test_utils.py - youtube-dl - Another place where youtube-dl lives on

test_utils.py (68992B)
      1 #!/usr/bin/env python
      2 # coding: utf-8
      3 
      4 from __future__ import unicode_literals
      5 
      6 # Allow direct execution
      7 import os
      8 import sys
      9 import unittest
     10 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
     11 
     12 
     13 # Various small unit tests
     14 import io
     15 import json
     16 import xml.etree.ElementTree
     17 
     18 from youtube_dl.utils import (
     19     age_restricted,
     20     args_to_str,
     21     encode_base_n,
     22     caesar,
     23     clean_html,
     24     clean_podcast_url,
     25     date_from_str,
     26     DateRange,
     27     detect_exe_version,
     28     determine_ext,
     29     dict_get,
     30     encode_compat_str,
     31     encodeFilename,
     32     escape_rfc3986,
     33     escape_url,
     34     extract_attributes,
     35     ExtractorError,
     36     find_xpath_attr,
     37     fix_xml_ampersands,
     38     float_or_none,
     39     get_element_by_class,
     40     get_element_by_attribute,
     41     get_elements_by_class,
     42     get_elements_by_attribute,
     43     InAdvancePagedList,
     44     int_or_none,
     45     intlist_to_bytes,
     46     is_html,
     47     js_to_json,
     48     limit_length,
     49     merge_dicts,
     50     mimetype2ext,
     51     month_by_name,
     52     multipart_encode,
     53     ohdave_rsa_encrypt,
     54     OnDemandPagedList,
     55     orderedSet,
     56     parse_age_limit,
     57     parse_duration,
     58     parse_filesize,
     59     parse_count,
     60     parse_iso8601,
     61     parse_resolution,
     62     parse_bitrate,
     63     pkcs1pad,
     64     read_batch_urls,
     65     sanitize_filename,
     66     sanitize_path,
     67     sanitize_url,
     68     expand_path,
     69     prepend_extension,
     70     replace_extension,
     71     remove_start,
     72     remove_end,
     73     remove_quotes,
     74     rot47,
     75     shell_quote,
     76     smuggle_url,
     77     str_to_int,
     78     strip_jsonp,
     79     strip_or_none,
     80     subtitles_filename,
     81     timeconvert,
     82     unescapeHTML,
     83     unified_strdate,
     84     unified_timestamp,
     85     unsmuggle_url,
     86     uppercase_escape,
     87     lowercase_escape,
     88     url_basename,
     89     url_or_none,
     90     base_url,
     91     urljoin,
     92     urlencode_postdata,
     93     urshift,
     94     update_url_query,
     95     version_tuple,
     96     xpath_with_ns,
     97     xpath_element,
     98     xpath_text,
     99     xpath_attr,
    100     render_table,
    101     match_str,
    102     parse_dfxp_time_expr,
    103     dfxp2srt,
    104     cli_option,
    105     cli_valueless_option,
    106     cli_bool_option,
    107     parse_codecs,
    108 )
    109 from youtube_dl.compat import (
    110     compat_chr,
    111     compat_etree_fromstring,
    112     compat_getenv,
    113     compat_os_name,
    114     compat_setenv,
    115     compat_urlparse,
    116     compat_parse_qs,
    117 )
    118 
    119 
    120 class TestUtil(unittest.TestCase):
    121     def test_timeconvert(self):
    122         self.assertTrue(timeconvert('') is None)
    123         self.assertTrue(timeconvert('bougrg') is None)
    124 
    125     def test_sanitize_filename(self):
    126         self.assertEqual(sanitize_filename('abc'), 'abc')
    127         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
    128 
    129         self.assertEqual(sanitize_filename('123'), '123')
    130 
    131         self.assertEqual('abc_de', sanitize_filename('abc/de'))
    132         self.assertFalse('/' in sanitize_filename('abc/de///'))
    133 
    134         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
    135         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
    136         self.assertEqual('yes no', sanitize_filename('yes? no'))
    137         self.assertEqual('this - that', sanitize_filename('this: that'))
    138 
    139         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
    140         aumlaut = 'ä'
    141         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
    142         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
    143         self.assertEqual(sanitize_filename(tests), tests)
    144 
    145         self.assertEqual(
    146             sanitize_filename('New World record at 0:12:34'),
    147             'New World record at 0_12_34')
    148 
    149         self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
    150         self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
    151         self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
    152         self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
    153 
    154         forbidden = '"\0\\/'
    155         for fc in forbidden:
    156             for fbc in forbidden:
    157                 self.assertTrue(fbc not in sanitize_filename(fc))
    158 
    159     def test_sanitize_filename_restricted(self):
    160         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
    161         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
    162 
    163         self.assertEqual(sanitize_filename('123', restricted=True), '123')
    164 
    165         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
    166         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
    167 
    168         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
    169         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
    170         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
    171         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
    172 
    173         tests = 'aäb\u4e2d\u56fd\u7684c'
    174         self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
    175         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
    176 
    177         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
    178         for fc in forbidden:
    179             for fbc in forbidden:
    180                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
    181 
    182         # Handle a common case more neatly
    183         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
    184         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
    185         # .. but make sure the file name is never empty
    186         self.assertTrue(sanitize_filename('-', restricted=True) != '')
    187         self.assertTrue(sanitize_filename(':', restricted=True) != '')
    188 
    189         self.assertEqual(sanitize_filename(
    190             'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
    191             'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy')
    192 
    193     def test_sanitize_ids(self):
    194         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
    195         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
    196         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
    197 
    198     def test_sanitize_path(self):
    199         if sys.platform != 'win32':
    200             return
    201 
    202         self.assertEqual(sanitize_path('abc'), 'abc')
    203         self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
    204         self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
    205         self.assertEqual(sanitize_path('abc|def'), 'abc#def')
    206         self.assertEqual(sanitize_path('<>:"|?*'), '#######')
    207         self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
    208         self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
    209 
    210         self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
    211         self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
    212 
    213         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
    214         self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
    215         self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
    216         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
    217 
    218         self.assertEqual(
    219             sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
    220             'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
    221 
    222         self.assertEqual(
    223             sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
    224             'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
    225         self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
    226         self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
    227         self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
    228 
    229         self.assertEqual(sanitize_path('../abc'), '..\\abc')
    230         self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
    231         self.assertEqual(sanitize_path('./abc'), 'abc')
    232         self.assertEqual(sanitize_path('./../abc'), '..\\abc')
    233 
    234     def test_sanitize_url(self):
    235         self.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
    236         self.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
    237         self.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
    238         self.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')
    239 
    240     def test_expand_path(self):
    241         def env(var):
