test/test_utils.py

   1 #!/usr/bin/env python3
   2
   3 # Allow direct execution
   4 import os
   5 import sys
   6 import unittest
   7
   8 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
   9
  10
  11 import contextlib
  12 import io
  13 import itertools
  14 import json
  15 import xml.etree.ElementTree
  16
  17 from yt_dlp.compat import (
  18     compat_etree_fromstring,
  19     compat_HTMLParseError,
  20     compat_os_name,
  21 )
  22 from yt_dlp.utils import (
  23     Config,
  24     DateRange,
  25     ExtractorError,
  26     InAdvancePagedList,
  27     LazyList,
  28     OnDemandPagedList,
  29     age_restricted,
  30     args_to_str,
  31     base_url,
  32     caesar,
  33     clean_html,
  34     clean_podcast_url,
  35     cli_bool_option,
  36     cli_option,
  37     cli_valueless_option,
  38     date_from_str,
  39     datetime_from_str,
  40     detect_exe_version,
  41     determine_ext,
  42     determine_file_encoding,
  43     dfxp2srt,
  44     dict_get,
  45     encode_base_n,
  46     encode_compat_str,
  47     encodeFilename,
  48     escape_rfc3986,
  49     escape_url,
  50     expand_path,
  51     extract_attributes,
  52     find_xpath_attr,
  53     fix_xml_ampersands,
  54     float_or_none,
  55     format_bytes,
  56     get_compatible_ext,
  57     get_element_by_attribute,
  58     get_element_by_class,
  59     get_element_html_by_attribute,
  60     get_element_html_by_class,
  61     get_element_text_and_html_by_tag,
  62     get_elements_by_attribute,
  63     get_elements_by_class,
  64     get_elements_html_by_attribute,
  65     get_elements_html_by_class,
  66     get_elements_text_and_html_by_attribute,
  67     int_or_none,
  68     intlist_to_bytes,
  69     iri_to_uri,
  70     is_html,
  71     js_to_json,
  72     limit_length,
  73     locked_file,
  74     lowercase_escape,
  75     match_str,
  76     merge_dicts,
  77     mimetype2ext,
  78     month_by_name,
  79     multipart_encode,
  80     ohdave_rsa_encrypt,
  81     orderedSet,
  82     parse_age_limit,
  83     parse_bitrate,
  84     parse_codecs,
  85     parse_count,
  86     parse_dfxp_time_expr,
  87     parse_duration,
  88     parse_filesize,
  89     parse_iso8601,
  90     parse_qs,
  91     parse_resolution,
  92     pkcs1pad,
  93     prepend_extension,
  94     read_batch_urls,
  95     remove_end,
  96     remove_quotes,
  97     remove_start,
  98     render_table,
  99     replace_extension,
 100     rot47,
 101     sanitize_filename,
 102     sanitize_path,
 103     sanitize_url,
 104     sanitized_Request,
 105     shell_quote,
 106     smuggle_url,
 107     str_to_int,
 108     strip_jsonp,
 109     strip_or_none,
 110     subtitles_filename,
 111     timeconvert,
 112     traverse_obj,
 113     unescapeHTML,
 114     unified_strdate,
 115     unified_timestamp,
 116     unsmuggle_url,
 117     update_url_query,
 118     uppercase_escape,
 119     url_basename,
 120     url_or_none,
 121     urlencode_postdata,
 122     urljoin,
 123     urshift,
 124     version_tuple,
 125     xpath_attr,
 126     xpath_element,
 127     xpath_text,
 128     xpath_with_ns,
 129 )
 130
 131
 132 class TestUtil(unittest.TestCase):
 133     def test_timeconvert(self):
 134         self.assertTrue(timeconvert('') is None)
 135         self.assertTrue(timeconvert('bougrg') is None)
 136
 137     def test_sanitize_filename(self):
 138         self.assertEqual(sanitize_filename(''), '')
 139         self.assertEqual(sanitize_filename('abc'), 'abc')
 140         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
 141
 142         self.assertEqual(sanitize_filename('123'), '123')
 143
 144         self.assertEqual('abc⧸de', sanitize_filename('abc/de'))
 145         self.assertFalse('/' in sanitize_filename('abc/de///'))
 146
 147         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', is_id=False))
 148         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', is_id=False))
 149         self.assertEqual('yes no', sanitize_filename('yes? no', is_id=False))
 150         self.assertEqual('this - that', sanitize_filename('this: that', is_id=False))
 151
 152         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
 153         aumlaut = 'ä'
 154         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
 155         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
 156         self.assertEqual(sanitize_filename(tests), tests)
 157
 158         self.assertEqual(
 159             sanitize_filename('New World record at 0:12:34'),
 160             'New World record at 0_12_34')
 161
 162         self.assertEqual(sanitize_filename('--gasdgf'), '--gasdgf')
 163         self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
 164         self.assertEqual(sanitize_filename('--gasdgf', is_id=False), '_-gasdgf')
 165         self.assertEqual(sanitize_filename('.gasdgf'), '.gasdgf')
 166         self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
 167         self.assertEqual(sanitize_filename('.gasdgf', is_id=False), 'gasdgf')
 168
 169         forbidden = '"\0\\/'
 170         for fc in forbidden:
 171             for fbc in forbidden:
 172                 self.assertTrue(fbc not in sanitize_filename(fc))
 173
 174     def test_sanitize_filename_restricted(self):
 175         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
 176         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
 177
 178         self.assertEqual(sanitize_filename('123', restricted=True), '123')
 179
 180         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
 181         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
 182
 183         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
 184         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
 185         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
 186         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
 187
 188         tests = 'aäb\u4e2d\u56fd\u7684c'
 189         self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
 190         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
 191
 192         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
 193         for fc in forbidden:
 194             for fbc in forbidden:
 195                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
 196
 197         # Handle a common case more neatly
 198         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
 199         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
 200         # .. but make sure the file name is never empty
 201         self.assertTrue(sanitize_filename('-', restricted=True) != '')
 202         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 203
 204         self.assertEqual(sanitize_filename(
 205             'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
 206             'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy')
 207
 208     def test_sanitize_ids(self):
 209         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 210         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 211         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 212
 213     def test_sanitize_path(self):
 214         if sys.platform != 'win32':
 215             return
 216
 217         self.assertEqual(sanitize_path('abc'), 'abc')
 218         self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
 219         self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
 220         self.assertEqual(sanitize_path('abc|def'), 'abc#def')
 221         self.assertEqual(sanitize_path('<>:"|?*'), '#######')
 222         self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
 223         self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
 224
 225         self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
 226         self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
 227
 228         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 229         self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
 230         self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
 231         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 232
 233         self.assertEqual(
 234             sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
 235             'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
 236
 237         self.assertEqual(
 238             sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
 239             'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
 240         self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
 241         self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
 242         self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
 243
 244         self.assertEqual(sanitize_path('../abc'), '..\\abc')
 245         self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
 246         self.assertEqual(sanitize_path('./abc'), 'abc')
 247         self.assertEqual(sanitize_path('./../abc'), '..\\abc')
 248
 249     def test_sanitize_url(self):
 250         self.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
 251         self.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
 252         self.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
 253         self.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')
 254         self.assertEqual(sanitize_url('foo bar'), 'foo bar')
 255
 256     def test_extract_basic_auth(self):
 257         auth_header = lambda url: sanitized_Request(url).get_header('Authorization')
 258         self.assertFalse(auth_header('http://foo.bar'))
 259         self.assertFalse(auth_header('http://:foo.bar'))
 260         self.assertEqual(auth_header('http://@foo.bar'), 'Basic Og==')
 261         self.assertEqual(auth_header('http://:pass@foo.bar'), 'Basic OnBhc3M=')
 262         self.assertEqual(auth_header('http://user:@foo.bar'), 'Basic dXNlcjo=')
 263         self.assertEqual(auth_header('http://user:pass@foo.bar'), 'Basic dXNlcjpwYXNz')
 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
 662     def test_fix_xml_ampersands(self):
 663         self.assertEqual(
 664             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 665         self.assertEqual(
 666             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 667             '"&amp;x=y&amp;wrong;&amp;z=a')
 668         self.assertEqual(
 669             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 670             '&amp;&apos;&gt;&lt;&quot;')
 671         self.assertEqual(
 672             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 673         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 674
 675     def test_paged_list(self):
