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
   9 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
  10
  11
  12 import contextlib
  13 import io
  14 import itertools
  15 import json
  16 import xml.etree.ElementTree
  17
  18 from yt_dlp.compat import (
  19     compat_etree_fromstring,
  20     compat_HTMLParseError,
  21     compat_os_name,
  22 )
  23 from yt_dlp.utils import (
  24     Config,
  25     DateRange,
  26     ExtractorError,
  27     InAdvancePagedList,
  28     LazyList,
  29     OnDemandPagedList,
  30     age_restricted,
  31     args_to_str,
  32     base_url,
  33     caesar,
  34     clean_html,
  35     clean_podcast_url,
  36     cli_bool_option,
  37     cli_option,
  38     cli_valueless_option,
  39     date_from_str,
  40     datetime_from_str,
  41     detect_exe_version,
  42     determine_ext,
  43     determine_file_encoding,
  44     dfxp2srt,
  45     dict_get,
  46     encode_base_n,
  47     encode_compat_str,
  48     encodeFilename,
  49     escape_rfc3986,
  50     escape_url,
  51     expand_path,
  52     extract_attributes,
  53     find_xpath_attr,
  54     fix_xml_ampersands,
  55     float_or_none,
  56     format_bytes,
  57     get_compatible_ext,
  58     get_element_by_attribute,
  59     get_element_by_class,
  60     get_element_html_by_attribute,
  61     get_element_html_by_class,
  62     get_element_text_and_html_by_tag,
  63     get_elements_by_attribute,
  64     get_elements_by_class,
  65     get_elements_html_by_attribute,
  66     get_elements_html_by_class,
  67     get_elements_text_and_html_by_attribute,
  68     int_or_none,
  69     intlist_to_bytes,
  70     iri_to_uri,
  71     is_html,
  72     js_to_json,
  73     limit_length,
  74     locked_file,
  75     lowercase_escape,
  76     match_str,
  77     merge_dicts,
  78     mimetype2ext,
  79     month_by_name,
  80     multipart_encode,
  81     ohdave_rsa_encrypt,
  82     orderedSet,
  83     parse_age_limit,
  84     parse_bitrate,
  85     parse_codecs,
  86     parse_count,
  87     parse_dfxp_time_expr,
  88     parse_duration,
  89     parse_filesize,
  90     parse_iso8601,
  91     parse_qs,
  92     parse_resolution,
  93     pkcs1pad,
  94     prepend_extension,
  95     read_batch_urls,
  96     remove_end,
  97     remove_quotes,
  98     remove_start,
  99     render_table,
 100     replace_extension,
 101     rot47,
 102     sanitize_filename,
 103     sanitize_path,
 104     sanitize_url,
 105     sanitized_Request,
 106     shell_quote,
 107     smuggle_url,
 108     str_or_none,
 109     str_to_int,
 110     strip_jsonp,
 111     strip_or_none,
 112     subtitles_filename,
 113     timeconvert,
 114     traverse_obj,
 115     unescapeHTML,
 116     unified_strdate,
 117     unified_timestamp,
 118     unsmuggle_url,
 119     update_url_query,
 120     uppercase_escape,
 121     url_basename,
 122     url_or_none,
 123     urlencode_postdata,
 124     urljoin,
 125     urshift,
 126     version_tuple,
 127     xpath_attr,
 128     xpath_element,
 129     xpath_text,
 130     xpath_with_ns,
 131 )
 132
 133
 134 class TestUtil(unittest.TestCase):
 135     def test_timeconvert(self):
 136         self.assertTrue(timeconvert('') is None)
 137         self.assertTrue(timeconvert('bougrg') is None)
 138
 139     def test_sanitize_filename(self):
 140         self.assertEqual(sanitize_filename(''), '')
 141         self.assertEqual(sanitize_filename('abc'), 'abc')
 142         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
 143
 144         self.assertEqual(sanitize_filename('123'), '123')
 145
 146         self.assertEqual('abc⧸de', sanitize_filename('abc/de'))
 147         self.assertFalse('/' in sanitize_filename('abc/de///'))
 148
 149         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', is_id=False))
 150         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', is_id=False))
 151         self.assertEqual('yes no', sanitize_filename('yes? no', is_id=False))
 152         self.assertEqual('this - that', sanitize_filename('this: that', is_id=False))
 153
 154         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
 155         aumlaut = 'ä'
 156         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
 157         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
 158         self.assertEqual(sanitize_filename(tests), tests)
 159
 160         self.assertEqual(
 161             sanitize_filename('New World record at 0:12:34'),
 162             'New World record at 0_12_34')
 163
 164         self.assertEqual(sanitize_filename('--gasdgf'), '--gasdgf')
 165         self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
 166         self.assertEqual(sanitize_filename('--gasdgf', is_id=False), '_-gasdgf')
 167         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
 171         forbidden = '"\0\\/'
 172         for fc in forbidden:
 173             for fbc in forbidden:
 174                 self.assertTrue(fbc not in sanitize_filename(fc))
 175
 176     def test_sanitize_filename_restricted(self):
 177         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
 178         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
 179
 180         self.assertEqual(sanitize_filename('123', restricted=True), '123')
 181
 182         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
 183         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
 184
 185         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
 186         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
 187         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
 188         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
 189
 190         tests = 'aäb\u4e2d\u56fd\u7684c'
 191         self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
 192         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
 193
 194         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
 195         for fc in forbidden:
 196             for fbc in forbidden:
 197                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
 198
 199         # Handle a common case more neatly
 200         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
 201         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
 202         # .. but make sure the file name is never empty
 203         self.assertTrue(sanitize_filename('-', restricted=True) != '')
 204         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 205
 206         self.assertEqual(sanitize_filename(
 207             'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
 208             'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy')
 209
 210     def test_sanitize_ids(self):
 211         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 212         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 213         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 214
 215     def test_sanitize_path(self):
 216         if sys.platform != 'win32':
 217             return
 218
 219         self.assertEqual(sanitize_path('abc'), 'abc')
 220         self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
 221         self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
 222         self.assertEqual(sanitize_path('abc|def'), 'abc#def')
 223         self.assertEqual(sanitize_path('<>:"|?*'), '#######')
 224         self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
 225         self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
 226
 227         self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
 228         self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
 229
 230         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 231         self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
 232         self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
 233         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 234
 235         self.assertEqual(
 236             sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
 237             'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
 238
 239         self.assertEqual(
 240             sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
 241             'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
 242         self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
 243         self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
 244         self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
 245
 246         self.assertEqual(sanitize_path('../abc'), '..\\abc')
 247         self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
 248         self.assertEqual(sanitize_path('./abc'), 'abc')
 249         self.assertEqual(sanitize_path('./../abc'), '..\\abc')
 250
 251     def test_sanitize_url(self):
 252         self.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
 253         self.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
 254         self.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
 255         self.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')
 256         self.assertEqual(sanitize_url('foo bar'), 'foo bar')
 257
 258     def test_extract_basic_auth(self):
 259         auth_header = lambda url: sanitized_Request(url).get_header('Authorization')
 260         self.assertFalse(auth_header('http://foo.bar'))
 261         self.assertFalse(auth_header('http://:foo.bar'))
 262         self.assertEqual(auth_header('http://@foo.bar'), 'Basic Og==')
 263         self.assertEqual(auth_header('http://:pass@foo.bar'), 'Basic OnBhc3M=')
 264         self.assertEqual(auth_header('http://user:@foo.bar'), 'Basic dXNlcjo=')
 265         self.assertEqual(auth_header('http://user:pass@foo.bar'), 'Basic dXNlcjpwYXNz')
 266
 267     def test_expand_path(self):
 268         def env(var):
 269             return f'%{var}%' if sys.platform == 'win32' else f'${var}'
 270
 271         os.environ['yt_dlp_EXPATH_PATH'] = 'expanded'
 272         self.assertEqual(expand_path(env('yt_dlp_EXPATH_PATH')), 'expanded')
 273
 274         old_home = os.environ.get('HOME')
 275         test_str = R'C:\Documents and Settings\тест\Application Data'
 276         try:
 277             os.environ['HOME'] = test_str
 278             self.assertEqual(expand_path(env('HOME')), os.getenv('HOME'))
 279             self.assertEqual(expand_path('~'), os.getenv('HOME'))
 280             self.assertEqual(
 281                 expand_path('~/%s' % env('yt_dlp_EXPATH_PATH')),
 282                 '%s/expanded' % os.getenv('HOME'))
 283         finally:
 284             os.environ['HOME'] = old_home or ''
 285
 286     def test_prepend_extension(self):
 287         self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
 288         self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
 289         self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 290         self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
 291         self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
 292         self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
 293
 294     def test_replace_extension(self):
 295         self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
 296         self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
 297         self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 298         self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
 299         self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
 300         self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
 301
 302     def test_subtitles_filename(self):
 303         self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt'), 'abc.en.vtt')
 304         self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt', 'ext'), 'abc.en.vtt')
 305         self.assertEqual(subtitles_filename('abc.unexpected_ext', 'en', 'vtt', 'ext'), 'abc.unexpected_ext.en.vtt')
 306
 307     def test_remove_start(self):
 308         self.assertEqual(remove_start(None, 'A - '), None)
 309         self.assertEqual(remove_start('A - B', 'A - '), 'B')
 310         self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
 311
 312     def test_remove_end(self):
 313         self.assertEqual(remove_end(None, ' - B'), None)
 314         self.assertEqual(remove_end('A - B', ' - B'), 'A')
 315         self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
 316
 317     def test_remove_quotes(self):
 318         self.assertEqual(remove_quotes(None), None)
 319         self.assertEqual(remove_quotes('"'), '"')
 320         self.assertEqual(remove_quotes("'"), "'")
 321         self.assertEqual(remove_quotes(';'), ';')
 322         self.assertEqual(remove_quotes('";'), '";')
 323         self.assertEqual(remove_quotes('""'), '')
 324         self.assertEqual(remove_quotes('";"'), ';')
 325
 326     def test_ordered_set(self):
 327         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
 328         self.assertEqual(orderedSet([]), [])
 329         self.assertEqual(orderedSet([1]), [1])
 330         # keep the list ordered
 331         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
 332
 333     def test_unescape_html(self):
 334         self.assertEqual(unescapeHTML('%20;'), '%20;')
 335         self.assertEqual(unescapeHTML('&#x2F;'), '/')
 336         self.assertEqual(unescapeHTML('&#47;'), '/')
 337         self.assertEqual(unescapeHTML('&eacute;'), 'é')
 338         self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
 339         self.assertEqual(unescapeHTML('&a&quot;'), '&a"')
 340         # HTML5 entities
 341         self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')
 342
 343     def test_date_from_str(self):
 344         self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
 345         self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
 346         self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
 347         self.assertEqual(date_from_str('20200229+365day'), date_from_str('20200229+1year'))
 348         self.assertEqual(date_from_str('20210131+28day'), date_from_str('20210131+1month'))
 349
 350     def test_datetime_from_str(self):
 351         self.assertEqual(datetime_from_str('yesterday', precision='day'), datetime_from_str('now-1day', precision='auto'))
 352         self.assertEqual(datetime_from_str('now+7day', precision='day'), datetime_from_str('now+1week', precision='auto'))
 353         self.assertEqual(datetime_from_str('now+14day', precision='day'), datetime_from_str('now+2week', precision='auto'))
 354         self.assertEqual(datetime_from_str('20200229+365day', precision='day'), datetime_from_str('20200229+1year', precision='auto'))
 355         self.assertEqual(datetime_from_str('20210131+28day', precision='day'), datetime_from_str('20210131+1month', precision='auto'))
 356         self.assertEqual(datetime_from_str('20210131+59day', precision='day'), datetime_from_str('20210131+2month', precision='auto'))
 357         self.assertEqual(datetime_from_str('now+1day', precision='hour'), datetime_from_str('now+24hours', precision='auto'))
 358         self.assertEqual(datetime_from_str('now+23hours', precision='hour'), datetime_from_str('now+23hours', precision='auto'))
 359
 360     def test_daterange(self):
 361         _20century = DateRange("19000101", "20000101")
