test/test_utils.py

   1 #!/usr/bin/env python3
   2
   3 # Allow direct execution
   4 import os
   5 import re
   6 import sys
   7 import unittest
   8 import warnings
   9
  10 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
  11
  12
  13 import contextlib
  14 import io
  15 import itertools
  16 import json
  17 import xml.etree.ElementTree
  18
  19 from yt_dlp.compat import (
  20     compat_etree_fromstring,
  21     compat_HTMLParseError,
  22     compat_os_name,
  23 )
  24 from yt_dlp.utils import (
  25     Config,
  26     DateRange,
  27     ExtractorError,
  28     InAdvancePagedList,
  29     LazyList,
  30     OnDemandPagedList,
  31     age_restricted,
  32     args_to_str,
  33     base_url,
  34     caesar,
  35     clean_html,
  36     clean_podcast_url,
  37     cli_bool_option,
  38     cli_option,
  39     cli_valueless_option,
  40     date_from_str,
  41     datetime_from_str,
  42     detect_exe_version,
  43     determine_ext,
  44     determine_file_encoding,
  45     dfxp2srt,
  46     dict_get,
  47     encode_base_n,
  48     encode_compat_str,
  49     encodeFilename,
  50     expand_path,
  51     extract_attributes,
  52     extract_basic_auth,
  53     find_xpath_attr,
  54     fix_xml_ampersands,
  55     float_or_none,
  56     format_bytes,
  57     get_compatible_ext,
  58     get_element_by_attribute,
  59     get_element_by_class,
  60     get_element_html_by_attribute,
  61     get_element_html_by_class,
  62     get_element_text_and_html_by_tag,
  63     get_elements_by_attribute,
  64     get_elements_by_class,
  65     get_elements_html_by_attribute,
  66     get_elements_html_by_class,
  67     get_elements_text_and_html_by_attribute,
  68     int_or_none,
  69     intlist_to_bytes,
  70     iri_to_uri,
  71     is_html,
  72     js_to_json,
  73     limit_length,
  74     locked_file,
  75     lowercase_escape,
  76     match_str,
  77     merge_dicts,
  78     mimetype2ext,
  79     month_by_name,
  80     multipart_encode,
  81     ohdave_rsa_encrypt,
  82     orderedSet,
  83     parse_age_limit,
  84     parse_bitrate,
  85     parse_codecs,
  86     parse_count,
  87     parse_dfxp_time_expr,
  88     parse_duration,
  89     parse_filesize,
  90     parse_iso8601,
  91     parse_qs,
  92     parse_resolution,
  93     pkcs1pad,
  94     prepend_extension,
  95     read_batch_urls,
  96     remove_end,
  97     remove_quotes,
  98     remove_start,
  99     render_table,
 100     replace_extension,
 101     rot47,
 102     sanitize_filename,
 103     sanitize_path,
 104     sanitize_url,
 105     shell_quote,
 106     smuggle_url,
 107     str_or_none,
 108     str_to_int,
 109     strip_jsonp,
 110     strip_or_none,
 111     subtitles_filename,
 112     timeconvert,
 113     traverse_obj,
 114     try_call,
 115     unescapeHTML,
 116     unified_strdate,
 117     unified_timestamp,
 118     unsmuggle_url,
 119     update_url_query,
 120     uppercase_escape,
 121     url_basename,
 122     url_or_none,
 123     urlencode_postdata,
 124     urljoin,
 125     urshift,
 126     variadic,
 127     version_tuple,
 128     xpath_attr,
 129     xpath_element,
 130     xpath_text,
 131     xpath_with_ns,
 132 )
 133 from yt_dlp.utils.networking import (
 134     HTTPHeaderDict,
 135     escape_rfc3986,
 136     normalize_url,
 137     remove_dot_segments,
 138 )
 139
 140
 141 class TestUtil(unittest.TestCase):
 142     def test_timeconvert(self):
 143         self.assertTrue(timeconvert('') is None)
 144         self.assertTrue(timeconvert('bougrg') is None)
 145
 146     def test_sanitize_filename(self):
 147         self.assertEqual(sanitize_filename(''), '')
 148         self.assertEqual(sanitize_filename('abc'), 'abc')
 149         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
 150
 151         self.assertEqual(sanitize_filename('123'), '123')
 152
 153         self.assertEqual('abc⧸de', sanitize_filename('abc/de'))
 154         self.assertFalse('/' in sanitize_filename('abc/de///'))
 155
 156         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', is_id=False))
 157         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', is_id=False))
 158         self.assertEqual('yes no', sanitize_filename('yes? no', is_id=False))
 159         self.assertEqual('this - that', sanitize_filename('this: that', is_id=False))
 160
 161         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
 162         aumlaut = 'ä'
 163         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
 164         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
 165         self.assertEqual(sanitize_filename(tests), tests)
 166
 167         self.assertEqual(
 168             sanitize_filename('New World record at 0:12:34'),
 169             'New World record at 0_12_34')
 170
 171         self.assertEqual(sanitize_filename('--gasdgf'), '--gasdgf')
 172         self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
 173         self.assertEqual(sanitize_filename('--gasdgf', is_id=False), '_-gasdgf')
 174         self.assertEqual(sanitize_filename('.gasdgf'), '.gasdgf')
 175         self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
 176         self.assertEqual(sanitize_filename('.gasdgf', is_id=False), 'gasdgf')
 177
 178         forbidden = '"\0\\/'
 179         for fc in forbidden:
 180             for fbc in forbidden:
 181                 self.assertTrue(fbc not in sanitize_filename(fc))
 182
 183     def test_sanitize_filename_restricted(self):
 184         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
 185         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
 186
 187         self.assertEqual(sanitize_filename('123', restricted=True), '123')
 188
 189         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
 190         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
 191
 192         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
 193         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
 194         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
 195         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
 196
 197         tests = 'aäb\u4e2d\u56fd\u7684c'
 198         self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
 199         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
 200
 201         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
 202         for fc in forbidden:
 203             for fbc in forbidden:
 204                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
 205
 206         # Handle a common case more neatly
 207         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
 208         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
 209         # .. but make sure the file name is never empty
 210         self.assertTrue(sanitize_filename('-', restricted=True) != '')
 211         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 212
 213         self.assertEqual(sanitize_filename(
 214             'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
 215             'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy')
 216
 217     def test_sanitize_ids(self):
 218         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 219         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 220         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 221
 222     def test_sanitize_path(self):
 223         if sys.platform != 'win32':
 224             return
 225
 226         self.assertEqual(sanitize_path('abc'), 'abc')
 227         self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
 228         self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
 229         self.assertEqual(sanitize_path('abc|def'), 'abc#def')
 230         self.assertEqual(sanitize_path('<>:"|?*'), '#######')
 231         self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
 232         self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
 233
 234         self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
 235         self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
 236
 237         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 238         self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
 239         self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
 240         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 241
 242         self.assertEqual(
 243             sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
 244             'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
 245
 246         self.assertEqual(
 247             sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
 248             'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
 249         self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
 250         self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
 251         self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
 252
 253         self.assertEqual(sanitize_path('../abc'), '..\\abc')
 254         self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
 255         self.assertEqual(sanitize_path('./abc'), 'abc')
 256         self.assertEqual(sanitize_path('./../abc'), '..\\abc')
 257
 258     def test_sanitize_url(self):
 259         self.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
 260         self.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
 261         self.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
 262         self.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')
 263         self.assertEqual(sanitize_url('foo bar'), 'foo bar')
 264
 265     def test_expand_path(self):
 266         def env(var):
 267             return f'%{var}%' if sys.platform == 'win32' else f'${var}'
 268
 269         os.environ['yt_dlp_EXPATH_PATH'] = 'expanded'
 270         self.assertEqual(expand_path(env('yt_dlp_EXPATH_PATH')), 'expanded')
 271
 272         old_home = os.environ.get('HOME')
 273         test_str = R'C:\Documents and Settings\тест\Application Data'
 274         try:
 275             os.environ['HOME'] = test_str
 276             self.assertEqual(expand_path(env('HOME')), os.getenv('HOME'))
 277             self.assertEqual(expand_path('~'), os.getenv('HOME'))
 278             self.assertEqual(
 279                 expand_path('~/%s' % env('yt_dlp_EXPATH_PATH')),
 280                 '%s/expanded' % os.getenv('HOME'))
 281         finally:
 282             os.environ['HOME'] = old_home or ''
 283
 284     def test_prepend_extension(self):
 285         self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
 286         self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
 287         self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 288         self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
 289         self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
 290         self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
 291
 292     def test_replace_extension(self):
 293         self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
 294         self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
 295         self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 296         self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
 297         self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
 298         self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
 299
 300     def test_subtitles_filename(self):
 301         self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt'), 'abc.en.vtt')
 302         self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt', 'ext'), 'abc.en.vtt')
 303         self.assertEqual(subtitles_filename('abc.unexpected_ext', 'en', 'vtt', 'ext'), 'abc.unexpected_ext.en.vtt')
 304
 305     def test_remove_start(self):
 306         self.assertEqual(remove_start(None, 'A - '), None)
 307         self.assertEqual(remove_start('A - B', 'A - '), 'B')
 308         self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
 309
 310     def test_remove_end(self):
 311         self.assertEqual(remove_end(None, ' - B'), None)
 312         self.assertEqual(remove_end('A - B', ' - B'), 'A')
 313         self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
 314
 315     def test_remove_quotes(self):
 316         self.assertEqual(remove_quotes(None), None)
 317         self.assertEqual(remove_quotes('"'), '"')
 318         self.assertEqual(remove_quotes("'"), "'")
 319         self.assertEqual(remove_quotes(';'), ';')
 320         self.assertEqual(remove_quotes('";'), '";')
 321         self.assertEqual(remove_quotes('""'), '')
 322         self.assertEqual(remove_quotes('";"'), ';')
 323
 324     def test_ordered_set(self):
 325         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
 326         self.assertEqual(orderedSet([]), [])
 327         self.assertEqual(orderedSet([1]), [1])
 328         # keep the list ordered
 329         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
 330
 331     def test_unescape_html(self):
 332         self.assertEqual(unescapeHTML('%20;'), '%20;')
 333         self.assertEqual(unescapeHTML('&#x2F;'), '/')
 334         self.assertEqual(unescapeHTML('&#47;'), '/')
 335         self.assertEqual(unescapeHTML('&eacute;'), 'é')
 336         self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
 337         self.assertEqual(unescapeHTML('&a&quot;'), '&a"')
 338         # HTML5 entities
 339         self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')
 340
 341     def test_date_from_str(self):
 342         self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
 343         self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
 344         self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
 345         self.assertEqual(date_from_str('20200229+365day'), date_from_str('20200229+1year'))
 346         self.assertEqual(date_from_str('20210131+28day'), date_from_str('20210131+1month'))
 347
 348     def test_datetime_from_str(self):
 349         self.assertEqual(datetime_from_str('yesterday', precision='day'), datetime_from_str('now-1day', precision='auto'))
 350         self.assertEqual(datetime_from_str('now+7day', precision='day'), datetime_from_str('now+1week', precision='auto'))
 351         self.assertEqual(datetime_from_str('now+14day', precision='day'), datetime_from_str('now+2week', precision='auto'))
 352         self.assertEqual(datetime_from_str('20200229+365day', precision='day'), datetime_from_str('20200229+1year', precision='auto'))
 353         self.assertEqual(datetime_from_str('20210131+28day', precision='day'), datetime_from_str('20210131+1month', precision='auto'))
 354         self.assertEqual(datetime_from_str('20210131+59day', precision='day'), datetime_from_str('20210131+2month', precision='auto'))
 355         self.assertEqual(datetime_from_str('now+1day', precision='hour'), datetime_from_str('now+24hours', precision='auto'))
 356         self.assertEqual(datetime_from_str('now+23hours', precision='hour'), datetime_from_str('now+23hours', precision='auto'))
 357
 358     def test_daterange(self):
 359         _20century = DateRange("19000101", "20000101")
 360         self.assertFalse("17890714" in _20century)
 361         _ac = DateRange("00010101")
 362         self.assertTrue("19690721" in _ac)
 363         _firstmilenium = DateRange(end="10000101")
 364         self.assertTrue("07110427" in _firstmilenium)
 365
 366     def test_unified_dates(self):
 367         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
 368         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
 369         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
 370         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
 371         self.assertEqual(unified_strdate('1968 12 10'), '19681210')
 372         self.assertEqual(unified_strdate('1968-12-10'), '19681210')
