def test_html_to_text(self):
s = '''
<html>
<head>
<title>Title</title>
<style>
p
{
font-family:"Times New Roman";
font-size:20px;
}
</style>
</head>
<body>
<script type="text/javascript">
$.document()
</script>
Body
<p>Paragraph <strong>here</strong></p>
More text
</body>
</html>
'''
t = html_to_text(s, encoding='ascii')
t = t.strip()
self.assertTrue(t.startswith('Body'))
self.assertTrue(t.endswith('text'))
def test_html_to_text(self):
s = '''
<html>
<head>
<title>Title</title>
<style>
p
{
font-family:"Times New Roman";
font-size:20px;
}
</style>
</head>
<body>
<script type="text/javascript">
$.document()
</script>
Body
<p>Paragraph <strong>here</strong></p>
More text
</body>
</html>
'''
t = html_to_text(s, encoding = 'ascii')
t = t.strip()
self.assertTrue(t.startswith('Body'))
self.assertTrue(t.endswith('text'))
def test_html_to_text_empty(self):
s = ''
t = html_to_text(s, encoding='ascii')
self.assertTrue(re.match('\s*', t))
def test_html_to_text_empty(self):
s = ''
t = html_to_text(s, encoding = 'ascii')
self.assertTrue(re.match('\s*', t))
