_ Git - youtube-dl/blob - test/test_utils.py

   1 #!/usr/bin/env python
   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 json
  16 import xml.etree.ElementTree
  17
  18 from youtube_dl.utils import (
  19     age_restricted,
  20     args_to_str,
  21     encode_base_n,
  22     clean_html,
  23     date_from_str,
  24     DateRange,
  25     detect_exe_version,
  26     determine_ext,
  27     dict_get,
  28     encode_compat_str,
  29     encodeFilename,
  30     escape_rfc3986,
  31     escape_url,
  32     extract_attributes,
  33     ExtractorError,
  34     find_xpath_attr,
  35     fix_xml_ampersands,
  36     get_element_by_class,
  37     InAdvancePagedList,
  38     intlist_to_bytes,
  39     is_html,
  40     js_to_json,
  41     limit_length,
  42     mimetype2ext,
  43     ohdave_rsa_encrypt,
  44     OnDemandPagedList,
  45     orderedSet,
  46     parse_age_limit,
  47     parse_duration,
  48     parse_filesize,
  49     parse_count,
  50     parse_iso8601,
  51     read_batch_urls,
  52     sanitize_filename,
  53     sanitize_path,
  54     prepend_extension,
  55     replace_extension,
  56     remove_start,
  57     remove_end,
  58     remove_quotes,
  59     shell_quote,
  60     smuggle_url,
  61     str_to_int,
  62     strip_jsonp,
  63     timeconvert,
  64     unescapeHTML,
  65     unified_strdate,
  66     unified_timestamp,
  67     unsmuggle_url,
  68     uppercase_escape,
  69     lowercase_escape,
  70     url_basename,
  71     urlencode_postdata,
  72     urshift,
  73     update_url_query,
  74     version_tuple,
  75     xpath_with_ns,
  76     xpath_element,
  77     xpath_text,
  78     xpath_attr,
  79     render_table,
  80     match_str,
  81     parse_dfxp_time_expr,
  82     dfxp2srt,
  83     cli_option,
  84     cli_valueless_option,
  85     cli_bool_option,
  86     parse_codecs,
  87 )
  88 from youtube_dl.compat import (
  89     compat_chr,
  90     compat_etree_fromstring,
  91     compat_urlparse,
  92     compat_parse_qs,
  93 )
  94
  95
  96 class TestUtil(unittest.TestCase):
  97     def test_timeconvert(self):
  98         self.assertTrue(timeconvert('') is None)
  99         self.assertTrue(timeconvert('bougrg') is None)
 100
 101     def test_sanitize_filename(self):
 102         self.assertEqual(sanitize_filename('abc'), 'abc')
 103         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
 104
 105         self.assertEqual(sanitize_filename('123'), '123')
 106
 107         self.assertEqual('abc_de', sanitize_filename('abc/de'))
 108         self.assertFalse('/' in sanitize_filename('abc/de///'))
 109
 110         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
 111         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
 112         self.assertEqual('yes no', sanitize_filename('yes? no'))
 113         self.assertEqual('this - that', sanitize_filename('this: that'))
 114
 115         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
 116         aumlaut = 'ä'
 117         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
 118         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
 119         self.assertEqual(sanitize_filename(tests), tests)
 120
 121         self.assertEqual(
 122             sanitize_filename('New World record at 0:12:34'),
 123             'New World record at 0_12_34')
 124
 125         self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
 126         self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
 127         self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
 128         self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
 129
 130         forbidden = '"\0\\/'
 131         for fc in forbidden:
 132             for fbc in forbidden:
 133                 self.assertTrue(fbc not in sanitize_filename(fc))
 134
 135     def test_sanitize_filename_restricted(self):
 136         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
 137         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
 138
 139         self.assertEqual(sanitize_filename('123', restricted=True), '123')
 140
 141         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
 142         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
 143
 144         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
 145         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
 146         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
 147         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
 148
 149         tests = 'aäb\u4e2d\u56fd\u7684c'
 150         self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
 151         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
 152
 153         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
 154         for fc in forbidden:
 155             for fbc in forbidden:
 156                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
 157
 158         # Handle a common case more neatly
 159         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
 160         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
 161         # .. but make sure the file name is never empty
 162         self.assertTrue(sanitize_filename('-', restricted=True) != '')
 163         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 164
 165         self.assertEqual(sanitize_filename(
 166             'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
 167             'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYPssaaaaaaaeceeeeiiiionooooooooeuuuuuypy')
 168
 169     def test_sanitize_ids(self):
 170         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 171         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 172         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 173
 174     def test_sanitize_path(self):
 175         if sys.platform != 'win32':
 176             return
 177
 178         self.assertEqual(sanitize_path('abc'), 'abc')
 179         self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
 180         self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
 181         self.assertEqual(sanitize_path('abc|def'), 'abc#def')
