df16c98366
공공기관2 작업 중. _temp 몽타주(재생성가능)는 제외. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
28 lines
1.3 KiB
Python
28 lines
1.3 KiB
Python
# -*- coding: utf-8 -*-
|
|
import urllib.request, ssl, re
|
|
from bs4 import BeautifulSoup
|
|
ctx=ssl.create_default_context(); ctx.check_hostname=False; ctx.verify_mode=ssl.CERT_NONE
|
|
HDR={'User-Agent':'Mozilla/5.0 (Windows NT 10.0; Win64; x64) Chrome/124 Safari/537.36'}
|
|
def fetch(u): return urllib.request.urlopen(urllib.request.Request(u,headers=HDR),timeout=20,context=ctx).read()
|
|
|
|
u='https://www.komipo.co.kr/kor/content/50/main.do?mnCd=FN02010802'
|
|
soup=BeautifulSoup(fetch(u),'html.parser')
|
|
# locate banner01 and a character img, print ancestor chain classes
|
|
def chain(img):
|
|
out=[]
|
|
p=img
|
|
for _ in range(8):
|
|
p=p.parent
|
|
if not p or p.name=='[document]': break
|
|
out.append(p.name+'.'+'.'.join(p.get('class') or []) + ('#'+p.get('id') if p.get('id') else ''))
|
|
return ' < '.join(out)
|
|
for im in soup.find_all('img'):
|
|
src=im.get('src','')
|
|
if 'ecomi_main_character' in src or 'banner01' in src:
|
|
print(src[:50])
|
|
print(' ',chain(im));print()
|
|
# look for editor content div (where real content lives)
|
|
for sel in ['div.con_area','div.cont_area','div.editor','div.fr-view','div.cont','div.view_cont','div.bbsView','div.sub_cont','div.subContents','.cms_content','.contents .con']:
|
|
el=soup.select_one(sel)
|
|
if el: print('FOUND',sel,'imgs',len(el.find_all('img')))
|