_ 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     caesar,
  23     clean_html,
  24     date_from_str,
  25     DateRange,
  26     detect_exe_version,
  27     determine_ext,
  28     dict_get,
  29     encode_compat_str,
  30     encodeFilename,
  31     escape_rfc3986,
  32     escape_url,
  33     extract_attributes,
  34     ExtractorError,
  35     find_xpath_attr,
  36     fix_xml_ampersands,
  37     float_or_none,
  38     get_element_by_class,
  39     get_element_by_attribute,
  40     get_elements_by_class,
  41     get_elements_by_attribute,
  42     InAdvancePagedList,
  43     int_or_none,
  44     intlist_to_bytes,
  45     is_html,
  46     js_to_json,
  47     limit_length,
  48     merge_dicts,
  49     mimetype2ext,
  50     month_by_name,
  51     multipart_encode,
  52     ohdave_rsa_encrypt,
  53     OnDemandPagedList,
  54     orderedSet,
  55     parse_age_limit,
  56     parse_duration,
  57     parse_filesize,
  58     parse_count,
  59     parse_iso8601,
  60     parse_resolution,
  61     parse_bitrate,
  62     pkcs1pad,
  63     read_batch_urls,
  64     sanitize_filename,
  65     sanitize_path,
  66     sanitize_url,
  67     expand_path,
  68     prepend_extension,
  69     replace_extension,
  70     remove_start,
  71     remove_end,
  72     remove_quotes,
  73     rot47,
  74     shell_quote,
  75     smuggle_url,
  76     str_to_int,
  77     strip_jsonp,
  78     strip_or_none,
  79     subtitles_filename,
  80     timeconvert,
  81     unescapeHTML,
  82     unified_strdate,
  83     unified_timestamp,
  84     unsmuggle_url,
  85     uppercase_escape,
  86     lowercase_escape,
  87     url_basename,
  88     url_or_none,
  89     base_url,
  90     urljoin,
  91     urlencode_postdata,
  92     urshift,
  93     update_url_query,
  94     version_tuple,
  95     xpath_with_ns,
  96     xpath_element,
  97     xpath_text,
  98     xpath_attr,
  99     render_table,
 100     match_str,
 101     parse_dfxp_time_expr,
 102     dfxp2srt,
 103     cli_option,
 104     cli_valueless_option,
 105     cli_bool_option,
 106     parse_codecs,
 107 )
 108 from youtube_dl.compat import (
 109     compat_chr,
 110     compat_etree_fromstring,
 111     compat_getenv,
 112     compat_os_name,
 113     compat_setenv,
 114     compat_urlparse,
 115     compat_parse_qs,
 116 )
 117
 118
 119 class TestUtil(unittest.TestCase):
 120     def test_timeconvert(self):
 121         self.assertTrue(timeconvert('') is None)
 122         self.assertTrue(timeconvert('bougrg') is None)
 123
 124     def test_sanitize_filename(self):
 125         self.assertEqual(sanitize_filename('abc'), 'abc')
 126         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
 127
 128         self.assertEqual(sanitize_filename('123'), '123')
 129
 130         self.assertEqual('abc_de', sanitize_filename('abc/de'))
 131         self.assertFalse('/' in sanitize_filename('abc/de///'))
 132
 133         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
 134         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
 135         self.assertEqual('yes no', sanitize_filename('yes? no'))
 136         self.assertEqual('this - that', sanitize_filename('this: that'))
 137
 138         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
 139         aumlaut = 'ä'
 140         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
 141         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
 142         self.assertEqual(sanitize_filename(tests), tests)
 143
 144         self.assertEqual(
 145             sanitize_filename('New World record at 0:12:34'),
 146             'New World record at 0_12_34')
 147
 148         self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
 149         self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
 150         self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
 151         self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
 152
 153         forbidden = '"\0\\/'
 154         for fc in forbidden:
 155             for fbc in forbidden:
 156                 self.assertTrue(fbc not in sanitize_filename(fc))
 157
 158     def test_sanitize_filename_restricted(self):
 159         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
 160         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
 161
 162         self.assertEqual(sanitize_filename('123', restricted=True), '123')
 163
 164         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
 165         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
 166
 167         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
 168         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
 169         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
 170         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
 171
 172         tests = 'aäb\u4e2d\u56fd\u7684c'
 173         self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
 174         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
 175
 176         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
 177         for fc in forbidden:
 178             for fbc in forbidden:
 179                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
 180
 181         # Handle a common case more neatly
 182         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
 183         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
 184         # .. but make sure the file name is never empty
 185         self.assertTrue(sanitize_filename('-', restricted=True) != '')
 186         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 187
 188         self.assertEqual(sanitize_filename(
 189             'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
 190             'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy')
 191
 192     def test_sanitize_ids(self):
 193         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 194         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 195         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 196
 197     def test_sanitize_path(self):
 198         if sys.platform != 'win32':
 199             return
 200
 201         self.assertEqual(sanitize_path('abc'), 'abc')
 202         self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
 203         self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
 204         self.assertEqual(sanitize_path('abc|def'), 'abc#def')
 205         self.assertEqual(sanitize_path('<>:"|?*'), '#######')
 206         self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
 207         self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
 208
 209         self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
 210         self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
 211
 212         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 213         self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
 214         self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
 215         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 216
 217         self.assertEqual(
 218             sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
 219             'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
 220
 221         self.assertEqual(
 222             sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
 223             'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
 224         self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
 225         self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
 226         self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
 227
 228         self.assertEqual(sanitize_path('../abc'), '..\\abc')
 229         self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
 230         self.assertEqual(sanitize_path('./abc'), 'abc')
 231         self.assertEqual(sanitize_path('./../abc'), '..\\abc')
 232
 233     def test_sanitize_url(self):
 234         self.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
 235         self.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
 236         self.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
 237         self.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')
 238
 239     def test_expand_path(self):
 240         def env(var):
 241             return '%{0}%'.format(var) if sys.platform == 'win32' else '${0}'.format(var)
 242
 243         compat_setenv('YOUTUBE_DL_EXPATH_PATH', 'expanded')
 244         self.assertEqual(expand_path(env('YOUTUBE_DL_EXPATH_PATH')), 'expanded')
 245         self.assertEqual(expand_path(env('HOME')), compat_getenv('HOME'))
 246         self.assertEqual(expand_path('~'), compat_getenv('HOME'))