 182         self.assertEqual(sanitize_path('<>:"|?*'), '#######')
 183         self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
 184         self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
 185
 186         self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
 187         self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
 188
 189         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 190         self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
 191         self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
 192         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 193
 194         self.assertEqual(
 195             sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
 196             'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
 197
 198         self.assertEqual(
 199             sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
 200             'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
 201         self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
 202         self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
 203         self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
 204
 205         self.assertEqual(sanitize_path('../abc'), '..\\abc')
 206         self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
 207         self.assertEqual(sanitize_path('./abc'), 'abc')
 208         self.assertEqual(sanitize_path('./../abc'), '..\\abc')
 209
 210     def test_prepend_extension(self):
 211         self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
 212         self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
 213         self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 214         self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
 215         self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
 216         self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
 217
 218     def test_replace_extension(self):
 219         self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
 220         self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
 221         self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 222         self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
 223         self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
 224         self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
 225
 226     def test_remove_start(self):
 227         self.assertEqual(remove_start(None, 'A - '), None)
 228         self.assertEqual(remove_start('A - B', 'A - '), 'B')
 229         self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
 230
 231     def test_remove_end(self):
 232         self.assertEqual(remove_end(None, ' - B'), None)
 233         self.assertEqual(remove_end('A - B', ' - B'), 'A')
 234         self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
 235
 236     def test_remove_quotes(self):
 237         self.assertEqual(remove_quotes(None), None)
 238         self.assertEqual(remove_quotes('"'), '"')
 239         self.assertEqual(remove_quotes("'"), "'")
 240         self.assertEqual(remove_quotes(';'), ';')
 241         self.assertEqual(remove_quotes('";'), '";')
 242         self.assertEqual(remove_quotes('""'), '')
 243         self.assertEqual(remove_quotes('";"'), ';')
 244
 245     def test_ordered_set(self):
 246         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
 247         self.assertEqual(orderedSet([]), [])
 248         self.assertEqual(orderedSet([1]), [1])
 249         # keep the list ordered
 250         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
 251
 252     def test_unescape_html(self):
 253         self.assertEqual(unescapeHTML('%20;'), '%20;')
 254         self.assertEqual(unescapeHTML('&#x2F;'), '/')
 255         self.assertEqual(unescapeHTML('&#47;'), '/')
 256         self.assertEqual(unescapeHTML('&eacute;'), 'é')
 257         self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
 258         # HTML5 entities
 259         self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')
 260
 261     def test_date_from_str(self):
 262         self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
 263         self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
 264         self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
 265         self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
 266         self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
 267
 268     def test_daterange(self):
 269         _20century = DateRange("19000101", "20000101")
 270         self.assertFalse("17890714" in _20century)
 271         _ac = DateRange("00010101")
 272         self.assertTrue("19690721" in _ac)
 273         _firstmilenium = DateRange(end="10000101")
 274         self.assertTrue("07110427" in _firstmilenium)
 275
 276     def test_unified_dates(self):
 277         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
 278         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
 279         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
 280         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
 281         self.assertEqual(unified_strdate('1968 12 10'), '19681210')
 282         self.assertEqual(unified_strdate('1968-12-10'), '19681210')
 283         self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
 284         self.assertEqual(
 285             unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
 286             '20141126')
 287         self.assertEqual(
 288             unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
 289             '20150202')
 290         self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
 291         self.assertEqual(unified_strdate('25-09-2014'), '20140925')
 292         self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
 293         self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
 294
 295     def test_unified_timestamps(self):
 296         self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
 297         self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
 298         self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
