def test_clean_with_article(self):
s = '<html><head></head><body><article>Hello! I am a test</article></body></html>'
s1 = '<html><head></head><body><div>dsfasfadfasdfasdf</div><article>Hello! I am a test</article></body></html>'
s2 = '<html><head></head><body><article><video></video>Hello! I am a test</article></body></html>'
self.assertEqual(condense(clean(s)), condense(s))
self.assertEqual(condense(clean(s1)), condense(s))
self.assertEqual(condense(clean(s2)), condense(s))
def test_clean_with_article(self):
s = '<html><head></head><body><article>Hello! I am a test</article></body></html>'
s1 = '<html><head></head><body><div>dsfasfadfasdfasdf</div><article>Hello! I am a test</article></body></html>'
s2 = '<html><head></head><body><article><video></video>Hello! I am a test</article></body></html>'
self.assertEqual(condense(clean(s)), condense(s))
self.assertEqual(condense(clean(s1)), condense(s))
self.assertEqual(condense(clean(s2)), condense(s))
def test_clean_empty_img(self):
s = '''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<img src=""></img>
</body>
</html>
'''
s1 = '''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<img src=""></img>
<img></img>
<img/>
</body>
</html>
'''
self.assertEqual(condense(clean(s1)), condense(clean(s)))
def test_html_to_xhtml(self):
s = u'<!DOCTYPE html><html xmlns="http://www.w3.org/1999/xhtml"><head></head><body><div id="Test">Hello</div><br /><br /></body></html>'
s1 = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<DIV ID="Test">Hello</div>
<br>
<br>
</body>
</html>
'''
self.assertEqual(condense(html_to_xhtml(clean(s1))), s)
def test_clean_empty_img(self):
s = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<img src=""></img>
</body>
</html>
'''
s1 = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<img src=""></img>
<img></img>
<img/>
</body>
</html>
'''
self.assertEqual(condense(clean(s1)), condense(clean(s)))
def test_html_to_xhtml(self):
s = u'<!DOCTYPE html><html xmlns="http://www.w3.org/1999/xhtml"><head></head><body><div id="Test">Hello</div><br /><br /></body></html>'
s1 = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<DIV ID="Test">Hello</div>
<br>
<br>
</body>
</html>
'''
self.assertEqual(condense(html_to_xhtml(clean(s1))), s)
def test_clean_tags_full_html(self):
s = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<div>Hello </div>
</body>
</html>
'''
s1 = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<div>Hello </div>
<script>Uh oh...it's an evil script!</script>
</body>
</html>
'''
s2 = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<div>Hello </div>
</body>
<script>Uh oh...it's an evil script again!</script>
</html>
'''
s3 = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<div>Hello </div>
</body>
<video>Play me!</video>
</html>
'''
s4 = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<video>
<div>Hello </div>
</video>
</body>
<video>Play me!</video>
</html>
'''
s5 = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<video>
<div>Hello </div>
</video>
</body>
<video>Play me!</video>
</html>
'''
self.assertEqual(condense(clean(s)), condense(s))
self.assertEqual(condense(clean(s1)), condense(s))
self.assertEqual(condense(clean(s2)), condense(s))
self.assertEqual(condense(clean(s3)), condense(s))
self.assertEqual(condense(clean(s4)), condense(s))
self.assertEqual(condense(clean(s5)), condense(s))
def test_clean_tags_full_html(self):
s = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<div>Hello </div>
</body>
</html>
'''
s1 = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<div>Hello </div>
<script>Uh oh...it's an evil script!</script>
</body>
</html>
'''
s2 = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<div>Hello </div>
</body>
<script>Uh oh...it's an evil script again!</script>
</html>
'''
s3 = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<div>Hello </div>
</body>
<video>Play me!</video>
</html>
'''
s4 = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<video>
<div>Hello </div>
</video>
</body>
<video>Play me!</video>
</html>
'''
s5 = u'''
<!DOCTYPE html>
<html>
<head>
</head>
<body>
<video>
<div>Hello </div>
</video>
</body>
<video>Play me!</video>
</html>
'''
self.assertEqual(condense(clean(s)), condense(s))
self.assertEqual(condense(clean(s1)), condense(s))
self.assertEqual(condense(clean(s2)), condense(s))
self.assertEqual(condense(clean(s3)), condense(s))
self.assertEqual(condense(clean(s4)), condense(s))
self.assertEqual(condense(clean(s5)), condense(s))