 247         self.assertEqual(
 248             expand_path('~/%s' % env('YOUTUBE_DL_EXPATH_PATH')),
 249             '%s/expanded' % compat_getenv('HOME'))
 250
 251     def test_prepend_extension(self):
 252         self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
 253         self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
 254         self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 255         self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
 256         self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
 257         self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
 258
 259     def test_replace_extension(self):
 260         self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
 261         self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
 262         self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 263         self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
 264         self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
 265         self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
 266
 267     def test_subtitles_filename(self):
 268         self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt'), 'abc.en.vtt')
 269         self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt', 'ext'), 'abc.en.vtt')
 270         self.assertEqual(subtitles_filename('abc.unexpected_ext', 'en', 'vtt', 'ext'), 'abc.unexpected_ext.en.vtt')
 271
 272     def test_remove_start(self):
 273         self.assertEqual(remove_start(None, 'A - '), None)
 274         self.assertEqual(remove_start('A - B', 'A - '), 'B')
 275         self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
 276
 277     def test_remove_end(self):
 278         self.assertEqual(remove_end(None, ' - B'), None)
 279         self.assertEqual(remove_end('A - B', ' - B'), 'A')
 280         self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
 281
 282     def test_remove_quotes(self):
 283         self.assertEqual(remove_quotes(None), None)
 284         self.assertEqual(remove_quotes('"'), '"')
 285         self.assertEqual(remove_quotes("'"), "'")
 286         self.assertEqual(remove_quotes(';'), ';')
 287         self.assertEqual(remove_quotes('";'), '";')
 288         self.assertEqual(remove_quotes('""'), '')
 289         self.assertEqual(remove_quotes('";"'), ';')
 290
 291     def test_ordered_set(self):
 292         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
 293         self.assertEqual(orderedSet([]), [])
 294         self.assertEqual(orderedSet([1]), [1])
 295         # keep the list ordered
 296         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
 297
 298     def test_unescape_html(self):
 299         self.assertEqual(unescapeHTML('%20;'), '%20;')
 300         self.assertEqual(unescapeHTML('&#x2F;'), '/')
 301         self.assertEqual(unescapeHTML('&#47;'), '/')
 302         self.assertEqual(unescapeHTML('&eacute;'), 'é')
 303         self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
 304         self.assertEqual(unescapeHTML('&a&quot;'), '&a"')
 305         # HTML5 entities
 306         self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')
 307
 308     def test_date_from_str(self):
 309         self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
 310         self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
 311         self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
 312         self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
 313         self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
 314
 315     def test_daterange(self):
 316         _20century = DateRange("19000101", "20000101")
 317         self.assertFalse("17890714" in _20century)
 318         _ac = DateRange("00010101")
 319         self.assertTrue("19690721" in _ac)
 320         _firstmilenium = DateRange(end="10000101")
 321         self.assertTrue("07110427" in _firstmilenium)
 322
 323     def test_unified_dates(self):
 324         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
 325         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
 326         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
 327         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
 328         self.assertEqual(unified_strdate('1968 12 10'), '19681210')
 329         self.assertEqual(unified_strdate('1968-12-10'), '19681210')
 330         self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
 331         self.assertEqual(
 332             unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
 333             '20141126')
 334         self.assertEqual(
 335             unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
 336             '20150202')
 337         self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
 338         self.assertEqual(unified_strdate('25-09-2014'), '20140925')
 339         self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
 340         self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
 341         self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
 342         self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
 343         self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
 344         self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
 345         self.assertEqual(unified_strdate('November 3rd, 2019'), '20191103')
 346         self.assertEqual(unified_strdate('October 23rd, 2005'), '20051023')
 347
 348     def test_unified_timestamps(self):
 349         self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
 350         self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
 351         self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
 352         self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
 353         self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
 354         self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
 355         self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
 356         self.assertEqual(
 357             unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
 358             1417001400)
 359         self.assertEqual(
 360             unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
 361             1422902860)
 362         self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
 363         self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
 364         self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
 365         self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
 366         self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
 367         self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
 368         self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
 369         self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
 370         self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
 371         self.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363)
 372
 373     def test_determine_ext(self):
 374         self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
 375         self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
 376         self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
 377         self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
 378         self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
 379         self.assertEqual(determine_ext('foobar', None), None)
 380
 381     def test_find_xpath_attr(self):
 382         testxml = '''<root>
 383             <node/>
 384             <node x="a"/>
 385             <node x="a" y="c" />
 386             <node x="b" y="d" />
 387             <node x="" />
 388         </root>'''
 389         doc = compat_etree_fromstring(testxml)
 390
 391         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
 392         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
 393         self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
 394         self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
 395         self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
 396         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
 397         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
 398         self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
 399         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
 400         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
 401         self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
 402
 403     def test_xpath_with_ns(self):
 404         testxml = '''<root xmlns:media="http://example.com/">
 405             <media:song>