 373         self.assertEqual(unified_strdate('31-07-2022 20:00'), '20220731')
 374         self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
 375         self.assertEqual(
 376             unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
 377             '20141126')
 378         self.assertEqual(
 379             unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
 380             '20150202')
 381         self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
 382         self.assertEqual(unified_strdate('25-09-2014'), '20140925')
 383         self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
 384         self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
 385         self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
 386         self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
 387         self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
 388         self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
 389         self.assertEqual(unified_strdate('November 3rd, 2019'), '20191103')
 390         self.assertEqual(unified_strdate('October 23rd, 2005'), '20051023')
 391
 392     def test_unified_timestamps(self):
 393         self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
 394         self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
 395         self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
 396         self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
 397         self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
 398         self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
 399         self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
 400         self.assertEqual(
 401             unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
 402             1417001400)
 403         self.assertEqual(
 404             unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
 405             1422902860)
 406         self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
 407         self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
 408         self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
 409         self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
 410         self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
 411         self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
 412         self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
 413         self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
 414         self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
 415         self.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363)
 416
 417         self.assertEqual(unified_timestamp('December 31 1969 20:00:01 EDT'), 1)
 418         self.assertEqual(unified_timestamp('Wednesday 31 December 1969 18:01:26 MDT'), 86)
 419         self.assertEqual(unified_timestamp('12/31/1969 20:01:18 EDT', False), 78)
 420
 421     def test_determine_ext(self):
 422         self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
 423         self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
 424         self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
 425         self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
 426         self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
 427         self.assertEqual(determine_ext('foobar', None), None)
 428
 429     def test_find_xpath_attr(self):
 430         testxml = '''<root>
 431             <node/>
 432             <node x="a"/>
 433             <node x="a" y="c" />
 434             <node x="b" y="d" />
 435             <node x="" />
 436         </root>'''
 437         doc = compat_etree_fromstring(testxml)
 438
 439         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
 440         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
 441         self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
 442         self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
 443         self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
 444         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
 445         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
 446         self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
 447         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
 448         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
 449         self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
 450
 451     def test_xpath_with_ns(self):
 452         testxml = '''<root xmlns:media="http://example.com/">
 453             <media:song>
 454                 <media:author>The Author</media:author>
 455                 <url>http://server.com/download.mp3</url>
 456             </media:song>
 457         </root>'''
 458         doc = compat_etree_fromstring(testxml)
 459         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
 460         self.assertTrue(find('media:song') is not None)
 461         self.assertEqual(find('media:song/media:author').text, 'The Author')
 462         self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
 463
 464     def test_xpath_element(self):
 465         doc = xml.etree.ElementTree.Element('root')
 466         div = xml.etree.ElementTree.SubElement(doc, 'div')
 467         p = xml.etree.ElementTree.SubElement(div, 'p')
 468         p.text = 'Foo'
 469         self.assertEqual(xpath_element(doc, 'div/p'), p)
 470         self.assertEqual(xpath_element(doc, ['div/p']), p)
 471         self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
 472         self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
 473         self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
 474         self.assertTrue(xpath_element(doc, 'div/bar') is None)
 475         self.assertTrue(xpath_element(doc, ['div/bar']) is None)
 476         self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
 477         self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
 478         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
 479         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
 480
 481     def test_xpath_text(self):
 482         testxml = '''<root>
 483             <div>
 484                 <p>Foo</p>
 485             </div>
 486         </root>'''
 487         doc = compat_etree_fromstring(testxml)
 488         self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
 489         self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
 490         self.assertTrue(xpath_text(doc, 'div/bar') is None)
 491         self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
 492
 493     def test_xpath_attr(self):
 494         testxml = '''<root>
 495             <div>
 496                 <p x="a">Foo</p>
 497             </div>
 498         </root>'''
 499         doc = compat_etree_fromstring(testxml)
 500         self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
 501         self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
 502         self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
 503         self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
 504         self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
 505         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
 506         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
 507
 508     def test_smuggle_url(self):
 509         data = {"ö": "ö", "abc": [3]}
 510         url = 'https://foo.bar/baz?x=y#a'
 511         smug_url = smuggle_url(url, data)
 512         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
 513         self.assertEqual(url, unsmug_url)
 514         self.assertEqual(data, unsmug_data)
 515
 516         res_url, res_data = unsmuggle_url(url)
 517         self.assertEqual(res_url, url)
 518         self.assertEqual(res_data, None)
 519
 520         smug_url = smuggle_url(url, {'a': 'b'})
 521         smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
 522         res_url, res_data = unsmuggle_url(smug_smug_url)
 523         self.assertEqual(res_url, url)
 524         self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
 525
 526     def test_shell_quote(self):
 527         args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
 528         self.assertEqual(
 529             shell_quote(args),
 530             """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
 531
 532     def test_float_or_none(self):
 533         self.assertEqual(float_or_none('42.42'), 42.42)
 534         self.assertEqual(float_or_none('42'), 42.0)
 535         self.assertEqual(float_or_none(''), None)
 536         self.assertEqual(float_or_none(None), None)
 537         self.assertEqual(float_or_none([]), None)
 538         self.assertEqual(float_or_none(set()), None)
 539
 540     def test_int_or_none(self):
 541         self.assertEqual(int_or_none('42'), 42)
 542         self.assertEqual(int_or_none(''), None)
 543         self.assertEqual(int_or_none(None), None)
 544         self.assertEqual(int_or_none([]), None)
 545         self.assertEqual(int_or_none(set()), None)
 546
 547     def test_str_to_int(self):
 548         self.assertEqual(str_to_int('123,456'), 123456)
 549         self.assertEqual(str_to_int('123.456'), 123456)
 550         self.assertEqual(str_to_int(523), 523)
 551         self.assertEqual(str_to_int('noninteger'), None)
 552         self.assertEqual(str_to_int([]), None)
 553
 554     def test_url_basename(self):
 555         self.assertEqual(url_basename('http://foo.de/'), '')
 556         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
 557         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
 558         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
 559         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
 560         self.assertEqual(
 561             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
 562             'trailer.mp4')
 563
 564     def test_base_url(self):
 565         self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
 566         self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
 567         self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
 568         self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
 569         self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
 570         self.assertEqual(base_url('http://foo.de/bar/baz&x=z&w=y/x/c'), 'http://foo.de/bar/baz&x=z&w=y/x/')
 571
 572     def test_urljoin(self):
 573         self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 574         self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 575         self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 576         self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 577         self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
 578         self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 579         self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 580         self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 581         self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 582         self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
 583         self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 584         self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
 585         self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 586         self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 587         self.assertEqual(urljoin('http://foo.de/', None), None)
 588         self.assertEqual(urljoin('http://foo.de/', ''), None)
 589         self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
 590         self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
 591         self.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de')
 592         self.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de')
 593
 594     def test_url_or_none(self):
 595         self.assertEqual(url_or_none(None), None)
 596         self.assertEqual(url_or_none(''), None)
 597         self.assertEqual(url_or_none('foo'), None)
 598         self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
 599         self.assertEqual(url_or_none('https://foo.de'), 'https://foo.de')
 600         self.assertEqual(url_or_none('http$://foo.de'), None)
 601         self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
 602         self.assertEqual(url_or_none('//foo.de'), '//foo.de')
 603         self.assertEqual(url_or_none('s3://foo.de'), None)
 604         self.assertEqual(url_or_none('rtmpte://foo.de'), 'rtmpte://foo.de')
 605         self.assertEqual(url_or_none('mms://foo.de'), 'mms://foo.de')
 606         self.assertEqual(url_or_none('rtspu://foo.de'), 'rtspu://foo.de')
 607         self.assertEqual(url_or_none('ftps://foo.de'), 'ftps://foo.de')
 608
 609     def test_parse_age_limit(self):
 610         self.assertEqual(parse_age_limit(None), None)
 611         self.assertEqual(parse_age_limit(False), None)
 612         self.assertEqual(parse_age_limit('invalid'), None)
 613         self.assertEqual(parse_age_limit(0), 0)
 614         self.assertEqual(parse_age_limit(18), 18)
 615         self.assertEqual(parse_age_limit(21), 21)
 616         self.assertEqual(parse_age_limit(22), None)
 617         self.assertEqual(parse_age_limit('18'), 18)
 618         self.assertEqual(parse_age_limit('18+'), 18)
 619         self.assertEqual(parse_age_limit('PG-13'), 13)
 620         self.assertEqual(parse_age_limit('TV-14'), 14)
 621         self.assertEqual(parse_age_limit('TV-MA'), 17)
 622         self.assertEqual(parse_age_limit('TV14'), 14)
 623         self.assertEqual(parse_age_limit('TV_G'), 0)
 624
 625     def test_parse_duration(self):
 626         self.assertEqual(parse_duration(None), None)
 627         self.assertEqual(parse_duration(False), None)
 628         self.assertEqual(parse_duration('invalid'), None)
 629         self.assertEqual(parse_duration('1'), 1)
 630         self.assertEqual(parse_duration('1337:12'), 80232)
 631         self.assertEqual(parse_duration('9:12:43'), 33163)
 632         self.assertEqual(parse_duration('12:00'), 720)
 633         self.assertEqual(parse_duration('00:01:01'), 61)
 634         self.assertEqual(parse_duration('x:y'), None)
 635         self.assertEqual(parse_duration('3h11m53s'), 11513)
 636         self.assertEqual(parse_duration('3h 11m 53s'), 11513)
 637         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
 638         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
 639         self.assertEqual(parse_duration('3 hours, 11 minutes, 53 seconds'), 11513)
 640         self.assertEqual(parse_duration('3 hours, 11 mins, 53 secs'), 11513)
 641         self.assertEqual(parse_duration('62m45s'), 3765)
 642         self.assertEqual(parse_duration('6m59s'), 419)
 643         self.assertEqual(parse_duration('49s'), 49)
 644         self.assertEqual(parse_duration('0h0m0s'), 0)
 645         self.assertEqual(parse_duration('0m0s'), 0)
 646         self.assertEqual(parse_duration('0s'), 0)
 647         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
 648         self.assertEqual(parse_duration('T30M38S'), 1838)
 649         self.assertEqual(parse_duration('5 s'), 5)
 650         self.assertEqual(parse_duration('3 min'), 180)
 651         self.assertEqual(parse_duration('2.5 hours'), 9000)
 652         self.assertEqual(parse_duration('02:03:04'), 7384)
 653         self.assertEqual(parse_duration('01:02:03:04'), 93784)
 654         self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
 655         self.assertEqual(parse_duration('87 Min.'), 5220)
 656         self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
 657         self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
 658         self.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)
 659         self.assertEqual(parse_duration('01:02:03:050'), 3723.05)
 660         self.assertEqual(parse_duration('103:050'), 103.05)
