+ # Ignore JavaScript code as well
+ on = js_to_json('''{
+ "x": 1,
+ y: "a",
+ z: some.code
+ }''')
+ d = json.loads(on)
+ self.assertEqual(d['x'], 1)
+ self.assertEqual(d['y'], 'a')
+
+ on = js_to_json('["abc", "def",]')
+ self.assertEqual(json.loads(on), ['abc', 'def'])
+
+ on = js_to_json('{"abc": "def",}')
+ self.assertEqual(json.loads(on), {'abc': 'def'})
+
+ on = js_to_json('{ 0: /* " \n */ ",]" , }')
+ self.assertEqual(json.loads(on), {'0': ',]'})
+
+ on = js_to_json(r'["<p>x<\/p>"]')
+ self.assertEqual(json.loads(on), ['<p>x</p>'])
+
+ on = js_to_json(r'["\xaa"]')
+ self.assertEqual(json.loads(on), ['\u00aa'])
+
+ on = js_to_json("['a\\\nb']")
+ self.assertEqual(json.loads(on), ['ab'])
+
+ on = js_to_json('{0xff:0xff}')
+ self.assertEqual(json.loads(on), {'255': 255})
+
+ on = js_to_json('{077:077}')
+ self.assertEqual(json.loads(on), {'63': 63})
+
+ on = js_to_json('{42:42}')
+ 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(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"')
+
+ def test_intlist_to_bytes(self):
+ self.assertEqual(
+ intlist_to_bytes([0, 1, 127, 128, 255]),
+ b'\x00\x01\x7f\x80\xff')
+
+ def test_args_to_str(self):
+ self.assertEqual(
+ args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
+ 'foo ba/r -baz \'2 be\' \'\''
+ )
+
+ def test_parse_filesize(self):
+ self.assertEqual(parse_filesize(None), None)
+ self.assertEqual(parse_filesize(''), None)
+ self.assertEqual(parse_filesize('91 B'), 91)
+ self.assertEqual(parse_filesize('foobar'), None)
+ 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('1,24 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))
+ self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
+
+ def test_detect_exe_version(self):
+ self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
+built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
+configuration: --prefix=/usr --extra-'''), '1.2.1')
+ self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
+built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
+ self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
+Trying to open render node...
+Success at /dev/dri/renderD128.
+ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
+
+ def test_age_restricted(self):
+ self.assertFalse(age_restricted(None, 10)) # unrestricted content
+ self.assertFalse(age_restricted(1, None)) # unrestricted policy
+ self.assertFalse(age_restricted(8, 10))
+ self.assertTrue(age_restricted(18, 14))
+ self.assertFalse(age_restricted(18, 18))
+
+ def test_is_html(self):
+ self.assertFalse(is_html(b'\x49\x44\x43<html'))
+ self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
+ self.assertTrue(is_html( # UTF-8 with BOM
+ b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
+ self.assertTrue(is_html( # UTF-16-LE
+ b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
+ ))
+ self.assertTrue(is_html( # UTF-16-BE
+ b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
+ ))
+ self.assertTrue(is_html( # UTF-32-BE
+ b'\x00\x00\xFE\xFF\x00\x00\x00<\x00\x00\x00h\x00\x00\x00t\x00\x00\x00m\x00\x00\x00l\x00\x00\x00>\x00\x00\x00\xe4'))
+ self.assertTrue(is_html( # UTF-32-LE
+ b'\xFF\xFE\x00\x00<\x00\x00\x00h\x00\x00\x00t\x00\x00\x00m\x00\x00\x00l\x00\x00\x00>\x00\x00\x00\xe4\x00\x00\x00'))
+
+ def test_render_table(self):
+ self.assertEqual(
+ render_table(
+ ['a', 'bcd'],
+ [[123, 4], [9999, 51]]),
+ 'a bcd\n'
+ '123 4\n'
+ '9999 51')
+
+ def test_match_str(self):
+ self.assertRaises(ValueError, match_str, 'xy>foobar', {})
+ self.assertFalse(match_str('xy', {'x': 1200}))
+ self.assertTrue(match_str('!xy', {'x': 1200}))
+ self.assertTrue(match_str('x', {'x': 1200}))
+ self.assertFalse(match_str('!x', {'x': 1200}))
+ self.assertTrue(match_str('x', {'x': 0}))
+ self.assertFalse(match_str('x>0', {'x': 0}))
+ self.assertFalse(match_str('x>0', {}))
+ self.assertTrue(match_str('x>?0', {}))
+ self.assertTrue(match_str('x>1K', {'x': 1200}))
+ self.assertFalse(match_str('x>2K', {'x': 1200}))
+ self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
+ self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
+ self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
+ self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
+ self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
+ self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
+ self.assertFalse(match_str(
+ 'like_count > 100 & dislike_count <? 50 & description',
+ {'like_count': 90, 'description': 'foo'}))
+ self.assertTrue(match_str(
+ 'like_count > 100 & dislike_count <? 50 & description',
+ {'like_count': 190, 'description': 'foo'}))