 362         self.assertFalse("17890714" in _20century)
 363         _ac = DateRange("00010101")
 364         self.assertTrue("19690721" in _ac)
 365         _firstmilenium = DateRange(end="10000101")
 366         self.assertTrue("07110427" in _firstmilenium)
 367
 368     def test_unified_dates(self):
 369         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
 370         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
 371         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
 372         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
 373         self.assertEqual(unified_strdate('1968 12 10'), '19681210')
 374         self.assertEqual(unified_strdate('1968-12-10'), '19681210')
 375         self.assertEqual(unified_strdate('31-07-2022 20:00'), '20220731')
 376         self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
 377         self.assertEqual(
 378             unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
 379             '20141126')
 380         self.assertEqual(
 381             unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
 382             '20150202')
 383         self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
 384         self.assertEqual(unified_strdate('25-09-2014'), '20140925')
 385         self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
 386         self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
 387         self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
 388         self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
 389         self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
 390         self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
 391         self.assertEqual(unified_strdate('November 3rd, 2019'), '20191103')
 392         self.assertEqual(unified_strdate('October 23rd, 2005'), '20051023')
 393
 394     def test_unified_timestamps(self):
 395         self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
 396         self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
 397         self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
 398         self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
 399         self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
 400         self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
 401         self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
 402         self.assertEqual(
 403             unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
 404             1417001400)
 405         self.assertEqual(
 406             unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
 407             1422902860)
 408         self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
 409         self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
 410         self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
 411         self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
 412         self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
 413         self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
 414         self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
 415         self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
 416         self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
 417         self.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363)
 418
 419         self.assertEqual(unified_timestamp('December 31 1969 20:00:01 EDT'), 1)
 420         self.assertEqual(unified_timestamp('Wednesday 31 December 1969 18:01:26 MDT'), 86)
 421         self.assertEqual(unified_timestamp('12/31/1969 20:01:18 EDT', False), 78)
 422
 423     def test_determine_ext(self):
 424         self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
 425         self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
 426         self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
 427         self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
 428         self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
 429         self.assertEqual(determine_ext('foobar', None), None)
 430
 431     def test_find_xpath_attr(self):
 432         testxml = '''<root>
 433             <node/>
 434             <node x="a"/>
 435             <node x="a" y="c" />
 436             <node x="b" y="d" />
 437             <node x="" />
 438         </root>'''
 439         doc = compat_etree_fromstring(testxml)
 440
 441         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
 442         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
 443         self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
 444         self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
 445         self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
 446         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
 447         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
 448         self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
 449         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
 450         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
 451         self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
 452
 453     def test_xpath_with_ns(self):
 454         testxml = '''<root xmlns:media="http://example.com/">
 455             <media:song>
 456                 <media:author>The Author</media:author>
 457                 <url>http://server.com/download.mp3</url>
 458             </media:song>
 459         </root>'''
 460         doc = compat_etree_fromstring(testxml)
 461         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
 462         self.assertTrue(find('media:song') is not None)
 463         self.assertEqual(find('media:song/media:author').text, 'The Author')
 464         self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
 465
 466     def test_xpath_element(self):
 467         doc = xml.etree.ElementTree.Element('root')
 468         div = xml.etree.ElementTree.SubElement(doc, 'div')
 469         p = xml.etree.ElementTree.SubElement(div, 'p')
 470         p.text = 'Foo'
 471         self.assertEqual(xpath_element(doc, 'div/p'), p)
 472         self.assertEqual(xpath_element(doc, ['div/p']), p)
 473         self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
 474         self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
 475         self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
 476         self.assertTrue(xpath_element(doc, 'div/bar') is None)
 477         self.assertTrue(xpath_element(doc, ['div/bar']) is None)
 478         self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
 479         self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
 480         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
 481         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
 482
 483     def test_xpath_text(self):
 484         testxml = '''<root>
 485             <div>
 486                 <p>Foo</p>
 487             </div>
 488         </root>'''
 489         doc = compat_etree_fromstring(testxml)
 490         self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
 491         self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
 492         self.assertTrue(xpath_text(doc, 'div/bar') is None)
 493         self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
 494
 495     def test_xpath_attr(self):
 496         testxml = '''<root>
 497             <div>
 498                 <p x="a">Foo</p>
 499             </div>
 500         </root>'''
 501         doc = compat_etree_fromstring(testxml)
 502         self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
 503         self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
 504         self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
 505         self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
 506         self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
 507         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
 508         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
 509
 510     def test_smuggle_url(self):
 511         data = {"ö": "ö", "abc": [3]}
 512         url = 'https://foo.bar/baz?x=y#a'
 513         smug_url = smuggle_url(url, data)
 514         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
 515         self.assertEqual(url, unsmug_url)
 516         self.assertEqual(data, unsmug_data)
 517
 518         res_url, res_data = unsmuggle_url(url)
 519         self.assertEqual(res_url, url)
 520         self.assertEqual(res_data, None)
 521
 522         smug_url = smuggle_url(url, {'a': 'b'})
 523         smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
 524         res_url, res_data = unsmuggle_url(smug_smug_url)
 525         self.assertEqual(res_url, url)
 526         self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
 527
 528     def test_shell_quote(self):
 529         args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
 530         self.assertEqual(
 531             shell_quote(args),
 532             """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
 533
 534     def test_float_or_none(self):
 535         self.assertEqual(float_or_none('42.42'), 42.42)
 536         self.assertEqual(float_or_none('42'), 42.0)
 537         self.assertEqual(float_or_none(''), None)
 538         self.assertEqual(float_or_none(None), None)
 539         self.assertEqual(float_or_none([]), None)
 540         self.assertEqual(float_or_none(set()), None)
 541
 542     def test_int_or_none(self):
 543         self.assertEqual(int_or_none('42'), 42)
 544         self.assertEqual(int_or_none(''), None)
 545         self.assertEqual(int_or_none(None), None)
 546         self.assertEqual(int_or_none([]), None)
 547         self.assertEqual(int_or_none(set()), None)
 548
 549     def test_str_to_int(self):
 550         self.assertEqual(str_to_int('123,456'), 123456)
 551         self.assertEqual(str_to_int('123.456'), 123456)
 552         self.assertEqual(str_to_int(523), 523)
 553         self.assertEqual(str_to_int('noninteger'), None)
 554         self.assertEqual(str_to_int([]), None)
 555
 556     def test_url_basename(self):
 557         self.assertEqual(url_basename('http://foo.de/'), '')
 558         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
 559         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
 560         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
 561         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
 562         self.assertEqual(
 563             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
 564             'trailer.mp4')
 565
 566     def test_base_url(self):
 567         self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
 568         self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
 569         self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
 570         self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
 571         self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
 572         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/')
 573
 574     def test_urljoin(self):
 575         self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 576         self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 577         self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 578         self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 579         self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
 580         self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 581         self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 582         self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 583         self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 584         self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
 585         self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 586         self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
 587         self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 588         self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 589         self.assertEqual(urljoin('http://foo.de/', None), None)
 590         self.assertEqual(urljoin('http://foo.de/', ''), None)
 591         self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
 592         self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
 593         self.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de')
 594         self.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de')
 595
 596     def test_url_or_none(self):
 597         self.assertEqual(url_or_none(None), None)
 598         self.assertEqual(url_or_none(''), None)
 599         self.assertEqual(url_or_none('foo'), None)
 600         self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
 601         self.assertEqual(url_or_none('https://foo.de'), 'https://foo.de')
 602         self.assertEqual(url_or_none('http$://foo.de'), None)
 603         self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
 604         self.assertEqual(url_or_none('//foo.de'), '//foo.de')
 605         self.assertEqual(url_or_none('s3://foo.de'), None)
 606         self.assertEqual(url_or_none('rtmpte://foo.de'), 'rtmpte://foo.de')
 607         self.assertEqual(url_or_none('mms://foo.de'), 'mms://foo.de')
 608         self.assertEqual(url_or_none('rtspu://foo.de'), 'rtspu://foo.de')
 609         self.assertEqual(url_or_none('ftps://foo.de'), 'ftps://foo.de')
 610
 611     def test_parse_age_limit(self):
 612         self.assertEqual(parse_age_limit(None), None)
 613         self.assertEqual(parse_age_limit(False), None)
 614         self.assertEqual(parse_age_limit('invalid'), None)
 615         self.assertEqual(parse_age_limit(0), 0)
 616         self.assertEqual(parse_age_limit(18), 18)
 617         self.assertEqual(parse_age_limit(21), 21)
 618         self.assertEqual(parse_age_limit(22), None)
 619         self.assertEqual(parse_age_limit('18'), 18)
 620         self.assertEqual(parse_age_limit('18+'), 18)
 621         self.assertEqual(parse_age_limit('PG-13'), 13)
 622         self.assertEqual(parse_age_limit('TV-14'), 14)
 623         self.assertEqual(parse_age_limit('TV-MA'), 17)
 624         self.assertEqual(parse_age_limit('TV14'), 14)
 625         self.assertEqual(parse_age_limit('TV_G'), 0)
 626
 627     def test_parse_duration(self):
 628         self.assertEqual(parse_duration(None), None)
 629         self.assertEqual(parse_duration(False), None)
 630         self.assertEqual(parse_duration('invalid'), None)
 631         self.assertEqual(parse_duration('1'), 1)
 632         self.assertEqual(parse_duration('1337:12'), 80232)
 633         self.assertEqual(parse_duration('9:12:43'), 33163)
 634         self.assertEqual(parse_duration('12:00'), 720)
 635         self.assertEqual(parse_duration('00:01:01'), 61)
 636         self.assertEqual(parse_duration('x:y'), None)
 637         self.assertEqual(parse_duration('3h11m53s'), 11513)
 638         self.assertEqual(parse_duration('3h 11m 53s'), 11513)
 639         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
 640         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
 641         self.assertEqual(parse_duration('3 hours, 11 minutes, 53 seconds'), 11513)
