# -*- coding: utf-8 -*- import sys,io sys.stdout=io.TextIOWrapper(sys.stdout.buffer,encoding='utf-8') from playwright.sync_api import sync_playwright ROWS={ 'r16_자체공시a':'https://www.ksponco.or.kr/menu.es?mid=a10109020800', 'r17_자체공시b':'https://www.ksponco.or.kr/menu.es?mid=a10109020900', 'r54_협력기관':'https://www.ksponco.or.kr/menu.es?mid=a10305000000', 'r56_홍보동영상':'https://www.ksponco.or.kr/menu.es?mid=a10307000000', 'r104_ESG경영':'https://www.ksponco.or.kr/menu.es?mid=a10506000000', 'r42_고객서비스헌장':'https://www.ksponco.or.kr/menu.es?mid=a10202010000', 'r111_윤리경영활동':'https://www.ksponco.or.kr/menu.es?mid=a10502050000', 'r132_홈피이용안내':'https://www.ksponco.or.kr/menu.es?mid=a10705000000', } dec=r"logo|icon|btn|mark_|sns|footer|header|banner|bg_|sprite|blank|no_img|arrow|family|common|opentype|btn_|ico_" with sync_playwright() as p: br=p.chromium.launch(headless=True); ctx=br.new_context(ignore_https_errors=True,viewport={'width':1280,'height':2200}) for nm,u in ROWS.items(): pg=ctx.new_page() try: pg.goto(u,wait_until='networkidle',timeout=30000); pg.wait_for_timeout(1200) r=pg.evaluate("""(dec)=>{ const re=new RegExp(dec,'i'); let cont=document.querySelector('#contents,.contents,#content,.board_view,.view_cont,.cont,.sub_cont,.bbs_view')||document.body; let imgs=[...cont.querySelectorAll('img')].filter(im=>im.naturalWidth>=150&&im.naturalHeight>=110&&!re.test(im.src)).map(im=>im.src.split('/').pop().slice(0,26)+' '+im.naturalWidth+'x'+im.naturalHeight); let vid=[...cont.querySelectorAll('video,iframe')].filter(v=>/youtube|youtu\.be|vimeo|player/i.test(v.src||'')).length; return {n:imgs.length,imgs:imgs.slice(0,8),vid}; }""",dec) print(nm,'::',r) except Exception as e: print(nm,'ERR',str(e)[:40]) pg.close() br.close()