使用Python读取LabVIEW TDMS 格式文件转成Excel格式+多进程版本

使用Python读取LabVIEW TDMS 格式文件转成Excel格式+多进程版本

前言：

更新，一直想试试进程池，因为看了B站的课程后，发现进程池比传统的multiprocess模块要方便多了，还快多了，就一直想试试，之前没有合适的例子，强化学习的部分，因为涉及到的东西太多，也没法当个简单的例子去测试。
刚好昨天做这个转换的时间非常长，典型的计算资源消耗型的任务，因此想着能不能利用起进程池，走一次流程。

试了一下果然效果拔群！

直接用nptdms模块读取的时候，如果你的tdms文件 not valid的话，会报错：

1). ValueError: Segment does not start with TDSm, but with $ '
2). ValueError: Segment does not start with TDSm, but with " %

原因是：

It sounds like you are trying to read files that are not valid TDMS files, as they don’t start with the first four TDSm bytes that should be at the beginning of all TDMS files.

参考链接：
https://github.com/adamreeve/npTDMS/issues/123

因此我只好选择最基本的文件读取方式，按行读取，手动处理为pandas的格式，然后保存为excel的文件格式。

背景

Labview的TMDS格式文件在很多偏硬件相关的工作中经常出现，有时候面临将TDMS转成Excel通用格式的情况，正常来讲Excel只要装有TDMS插件读取是没有问题的，然而面对大量TDMS文件的时候，手动一个个将TDMS文件转成Excel文件显然很浪费时间，以下介绍使用Python配合pandas，xlsxwriter将TMDS文件批量转成Excel文件的方法：

tmds文件样式

其实就是数字+制表符+换行符的一个内容，按照最基本的文本读取，加上替换，就能完成格式转换。

974.832	1012.613	1016.344	1012.713	1012.009	1015.501	1000.094	1012.158	1011.943	1026.974	1011.168	1011.470	974.860	1011.517	1012.005	1062.471	1011.013	970.696	1012.960	1011.206	1011.439	1003.868	1011.456	1013.407	1011.972	1010.859	1011.276	1024.039	1039.353	1011.088	1012.036	1010.735	1011.564	1011.131	1011.837	1012.786
974.865	1012.567	1016.372	1012.719	1011.996	1015.527	1000.109	1012.148	1011.952	1026.951	1011.189	1011.471	974.846	1011.503	1012.065	1062.495	1011.044	970.726	1012.951	1011.242	1011.460	1003.885	1011.459	1013.423	1011.970	1010.856	1011.310	1024.043	1039.399	1011.112	1012.026	1010.742	1011.571	1011.164	1011.808	1012.732
974.862	1012.610	1016.382	1012.736	1012.038	1015.497	1000.135	1012.132	1011.952	1026.975	1011.145	1011.454	974.825	1011.512	1012.047	1062.503	1011.046	970.711	1012.946	1011.243	1011.487	1003.897	1011.468	1013.414	1012.003	1010.875	1011.280	1024.038	1039.380	1011.098	1012.008	1010.776	1011.573	1011.166	1011.865	1012.772
974.841	1012.577	1016.365	1012.700	1012.008	1015.509	1000.092	1012.156	1011.984	1026.962	1011.171	1011.443	974.826	1011.567	1012.039	1062.488	1011.032	970.715	1012.983	1011.187	1011.442	1003.905	1011.448	1013.408	1011.974	1010.837	1011.264	1024.038	1039.415	1011.086	1012.042	1010.779	1011.574	1011.141	1011.826	1012.772

代码：

import xlwt
import codecs
import os
"""
├─tdms_scripts
│  ├─excel
│  └─tdms
│──tmds2excel.py
"""

