test/test_utils.py

   1 #!/usr/bin/env python3
   2
   3 # Allow direct execution
   4 import os
   5 import sys
   6 import unittest
   7 import warnings
   8 import datetime as dt
   9
  10 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
  11
  12
  13 import contextlib
  14 import io
  15 import itertools
  16 import json
  17 import subprocess
  18 import xml.etree.ElementTree
  19
  20 from yt_dlp.compat import (
  21     compat_etree_fromstring,
  22     compat_HTMLParseError,
  23     compat_os_name,
  24 )
  25 from yt_dlp.utils import (
  26     Config,
  27     DateRange,
  28     ExtractorError,
  29     InAdvancePagedList,
  30     LazyList,
  31     NO_DEFAULT,
  32     OnDemandPagedList,
  33     Popen,
  34     age_restricted,
  35     args_to_str,
  36     base_url,
  37     caesar,
  38     clean_html,
  39     clean_podcast_url,
  40     cli_bool_option,
  41     cli_option,
  42     cli_valueless_option,
  43     date_from_str,
  44     datetime_from_str,
  45     detect_exe_version,
  46     determine_ext,
  47     determine_file_encoding,
  48     dfxp2srt,
  49     encode_base_n,
  50     encode_compat_str,
  51     encodeFilename,
  52     expand_path,
  53     extract_attributes,
  54     extract_basic_auth,
  55     find_xpath_attr,
  56     fix_xml_ampersands,
  57     float_or_none,
  58     format_bytes,
  59     get_compatible_ext,
  60     get_element_by_attribute,
  61     get_element_by_class,
  62     get_element_html_by_attribute,
  63     get_element_html_by_class,
  64     get_element_text_and_html_by_tag,
  65     get_elements_by_attribute,
  66     get_elements_by_class,
  67     get_elements_html_by_attribute,
  68     get_elements_html_by_class,
  69     get_elements_text_and_html_by_attribute,
  70     int_or_none,
  71     intlist_to_bytes,
  72     iri_to_uri,
  73     is_html,
  74     js_to_json,
  75     limit_length,
  76     locked_file,
  77     lowercase_escape,
  78     match_str,
  79     merge_dicts,
  80     mimetype2ext,
  81     month_by_name,
  82     multipart_encode,
  83     ohdave_rsa_encrypt,
  84     orderedSet,
  85     parse_age_limit,
  86     parse_bitrate,
  87     parse_codecs,
  88     parse_count,
  89     parse_dfxp_time_expr,
  90     parse_duration,
  91     parse_filesize,
  92     parse_iso8601,
  93     parse_qs,
  94     parse_resolution,
  95     pkcs1pad,
  96     prepend_extension,
  97     read_batch_urls,
  98     remove_end,
  99     remove_quotes,
 100     remove_start,
 101     render_table,
 102     replace_extension,
 103     rot47,
 104     sanitize_filename,
 105     sanitize_path,
 106     sanitize_url,
 107     shell_quote,
 108     smuggle_url,
 109     str_to_int,
 110     strip_jsonp,
 111     strip_or_none,
 112     subtitles_filename,
 113     timeconvert,
 114     try_call,
 115     unescapeHTML,
 116     unified_strdate,
 117     unified_timestamp,
 118     unsmuggle_url,
 119     update_url_query,
 120     uppercase_escape,
 121     url_basename,
 122     url_or_none,
 123     urlencode_postdata,
 124     urljoin,
 125     urshift,
 126     variadic,
 127     version_tuple,
 128     xpath_attr,
 129     xpath_element,
 130     xpath_text,
 131     xpath_with_ns,
 132 )
 133 from yt_dlp.utils.networking import (
 134     HTTPHeaderDict,
 135     escape_rfc3986,
 136     normalize_url,
 137     remove_dot_segments,
 138 )
 139
 140
 141 class TestUtil(unittest.TestCase):
 142     def test_timeconvert(self):
 143         self.assertTrue(timeconvert('') is None)
 144         self.assertTrue(timeconvert('bougrg') is None)
 145
 146     def test_sanitize_filename(self):
 147         self.assertEqual(sanitize_filename(''), '')
 148         self.assertEqual(sanitize_filename('abc'), 'abc')
 149         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
 150
 151         self.assertEqual(sanitize_filename('123'), '123')
 152
 153         self.assertEqual('abc⧸de', sanitize_filename('abc/de'))
 154         self.assertFalse('/' in sanitize_filename('abc/de///'))
 155
 156         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', is_id=False))
 157         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', is_id=False))
 158         self.assertEqual('yes no', sanitize_filename('yes? no', is_id=False))
 159         self.assertEqual('this - that', sanitize_filename('this: that', is_id=False))
 160
 161         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
 162         aumlaut = 'ä'
 163         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
 164         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
 165         self.assertEqual(sanitize_filename(tests), tests)
 166
 167         self.assertEqual(
 168             sanitize_filename('New World record at 0:12:34'),
 169             'New World record at 0_12_34')
 170
 171         self.assertEqual(sanitize_filename('--gasdgf'), '--gasdgf')
 172         self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
 173         self.assertEqual(sanitize_filename('--gasdgf', is_id=False), '_-gasdgf')
 174         self.assertEqual(sanitize_filename('.gasdgf'), '.gasdgf')
 175         self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
 176         self.assertEqual(sanitize_filename('.gasdgf', is_id=False), 'gasdgf')
 177
 178         forbidden = '"\0\\/'
 179         for fc in forbidden:
 180             for fbc in forbidden:
 181                 self.assertTrue(fbc not in sanitize_filename(fc))
 182
 183     def test_sanitize_filename_restricted(self):
 184         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
 185         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
 186
 187         self.assertEqual(sanitize_filename('123', restricted=True), '123')
 188
 189         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
 190         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
 191
 192         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
 193         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
 194         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
 195         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
 196
 197         tests = 'aäb\u4e2d\u56fd\u7684c'
 198         self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
 199         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
 200
 201         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
 202         for fc in forbidden:
 203             for fbc in forbidden:
 204                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
 205
 206         # Handle a common case more neatly
 207         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
 208         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
 209         # .. but make sure the file name is never empty
 210         self.assertTrue(sanitize_filename('-', restricted=True) != '')
 211         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 212
 213         self.assertEqual(sanitize_filename(
 214             'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
 215             'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy')
 216
 217     def test_sanitize_ids(self):
 218         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 219         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 220         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 221
 222     def test_sanitize_path(self):
 223         if sys.platform != 'win32':
 224             return
 225
 226         self.assertEqual(sanitize_path('abc'), 'abc')
 227         self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
 228         self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
 229         self.assertEqual(sanitize_path('abc|def'), 'abc#def')
 230         self.assertEqual(sanitize_path('<>:"|?*'), '#######')
 231         self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
 232         self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
 233
 234         self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
 235         self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
 236
 237         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 238         self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
 239         self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
 240         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 241
 242         self.assertEqual(
 243             sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
 244             'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
 245
 246         self.assertEqual(
 247             sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
 248             'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
 249         self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
 250         self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
 251         self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
 252
 253         self.assertEqual(sanitize_path('../abc'), '..\\abc')
 254         self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
 255         self.assertEqual(sanitize_path('./abc'), 'abc')
 256         self.assertEqual(sanitize_path('./../abc'), '..\\abc')
 257
 258     def test_sanitize_url(self):
 259         self.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
 260         self.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
 261         self.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
 262         self.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')
 263         self.assertEqual(sanitize_url('foo bar'), 'foo bar')
 264
 265     def test_expand_path(self):
 266         def env(var):
 267             return f'%{var}%' if sys.platform == 'win32' else f'${var}'
 268
 269         os.environ['yt_dlp_EXPATH_PATH'] = 'expanded'
 270         self.assertEqual(expand_path(env('yt_dlp_EXPATH_PATH')), 'expanded')
 271
 272         old_home = os.environ.get('HOME')
 273         test_str = R'C:\Documents and Settings\тест\Application Data'
 274         try:
 275             os.environ['HOME'] = test_str
 276             self.assertEqual(expand_path(env('HOME')), os.getenv('HOME'))
 277             self.assertEqual(expand_path('~'), os.getenv('HOME'))
 278             self.assertEqual(
 279                 expand_path('~/%s' % env('yt_dlp_EXPATH_PATH')),
 280                 '%s/expanded' % os.getenv('HOME'))
 281         finally:
 282             os.environ['HOME'] = old_home or ''
 283
 284     def test_prepend_extension(self):
 285         self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
 286         self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
 287         self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 288         self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
 289         self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
 290         self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
 291
 292     def test_replace_extension(self):
 293         self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
 294         self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
 295         self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 296         self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
 297         self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
 298         self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
 299
 300     def test_subtitles_filename(self):
 301         self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt'), 'abc.en.vtt')
 302         self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt', 'ext'), 'abc.en.vtt')
 303         self.assertEqual(subtitles_filename('abc.unexpected_ext', 'en', 'vtt', 'ext'), 'abc.unexpected_ext.en.vtt')
 304
 305     def test_remove_start(self):
 306         self.assertEqual(remove_start(None, 'A - '), None)
 307         self.assertEqual(remove_start('A - B', 'A - '), 'B')
 308         self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
 309
 310     def test_remove_end(self):
 311         self.assertEqual(remove_end(None, ' - B'), None)
 312         self.assertEqual(remove_end('A - B', ' - B'), 'A')
 313         self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
 314
 315     def test_remove_quotes(self):
 316         self.assertEqual(remove_quotes(None), None)
 317         self.assertEqual(remove_quotes('"'), '"')
 318         self.assertEqual(remove_quotes("'"), "'")
 319         self.assertEqual(remove_quotes(';'), ';')
 320         self.assertEqual(remove_quotes('";'), '";')
 321         self.assertEqual(remove_quotes('""'), '')
 322         self.assertEqual(remove_quotes('";"'), ';')
 323
 324     def test_ordered_set(self):