 299         self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
 300         self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
 301         self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
 302         self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
 303         self.assertEqual(
 304             unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
 305             1417001400)
 306         self.assertEqual(
 307             unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
 308             1422902860)
 309         self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
 310         self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
 311         self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
 312         self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
 313         self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
 314
 315     def test_determine_ext(self):
 316         self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
 317         self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
 318         self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
 319         self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
 320         self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
 321
 322     def test_find_xpath_attr(self):
 323         testxml = '''<root>
 324             <node/>
 325             <node x="a"/>
 326             <node x="a" y="c" />
 327             <node x="b" y="d" />
 328             <node x="" />
 329         </root>'''
 330         doc = compat_etree_fromstring(testxml)
 331
 332         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
 333         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
 334         self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
 335         self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
 336         self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
 337         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
 338         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
 339         self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
 340         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
 341         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
 342         self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
 343
 344     def test_xpath_with_ns(self):
 345         testxml = '''<root xmlns:media="http://example.com/">
 346             <media:song>
 347                 <media:author>The Author</media:author>
 348                 <url>http://server.com/download.mp3</url>
 349             </media:song>
 350         </root>'''
 351         doc = compat_etree_fromstring(testxml)
 352         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
 353         self.assertTrue(find('media:song') is not None)
 354         self.assertEqual(find('media:song/media:author').text, 'The Author')
 355         self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
 356
 357     def test_xpath_element(self):
 358         doc = xml.etree.ElementTree.Element('root')
 359         div = xml.etree.ElementTree.SubElement(doc, 'div')
 360         p = xml.etree.ElementTree.SubElement(div, 'p')
 361         p.text = 'Foo'
 362         self.assertEqual(xpath_element(doc, 'div/p'), p)
 363         self.assertEqual(xpath_element(doc, ['div/p']), p)
 364         self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
 365         self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
 366         self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
 367         self.assertTrue(xpath_element(doc, 'div/bar') is None)
 368         self.assertTrue(xpath_element(doc, ['div/bar']) is None)
 369         self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
 370         self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
 371         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
 372         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
 373
 374     def test_xpath_text(self):
 375         testxml = '''<root>
 376             <div>
 377                 <p>Foo</p>
 378             </div>
 379         </root>'''
 380         doc = compat_etree_fromstring(testxml)
 381         self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
 382         self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
 383         self.assertTrue(xpath_text(doc, 'div/bar') is None)
 384         self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
 385
 386     def test_xpath_attr(self):
 387         testxml = '''<root>
 388             <div>
 389                 <p x="a">Foo</p>
 390             </div>
 391         </root>'''
 392         doc = compat_etree_fromstring(testxml)
 393         self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
 394         self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
 395         self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
 396         self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
 397         self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
 398         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
 399         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
 400
 401     def test_smuggle_url(self):
 402         data = {"ö": "ö", "abc": [3]}
 403         url = 'https://foo.bar/baz?x=y#a'
 404         smug_url = smuggle_url(url, data)
 405         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
 406         self.assertEqual(url, unsmug_url)
 407         self.assertEqual(data, unsmug_data)
 408
 409         res_url, res_data = unsmuggle_url(url)
 410         self.assertEqual(res_url, url)
 411         self.assertEqual(res_data, None)
 412
 413         smug_url = smuggle_url(url, {'a': 'b'})
 414         smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
 415         res_url, res_data = unsmuggle_url(smug_smug_url)
 416         self.assertEqual(res_url, url)
 417         self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
 418
 419     def test_shell_quote(self):
 420         args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
