modify scripts

This commit is contained in:
2025-03-15 17:39:25 +08:00
parent 3b76c00500
commit 7c0cf12704
4 changed files with 117 additions and 12 deletions

View File

@ -18,7 +18,7 @@ config.setup_logging()
debug = False
force = False
pdf_base_dir = "/root/hostdir/stock_data/pdfs" # 下载 PDF 存放目录
pdf_base_dir = f"{config.global_host_data_dir}/pdfs" # 下载 PDF 存放目录
map_pdf_page = {
@ -128,15 +128,18 @@ def download_pdf_stock_general(parse_func, tbl_name, querystr='', s_date=start_d
if file_path and os.path.isfile(file_path):
logging.info(f'{file_path} already exists. skipping...')
continue
# 获取pdf链接地址
if url:
pdf_url = em.fetch_pdf_link(url)
if pdf_url:
# 下载 PDF
down = em.download_pdf(pdf_url, file_path)
if down:
logging.info(f'saved file {file_path}')
# 获取pdf链接地址
pdf_url = em.fetch_pdf_link(url)
if pdf_url:
# 下载 PDF
down = em.download_pdf(pdf_url, file_path)
if down:
logging.info(f'saved file {file_path}')
else:
logging.warning(f'download pdf file error. file_path: {pdf_url}, save_path: {file_path}')
else:
logging.warning(f'cannot get pdf link. url: {url}, save_path: {file_path}')
time.sleep(1) # 避免请求过快