 325         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
 326         self.assertEqual(orderedSet([]), [])
 327         self.assertEqual(orderedSet([1]), [1])
 328         # keep the list ordered
 329         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
 330
 331     def test_unescape_html(self):
 332         self.assertEqual(unescapeHTML('%20;'), '%20;')
 333         self.assertEqual(unescapeHTML('&#x2F;'), '/')
 334         self.assertEqual(unescapeHTML('&#47;'), '/')
 335         self.assertEqual(unescapeHTML('&eacute;'), 'é')
 336         self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
 337         self.assertEqual(unescapeHTML('&a&quot;'), '&a"')
 338         # HTML5 entities
 339         self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')
 340
 341     def test_date_from_str(self):
 342         self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
 343         self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
 344         self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
 345         self.assertEqual(date_from_str('20200229+365day'), date_from_str('20200229+1year'))
 346         self.assertEqual(date_from_str('20210131+28day'), date_from_str('20210131+1month'))
 347
 348     def test_datetime_from_str(self):
 349         self.assertEqual(datetime_from_str('yesterday', precision='day'), datetime_from_str('now-1day', precision='auto'))
 350         self.assertEqual(datetime_from_str('now+7day', precision='day'), datetime_from_str('now+1week', precision='auto'))
 351         self.assertEqual(datetime_from_str('now+14day', precision='day'), datetime_from_str('now+2week', precision='auto'))
 352         self.assertEqual(datetime_from_str('20200229+365day', precision='day'), datetime_from_str('20200229+1year', precision='auto'))
 353         self.assertEqual(datetime_from_str('20210131+28day', precision='day'), datetime_from_str('20210131+1month', precision='auto'))
 354         self.assertEqual(datetime_from_str('20210131+59day', precision='day'), datetime_from_str('20210131+2month', precision='auto'))
 355         self.assertEqual(datetime_from_str('now+1day', precision='hour'), datetime_from_str('now+24hours', precision='auto'))
 356         self.assertEqual(datetime_from_str('now+23hours', precision='hour'), datetime_from_str('now+23hours', precision='auto'))
 357
 358     def test_daterange(self):
 359         _20century = DateRange("19000101", "20000101")
 360         self.assertFalse("17890714" in _20century)
 361         _ac = DateRange("00010101")
 362         self.assertTrue("19690721" in _ac)
 363         _firstmilenium = DateRange(end="10000101")
 364         self.assertTrue("07110427" in _firstmilenium)
 365
 366     def test_unified_dates(self):
 367         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
 368         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
 369         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
 370         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
 371         self.assertEqual(unified_strdate('1968 12 10'), '19681210')
 372         self.assertEqual(unified_strdate('1968-12-10'), '19681210')
 373         self.assertEqual(unified_strdate('31-07-2022 20:00'), '20220731')
 374         self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
 375         self.assertEqual(
 376             unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
 377             '20141126')
 378         self.assertEqual(
 379             unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
 380             '20150202')
 381         self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
 382         self.assertEqual(unified_strdate('25-09-2014'), '20140925')
 383         self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
 384         self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
 385         self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
 386         self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
 387         self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
 388         self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
 389         self.assertEqual(unified_strdate('November 3rd, 2019'), '20191103')
 390         self.assertEqual(unified_strdate('October 23rd, 2005'), '20051023')
 391
 392     def test_unified_timestamps(self):
 393         self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
 394         self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
 395         self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
 396         self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
 397         self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
 398         self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
 399         self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
 400         self.assertEqual(
 401             unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
 402             1417001400)
 403         self.assertEqual(
 404             unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
 405             1422902860)
 406         self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
 407         self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
 408         self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
 409         self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
 410         self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
 411         self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
 412         self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
 413         self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
 414         self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
 415         self.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363)
 416
 417         self.assertEqual(unified_timestamp('December 31 1969 20:00:01 EDT'), 1)
 418         self.assertEqual(unified_timestamp('Wednesday 31 December 1969 18:01:26 MDT'), 86)
 419         self.assertEqual(unified_timestamp('12/31/1969 20:01:18 EDT', False), 78)
 420
 421     def test_determine_ext(self):
 422         self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
 423         self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
 424         self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
 425         self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
 426         self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
 427         self.assertEqual(determine_ext('foobar', None), None)
 428
 429     def test_find_xpath_attr(self):
 430         testxml = '''<root>
 431             <node/>
 432             <node x="a"/>
 433             <node x="a" y="c" />
 434             <node x="b" y="d" />
 435             <node x="" />
 436         </root>'''
 437         doc = compat_etree_fromstring(testxml)
 438
 439         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
 440         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
 441         self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
 442         self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
 443         self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
 444         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
 445         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
 446         self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
 447         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
 448         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
 449         self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
 450
 451     def test_xpath_with_ns(self):
 452         testxml = '''<root xmlns:media="http://example.com/">
 453             <media:song>
 454                 <media:author>The Author</media:author>
 455                 <url>http://server.com/download.mp3</url>
 456             </media:song>
 457         </root>'''
 458         doc = compat_etree_fromstring(testxml)
 459         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
 460         self.assertTrue(find('media:song') is not None)
 461         self.assertEqual(find('media:song/media:author').text, 'The Author')
 462         self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
 463
 464     def test_xpath_element(self):
 465         doc = xml.etree.ElementTree.Element('root')
 466         div = xml.etree.ElementTree.SubElement(doc, 'div')
 467         p = xml.etree.ElementTree.SubElement(div, 'p')
 468         p.text = 'Foo'
 469         self.assertEqual(xpath_element(doc, 'div/p'), p)
 470         self.assertEqual(xpath_element(doc, ['div/p']), p)
 471         self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
 472         self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
 473         self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
 474         self.assertTrue(xpath_element(doc, 'div/bar') is None)
 475         self.assertTrue(xpath_element(doc, ['div/bar']) is None)
 476         self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
 477         self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
 478         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
 479         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
 480
 481     def test_xpath_text(self):
 482         testxml = '''<root>
 483             <div>
 484                 <p>Foo</p>
 485             </div>
 486         </root>'''
 487         doc = compat_etree_fromstring(testxml)
 488         self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
 489         self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
 490         self.assertTrue(xpath_text(doc, 'div/bar') is None)
 491         self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
 492
 493     def test_xpath_attr(self):
 494         testxml = '''<root>
 495             <div>
 496                 <p x="a">Foo</p>
 497             </div>
 498         </root>'''
 499         doc = compat_etree_fromstring(testxml)
 500         self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
 501         self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
 502         self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
 503         self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
 504         self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
 505         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
 506         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
 507
 508     def test_smuggle_url(self):
 509         data = {"ö": "ö", "abc": [3]}
 510         url = 'https://foo.bar/baz?x=y#a'
 511         smug_url = smuggle_url(url, data)
 512         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
 513         self.assertEqual(url, unsmug_url)
 514         self.assertEqual(data, unsmug_data)
 515
 516         res_url, res_data = unsmuggle_url(url)
 517         self.assertEqual(res_url, url)
 518         self.assertEqual(res_data, None)
 519
 520         smug_url = smuggle_url(url, {'a': 'b'})
 521         smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
 522         res_url, res_data = unsmuggle_url(smug_smug_url)
 523         self.assertEqual(res_url, url)
 524         self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
 525
 526     def test_shell_quote(self):
 527         args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
 528         self.assertEqual(
 529             shell_quote(args),
 530             """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
 531
 532     def test_float_or_none(self):
 533         self.assertEqual(float_or_none('42.42'), 42.42)
 534         self.assertEqual(float_or_none('42'), 42.0)
 535         self.assertEqual(float_or_none(''), None)
 536         self.assertEqual(float_or_none(None), None)
 537         self.assertEqual(float_or_none([]), None)
 538         self.assertEqual(float_or_none(set()), None)
 539
 540     def test_int_or_none(self):
 541         self.assertEqual(int_or_none('42'), 42)
 542         self.assertEqual(int_or_none(''), None)
 543         self.assertEqual(int_or_none(None), None)
 544         self.assertEqual(int_or_none([]), None)
 545         self.assertEqual(int_or_none(set()), None)
 546
 547     def test_str_to_int(self):
 548         self.assertEqual(str_to_int('123,456'), 123456)
 549         self.assertEqual(str_to_int('123.456'), 123456)
 550         self.assertEqual(str_to_int(523), 523)
 551         self.assertEqual(str_to_int('noninteger'), None)
 552         self.assertEqual(str_to_int([]), None)
 553
 554     def test_url_basename(self):
 555         self.assertEqual(url_basename('http://foo.de/'), '')
