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

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