import numpy as np
import scipy.io as scio
import heartpy as hp
import libs.SkinVector.SkinVector as SkinVector
import CsvAccess


#20

import os   #清除残余文件

#https://scikit-learn.org/stable/modules/model_persistence.html
if os.path.exists('feature.csv'):
    os.remove('feature.csv')




# 文件载入
filename_1 = 'n08'
data1 = scio.loadmat('data\\心电\\10_12211013\\' + filename_1 + '.mat')

filename_2 = 'n08'
data2 = scio.loadmat('data\\皮电\\101221\\' + filename_2 + '.mat')

# 统计数据长度
data_ecg = np.array(data1[filename_1])
data_skin = np.array(data2[filename_2])


ecg_length = data_ecg.shape[1]
skin_length = data_skin.shape[0]

#强制转换
data_skin = data_skin.reshape(-1,1)


fs_ecg = 514  # 采样频率
fs_skin = 50

#####################测试环节(不要删除,因为要写入表头)##################

#初始载入一些数据
data_ecg = np.array(data1[filename_1][0][0:1000])
data_skin_temp = data_skin[0:1000]
# # 心电特征提取测试
data = hp.scale_data(data_ecg)
working_data, measures = hp.process(data, fs_ecg)

#皮电提取测试
vector = SkinVector.get_vector(data_skin_temp)

# # 将两个字典嵌套为一个字典写入文件
#
# dict = {'ecg': measures, 'skin': vector}
#
# file = open('cache.pkl', 'w')
# for k,v in dict.items():
#     file.write(k + ' ')
#     for k1, v1 in v.items():
#         file.write(k1 + ' ' + str(v1) + ' ')
#     file.write(' \n')
# file.close()


# 将表头写入文件
heads = CsvAccess.dict_key_to_list('number', 'label', measures, vector)  # 写入表头
CsvAccess.write('feature.csv', heads)



############### 特征提取部分###########################3

# 参数设置
time = ecg_length / fs_ecg  # 计算采集时间
sect_time = 180  # 切片时间单位秒

n_sect = time / sect_time  # 计算时间片数量

offset = 30000  # 切片偏移量

for section in range(int(n_sect) - 1):
    # 计算切片起始点与终点
    end_ecg = offset + sect_time * fs_ecg * (section + 1)
    start_ecg = end_ecg - sect_time * fs_ecg

    end_skin = sect_time * fs_skin * (section + 1)
    start_skin = end_skin - sect_time * fs_skin


    # 通过上述计算进行切片
    data_ecg = np.array(data1[filename_1][0][start_ecg:end_ecg])
    data_skin_split = np.array(data_skin[start_skin:end_skin])


    # 计算心电的特征
    data = hp.scale_data(data_ecg)

    working_data, measures = hp.process(data, fs_ecg)

    # 计算皮电的特征
    vector = SkinVector.get_vector(data_skin_split)


    # 将特征值和标注转成list类型
    vaule = CsvAccess.dict_value_to_list(section, 'sad', measures, vector)
    # 写入文件
    CsvAccess.write('feature.csv', vaule)

print('\n特征已经提取并写入成功')

#从文件读取结果 (测试)
#print(type(get_dict()))


# fs=514 # 采样率
# lowcut=0.5
# highcut=85
# fs=fs*0.5
# b, a = filter.butter(6,[lowcut/fs,highcut/fs],btype='bandpass')
# data_ecg = filter.filtfilt(b, a, data_ecg)