 676         def testPL(size, pagesize, sliceargs, expected):
 677             def get_page(pagenum):
 678                 firstid = pagenum * pagesize
 679                 upto = min(size, pagenum * pagesize + pagesize)
 680                 yield from range(firstid, upto)
 681
 682             pl = OnDemandPagedList(get_page, pagesize)
 683             got = pl.getslice(*sliceargs)
 684             self.assertEqual(got, expected)
 685
 686             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
 687             got = iapl.getslice(*sliceargs)
 688             self.assertEqual(got, expected)
 689
 690         testPL(5, 2, (), [0, 1, 2, 3, 4])
 691         testPL(5, 2, (1,), [1, 2, 3, 4])
 692         testPL(5, 2, (2,), [2, 3, 4])
 693         testPL(5, 2, (4,), [4])
 694         testPL(5, 2, (0, 3), [0, 1, 2])
 695         testPL(5, 2, (1, 4), [1, 2, 3])
 696         testPL(5, 2, (2, 99), [2, 3, 4])
 697         testPL(5, 2, (20, 99), [])
 698
 699     def test_read_batch_urls(self):
 700         f = io.StringIO('''\xef\xbb\xbf foo
 701             bar\r
 702             baz
 703             # More after this line\r
 704             ; or after this
 705             bam''')
 706         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
 707
 708     def test_urlencode_postdata(self):
 709         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
 710         self.assertTrue(isinstance(data, bytes))
 711
 712     def test_update_url_query(self):
 713         self.assertEqual(parse_qs(update_url_query(
 714             'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
 715             parse_qs('http://example.com/path?quality=HD&format=mp4'))
 716         self.assertEqual(parse_qs(update_url_query(
 717             'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
 718             parse_qs('http://example.com/path?system=LINUX&system=WINDOWS'))
 719         self.assertEqual(parse_qs(update_url_query(
 720             'http://example.com/path', {'fields': 'id,formats,subtitles'})),
 721             parse_qs('http://example.com/path?fields=id,formats,subtitles'))
 722         self.assertEqual(parse_qs(update_url_query(
 723             'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
 724             parse_qs('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
 725         self.assertEqual(parse_qs(update_url_query(
 726             'http://example.com/path?manifest=f4m', {'manifest': []})),
 727             parse_qs('http://example.com/path'))
 728         self.assertEqual(parse_qs(update_url_query(
 729             'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
 730             parse_qs('http://example.com/path?system=LINUX'))
 731         self.assertEqual(parse_qs(update_url_query(
 732             'http://example.com/path', {'fields': b'id,formats,subtitles'})),
 733             parse_qs('http://example.com/path?fields=id,formats,subtitles'))
 734         self.assertEqual(parse_qs(update_url_query(
 735             'http://example.com/path', {'width': 1080, 'height': 720})),
 736             parse_qs('http://example.com/path?width=1080&height=720'))
 737         self.assertEqual(parse_qs(update_url_query(
 738             'http://example.com/path', {'bitrate': 5020.43})),
 739             parse_qs('http://example.com/path?bitrate=5020.43'))
 740         self.assertEqual(parse_qs(update_url_query(
 741             'http://example.com/path', {'test': '第二行тест'})),
 742             parse_qs('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
 743
 744     def test_multipart_encode(self):
 745         self.assertEqual(
 746             multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0],
 747             b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
 748         self.assertEqual(
 749             multipart_encode({'欄位'.encode(): '值'.encode()}, boundary='AAAAAA')[0],
 750             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')
 751         self.assertRaises(
 752             ValueError, multipart_encode, {b'field': b'value'}, boundary='value')
 753
 754     def test_dict_get(self):
 755         FALSE_VALUES = {
 756             'none': None,
 757             'false': False,
 758             'zero': 0,
 759             'empty_string': '',
 760             'empty_list': [],
 761         }
 762         d = FALSE_VALUES.copy()
 763         d['a'] = 42
 764         self.assertEqual(dict_get(d, 'a'), 42)
 765         self.assertEqual(dict_get(d, 'b'), None)
 766         self.assertEqual(dict_get(d, 'b', 42), 42)
 767         self.assertEqual(dict_get(d, ('a', )), 42)
 768         self.assertEqual(dict_get(d, ('b', 'a', )), 42)
 769         self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
 770         self.assertEqual(dict_get(d, ('b', 'c', )), None)
 771         self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
 772         for key, false_value in FALSE_VALUES.items():
 773             self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
 774             self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
 775
 776     def test_merge_dicts(self):
 777         self.assertEqual(merge_dicts({'a': 1}, {'b': 2}), {'a': 1, 'b': 2})
 778         self.assertEqual(merge_dicts({'a': 1}, {'a': 2}), {'a': 1})
 779         self.assertEqual(merge_dicts({'a': 1}, {'a': None}), {'a': 1})
 780         self.assertEqual(merge_dicts({'a': 1}, {'a': ''}), {'a': 1})
 781         self.assertEqual(merge_dicts({'a': 1}, {}), {'a': 1})
 782         self.assertEqual(merge_dicts({'a': None}, {'a': 1}), {'a': 1})
 783         self.assertEqual(merge_dicts({'a': ''}, {'a': 1}), {'a': ''})
 784         self.assertEqual(merge_dicts({'a': ''}, {'a': 'abc'}), {'a': 'abc'})
 785         self.assertEqual(merge_dicts({'a': None}, {'a': ''}, {'a': 'abc'}), {'a': 'abc'})
 786
 787     def test_encode_compat_str(self):
 788         self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
 789         self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
 790
 791     def test_parse_iso8601(self):
 792         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
 793         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
 794         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
 795         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
 796         self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
 797         self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
 798
 799     def test_strip_jsonp(self):
 800         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
 801         d = json.loads(stripped)
 802         self.assertEqual(d, [{"id": "532cb", "x": 3}])
 803
 804         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
 805         d = json.loads(stripped)
 806         self.assertEqual(d, {'STATUS': 'OK'})
 807
 808         stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
 809         d = json.loads(stripped)
 810         self.assertEqual(d, {'status': 'success'})
 811
 812         stripped = strip_jsonp('window.cb && window.cb({"status": "success"});')
 813         d = json.loads(stripped)
 814         self.assertEqual(d, {'status': 'success'})
 815
 816         stripped = strip_jsonp('window.cb && cb({"status": "success"});')
 817         d = json.loads(stripped)
 818         self.assertEqual(d, {'status': 'success'})
 819
 820         stripped = strip_jsonp('({"status": "success"});')
 821         d = json.loads(stripped)
 822         self.assertEqual(d, {'status': 'success'})
 823
 824     def test_strip_or_none(self):
 825         self.assertEqual(strip_or_none(' abc'), 'abc')
 826         self.assertEqual(strip_or_none('abc '), 'abc')
 827         self.assertEqual(strip_or_none(' abc '), 'abc')
 828         self.assertEqual(strip_or_none('\tabc\t'), 'abc')
 829         self.assertEqual(strip_or_none('\n\tabc\n\t'), 'abc')
 830         self.assertEqual(strip_or_none('abc'), 'abc')
 831         self.assertEqual(strip_or_none(''), '')
 832         self.assertEqual(strip_or_none(None), None)
 833         self.assertEqual(strip_or_none(42), None)
 834         self.assertEqual(strip_or_none([]), None)
 835
 836     def test_uppercase_escape(self):
 837         self.assertEqual(uppercase_escape('aä'), 'aä')
 838         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
 839
 840     def test_lowercase_escape(self):
 841         self.assertEqual(lowercase_escape('aä'), 'aä')
 842         self.assertEqual(lowercase_escape('\\u0026'), '&')
 843
 844     def test_limit_length(self):
 845         self.assertEqual(limit_length(None, 12), None)
 846         self.assertEqual(limit_length('foo', 12), 'foo')
 847         self.assertTrue(
 848             limit_length('foo bar baz asd', 12).startswith('foo bar'))
 849         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
 850
 851     def test_mimetype2ext(self):
 852         self.assertEqual(mimetype2ext(None), None)
 853         self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
 854         self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
 855         self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
 856         self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
 857         self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
 858         self.assertEqual(mimetype2ext('audio/x-wav'), 'wav')
 859         self.assertEqual(mimetype2ext('audio/x-wav;codec=pcm'), 'wav')
 860
 861     def test_month_by_name(self):
 862         self.assertEqual(month_by_name(None), None)
 863         self.assertEqual(month_by_name('December', 'en'), 12)
 864         self.assertEqual(month_by_name('décembre', 'fr'), 12)
 865         self.assertEqual(month_by_name('December'), 12)
 866         self.assertEqual(month_by_name('décembre'), None)
 867         self.assertEqual(month_by_name('Unknown', 'unknown'), None)
 868
 869     def test_parse_codecs(self):
 870         self.assertEqual(parse_codecs(''), {})
 871         self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
 872             'vcodec': 'avc1.77.30',
 873             'acodec': 'mp4a.40.2',
 874             'dynamic_range': None,
 875         })
 876         self.assertEqual(parse_codecs('mp4a.40.2'), {
 877             'vcodec': 'none',
 878             'acodec': 'mp4a.40.2',
 879             'dynamic_range': None,
 880         })
 881         self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
 882             'vcodec': 'avc1.42001e',
 883             'acodec': 'mp4a.40.5',
 884             'dynamic_range': None,
 885         })
 886         self.assertEqual(parse_codecs('avc3.640028'), {
 887             'vcodec': 'avc3.640028',
 888             'acodec': 'none',
