def test_check_for_meta_with_content_split_bad(self):
"""
В контенте не 2 параметра
:return:
"""
result = self._prepare_mock()
result.__getitem__ = mock.Mock(return_value="content")
with mock.patch.object(re, 'search', mock.Mock()) as research, \
mock.patch.object(BeautifulSoup, 'find', return_value=result):
check_for_meta("content", "url")
self.assertFalse(research.called)
def test_check_for_meta_with_content_split_bad(self):
"""
В контенте не 2 параметра
:return:
"""
result = self._prepare_mock()
result.__getitem__ = mock.Mock(return_value="content")
with mock.patch.object(re, 'search', mock.Mock()) as research, \
mock.patch.object(BeautifulSoup, 'find', return_value=result):
check_for_meta("content", "url")
self.assertFalse(research.called)
def test_check_for_meta_with_content_split_bad(self):
"""
В контенте не 2 параметра
:return:
"""
result = mock.MagicMock(name="result")
result.attrs = {
"content": True,
"http-equiv": "refresh"
}
result.__getitem__ = mock.Mock(return_value="content")
with mock.patch.object(re, 'search', mock.Mock()) as research:
with mock.patch.object(BeautifulSoup, 'find', return_value=result):
check_for_meta("content", "url")
assert research.called is False
def test_check_for_meta_no_meta(self):
content = self.get_default_html()
url_prefix = 'http://mail.ru'
url = lib.check_for_meta(content, url_prefix)
assert url is None, 'url found but no url was specified'
def test_check_for_meta_no_meta(self):
"""
Result is None
"""
result = None
with mock.patch.object(BeautifulSoup, "find", return_value=result):
self.assertIsNone(check_for_meta("content", "url"))
def test_check_for_meta(self):
content = '<meta http-equiv="refresh" content="content; url=redirect_url">'
url = 'http://url.ru'
redirect_url = 'http://redirect-url.ru'
with mock.patch('source.lib.to_unicode', mock.Mock(return_value=redirect_url)):
self.assertEquals(lib.urljoin(url, redirect_url), lib.check_for_meta(content, url))
def test_check_for_meta_not_content(self):
content = "meta: not content"
url = 'url'
result = mock.MagicMock()
with mock.patch('source.lib.BeautifulSoup.find', mock.Mock(return_value=result)):
self.assertEquals(None, lib.check_for_meta(content, url))
def test_check_for_meta_no_meta(self):
"""
find("meta") returns None
"""
ret = None
with mock.patch.object(BeautifulSoup, 'find', return_value=ret):
check = check_for_meta("content", "url")
self.assertIsNone(check)
def test_check_for_meta_no_httpequiv_attr(self):
result = mock.MagicMock(name="result")
result.attrs = {
"content": True,
}
with mock.patch.object(BeautifulSoup, 'find', return_value=result):
check = check_for_meta("content", "url")
self.assertIsNone(check)
def test_check_for_meta_no_meta(self):
"""
find("meta") returns None
"""
ret = None
with mock.patch.object(BeautifulSoup, 'find', return_value=ret):
check = check_for_meta("content", "url")
self.assertIsNone(check)
def test_check_for_meta(self):
content = self.get_redirect_html('page.html')
url_prefix = 'http://mail.ru'
url = lib.check_for_meta(content, url_prefix)
assert url, 'No meta tag found'
assert url.startswith(url_prefix), 'supplied url prefix not took in account'
def test_check_for_meta_httpequiv_no_refresh(self):
"""
http-equiv is /no refresh/
"""
result = mock.MagicMock(name="result")
result.attrs = {"content": True, 'http-equiv': "no refresh"}
with mock.patch.object(BeautifulSoup, 'find', return_value=result):
check = check_for_meta("content", "url")
self.assertIsNone(check)
def test_check_for_meta(self):
meta_url = 'test.url'
url = 'http://test/'
content = """
<html>
<head>
<meta http-equiv="refresh" content="5; url=%s">
</head>
</html>""" % meta_url
self.assertEqual(lib.check_for_meta(content, url), url+meta_url)
def test_check_for_meta(self):
meta_url = 'test.url'
url = 'http://test/'