 661         self.assertEqual(parse_duration('1HR 3MIN'), 3780)
 662         self.assertEqual(parse_duration('2hrs 3mins'), 7380)
 663
 664     def test_fix_xml_ampersands(self):
 665         self.assertEqual(
 666             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 667         self.assertEqual(
 668             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 669             '"&amp;x=y&amp;wrong;&amp;z=a')
 670         self.assertEqual(
 671             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 672             '&amp;&apos;&gt;&lt;&quot;')
 673         self.assertEqual(
 674             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 675         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 676
 677     def test_paged_list(self):
 678         def testPL(size, pagesize, sliceargs, expected):
 679             def get_page(pagenum):
 680                 firstid = pagenum * pagesize
 681                 upto = min(size, pagenum * pagesize + pagesize)
 682                 yield from range(firstid, upto)
 683
 684             pl = OnDemandPagedList(get_page, pagesize)
 685             got = pl.getslice(*sliceargs)
 686             self.assertEqual(got, expected)
 687
 688             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
 689             got = iapl.getslice(*sliceargs)
 690             self.assertEqual(got, expected)
 691
 692         testPL(5, 2, (), [0, 1, 2, 3, 4])
 693         testPL(5, 2, (1,), [1, 2, 3, 4])
 694         testPL(5, 2, (2,), [2, 3, 4])
 695         testPL(5, 2, (4,), [4])
 696         testPL(5, 2, (0, 3), [0, 1, 2])
 697         testPL(5, 2, (1, 4), [1, 2, 3])
 698         testPL(5, 2, (2, 99), [2, 3, 4])
 699         testPL(5, 2, (20, 99), [])
 700
 701     def test_read_batch_urls(self):
 702         f = io.StringIO('''\xef\xbb\xbf foo
 703             bar\r
 704             baz
 705             # More after this line\r
 706             ; or after this
 707             bam''')
 708         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
 709
 710     def test_urlencode_postdata(self):
 711         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
 712         self.assertTrue(isinstance(data, bytes))
 713
 714     def test_update_url_query(self):
 715         self.assertEqual(parse_qs(update_url_query(
 716             'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
 717             parse_qs('http://example.com/path?quality=HD&format=mp4'))
 718         self.assertEqual(parse_qs(update_url_query(
 719             'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
 720             parse_qs('http://example.com/path?system=LINUX&system=WINDOWS'))
 721         self.assertEqual(parse_qs(update_url_query(
 722             'http://example.com/path', {'fields': 'id,formats,subtitles'})),
 723             parse_qs('http://example.com/path?fields=id,formats,subtitles'))
 724         self.assertEqual(parse_qs(update_url_query(
 725             'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
 726             parse_qs('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
 727         self.assertEqual(parse_qs(update_url_query(
 728             'http://example.com/path?manifest=f4m', {'manifest': []})),
 729             parse_qs('http://example.com/path'))
 730         self.assertEqual(parse_qs(update_url_query(
 731             'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
 732             parse_qs('http://example.com/path?system=LINUX'))
 733         self.assertEqual(parse_qs(update_url_query(
 734             'http://example.com/path', {'fields': b'id,formats,subtitles'})),
 735             parse_qs('http://example.com/path?fields=id,formats,subtitles'))
 736         self.assertEqual(parse_qs(update_url_query(
 737             'http://example.com/path', {'width': 1080, 'height': 720})),
 738             parse_qs('http://example.com/path?width=1080&height=720'))
 739         self.assertEqual(parse_qs(update_url_query(
 740             'http://example.com/path', {'bitrate': 5020.43})),
 741             parse_qs('http://example.com/path?bitrate=5020.43'))
 742         self.assertEqual(parse_qs(update_url_query(
 743             'http://example.com/path', {'test': '第二行тест'})),
 744             parse_qs('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
 745
 746     def test_multipart_encode(self):
 747         self.assertEqual(
 748             multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0],
 749             b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
 750         self.assertEqual(
 751             multipart_encode({'欄位'.encode(): '值'.encode()}, boundary='AAAAAA')[0],
 752             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')
 753         self.assertRaises(
 754             ValueError, multipart_encode, {b'field': b'value'}, boundary='value')
 755
 756     def test_dict_get(self):
 757         FALSE_VALUES = {
 758             'none': None,
 759             'false': False,
 760             'zero': 0,
 761             'empty_string': '',
 762             'empty_list': [],
 763         }
 764         d = FALSE_VALUES.copy()
 765         d['a'] = 42
 766         self.assertEqual(dict_get(d, 'a'), 42)
 767         self.assertEqual(dict_get(d, 'b'), None)
 768         self.assertEqual(dict_get(d, 'b', 42), 42)
 769         self.assertEqual(dict_get(d, ('a', )), 42)
 770         self.assertEqual(dict_get(d, ('b', 'a', )), 42)
 771         self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
 772         self.assertEqual(dict_get(d, ('b', 'c', )), None)
 773         self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
 774         for key, false_value in FALSE_VALUES.items():
 775             self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
 776             self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
 777
 778     def test_merge_dicts(self):
 779         self.assertEqual(merge_dicts({'a': 1}, {'b': 2}), {'a': 1, 'b': 2})
 780         self.assertEqual(merge_dicts({'a': 1}, {'a': 2}), {'a': 1})
 781         self.assertEqual(merge_dicts({'a': 1}, {'a': None}), {'a': 1})
 782         self.assertEqual(merge_dicts({'a': 1}, {'a': ''}), {'a': 1})
 783         self.assertEqual(merge_dicts({'a': 1}, {}), {'a': 1})
 784         self.assertEqual(merge_dicts({'a': None}, {'a': 1}), {'a': 1})
 785         self.assertEqual(merge_dicts({'a': ''}, {'a': 1}), {'a': ''})
 786         self.assertEqual(merge_dicts({'a': ''}, {'a': 'abc'}), {'a': 'abc'})
 787         self.assertEqual(merge_dicts({'a': None}, {'a': ''}, {'a': 'abc'}), {'a': 'abc'})
 788
 789     def test_encode_compat_str(self):
 790         self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
 791         self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
 792
 793     def test_parse_iso8601(self):
 794         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
 795         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
 796         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
 797         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
 798         self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
 799         self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
 800
 801     def test_strip_jsonp(self):
 802         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
 803         d = json.loads(stripped)
 804         self.assertEqual(d, [{"id": "532cb", "x": 3}])
 805
 806         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
 807         d = json.loads(stripped)
 808         self.assertEqual(d, {'STATUS': 'OK'})
 809
 810         stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
 811         d = json.loads(stripped)
 812         self.assertEqual(d, {'status': 'success'})
 813
 814         stripped = strip_jsonp('window.cb && window.cb({"status": "success"});')
 815         d = json.loads(stripped)
 816         self.assertEqual(d, {'status': 'success'})
 817
 818         stripped = strip_jsonp('window.cb && cb({"status": "success"});')
 819         d = json.loads(stripped)
 820         self.assertEqual(d, {'status': 'success'})
 821
 822         stripped = strip_jsonp('({"status": "success"});')
 823         d = json.loads(stripped)
 824         self.assertEqual(d, {'status': 'success'})
 825
 826     def test_strip_or_none(self):
 827         self.assertEqual(strip_or_none(' abc'), 'abc')
 828         self.assertEqual(strip_or_none('abc '), 'abc')
 829         self.assertEqual(strip_or_none(' abc '), 'abc')
 830         self.assertEqual(strip_or_none('\tabc\t'), 'abc')
 831         self.assertEqual(strip_or_none('\n\tabc\n\t'), 'abc')
 832         self.assertEqual(strip_or_none('abc'), 'abc')
 833         self.assertEqual(strip_or_none(''), '')
 834         self.assertEqual(strip_or_none(None), None)
 835         self.assertEqual(strip_or_none(42), None)
 836         self.assertEqual(strip_or_none([]), None)
 837
 838     def test_uppercase_escape(self):
 839         self.assertEqual(uppercase_escape('aä'), 'aä')
 840         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
 841
 842     def test_lowercase_escape(self):
 843         self.assertEqual(lowercase_escape('aä'), 'aä')
 844         self.assertEqual(lowercase_escape('\\u0026'), '&')
 845
 846     def test_limit_length(self):
 847         self.assertEqual(limit_length(None, 12), None)
 848         self.assertEqual(limit_length('foo', 12), 'foo')
 849         self.assertTrue(
 850             limit_length('foo bar baz asd', 12).startswith('foo bar'))
 851         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
 852
 853     def test_mimetype2ext(self):
 854         self.assertEqual(mimetype2ext(None), None)
 855         self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
 856         self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
 857         self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
 858         self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
 859         self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
 860         self.assertEqual(mimetype2ext('audio/x-wav'), 'wav')
 861         self.assertEqual(mimetype2ext('audio/x-wav;codec=pcm'), 'wav')
 862
 863     def test_month_by_name(self):
 864         self.assertEqual(month_by_name(None), None)
 865         self.assertEqual(month_by_name('December', 'en'), 12)
 866         self.assertEqual(month_by_name('décembre', 'fr'), 12)
 867         self.assertEqual(month_by_name('December'), 12)
 868         self.assertEqual(month_by_name('décembre'), None)
 869         self.assertEqual(month_by_name('Unknown', 'unknown'), None)
 870
 871     def test_parse_codecs(self):
 872         self.assertEqual(parse_codecs(''), {})
 873         self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
 874             'vcodec': 'avc1.77.30',
 875             'acodec': 'mp4a.40.2',
 876             'dynamic_range': None,
 877         })
 878         self.assertEqual(parse_codecs('mp4a.40.2'), {
 879             'vcodec': 'none',
 880             'acodec': 'mp4a.40.2',
 881             'dynamic_range': None,
 882         })
 883         self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
 884             'vcodec': 'avc1.42001e',
 885             'acodec': 'mp4a.40.5',
 886             'dynamic_range': None,
 887         })
 888         self.assertEqual(parse_codecs('avc3.640028'), {
 889             'vcodec': 'avc3.640028',
 890             'acodec': 'none',
 891             'dynamic_range': None,
 892         })
 893         self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
 894             'vcodec': 'h264',
 895             'acodec': 'aac',
 896             'dynamic_range': None,
 897         })
 898         self.assertEqual(parse_codecs('av01.0.05M.08'), {
 899             'vcodec': 'av01.0.05M.08',
 900             'acodec': 'none',
 901             'dynamic_range': None,
 902         })
 903         self.assertEqual(parse_codecs('vp9.2'), {
 904             'vcodec': 'vp9.2',
 905             'acodec': 'none',
 906             'dynamic_range': 'HDR10',
 907         })
 908         self.assertEqual(parse_codecs('av01.0.12M.10.0.110.09.16.09.0'), {
 909             'vcodec': 'av01.0.12M.10.0.110.09.16.09.0',
 910             'acodec': 'none',
 911             'dynamic_range': 'HDR10',
 912         })
 913         self.assertEqual(parse_codecs('dvhe'), {
 914             'vcodec': 'dvhe',
 915             'acodec': 'none',
 916             'dynamic_range': 'DV',
 917         })
 918         self.assertEqual(parse_codecs('theora, vorbis'), {
 919             'vcodec': 'theora',
 920             'acodec': 'vorbis',
 921             'dynamic_range': None,
 922         })
 923         self.assertEqual(parse_codecs('unknownvcodec, unknownacodec'), {
 924             'vcodec': 'unknownvcodec',
 925             'acodec': 'unknownacodec',
 926         })
 927         self.assertEqual(parse_codecs('unknown'), {})
 928
 929     def test_escape_rfc3986(self):
 930         reserved = "!*'();:@&=+$,/?#[]"
 931         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
 932         self.assertEqual(escape_rfc3986(reserved), reserved)
 933         self.assertEqual(escape_rfc3986(unreserved), unreserved)
 934         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
 935         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
 936         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
 937         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
 938
 939     def test_normalize_url(self):
 940         self.assertEqual(
 941             normalize_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
 942             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
 943         )
 944         self.assertEqual(
 945             normalize_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
 946             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
 947         )
 948         self.assertEqual(
 949             normalize_url('http://тест.рф/фрагмент'),
 950             'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
 951         )
 952         self.assertEqual(
 953             normalize_url('http://тест.рф/абв?абв=абв#абв'),
 954             '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'
 955         )
 956         self.assertEqual(normalize_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
 957
 958         self.assertEqual(normalize_url('http://www.example.com/../a/b/../c/./d.html'), 'http://www.example.com/a/c/d.html')
 959
 960     def test_remove_dot_segments(self):
 961         self.assertEqual(remove_dot_segments('/a/b/c/./../../g'), '/a/g')
 962         self.assertEqual(remove_dot_segments('mid/content=5/../6'), 'mid/6')
 963         self.assertEqual(remove_dot_segments('/ad/../cd'), '/cd')
 964         self.assertEqual(remove_dot_segments('/ad/../cd/'), '/cd/')
 965         self.assertEqual(remove_dot_segments('/..'), '/')
 966         self.assertEqual(remove_dot_segments('/./'), '/')
 967         self.assertEqual(remove_dot_segments('/./a'), '/a')
 968         self.assertEqual(remove_dot_segments('/abc/./.././d/././e/.././f/./../../ghi'), '/ghi')
 969         self.assertEqual(remove_dot_segments('/'), '/')
 970         self.assertEqual(remove_dot_segments('/t'), '/t')
 971         self.assertEqual(remove_dot_segments('t'), 't')
 972         self.assertEqual(remove_dot_segments(''), '')
 973         self.assertEqual(remove_dot_segments('/../a/b/c'), '/a/b/c')
 974         self.assertEqual(remove_dot_segments('../a'), 'a')
 975         self.assertEqual(remove_dot_segments('./a'), 'a')
 976         self.assertEqual(remove_dot_segments('.'), '')
 977         self.assertEqual(remove_dot_segments('////'), '////')
 978
 979     def test_js_to_json_vars_strings(self):
 980         self.assertDictEqual(
 981             json.loads(js_to_json(
 982                 '''{
 983                     'null': a,