 889             'dynamic_range': None,
 890         })
 891         self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
 892             'vcodec': 'h264',
 893             'acodec': 'aac',
 894             'dynamic_range': None,
 895         })
 896         self.assertEqual(parse_codecs('av01.0.05M.08'), {
 897             'vcodec': 'av01.0.05M.08',
 898             'acodec': 'none',
 899             'dynamic_range': None,
 900         })
 901         self.assertEqual(parse_codecs('vp9.2'), {
 902             'vcodec': 'vp9.2',
 903             'acodec': 'none',
 904             'dynamic_range': 'HDR10',
 905         })
 906         self.assertEqual(parse_codecs('av01.0.12M.10.0.110.09.16.09.0'), {
 907             'vcodec': 'av01.0.12M.10.0.110.09.16.09.0',
 908             'acodec': 'none',
 909             'dynamic_range': 'HDR10',
 910         })
 911         self.assertEqual(parse_codecs('dvhe'), {
 912             'vcodec': 'dvhe',
 913             'acodec': 'none',
 914             'dynamic_range': 'DV',
 915         })
 916         self.assertEqual(parse_codecs('theora, vorbis'), {
 917             'vcodec': 'theora',
 918             'acodec': 'vorbis',
 919             'dynamic_range': None,
 920         })
 921         self.assertEqual(parse_codecs('unknownvcodec, unknownacodec'), {
 922             'vcodec': 'unknownvcodec',
 923             'acodec': 'unknownacodec',
 924         })
 925         self.assertEqual(parse_codecs('unknown'), {})
 926
 927     def test_escape_rfc3986(self):
 928         reserved = "!*'();:@&=+$,/?#[]"
 929         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
 930         self.assertEqual(escape_rfc3986(reserved), reserved)
 931         self.assertEqual(escape_rfc3986(unreserved), unreserved)
 932         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
 933         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
 934         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
 935         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
 936
 937     def test_escape_url(self):
 938         self.assertEqual(
 939             escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
 940             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
 941         )
 942         self.assertEqual(
 943             escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
 944             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
 945         )
 946         self.assertEqual(
 947             escape_url('http://тест.рф/фрагмент'),
 948             'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
 949         )
 950         self.assertEqual(
 951             escape_url('http://тест.рф/абв?абв=абв#абв'),
 952             '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'
 953         )
 954         self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
 955
 956     def test_js_to_json_realworld(self):
 957         inp = '''{
 958             'clip':{'provider':'pseudo'}
 959         }'''
 960         self.assertEqual(js_to_json(inp), '''{
 961             "clip":{"provider":"pseudo"}
 962         }''')
 963         json.loads(js_to_json(inp))
 964
 965         inp = '''{
 966             'playlist':[{'controls':{'all':null}}]
 967         }'''
 968         self.assertEqual(js_to_json(inp), '''{
 969             "playlist":[{"controls":{"all":null}}]
 970         }''')
 971
 972         inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
 973         self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
 974
 975         inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
 976         json_code = js_to_json(inp)
 977         self.assertEqual(json.loads(json_code), json.loads(inp))
 978
 979         inp = '''{
 980             0:{src:'skipped', type: 'application/dash+xml'},
 981             1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
 982         }'''
 983         self.assertEqual(js_to_json(inp), '''{
 984             "0":{"src":"skipped", "type": "application/dash+xml"},
 985             "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
 986         }''')
 987
 988         inp = '''{"foo":101}'''
 989         self.assertEqual(js_to_json(inp), '''{"foo":101}''')
 990
 991         inp = '''{"duration": "00:01:07"}'''
 992         self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
 993
 994         inp = '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}'''
 995         self.assertEqual(js_to_json(inp), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''')
 996
 997     def test_js_to_json_edgecases(self):
 998         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
 999         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
1000
1001         on = js_to_json('{"abc": true}')
1002         self.assertEqual(json.loads(on), {'abc': True})
1003
1004         # Ignore JavaScript code as well
1005         on = js_to_json('''{
1006             "x": 1,
1007             y: "a",
1008             z: some.code
1009         }''')
1010         d = json.loads(on)
1011         self.assertEqual(d['x'], 1)
1012         self.assertEqual(d['y'], 'a')
1013
1014         # Just drop ! prefix for now though this results in a wrong value
1015         on = js_to_json('''{
1016             a: !0,
1017             b: !1,
1018             c: !!0,
1019             d: !!42.42,
1020             e: !!![],
1021             f: !"abc",
1022             g: !"",
1023             !42: 42
1024         }''')
1025         self.assertEqual(json.loads(on), {
1026             'a': 0,
1027             'b': 1,
1028             'c': 0,
1029             'd': 42.42,
1030             'e': [],
1031             'f': "abc",
1032             'g': "",
1033             '42': 42
1034         })
1035
1036         on = js_to_json('["abc", "def",]')
1037         self.assertEqual(json.loads(on), ['abc', 'def'])
1038
1039         on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
1040         self.assertEqual(json.loads(on), ['abc', 'def'])
1041
1042         on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
1043         self.assertEqual(json.loads(on), ['abc', 'def'])
1044
1045         on = js_to_json('{"abc": "def",}')
1046         self.assertEqual(json.loads(on), {'abc': 'def'})
1047
1048         on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
1049         self.assertEqual(json.loads(on), {'abc': 'def'})
1050
1051         on = js_to_json('{ 0: /* " \n */ ",]" , }')
1052         self.assertEqual(json.loads(on), {'0': ',]'})
1053
1054         on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
1055         self.assertEqual(json.loads(on), {'0': ',]'})
1056
1057         on = js_to_json('{ 0: // comment\n1 }')
1058         self.assertEqual(json.loads(on), {'0': 1})
1059
1060         on = js_to_json(r'["<p>x<\/p>"]')
1061         self.assertEqual(json.loads(on), ['<p>x</p>'])
1062
1063         on = js_to_json(r'["\xaa"]')
1064         self.assertEqual(json.loads(on), ['\u00aa'])
1065
1066         on = js_to_json("['a\\\nb']")
1067         self.assertEqual(json.loads(on), ['ab'])
1068
1069         on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
1070         self.assertEqual(json.loads(on), ['ab'])
1071
1072         on = js_to_json('{0xff:0xff}')
1073         self.assertEqual(json.loads(on), {'255': 255})
1074
1075         on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
1076         self.assertEqual(json.loads(on), {'255': 255})
1077
1078         on = js_to_json('{077:077}')
1079         self.assertEqual(json.loads(on), {'63': 63})
1080
1081         on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
1082         self.assertEqual(json.loads(on), {'63': 63})
1083
1084         on = js_to_json('{42:42}')
1085         self.assertEqual(json.loads(on), {'42': 42})
1086
1087         on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
1088         self.assertEqual(json.loads(on), {'42': 42})
1089
1090         on = js_to_json('{42:4.2e1}')
1091         self.assertEqual(json.loads(on), {'42': 42.0})
1092
1093         on = js_to_json('{ "0x40": "0x40" }')
1094         self.assertEqual(json.loads(on), {'0x40': '0x40'})
1095
1096         on = js_to_json('{ "040": "040" }')
1097         self.assertEqual(json.loads(on), {'040': '040'})
1098
1099         on = js_to_json('[1,//{},\n2]')
1100         self.assertEqual(json.loads(on), [1, 2])
1101
1102     def test_js_to_json_malformed(self):
1103         self.assertEqual(js_to_json('42a1'), '42"a1"')
1104         self.assertEqual(js_to_json('42a-1'), '42"a"-1')
1105
1106     def test_extract_attributes(self):
1107         self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
1108         self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
1109         self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
1110         self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
1111         self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
1112         self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
1113         self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
1114         self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'})  # XML
1115         self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
1116         self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'})  # HTML 3.2
1117         self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'})  # HTML 4.0
1118         self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
1119         self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
1120         self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
1121         self.assertEqual(extract_attributes('<e x >'), {'x': None})
1122         self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
1123         self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
1124         self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
1125         self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
1126         self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
1127         self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
1128         self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
1129         self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'})  # Names lowercased
1130         self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
1131         self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
1132         self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
1133         self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
1134         self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
1135         # "Narrow" Python builds don't support unicode code points outside BMP.