    242             return '%{0}%'.format(var) if sys.platform == 'win32' else '${0}'.format(var)
    243 
    244         compat_setenv('YOUTUBE_DL_EXPATH_PATH', 'expanded')
    245         self.assertEqual(expand_path(env('YOUTUBE_DL_EXPATH_PATH')), 'expanded')
    246         self.assertEqual(expand_path(env('HOME')), compat_getenv('HOME'))
    247         self.assertEqual(expand_path('~'), compat_getenv('HOME'))
    248         self.assertEqual(
    249             expand_path('~/%s' % env('YOUTUBE_DL_EXPATH_PATH')),
    250             '%s/expanded' % compat_getenv('HOME'))
    251 
    252     def test_prepend_extension(self):
    253         self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
    254         self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
    255         self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
    256         self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
    257         self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
    258         self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
    259 
    260     def test_replace_extension(self):
    261         self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
    262         self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
    263         self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
    264         self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
    265         self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
    266         self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
    267 
    268     def test_subtitles_filename(self):
    269         self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt'), 'abc.en.vtt')
    270         self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt', 'ext'), 'abc.en.vtt')
    271         self.assertEqual(subtitles_filename('abc.unexpected_ext', 'en', 'vtt', 'ext'), 'abc.unexpected_ext.en.vtt')
    272 
    273     def test_remove_start(self):
    274         self.assertEqual(remove_start(None, 'A - '), None)
    275         self.assertEqual(remove_start('A - B', 'A - '), 'B')
    276         self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
    277 
    278     def test_remove_end(self):
    279         self.assertEqual(remove_end(None, ' - B'), None)
    280         self.assertEqual(remove_end('A - B', ' - B'), 'A')
    281         self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
    282 
    283     def test_remove_quotes(self):
    284         self.assertEqual(remove_quotes(None), None)
    285         self.assertEqual(remove_quotes('"'), '"')
    286         self.assertEqual(remove_quotes("'"), "'")
    287         self.assertEqual(remove_quotes(';'), ';')
    288         self.assertEqual(remove_quotes('";'), '";')
    289         self.assertEqual(remove_quotes('""'), '')
    290         self.assertEqual(remove_quotes('";"'), ';')
    291 
    292     def test_ordered_set(self):
    293         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
    294         self.assertEqual(orderedSet([]), [])
    295         self.assertEqual(orderedSet([1]), [1])
    296         # keep the list ordered
    297         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
    298 
    299     def test_unescape_html(self):
    300         self.assertEqual(unescapeHTML('%20;'), '%20;')
    301         self.assertEqual(unescapeHTML('&#x2F;'), '/')
    302         self.assertEqual(unescapeHTML('&#47;'), '/')
    303         self.assertEqual(unescapeHTML('&eacute;'), 'é')
    304         self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
    305         self.assertEqual(unescapeHTML('&a&quot;'), '&a"')
    306         # HTML5 entities
    307         self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')
    308 
    309     def test_date_from_str(self):
    310         self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
    311         self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
    312         self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
    313         self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
    314         self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
    315 
    316     def test_daterange(self):
    317         _20century = DateRange("19000101", "20000101")
    318         self.assertFalse("17890714" in _20century)
    319         _ac = DateRange("00010101")
    320         self.assertTrue("19690721" in _ac)
    321         _firstmilenium = DateRange(end="10000101")
    322         self.assertTrue("07110427" in _firstmilenium)
    323 
    324     def test_unified_dates(self):
    325         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
    326         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
    327         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
    328         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
    329         self.assertEqual(unified_strdate('1968 12 10'), '19681210')
    330         self.assertEqual(unified_strdate('1968-12-10'), '19681210')
    331         self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
    332         self.assertEqual(
    333             unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
    334             '20141126')
    335         self.assertEqual(
    336             unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
    337             '20150202')
    338         self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
    339         self.assertEqual(unified_strdate('25-09-2014'), '20140925')
    340         self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
    341         self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
    342         self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
    343         self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
    344         self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
    345         self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
    346         self.assertEqual(unified_strdate('November 3rd, 2019'), '20191103')
    347         self.assertEqual(unified_strdate('October 23rd, 2005'), '20051023')
    348 
    349     def test_unified_timestamps(self):
    350         self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
    351         self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
    352         self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
    353         self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
    354         self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
    355         self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
    356         self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
    357         self.assertEqual(
    358             unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
    359             1417001400)
    360         self.assertEqual(
    361             unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
    362             1422902860)
    363         self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
    364         self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
    365         self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
    366         self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
    367         self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
    368         self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
    369         self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
    370         self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
    371         self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
    372         self.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363)
    373 
    374     def test_determine_ext(self):
    375         self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
    376         self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
    377         self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
    378         self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
    379         self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
    380         self.assertEqual(determine_ext('foobar', None), None)
    381 
    382     def test_find_xpath_attr(self):
    383         testxml = '''<root>
    384             <node/>
    385             <node x="a"/>
    386             <node x="a" y="c" />
    387             <node x="b" y="d" />
    388             <node x="" />
    389         </root>'''
    390         doc = compat_etree_fromstring(testxml)
    391 
    392         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
    393         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
    394         self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
    395         self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
    396         self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
    397         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
    398         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
    399         self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
    400         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
    401         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
    402         self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
    403 
    404     def test_xpath_with_ns(self):
    405         testxml = '''<root xmlns:media="http://example.com/">
    406             <media:song>
    407                 <media:author>The Author</media:author>