 642         self.assertEqual(parse_duration('3 hours, 11 mins, 53 secs'), 11513)
 643         self.assertEqual(parse_duration('62m45s'), 3765)
 644         self.assertEqual(parse_duration('6m59s'), 419)
 645         self.assertEqual(parse_duration('49s'), 49)
 646         self.assertEqual(parse_duration('0h0m0s'), 0)
 647         self.assertEqual(parse_duration('0m0s'), 0)
 648         self.assertEqual(parse_duration('0s'), 0)
 649         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
 650         self.assertEqual(parse_duration('T30M38S'), 1838)
 651         self.assertEqual(parse_duration('5 s'), 5)
 652         self.assertEqual(parse_duration('3 min'), 180)
 653         self.assertEqual(parse_duration('2.5 hours'), 9000)
 654         self.assertEqual(parse_duration('02:03:04'), 7384)
 655         self.assertEqual(parse_duration('01:02:03:04'), 93784)
 656         self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
 657         self.assertEqual(parse_duration('87 Min.'), 5220)
 658         self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
 659         self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
 660         self.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)
 661         self.assertEqual(parse_duration('01:02:03:050'), 3723.05)
 662         self.assertEqual(parse_duration('103:050'), 103.05)
 663
 664     def test_fix_xml_ampersands(self):
 665         self.assertEqual(
 666             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 667         self.assertEqual(
 668             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 669             '"&amp;x=y&amp;wrong;&amp;z=a')
 670         self.assertEqual(
 671             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 672             '&amp;&apos;&gt;&lt;&quot;')
 673         self.assertEqual(
 674             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 675         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 676
 677     def test_paged_list(self):
 678         def testPL(size, pagesize, sliceargs, expected):
 679             def get_page(pagenum):
 680                 firstid = pagenum * pagesize
 681                 upto = min(size, pagenum * pagesize + pagesize)
 682                 yield from range(firstid, upto)
 683
 684             pl = OnDemandPagedList(get_page, pagesize)
 685             got = pl.getslice(*sliceargs)
 686             self.assertEqual(got, expected)
 687
 688             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
 689             got = iapl.getslice(*sliceargs)
 690             self.assertEqual(got, expected)
 691
 692         testPL(5, 2, (), [0, 1, 2, 3, 4])
 693         testPL(5, 2, (1,), [1, 2, 3, 4])
 694         testPL(5, 2, (2,), [2, 3, 4])
 695         testPL(5, 2, (4,), [4])
 696         testPL(5, 2, (0, 3), [0, 1, 2])
 697         testPL(5, 2, (1, 4), [1, 2, 3])
 698         testPL(5, 2, (2, 99), [2, 3, 4])
 699         testPL(5, 2, (20, 99), [])
 700
 701     def test_read_batch_urls(self):
 702         f = io.StringIO('''\xef\xbb\xbf foo
 703             bar\r
 704             baz
 705             # More after this line\r
 706             ; or after this
 707             bam''')
 708         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
 709
 710     def test_urlencode_postdata(self):
 711         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
 712         self.assertTrue(isinstance(data, bytes))
 713
 714     def test_update_url_query(self):
 715         self.assertEqual(parse_qs(update_url_query(
 716             'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
 717             parse_qs('http://example.com/path?quality=HD&format=mp4'))
 718         self.assertEqual(parse_qs(update_url_query(
 719             'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
 720             parse_qs('http://example.com/path?system=LINUX&system=WINDOWS'))
 721         self.assertEqual(parse_qs(update_url_query(
 722             'http://example.com/path', {'fields': 'id,formats,subtitles'})),
 723             parse_qs('http://example.com/path?fields=id,formats,subtitles'))
 724         self.assertEqual(parse_qs(update_url_query(
 725             'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
 726             parse_qs('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
 727         self.assertEqual(parse_qs(update_url_query(
 728             'http://example.com/path?manifest=f4m', {'manifest': []})),
 729             parse_qs('http://example.com/path'))
 730         self.assertEqual(parse_qs(update_url_query(
 731             'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
 732             parse_qs('http://example.com/path?system=LINUX'))
 733         self.assertEqual(parse_qs(update_url_query(
 734             'http://example.com/path', {'fields': b'id,formats,subtitles'})),
 735             parse_qs('http://example.com/path?fields=id,formats,subtitles'))
 736         self.assertEqual(parse_qs(update_url_query(
 737             'http://example.com/path', {'width': 1080, 'height': 720})),
 738             parse_qs('http://example.com/path?width=1080&height=720'))
 739         self.assertEqual(parse_qs(update_url_query(
 740             'http://example.com/path', {'bitrate': 5020.43})),
 741             parse_qs('http://example.com/path?bitrate=5020.43'))
 742         self.assertEqual(parse_qs(update_url_query(
 743             'http://example.com/path', {'test': '第二行тест'})),
 744             parse_qs('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
 745
 746     def test_multipart_encode(self):
 747         self.assertEqual(
 748             multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0],
 749             b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
 750         self.assertEqual(
 751             multipart_encode({'欄位'.encode(): '值'.encode()}, boundary='AAAAAA')[0],
 752             b'--AAAAAA\r\nContent-Disposition: form-data; name="\xe6\xac\x84\xe4\xbd\x8d"\r\n\r\n\xe5\x80\xbc\r\n--AAAAAA--\r\n')
 753         self.assertRaises(
 754             ValueError, multipart_encode, {b'field': b'value'}, boundary='value')
 755
 756     def test_dict_get(self):
 757         FALSE_VALUES = {
 758             'none': None,
 759             'false': False,
 760             'zero': 0,
 761             'empty_string': '',
 762             'empty_list': [],
 763         }
 764         d = FALSE_VALUES.copy()
 765         d['a'] = 42
 766         self.assertEqual(dict_get(d, 'a'), 42)
 767         self.assertEqual(dict_get(d, 'b'), None)
 768         self.assertEqual(dict_get(d, 'b', 42), 42)
 769         self.assertEqual(dict_get(d, ('a', )), 42)
 770         self.assertEqual(dict_get(d, ('b', 'a', )), 42)
 771         self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
 772         self.assertEqual(dict_get(d, ('b', 'c', )), None)
 773         self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
 774         for key, false_value in FALSE_VALUES.items():
 775             self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
 776             self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
 777
 778     def test_merge_dicts(self):
 779         self.assertEqual(merge_dicts({'a': 1}, {'b': 2}), {'a': 1, 'b': 2})
 780         self.assertEqual(merge_dicts({'a': 1}, {'a': 2}), {'a': 1})
 781         self.assertEqual(merge_dicts({'a': 1}, {'a': None}), {'a': 1})
 782         self.assertEqual(merge_dicts({'a': 1}, {'a': ''}), {'a': 1})
 783         self.assertEqual(merge_dicts({'a': 1}, {}), {'a': 1})
 784         self.assertEqual(merge_dicts({'a': None}, {'a': 1}), {'a': 1})
 785         self.assertEqual(merge_dicts({'a': ''}, {'a': 1}), {'a': ''})
 786         self.assertEqual(merge_dicts({'a': ''}, {'a': 'abc'}), {'a': 'abc'})
 787         self.assertEqual(merge_dicts({'a': None}, {'a': ''}, {'a': 'abc'}), {'a': 'abc'})
 788
 789     def test_encode_compat_str(self):
 790         self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
 791         self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
 792
 793     def test_parse_iso8601(self):
 794         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
 795         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
 796         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
 797         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
 798         self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
 799         self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
 800
 801     def test_strip_jsonp(self):
 802         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
 803         d = json.loads(stripped)
 804         self.assertEqual(d, [{"id": "532cb", "x": 3}])
 805
 806         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
 807         d = json.loads(stripped)
 808         self.assertEqual(d, {'STATUS': 'OK'})
 809
 810         stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
 811         d = json.loads(stripped)
 812         self.assertEqual(d, {'status': 'success'})
 813
 814         stripped = strip_jsonp('window.cb && window.cb({"status": "success"});')
 815         d = json.loads(stripped)
 816         self.assertEqual(d, {'status': 'success'})
 817
 818         stripped = strip_jsonp('window.cb && cb({"status": "success"});')
 819         d = json.loads(stripped)
 820         self.assertEqual(d, {'status': 'success'})
 821
 822         stripped = strip_jsonp('({"status": "success"});')
 823         d = json.loads(stripped)
 824         self.assertEqual(d, {'status': 'success'})
 825
 826     def test_strip_or_none(self):
 827         self.assertEqual(strip_or_none(' abc'), 'abc')
 828         self.assertEqual(strip_or_none('abc '), 'abc')
 829         self.assertEqual(strip_or_none(' abc '), 'abc')
 830         self.assertEqual(strip_or_none('\tabc\t'), 'abc')
 831         self.assertEqual(strip_or_none('\n\tabc\n\t'), 'abc')
 832         self.assertEqual(strip_or_none('abc'), 'abc')
 833         self.assertEqual(strip_or_none(''), '')
 834         self.assertEqual(strip_or_none(None), None)
 835         self.assertEqual(strip_or_none(42), None)
 836         self.assertEqual(strip_or_none([]), None)
 837
 838     def test_uppercase_escape(self):
 839         self.assertEqual(uppercase_escape('aä'), 'aä')
 840         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
 841
 842     def test_lowercase_escape(self):
 843         self.assertEqual(lowercase_escape('aä'), 'aä')
 844         self.assertEqual(lowercase_escape('\\u0026'), '&')
 845
 846     def test_limit_length(self):
 847         self.assertEqual(limit_length(None, 12), None)
 848         self.assertEqual(limit_length('foo', 12), 'foo')
 849         self.assertTrue(
 850             limit_length('foo bar baz asd', 12).startswith('foo bar'))
 851         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
 852
 853     def test_mimetype2ext(self):
 854         self.assertEqual(mimetype2ext(None), None)
 855         self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
 856         self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
 857         self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
 858         self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
 859         self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
 860         self.assertEqual(mimetype2ext('audio/x-wav'), 'wav')
 861         self.assertEqual(mimetype2ext('audio/x-wav;codec=pcm'), 'wav')
 862
 863     def test_month_by_name(self):
 864         self.assertEqual(month_by_name(None), None)
 865         self.assertEqual(month_by_name('December', 'en'), 12)
 866         self.assertEqual(month_by_name('décembre', 'fr'), 12)
 867         self.assertEqual(month_by_name('December'), 12)
 868         self.assertEqual(month_by_name('décembre'), None)
 869         self.assertEqual(month_by_name('Unknown', 'unknown'), None)
 870
 871     def test_parse_codecs(self):
 872         self.assertEqual(parse_codecs(''), {})
 873         self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
 874             'vcodec': 'avc1.77.30',
 875             'acodec': 'mp4a.40.2',
 876             'dynamic_range': None,
 877         })
 878         self.assertEqual(parse_codecs('mp4a.40.2'), {
 879             'vcodec': 'none',
 880             'acodec': 'mp4a.40.2',
 881             'dynamic_range': None,
 882         })
 883         self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
 884             'vcodec': 'avc1.42001e',
 885             'acodec': 'mp4a.40.5',
 886             'dynamic_range': None,
 887         })
 888         self.assertEqual(parse_codecs('avc3.640028'), {
 889             'vcodec': 'avc3.640028',
 890             'acodec': 'none',
 891             'dynamic_range': None,
 892         })
 893         self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
 894             'vcodec': 'h264',
 895             'acodec': 'aac',
 896             'dynamic_range': None,
 897         })
 898         self.assertEqual(parse_codecs('av01.0.05M.08'), {
 899             'vcodec': 'av01.0.05M.08',
 900             'acodec': 'none',
 901             'dynamic_range': None,
 902         })
 903         self.assertEqual(parse_codecs('vp9.2'), {
 904             'vcodec': 'vp9.2',
 905             'acodec': 'none',
 906             'dynamic_range': 'HDR10',
 907         })
 908         self.assertEqual(parse_codecs('av01.0.12M.10.0.110.09.16.09.0'), {
 909             'vcodec': 'av01.0.12M.10.0.110.09.16.09.0',
 910             'acodec': 'none',
 911             'dynamic_range': 'HDR10',
 912         })
 913         self.assertEqual(parse_codecs('dvhe'), {
 914             'vcodec': 'dvhe',
 915             'acodec': 'none',
 916             'dynamic_range': 'DV',
 917         })
 918         self.assertEqual(parse_codecs('theora, vorbis'), {
 919             'vcodec': 'theora',
 920             'acodec': 'vorbis',
 921             'dynamic_range': None,
 922         })
 923         self.assertEqual(parse_codecs('unknownvcodec, unknownacodec'), {
 924             'vcodec': 'unknownvcodec',
 925             'acodec': 'unknownacodec',
 926         })
 927         self.assertEqual(parse_codecs('unknown'), {})
 928
 929     def test_escape_rfc3986(self):
 930         reserved = "!*'();:@&=+$,/?#[]"
 931         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
 932         self.assertEqual(escape_rfc3986(reserved), reserved)
 933         self.assertEqual(escape_rfc3986(unreserved), unreserved)
 934         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
 935         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
 936         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
 937         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
 938
 939     def test_escape_url(self):
 940         self.assertEqual(
 941             escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
 942             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
 943         )
 944         self.assertEqual(
 945             escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
 946             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
 947         )
 948         self.assertEqual(
 949             escape_url('http://тест.рф/фрагмент'),
 950             'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
