Script para pegar a programação do dia da TV fechada (cabo) no site http://www.hagah.com.br

O script faz uso do BeautifulSoup para fazer o parsing da página. 

{{{#!python
#!/usr/bin/env python
# -*- coding: utf-8 -*-
import urllib2
from BeautifulSoup import BeautifulSoup
from datetime import datetime
x = datetime.now()
hoje = x.strftime('%d/%m/%Y')
print hoje

canais = [
["MDO","A & E Mundo"],
#["APL","Animal Planet"],
#["LOC","Animax"],
["UNO","AXN"],
#["BAN","Bandeirantes"],
#["BBC","BBC World"],
#["BIT","Bloomberg"],
#["BMG","Boomerang"],
#["CBR","Canal Brasil"],
#["FUT","Canal Futura"],
#["CAR","Cartoon Network"],
#["MAX","Cinemax"],
#["MXE","Cinemax e"],
#["CLI","Clima Tempo"],
#["CNE","CNN Espanhol"],
#["CNN","CNN International"],
#["CNT","CNT"],
#["DWL","Deutsche Welle"],
#["DIS","Discovery Channel"],
#["HEA","Discovery Home & Health"],
#["DIK","Discovery Kids"],
#["TRV","Discovery Travel & Living"],
#["DNY","Disney Channel"],
#["EET","E! Entertainment "],
#["ESB","ESPN Brasil"],
#["ESP","ESPN International"],
#["EUR","EuroChannel"],
#["FAS","Fashion TV"],
#["GLS","For Man"],
#["FOX","Fox"],
#["FLI","Fox Life"],
#["FNE","Fox News"],
["CFX","FX"],
#["GLN","Globo News"],
#["GNT","GNT"],
["HAL","Hallmark"],
#["HBO","HBO"],
#["HBE","HBO e"],
#["HFA","HBO Family"],
#["HFE","HBO Family e"],
#["HPL","HBO Plus"],
#["HPE","HBO Plus e"],
#["KID","Jetix"],
#["MAP","Max Prime"],
#["MPE","Max Prime e"],
#["MGM","MGM"],
#["MTV","MTV Brasil"],
#["MTH","MTV Hits"],
#["MSW","Multishow"],
#["SUP","National Geographic"],
#["NIC","Nickelodeon"],
["TRA","People+Arts"],
#["PLA","Playboy TV"],
#["RAI","RAI (italiano)"],
#["POA","RBS TV"],
#["REC","Record"],
#["RTV","Rede TV"],
#["VDA","Rede Vida"],
#["NAC","Sesc TV"],
#["HOT","Sexy Hot"],
#["SHO","Shoptime"],
#["SIC","SIC Internacional"],
["SET","Sony Entertainment Television"],
#["SPE","Speed Channel"],
#["SPO","SPORTV"],
#["SP2","SPORTV 2"],
#["TC2","Telecine Action"],
#["TC5","Telecine Cult"],
#["TC3","Telecine Light"],
#["TC4","Telecine Pipoca"],
#["TC1","Telecine Premium"],
#["TGC","The Golf Channel"],
#["HIS","The History Channel"],
#["TNT","TNT"],
#["TCM","Turner Classic Movies"],
#["CAM","TV Câmara"],
#["CNV","TV Canção Nova"],
#["CUL","TV Cultura"],
#["ESC","TV Escola"],
#["JUS","TV Justiça"],
#["RTB","TV Rá Tim Bum"],
#["SEN","TV Senado"],
#["TVV","TV Terra Viva"],
#["TED","TVE Brasil"],
#["TVE","TVE Internacional"],
#["TV5","TV5MONDE"],
#["USA","Universal Channel"],
#["VH1","VH1"],
["WBT","Warner"]
]

for canal in canais:
    url="http://www.hagah.com.br/programacao-tv/jsp/default.jsp?uf=1&local=1&regionId=1&action=programacao_canal&canal=%s&operadora=14&data=%s" % (canal[0],hoje)
    pagina = urllib2.urlopen(url)
    print "***** %s *****" % canal[1]
    conteudo = BeautifulSoup(pagina.read(), fromEncoding="iso-8858-1")
    tabela = conteudo.findAll(id='grade canal')
    linhas = conteudo.findAll('tr')
    for linha in linhas: 
        dado = linha.findAll('td')
        hora = dado[0]
        if dado[1].findAll('strong'):
            programa=dado[1].findAll('strong')
	    for item in programa:
	        if item:
                    print "%s |%s " % (hora.contents[0],item.contents[0])
    print "----------------------------------------------------------------------------------------------------"
}}}