import os import win32com.client as win32 import shutil import openpyxl import os import win32com.client as win32 import shutil import openpyxl import time import time #####part one ws = {'资产负债表':0,'利润表':0,'纳税申报表':0,'纳税申报表':0 } mydir = 'bak' #基本逻辑就是分两种情况,文件夹有和无,有时,为了避免里面数据污染,删了重建一个;没有则是正常建一个就好 if os.path.exists(mydir) == 0 : os.mkdir(mydir) else: shutil.rmtree('bak') os.mkdir(mydir) file = [] path = os.getcwd() # 转换前程序所在文件目录 path_bak = os.getcwd() + r'/bak/' pathdir = os.listdir(path) #print(pathdir) for s in pathdir: newdir = os.path.join(path, s) if os.path.isfile(newdir): os.chdir(path_bak) pathdir_bak = os.listdir(path_bak) # print(pathdir_bak) if os.path.splitext(newdir)[1] == ".xls": t = os.path.split(s)[1] + r'x' if os.path.exists(t): # print(os.path.exists(os.path.split(s)[1])) break else: excel = win32.gencache.EnsureDispatch('Excel.Application') wb = excel.Workbooks.Open(newdir) # print(newdir) path_dir = os.path.split(newdir)[0] path_file = os.path.split(newdir)[1] # print(path_dir) # print(path_file) wb.SaveAs(path_dir + '/bak/' + path_file + 'x', FileFormat=51) # FileFormat = 51 is for .xlsx extension wb.Close() # FileFormat = 56 is for .xls extension excel.Application.Quit() elif os.path.splitext(newdir)[1] == ".xlsx": file.append(newdir) elif os.path.isdir(newdir): continue for i in range(len(file)): pcm_file = file[i] path_0 = os.path.split(file[i])[0] path_1 = os.path.split(file[i])[1] oldname = path_0 + '/' + path_1 newname = path_0 + r'/bak/' + path_1 shutil.copyfile(oldname, newname) #time.sleep(5) #暂停10秒 #####第二步将文件目录切换至bak文件夹中,,切片u v w file_deal = [] path_deal = os.getcwd() # 转换前程序所在文件目录 pathdir_deal = os.listdir(path_deal) for u in pathdir_bak: newdir_deal = os.path.join(path_deal, u) file_deal.append(newdir_deal) # print(file_deal) for j in range(len(file_deal)): pcm_file_deal = file_deal[j] path_0_deal = os.path.split(file_deal[j])[0] #文件所在目录 path_1_deal = os.path.split(file_deal[j])[1] #文件名称带后缀 # print(pcm_file_deal) # print(path_0_deal) #print(path_1_deal) wb = openpyxl.load_workbook(path_1_deal) allSheets = wb.get_sheet_names() for i in range(len(allSheets)): sheet = wb.get_sheet_by_name(allSheets[i]) for row in sheet.iter_rows(): for cell in row: if str(str(cell.value).replace(' ', '')).find("资产负债表") != -1: sheet_selected = wb.get_sheet_by_name(allSheets[i]) #print(sheet_selected) #print(path_1_deal) #fn_fuzhaibiao = path_1_deal #print(fn_fuzhaibiao) ws['资产负债表'] = wb.get_sheet_by_name(sheet_selected.title) continue elif str(str(cell.value).replace(' ', '')).find("利润表") != -1 or str(str(cell.value).replace(' ', '')).find("损益表") != -1: sheet_selected = wb.get_sheet_by_name(allSheets[i]) #print(sheet_selected) #print(path_1_deal) #fn_lirunbiao = path_1_deal ws['利润表'] = wb.get_sheet_by_name(sheet_selected.title) continue elif str(str(cell.value).replace(' ', '')).find("一般纳税人适用") != -1 or str(str(cell.value).replace(' ', '')).find("小规模纳税人适用") != -1: sheet_selected = wb.get_sheet_by_name(allSheets[i]) #print(sheet_selected) #print(path_1_deal) #fn_nashuishenbaobiao = path_1_deal ws['纳税申报表'] = wb.get_sheet_by_name(sheet_selected.title) continue elif str(str(cell.value).replace(' ', '')).find("财务状况表") != -1 or str(str(cell.value).replace(' ', '')).find("***子表开始") != -1: sheet_selected = wb.get_sheet_by_name(allSheets[i]) #print(sheet_selected) #print(path_1_deal) #fn_nashuishenbaobiao = path_1_deal ws['财务状况表'] = wb.get_sheet_by_name(sheet_selected.title) continue ################part two def getdate(ws_source,date_row, date_column): row_result = 0 column_result = 0 result = 0 for row in ws_source.iter_rows(): for cell in row: for i in range(len(date_row)): if str(str(cell.value).replace(' ', '')).find(date_row[i]) != -1 and len(str(cell.value).replace(' ', '')) == len(date_row[i]) : row_result = cell.row #print(row_result) break for row in ws_source.iter_rows(): for cell in row: for i in range(len(date_column)): if str(str(cell.value).replace(' ', '')).find(date_column[i]) != -1 and len(str(cell.value).replace(' ', '')) == len(date_column[i]) : column_result = cell.column #print(column_result) break if row_result == 0 or column_result == 0: result = 0 else: if ws_source.cell(row = row_result , column = column_result).value != None: #print(ws_source.cell(row=row_result, column=column_result).value) result = ws_source.cell(row=row_result, column=column_result).value return result def writedate(ws_source,date_row, date_column,data): row_result = 0 column_result = 0 result = 0 for row in ws_source.iter_rows(): for cell in row: for i in range(len(date_row)): if str(str(cell.value).replace(' ', '')).find(date_row[i]) != -1 and len(str(cell.value).replace(' ', '')) == len(date_row[i]) : row_result = cell.row #print(row_result) break for row in ws_source.iter_rows(): for cell in row: for i in range(len(date_column)): if str(str(cell.value).replace(' ', '')).find(date_column[i]) != -1 and len(str(cell.value).replace(' ', '')) == len(date_column[i]) : column_result = cell.column #print(column_result) break ws_source.cell(row=row_result, column=column_result).value = data #print("********") print(ws_source.cell(row=row_result, column=column_result).value) #print("********") ##############################part three if __name__ == '__main__': print(ws['资产负债表']) #资产负债表 print(getdate(ws['资产负债表'],['存货'],['年初余额','年初数'])); writedate(ws['财务状况表'], ['一、年初存货'], ['1-本季'], getdate(ws['资产负债表'],['存货'],['年初余额','年初数']))