b0acd1e705
- 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>
79 lines
3.2 KiB
Python
79 lines
3.2 KiB
Python
# -*- coding: utf-8 -*-
|
|
"""3.한국의약품안전관리원 GNB 권위트리(.allmenu p.depth1/ul.depth2/depth3/depth4) 재구조화 → _gnb/3.json"""
|
|
import sys, io, json, warnings
|
|
sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding='utf-8')
|
|
warnings.filterwarnings('ignore')
|
|
from playwright.sync_api import sync_playwright
|
|
|
|
JS = r"""
|
|
() => {
|
|
const cont = document.querySelector('.allmenu');
|
|
if(!cont) return {err:'no allmenu'};
|
|
function childUl(li, cls){
|
|
for(const u of li.children){ if(u.tagName==='UL' && u.classList.contains(cls)) return u; }
|
|
return null;
|
|
}
|
|
function directA(li){
|
|
for(const a of li.children){ if(a.tagName==='A') return a; }
|
|
// sometimes wrapped
|
|
const a=li.querySelector(':scope > a'); return a;
|
|
}
|
|
const rows=[];
|
|
// top blocks: li containing p.depth1
|
|
const blocks = [...cont.querySelectorAll('p.depth1')].map(p=>p.closest('li'));
|
|
blocks.forEach(b=>{
|
|
const p1=b.querySelector(':scope > p.depth1');
|
|
const d1label=(p1.textContent||'').trim().replace(/\s+/g,' ');
|
|
rows.push({depth:1, label:d1label, href:''});
|
|
const u2=childUl(b,'depth2'); if(!u2) return;
|
|
[...u2.children].filter(li=>li.tagName==='LI').forEach(li2=>{
|
|
const a2=directA(li2); if(!a2) return;
|
|
const l2=(a2.textContent||'').trim().replace(/\s+/g,' ');
|
|
const h2=a2.getAttribute('href')||'';
|
|
rows.push({depth:2, label:l2, href:h2});
|
|
const u3=childUl(li2,'depth3'); if(!u3) return;
|
|
[...u3.children].filter(li=>li.tagName==='LI').forEach(li3=>{
|
|
const a3=directA(li3); if(!a3) return;
|
|
const l3=(a3.textContent||'').trim().replace(/\s+/g,' ');
|
|
const h3=a3.getAttribute('href')||'';
|
|
rows.push({depth:3, label:l3, href:h3});
|
|
const u4=childUl(li3,'depth4'); if(!u4) return;
|
|
[...u4.children].filter(li=>li.tagName==='LI').forEach(li4=>{
|
|
const a4=directA(li4); if(!a4) return;
|
|
const l4=(a4.textContent||'').trim().replace(/\s+/g,' ');
|
|
const h4=a4.getAttribute('href')||'';
|
|
rows.push({depth:4, label:l4, href:h4});
|
|
});
|
|
});
|
|
});
|
|
});
|
|
return {rows};
|
|
}
|
|
"""
|
|
BASE='https://www.drugsafe.or.kr'
|
|
with sync_playwright() as p:
|
|
br=p.chromium.launch(headless=True)
|
|
ctx=br.new_context(ignore_https_errors=True, viewport={'width':1920,'height':1080})
|
|
pg=ctx.new_page()
|
|
pg.goto(BASE+'/', wait_until='domcontentloaded', timeout=40000)
|
|
pg.wait_for_timeout(2500)
|
|
res=pg.evaluate(JS)
|
|
br.close()
|
|
if 'err' in res: print('ERR', res['err']); sys.exit(1)
|
|
# absolutize hrefs, blank javascript:/#
|
|
out=[]
|
|
for r in res['rows']:
|
|
h=r['href'].strip()
|
|
if h.startswith('javascript:') or h=='#' or h=='': h=''
|
|
elif h.startswith('/'): h=BASE+h
|
|
elif h.startswith('http'): pass
|
|
else: h=BASE+'/'+h
|
|
out.append({'depth':r['depth'],'label':r['label'],'href':h})
|
|
data={'num':3,'name':'한국의약품안전관리원','base':BASE,'sel':'.allmenu(depth1/2/3/4)','gnb':out}
|
|
json.dump(data, open('_gnb/3.json','w',encoding='utf-8'), ensure_ascii=False, indent=1)
|
|
import collections
|
|
dc=collections.Counter(r['depth'] for r in out)
|
|
print('rows',len(out),'depthdist',dict(dc))
|
|
for r in out:
|
|
print(' '*(r['depth']-1)+f"d{r['depth']} | {r['label'][:42]} | {r['href'][:55]}")
|