From 2aca3fa30fac2aa5790f6b598dad7e2e59173012 Mon Sep 17 00:00:00 2001 From: zhangxiao <898441624@qq.com> Date: 星期五, 18 十月 2024 17:07:30 +0800 Subject: [PATCH] 修改python Excel合并脚本。由xlwings改为openpyxl --- app/utils/excelmerge/conformity.py | 89 ++++++++++++++++++++++++-------------------- 1 files changed, 49 insertions(+), 40 deletions(-) diff --git a/app/utils/excelmerge/conformity.py b/app/utils/excelmerge/conformity.py index 901357c..7276d06 100644 --- a/app/utils/excelmerge/conformity.py +++ b/app/utils/excelmerge/conformity.py @@ -1,37 +1,48 @@ -import xlwings as xw +from openpyxl import load_workbook +from openpyxl.utils import get_column_letter from datetime import datetime import os -import shutil def run_conformity(): - - source_folder = os.path.join('data', 'source') + # 鍔犺浇妯℃澘鏂囦欢 template_path = os.path.join('app', 'utils', 'excelmerge', '鍥界綉涓婃捣鐢靛姏鏁村悎妯$増.xlsx') + template_excel = load_workbook(template_path) EXCEL_FILES_PATH = os.path.join('data', 'output') - app = xw.App(visible=False) - templateExcel = xw.Book(template_path) + # 鑾峰彇宸ヤ綔琛� + sheet1 = template_excel['鎶�鏈洃鐫e伐浣滅粺璁¤〃'] + sheet2 = template_excel['鎶�鏈洃鐫e憡锛堥锛夎鍗曠粺璁¤〃'] + sheet3 = template_excel['鎶曚骇鍓嶆妧鏈洃鐫f姤鍛婄粺璁¤〃'] + sheet4 = template_excel['鎶�鏈洃鐫g粏鍒欏畬鍠勫缓璁�'] + sheet5 = template_excel['鍏稿瀷缁忛獙浜ゆ祦'] - sheet1 = templateExcel.sheets['鎶�鏈洃鐫e伐浣滅粺璁¤〃'] - sheet2 = templateExcel.sheets['鎶�鏈洃鐫e憡锛堥锛夎鍗曠粺璁¤〃'] - sheet3 = templateExcel.sheets['鎶曚骇鍓嶆妧鏈洃鐫f姤鍛婄粺璁¤〃'] - sheet4 = templateExcel.sheets['鎶�鏈洃鐫g粏鍒欏畬鍠勫缓璁�'] - sheet5 = templateExcel.sheets['鍏稿瀷缁忛獙浜ゆ祦'] - + # 鑾峰彇婧愭枃浠惰矾寰� + source_folder = os.path.join('data', 'source') source_files = [f for f in os.listdir(source_folder) if f.endswith('.xlsx') and not f.startswith('~$')] for file in source_files: - sourceExcel = xw.Book(os.path.join(source_folder, file)) - ssheet1 = sourceExcel.sheets['鎶�鏈洃鐫e伐浣滅粺璁¤〃'] - ssheet2 = sourceExcel.sheets['鎶�鏈洃鐫e憡锛堥锛夎鍗曠粺璁¤〃'] - ssheet3 = sourceExcel.sheets['鎶曚骇鍓嶆妧鏈洃鐫f姤鍛婄粺璁¤〃'] - ssheet4 = sourceExcel.sheets['鎶�鏈洃鐫g粏鍒欏畬鍠勫缓璁�'] - ssheet5 = sourceExcel.sheets['鍏稿瀷缁忛獙浜ゆ祦'] + source_path = os.path.join(source_folder, file) + source_excel = load_workbook(source_path) + + # 鑾峰彇婧愬伐浣滆〃 + ssheet1 = source_excel['鎶�鏈洃鐫e伐浣滅粺璁¤〃'] + ssheet2 = source_excel['鎶�鏈洃鐫e憡锛堥锛夎鍗曠粺璁¤〃'] + ssheet3 = source_excel['鎶曚骇鍓嶆妧鏈洃鐫f姤鍛婄粺璁¤〃'] + ssheet4 = source_excel['鎶�鏈洃鐫g粏鍒欏畬鍠勫缓璁�'] + ssheet5 = source_excel['鍏稿瀷缁忛獙浜ゆ祦'] + + # 娓呴櫎鏃犳晥绌虹櫧琛� for ssheet in [ssheet1, ssheet2, ssheet3, ssheet4, ssheet5]: - last_row_value = ssheet.range(f'A{ssheet.used_range.rows.count}').value - if isinstance(last_row_value, str): - ssheet.range(f'A{ssheet.used_range.rows.count}:AA{ssheet.used_range.rows.count}').delete() + last_row = ssheet.max_row + if last_row > 1: + for row in range(last_row, 1, -1): + if all(ssheet.cell(row=row, column=col).value is None or ssheet.cell(row=row, + column=col).value == '' + for col in range(1, ssheet.max_column + 1)): + ssheet.delete_rows(row) + + # 澶嶅埗鏁版嵁 for (ssheet, tsheet, start_point) in [ (ssheet1, sheet1, 4), (ssheet2, sheet2, 3), @@ -39,26 +50,24 @@ (ssheet4, sheet4, 4), (ssheet5, sheet5, 3), ]: - for row in range(start_point, ssheet.used_range.rows.count + 1): - a_cell_value = ssheet.range(f'A{row}').value - if not isinstance(a_cell_value, (int, float)): - continue - if not all(cell.value is None or cell.value == '' for cell in ssheet.range(f'B{row}:AA{row}')): - tsheet.range(f'B{tsheet.used_range.rows.count + 1}').value = ssheet.range(f'B{row}:AA{row}').value - sourceExcel.close() - for tsheet, start_point in [(sheet1, 4), (sheet2, 3), (sheet3, 3), (sheet4, 4), (sheet5, 3)]: - last_row = tsheet.used_range.rows.count - number = 1 - for i in range(start_point, last_row + 1): - tsheet.range(f'A{i}').value = number - number += 1 + for row in range(start_point, ssheet.max_row + 1): + a_cell_value = ssheet.cell(row=row, column=1).value + if isinstance(a_cell_value, (int, float)) and any( + ssheet.cell(row=row, column=col).value for col in range(2, ssheet.max_column + 1)): + tsheet.append([ssheet.cell(row=row, column=col).value for col in range(1, ssheet.max_column + 1)]) - timestamp = datetime.now().strftime('%Y-%m-%d_%H-%M-%S') + source_excel.close() + + # 鍦ㄧ洰鏍囧伐浣滆〃涓坊鍔犲簭鍙� + for tsheet, start_point in [(sheet1, 4), (sheet2, 3), (sheet3, 3), (sheet4, 4), (sheet5, 3)]: + last_row = tsheet.max_row + for i in range(start_point, last_row + 1): + tsheet.cell(row=i, column=1).value = i - start_point + 1 + + # 淇濆瓨鏂囦欢 + timestamp = datetime.now().strftime('%Y_%m_%d_%H_%M_%S') output_path = os.path.join(EXCEL_FILES_PATH, f'{timestamp}.xlsx') - templateExcel.save(output_path) - templateExcel.close() - app.quit() + template_excel.save(output_path) + template_excel.close() return timestamp - - -- Gitblit v1.8.0