test/test_utils.py

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