1136         try:
1137             chr(0x10000)
1138             supports_outside_bmp = True
1139         except ValueError:
1140             supports_outside_bmp = False
1141         if supports_outside_bmp:
1142             self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
1143         # Malformed HTML should not break attributes extraction on older Python
1144         self.assertEqual(extract_attributes('<mal"formed/>'), {})
1145
1146     def test_clean_html(self):
1147         self.assertEqual(clean_html('a:\nb'), 'a: b')
1148         self.assertEqual(clean_html('a:\n   "b"'), 'a: "b"')
1149         self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
1150
1151     def test_intlist_to_bytes(self):
1152         self.assertEqual(
1153             intlist_to_bytes([0, 1, 127, 128, 255]),
1154             b'\x00\x01\x7f\x80\xff')
1155
1156     def test_args_to_str(self):
1157         self.assertEqual(
1158             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
1159             'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""'
1160         )
1161
1162     def test_parse_filesize(self):
1163         self.assertEqual(parse_filesize(None), None)
1164         self.assertEqual(parse_filesize(''), None)
1165         self.assertEqual(parse_filesize('91 B'), 91)
1166         self.assertEqual(parse_filesize('foobar'), None)
1167         self.assertEqual(parse_filesize('2 MiB'), 2097152)
1168         self.assertEqual(parse_filesize('5 GB'), 5000000000)
1169         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
1170         self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
1171         self.assertEqual(parse_filesize('1,24 KB'), 1240)
1172         self.assertEqual(parse_filesize('1,24 kb'), 1240)
1173         self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
1174
1175     def test_parse_count(self):
1176         self.assertEqual(parse_count(None), None)
1177         self.assertEqual(parse_count(''), None)
1178         self.assertEqual(parse_count('0'), 0)
1179         self.assertEqual(parse_count('1000'), 1000)
1180         self.assertEqual(parse_count('1.000'), 1000)
1181         self.assertEqual(parse_count('1.1k'), 1100)
1182         self.assertEqual(parse_count('1.1 k'), 1100)
1183         self.assertEqual(parse_count('1,1 k'), 1100)
1184         self.assertEqual(parse_count('1.1kk'), 1100000)
1185         self.assertEqual(parse_count('1.1kk '), 1100000)
1186         self.assertEqual(parse_count('1,1kk'), 1100000)
1187         self.assertEqual(parse_count('100 views'), 100)
1188         self.assertEqual(parse_count('1,100 views'), 1100)
1189         self.assertEqual(parse_count('1.1kk views'), 1100000)
1190         self.assertEqual(parse_count('10M views'), 10000000)
1191         self.assertEqual(parse_count('has 10M views'), 10000000)
1192
1193     def test_parse_resolution(self):
1194         self.assertEqual(parse_resolution(None), {})
1195         self.assertEqual(parse_resolution(''), {})
1196         self.assertEqual(parse_resolution(' 1920x1080'), {'width': 1920, 'height': 1080})
1197         self.assertEqual(parse_resolution('1920×1080 '), {'width': 1920, 'height': 1080})
1198         self.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080})
1199         self.assertEqual(parse_resolution('720p'), {'height': 720})
1200         self.assertEqual(parse_resolution('4k'), {'height': 2160})
1201         self.assertEqual(parse_resolution('8K'), {'height': 4320})
1202         self.assertEqual(parse_resolution('pre_1920x1080_post'), {'width': 1920, 'height': 1080})
1203         self.assertEqual(parse_resolution('ep1x2'), {})
1204         self.assertEqual(parse_resolution('1920, 1080'), {'width': 1920, 'height': 1080})
1205
1206     def test_parse_bitrate(self):
1207         self.assertEqual(parse_bitrate(None), None)
1208         self.assertEqual(parse_bitrate(''), None)
1209         self.assertEqual(parse_bitrate('300kbps'), 300)
1210         self.assertEqual(parse_bitrate('1500kbps'), 1500)
1211         self.assertEqual(parse_bitrate('300 kbps'), 300)
1212
1213     def test_version_tuple(self):
1214         self.assertEqual(version_tuple('1'), (1,))
1215         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
1216         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
1217
1218     def test_detect_exe_version(self):
1219         self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
1220 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
1221 configuration: --prefix=/usr --extra-'''), '1.2.1')
1222         self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
1223 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
1224         self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
1225 Trying to open render node...
1226 Success at /dev/dri/renderD128.
1227 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
1228
1229     def test_age_restricted(self):
1230         self.assertFalse(age_restricted(None, 10))  # unrestricted content
1231         self.assertFalse(age_restricted(1, None))  # unrestricted policy
1232         self.assertFalse(age_restricted(8, 10))
1233         self.assertTrue(age_restricted(18, 14))
1234         self.assertFalse(age_restricted(18, 18))
1235
1236     def test_is_html(self):
1237         self.assertFalse(is_html(b'\x49\x44\x43<html'))
1238         self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
1239         self.assertTrue(is_html(  # UTF-8 with BOM
1240             b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
1241         self.assertTrue(is_html(  # UTF-16-LE
1242             b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
1243         ))
1244         self.assertTrue(is_html(  # UTF-16-BE
1245             b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
1246         ))
1247         self.assertTrue(is_html(  # UTF-32-BE
1248             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'))
1249         self.assertTrue(is_html(  # UTF-32-LE
1250             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'))
1251
1252     def test_render_table(self):
1253         self.assertEqual(
1254             render_table(
1255                 ['a', 'empty', 'bcd'],
1256                 [[123, '', 4], [9999, '', 51]]),
1257             'a    empty bcd\n'
1258             '123        4\n'
1259             '9999       51')
1260
1261         self.assertEqual(
1262             render_table(
1263                 ['a', 'empty', 'bcd'],
1264                 [[123, '', 4], [9999, '', 51]],
1265                 hide_empty=True),
1266             'a    bcd\n'
1267             '123  4\n'
1268             '9999 51')
1269
1270         self.assertEqual(
1271             render_table(
1272                 ['\ta', 'bcd'],
1273                 [['1\t23', 4], ['\t9999', 51]]),
1274             '   a bcd\n'
1275             '1 23 4\n'
1276             '9999 51')
1277
1278         self.assertEqual(
1279             render_table(
1280                 ['a', 'bcd'],
1281                 [[123, 4], [9999, 51]],
1282                 delim='-'),
1283             'a    bcd\n'
1284             '--------\n'
1285             '123  4\n'
1286             '9999 51')
1287
1288         self.assertEqual(
1289             render_table(
1290                 ['a', 'bcd'],
1291                 [[123, 4], [9999, 51]],
1292                 delim='-', extra_gap=2),
1293             'a      bcd\n'
1294             '----------\n'
1295             '123    4\n'
1296             '9999   51')
1297
1298     def test_match_str(self):
1299         # Unary
1300         self.assertFalse(match_str('xy', {'x': 1200}))
1301         self.assertTrue(match_str('!xy', {'x': 1200}))
1302         self.assertTrue(match_str('x', {'x': 1200}))
1303         self.assertFalse(match_str('!x', {'x': 1200}))
1304         self.assertTrue(match_str('x', {'x': 0}))
1305         self.assertTrue(match_str('is_live', {'is_live': True}))
1306         self.assertFalse(match_str('is_live', {'is_live': False}))
1307         self.assertFalse(match_str('is_live', {'is_live': None}))
1308         self.assertFalse(match_str('is_live', {}))
1309         self.assertFalse(match_str('!is_live', {'is_live': True}))
1310         self.assertTrue(match_str('!is_live', {'is_live': False}))
1311         self.assertTrue(match_str('!is_live', {'is_live': None}))
1312         self.assertTrue(match_str('!is_live', {}))
1313         self.assertTrue(match_str('title', {'title': 'abc'}))
1314         self.assertTrue(match_str('title', {'title': ''}))
1315         self.assertFalse(match_str('!title', {'title': 'abc'}))
1316         self.assertFalse(match_str('!title', {'title': ''}))
1317
1318         # Numeric
1319         self.assertFalse(match_str('x>0', {'x': 0}))
1320         self.assertFalse(match_str('x>0', {}))
1321         self.assertTrue(match_str('x>?0', {}))
1322         self.assertTrue(match_str('x>1K', {'x': 1200}))
1323         self.assertFalse(match_str('x>2K', {'x': 1200}))
1324         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
1325         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
1326         self.assertTrue(match_str('x > 1:0:0', {'x': 3700}))
1327
1328         # String
1329         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
1330         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
1331         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
1332         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
1333         self.assertTrue(match_str('y^=foo', {'y': 'foobar42'}))
1334         self.assertFalse(match_str('y!^=foo', {'y': 'foobar42'}))
1335         self.assertFalse(match_str('y^=bar', {'y': 'foobar42'}))
1336         self.assertTrue(match_str('y!^=bar', {'y': 'foobar42'}))
1337         self.assertRaises(ValueError, match_str, 'x^=42', {'x': 42})
1338         self.assertTrue(match_str('y*=bar', {'y': 'foobar42'}))
1339         self.assertFalse(match_str('y!*=bar', {'y': 'foobar42'}))
1340         self.assertFalse(match_str('y*=baz', {'y': 'foobar42'}))
1341         self.assertTrue(match_str('y!*=baz', {'y': 'foobar42'}))
1342         self.assertTrue(match_str('y$=42', {'y': 'foobar42'}))
1343         self.assertFalse(match_str('y$=43', {'y': 'foobar42'}))
1344
1345         # And
1346         self.assertFalse(match_str(
1347             'like_count > 100 & dislike_count <? 50 & description',
1348             {'like_count': 90, 'description': 'foo'}))
1349         self.assertTrue(match_str(
