# -*- coding: utf-8 -*- from playwright.sync_api import sync_playwright targets={ 'r222_홍보영상':'https://www.komipo.co.kr/kor/content/126/main.do?mnCd=FN0902', 'r40_정보보호':'https://www.komipo.co.kr/kor/content/446/main.do?mnCd=FN0107', 'r16_489':'https://www.komipo.co.kr/kor/content/489/main.do', 'r84_24':'https://www.komipo.co.kr/kor/content/24/main.do', 'r221_135':'https://www.komipo.co.kr/kor/content/135/main.do', } import os os.makedirs('_render',exist_ok=True) with sync_playwright() as p: b=p.chromium.launch() pg=b.new_page(viewport={'width':1100,'height':1500},ignore_https_errors=True) for n,u in targets.items(): try: pg.goto(u,timeout=30000,wait_until='networkidle') pg.wait_for_timeout(1500) # main content area: try several el=pg.query_selector('#content')or pg.query_selector('.contents')or pg.query_selector('#container')or pg.query_selector('main') (el or pg).screenshot(path=f'_render/{n}.png') vids=pg.eval_on_selector_all('video, iframe', 'els=>els.filter(e=>{s=(e.src||e.currentSrc||"");return /youtube|vimeo|\\.mp4|player/.test(s)}).map(e=>e.src||e.currentSrc)') bigimg=pg.eval_on_selector_all('img','els=>els.filter(i=>i.naturalWidth>=250&&i.naturalHeight>=170).map(i=>i.alt+"|"+i.src.split("/").pop()).slice(0,8)') txt=(pg.inner_text('body')or'')[:160].replace('\n',' ') print('==',n,'==') print(' vids:',vids[:4]) print(' bigimg:',bigimg) print(' txt:',txt) except Exception as e: print(n,'ERR',str(e)[:70]) b.close()