 406                 <media:author>The Author</media:author>
 407                 <url>http://server.com/download.mp3</url>
 408             </media:song>
 409         </root>'''
 410         doc = compat_etree_fromstring(testxml)
 411         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
 412         self.assertTrue(find('media:song') is not None)
 413         self.assertEqual(find('media:song/media:author').text, 'The Author')
 414         self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
 415
 416     def test_xpath_element(self):
 417         doc = xml.etree.ElementTree.Element('root')
 418         div = xml.etree.ElementTree.SubElement(doc, 'div')
 419         p = xml.etree.ElementTree.SubElement(div, 'p')
 420         p.text = 'Foo'
 421         self.assertEqual(xpath_element(doc, 'div/p'), p)
 422         self.assertEqual(xpath_element(doc, ['div/p']), p)
 423         self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
 424         self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
 425         self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
 426         self.assertTrue(xpath_element(doc, 'div/bar') is None)
 427         self.assertTrue(xpath_element(doc, ['div/bar']) is None)
 428         self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
 429         self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
 430         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
 431         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
 432
 433     def test_xpath_text(self):
 434         testxml = '''<root>
 435             <div>
 436                 <p>Foo</p>
 437             </div>
 438         </root>'''
 439         doc = compat_etree_fromstring(testxml)
 440         self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
 441         self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
 442         self.assertTrue(xpath_text(doc, 'div/bar') is None)
 443         self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
 444
 445     def test_xpath_attr(self):
 446         testxml = '''<root>
 447             <div>
 448                 <p x="a">Foo</p>
 449             </div>
 450         </root>'''
 451         doc = compat_etree_fromstring(testxml)
 452         self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
 453         self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
 454         self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
 455         self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
 456         self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
 457         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
 458         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
 459
 460     def test_smuggle_url(self):
 461         data = {"ö": "ö", "abc": [3]}
 462         url = 'https://foo.bar/baz?x=y#a'
 463         smug_url = smuggle_url(url, data)
 464         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
 465         self.assertEqual(url, unsmug_url)
 466         self.assertEqual(data, unsmug_data)
 467
 468         res_url, res_data = unsmuggle_url(url)
 469         self.assertEqual(res_url, url)
 470         self.assertEqual(res_data, None)
 471
 472         smug_url = smuggle_url(url, {'a': 'b'})
 473         smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
 474         res_url, res_data = unsmuggle_url(smug_smug_url)
 475         self.assertEqual(res_url, url)
 476         self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
 477
 478     def test_shell_quote(self):
 479         args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
 480         self.assertEqual(
 481             shell_quote(args),
 482             """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
 483
 484     def test_float_or_none(self):
 485         self.assertEqual(float_or_none('42.42'), 42.42)
 486         self.assertEqual(float_or_none('42'), 42.0)
 487         self.assertEqual(float_or_none(''), None)
 488         self.assertEqual(float_or_none(None), None)
 489         self.assertEqual(float_or_none([]), None)
 490         self.assertEqual(float_or_none(set()), None)
 491
 492     def test_int_or_none(self):
 493         self.assertEqual(int_or_none('42'), 42)
 494         self.assertEqual(int_or_none(''), None)
 495         self.assertEqual(int_or_none(None), None)
 496         self.assertEqual(int_or_none([]), None)
 497         self.assertEqual(int_or_none(set()), None)
 498
 499     def test_str_to_int(self):
 500         self.assertEqual(str_to_int('123,456'), 123456)
 501         self.assertEqual(str_to_int('123.456'), 123456)
 502         self.assertEqual(str_to_int(523), 523)
 503         # Python 3 has no long
 504         if sys.version_info < (3, 0):
 505             eval('self.assertEqual(str_to_int(123456L), 123456)')
 506         self.assertEqual(str_to_int('noninteger'), None)
 507         self.assertEqual(str_to_int([]), None)
 508
 509     def test_url_basename(self):
 510         self.assertEqual(url_basename('http://foo.de/'), '')
 511         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
 512         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
 513         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
 514         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
 515         self.assertEqual(
 516             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
 517             'trailer.mp4')
 518
 519     def test_base_url(self):
 520         self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
 521         self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
 522         self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
 523         self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
 524         self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
 525
 526     def test_urljoin(self):
 527         self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 528         self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 529         self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 530         self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 531         self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
 532         self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 533         self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 534         self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 535         self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 536         self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
 537         self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 538         self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
 539         self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 540         self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 541         self.assertEqual(urljoin('http://foo.de/', None), None)
 542         self.assertEqual(urljoin('http://foo.de/', ''), None)
 543         self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
 544         self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
 545         self.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de')
 546         self.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de')
 547
 548     def test_url_or_none(self):
 549         self.assertEqual(url_or_none(None), None)
 550         self.assertEqual(url_or_none(''), None)
 551         self.assertEqual(url_or_none('foo'), None)
 552         self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
 553         self.assertEqual(url_or_none('https://foo.de'), 'https://foo.de')
 554         self.assertEqual(url_or_none('http$://foo.de'), None)
 555         self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
 556         self.assertEqual(url_or_none('//foo.de'), '//foo.de')
 557
 558     def test_parse_age_limit(self):
 559         self.assertEqual(parse_age_limit(None), None)
 560         self.assertEqual(parse_age_limit(False), None)
 561         self.assertEqual(parse_age_limit('invalid'), None)
 562         self.assertEqual(parse_age_limit(0), 0)
 563         self.assertEqual(parse_age_limit(18), 18)
 564         self.assertEqual(parse_age_limit(21), 21)
 565         self.assertEqual(parse_age_limit(22), None)
 566         self.assertEqual(parse_age_limit('18'), 18)
 567         self.assertEqual(parse_age_limit('18+'), 18)
 568         self.assertEqual(parse_age_limit('PG-13'), 13)
 569         self.assertEqual(parse_age_limit('TV-14'), 14)
 570         self.assertEqual(parse_age_limit('TV-MA'), 17)
 571         self.assertEqual(parse_age_limit('TV14'), 14)
 572         self.assertEqual(parse_age_limit('TV_G'), 0)
