python-webrobot Code

#!/usr/bin/python
# -*- coding: iso-8859-15 -*-

def parse(URL, debug=False):

	# this parser doesn't work
	return {'directlink': '', 'filename': ''}

	from time import sleep
	from htmlparser import between, Eval
	from httpclient import HttpClient

	r = HttpClient(debug=debug)
	r.GET(URL)
	filename = between(r.Page, 'name="fname" value="', '"')
	
	# manually set JavaScript-set cookies
#	r.Cookies.append = {
#				'__utma':	# Google Analytics cookies
#				'__utmb':
#				'__utmc':
#				'__utmv':
#				'__utmz':
#				'ad_referer': '',
#				'aff': '1',
#				'lang': 'english'
#			}

	# "Continue" form
	form = r.Page.findForm(number=2)
		
	del form.input['method_premium']
	#form.input['referer'].value = URL
	form.input['method_free'].value = 'Free Download'
	
	r.Cookies['ad_referer'] = ''
	r.Cookies['__utma']='5195046.1779219333.1372885161.1372885161.1373715176.2'
	r.Cookies['__utmb']='5195046.4.10.1373715176'
	r.Cookies['__utmc']='5195046'
	r.Cookies['__utmz']='5195046.1372885161.1.1.utmcsr=watchseries.lt|utmccn=(referral)|utmcmd=referral|utmcct=/open/cale/3748860.html'
	
	# wait
	for i in range(5):
		sleep(1)
		print str(5-i)

	host = URL
	if 'http://' in host:
		host = host[7:]
	host = host[:host.find('/')]
	#print host
	r.POST(URL, form=form, AdditionalHeaders={'Origin':host, 'Referer':URL, 'Accept':'text/html,application/xhtml+xml,application/xml', 'Accept-Language':'en-US,en'})

#/cgi-bin/index_dlf.cgi?op=track_view&file_code=atqunkn8k2xy&rnd=0.3730618921108544

	# check
	if 'value="Continue"' in str(r.Page):
		print 'Warning: There is a "Continue" button on this page.'
	if not 'player_code' in str(r.Page):
		print 'Warning: There is no player on this page.'
	if not 'flashvars' in str(r.Page):
		print 'Warning: No flashvars on this page.'

	# find URL
	url = between(r.Page, 'file: "', '"')
	if url == '':
		url = between(r.Page, 'file:"', '"')

	if url == '':
		print 'Failed to extract URL'
	
	return {'directlink': url, 'filename': filename}

if __name__ == "__main__":
	from sys import argv
	print str( parse(argv[1]) )