 556         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
 557         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
 558         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
 559         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
 560         self.assertEqual(
 561             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
 562             'trailer.mp4')
 563
 564     def test_base_url(self):
 565         self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
 566         self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
 567         self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
 568         self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
 569         self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
 570         self.assertEqual(base_url('http://foo.de/bar/baz&x=z&w=y/x/c'), 'http://foo.de/bar/baz&x=z&w=y/x/')
 571
 572     def test_urljoin(self):
 573         self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 574         self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 575         self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 576         self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 577         self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
 578         self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 579         self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 580         self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 581         self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 582         self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
 583         self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 584         self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
 585         self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 586         self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 587         self.assertEqual(urljoin('http://foo.de/', None), None)
 588         self.assertEqual(urljoin('http://foo.de/', ''), None)
 589         self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
 590         self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
 591         self.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de')
 592         self.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de')
 593
 594     def test_url_or_none(self):
 595         self.assertEqual(url_or_none(None), None)
 596         self.assertEqual(url_or_none(''), None)
 597         self.assertEqual(url_or_none('foo'), None)
 598         self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
 599         self.assertEqual(url_or_none('https://foo.de'), 'https://foo.de')
 600         self.assertEqual(url_or_none('http$://foo.de'), None)
 601         self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
 602         self.assertEqual(url_or_none('//foo.de'), '//foo.de')
 603         self.assertEqual(url_or_none('s3://foo.de'), None)
 604         self.assertEqual(url_or_none('rtmpte://foo.de'), 'rtmpte://foo.de')
 605         self.assertEqual(url_or_none('mms://foo.de'), 'mms://foo.de')
 606         self.assertEqual(url_or_none('rtspu://foo.de'), 'rtspu://foo.de')
 607         self.assertEqual(url_or_none('ftps://foo.de'), 'ftps://foo.de')
 608
 609     def test_parse_age_limit(self):
 610         self.assertEqual(parse_age_limit(None), None)
 611         self.assertEqual(parse_age_limit(False), None)
 612         self.assertEqual(parse_age_limit('invalid'), None)
 613         self.assertEqual(parse_age_limit(0), 0)
 614         self.assertEqual(parse_age_limit(18), 18)
 615         self.assertEqual(parse_age_limit(21), 21)
 616         self.assertEqual(parse_age_limit(22), None)
 617         self.assertEqual(parse_age_limit('18'), 18)
 618         self.assertEqual(parse_age_limit('18+'), 18)
 619         self.assertEqual(parse_age_limit('PG-13'), 13)
 620         self.assertEqual(parse_age_limit('TV-14'), 14)
 621         self.assertEqual(parse_age_limit('TV-MA'), 17)
 622         self.assertEqual(parse_age_limit('TV14'), 14)
 623         self.assertEqual(parse_age_limit('TV_G'), 0)
 624
 625     def test_parse_duration(self):
 626         self.assertEqual(parse_duration(None), None)
 627         self.assertEqual(parse_duration(False), None)
 628         self.assertEqual(parse_duration('invalid'), None)
 629         self.assertEqual(parse_duration('1'), 1)
 630         self.assertEqual(parse_duration('1337:12'), 80232)
 631         self.assertEqual(parse_duration('9:12:43'), 33163)
 632         self.assertEqual(parse_duration('12:00'), 720)
 633         self.assertEqual(parse_duration('00:01:01'), 61)
 634         self.assertEqual(parse_duration('x:y'), None)
 635         self.assertEqual(parse_duration('3h11m53s'), 11513)
 636         self.assertEqual(parse_duration('3h 11m 53s'), 11513)
 637         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
 638         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
 639         self.assertEqual(parse_duration('3 hours, 11 minutes, 53 seconds'), 11513)
 640         self.assertEqual(parse_duration('3 hours, 11 mins, 53 secs'), 11513)
 641         self.assertEqual(parse_duration('62m45s'), 3765)
 642         self.assertEqual(parse_duration('6m59s'), 419)
 643         self.assertEqual(parse_duration('49s'), 49)
 644         self.assertEqual(parse_duration('0h0m0s'), 0)
 645         self.assertEqual(parse_duration('0m0s'), 0)
 646         self.assertEqual(parse_duration('0s'), 0)
 647         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
 648         self.assertEqual(parse_duration('T30M38S'), 1838)
 649         self.assertEqual(parse_duration('5 s'), 5)
 650         self.assertEqual(parse_duration('3 min'), 180)
 651         self.assertEqual(parse_duration('2.5 hours'), 9000)
 652         self.assertEqual(parse_duration('02:03:04'), 7384)
 653         self.assertEqual(parse_duration('01:02:03:04'), 93784)
 654         self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
 655         self.assertEqual(parse_duration('87 Min.'), 5220)
 656         self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
 657         self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
 658         self.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)
 659         self.assertEqual(parse_duration('01:02:03:050'), 3723.05)
 660         self.assertEqual(parse_duration('103:050'), 103.05)
 661         self.assertEqual(parse_duration('1HR 3MIN'), 3780)
 662         self.assertEqual(parse_duration('2hrs 3mins'), 7380)
 663
 664     def test_fix_xml_ampersands(self):
 665         self.assertEqual(
 666             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 667         self.assertEqual(
 668             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 669             '"&amp;x=y&amp;wrong;&amp;z=a')
 670         self.assertEqual(
 671             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 672             '&amp;&apos;&gt;&lt;&quot;')
 673         self.assertEqual(
 674             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 675         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 676
 677     def test_paged_list(self):
 678         def testPL(size, pagesize, sliceargs, expected):
 679             def get_page(pagenum):
 680                 firstid = pagenum * pagesize
 681                 upto = min(size, pagenum * pagesize + pagesize)
 682                 yield from range(firstid, upto)
 683
 684             pl = OnDemandPagedList(get_page, pagesize)
 685             got = pl.getslice(*sliceargs)
 686             self.assertEqual(got, expected)
 687
 688             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
 689             got = iapl.getslice(*sliceargs)
 690             self.assertEqual(got, expected)
 691
 692         testPL(5, 2, (), [0, 1, 2, 3, 4])
 693         testPL(5, 2, (1,), [1, 2, 3, 4])
 694         testPL(5, 2, (2,), [2, 3, 4])
 695         testPL(5, 2, (4,), [4])
 696         testPL(5, 2, (0, 3), [0, 1, 2])
 697         testPL(5, 2, (1, 4), [1, 2, 3])
 698         testPL(5, 2, (2, 99), [2, 3, 4])
 699         testPL(5, 2, (20, 99), [])
 700
 701     def test_read_batch_urls(self):
 702         f = io.StringIO('''\xef\xbb\xbf foo
 703             bar\r
 704             baz
 705             # More after this line\r
 706             ; or after this
 707             bam''')
 708         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
 709
 710     def test_urlencode_postdata(self):
 711         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
 712         self.assertTrue(isinstance(data, bytes))
 713
 714     def test_update_url_query(self):
 715         self.assertEqual(parse_qs(update_url_query(
 716             'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
 717             parse_qs('http://example.com/path?quality=HD&format=mp4'))
 718         self.assertEqual(parse_qs(update_url_query(
 719             'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
 720             parse_qs('http://example.com/path?system=LINUX&system=WINDOWS'))
 721         self.assertEqual(parse_qs(update_url_query(
 722             'http://example.com/path', {'fields': 'id,formats,subtitles'})),
 723             parse_qs('http://example.com/path?fields=id,formats,subtitles'))
 724         self.assertEqual(parse_qs(update_url_query(
 725             'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
 726             parse_qs('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
 727         self.assertEqual(parse_qs(update_url_query(
 728             'http://example.com/path?manifest=f4m', {'manifest': []})),
 729             parse_qs('http://example.com/path'))
 730         self.assertEqual(parse_qs(update_url_query(
 731             'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
 732             parse_qs('http://example.com/path?system=LINUX'))
 733         self.assertEqual(parse_qs(update_url_query(
 734             'http://example.com/path', {'fields': b'id,formats,subtitles'})),
 735             parse_qs('http://example.com/path?fields=id,formats,subtitles'))
 736         self.assertEqual(parse_qs(update_url_query(
 737             'http://example.com/path', {'width': 1080, 'height': 720})),
 738             parse_qs('http://example.com/path?width=1080&height=720'))
 739         self.assertEqual(parse_qs(update_url_query(
 740             'http://example.com/path', {'bitrate': 5020.43})),
 741             parse_qs('http://example.com/path?bitrate=5020.43'))
 742         self.assertEqual(parse_qs(update_url_query(
 743             'http://example.com/path', {'test': '第二行тест'})),
 744             parse_qs('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
 745
 746     def test_multipart_encode(self):
 747         self.assertEqual(
 748             multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0],
 749             b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
 750         self.assertEqual(
 751             multipart_encode({'欄位'.encode(): '值'.encode()}, boundary='AAAAAA')[0],
 752             b'--AAAAAA\r\nContent-Disposition: form-data; name="\xe6\xac\x84\xe4\xbd\x8d"\r\n\r\n\xe5\x80\xbc\r\n--AAAAAA--\r\n')
 753         self.assertRaises(
 754             ValueError, multipart_encode, {b'field': b'value'}, boundary='value')
 755
 756     def test_merge_dicts(self):
 757         self.assertEqual(merge_dicts({'a': 1}, {'b': 2}), {'a': 1, 'b': 2})
 758         self.assertEqual(merge_dicts({'a': 1}, {'a': 2}), {'a': 1})
 759         self.assertEqual(merge_dicts({'a': 1}, {'a': None}), {'a': 1})
 760         self.assertEqual(merge_dicts({'a': 1}, {'a': ''}), {'a': 1})
 761         self.assertEqual(merge_dicts({'a': 1}, {}), {'a': 1})
 762         self.assertEqual(merge_dicts({'a': None}, {'a': 1}), {'a': 1})
 763         self.assertEqual(merge_dicts({'a': ''}, {'a': 1}), {'a': ''})
 764         self.assertEqual(merge_dicts({'a': ''}, {'a': 'abc'}), {'a': 'abc'})
 765         self.assertEqual(merge_dicts({'a': None}, {'a': ''}, {'a': 'abc'}), {'a': 'abc'})
 766
 767     def test_encode_compat_str(self):
 768         self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
 769         self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
 770
 771     def test_parse_iso8601(self):
 772         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
 773         self.assertEqual(parse_iso8601('2014-03-23T23:04:26-07:00'), 1395641066)
 774         self.assertEqual(parse_iso8601('2014-03-23T23:04:26', timezone=dt.timedelta(hours=-7)), 1395641066)
 775         self.assertEqual(parse_iso8601('2014-03-23T23:04:26', timezone=NO_DEFAULT), None)
 776         # default does not override timezone in date_str
