1、合并相同表头表格
代码如下
#python pandas合并表格插件 #coding:gbk import os,time import pandas as pd class Excel_concat(object): def __init__(self,FILES_PATH,NEW_FILE_PATH,FILES): self.files_path = FILES_PATH self.new_file_path = NEW_FILE_PATH self.file_list = FILES #文件列表 self.one_sheet = [] #要合并的子sheet1 self.two_sheet = [] #要合并的子sheet2 self.rows = ['登记证明编号','关联的初始登记编号'] #要处理的表头 def rows_utils(self,sheet): #列数据处理，将int数字形式转化为文本格式（字符串） for i in self.rows : sheet[i] = [str(i) for i in sheet[i]] return sheet def read_excel(self,file): #读取excel返回pd对象 file_sheet0 = self.rows_utils(pd.read_excel(os.path.join(self.files_path,file), sheet_name=0,encoding='gbk')) file_sheet1 = self.rows_utils(pd.read_excel(os.path.join(self.files_path,file), sheet_name=1,encoding='gbk')) return file_sheet0,file_sheet1 def write_excel(self): writer = pd.ExcelWriter(self.new_file_path) pd.concat(self.one_sheet).to_excel(writer, 'sheet1', index=False) pd.concat(self.two_sheet).to_excel(writer, 'sheet2', index=False) writer.save() def run(self): for i in self.file_list: try: print(i) sh0, sh1 = self.read_excel(i) self.one_sheet.append(sh0) self.two_sheet.append(sh1) except Exception as e: print('文件内容错误',i) self.write_excel() def subutils(data): #正则处理空白字符串（空格，换行符等） import re data = re.sub('[s]', '', data) return data if __name__ == '__main__': start = time.time() FILES_PATH = r'D:xxxxxxxxxxxxxxxxxxxx' NEW_FILE_PATH = r'D:xxxxxxxxxxxxx ewxls1.xlsx' FILES = os.listdir(FILES_PATH) obj = Excel_concat(FILES_PATH,NEW_FILE_PATH,FILES[0:2]) print(FILES) print(len(FILES)) # obj.run() # pf1,pf2 = read_excel(FILES[0]) # print(pf1) # print(pf2) end = time.time() print('Running Time is: ', str(end - start) + 's')
2、反解XML流
代码如下
#反解xml流，经过base64加密加zip打包 import base64 from io import BytesIO import zipfile content = b'<?xml version="1.0" encoding="gbk"?>' b'<BODY><xmlResultByte>' b'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' b'</xmlResultByte></BODY>' zip1=base64.b64decode(content) fio = BytesIO(zip1) myzip = zipfile.ZipFile(file=fio) print(myzip.namelist()) print(myzip.read(myzip.namelist()[0]).decode())