近期对FTP及Excel数据处理的一些摸索

 1 #!/usr/bin/env python
 2 # -*- coding: utf-8 -*-
 3 import os,sys
 4 from datetime import datetime,timedelta
 5 
 6 from ftplib import FTP            #加载ftp模块
 7 import xlrd,xlwt
 8 
 9 
10 starttime = input("请输入想要下载文件的开始时间（格式：20170101）：\n")
11 # endtime = parse(input("请输入想要下载文件的结束时间：\n"))                       
12 
13 # input_time = int(input("请输入想要下载前多少天的数据：\n"))
14 
15 base_dir = os.path.dirname(os.path.abspath(__file__))
16 print(base_dir)
17 sys.path.append(base_dir)
18 
19 ftp=FTP()                         #设置变量
20 #ftp.set_debuglevel(2)             #打开调试级别2，显示详细信息
21 ftp.connect("IP地址",21)          #连接的ftp sever和端口
22 ftp.login("用户名","密码>")      #连接的用户名，密码
23 
24 
25 
26 if ftp.cwd("目录" + starttime + "/"):    #进入远程目录
27     #print(ftp.nlst())
28     #downloadlist = ftp.nlst()               
29     bufsize=1024                      #设置的缓冲区大小
30     
31     for filename in ftp.nlst():
32         
33         if "GHP" in filename and "0800_001.cmep" in filename:           #需要下载的文件
34             file_handle=open(base_dir + "/data/" + filename,"wb").write #以写模式在本地打开文件
35             #file_handle=open(filename,"wb").write
36             ftp.retrbinary("RETR "+filename,file_handle,bufsize) #接收服务器上文件并写入本地文件
37 #            ftp.set_debuglevel(0)             #关闭调试模式
38             print(filename + "已经下载完成！")
39             
40         
41     ftp.cwd("/")
42 
43 print("您想要的文件已经全部下载完成，欢迎下次使用！")
44 ftp.quit()                        #退出ftp

 1 import json
 2 import os,sys
 3 
 4 base_dir = os.path.dirname(os.path.abspath(__file__))
 5 print(base_dir)
 6 sys.path.append(base_dir)
 7 
 8 
 9 file_list = os.listdir(base_dir + "/data/")
10 # print(file_list)
11 
12 output_data = {"items":[]}
13 
14 for filename in file_list:
15     with open(base_dir + "/data/" + filename) as f:
16         lines = f.readlines() 
17         # print(lines)
18         line_number = []
19         for nums,line in enumerate(lines):
20             if line.split(",")[10] == "GJ" or line.split(",")[10] == "kWh":
21                 line_number.append(nums)
22                
23         out_filename = filename.split("_GHP")[0].split("SSTJEC_")[1]
24         print(out_filename)
25         data = {"item":out_filename,"download_line":line_number}
26         output_data["items"].append(data)
27         
28 with open("settings.json","w") as f1:
29     json.dump(output_data,f1)

 1 import os,sys,json
 2 import xlrd,xlwt
 3 
 4 #把当前文件所在目录添加到电脑的环境变量
 5 base_dir = os.path.dirname(os.path.abspath(__file__))
 6 print(base_dir)
 7 sys.path.append(base_dir)
 8 
 9 #当前文件所在目录data文件夹下的所有文件名以列表的形式存到file_list里
10 file_list = os.listdir(base_dir+"/data")
11 print(file_list)
12 
13 #打开当前文件所在目录的settings.json文件
14 with open("settings.json","r") as f1:
15     settings = json.load(f1)
16 
17 #以写的形式打开一个workbook
18 my_workbook = xlwt.Workbook()
19 
20 for filename in file_list:
21     year = filename.split("_")[-2][:4]
22     date = filename.split("_")[-2][4:8]
23 
24     #从json文件中获取到每个项目的配置文件，然后下载相应的行到Excel里
25     for setting in settings["items"]:
26         project_name = filename.split("_GHP")[0].split("SSTJEC_")[1]
27         if project_name == setting["item"]:
28             my_sheet = my_workbook.add_sheet(project_name + "_" + date)
29             with open(base_dir+"/data/"+filename) as f2:
30                 lines = f2.readlines()
31                 step = 0
32 
33                 #把所需要的数据写到Excel中
34                 for nums,line in enumerate(lines):
35                     if nums in setting["download_line"]:
36                         re_time = line.split(",")[6]
37                         out_file = line.split(",")[7]
38                         energy = line.split(",")[-2]
39                         unit = line.split(",")[10]
40 
41                         my_sheet.write(step,0,out_file)
42                         my_sheet.write(step,1,re_time)
43                         my_sheet.write(step,2,energy)
44                         my_sheet.write(step,3,unit)
45                         step += 1
46 
47 my_workbook.save(base_dir+"/handle/" + year +" cmep_to_excel.xls")

 1 import xlrd,xlwt
 2 import os,sys
 3 
 4 base_dir = os.path.dirname(os.path.abspath(__file__))
 5 print(base_dir)
 6 sys.path.append(base_dir)
 7 
 8 file_list = os.listdir(base_dir+"/handle/")
 9 print(file_list)
10 
11 my_workbook = xlwt.Workbook()
12 
13 for filename in file_list:
14     date,project = filename.split(" ")[0],filename.split(" ")[-1]
15 
16     if project == "cmep_to_excel.xls":
17 
18         myworkbook = xlrd.open_workbook((base_dir+"/handle/" + filename))
19         i = 0
20         o = 1
21         while o < len(myworkbook.sheets()):
22             frist_sheet = myworkbook.sheets()[i]
23             second_sheet = myworkbook.sheets()[o]
24             rows_num = second_sheet.nrows
25             print(rows_num)
26 
27             my_sheet = my_workbook.add_sheet(frist_sheet.cell(0,0).value.split("_GHP_")[0])
28             
29             #在每个表单的头一行添加表头
30             my_sheet.write(0,0,"数据类型")
31             my_sheet.write(0,1,"日期")
32             my_sheet.write(0,2,"数据")
33             my_sheet.write(0,3,"单位")
34             m = 1
35             n = 0
36             while n < rows_num:
37                 
38                 out_file = frist_sheet.cell(n,0).value
39                 re_time = frist_sheet.cell(n,1).value
40                 energy1 = float(frist_sheet.cell(n,2).value)
41                 unit = frist_sheet.cell(n,3).value
42 
43                 energy2 = float(second_sheet.cell(n,2).value)
44                 energy_output = energy2 - energy1
45 
46                 my_sheet.write(m,0,out_file)
47                 my_sheet.write(m,1,re_time)
48                 my_sheet.write(m,2,energy_output)
49                 my_sheet.write(m,3,unit)
50 
51                 m +=1
52                 n +=1
53             i += 2
54             o += 2
55 
56         my_workbook.save(base_dir+"/handle/" + date + " result.xls")