：{}".format(e))
commonResponse = com.CommonResponse(
com.errorCode, "{}".format(e),
dict([("obj", com.missingValues)]))
finally:
# 返回数据
self.write(commonResponse.__dict__)
def post(self):
global commonResponse
try:
jsonByte = self.request.body
jsonStr = jsonByte.decode("utf-8")
tccImageVo = json.loads(jsonStr, object_hook=dictToTccImageVo)
result = create_svg(tccImageVo)
commonResponse = com.CommonResponse(com.successCode,
com.successMsg,
dict([("obj", result)]))
except BaseException as e:
print("执行TCC画图失败：{}".format(e))
commonResponse = com.CommonResponse(
com.errorCode, "{}".format(e),
dict([("obj", com.missingValues)]))
finally:
# 返回数据
self.write(commonResponse.__dict__)
def post(self):
try:
jsonByte = self.request.body
jsonStr = jsonByte.decode("utf-8")
svmVo = json.loads(jsonStr, object_hook=sv.dictToSvmVo)
file_name = svmVo.testDataPath
model_name = svmVo.modelPath
# 1、导入测试数据
test_data = load_test_data(file_name)
# 2、导入SVM模型
svm_model = load_svm_model(model_name)
# 3、得到预测值
prediction = get_prediction(test_data, svm_model)
commonResponse = com.CommonResponse(com.successCode, com.successMsg, dict([("obj", prediction[0])]))
except BaseException as e:
print("执行支持向量机算法失败：{}".format(e))
commonResponse = com.CommonResponse(com.errorCode, "{}".format(e), dict([("obj", com.missingValues)]))
finally:
# 返回数据
self.write(commonResponse.__dict__)
def post(self):
try:
jsonByte = self.request.body
jsonStr = jsonByte.decode("utf-8")
pcaVo = json.loads(jsonStr, object_hook=dictToPcaVo)
feature_path = pcaVo.sampleDataPath
label_path = pcaVo.labelDataPath
test_path = pcaVo.forecastDataPath
data = loadDataSet(feature_path)
lowDDataMat, reconMat = pca(data, 3)
y = load_data(label_path)
w_ls = least_square(mat(lowDDataMat), y)
testData = loadDataSet(test_path)
testlowDDataMat, testreconMat = pca(testData, 3)
predict = get_prediction(mat(testlowDDataMat), mat(w_ls))
commonResponse = com.CommonResponse(com.successCode, com.successMsg, dict([("obj", predict[-1, 0].real)]))
except BaseException as e:
print("执行主成分分析失败：{}".format(e))
commonResponse = com.CommonResponse(com.errorCode, "{}".format(e), dict([("obj", com.missingValues)]))
finally:
# 返回数据
self.write(commonResponse.__dict__)
def post(self):
try:
jsonByte = self.request.body
jsonStr = jsonByte.decode("utf-8")
vectorWindVo = json.loads(jsonStr, object_hook=dictToVectorWindVo)
x = vectorWindVo.lons
y = vectorWindVo.lats
u = vectorWindVo.uwindValues
v = vectorWindVo.vwindValues
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1)
ax.barbs(x, y, u, v, length=4, pivot='middle')
plt.xticks([])
plt.yticks([])
plt.axis('off')
imageDir = vectorWindVo.imageDir
if os.path.exists(imageDir) == False:
os.makedirs(imageDir)
