playwright
2025年9月5日小于 1 分钟
from playwright.sync_api import sync_playwright
url = "https://www.tj.gov.cn/sy/zwdt/gqdt/202505/t20250521_6935789.html"
with sync_playwright() as p:
browser = p.chromium.launch(headless=True)
page = browser.new_page()
# 拦截请求,阻止图片、样式、字体、媒体资源加载
def block_resources(route, request):
if request.resource_type in ["image", "media", "stylesheet", "font", "script"]:
route.abort()
else:
route.continue_()
page.route("**/*", block_resources)
try:
page.goto(url, timeout=60000, wait_until="load")
html = page.content()
print(html)
except Exception as e:
print("页面加载失败:", e)
browser.close()