_ 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     get_element_by_attribute,
  38     get_elements_by_class,
  39     get_elements_by_attribute,
  40     InAdvancePagedList,
  41     intlist_to_bytes,
  42     is_html,
  43     js_to_json,
  44     limit_length,
  45     mimetype2ext,
  46     month_by_name,
  47     ohdave_rsa_encrypt,
  48     OnDemandPagedList,
  49     orderedSet,
  50     parse_age_limit,
  51     parse_duration,
  52     parse_filesize,
  53     parse_count,
  54     parse_iso8601,
  55     pkcs1pad,
  56     read_batch_urls,
  57     sanitize_filename,
  58     sanitize_path,
  59     expand_path,
  60     prepend_extension,
  61     replace_extension,
  62     remove_start,
  63     remove_end,
  64     remove_quotes,
  65     shell_quote,
  66     smuggle_url,
  67     str_to_int,
  68     strip_jsonp,
  69     timeconvert,
  70     unescapeHTML,
  71     unified_strdate,
  72     unified_timestamp,
  73     unsmuggle_url,
  74     uppercase_escape,
  75     lowercase_escape,
  76     url_basename,
  77     base_url,
  78     urljoin,
  79     urlencode_postdata,
  80     urshift,
  81     update_url_query,
  82     version_tuple,
  83     xpath_with_ns,
  84     xpath_element,
  85     xpath_text,
  86     xpath_attr,
  87     render_table,
  88     match_str,
  89     parse_dfxp_time_expr,
  90     dfxp2srt,
  91     cli_option,
  92     cli_valueless_option,
  93     cli_bool_option,
  94     parse_codecs,
  95 )
  96 from youtube_dl.compat import (
  97     compat_chr,
  98     compat_etree_fromstring,
  99     compat_getenv,
 100     compat_setenv,
 101     compat_urlparse,
 102     compat_parse_qs,
 103 )
 104
 105
 106 class TestUtil(unittest.TestCase):
 107     def test_timeconvert(self):
 108         self.assertTrue(timeconvert('') is None)
 109         self.assertTrue(timeconvert('bougrg') is None)
 110
 111     def test_sanitize_filename(self):
 112         self.assertEqual(sanitize_filename('abc'), 'abc')
 113         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
 114
 115         self.assertEqual(sanitize_filename('123'), '123')
 116
 117         self.assertEqual('abc_de', sanitize_filename('abc/de'))
 118         self.assertFalse('/' in sanitize_filename('abc/de///'))
 119
 120         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
 121         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
 122         self.assertEqual('yes no', sanitize_filename('yes? no'))
 123         self.assertEqual('this - that', sanitize_filename('this: that'))
 124
 125         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
 126         aumlaut = 'ä'
 127         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
 128         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
 129         self.assertEqual(sanitize_filename(tests), tests)
 130
 131         self.assertEqual(
 132             sanitize_filename('New World record at 0:12:34'),
 133             'New World record at 0_12_34')
 134
 135         self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
 136         self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
 137         self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
 138         self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
 139
 140         forbidden = '"\0\\/'
 141         for fc in forbidden:
 142             for fbc in forbidden:
 143                 self.assertTrue(fbc not in sanitize_filename(fc))
 144
 145     def test_sanitize_filename_restricted(self):
 146         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
 147         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
 148
 149         self.assertEqual(sanitize_filename('123', restricted=True), '123')
 150
 151         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
 152         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
 153
 154         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
 155         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
 156         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
 157         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
 158
 159         tests = 'aäb\u4e2d\u56fd\u7684c'
 160         self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
 161         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
 162
 163         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
 164         for fc in forbidden:
 165             for fbc in forbidden:
 166                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
 167
 168         # Handle a common case more neatly
 169         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
 170         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
 171         # .. but make sure the file name is never empty
 172         self.assertTrue(sanitize_filename('-', restricted=True) != '')
 173         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 174
 175         self.assertEqual(sanitize_filename(
 176             'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
 177             'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYPssaaaaaaaeceeeeiiiionooooooooeuuuuuypy')
 178
 179     def test_sanitize_ids(self):
 180         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 181         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 182         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 183
 184     def test_sanitize_path(self):
 185         if sys.platform != 'win32':
 186             return
 187
 188         self.assertEqual(sanitize_path('abc'), 'abc')
 189         self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
 190         self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
 191         self.assertEqual(sanitize_path('abc|def'), 'abc#def')
 192         self.assertEqual(sanitize_path('<>:"|?*'), '#######')
 193         self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
 194         self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
 195
 196         self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
 197         self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
 198
 199         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 200         self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
 201         self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
 202         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 203
 204         self.assertEqual(
 205             sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
 206             'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
 207
 208         self.assertEqual(
 209             sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
 210             'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
 211         self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
 212         self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
 213         self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
 214
 215         self.assertEqual(sanitize_path('../abc'), '..\\abc')
 216         self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
 217         self.assertEqual(sanitize_path('./abc'), 'abc')
 218         self.assertEqual(sanitize_path('./../abc'), '..\\abc')
 219
