getharim.py

getharim.py

# -*- coding: utf-8 -*-
import urllib2 as u
import re

def GetDataFromURL(url):
	req = u.Request(url)
	stream = u.urlopen(req).read()
	return stream

def GetHarim():
	
	harim_regex = r"<a\shref=\"bbsView\.php\?id=([0-9]+)&page=1&code=bbs_menu01\">"
	base_url = "http://www.pvv.co.kr/bbs/index.php?code=bbs_menu01"

	try:
		stream = GetDataFromURL(base_url)
		bbs_id = re.search(harim_regex, stream).group(1)

	except:
		print "[!] error in GetHarim() - base data parse"
		return None

	try:
		bbs_url = "http://www.pvv.co.kr/bbs/bbsView_body.php?id={}&code=bbs_menu01".format(bbs_id)
		bbs_regex = r"(\d{2}/\d{2})(.+)\s?<br\s/>"

		stream = GetDataFromURL(bbs_url)
		bbs_data = re.findall(bbs_regex, stream)

		return bbs_data

	except Exception as e:
		print e
		print "[!] error in GetHarim() - bbs data parse"
		return None

하림 진자 싫다...