# 改为你的tmds文件夹！的路径，比如0静止
root_path = r'F:\rl_code\tdms_scripts'
# 修改这里，第0个为1，第1个从1001开始
task_num = 0
# 创建一个excel文件夹
try:
    os.mkdir(os.path.join(root_path, 'excel'+str(task_num)))
except:
    pass
tdms_files = os.listdir(os.path.join(root_path, 'tdms'))
# 将文件名从小到大排序
tdms_files.sort()
# [print(t) for t in tdms_files]

file_count = 1 + task_num * 1000
for tdms_file in tdms_files:
    print("tdms_file:", tdms_file, end='\t')
    sheetName = 'Sheet1'
    start_row = 0
    start_col = 0

    wb = xlwt.Workbook(encoding = 'utf-8')
    ws = wb.add_sheet(sheetName)
    tdms_path = os.path.join(root_path, 'tdms', tdms_file)
    f = open(tdms_path, encoding = 'utf-8')

    row_excel = start_row
    for line in f:
        # 消除换行符
        line = line.strip('\n')
        # 以制表符'\t'作为切分对象，将长字符串，转为短字符串列表；
        # 如果原数据用空格切分，则line = line.split(' ')
        line = line.split('\t')

        col_excel = start_col
        len_line = len(line)
        for j in range(len_line):            
            ws.write(row_excel,col_excel,line[j])
            col_excel += 1
            excel_path = os.path.join(root_path, 'excel'+str(task_num), str(file_count)+'.xls')            
            # 如果用原名，用下面的句子
            # excel_path = os.path.join(root_path, 'excel', tdms_file.replace('tdms', 'xls'))            
            wb.save(excel_path)

        row_excel += 1
    
    f.close
    print("excel_path:", excel_path)
    file_count += 1

进程池版本：

由于单个文件处理时间过长，因此写了一个多进程版本的处理脚本。
可以CPU核心数倍加速处理！

from multiprocessing import Pool
import xlwt
import codecs
import os
import time
"""
├─tdms_scripts
│  ├─excel
│  └─tdms
│──tmds2excel.py
"""

# 改为你的tmds文件夹！的路径，比如0静止
root_path = r'F:\rl_code\calibration\tdms_scripts'
# 修改这里，第0个为1，第1个从1001开始
task_num = 0
# 创建一个excel文件夹
try:
    os.mkdir(os.path.join(root_path, 'excel'+str(task_num)))
except:
    pass
tdms_files = os.listdir(os.path.join(root_path, 'tdms'))
# 将文件名从小到大排序
tdms_files.sort()
# [print(t) for t in tdms_files]

file_count = 1 + task_num * 1000


def tdms2excel(tdms_index_file):
    st = time.time()

    index, tdms_file = tdms_index_file
    print("tdms_file:", tdms_file, end='\t')
    sheetName = 'Sheet1'
    start_row = 0
    start_col = 0
    wb = xlwt.Workbook(encoding = 'utf-8')
    ws = wb.add_sheet(sheetName)
    tdms_path = os.path.join(root_path, 'tdms', tdms_file)
    f = open(tdms_path, encoding = 'utf-8')

    row_excel = start_row
    for line in f:
        # 消除换行符
        line = line.strip('\n')
        # 以制表符'\t'作为切分对象，将长字符串，转为短字符串列表；
        # 如果原数据用空格切分，则line = line.split(' ')
        line = line.split('\t')
        # print("line:",  line)

        col_excel = start_col
        len_line = len(line)
        for j in range(len_line):            
            ws.write(row_excel, col_excel, line[j])
            col_excel += 1
            excel_path = os.path.join(root_path, 'excel'+str(task_num), str(index)+'.xls')            
            # 如果用原名，用下面的句子
            # excel_path = os.path.join(root_path, 'excel', tdms_file.replace('tdms', 'xls'))            
            wb.save(excel_path)

        row_excel += 1
    
    f.close
    print("excel_path:", excel_path, end='\t')
    print("trans_time:", time.time()-st)


if __name__ == '__main__':
    p = Pool()    
    p.map(tdms2excel, [[index, tdms_file] for index, tdms_file in enumerate(tdms_files)])

我的笔记本是八核2.6Ghz的CPU，单个进程的处理时间在20-30秒之间，我用了进程池之后，八个进程一起执行，可以很明显的看到每隔8个，有一个时间分层，八个小工轮着干的效果就体现出来了~