 220     def test_expand_path(self):
 221         def env(var):
 222             return '%{0}%'.format(var) if sys.platform == 'win32' else '${0}'.format(var)
 223
 224         compat_setenv('YOUTUBE_DL_EXPATH_PATH', 'expanded')
 225         self.assertEqual(expand_path(env('YOUTUBE_DL_EXPATH_PATH')), 'expanded')
 226         self.assertEqual(expand_path(env('HOME')), compat_getenv('HOME'))
 227         self.assertEqual(expand_path('~'), compat_getenv('HOME'))
 228         self.assertEqual(
 229             expand_path('~/%s' % env('YOUTUBE_DL_EXPATH_PATH')),
 230             '%s/expanded' % compat_getenv('HOME'))
 231
 232     def test_prepend_extension(self):
 233         self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
 234         self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
 235         self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 236         self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
 237         self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
 238         self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
 239
 240     def test_replace_extension(self):
 241         self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
 242         self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
 243         self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 244         self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
 245         self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
 246         self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
 247
 248     def test_remove_start(self):
 249         self.assertEqual(remove_start(None, 'A - '), None)
 250         self.assertEqual(remove_start('A - B', 'A - '), 'B')
 251         self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
 252
 253     def test_remove_end(self):
 254         self.assertEqual(remove_end(None, ' - B'), None)
 255         self.assertEqual(remove_end('A - B', ' - B'), 'A')
 256         self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
 257
 258     def test_remove_quotes(self):
 259         self.assertEqual(remove_quotes(None), None)
 260         self.assertEqual(remove_quotes('"'), '"')
 261         self.assertEqual(remove_quotes("'"), "'")
 262         self.assertEqual(remove_quotes(';'), ';')
 263         self.assertEqual(remove_quotes('";'), '";')
 264         self.assertEqual(remove_quotes('""'), '')
 265         self.assertEqual(remove_quotes('";"'), ';')
 266
 267     def test_ordered_set(self):
 268         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
 269         self.assertEqual(orderedSet([]), [])
 270         self.assertEqual(orderedSet([1]), [1])
 271         # keep the list ordered
 272         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
 273
 274     def test_unescape_html(self):
 275         self.assertEqual(unescapeHTML('%20;'), '%20;')
 276         self.assertEqual(unescapeHTML('&#x2F;'), '/')
 277         self.assertEqual(unescapeHTML('&#47;'), '/')
 278         self.assertEqual(unescapeHTML('&eacute;'), 'é')
 279         self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
 280         # HTML5 entities
 281         self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')
 282
 283     def test_date_from_str(self):
 284         self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
 285         self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
 286         self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
 287         self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
 288         self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
 289
 290     def test_daterange(self):
 291         _20century = DateRange("19000101", "20000101")
 292         self.assertFalse("17890714" in _20century)
 293         _ac = DateRange("00010101")
 294         self.assertTrue("19690721" in _ac)
 295         _firstmilenium = DateRange(end="10000101")
 296         self.assertTrue("07110427" in _firstmilenium)
 297
 298     def test_unified_dates(self):
 299         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
 300         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
 301         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
 302         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
 303         self.assertEqual(unified_strdate('1968 12 10'), '19681210')
 304         self.assertEqual(unified_strdate('1968-12-10'), '19681210')
 305         self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
 306         self.assertEqual(
 307             unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
 308             '20141126')
 309         self.assertEqual(
 310             unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
 311             '20150202')
 312         self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
 313         self.assertEqual(unified_strdate('25-09-2014'), '20140925')
 314         self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
 315         self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
 316         self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
 317         self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
 318         self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
 319         self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
 320
 321     def test_unified_timestamps(self):
 322         self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
 323         self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
 324         self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
 325         self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
 326         self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
 327         self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
 328         self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
 329         self.assertEqual(
 330             unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
 331             1417001400)
 332         self.assertEqual(
 333             unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
 334             1422902860)
 335         self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
 336         self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
 337         self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
 338         self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
 339         self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
 340         self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
 341         self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
 342
 343     def test_determine_ext(self):
 344         self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
 345         self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
 346         self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
 347         self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
 348         self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
 349
 350     def test_find_xpath_attr(self):
 351         testxml = '''<root>
 352             <node/>
 353             <node x="a"/>
 354             <node x="a" y="c" />
 355             <node x="b" y="d" />
 356             <node x="" />
