From 20bbed188fd7833d14801dc202fa3cdbb085ca2b Mon Sep 17 00:00:00 2001 From: xuyonghao <898441624@qq.com> Date: 星期五, 20 十二月 2024 11:51:34 +0800 Subject: [PATCH] 优化excel合并脚本 --- app/utils/excelmerge/conformity.py | 76 +++++++++++++++++--------------------- 1 files changed, 34 insertions(+), 42 deletions(-) diff --git a/app/utils/excelmerge/conformity.py b/app/utils/excelmerge/conformity.py index 128657d..6a609b3 100644 --- a/app/utils/excelmerge/conformity.py +++ b/app/utils/excelmerge/conformity.py @@ -3,6 +3,22 @@ import os +def clear_blank_rows(sheet): + last_row = sheet.max_row + for row in range(last_row, 1, -1): + if all(cell.value is None or cell.value == '' for cell in sheet[row]): + sheet.delete_rows(row) + + +def copy_data(source_sheet, target_sheet, start_row): + for row in range(start_row, source_sheet.max_row + 1): + a_cell_value = source_sheet.cell(row=row, column=1).value + if isinstance(a_cell_value, (int, float)) and any( + source_sheet.cell(row=row, column=col).value for col in range(4, source_sheet.max_column + 1)): + target_sheet.append( + [source_sheet.cell(row=row, column=col).value for col in range(1, source_sheet.max_column + 1)]) + + def run_conformity(): try: # 鍔犺浇妯℃澘鏂囦欢 @@ -10,14 +26,8 @@ template_excel = load_workbook(template_path) EXCEL_FILES_PATH = os.path.join('data', 'output') - # 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'] + template_sheets = {sheet.title: sheet for sheet in template_excel} - # 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('~$')] @@ -25,47 +35,29 @@ 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'] + # 鍔ㄦ�佽幏鍙栧伐浣滆〃 + source_sheets = {sheet.title: sheet for sheet in source_excel} - # 娓呴櫎鏃犳晥绌虹櫧琛� - for ssheet in [ssheet1, ssheet2, ssheet3, ssheet4, ssheet5]: - 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) + for name in template_sheets: + if name in source_sheets: + clear_blank_rows(source_sheets[name]) - # 澶嶅埗鏁版嵁 - for (ssheet, tsheet, start_point) in [ - (ssheet1, sheet1, 4), - (ssheet2, sheet2, 3), - (ssheet3, sheet3, 3), - (ssheet4, sheet4, 4), - (ssheet5, sheet5, 3), - ]: - 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)]) + for name, start_row in [('鎶�鏈洃鐫e伐浣滅粺璁¤〃', 4), ('鎶�鏈洃鐫e憡锛堥锛夎鍗曠粺璁¤〃', 3), + ('鎶曚骇鍓嶆妧鏈洃鐫f姤鍛婄粺璁¤〃', 3), ('鎶�鏈洃鐫g粏鍒欏畬鍠勫缓璁�', 4), + ('鍏稿瀷缁忛獙浜ゆ祦', 3)]: + if name in source_sheets and name in template_sheets: + copy_data(source_sheets[name], template_sheets[name], start_row) 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 + for name, start_row in [('鎶�鏈洃鐫e伐浣滅粺璁¤〃', 4), ('鎶�鏈洃鐫e憡锛堥锛夎鍗曠粺璁¤〃', 3), + ('鎶曚骇鍓嶆妧鏈洃鐫f姤鍛婄粺璁¤〃', 3), ('鎶�鏈洃鐫g粏鍒欏畬鍠勫缓璁�', 4), + ('鍏稿瀷缁忛獙浜ゆ祦', 3)]: + if name in template_sheets: + last_row = template_sheets[name].max_row + for i in range(start_row, last_row + 1): + template_sheets[name].cell(row=i, column=1).value = i - start_row + 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') template_excel.save(output_path) -- Gitblit v1.8.0