# -*- coding: utf-8 -*-
from playwright.sync_api import sync_playwright
import json
BASE='https://www.kipf.re.kr'
def clean(h):
    if not h: return ''
    h=h.split(';jsessionid')[0]
    if h.startswith('http'): return h
    if h.startswith('/'): return BASE+h
    return h
with sync_playwright() as p:
    b=p.chromium.launch();pg=b.new_page(ignore_https_errors=True)
    pg.goto('https://www.kipf.re.kr/kor/',timeout=40000,wait_until='domcontentloaded')
    pg.wait_for_selector('a.th_1st',timeout=20000)
    pg.wait_for_timeout(1500)
    tree=pg.evaluate(r'''()=>{
      function txt(a){return (a.innerText||a.textContent||'').trim().replace(/\s+/g,' ');}
      const tops=[];
      document.querySelectorAll('a.th_1st').forEach(a=>{
        const li=a.closest('li');
        const node={t:txt(a), h:a.getAttribute('href'), mids:[]};
        li.querySelectorAll('ul.depth2_ul>li').forEach(ml=>{
          const ma=ml.querySelector(':scope>a');
          const m={t:ma?txt(ma):'?', h:ma?ma.getAttribute('href'):'', leaves:[]};
          ml.querySelectorAll(':scope>ul.depth3_ul>li>a').forEach(la=>{
            m.leaves.push({t:txt(la), h:la.getAttribute('href')});
          });
          node.mids.push(m);
        });
        tops.push(node);
      });
      return tops;
    }''')
    b.close()
for t in tree:
    t['h']=clean(t['h'])
    for m in t['mids']:
        m['h']=clean(m['h'])
        for l in m['leaves']: l['h']=clean(l['h'])
json.dump(tree,open('_gnb_tree.json','w',encoding='utf-8'),ensure_ascii=False,indent=1)
# summary
tot=0
for t in tree:
    leafcnt=sum(len(m['leaves']) or 1 for m in t['mids'])
    tot+=leafcnt
    print(f"TOP {t['t']} | mids={len(t['mids'])} | rows~{leafcnt}")
    for m in t['mids']:
        print(f"   E {m['t']} ({len(m['leaves'])}F)  {m['h']}")
print('총 예상 행 ~',tot)