imageName = vectorWindVo.imageName
plt.savefig(imageDir + imageName,
format='svg',
bbox_inches='tight',
transparent=True,
dpi=600)
plt.clf()
plt.close(fig)
commonResponse = com.CommonResponse(
com.successCode, com.successMsg,
dict([("obj", imageDir + imageName)]))
except BaseException as e:
print("执行矢量风出图失败：{}".format(e))
commonResponse = com.CommonResponse(
com.errorCode, "{}".format(e),
dict([("obj", com.missingValues)]))
finally:
# 返回数据
self.write(commonResponse.__dict__)
def post(self):
try:
jsonByte = self.request.body
jsonStr = jsonByte.decode("utf-8")
multipleRegressionVo = json.loads(
jsonStr, object_hook=dictToMultipleRegressionVo)
# 1、导入数据
data_train = load_data(multipleRegressionVo.resultDataFilePath)
# 3、获取样本数据
sample_data = load_data(multipleRegressionVo.sampleDataFilePath)
test_data = multipleRegressionVo.testDataFilePath
# 4、对样本进行预测
result = multiple_regression(data_train, sample_data, test_data)
commonResponse = com.CommonResponse(com.successCode,
com.successMsg,
dict([("obj", result[0][0])]))
except BaseException as e:
print("执行多元回归算法失败：{}".format(e))
commonResponse = com.CommonResponse(
com.errorCode, "{}".format(e),
dict([("obj", com.missingValues)]))
finally:
# 返回数据
self.write(commonResponse.__dict__)
def post(self):
try:
jsonByte = self.request.body
jsonStr = jsonByte.decode("utf-8")
optimalSubsetVo = json.loads(jsonStr,
object_hook=dictToOptimalSubsetVo)
txtPath = optimalSubsetVo.sampleDataPath
feature_num = int(optimalSubsetVo.featureNum)
col_row_num = (feature_num // 100 + 1) * 100
ff1 = pd.read_csv(txtPath, sep='\\s+', header=None)
ff2 = pd.DataFrame(ff1)
nums = ff1.shape[0]
xxCol = col_row_num
xfCol = col_row_num
xfRow = col_row_num
n = feature_num
nsize = n // 10 + 1
n1 = n + 1
# 创建二维数组
ex = np.zeros(shape=(30, 30), dtype=float)
cc = np.zeros(shape=(30, 30), dtype=float)
coe = np.zeros(shape=(30, 30), dtype=float)
# 创建30*100的数组(指数先改为30*200)(高度场30*2800)
xx = np.zeros(shape=(30, xxCol), dtype=float)
# 创建100*100的数组(指数先改为200*200)(高度场2800*2800)
xf = np.zeros(shape=(xfRow, xfCol), dtype=float)
csca = []
ihh = []
for i in range(0, 30):
csca.append(0)
ihh.append(0)
csca = np.array(csca, dtype=float)
ihh = np.array(ihh)
x1 = []
x2 = []
x = []
for j in range(0, n1):
x1.append(0)
x2.append(0)
x.append(0)
x1 = np.array(x1, dtype=float)
x2 = np.array(x2, dtype=float)
x = np.array(x, dtype=float)
'''
ff1和ff2获取的二维数组数据有缺失的，比如此例中最后一行的最后一个数据，
fillna缺测数据填充为参数值
'''
f1 = ff2.fillna(32766)
f2 = ff2.fillna(-999)
# 气候指数，需要的训练集指数刚好是10的整数位，最后一组需要求最优子集测试回归的数据，刚好就少一列，需要补回来
# 前面几组，有1个特征向量占一行，不会出现最后一组的情况
if feature_num % 10 == 0:
f32766 = pd.DataFrame([[
32766, 32766, 32766, 32766, 32766, 32766, 32766, 32766,
32766, 32766
]])
# ignore_index=True True 添加时，需要索引，默认值是false
f1 = f1.append(f32766, ignore_index=True)