 421         self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
 422
 423     def test_str_to_int(self):
 424         self.assertEqual(str_to_int('123,456'), 123456)
 425         self.assertEqual(str_to_int('123.456'), 123456)
 426
 427     def test_url_basename(self):
 428         self.assertEqual(url_basename('http://foo.de/'), '')
 429         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
 430         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
 431         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
 432         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
 433         self.assertEqual(
 434             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
 435             'trailer.mp4')
 436
 437     def test_parse_age_limit(self):
 438         self.assertEqual(parse_age_limit(None), None)
 439         self.assertEqual(parse_age_limit(False), None)
 440         self.assertEqual(parse_age_limit('invalid'), None)
 441         self.assertEqual(parse_age_limit(0), 0)
 442         self.assertEqual(parse_age_limit(18), 18)
 443         self.assertEqual(parse_age_limit(21), 21)
 444         self.assertEqual(parse_age_limit(22), None)
 445         self.assertEqual(parse_age_limit('18'), 18)
 446         self.assertEqual(parse_age_limit('18+'), 18)
 447         self.assertEqual(parse_age_limit('PG-13'), 13)
 448         self.assertEqual(parse_age_limit('TV-14'), 14)
 449         self.assertEqual(parse_age_limit('TV-MA'), 17)
 450
 451     def test_parse_duration(self):
 452         self.assertEqual(parse_duration(None), None)
 453         self.assertEqual(parse_duration(False), None)
 454         self.assertEqual(parse_duration('invalid'), None)
 455         self.assertEqual(parse_duration('1'), 1)
 456         self.assertEqual(parse_duration('1337:12'), 80232)
 457         self.assertEqual(parse_duration('9:12:43'), 33163)
 458         self.assertEqual(parse_duration('12:00'), 720)
 459         self.assertEqual(parse_duration('00:01:01'), 61)
 460         self.assertEqual(parse_duration('x:y'), None)
 461         self.assertEqual(parse_duration('3h11m53s'), 11513)
 462         self.assertEqual(parse_duration('3h 11m 53s'), 11513)
 463         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
 464         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
 465         self.assertEqual(parse_duration('62m45s'), 3765)
 466         self.assertEqual(parse_duration('6m59s'), 419)
 467         self.assertEqual(parse_duration('49s'), 49)
 468         self.assertEqual(parse_duration('0h0m0s'), 0)
 469         self.assertEqual(parse_duration('0m0s'), 0)
 470         self.assertEqual(parse_duration('0s'), 0)
 471         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
 472         self.assertEqual(parse_duration('T30M38S'), 1838)
 473         self.assertEqual(parse_duration('5 s'), 5)
 474         self.assertEqual(parse_duration('3 min'), 180)
 475         self.assertEqual(parse_duration('2.5 hours'), 9000)
 476         self.assertEqual(parse_duration('02:03:04'), 7384)
 477         self.assertEqual(parse_duration('01:02:03:04'), 93784)
 478         self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
 479         self.assertEqual(parse_duration('87 Min.'), 5220)
 480         self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
 481
 482     def test_fix_xml_ampersands(self):
 483         self.assertEqual(
 484             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 485         self.assertEqual(
 486             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 487             '"&amp;x=y&amp;wrong;&amp;z=a')
 488         self.assertEqual(
 489             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 490             '&amp;&apos;&gt;&lt;&quot;')
 491         self.assertEqual(
 492             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 493         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 494
 495     def test_paged_list(self):
 496         def testPL(size, pagesize, sliceargs, expected):
 497             def get_page(pagenum):
 498                 firstid = pagenum * pagesize
 499                 upto = min(size, pagenum * pagesize + pagesize)
 500                 for i in range(firstid, upto):
 501                     yield i
 502
 503             pl = OnDemandPagedList(get_page, pagesize)
 504             got = pl.getslice(*sliceargs)
 505             self.assertEqual(got, expected)
 506
 507             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
 508             got = iapl.getslice(*sliceargs)
 509             self.assertEqual(got, expected)
 510
 511         testPL(5, 2, (), [0, 1, 2, 3, 4])
 512         testPL(5, 2, (1,), [1, 2, 3, 4])
 513         testPL(5, 2, (2,), [2, 3, 4])
 514         testPL(5, 2, (4,), [4])
 515         testPL(5, 2, (0, 3), [0, 1, 2])
 516         testPL(5, 2, (1, 4), [1, 2, 3])
 517         testPL(5, 2, (2, 99), [2, 3, 4])
 518         testPL(5, 2, (20, 99), [])
 519
 520     def test_read_batch_urls(self):
 521         f = io.StringIO('''\xef\xbb\xbf foo
 522             bar\r
 523             baz
 524             # More after this line\r
 525             ; or after this
 526             bam''')
 527         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
 528
 529     def test_urlencode_postdata(self):
 530         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
 531         self.assertTrue(isinstance(data, bytes))
 532
 533     def test_update_url_query(self):
 534         def query_dict(url):
 535             return compat_parse_qs(compat_urlparse.urlparse(url).query)
 536         self.assertEqual(query_dict(update_url_query(
 537             'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
 538             query_dict('http://example.com/path?quality=HD&format=mp4'))
 539         self.assertEqual(query_dict(update_url_query(
 540             'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
 541             query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
 542         self.assertEqual(query_dict(update_url_query(
 543             'http://example.com/path', {'fields': 'id,formats,subtitles'})),