 357         </root>'''
 358         doc = compat_etree_fromstring(testxml)
 359
 360         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
 361         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
 362         self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
 363         self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
 364         self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
 365         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
 366         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
 367         self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
 368         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
 369         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
 370         self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
 371
 372     def test_xpath_with_ns(self):
 373         testxml = '''<root xmlns:media="http://example.com/">
 374             <media:song>
 375                 <media:author>The Author</media:author>
 376                 <url>http://server.com/download.mp3</url>
 377             </media:song>
 378         </root>'''
 379         doc = compat_etree_fromstring(testxml)
 380         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
 381         self.assertTrue(find('media:song') is not None)
 382         self.assertEqual(find('media:song/media:author').text, 'The Author')
 383         self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
 384
 385     def test_xpath_element(self):
 386         doc = xml.etree.ElementTree.Element('root')
 387         div = xml.etree.ElementTree.SubElement(doc, 'div')
 388         p = xml.etree.ElementTree.SubElement(div, 'p')
 389         p.text = 'Foo'
 390         self.assertEqual(xpath_element(doc, 'div/p'), p)
 391         self.assertEqual(xpath_element(doc, ['div/p']), p)
 392         self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
 393         self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
 394         self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
 395         self.assertTrue(xpath_element(doc, 'div/bar') is None)
 396         self.assertTrue(xpath_element(doc, ['div/bar']) is None)
 397         self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
 398         self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
 399         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
 400         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
 401
 402     def test_xpath_text(self):
 403         testxml = '''<root>
 404             <div>
 405                 <p>Foo</p>
 406             </div>
 407         </root>'''
 408         doc = compat_etree_fromstring(testxml)
 409         self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
 410         self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
 411         self.assertTrue(xpath_text(doc, 'div/bar') is None)
 412         self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
 413
 414     def test_xpath_attr(self):
 415         testxml = '''<root>
 416             <div>
 417                 <p x="a">Foo</p>
 418             </div>
 419         </root>'''
 420         doc = compat_etree_fromstring(testxml)
 421         self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
 422         self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
 423         self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
 424         self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
 425         self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
 426         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
 427         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
 428
 429     def test_smuggle_url(self):
 430         data = {"ö": "ö", "abc": [3]}
 431         url = 'https://foo.bar/baz?x=y#a'
 432         smug_url = smuggle_url(url, data)
 433         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
 434         self.assertEqual(url, unsmug_url)
 435         self.assertEqual(data, unsmug_data)
 436
 437         res_url, res_data = unsmuggle_url(url)
 438         self.assertEqual(res_url, url)
 439         self.assertEqual(res_data, None)
 440
 441         smug_url = smuggle_url(url, {'a': 'b'})
 442         smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
 443         res_url, res_data = unsmuggle_url(smug_smug_url)
 444         self.assertEqual(res_url, url)
 445         self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
 446
 447     def test_shell_quote(self):
 448         args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
 449         self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
 450
 451     def test_str_to_int(self):
 452         self.assertEqual(str_to_int('123,456'), 123456)
 453         self.assertEqual(str_to_int('123.456'), 123456)
 454
 455     def test_url_basename(self):
 456         self.assertEqual(url_basename('http://foo.de/'), '')
 457         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
 458         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
 459         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
 460         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
 461         self.assertEqual(
 462             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
 463             'trailer.mp4')
 464
 465     def test_base_url(self):
 466         self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
 467         self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
 468         self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
 469         self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
 470         self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
 471
 472     def test_urljoin(self):
 473         self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 474         self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 475         self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 476         self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 477         self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
 478         self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 479         self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 480         self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 481         self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 482         self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
 483         self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 484         self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
 485         self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 486         self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
 487         self.assertEqual(urljoin('http://foo.de/', None), None)
 488         self.assertEqual(urljoin('http://foo.de/', ''), None)
 489         self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
 490         self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
 491
 492     def test_parse_age_limit(self):
 493         self.assertEqual(parse_age_limit(None), None)
 494         self.assertEqual(parse_age_limit(False), None)
 495         self.assertEqual(parse_age_limit('invalid'), None)
 496         self.assertEqual(parse_age_limit(0), 0)