 777         self.assertEqual(parse_iso8601('2014-03-23T23:04:26-07:00', timezone=dt.timedelta(hours=-10)), 1395641066)
 778         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
 779         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
 780         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
 781         self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
 782         self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
 783
 784     def test_strip_jsonp(self):
 785         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
 786         d = json.loads(stripped)
 787         self.assertEqual(d, [{"id": "532cb", "x": 3}])
 788
 789         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
 790         d = json.loads(stripped)
 791         self.assertEqual(d, {'STATUS': 'OK'})
 792
 793         stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
 794         d = json.loads(stripped)
 795         self.assertEqual(d, {'status': 'success'})
 796
 797         stripped = strip_jsonp('window.cb && window.cb({"status": "success"});')
 798         d = json.loads(stripped)
 799         self.assertEqual(d, {'status': 'success'})
 800
 801         stripped = strip_jsonp('window.cb && cb({"status": "success"});')
 802         d = json.loads(stripped)
 803         self.assertEqual(d, {'status': 'success'})
 804
 805         stripped = strip_jsonp('({"status": "success"});')
 806         d = json.loads(stripped)
 807         self.assertEqual(d, {'status': 'success'})
 808
 809     def test_strip_or_none(self):
 810         self.assertEqual(strip_or_none(' abc'), 'abc')
 811         self.assertEqual(strip_or_none('abc '), 'abc')
 812         self.assertEqual(strip_or_none(' abc '), 'abc')
 813         self.assertEqual(strip_or_none('\tabc\t'), 'abc')
 814         self.assertEqual(strip_or_none('\n\tabc\n\t'), 'abc')
 815         self.assertEqual(strip_or_none('abc'), 'abc')
 816         self.assertEqual(strip_or_none(''), '')
 817         self.assertEqual(strip_or_none(None), None)
 818         self.assertEqual(strip_or_none(42), None)
 819         self.assertEqual(strip_or_none([]), None)
 820
 821     def test_uppercase_escape(self):
 822         self.assertEqual(uppercase_escape('aä'), 'aä')
 823         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
 824
 825     def test_lowercase_escape(self):
 826         self.assertEqual(lowercase_escape('aä'), 'aä')
 827         self.assertEqual(lowercase_escape('\\u0026'), '&')
 828
 829     def test_limit_length(self):
 830         self.assertEqual(limit_length(None, 12), None)
 831         self.assertEqual(limit_length('foo', 12), 'foo')
 832         self.assertTrue(
 833             limit_length('foo bar baz asd', 12).startswith('foo bar'))
 834         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
 835
 836     def test_mimetype2ext(self):
 837         self.assertEqual(mimetype2ext(None), None)
 838         self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
 839         self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
 840         self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
 841         self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
 842         self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
 843         self.assertEqual(mimetype2ext('audio/x-wav'), 'wav')
 844         self.assertEqual(mimetype2ext('audio/x-wav;codec=pcm'), 'wav')
 845
 846     def test_month_by_name(self):
 847         self.assertEqual(month_by_name(None), None)
 848         self.assertEqual(month_by_name('December', 'en'), 12)
 849         self.assertEqual(month_by_name('décembre', 'fr'), 12)
 850         self.assertEqual(month_by_name('December'), 12)
 851         self.assertEqual(month_by_name('décembre'), None)
 852         self.assertEqual(month_by_name('Unknown', 'unknown'), None)
 853
 854     def test_parse_codecs(self):
 855         self.assertEqual(parse_codecs(''), {})
 856         self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
 857             'vcodec': 'avc1.77.30',
 858             'acodec': 'mp4a.40.2',
 859             'dynamic_range': None,
 860         })
 861         self.assertEqual(parse_codecs('mp4a.40.2'), {
 862             'vcodec': 'none',
 863             'acodec': 'mp4a.40.2',
 864             'dynamic_range': None,
 865         })
 866         self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
 867             'vcodec': 'avc1.42001e',
 868             'acodec': 'mp4a.40.5',
 869             'dynamic_range': None,
 870         })
 871         self.assertEqual(parse_codecs('avc3.640028'), {
 872             'vcodec': 'avc3.640028',
 873             'acodec': 'none',
 874             'dynamic_range': None,
 875         })
 876         self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
 877             'vcodec': 'h264',
 878             'acodec': 'aac',
 879             'dynamic_range': None,
 880         })
 881         self.assertEqual(parse_codecs('av01.0.05M.08'), {
 882             'vcodec': 'av01.0.05M.08',
 883             'acodec': 'none',
 884             'dynamic_range': None,
 885         })
 886         self.assertEqual(parse_codecs('vp9.2'), {
 887             'vcodec': 'vp9.2',
 888             'acodec': 'none',
 889             'dynamic_range': 'HDR10',
 890         })
 891         self.assertEqual(parse_codecs('av01.0.12M.10.0.110.09.16.09.0'), {
 892             'vcodec': 'av01.0.12M.10.0.110.09.16.09.0',
 893             'acodec': 'none',
 894             'dynamic_range': 'HDR10',
 895         })
 896         self.assertEqual(parse_codecs('dvhe'), {
 897             'vcodec': 'dvhe',
 898             'acodec': 'none',
 899             'dynamic_range': 'DV',
 900         })
 901         self.assertEqual(parse_codecs('theora, vorbis'), {
 902             'vcodec': 'theora',
 903             'acodec': 'vorbis',
 904             'dynamic_range': None,
 905         })
 906         self.assertEqual(parse_codecs('unknownvcodec, unknownacodec'), {
 907             'vcodec': 'unknownvcodec',
 908             'acodec': 'unknownacodec',
 909         })
 910         self.assertEqual(parse_codecs('unknown'), {})
 911
 912     def test_escape_rfc3986(self):
 913         reserved = "!*'();:@&=+$,/?#[]"
 914         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
 915         self.assertEqual(escape_rfc3986(reserved), reserved)
 916         self.assertEqual(escape_rfc3986(unreserved), unreserved)
 917         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
 918         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
 919         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
 920         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
 921
 922     def test_normalize_url(self):
 923         self.assertEqual(
 924             normalize_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
 925             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
 926         )
 927         self.assertEqual(
 928             normalize_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
 929             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
 930         )
 931         self.assertEqual(
 932             normalize_url('http://тест.рф/фрагмент'),
 933             'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
 934         )
 935         self.assertEqual(
 936             normalize_url('http://тест.рф/абв?абв=абв#абв'),
 937             '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'
 938         )
 939         self.assertEqual(normalize_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
 940
 941         self.assertEqual(normalize_url('http://www.example.com/../a/b/../c/./d.html'), 'http://www.example.com/a/c/d.html')
 942
 943     def test_remove_dot_segments(self):
 944         self.assertEqual(remove_dot_segments('/a/b/c/./../../g'), '/a/g')
 945         self.assertEqual(remove_dot_segments('mid/content=5/../6'), 'mid/6')
 946         self.assertEqual(remove_dot_segments('/ad/../cd'), '/cd')
 947         self.assertEqual(remove_dot_segments('/ad/../cd/'), '/cd/')
 948         self.assertEqual(remove_dot_segments('/..'), '/')
 949         self.assertEqual(remove_dot_segments('/./'), '/')
 950         self.assertEqual(remove_dot_segments('/./a'), '/a')
 951         self.assertEqual(remove_dot_segments('/abc/./.././d/././e/.././f/./../../ghi'), '/ghi')
 952         self.assertEqual(remove_dot_segments('/'), '/')
 953         self.assertEqual(remove_dot_segments('/t'), '/t')
 954         self.assertEqual(remove_dot_segments('t'), 't')
 955         self.assertEqual(remove_dot_segments(''), '')
 956         self.assertEqual(remove_dot_segments('/../a/b/c'), '/a/b/c')
 957         self.assertEqual(remove_dot_segments('../a'), 'a')
 958         self.assertEqual(remove_dot_segments('./a'), 'a')
 959         self.assertEqual(remove_dot_segments('.'), '')
 960         self.assertEqual(remove_dot_segments('////'), '////')
 961
 962     def test_js_to_json_vars_strings(self):
 963         self.assertDictEqual(
 964             json.loads(js_to_json(
 965                 '''{
 966                     'null': a,
 967                     'nullStr': b,
 968                     'true': c,
 969                     'trueStr': d,
 970                     'false': e,
 971                     'falseStr': f,
 972                     'unresolvedVar': g,
 973                 }''',
 974                 {
 975                     'a': 'null',
 976                     'b': '"null"',
 977                     'c': 'true',
 978                     'd': '"true"',
 979                     'e': 'false',
 980                     'f': '"false"',
 981                     'g': 'var',
 982                 }
 983             )),
 984             {
 985                 'null': None,
 986                 'nullStr': 'null',
 987                 'true': True,
 988                 'trueStr': 'true',
 989                 'false': False,
 990                 'falseStr': 'false',
 991                 'unresolvedVar': 'var'
 992             }
 993         )
 994
 995         self.assertDictEqual(
 996             json.loads(js_to_json(
 997                 '''{
 998                     'int': a,
 999                     'intStr': b,
1000                     'float': c,
1001                     'floatStr': d,
1002                 }''',
1003                 {
1004                     'a': '123',
1005                     'b': '"123"',
1006                     'c': '1.23',
1007                     'd': '"1.23"',
1008                 }
1009             )),
1010             {
1011                 'int': 123,
1012                 'intStr': '123',
1013                 'float': 1.23,
1014                 'floatStr': '1.23',
1015             }
1016         )
1017
1018         self.assertDictEqual(
1019             json.loads(js_to_json(
1020                 '''{
1021                     'object': a,
1022                     'objectStr': b,
1023                     'array': c,
1024                     'arrayStr': d,
1025                 }''',
1026                 {
1027                     'a': '{}',
1028                     'b': '"{}"',
1029                     'c': '[]',
1030                     'd': '"[]"',
1031                 }
1032             )),
1033             {
1034                 'object': {},
1035                 'objectStr': '{}',
1036                 'array': [],
1037                 'arrayStr': '[]',
1038             }
1039         )
1040
1041     def test_js_to_json_realworld(self):
1042         inp = '''{
1043             'clip':{'provider':'pseudo'}
1044         }'''
1045         self.assertEqual(js_to_json(inp), '''{
1046             "clip":{"provider":"pseudo"}
1047         }''')
1048         json.loads(js_to_json(inp))
1049
1050         inp = '''{
1051             'playlist':[{'controls':{'all':null}}]
1052         }'''
1053         self.assertEqual(js_to_json(inp), '''{
1054             "playlist":[{"controls":{"all":null}}]
1055         }''')
1056
1057         inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
1058         self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
1059
1060         inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
1061         json_code = js_to_json(inp)
1062         self.assertEqual(json.loads(json_code), json.loads(inp))
1063
1064         inp = '''{
1065             0:{src:'skipped', type: 'application/dash+xml'},
1066             1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
1067         }'''
1068         self.assertEqual(js_to_json(inp), '''{
1069             "0":{"src":"skipped", "type": "application/dash+xml"},
1070             "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