 544             query_dict('http://example.com/path?fields=id,formats,subtitles'))
 545         self.assertEqual(query_dict(update_url_query(
 546             'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
 547             query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
 548         self.assertEqual(query_dict(update_url_query(
 549             'http://example.com/path?manifest=f4m', {'manifest': []})),
 550             query_dict('http://example.com/path'))
 551         self.assertEqual(query_dict(update_url_query(
 552             'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
 553             query_dict('http://example.com/path?system=LINUX'))
 554         self.assertEqual(query_dict(update_url_query(
 555             'http://example.com/path', {'fields': b'id,formats,subtitles'})),
 556             query_dict('http://example.com/path?fields=id,formats,subtitles'))
 557         self.assertEqual(query_dict(update_url_query(
 558             'http://example.com/path', {'width': 1080, 'height': 720})),
 559             query_dict('http://example.com/path?width=1080&height=720'))
 560         self.assertEqual(query_dict(update_url_query(
 561             'http://example.com/path', {'bitrate': 5020.43})),
 562             query_dict('http://example.com/path?bitrate=5020.43'))
 563         self.assertEqual(query_dict(update_url_query(
 564             'http://example.com/path', {'test': '第二行тест'})),
 565             query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
 566
 567     def test_dict_get(self):
 568         FALSE_VALUES = {
 569             'none': None,
 570             'false': False,
 571             'zero': 0,
 572             'empty_string': '',
 573             'empty_list': [],
 574         }
 575         d = FALSE_VALUES.copy()
 576         d['a'] = 42
 577         self.assertEqual(dict_get(d, 'a'), 42)
 578         self.assertEqual(dict_get(d, 'b'), None)
 579         self.assertEqual(dict_get(d, 'b', 42), 42)
 580         self.assertEqual(dict_get(d, ('a', )), 42)
 581         self.assertEqual(dict_get(d, ('b', 'a', )), 42)
 582         self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
 583         self.assertEqual(dict_get(d, ('b', 'c', )), None)
 584         self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
 585         for key, false_value in FALSE_VALUES.items():
 586             self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
 587             self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
 588
 589     def test_encode_compat_str(self):
 590         self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
 591         self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
 592
 593     def test_parse_iso8601(self):
 594         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
 595         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
 596         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
 597         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
 598         self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
 599         self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
 600
 601     def test_strip_jsonp(self):
 602         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
 603         d = json.loads(stripped)
 604         self.assertEqual(d, [{"id": "532cb", "x": 3}])
 605
 606         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
 607         d = json.loads(stripped)
 608         self.assertEqual(d, {'STATUS': 'OK'})
 609
 610         stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
 611         d = json.loads(stripped)
 612         self.assertEqual(d, {'status': 'success'})
 613
 614     def test_uppercase_escape(self):
 615         self.assertEqual(uppercase_escape('aä'), 'aä')
 616         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
 617
 618     def test_lowercase_escape(self):
 619         self.assertEqual(lowercase_escape('aä'), 'aä')
 620         self.assertEqual(lowercase_escape('\\u0026'), '&')
 621
 622     def test_limit_length(self):
 623         self.assertEqual(limit_length(None, 12), None)
 624         self.assertEqual(limit_length('foo', 12), 'foo')
 625         self.assertTrue(
 626             limit_length('foo bar baz asd', 12).startswith('foo bar'))
 627         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
 628
 629     def test_mimetype2ext(self):
 630         self.assertEqual(mimetype2ext(None), None)
 631         self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
 632         self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
 633         self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
 634         self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
 635         self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
 636
 637     def test_parse_codecs(self):
 638         self.assertEqual(parse_codecs(''), {})
 639         self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
 640             'vcodec': 'avc1.77.30',
 641             'acodec': 'mp4a.40.2',
 642         })
 643         self.assertEqual(parse_codecs('mp4a.40.2'), {
 644             'vcodec': 'none',
 645             'acodec': 'mp4a.40.2',
 646         })
 647         self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
 648             'vcodec': 'avc1.42001e',
 649             'acodec': 'mp4a.40.5',
 650         })
 651         self.assertEqual(parse_codecs('avc3.640028'), {
 652             'vcodec': 'avc3.640028',
 653             'acodec': 'none',
 654         })
 655         self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
 656             'vcodec': 'h264',
 657             'acodec': 'aac',
 658         })
 659
 660     def test_escape_rfc3986(self):
 661         reserved = "!*'();:@&=+$,/?#[]"
 662         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
 663         self.assertEqual(escape_rfc3986(reserved), reserved)
 664         self.assertEqual(escape_rfc3986(unreserved), unreserved)
 665         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
 666         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
 667         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