1350             'like_count > 100 & dislike_count <? 50 & description',
1351             {'like_count': 190, 'description': 'foo'}))
1352         self.assertFalse(match_str(
1353             'like_count > 100 & dislike_count <? 50 & description',
1354             {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
1355         self.assertFalse(match_str(
1356             'like_count > 100 & dislike_count <? 50 & description',
1357             {'like_count': 190, 'dislike_count': 10}))
1358
1359         # Regex
1360         self.assertTrue(match_str(r'x~=\bbar', {'x': 'foo bar'}))
1361         self.assertFalse(match_str(r'x~=\bbar.+', {'x': 'foo bar'}))
1362         self.assertFalse(match_str(r'x~=^FOO', {'x': 'foo bar'}))
1363         self.assertTrue(match_str(r'x~=(?i)^FOO', {'x': 'foo bar'}))
1364
1365         # Quotes
1366         self.assertTrue(match_str(r'x^="foo"', {'x': 'foo "bar"'}))
1367         self.assertFalse(match_str(r'x^="foo  "', {'x': 'foo "bar"'}))
1368         self.assertFalse(match_str(r'x$="bar"', {'x': 'foo "bar"'}))
1369         self.assertTrue(match_str(r'x$=" \"bar\""', {'x': 'foo "bar"'}))
1370
1371         # Escaping &
1372         self.assertFalse(match_str(r'x=foo & bar', {'x': 'foo & bar'}))
1373         self.assertTrue(match_str(r'x=foo \& bar', {'x': 'foo & bar'}))
1374         self.assertTrue(match_str(r'x=foo \& bar & x^=foo', {'x': 'foo & bar'}))
1375         self.assertTrue(match_str(r'x="foo \& bar" & x^=foo', {'x': 'foo & bar'}))
1376
1377         # Example from docs
1378         self.assertTrue(match_str(
1379             r"!is_live & like_count>?100 & description~='(?i)\bcats \& dogs\b'",
1380             {'description': 'Raining Cats & Dogs'}))
1381
1382         # Incomplete
1383         self.assertFalse(match_str('id!=foo', {'id': 'foo'}, True))
1384         self.assertTrue(match_str('x', {'id': 'foo'}, True))
1385         self.assertTrue(match_str('!x', {'id': 'foo'}, True))
1386         self.assertFalse(match_str('x', {'id': 'foo'}, False))
1387
1388     def test_parse_dfxp_time_expr(self):
1389         self.assertEqual(parse_dfxp_time_expr(None), None)
1390         self.assertEqual(parse_dfxp_time_expr(''), None)
1391         self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1392         self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1393         self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1394         self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1395         self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1396
1397     def test_dfxp2srt(self):
1398         dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
1399             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1400             <body>
1401                 <div xml:lang="en">
1402                     <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1403                     <p begin="1" end="2">第二行<br/>♪♪</p>
1404                     <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1405                     <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1406                     <p begin="-1" end="-1">Ignore, two</p>
1407                     <p begin="3" dur="-1">Ignored, three</p>
1408                 </div>
1409             </body>
1410             </tt>'''.encode()
1411         srt_data = '''1
1412 00:00:00,000 --> 00:00:01,000
1413 The following line contains Chinese characters and special symbols
1414
1415 2
1416 00:00:01,000 --> 00:00:02,000
1417 第二行
1418 ♪♪
1419
1420 3
1421 00:00:02,000 --> 00:00:03,000
1422 Third
1423 Line
1424
1425 '''
1426         self.assertEqual(dfxp2srt(dfxp_data), srt_data)
1427
1428         dfxp_data_no_default_namespace = b'''<?xml version="1.0" encoding="UTF-8"?>
1429             <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1430             <body>
1431                 <div xml:lang="en">
1432                     <p begin="0" end="1">The first line</p>
1433                 </div>
1434             </body>
1435             </tt>'''
1436         srt_data = '''1
1437 00:00:00,000 --> 00:00:01,000
1438 The first line
1439
1440 '''
1441         self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
1442
1443         dfxp_data_with_style = b'''<?xml version="1.0" encoding="utf-8"?>
1444 <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">
1445   <head>
1446     <styling>
1447       <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1448       <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1449       <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1450       <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1451     </styling>
1452   </head>
1453   <body tts:textAlign="center" style="s0">
1454     <div>
1455       <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1456       <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>
1457       <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1458       <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>
1459     </div>
1460   </body>
1461 </tt>'''
1462         srt_data = '''1
1463 00:00:02,080 --> 00:00:05,840
1464 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1465
1466 2
1467 00:00:02,080 --> 00:00:05,840
1468 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1469 </font>part 2</font></b>
1470
1471 3
1472 00:00:05,840 --> 00:00:09,560
1473 <u><font color="lime">line 3
1474 part 3</font></u>
1475
1476 4
1477 00:00:09,560 --> 00:00:12,360
1478 <i><u><font color="yellow"><font color="lime">inner
1479  </font>style</font></u></i>
1480
1481 '''
1482         self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)
1483
1484         dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?>
1485             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1486             <body>
1487                 <div xml:lang="en">
1488                     <p begin="0" end="1">Line 1</p>
1489                     <p begin="1" end="2">第二行</p>
1490                 </div>
1491             </body>
1492             </tt>'''.encode('utf-16')
1493         srt_data = '''1
1494 00:00:00,000 --> 00:00:01,000
1495 Line 1
1496
1497 2
1498 00:00:01,000 --> 00:00:02,000
1499 第二行
1500
1501 '''
1502         self.assertEqual(dfxp2srt(dfxp_data_non_utf8), srt_data)
1503
1504     def test_cli_option(self):
1505         self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1506         self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
1507         self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1508         self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
1509
1510     def test_cli_valueless_option(self):
1511         self.assertEqual(cli_valueless_option(
1512             {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1513         self.assertEqual(cli_valueless_option(
1514             {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
1515         self.assertEqual(cli_valueless_option(
1516             {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1517         self.assertEqual(cli_valueless_option(
1518             {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
1519         self.assertEqual(cli_valueless_option(
1520             {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
1521         self.assertEqual(cli_valueless_option(
1522             {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1523
1524     def test_cli_bool_option(self):
1525         self.assertEqual(
1526             cli_bool_option(
1527                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1528             ['--no-check-certificate', 'true'])
1529         self.assertEqual(
1530             cli_bool_option(
1531                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1532             ['--no-check-certificate=true'])
1533         self.assertEqual(
1534             cli_bool_option(
1535                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1536             ['--check-certificate', 'false'])
1537         self.assertEqual(
1538             cli_bool_option(
1539                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1540             ['--check-certificate=false'])
1541         self.assertEqual(
1542             cli_bool_option(
1543                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1544             ['--check-certificate', 'true'])
1545         self.assertEqual(
1546             cli_bool_option(
1547                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1548             ['--check-certificate=true'])
1549         self.assertEqual(
1550             cli_bool_option(
1551                 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1552             [])
1553
1554     def test_ohdave_rsa_encrypt(self):
1555         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1556         e = 65537
1557
1558         self.assertEqual(
1559             ohdave_rsa_encrypt(b'aa111222', e, N),
1560             '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1561
1562     def test_pkcs1pad(self):
1563         data = [1, 2, 3]
1564         padded_data = pkcs1pad(data, 32)
1565         self.assertEqual(padded_data[:2], [0, 2])
1566         self.assertEqual(padded_data[28:], [0, 1, 2, 3])
1567
1568         self.assertRaises(ValueError, pkcs1pad, data, 8)
1569
1570     def test_encode_base_n(self):
1571         self.assertEqual(encode_base_n(0, 30), '0')
1572         self.assertEqual(encode_base_n(80, 30), '2k')
1573
1574         custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1575         self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1576         self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1577
1578         self.assertRaises(ValueError, encode_base_n, 0, 70)
1579         self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1580
1581     def test_caesar(self):
1582         self.assertEqual(caesar('ace', 'abcdef', 2), 'cea')
1583         self.assertEqual(caesar('cea', 'abcdef', -2), 'ace')
1584         self.assertEqual(caesar('ace', 'abcdef', -2), 'eac')
1585         self.assertEqual(caesar('eac', 'abcdef', 2), 'ace')
1586         self.assertEqual(caesar('ace', 'abcdef', 0), 'ace')
1587         self.assertEqual(caesar('xyz', 'abcdef', 2), 'xyz')
1588         self.assertEqual(caesar('abc', 'acegik', 2), 'ebg')
1589         self.assertEqual(caesar('ebg', 'acegik', -2), 'abc')
1590
1591     def test_rot47(self):
1592         self.assertEqual(rot47('yt-dlp'), r'JE\5=A')
1593         self.assertEqual(rot47('YT-DLP'), r'*%\s{!')