    408                 <url>http://server.com/download.mp3</url>
    409             </media:song>
    410         </root>'''
    411         doc = compat_etree_fromstring(testxml)
    412         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
    413         self.assertTrue(find('media:song') is not None)
    414         self.assertEqual(find('media:song/media:author').text, 'The Author')
    415         self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
    416 
    417     def test_xpath_element(self):
    418         doc = xml.etree.ElementTree.Element('root')
    419         div = xml.etree.ElementTree.SubElement(doc, 'div')
    420         p = xml.etree.ElementTree.SubElement(div, 'p')
    421         p.text = 'Foo'
    422         self.assertEqual(xpath_element(doc, 'div/p'), p)
    423         self.assertEqual(xpath_element(doc, ['div/p']), p)
    424         self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
    425         self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
    426         self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
    427         self.assertTrue(xpath_element(doc, 'div/bar') is None)
    428         self.assertTrue(xpath_element(doc, ['div/bar']) is None)
    429         self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
    430         self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
    431         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
    432         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
    433 
    434     def test_xpath_text(self):
    435         testxml = '''<root>
    436             <div>
    437                 <p>Foo</p>
    438             </div>
    439         </root>'''
    440         doc = compat_etree_fromstring(testxml)
    441         self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
    442         self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
    443         self.assertTrue(xpath_text(doc, 'div/bar') is None)
    444         self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
    445 
    446     def test_xpath_attr(self):
    447         testxml = '''<root>
    448             <div>
    449                 <p x="a">Foo</p>
    450             </div>
    451         </root>'''
    452         doc = compat_etree_fromstring(testxml)
    453         self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
    454         self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
    455         self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
    456         self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
    457         self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
    458         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
    459         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
    460 
    461     def test_smuggle_url(self):
    462         data = {"ö": "ö", "abc": [3]}
    463         url = 'https://foo.bar/baz?x=y#a'
    464         smug_url = smuggle_url(url, data)
    465         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
    466         self.assertEqual(url, unsmug_url)
    467         self.assertEqual(data, unsmug_data)
    468 
    469         res_url, res_data = unsmuggle_url(url)
    470         self.assertEqual(res_url, url)
    471         self.assertEqual(res_data, None)
    472 
    473         smug_url = smuggle_url(url, {'a': 'b'})
    474         smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
    475         res_url, res_data = unsmuggle_url(smug_smug_url)
    476         self.assertEqual(res_url, url)
    477         self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
    478 
    479     def test_shell_quote(self):
    480         args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
    481         self.assertEqual(
    482             shell_quote(args),
    483             """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
    484 
    485     def test_float_or_none(self):
    486         self.assertEqual(float_or_none('42.42'), 42.42)
    487         self.assertEqual(float_or_none('42'), 42.0)
    488         self.assertEqual(float_or_none(''), None)
    489         self.assertEqual(float_or_none(None), None)
    490         self.assertEqual(float_or_none([]), None)
    491         self.assertEqual(float_or_none(set()), None)
    492 
    493     def test_int_or_none(self):
    494         self.assertEqual(int_or_none('42'), 42)
    495         self.assertEqual(int_or_none(''), None)
    496         self.assertEqual(int_or_none(None), None)
    497         self.assertEqual(int_or_none([]), None)
    498         self.assertEqual(int_or_none(set()), None)
    499 
    500     def test_str_to_int(self):
    501         self.assertEqual(str_to_int('123,456'), 123456)
    502         self.assertEqual(str_to_int('123.456'), 123456)
    503         self.assertEqual(str_to_int(523), 523)
    504         # Python 3 has no long
    505         if sys.version_info < (3, 0):
    506             eval('self.assertEqual(str_to_int(123456L), 123456)')
    507         self.assertEqual(str_to_int('noninteger'), None)
    508         self.assertEqual(str_to_int([]), None)
    509 
    510     def test_url_basename(self):
    511         self.assertEqual(url_basename('http://foo.de/'), '')
    512         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
    513         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
    514         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
    515         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
    516         self.assertEqual(
    517             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
    518             'trailer.mp4')
    519 
    520     def test_base_url(self):
    521         self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
    522         self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
    523         self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
    524         self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
    525         self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
    526 
    527     def test_urljoin(self):
    528         self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
    529         self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
    530         self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
    531         self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
    532         self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
    533         self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
    534         self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
    535         self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
    536         self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
    537         self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
    538         self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
    539         self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
    540         self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
    541         self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
    542         self.assertEqual(urljoin('http://foo.de/', None), None)
    543         self.assertEqual(urljoin('http://foo.de/', ''), None)
    544         self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
    545         self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
    546         self.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de')
    547         self.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de')
    548 
    549     def test_url_or_none(self):
    550         self.assertEqual(url_or_none(None), None)
    551         self.assertEqual(url_or_none(''), None)
    552         self.assertEqual(url_or_none('foo'), None)
    553         self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
    554         self.assertEqual(url_or_none('https://foo.de'), 'https://foo.de')
    555         self.assertEqual(url_or_none('http$://foo.de'), None)
    556         self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
    557         self.assertEqual(url_or_none('//foo.de'), '//foo.de')
    558         self.assertEqual(url_or_none('s3://foo.de'), None)
    559         self.assertEqual(url_or_none('rtmpte://foo.de'), 'rtmpte://foo.de')
    560         self.assertEqual(url_or_none('mms://foo.de'), 'mms://foo.de')
    561         self.assertEqual(url_or_none('rtspu://foo.de'), 'rtspu://foo.de')
    562         self.assertEqual(url_or_none('ftps://foo.de'), 'ftps://foo.de')
    563 
    564     def test_parse_age_limit(self):
    565         self.assertEqual(parse_age_limit(None), None)
    566         self.assertEqual(parse_age_limit(False), None)
    567         self.assertEqual(parse_age_limit('invalid'), None)
    568         self.assertEqual(parse_age_limit(0), 0)
    569         self.assertEqual(parse_age_limit(18), 18)
    570         self.assertEqual(parse_age_limit(21), 21)
    571         self.assertEqual(parse_age_limit(22), None)
    572         self.assertEqual(parse_age_limit('18'), 18)
    573         self.assertEqual(parse_age_limit('18+'), 18)
    574         self.assertEqual(parse_age_limit('PG-13'), 13)
    575         self.assertEqual(parse_age_limit('TV-14'), 14)
    576         self.assertEqual(parse_age_limit('TV-MA'), 17)
    577         self.assertEqual(parse_age_limit('TV14'), 14)
    578         self.assertEqual(parse_age_limit('TV_G'), 0)
    579 
    580     def test_parse_duration(self):
