def setUp(self):
self.parser = PageProcessor()
self.parser._theme = ['lorem', 'ipsum']
self.parser._my_data.base_url = "http://www.test.fr"
class PageProcessor_test(unittest.TestCase):
def setUp(self):
self.parser = PageProcessor()
self.parser._theme = ['lorem', 'ipsum']
self.parser._my_data.base_url = "http://www.test.fr"
def test_keyword(self):
test_page = '<html><head><meta name="keywords" content="test, bla,\
python" /></head></html>'
keywords = self.parser._parse('http://localhost',
test_page)['keywords']
self.assertEqual(keywords, ['test', 'bla', 'python'])
def test_none_keyword(self):
test_page = '<html><head><meta name="keywords"\
content="None" /></head></html>'
keywords = self.parser._parse('http://localhost',
test_page)['keywords']
self.assertEqual(keywords, [])
def test_link(self):
test_page = u'<html><head><title>Test</title></head><body>\
<a href="/local" /></body></html>'
links = self.parser._parse('http://localhost', test_page)['links']
self.assertEqual(links, ['/local'])
def test_calculate_score(self):
test_page = '<html><body><p>Lorem Ipsum</p></body></html>'
self.parser._parse('http://localhost', test_page)
self.assertEqual(self.parser.calculate_score(test_page),1)
def test_calculate_score2(self):
test_page = '<html><body><p>Georges Brassens</p></body></html>'
self.parser._parse('http://localhost', test_page)
self.assertEqual(self.parser.calculate_score(test_page),0)
def test_data(self):
test_page = '<html><body><p>Georges Brassens</p></body></html>'
text_content = \
self.parser._parse('http://localhost', test_page)['text_content']
self.assertEqual(text_content, 'Georges Brassens')
def test_anchor(self):
test_page = u'<html><head><title>Test</title></head><body>\
<a href="/local">Bouh</a></body></html>'
link, anchor = self.parser._parse('http://localhost',
test_page)['anchors'][0]
self.assertEqual(anchor, 'bouh')
def test_script(self):
test_page = u'<html><head><title>Test</title></head><body>\
<script>LFDKNG</script><a href="/local">Bouh</a></body></html>'
link, anchor = self.parser._parse('http://localhost',
test_page)['anchors'][0]
self.assertEqual(anchor, 'bouh')
