df16c98366
공공기관2 작업 중. _temp 몽타주(재생성가능)는 제외. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
43 lines
2.1 KiB
Python
43 lines
2.1 KiB
Python
# -*- coding: utf-8 -*-
|
|
import sys, io, json
|
|
sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding='utf-8')
|
|
from playwright.sync_api import sync_playwright
|
|
|
|
URLS = [
|
|
"https://www.kesco.or.kr/contents/selectContents.do?contents_id=CKK00006", # 비전핵심가치(공사소개)
|
|
"https://www.kesco.or.kr/contents/selectContents.do?contents_id=MEB00002", # 사업분야
|
|
]
|
|
JS = """
|
|
() => {
|
|
// LNB 컨테이너 후보
|
|
const cands=[];
|
|
document.querySelectorAll('#snb,.snb,.lnb,#lnb,.sub_lnb,[class*=snb],[class*=lnb],[class*=left_menu],aside').forEach(el=>{
|
|
const a=el.querySelectorAll('a').length;
|
|
if(a>=2) cands.push({sel:(el.id?'#'+el.id:'.'+String(el.className).split(' ')[0]), a, d4:el.querySelectorAll('ul.depth4,.depth4').length});
|
|
});
|
|
cands.sort((x,y)=>y.a-x.a);
|
|
// 전체 depth ul 클래스 종류
|
|
const ulcls=[...new Set([...document.querySelectorAll('ul')].map(u=>String(u.className)).filter(c=>/depth/i.test(c)))];
|
|
// depth4 항목들
|
|
const d4=[...document.querySelectorAll('ul.depth4')].map(u=>[...u.querySelectorAll(':scope>li>a')].map(a=>a.textContent.trim()+'::'+a.getAttribute('href')));
|
|
// 가장 앵커많은 LNB의 전체 트리(ancestor ul depth)
|
|
const lnb=cands.length?document.querySelector(cands[0].sel):null;
|
|
let tree=[];
|
|
if(lnb){ lnb.querySelectorAll('a').forEach(a=>{let d=0,p=a.parentElement;while(p&&p!==lnb){if(p.tagName==='UL')d++;p=p.parentElement;}tree.push([d,a.textContent.trim(),a.getAttribute('href')]);}); }
|
|
return {cands:cands.slice(0,5), ulcls, d4, tree};
|
|
}
|
|
"""
|
|
with sync_playwright() as p:
|
|
br=p.chromium.launch(headless=True)
|
|
pg=br.new_context(ignore_https_errors=True, viewport={'width':1600,'height':1000}).new_page()
|
|
for u in URLS:
|
|
pg.goto(u, wait_until='domcontentloaded', timeout=30000); pg.wait_for_timeout(2000)
|
|
r=pg.evaluate(JS)
|
|
print('\n===', u[-12:], '===')
|
|
print('LNB cands:', r['cands'])
|
|
print('ul depth classes:', r['ulcls'])
|
|
print('depth4 items:', json.dumps(r['d4'], ensure_ascii=False))
|
|
print('LNB tree(앞30):')
|
|
for t in r['tree'][:30]: print(' ', t)
|
|
br.close()
|