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>
33 lines
1.6 KiB
Python
33 lines
1.6 KiB
Python
# -*- coding: utf-8 -*-
|
|
"""22행+ 게시판 상세 렌더해 '공공누리 제N유형' 텍스트 라이선스 검출 → O/P/Q 판정(읽기, 결과 json)."""
|
|
import openpyxl, re, json
|
|
from playwright.sync_api import sync_playwright
|
|
KOGL=re.compile(r'공공누리\s*제?\s*([1-4])\s*유형')
|
|
ws=openpyxl.load_workbook('한국조세재정연구원.xlsx').active
|
|
targets=[]
|
|
for r in range(22,ws.max_row+1):
|
|
if all(ws.cell(r,c).value in (None,'') for c in range(2,16)): continue
|
|
if ws.cell(r,12).value=='게시판':
|
|
lab=ws.cell(r,6).value or ws.cell(r,5).value or ws.cell(r,4).value
|
|
targets.append((r,lab,str(ws.cell(r,11).value)))
|
|
print('게시판 대상:',len(targets))
|
|
out={}
|
|
with sync_playwright() as p:
|
|
b=p.chromium.launch();pg=b.new_page(ignore_https_errors=True)
|
|
for r,lab,url in targets:
|
|
types=set()
|
|
try:
|
|
pg.goto(url,timeout=30000,wait_until='networkidle');pg.wait_for_timeout(700)
|
|
a=pg.query_selector('a[onclick*=fn_]') or pg.query_selector('table tbody tr a') or pg.query_selector('.lst a')
|
|
if a:
|
|
a.click();pg.wait_for_timeout(1500)
|
|
txt=pg.inner_text('body')
|
|
for m in KOGL.finditer(txt): types.add(int(m.group(1)))
|
|
res='없음' if not types else ','.join(f'{t}유형' for t in sorted(types))
|
|
except Exception as e:
|
|
res='ERR:'+str(e)[:25]
|
|
out[r]={'label':lab,'O':res}
|
|
print(f'r{r} {lab[:22]:22} -> {res}')
|
|
b.close()
|
|
json.dump(out,open('_kogl_result.json','w',encoding='utf-8'),ensure_ascii=False,indent=1)
|