1594
1595     def test_urshift(self):
1596         self.assertEqual(urshift(3, 1), 1)
1597         self.assertEqual(urshift(-3, 1), 2147483646)
1598
1599     GET_ELEMENT_BY_CLASS_TEST_STRING = '''
1600         <span class="foo bar">nice</span>
1601     '''
1602
1603     def test_get_element_by_class(self):
1604         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1605
1606         self.assertEqual(get_element_by_class('foo', html), 'nice')
1607         self.assertEqual(get_element_by_class('no-such-class', html), None)
1608
1609     def test_get_element_html_by_class(self):
1610         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1611
1612         self.assertEqual(get_element_html_by_class('foo', html), html.strip())
1613         self.assertEqual(get_element_by_class('no-such-class', html), None)
1614
1615     GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING = '''
1616         <div itemprop="author" itemscope>foo</div>
1617     '''
1618
1619     def test_get_element_by_attribute(self):
1620         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1621
1622         self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
1623         self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
1624         self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
1625
1626         html = self.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
1627
1628         self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo')
1629
1630     def test_get_element_html_by_attribute(self):
1631         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1632
1633         self.assertEqual(get_element_html_by_attribute('class', 'foo bar', html), html.strip())
1634         self.assertEqual(get_element_html_by_attribute('class', 'foo', html), None)
1635         self.assertEqual(get_element_html_by_attribute('class', 'no-such-foo', html), None)
1636
1637         html = self.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
1638
1639         self.assertEqual(get_element_html_by_attribute('itemprop', 'author', html), html.strip())
1640
1641     GET_ELEMENTS_BY_CLASS_TEST_STRING = '''
1642         <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1643     '''
1644     GET_ELEMENTS_BY_CLASS_RES = ['<span class="foo bar">nice</span>', '<span class="foo bar">also nice</span>']
1645
1646     def test_get_elements_by_class(self):
1647         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1648
1649         self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
1650         self.assertEqual(get_elements_by_class('no-such-class', html), [])
1651
1652     def test_get_elements_html_by_class(self):
1653         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1654
1655         self.assertEqual(get_elements_html_by_class('foo', html), self.GET_ELEMENTS_BY_CLASS_RES)
1656         self.assertEqual(get_elements_html_by_class('no-such-class', html), [])
1657
1658     def test_get_elements_by_attribute(self):
1659         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1660
1661         self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
1662         self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
1663         self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
1664
1665     def test_get_elements_html_by_attribute(self):
1666         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1667
1668         self.assertEqual(get_elements_html_by_attribute('class', 'foo bar', html), self.GET_ELEMENTS_BY_CLASS_RES)
1669         self.assertEqual(get_elements_html_by_attribute('class', 'foo', html), [])
1670         self.assertEqual(get_elements_html_by_attribute('class', 'no-such-foo', html), [])
1671
1672     def test_get_elements_text_and_html_by_attribute(self):
1673         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1674
1675         self.assertEqual(
1676             list(get_elements_text_and_html_by_attribute('class', 'foo bar', html)),
1677             list(zip(['nice', 'also nice'], self.GET_ELEMENTS_BY_CLASS_RES)))
1678         self.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'foo', html)), [])
1679         self.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'no-such-foo', html)), [])
1680
1681     GET_ELEMENT_BY_TAG_TEST_STRING = '''
1682     random text lorem ipsum</p>
1683     <div>
1684         this should be returned
1685         <span>this should also be returned</span>
1686         <div>
1687             this should also be returned
1688         </div>
1689         closing tag above should not trick, so this should also be returned
1690     </div>
1691     but this text should not be returned
1692     '''
1693     GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML = GET_ELEMENT_BY_TAG_TEST_STRING.strip()[32:276]
1694     GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT = GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML[5:-6]
1695     GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML = GET_ELEMENT_BY_TAG_TEST_STRING.strip()[78:119]
1696     GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT = GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML[6:-7]
1697
1698     def test_get_element_text_and_html_by_tag(self):
1699         html = self.GET_ELEMENT_BY_TAG_TEST_STRING
1700
1701         self.assertEqual(
1702             get_element_text_and_html_by_tag('div', html),
1703             (self.GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT, self.GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML))
1704         self.assertEqual(
1705             get_element_text_and_html_by_tag('span', html),
1706             (self.GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT, self.GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML))
1707         self.assertRaises(compat_HTMLParseError, get_element_text_and_html_by_tag, 'article', html)
1708
1709     def test_iri_to_uri(self):
1710         self.assertEqual(
1711             iri_to_uri('https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b'),
1712             'https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b')  # Same
1713         self.assertEqual(
1714             iri_to_uri('https://www.google.com/search?q=Käsesoßenrührlöffel'),  # German for cheese sauce stirring spoon
1715             'https://www.google.com/search?q=K%C3%A4seso%C3%9Fenr%C3%BChrl%C3%B6ffel')
1716         self.assertEqual(
1717             iri_to_uri('https://www.google.com/search?q=lt<+gt>+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#'),
1718             'https://www.google.com/search?q=lt%3C+gt%3E+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#')
1719         self.assertEqual(
1720             iri_to_uri('http://правозащита38.рф/category/news/'),
1721             'http://xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/')
1722         self.assertEqual(
1723             iri_to_uri('http://www.правозащита38.рф/category/news/'),
1724             'http://www.xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/')
1725         self.assertEqual(
1726             iri_to_uri('https://i❤.ws/emojidomain/👍👏🤝💪'),
1727             'https://xn--i-7iq.ws/emojidomain/%F0%9F%91%8D%F0%9F%91%8F%F0%9F%A4%9D%F0%9F%92%AA')
1728         self.assertEqual(
1729             iri_to_uri('http://日本語.jp/'),
1730             'http://xn--wgv71a119e.jp/')
1731         self.assertEqual(
1732             iri_to_uri('http://导航.中国/'),
1733             'http://xn--fet810g.xn--fiqs8s/')
1734
1735     def test_clean_podcast_url(self):
1736         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')
1737         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')
1738
1739     def test_LazyList(self):
1740         it = list(range(10))
1741
1742         self.assertEqual(list(LazyList(it)), it)
1743         self.assertEqual(LazyList(it).exhaust(), it)
1744         self.assertEqual(LazyList(it)[5], it[5])
1745
1746         self.assertEqual(LazyList(it)[5:], it[5:])
1747         self.assertEqual(LazyList(it)[:5], it[:5])
1748         self.assertEqual(LazyList(it)[::2], it[::2])
1749         self.assertEqual(LazyList(it)[1::2], it[1::2])
1750         self.assertEqual(LazyList(it)[5::-1], it[5::-1])
1751         self.assertEqual(LazyList(it)[6:2:-2], it[6:2:-2])
1752         self.assertEqual(LazyList(it)[::-1], it[::-1])
1753
1754         self.assertTrue(LazyList(it))
1755         self.assertFalse(LazyList(range(0)))
1756         self.assertEqual(len(LazyList(it)), len(it))
1757         self.assertEqual(repr(LazyList(it)), repr(it))
1758         self.assertEqual(str(LazyList(it)), str(it))
1759
1760         self.assertEqual(list(LazyList(it, reverse=True)), it[::-1])
1761         self.assertEqual(list(reversed(LazyList(it))[::-1]), it)
1762         self.assertEqual(list(reversed(LazyList(it))[1:3:7]), it[::-1][1:3:7])
1763
1764     def test_LazyList_laziness(self):
1765
1766         def test(ll, idx, val, cache):
1767             self.assertEqual(ll[idx], val)
1768             self.assertEqual(ll._cache, list(cache))
1769
1770         ll = LazyList(range(10))
1771         test(ll, 0, 0, range(1))
1772         test(ll, 5, 5, range(6))
1773         test(ll, -3, 7, range(10))
1774
1775         ll = LazyList(range(10), reverse=True)
1776         test(ll, -1, 0, range(1))