    581         self.assertEqual(parse_duration(None), None)
    582         self.assertEqual(parse_duration(False), None)
    583         self.assertEqual(parse_duration('invalid'), None)
    584         self.assertEqual(parse_duration('1'), 1)
    585         self.assertEqual(parse_duration('1337:12'), 80232)
    586         self.assertEqual(parse_duration('9:12:43'), 33163)
    587         self.assertEqual(parse_duration('12:00'), 720)
    588         self.assertEqual(parse_duration('00:01:01'), 61)
    589         self.assertEqual(parse_duration('x:y'), None)
    590         self.assertEqual(parse_duration('3h11m53s'), 11513)
    591         self.assertEqual(parse_duration('3h 11m 53s'), 11513)
    592         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
    593         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
    594         self.assertEqual(parse_duration('62m45s'), 3765)
    595         self.assertEqual(parse_duration('6m59s'), 419)
    596         self.assertEqual(parse_duration('49s'), 49)
    597         self.assertEqual(parse_duration('0h0m0s'), 0)
    598         self.assertEqual(parse_duration('0m0s'), 0)
    599         self.assertEqual(parse_duration('0s'), 0)
    600         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
    601         self.assertEqual(parse_duration('T30M38S'), 1838)
    602         self.assertEqual(parse_duration('5 s'), 5)
    603         self.assertEqual(parse_duration('3 min'), 180)
    604         self.assertEqual(parse_duration('2.5 hours'), 9000)
    605         self.assertEqual(parse_duration('02:03:04'), 7384)
    606         self.assertEqual(parse_duration('01:02:03:04'), 93784)
    607         self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
    608         self.assertEqual(parse_duration('87 Min.'), 5220)
    609         self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
    610         self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
    611         self.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)
    612 
    613     def test_fix_xml_ampersands(self):
    614         self.assertEqual(
    615             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
    616         self.assertEqual(
    617             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
    618             '"&amp;x=y&amp;wrong;&amp;z=a')
    619         self.assertEqual(
    620             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
    621             '&amp;&apos;&gt;&lt;&quot;')
    622         self.assertEqual(
    623             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
    624         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
    625 
    626     def test_paged_list(self):
    627         def testPL(size, pagesize, sliceargs, expected):
    628             def get_page(pagenum):
    629                 firstid = pagenum * pagesize
    630                 upto = min(size, pagenum * pagesize + pagesize)
    631                 for i in range(firstid, upto):
    632                     yield i
    633 
    634             pl = OnDemandPagedList(get_page, pagesize)
    635             got = pl.getslice(*sliceargs)
    636             self.assertEqual(got, expected)
    637 
    638             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
    639             got = iapl.getslice(*sliceargs)
    640             self.assertEqual(got, expected)
    641 
    642         testPL(5, 2, (), [0, 1, 2, 3, 4])
    643         testPL(5, 2, (1,), [1, 2, 3, 4])
    644         testPL(5, 2, (2,), [2, 3, 4])
    645         testPL(5, 2, (4,), [4])
    646         testPL(5, 2, (0, 3), [0, 1, 2])
    647         testPL(5, 2, (1, 4), [1, 2, 3])
    648         testPL(5, 2, (2, 99), [2, 3, 4])
    649         testPL(5, 2, (20, 99), [])
    650 
    651     def test_read_batch_urls(self):
    652         f = io.StringIO('''\xef\xbb\xbf foo
    653             bar\r
    654             baz
    655             # More after this line\r
    656             ; or after this
    657             bam''')
    658         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
    659 
    660     def test_urlencode_postdata(self):
    661         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
    662         self.assertTrue(isinstance(data, bytes))
    663 
    664     def test_update_url_query(self):
    665         def query_dict(url):
    666             return compat_parse_qs(compat_urlparse.urlparse(url).query)
    667         self.assertEqual(query_dict(update_url_query(
    668             'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
    669             query_dict('http://example.com/path?quality=HD&format=mp4'))
    670         self.assertEqual(query_dict(update_url_query(
    671             'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
    672             query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
    673         self.assertEqual(query_dict(update_url_query(
    674             'http://example.com/path', {'fields': 'id,formats,subtitles'})),
    675             query_dict('http://example.com/path?fields=id,formats,subtitles'))
    676         self.assertEqual(query_dict(update_url_query(
    677             'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
    678             query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
    679         self.assertEqual(query_dict(update_url_query(
    680             'http://example.com/path?manifest=f4m', {'manifest': []})),
    681             query_dict('http://example.com/path'))
    682         self.assertEqual(query_dict(update_url_query(
    683             'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
    684             query_dict('http://example.com/path?system=LINUX'))
    685         self.assertEqual(query_dict(update_url_query(
    686             'http://example.com/path', {'fields': b'id,formats,subtitles'})),
    687             query_dict('http://example.com/path?fields=id,formats,subtitles'))
    688         self.assertEqual(query_dict(update_url_query(
    689             'http://example.com/path', {'width': 1080, 'height': 720})),
    690             query_dict('http://example.com/path?width=1080&height=720'))
    691         self.assertEqual(query_dict(update_url_query(
    692             'http://example.com/path', {'bitrate': 5020.43})),
    693             query_dict('http://example.com/path?bitrate=5020.43'))
    694         self.assertEqual(query_dict(update_url_query(
    695             'http://example.com/path', {'test': '第二行тест'})),
    696             query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
    697 
    698     def test_multipart_encode(self):
    699         self.assertEqual(
    700             multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0],
    701             b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
    702         self.assertEqual(
    703             multipart_encode({'欄位'.encode('utf-8'): '值'.encode('utf-8')}, boundary='AAAAAA')[0],
    704             b'--AAAAAA\r\nContent-Disposition: form-data; name="\xe6\xac\x84\xe4\xbd\x8d"\r\n\r\n\xe5\x80\xbc\r\n--AAAAAA--\r\n')
    705         self.assertRaises(
    706             ValueError, multipart_encode, {b'field': b'value'}, boundary='value')
    707 
    708     def test_dict_get(self):
    709         FALSE_VALUES = {
    710             'none': None,
    711             'false': False,
    712             'zero': 0,
    713             'empty_string': '',
    714             'empty_list': [],
    715         }
    716         d = FALSE_VALUES.copy()
    717         d['a'] = 42
    718         self.assertEqual(dict_get(d, 'a'), 42)
    719         self.assertEqual(dict_get(d, 'b'), None)
    720         self.assertEqual(dict_get(d, 'b', 42), 42)
    721         self.assertEqual(dict_get(d, ('a', )), 42)
    722         self.assertEqual(dict_get(d, ('b', 'a', )), 42)
    723         self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
    724         self.assertEqual(dict_get(d, ('b', 'c', )), None)
    725         self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
    726         for key, false_value in FALSE_VALUES.items():
    727             self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
    728             self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
    729 
    730     def test_merge_dicts(self):
    731         self.assertEqual(merge_dicts({'a': 1}, {'b': 2}), {'a': 1, 'b': 2})
    732         self.assertEqual(merge_dicts({'a': 1}, {'a': 2}), {'a': 1})
    733         self.assertEqual(merge_dicts({'a': 1}, {'a': None}), {'a': 1})
    734         self.assertEqual(merge_dicts({'a': 1}, {'a': ''}), {'a': 1})
    735         self.assertEqual(merge_dicts({'a': 1}, {}), {'a': 1})
    736         self.assertEqual(merge_dicts({'a': None}, {'a': 1}), {'a': 1})
    737         self.assertEqual(merge_dicts({'a': ''}, {'a': 1}), {'a': ''})
    738         self.assertEqual(merge_dicts({'a': ''}, {'a': 'abc'}), {'a': 'abc'})
    739         self.assertEqual(merge_dicts({'a': None}, {'a': ''}, {'a': 'abc'}), {'a': 'abc'})
    740 
    741     def test_encode_compat_str(self):
    742         self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
    743         self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
    744 
    745     def test_parse_iso8601(self):
    746         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
    747         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
    748         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
