is_html,
js_to_json,
limit_length,
+ mimetype2ext,
+ month_by_name,
ohdave_rsa_encrypt,
OnDemandPagedList,
orderedSet,
uppercase_escape,
lowercase_escape,
url_basename,
+ base_url,
+ urljoin,
urlencode_postdata,
urshift,
update_url_query,
self.assertEqual(unified_strdate('25-09-2014'), '20140925')
self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
+ self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
+ self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
+ self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
+ self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
def test_unified_timestamps(self):
self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
+ self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
def test_determine_ext(self):
self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
'trailer.mp4')
+ def test_base_url(self):
+ self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
+ self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
+ self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
+ self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
+ self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
+
+ def test_urljoin(self):
+ self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
+ self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
+ self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
+ self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
+ self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
+ self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
+ self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
+ self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
+ self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
+ self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
+ self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
+ self.assertEqual(urljoin('http://foo.de/', None), None)
+ self.assertEqual(urljoin('http://foo.de/', ''), None)
+ self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
+ self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
+
def test_parse_age_limit(self):
self.assertEqual(parse_age_limit(None), None)
self.assertEqual(parse_age_limit(False), None)
self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
self.assertEqual(parse_duration('87 Min.'), 5220)
self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
+ self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
def test_fix_xml_ampersands(self):
self.assertEqual(
limit_length('foo bar baz asd', 12).startswith('foo bar'))
self.assertTrue('...' in limit_length('foo bar baz asd', 12))
+ def test_mimetype2ext(self):
+ self.assertEqual(mimetype2ext(None), None)
+ self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
+ self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
+ self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
+ self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
+ self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
+
+ def test_month_by_name(self):
+ self.assertEqual(month_by_name(None), None)
+ self.assertEqual(month_by_name('December', 'en'), 12)
+ self.assertEqual(month_by_name('décembre', 'fr'), 12)
+ self.assertEqual(month_by_name('December'), 12)
+ self.assertEqual(month_by_name('décembre'), None)
+ self.assertEqual(month_by_name('Unknown', 'unknown'), None)
+
def test_parse_codecs(self):
self.assertEqual(parse_codecs(''), {})
self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
inp = '''{"foo":101}'''
self.assertEqual(js_to_json(inp), '''{"foo":101}''')
+ inp = '''{"duration": "00:01:07"}'''
+ self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
+
def test_js_to_json_edgecases(self):
on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
on = js_to_json('["abc", "def",]')
self.assertEqual(json.loads(on), ['abc', 'def'])
+ on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
+ self.assertEqual(json.loads(on), ['abc', 'def'])
+
+ on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
+ self.assertEqual(json.loads(on), ['abc', 'def'])
+
on = js_to_json('{"abc": "def",}')
self.assertEqual(json.loads(on), {'abc': 'def'})
+ on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
+ self.assertEqual(json.loads(on), {'abc': 'def'})
+
on = js_to_json('{ 0: /* " \n */ ",]" , }')
self.assertEqual(json.loads(on), {'0': ',]'})
+ on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
+ self.assertEqual(json.loads(on), {'0': ',]'})
+
+ on = js_to_json('{ 0: // comment\n1 }')
+ self.assertEqual(json.loads(on), {'0': 1})
+
on = js_to_json(r'["<p>x<\/p>"]')
self.assertEqual(json.loads(on), ['<p>x</p>'])
on = js_to_json("['a\\\nb']")
self.assertEqual(json.loads(on), ['ab'])
+ on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
+ self.assertEqual(json.loads(on), ['ab'])
+
on = js_to_json('{0xff:0xff}')
self.assertEqual(json.loads(on), {'255': 255})
+ on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
+ self.assertEqual(json.loads(on), {'255': 255})
+
on = js_to_json('{077:077}')
self.assertEqual(json.loads(on), {'63': 63})
+ on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
+ self.assertEqual(json.loads(on), {'63': 63})
+
on = js_to_json('{42:42}')
self.assertEqual(json.loads(on), {'42': 42})
+ on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
+ self.assertEqual(json.loads(on), {'42': 42})
+
def test_extract_attributes(self):
self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
self.assertEqual(parse_filesize('2 MiB'), 2097152)
self.assertEqual(parse_filesize('5 GB'), 5000000000)
self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
+ self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
self.assertEqual(parse_filesize('1,24 KB'), 1240)
+ self.assertEqual(parse_filesize('1,24 kb'), 1240)
+ self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
def test_parse_count(self):
self.assertEqual(parse_count(None), None)
self.assertEqual(get_element_by_class('foo', html), 'nice')
self.assertEqual(get_element_by_class('no-such-class', html), None)
+
if __name__ == '__main__':
unittest.main()