 497         self.assertEqual(parse_age_limit(18), 18)
 498         self.assertEqual(parse_age_limit(21), 21)
 499         self.assertEqual(parse_age_limit(22), None)
 500         self.assertEqual(parse_age_limit('18'), 18)
 501         self.assertEqual(parse_age_limit('18+'), 18)
 502         self.assertEqual(parse_age_limit('PG-13'), 13)
 503         self.assertEqual(parse_age_limit('TV-14'), 14)
 504         self.assertEqual(parse_age_limit('TV-MA'), 17)
 505
 506     def test_parse_duration(self):
 507         self.assertEqual(parse_duration(None), None)
 508         self.assertEqual(parse_duration(False), None)
 509         self.assertEqual(parse_duration('invalid'), None)
 510         self.assertEqual(parse_duration('1'), 1)
 511         self.assertEqual(parse_duration('1337:12'), 80232)
 512         self.assertEqual(parse_duration('9:12:43'), 33163)
 513         self.assertEqual(parse_duration('12:00'), 720)
 514         self.assertEqual(parse_duration('00:01:01'), 61)
 515         self.assertEqual(parse_duration('x:y'), None)
 516         self.assertEqual(parse_duration('3h11m53s'), 11513)
 517         self.assertEqual(parse_duration('3h 11m 53s'), 11513)
 518         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
 519         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
 520         self.assertEqual(parse_duration('62m45s'), 3765)
 521         self.assertEqual(parse_duration('6m59s'), 419)
 522         self.assertEqual(parse_duration('49s'), 49)
 523         self.assertEqual(parse_duration('0h0m0s'), 0)
 524         self.assertEqual(parse_duration('0m0s'), 0)
 525         self.assertEqual(parse_duration('0s'), 0)
 526         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
 527         self.assertEqual(parse_duration('T30M38S'), 1838)
 528         self.assertEqual(parse_duration('5 s'), 5)
 529         self.assertEqual(parse_duration('3 min'), 180)
 530         self.assertEqual(parse_duration('2.5 hours'), 9000)
 531         self.assertEqual(parse_duration('02:03:04'), 7384)
 532         self.assertEqual(parse_duration('01:02:03:04'), 93784)
 533         self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
 534         self.assertEqual(parse_duration('87 Min.'), 5220)
 535         self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
 536         self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
 537
 538     def test_fix_xml_ampersands(self):
 539         self.assertEqual(
 540             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 541         self.assertEqual(
 542             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 543             '"&amp;x=y&amp;wrong;&amp;z=a')
 544         self.assertEqual(
 545             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 546             '&amp;&apos;&gt;&lt;&quot;')
 547         self.assertEqual(
 548             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 549         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 550
 551     def test_paged_list(self):
 552         def testPL(size, pagesize, sliceargs, expected):
 553             def get_page(pagenum):
 554                 firstid = pagenum * pagesize
 555                 upto = min(size, pagenum * pagesize + pagesize)
 556                 for i in range(firstid, upto):
 557                     yield i
 558
 559             pl = OnDemandPagedList(get_page, pagesize)
 560             got = pl.getslice(*sliceargs)
 561             self.assertEqual(got, expected)
 562
 563             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
 564             got = iapl.getslice(*sliceargs)
 565             self.assertEqual(got, expected)
 566
 567         testPL(5, 2, (), [0, 1, 2, 3, 4])
 568         testPL(5, 2, (1,), [1, 2, 3, 4])
 569         testPL(5, 2, (2,), [2, 3, 4])
 570         testPL(5, 2, (4,), [4])
 571         testPL(5, 2, (0, 3), [0, 1, 2])
 572         testPL(5, 2, (1, 4), [1, 2, 3])
 573         testPL(5, 2, (2, 99), [2, 3, 4])
 574         testPL(5, 2, (20, 99), [])
 575
 576     def test_read_batch_urls(self):
 577         f = io.StringIO('''\xef\xbb\xbf foo
 578             bar\r
 579             baz
 580             # More after this line\r
 581             ; or after this
 582             bam''')
 583         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
 584
 585     def test_urlencode_postdata(self):
 586         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
 587         self.assertTrue(isinstance(data, bytes))
 588
 589     def test_update_url_query(self):
 590         def query_dict(url):
 591             return compat_parse_qs(compat_urlparse.urlparse(url).query)
 592         self.assertEqual(query_dict(update_url_query(
 593             'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
 594             query_dict('http://example.com/path?quality=HD&format=mp4'))
 595         self.assertEqual(query_dict(update_url_query(
 596             'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
 597             query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
 598         self.assertEqual(query_dict(update_url_query(
 599             'http://example.com/path', {'fields': 'id,formats,subtitles'})),
 600             query_dict('http://example.com/path?fields=id,formats,subtitles'))
 601         self.assertEqual(query_dict(update_url_query(
 602             'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
 603             query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
 604         self.assertEqual(query_dict(update_url_query(
 605             'http://example.com/path?manifest=f4m', {'manifest': []})),
 606             query_dict('http://example.com/path'))
 607         self.assertEqual(query_dict(update_url_query(
 608             'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
 609             query_dict('http://example.com/path?system=LINUX'))
 610         self.assertEqual(query_dict(update_url_query(
 611             'http://example.com/path', {'fields': b'id,formats,subtitles'})),
 612             query_dict('http://example.com/path?fields=id,formats,subtitles'))
 613         self.assertEqual(query_dict(update_url_query(
 614             'http://example.com/path', {'width': 1080, 'height': 720})),
 615             query_dict('http://example.com/path?width=1080&height=720'))
 616         self.assertEqual(query_dict(update_url_query(
 617             'http://example.com/path', {'bitrate': 5020.43})),
 618             query_dict('http://example.com/path?bitrate=5020.43'))
 619         self.assertEqual(query_dict(update_url_query(
 620             'http://example.com/path', {'test': '第二行тест'})),
 621             query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
 622
 623     def test_dict_get(self):
 624         FALSE_VALUES = {
 625             'none': None,
 626             'false': False,
 627             'zero': 0,
 628             'empty_string': '',
 629             'empty_list': [],
 630         }
 631         d = FALSE_VALUES.copy()
 632         d['a'] = 42
 633         self.assertEqual(dict_get(d, 'a'), 42)
 634         self.assertEqual(dict_get(d, 'b'), None)
 635         self.assertEqual(dict_get(d, 'b', 42), 42)
 636         self.assertEqual(dict_get(d, ('a', )), 42)
