Get latest HANRel TV shows from bitsnoop.com.

bitsnoop_kr_show.py

# -*- coding: utf-8 -*-
# <nbformat>3.0</nbformat>

# <codecell>

import requests, bs4, re

# <codecell>

re_date = re.compile(r'(?P<date>13[0-1]{1}[0-9]{1}[0-9]{1}[0-9]{1})')

def getPage(page=1):
    headers = { "user-agent":"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_7_3) AppleWebKit/535.19 (KHTML, like Gecko) Chrome/18.0.1025.151 Safari/535.19" }
    session = requests.Session()
    r = session.get("http://bitsnoop.com/search/all/HANRel/c/d/%d/" % page, headers=headers)
    b = bs4.BeautifulSoup(r.text)
    return b

# <codecell>

def torrents():
    tor= []
    for page in xrange(1,3):
        b = getPage(1)
        for i in b.find_all("a"):
            if i["href"].find(".html")>=0 and len(i["href"])>30:
                tor.append("http://bitsnoop.com"+i["href"])
    return tor

def getDatedTorrents(t):
    DATA = {}
    for i in t:
        d = re_date.search(i)
        if d!=None:
            dd = d.group("date")
            DATA[dd] = list( set(DATA.get(dd, []) + [ i ]) )
    return DATA

# <codecell>

t = torrents()
DATA = getDatedTorrents(t)
dates = sorted(DATA.keys(),reverse=True)
for i in dates:
    print i
    for j in DATA[i]:
        print "\t", j