 573
 574     def test_parse_duration(self):
 575         self.assertEqual(parse_duration(None), None)
 576         self.assertEqual(parse_duration(False), None)
 577         self.assertEqual(parse_duration('invalid'), None)
 578         self.assertEqual(parse_duration('1'), 1)
 579         self.assertEqual(parse_duration('1337:12'), 80232)
 580         self.assertEqual(parse_duration('9:12:43'), 33163)
 581         self.assertEqual(parse_duration('12:00'), 720)
 582         self.assertEqual(parse_duration('00:01:01'), 61)
 583         self.assertEqual(parse_duration('x:y'), None)
 584         self.assertEqual(parse_duration('3h11m53s'), 11513)
 585         self.assertEqual(parse_duration('3h 11m 53s'), 11513)
 586         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
 587         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
 588         self.assertEqual(parse_duration('62m45s'), 3765)
 589         self.assertEqual(parse_duration('6m59s'), 419)
 590         self.assertEqual(parse_duration('49s'), 49)
 591         self.assertEqual(parse_duration('0h0m0s'), 0)
 592         self.assertEqual(parse_duration('0m0s'), 0)
 593         self.assertEqual(parse_duration('0s'), 0)
 594         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
 595         self.assertEqual(parse_duration('T30M38S'), 1838)
 596         self.assertEqual(parse_duration('5 s'), 5)
 597         self.assertEqual(parse_duration('3 min'), 180)
 598         self.assertEqual(parse_duration('2.5 hours'), 9000)
 599         self.assertEqual(parse_duration('02:03:04'), 7384)
 600         self.assertEqual(parse_duration('01:02:03:04'), 93784)
 601         self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
 602         self.assertEqual(parse_duration('87 Min.'), 5220)
 603         self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
 604         self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
 605         self.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)
 606
 607     def test_fix_xml_ampersands(self):
 608         self.assertEqual(
 609             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 610         self.assertEqual(
 611             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 612             '"&amp;x=y&amp;wrong;&amp;z=a')
 613         self.assertEqual(
 614             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 615             '&amp;&apos;&gt;&lt;&quot;')
 616         self.assertEqual(
 617             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 618         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 619
 620     def test_paged_list(self):
 621         def testPL(size, pagesize, sliceargs, expected):
 622             def get_page(pagenum):
 623                 firstid = pagenum * pagesize
 624                 upto = min(size, pagenum * pagesize + pagesize)
 625                 for i in range(firstid, upto):
 626                     yield i
 627
 628             pl = OnDemandPagedList(get_page, pagesize)
 629             got = pl.getslice(*sliceargs)
 630             self.assertEqual(got, expected)
 631
 632             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
 633             got = iapl.getslice(*sliceargs)
 634             self.assertEqual(got, expected)
 635
 636         testPL(5, 2, (), [0, 1, 2, 3, 4])
 637         testPL(5, 2, (1,), [1, 2, 3, 4])
 638         testPL(5, 2, (2,), [2, 3, 4])
 639         testPL(5, 2, (4,), [4])
 640         testPL(5, 2, (0, 3), [0, 1, 2])
 641         testPL(5, 2, (1, 4), [1, 2, 3])
 642         testPL(5, 2, (2, 99), [2, 3, 4])
 643         testPL(5, 2, (20, 99), [])
 644
 645     def test_read_batch_urls(self):
 646         f = io.StringIO('''\xef\xbb\xbf foo
 647             bar\r
 648             baz
 649             # More after this line\r
 650             ; or after this
 651             bam''')
 652         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
 653
 654     def test_urlencode_postdata(self):
 655         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
 656         self.assertTrue(isinstance(data, bytes))
 657
 658     def test_update_url_query(self):
 659         def query_dict(url):
 660             return compat_parse_qs(compat_urlparse.urlparse(url).query)
 661         self.assertEqual(query_dict(update_url_query(
 662             'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
 663             query_dict('http://example.com/path?quality=HD&format=mp4'))
 664         self.assertEqual(query_dict(update_url_query(
 665             'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
 666             query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
 667         self.assertEqual(query_dict(update_url_query(
 668             'http://example.com/path', {'fields': 'id,formats,subtitles'})),
 669             query_dict('http://example.com/path?fields=id,formats,subtitles'))
 670         self.assertEqual(query_dict(update_url_query(
 671             'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
 672             query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
 673         self.assertEqual(query_dict(update_url_query(
 674             'http://example.com/path?manifest=f4m', {'manifest': []})),
 675             query_dict('http://example.com/path'))
 676         self.assertEqual(query_dict(update_url_query(
 677             'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
 678             query_dict('http://example.com/path?system=LINUX'))
 679         self.assertEqual(query_dict(update_url_query(
 680             'http://example.com/path', {'fields': b'id,formats,subtitles'})),
 681             query_dict('http://example.com/path?fields=id,formats,subtitles'))
 682         self.assertEqual(query_dict(update_url_query(
 683             'http://example.com/path', {'width': 1080, 'height': 720})),
 684             query_dict('http://example.com/path?width=1080&height=720'))
 685         self.assertEqual(query_dict(update_url_query(
 686             'http://example.com/path', {'bitrate': 5020.43})),
 687             query_dict('http://example.com/path?bitrate=5020.43'))
 688         self.assertEqual(query_dict(update_url_query(
 689             'http://example.com/path', {'test': '第二行тест'})),
 690             query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
 691
 692     def test_multipart_encode(self):
 693         self.assertEqual(
 694             multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0],
 695             b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
 696         self.assertEqual(
 697             multipart_encode({'欄位'.encode('utf-8'): '值'.encode('utf-8')}, boundary='AAAAAA')[0],
 698             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')
 699         self.assertRaises(
 700             ValueError, multipart_encode, {b'field': b'value'}, boundary='value')
 701
 702     def test_dict_get(self):
 703         FALSE_VALUES = {
 704             'none': None,
 705             'false': False,
 706             'zero': 0,
 707             'empty_string': '',
 708             'empty_list': [],
 709         }
 710         d = FALSE_VALUES.copy()
 711         d['a'] = 42
 712         self.assertEqual(dict_get(d, 'a'), 42)
 713         self.assertEqual(dict_get(d, 'b'), None)
 714         self.assertEqual(dict_get(d, 'b', 42), 42)
 715         self.assertEqual(dict_get(d, ('a', )), 42)
 716         self.assertEqual(dict_get(d, ('b', 'a', )), 42)
 717         self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
 718         self.assertEqual(dict_get(d, ('b', 'c', )), None)
 719         self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
 720         for key, false_value in FALSE_VALUES.items():
 721             self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
 722             self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
 723
 724     def test_merge_dicts(self):
 725         self.assertEqual(merge_dicts({'a': 1}, {'b': 2}), {'a': 1, 'b': 2})
 726         self.assertEqual(merge_dicts({'a': 1}, {'a': 2}), {'a': 1})
 727         self.assertEqual(merge_dicts({'a': 1}, {'a': None}), {'a': 1})
 728         self.assertEqual(merge_dicts({'a': 1}, {'a': ''}), {'a': 1})
 729         self.assertEqual(merge_dicts({'a': 1}, {}), {'a': 1})
 730         self.assertEqual(merge_dicts({'a': None}, {'a': 1}), {'a': 1})
 731         self.assertEqual(merge_dicts({'a': ''}, {'a': 1}), {'a': ''})
 732         self.assertEqual(merge_dicts({'a': ''}, {'a': 'abc'}), {'a': 'abc'})
 733         self.assertEqual(merge_dicts({'a': None}, {'a': ''}, {'a': 'abc'}), {'a': 'abc'})
 734
 735     def test_encode_compat_str(self):
