test/test_utils.py

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