 951         )
 952         self.assertEqual(
 953             escape_url('http://тест.рф/абв?абв=абв#абв'),
 954             'http://xn--e1aybc.xn--p1ai/%D0%B0%D0%B1%D0%B2?%D0%B0%D0%B1%D0%B2=%D0%B0%D0%B1%D0%B2#%D0%B0%D0%B1%D0%B2'
 955         )
 956         self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
 957
 958     def test_js_to_json_vars_strings(self):
 959         self.assertDictEqual(
 960             json.loads(js_to_json(
 961                 '''{
 962                     'null': a,
 963                     'nullStr': b,
 964                     'true': c,
 965                     'trueStr': d,
 966                     'false': e,
 967                     'falseStr': f,
 968                     'unresolvedVar': g,
 969                 }''',
 970                 {
 971                     'a': 'null',
 972                     'b': '"null"',
 973                     'c': 'true',
 974                     'd': '"true"',
 975                     'e': 'false',
 976                     'f': '"false"',
 977                     'g': 'var',
 978                 }
 979             )),
 980             {
 981                 'null': None,
 982                 'nullStr': 'null',
 983                 'true': True,
 984                 'trueStr': 'true',
 985                 'false': False,
 986                 'falseStr': 'false',
 987                 'unresolvedVar': 'var'
 988             }
 989         )
 990
 991         self.assertDictEqual(
 992             json.loads(js_to_json(
 993                 '''{
 994                     'int': a,
 995                     'intStr': b,
 996                     'float': c,
 997                     'floatStr': d,
 998                 }''',
 999                 {
1000                     'a': '123',
1001                     'b': '"123"',
1002                     'c': '1.23',
1003                     'd': '"1.23"',
1004                 }
1005             )),
1006             {
1007                 'int': 123,
1008                 'intStr': '123',
1009                 'float': 1.23,
1010                 'floatStr': '1.23',
1011             }
1012         )
1013
1014         self.assertDictEqual(
1015             json.loads(js_to_json(
1016                 '''{
1017                     'object': a,
1018                     'objectStr': b,
1019                     'array': c,
1020                     'arrayStr': d,
1021                 }''',
1022                 {
1023                     'a': '{}',
1024                     'b': '"{}"',
1025                     'c': '[]',
1026                     'd': '"[]"',
1027                 }
1028             )),
1029             {
1030                 'object': {},
1031                 'objectStr': '{}',
1032                 'array': [],
1033                 'arrayStr': '[]',
1034             }
1035         )
1036
1037     def test_js_to_json_realworld(self):
1038         inp = '''{
1039             'clip':{'provider':'pseudo'}
1040         }'''
1041         self.assertEqual(js_to_json(inp), '''{
1042             "clip":{"provider":"pseudo"}
1043         }''')
1044         json.loads(js_to_json(inp))
1045
1046         inp = '''{
1047             'playlist':[{'controls':{'all':null}}]
1048         }'''
1049         self.assertEqual(js_to_json(inp), '''{
1050             "playlist":[{"controls":{"all":null}}]
1051         }''')
1052
1053         inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
1054         self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
1055
1056         inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
1057         json_code = js_to_json(inp)
1058         self.assertEqual(json.loads(json_code), json.loads(inp))
1059
1060         inp = '''{
1061             0:{src:'skipped', type: 'application/dash+xml'},
1062             1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
1063         }'''
1064         self.assertEqual(js_to_json(inp), '''{
1065             "0":{"src":"skipped", "type": "application/dash+xml"},
1066             "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
1067         }''')
1068
1069         inp = '''{"foo":101}'''
1070         self.assertEqual(js_to_json(inp), '''{"foo":101}''')
1071
1072         inp = '''{"duration": "00:01:07"}'''
1073         self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
1074
1075         inp = '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}'''
1076         self.assertEqual(js_to_json(inp), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''')
1077
1078     def test_js_to_json_edgecases(self):
1079         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
1080         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
1081
1082         on = js_to_json('{"abc": true}')
1083         self.assertEqual(json.loads(on), {'abc': True})
1084
1085         # Ignore JavaScript code as well
1086         on = js_to_json('''{
1087             "x": 1,
1088             y: "a",
1089             z: some.code
1090         }''')
1091         d = json.loads(on)
1092         self.assertEqual(d['x'], 1)
1093         self.assertEqual(d['y'], 'a')
1094
1095         # Just drop ! prefix for now though this results in a wrong value
1096         on = js_to_json('''{
1097             a: !0,
1098             b: !1,
1099             c: !!0,
1100             d: !!42.42,
1101             e: !!![],
1102             f: !"abc",
1103             g: !"",
1104             !42: 42
1105         }''')
1106         self.assertEqual(json.loads(on), {
1107             'a': 0,
1108             'b': 1,
1109             'c': 0,
1110             'd': 42.42,
1111             'e': [],
1112             'f': "abc",
1113             'g': "",
1114             '42': 42
1115         })
1116
1117         on = js_to_json('["abc", "def",]')
1118         self.assertEqual(json.loads(on), ['abc', 'def'])
1119
1120         on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
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('{"abc": "def",}')
1127         self.assertEqual(json.loads(on), {'abc': 'def'})
1128
1129         on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
1130         self.assertEqual(json.loads(on), {'abc': 'def'})
1131
1132         on = js_to_json('{ 0: /* " \n */ ",]" , }')
1133         self.assertEqual(json.loads(on), {'0': ',]'})
1134
1135         on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
1136         self.assertEqual(json.loads(on), {'0': ',]'})
1137
1138         on = js_to_json('{ 0: // comment\n1 }')
1139         self.assertEqual(json.loads(on), {'0': 1})
1140
1141         on = js_to_json(r'["<p>x<\/p>"]')
1142         self.assertEqual(json.loads(on), ['<p>x</p>'])
1143
1144         on = js_to_json(r'["\xaa"]')
1145         self.assertEqual(json.loads(on), ['\u00aa'])
1146
1147         on = js_to_json("['a\\\nb']")
1148         self.assertEqual(json.loads(on), ['ab'])
1149
1150         on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
1151         self.assertEqual(json.loads(on), ['ab'])
1152
1153         on = js_to_json('{0xff:0xff}')
1154         self.assertEqual(json.loads(on), {'255': 255})
1155
1156         on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
1157         self.assertEqual(json.loads(on), {'255': 255})
1158
1159         on = js_to_json('{077:077}')
1160         self.assertEqual(json.loads(on), {'63': 63})
1161
1162         on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
1163         self.assertEqual(json.loads(on), {'63': 63})
1164
1165         on = js_to_json('{42:42}')
1166         self.assertEqual(json.loads(on), {'42': 42})
1167
1168         on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
1169         self.assertEqual(json.loads(on), {'42': 42})
1170
1171         on = js_to_json('{42:4.2e1}')
1172         self.assertEqual(json.loads(on), {'42': 42.0})
1173
1174         on = js_to_json('{ "0x40": "0x40" }')
1175         self.assertEqual(json.loads(on), {'0x40': '0x40'})
1176
1177         on = js_to_json('{ "040": "040" }')
1178         self.assertEqual(json.loads(on), {'040': '040'})
1179
1180         on = js_to_json('[1,//{},\n2]')
1181         self.assertEqual(json.loads(on), [1, 2])
1182
1183         on = js_to_json(R'"\^\$\#"')
1184         self.assertEqual(json.loads(on), R'^$#', msg='Unnecessary escapes should be stripped')
1185
1186         on = js_to_json('\'"\\""\'')
1187         self.assertEqual(json.loads(on), '"""', msg='Unnecessary quote escape should be escaped')
1188
1189     def test_js_to_json_malformed(self):
1190         self.assertEqual(js_to_json('42a1'), '42"a1"')
1191         self.assertEqual(js_to_json('42a-1'), '42"a"-1')
1192
1193     def test_extract_attributes(self):
1194         self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
1195         self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
1196         self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
1197         self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
1198         self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
1199         self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
1200         self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
1201         self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'})  # XML
1202         self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
1203         self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'})  # HTML 3.2
1204         self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'})  # HTML 4.0
1205         self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
1206         self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
1207         self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
1208         self.assertEqual(extract_attributes('<e x >'), {'x': None})
1209         self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
1210         self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
1211         self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
1212         self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
1213         self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
1214         self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
1215         self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
1216         self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'})  # Names lowercased
1217         self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
1218         self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
1219         self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
1220         self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
1221         self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
1222         # "Narrow" Python builds don't support unicode code points outside BMP.
1223         try:
1224             chr(0x10000)
1225             supports_outside_bmp = True
1226         except ValueError:
1227             supports_outside_bmp = False
1228         if supports_outside_bmp:
1229             self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
1230         # Malformed HTML should not break attributes extraction on older Python
1231         self.assertEqual(extract_attributes('<mal"formed/>'), {})
1232
1233     def test_clean_html(self):
1234         self.assertEqual(clean_html('a:\nb'), 'a: b')
1235         self.assertEqual(clean_html('a:\n   "b"'), 'a: "b"')
1236         self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
1237
1238     def test_intlist_to_bytes(self):
1239         self.assertEqual(
1240             intlist_to_bytes([0, 1, 127, 128, 255]),
1241             b'\x00\x01\x7f\x80\xff')
1242
1243     def test_args_to_str(self):
1244         self.assertEqual(
1245             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
1246             'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""'
1247         )
1248
1249     def test_parse_filesize(self):
1250         self.assertEqual(parse_filesize(None), None)
1251         self.assertEqual(parse_filesize(''), None)
1252         self.assertEqual(parse_filesize('91 B'), 91)
1253         self.assertEqual(parse_filesize('foobar'), None)
1254         self.assertEqual(parse_filesize('2 MiB'), 2097152)
1255         self.assertEqual(parse_filesize('5 GB'), 5000000000)
1256         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
1257         self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
1258         self.assertEqual(parse_filesize('1,24 KB'), 1240)
1259         self.assertEqual(parse_filesize('1,24 kb'), 1240)
1260         self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
1261
1262     def test_parse_count(self):
1263         self.assertEqual(parse_count(None), None)
1264         self.assertEqual(parse_count(''), None)
1265         self.assertEqual(parse_count('0'), 0)
1266         self.assertEqual(parse_count('1000'), 1000)
1267         self.assertEqual(parse_count('1.000'), 1000)
1268         self.assertEqual(parse_count('1.1k'), 1100)
1269         self.assertEqual(parse_count('1.1 k'), 1100)
1270         self.assertEqual(parse_count('1,1 k'), 1100)
1271         self.assertEqual(parse_count('1.1kk'), 1100000)
1272         self.assertEqual(parse_count('1.1kk '), 1100000)
1273         self.assertEqual(parse_count('1,1kk'), 1100000)
1274         self.assertEqual(parse_count('100 views'), 100)
1275         self.assertEqual(parse_count('1,100 views'), 1100)
1276         self.assertEqual(parse_count('1.1kk views'), 1100000)
1277         self.assertEqual(parse_count('10M views'), 10000000)
1278         self.assertEqual(parse_count('has 10M views'), 10000000)
1279
1280     def test_parse_resolution(self):
1281         self.assertEqual(parse_resolution(None), {})
1282         self.assertEqual(parse_resolution(''), {})
1283         self.assertEqual(parse_resolution(' 1920x1080'), {'width': 1920, 'height': 1080})
1284         self.assertEqual(parse_resolution('1920×1080 '), {'width': 1920, 'height': 1080})
1285         self.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080})
1286         self.assertEqual(parse_resolution('720p'), {'height': 720})
1287         self.assertEqual(parse_resolution('4k'), {'height': 2160})
1288         self.assertEqual(parse_resolution('8K'), {'height': 4320})
1289         self.assertEqual(parse_resolution('pre_1920x1080_post'), {'width': 1920, 'height': 1080})
1290         self.assertEqual(parse_resolution('ep1x2'), {})
1291         self.assertEqual(parse_resolution('1920, 1080'), {'width': 1920, 'height': 1080})
1292
1293     def test_parse_bitrate(self):
1294         self.assertEqual(parse_bitrate(None), None)
1295         self.assertEqual(parse_bitrate(''), None)
1296         self.assertEqual(parse_bitrate('300kbps'), 300)
1297         self.assertEqual(parse_bitrate('1500kbps'), 1500)
1298         self.assertEqual(parse_bitrate('300 kbps'), 300)
1299
1300     def test_version_tuple(self):
1301         self.assertEqual(version_tuple('1'), (1,))
1302         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
1303         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
1304
1305     def test_detect_exe_version(self):
1306         self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
1307 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
1308 configuration: --prefix=/usr --extra-'''), '1.2.1')