 984                     'nullStr': b,
 985                     'true': c,
 986                     'trueStr': d,
 987                     'false': e,
 988                     'falseStr': f,
 989                     'unresolvedVar': g,
 990                 }''',
 991                 {
 992                     'a': 'null',
 993                     'b': '"null"',
 994                     'c': 'true',
 995                     'd': '"true"',
 996                     'e': 'false',
 997                     'f': '"false"',
 998                     'g': 'var',
 999                 }
1000             )),
1001             {
1002                 'null': None,
1003                 'nullStr': 'null',
1004                 'true': True,
1005                 'trueStr': 'true',
1006                 'false': False,
1007                 'falseStr': 'false',
1008                 'unresolvedVar': 'var'
1009             }
1010         )
1011
1012         self.assertDictEqual(
1013             json.loads(js_to_json(
1014                 '''{
1015                     'int': a,
1016                     'intStr': b,
1017                     'float': c,
1018                     'floatStr': d,
1019                 }''',
1020                 {
1021                     'a': '123',
1022                     'b': '"123"',
1023                     'c': '1.23',
1024                     'd': '"1.23"',
1025                 }
1026             )),
1027             {
1028                 'int': 123,
1029                 'intStr': '123',
1030                 'float': 1.23,
1031                 'floatStr': '1.23',
1032             }
1033         )
1034
1035         self.assertDictEqual(
1036             json.loads(js_to_json(
1037                 '''{
1038                     'object': a,
1039                     'objectStr': b,
1040                     'array': c,
1041                     'arrayStr': d,
1042                 }''',
1043                 {
1044                     'a': '{}',
1045                     'b': '"{}"',
1046                     'c': '[]',
1047                     'd': '"[]"',
1048                 }
1049             )),
1050             {
1051                 'object': {},
1052                 'objectStr': '{}',
1053                 'array': [],
1054                 'arrayStr': '[]',
1055             }
1056         )
1057
1058     def test_js_to_json_realworld(self):
1059         inp = '''{
1060             'clip':{'provider':'pseudo'}
1061         }'''
1062         self.assertEqual(js_to_json(inp), '''{
1063             "clip":{"provider":"pseudo"}
1064         }''')
1065         json.loads(js_to_json(inp))
1066
1067         inp = '''{
1068             'playlist':[{'controls':{'all':null}}]
1069         }'''
1070         self.assertEqual(js_to_json(inp), '''{
1071             "playlist":[{"controls":{"all":null}}]
1072         }''')
1073
1074         inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
1075         self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
1076
1077         inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
1078         json_code = js_to_json(inp)
1079         self.assertEqual(json.loads(json_code), json.loads(inp))
1080
1081         inp = '''{
1082             0:{src:'skipped', type: 'application/dash+xml'},
1083             1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
1084         }'''
1085         self.assertEqual(js_to_json(inp), '''{
1086             "0":{"src":"skipped", "type": "application/dash+xml"},
1087             "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
1088         }''')
1089
1090         inp = '''{"foo":101}'''
1091         self.assertEqual(js_to_json(inp), '''{"foo":101}''')
1092
1093         inp = '''{"duration": "00:01:07"}'''
1094         self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
1095
1096         inp = '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}'''
1097         self.assertEqual(js_to_json(inp), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''')
1098
1099     def test_js_to_json_edgecases(self):
1100         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
1101         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
1102
1103         on = js_to_json('{"abc": true}')
1104         self.assertEqual(json.loads(on), {'abc': True})
1105
1106         # Ignore JavaScript code as well
1107         on = js_to_json('''{
1108             "x": 1,
1109             y: "a",
1110             z: some.code
1111         }''')
1112         d = json.loads(on)
1113         self.assertEqual(d['x'], 1)
1114         self.assertEqual(d['y'], 'a')
1115
1116         # Just drop ! prefix for now though this results in a wrong value
1117         on = js_to_json('''{
1118             a: !0,
1119             b: !1,
1120             c: !!0,
1121             d: !!42.42,
1122             e: !!![],
1123             f: !"abc",
1124             g: !"",
1125             !42: 42
1126         }''')
1127         self.assertEqual(json.loads(on), {
1128             'a': 0,
1129             'b': 1,
1130             'c': 0,
1131             'd': 42.42,
1132             'e': [],
1133             'f': "abc",
1134             'g': "",
1135             '42': 42
1136         })
1137
1138         on = js_to_json('["abc", "def",]')
1139         self.assertEqual(json.loads(on), ['abc', 'def'])
1140
1141         on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
1142         self.assertEqual(json.loads(on), ['abc', 'def'])
1143
1144         on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
1145         self.assertEqual(json.loads(on), ['abc', 'def'])
1146
1147         on = js_to_json('{"abc": "def",}')
1148         self.assertEqual(json.loads(on), {'abc': 'def'})
1149
1150         on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
1151         self.assertEqual(json.loads(on), {'abc': 'def'})
1152
1153         on = js_to_json('{ 0: /* " \n */ ",]" , }')
1154         self.assertEqual(json.loads(on), {'0': ',]'})
1155
1156         on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
1157         self.assertEqual(json.loads(on), {'0': ',]'})
1158
1159         on = js_to_json('{ 0: // comment\n1 }')
1160         self.assertEqual(json.loads(on), {'0': 1})
1161
1162         on = js_to_json(r'["<p>x<\/p>"]')
1163         self.assertEqual(json.loads(on), ['<p>x</p>'])
1164
1165         on = js_to_json(r'["\xaa"]')
1166         self.assertEqual(json.loads(on), ['\u00aa'])
1167
1168         on = js_to_json("['a\\\nb']")
1169         self.assertEqual(json.loads(on), ['ab'])
1170
1171         on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
1172         self.assertEqual(json.loads(on), ['ab'])
1173
1174         on = js_to_json('{0xff:0xff}')
1175         self.assertEqual(json.loads(on), {'255': 255})
1176
1177         on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
1178         self.assertEqual(json.loads(on), {'255': 255})
1179
1180         on = js_to_json('{077:077}')
1181         self.assertEqual(json.loads(on), {'63': 63})
1182
1183         on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
1184         self.assertEqual(json.loads(on), {'63': 63})
1185
1186         on = js_to_json('{42:42}')
1187         self.assertEqual(json.loads(on), {'42': 42})
1188
1189         on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
1190         self.assertEqual(json.loads(on), {'42': 42})
1191
1192         on = js_to_json('{42:4.2e1}')
1193         self.assertEqual(json.loads(on), {'42': 42.0})
1194
1195         on = js_to_json('{ "0x40": "0x40" }')
1196         self.assertEqual(json.loads(on), {'0x40': '0x40'})
1197
1198         on = js_to_json('{ "040": "040" }')
1199         self.assertEqual(json.loads(on), {'040': '040'})
1200
1201         on = js_to_json('[1,//{},\n2]')
1202         self.assertEqual(json.loads(on), [1, 2])
1203
1204         on = js_to_json(R'"\^\$\#"')
1205         self.assertEqual(json.loads(on), R'^$#', msg='Unnecessary escapes should be stripped')
1206
1207         on = js_to_json('\'"\\""\'')
1208         self.assertEqual(json.loads(on), '"""', msg='Unnecessary quote escape should be escaped')
1209
1210     def test_js_to_json_malformed(self):
1211         self.assertEqual(js_to_json('42a1'), '42"a1"')
1212         self.assertEqual(js_to_json('42a-1'), '42"a"-1')
1213
1214     def test_js_to_json_template_literal(self):
1215         self.assertEqual(js_to_json('`Hello ${name}`', {'name': '"world"'}), '"Hello world"')
1216         self.assertEqual(js_to_json('`${name}${name}`', {'name': '"X"'}), '"XX"')
1217         self.assertEqual(js_to_json('`${name}${name}`', {'name': '5'}), '"55"')
1218         self.assertEqual(js_to_json('`${name}"${name}"`', {'name': '5'}), '"5\\"5\\""')
1219         self.assertEqual(js_to_json('`${name}`', {}), '"name"')
1220
1221     def test_js_to_json_map_array_constructors(self):
1222         self.assertEqual(json.loads(js_to_json('new Map([["a", 5]])')), {'a': 5})
1223         self.assertEqual(json.loads(js_to_json('Array(5, 10)')), [5, 10])
1224         self.assertEqual(json.loads(js_to_json('new Array(15,5)')), [15, 5])
1225         self.assertEqual(json.loads(js_to_json('new Map([Array(5, 10),new Array(15,5)])')), {'5': 10, '15': 5})
1226
1227     def test_extract_attributes(self):
1228         self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
1229         self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
1230         self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
1231         self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
1232         self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
1233         self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
1234         self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
1235         self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'})  # XML
1236         self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
1237         self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'})  # HTML 3.2
1238         self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'})  # HTML 4.0
1239         self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
1240         self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
1241         self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
1242         self.assertEqual(extract_attributes('<e x >'), {'x': None})
1243         self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
1244         self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
1245         self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
1246         self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
1247         self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
1248         self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
1249         self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
1250         self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'})  # Names lowercased
1251         self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
1252         self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
1253         self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
1254         self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
1255         self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
1256         # "Narrow" Python builds don't support unicode code points outside BMP.
1257         try:
1258             chr(0x10000)
1259             supports_outside_bmp = True
1260         except ValueError:
1261             supports_outside_bmp = False
1262         if supports_outside_bmp:
1263             self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
1264         # Malformed HTML should not break attributes extraction on older Python
1265         self.assertEqual(extract_attributes('<mal"formed/>'), {})
1266
1267     def test_clean_html(self):
1268         self.assertEqual(clean_html('a:\nb'), 'a: b')
1269         self.assertEqual(clean_html('a:\n   "b"'), 'a: "b"')
1270         self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
1271
1272     def test_intlist_to_bytes(self):
1273         self.assertEqual(
1274             intlist_to_bytes([0, 1, 127, 128, 255]),
1275             b'\x00\x01\x7f\x80\xff')
1276
1277     def test_args_to_str(self):
1278         self.assertEqual(
1279             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
1280             'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""'
1281         )
1282
1283     def test_parse_filesize(self):
1284         self.assertEqual(parse_filesize(None), None)
1285         self.assertEqual(parse_filesize(''), None)
1286         self.assertEqual(parse_filesize('91 B'), 91)
1287         self.assertEqual(parse_filesize('foobar'), None)
1288         self.assertEqual(parse_filesize('2 MiB'), 2097152)
1289         self.assertEqual(parse_filesize('5 GB'), 5000000000)
1290         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
1291         self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
1292         self.assertEqual(parse_filesize('1,24 KB'), 1240)
1293         self.assertEqual(parse_filesize('1,24 kb'), 1240)
1294         self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
1295
1296     def test_parse_count(self):
1297         self.assertEqual(parse_count(None), None)
1298         self.assertEqual(parse_count(''), None)
1299         self.assertEqual(parse_count('0'), 0)
1300         self.assertEqual(parse_count('1000'), 1000)
1301         self.assertEqual(parse_count('1.000'), 1000)
1302         self.assertEqual(parse_count('1.1k'), 1100)
1303         self.assertEqual(parse_count('1.1 k'), 1100)
1304         self.assertEqual(parse_count('1,1 k'), 1100)
1305         self.assertEqual(parse_count('1.1kk'), 1100000)
1306         self.assertEqual(parse_count('1.1kk '), 1100000)
1307         self.assertEqual(parse_count('1,1kk'), 1100000)
1308         self.assertEqual(parse_count('100 views'), 100)
1309         self.assertEqual(parse_count('1,100 views'), 1100)
1310         self.assertEqual(parse_count('1.1kk views'), 1100000)
1311         self.assertEqual(parse_count('10M views'), 10000000)
1312         self.assertEqual(parse_count('has 10M views'), 10000000)
1313
1314     def test_parse_resolution(self):
1315         self.assertEqual(parse_resolution(None), {})
1316         self.assertEqual(parse_resolution(''), {})
1317         self.assertEqual(parse_resolution(' 1920x1080'), {'width': 1920, 'height': 1080})
1318         self.assertEqual(parse_resolution('1920×1080 '), {'width': 1920, 'height': 1080})
1319         self.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080})
1320         self.assertEqual(parse_resolution('720p'), {'height': 720})
1321         self.assertEqual(parse_resolution('4k'), {'height': 2160})
1322         self.assertEqual(parse_resolution('8K'), {'height': 4320})
1323         self.assertEqual(parse_resolution('pre_1920x1080_post'), {'width': 1920, 'height': 1080})
1324         self.assertEqual(parse_resolution('ep1x2'), {})
1325         self.assertEqual(parse_resolution('1920, 1080'), {'width': 1920, 'height': 1080})
1326
1327     def test_parse_bitrate(self):
1328         self.assertEqual(parse_bitrate(None), None)
1329         self.assertEqual(parse_bitrate(''), None)
1330         self.assertEqual(parse_bitrate('300kbps'), 300)
1331         self.assertEqual(parse_bitrate('1500kbps'), 1500)
1332         self.assertEqual(parse_bitrate('300 kbps'), 300)
1333
1334     def test_version_tuple(self):
1335         self.assertEqual(version_tuple('1'), (1,))
1336         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
1337         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
1338
1339     def test_detect_exe_version(self):
1340         self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
1341 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
1342 configuration: --prefix=/usr --extra-'''), '1.2.1')