    749         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
    750         self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
    751         self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
    752 
    753     def test_strip_jsonp(self):
    754         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
    755         d = json.loads(stripped)
    756         self.assertEqual(d, [{"id": "532cb", "x": 3}])
    757 
    758         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
    759         d = json.loads(stripped)
    760         self.assertEqual(d, {'STATUS': 'OK'})
    761 
    762         stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
    763         d = json.loads(stripped)
    764         self.assertEqual(d, {'status': 'success'})
    765 
    766         stripped = strip_jsonp('window.cb && window.cb({"status": "success"});')
    767         d = json.loads(stripped)
    768         self.assertEqual(d, {'status': 'success'})
    769 
    770         stripped = strip_jsonp('window.cb && cb({"status": "success"});')
    771         d = json.loads(stripped)
    772         self.assertEqual(d, {'status': 'success'})
    773 
    774         stripped = strip_jsonp('({"status": "success"});')
    775         d = json.loads(stripped)
    776         self.assertEqual(d, {'status': 'success'})
    777 
    778     def test_strip_or_none(self):
    779         self.assertEqual(strip_or_none(' abc'), 'abc')
    780         self.assertEqual(strip_or_none('abc '), 'abc')
    781         self.assertEqual(strip_or_none(' abc '), 'abc')
    782         self.assertEqual(strip_or_none('\tabc\t'), 'abc')
    783         self.assertEqual(strip_or_none('\n\tabc\n\t'), 'abc')
    784         self.assertEqual(strip_or_none('abc'), 'abc')
    785         self.assertEqual(strip_or_none(''), '')
    786         self.assertEqual(strip_or_none(None), None)
    787         self.assertEqual(strip_or_none(42), None)
    788         self.assertEqual(strip_or_none([]), None)
    789 
    790     def test_uppercase_escape(self):
    791         self.assertEqual(uppercase_escape('aä'), 'aä')
    792         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
    793 
    794     def test_lowercase_escape(self):
    795         self.assertEqual(lowercase_escape('aä'), 'aä')
    796         self.assertEqual(lowercase_escape('\\u0026'), '&')
    797 
    798     def test_limit_length(self):
    799         self.assertEqual(limit_length(None, 12), None)
    800         self.assertEqual(limit_length('foo', 12), 'foo')
    801         self.assertTrue(
    802             limit_length('foo bar baz asd', 12).startswith('foo bar'))
    803         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
    804 
    805     def test_mimetype2ext(self):
    806         self.assertEqual(mimetype2ext(None), None)
    807         self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
    808         self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
    809         self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
    810         self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
    811         self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
    812         self.assertEqual(mimetype2ext('audio/x-wav'), 'wav')
    813         self.assertEqual(mimetype2ext('audio/x-wav;codec=pcm'), 'wav')
    814 
    815     def test_month_by_name(self):
    816         self.assertEqual(month_by_name(None), None)
    817         self.assertEqual(month_by_name('December', 'en'), 12)
    818         self.assertEqual(month_by_name('décembre', 'fr'), 12)
    819         self.assertEqual(month_by_name('December'), 12)
    820         self.assertEqual(month_by_name('décembre'), None)
    821         self.assertEqual(month_by_name('Unknown', 'unknown'), None)
    822 
    823     def test_parse_codecs(self):
    824         self.assertEqual(parse_codecs(''), {})
    825         self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
    826             'vcodec': 'avc1.77.30',
    827             'acodec': 'mp4a.40.2',
    828         })
    829         self.assertEqual(parse_codecs('mp4a.40.2'), {
    830             'vcodec': 'none',
    831             'acodec': 'mp4a.40.2',
    832         })
    833         self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
    834             'vcodec': 'avc1.42001e',
    835             'acodec': 'mp4a.40.5',
    836         })
    837         self.assertEqual(parse_codecs('avc3.640028'), {
    838             'vcodec': 'avc3.640028',
    839             'acodec': 'none',
    840         })
    841         self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
    842             'vcodec': 'h264',
    843             'acodec': 'aac',
    844         })
    845         self.assertEqual(parse_codecs('av01.0.05M.08'), {
    846             'vcodec': 'av01.0.05M.08',
    847             'acodec': 'none',
    848         })
    849         self.assertEqual(parse_codecs('theora, vorbis'), {
    850             'vcodec': 'theora',
    851             'acodec': 'vorbis',
    852         })
    853         self.assertEqual(parse_codecs('unknownvcodec, unknownacodec'), {
    854             'vcodec': 'unknownvcodec',
    855             'acodec': 'unknownacodec',
    856         })
    857         self.assertEqual(parse_codecs('unknown'), {})
    858 
    859     def test_escape_rfc3986(self):
    860         reserved = "!*'();:@&=+$,/?#[]"
    861         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
    862         self.assertEqual(escape_rfc3986(reserved), reserved)
    863         self.assertEqual(escape_rfc3986(unreserved), unreserved)
    864         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
    865         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
    866         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
    867         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
    868 
    869     def test_escape_url(self):
    870         self.assertEqual(
    871             escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
    872             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
    873         )
    874         self.assertEqual(
    875             escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
    876             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
    877         )
    878         self.assertEqual(
    879             escape_url('http://тест.рф/фрагмент'),
    880             'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
    881         )
    882         self.assertEqual(
    883             escape_url('http://тест.рф/абв?абв=абв#абв'),
    884             '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'
    885         )
    886         self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
    887 
    888     def test_js_to_json_realworld(self):
    889         inp = '''{
    890             'clip':{'provider':'pseudo'}
    891         }'''
    892         self.assertEqual(js_to_json(inp), '''{
    893             "clip":{"provider":"pseudo"}
    894         }''')
    895         json.loads(js_to_json(inp))
    896 
    897         inp = '''{
    898             'playlist':[{'controls':{'all':null}}]
    899         }'''
    900         self.assertEqual(js_to_json(inp), '''{
    901             "playlist":[{"controls":{"all":null}}]
    902         }''')
    903 
    904         inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
    905         self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
    906 
    907         inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
    908         json_code = js_to_json(inp)
    909         self.assertEqual(json.loads(json_code), json.loads(inp))
    910 
    911         inp = '''{
    912             0:{src:'skipped', type: 'application/dash+xml'},
    913             1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
    914         }'''
    915         self.assertEqual(js_to_json(inp), '''{
    916             "0":{"src":"skipped", "type": "application/dash+xml"},
    917             "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
    918         }''')
    919 
    920         inp = '''{"foo":101}'''
    921         self.assertEqual(js_to_json(inp), '''{"foo":101}''')
    922 
    923         inp = '''{"duration": "00:01:07"}'''
    924         self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
    925 
    926         inp = '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}'''
    927         self.assertEqual(js_to_json(inp), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''')
    928 
    929     def test_js_to_json_edgecases(self):
    930         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
    931         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
    932 
    933         on = js_to_json('{"abc": true}')
    934         self.assertEqual(json.loads(on), {'abc': True})
    935 
    936         # Ignore JavaScript code as well
    937         on = js_to_json('''{
    938             "x": 1,
    939             y: "a",
    940             z: some.code
    941         }''')
    942         d = json.loads(on)
    943         self.assertEqual(d['x'], 1)
    944         self.assertEqual(d['y'], 'a')
    945 
    946         # Just drop ! prefix for now though this results in a wrong value
    947         on = js_to_json('''{
    948             a: !0,
    949             b: !1,
    950             c: !!0,
    951             d: !!42.42,
    952             e: !!![],
    953             f: !"abc",