1309         self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
1310 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
1311         self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
1312 Trying to open render node...
1313 Success at /dev/dri/renderD128.
1314 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
1315
1316     def test_age_restricted(self):
1317         self.assertFalse(age_restricted(None, 10))  # unrestricted content
1318         self.assertFalse(age_restricted(1, None))  # unrestricted policy
1319         self.assertFalse(age_restricted(8, 10))
1320         self.assertTrue(age_restricted(18, 14))
1321         self.assertFalse(age_restricted(18, 18))
1322
1323     def test_is_html(self):
1324         self.assertFalse(is_html(b'\x49\x44\x43<html'))
1325         self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
1326         self.assertTrue(is_html(  # UTF-8 with BOM
1327             b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
1328         self.assertTrue(is_html(  # UTF-16-LE
1329             b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
1330         ))
1331         self.assertTrue(is_html(  # UTF-16-BE
1332             b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
1333         ))
1334         self.assertTrue(is_html(  # UTF-32-BE
1335             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'))
1336         self.assertTrue(is_html(  # UTF-32-LE
1337             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'))
1338
1339     def test_render_table(self):
1340         self.assertEqual(
1341             render_table(
1342                 ['a', 'empty', 'bcd'],
1343                 [[123, '', 4], [9999, '', 51]]),
1344             'a    empty bcd\n'
1345             '123        4\n'
1346             '9999       51')
1347
1348         self.assertEqual(
1349             render_table(
1350                 ['a', 'empty', 'bcd'],
1351                 [[123, '', 4], [9999, '', 51]],
1352                 hide_empty=True),
1353             'a    bcd\n'
1354             '123  4\n'
1355             '9999 51')
1356
1357         self.assertEqual(
1358             render_table(
1359                 ['\ta', 'bcd'],
1360                 [['1\t23', 4], ['\t9999', 51]]),
1361             '   a bcd\n'
1362             '1 23 4\n'
1363             '9999 51')
1364
1365         self.assertEqual(
1366             render_table(
1367                 ['a', 'bcd'],
1368                 [[123, 4], [9999, 51]],
1369                 delim='-'),
1370             'a    bcd\n'
1371             '--------\n'
1372             '123  4\n'
1373             '9999 51')
1374
1375         self.assertEqual(
1376             render_table(
1377                 ['a', 'bcd'],
1378                 [[123, 4], [9999, 51]],
1379                 delim='-', extra_gap=2),
1380             'a      bcd\n'
1381             '----------\n'
1382             '123    4\n'
1383             '9999   51')
1384
1385     def test_match_str(self):
1386         # Unary
1387         self.assertFalse(match_str('xy', {'x': 1200}))
1388         self.assertTrue(match_str('!xy', {'x': 1200}))
1389         self.assertTrue(match_str('x', {'x': 1200}))
1390         self.assertFalse(match_str('!x', {'x': 1200}))
1391         self.assertTrue(match_str('x', {'x': 0}))
1392         self.assertTrue(match_str('is_live', {'is_live': True}))
1393         self.assertFalse(match_str('is_live', {'is_live': False}))
1394         self.assertFalse(match_str('is_live', {'is_live': None}))
1395         self.assertFalse(match_str('is_live', {}))
1396         self.assertFalse(match_str('!is_live', {'is_live': True}))
1397         self.assertTrue(match_str('!is_live', {'is_live': False}))
1398         self.assertTrue(match_str('!is_live', {'is_live': None}))
1399         self.assertTrue(match_str('!is_live', {}))
1400         self.assertTrue(match_str('title', {'title': 'abc'}))
1401         self.assertTrue(match_str('title', {'title': ''}))
1402         self.assertFalse(match_str('!title', {'title': 'abc'}))
1403         self.assertFalse(match_str('!title', {'title': ''}))
1404
1405         # Numeric
1406         self.assertFalse(match_str('x>0', {'x': 0}))
1407         self.assertFalse(match_str('x>0', {}))
1408         self.assertTrue(match_str('x>?0', {}))
1409         self.assertTrue(match_str('x>1K', {'x': 1200}))
1410         self.assertFalse(match_str('x>2K', {'x': 1200}))
1411         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
1412         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
1413         self.assertTrue(match_str('x > 1:0:0', {'x': 3700}))
1414
1415         # String
1416         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
1417         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
1418         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
1419         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
1420         self.assertTrue(match_str('y^=foo', {'y': 'foobar42'}))
1421         self.assertFalse(match_str('y!^=foo', {'y': 'foobar42'}))
1422         self.assertFalse(match_str('y^=bar', {'y': 'foobar42'}))
1423         self.assertTrue(match_str('y!^=bar', {'y': 'foobar42'}))
1424         self.assertRaises(ValueError, match_str, 'x^=42', {'x': 42})
1425         self.assertTrue(match_str('y*=bar', {'y': 'foobar42'}))
1426         self.assertFalse(match_str('y!*=bar', {'y': 'foobar42'}))
1427         self.assertFalse(match_str('y*=baz', {'y': 'foobar42'}))
1428         self.assertTrue(match_str('y!*=baz', {'y': 'foobar42'}))
1429         self.assertTrue(match_str('y$=42', {'y': 'foobar42'}))
1430         self.assertFalse(match_str('y$=43', {'y': 'foobar42'}))
1431
1432         # And
1433         self.assertFalse(match_str(
1434             'like_count > 100 & dislike_count <? 50 & description',
1435             {'like_count': 90, 'description': 'foo'}))
1436         self.assertTrue(match_str(
1437             'like_count > 100 & dislike_count <? 50 & description',
1438             {'like_count': 190, 'description': 'foo'}))
1439         self.assertFalse(match_str(
1440             'like_count > 100 & dislike_count <? 50 & description',
1441             {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
1442         self.assertFalse(match_str(
1443             'like_count > 100 & dislike_count <? 50 & description',
1444             {'like_count': 190, 'dislike_count': 10}))
1445
1446         # Regex
1447         self.assertTrue(match_str(r'x~=\bbar', {'x': 'foo bar'}))
1448         self.assertFalse(match_str(r'x~=\bbar.+', {'x': 'foo bar'}))
1449         self.assertFalse(match_str(r'x~=^FOO', {'x': 'foo bar'}))
1450         self.assertTrue(match_str(r'x~=(?i)^FOO', {'x': 'foo bar'}))
1451
1452         # Quotes
1453         self.assertTrue(match_str(r'x^="foo"', {'x': 'foo "bar"'}))
1454         self.assertFalse(match_str(r'x^="foo  "', {'x': 'foo "bar"'}))
1455         self.assertFalse(match_str(r'x$="bar"', {'x': 'foo "bar"'}))
1456         self.assertTrue(match_str(r'x$=" \"bar\""', {'x': 'foo "bar"'}))
1457
1458         # Escaping &
1459         self.assertFalse(match_str(r'x=foo & bar', {'x': 'foo & bar'}))
1460         self.assertTrue(match_str(r'x=foo \& bar', {'x': 'foo & bar'}))
1461         self.assertTrue(match_str(r'x=foo \& bar & x^=foo', {'x': 'foo & bar'}))
1462         self.assertTrue(match_str(r'x="foo \& bar" & x^=foo', {'x': 'foo & bar'}))
1463
1464         # Example from docs
1465         self.assertTrue(match_str(
1466             r"!is_live & like_count>?100 & description~='(?i)\bcats \& dogs\b'",
1467             {'description': 'Raining Cats & Dogs'}))
1468
1469         # Incomplete
1470         self.assertFalse(match_str('id!=foo', {'id': 'foo'}, True))
1471         self.assertTrue(match_str('x', {'id': 'foo'}, True))
1472         self.assertTrue(match_str('!x', {'id': 'foo'}, True))
1473         self.assertFalse(match_str('x', {'id': 'foo'}, False))
1474
1475     def test_parse_dfxp_time_expr(self):
1476         self.assertEqual(parse_dfxp_time_expr(None), None)
1477         self.assertEqual(parse_dfxp_time_expr(''), None)
1478         self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1479         self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1480         self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1481         self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1482         self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1483
1484     def test_dfxp2srt(self):
1485         dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
1486             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1487             <body>
1488                 <div xml:lang="en">
1489                     <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1490                     <p begin="1" end="2">第二行<br/>♪♪</p>
1491                     <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1492                     <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1493                     <p begin="-1" end="-1">Ignore, two</p>
1494                     <p begin="3" dur="-1">Ignored, three</p>
1495                 </div>
1496             </body>
1497             </tt>'''.encode()
1498         srt_data = '''1
1499 00:00:00,000 --> 00:00:01,000
1500 The following line contains Chinese characters and special symbols
1501
1502 2
1503 00:00:01,000 --> 00:00:02,000
1504 第二行
1505 ♪♪
1506
1507 3
1508 00:00:02,000 --> 00:00:03,000
1509 Third
1510 Line
1511
1512 '''
1513         self.assertEqual(dfxp2srt(dfxp_data), srt_data)
1514
1515         dfxp_data_no_default_namespace = b'''<?xml version="1.0" encoding="UTF-8"?>
1516             <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1517             <body>
1518                 <div xml:lang="en">
1519                     <p begin="0" end="1">The first line</p>
1520                 </div>
1521             </body>
1522             </tt>'''
1523         srt_data = '''1
1524 00:00:00,000 --> 00:00:01,000
1525 The first line
1526
1527 '''
1528         self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
1529
1530         dfxp_data_with_style = b'''<?xml version="1.0" encoding="utf-8"?>
1531 <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">
1532   <head>
1533     <styling>
1534       <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1535       <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1536       <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1537       <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1538     </styling>
1539   </head>
1540   <body tts:textAlign="center" style="s0">
1541     <div>
1542       <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1543       <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>
1544       <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1545       <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>
1546     </div>
1547   </body>
1548 </tt>'''
1549         srt_data = '''1
1550 00:00:02,080 --> 00:00:05,840
1551 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1552
1553 2
1554 00:00:02,080 --> 00:00:05,840
1555 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1556 </font>part 2</font></b>
1557
1558 3
1559 00:00:05,840 --> 00:00:09,560
1560 <u><font color="lime">line 3
1561 part 3</font></u>
1562
1563 4
1564 00:00:09,560 --> 00:00:12,360
1565 <i><u><font color="yellow"><font color="lime">inner
1566  </font>style</font></u></i>
1567
1568 '''
1569         self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)
1570
1571         dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?>
1572             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1573             <body>
1574                 <div xml:lang="en">
1575                     <p begin="0" end="1">Line 1</p>
1576                     <p begin="1" end="2">第二行</p>
1577                 </div>
1578             </body>
1579             </tt>'''.encode('utf-16')
1580         srt_data = '''1
1581 00:00:00,000 --> 00:00:01,000
1582 Line 1
1583
1584 2
1585 00:00:01,000 --> 00:00:02,000
1586 第二行
1587
1588 '''
1589         self.assertEqual(dfxp2srt(dfxp_data_non_utf8), srt_data)
1590
1591     def test_cli_option(self):
1592         self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1593         self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
1594         self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1595         self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
1596
1597     def test_cli_valueless_option(self):
1598         self.assertEqual(cli_valueless_option(
1599             {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1600         self.assertEqual(cli_valueless_option(
1601             {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
1602         self.assertEqual(cli_valueless_option(
1603             {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1604         self.assertEqual(cli_valueless_option(
1605             {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
1606         self.assertEqual(cli_valueless_option(
1607             {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
1608         self.assertEqual(cli_valueless_option(
1609             {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1610
1611     def test_cli_bool_option(self):
1612         self.assertEqual(
1613             cli_bool_option(
1614                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1615             ['--no-check-certificate', 'true'])
1616         self.assertEqual(
1617             cli_bool_option(
1618                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1619             ['--no-check-certificate=true'])
1620         self.assertEqual(
1621             cli_bool_option(
1622                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1623             ['--check-certificate', 'false'])
1624         self.assertEqual(
1625             cli_bool_option(
1626                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1627             ['--check-certificate=false'])
1628         self.assertEqual(
1629             cli_bool_option(
1630                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1631             ['--check-certificate', 'true'])
1632         self.assertEqual(
1633             cli_bool_option(
1634                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1635             ['--check-certificate=true'])
1636         self.assertEqual(
1637             cli_bool_option(
1638                 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1639             [])
1640
1641     def test_ohdave_rsa_encrypt(self):
1642         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1643         e = 65537
1644
1645         self.assertEqual(
1646             ohdave_rsa_encrypt(b'aa111222', e, N),
1647             '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1648
1649     def test_pkcs1pad(self):
1650         data = [1, 2, 3]
1651         padded_data = pkcs1pad(data, 32)
1652         self.assertEqual(padded_data[:2], [0, 2])
1653         self.assertEqual(padded_data[28:], [0, 1, 2, 3])
1654
1655         self.assertRaises(ValueError, pkcs1pad, data, 8)
1656
1657     def test_encode_base_n(self):
1658         self.assertEqual(encode_base_n(0, 30), '0')
1659         self.assertEqual(encode_base_n(80, 30), '2k')
1660
1661         custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1662         self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1663         self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1664
1665         self.assertRaises(ValueError, encode_base_n, 0, 70)
1666         self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1667
1668     def test_caesar(self):
1669         self.assertEqual(caesar('ace', 'abcdef', 2), 'cea')
1670         self.assertEqual(caesar('cea', 'abcdef', -2), 'ace')
1671         self.assertEqual(caesar('ace', 'abcdef', -2), 'eac')
1672         self.assertEqual(caesar('eac', 'abcdef', 2), 'ace')
1673         self.assertEqual(caesar('ace', 'abcdef', 0), 'ace')
1674         self.assertEqual(caesar('xyz', 'abcdef', 2), 'xyz')
1675         self.assertEqual(caesar('abc', 'acegik', 2), 'ebg')
1676         self.assertEqual(caesar('ebg', 'acegik', -2), 'abc')
1677
1678     def test_rot47(self):
1679         self.assertEqual(rot47('yt-dlp'), r'JE\5=A')
1680         self.assertEqual(rot47('YT-DLP'), r'*%\s{!')