f0 = pd.DataFrame([[
-999, -999, -999, -999, -999, -999, -999, -999, -999, -999
]])
f2 = f2.append(f0, ignore_index=True)
dat = []
xy = []
for num in range(0, nums):
f12 = []
f13 = []
# shape查看数组维数，shape[0]=11，shape[1]=143 或 2702
for feature_index in range(0, f1.shape[1]):
f12.append(f1[feature_index][num])
f13.append(f2[feature_index][num])
for bb in f12:
if bb == 32766:
for bb1 in range(0, f12.count(32766)):
f12.remove(bb)
f12 = np.array(f12)
f13 = np.array(f13)
dat.append(f12)
# xy.txt读取的数据，不满数组的数据补0操作
xy.append(f13)
dat = np.array(dat)
dat_mean = []
for i in range(0, len(dat)):
dat_mean.append(np.nanmean(dat[i][0:n]))
a1 = []
for i in range(0, nums):
for j in range(0, nums):
s = 0
for k in range(0, n):
s += (xy[i][k] * xy[j][k])
s -= (n * dat_mean[i] * dat_mean[j])
a1.append(s)
a = []
for i in range(0, nums):
a2 = []
# 魔法数值 6 ,气候指数修改为nums
for j in range(i * nums, i * nums + nums):
a2.append(a1[j])
a.append(a2)
a = np.array(a)
sto = a[nums - 1][nums - 1]
# **求次方
ih = 2**(nums - 2) - 1
ks = []
for i in range(0, ih + 1):
ks.append(0)
ks[0] = 1
for k in range(1, nums - 2):
j = 2**k
ks[j - 1] = k + 1
for i in range(0, j - 1):
m = ks[j - i - 2]
ks[j + i] = -m
ks[ih] = 5
nb = 0
rss = []
ma = []
for i in range(0, nums - 1):
rss.append(10.0**20)
ma.append(0)
rss = np.array(rss)
ma = np.array(ma)
def csc():
le = np.array([[0 for col in range(3)] for row in range(3)],
dtype=float)
p = []
q = []
for i in range(0, 3):
p.append(0)
q.append(0)
p = np.array(p, dtype=float)
q = np.array(q, dtype=float)
v = 0.0
u = 0.0
for i in range(0, n - 1):
u += abs(x1[i + 1] - x1[i]) / (n - 1)
v += abs(x[i + 1] - x[i]) / (n - 1)
for i in range(0, 3):
for j in range(0, 3):
le[i][j] = 0.0
for i in range(0, n - 1):
xp = x[i + 1] - x[i]
xp1 = x1[i + 1] - x1[i]
if (xp1 >= u and xp >= v):
le[0][0] += 1
if (xp1 >= u and xp <= 0):
le[2][0] += 1
if (xp1 >= u and xp > 0 and xp < v):
le[1][0] += 1
if (xp1 > 0 and xp1 < u and xp >= v):
le[0][1] += 1
if (xp1 > 0 and xp1 < u and xp > 0 and xp < v):
le[1][1] += 1
if (xp1 > 0 and xp1 < u and xp <= 0):
le[2][1] += 1
if (xp1 <= 0 and xp >= v):
le[0][2] += 1
if (xp1 <= 0 and xp > 0 and xp < v):
le[1][2] += 1
if (xp1 <= 0 and xp <= 0):
le[2][2] += 1
le = np.array(le)
le = le.T
r1 = 0.0
r2 = 0.0
r3 = 0.0
for i in range(0, 3):
p[i] = 0.0
q[i] = 0.0
for i in range(0, 3):
for j in range(0, 3):
if le[j][i] != 0.0:
r1 += (le[j][i] * (math.log(float(le[j][i]))))
for i in range(0, 3):
for j in range(0, 3):
p[i] += le[j][i]
q[i] += le[i][j]
for i in range(0, 3):
if (q[i] != 0.0):
r2 += (q[i] * (math.log(q[i])))
if (p[i] != 0.0):
r3 += (p[i] * (math.log(p[i])))
s1 = 2 * (r1 + (n - 1) * (math.log(float(n - 1))) - (r2 + r3))
xm1 = 0.0
for i in range(0, n):