 736         self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
 737         self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
 738
 739     def test_parse_iso8601(self):
 740         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
 741         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
 742         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
 743         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
 744         self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
 745         self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
 746
 747     def test_strip_jsonp(self):
 748         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
 749         d = json.loads(stripped)
 750         self.assertEqual(d, [{"id": "532cb", "x": 3}])
 751
 752         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
 753         d = json.loads(stripped)
 754         self.assertEqual(d, {'STATUS': 'OK'})
 755
 756         stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
 757         d = json.loads(stripped)
 758         self.assertEqual(d, {'status': 'success'})
 759
 760         stripped = strip_jsonp('window.cb && window.cb({"status": "success"});')
 761         d = json.loads(stripped)
 762         self.assertEqual(d, {'status': 'success'})
 763
 764         stripped = strip_jsonp('window.cb && cb({"status": "success"});')
 765         d = json.loads(stripped)
 766         self.assertEqual(d, {'status': 'success'})
 767
 768         stripped = strip_jsonp('({"status": "success"});')
 769         d = json.loads(stripped)
 770         self.assertEqual(d, {'status': 'success'})
 771
 772     def test_strip_or_none(self):
 773         self.assertEqual(strip_or_none(' abc'), 'abc')
 774         self.assertEqual(strip_or_none('abc '), 'abc')
 775         self.assertEqual(strip_or_none(' abc '), 'abc')
 776         self.assertEqual(strip_or_none('\tabc\t'), 'abc')
 777         self.assertEqual(strip_or_none('\n\tabc\n\t'), 'abc')
 778         self.assertEqual(strip_or_none('abc'), 'abc')
 779         self.assertEqual(strip_or_none(''), '')
 780         self.assertEqual(strip_or_none(None), None)
 781         self.assertEqual(strip_or_none(42), None)
 782         self.assertEqual(strip_or_none([]), None)
 783
 784     def test_uppercase_escape(self):
 785         self.assertEqual(uppercase_escape('aä'), 'aä')
 786         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
 787
 788     def test_lowercase_escape(self):
 789         self.assertEqual(lowercase_escape('aä'), 'aä')
 790         self.assertEqual(lowercase_escape('\\u0026'), '&')
 791
 792     def test_limit_length(self):
 793         self.assertEqual(limit_length(None, 12), None)
 794         self.assertEqual(limit_length('foo', 12), 'foo')
 795         self.assertTrue(
 796             limit_length('foo bar baz asd', 12).startswith('foo bar'))
 797         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
 798
 799     def test_mimetype2ext(self):
 800         self.assertEqual(mimetype2ext(None), None)
 801         self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
 802         self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
 803         self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
 804         self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
 805         self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
 806
 807     def test_month_by_name(self):
 808         self.assertEqual(month_by_name(None), None)
 809         self.assertEqual(month_by_name('December', 'en'), 12)
 810         self.assertEqual(month_by_name('décembre', 'fr'), 12)
 811         self.assertEqual(month_by_name('December'), 12)
 812         self.assertEqual(month_by_name('décembre'), None)
 813         self.assertEqual(month_by_name('Unknown', 'unknown'), None)
 814
 815     def test_parse_codecs(self):
 816         self.assertEqual(parse_codecs(''), {})
 817         self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
 818             'vcodec': 'avc1.77.30',
 819             'acodec': 'mp4a.40.2',
 820         })
 821         self.assertEqual(parse_codecs('mp4a.40.2'), {
 822             'vcodec': 'none',
 823             'acodec': 'mp4a.40.2',
 824         })
 825         self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
 826             'vcodec': 'avc1.42001e',
 827             'acodec': 'mp4a.40.5',
 828         })
 829         self.assertEqual(parse_codecs('avc3.640028'), {
 830             'vcodec': 'avc3.640028',
 831             'acodec': 'none',
 832         })
 833         self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
 834             'vcodec': 'h264',
 835             'acodec': 'aac',
 836         })
 837         self.assertEqual(parse_codecs('av01.0.05M.08'), {
 838             'vcodec': 'av01.0.05M.08',
 839             'acodec': 'none',
 840         })
 841         self.assertEqual(parse_codecs('theora, vorbis'), {
 842             'vcodec': 'theora',
 843             'acodec': 'vorbis',
 844         })
 845         self.assertEqual(parse_codecs('unknownvcodec, unknownacodec'), {
 846             'vcodec': 'unknownvcodec',
 847             'acodec': 'unknownacodec',
 848         })
 849         self.assertEqual(parse_codecs('unknown'), {})
 850
 851     def test_escape_rfc3986(self):
 852         reserved = "!*'();:@&=+$,/?#[]"
 853         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
 854         self.assertEqual(escape_rfc3986(reserved), reserved)
 855         self.assertEqual(escape_rfc3986(unreserved), unreserved)
 856         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
 857         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
 858         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
 859         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
 860
 861     def test_escape_url(self):
 862         self.assertEqual(
 863             escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
 864             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
 865         )
 866         self.assertEqual(
 867             escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
 868             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
 869         )
 870         self.assertEqual(
 871             escape_url('http://тест.рф/фрагмент'),
 872             'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
 873         )
 874         self.assertEqual(
 875             escape_url('http://тест.рф/абв?абв=абв#абв'),
 876             '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'
 877         )
 878         self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
 879
 880     def test_js_to_json_realworld(self):
 881         inp = '''{
 882             'clip':{'provider':'pseudo'}
 883         }'''
 884         self.assertEqual(js_to_json(inp), '''{
 885             "clip":{"provider":"pseudo"}
 886         }''')
 887         json.loads(js_to_json(inp))
 888
 889         inp = '''{
 890             'playlist':[{'controls':{'all':null}}]
 891         }'''
 892         self.assertEqual(js_to_json(inp), '''{
 893             "playlist":[{"controls":{"all":null}}]
 894         }''')
 895
 896         inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
 897         self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
 898
 899         inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
 900         json_code = js_to_json(inp)
 901         self.assertEqual(json.loads(json_code), json.loads(inp))
 902
 903         inp = '''{
 904             0:{src:'skipped', type: 'application/dash+xml'},
 905             1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
 906         }'''
 907         self.assertEqual(js_to_json(inp), '''{
 908             "0":{"src":"skipped", "type": "application/dash+xml"},
 909             "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
 910         }''')
 911
 912         inp = '''{"foo":101}'''
 913         self.assertEqual(js_to_json(inp), '''{"foo":101}''')
 914
 915         inp = '''{"duration": "00:01:07"}'''
 916         self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
 917
 918         inp = '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}'''
 919         self.assertEqual(js_to_json(inp), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''')
 920
 921     def test_js_to_json_edgecases(self):
 922         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
 923         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
 924
 925         on = js_to_json('{"abc": true}')
 926         self.assertEqual(json.loads(on), {'abc': True})
 927
 928         # Ignore JavaScript code as well