1343         self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
1344 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
1345         self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
1346 Trying to open render node...
1347 Success at /dev/dri/renderD128.
1348 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
1349
1350     def test_age_restricted(self):
1351         self.assertFalse(age_restricted(None, 10))  # unrestricted content
1352         self.assertFalse(age_restricted(1, None))  # unrestricted policy
1353         self.assertFalse(age_restricted(8, 10))
1354         self.assertTrue(age_restricted(18, 14))
1355         self.assertFalse(age_restricted(18, 18))
1356
1357     def test_is_html(self):
1358         self.assertFalse(is_html(b'\x49\x44\x43<html'))
1359         self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
1360         self.assertTrue(is_html(  # UTF-8 with BOM
1361             b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
1362         self.assertTrue(is_html(  # UTF-16-LE
1363             b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
1364         ))
1365         self.assertTrue(is_html(  # UTF-16-BE
1366             b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
1367         ))
1368         self.assertTrue(is_html(  # UTF-32-BE
1369             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'))
1370         self.assertTrue(is_html(  # UTF-32-LE
1371             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'))
1372
1373     def test_render_table(self):
1374         self.assertEqual(
1375             render_table(
1376                 ['a', 'empty', 'bcd'],
1377                 [[123, '', 4], [9999, '', 51]]),
1378             'a    empty bcd\n'
1379             '123        4\n'
1380             '9999       51')
1381
1382         self.assertEqual(
1383             render_table(
1384                 ['a', 'empty', 'bcd'],
1385                 [[123, '', 4], [9999, '', 51]],
1386                 hide_empty=True),
1387             'a    bcd\n'
1388             '123  4\n'
1389             '9999 51')
1390
1391         self.assertEqual(
1392             render_table(
1393                 ['\ta', 'bcd'],
1394                 [['1\t23', 4], ['\t9999', 51]]),
1395             '   a bcd\n'
1396             '1 23 4\n'
1397             '9999 51')
1398
1399         self.assertEqual(
1400             render_table(
1401                 ['a', 'bcd'],
1402                 [[123, 4], [9999, 51]],
1403                 delim='-'),
1404             'a    bcd\n'
1405             '--------\n'
1406             '123  4\n'
1407             '9999 51')
1408
1409         self.assertEqual(
1410             render_table(
1411                 ['a', 'bcd'],
1412                 [[123, 4], [9999, 51]],
1413                 delim='-', extra_gap=2),
1414             'a      bcd\n'
1415             '----------\n'
1416             '123    4\n'
1417             '9999   51')
1418
1419     def test_match_str(self):
1420         # Unary
1421         self.assertFalse(match_str('xy', {'x': 1200}))
1422         self.assertTrue(match_str('!xy', {'x': 1200}))
1423         self.assertTrue(match_str('x', {'x': 1200}))
1424         self.assertFalse(match_str('!x', {'x': 1200}))
1425         self.assertTrue(match_str('x', {'x': 0}))
1426         self.assertTrue(match_str('is_live', {'is_live': True}))
1427         self.assertFalse(match_str('is_live', {'is_live': False}))
1428         self.assertFalse(match_str('is_live', {'is_live': None}))
1429         self.assertFalse(match_str('is_live', {}))
1430         self.assertFalse(match_str('!is_live', {'is_live': True}))
1431         self.assertTrue(match_str('!is_live', {'is_live': False}))
1432         self.assertTrue(match_str('!is_live', {'is_live': None}))
1433         self.assertTrue(match_str('!is_live', {}))
1434         self.assertTrue(match_str('title', {'title': 'abc'}))
1435         self.assertTrue(match_str('title', {'title': ''}))
1436         self.assertFalse(match_str('!title', {'title': 'abc'}))
1437         self.assertFalse(match_str('!title', {'title': ''}))
1438
1439         # Numeric
1440         self.assertFalse(match_str('x>0', {'x': 0}))
1441         self.assertFalse(match_str('x>0', {}))
1442         self.assertTrue(match_str('x>?0', {}))
1443         self.assertTrue(match_str('x>1K', {'x': 1200}))
1444         self.assertFalse(match_str('x>2K', {'x': 1200}))
1445         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
1446         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
1447         self.assertTrue(match_str('x > 1:0:0', {'x': 3700}))
1448
1449         # String
1450         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
1451         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
1452         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
1453         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
1454         self.assertTrue(match_str('y^=foo', {'y': 'foobar42'}))
1455         self.assertFalse(match_str('y!^=foo', {'y': 'foobar42'}))
1456         self.assertFalse(match_str('y^=bar', {'y': 'foobar42'}))
1457         self.assertTrue(match_str('y!^=bar', {'y': 'foobar42'}))
1458         self.assertRaises(ValueError, match_str, 'x^=42', {'x': 42})
1459         self.assertTrue(match_str('y*=bar', {'y': 'foobar42'}))
1460         self.assertFalse(match_str('y!*=bar', {'y': 'foobar42'}))
1461         self.assertFalse(match_str('y*=baz', {'y': 'foobar42'}))
1462         self.assertTrue(match_str('y!*=baz', {'y': 'foobar42'}))
1463         self.assertTrue(match_str('y$=42', {'y': 'foobar42'}))
1464         self.assertFalse(match_str('y$=43', {'y': 'foobar42'}))
1465
1466         # And
1467         self.assertFalse(match_str(
1468             'like_count > 100 & dislike_count <? 50 & description',
1469             {'like_count': 90, 'description': 'foo'}))
1470         self.assertTrue(match_str(
1471             'like_count > 100 & dislike_count <? 50 & description',
1472             {'like_count': 190, 'description': 'foo'}))
1473         self.assertFalse(match_str(
1474             'like_count > 100 & dislike_count <? 50 & description',
1475             {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
1476         self.assertFalse(match_str(
1477             'like_count > 100 & dislike_count <? 50 & description',
1478             {'like_count': 190, 'dislike_count': 10}))
1479
1480         # Regex
1481         self.assertTrue(match_str(r'x~=\bbar', {'x': 'foo bar'}))
1482         self.assertFalse(match_str(r'x~=\bbar.+', {'x': 'foo bar'}))
1483         self.assertFalse(match_str(r'x~=^FOO', {'x': 'foo bar'}))
1484         self.assertTrue(match_str(r'x~=(?i)^FOO', {'x': 'foo bar'}))
1485
1486         # Quotes
1487         self.assertTrue(match_str(r'x^="foo"', {'x': 'foo "bar"'}))
1488         self.assertFalse(match_str(r'x^="foo  "', {'x': 'foo "bar"'}))
1489         self.assertFalse(match_str(r'x$="bar"', {'x': 'foo "bar"'}))
1490         self.assertTrue(match_str(r'x$=" \"bar\""', {'x': 'foo "bar"'}))
1491
1492         # Escaping &
1493         self.assertFalse(match_str(r'x=foo & bar', {'x': 'foo & bar'}))
1494         self.assertTrue(match_str(r'x=foo \& bar', {'x': 'foo & bar'}))
1495         self.assertTrue(match_str(r'x=foo \& bar & x^=foo', {'x': 'foo & bar'}))
1496         self.assertTrue(match_str(r'x="foo \& bar" & x^=foo', {'x': 'foo & bar'}))
1497
1498         # Example from docs
1499         self.assertTrue(match_str(
1500             r"!is_live & like_count>?100 & description~='(?i)\bcats \& dogs\b'",
1501             {'description': 'Raining Cats & Dogs'}))
1502
1503         # Incomplete
1504         self.assertFalse(match_str('id!=foo', {'id': 'foo'}, True))
1505         self.assertTrue(match_str('x', {'id': 'foo'}, True))
1506         self.assertTrue(match_str('!x', {'id': 'foo'}, True))
1507         self.assertFalse(match_str('x', {'id': 'foo'}, False))
1508
1509     def test_parse_dfxp_time_expr(self):
1510         self.assertEqual(parse_dfxp_time_expr(None), None)
1511         self.assertEqual(parse_dfxp_time_expr(''), None)
1512         self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1513         self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1514         self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1515         self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1516         self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1517
1518     def test_dfxp2srt(self):
1519         dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
1520             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1521             <body>
1522                 <div xml:lang="en">
1523                     <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1524                     <p begin="1" end="2">第二行<br/>♪♪</p>
1525                     <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1526                     <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1527                     <p begin="-1" end="-1">Ignore, two</p>
1528                     <p begin="3" dur="-1">Ignored, three</p>
1529                 </div>
1530             </body>
1531             </tt>'''.encode()
1532         srt_data = '''1
1533 00:00:00,000 --> 00:00:01,000
1534 The following line contains Chinese characters and special symbols
1535
1536 2
1537 00:00:01,000 --> 00:00:02,000
1538 第二行
1539 ♪♪
1540
1541 3
1542 00:00:02,000 --> 00:00:03,000
1543 Third
1544 Line
1545
1546 '''
1547         self.assertEqual(dfxp2srt(dfxp_data), srt_data)
1548
1549         dfxp_data_no_default_namespace = b'''<?xml version="1.0" encoding="UTF-8"?>
1550             <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1551             <body>
1552                 <div xml:lang="en">
1553                     <p begin="0" end="1">The first line</p>
1554                 </div>
1555             </body>
1556             </tt>'''
1557         srt_data = '''1
1558 00:00:00,000 --> 00:00:01,000
1559 The first line
1560
1561 '''
1562         self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
1563
1564         dfxp_data_with_style = b'''<?xml version="1.0" encoding="utf-8"?>
1565 <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">
1566   <head>
1567     <styling>
1568       <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1569       <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1570       <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1571       <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1572     </styling>
1573   </head>
1574   <body tts:textAlign="center" style="s0">
1575     <div>
1576       <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1577       <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>
1578       <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1579       <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>
1580     </div>
1581   </body>
1582 </tt>'''
1583         srt_data = '''1
1584 00:00:02,080 --> 00:00:05,840
1585 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1586
1587 2
1588 00:00:02,080 --> 00:00:05,840
1589 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1590 </font>part 2</font></b>
1591
1592 3
1593 00:00:05,840 --> 00:00:09,560
1594 <u><font color="lime">line 3
1595 part 3</font></u>
1596
1597 4
1598 00:00:09,560 --> 00:00:12,360
1599 <i><u><font color="yellow"><font color="lime">inner
1600  </font>style</font></u></i>
1601
1602 '''
1603         self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)
1604
1605         dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?>
1606             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1607             <body>
1608                 <div xml:lang="en">
1609                     <p begin="0" end="1">Line 1</p>
1610                     <p begin="1" end="2">第二行</p>
1611                 </div>
1612             </body>
1613             </tt>'''.encode('utf-16')
1614         srt_data = '''1
1615 00:00:00,000 --> 00:00:01,000
1616 Line 1
1617
1618 2
1619 00:00:01,000 --> 00:00:02,000
1620 第二行
1621
1622 '''
1623         self.assertEqual(dfxp2srt(dfxp_data_non_utf8), srt_data)
1624
1625     def test_cli_option(self):
1626         self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1627         self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
1628         self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1629         self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
1630
1631     def test_cli_valueless_option(self):
1632         self.assertEqual(cli_valueless_option(
1633             {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1634         self.assertEqual(cli_valueless_option(
1635             {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
1636         self.assertEqual(cli_valueless_option(
1637             {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1638         self.assertEqual(cli_valueless_option(
1639             {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
1640         self.assertEqual(cli_valueless_option(
1641             {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
1642         self.assertEqual(cli_valueless_option(
1643             {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1644
1645     def test_cli_bool_option(self):
1646         self.assertEqual(
1647             cli_bool_option(
1648                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1649             ['--no-check-certificate', 'true'])
1650         self.assertEqual(
1651             cli_bool_option(
1652                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1653             ['--no-check-certificate=true'])
1654         self.assertEqual(
1655             cli_bool_option(
1656                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1657             ['--check-certificate', 'false'])
1658         self.assertEqual(
1659             cli_bool_option(
1660                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1661             ['--check-certificate=false'])
1662         self.assertEqual(
1663             cli_bool_option(
1664                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1665             ['--check-certificate', 'true'])
1666         self.assertEqual(
1667             cli_bool_option(
1668                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1669             ['--check-certificate=true'])
1670         self.assertEqual(
1671             cli_bool_option(
1672                 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1673             [])
1674
1675     def test_ohdave_rsa_encrypt(self):
1676         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1677         e = 65537
1678
1679         self.assertEqual(
1680             ohdave_rsa_encrypt(b'aa111222', e, N),
1681             '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1682
1683     def test_pkcs1pad(self):
1684         data = [1, 2, 3]
1685         padded_data = pkcs1pad(data, 32)
1686         self.assertEqual(padded_data[:2], [0, 2])
1687         self.assertEqual(padded_data[28:], [0, 1, 2, 3])
1688
1689         self.assertRaises(ValueError, pkcs1pad, data, 8)
1690
1691     def test_encode_base_n(self):
1692         self.assertEqual(encode_base_n(0, 30), '0')
1693         self.assertEqual(encode_base_n(80, 30), '2k')
1694
1695         custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1696         self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1697         self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1698
1699         self.assertRaises(ValueError, encode_base_n, 0, 70)
1700         self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1701
1702     def test_caesar(self):
1703         self.assertEqual(caesar('ace', 'abcdef', 2), 'cea')
1704         self.assertEqual(caesar('cea', 'abcdef', -2), 'ace')
1705         self.assertEqual(caesar('ace', 'abcdef', -2), 'eac')
1706         self.assertEqual(caesar('eac', 'abcdef', 2), 'ace')
1707         self.assertEqual(caesar('ace', 'abcdef', 0), 'ace')
1708         self.assertEqual(caesar('xyz', 'abcdef', 2), 'xyz')
1709         self.assertEqual(caesar('abc', 'acegik', 2), 'ebg')
1710         self.assertEqual(caesar('ebg', 'acegik', -2), 'abc')
1711
1712     def test_rot47(self):
1713         self.assertEqual(rot47('yt-dlp'), r'JE\5=A')
1714         self.assertEqual(rot47('YT-DLP'), r'*%\s{!')
