def test_formatter_hostFromUrl_method_returns_correct_result(self):
f = Formatter("http://www.elpais.es")
res = f.hostFromUrl()
f2 = Formatter("http://madrid.elpais.es")
res2 = f.hostFromUrl()
self.assertEqual("elpais", res)
self.assertEqual("elpais", res2)
def fs_formatting(self, hd_size=256, cluster_size=4096):
"""Format FS"""
if 20 <= hd_size <= 1024 and (not cluster_size & (cluster_size - 1)
and 512 <= cluster_size <= 32768):
formatter = Formatter(hd_size, cluster_size)
formatter.formatting()
self.__init__()
else:
raise FSExeption("Incorrect HD size or cluster's size!")
def test_formatter_isUrl_method_returns_correct_result(self):
f = Formatter("http://www.elpais.es")
res = f.isUrl()
f2 = Formatter("http://madrid.elpais.es")
res2 = f2.isUrl()
f3 = Formatter("/noticias.html")
res3 = f3.isUrl()
self.assertEqual(True, res)
self.assertEqual(True, res2)
self.assertEqual(False, res3)
def urlsLevel1Host(self, **kwargs):
if ("url" in kwargs):
url = kwargs["url"]
else:
url = self.url
f = Formatter(url)
page = requests.get(url)
tree = html.fromstring(page.content)
hrefs = tree.xpath('//a//@href')
regex = '^http[s]?://(www\.)?([a-z]*\.)?'+f.hostFromUrl()
for h in hrefs:
if h not in self.urls:
match = re.search(regex, h)
if match:
self.urls.append(h)
def test_formatter_isPath_method_returns_correct_result(self):
f = Formatter("/noticias.html")
res = f.isPath()
self.assertEqual(True, res)
def test_formatter_drop_accents_whitespaces_method_returns_correct_result(self):
f = Formatter("Balón lala")
res = f.drop_accents_whitespaces()
self.assertEqual("Balon-lala", res)
self.assertEqual("balon-lala", res.lower())
def test_formatter_drop_whitespaces_method_returns_correct_result(self):
f = Formatter("En un lugar de la Mancha")
res = f.drop_whitespaces()
self.assertEqual("En-un-lugar-de-la-Mancha", res)
def test_formatter_drop_accents3_method_returns_correct_result(self):
f = Formatter("Balón")
res = f.drop_accents()
self.assertEqual("Balon", res)
def test_formatter_drop_accents2_method_returns_correct_result(self):
f = Formatter("Balon")
self.assertEqual("Balon", f.string)