 929         on = js_to_json('''{
 930             "x": 1,
 931             y: "a",
 932             z: some.code
 933         }''')
 934         d = json.loads(on)
 935         self.assertEqual(d['x'], 1)
 936         self.assertEqual(d['y'], 'a')
 937
 938         on = js_to_json('["abc", "def",]')
 939         self.assertEqual(json.loads(on), ['abc', 'def'])
 940
 941         on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
 942         self.assertEqual(json.loads(on), ['abc', 'def'])
 943
 944         on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
 945         self.assertEqual(json.loads(on), ['abc', 'def'])
 946
 947         on = js_to_json('{"abc": "def",}')
 948         self.assertEqual(json.loads(on), {'abc': 'def'})
 949
 950         on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
 951         self.assertEqual(json.loads(on), {'abc': 'def'})
 952
 953         on = js_to_json('{ 0: /* " \n */ ",]" , }')
 954         self.assertEqual(json.loads(on), {'0': ',]'})
 955
 956         on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
 957         self.assertEqual(json.loads(on), {'0': ',]'})
 958
 959         on = js_to_json('{ 0: // comment\n1 }')
 960         self.assertEqual(json.loads(on), {'0': 1})
 961
 962         on = js_to_json(r'["<p>x<\/p>"]')
 963         self.assertEqual(json.loads(on), ['<p>x</p>'])
 964
 965         on = js_to_json(r'["\xaa"]')
 966         self.assertEqual(json.loads(on), ['\u00aa'])
 967
 968         on = js_to_json("['a\\\nb']")
 969         self.assertEqual(json.loads(on), ['ab'])
 970
 971         on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
 972         self.assertEqual(json.loads(on), ['ab'])
 973
 974         on = js_to_json('{0xff:0xff}')
 975         self.assertEqual(json.loads(on), {'255': 255})
 976
 977         on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
 978         self.assertEqual(json.loads(on), {'255': 255})
 979
 980         on = js_to_json('{077:077}')
 981         self.assertEqual(json.loads(on), {'63': 63})
 982
 983         on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
 984         self.assertEqual(json.loads(on), {'63': 63})
 985
 986         on = js_to_json('{42:42}')
 987         self.assertEqual(json.loads(on), {'42': 42})
 988
 989         on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
 990         self.assertEqual(json.loads(on), {'42': 42})
 991
 992         on = js_to_json('{42:4.2e1}')
 993         self.assertEqual(json.loads(on), {'42': 42.0})
 994
 995     def test_js_to_json_malformed(self):
 996         self.assertEqual(js_to_json('42a1'), '42"a1"')
 997         self.assertEqual(js_to_json('42a-1'), '42"a"-1')
 998
 999     def test_extract_attributes(self):
1000         self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
1001         self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
1002         self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
1003         self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
1004         self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
1005         self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
1006         self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
1007         self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'})  # XML
1008         self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
1009         self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'})  # HTML 3.2
1010         self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'})  # HTML 4.0
1011         self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
1012         self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
1013         self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
1014         self.assertEqual(extract_attributes('<e x >'), {'x': None})
1015         self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
1016         self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
1017         self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
1018         self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
1019         self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
1020         self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
1021         self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
1022         self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'})  # Names lowercased
1023         self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
1024         self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
1025         self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
1026         self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
1027         self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
1028         # "Narrow" Python builds don't support unicode code points outside BMP.
1029         try:
1030             compat_chr(0x10000)
1031             supports_outside_bmp = True
1032         except ValueError:
1033             supports_outside_bmp = False
1034         if supports_outside_bmp:
1035             self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
1036         # Malformed HTML should not break attributes extraction on older Python
1037         self.assertEqual(extract_attributes('<mal"formed/>'), {})
1038
1039     def test_clean_html(self):
1040         self.assertEqual(clean_html('a:\nb'), 'a: b')
1041         self.assertEqual(clean_html('a:\n   "b"'), 'a:    "b"')
1042         self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
1043
1044     def test_intlist_to_bytes(self):
1045         self.assertEqual(
1046             intlist_to_bytes([0, 1, 127, 128, 255]),
1047             b'\x00\x01\x7f\x80\xff')
1048
1049     def test_args_to_str(self):
1050         self.assertEqual(
1051             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
1052             'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""'
1053         )
1054
1055     def test_parse_filesize(self):
1056         self.assertEqual(parse_filesize(None), None)
1057         self.assertEqual(parse_filesize(''), None)
1058         self.assertEqual(parse_filesize('91 B'), 91)
1059         self.assertEqual(parse_filesize('foobar'), None)
1060         self.assertEqual(parse_filesize('2 MiB'), 2097152)
1061         self.assertEqual(parse_filesize('5 GB'), 5000000000)
1062         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
1063         self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
1064         self.assertEqual(parse_filesize('1,24 KB'), 1240)
1065         self.assertEqual(parse_filesize('1,24 kb'), 1240)
1066         self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
1067
1068     def test_parse_count(self):
1069         self.assertEqual(parse_count(None), None)
1070         self.assertEqual(parse_count(''), None)
1071         self.assertEqual(parse_count('0'), 0)
1072         self.assertEqual(parse_count('1000'), 1000)
1073         self.assertEqual(parse_count('1.000'), 1000)
1074         self.assertEqual(parse_count('1.1k'), 1100)
1075         self.assertEqual(parse_count('1.1kk'), 1100000)
1076         self.assertEqual(parse_count('1.1kk '), 1100000)
1077         self.assertEqual(parse_count('1.1kk views'), 1100000)
1078
1079     def test_parse_resolution(self):
1080         self.assertEqual(parse_resolution(None), {})
1081         self.assertEqual(parse_resolution(''), {})
1082         self.assertEqual(parse_resolution('1920x1080'), {'width': 1920, 'height': 1080})
1083         self.assertEqual(parse_resolution('1920×1080'), {'width': 1920, 'height': 1080})
1084         self.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080})
1085         self.assertEqual(parse_resolution('720p'), {'height': 720})
1086         self.assertEqual(parse_resolution('4k'), {'height': 2160})
1087         self.assertEqual(parse_resolution('8K'), {'height': 4320})
1088
1089     def test_parse_bitrate(self):
1090         self.assertEqual(parse_bitrate(None), None)
1091         self.assertEqual(parse_bitrate(''), None)
1092         self.assertEqual(parse_bitrate('300kbps'), 300)
1093         self.assertEqual(parse_bitrate('1500kbps'), 1500)
1094         self.assertEqual(parse_bitrate('300 kbps'), 300)
1095
1096     def test_version_tuple(self):
1097         self.assertEqual(version_tuple('1'), (1,))
1098         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
1099         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
1100
1101     def test_detect_exe_version(self):
1102         self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
1103 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
1104 configuration: --prefix=/usr --extra-'''), '1.2.1')
