"https://oss.henan.gov.cn/sbgt-wztipt/attachment/hntjj/hntj/lib/tjnj/2023nj/zk/lefte.htm",2024:
"https://oss.henan.gov.cn/sbgt-wztipt/attachment/hntjj/hntj/lib/tjnj/2024nj/zk/indexce.htm"
=================
timeout=20)resp.raise_for_status()resp.encoding
'gb2312'except
href.startswith("http"):links.append(href)else:base
f"https://oss.henan.gov.cn/sbgt-wztipt/attachment/hntjj/hntj/lib/tjnj/{year}nj/zk/"links.append(base
linksdef
save_path):"""下载文件，如果已存在则跳过"""if
and
timeout=30)resp.raise_for_status()with
open(save_path,
f:f.write(resp.content)print(f"[OK]
下载成功:
clean_and_merge_by_category(year_folder,
for
{year_folder}")returncategory_dict
defaultdict(list)#
match.groups()category_dict[main_cat].append((f,
sub_cat))os.makedirs(output_dir,
exist_ok=True)for
category_dict.items():sheets_to_write
[]for
df.empty:continuesheet_name_clean
'',
sub_cat)[:31]sheets_to_write.append((df,
Exception
sheets_to_write:sheets_to_write.append((pd.DataFrame({"提示":
["此大类下无有效数据"]}),
os.path.exists(output_file):os.remove(output_file)with
engine='openpyxl')
sheets_to_write:df.to_excel(writer,
index=False,
"excel")os.makedirs(year_folder,
exist_ok=True)for
os.path.basename(link)save_path
save_path)output_dir
str(year))clean_and_merge_by_category(year_folder,
所有年份下载与整合完成！")
id="用大白话理解核心逻辑">用大白话理解核心逻辑
