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