1105         self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
1106 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
1107         self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
1108 Trying to open render node...
1109 Success at /dev/dri/renderD128.
1110 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
1111
1112     def test_age_restricted(self):
1113         self.assertFalse(age_restricted(None, 10))  # unrestricted content
1114         self.assertFalse(age_restricted(1, None))  # unrestricted policy
1115         self.assertFalse(age_restricted(8, 10))
1116         self.assertTrue(age_restricted(18, 14))
1117         self.assertFalse(age_restricted(18, 18))
1118
1119     def test_is_html(self):
1120         self.assertFalse(is_html(b'\x49\x44\x43<html'))
1121         self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
1122         self.assertTrue(is_html(  # UTF-8 with BOM
1123             b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
1124         self.assertTrue(is_html(  # UTF-16-LE
1125             b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
1126         ))
1127         self.assertTrue(is_html(  # UTF-16-BE
1128             b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
1129         ))
1130         self.assertTrue(is_html(  # UTF-32-BE
1131             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'))
1132         self.assertTrue(is_html(  # UTF-32-LE
1133             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'))
1134
1135     def test_render_table(self):
1136         self.assertEqual(
1137             render_table(
1138                 ['a', 'bcd'],
1139                 [[123, 4], [9999, 51]]),
1140             'a    bcd\n'
1141             '123  4\n'
1142             '9999 51')
1143
1144     def test_match_str(self):
1145         self.assertRaises(ValueError, match_str, 'xy>foobar', {})
1146         self.assertFalse(match_str('xy', {'x': 1200}))
1147         self.assertTrue(match_str('!xy', {'x': 1200}))
1148         self.assertTrue(match_str('x', {'x': 1200}))
1149         self.assertFalse(match_str('!x', {'x': 1200}))
1150         self.assertTrue(match_str('x', {'x': 0}))
1151         self.assertFalse(match_str('x>0', {'x': 0}))
1152         self.assertFalse(match_str('x>0', {}))
1153         self.assertTrue(match_str('x>?0', {}))
1154         self.assertTrue(match_str('x>1K', {'x': 1200}))
1155         self.assertFalse(match_str('x>2K', {'x': 1200}))
1156         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
1157         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
1158         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
1159         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
1160         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
1161         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
1162         self.assertFalse(match_str(
1163             'like_count > 100 & dislike_count <? 50 & description',
1164             {'like_count': 90, 'description': 'foo'}))
1165         self.assertTrue(match_str(
1166             'like_count > 100 & dislike_count <? 50 & description',
1167             {'like_count': 190, 'description': 'foo'}))
1168         self.assertFalse(match_str(
1169             'like_count > 100 & dislike_count <? 50 & description',
1170             {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
1171         self.assertFalse(match_str(
1172             'like_count > 100 & dislike_count <? 50 & description',
1173             {'like_count': 190, 'dislike_count': 10}))
1174         self.assertTrue(match_str('is_live', {'is_live': True}))
1175         self.assertFalse(match_str('is_live', {'is_live': False}))
1176         self.assertFalse(match_str('is_live', {'is_live': None}))
1177         self.assertFalse(match_str('is_live', {}))
1178         self.assertFalse(match_str('!is_live', {'is_live': True}))
1179         self.assertTrue(match_str('!is_live', {'is_live': False}))
1180         self.assertTrue(match_str('!is_live', {'is_live': None}))
1181         self.assertTrue(match_str('!is_live', {}))
1182         self.assertTrue(match_str('title', {'title': 'abc'}))
1183         self.assertTrue(match_str('title', {'title': ''}))
1184         self.assertFalse(match_str('!title', {'title': 'abc'}))
1185         self.assertFalse(match_str('!title', {'title': ''}))
1186
1187     def test_parse_dfxp_time_expr(self):
1188         self.assertEqual(parse_dfxp_time_expr(None), None)
1189         self.assertEqual(parse_dfxp_time_expr(''), None)
1190         self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1191         self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1192         self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1193         self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1194         self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1195
1196     def test_dfxp2srt(self):
1197         dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
1198             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1199             <body>
1200                 <div xml:lang="en">
1201                     <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1202                     <p begin="1" end="2">第二行<br/>♪♪</p>
1203                     <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1204                     <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1205                     <p begin="-1" end="-1">Ignore, two</p>
1206                     <p begin="3" dur="-1">Ignored, three</p>
1207                 </div>
1208             </body>
1209             </tt>'''.encode('utf-8')
1210         srt_data = '''1
1211 00:00:00,000 --> 00:00:01,000
1212 The following line contains Chinese characters and special symbols
1213
1214 2
1215 00:00:01,000 --> 00:00:02,000
1216 第二行
1217 ♪♪
1218
1219 3
1220 00:00:02,000 --> 00:00:03,000
1221 Third
1222 Line
1223
1224 '''
1225         self.assertEqual(dfxp2srt(dfxp_data), srt_data)
1226
1227         dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
1228             <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1229             <body>
1230                 <div xml:lang="en">
1231                     <p begin="0" end="1">The first line</p>
1232                 </div>
1233             </body>
1234             </tt>'''.encode('utf-8')
1235         srt_data = '''1
1236 00:00:00,000 --> 00:00:01,000
1237 The first line
1238
1239 '''
1240         self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
1241
1242         dfxp_data_with_style = '''<?xml version="1.0" encoding="utf-8"?>
1243 <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">
1244   <head>
1245     <styling>
1246       <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1247       <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1248       <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1249       <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1250     </styling>
1251   </head>
1252   <body tts:textAlign="center" style="s0">
1253     <div>
1254       <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1255       <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>
1256       <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1257       <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>