1071         }''')
1072
1073         inp = '''{"foo":101}'''
1074         self.assertEqual(js_to_json(inp), '''{"foo":101}''')
1075
1076         inp = '''{"duration": "00:01:07"}'''
1077         self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
1078
1079         inp = '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}'''
1080         self.assertEqual(js_to_json(inp), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''')
1081
1082     def test_js_to_json_edgecases(self):
1083         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
1084         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
1085
1086         on = js_to_json('{"abc": true}')
1087         self.assertEqual(json.loads(on), {'abc': True})
1088
1089         # Ignore JavaScript code as well
1090         on = js_to_json('''{
1091             "x": 1,
1092             y: "a",
1093             z: some.code
1094         }''')
1095         d = json.loads(on)
1096         self.assertEqual(d['x'], 1)
1097         self.assertEqual(d['y'], 'a')
1098
1099         # Just drop ! prefix for now though this results in a wrong value
1100         on = js_to_json('''{
1101             a: !0,
1102             b: !1,
1103             c: !!0,
1104             d: !!42.42,
1105             e: !!![],
1106             f: !"abc",
1107             g: !"",
1108             !42: 42
1109         }''')
1110         self.assertEqual(json.loads(on), {
1111             'a': 0,
1112             'b': 1,
1113             'c': 0,
1114             'd': 42.42,
1115             'e': [],
1116             'f': "abc",
1117             'g': "",
1118             '42': 42
1119         })
1120
1121         on = js_to_json('["abc", "def",]')
1122         self.assertEqual(json.loads(on), ['abc', 'def'])
1123
1124         on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
1125         self.assertEqual(json.loads(on), ['abc', 'def'])
1126
1127         on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
1128         self.assertEqual(json.loads(on), ['abc', 'def'])
1129
1130         on = js_to_json('{"abc": "def",}')
1131         self.assertEqual(json.loads(on), {'abc': 'def'})
1132
1133         on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
1134         self.assertEqual(json.loads(on), {'abc': 'def'})
1135
1136         on = js_to_json('{ 0: /* " \n */ ",]" , }')
1137         self.assertEqual(json.loads(on), {'0': ',]'})
1138
1139         on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
1140         self.assertEqual(json.loads(on), {'0': ',]'})
1141
1142         on = js_to_json('{ 0: // comment\n1 }')
1143         self.assertEqual(json.loads(on), {'0': 1})
1144
1145         on = js_to_json(r'["<p>x<\/p>"]')
1146         self.assertEqual(json.loads(on), ['<p>x</p>'])
1147
1148         on = js_to_json(r'["\xaa"]')
1149         self.assertEqual(json.loads(on), ['\u00aa'])
1150
1151         on = js_to_json("['a\\\nb']")
1152         self.assertEqual(json.loads(on), ['ab'])
1153
1154         on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
1155         self.assertEqual(json.loads(on), ['ab'])
1156
1157         on = js_to_json('{0xff:0xff}')
1158         self.assertEqual(json.loads(on), {'255': 255})
1159
1160         on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
1161         self.assertEqual(json.loads(on), {'255': 255})
1162
1163         on = js_to_json('{077:077}')
1164         self.assertEqual(json.loads(on), {'63': 63})
1165
1166         on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
1167         self.assertEqual(json.loads(on), {'63': 63})
1168
1169         on = js_to_json('{42:42}')
1170         self.assertEqual(json.loads(on), {'42': 42})
1171
1172         on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
1173         self.assertEqual(json.loads(on), {'42': 42})
1174
1175         on = js_to_json('{42:4.2e1}')
1176         self.assertEqual(json.loads(on), {'42': 42.0})
1177
1178         on = js_to_json('{ "0x40": "0x40" }')
1179         self.assertEqual(json.loads(on), {'0x40': '0x40'})
1180
1181         on = js_to_json('{ "040": "040" }')
1182         self.assertEqual(json.loads(on), {'040': '040'})
1183
1184         on = js_to_json('[1,//{},\n2]')
1185         self.assertEqual(json.loads(on), [1, 2])
1186
1187         on = js_to_json(R'"\^\$\#"')
1188         self.assertEqual(json.loads(on), R'^$#', msg='Unnecessary escapes should be stripped')
1189
1190         on = js_to_json('\'"\\""\'')
1191         self.assertEqual(json.loads(on), '"""', msg='Unnecessary quote escape should be escaped')
1192
1193         on = js_to_json('[new Date("spam"), \'("eggs")\']')
1194         self.assertEqual(json.loads(on), ['spam', '("eggs")'], msg='Date regex should match a single string')
1195
1196     def test_js_to_json_malformed(self):
1197         self.assertEqual(js_to_json('42a1'), '42"a1"')
1198         self.assertEqual(js_to_json('42a-1'), '42"a"-1')
1199
1200     def test_js_to_json_template_literal(self):
1201         self.assertEqual(js_to_json('`Hello ${name}`', {'name': '"world"'}), '"Hello world"')
1202         self.assertEqual(js_to_json('`${name}${name}`', {'name': '"X"'}), '"XX"')
1203         self.assertEqual(js_to_json('`${name}${name}`', {'name': '5'}), '"55"')
1204         self.assertEqual(js_to_json('`${name}"${name}"`', {'name': '5'}), '"5\\"5\\""')
1205         self.assertEqual(js_to_json('`${name}`', {}), '"name"')
1206
1207     def test_js_to_json_common_constructors(self):
1208         self.assertEqual(json.loads(js_to_json('new Map([["a", 5]])')), {'a': 5})
1209         self.assertEqual(json.loads(js_to_json('Array(5, 10)')), [5, 10])
1210         self.assertEqual(json.loads(js_to_json('new Array(15,5)')), [15, 5])
1211         self.assertEqual(json.loads(js_to_json('new Map([Array(5, 10),new Array(15,5)])')), {'5': 10, '15': 5})
1212         self.assertEqual(json.loads(js_to_json('new Date("123")')), "123")
1213         self.assertEqual(json.loads(js_to_json('new Date(\'2023-10-19\')')), "2023-10-19")
1214
1215     def test_extract_attributes(self):
1216         self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
1217         self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
1218         self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
1219         self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
1220         self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
1221         self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
1222         self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
1223         self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'})  # XML
1224         self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
1225         self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'})  # HTML 3.2
1226         self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'})  # HTML 4.0
1227         self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
1228         self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
1229         self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
1230         self.assertEqual(extract_attributes('<e x >'), {'x': None})
1231         self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
1232         self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
1233         self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
1234         self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
1235         self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
1236         self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
1237         self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
1238         self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'})  # Names lowercased
1239         self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
1240         self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
1241         self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
1242         self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
1243         self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
1244         # "Narrow" Python builds don't support unicode code points outside BMP.
1245         try:
1246             chr(0x10000)
1247             supports_outside_bmp = True
1248         except ValueError:
1249             supports_outside_bmp = False
1250         if supports_outside_bmp:
1251             self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
1252         # Malformed HTML should not break attributes extraction on older Python
1253         self.assertEqual(extract_attributes('<mal"formed/>'), {})
1254
1255     def test_clean_html(self):
1256         self.assertEqual(clean_html('a:\nb'), 'a: b')
1257         self.assertEqual(clean_html('a:\n   "b"'), 'a: "b"')
1258         self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
1259
1260     def test_intlist_to_bytes(self):
1261         self.assertEqual(
1262             intlist_to_bytes([0, 1, 127, 128, 255]),
1263             b'\x00\x01\x7f\x80\xff')
1264
1265     def test_args_to_str(self):
1266         self.assertEqual(
1267             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
1268             'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""'
1269         )
1270
1271     def test_parse_filesize(self):
1272         self.assertEqual(parse_filesize(None), None)
1273         self.assertEqual(parse_filesize(''), None)
1274         self.assertEqual(parse_filesize('91 B'), 91)
1275         self.assertEqual(parse_filesize('foobar'), None)
1276         self.assertEqual(parse_filesize('2 MiB'), 2097152)
1277         self.assertEqual(parse_filesize('5 GB'), 5000000000)
1278         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
1279         self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
1280         self.assertEqual(parse_filesize('1,24 KB'), 1240)
1281         self.assertEqual(parse_filesize('1,24 kb'), 1240)
1282         self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
1283
1284     def test_parse_count(self):
1285         self.assertEqual(parse_count(None), None)
1286         self.assertEqual(parse_count(''), None)
1287         self.assertEqual(parse_count('0'), 0)
1288         self.assertEqual(parse_count('1000'), 1000)
1289         self.assertEqual(parse_count('1.000'), 1000)
1290         self.assertEqual(parse_count('1.1k'), 1100)
1291         self.assertEqual(parse_count('1.1 k'), 1100)
1292         self.assertEqual(parse_count('1,1 k'), 1100)
1293         self.assertEqual(parse_count('1.1kk'), 1100000)
1294         self.assertEqual(parse_count('1.1kk '), 1100000)
1295         self.assertEqual(parse_count('1,1kk'), 1100000)
1296         self.assertEqual(parse_count('100 views'), 100)
1297         self.assertEqual(parse_count('1,100 views'), 1100)
1298         self.assertEqual(parse_count('1.1kk views'), 1100000)
1299         self.assertEqual(parse_count('10M views'), 10000000)
1300         self.assertEqual(parse_count('has 10M views'), 10000000)
1301
1302     def test_parse_resolution(self):
1303         self.assertEqual(parse_resolution(None), {})
1304         self.assertEqual(parse_resolution(''), {})
1305         self.assertEqual(parse_resolution(' 1920x1080'), {'width': 1920, 'height': 1080})
1306         self.assertEqual(parse_resolution('1920×1080 '), {'width': 1920, 'height': 1080})
1307         self.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080})
1308         self.assertEqual(parse_resolution('720p'), {'height': 720})
1309         self.assertEqual(parse_resolution('4k'), {'height': 2160})
1310         self.assertEqual(parse_resolution('8K'), {'height': 4320})
1311         self.assertEqual(parse_resolution('pre_1920x1080_post'), {'width': 1920, 'height': 1080})
1312         self.assertEqual(parse_resolution('ep1x2'), {})
1313         self.assertEqual(parse_resolution('1920, 1080'), {'width': 1920, 'height': 1080})
1314
1315     def test_parse_bitrate(self):
1316         self.assertEqual(parse_bitrate(None), None)
1317         self.assertEqual(parse_bitrate(''), None)
1318         self.assertEqual(parse_bitrate('300kbps'), 300)
1319         self.assertEqual(parse_bitrate('1500kbps'), 1500)
1320         self.assertEqual(parse_bitrate('300 kbps'), 300)
1321
1322     def test_version_tuple(self):
1323         self.assertEqual(version_tuple('1'), (1,))
1324         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
1325         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
1326
1327     def test_detect_exe_version(self):
1328         self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
1329 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
1330 configuration: --prefix=/usr --extra-'''), '1.2.1')
