self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
+ self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
def test_determine_ext(self):
self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
d = json.loads(stripped)
self.assertEqual(d, {'status': 'success'})
+ stripped = strip_jsonp('window.cb && window.cb({"status": "success"});')
+ d = json.loads(stripped)
+ self.assertEqual(d, {'status': 'success'})
+
+ stripped = strip_jsonp('window.cb && cb({"status": "success"});')
+ d = json.loads(stripped)
+ self.assertEqual(d, {'status': 'success'})
+
def test_uppercase_escape(self):
self.assertEqual(uppercase_escape('aä'), 'aä')
self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
supports_outside_bmp = False
if supports_outside_bmp:
self.assertEqual(extract_attributes('<e x="Smile 😀!">'), {'x': 'Smile \U0001f600!'})
+ # Malformed HTML should not break attributes extraction on older Python
+ self.assertEqual(extract_attributes('<mal"formed/>'), {})
def test_clean_html(self):
self.assertEqual(clean_html('a:\nb'), 'a: b')
self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
+ html = '''
+ <div itemprop="author" itemscope>foo</div>
+ '''
+
+ self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo')
+
def test_get_elements_by_class(self):
html = '''
<span class="foo bar">nice</span><span class="foo bar">also nice</span>