 637         self.assertEqual(dict_get(d, ('b', 'a', )), 42)
 638         self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
 639         self.assertEqual(dict_get(d, ('b', 'c', )), None)
 640         self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
 641         for key, false_value in FALSE_VALUES.items():
 642             self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
 643             self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
 644
 645     def test_encode_compat_str(self):
 646         self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
 647         self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
 648
 649     def test_parse_iso8601(self):
 650         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
 651         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
 652         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
 653         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
 654         self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
 655         self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
 656
 657     def test_strip_jsonp(self):
 658         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
 659         d = json.loads(stripped)
 660         self.assertEqual(d, [{"id": "532cb", "x": 3}])
 661
 662         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
 663         d = json.loads(stripped)
 664         self.assertEqual(d, {'STATUS': 'OK'})
 665
 666         stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
 667         d = json.loads(stripped)
 668         self.assertEqual(d, {'status': 'success'})
 669
 670     def test_uppercase_escape(self):
 671         self.assertEqual(uppercase_escape('aä'), 'aä')
 672         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
 673
 674     def test_lowercase_escape(self):
 675         self.assertEqual(lowercase_escape('aä'), 'aä')
 676         self.assertEqual(lowercase_escape('\\u0026'), '&')
 677
 678     def test_limit_length(self):
 679         self.assertEqual(limit_length(None, 12), None)
 680         self.assertEqual(limit_length('foo', 12), 'foo')
 681         self.assertTrue(
 682             limit_length('foo bar baz asd', 12).startswith('foo bar'))
 683         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
 684
 685     def test_mimetype2ext(self):
 686         self.assertEqual(mimetype2ext(None), None)
 687         self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
 688         self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
 689         self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
 690         self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
 691         self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
 692
 693     def test_month_by_name(self):
 694         self.assertEqual(month_by_name(None), None)
 695         self.assertEqual(month_by_name('December', 'en'), 12)
 696         self.assertEqual(month_by_name('décembre', 'fr'), 12)
 697         self.assertEqual(month_by_name('December'), 12)
 698         self.assertEqual(month_by_name('décembre'), None)
 699         self.assertEqual(month_by_name('Unknown', 'unknown'), None)
 700
 701     def test_parse_codecs(self):
 702         self.assertEqual(parse_codecs(''), {})
 703         self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
 704             'vcodec': 'avc1.77.30',
 705             'acodec': 'mp4a.40.2',
 706         })
 707         self.assertEqual(parse_codecs('mp4a.40.2'), {
 708             'vcodec': 'none',
 709             'acodec': 'mp4a.40.2',
 710         })
 711         self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
 712             'vcodec': 'avc1.42001e',
 713             'acodec': 'mp4a.40.5',
 714         })
 715         self.assertEqual(parse_codecs('avc3.640028'), {
 716             'vcodec': 'avc3.640028',
 717             'acodec': 'none',
 718         })
 719         self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
 720             'vcodec': 'h264',
 721             'acodec': 'aac',
 722         })
 723
 724     def test_escape_rfc3986(self):
 725         reserved = "!*'();:@&=+$,/?#[]"
 726         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
 727         self.assertEqual(escape_rfc3986(reserved), reserved)
 728         self.assertEqual(escape_rfc3986(unreserved), unreserved)
 729         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
 730         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
 731         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
 732         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
 733
 734     def test_escape_url(self):
 735         self.assertEqual(
 736             escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
 737             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
 738         )
 739         self.assertEqual(
 740             escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
 741             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
 742         )
 743         self.assertEqual(
 744             escape_url('http://тест.рф/фрагмент'),
 745             'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
 746         )
 747         self.assertEqual(
 748             escape_url('http://тест.рф/абв?абв=абв#абв'),
 749             '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'
 750         )
 751         self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
 752
 753     def test_js_to_json_realworld(self):
 754         inp = '''{
 755             'clip':{'provider':'pseudo'}
 756         }'''
 757         self.assertEqual(js_to_json(inp), '''{
 758             "clip":{"provider":"pseudo"}
 759         }''')
 760         json.loads(js_to_json(inp))
 761
 762         inp = '''{
 763             'playlist':[{'controls':{'all':null}}]
 764         }'''
 765         self.assertEqual(js_to_json(inp), '''{
 766             "playlist":[{"controls":{"all":null}}]
 767         }''')
 768
 769         inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
 770         self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
 771
 772         inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
 773         json_code = js_to_json(inp)
 774         self.assertEqual(json.loads(json_code), json.loads(inp))
 775
 776         inp = '''{
 777             0:{src:'skipped', type: 'application/dash+xml'},
 778             1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
 779         }'''
 780         self.assertEqual(js_to_json(inp), '''{
 781             "0":{"src":"skipped", "type": "application/dash+xml"},
 782             "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
 783         }''')
 784
 785         inp = '''{"foo":101}'''
 786         self.assertEqual(js_to_json(inp), '''{"foo":101}''')
 787
 788         inp = '''{"duration": "00:01:07"}'''
 789         self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
 790
 791     def test_js_to_json_edgecases(self):
 792         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
 793         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
 794
 795         on = js_to_json('{"abc": true}')
 796         self.assertEqual(json.loads(on), {'abc': True})
 797