+ self.assertFalse(match_str(
+ 'like_count > 100 & dislike_count <? 50 & description',
+ {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
+ self.assertFalse(match_str(
+ 'like_count > 100 & dislike_count <? 50 & description',
+ {'like_count': 190, 'dislike_count': 10}))
+
+ def test_parse_dfxp_time_expr(self):
+ self.assertEqual(parse_dfxp_time_expr(None), None)
+ self.assertEqual(parse_dfxp_time_expr(''), None)
+ self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
+ self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
+ self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
+ self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
+ self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
+
+ def test_dfxp2srt(self):
+ dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
+ <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
+ <body>
+ <div xml:lang="en">
+ <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
+ <p begin="1" end="2">第二行<br/>♪♪</p>
+ <p begin="2" dur="1"><span>Third<br/>Line</span></p>
+ <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
+ <p begin="-1" end="-1">Ignore, two</p>
+ <p begin="3" dur="-1">Ignored, three</p>
+ </div>
+ </body>
+ </tt>'''
+ srt_data = '''1
+00:00:00,000 --> 00:00:01,000
+The following line contains Chinese characters and special symbols
+
+2
+00:00:01,000 --> 00:00:02,000
+第二行
+♪♪
+
+3
+00:00:02,000 --> 00:00:03,000
+Third
+Line
+
+'''
+ self.assertEqual(dfxp2srt(dfxp_data), srt_data)
+
+ dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
+ <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
+ <body>
+ <div xml:lang="en">
+ <p begin="0" end="1">The first line</p>
+ </div>
+ </body>
+ </tt>'''
+ srt_data = '''1
+00:00:00,000 --> 00:00:01,000
+The first line
+
+'''
+ self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
+
+ def test_cli_option(self):
+ self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
+ self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
+ self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
+ self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
+
+ def test_cli_valueless_option(self):
+ self.assertEqual(cli_valueless_option(
+ {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
+ self.assertEqual(cli_valueless_option(
+ {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
+ self.assertEqual(cli_valueless_option(
+ {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
+ self.assertEqual(cli_valueless_option(
+ {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
+ self.assertEqual(cli_valueless_option(
+ {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
+ self.assertEqual(cli_valueless_option(
+ {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
+
+ def test_cli_bool_option(self):
+ self.assertEqual(
+ cli_bool_option(
+ {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
+ ['--no-check-certificate', 'true'])
+ self.assertEqual(
+ cli_bool_option(
+ {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
+ ['--no-check-certificate=true'])
+ self.assertEqual(
+ cli_bool_option(
+ {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
+ ['--check-certificate', 'false'])
+ self.assertEqual(
+ cli_bool_option(
+ {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
+ ['--check-certificate=false'])
+ self.assertEqual(
+ cli_bool_option(
+ {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
+ ['--check-certificate', 'true'])
+ self.assertEqual(
+ cli_bool_option(
+ {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
+ ['--check-certificate=true'])
+
+ def test_ohdave_rsa_encrypt(self):
+ N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
+ e = 65537
+
+ self.assertEqual(
+ ohdave_rsa_encrypt(b'aa111222', e, N),
+ '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
+
+ def test_encode_base_n(self):
+ self.assertEqual(encode_base_n(0, 30), '0')
+ self.assertEqual(encode_base_n(80, 30), '2k')
+
+ custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
+ self.assertEqual(encode_base_n(0, 30, custom_table), '9')
+ self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
+
+ self.assertRaises(ValueError, encode_base_n, 0, 70)
+ self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
+
+ def test_urshift(self):
+ self.assertEqual(urshift(3, 1), 1)
+ self.assertEqual(urshift(-3, 1), 2147483646)
+
+ def test_get_element_by_class(self):
+ html = '''
+ <span class="foo bar">nice</span>
+ '''
+
+ self.assertEqual(get_element_by_class('foo', html), 'nice')
+ self.assertEqual(get_element_by_class('no-such-class', html), None)