1777         test(ll, 3, 6, range(10))
1778
1779         ll = LazyList(itertools.count())
1780         test(ll, 10, 10, range(11))
1781         ll = reversed(ll)
1782         test(ll, -15, 14, range(15))
1783
1784     def test_format_bytes(self):
1785         self.assertEqual(format_bytes(0), '0.00B')
1786         self.assertEqual(format_bytes(1000), '1000.00B')
1787         self.assertEqual(format_bytes(1024), '1.00KiB')
1788         self.assertEqual(format_bytes(1024**2), '1.00MiB')
1789         self.assertEqual(format_bytes(1024**3), '1.00GiB')
1790         self.assertEqual(format_bytes(1024**4), '1.00TiB')
1791         self.assertEqual(format_bytes(1024**5), '1.00PiB')
1792         self.assertEqual(format_bytes(1024**6), '1.00EiB')
1793         self.assertEqual(format_bytes(1024**7), '1.00ZiB')
1794         self.assertEqual(format_bytes(1024**8), '1.00YiB')
1795         self.assertEqual(format_bytes(1024**9), '1024.00YiB')
1796
1797     def test_hide_login_info(self):
1798         self.assertEqual(Config.hide_login_info(['-u', 'foo', '-p', 'bar']),
1799                          ['-u', 'PRIVATE', '-p', 'PRIVATE'])
1800         self.assertEqual(Config.hide_login_info(['-u']), ['-u'])
1801         self.assertEqual(Config.hide_login_info(['-u', 'foo', '-u', 'bar']),
1802                          ['-u', 'PRIVATE', '-u', 'PRIVATE'])
1803         self.assertEqual(Config.hide_login_info(['--username=foo']),
1804                          ['--username=PRIVATE'])
1805
1806     def test_locked_file(self):
1807         TEXT = 'test_locked_file\n'
1808         FILE = 'test_locked_file.ytdl'
1809         MODES = 'war'  # Order is important
1810
1811         try:
1812             for lock_mode in MODES:
1813                 with locked_file(FILE, lock_mode, False) as f:
1814                     if lock_mode == 'r':
1815                         self.assertEqual(f.read(), TEXT * 2, 'Wrong file content')
1816                     else:
1817                         f.write(TEXT)
1818                     for test_mode in MODES:
1819                         testing_write = test_mode != 'r'
1820                         try:
1821                             with locked_file(FILE, test_mode, False):
1822                                 pass
1823                         except (BlockingIOError, PermissionError):
1824                             if not testing_write:  # FIXME
1825                                 print(f'Known issue: Exclusive lock ({lock_mode}) blocks read access ({test_mode})')
1826                                 continue
1827                             self.assertTrue(testing_write, f'{test_mode} is blocked by {lock_mode}')
1828                         else:
1829                             self.assertFalse(testing_write, f'{test_mode} is not blocked by {lock_mode}')
1830         finally:
1831             with contextlib.suppress(OSError):
1832                 os.remove(FILE)
1833
1834     def test_determine_file_encoding(self):
1835         self.assertEqual(determine_file_encoding(b''), (None, 0))
1836         self.assertEqual(determine_file_encoding(b'--verbose -x --audio-format mkv\n'), (None, 0))
1837
1838         self.assertEqual(determine_file_encoding(b'\xef\xbb\xbf'), ('utf-8', 3))
1839         self.assertEqual(determine_file_encoding(b'\x00\x00\xfe\xff'), ('utf-32-be', 4))
1840         self.assertEqual(determine_file_encoding(b'\xff\xfe'), ('utf-16-le', 2))
1841
1842         self.assertEqual(determine_file_encoding(b'\xff\xfe# coding: utf-8\n--verbose'), ('utf-16-le', 2))
1843
1844         self.assertEqual(determine_file_encoding(b'# coding: utf-8\n--verbose'), ('utf-8', 0))
1845         self.assertEqual(determine_file_encoding(b'# coding: someencodinghere-12345\n--verbose'), ('someencodinghere-12345', 0))
1846
1847         self.assertEqual(determine_file_encoding(b'#coding:utf-8\n--verbose'), ('utf-8', 0))
1848         self.assertEqual(determine_file_encoding(b'#  coding:   utf-8   \r\n--verbose'), ('utf-8', 0))
1849
1850         self.assertEqual(determine_file_encoding('# coding: utf-32-be'.encode('utf-32-be')), ('utf-32-be', 0))
1851         self.assertEqual(determine_file_encoding('# coding: utf-16-le'.encode('utf-16-le')), ('utf-16-le', 0))
1852
1853     def test_get_compatible_ext(self):
1854         self.assertEqual(get_compatible_ext(
1855             vcodecs=[None], acodecs=[None, None], vexts=['mp4'], aexts=['m4a', 'm4a']), 'mkv')
1856         self.assertEqual(get_compatible_ext(
1857             vcodecs=[None], acodecs=[None], vexts=['flv'], aexts=['flv']), 'flv')
1858
1859         self.assertEqual(get_compatible_ext(
1860             vcodecs=[None], acodecs=[None], vexts=['mp4'], aexts=['m4a']), 'mp4')
1861         self.assertEqual(get_compatible_ext(
1862             vcodecs=[None], acodecs=[None], vexts=['mp4'], aexts=['webm']), 'mkv')
1863         self.assertEqual(get_compatible_ext(
1864             vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['m4a']), 'mkv')
1865         self.assertEqual(get_compatible_ext(
1866             vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['webm']), 'webm')
1867
1868         self.assertEqual(get_compatible_ext(
1869             vcodecs=['h264'], acodecs=['mp4a'], vexts=['mov'], aexts=['m4a']), 'mp4')
1870         self.assertEqual(get_compatible_ext(
1871             vcodecs=['av01.0.12M.08'], acodecs=['opus'], vexts=['mp4'], aexts=['webm']), 'webm')
1872
1873         self.assertEqual(get_compatible_ext(
1874             vcodecs=['vp9'], acodecs=['opus'], vexts=['webm'], aexts=['webm'], preferences=['flv', 'mp4']), 'mp4')
1875         self.assertEqual(get_compatible_ext(
1876             vcodecs=['av1'], acodecs=['mp4a'], vexts=['webm'], aexts=['m4a'], preferences=('webm', 'mkv')), 'mkv')
1877
1878     def test_traverse_obj(self):
1879         _TEST_DATA = {
1880             100: 100,
1881             1.2: 1.2,
1882             'str': 'str',
1883             'None': None,
1884             '...': ...,
1885             'urls': [
1886                 {'index': 0, 'url': 'https://www.example.com/0'},
1887                 {'index': 1, 'url': 'https://www.example.com/1'},
1888             ],
1889             'data': (
1890                 {'index': 2},
1891                 {'index': 3},
1892             ),
1893         }
1894
1895         # Test base functionality
1896         self.assertEqual(traverse_obj(_TEST_DATA, ('str',)), 'str',
1897                          msg='allow tuple path')
1898         self.assertEqual(traverse_obj(_TEST_DATA, ['str']), 'str',
1899                          msg='allow list path')
1900         self.assertEqual(traverse_obj(_TEST_DATA, (value for value in ("str",))), 'str',
1901                          msg='allow iterable path')
1902         self.assertEqual(traverse_obj(_TEST_DATA, 'str'), 'str',
1903                          msg='single items should be treated as a path')
1904         self.assertEqual(traverse_obj(_TEST_DATA, None), _TEST_DATA)
1905         self.assertEqual(traverse_obj(_TEST_DATA, 100), 100)
1906         self.assertEqual(traverse_obj(_TEST_DATA, 1.2), 1.2)
1907
1908         # Test Ellipsis behavior
1909         self.assertCountEqual(traverse_obj(_TEST_DATA, ...),
1910                               (item for item in _TEST_DATA.values() if item is not None),
1911                               msg='`...` should give all values except `None`')
1912         self.assertCountEqual(traverse_obj(_TEST_DATA, ('urls', 0, ...)), _TEST_DATA['urls'][0].values(),
1913                               msg='`...` selection for dicts should select all values')
1914         self.assertEqual(traverse_obj(_TEST_DATA, (..., ..., 'url')),
1915                          ['https://www.example.com/0', 'https://www.example.com/1'],
1916                          msg='nested `...` queries should work')
1917         self.assertCountEqual(traverse_obj(_TEST_DATA, (..., ..., 'index')), range(4),
1918                               msg='`...` query result should be flattened')
1919
1920         # Test function as key
1921         self.assertEqual(traverse_obj(_TEST_DATA, lambda x, y: x == 'urls' and isinstance(y, list)),
1922                          [_TEST_DATA['urls']],
1923                          msg='function as query key should perform a filter based on (key, value)')
1924         self.assertCountEqual(traverse_obj(_TEST_DATA, lambda _, x: isinstance(x[0], str)), {'str'},
1925                               msg='exceptions in the query function should be catched')
1926
1927         # Test alternative paths
1928         self.assertEqual(traverse_obj(_TEST_DATA, 'fail', 'str'), 'str',
1929                          msg='multiple `path_list` should be treated as alternative paths')
1930         self.assertEqual(traverse_obj(_TEST_DATA, 'str', 100), 'str',
1931                          msg='alternatives should exit early')
1932         self.assertEqual(traverse_obj(_TEST_DATA, 'fail', 'fail'), None,
1933                          msg='alternatives should return `default` if exhausted')
1934
1935         # Test branch and path nesting
1936         self.assertEqual(traverse_obj(_TEST_DATA, ('urls', (3, 0), 'url')), ['https://www.example.com/0'],
1937                          msg='tuple as key should be treated as branches')