1331         self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
1332 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
1333         self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
1334 Trying to open render node...
1335 Success at /dev/dri/renderD128.
1336 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
1337
1338     def test_age_restricted(self):
1339         self.assertFalse(age_restricted(None, 10))  # unrestricted content
1340         self.assertFalse(age_restricted(1, None))  # unrestricted policy
1341         self.assertFalse(age_restricted(8, 10))
1342         self.assertTrue(age_restricted(18, 14))
1343         self.assertFalse(age_restricted(18, 18))
1344
1345     def test_is_html(self):
1346         self.assertFalse(is_html(b'\x49\x44\x43<html'))
1347         self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
1348         self.assertTrue(is_html(  # UTF-8 with BOM
1349             b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
1350         self.assertTrue(is_html(  # UTF-16-LE
1351             b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
1352         ))
1353         self.assertTrue(is_html(  # UTF-16-BE
1354             b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
1355         ))
1356         self.assertTrue(is_html(  # UTF-32-BE
1357             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'))
1358         self.assertTrue(is_html(  # UTF-32-LE
1359             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'))
1360
1361     def test_render_table(self):
1362         self.assertEqual(
1363             render_table(
1364                 ['a', 'empty', 'bcd'],
1365                 [[123, '', 4], [9999, '', 51]]),
1366             'a    empty bcd\n'
1367             '123        4\n'
1368             '9999       51')
1369
1370         self.assertEqual(
1371             render_table(
1372                 ['a', 'empty', 'bcd'],
1373                 [[123, '', 4], [9999, '', 51]],
1374                 hide_empty=True),
1375             'a    bcd\n'
1376             '123  4\n'
1377             '9999 51')
1378
1379         self.assertEqual(
1380             render_table(
1381                 ['\ta', 'bcd'],
1382                 [['1\t23', 4], ['\t9999', 51]]),
1383             '   a bcd\n'
1384             '1 23 4\n'
1385             '9999 51')
1386
1387         self.assertEqual(
1388             render_table(
1389                 ['a', 'bcd'],
1390                 [[123, 4], [9999, 51]],
1391                 delim='-'),
1392             'a    bcd\n'
1393             '--------\n'
1394             '123  4\n'
1395             '9999 51')
1396
1397         self.assertEqual(
1398             render_table(
1399                 ['a', 'bcd'],
1400                 [[123, 4], [9999, 51]],
1401                 delim='-', extra_gap=2),
1402             'a      bcd\n'
1403             '----------\n'
1404             '123    4\n'
1405             '9999   51')
1406
1407     def test_match_str(self):
1408         # Unary
1409         self.assertFalse(match_str('xy', {'x': 1200}))
1410         self.assertTrue(match_str('!xy', {'x': 1200}))
1411         self.assertTrue(match_str('x', {'x': 1200}))
1412         self.assertFalse(match_str('!x', {'x': 1200}))
1413         self.assertTrue(match_str('x', {'x': 0}))
1414         self.assertTrue(match_str('is_live', {'is_live': True}))
1415         self.assertFalse(match_str('is_live', {'is_live': False}))
1416         self.assertFalse(match_str('is_live', {'is_live': None}))
1417         self.assertFalse(match_str('is_live', {}))
1418         self.assertFalse(match_str('!is_live', {'is_live': True}))
1419         self.assertTrue(match_str('!is_live', {'is_live': False}))
1420         self.assertTrue(match_str('!is_live', {'is_live': None}))
1421         self.assertTrue(match_str('!is_live', {}))
1422         self.assertTrue(match_str('title', {'title': 'abc'}))
1423         self.assertTrue(match_str('title', {'title': ''}))
1424         self.assertFalse(match_str('!title', {'title': 'abc'}))
1425         self.assertFalse(match_str('!title', {'title': ''}))
1426
1427         # Numeric
1428         self.assertFalse(match_str('x>0', {'x': 0}))
1429         self.assertFalse(match_str('x>0', {}))
1430         self.assertTrue(match_str('x>?0', {}))
1431         self.assertTrue(match_str('x>1K', {'x': 1200}))
1432         self.assertFalse(match_str('x>2K', {'x': 1200}))
1433         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
1434         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
1435         self.assertTrue(match_str('x > 1:0:0', {'x': 3700}))
1436
1437         # String
1438         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
1439         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
1440         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
1441         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
1442         self.assertTrue(match_str('y^=foo', {'y': 'foobar42'}))
1443         self.assertFalse(match_str('y!^=foo', {'y': 'foobar42'}))
1444         self.assertFalse(match_str('y^=bar', {'y': 'foobar42'}))
1445         self.assertTrue(match_str('y!^=bar', {'y': 'foobar42'}))
1446         self.assertRaises(ValueError, match_str, 'x^=42', {'x': 42})
1447         self.assertTrue(match_str('y*=bar', {'y': 'foobar42'}))
1448         self.assertFalse(match_str('y!*=bar', {'y': 'foobar42'}))
1449         self.assertFalse(match_str('y*=baz', {'y': 'foobar42'}))
1450         self.assertTrue(match_str('y!*=baz', {'y': 'foobar42'}))
1451         self.assertTrue(match_str('y$=42', {'y': 'foobar42'}))
1452         self.assertFalse(match_str('y$=43', {'y': 'foobar42'}))
1453
1454         # And
1455         self.assertFalse(match_str(
1456             'like_count > 100 & dislike_count <? 50 & description',
1457             {'like_count': 90, 'description': 'foo'}))
1458         self.assertTrue(match_str(
1459             'like_count > 100 & dislike_count <? 50 & description',
1460             {'like_count': 190, 'description': 'foo'}))
1461         self.assertFalse(match_str(
1462             'like_count > 100 & dislike_count <? 50 & description',
1463             {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
1464         self.assertFalse(match_str(
1465             'like_count > 100 & dislike_count <? 50 & description',
1466             {'like_count': 190, 'dislike_count': 10}))
1467
1468         # Regex
1469         self.assertTrue(match_str(r'x~=\bbar', {'x': 'foo bar'}))
1470         self.assertFalse(match_str(r'x~=\bbar.+', {'x': 'foo bar'}))
1471         self.assertFalse(match_str(r'x~=^FOO', {'x': 'foo bar'}))
1472         self.assertTrue(match_str(r'x~=(?i)^FOO', {'x': 'foo bar'}))
1473
1474         # Quotes
1475         self.assertTrue(match_str(r'x^="foo"', {'x': 'foo "bar"'}))
1476         self.assertFalse(match_str(r'x^="foo  "', {'x': 'foo "bar"'}))
1477         self.assertFalse(match_str(r'x$="bar"', {'x': 'foo "bar"'}))
1478         self.assertTrue(match_str(r'x$=" \"bar\""', {'x': 'foo "bar"'}))
1479
1480         # Escaping &
1481         self.assertFalse(match_str(r'x=foo & bar', {'x': 'foo & bar'}))
1482         self.assertTrue(match_str(r'x=foo \& bar', {'x': 'foo & bar'}))
1483         self.assertTrue(match_str(r'x=foo \& bar & x^=foo', {'x': 'foo & bar'}))
1484         self.assertTrue(match_str(r'x="foo \& bar" & x^=foo', {'x': 'foo & bar'}))
1485
1486         # Example from docs
1487         self.assertTrue(match_str(
1488             r"!is_live & like_count>?100 & description~='(?i)\bcats \& dogs\b'",
1489             {'description': 'Raining Cats & Dogs'}))
1490
1491         # Incomplete
1492         self.assertFalse(match_str('id!=foo', {'id': 'foo'}, True))
1493         self.assertTrue(match_str('x', {'id': 'foo'}, True))
1494         self.assertTrue(match_str('!x', {'id': 'foo'}, True))
1495         self.assertFalse(match_str('x', {'id': 'foo'}, False))
1496
1497     def test_parse_dfxp_time_expr(self):
1498         self.assertEqual(parse_dfxp_time_expr(None), None)
1499         self.assertEqual(parse_dfxp_time_expr(''), None)
1500         self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1501         self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1502         self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1503         self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1504         self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1505
1506     def test_dfxp2srt(self):
1507         dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
1508             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1509             <body>
1510                 <div xml:lang="en">
1511                     <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1512                     <p begin="1" end="2">第二行<br/>♪♪</p>
1513                     <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1514                     <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1515                     <p begin="-1" end="-1">Ignore, two</p>
1516                     <p begin="3" dur="-1">Ignored, three</p>
1517                 </div>
1518             </body>
1519             </tt>'''.encode()
1520         srt_data = '''1
1521 00:00:00,000 --> 00:00:01,000
1522 The following line contains Chinese characters and special symbols
1523
1524 2
1525 00:00:01,000 --> 00:00:02,000
1526 第二行
1527 ♪♪
1528
1529 3
1530 00:00:02,000 --> 00:00:03,000
1531 Third
1532 Line
1533
1534 '''
1535         self.assertEqual(dfxp2srt(dfxp_data), srt_data)
1536
1537         dfxp_data_no_default_namespace = b'''<?xml version="1.0" encoding="UTF-8"?>
1538             <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1539             <body>
1540                 <div xml:lang="en">
1541                     <p begin="0" end="1">The first line</p>
1542                 </div>
1543             </body>
1544             </tt>'''
1545         srt_data = '''1
1546 00:00:00,000 --> 00:00:01,000
1547 The first line
1548
1549 '''
1550         self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
1551
1552         dfxp_data_with_style = b'''<?xml version="1.0" encoding="utf-8"?>
1553 <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">
1554   <head>
1555     <styling>
1556       <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1557       <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1558       <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1559       <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1560     </styling>
1561   </head>
1562   <body tts:textAlign="center" style="s0">
1563     <div>
1564       <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1565       <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>
1566       <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1567       <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>
1568     </div>
1569   </body>
1570 </tt>'''
1571         srt_data = '''1
1572 00:00:02,080 --> 00:00:05,840
1573 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1574
1575 2
1576 00:00:02,080 --> 00:00:05,840
1577 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1578 </font>part 2</font></b>
1579
1580 3
1581 00:00:05,840 --> 00:00:09,560
1582 <u><font color="lime">line 3
1583 part 3</font></u>
1584
1585 4
1586 00:00:09,560 --> 00:00:12,360
1587 <i><u><font color="yellow"><font color="lime">inner
1588  </font>style</font></u></i>
1589
1590 '''
1591         self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)
1592
1593         dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?>
1594             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1595             <body>
1596                 <div xml:lang="en">
1597                     <p begin="0" end="1">Line 1</p>
1598                     <p begin="1" end="2">第二行</p>
1599                 </div>
1600             </body>
1601             </tt>'''.encode('utf-16')
1602         srt_data = '''1
1603 00:00:00,000 --> 00:00:01,000
1604 Line 1
1605
1606 2
1607 00:00:01,000 --> 00:00:02,000
1608 第二行
1609
1610 '''
1611         self.assertEqual(dfxp2srt(dfxp_data_non_utf8), srt_data)
1612
1613     def test_cli_option(self):
1614         self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1615         self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
1616         self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1617         self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
1618
1619     def test_cli_valueless_option(self):
1620         self.assertEqual(cli_valueless_option(
1621             {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1622         self.assertEqual(cli_valueless_option(
1623             {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
1624         self.assertEqual(cli_valueless_option(
1625             {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1626         self.assertEqual(cli_valueless_option(
1627             {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
1628         self.assertEqual(cli_valueless_option(
1629             {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
1630         self.assertEqual(cli_valueless_option(
1631             {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1632
1633     def test_cli_bool_option(self):
1634         self.assertEqual(
1635             cli_bool_option(
1636                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1637             ['--no-check-certificate', 'true'])
1638         self.assertEqual(
1639             cli_bool_option(
1640                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1641             ['--no-check-certificate=true'])
1642         self.assertEqual(
1643             cli_bool_option(
1644                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1645             ['--check-certificate', 'false'])
1646         self.assertEqual(
1647             cli_bool_option(
1648                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1649             ['--check-certificate=false'])
1650         self.assertEqual(
1651             cli_bool_option(
1652                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1653             ['--check-certificate', 'true'])
1654         self.assertEqual(
1655             cli_bool_option(
1656                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1657             ['--check-certificate=true'])
1658         self.assertEqual(
1659             cli_bool_option(
1660                 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1661             [])
1662
1663     def test_ohdave_rsa_encrypt(self):
1664         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1665         e = 65537
1666
1667         self.assertEqual(
1668             ohdave_rsa_encrypt(b'aa111222', e, N),
1669             '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1670
1671     def test_pkcs1pad(self):
1672         data = [1, 2, 3]
1673         padded_data = pkcs1pad(data, 32)
1674         self.assertEqual(padded_data[:2], [0, 2])
1675         self.assertEqual(padded_data[28:], [0, 1, 2, 3])
1676
1677         self.assertRaises(ValueError, pkcs1pad, data, 8)
1678
1679     def test_encode_base_n(self):
1680         self.assertEqual(encode_base_n(0, 30), '0')
1681         self.assertEqual(encode_base_n(80, 30), '2k')
1682
1683         custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1684         self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1685         self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1686
1687         self.assertRaises(ValueError, encode_base_n, 0, 70)
1688         self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1689
1690     def test_caesar(self):
1691         self.assertEqual(caesar('ace', 'abcdef', 2), 'cea')
1692         self.assertEqual(caesar('cea', 'abcdef', -2), 'ace')
1693         self.assertEqual(caesar('ace', 'abcdef', -2), 'eac')
1694         self.assertEqual(caesar('eac', 'abcdef', 2), 'ace')
1695         self.assertEqual(caesar('ace', 'abcdef', 0), 'ace')
1696         self.assertEqual(caesar('xyz', 'abcdef', 2), 'xyz')
1697         self.assertEqual(caesar('abc', 'acegik', 2), 'ebg')
1698         self.assertEqual(caesar('ebg', 'acegik', -2), 'abc')
1699
1700     def test_rot47(self):
1701         self.assertEqual(rot47('yt-dlp'), r'JE\5=A')
1702         self.assertEqual(rot47('YT-DLP'), r'*%\s{!')