 798         # Ignore JavaScript code as well
 799         on = js_to_json('''{
 800             "x": 1,
 801             y: "a",
 802             z: some.code
 803         }''')
 804         d = json.loads(on)
 805         self.assertEqual(d['x'], 1)
 806         self.assertEqual(d['y'], 'a')
 807
 808         on = js_to_json('["abc", "def",]')
 809         self.assertEqual(json.loads(on), ['abc', 'def'])
 810
 811         on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
 812         self.assertEqual(json.loads(on), ['abc', 'def'])
 813
 814         on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
 815         self.assertEqual(json.loads(on), ['abc', 'def'])
 816
 817         on = js_to_json('{"abc": "def",}')
 818         self.assertEqual(json.loads(on), {'abc': 'def'})
 819
 820         on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
 821         self.assertEqual(json.loads(on), {'abc': 'def'})
 822
 823         on = js_to_json('{ 0: /* " \n */ ",]" , }')
 824         self.assertEqual(json.loads(on), {'0': ',]'})
 825
 826         on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
 827         self.assertEqual(json.loads(on), {'0': ',]'})
 828
 829         on = js_to_json('{ 0: // comment\n1 }')
 830         self.assertEqual(json.loads(on), {'0': 1})
 831
 832         on = js_to_json(r'["<p>x<\/p>"]')
 833         self.assertEqual(json.loads(on), ['<p>x</p>'])
 834
 835         on = js_to_json(r'["\xaa"]')
 836         self.assertEqual(json.loads(on), ['\u00aa'])
 837
 838         on = js_to_json("['a\\\nb']")
 839         self.assertEqual(json.loads(on), ['ab'])
 840
 841         on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
 842         self.assertEqual(json.loads(on), ['ab'])
 843
 844         on = js_to_json('{0xff:0xff}')
 845         self.assertEqual(json.loads(on), {'255': 255})
 846
 847         on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
 848         self.assertEqual(json.loads(on), {'255': 255})
 849
 850         on = js_to_json('{077:077}')
 851         self.assertEqual(json.loads(on), {'63': 63})
 852
 853         on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
 854         self.assertEqual(json.loads(on), {'63': 63})
 855
 856         on = js_to_json('{42:42}')
 857         self.assertEqual(json.loads(on), {'42': 42})
 858
 859         on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
 860         self.assertEqual(json.loads(on), {'42': 42})
 861
 862     def test_extract_attributes(self):
 863         self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
 864         self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
 865         self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
 866         self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
 867         self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
 868         self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
 869         self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
 870         self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'})  # XML
 871         self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
 872         self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'})  # HTML 3.2
 873         self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'})  # HTML 4.0
 874         self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
 875         self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
 876         self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
 877         self.assertEqual(extract_attributes('<e x >'), {'x': None})
 878         self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
 879         self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
 880         self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
 881         self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
 882         self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
 883         self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
 884         self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
 885         self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'})  # Names lowercased
 886         self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
 887         self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
 888         self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
 889         self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
 890         self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
 891         # "Narrow" Python builds don't support unicode code points outside BMP.
 892         try:
 893             compat_chr(0x10000)
 894             supports_outside_bmp = True
 895         except ValueError:
 896             supports_outside_bmp = False
 897         if supports_outside_bmp:
 898             self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
 899
 900     def test_clean_html(self):
 901         self.assertEqual(clean_html('a:\nb'), 'a: b')
 902         self.assertEqual(clean_html('a:\n   "b"'), 'a:    "b"')
 903         self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
 904
 905     def test_intlist_to_bytes(self):
 906         self.assertEqual(
 907             intlist_to_bytes([0, 1, 127, 128, 255]),
 908             b'\x00\x01\x7f\x80\xff')
 909
 910     def test_args_to_str(self):
 911         self.assertEqual(
 912             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
 913             'foo ba/r -baz \'2 be\' \'\''
 914         )
 915
 916     def test_parse_filesize(self):
 917         self.assertEqual(parse_filesize(None), None)
 918         self.assertEqual(parse_filesize(''), None)
 919         self.assertEqual(parse_filesize('91 B'), 91)
 920         self.assertEqual(parse_filesize('foobar'), None)
 921         self.assertEqual(parse_filesize('2 MiB'), 2097152)
 922         self.assertEqual(parse_filesize('5 GB'), 5000000000)
 923         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
 924         self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
 925         self.assertEqual(parse_filesize('1,24 KB'), 1240)
 926         self.assertEqual(parse_filesize('1,24 kb'), 1240)
 927         self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
 928
 929     def test_parse_count(self):
 930         self.assertEqual(parse_count(None), None)
 931         self.assertEqual(parse_count(''), None)
 932         self.assertEqual(parse_count('0'), 0)
 933         self.assertEqual(parse_count('1000'), 1000)
 934         self.assertEqual(parse_count('1.000'), 1000)
 935         self.assertEqual(parse_count('1.1k'), 1100)
 936         self.assertEqual(parse_count('1.1kk'), 1100000)
 937         self.assertEqual(parse_count('1.1kk '), 1100000)
 938         self.assertEqual(parse_count('1.1kk views'), 1100000)
 939
 940     def test_version_tuple(self):
 941         self.assertEqual(version_tuple('1'), (1,))
 942         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
 943         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
 944
 945     def test_detect_exe_version(self):
 946         self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
 947 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
 948 configuration: --prefix=/usr --extra-'''), '1.2.1')
