DB_JOB/작업파일/완료/공공기관/_gnb_relabel.py

# -*- coding: utf-8 -*-
import openpyxl, glob, os, re, sys, json, shutil
from copy import copy
from openpyxl.styles import PatternFill
BASE=os.path.dirname(os.path.abspath(__file__))
NO=int(sys.argv[1]); APPLY='--apply' in sys.argv
BLUE=PatternFill(fill_type='solid', fgColor='FFBDD7EE')
MAXC=27
trees=json.load(open(os.path.join(BASE,'_gnb_trees.json'),encoding='utf-8'))
g=trees[str(NO)]; base=g['base']
def findfile(no):
    for d in glob.glob(os.path.join(BASE,'*')):
        b=os.path.basename(d)
        if os.path.isdir(d) and b.startswith(str(no)+'.'):
            xs=[p for p in glob.glob(os.path.join(d,'*.xlsx')) if not os.path.basename(p).startswith(('_','~')) and 'conflict' not in os.path.basename(p)]
            if xs: return xs[0]
fn=findfile(NO)
def norm(u):
    if not u: return None
    u=str(u).split('#')[0]
    pg=re.search(r'pgmid=([A-Za-z0-9]+)', u, re.I)   # 심평원 등: pgmid가 경로보다 식별자
    if pg: return 'pgmid:'+pg.group(1).lower()
    m=re.search(r'https?://[^/]+(/[^?]*)', u)
    path=(m.group(1) if m else u).rstrip('/').lower()
    qs=u.split('?',1)[1] if '?' in u else ''
    params=[]
    for kv in qs.split('&'):
        if not kv or '=' not in kv: continue
        k=kv.split('=',1)[0].lower()
        if k.startswith('wt.') or k.startswith('utm') or k in ('returnurl','redirect'): continue
        params.append(kv.lower())
    return path+('?'+'&'.join(sorted(params)) if params else '')
# build GNB ordered leaf list: walk tree, track D/E/F path; a node is a 'row' if it has href(leaf-ish)
tree=g['tree']
rows_gnb=[]  # (D,E,F,G,normpath)
path=[None]*8  # idx by lvl 1..7
for x in tree:
    lvl=x['lvl']; t=x['t']; href=x['href']
    if lvl>=len(path): continue
    path[lvl]=t
    for k in range(lvl+1,len(path)): path[k]=None
    if href and href not in ('#','#a') and 'javascript' not in href:
        full=href if href.startswith('http') else base+href
        if base.split('//')[1].split('/')[0] in full:   # internal only
            rows_gnb.append((path[1],path[2],path[3],path[4],norm(full)))
gmap={}
for D,E,F,G,np in rows_gnb:
    if np and np not in gmap: gmap[np]=(D,E,F,G)
print('GNB 내부리프 %d개 (유니크 %d)'%(len(rows_gnb),len(gmap)))

wb=openpyxl.load_workbook(fn); ws=wb.active
def eff(r,c):
    v=ws.cell(r,c).value
    if v is not None: return v
    for mr in ws.merged_cells.ranges:
        if mr.min_col==c and mr.min_row<=r<=mr.max_row: return ws.cell(mr.min_row,c).value
    return None
data=[r for r in range(3,ws.max_row+1) if any(ws.cell(r,c).value not in (None,'') for c in (2,4,5,6,7,11))]
matched=0
for r in data:
    if norm(ws.cell(r,11).value) in gmap: matched+=1
print('시트 %d행 중 GNB매칭 %d행 (%.0f%%)'%(len(data),matched,100*matched/len(data)))
gnb_order=list(dict.fromkeys(D for D,_,_,_,_ in rows_gnb))
print('--- GNB 대분류 순서 ---', gnb_order)
print('--- 현재 시트 대분류 ---', list(dict.fromkeys(eff(r,4) for r in data if eff(r,4)))[:15])