xm1 += (x[i] / float(n))
qk = 0.0
qx = 0.0
for i in range(0, n):
qk += ((x[i] - x1[i]) * (x[i] - x1[i]) / n)
qx += ((x[i] - xm1) * (x[i] - xm1) / n)
s2 = (n - (k + 1)) * (1 - qk / qx)
csc3 = s1 + s2
return csc3
def reg2(it):
kk = it - 1
if (abs(a[kk][kk]) > (10**(-12))):
for j in range(0, nums):
for i in range(0, nums):
if (i != kk and j != kk):
a[i][j] -= a[kk][j] * a[i][kk] / a[kk][kk]
else:
continue
for j in range(0, nums):
if (j != kk):
a[j][kk] = a[j][kk] / a[kk][kk]
a[kk][j] = -a[kk][j] / a[kk][kk]
else:
continue
a[kk][kk] = 1 / a[kk][kk]
ma[kk] = 0
if (a[nums - 1][nums - 1] <= rss[nb - 1]):
rss[nb - 1] = a[nums - 1][nums - 1]
for j in range(0, nums - 1):
ex[j][nb - 1] = ma[j]
coe[j][nb - 1] = a[nums - 1][j]
else:
return ex, coe, ma, rss, a, nb, ts
return ex, coe, ma, rss, a, nb, ts
def reg1(it):
kk = it - 1
if (abs(a[kk][kk]) > (10**(-12))):
for j in range(0, nums):
for i in range(0, nums):
if (i != kk and j != kk):
a[i][j] -= a[kk][j] * a[i][kk] / a[kk][kk]
else:
continue
for j in range(0, nums):
if (j != kk):
a[j][kk] = a[j][kk] / a[kk][kk]
a[kk][j] = -a[kk][j] / a[kk][kk]
else:
continue
a[kk][kk] = 1 / a[kk][kk]
ma[kk] = 1
if (a[nums - 1][nums - 1] <= rss[nb - 1]):
rss[nb - 1] = a[nums - 1][nums - 1]
for j in range(0, nums - 1):
ex[j][nb - 1] = ma[j]
coe[j][nb - 1] = a[nums - 1][j]
else:
return ex, coe, ma, rss, a, nb, ts
return ex, coe, ma, rss, a, nb, ts
ts = 0
for m in range(0, ih + 1):
it = abs(ks[m])
if ks[m] > 0:
nb += 1
ts += 1
ex, coe, ma, rss, a, nb, ts = reg1(it)
else:
nb -= 1
ts += 1
ex, coe, ma, rss, a, nb, ts = reg2(it)
for m in range(0, ih):
it = abs((-1) * ks[ih - 1 - m])
if (-1) * ks[ih - 1 - m] > 0:
nb += 1
ts += 1
ex, coe, ma, rss, a, nb, ts = reg1(it)
else:
nb -= 1
ts += 1
ex, coe, ma, rss, a, nb, ts = reg2(it)
for k in range(0, nums - 1):
s = rss[k]
c = 0.0
for j in range(0, nums - 1):
if ex[j][k] != 0:
c += (coe[j][k] * dat_mean[j])
else:
continue
c = dat_mean[nums - 1] - c
mm = 0
for j in range(0, nums - 1):
if ex[j][k] != 0.0:
mm += 1
cc[mm - 1][k] = coe[j][k]
for i in range(0, n1):
xx[mm - 1][i] = xy[j][i]
for i in range(0, n1):
xf[i][k] = c
for j in range(0, mm):
xf[i][k] = xf[i][k] + cc[j][k] * xx[j][i]
x[i] = xy[nums - 1][i]
x1[i] = xf[i][k]
for j in range(0, n):
x2[j] = xf[j][k] - x[j]
w = 0.0
for j in range(0, n):
w += x2[j]**2
v1 = np.sqrt(w / n)
csca[k] = csc()
aa = 0.0
for i in range(0, nums - 1):
if (csca[i] > aa):
aa = csca[i]
for i in range(0, nums - 1):
if csca[i] == aa:
ihh[0] = i
kk1 = ihh[0]
for j in range(0, nums - 1):
if (j == kk1):
for i in range(0, n1):
x1[i] = xf[i][j]
else:
continue
w = 0.0
for i in range(0, n):
w += (x[i] - x1[1])**2
v = np.sqrt(w / n)
commonResponse = com.CommonResponse(