1715
1716     def test_urshift(self):
1717         self.assertEqual(urshift(3, 1), 1)
1718         self.assertEqual(urshift(-3, 1), 2147483646)
1719
1720     GET_ELEMENT_BY_CLASS_TEST_STRING = '''
1721         <span class="foo bar">nice</span>
1722     '''
1723
1724     def test_get_element_by_class(self):
1725         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1726
1727         self.assertEqual(get_element_by_class('foo', html), 'nice')
1728         self.assertEqual(get_element_by_class('no-such-class', html), None)
1729
1730     def test_get_element_html_by_class(self):
1731         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1732
1733         self.assertEqual(get_element_html_by_class('foo', html), html.strip())
1734         self.assertEqual(get_element_by_class('no-such-class', html), None)
1735
1736     GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING = '''
1737         <div itemprop="author" itemscope>foo</div>
1738     '''
1739
1740     def test_get_element_by_attribute(self):
1741         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1742
1743         self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
1744         self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
1745         self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
1746
1747         html = self.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
1748
1749         self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo')
1750
1751     def test_get_element_html_by_attribute(self):
1752         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1753
1754         self.assertEqual(get_element_html_by_attribute('class', 'foo bar', html), html.strip())
1755         self.assertEqual(get_element_html_by_attribute('class', 'foo', html), None)
1756         self.assertEqual(get_element_html_by_attribute('class', 'no-such-foo', html), None)
1757
1758         html = self.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
1759
1760         self.assertEqual(get_element_html_by_attribute('itemprop', 'author', html), html.strip())
1761
1762     GET_ELEMENTS_BY_CLASS_TEST_STRING = '''
1763         <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1764     '''
1765     GET_ELEMENTS_BY_CLASS_RES = ['<span class="foo bar">nice</span>', '<span class="foo bar">also nice</span>']
1766
1767     def test_get_elements_by_class(self):
1768         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1769
1770         self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
1771         self.assertEqual(get_elements_by_class('no-such-class', html), [])
1772
1773     def test_get_elements_html_by_class(self):
1774         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1775
1776         self.assertEqual(get_elements_html_by_class('foo', html), self.GET_ELEMENTS_BY_CLASS_RES)
1777         self.assertEqual(get_elements_html_by_class('no-such-class', html), [])
1778
1779     def test_get_elements_by_attribute(self):
1780         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1781
1782         self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
1783         self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
1784         self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
1785
1786     def test_get_elements_html_by_attribute(self):
1787         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1788
1789         self.assertEqual(get_elements_html_by_attribute('class', 'foo bar', html), self.GET_ELEMENTS_BY_CLASS_RES)
1790         self.assertEqual(get_elements_html_by_attribute('class', 'foo', html), [])
1791         self.assertEqual(get_elements_html_by_attribute('class', 'no-such-foo', html), [])
1792
1793     def test_get_elements_text_and_html_by_attribute(self):
1794         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1795
1796         self.assertEqual(
1797             list(get_elements_text_and_html_by_attribute('class', 'foo bar', html)),
1798             list(zip(['nice', 'also nice'], self.GET_ELEMENTS_BY_CLASS_RES)))
1799         self.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'foo', html)), [])
1800         self.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'no-such-foo', html)), [])
1801
1802         self.assertEqual(list(get_elements_text_and_html_by_attribute(
1803             'class', 'foo', '<a class="foo">nice</a><span class="foo">nice</span>', tag='a')), [('nice', '<a class="foo">nice</a>')])
1804
1805     GET_ELEMENT_BY_TAG_TEST_STRING = '''
1806     random text lorem ipsum</p>
1807     <div>
1808         this should be returned
1809         <span>this should also be returned</span>
1810         <div>
1811             this should also be returned
1812         </div>
1813         closing tag above should not trick, so this should also be returned
1814     </div>
1815     but this text should not be returned
1816     '''
1817     GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML = GET_ELEMENT_BY_TAG_TEST_STRING.strip()[32:276]
1818     GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT = GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML[5:-6]
1819     GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML = GET_ELEMENT_BY_TAG_TEST_STRING.strip()[78:119]
1820     GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT = GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML[6:-7]
1821
1822     def test_get_element_text_and_html_by_tag(self):
1823         html = self.GET_ELEMENT_BY_TAG_TEST_STRING
1824
1825         self.assertEqual(
1826             get_element_text_and_html_by_tag('div', html),
1827             (self.GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT, self.GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML))
1828         self.assertEqual(
1829             get_element_text_and_html_by_tag('span', html),
1830             (self.GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT, self.GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML))
1831         self.assertRaises(compat_HTMLParseError, get_element_text_and_html_by_tag, 'article', html)
1832
1833     def test_iri_to_uri(self):
1834         self.assertEqual(
1835             iri_to_uri('https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b'),
1836             'https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b')  # Same
1837         self.assertEqual(
1838             iri_to_uri('https://www.google.com/search?q=Käsesoßenrührlöffel'),  # German for cheese sauce stirring spoon
1839             'https://www.google.com/search?q=K%C3%A4seso%C3%9Fenr%C3%BChrl%C3%B6ffel')
1840         self.assertEqual(
1841             iri_to_uri('https://www.google.com/search?q=lt<+gt>+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#'),
1842             'https://www.google.com/search?q=lt%3C+gt%3E+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#')
1843         self.assertEqual(
1844             iri_to_uri('http://правозащита38.рф/category/news/'),
1845             'http://xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/')
1846         self.assertEqual(
1847             iri_to_uri('http://www.правозащита38.рф/category/news/'),
1848             'http://www.xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/')
1849         self.assertEqual(
1850             iri_to_uri('https://i❤.ws/emojidomain/👍👏🤝💪'),
1851             'https://xn--i-7iq.ws/emojidomain/%F0%9F%91%8D%F0%9F%91%8F%F0%9F%A4%9D%F0%9F%92%AA')
1852         self.assertEqual(
1853             iri_to_uri('http://日本語.jp/'),
1854             'http://xn--wgv71a119e.jp/')
1855         self.assertEqual(
1856             iri_to_uri('http://导航.中国/'),
1857             'http://xn--fet810g.xn--fiqs8s/')
1858
1859     def test_clean_podcast_url(self):
1860         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')
1861         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')
1862         self.assertEqual(clean_podcast_url('https://pdst.fm/e/2.gum.fm/chtbl.com/track/chrt.fm/track/34D33/pscrb.fm/rss/p/traffic.megaphone.fm/ITLLC7765286967.mp3?updated=1687282661'), 'https://traffic.megaphone.fm/ITLLC7765286967.mp3?updated=1687282661')
1863         self.assertEqual(clean_podcast_url('https://pdst.fm/e/https://mgln.ai/e/441/www.buzzsprout.com/1121972/13019085-ep-252-the-deep-life-stack.mp3'), 'https://www.buzzsprout.com/1121972/13019085-ep-252-the-deep-life-stack.mp3')
1864
1865     def test_LazyList(self):
1866         it = list(range(10))
1867
1868         self.assertEqual(list(LazyList(it)), it)
1869         self.assertEqual(LazyList(it).exhaust(), it)
1870         self.assertEqual(LazyList(it)[5], it[5])
1871
1872         self.assertEqual(LazyList(it)[5:], it[5:])
1873         self.assertEqual(LazyList(it)[:5], it[:5])
1874         self.assertEqual(LazyList(it)[::2], it[::2])
1875         self.assertEqual(LazyList(it)[1::2], it[1::2])
1876         self.assertEqual(LazyList(it)[5::-1], it[5::-1])
1877         self.assertEqual(LazyList(it)[6:2:-2], it[6:2:-2])
1878         self.assertEqual(LazyList(it)[::-1], it[::-1])
1879
1880         self.assertTrue(LazyList(it))
1881         self.assertFalse(LazyList(range(0)))
1882         self.assertEqual(len(LazyList(it)), len(it))
1883         self.assertEqual(repr(LazyList(it)), repr(it))
1884         self.assertEqual(str(LazyList(it)), str(it))
1885
1886         self.assertEqual(list(LazyList(it, reverse=True)), it[::-1])
1887         self.assertEqual(list(reversed(LazyList(it))[::-1]), it)
1888         self.assertEqual(list(reversed(LazyList(it))[1:3:7]), it[::-1][1:3:7])
1889
1890     def test_LazyList_laziness(self):
1891
1892         def test(ll, idx, val, cache):
1893             self.assertEqual(ll[idx], val)
1894             self.assertEqual(ll._cache, list(cache))
1895
1896         ll = LazyList(range(10))
1897         test(ll, 0, 0, range(1))
1898         test(ll, 5, 5, range(6))
1899         test(ll, -3, 7, range(10))
1900
1901         ll = LazyList(range(10), reverse=True)
1902         test(ll, -1, 0, range(1))
1903         test(ll, 3, 6, range(10))
1904
1905         ll = LazyList(itertools.count())
1906         test(ll, 10, 10, range(11))
1907         ll = reversed(ll)
1908         test(ll, -15, 14, range(15))
1909
1910     def test_format_bytes(self):
1911         self.assertEqual(format_bytes(0), '0.00B')
1912         self.assertEqual(format_bytes(1000), '1000.00B')
1913         self.assertEqual(format_bytes(1024), '1.00KiB')
1914         self.assertEqual(format_bytes(1024**2), '1.00MiB')
1915         self.assertEqual(format_bytes(1024**3), '1.00GiB')
1916         self.assertEqual(format_bytes(1024**4), '1.00TiB')
1917         self.assertEqual(format_bytes(1024**5), '1.00PiB')
1918         self.assertEqual(format_bytes(1024**6), '1.00EiB')
1919         self.assertEqual(format_bytes(1024**7), '1.00ZiB')
1920         self.assertEqual(format_bytes(1024**8), '1.00YiB')
1921         self.assertEqual(format_bytes(1024**9), '1024.00YiB')
1922
1923     def test_hide_login_info(self):
1924         self.assertEqual(Config.hide_login_info(['-u', 'foo', '-p', 'bar']),
1925                          ['-u', 'PRIVATE', '-p', 'PRIVATE'])
1926         self.assertEqual(Config.hide_login_info(['-u']), ['-u'])
1927         self.assertEqual(Config.hide_login_info(['-u', 'foo', '-u', 'bar']),
1928                          ['-u', 'PRIVATE', '-u', 'PRIVATE'])
1929         self.assertEqual(Config.hide_login_info(['--username=foo']),
1930                          ['--username=PRIVATE'])
1931
1932     def test_locked_file(self):
1933         TEXT = 'test_locked_file\n'
1934         FILE = 'test_locked_file.ytdl'
1935         MODES = 'war'  # Order is important
1936
1937         try:
1938             for lock_mode in MODES:
1939                 with locked_file(FILE, lock_mode, False) as f:
1940                     if lock_mode == 'r':
1941                         self.assertEqual(f.read(), TEXT * 2, 'Wrong file content')
1942                     else:
1943                         f.write(TEXT)
1944                     for test_mode in MODES:
1945                         testing_write = test_mode != 'r'
1946                         try:
1947                             with locked_file(FILE, test_mode, False):
1948                                 pass
1949                         except (BlockingIOError, PermissionError):
1950                             if not testing_write:  # FIXME
1951                                 print(f'Known issue: Exclusive lock ({lock_mode}) blocks read access ({test_mode})')
1952                                 continue
1953                             self.assertTrue(testing_write, f'{test_mode} is blocked by {lock_mode}')
1954                         else:
1955                             self.assertFalse(testing_write, f'{test_mode} is not blocked by {lock_mode}')
1956         finally:
1957             with contextlib.suppress(OSError):
1958                 os.remove(FILE)
1959
1960     def test_determine_file_encoding(self):
1961         self.assertEqual(determine_file_encoding(b''), (None, 0))
1962         self.assertEqual(determine_file_encoding(b'--verbose -x --audio-format mkv\n'), (None, 0))
1963
1964         self.assertEqual(determine_file_encoding(b'\xef\xbb\xbf'), ('utf-8', 3))
1965         self.assertEqual(determine_file_encoding(b'\x00\x00\xfe\xff'), ('utf-32-be', 4))
1966         self.assertEqual(determine_file_encoding(b'\xff\xfe'), ('utf-16-le', 2))
1967
1968         self.assertEqual(determine_file_encoding(b'\xff\xfe# coding: utf-8\n--verbose'), ('utf-16-le', 2))
1969
1970         self.assertEqual(determine_file_encoding(b'# coding: utf-8\n--verbose'), ('utf-8', 0))
1971         self.assertEqual(determine_file_encoding(b'# coding: someencodinghere-12345\n--verbose'), ('someencodinghere-12345', 0))
1972
1973         self.assertEqual(determine_file_encoding(b'#coding:utf-8\n--verbose'), ('utf-8', 0))
1974         self.assertEqual(determine_file_encoding(b'#  coding:   utf-8   \r\n--verbose'), ('utf-8', 0))
1975
1976         self.assertEqual(determine_file_encoding('# coding: utf-32-be'.encode('utf-32-be')), ('utf-32-be', 0))
1977         self.assertEqual(determine_file_encoding('# coding: utf-16-le'.encode('utf-16-le')), ('utf-16-le', 0))
1978
1979     def test_get_compatible_ext(self):
1980         self.assertEqual(get_compatible_ext(
1981             vcodecs=[None], acodecs=[None, None], vexts=['mp4'], aexts=['m4a', 'm4a']), 'mkv')