content = """
<html>
<head>
<meta http-equiv="refresh" content="5; url=%s">
</head>
</html>""" % meta_url
self.assertEqual(lib.check_for_meta(content, url), url + meta_url)
def test_check_for_meta_no_http_equiv(self):
"""
No http_equiv in result
"""
result = mock.MagicMock(name="result")
result.attrs = {
"content": "contentDummy",
}
with mock.patch.object(BeautifulSoup, "find", return_value=result):
self.assertIsNone(check_for_meta("content", "url"))
def test_check_for_meta_not_refresh(self):
content = "meta: http-equiv=Content-Type content=content"
url = 'url'
result = mock.MagicMock()
result.attrs = {
'content': 'content',
'http-equiv': 'Content-Type',
}
with mock.patch('source.lib.BeautifulSoup.find', mock.Mock(return_value=result)):
self.assertEquals(None, lib.check_for_meta(content, url))
def test_check_for_meta_correct(self):
"""
Весь путь в check_for_meta
:return:
"""
url = "localhost/lal?what_are_you_doing=dont_know"
result = self._prepare_mock()
result.__getitem__ = mock.Mock(return_value="wait;url=" + url)
with mock.patch.object(BeautifulSoup, 'find', return_value=result):
check = check_for_meta("content", "url")
self.assertEquals(check, url)
def test_check_for_meta_without_httpequiv_attr(self):
"""
http-equiv unspecified
"""
result = mock.MagicMock(name="result")
result.attrs = {
"content": True,
}
with mock.patch.object(BeautifulSoup, 'find', return_value=result):
check = check_for_meta("content", "url")
self.assertIsNone(check)
def test_check_for_meta_not_result(self):
content = """
<!DOCTYPE html>
<html>
<head></head>
<body></body>
</html>
"""
url = 'url'
self.assertEquals(None, lib.check_for_meta(content, url))
def test_check_for_meta_correct(self):
"""
Весь путь в check_for_meta
:return:
"""
url = "localhost/lal?what_are_you_doing=dont_know"
result = self._prepare_mock()
result.__getitem__ = mock.Mock(return_value="wait;url=" + url)
with mock.patch.object(BeautifulSoup, 'find', return_value=result):
check = check_for_meta("content", "url")
self.assertEquals(check, url)
def test_check_for_meta_httpequiv_no_refresh(self):
"""
http-equiv is /no refresh/
"""
result = mock.MagicMock(name="result")
result.attrs = {
"content": True,
'http-equiv': "no refresh"
}
with mock.patch.object(BeautifulSoup, 'find', return_value=result):
check = check_for_meta("content", "url")
self.assertIsNone(check)
def test_check_for_meta_wrong_length(self):
"""
Splitted content length is wrong
"""
result = mock.MagicMock(name="result")
result.attrs = {
"content": "dummy;dummy;dummy",
"http-equiv": "refresh"
}
result.__getitem__ = mock.Mock(return_value=result.attrs["content"])
with mock.patch.object(BeautifulSoup, "find", return_value=result):
self.assertIsNone(check_for_meta("content", "url"))
def test_check_for_meta_not_search_url(self):
content = """
<!DOCTYPE html>
<html>
<head>
<meta http-equiv="refresh" content="5; url=">
</head>
<body></body>
</html>
"""
url = 'url'
self.assertEquals(None, lib.check_for_meta(content, url))
def test_check_for_meta_not_refresh(self):
content = """
<!DOCTYPE html>
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
</head>
<body></body>
</html>
"""
url = 'url'
self.assertEquals(None, lib.check_for_meta(content, url))
def test_check_for_meta_len_splitted_not_equals_2(self):
content = """
<!DOCTYPE html>
<html>
<head>
<meta http-equiv="ReFresh" content="5" />
</head>
<body></body>
</html>
"""
url = 'url'
self.assertEquals(None, lib.check_for_meta(content, url))
def test_check_for_meta_not_http_equiv(self):
content = """
<!DOCTYPE html>
<html>
<head>
<meta name="refresh" content="content" />
</head>
<body></body>
</html>
"""
url = 'url'
self.assertEquals(None, lib.check_for_meta(content, url))
def test_check_for_meta_dfs(self):
"""
DFS -> 100% line coverage, but bad branch cov.