1258     </div>
1259   </body>
1260 </tt>'''.encode('utf-8')
1261         srt_data = '''1
1262 00:00:02,080 --> 00:00:05,839
1263 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1264
1265 2
1266 00:00:02,080 --> 00:00:05,839
1267 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1268 </font>part 2</font></b>
1269
1270 3
1271 00:00:05,839 --> 00:00:09,560
1272 <u><font color="lime">line 3
1273 part 3</font></u>
1274
1275 4
1276 00:00:09,560 --> 00:00:12,359
1277 <i><u><font color="yellow"><font color="lime">inner
1278  </font>style</font></u></i>
1279
1280 '''
1281         self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)
1282
1283         dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?>
1284             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1285             <body>
1286                 <div xml:lang="en">
1287                     <p begin="0" end="1">Line 1</p>
1288                     <p begin="1" end="2">第二行</p>
1289                 </div>
1290             </body>
1291             </tt>'''.encode('utf-16')
1292         srt_data = '''1
1293 00:00:00,000 --> 00:00:01,000
1294 Line 1
1295
1296 2
1297 00:00:01,000 --> 00:00:02,000
1298 第二行
1299
1300 '''
1301         self.assertEqual(dfxp2srt(dfxp_data_non_utf8), srt_data)
1302
1303     def test_cli_option(self):
1304         self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1305         self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
1306         self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1307         self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
1308
1309     def test_cli_valueless_option(self):
1310         self.assertEqual(cli_valueless_option(
1311             {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1312         self.assertEqual(cli_valueless_option(
1313             {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
1314         self.assertEqual(cli_valueless_option(
1315             {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1316         self.assertEqual(cli_valueless_option(
1317             {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
1318         self.assertEqual(cli_valueless_option(
1319             {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
1320         self.assertEqual(cli_valueless_option(
1321             {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1322
1323     def test_cli_bool_option(self):
1324         self.assertEqual(
1325             cli_bool_option(
1326                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1327             ['--no-check-certificate', 'true'])
1328         self.assertEqual(
1329             cli_bool_option(
1330                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1331             ['--no-check-certificate=true'])
1332         self.assertEqual(
1333             cli_bool_option(
1334                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1335             ['--check-certificate', 'false'])
1336         self.assertEqual(
1337             cli_bool_option(
1338                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1339             ['--check-certificate=false'])
1340         self.assertEqual(
1341             cli_bool_option(
1342                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1343             ['--check-certificate', 'true'])
1344         self.assertEqual(
1345             cli_bool_option(
1346                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1347             ['--check-certificate=true'])
1348         self.assertEqual(
1349             cli_bool_option(
1350                 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1351             [])
1352
1353     def test_ohdave_rsa_encrypt(self):
1354         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1355         e = 65537
1356
1357         self.assertEqual(
1358             ohdave_rsa_encrypt(b'aa111222', e, N),
1359             '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1360
1361     def test_pkcs1pad(self):
1362         data = [1, 2, 3]
1363         padded_data = pkcs1pad(data, 32)
1364         self.assertEqual(padded_data[:2], [0, 2])
1365         self.assertEqual(padded_data[28:], [0, 1, 2, 3])
1366
1367         self.assertRaises(ValueError, pkcs1pad, data, 8)
1368
1369     def test_encode_base_n(self):
1370         self.assertEqual(encode_base_n(0, 30), '0')
1371         self.assertEqual(encode_base_n(80, 30), '2k')
1372
1373         custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1374         self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1375         self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1376
1377         self.assertRaises(ValueError, encode_base_n, 0, 70)
1378         self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1379
1380     def test_caesar(self):
1381         self.assertEqual(caesar('ace', 'abcdef', 2), 'cea')
1382         self.assertEqual(caesar('cea', 'abcdef', -2), 'ace')
1383         self.assertEqual(caesar('ace', 'abcdef', -2), 'eac')
1384         self.assertEqual(caesar('eac', 'abcdef', 2), 'ace')
1385         self.assertEqual(caesar('ace', 'abcdef', 0), 'ace')
1386         self.assertEqual(caesar('xyz', 'abcdef', 2), 'xyz')
1387         self.assertEqual(caesar('abc', 'acegik', 2), 'ebg')
1388         self.assertEqual(caesar('ebg', 'acegik', -2), 'abc')
1389
1390     def test_rot47(self):
1391         self.assertEqual(rot47('youtube-dl'), r'J@FEF36\5=')
1392         self.assertEqual(rot47('YOUTUBE-DL'), r'*~&%&qt\s{')
1393
1394     def test_urshift(self):
1395         self.assertEqual(urshift(3, 1), 1)
1396         self.assertEqual(urshift(-3, 1), 2147483646)
1397
1398     def test_get_element_by_class(self):
1399         html = '''
1400             <span class="foo bar">nice</span>
1401         '''
1402
1403         self.assertEqual(get_element_by_class('foo', html), 'nice')
1404         self.assertEqual(get_element_by_class('no-such-class', html), None)
1405
1406     def test_get_element_by_attribute(self):
1407         html = '''
1408             <span class="foo bar">nice</span>
1409         '''
1410
1411         self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
1412         self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
1413         self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
1414
1415         html = '''
1416             <div itemprop="author" itemscope>foo</div>
1417         '''
1418
1419         self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo')
1420
1421     def test_get_elements_by_class(self):
1422         html = '''
1423             <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1424         '''
1425
1426         self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
1427         self.assertEqual(get_elements_by_class('no-such-class', html), [])
1428
1429     def test_get_elements_by_attribute(self):
1430         html = '''
1431             <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1432         '''
1433
1434         self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
1435         self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
1436         self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
1437
1438
1439 if __name__ == '__main__':
1440     unittest.main()