ExtractorError,
find_xpath_attr,
fix_xml_ampersands,
+ float_or_none,
get_element_by_class,
get_element_by_attribute,
get_elements_by_class,
get_elements_by_attribute,
InAdvancePagedList,
+ int_or_none,
intlist_to_bytes,
is_html,
js_to_json,
limit_length,
+ merge_dicts,
mimetype2ext,
month_by_name,
multipart_encode,
parse_filesize,
parse_count,
parse_iso8601,
+ parse_resolution,
+ parse_bitrate,
pkcs1pad,
read_batch_urls,
sanitize_filename,
sanitize_path,
+ sanitize_url,
expand_path,
prepend_extension,
replace_extension,
smuggle_url,
str_to_int,
strip_jsonp,
+ strip_or_none,
timeconvert,
unescapeHTML,
unified_strdate,
uppercase_escape,
lowercase_escape,
url_basename,
+ url_or_none,
base_url,
urljoin,
urlencode_postdata,
self.assertEqual(sanitize_filename(
'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
- 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYPssaaaaaaaeceeeeiiiionooooooooeuuuuuypy')
+ 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy')
def test_sanitize_ids(self):
self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
self.assertEqual(sanitize_path('./abc'), 'abc')
self.assertEqual(sanitize_path('./../abc'), '..\\abc')
+ def test_sanitize_url(self):
+ self.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
+ self.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
+ self.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
+ self.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')
+
def test_expand_path(self):
def env(var):
return '%{0}%'.format(var) if sys.platform == 'win32' else '${0}'.format(var)
self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
+ self.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363)
def test_determine_ext(self):
self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
+ self.assertEqual(determine_ext('foobar', None), None)
def test_find_xpath_attr(self):
testxml = '''<root>
shell_quote(args),
"""ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
+ def test_float_or_none(self):
+ self.assertEqual(float_or_none('42.42'), 42.42)
+ self.assertEqual(float_or_none('42'), 42.0)
+ self.assertEqual(float_or_none(''), None)
+ self.assertEqual(float_or_none(None), None)
+ self.assertEqual(float_or_none([]), None)
+ self.assertEqual(float_or_none(set()), None)
+
+ def test_int_or_none(self):
+ self.assertEqual(int_or_none('42'), 42)
+ self.assertEqual(int_or_none(''), None)
+ self.assertEqual(int_or_none(None), None)
+ self.assertEqual(int_or_none([]), None)
+ self.assertEqual(int_or_none(set()), None)
+
def test_str_to_int(self):
self.assertEqual(str_to_int('123,456'), 123456)
self.assertEqual(str_to_int('123.456'), 123456)
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')
+ self.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de')
+ self.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de')
+
+ def test_url_or_none(self):
+ self.assertEqual(url_or_none(None), None)
+ self.assertEqual(url_or_none(''), None)
+ self.assertEqual(url_or_none('foo'), None)
+ self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
+ self.assertEqual(url_or_none('https://foo.de'), 'https://foo.de')
+ self.assertEqual(url_or_none('http$://foo.de'), None)
+ self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
+ self.assertEqual(url_or_none('//foo.de'), '//foo.de')
def test_parse_age_limit(self):
self.assertEqual(parse_age_limit(None), None)
self.assertEqual(parse_age_limit('PG-13'), 13)
self.assertEqual(parse_age_limit('TV-14'), 14)
self.assertEqual(parse_age_limit('TV-MA'), 17)
+ self.assertEqual(parse_age_limit('TV14'), 14)
+ self.assertEqual(parse_age_limit('TV_G'), 0)
def test_parse_duration(self):
self.assertEqual(parse_duration(None), None)
self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
+ def test_merge_dicts(self):
+ self.assertEqual(merge_dicts({'a': 1}, {'b': 2}), {'a': 1, 'b': 2})
+ self.assertEqual(merge_dicts({'a': 1}, {'a': 2}), {'a': 1})
+ self.assertEqual(merge_dicts({'a': 1}, {'a': None}), {'a': 1})
+ self.assertEqual(merge_dicts({'a': 1}, {'a': ''}), {'a': 1})
+ self.assertEqual(merge_dicts({'a': 1}, {}), {'a': 1})
