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