这个比手动打开八个脚本，利用Python新开的脚本自动在另外的CPU核心运行的特性，利用其他CPU核心的方法，要优雅多了~

运行效果：

tdms_file: 1_2021_02_21_22_41_15.tdms excel_path: …cel0\0.xls single_time: 27.0803 total_time: 27.0957
tdms_file: 1_2021_02_21_22_41_58.tdms excel_path: …cel0\4.xls single_time: 27.4848 total_time: 27.5031
tdms_file: 1_2021_02_21_22_41_39.tdms excel_path: …cel0\2.xls single_time: 28.6732 total_time: 28.6935
tdms_file: 1_2021_02_21_22_41_48.tdms excel_path: …cel0\3.xls single_time: 29.9796 total_time: 29.9925
tdms_file: 1_2021_02_21_22_41_26.tdms excel_path: …cel0\1.xls single_time: 30.8471 total_time: 30.8605
tdms_file: 1_2021_02_21_22_42_24.tdms excel_path: …cel0\7.xls single_time: 31.049 total_time: 31.0644
tdms_file: 1_2021_02_21_22_42_14.tdms excel_path: …cel0\6.xls single_time: 32.2027 total_time: 32.2215
tdms_file: 1_2021_02_21_22_42_05.tdms excel_path: …cel0\5.xls single_time: 33.8177 total_time: 33.8405
tdms_file: 1_2021_02_21_22_42_32.tdms excel_path: …cel0\8.xls single_time: 33.5036 total_time: 60.6003
tdms_file: 1_2021_02_21_22_43_05.tdms excel_path: …el0\12.xls single_time: 29.8122 total_time: 60.6732
tdms_file: 1_2021_02_21_22_44_03.tdms excel_path: …el0\14.xls single_time: 28.4923 total_time: 60.7139
tdms_file: 1_2021_02_21_22_42_40.tdms excel_path: …cel0\9.xls single_time: 33.4553 total_time: 60.9589
tdms_file: 1_2021_02_21_22_43_54.tdms excel_path: …el0\13.xls single_time: 29.904 total_time: 60.9688
tdms_file: 1_2021_02_21_22_44_14.tdms excel_path: …el0\15.xls single_time: 28.3613 total_time: 62.2024
tdms_file: 1_2021_02_21_22_42_57.tdms excel_path: …el0\11.xls single_time: 34.8927 total_time: 64.8862
tdms_file: 1_2021_02_21_22_42_49.tdms excel_path: …el0\10.xls single_time: 39.0403 total_time: 67.7338
tdms_file: 1_2021_02_21_22_44_58.tdms excel_path: …el0\20.xls single_time: 25.6077 total_time: 86.577
tdms_file: 1_2021_02_21_22_44_36.tdms excel_path: …el0\17.xls single_time: 30.7522 total_time: 91.4259
tdms_file: 1_2021_02_21_22_44_44.tdms excel_path: …el0\18.xls single_time: 33.2317 total_time: 93.9461
tdms_file: 1_2021_02_21_22_44_51.tdms excel_path: …el0\19.xls single_time: 33.3145 total_time: 94.2744
tdms_file: 1_2021_02_21_22_44_28.tdms excel_path: …el0\16.xls single_time: 34.4439 total_time: 95.0447
tdms_file: 1_2021_02_21_22_45_12.tdms excel_path: …el0\22.xls single_time: 33.2763 total_time: 98.1626
tdms_file: 1_2021_02_21_22_45_05.tdms excel_path: …el0\21.xls single_time: 37.2711 total_time: 99.4745
tdms_file: 1_2021_02_21_22_45_19.tdms excel_path: …el0\23.xls single_time: 34.0878 total_time: 101.8221
tdms_file: 1_2021_02_21_22_45_27.tdms excel_path: …el0\24.xls single_time: 30.2179 total_time: 116.7959
tdms_file: 1_2021_02_21_22_45_35.tdms excel_path: …el0\25.xls single_time: 25.7747 total_time: 117.2011
tdms_file: 1_2021_02_21_22_45_56.tdms excel_path: …el0\28.xls single_time: 25.1771 total_time: 120.2228
tdms_file: 1_2021_02_21_22_45_42.tdms excel_path: …el0\26.xls single_time: 26.8273 total_time: 120.7738
tdms_file: 1_2021_02_21_22_45_50.tdms excel_path: …el0\27.xls single_time: 27.3055 total_time: 121.5819
tdms_file: 1_2021_02_21_22_46_03.tdms excel_path: …el0\29.xls single_time: 23.8782 total_time: 122.0407

联系方式

ps: 欢迎做强化的同学加群一起学习：

深度强化学习-DRL：799378128

欢迎关注知乎帐号：未入门的炼丹学徒

CSDN帐号：https://blog.csdn.net/hehedadaq