# -*- coding: utf-8 -*- """순창 LNB 4차하위 전개(자동차등록안내식): M>1 페이지(=LNB 부모메뉴, URL은 첫자식)를 좌측메뉴 자식들로 펼침. 부모행 M→1·lc+1=첫자식, 나머지 자식 행 삽입, lc병합, process_row. 가드: 부모라벨==행라벨 & 자식수>=2. 사용: python -X utf8 _순창_LNB하위.py [--write] """ import sys, os, re, shutil, importlib.util from urllib.parse import urljoin import openpyxl from openpyxl.worksheet.hyperlink import Hyperlink HERE=os.path.dirname(os.path.abspath(__file__)) s=importlib.util.spec_from_file_location('m',os.path.join(HERE,'_jeonbuk_phase234_all.py')) M=importlib.util.module_from_spec(s); s.loader.exec_module(M) XLSX=r'D:\01.프로젝트\DB수집\작업파일\광역_사이트맵\전북특별자치도\7.순창군\전북특별자치도_순창군.xlsx' BASE='https://www.sunchang.go.kr' SESS=M.make_session() def clean(t): return re.sub(r'하위메뉴.*','',t or '').strip() def muid(u): mm=re.search(r'menuUid=([0-9a-f]+)',u or ''); return mm.group(1) if mm else None def leafcol(ws,r): for c in range(10,3,-1): if ws.cell(r,c).value not in (None,''): return c return None def extract(url): soup=M.fetch(SESS,url) if soup is None: return None,None lnb=soup.select_one('.lnb') if not lnb: return None,None mid=muid(url) target=next((a for a in lnb.find_all('a',href=True) if mid and mid in a['href']),None) if not target: return None,None li=target.find_parent('li'); pul=li.find_parent('ul') if li else None if not pul: return None,None pli=pul.find_parent('li'); plabel=clean(pli.find('a').get_text()) if (pli and pli.find('a')) else None ch=[(clean(c.find('a').get_text()), urljoin(BASE,c.find('a')['href'])) for c in pul.find_all('li',recursive=False) if c.find('a',href=True)] return plabel,ch def remap_insert(ws, ins, n): ranges=[(x.min_col,x.min_row,x.max_col,x.max_row) for x in list(ws.merged_cells.ranges)] for x in list(ws.merged_cells.ranges): ws.unmerge_cells(str(x)) ws.insert_rows(ins,n) for mc,mr,xc,xr in ranges: if mr>=ins: mr+=n; xr+=n elif xr>=ins: xr+=n ws.merge_cells(start_row=mr,start_column=mc,end_row=xr,end_column=xc) def main(): write='--write' in sys.argv wb=openpyxl.load_workbook(XLSX); ws=wb.active C=ws.cell(3,3).value cands=[] for r in range(3,ws.max_row+1): if ws.cell(r,12).value!='페이지': continue Mv=ws.cell(r,13).value if not(isinstance(Mv,int) and Mv>1): continue lc=leafcol(ws,r) if not lc or lc>=10: continue if ws.cell(r,lc+1).value not in (None,''): continue # 이미 전개됨 K=ws.cell(r,11).value if not(isinstance(K,str) and 'menuUid=' in K): continue cands.append((r,lc,Mv,K,ws.cell(r,lc).value)) plan=[] for r,lc,Mv,K,lab in cands: plabel,ch=extract(K) if plabel and clean(lab)==plabel and ch and len(ch)>=2: plan.append((r,lc,Mv,ch,lab)) else: print(' SKIP r%d %s (부모=%s 자식=%s)'%(r,lab,plabel,len(ch) if ch else None)) print('전개대상 %d개 (가드통과)'%len(plan)) if not write: for r,lc,Mv,ch,lab in plan: print(' r%d %s M%s→%d자식'%(r,lab,Mv,len(ch))) return shutil.copy(XLSX, XLSX.replace('.xlsx','_backup_LNB하위전.xlsx')) tot=0 for r,lc,Mv,ch,lab in sorted(plan,key=lambda t:-t[0]): # bottom-up # 부모행 = 첫자식 ws.cell(r,lc+1).value=ch[0][0] ws.cell(r,11).value=ch[0][1] ws.cell(r,13).value=1 rest=ch[1:]; n=len(rest) if n: remap_insert(ws,r+1,n) for i,(clab,curl) in enumerate(rest): rr=r+1+i out=M.process_row(SESS,curl,M.BODY_SEL) ws.cell(rr,3).value=C ws.cell(rr,lc+1).value=clab ws.cell(rr,11).value=curl ws.cell(rr,12).value=out['L'] or '페이지' ws.cell(rr,13).value=out['M'] if out['M']!='' else 1 ws.cell(rr,14).value=out['N'] ws.cell(rr,15).value=out['O']; ws.cell(rr,16).value=out['P']; ws.cell(rr,17).value=out['Q'] ws.row_dimensions[rr].height=15 ws.merge_cells(start_row=r,start_column=lc,end_row=r+n,end_column=lc) tot+=n print(' r%d %s +%d'%(r,lab,n)) for r in range(3,ws.max_row+1): v=ws.cell(r,11).value; cell=ws.cell(r,11) if isinstance(v,str) and v.startswith('http'): cell.hyperlink=Hyperlink(ref=cell.coordinate,target=v) else: cell.hyperlink=None wb.save(XLSX) print('적용: +%d행 · 하이퍼링크재구성 · 백업'%tot) if __name__=='__main__': main()