 949         self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
 950 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
 951         self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
 952 Trying to open render node...
 953 Success at /dev/dri/renderD128.
 954 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
 955
 956     def test_age_restricted(self):
 957         self.assertFalse(age_restricted(None, 10))  # unrestricted content
 958         self.assertFalse(age_restricted(1, None))  # unrestricted policy
 959         self.assertFalse(age_restricted(8, 10))
 960         self.assertTrue(age_restricted(18, 14))
 961         self.assertFalse(age_restricted(18, 18))
 962
 963     def test_is_html(self):
 964         self.assertFalse(is_html(b'\x49\x44\x43<html'))
 965         self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
 966         self.assertTrue(is_html(  # UTF-8 with BOM
 967             b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
 968         self.assertTrue(is_html(  # UTF-16-LE
 969             b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
 970         ))
 971         self.assertTrue(is_html(  # UTF-16-BE
 972             b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
 973         ))
 974         self.assertTrue(is_html(  # UTF-32-BE
 975             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'))
 976         self.assertTrue(is_html(  # UTF-32-LE
 977             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'))
 978
 979     def test_render_table(self):
 980         self.assertEqual(
 981             render_table(
 982                 ['a', 'bcd'],
 983                 [[123, 4], [9999, 51]]),
 984             'a    bcd\n'
 985             '123  4\n'
 986             '9999 51')
 987
 988     def test_match_str(self):
 989         self.assertRaises(ValueError, match_str, 'xy>foobar', {})
 990         self.assertFalse(match_str('xy', {'x': 1200}))
 991         self.assertTrue(match_str('!xy', {'x': 1200}))
 992         self.assertTrue(match_str('x', {'x': 1200}))
 993         self.assertFalse(match_str('!x', {'x': 1200}))
 994         self.assertTrue(match_str('x', {'x': 0}))
 995         self.assertFalse(match_str('x>0', {'x': 0}))
 996         self.assertFalse(match_str('x>0', {}))
 997         self.assertTrue(match_str('x>?0', {}))
 998         self.assertTrue(match_str('x>1K', {'x': 1200}))
 999         self.assertFalse(match_str('x>2K', {'x': 1200}))
1000         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
1001         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
1002         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
1003         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
1004         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
1005         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
1006         self.assertFalse(match_str(
1007             'like_count > 100 & dislike_count <? 50 & description',
1008             {'like_count': 90, 'description': 'foo'}))
1009         self.assertTrue(match_str(
1010             'like_count > 100 & dislike_count <? 50 & description',
1011             {'like_count': 190, 'description': 'foo'}))
1012         self.assertFalse(match_str(
1013             'like_count > 100 & dislike_count <? 50 & description',
1014             {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
1015         self.assertFalse(match_str(
1016             'like_count > 100 & dislike_count <? 50 & description',
1017             {'like_count': 190, 'dislike_count': 10}))
1018
1019     def test_parse_dfxp_time_expr(self):
1020         self.assertEqual(parse_dfxp_time_expr(None), None)
1021         self.assertEqual(parse_dfxp_time_expr(''), None)
1022         self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1023         self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1024         self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1025         self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1026         self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1027
1028     def test_dfxp2srt(self):
1029         dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
1030             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1031             <body>
1032                 <div xml:lang="en">
1033                     <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1034                     <p begin="1" end="2">第二行<br/>♪♪</p>
1035                     <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1036                     <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1037                     <p begin="-1" end="-1">Ignore, two</p>
1038                     <p begin="3" dur="-1">Ignored, three</p>
1039                 </div>
1040             </body>
1041             </tt>'''
1042         srt_data = '''1
1043 00:00:00,000 --> 00:00:01,000
1044 The following line contains Chinese characters and special symbols
1045
1046 2
1047 00:00:01,000 --> 00:00:02,000
1048 第二行
1049 ♪♪
1050
1051 3
1052 00:00:02,000 --> 00:00:03,000
1053 Third
1054 Line
1055
1056 '''
1057         self.assertEqual(dfxp2srt(dfxp_data), srt_data)
1058
1059         dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
1060             <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1061             <body>
1062                 <div xml:lang="en">
1063                     <p begin="0" end="1">The first line</p>
1064                 </div>
1065             </body>
1066             </tt>'''
1067         srt_data = '''1
1068 00:00:00,000 --> 00:00:01,000
1069 The first line
1070
1071 '''
1072         self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
1073
1074         dfxp_data_with_style = '''<?xml version="1.0" encoding="utf-8"?>
1075 <tt xmlns="http://www.w3.org/2006/10/ttaf1" xmlns:ttp="http://www.w3.org/2006/10/ttaf1#parameter" ttp:timeBase="media" xmlns:tts="http://www.w3.org/2006/10/ttaf1#style" xml:lang="en" xmlns:ttm="http://www.w3.org/2006/10/ttaf1#metadata">
1076   <head>
1077     <styling>
1078       <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1079       <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1080       <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1081       <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1082     </styling>