    954             g: !"",
    955             !42: 42
    956         }''')
    957         self.assertEqual(json.loads(on), {
    958             'a': 0,
    959             'b': 1,
    960             'c': 0,
    961             'd': 42.42,
    962             'e': [],
    963             'f': "abc",
    964             'g': "",
    965             '42': 42
    966         })
    967 
    968         on = js_to_json('["abc", "def",]')
    969         self.assertEqual(json.loads(on), ['abc', 'def'])
    970 
    971         on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
    972         self.assertEqual(json.loads(on), ['abc', 'def'])
    973 
    974         on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
    975         self.assertEqual(json.loads(on), ['abc', 'def'])
    976 
    977         on = js_to_json('{"abc": "def",}')
    978         self.assertEqual(json.loads(on), {'abc': 'def'})
    979 
    980         on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
    981         self.assertEqual(json.loads(on), {'abc': 'def'})
    982 
    983         on = js_to_json('{ 0: /* " \n */ ",]" , }')
    984         self.assertEqual(json.loads(on), {'0': ',]'})
    985 
    986         on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
    987         self.assertEqual(json.loads(on), {'0': ',]'})
    988 
    989         on = js_to_json('{ 0: // comment\n1 }')
    990         self.assertEqual(json.loads(on), {'0': 1})
    991 
    992         on = js_to_json(r'["<p>x<\/p>"]')
    993         self.assertEqual(json.loads(on), ['<p>x</p>'])
    994 
    995         on = js_to_json(r'["\xaa"]')
    996         self.assertEqual(json.loads(on), ['\u00aa'])
    997 
    998         on = js_to_json("['a\\\nb']")
    999         self.assertEqual(json.loads(on), ['ab'])
   1000 
   1001         on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
   1002         self.assertEqual(json.loads(on), ['ab'])
   1003 
   1004         on = js_to_json('{0xff:0xff}')
   1005         self.assertEqual(json.loads(on), {'255': 255})
   1006 
   1007         on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
   1008         self.assertEqual(json.loads(on), {'255': 255})
   1009 
   1010         on = js_to_json('{077:077}')
   1011         self.assertEqual(json.loads(on), {'63': 63})
   1012 
   1013         on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
   1014         self.assertEqual(json.loads(on), {'63': 63})
   1015 
   1016         on = js_to_json('{42:42}')
   1017         self.assertEqual(json.loads(on), {'42': 42})
   1018 
   1019         on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
   1020         self.assertEqual(json.loads(on), {'42': 42})
   1021 
   1022         on = js_to_json('{42:4.2e1}')
   1023         self.assertEqual(json.loads(on), {'42': 42.0})
   1024 
   1025         on = js_to_json('{ "0x40": "0x40" }')
   1026         self.assertEqual(json.loads(on), {'0x40': '0x40'})
   1027 
   1028         on = js_to_json('{ "040": "040" }')
   1029         self.assertEqual(json.loads(on), {'040': '040'})
   1030 
   1031     def test_js_to_json_malformed(self):
   1032         self.assertEqual(js_to_json('42a1'), '42"a1"')
   1033         self.assertEqual(js_to_json('42a-1'), '42"a"-1')
   1034 
   1035     def test_extract_attributes(self):
   1036         self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
   1037         self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
   1038         self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
   1039         self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
   1040         self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
   1041         self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
   1042         self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
   1043         self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'})  # XML
   1044         self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
   1045         self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'})  # HTML 3.2
   1046         self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'})  # HTML 4.0
   1047         self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
   1048         self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
   1049         self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
   1050         self.assertEqual(extract_attributes('<e x >'), {'x': None})
   1051         self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
   1052         self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
   1053         self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
   1054         self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
   1055         self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
   1056         self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
   1057         self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
   1058         self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'})  # Names lowercased
   1059         self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
   1060         self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
   1061         self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
   1062         self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
   1063         self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
   1064         # "Narrow" Python builds don't support unicode code points outside BMP.
   1065         try:
   1066             compat_chr(0x10000)
   1067             supports_outside_bmp = True
   1068         except ValueError:
   1069             supports_outside_bmp = False
   1070         if supports_outside_bmp:
   1071             self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
   1072         # Malformed HTML should not break attributes extraction on older Python
   1073         self.assertEqual(extract_attributes('<mal"formed/>'), {})
   1074 
   1075     def test_clean_html(self):
   1076         self.assertEqual(clean_html('a:\nb'), 'a: b')
   1077         self.assertEqual(clean_html('a:\n   "b"'), 'a:    "b"')
   1078         self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
   1079 
   1080     def test_intlist_to_bytes(self):
   1081         self.assertEqual(
   1082             intlist_to_bytes([0, 1, 127, 128, 255]),
   1083             b'\x00\x01\x7f\x80\xff')
   1084 
   1085     def test_args_to_str(self):
   1086         self.assertEqual(
   1087             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
   1088             'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""'
   1089         )
   1090 
   1091     def test_parse_filesize(self):
   1092         self.assertEqual(parse_filesize(None), None)
   1093         self.assertEqual(parse_filesize(''), None)
   1094         self.assertEqual(parse_filesize('91 B'), 91)
   1095         self.assertEqual(parse_filesize('foobar'), None)
   1096         self.assertEqual(parse_filesize('2 MiB'), 2097152)
   1097         self.assertEqual(parse_filesize('5 GB'), 5000000000)
   1098         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
   1099         self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
   1100         self.assertEqual(parse_filesize('1,24 KB'), 1240)
   1101         self.assertEqual(parse_filesize('1,24 kb'), 1240)
   1102         self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
   1103 
   1104     def test_parse_count(self):
   1105         self.assertEqual(parse_count(None), None)
   1106         self.assertEqual(parse_count(''), None)
   1107         self.assertEqual(parse_count('0'), 0)
   1108         self.assertEqual(parse_count('1000'), 1000)
   1109         self.assertEqual(parse_count('1.000'), 1000)
   1110         self.assertEqual(parse_count('1.1k'), 1100)
   1111         self.assertEqual(parse_count('1.1kk'), 1100000)
   1112         self.assertEqual(parse_count('1.1kk '), 1100000)
   1113         self.assertEqual(parse_count('1.1kk views'), 1100000)
   1114 
   1115     def test_parse_resolution(self):
   1116         self.assertEqual(parse_resolution(None), {})
   1117         self.assertEqual(parse_resolution(''), {})
   1118         self.assertEqual(parse_resolution('1920x1080'), {'width': 1920, 'height': 1080})
   1119         self.assertEqual(parse_resolution('1920×1080'), {'width': 1920, 'height': 1080})
   1120         self.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080})
   1121         self.assertEqual(parse_resolution('720p'), {'height': 720})
   1122         self.assertEqual(parse_resolution('4k'), {'height': 2160})
   1123         self.assertEqual(parse_resolution('8K'), {'height': 4320})
   1124 
   1125     def test_parse_bitrate(self):
   1126         self.assertEqual(parse_bitrate(None), None)
   1127         self.assertEqual(parse_bitrate(''), None)
   1128         self.assertEqual(parse_bitrate('300kbps'), 300)
   1129         self.assertEqual(parse_bitrate('1500kbps'), 1500)
   1130         self.assertEqual(parse_bitrate('300 kbps'), 300)
   1131 
   1132     def test_version_tuple(self):
   1133         self.assertEqual(version_tuple('1'), (1,))
   1134         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
   1135         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
   1136 
   1137     def test_detect_exe_version(self):
   1138         self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
   1139 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
   1140 configuration: --prefix=/usr --extra-'''), '1.2.1')
