df16c98366
공공기관2 작업 중. _temp 몽타주(재생성가능)는 제외. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
61 lines
2.3 KiB
Python
61 lines
2.3 KiB
Python
# -*- coding: utf-8 -*-
|
|
"""전기안전공사 LNB(.sub_left_menu) 4단 권위트리 재추출 → _gnb\1.json.
|
|
대분류(GNB d1) > 섹션(LNB d0=E) > 항목(d1=F) > 세부(d2=G)."""
|
|
import sys, io, os, json, re
|
|
sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding='utf-8')
|
|
from playwright.sync_api import sync_playwright
|
|
|
|
GNB = r"D:\01.프로젝트\DB수집\작업파일\공공기관2\_gnb\1.json"
|
|
old = json.load(open(GNB, encoding='utf-8'))
|
|
# 대분류(depth1) 목록
|
|
majors = [(r['label'], r['href']) for r in old['gnb'] if r['depth'] == 1]
|
|
|
|
LNB_JS = """
|
|
() => {
|
|
const lnb = document.querySelector('.sub_left_menu, #snb, .snb, .lnb');
|
|
if(!lnb) return {rows:[]};
|
|
const rows=[];
|
|
lnb.querySelectorAll('a').forEach(a=>{
|
|
let d=0,p=a.parentElement;
|
|
while(p&&p!==lnb){ if(p.tagName==='UL')d++; p=p.parentElement; }
|
|
let label=(a.textContent||'').trim().replace(/\\s+/g,' ');
|
|
label=label.replace(/하위메뉴(닫기|열기)$/,'').trim();
|
|
let href=a.getAttribute('href')||'';
|
|
if(href==='#'||href.startsWith('javascript')) href='';
|
|
else if(href && !href.startsWith('http')) href='https://www.kesco.or.kr'+ (href.startsWith('/')?href:'/'+href);
|
|
if(label) rows.push({d, label, href});
|
|
});
|
|
return {rows};
|
|
}
|
|
"""
|
|
|
|
def clean(label):
|
|
return label
|
|
|
|
out_rows = []
|
|
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 name, href in majors:
|
|
try:
|
|
pg.goto(href, wait_until='domcontentloaded', timeout=30000)
|
|
pg.wait_for_timeout(1800)
|
|
r = pg.evaluate(LNB_JS)
|
|
except Exception as e:
|
|
print(f"[{name}] 실패 {e}");
|
|
out_rows.append({'depth':1,'label':name,'href':href}); continue
|
|
lnb = r['rows']
|
|
# 대분류 행
|
|
out_rows.append({'depth':1, 'label':name, 'href':href})
|
|
# LNB d0->depth2, d1->depth3, d2->depth4 ...
|
|
for n in lnb:
|
|
out_rows.append({'depth': 2 + n['d'], 'label': n['label'], 'href': n['href']})
|
|
d1cnt = sum(1 for n in lnb if n['d']==0)
|
|
print(f"[{name}] LNB {len(lnb)}항목 (섹션 {d1cnt})")
|
|
br.close()
|
|
|
|
old['gnb'] = out_rows
|
|
old['base'] = 'https://www.kesco.or.kr'
|
|
json.dump(old, open(GNB, 'w', encoding='utf-8'), ensure_ascii=False, indent=1)
|
|
print('총', len(out_rows), '행 →', GNB)
|