1681
1682     def test_urshift(self):
1683         self.assertEqual(urshift(3, 1), 1)
1684         self.assertEqual(urshift(-3, 1), 2147483646)
1685
1686     GET_ELEMENT_BY_CLASS_TEST_STRING = '''
1687         <span class="foo bar">nice</span>
1688     '''
1689
1690     def test_get_element_by_class(self):
1691         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1692
1693         self.assertEqual(get_element_by_class('foo', html), 'nice')
1694         self.assertEqual(get_element_by_class('no-such-class', html), None)
1695
1696     def test_get_element_html_by_class(self):
1697         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1698
1699         self.assertEqual(get_element_html_by_class('foo', html), html.strip())
1700         self.assertEqual(get_element_by_class('no-such-class', html), None)
1701
1702     GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING = '''
1703         <div itemprop="author" itemscope>foo</div>
1704     '''
1705
1706     def test_get_element_by_attribute(self):
1707         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1708
1709         self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
1710         self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
1711         self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
1712
1713         html = self.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
1714
1715         self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo')
1716
1717     def test_get_element_html_by_attribute(self):
1718         html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1719
1720         self.assertEqual(get_element_html_by_attribute('class', 'foo bar', html), html.strip())
1721         self.assertEqual(get_element_html_by_attribute('class', 'foo', html), None)
1722         self.assertEqual(get_element_html_by_attribute('class', 'no-such-foo', html), None)
1723
1724         html = self.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
1725
1726         self.assertEqual(get_element_html_by_attribute('itemprop', 'author', html), html.strip())
1727
1728     GET_ELEMENTS_BY_CLASS_TEST_STRING = '''
1729         <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1730     '''
1731     GET_ELEMENTS_BY_CLASS_RES = ['<span class="foo bar">nice</span>', '<span class="foo bar">also nice</span>']
1732
1733     def test_get_elements_by_class(self):
1734         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1735
1736         self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
1737         self.assertEqual(get_elements_by_class('no-such-class', html), [])
1738
1739     def test_get_elements_html_by_class(self):
1740         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1741
1742         self.assertEqual(get_elements_html_by_class('foo', html), self.GET_ELEMENTS_BY_CLASS_RES)
1743         self.assertEqual(get_elements_html_by_class('no-such-class', html), [])
1744
1745     def test_get_elements_by_attribute(self):
1746         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1747
1748         self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
1749         self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
1750         self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
1751
1752     def test_get_elements_html_by_attribute(self):
1753         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1754
1755         self.assertEqual(get_elements_html_by_attribute('class', 'foo bar', html), self.GET_ELEMENTS_BY_CLASS_RES)
1756         self.assertEqual(get_elements_html_by_attribute('class', 'foo', html), [])
1757         self.assertEqual(get_elements_html_by_attribute('class', 'no-such-foo', html), [])
1758
1759     def test_get_elements_text_and_html_by_attribute(self):
1760         html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1761
1762         self.assertEqual(
1763             list(get_elements_text_and_html_by_attribute('class', 'foo bar', html)),
1764             list(zip(['nice', 'also nice'], self.GET_ELEMENTS_BY_CLASS_RES)))
1765         self.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'foo', html)), [])
1766         self.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'no-such-foo', html)), [])
1767
1768         self.assertEqual(list(get_elements_text_and_html_by_attribute(
1769             'class', 'foo', '<a class="foo">nice</a><span class="foo">nice</span>', tag='a')), [('nice', '<a class="foo">nice</a>')])
1770
1771     GET_ELEMENT_BY_TAG_TEST_STRING = '''
1772     random text lorem ipsum</p>
1773     <div>
1774         this should be returned
1775         <span>this should also be returned</span>
1776         <div>
1777             this should also be returned
1778         </div>
1779         closing tag above should not trick, so this should also be returned
1780     </div>
1781     but this text should not be returned
1782     '''
1783     GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML = GET_ELEMENT_BY_TAG_TEST_STRING.strip()[32:276]
1784     GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT = GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML[5:-6]
1785     GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML = GET_ELEMENT_BY_TAG_TEST_STRING.strip()[78:119]
1786     GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT = GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML[6:-7]
1787
1788     def test_get_element_text_and_html_by_tag(self):
1789         html = self.GET_ELEMENT_BY_TAG_TEST_STRING
1790
1791         self.assertEqual(
1792             get_element_text_and_html_by_tag('div', html),
1793             (self.GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT, self.GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML))
1794         self.assertEqual(
1795             get_element_text_and_html_by_tag('span', html),
1796             (self.GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT, self.GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML))
1797         self.assertRaises(compat_HTMLParseError, get_element_text_and_html_by_tag, 'article', html)
1798
1799     def test_iri_to_uri(self):
1800         self.assertEqual(
1801             iri_to_uri('https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b'),
1802             'https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b')  # Same
1803         self.assertEqual(
1804             iri_to_uri('https://www.google.com/search?q=Käsesoßenrührlöffel'),  # German for cheese sauce stirring spoon
1805             'https://www.google.com/search?q=K%C3%A4seso%C3%9Fenr%C3%BChrl%C3%B6ffel')
1806         self.assertEqual(
1807             iri_to_uri('https://www.google.com/search?q=lt<+gt>+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#'),
1808             'https://www.google.com/search?q=lt%3C+gt%3E+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#')
1809         self.assertEqual(
1810             iri_to_uri('http://правозащита38.рф/category/news/'),
1811             'http://xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/')
1812         self.assertEqual(
1813             iri_to_uri('http://www.правозащита38.рф/category/news/'),
1814             'http://www.xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/')
1815         self.assertEqual(
1816             iri_to_uri('https://i❤.ws/emojidomain/👍👏🤝💪'),
1817             'https://xn--i-7iq.ws/emojidomain/%F0%9F%91%8D%F0%9F%91%8F%F0%9F%A4%9D%F0%9F%92%AA')
1818         self.assertEqual(
1819             iri_to_uri('http://日本語.jp/'),
1820             'http://xn--wgv71a119e.jp/')
1821         self.assertEqual(
1822             iri_to_uri('http://导航.中国/'),
1823             'http://xn--fet810g.xn--fiqs8s/')
1824
1825     def test_clean_podcast_url(self):
1826         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')
1827         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')
1828
1829     def test_LazyList(self):
1830         it = list(range(10))
1831
1832         self.assertEqual(list(LazyList(it)), it)
1833         self.assertEqual(LazyList(it).exhaust(), it)
1834         self.assertEqual(LazyList(it)[5], it[5])
1835
1836         self.assertEqual(LazyList(it)[5:], it[5:])
1837         self.assertEqual(LazyList(it)[:5], it[:5])
1838         self.assertEqual(LazyList(it)[::2], it[::2])
1839         self.assertEqual(LazyList(it)[1::2], it[1::2])
1840         self.assertEqual(LazyList(it)[5::-1], it[5::-1])
1841         self.assertEqual(LazyList(it)[6:2:-2], it[6:2:-2])
1842         self.assertEqual(LazyList(it)[::-1], it[::-1])
1843
1844         self.assertTrue(LazyList(it))
1845         self.assertFalse(LazyList(range(0)))
1846         self.assertEqual(len(LazyList(it)), len(it))
1847         self.assertEqual(repr(LazyList(it)), repr(it))
1848         self.assertEqual(str(LazyList(it)), str(it))
1849
1850         self.assertEqual(list(LazyList(it, reverse=True)), it[::-1])
1851         self.assertEqual(list(reversed(LazyList(it))[::-1]), it)
1852         self.assertEqual(list(reversed(LazyList(it))[1:3:7]), it[::-1][1:3:7])
1853
1854     def test_LazyList_laziness(self):
1855
1856         def test(ll, idx, val, cache):
1857             self.assertEqual(ll[idx], val)
1858             self.assertEqual(ll._cache, list(cache))
1859