   1141         self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
   1142 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
   1143         self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
   1144 Trying to open render node...
   1145 Success at /dev/dri/renderD128.
   1146 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
   1147 
   1148     def test_age_restricted(self):
   1149         self.assertFalse(age_restricted(None, 10))  # unrestricted content
   1150         self.assertFalse(age_restricted(1, None))  # unrestricted policy
   1151         self.assertFalse(age_restricted(8, 10))
   1152         self.assertTrue(age_restricted(18, 14))
   1153         self.assertFalse(age_restricted(18, 18))
   1154 
   1155     def test_is_html(self):
   1156         self.assertFalse(is_html(b'\x49\x44\x43<html'))
   1157         self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
   1158         self.assertTrue(is_html(  # UTF-8 with BOM
   1159             b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
   1160         self.assertTrue(is_html(  # UTF-16-LE
   1161             b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
   1162         ))
   1163         self.assertTrue(is_html(  # UTF-16-BE
   1164             b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
   1165         ))
   1166         self.assertTrue(is_html(  # UTF-32-BE
   1167             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'))
   1168         self.assertTrue(is_html(  # UTF-32-LE
   1169             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'))
   1170 
   1171     def test_render_table(self):
   1172         self.assertEqual(
   1173             render_table(
   1174                 ['a', 'bcd'],
   1175                 [[123, 4], [9999, 51]]),
   1176             'a    bcd\n'
   1177             '123  4\n'
   1178             '9999 51')
   1179 
   1180     def test_match_str(self):
   1181         self.assertRaises(ValueError, match_str, 'xy>foobar', {})
   1182         self.assertFalse(match_str('xy', {'x': 1200}))
   1183         self.assertTrue(match_str('!xy', {'x': 1200}))
   1184         self.assertTrue(match_str('x', {'x': 1200}))
   1185         self.assertFalse(match_str('!x', {'x': 1200}))
   1186         self.assertTrue(match_str('x', {'x': 0}))
   1187         self.assertFalse(match_str('x>0', {'x': 0}))
   1188         self.assertFalse(match_str('x>0', {}))
   1189         self.assertTrue(match_str('x>?0', {}))
   1190         self.assertTrue(match_str('x>1K', {'x': 1200}))
   1191         self.assertFalse(match_str('x>2K', {'x': 1200}))
   1192         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
   1193         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
   1194         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
   1195         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
   1196         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
   1197         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
   1198         self.assertFalse(match_str(
   1199             'like_count > 100 & dislike_count <? 50 & description',
   1200             {'like_count': 90, 'description': 'foo'}))
   1201         self.assertTrue(match_str(
   1202             'like_count > 100 & dislike_count <? 50 & description',
   1203             {'like_count': 190, 'description': 'foo'}))
   1204         self.assertFalse(match_str(
   1205             'like_count > 100 & dislike_count <? 50 & description',
   1206             {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
   1207         self.assertFalse(match_str(
   1208             'like_count > 100 & dislike_count <? 50 & description',
   1209             {'like_count': 190, 'dislike_count': 10}))
   1210         self.assertTrue(match_str('is_live', {'is_live': True}))
   1211         self.assertFalse(match_str('is_live', {'is_live': False}))
   1212         self.assertFalse(match_str('is_live', {'is_live': None}))
   1213         self.assertFalse(match_str('is_live', {}))
   1214         self.assertFalse(match_str('!is_live', {'is_live': True}))
   1215         self.assertTrue(match_str('!is_live', {'is_live': False}))
   1216         self.assertTrue(match_str('!is_live', {'is_live': None}))
   1217         self.assertTrue(match_str('!is_live', {}))
   1218         self.assertTrue(match_str('title', {'title': 'abc'}))
   1219         self.assertTrue(match_str('title', {'title': ''}))
   1220         self.assertFalse(match_str('!title', {'title': 'abc'}))
   1221         self.assertFalse(match_str('!title', {'title': ''}))
   1222 
   1223     def test_parse_dfxp_time_expr(self):
   1224         self.assertEqual(parse_dfxp_time_expr(None), None)
   1225         self.assertEqual(parse_dfxp_time_expr(''), None)
   1226         self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
   1227         self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
   1228         self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
   1229         self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
   1230         self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
   1231 
   1232     def test_dfxp2srt(self):
   1233         dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
   1234             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
   1235             <body>
   1236                 <div xml:lang="en">
   1237                     <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
   1238                     <p begin="1" end="2">第二行<br/>♪♪</p>
   1239                     <p begin="2" dur="1"><span>Third<br/>Line</span></p>
   1240                     <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
   1241                     <p begin="-1" end="-1">Ignore, two</p>
   1242                     <p begin="3" dur="-1">Ignored, three</p>
   1243                 </div>
   1244             </body>
   1245             </tt>'''.encode('utf-8')
   1246         srt_data = '''1
   1247 00:00:00,000 --> 00:00:01,000
   1248 The following line contains Chinese characters and special symbols
   1249 
   1250 2
   1251 00:00:01,000 --> 00:00:02,000
   1252 第二行
   1253 ♪♪
   1254 
   1255 3
   1256 00:00:02,000 --> 00:00:03,000
   1257 Third
   1258 Line
   1259 
   1260 '''
   1261         self.assertEqual(dfxp2srt(dfxp_data), srt_data)
   1262 
   1263         dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
   1264             <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
   1265             <body>
   1266                 <div xml:lang="en">
   1267                     <p begin="0" end="1">The first line</p>
   1268                 </div>
   1269             </body>
   1270             </tt>'''.encode('utf-8')
   1271         srt_data = '''1
   1272 00:00:00,000 --> 00:00:01,000
   1273 The first line
   1274 
   1275 '''
   1276         self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
   1277 
   1278         dfxp_data_with_style = '''<?xml version="1.0" encoding="utf-8"?>
   1279 <tt xmlns="http://www.w3.org/2006/10/ttaf1" xmlns:ttp="http://www.w3.org/2006/10/ttaf1#parameter" ttp:timeBase="media" xmlns:tts="http://www.w3.org/2006/10/ttaf1#style" xml:lang="en" xmlns:ttm="http://www.w3.org/2006/10/ttaf1#metadata">
   1280   <head>
   1281     <styling>
   1282       <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
   1283       <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
   1284       <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
   1285       <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
   1286     </styling>
   1287   </head>
   1288   <body tts:textAlign="center" style="s0">
   1289     <div>
   1290       <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
   1291       <p style="s2" begin="00:00:02.08" id="p0" end="00:00:05.84"><span tts:color="lime">part 1<br /></span><span tts:color="cyan">part 2</span></p>
   1292       <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
   1293       <p style="s1" tts:textDecoration="underline" begin="00:00:09.56" id="p2" end="00:00:12.36"><span style="s2" tts:color="lime">inner<br /> </span>style</p>
   1294     </div>
   1295   </body>
   1296 </tt>'''.encode('utf-8')
   1297         srt_data = '''1
   1298 00:00:02,080 --> 00:00:05,839
   1299 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
   1300 
   1301 2
   1302 00:00:02,080 --> 00:00:05,839
