args_to_str,
encode_base_n,
clean_html,
+ date_from_str,
DateRange,
detect_exe_version,
determine_ext,
encodeFilename,
escape_rfc3986,
escape_url,
+ extract_attributes,
ExtractorError,
find_xpath_attr,
fix_xml_ampersands,
orderedSet,
parse_duration,
parse_filesize,
+ parse_count,
parse_iso8601,
read_batch_urls,
sanitize_filename,
cli_bool_option,
)
from youtube_dl.compat import (
+ compat_chr,
compat_etree_fromstring,
compat_urlparse,
compat_parse_qs,
self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
- tests = 'a\xe4b\u4e2d\u56fd\u7684c'
- self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
+ tests = 'aäb\u4e2d\u56fd\u7684c'
+ self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
self.assertTrue(sanitize_filename('-', restricted=True) != '')
self.assertTrue(sanitize_filename(':', restricted=True) != '')
+ self.assertEqual(sanitize_filename(
+ 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ', restricted=True),
+ 'AAAAAAAECEEEEIIIIDNOOOOOOUUUUYPssaaaaaaaeceeeeiiiionoooooouuuuypy')
+ pass
+
def test_sanitize_ids(self):
self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
self.assertEqual(unescapeHTML('é'), 'é')
self.assertEqual(unescapeHTML('�'), '�')
+ def test_date_from_str(self):
+ self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
+ self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
+ self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
+ self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
+ self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
+
def test_daterange(self):
_20century = DateRange("19000101", "20000101")
self.assertFalse("17890714" in _20century)
self.assertEqual(parse_duration('01:02:03:04'), 93784)
self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
self.assertEqual(parse_duration('87 Min.'), 5220)
+ self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
def test_fix_xml_ampersands(self):
self.assertEqual(
)
self.assertEqual(
escape_url('http://тест.рф/фрагмент'),
- 'http://тест.рф/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
+ 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
)
self.assertEqual(
escape_url('http://тест.рф/абв?абв=абв#абв'),
- 'http://тест.рф/%D0%B0%D0%B1%D0%B2?%D0%B0%D0%B1%D0%B2=%D0%B0%D0%B1%D0%B2#%D0%B0%D0%B1%D0%B2'
+ 'http://xn--e1aybc.xn--p1ai/%D0%B0%D0%B1%D0%B2?%D0%B0%D0%B1%D0%B2=%D0%B0%D0%B1%D0%B2#%D0%B0%D0%B1%D0%B2'
)
self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
on = js_to_json('{"abc": "def",}')
self.assertEqual(json.loads(on), {'abc': 'def'})
+ 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(extract_attributes('<e x=y>'), {'x': 'y'})
+ self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
+ self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
+ self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
+ self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
+ self.assertEqual(extract_attributes('<e x="&">'), {'x': '&'}) # XML
+ self.assertEqual(extract_attributes('<e x=""">'), {'x': '"'})
+ self.assertEqual(extract_attributes('<e x="£">'), {'x': '£'}) # HTML 3.2
+ self.assertEqual(extract_attributes('<e x="λ">'), {'x': 'λ'}) # HTML 4.0
+ self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
+ self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
+ self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
+ self.assertEqual(extract_attributes('<e x >'), {'x': None})
+ self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
+ self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
+ self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
+ self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
+ self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
+ self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
+ self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
+ self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}) # Names lowercased
+ self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
+ self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
+ self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
+ self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
+ self.assertEqual(extract_attributes('<e x="décomposé">'), {'x': 'décompose\u0301'})
+ # "Narrow" Python builds don't support unicode code points outside BMP.
+ try:
+ compat_chr(0x10000)
+ supports_outside_bmp = True
+ except ValueError:
+ supports_outside_bmp = False
+ if supports_outside_bmp:
+ self.assertEqual(extract_attributes('<e x="Smile 😀!">'), {'x': 'Smile \U0001f600!'})
+
def test_clean_html(self):
self.assertEqual(clean_html('a:\nb'), 'a: b')
self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
self.assertEqual(parse_filesize('1,24 KB'), 1240)
+ def test_parse_count(self):
+ self.assertEqual(parse_count(None), None)
+ self.assertEqual(parse_count(''), None)
+ self.assertEqual(parse_count('0'), 0)
+ self.assertEqual(parse_count('1000'), 1000)
+ self.assertEqual(parse_count('1.000'), 1000)
+ self.assertEqual(parse_count('1.1k'), 1100)
+ self.assertEqual(parse_count('1.1kk'), 1100000)
+ self.assertEqual(parse_count('1.1kk '), 1100000)
+ self.assertEqual(parse_count('1.1kk views'), 1100000)
+
def test_version_tuple(self):
self.assertEqual(version_tuple('1'), (1,))
self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))