"""
result = mock.MagicMock(name="result")
result.attrs = {
"content": "wat;url=",
"http-equiv": "refresh",
}
url = 'example.com/hell.php?what=123'
result.__getitem__ = mock.Mock(return_value=result.attrs["content"] + url)
with mock.patch.object(BeautifulSoup, "find", return_value=result):
check = check_for_meta("content", "url")
self.assertEquals(check, url)
def test_check_for_meta(self):
url = 'http://url.ru'
redirect_url = 'http://redirect-url.ru'
content = """
<!DOCTYPE html>
<html>
<head>
<meta http-equiv="refresh" content="5; url=""" + redirect_url + """">
</head>
<body></body>
</html>
"""
with mock.patch('source.lib.to_unicode', mock.Mock(return_value=redirect_url)):
self.assertEquals(lib.urljoin(url, redirect_url), lib.check_for_meta(content, url))
def test_check_for_meta_wrong_search(self):
"""
re.search returned none
"""
result = mock.MagicMock(name="result")
result.attrs = {
"content": "wat;url=",
"http-equiv": "refresh",
}
url = 'example.com/hell.php?what=123'
result.__getitem__ = mock.Mock(return_value=result.attrs["content"] + url)
with mock.patch.object(BeautifulSoup, "find", return_value=result):
with mock.patch("source.lib.urljoin", mock.Mock()):
self.assertIsNone(check_for_meta("content", "url"))
def test_check_for_meta_correct_content_correct_research(self):
"""
Весь путь в check_for_meta
:return:
"""
result = mock.MagicMock(name="result")
result.attrs = {
"content": True,
"http-equiv": "refresh"
}
url = "localhost/lal?what_are_you_doing=dont_know"
result.__getitem__ = mock.Mock(return_value="wait;url=" + url)
with mock.patch.object(BeautifulSoup, 'find', return_value=result):
check = check_for_meta("content", "url")
self.assertEquals(check, url)
def test_check_for_meta_no_equal_sign(self):
content = r"""
<html>
<head>
<meta http-equiv="refresh" content="5; url">
</head>
<body>
</body>
</html>"""
url_prefix = 'http://mail.ru'
url = lib.check_for_meta(content, url_prefix)
assert url is None, 'url found but no url was specified (no equal sign)'
def test_check_for_meta_bad_html(self):
content = """
<html>
<head>
<meta http-equiv="refresh" content="5url={0}">
</head>
<body>
</body>
</html>"""
url_prefix = 'http://mail.ru'
url = lib.check_for_meta(content, url_prefix)
self.assertEquals(url, None, 'such bad html should be parsed')
def test_check_for_meta_cant_research(self):
"""
re.search ничего не вернул
:return:
"""
result = mock.MagicMock(name="result")
result.attrs = {
"content": True,
"http-equiv": "refresh"
}
url = "localhost/lal?what_are_you_doing=dont_know"
result.__getitem__ = mock.Mock(return_value="wait;url=" + url)
with mock.patch("source.lib.urljoin", mock.Mock()) as urljoin:
with mock.patch.object(BeautifulSoup, 'find', return_value=result):
check = check_for_meta("content", "url")
self.assertFalse(urljoin.called)
self.assertIsNone(check)
def test_check_for_bad_meta(self):
meta = '<html><head><meta http-equiv="refresh" content="0;URL=''http://thetudors.example.com/''; sdfsdfsdf" /></head></html>'
self.assertIsNone(check_for_meta(meta, ''))
def test_check_for_meta_wrong_url(self):
self.assertEqual(
check_for_meta('<html><head><meta http-equiv="refresh" content="0;URwqeL=''http://thetudors.example.com/''" /></head></html>', ''),
None
)
def test_check_for_meta_wrong(self):
self.assertEqual(check_for_meta('', ''), None)