 668         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
 669
 670     def test_escape_url(self):
 671         self.assertEqual(
 672             escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
 673             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
 674         )
 675         self.assertEqual(
 676             escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
 677             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
 678         )
 679         self.assertEqual(
 680             escape_url('http://тест.рф/фрагмент'),
 681             'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
 682         )
 683         self.assertEqual(
 684             escape_url('http://тест.рф/абв?абв=абв#абв'),
 685             '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'
 686         )
 687         self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
 688
 689     def test_js_to_json_realworld(self):
 690         inp = '''{
 691             'clip':{'provider':'pseudo'}
 692         }'''
 693         self.assertEqual(js_to_json(inp), '''{
 694             "clip":{"provider":"pseudo"}
 695         }''')
 696         json.loads(js_to_json(inp))
 697
 698         inp = '''{
 699             'playlist':[{'controls':{'all':null}}]
 700         }'''
 701         self.assertEqual(js_to_json(inp), '''{
 702             "playlist":[{"controls":{"all":null}}]
 703         }''')
 704
 705         inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
 706         self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
 707
 708         inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
 709         json_code = js_to_json(inp)
 710         self.assertEqual(json.loads(json_code), json.loads(inp))
 711
 712         inp = '''{
 713             0:{src:'skipped', type: 'application/dash+xml'},
 714             1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
 715         }'''
 716         self.assertEqual(js_to_json(inp), '''{
 717             "0":{"src":"skipped", "type": "application/dash+xml"},
 718             "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
 719         }''')
 720
 721         inp = '''{"foo":101}'''
 722         self.assertEqual(js_to_json(inp), '''{"foo":101}''')
 723
 724         inp = '''{"duration": "00:01:07"}'''
 725         self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
 726
 727     def test_js_to_json_edgecases(self):
 728         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
 729         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
 730
 731         on = js_to_json('{"abc": true}')
 732         self.assertEqual(json.loads(on), {'abc': True})
 733
 734         # Ignore JavaScript code as well
 735         on = js_to_json('''{
 736             "x": 1,
 737             y: "a",
 738             z: some.code
 739         }''')
 740         d = json.loads(on)
 741         self.assertEqual(d['x'], 1)
 742         self.assertEqual(d['y'], 'a')
 743
 744         on = js_to_json('["abc", "def",]')
 745         self.assertEqual(json.loads(on), ['abc', 'def'])
 746
 747         on = js_to_json('{"abc": "def",}')
 748         self.assertEqual(json.loads(on), {'abc': 'def'})
 749
 750         on = js_to_json('{ 0: /* " \n */ ",]" , }')
 751         self.assertEqual(json.loads(on), {'0': ',]'})
 752
 753         on = js_to_json(r'["<p>x<\/p>"]')
 754         self.assertEqual(json.loads(on), ['<p>x</p>'])
 755
 756         on = js_to_json(r'["\xaa"]')
 757         self.assertEqual(json.loads(on), ['\u00aa'])
 758
 759         on = js_to_json("['a\\\nb']")
 760         self.assertEqual(json.loads(on), ['ab'])
 761
 762         on = js_to_json('{0xff:0xff}')
 763         self.assertEqual(json.loads(on), {'255': 255})
 764
 765         on = js_to_json('{077:077}')
 766         self.assertEqual(json.loads(on), {'63': 63})
 767
 768         on = js_to_json('{42:42}')
 769         self.assertEqual(json.loads(on), {'42': 42})
 770
 771     def test_extract_attributes(self):
 772         self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
 773         self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
 774         self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
 775         self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
 776         self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
 777         self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
 778         self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
 779         self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'})  # XML
 780         self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
 781         self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'})  # HTML 3.2
 782         self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'})  # HTML 4.0
 783         self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
 784         self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
 785         self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
 786         self.assertEqual(extract_attributes('<e x >'), {'x': None})
 787         self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
 788         self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
 789         self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
 790         self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
 791         self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
 792         self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
 793         self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
 794         self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'})  # Names lowercased
 795         self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
 796         self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
 797         self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
 798         self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
 799         self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
 800         # "Narrow" Python builds don't support unicode code points outside BMP.
