def test_count(self):
from djangocg.test.html import parse_html
# equal html contains each other one time
dom1 = parse_html('<p>foo')
dom2 = parse_html('<p>foo</p>')
self.assertEqual(dom1.count(dom2), 1)
self.assertEqual(dom2.count(dom1), 1)
dom2 = parse_html('<p>foo</p><p>bar</p>')
self.assertEqual(dom2.count(dom1), 1)
dom2 = parse_html('<p>foo foo</p><p>foo</p>')
self.assertEqual(dom2.count('foo'), 3)
dom2 = parse_html('<p class="bar">foo</p>')
self.assertEqual(dom2.count('bar'), 0)
self.assertEqual(dom2.count('class'), 0)
self.assertEqual(dom2.count('p'), 0)
self.assertEqual(dom2.count('o'), 2)
dom2 = parse_html('<p>foo</p><p>foo</p>')
self.assertEqual(dom2.count(dom1), 2)
dom2 = parse_html('<div><p>foo<input type=""></p><p>foo</p></div>')
self.assertEqual(dom2.count(dom1), 1)
dom2 = parse_html('<div><div><p>foo</p></div></div>')
self.assertEqual(dom2.count(dom1), 1)
dom2 = parse_html('<p>foo<p>foo</p></p>')
self.assertEqual(dom2.count(dom1), 1)
dom2 = parse_html('<p>foo<p>bar</p></p>')
self.assertEqual(dom2.count(dom1), 0)
def test_parsing_errors(self):
from djangocg.test.html import HTMLParseError, parse_html
with self.assertRaises(AssertionError):
self.assertHTMLEqual('<p>', '')
with self.assertRaises(AssertionError):
self.assertHTMLEqual('', '<p>')
with self.assertRaises(HTMLParseError):
parse_html('</p>')
def test_parse_html_in_script(self):
from djangocg.test.html import parse_html
parse_html('<script>var a = "<p" + ">";</script>');
parse_html('''
<script>
var js_sha_link='<p>***</p>';
</script>
''')
# script content will be parsed to text
dom = parse_html('''
<script><p>foo</p> '</scr'+'ipt>' <span>bar</span></script>
''')
self.assertEqual(len(dom.children), 1)
self.assertEqual(dom.children[0], "<p>foo</p> '</scr'+'ipt>' <span>bar</span>")
def assert_and_parse_html(self, html, user_msg, msg):
try:
dom = parse_html(html)
except HTMLParseError as e:
standardMsg = '%s\n%s' % (msg, e.msg)
self.fail(self._formatMessage(user_msg, standardMsg))
return dom
def test_self_closing_tags(self):
from djangocg.test.html import parse_html
self_closing_tags = ('br' , 'hr', 'input', 'img', 'meta', 'spacer',
'link', 'frame', 'base', 'col')
for tag in self_closing_tags:
dom = parse_html('<p>Hello <%s> world</p>' % tag)
self.assertEqual(len(dom.children), 3)
self.assertEqual(dom[0], 'Hello')
self.assertEqual(dom[1].name, tag)
self.assertEqual(dom[2], 'world')
dom = parse_html('<p>Hello <%s /> world</p>' % tag)
self.assertEqual(len(dom.children), 3)
self.assertEqual(dom[0], 'Hello')
self.assertEqual(dom[1].name, tag)
self.assertEqual(dom[2], 'world')
def test_html_parser(self):
from djangocg.test.html import parse_html
element = parse_html('<div><p>Hello</p></div>')
self.assertEqual(len(element.children), 1)
self.assertEqual(element.children[0].name, 'p')
self.assertEqual(element.children[0].children[0], 'Hello')
parse_html('<p>')
parse_html('<p attr>')
dom = parse_html('<p>foo')
self.assertEqual(len(dom.children), 1)
self.assertEqual(dom.name, 'p')
self.assertEqual(dom[0], 'foo')
def test_html_contain(self):
from djangocg.test.html import parse_html
# equal html contains each other
dom1 = parse_html('<p>foo')
dom2 = parse_html('<p>foo</p>')
self.assertTrue(dom1 in dom2)
self.assertTrue(dom2 in dom1)
dom2 = parse_html('<div><p>foo</p></div>')
self.assertTrue(dom1 in dom2)
self.assertTrue(dom2 not in dom1)
self.assertFalse('<p>foo</p>' in dom2)
self.assertTrue('foo' in dom2)
# when a root element is used ...
dom1 = parse_html('<p>foo</p><p>bar</p>')
dom2 = parse_html('<p>foo</p><p>bar</p>')
self.assertTrue(dom1 in dom2)
dom1 = parse_html('<p>foo</p>')
self.assertTrue(dom1 in dom2)
dom1 = parse_html('<p>bar</p>')
self.assertTrue(dom1 in dom2)