1938         self.assertEqual(traverse_obj(_TEST_DATA, ('urls', [3, 0], 'url')), ['https://www.example.com/0'],
1939                          msg='list as key should be treated as branches')
1940         self.assertEqual(traverse_obj(_TEST_DATA, ('urls', ((1, 'fail'), (0, 'url')))), ['https://www.example.com/0'],
1941                          msg='double nesting in path should be treated as paths')
1942         self.assertEqual(traverse_obj(['0', [1, 2]], [(0, 1), 0]), [1],
1943                          msg='do not fail early on branching')
1944         self.assertCountEqual(traverse_obj(_TEST_DATA, ('urls', ((1, ('fail', 'url')), (0, 'url')))),
1945                               ['https://www.example.com/0', 'https://www.example.com/1'],
1946                               msg='tripple nesting in path should be treated as branches')
1947         self.assertEqual(traverse_obj(_TEST_DATA, ('urls', ('fail', (..., 'url')))),
1948                          ['https://www.example.com/0', 'https://www.example.com/1'],
1949                          msg='ellipsis as branch path start gets flattened')
1950
1951         # Test dictionary as key
1952         self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2}), {0: 100, 1: 1.2},
1953                          msg='dict key should result in a dict with the same keys')
1954         self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', 0, 'url')}),
1955                          {0: 'https://www.example.com/0'},
1956                          msg='dict key should allow paths')
1957         self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', (3, 0), 'url')}),
1958                          {0: ['https://www.example.com/0']},
1959                          msg='tuple in dict path should be treated as branches')
1960         self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', ((1, 'fail'), (0, 'url')))}),
1961                          {0: ['https://www.example.com/0']},
1962                          msg='double nesting in dict path should be treated as paths')
1963         self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', ((1, ('fail', 'url')), (0, 'url')))}),
1964                          {0: ['https://www.example.com/1', 'https://www.example.com/0']},
1965                          msg='tripple nesting in dict path should be treated as branches')
1966         self.assertEqual(traverse_obj({}, {0: 1}, default=...), {0: ...},
1967                          msg='do not remove `None` values when dict key')
1968
1969         # Testing default parameter behavior
1970         _DEFAULT_DATA = {'None': None, 'int': 0, 'list': []}
1971         self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail'), None,
1972                          msg='default value should be `None`')
1973         self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail', 'fail', default=...), ...,
1974                          msg='chained fails should result in default')
1975         self.assertEqual(traverse_obj(_DEFAULT_DATA, 'None', 'int'), 0,
1976                          msg='should not short cirquit on `None`')
1977         self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail', default=1), 1,
1978                          msg='invalid dict key should result in `default`')
1979         self.assertEqual(traverse_obj(_DEFAULT_DATA, 'None', default=1), 1,
1980                          msg='`None` is a deliberate sentinel and should become `default`')
1981         self.assertEqual(traverse_obj(_DEFAULT_DATA, ('list', 10)), None,
1982                          msg='`IndexError` should result in `default`')
1983         self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail'), default=1), 1,
1984                          msg='if branched but not successfull return `default`, not `[]`')
1985
1986         # Testing expected_type behavior
1987         _EXPECTED_TYPE_DATA = {'str': 'str', 'int': 0}
1988         self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=str), 'str',
1989                          msg='accept matching `expected_type` type')
1990         self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=int), None,
1991                          msg='reject non matching `expected_type` type')
1992         self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'int', expected_type=lambda x: str(x)), '0',
1993                          msg='transform type using type function')
1994         self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str',
1995                                       expected_type=lambda _: 1 / 0), None,
1996                          msg='wrap expected_type fuction in try_call')
1997         self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, ..., expected_type=str), ['str'],
1998                          msg='eliminate items that expected_type fails on')
1999
2000         # Test get_all behavior
2001         _GET_ALL_DATA = {'key': [0, 1, 2]}
2002         self.assertEqual(traverse_obj(_GET_ALL_DATA, ('key', ...), get_all=False), 0,
2003                          msg='if not `get_all`, return only first matching value')
2004         self.assertEqual(traverse_obj(_GET_ALL_DATA, ..., get_all=False), [0, 1, 2],
2005                          msg='do not overflatten if not `get_all`')
2006
2007         # Test casesense behavior
2008         _CASESENSE_DATA = {
2009             'KeY': 'value0',
2010             0: {
2011                 'KeY': 'value1',
2012                 0: {'KeY': 'value2'},
2013             },
2014         }
2015         self.assertEqual(traverse_obj(_CASESENSE_DATA, 'key'), None,
2016                          msg='dict keys should be case sensitive unless `casesense`')
2017         self.assertEqual(traverse_obj(_CASESENSE_DATA, 'keY',
2018                                       casesense=False), 'value0',
2019                          msg='allow non matching key case if `casesense`')
2020         self.assertEqual(traverse_obj(_CASESENSE_DATA, (0, ('keY',)),
2021                                       casesense=False), ['value1'],
2022                          msg='allow non matching key case in branch if `casesense`')
2023         self.assertEqual(traverse_obj(_CASESENSE_DATA, (0, ((0, 'keY'),)),
2024                                       casesense=False), ['value2'],
2025                          msg='allow non matching key case in branch path if `casesense`')
2026
2027         # Test traverse_string behavior
2028         _TRAVERSE_STRING_DATA = {'str': 'str', 1.2: 1.2}
2029         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', 0)), None,
2030                          msg='do not traverse into string if not `traverse_string`')
2031         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', 0),
2032                                       traverse_string=True), 's',
2033                          msg='traverse into string if `traverse_string`')
2034         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, (1.2, 1),
2035                                       traverse_string=True), '.',
2036                          msg='traverse into converted data if `traverse_string`')
2037         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', ...),
2038                                       traverse_string=True), list('str'),
2039                          msg='`...` branching into string should result in list')
2040         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', (0, 2)),
2041                                       traverse_string=True), ['s', 'r'],
2042                          msg='branching into string should result in list')
2043         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', lambda _, x: x),
2044                                       traverse_string=True), list('str'),
2045                          msg='function branching into string should result in list')
2046
2047         # Test is_user_input behavior
2048         _IS_USER_INPUT_DATA = {'range8': list(range(8))}
2049         self.assertEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', '3'),
2050                                       is_user_input=True), 3,
2051                          msg='allow for string indexing if `is_user_input`')
2052         self.assertCountEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', '3:'),
2053                                            is_user_input=True), tuple(range(8))[3:],
2054                               msg='allow for string slice if `is_user_input`')
2055         self.assertCountEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', ':4:2'),
2056                                            is_user_input=True), tuple(range(8))[:4:2],
2057                               msg='allow step in string slice if `is_user_input`')
2058         self.assertCountEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', ':'),
2059                                            is_user_input=True), range(8),
2060                               msg='`:` should be treated as `...` if `is_user_input`')
2061         with self.assertRaises(TypeError, msg='too many params should result in error'):
2062             traverse_obj(_IS_USER_INPUT_DATA, ('range8', ':::'), is_user_input=True)
2063
2064
2065 if __name__ == '__main__':
2066     unittest.main()