1860         ll = LazyList(range(10))
1861         test(ll, 0, 0, range(1))
1862         test(ll, 5, 5, range(6))
1863         test(ll, -3, 7, range(10))
1864
1865         ll = LazyList(range(10), reverse=True)
1866         test(ll, -1, 0, range(1))
1867         test(ll, 3, 6, range(10))
1868
1869         ll = LazyList(itertools.count())
1870         test(ll, 10, 10, range(11))
1871         ll = reversed(ll)
1872         test(ll, -15, 14, range(15))
1873
1874     def test_format_bytes(self):
1875         self.assertEqual(format_bytes(0), '0.00B')
1876         self.assertEqual(format_bytes(1000), '1000.00B')
1877         self.assertEqual(format_bytes(1024), '1.00KiB')
1878         self.assertEqual(format_bytes(1024**2), '1.00MiB')
1879         self.assertEqual(format_bytes(1024**3), '1.00GiB')
1880         self.assertEqual(format_bytes(1024**4), '1.00TiB')
1881         self.assertEqual(format_bytes(1024**5), '1.00PiB')
1882         self.assertEqual(format_bytes(1024**6), '1.00EiB')
1883         self.assertEqual(format_bytes(1024**7), '1.00ZiB')
1884         self.assertEqual(format_bytes(1024**8), '1.00YiB')
1885         self.assertEqual(format_bytes(1024**9), '1024.00YiB')
1886
1887     def test_hide_login_info(self):
1888         self.assertEqual(Config.hide_login_info(['-u', 'foo', '-p', 'bar']),
1889                          ['-u', 'PRIVATE', '-p', 'PRIVATE'])
1890         self.assertEqual(Config.hide_login_info(['-u']), ['-u'])
1891         self.assertEqual(Config.hide_login_info(['-u', 'foo', '-u', 'bar']),
1892                          ['-u', 'PRIVATE', '-u', 'PRIVATE'])
1893         self.assertEqual(Config.hide_login_info(['--username=foo']),
1894                          ['--username=PRIVATE'])
1895
1896     def test_locked_file(self):
1897         TEXT = 'test_locked_file\n'
1898         FILE = 'test_locked_file.ytdl'
1899         MODES = 'war'  # Order is important
1900
1901         try:
1902             for lock_mode in MODES:
1903                 with locked_file(FILE, lock_mode, False) as f:
1904                     if lock_mode == 'r':
1905                         self.assertEqual(f.read(), TEXT * 2, 'Wrong file content')
1906                     else:
1907                         f.write(TEXT)
1908                     for test_mode in MODES:
1909                         testing_write = test_mode != 'r'
1910                         try:
1911                             with locked_file(FILE, test_mode, False):
1912                                 pass
1913                         except (BlockingIOError, PermissionError):
1914                             if not testing_write:  # FIXME
1915                                 print(f'Known issue: Exclusive lock ({lock_mode}) blocks read access ({test_mode})')
1916                                 continue
1917                             self.assertTrue(testing_write, f'{test_mode} is blocked by {lock_mode}')
1918                         else:
1919                             self.assertFalse(testing_write, f'{test_mode} is not blocked by {lock_mode}')
1920         finally:
1921             with contextlib.suppress(OSError):
1922                 os.remove(FILE)
1923
1924     def test_determine_file_encoding(self):
1925         self.assertEqual(determine_file_encoding(b''), (None, 0))
1926         self.assertEqual(determine_file_encoding(b'--verbose -x --audio-format mkv\n'), (None, 0))
1927
1928         self.assertEqual(determine_file_encoding(b'\xef\xbb\xbf'), ('utf-8', 3))
1929         self.assertEqual(determine_file_encoding(b'\x00\x00\xfe\xff'), ('utf-32-be', 4))
1930         self.assertEqual(determine_file_encoding(b'\xff\xfe'), ('utf-16-le', 2))
1931
1932         self.assertEqual(determine_file_encoding(b'\xff\xfe# coding: utf-8\n--verbose'), ('utf-16-le', 2))
1933
1934         self.assertEqual(determine_file_encoding(b'# coding: utf-8\n--verbose'), ('utf-8', 0))
1935         self.assertEqual(determine_file_encoding(b'# coding: someencodinghere-12345\n--verbose'), ('someencodinghere-12345', 0))
1936
1937         self.assertEqual(determine_file_encoding(b'#coding:utf-8\n--verbose'), ('utf-8', 0))
1938         self.assertEqual(determine_file_encoding(b'#  coding:   utf-8   \r\n--verbose'), ('utf-8', 0))
1939
1940         self.assertEqual(determine_file_encoding('# coding: utf-32-be'.encode('utf-32-be')), ('utf-32-be', 0))
1941         self.assertEqual(determine_file_encoding('# coding: utf-16-le'.encode('utf-16-le')), ('utf-16-le', 0))
1942
1943     def test_get_compatible_ext(self):
1944         self.assertEqual(get_compatible_ext(
1945             vcodecs=[None], acodecs=[None, None], vexts=['mp4'], aexts=['m4a', 'm4a']), 'mkv')
1946         self.assertEqual(get_compatible_ext(
1947             vcodecs=[None], acodecs=[None], vexts=['flv'], aexts=['flv']), 'flv')
1948
1949         self.assertEqual(get_compatible_ext(
1950             vcodecs=[None], acodecs=[None], vexts=['mp4'], aexts=['m4a']), 'mp4')
1951         self.assertEqual(get_compatible_ext(
1952             vcodecs=[None], acodecs=[None], vexts=['mp4'], aexts=['webm']), 'mkv')
1953         self.assertEqual(get_compatible_ext(
1954             vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['m4a']), 'mkv')
1955         self.assertEqual(get_compatible_ext(
1956             vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['webm']), 'webm')
1957         self.assertEqual(get_compatible_ext(
1958             vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['weba']), 'webm')
1959
1960         self.assertEqual(get_compatible_ext(
1961             vcodecs=['h264'], acodecs=['mp4a'], vexts=['mov'], aexts=['m4a']), 'mp4')
1962         self.assertEqual(get_compatible_ext(
1963             vcodecs=['av01.0.12M.08'], acodecs=['opus'], vexts=['mp4'], aexts=['webm']), 'webm')
1964
1965         self.assertEqual(get_compatible_ext(
1966             vcodecs=['vp9'], acodecs=['opus'], vexts=['webm'], aexts=['webm'], preferences=['flv', 'mp4']), 'mp4')
1967         self.assertEqual(get_compatible_ext(
1968             vcodecs=['av1'], acodecs=['mp4a'], vexts=['webm'], aexts=['m4a'], preferences=('webm', 'mkv')), 'mkv')
1969
1970     def test_traverse_obj(self):
1971         _TEST_DATA = {
1972             100: 100,
1973             1.2: 1.2,
1974             'str': 'str',
1975             'None': None,
1976             '...': ...,
1977             'urls': [
1978                 {'index': 0, 'url': 'https://www.example.com/0'},
1979                 {'index': 1, 'url': 'https://www.example.com/1'},
1980             ],
1981             'data': (
1982                 {'index': 2},
1983                 {'index': 3},
1984             ),
1985             'dict': {},
1986         }
1987
1988         # Test base functionality
1989         self.assertEqual(traverse_obj(_TEST_DATA, ('str',)), 'str',
1990                          msg='allow tuple path')
1991         self.assertEqual(traverse_obj(_TEST_DATA, ['str']), 'str',
1992                          msg='allow list path')
1993         self.assertEqual(traverse_obj(_TEST_DATA, (value for value in ("str",))), 'str',
1994                          msg='allow iterable path')
1995         self.assertEqual(traverse_obj(_TEST_DATA, 'str'), 'str',
1996                          msg='single items should be treated as a path')
1997         self.assertEqual(traverse_obj(_TEST_DATA, None), _TEST_DATA)
1998         self.assertEqual(traverse_obj(_TEST_DATA, 100), 100)
1999         self.assertEqual(traverse_obj(_TEST_DATA, 1.2), 1.2)
2000
2001         # Test Ellipsis behavior
2002         self.assertCountEqual(traverse_obj(_TEST_DATA, ...),
2003                               (item for item in _TEST_DATA.values() if item not in (None, {})),
2004                               msg='`...` should give all non discarded values')
2005         self.assertCountEqual(traverse_obj(_TEST_DATA, ('urls', 0, ...)), _TEST_DATA['urls'][0].values(),
2006                               msg='`...` selection for dicts should select all values')
2007         self.assertEqual(traverse_obj(_TEST_DATA, (..., ..., 'url')),
2008                          ['https://www.example.com/0', 'https://www.example.com/1'],
2009                          msg='nested `...` queries should work')
2010         self.assertCountEqual(traverse_obj(_TEST_DATA, (..., ..., 'index')), range(4),
2011                               msg='`...` query result should be flattened')
2012
2013         # Test function as key
2014         self.assertEqual(traverse_obj(_TEST_DATA, lambda x, y: x == 'urls' and isinstance(y, list)),
2015                          [_TEST_DATA['urls']],
2016                          msg='function as query key should perform a filter based on (key, value)')
2017         self.assertCountEqual(traverse_obj(_TEST_DATA, lambda _, x: isinstance(x[0], str)), {'str'},
2018                               msg='exceptions in the query function should be catched')
2019         if __debug__:
2020             with self.assertRaises(Exception, msg='Wrong function signature should raise in debug'):
2021                 traverse_obj(_TEST_DATA, lambda a: ...)
2022             with self.assertRaises(Exception, msg='Wrong function signature should raise in debug'):
2023                 traverse_obj(_TEST_DATA, lambda a, b, c: ...)