if not APPLY:
    print('\nDRY. --apply 로 in-place relabel 실행.'); sys.exit(0)

mtime_before=os.path.getmtime(fn)
# capture records (keep sheet order), relabel matched D/E/F
stylesnap={(r,c):copy(ws.cell(r,c)._style) for r in data for c in range(1,MAXC+1)}
recs=[]
for r in data:
    rec={'src':r,'mark':set()}
    np=norm(ws.cell(r,11).value)
    if np in gmap:
        D,E,F,G=gmap[np]
        rec['D'],rec['E'],rec['F']=D,E,F or None
        rec[7]=G or None  # GNB leaf G(4단)
        if (eff(r,4),eff(r,5),ws.cell(r,6).value,ws.cell(r,7).value)!=(D,E,F,G): rec['mark'].update([4,5,6,7])
    else:
        rec['D']=eff(r,4); rec['E']=eff(r,5); rec['F']=ws.cell(r,6).value; rec[7]=ws.cell(r,7).value
    for c in range(8,MAXC+1): rec[c]=ws.cell(r,c).value
    rec['url']=ws.cell(r,11).value
    hl=ws.cell(r,11).hyperlink; rec['hlink']=hl.target if hl else rec['url']
    recs.append(rec)
# reorder: by GNB 대분류 order; within keep sheet order; unmatched D keep at end in sheet order
order_idx={d:i for i,d in enumerate(gnb_order)}
for i,rc in enumerate(recs): rc['oi']=i
recs.sort(key=lambda rc:(order_idx.get(rc['D'],999), rc['oi']))
# write
for mr in list(ws.merged_cells.ranges):
    if mr.min_row>=3: ws.unmerge_cells(str(mr))
n=len(recs)
for i,rc in enumerate(recs):
    r=3+i; src=rc['src']
    for c in range(1,MAXC+1): ws.cell(r,c)._style=copy(stylesnap[(src,c)])
    ws.cell(r,2).value=i+1; ws.cell(r,3).value=ws.cell(src,3).value
    ws.cell(r,4).value=rc['D']; ws.cell(r,5).value=rc['E']; ws.cell(r,6).value=rc['F']
    for c in range(7,MAXC+1): ws.cell(r,c).value=rc.get(c)
    kc=ws.cell(r,11); kc.value=rc['url']; kc.hyperlink=rc['hlink'] if rc['url'] else None
    for c in rc['mark']: ws.cell(r,c).fill=BLUE
    ws.row_dimensions[r].height=15
for r in range(3+n, ws.max_row+1):
    for c in range(1,MAXC+1):
        cell=ws.cell(r,c); cell.value=None; cell.hyperlink=None; cell.fill=PatternFill(fill_type=None)
def remerge(col,scope):
    r=3
    while r<3+n:
        v=ws.cell(r,col).value
        if v is None or v=='': r+=1; continue
        r2=r
        while r2+1<3+n and ws.cell(r2+1,col).value==v and all(ws.cell(r2+1,s).value==ws.cell(r,s).value for s in scope): r2+=1
        if r2>r:
            for rr in range(r+1,r2+1): ws.cell(rr,col).value=None
            ws.merge_cells(start_row=r,start_column=col,end_row=r2,end_column=col)
        r=r2+1
remerge(7,[4,5,6]); remerge(6,[4,5]); remerge(5,[4]); remerge(4,[])
if os.path.getmtime(fn)!=mtime_before:
    print('ABORT: file changed'); sys.exit(1)
bdir=os.path.join(os.path.dirname(fn),'_backup'); os.makedirs(bdir,exist_ok=True)
shutil.copy(fn, os.path.join(bdir, os.path.basename(fn).replace('.xlsx','_backup_GNB재라벨전.xlsx')))
wb.save(fn)
# verify
bs=[ws.cell(r,2).value for r in range(3,3+n)]
print('saved %d행 B연속%s'%(n,bs==list(range(1,n+1))))