 801         try:
 802             compat_chr(0x10000)
 803             supports_outside_bmp = True
 804         except ValueError:
 805             supports_outside_bmp = False
 806         if supports_outside_bmp:
 807             self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
 808
 809     def test_clean_html(self):
 810         self.assertEqual(clean_html('a:\nb'), 'a: b')
 811         self.assertEqual(clean_html('a:\n   "b"'), 'a:    "b"')
 812
 813     def test_intlist_to_bytes(self):
 814         self.assertEqual(
 815             intlist_to_bytes([0, 1, 127, 128, 255]),
 816             b'\x00\x01\x7f\x80\xff')
 817
 818     def test_args_to_str(self):
 819         self.assertEqual(
 820             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
 821             'foo ba/r -baz \'2 be\' \'\''
 822         )
 823
 824     def test_parse_filesize(self):
 825         self.assertEqual(parse_filesize(None), None)
 826         self.assertEqual(parse_filesize(''), None)
 827         self.assertEqual(parse_filesize('91 B'), 91)
 828         self.assertEqual(parse_filesize('foobar'), None)
 829         self.assertEqual(parse_filesize('2 MiB'), 2097152)
 830         self.assertEqual(parse_filesize('5 GB'), 5000000000)
 831         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
 832         self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
 833         self.assertEqual(parse_filesize('1,24 KB'), 1240)
 834         self.assertEqual(parse_filesize('1,24 kb'), 1240)
 835         self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
 836
 837     def test_parse_count(self):
 838         self.assertEqual(parse_count(None), None)
 839         self.assertEqual(parse_count(''), None)
 840         self.assertEqual(parse_count('0'), 0)
 841         self.assertEqual(parse_count('1000'), 1000)
 842         self.assertEqual(parse_count('1.000'), 1000)
 843         self.assertEqual(parse_count('1.1k'), 1100)
 844         self.assertEqual(parse_count('1.1kk'), 1100000)
 845         self.assertEqual(parse_count('1.1kk '), 1100000)
 846         self.assertEqual(parse_count('1.1kk views'), 1100000)
 847
 848     def test_version_tuple(self):
 849         self.assertEqual(version_tuple('1'), (1,))
 850         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
 851         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
 852
 853     def test_detect_exe_version(self):
 854         self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
 855 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
 856 configuration: --prefix=/usr --extra-'''), '1.2.1')
 857         self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
 858 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
 859         self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
 860 Trying to open render node...
 861 Success at /dev/dri/renderD128.
 862 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
 863
 864     def test_age_restricted(self):
 865         self.assertFalse(age_restricted(None, 10))  # unrestricted content
 866         self.assertFalse(age_restricted(1, None))  # unrestricted policy
 867         self.assertFalse(age_restricted(8, 10))
 868         self.assertTrue(age_restricted(18, 14))
 869         self.assertFalse(age_restricted(18, 18))
 870
 871     def test_is_html(self):
 872         self.assertFalse(is_html(b'\x49\x44\x43<html'))
 873         self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
 874         self.assertTrue(is_html(  # UTF-8 with BOM
 875             b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
 876         self.assertTrue(is_html(  # UTF-16-LE
 877             b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
 878         ))
 879         self.assertTrue(is_html(  # UTF-16-BE
 880             b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
 881         ))
 882         self.assertTrue(is_html(  # UTF-32-BE
 883             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'))
 884         self.assertTrue(is_html(  # UTF-32-LE
 885             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'))
 886
 887     def test_render_table(self):
 888         self.assertEqual(
 889             render_table(
 890                 ['a', 'bcd'],
 891                 [[123, 4], [9999, 51]]),
 892             'a    bcd\n'
 893             '123  4\n'
 894             '9999 51')
 895
 896     def test_match_str(self):
 897         self.assertRaises(ValueError, match_str, 'xy>foobar', {})
 898         self.assertFalse(match_str('xy', {'x': 1200}))
 899         self.assertTrue(match_str('!xy', {'x': 1200}))
 900         self.assertTrue(match_str('x', {'x': 1200}))
 901         self.assertFalse(match_str('!x', {'x': 1200}))
 902         self.assertTrue(match_str('x', {'x': 0}))
 903         self.assertFalse(match_str('x>0', {'x': 0}))
 904         self.assertFalse(match_str('x>0', {}))
 905         self.assertTrue(match_str('x>?0', {}))
 906         self.assertTrue(match_str('x>1K', {'x': 1200}))
 907         self.assertFalse(match_str('x>2K', {'x': 1200}))
 908         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
 909         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
 910         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
 911         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
 912         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
 913         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
 914         self.assertFalse(match_str(
 915             'like_count > 100 & dislike_count <? 50 & description',
 916             {'like_count': 90, 'description': 'foo'}))
 917         self.assertTrue(match_str(
 918             'like_count > 100 & dislike_count <? 50 & description',
 919             {'like_count': 190, 'description': 'foo'}))
 920         self.assertFalse(match_str(
 921             'like_count > 100 & dislike_count <? 50 & description',