2024
2025         # Test set as key (transformation/type, like `expected_type`)
2026         self.assertEqual(traverse_obj(_TEST_DATA, (..., {str.upper}, )), ['STR'],
2027                          msg='Function in set should be a transformation')
2028         self.assertEqual(traverse_obj(_TEST_DATA, (..., {str})), ['str'],
2029                          msg='Type in set should be a type filter')
2030         self.assertEqual(traverse_obj(_TEST_DATA, {dict}), _TEST_DATA,
2031                          msg='A single set should be wrapped into a path')
2032         self.assertEqual(traverse_obj(_TEST_DATA, (..., {str.upper})), ['STR'],
2033                          msg='Transformation function should not raise')
2034         self.assertEqual(traverse_obj(_TEST_DATA, (..., {str_or_none})),
2035                          [item for item in map(str_or_none, _TEST_DATA.values()) if item is not None],
2036                          msg='Function in set should be a transformation')
2037         if __debug__:
2038             with self.assertRaises(Exception, msg='Sets with length != 1 should raise in debug'):
2039                 traverse_obj(_TEST_DATA, set())
2040             with self.assertRaises(Exception, msg='Sets with length != 1 should raise in debug'):
2041                 traverse_obj(_TEST_DATA, {str.upper, str})
2042
2043         # Test alternative paths
2044         self.assertEqual(traverse_obj(_TEST_DATA, 'fail', 'str'), 'str',
2045                          msg='multiple `paths` should be treated as alternative paths')
2046         self.assertEqual(traverse_obj(_TEST_DATA, 'str', 100), 'str',
2047                          msg='alternatives should exit early')
2048         self.assertEqual(traverse_obj(_TEST_DATA, 'fail', 'fail'), None,
2049                          msg='alternatives should return `default` if exhausted')
2050         self.assertEqual(traverse_obj(_TEST_DATA, (..., 'fail'), 100), 100,
2051                          msg='alternatives should track their own branching return')
2052         self.assertEqual(traverse_obj(_TEST_DATA, ('dict', ...), ('data', ...)), list(_TEST_DATA['data']),
2053                          msg='alternatives on empty objects should search further')
2054
2055         # Test branch and path nesting
2056         self.assertEqual(traverse_obj(_TEST_DATA, ('urls', (3, 0), 'url')), ['https://www.example.com/0'],
2057                          msg='tuple as key should be treated as branches')
2058         self.assertEqual(traverse_obj(_TEST_DATA, ('urls', [3, 0], 'url')), ['https://www.example.com/0'],
2059                          msg='list as key should be treated as branches')
2060         self.assertEqual(traverse_obj(_TEST_DATA, ('urls', ((1, 'fail'), (0, 'url')))), ['https://www.example.com/0'],
2061                          msg='double nesting in path should be treated as paths')
2062         self.assertEqual(traverse_obj(['0', [1, 2]], [(0, 1), 0]), [1],
2063                          msg='do not fail early on branching')
2064         self.assertCountEqual(traverse_obj(_TEST_DATA, ('urls', ((1, ('fail', 'url')), (0, 'url')))),
2065                               ['https://www.example.com/0', 'https://www.example.com/1'],
2066                               msg='tripple nesting in path should be treated as branches')
2067         self.assertEqual(traverse_obj(_TEST_DATA, ('urls', ('fail', (..., 'url')))),
2068                          ['https://www.example.com/0', 'https://www.example.com/1'],
2069                          msg='ellipsis as branch path start gets flattened')
2070
2071         # Test dictionary as key
2072         self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2}), {0: 100, 1: 1.2},
2073                          msg='dict key should result in a dict with the same keys')
2074         self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', 0, 'url')}),
2075                          {0: 'https://www.example.com/0'},
2076                          msg='dict key should allow paths')
2077         self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', (3, 0), 'url')}),
2078                          {0: ['https://www.example.com/0']},
2079                          msg='tuple in dict path should be treated as branches')
2080         self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', ((1, 'fail'), (0, 'url')))}),
2081                          {0: ['https://www.example.com/0']},
2082                          msg='double nesting in dict path should be treated as paths')
2083         self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', ((1, ('fail', 'url')), (0, 'url')))}),
2084                          {0: ['https://www.example.com/1', 'https://www.example.com/0']},
2085                          msg='tripple nesting in dict path should be treated as branches')
2086         self.assertEqual(traverse_obj(_TEST_DATA, {0: 'fail'}), {},
2087                          msg='remove `None` values when top level dict key fails')
2088         self.assertEqual(traverse_obj(_TEST_DATA, {0: 'fail'}, default=...), {0: ...},
2089                          msg='use `default` if key fails and `default`')
2090         self.assertEqual(traverse_obj(_TEST_DATA, {0: 'dict'}), {},
2091                          msg='remove empty values when dict key')
2092         self.assertEqual(traverse_obj(_TEST_DATA, {0: 'dict'}, default=...), {0: ...},
2093                          msg='use `default` when dict key and `default`')
2094         self.assertEqual(traverse_obj(_TEST_DATA, {0: {0: 'fail'}}), {},
2095                          msg='remove empty values when nested dict key fails')
2096         self.assertEqual(traverse_obj(None, {0: 'fail'}), {},
2097                          msg='default to dict if pruned')
2098         self.assertEqual(traverse_obj(None, {0: 'fail'}, default=...), {0: ...},
2099                          msg='default to dict if pruned and default is given')
2100         self.assertEqual(traverse_obj(_TEST_DATA, {0: {0: 'fail'}}, default=...), {0: {0: ...}},
2101                          msg='use nested `default` when nested dict key fails and `default`')
2102         self.assertEqual(traverse_obj(_TEST_DATA, {0: ('dict', ...)}), {},
2103                          msg='remove key if branch in dict key not successful')
2104
2105         # Testing default parameter behavior
2106         _DEFAULT_DATA = {'None': None, 'int': 0, 'list': []}
2107         self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail'), None,
2108                          msg='default value should be `None`')
2109         self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail', 'fail', default=...), ...,
2110                          msg='chained fails should result in default')
2111         self.assertEqual(traverse_obj(_DEFAULT_DATA, 'None', 'int'), 0,
2112                          msg='should not short cirquit on `None`')
2113         self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail', default=1), 1,
2114                          msg='invalid dict key should result in `default`')
2115         self.assertEqual(traverse_obj(_DEFAULT_DATA, 'None', default=1), 1,
2116                          msg='`None` is a deliberate sentinel and should become `default`')
2117         self.assertEqual(traverse_obj(_DEFAULT_DATA, ('list', 10)), None,
2118                          msg='`IndexError` should result in `default`')
2119         self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail'), default=1), 1,
2120                          msg='if branched but not successful return `default` if defined, not `[]`')
2121         self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail'), default=None), None,
2122                          msg='if branched but not successful return `default` even if `default` is `None`')
2123         self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail')), [],
2124                          msg='if branched but not successful return `[]`, not `default`')
2125         self.assertEqual(traverse_obj(_DEFAULT_DATA, ('list', ...)), [],
2126                          msg='if branched but object is empty return `[]`, not `default`')
2127         self.assertEqual(traverse_obj(None, ...), [],
2128                          msg='if branched but object is `None` return `[]`, not `default`')
2129         self.assertEqual(traverse_obj({0: None}, (0, ...)), [],
2130                          msg='if branched but state is `None` return `[]`, not `default`')
2131
2132         branching_paths = [
2133             ('fail', ...),
2134             (..., 'fail'),
2135             100 * ('fail',) + (...,),
2136             (...,) + 100 * ('fail',),
2137         ]
2138         for branching_path in branching_paths:
2139             self.assertEqual(traverse_obj({}, branching_path), [],
2140                              msg='if branched but state is `None`, return `[]` (not `default`)')
2141             self.assertEqual(traverse_obj({}, 'fail', branching_path), [],
2142                              msg='if branching in last alternative and previous did not match, return `[]` (not `default`)')
2143             self.assertEqual(traverse_obj({0: 'x'}, 0, branching_path), 'x',
2144                              msg='if branching in last alternative and previous did match, return single value')
2145             self.assertEqual(traverse_obj({0: 'x'}, branching_path, 0), 'x',
2146                              msg='if branching in first alternative and non-branching path does match, return single value')
2147             self.assertEqual(traverse_obj({}, branching_path, 'fail'), None,
2148                              msg='if branching in first alternative and non-branching path does not match, return `default`')
2149
2150         # Testing expected_type behavior
2151         _EXPECTED_TYPE_DATA = {'str': 'str', 'int': 0}
2152         self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=str),
2153                          'str', msg='accept matching `expected_type` type')
2154         self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=int),
2155                          None, msg='reject non matching `expected_type` type')
2156         self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'int', expected_type=lambda x: str(x)),
2157                          '0', msg='transform type using type function')
2158         self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=lambda _: 1 / 0),
2159                          None, msg='wrap expected_type fuction in try_call')
2160         self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, ..., expected_type=str),
2161                          ['str'], msg='eliminate items that expected_type fails on')
2162         self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2}, expected_type=int),
2163                          {0: 100}, msg='type as expected_type should filter dict values')
2164         self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2, 2: 'None'}, expected_type=str_or_none),
2165                          {0: '100', 1: '1.2'}, msg='function as expected_type should transform dict values')
2166         self.assertEqual(traverse_obj(_TEST_DATA, ({0: 1.2}, 0, {int_or_none}), expected_type=int),
2167                          1, msg='expected_type should not filter non final dict values')
2168         self.assertEqual(traverse_obj(_TEST_DATA, {0: {0: 100, 1: 'str'}}, expected_type=int),
2169                          {0: {0: 100}}, msg='expected_type should transform deep dict values')
2170         self.assertEqual(traverse_obj(_TEST_DATA, [({0: '...'}, {0: '...'})], expected_type=type(...)),
2171                          [{0: ...}, {0: ...}], msg='expected_type should transform branched dict values')
2172         self.assertEqual(traverse_obj({1: {3: 4}}, [(1, 2), 3], expected_type=int),
2173                          [4], msg='expected_type regression for type matching in tuple branching')
2174         self.assertEqual(traverse_obj(_TEST_DATA, ['data', ...], expected_type=int),
2175                          [], msg='expected_type regression for type matching in dict result')
2176
2177         # Test get_all behavior
2178         _GET_ALL_DATA = {'key': [0, 1, 2]}
2179         self.assertEqual(traverse_obj(_GET_ALL_DATA, ('key', ...), get_all=False), 0,
2180                          msg='if not `get_all`, return only first matching value')
2181         self.assertEqual(traverse_obj(_GET_ALL_DATA, ..., get_all=False), [0, 1, 2],
2182                          msg='do not overflatten if not `get_all`')
2183
2184         # Test casesense behavior
2185         _CASESENSE_DATA = {
2186             'KeY': 'value0',
2187             0: {
2188                 'KeY': 'value1',
2189                 0: {'KeY': 'value2'},
2190             },
2191         }
2192         self.assertEqual(traverse_obj(_CASESENSE_DATA, 'key'), None,
2193                          msg='dict keys should be case sensitive unless `casesense`')
2194         self.assertEqual(traverse_obj(_CASESENSE_DATA, 'keY',
2195                                       casesense=False), 'value0',
2196                          msg='allow non matching key case if `casesense`')
2197         self.assertEqual(traverse_obj(_CASESENSE_DATA, (0, ('keY',)),
2198                                       casesense=False), ['value1'],
2199                          msg='allow non matching key case in branch if `casesense`')
2200         self.assertEqual(traverse_obj(_CASESENSE_DATA, (0, ((0, 'keY'),)),
2201                                       casesense=False), ['value2'],
2202                          msg='allow non matching key case in branch path if `casesense`')
2203
2204         # Test traverse_string behavior
2205         _TRAVERSE_STRING_DATA = {'str': 'str', 1.2: 1.2}
2206         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', 0)), None,
2207                          msg='do not traverse into string if not `traverse_string`')
2208         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', 0),
2209                                       traverse_string=True), 's',
2210                          msg='traverse into string if `traverse_string`')
2211         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, (1.2, 1),
2212                                       traverse_string=True), '.',
2213                          msg='traverse into converted data if `traverse_string`')
2214         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', ...),
2215                                       traverse_string=True), 'str',
2216                          msg='`...` should result in string (same value) if `traverse_string`')
2217         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', slice(0, None, 2)),
2218                                       traverse_string=True), 'sr',
2219                          msg='`slice` should result in string if `traverse_string`')
2220         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', lambda i, v: i or v == "s"),
2221                                       traverse_string=True), 'str',
2222                          msg='function should result in string if `traverse_string`')
2223         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', (0, 2)),
2224                                       traverse_string=True), ['s', 'r'],
2225                          msg='branching should result in list if `traverse_string`')
2226
2227         # Test is_user_input behavior
2228         _IS_USER_INPUT_DATA = {'range8': list(range(8))}
2229         self.assertEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', '3'),
2230                                       is_user_input=True), 3,
2231                          msg='allow for string indexing if `is_user_input`')
2232         self.assertCountEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', '3:'),
2233                                            is_user_input=True), tuple(range(8))[3:],
2234                               msg='allow for string slice if `is_user_input`')
2235         self.assertCountEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', ':4:2'),
2236                                            is_user_input=True), tuple(range(8))[:4:2],
2237                               msg='allow step in string slice if `is_user_input`')
2238         self.assertCountEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', ':'),
2239                                            is_user_input=True), range(8),
2240                               msg='`:` should be treated as `...` if `is_user_input`')
2241         with self.assertRaises(TypeError, msg='too many params should result in error'):
2242             traverse_obj(_IS_USER_INPUT_DATA, ('range8', ':::'), is_user_input=True)
2243
2244         # Test re.Match as input obj
2245         mobj = re.fullmatch(r'0(12)(?P<group>3)(4)?', '0123')
2246         self.assertEqual(traverse_obj(mobj, ...), [x for x in mobj.groups() if x is not None],
2247                          msg='`...` on a `re.Match` should give its `groups()`')
2248         self.assertEqual(traverse_obj(mobj, lambda k, _: k in (0, 2)), ['0123', '3'],
2249                          msg='function on a `re.Match` should give groupno, value starting at 0')
2250         self.assertEqual(traverse_obj(mobj, 'group'), '3',
2251                          msg='str key on a `re.Match` should give group with that name')
2252         self.assertEqual(traverse_obj(mobj, 2), '3',
2253                          msg='int key on a `re.Match` should give group with that name')
2254         self.assertEqual(traverse_obj(mobj, 'gRoUp', casesense=False), '3',
2255                          msg='str key on a `re.Match` should respect casesense')
2256         self.assertEqual(traverse_obj(mobj, 'fail'), None,
2257                          msg='failing str key on a `re.Match` should return `default`')
2258         self.assertEqual(traverse_obj(mobj, 'gRoUpS', casesense=False), None,
2259                          msg='failing str key on a `re.Match` should return `default`')
2260         self.assertEqual(traverse_obj(mobj, 8), None,
2261                          msg='failing int key on a `re.Match` should return `default`')
2262         self.assertEqual(traverse_obj(mobj, lambda k, _: k in (0, 'group')), ['0123', '3'],
2263                          msg='function on a `re.Match` should give group name as well')
2264
2265
2266 if __name__ == '__main__':
2267     unittest.main()