26 lines
1.0 KiB
Python
26 lines
1.0 KiB
Python
import requests, io
|
|
from bs4 import BeautifulSoup
|
|
hd={"User-Agent":"Mozilla/5.0"}
|
|
urls=["https://www.kofpi.or.kr/public/publicInfo_03.do",
|
|
"https://www.kofpi.or.kr/public/publicInfo_03_02.do",
|
|
"https://www.kofpi.or.kr/public/publicInfo_03_03.do"]
|
|
out=io.open("_tabs_out.txt","w",encoding="utf-8")
|
|
for u in urls:
|
|
try:
|
|
r=requests.get(u,headers=hd,timeout=20)
|
|
r.encoding=r.apparent_encoding
|
|
s=BeautifulSoup(r.content,"html.parser")
|
|
out.write(f"\n===== {u} status={r.status_code} final={r.url}\n")
|
|
for tab in s.select("div.tab01, div[class*=tab01], ul.tab, div[class*=tab]"):
|
|
cls=tab.get("class")
|
|
links=tab.select("li a")
|
|
if links:
|
|
out.write(f" TABDIV class={cls}\n")
|
|
for a in links:
|
|
out.write(f" li.{a.find_parent('li').get('class')} | {a.get_text(strip=True)} -> {a.get('href')}\n")
|
|
break
|
|
except Exception as e:
|
|
out.write(f"\n===== {u} ERROR {e}\n")
|
|
out.close()
|
|
print("done")
|