   1303 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
   1304 </font>part 2</font></b>
   1305 
   1306 3
   1307 00:00:05,839 --> 00:00:09,560
   1308 <u><font color="lime">line 3
   1309 part 3</font></u>
   1310 
   1311 4
   1312 00:00:09,560 --> 00:00:12,359
   1313 <i><u><font color="yellow"><font color="lime">inner
   1314  </font>style</font></u></i>
   1315 
   1316 '''
   1317         self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)
   1318 
   1319         dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?>
   1320             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
   1321             <body>
   1322                 <div xml:lang="en">
   1323                     <p begin="0" end="1">Line 1</p>
   1324                     <p begin="1" end="2">第二行</p>
   1325                 </div>
   1326             </body>
   1327             </tt>'''.encode('utf-16')
   1328         srt_data = '''1
   1329 00:00:00,000 --> 00:00:01,000
   1330 Line 1
   1331 
   1332 2
   1333 00:00:01,000 --> 00:00:02,000
   1334 第二行
   1335 
   1336 '''
   1337         self.assertEqual(dfxp2srt(dfxp_data_non_utf8), srt_data)
   1338 
   1339     def test_cli_option(self):
   1340         self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
   1341         self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
   1342         self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
   1343         self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
   1344 
   1345     def test_cli_valueless_option(self):
   1346         self.assertEqual(cli_valueless_option(
   1347             {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
   1348         self.assertEqual(cli_valueless_option(
   1349             {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
   1350         self.assertEqual(cli_valueless_option(
   1351             {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
   1352         self.assertEqual(cli_valueless_option(
   1353             {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
   1354         self.assertEqual(cli_valueless_option(
   1355             {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
   1356         self.assertEqual(cli_valueless_option(
   1357             {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
   1358 
   1359     def test_cli_bool_option(self):
   1360         self.assertEqual(
   1361             cli_bool_option(
   1362                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
   1363             ['--no-check-certificate', 'true'])
   1364         self.assertEqual(
   1365             cli_bool_option(
   1366                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
   1367             ['--no-check-certificate=true'])
   1368         self.assertEqual(
   1369             cli_bool_option(
   1370                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
   1371             ['--check-certificate', 'false'])
   1372         self.assertEqual(
   1373             cli_bool_option(
   1374                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
   1375             ['--check-certificate=false'])
   1376         self.assertEqual(
   1377             cli_bool_option(
   1378                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
   1379             ['--check-certificate', 'true'])
   1380         self.assertEqual(
   1381             cli_bool_option(
   1382                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
   1383             ['--check-certificate=true'])
   1384         self.assertEqual(
   1385             cli_bool_option(
   1386                 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
   1387             [])
   1388 
   1389     def test_ohdave_rsa_encrypt(self):
   1390         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
   1391         e = 65537
   1392 
   1393         self.assertEqual(
   1394             ohdave_rsa_encrypt(b'aa111222', e, N),
   1395             '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
   1396 
   1397     def test_pkcs1pad(self):
   1398         data = [1, 2, 3]
   1399         padded_data = pkcs1pad(data, 32)
   1400         self.assertEqual(padded_data[:2], [0, 2])
   1401         self.assertEqual(padded_data[28:], [0, 1, 2, 3])
   1402 
   1403         self.assertRaises(ValueError, pkcs1pad, data, 8)
   1404 
   1405     def test_encode_base_n(self):
   1406         self.assertEqual(encode_base_n(0, 30), '0')
   1407         self.assertEqual(encode_base_n(80, 30), '2k')
   1408 
   1409         custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
   1410         self.assertEqual(encode_base_n(0, 30, custom_table), '9')
   1411         self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
   1412 
   1413         self.assertRaises(ValueError, encode_base_n, 0, 70)
   1414         self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
   1415 
   1416     def test_caesar(self):
   1417         self.assertEqual(caesar('ace', 'abcdef', 2), 'cea')
   1418         self.assertEqual(caesar('cea', 'abcdef', -2), 'ace')
   1419         self.assertEqual(caesar('ace', 'abcdef', -2), 'eac')
   1420         self.assertEqual(caesar('eac', 'abcdef', 2), 'ace')
   1421         self.assertEqual(caesar('ace', 'abcdef', 0), 'ace')
   1422         self.assertEqual(caesar('xyz', 'abcdef', 2), 'xyz')
   1423         self.assertEqual(caesar('abc', 'acegik', 2), 'ebg')
   1424         self.assertEqual(caesar('ebg', 'acegik', -2), 'abc')
   1425 
   1426     def test_rot47(self):
   1427         self.assertEqual(rot47('youtube-dl'), r'J@FEF36\5=')
   1428         self.assertEqual(rot47('YOUTUBE-DL'), r'*~&%&qt\s{')
   1429 
   1430     def test_urshift(self):
   1431         self.assertEqual(urshift(3, 1), 1)
   1432         self.assertEqual(urshift(-3, 1), 2147483646)
   1433 
   1434     def test_get_element_by_class(self):
   1435         html = '''
   1436             <span class="foo bar">nice</span>
   1437         '''
   1438 
   1439         self.assertEqual(get_element_by_class('foo', html), 'nice')
   1440         self.assertEqual(get_element_by_class('no-such-class', html), None)
   1441 
   1442     def test_get_element_by_attribute(self):
   1443         html = '''
   1444             <span class="foo bar">nice</span>
   1445         '''
   1446 
   1447         self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
   1448         self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
   1449         self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
   1450 
   1451         html = '''
   1452             <div itemprop="author" itemscope>foo</div>
   1453         '''
   1454 
   1455         self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo')
   1456 
   1457     def test_get_elements_by_class(self):
   1458         html = '''
   1459             <span class="foo bar">nice</span><span class="foo bar">also nice</span>
   1460         '''
   1461 
   1462         self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
   1463         self.assertEqual(get_elements_by_class('no-such-class', html), [])
   1464 
   1465     def test_get_elements_by_attribute(self):
   1466         html = '''
   1467             <span class="foo bar">nice</span><span class="foo bar">also nice</span>
   1468         '''
   1469 
   1470         self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
   1471         self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
   1472         self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
   1473 
   1474     def test_clean_podcast_url(self):
   1475         self.assertEqual(clean_podcast_url('https://www.podtrac.com/pts/redirect.mp3/chtbl.com/track/5899E/traffic.megaphone.fm/HSW7835899191.mp3'), 'https://traffic.megaphone.fm/HSW7835899191.mp3')
   1476         self.assertEqual(clean_podcast_url('https://play.podtrac.com/npr-344098539/edge1.pod.npr.org/anon.npr-podcasts/podcast/npr/waitwait/2020/10/20201003_waitwait_wwdtmpodcast201003-015621a5-f035-4eca-a9a1-7c118d90bc3c.mp3'), 'https://edge1.pod.npr.org/anon.npr-podcasts/podcast/npr/waitwait/2020/10/20201003_waitwait_wwdtmpodcast201003-015621a5-f035-4eca-a9a1-7c118d90bc3c.mp3')
   1477 
   1478 
   1479 if __name__ == '__main__':
   1480     unittest.main()
	youtube-dl Another place where youtube-dl lives on
	git clone git://git.oshgnacknak.de/youtube-dl.git
	Log \| Files \| Refs \| README \| LICENSE