Commit 3353db981b5e100e367b60fb2d4c356825161a10

Enconding problems, save things on a file too


22	22	cookie_jar = cookielib.CookieJar()
23	23	opener = urllib2.build_opener(urllib2.HTTPCookieProcessor(cookie_jar))
24	24
25		output = file('results.txt')
	25	output = file('results.txt', 'wb')
26	26
27	27	for source in sources:
28	28	print source + ':'
	29	output.write(source + '\n')
	30
29	31	content = csv.reader(open(source + '.csv'), delimiter=';')
30	32
31	33	# The data have two headers
…	…
35	35	content.next()
36	36	for row in content:
37	37	print row[0], row[1]
38		#print >>output, row[0], row[1]
	38	output.write('%s %s\n' % (row[0], row[1]))
39	39
40	40	# remove ".", "/" and "-". Do not translate any other characters.
41	41	cnpj_clean = row[0].translate(None, './-')
…	…
87	87
88	88	soup = BeautifulSoup(data)
89	89	rows = soup.findAll('tr')
	90
90	91	# 0: Empty
91	92	# 1: Donator name
92	93	# 2: Empty
…	…
114	114
115	115	for data in rows[2:]:
116	116	donation = '%s = %s (%s %s) %s' % (
117		unicode(data.contents[7].contents[0]).strip(),
118		unicode(data.contents[13].contents[2]).strip(),
119		unicode(data.contents[15].contents[0]).strip(),
120		unicode(data.contents[17].contents[0]).strip(),
121		unicode(data.contents[21].contents[0]).strip())
	117	data.contents[7].contents[0].strip(),
	118	data.contents[13].contents[2].strip(),
	119	data.contents[15].contents[0].strip(),
	120	data.contents[17].contents[0].strip(),
	121	data.contents[21].contents[0].strip())
122	122	print donation
123		#print >>output, donation
	123	output.write(donation + '\n')
124	124	print
125		#print >>output, ''
	125	output.write('\n')
	126	output.flush()
	127
126	128	print
127		#print >>output, ''
	129	output.write('\n')