+ self.assertEqual(merge_dicts({'a': None}, {'a': 1}), {'a': 1})
+ self.assertEqual(merge_dicts({'a': ''}, {'a': 1}), {'a': ''})
+ self.assertEqual(merge_dicts({'a': ''}, {'a': 'abc'}), {'a': 'abc'})
+ self.assertEqual(merge_dicts({'a': None}, {'a': ''}, {'a': 'abc'}), {'a': 'abc'})
+
def test_encode_compat_str(self):
self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
d = json.loads(stripped)
self.assertEqual(d, {'status': 'success'})
+ stripped = strip_jsonp('({"status": "success"});')
+ d = json.loads(stripped)
+ self.assertEqual(d, {'status': 'success'})
+
+ def test_strip_or_none(self):
+ self.assertEqual(strip_or_none(' abc'), 'abc')
+ self.assertEqual(strip_or_none('abc '), 'abc')
+ self.assertEqual(strip_or_none(' abc '), 'abc')
+ self.assertEqual(strip_or_none('\tabc\t'), 'abc')
+ self.assertEqual(strip_or_none('\n\tabc\n\t'), 'abc')
+ self.assertEqual(strip_or_none('abc'), 'abc')
+ self.assertEqual(strip_or_none(''), '')
+ self.assertEqual(strip_or_none(None), None)
+ self.assertEqual(strip_or_none(42), None)
+ self.assertEqual(strip_or_none([]), None)
+
def test_uppercase_escape(self):
self.assertEqual(uppercase_escape('aä'), 'aä')
self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
'vcodec': 'h264',
'acodec': 'aac',
})
+ self.assertEqual(parse_codecs('av01.0.05M.08'), {
+ 'vcodec': 'av01.0.05M.08',
+ 'acodec': 'none',
+ })
def test_escape_rfc3986(self):
reserved = "!*'();:@&=+$,/?#[]"
inp = '''{"duration": "00:01:07"}'''
self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
+ inp = '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}'''
+ self.assertEqual(js_to_json(inp), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''')
+
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('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
self.assertEqual(json.loads(on), {'42': 42})
+ on = js_to_json('{42:4.2e1}')
+ self.assertEqual(json.loads(on), {'42': 42.0})
+
+ def test_js_to_json_malformed(self):
+ self.assertEqual(js_to_json('42a1'), '42"a1"')
+ self.assertEqual(js_to_json('42a-1'), '42"a"-1')
+
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_count('1.1kk '), 1100000)
self.assertEqual(parse_count('1.1kk views'), 1100000)
+ def test_parse_resolution(self):
+ self.assertEqual(parse_resolution(None), {})
+ self.assertEqual(parse_resolution(''), {})
+ self.assertEqual(parse_resolution('1920x1080'), {'width': 1920, 'height': 1080})
+ self.assertEqual(parse_resolution('1920×1080'), {'width': 1920, 'height': 1080})
+ self.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080})
+ self.assertEqual(parse_resolution('720p'), {'height': 720})
+ self.assertEqual(parse_resolution('4k'), {'height': 2160})
+ self.assertEqual(parse_resolution('8K'), {'height': 4320})
+
+ def test_parse_bitrate(self):
+ self.assertEqual(parse_bitrate(None), None)
+ self.assertEqual(parse_bitrate(''), None)
+ self.assertEqual(parse_bitrate('300kbps'), 300)
+ self.assertEqual(parse_bitrate('1500kbps'), 1500)
+ self.assertEqual(parse_bitrate('300 kbps'), 300)
+
def test_version_tuple(self):
self.assertEqual(version_tuple('1'), (1,))
self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
self.assertFalse(match_str(
'like_count > 100 & dislike_count <? 50 & description',
{'like_count': 190, 'dislike_count': 10}))
+ self.assertTrue(match_str('is_live', {'is_live': True}))
+ self.assertFalse(match_str('is_live', {'is_live': False}))
+ self.assertFalse(match_str('is_live', {'is_live': None}))
+ self.assertFalse(match_str('is_live', {}))
+ self.assertFalse(match_str('!is_live', {'is_live': True}))
+ self.assertTrue(match_str('!is_live', {'is_live': False}))
+ self.assertTrue(match_str('!is_live', {'is_live': None}))
+ self.assertTrue(match_str('!is_live', {}))
+ self.assertTrue(match_str('title', {'title': 'abc'}))
+ self.assertTrue(match_str('title', {'title': ''}))
+ self.assertFalse(match_str('!title', {'title': 'abc'}))
+ self.assertFalse(match_str('!title', {'title': ''}))
def test_parse_dfxp_time_expr(self):
self.assertEqual(parse_dfxp_time_expr(None), None)