1703
1704     def test_urshift(self):
1705         self.assertEqual(urshift(3, 1), 1)
1706         self.assertEqual(urshift(-3, 1), 2147483646)
1707
1708     GET_ELEMENT_BY_CLASS_TEST_STRING = '''
1709         <span class="foo bar">nice</span>
1710     '''
1711
1712     def test_get_element_by_class(self):
1713         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1714
1715         self.assertEqual(get_element_by_class('foo', html), 'nice')
1716         self.assertEqual(get_element_by_class('no-such-class', html), None)
1717
1718     def test_get_element_html_by_class(self):
1719         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1720
1721         self.assertEqual(get_element_html_by_class('foo', html), html.strip())
1722         self.assertEqual(get_element_by_class('no-such-class', html), None)
1723
1724     GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING = '''
1725         <div itemprop="author" itemscope>foo</div>
1726     '''
1727
1728     def test_get_element_by_attribute(self):
1729         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1730
1731         self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
1732         self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
1733         self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
1734
1735         html = self.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
1736
1737         self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo')
1738
1739     def test_get_element_html_by_attribute(self):
1740         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1741
1742         self.assertEqual(get_element_html_by_attribute('class', 'foo bar', html), html.strip())
1743         self.assertEqual(get_element_html_by_attribute('class', 'foo', html), None)
1744         self.assertEqual(get_element_html_by_attribute('class', 'no-such-foo', html), None)
1745
1746         html = self.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
1747
1748         self.assertEqual(get_element_html_by_attribute('itemprop', 'author', html), html.strip())
1749
1750     GET_ELEMENTS_BY_CLASS_TEST_STRING = '''
1751         <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1752     '''
1753     GET_ELEMENTS_BY_CLASS_RES = ['<span class="foo bar">nice</span>', '<span class="foo bar">also nice</span>']
1754
1755     def test_get_elements_by_class(self):
1756         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1757
1758         self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
1759         self.assertEqual(get_elements_by_class('no-such-class', html), [])
1760
1761     def test_get_elements_html_by_class(self):
1762         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1763
1764         self.assertEqual(get_elements_html_by_class('foo', html), self.GET_ELEMENTS_BY_CLASS_RES)
1765         self.assertEqual(get_elements_html_by_class('no-such-class', html), [])
1766
1767     def test_get_elements_by_attribute(self):
1768         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1769
1770         self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
1771         self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
1772         self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
1773
1774     def test_get_elements_html_by_attribute(self):
1775         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1776
1777         self.assertEqual(get_elements_html_by_attribute('class', 'foo bar', html), self.GET_ELEMENTS_BY_CLASS_RES)
1778         self.assertEqual(get_elements_html_by_attribute('class', 'foo', html), [])
1779         self.assertEqual(get_elements_html_by_attribute('class', 'no-such-foo', html), [])
1780
1781     def test_get_elements_text_and_html_by_attribute(self):
1782         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1783
1784         self.assertEqual(
1785             list(get_elements_text_and_html_by_attribute('class', 'foo bar', html)),
1786             list(zip(['nice', 'also nice'], self.GET_ELEMENTS_BY_CLASS_RES)))
1787         self.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'foo', html)), [])
1788         self.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'no-such-foo', html)), [])
1789
1790         self.assertEqual(list(get_elements_text_and_html_by_attribute(
1791             'class', 'foo', '<a class="foo">nice</a><span class="foo">nice</span>', tag='a')), [('nice', '<a class="foo">nice</a>')])
1792
1793     GET_ELEMENT_BY_TAG_TEST_STRING = '''
1794     random text lorem ipsum</p>
1795     <div>
1796         this should be returned
1797         <span>this should also be returned</span>
1798         <div>
1799             this should also be returned
1800         </div>
1801         closing tag above should not trick, so this should also be returned
1802     </div>
1803     but this text should not be returned
1804     '''
1805     GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML = GET_ELEMENT_BY_TAG_TEST_STRING.strip()[32:276]
1806     GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT = GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML[5:-6]
1807     GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML = GET_ELEMENT_BY_TAG_TEST_STRING.strip()[78:119]
1808     GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT = GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML[6:-7]
1809
1810     def test_get_element_text_and_html_by_tag(self):
1811         html = self.GET_ELEMENT_BY_TAG_TEST_STRING
1812
1813         self.assertEqual(
1814             get_element_text_and_html_by_tag('div', html),
1815             (self.GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT, self.GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML))
1816         self.assertEqual(
1817             get_element_text_and_html_by_tag('span', html),
1818             (self.GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT, self.GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML))
1819         self.assertRaises(compat_HTMLParseError, get_element_text_and_html_by_tag, 'article', html)
1820
1821     def test_iri_to_uri(self):
1822         self.assertEqual(
1823             iri_to_uri('https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b'),
1824             'https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b')  # Same
1825         self.assertEqual(
1826             iri_to_uri('https://www.google.com/search?q=Käsesoßenrührlöffel'),  # German for cheese sauce stirring spoon
1827             'https://www.google.com/search?q=K%C3%A4seso%C3%9Fenr%C3%BChrl%C3%B6ffel')
1828         self.assertEqual(
1829             iri_to_uri('https://www.google.com/search?q=lt<+gt>+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#'),
1830             'https://www.google.com/search?q=lt%3C+gt%3E+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#')
1831         self.assertEqual(
1832             iri_to_uri('http://правозащита38.рф/category/news/'),
1833             'http://xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/')
1834         self.assertEqual(
1835             iri_to_uri('http://www.правозащита38.рф/category/news/'),
1836             'http://www.xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/')
1837         self.assertEqual(
1838             iri_to_uri('https://i❤.ws/emojidomain/👍👏🤝💪'),
1839             'https://xn--i-7iq.ws/emojidomain/%F0%9F%91%8D%F0%9F%91%8F%F0%9F%A4%9D%F0%9F%92%AA')
1840         self.assertEqual(
1841             iri_to_uri('http://日本語.jp/'),
1842             'http://xn--wgv71a119e.jp/')
1843         self.assertEqual(
1844             iri_to_uri('http://导航.中国/'),
1845             'http://xn--fet810g.xn--fiqs8s/')
1846
1847     def test_clean_podcast_url(self):
1848         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')
1849         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')
1850         self.assertEqual(clean_podcast_url('https://pdst.fm/e/2.gum.fm/chtbl.com/track/chrt.fm/track/34D33/pscrb.fm/rss/p/traffic.megaphone.fm/ITLLC7765286967.mp3?updated=1687282661'), 'https://traffic.megaphone.fm/ITLLC7765286967.mp3?updated=1687282661')
1851         self.assertEqual(clean_podcast_url('https://pdst.fm/e/https://mgln.ai/e/441/www.buzzsprout.com/1121972/13019085-ep-252-the-deep-life-stack.mp3'), 'https://www.buzzsprout.com/1121972/13019085-ep-252-the-deep-life-stack.mp3')
1852
1853     def test_LazyList(self):
1854         it = list(range(10))
1855
1856         self.assertEqual(list(LazyList(it)), it)
1857         self.assertEqual(LazyList(it).exhaust(), it)
1858         self.assertEqual(LazyList(it)[5], it[5])
1859
1860         self.assertEqual(LazyList(it)[5:], it[5:])
1861         self.assertEqual(LazyList(it)[:5], it[:5])
1862         self.assertEqual(LazyList(it)[::2], it[::2])
1863         self.assertEqual(LazyList(it)[1::2], it[1::2])
1864         self.assertEqual(LazyList(it)[5::-1], it[5::-1])
1865         self.assertEqual(LazyList(it)[6:2:-2], it[6:2:-2])
1866         self.assertEqual(LazyList(it)[::-1], it[::-1])
1867
1868         self.assertTrue(LazyList(it))
1869         self.assertFalse(LazyList(range(0)))
1870         self.assertEqual(len(LazyList(it)), len(it))
1871         self.assertEqual(repr(LazyList(it)), repr(it))
1872         self.assertEqual(str(LazyList(it)), str(it))
1873
1874         self.assertEqual(list(LazyList(it, reverse=True)), it[::-1])
1875         self.assertEqual(list(reversed(LazyList(it))[::-1]), it)
1876         self.assertEqual(list(reversed(LazyList(it))[1:3:7]), it[::-1][1:3:7])
1877
1878     def test_LazyList_laziness(self):
1879
1880         def test(ll, idx, val, cache):
1881             self.assertEqual(ll[idx], val)
1882             self.assertEqual(ll._cache, list(cache))
1883
1884         ll = LazyList(range(10))
1885         test(ll, 0, 0, range(1))
1886         test(ll, 5, 5, range(6))
1887         test(ll, -3, 7, range(10))
1888
1889         ll = LazyList(range(10), reverse=True)
1890         test(ll, -1, 0, range(1))
1891         test(ll, 3, 6, range(10))
1892
1893         ll = LazyList(itertools.count())
1894         test(ll, 10, 10, range(11))
1895         ll = reversed(ll)
1896         test(ll, -15, 14, range(15))
1897
1898     def test_format_bytes(self):
1899         self.assertEqual(format_bytes(0), '0.00B')
1900         self.assertEqual(format_bytes(1000), '1000.00B')
1901         self.assertEqual(format_bytes(1024), '1.00KiB')
1902         self.assertEqual(format_bytes(1024**2), '1.00MiB')
1903         self.assertEqual(format_bytes(1024**3), '1.00GiB')
1904         self.assertEqual(format_bytes(1024**4), '1.00TiB')
1905         self.assertEqual(format_bytes(1024**5), '1.00PiB')
1906         self.assertEqual(format_bytes(1024**6), '1.00EiB')
1907         self.assertEqual(format_bytes(1024**7), '1.00ZiB')