1982         self.assertEqual(get_compatible_ext(
1983             vcodecs=[None], acodecs=[None], vexts=['flv'], aexts=['flv']), 'flv')
1984
1985         self.assertEqual(get_compatible_ext(
1986             vcodecs=[None], acodecs=[None], vexts=['mp4'], aexts=['m4a']), 'mp4')
1987         self.assertEqual(get_compatible_ext(
1988             vcodecs=[None], acodecs=[None], vexts=['mp4'], aexts=['webm']), 'mkv')
1989         self.assertEqual(get_compatible_ext(
1990             vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['m4a']), 'mkv')
1991         self.assertEqual(get_compatible_ext(
1992             vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['webm']), 'webm')
1993         self.assertEqual(get_compatible_ext(
1994             vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['weba']), 'webm')
1995
1996         self.assertEqual(get_compatible_ext(
1997             vcodecs=['h264'], acodecs=['mp4a'], vexts=['mov'], aexts=['m4a']), 'mp4')
1998         self.assertEqual(get_compatible_ext(
1999             vcodecs=['av01.0.12M.08'], acodecs=['opus'], vexts=['mp4'], aexts=['webm']), 'webm')
2000
2001         self.assertEqual(get_compatible_ext(
2002             vcodecs=['vp9'], acodecs=['opus'], vexts=['webm'], aexts=['webm'], preferences=['flv', 'mp4']), 'mp4')
2003         self.assertEqual(get_compatible_ext(
2004             vcodecs=['av1'], acodecs=['mp4a'], vexts=['webm'], aexts=['m4a'], preferences=('webm', 'mkv')), 'mkv')
2005
2006     def test_try_call(self):
2007         def total(*x, **kwargs):
2008             return sum(x) + sum(kwargs.values())
2009
2010         self.assertEqual(try_call(None), None,
2011                          msg='not a fn should give None')
2012         self.assertEqual(try_call(lambda: 1), 1,
2013                          msg='int fn with no expected_type should give int')
2014         self.assertEqual(try_call(lambda: 1, expected_type=int), 1,
2015                          msg='int fn with expected_type int should give int')
2016         self.assertEqual(try_call(lambda: 1, expected_type=dict), None,
2017                          msg='int fn with wrong expected_type should give None')
2018         self.assertEqual(try_call(total, args=(0, 1, 0, ), expected_type=int), 1,
2019                          msg='fn should accept arglist')
2020         self.assertEqual(try_call(total, kwargs={'a': 0, 'b': 1, 'c': 0}, expected_type=int), 1,
2021                          msg='fn should accept kwargs')
2022         self.assertEqual(try_call(lambda: 1, expected_type=dict), None,
2023                          msg='int fn with no expected_type should give None')
2024         self.assertEqual(try_call(lambda x: {}, total, args=(42, ), expected_type=int), 42,
2025                          msg='expect first int result with expected_type int')
2026
2027     def test_variadic(self):
2028         self.assertEqual(variadic(None), (None, ))
2029         self.assertEqual(variadic('spam'), ('spam', ))
2030         self.assertEqual(variadic('spam', allowed_types=dict), 'spam')
2031         with warnings.catch_warnings():
2032             warnings.simplefilter('ignore')
2033             self.assertEqual(variadic('spam', allowed_types=[dict]), 'spam')
2034
2035     def test_traverse_obj(self):
2036         _TEST_DATA = {
2037             100: 100,
2038             1.2: 1.2,
2039             'str': 'str',
2040             'None': None,
2041             '...': ...,
2042             'urls': [
2043                 {'index': 0, 'url': 'https://www.example.com/0'},
2044                 {'index': 1, 'url': 'https://www.example.com/1'},
2045             ],
2046             'data': (
2047                 {'index': 2},
2048                 {'index': 3},
2049             ),
2050             'dict': {},
2051         }
2052
2053         # Test base functionality
2054         self.assertEqual(traverse_obj(_TEST_DATA, ('str',)), 'str',
2055                          msg='allow tuple path')
2056         self.assertEqual(traverse_obj(_TEST_DATA, ['str']), 'str',
2057                          msg='allow list path')
2058         self.assertEqual(traverse_obj(_TEST_DATA, (value for value in ("str",))), 'str',
2059                          msg='allow iterable path')
2060         self.assertEqual(traverse_obj(_TEST_DATA, 'str'), 'str',
2061                          msg='single items should be treated as a path')
2062         self.assertEqual(traverse_obj(_TEST_DATA, None), _TEST_DATA)
2063         self.assertEqual(traverse_obj(_TEST_DATA, 100), 100)
2064         self.assertEqual(traverse_obj(_TEST_DATA, 1.2), 1.2)
2065
2066         # Test Ellipsis behavior
2067         self.assertCountEqual(traverse_obj(_TEST_DATA, ...),
2068                               (item for item in _TEST_DATA.values() if item not in (None, {})),
2069                               msg='`...` should give all non discarded values')
2070         self.assertCountEqual(traverse_obj(_TEST_DATA, ('urls', 0, ...)), _TEST_DATA['urls'][0].values(),
2071                               msg='`...` selection for dicts should select all values')
2072         self.assertEqual(traverse_obj(_TEST_DATA, (..., ..., 'url')),
2073                          ['https://www.example.com/0', 'https://www.example.com/1'],
2074                          msg='nested `...` queries should work')
2075         self.assertCountEqual(traverse_obj(_TEST_DATA, (..., ..., 'index')), range(4),
2076                               msg='`...` query result should be flattened')
2077         self.assertEqual(traverse_obj(iter(range(4)), ...), list(range(4)),
2078                          msg='`...` should accept iterables')
2079
2080         # Test function as key
2081         self.assertEqual(traverse_obj(_TEST_DATA, lambda x, y: x == 'urls' and isinstance(y, list)),
2082                          [_TEST_DATA['urls']],
2083                          msg='function as query key should perform a filter based on (key, value)')
2084         self.assertCountEqual(traverse_obj(_TEST_DATA, lambda _, x: isinstance(x[0], str)), {'str'},
2085                               msg='exceptions in the query function should be catched')
2086         self.assertEqual(traverse_obj(iter(range(4)), lambda _, x: x % 2 == 0), [0, 2],
2087                          msg='function key should accept iterables')
2088         if __debug__:
2089             with self.assertRaises(Exception, msg='Wrong function signature should raise in debug'):
2090                 traverse_obj(_TEST_DATA, lambda a: ...)
2091             with self.assertRaises(Exception, msg='Wrong function signature should raise in debug'):
2092                 traverse_obj(_TEST_DATA, lambda a, b, c: ...)
2093
2094         # Test set as key (transformation/type, like `expected_type`)
2095         self.assertEqual(traverse_obj(_TEST_DATA, (..., {str.upper}, )), ['STR'],
2096                          msg='Function in set should be a transformation')
2097         self.assertEqual(traverse_obj(_TEST_DATA, (..., {str})), ['str'],
2098                          msg='Type in set should be a type filter')
2099         self.assertEqual(traverse_obj(_TEST_DATA, {dict}), _TEST_DATA,
2100                          msg='A single set should be wrapped into a path')
2101         self.assertEqual(traverse_obj(_TEST_DATA, (..., {str.upper})), ['STR'],
2102                          msg='Transformation function should not raise')
2103         self.assertEqual(traverse_obj(_TEST_DATA, (..., {str_or_none})),
2104                          [item for item in map(str_or_none, _TEST_DATA.values()) if item is not None],
2105                          msg='Function in set should be a transformation')
2106         if __debug__:
2107             with self.assertRaises(Exception, msg='Sets with length != 1 should raise in debug'):
2108                 traverse_obj(_TEST_DATA, set())
2109             with self.assertRaises(Exception, msg='Sets with length != 1 should raise in debug'):
2110                 traverse_obj(_TEST_DATA, {str.upper, str})
2111
2112         # Test `slice` as a key
2113         _SLICE_DATA = [0, 1, 2, 3, 4]
2114         self.assertEqual(traverse_obj(_TEST_DATA, ('dict', slice(1))), None,
2115                          msg='slice on a dictionary should not throw')
2116         self.assertEqual(traverse_obj(_SLICE_DATA, slice(1)), _SLICE_DATA[:1],
2117                          msg='slice key should apply slice to sequence')
2118         self.assertEqual(traverse_obj(_SLICE_DATA, slice(1, 2)), _SLICE_DATA[1:2],
2119                          msg='slice key should apply slice to sequence')
2120         self.assertEqual(traverse_obj(_SLICE_DATA, slice(1, 4, 2)), _SLICE_DATA[1:4:2],
2121                          msg='slice key should apply slice to sequence')
2122
2123         # Test alternative paths
2124         self.assertEqual(traverse_obj(_TEST_DATA, 'fail', 'str'), 'str',
2125                          msg='multiple `paths` should be treated as alternative paths')
2126         self.assertEqual(traverse_obj(_TEST_DATA, 'str', 100), 'str',
2127                          msg='alternatives should exit early')
2128         self.assertEqual(traverse_obj(_TEST_DATA, 'fail', 'fail'), None,
2129                          msg='alternatives should return `default` if exhausted')
2130         self.assertEqual(traverse_obj(_TEST_DATA, (..., 'fail'), 100), 100,
2131                          msg='alternatives should track their own branching return')
2132         self.assertEqual(traverse_obj(_TEST_DATA, ('dict', ...), ('data', ...)), list(_TEST_DATA['data']),
2133                          msg='alternatives on empty objects should search further')
2134
2135         # Test branch and path nesting
2136         self.assertEqual(traverse_obj(_TEST_DATA, ('urls', (3, 0), 'url')), ['https://www.example.com/0'],
2137                          msg='tuple as key should be treated as branches')
2138         self.assertEqual(traverse_obj(_TEST_DATA, ('urls', [3, 0], 'url')), ['https://www.example.com/0'],
2139                          msg='list as key should be treated as branches')
2140         self.assertEqual(traverse_obj(_TEST_DATA, ('urls', ((1, 'fail'), (0, 'url')))), ['https://www.example.com/0'],
2141                          msg='double nesting in path should be treated as paths')
2142         self.assertEqual(traverse_obj(['0', [1, 2]], [(0, 1), 0]), [1],
2143                          msg='do not fail early on branching')
2144         self.assertCountEqual(traverse_obj(_TEST_DATA, ('urls', ((1, ('fail', 'url')), (0, 'url')))),
2145                               ['https://www.example.com/0', 'https://www.example.com/1'],
2146                               msg='tripple nesting in path should be treated as branches')
2147         self.assertEqual(traverse_obj(_TEST_DATA, ('urls', ('fail', (..., 'url')))),
2148                          ['https://www.example.com/0', 'https://www.example.com/1'],
2149                          msg='ellipsis as branch path start gets flattened')
2150
2151         # Test dictionary as key
2152         self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2}), {0: 100, 1: 1.2},
2153                          msg='dict key should result in a dict with the same keys')
2154         self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', 0, 'url')}),
2155                          {0: 'https://www.example.com/0'},
2156                          msg='dict key should allow paths')
2157         self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', (3, 0), 'url')}),
2158                          {0: ['https://www.example.com/0']},
2159                          msg='tuple in dict path should be treated as branches')
2160         self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', ((1, 'fail'), (0, 'url')))}),
2161                          {0: ['https://www.example.com/0']},
2162                          msg='double nesting in dict path should be treated as paths')
2163         self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', ((1, ('fail', 'url')), (0, 'url')))}),
2164                          {0: ['https://www.example.com/1', 'https://www.example.com/0']},
2165                          msg='tripple nesting in dict path should be treated as branches')
2166         self.assertEqual(traverse_obj(_TEST_DATA, {0: 'fail'}), {},
2167                          msg='remove `None` values when top level dict key fails')
2168         self.assertEqual(traverse_obj(_TEST_DATA, {0: 'fail'}, default=...), {0: ...},
2169                          msg='use `default` if key fails and `default`')
2170         self.assertEqual(traverse_obj(_TEST_DATA, {0: 'dict'}), {},
2171                          msg='remove empty values when dict key')
2172         self.assertEqual(traverse_obj(_TEST_DATA, {0: 'dict'}, default=...), {0: ...},
2173                          msg='use `default` when dict key and `default`')
2174         self.assertEqual(traverse_obj(_TEST_DATA, {0: {0: 'fail'}}), {},
2175                          msg='remove empty values when nested dict key fails')
2176         self.assertEqual(traverse_obj(None, {0: 'fail'}), {},
2177                          msg='default to dict if pruned')
2178         self.assertEqual(traverse_obj(None, {0: 'fail'}, default=...), {0: ...},
2179                          msg='default to dict if pruned and default is given')
2180         self.assertEqual(traverse_obj(_TEST_DATA, {0: {0: 'fail'}}, default=...), {0: {0: ...}},
2181                          msg='use nested `default` when nested dict key fails and `default`')
2182         self.assertEqual(traverse_obj(_TEST_DATA, {0: ('dict', ...)}), {},
2183                          msg='remove key if branch in dict key not successful')
2184
2185         # Testing default parameter behavior
2186         _DEFAULT_DATA = {'None': None, 'int': 0, 'list': []}
2187         self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail'), None,
2188                          msg='default value should be `None`')
2189         self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail', 'fail', default=...), ...,
2190                          msg='chained fails should result in default')
2191         self.assertEqual(traverse_obj(_DEFAULT_DATA, 'None', 'int'), 0,
2192                          msg='should not short cirquit on `None`')