com.successCode, com.successMsg,
dict([("obj", '%.2f' % x1[n])]))
except BaseException as e:
print("执行最优子集失败：{}".format(e))
commonResponse = com.CommonResponse(
com.errorCode, "{}".format(e),
dict([("obj", com.missingValues)]))
finally:
# 返回数据
self.write(commonResponse.__dict__)
def post(self):
try:
jsonByte = self.request.body
jsonStr = jsonByte.decode("utf-8")
svgDto = json.loads(jsonStr, object_hook=dictToSvgDto)
# 读取txt文件
lat = svgDto.lats
lon = svgDto.lons
val = svgDto.values
minLon = svgDto.areaMinLon
maxLon = svgDto.areaMaxLon
minLat = svgDto.areaMinLat
maxLat = svgDto.areaMaxLat
lonLength = svgDto.lonLength
latLength = svgDto.latLength
# 经纬度处理
olon = np.linspace(minLon, maxLon, lonLength)
olat = np.linspace(minLat, maxLat, latLength)
# 转成地理经纬度
olon, olat = np.meshgrid(olon, olat)
# 插值处理
func = Rbf(lon, lat, val, function='linear')
rhu_data_new = func(olon, olat)
# 自定义色卡处理
levels = svgDto.levels
my_cmap = colors.ListedColormap(svgDto.colorMap)
my_cmap.set_under(svgDto.underColor)
my_cmap.set_over(svgDto.overColor)
norm3 = colors.BoundaryNorm(levels, my_cmap.N)
# 投影方式
projection = ccrs.Mercator()
# 创建图实例 创建设置像素
fig = plt.figure(dpi=300)
ax = fig.add_subplot(111, projection=projection)
# 设定要显示的边界区域
ax.set_extent([minLon, maxLon, minLat, maxLat], crs=projection)
# 等值面处理
cs = ax.contourf(olon,
olat,
rhu_data_new,
levels=levels,
cmap=my_cmap,
norm=norm3,
extend='both',
transform=projection,
alpha=1)
# 读取shp文件，获取要裁剪的区域clip(环流图不用裁剪)
shp_name = svgDto.shpPath
if shp_name:
sf = shapefile.Reader(shp_name, encoding="gbk")
for shp_record in sf.shapeRecords():
vertices = []
codes = []
pts = shp_record.shape.points
prt = list(shp_record.shape.parts) + [len(pts)]
for i in range(len(prt) - 1):
for j in range(prt[i], prt[i + 1]):
vertices.append((pts[j][0], pts[j][1]))
codes += [Path.MOVETO]
codes += [Path.LINETO] * (prt[i + 1] - prt[i] - 2)
codes += [Path.CLOSEPOLY]
clip = Path(vertices, codes)
clip = PathPatch(clip, transform=ax.transData)
# 裁剪显示区域
for contour in cs.collections:
contour.set_clip_path(clip)
# cartopy 设置背景透明及去除边框
ax.outline_patch.set_visible(False)
ax.background_patch.set_visible(False)
#判断目录是否存在，不存在则创建
save_path = svgDto.savePath
if os.path.exists(save_path) == False:
os.makedirs(save_path)
image_name = svgDto.imageName
# 保存图片，tight让保存的图片更加紧促
plt.savefig(save_path + image_name,
bbox_inches='tight',
transparent=True,
pad_inches=0)
plt.clf()
plt.close(fig)
commonResponse = com.CommonResponse(
com.successCode, com.successMsg,
dict([("obj", save_path + image_name)]))
except BaseException as e:
print("执行svg出图失败：{}".format(e))
commonResponse = com.CommonResponse(
com.errorCode, "{}".format(e),
dict([("obj", com.missingValues)]))
finally:
# 返回数据
self.write(commonResponse.__dict__)
