# -*- coding: utf-8 -*- from playwright.sync_api import sync_playwright targets={ 'r166_기후변화':'https://www.komipo.co.kr/esg/content/102/main.do?mnCd=ESG020201', 'r179_인권위원회':'https://www.komipo.co.kr/esg/content/393/main.do?mnCd=ESG030301', 'r222_홍보영상':'https://www.komipo.co.kr/kor/content/126/main.do?mnCd=FN0902', 'r223_KOMIPOMEDIA':'https://www.komipo.co.kr/kor/content/215/main.do?mnCd=FN090202', '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':1280,'height':1400},ignore_https_errors=True) for n,u in targets.items(): try: pg.goto(u,timeout=30000,wait_until='networkidle') pg.wait_for_timeout(1500) el=pg.query_selector('#content') or pg.query_selector('div.contents') info={} if el: el.screenshot(path=f'_render/{n}.png') # detect media vids=pg.eval_on_selector_all('#content video, #content iframe', 'els=>els.map(e=>({t:e.tagName,src:e.src||e.currentSrc||""}))') imgs=pg.eval_on_selector_all('#content img', 'els=>els.filter(i=>i.naturalWidth>=200&&i.naturalHeight>=150).map(i=>({w:i.naturalWidth,h:i.naturalHeight,alt:i.alt,src:i.src.split("/").pop()}))') print(n,'| media:',vids,'| bigimgs:',len(imgs)) for im in imgs[:6]: print(' ',im['w'],'x',im['h'],im['alt'][:40],im['src'][:30]) except Exception as e: print(n,'ERR',str(e)[:60]) b.close()