DB_JOB/작업파일/공공기관3/_probe_gnb.py
hehihoho3 b0acd1e705 백업: 셀병합 정상화(D~J)+KDHC 옵션2/행높이17, 오늘(06-21) 제출 17곳 검수·압축
- 공공기관2/3 작업본 + 오늘 제출 17곳 D~J 카테고리 셀병합 정상화
- 한국지역난방공사 옵션2(고아셀 F98 수정)+전행 높이17
- 제출_프리랜서2_2026-06-21.zip 생성(17개 xlsx, 2,468행)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-21 13:21:26 +09:00

114 lines
4.8 KiB
Python

# -*- coding: utf-8 -*-
"""13개 사이트 GNB 메가메뉴 자동탐지 프로빙 → _gnb/probe.json + 콘솔요약."""
import sys, io, os, json, re, warnings
sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding='utf-8')
warnings.filterwarnings('ignore')
from playwright.sync_api import sync_playwright
GNB_DIR = r"D:\01.프로젝트\DB수집\작업파일\공공기관3\_gnb"
os.makedirs(GNB_DIR, exist_ok=True)
SITES = {
1: ("한국원자력환경공단", "https://www.korad.or.kr"),
2: ("한국의료분쟁조정중재원", "https://www.k-medi.or.kr"),
3: ("한국의약품안전관리원", "https://www.drugsafe.or.kr"),
4: ("한국인터넷진흥원", "https://www.kisa.or.kr"),
5: ("한국임업진흥원", "https://www.kofpi.or.kr"),
6: ("한국자산관리공사", "https://www.kamco.or.kr"),
7: ("한국잡월드", "https://www.koreajobworld.or.kr"),
8: ("한국장기조직기증원", "https://www.koda1458.kr"),
9: ("한국장애인개발원", "https://www.koddi.or.kr"),
10: ("한국장애인고용공단", "https://www.kead.or.kr"),
11: ("한국장학재단", "https://www.kosaf.go.kr"),
12: ("한국재정정보원", "https://www.fis.kr"),
13: ("한국저작권보호원", "https://www.kcopa.or.kr"),
}
# 후보 GNB 컨테이너 셀렉터 (계층 앵커 최다인 것 선택)
CAND_SELS = ["#gnb","ul#gnb","#gnbmenu",".gnb",".gnb_area",".gnb_wrap","nav .gnb",
"#header nav ul","header nav","#nav",".nav",".lnb_all",".allmenu",
".all_menu","#allMenu",".total_menu",".totalmenu",".menu_all",
"ul.depth1","ul.dep1",".gnb ul",".navi",".navbar","#snb"]
PROBE_JS = """
(sels) => {
function topUls(el){
return [...el.querySelectorAll('ul')].filter(ul=>{
let p=ul.parentElement;
while(p&&p!==el){ if(p.tagName==='UL') return false; p=p.parentElement; }
return true; });
}
function countTree(el){
// count anchors with text inside nested uls (proxy for menu richness) + max depth
let n=0, md=0;
function walk(ul, depth){
[...ul.children].filter(li=>li.tagName==='LI').forEach(li=>{
let a=null;
for(const x of li.querySelectorAll('a')){
let p=x.parentElement, inside=false;
while(p&&p!==li){ if(p.tagName==='UL'){inside=true;break;} p=p.parentElement;}
if(!inside && (x.textContent||'').trim()){a=x;break;}
}
if(a){ n++; md=Math.max(md,depth); }
topUls(li).forEach(s=>walk(s,depth+1));
});
}
(el.tagName==='UL'?[el]:topUls(el)).forEach(u=>walk(u,1));
return {n, md};
}
const out=[];
for(const sel of sels){
let best=null;
document.querySelectorAll(sel).forEach(el=>{
const r=countTree(el);
if(!best || r.n>best.n) best=r;
});
if(best && best.n>=3) out.push({sel, n:best.n, md:best.md});
}
out.sort((a,b)=> (b.n*10+b.md)-(a.n*10+a.md));
// d1 labels of the winner
let d1=[];
if(out.length){
const el=document.querySelector(out[0].sel);
const uls = el.tagName==='UL'?[el]:topUls(el);
uls.forEach(u=>[...u.children].filter(li=>li.tagName==='LI').forEach(li=>{
for(const a of li.querySelectorAll('a')){
let p=a.parentElement, inside=false;
while(p&&p!==li){ if(p.tagName==='UL'){inside=true;break;} p=p.parentElement;}
if(!inside && (a.textContent||'').trim()){ d1.push(a.textContent.trim().replace(/\\s+/g,' ')); break;}
}
}));
}
return {cands: out.slice(0,5), d1: d1.slice(0,15)};
}
"""
def run():
res={}
with sync_playwright() as p:
br=p.chromium.launch(headless=True)
ctx=br.new_context(ignore_https_errors=True, viewport={'width':1920,'height':1080},
user_agent='Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 Chrome/120 Safari/537.36')
for num,(name,base) in SITES.items():
pg=ctx.new_page()
info={'num':num,'name':name,'base':base}
try:
pg.goto(base+'/', wait_until='domcontentloaded', timeout=35000)
pg.wait_for_timeout(2800)
r=pg.evaluate(PROBE_JS, CAND_SELS)
info.update(r); info['ok']=True
info['final_url']=pg.url
except Exception as e:
info['ok']=False; info['err']=f'{type(e).__name__}:{e}'
res[num]=info
pg.close()
c=info.get('cands',[])
top=c[0] if c else None
print(f"[{num:2}.{name}] {'OK' if info.get('ok') else 'FAIL'} top={top} d1#={len(info.get('d1',[]))}")
br.close()
io.open(os.path.join(GNB_DIR,'probe.json'),'w',encoding='utf-8').write(json.dumps(res,ensure_ascii=False,indent=1))
print('saved probe.json')
if __name__=='__main__':
run()