 922             {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
 923         self.assertFalse(match_str(
 924             'like_count > 100 & dislike_count <? 50 & description',
 925             {'like_count': 190, 'dislike_count': 10}))
 926
 927     def test_parse_dfxp_time_expr(self):
 928         self.assertEqual(parse_dfxp_time_expr(None), None)
 929         self.assertEqual(parse_dfxp_time_expr(''), None)
 930         self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
 931         self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
 932         self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
 933         self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
 934         self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
 935
 936     def test_dfxp2srt(self):
 937         dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
 938             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
 939             <body>
 940                 <div xml:lang="en">
 941                     <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
 942                     <p begin="1" end="2">第二行<br/>♪♪</p>
 943                     <p begin="2" dur="1"><span>Third<br/>Line</span></p>
 944                     <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
 945                     <p begin="-1" end="-1">Ignore, two</p>
 946                     <p begin="3" dur="-1">Ignored, three</p>
 947                 </div>
 948             </body>
 949             </tt>'''
 950         srt_data = '''1
 951 00:00:00,000 --> 00:00:01,000
 952 The following line contains Chinese characters and special symbols
 953
 954 2
 955 00:00:01,000 --> 00:00:02,000
 956 第二行
 957 ♪♪
 958
 959 3
 960 00:00:02,000 --> 00:00:03,000
 961 Third
 962 Line
 963
 964 '''
 965         self.assertEqual(dfxp2srt(dfxp_data), srt_data)
 966
 967         dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
 968             <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
 969             <body>
 970                 <div xml:lang="en">
 971                     <p begin="0" end="1">The first line</p>
 972                 </div>
 973             </body>
 974             </tt>'''
 975         srt_data = '''1
 976 00:00:00,000 --> 00:00:01,000
 977 The first line
 978
 979 '''
 980         self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
 981
 982     def test_cli_option(self):
 983         self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
 984         self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
 985         self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
 986         self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
 987
 988     def test_cli_valueless_option(self):
 989         self.assertEqual(cli_valueless_option(
 990             {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
 991         self.assertEqual(cli_valueless_option(
 992             {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
 993         self.assertEqual(cli_valueless_option(
 994             {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
 995         self.assertEqual(cli_valueless_option(
 996             {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
 997         self.assertEqual(cli_valueless_option(
 998             {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
 999         self.assertEqual(cli_valueless_option(
1000             {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1001
1002     def test_cli_bool_option(self):
1003         self.assertEqual(
1004             cli_bool_option(
1005                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1006             ['--no-check-certificate', 'true'])
1007         self.assertEqual(
1008             cli_bool_option(
1009                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1010             ['--no-check-certificate=true'])
1011         self.assertEqual(
1012             cli_bool_option(
1013                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1014             ['--check-certificate', 'false'])
1015         self.assertEqual(
1016             cli_bool_option(
1017                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1018             ['--check-certificate=false'])
1019         self.assertEqual(
1020             cli_bool_option(
1021                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1022             ['--check-certificate', 'true'])
1023         self.assertEqual(
1024             cli_bool_option(
1025                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1026             ['--check-certificate=true'])
1027
1028     def test_ohdave_rsa_encrypt(self):
1029         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1030         e = 65537
1031
1032         self.assertEqual(
1033             ohdave_rsa_encrypt(b'aa111222', e, N),
1034             '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1035
1036     def test_encode_base_n(self):
1037         self.assertEqual(encode_base_n(0, 30), '0')
1038         self.assertEqual(encode_base_n(80, 30), '2k')
1039
1040         custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1041         self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1042         self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1043
1044         self.assertRaises(ValueError, encode_base_n, 0, 70)
1045         self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1046
1047     def test_urshift(self):
1048         self.assertEqual(urshift(3, 1), 1)
1049         self.assertEqual(urshift(-3, 1), 2147483646)
1050
1051     def test_get_element_by_class(self):
1052         html = '''
1053             <span class="foo bar">nice</span>
1054         '''
1055
1056         self.assertEqual(get_element_by_class('foo', html), 'nice')
1057         self.assertEqual(get_element_by_class('no-such-class', html), None)
1058
1059 if __name__ == '__main__':
1060     unittest.main()