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