1908         self.assertEqual(format_bytes(1024**8), '1.00YiB')
1909         self.assertEqual(format_bytes(1024**9), '1024.00YiB')
1910
1911     def test_hide_login_info(self):
1912         self.assertEqual(Config.hide_login_info(['-u', 'foo', '-p', 'bar']),
1913                          ['-u', 'PRIVATE', '-p', 'PRIVATE'])
1914         self.assertEqual(Config.hide_login_info(['-u']), ['-u'])
1915         self.assertEqual(Config.hide_login_info(['-u', 'foo', '-u', 'bar']),
1916                          ['-u', 'PRIVATE', '-u', 'PRIVATE'])
1917         self.assertEqual(Config.hide_login_info(['--username=foo']),
1918                          ['--username=PRIVATE'])
1919
1920     def test_locked_file(self):
1921         TEXT = 'test_locked_file\n'
1922         FILE = 'test_locked_file.ytdl'
1923         MODES = 'war'  # Order is important
1924
1925         try:
1926             for lock_mode in MODES:
1927                 with locked_file(FILE, lock_mode, False) as f:
1928                     if lock_mode == 'r':
1929                         self.assertEqual(f.read(), TEXT * 2, 'Wrong file content')
1930                     else:
1931                         f.write(TEXT)
1932                     for test_mode in MODES:
1933                         testing_write = test_mode != 'r'
1934                         try:
1935                             with locked_file(FILE, test_mode, False):
1936                                 pass
1937                         except (BlockingIOError, PermissionError):
1938                             if not testing_write:  # FIXME
1939                                 print(f'Known issue: Exclusive lock ({lock_mode}) blocks read access ({test_mode})')
1940                                 continue
1941                             self.assertTrue(testing_write, f'{test_mode} is blocked by {lock_mode}')
1942                         else:
1943                             self.assertFalse(testing_write, f'{test_mode} is not blocked by {lock_mode}')
1944         finally:
1945             with contextlib.suppress(OSError):
1946                 os.remove(FILE)
1947
1948     def test_determine_file_encoding(self):
1949         self.assertEqual(determine_file_encoding(b''), (None, 0))
1950         self.assertEqual(determine_file_encoding(b'--verbose -x --audio-format mkv\n'), (None, 0))
1951
1952         self.assertEqual(determine_file_encoding(b'\xef\xbb\xbf'), ('utf-8', 3))
1953         self.assertEqual(determine_file_encoding(b'\x00\x00\xfe\xff'), ('utf-32-be', 4))
1954         self.assertEqual(determine_file_encoding(b'\xff\xfe'), ('utf-16-le', 2))
1955
1956         self.assertEqual(determine_file_encoding(b'\xff\xfe# coding: utf-8\n--verbose'), ('utf-16-le', 2))
1957
1958         self.assertEqual(determine_file_encoding(b'# coding: utf-8\n--verbose'), ('utf-8', 0))
1959         self.assertEqual(determine_file_encoding(b'# coding: someencodinghere-12345\n--verbose'), ('someencodinghere-12345', 0))
1960
1961         self.assertEqual(determine_file_encoding(b'#coding:utf-8\n--verbose'), ('utf-8', 0))
1962         self.assertEqual(determine_file_encoding(b'#  coding:   utf-8   \r\n--verbose'), ('utf-8', 0))
1963
1964         self.assertEqual(determine_file_encoding('# coding: utf-32-be'.encode('utf-32-be')), ('utf-32-be', 0))
1965         self.assertEqual(determine_file_encoding('# coding: utf-16-le'.encode('utf-16-le')), ('utf-16-le', 0))
1966
1967     def test_get_compatible_ext(self):
1968         self.assertEqual(get_compatible_ext(
1969             vcodecs=[None], acodecs=[None, None], vexts=['mp4'], aexts=['m4a', 'm4a']), 'mkv')
1970         self.assertEqual(get_compatible_ext(
1971             vcodecs=[None], acodecs=[None], vexts=['flv'], aexts=['flv']), 'flv')
1972
1973         self.assertEqual(get_compatible_ext(
1974             vcodecs=[None], acodecs=[None], vexts=['mp4'], aexts=['m4a']), 'mp4')
1975         self.assertEqual(get_compatible_ext(
1976             vcodecs=[None], acodecs=[None], vexts=['mp4'], aexts=['webm']), 'mkv')
1977         self.assertEqual(get_compatible_ext(
1978             vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['m4a']), 'mkv')
1979         self.assertEqual(get_compatible_ext(
1980             vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['webm']), 'webm')
1981         self.assertEqual(get_compatible_ext(
1982             vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['weba']), 'webm')
1983
1984         self.assertEqual(get_compatible_ext(
1985             vcodecs=['h264'], acodecs=['mp4a'], vexts=['mov'], aexts=['m4a']), 'mp4')
1986         self.assertEqual(get_compatible_ext(
1987             vcodecs=['av01.0.12M.08'], acodecs=['opus'], vexts=['mp4'], aexts=['webm']), 'webm')
1988
1989         self.assertEqual(get_compatible_ext(
1990             vcodecs=['vp9'], acodecs=['opus'], vexts=['webm'], aexts=['webm'], preferences=['flv', 'mp4']), 'mp4')
1991         self.assertEqual(get_compatible_ext(
1992             vcodecs=['av1'], acodecs=['mp4a'], vexts=['webm'], aexts=['m4a'], preferences=('webm', 'mkv')), 'mkv')
1993
1994     def test_try_call(self):
1995         def total(*x, **kwargs):
1996             return sum(x) + sum(kwargs.values())
1997
1998         self.assertEqual(try_call(None), None,
1999                          msg='not a fn should give None')
2000         self.assertEqual(try_call(lambda: 1), 1,
2001                          msg='int fn with no expected_type should give int')
2002         self.assertEqual(try_call(lambda: 1, expected_type=int), 1,
2003                          msg='int fn with expected_type int should give int')
2004         self.assertEqual(try_call(lambda: 1, expected_type=dict), None,
2005                          msg='int fn with wrong expected_type should give None')
2006         self.assertEqual(try_call(total, args=(0, 1, 0, ), expected_type=int), 1,
2007                          msg='fn should accept arglist')
2008         self.assertEqual(try_call(total, kwargs={'a': 0, 'b': 1, 'c': 0}, expected_type=int), 1,
2009                          msg='fn should accept kwargs')
2010         self.assertEqual(try_call(lambda: 1, expected_type=dict), None,
2011                          msg='int fn with no expected_type should give None')
2012         self.assertEqual(try_call(lambda x: {}, total, args=(42, ), expected_type=int), 42,
2013                          msg='expect first int result with expected_type int')
2014
2015     def test_variadic(self):
2016         self.assertEqual(variadic(None), (None, ))
2017         self.assertEqual(variadic('spam'), ('spam', ))
2018         self.assertEqual(variadic('spam', allowed_types=dict), 'spam')
2019         with warnings.catch_warnings():
2020             warnings.simplefilter('ignore')
2021             self.assertEqual(variadic('spam', allowed_types=[dict]), 'spam')
2022
2023     def test_http_header_dict(self):
2024         headers = HTTPHeaderDict()
2025         headers['ytdl-test'] = b'0'
2026         self.assertEqual(list(headers.items()), [('Ytdl-Test', '0')])
2027         headers['ytdl-test'] = 1
2028         self.assertEqual(list(headers.items()), [('Ytdl-Test', '1')])
2029         headers['Ytdl-test'] = '2'
2030         self.assertEqual(list(headers.items()), [('Ytdl-Test', '2')])
2031         self.assertTrue('ytDl-Test' in headers)
2032         self.assertEqual(str(headers), str(dict(headers)))
2033         self.assertEqual(repr(headers), str(dict(headers)))
2034
2035         headers.update({'X-dlp': 'data'})
2036         self.assertEqual(set(headers.items()), {('Ytdl-Test', '2'), ('X-Dlp', 'data')})
2037         self.assertEqual(dict(headers), {'Ytdl-Test': '2', 'X-Dlp': 'data'})
2038         self.assertEqual(len(headers), 2)
2039         self.assertEqual(headers.copy(), headers)
2040         headers2 = HTTPHeaderDict({'X-dlp': 'data3'}, **headers, **{'X-dlp': 'data2'})
2041         self.assertEqual(set(headers2.items()), {('Ytdl-Test', '2'), ('X-Dlp', 'data2')})
2042         self.assertEqual(len(headers2), 2)
2043         headers2.clear()
2044         self.assertEqual(len(headers2), 0)
2045
2046         # ensure we prefer latter headers
2047         headers3 = HTTPHeaderDict({'Ytdl-TeSt': 1}, {'Ytdl-test': 2})
2048         self.assertEqual(set(headers3.items()), {('Ytdl-Test', '2')})
2049         del headers3['ytdl-tesT']
2050         self.assertEqual(dict(headers3), {})
2051
2052         headers4 = HTTPHeaderDict({'ytdl-test': 'data;'})
2053         self.assertEqual(set(headers4.items()), {('Ytdl-Test', 'data;')})
2054
2055         # common mistake: strip whitespace from values
2056         # https://github.com/yt-dlp/yt-dlp/issues/8729
2057         headers5 = HTTPHeaderDict({'ytdl-test': ' data; '})
2058         self.assertEqual(set(headers5.items()), {('Ytdl-Test', 'data;')})
2059
2060     def test_extract_basic_auth(self):
2061         assert extract_basic_auth('http://:foo.bar') == ('http://:foo.bar', None)
2062         assert extract_basic_auth('http://foo.bar') == ('http://foo.bar', None)
2063         assert extract_basic_auth('http://@foo.bar') == ('http://foo.bar', 'Basic Og==')
2064         assert extract_basic_auth('http://:pass@foo.bar') == ('http://foo.bar', 'Basic OnBhc3M=')
2065         assert extract_basic_auth('http://user:@foo.bar') == ('http://foo.bar', 'Basic dXNlcjo=')
2066         assert extract_basic_auth('http://user:pass@foo.bar') == ('http://foo.bar', 'Basic dXNlcjpwYXNz')
2067
2068     @unittest.skipUnless(compat_os_name == 'nt', 'Only relevant on Windows')
2069     def test_windows_escaping(self):
2070         tests = [
2071             'test"&',
2072             '%CMDCMDLINE:~-1%&',
2073             'a\nb',
2074             '"',
2075             '\\',
2076             '!',
2077             '^!',
2078             'a \\ b',
2079             'a \\" b',
2080             'a \\ b\\',
2081             # We replace \r with \n
2082             ('a\r\ra', 'a\n\na'),
2083         ]
2084
2085         def run_shell(args):
2086             stdout, stderr, error = Popen.run(
2087                 args, text=True, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
2088             assert not stderr
2089             assert not error
2090             return stdout
2091
2092         for argument in tests:
2093             if isinstance(argument, str):
2094                 expected = argument
2095             else:
2096                 argument, expected = argument
2097
2098             args = [sys.executable, '-c', 'import sys; print(end=sys.argv[1])', argument, 'end']
2099             assert run_shell(args) == expected
2100             assert run_shell(shell_quote(args, shell=True)) == expected
2101
2102
2103 if __name__ == '__main__':
2104     unittest.main()