test/test_utils.py

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