2193         self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail', default=1), 1,
2194                          msg='invalid dict key should result in `default`')
2195         self.assertEqual(traverse_obj(_DEFAULT_DATA, 'None', default=1), 1,
2196                          msg='`None` is a deliberate sentinel and should become `default`')
2197         self.assertEqual(traverse_obj(_DEFAULT_DATA, ('list', 10)), None,
2198                          msg='`IndexError` should result in `default`')
2199         self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail'), default=1), 1,
2200                          msg='if branched but not successful return `default` if defined, not `[]`')
2201         self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail'), default=None), None,
2202                          msg='if branched but not successful return `default` even if `default` is `None`')
2203         self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail')), [],
2204                          msg='if branched but not successful return `[]`, not `default`')
2205         self.assertEqual(traverse_obj(_DEFAULT_DATA, ('list', ...)), [],
2206                          msg='if branched but object is empty return `[]`, not `default`')
2207         self.assertEqual(traverse_obj(None, ...), [],
2208                          msg='if branched but object is `None` return `[]`, not `default`')
2209         self.assertEqual(traverse_obj({0: None}, (0, ...)), [],
2210                          msg='if branched but state is `None` return `[]`, not `default`')
2211
2212         branching_paths = [
2213             ('fail', ...),
2214             (..., 'fail'),
2215             100 * ('fail',) + (...,),
2216             (...,) + 100 * ('fail',),
2217         ]
2218         for branching_path in branching_paths:
2219             self.assertEqual(traverse_obj({}, branching_path), [],
2220                              msg='if branched but state is `None`, return `[]` (not `default`)')
2221             self.assertEqual(traverse_obj({}, 'fail', branching_path), [],
2222                              msg='if branching in last alternative and previous did not match, return `[]` (not `default`)')
2223             self.assertEqual(traverse_obj({0: 'x'}, 0, branching_path), 'x',
2224                              msg='if branching in last alternative and previous did match, return single value')
2225             self.assertEqual(traverse_obj({0: 'x'}, branching_path, 0), 'x',
2226                              msg='if branching in first alternative and non-branching path does match, return single value')
2227             self.assertEqual(traverse_obj({}, branching_path, 'fail'), None,
2228                              msg='if branching in first alternative and non-branching path does not match, return `default`')
2229
2230         # Testing expected_type behavior
2231         _EXPECTED_TYPE_DATA = {'str': 'str', 'int': 0}
2232         self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=str),
2233                          'str', msg='accept matching `expected_type` type')
2234         self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=int),
2235                          None, msg='reject non matching `expected_type` type')
2236         self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'int', expected_type=lambda x: str(x)),
2237                          '0', msg='transform type using type function')
2238         self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=lambda _: 1 / 0),
2239                          None, msg='wrap expected_type fuction in try_call')
2240         self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, ..., expected_type=str),
2241                          ['str'], msg='eliminate items that expected_type fails on')
2242         self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2}, expected_type=int),
2243                          {0: 100}, msg='type as expected_type should filter dict values')
2244         self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2, 2: 'None'}, expected_type=str_or_none),
2245                          {0: '100', 1: '1.2'}, msg='function as expected_type should transform dict values')
2246         self.assertEqual(traverse_obj(_TEST_DATA, ({0: 1.2}, 0, {int_or_none}), expected_type=int),
2247                          1, msg='expected_type should not filter non final dict values')
2248         self.assertEqual(traverse_obj(_TEST_DATA, {0: {0: 100, 1: 'str'}}, expected_type=int),
2249                          {0: {0: 100}}, msg='expected_type should transform deep dict values')
2250         self.assertEqual(traverse_obj(_TEST_DATA, [({0: '...'}, {0: '...'})], expected_type=type(...)),
2251                          [{0: ...}, {0: ...}], msg='expected_type should transform branched dict values')
2252         self.assertEqual(traverse_obj({1: {3: 4}}, [(1, 2), 3], expected_type=int),
2253                          [4], msg='expected_type regression for type matching in tuple branching')
2254         self.assertEqual(traverse_obj(_TEST_DATA, ['data', ...], expected_type=int),
2255                          [], msg='expected_type regression for type matching in dict result')
2256
2257         # Test get_all behavior
2258         _GET_ALL_DATA = {'key': [0, 1, 2]}
2259         self.assertEqual(traverse_obj(_GET_ALL_DATA, ('key', ...), get_all=False), 0,
2260                          msg='if not `get_all`, return only first matching value')
2261         self.assertEqual(traverse_obj(_GET_ALL_DATA, ..., get_all=False), [0, 1, 2],
2262                          msg='do not overflatten if not `get_all`')
2263
2264         # Test casesense behavior
2265         _CASESENSE_DATA = {
2266             'KeY': 'value0',
2267             0: {
2268                 'KeY': 'value1',
2269                 0: {'KeY': 'value2'},
2270             },
2271         }
2272         self.assertEqual(traverse_obj(_CASESENSE_DATA, 'key'), None,
2273                          msg='dict keys should be case sensitive unless `casesense`')
2274         self.assertEqual(traverse_obj(_CASESENSE_DATA, 'keY',
2275                                       casesense=False), 'value0',
2276                          msg='allow non matching key case if `casesense`')
2277         self.assertEqual(traverse_obj(_CASESENSE_DATA, (0, ('keY',)),
2278                                       casesense=False), ['value1'],
2279                          msg='allow non matching key case in branch if `casesense`')
2280         self.assertEqual(traverse_obj(_CASESENSE_DATA, (0, ((0, 'keY'),)),
2281                                       casesense=False), ['value2'],
2282                          msg='allow non matching key case in branch path if `casesense`')
2283
2284         # Test traverse_string behavior
2285         _TRAVERSE_STRING_DATA = {'str': 'str', 1.2: 1.2}
2286         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', 0)), None,
2287                          msg='do not traverse into string if not `traverse_string`')
2288         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', 0),
2289                                       traverse_string=True), 's',
2290                          msg='traverse into string if `traverse_string`')
2291         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, (1.2, 1),
2292                                       traverse_string=True), '.',
2293                          msg='traverse into converted data if `traverse_string`')
2294         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', ...),
2295                                       traverse_string=True), 'str',
2296                          msg='`...` should result in string (same value) if `traverse_string`')
2297         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', slice(0, None, 2)),
2298                                       traverse_string=True), 'sr',
2299                          msg='`slice` should result in string if `traverse_string`')
2300         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', lambda i, v: i or v == "s"),
2301                                       traverse_string=True), 'str',
2302                          msg='function should result in string if `traverse_string`')
2303         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', (0, 2)),
2304                                       traverse_string=True), ['s', 'r'],
2305                          msg='branching should result in list if `traverse_string`')
2306         self.assertEqual(traverse_obj({}, (0, ...), traverse_string=True), [],
2307                          msg='branching should result in list if `traverse_string`')
2308         self.assertEqual(traverse_obj({}, (0, lambda x, y: True), traverse_string=True), [],
2309                          msg='branching should result in list if `traverse_string`')
2310         self.assertEqual(traverse_obj({}, (0, slice(1)), traverse_string=True), [],
2311                          msg='branching should result in list if `traverse_string`')
2312
2313         # Test is_user_input behavior
2314         _IS_USER_INPUT_DATA = {'range8': list(range(8))}
2315         self.assertEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', '3'),
2316                                       is_user_input=True), 3,
2317                          msg='allow for string indexing if `is_user_input`')
2318         self.assertCountEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', '3:'),
2319                                            is_user_input=True), tuple(range(8))[3:],
2320                               msg='allow for string slice if `is_user_input`')
2321         self.assertCountEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', ':4:2'),
2322                                            is_user_input=True), tuple(range(8))[:4:2],
2323                               msg='allow step in string slice if `is_user_input`')
2324         self.assertCountEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', ':'),
2325                                            is_user_input=True), range(8),
2326                               msg='`:` should be treated as `...` if `is_user_input`')
2327         with self.assertRaises(TypeError, msg='too many params should result in error'):
2328             traverse_obj(_IS_USER_INPUT_DATA, ('range8', ':::'), is_user_input=True)
2329
2330         # Test re.Match as input obj
2331         mobj = re.fullmatch(r'0(12)(?P<group>3)(4)?', '0123')
2332         self.assertEqual(traverse_obj(mobj, ...), [x for x in mobj.groups() if x is not None],
2333                          msg='`...` on a `re.Match` should give its `groups()`')
2334         self.assertEqual(traverse_obj(mobj, lambda k, _: k in (0, 2)), ['0123', '3'],
2335                          msg='function on a `re.Match` should give groupno, value starting at 0')
2336         self.assertEqual(traverse_obj(mobj, 'group'), '3',
2337                          msg='str key on a `re.Match` should give group with that name')
2338         self.assertEqual(traverse_obj(mobj, 2), '3',
2339                          msg='int key on a `re.Match` should give group with that name')
2340         self.assertEqual(traverse_obj(mobj, 'gRoUp', casesense=False), '3',
2341                          msg='str key on a `re.Match` should respect casesense')
2342         self.assertEqual(traverse_obj(mobj, 'fail'), None,
2343                          msg='failing str key on a `re.Match` should return `default`')
2344         self.assertEqual(traverse_obj(mobj, 'gRoUpS', casesense=False), None,
2345                          msg='failing str key on a `re.Match` should return `default`')
2346         self.assertEqual(traverse_obj(mobj, 8), None,
2347                          msg='failing int key on a `re.Match` should return `default`')
2348         self.assertEqual(traverse_obj(mobj, lambda k, _: k in (0, 'group')), ['0123', '3'],
2349                          msg='function on a `re.Match` should give group name as well')
2350
2351     def test_http_header_dict(self):
2352         headers = HTTPHeaderDict()
2353         headers['ytdl-test'] = b'0'
2354         self.assertEqual(list(headers.items()), [('Ytdl-Test', '0')])
2355         headers['ytdl-test'] = 1
2356         self.assertEqual(list(headers.items()), [('Ytdl-Test', '1')])
2357         headers['Ytdl-test'] = '2'
2358         self.assertEqual(list(headers.items()), [('Ytdl-Test', '2')])
2359         self.assertTrue('ytDl-Test' in headers)
2360         self.assertEqual(str(headers), str(dict(headers)))
2361         self.assertEqual(repr(headers), str(dict(headers)))
2362
2363         headers.update({'X-dlp': 'data'})
2364         self.assertEqual(set(headers.items()), {('Ytdl-Test', '2'), ('X-Dlp', 'data')})
2365         self.assertEqual(dict(headers), {'Ytdl-Test': '2', 'X-Dlp': 'data'})
2366         self.assertEqual(len(headers), 2)
2367         self.assertEqual(headers.copy(), headers)
2368         headers2 = HTTPHeaderDict({'X-dlp': 'data3'}, **headers, **{'X-dlp': 'data2'})
2369         self.assertEqual(set(headers2.items()), {('Ytdl-Test', '2'), ('X-Dlp', 'data2')})
2370         self.assertEqual(len(headers2), 2)
2371         headers2.clear()
2372         self.assertEqual(len(headers2), 0)
2373
2374         # ensure we prefer latter headers
2375         headers3 = HTTPHeaderDict({'Ytdl-TeSt': 1}, {'Ytdl-test': 2})
2376         self.assertEqual(set(headers3.items()), {('Ytdl-Test', '2')})
2377         del headers3['ytdl-tesT']
2378         self.assertEqual(dict(headers3), {})
2379
2380         headers4 = HTTPHeaderDict({'ytdl-test': 'data;'})
2381         self.assertEqual(set(headers4.items()), {('Ytdl-Test', 'data;')})
2382
2383     def test_extract_basic_auth(self):
2384         assert extract_basic_auth('http://:foo.bar') == ('http://:foo.bar', None)
2385         assert extract_basic_auth('http://foo.bar') == ('http://foo.bar', None)
2386         assert extract_basic_auth('http://@foo.bar') == ('http://foo.bar', 'Basic Og==')
2387         assert extract_basic_auth('http://:pass@foo.bar') == ('http://foo.bar', 'Basic OnBhc3M=')
2388         assert extract_basic_auth('http://user:@foo.bar') == ('http://foo.bar', 'Basic dXNlcjo=')
2389         assert extract_basic_auth('http://user:pass@foo.bar') == ('http://foo.bar', 'Basic dXNlcjpwYXNz')
2390
2391
2392 if __name__ == '__main__':
2393     unittest.main()