df16c98366
공공기관2 작업 중. _temp 몽타주(재생성가능)는 제외. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
121 lines
5.4 KiB
Python
121 lines
5.4 KiB
Python
# -*- coding: utf-8 -*-
|
|
# 충북형(eGov 메가메뉴) 범용 수집: 트리 + 본문탭 + L/M/O
|
|
# 사용: python _cb_crawl.py <host> <sitemap_url> <out_nodes.json>
|
|
import sys,io,json,re,time,urllib.request,ssl
|
|
from bs4 import BeautifulSoup
|
|
sys.stdout=io.TextIOWrapper(sys.stdout.buffer,encoding='utf-8')
|
|
ctx=ssl.create_default_context(); ctx.check_hostname=False; ctx.verify_mode=ssl.CERT_NONE
|
|
HOST=sys.argv[1]; SITEMAP=sys.argv[2]; OUT=sys.argv[3]
|
|
BASE='https://'+HOST
|
|
def absu(u):
|
|
if not u: return u
|
|
if u.startswith('http'): return u
|
|
if u.startswith('/'): return BASE+u
|
|
return BASE+'/'+u
|
|
_cache={}
|
|
def get(u):
|
|
if u in _cache: return _cache[u]
|
|
h=''
|
|
for _ in range(2):
|
|
try:
|
|
req=urllib.request.Request(u,headers={'User-Agent':'Mozilla/5.0'})
|
|
h=urllib.request.urlopen(req,timeout=25,context=ctx).read().decode('utf-8','ignore'); break
|
|
except Exception: time.sleep(1)
|
|
_cache[u]=h; return h
|
|
def kind(u):
|
|
if not u: return 'none'
|
|
if u.startswith('http') and HOST not in u: return '사이트'
|
|
if 'eminwon' in u: return '사이트'
|
|
if re.search(r'selectBbsNttList|selectOthbcInfoList|BbsNtt',u): return '게시판'
|
|
return '페이지'
|
|
def board_M(html):
|
|
m=re.search(r'게시물[^\d]{0,15}([\d,]+)', html)
|
|
if m:
|
|
try: return int(m.group(1).replace(',',''))
|
|
except: return 0
|
|
return 0
|
|
def has_opentype(html): # full-page(빠른 1차); 본문한정은 _cb_kogl로 후처리
|
|
mm=re.findall(r'img_opentype0?(\d)', html); return sorted(set(int(x) for x in mm)) if mm else []
|
|
def page_tabs(html, self_url):
|
|
# 본문 콘텐츠 탭만: 탭바가 '자기 페이지 key'를 포함(self-referential)해야 인정 → 전역위젯 배제
|
|
s=BeautifulSoup(html,'html.parser')
|
|
cont=s.select_one('#contents,.contents,.sub_content,.sub_contents,#content,.content_in') or s
|
|
selfk=re.search(r'key=(\d+)',self_url); selfk=selfk.group(1) if selfk else None
|
|
if not selfk: return []
|
|
for ul in cont.find_all('ul'):
|
|
cls=' '.join(ul.get('class') or [])
|
|
if not re.search(r'tab',cls,re.I): continue
|
|
anchors=ul.find_all('a')
|
|
if len(anchors)<2: continue
|
|
keys=[re.search(r'key=(\d+)',a.get('href') or '') for a in anchors]
|
|
keys=[k.group(1) for k in keys if k]
|
|
# 가드: 탭앵커가 전부 key= 형식이고 self key를 포함해야 진짜 본문탭
|
|
if selfk not in keys: continue
|
|
if len(keys)<len(anchors)-1: continue # 대부분 key링크여야(아이콘/외부 위젯 배제)
|
|
out=[]
|
|
for a in anchors:
|
|
lab=a.get_text(strip=True); href=a.get('href')
|
|
if not lab: continue
|
|
k=re.search(r'key=(\d+)',href or '')
|
|
if not k: continue
|
|
if k.group(1)==selfk: continue # 자기자신(전체/현재) 제외
|
|
out.append((lab,href))
|
|
if out: return out
|
|
return []
|
|
# 1) 트리 파싱 (depth1~4 megamenu)
|
|
h=get(SITEMAP); s=BeautifulSoup(h,'html.parser')
|
|
tree=[]
|
|
for li1 in s.select('li.depth1_item'):
|
|
a1=li1.select_one('a.depth1_text');
|
|
if not a1: continue
|
|
n1=a1.get_text(strip=True)
|
|
for li2 in li1.select('li.depth2_item'):
|
|
a2=li2.select_one('a.depth2_text');
|
|
if not a2: continue
|
|
n2=a2.get_text(strip=True)
|
|
d3=li2.select('li.depth3_item')
|
|
if not d3: tree.append((n1,n2,'','',a2.get('href'))); continue
|
|
for li3 in d3:
|
|
a3=li3.select_one('a.depth3_text');
|
|
if not a3: continue
|
|
n3=a3.get_text(strip=True)
|
|
d4=li3.select('li.depth4_item a.depth4_text')
|
|
if not d4: tree.append((n1,n2,n3,'',a3.get('href')))
|
|
else:
|
|
for a4 in d4: tree.append((n1,n2,n3,a4.get_text(strip=True),a4.get('href')))
|
|
print('트리노드',len(tree),flush=True)
|
|
# 2) 분류 + 탭전개
|
|
def classify(u):
|
|
u=absu(u); k=kind(u); M=None; O=[]; det=None
|
|
if k=='게시판':
|
|
hh=get(u); M=board_M(hh); O=has_opentype(hh)
|
|
ss=BeautifulSoup(hh,'html.parser')
|
|
for a in ss.find_all('a'):
|
|
href=a.get('href') or ''
|
|
if re.search(r'NttView|nttNo=|addBbsNttView',href): det=absu(href); break
|
|
elif k=='페이지':
|
|
O=has_opentype(get(u))
|
|
return k,M,O,det
|
|
nodes=[]
|
|
menu_urls=set(absu(t[4]) for t in tree if t[4]) # 메뉴에 이미 있는 url(탭 중복 방지)
|
|
seen_tab=set()
|
|
for i,(d1,d2,d3,d4,url) in enumerate(tree):
|
|
u=absu(url); k,M,O,det=classify(url)
|
|
if k=='none' and not u: continue # url 없는 빈 헤더 노드 제외
|
|
nodes.append({'d':[d1,d2,d3,d4,''],'url':u,'kind':k,'M':M,'O':O,'det':det})
|
|
if k=='페이지' and re.search(r'(contents|sub)\.do',u):
|
|
for lab,href in page_tabs(get(u),u):
|
|
tu=absu(href)
|
|
if tu in menu_urls or tu in seen_tab: continue # 메뉴/기존탭 중복 제외
|
|
seen_tab.add(tu)
|
|
tk,tM,tO,tdet=classify(href)
|
|
dd=[d1,d2,d3,d4,'']; depth=max((idx for idx in range(4) if dd[idx]),default=0)
|
|
child=dd[:]
|
|
if depth>=3: child[4]=lab
|
|
else: child[depth+1]=lab
|
|
nodes.append({'d':child,'url':tu,'kind':tk,'M':tM,'O':tO,'det':tdet,'tab':True})
|
|
if i%25==0: print('%d/%d'%(i+1,len(tree)),flush=True)
|
|
json.dump(nodes,open(OUT,'w',encoding='utf-8'),ensure_ascii=False)
|
|
from collections import Counter
|
|
print('총행',len(nodes),'| L',dict(Counter(n['kind'] for n in nodes)),'| 탭',sum(1 for n in nodes if n.get('tab')),'| O',sum(1 for n in nodes if n['O']))
|