1083   </head>
1084   <body tts:textAlign="center" style="s0">
1085     <div>
1086       <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1087       <p style="s2" begin="00:00:02.08" id="p0" end="00:00:05.84"><span tts:color="lime">part 1<br /></span><span tts:color="cyan">part 2</span></p>
1088       <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1089       <p style="s1" tts:textDecoration="underline" begin="00:00:09.56" id="p2" end="00:00:12.36"><span style="s2" tts:color="lime">inner<br /> </span>style</p>
1090     </div>
1091   </body>
1092 </tt>'''
1093         srt_data = '''1
1094 00:00:02,080 --> 00:00:05,839
1095 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1096
1097 2
1098 00:00:02,080 --> 00:00:05,839
1099 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1100 </font>part 2</font></b>
1101
1102 3
1103 00:00:05,839 --> 00:00:09,560
1104 <u><font color="lime">line 3
1105 part 3</font></u>
1106
1107 4
1108 00:00:09,560 --> 00:00:12,359
1109 <i><u><font color="yellow"><font color="lime">inner
1110  </font>style</font></u></i>
1111
1112 '''
1113         self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)
1114
1115     def test_cli_option(self):
1116         self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1117         self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
1118         self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1119         self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
1120
1121     def test_cli_valueless_option(self):
1122         self.assertEqual(cli_valueless_option(
1123             {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1124         self.assertEqual(cli_valueless_option(
1125             {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
1126         self.assertEqual(cli_valueless_option(
1127             {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1128         self.assertEqual(cli_valueless_option(
1129             {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
1130         self.assertEqual(cli_valueless_option(
1131             {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
1132         self.assertEqual(cli_valueless_option(
1133             {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1134
1135     def test_cli_bool_option(self):
1136         self.assertEqual(
1137             cli_bool_option(
1138                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1139             ['--no-check-certificate', 'true'])
1140         self.assertEqual(
1141             cli_bool_option(
1142                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1143             ['--no-check-certificate=true'])
1144         self.assertEqual(
1145             cli_bool_option(
1146                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1147             ['--check-certificate', 'false'])
1148         self.assertEqual(
1149             cli_bool_option(
1150                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1151             ['--check-certificate=false'])
1152         self.assertEqual(
1153             cli_bool_option(
1154                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1155             ['--check-certificate', 'true'])
1156         self.assertEqual(
1157             cli_bool_option(
1158                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1159             ['--check-certificate=true'])
1160
1161     def test_ohdave_rsa_encrypt(self):
1162         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1163         e = 65537
1164
1165         self.assertEqual(
1166             ohdave_rsa_encrypt(b'aa111222', e, N),
1167             '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1168
1169     def test_pkcs1pad(self):
1170         data = [1, 2, 3]
1171         padded_data = pkcs1pad(data, 32)
1172         self.assertEqual(padded_data[:2], [0, 2])
1173         self.assertEqual(padded_data[28:], [0, 1, 2, 3])
1174
1175         self.assertRaises(ValueError, pkcs1pad, data, 8)
1176
1177     def test_encode_base_n(self):
1178         self.assertEqual(encode_base_n(0, 30), '0')
1179         self.assertEqual(encode_base_n(80, 30), '2k')
1180
1181         custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1182         self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1183         self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1184
1185         self.assertRaises(ValueError, encode_base_n, 0, 70)
1186         self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1187
1188     def test_urshift(self):
1189         self.assertEqual(urshift(3, 1), 1)
1190         self.assertEqual(urshift(-3, 1), 2147483646)
1191
1192     def test_get_element_by_class(self):
1193         html = '''
1194             <span class="foo bar">nice</span>
1195         '''
1196
1197         self.assertEqual(get_element_by_class('foo', html), 'nice')
1198         self.assertEqual(get_element_by_class('no-such-class', html), None)
1199
1200     def test_get_element_by_attribute(self):
1201         html = '''
1202             <span class="foo bar">nice</span>
1203         '''
1204
1205         self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
1206         self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
1207         self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
1208
1209     def test_get_elements_by_class(self):
1210         html = '''
1211             <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1212         '''
1213
1214         self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
1215         self.assertEqual(get_elements_by_class('no-such-class', html), [])
1216
1217     def test_get_elements_by_attribute(self):
1218         html = '''
1219             <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1220         '''
1221
1222         self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
1223         self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
1224         self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
1225
1226
1227 if __name__ == '__main__':
1228     unittest.main()