_ 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     DateRange,
  24     detect_exe_version,
  25     determine_ext,
  26     dict_get,
  27     encode_compat_str,
  28     encodeFilename,
  29     escape_rfc3986,
  30     escape_url,
  31     extract_attributes,
  32     ExtractorError,
  33     find_xpath_attr,
  34     fix_xml_ampersands,
  35     InAdvancePagedList,
  36     intlist_to_bytes,
  37     is_html,
  38     js_to_json,
  39     limit_length,
  40     ohdave_rsa_encrypt,
  41     OnDemandPagedList,
  42     orderedSet,
  43     parse_duration,
  44     parse_filesize,
  45     parse_iso8601,
  46     read_batch_urls,
  47     sanitize_filename,
  48     sanitize_path,
  49     prepend_extension,
  50     replace_extension,
  51     remove_quotes,
  52     shell_quote,
  53     smuggle_url,
  54     str_to_int,
  55     strip_jsonp,
  56     struct_unpack,
  57     timeconvert,
  58     unescapeHTML,
  59     unified_strdate,
  60     unsmuggle_url,
  61     uppercase_escape,
  62     lowercase_escape,
  63     url_basename,
  64     urlencode_postdata,
  65     version_tuple,
  66     xpath_with_ns,
  67     xpath_element,
  68     xpath_text,
  69     xpath_attr,
  70     render_table,
  71     match_str,
  72     parse_dfxp_time_expr,
  73     dfxp2srt,
  74     cli_option,
  75     cli_valueless_option,
  76     cli_bool_option,
  77 )
  78 from youtube_dl.compat import (
  79     compat_chr,
  80     compat_etree_fromstring,
  81 )
  82
  83
  84 class TestUtil(unittest.TestCase):
  85     def test_timeconvert(self):
  86         self.assertTrue(timeconvert('') is None)
  87         self.assertTrue(timeconvert('bougrg') is None)
  88
  89     def test_sanitize_filename(self):
  90         self.assertEqual(sanitize_filename('abc'), 'abc')
  91         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
  92
  93         self.assertEqual(sanitize_filename('123'), '123')
  94
  95         self.assertEqual('abc_de', sanitize_filename('abc/de'))
  96         self.assertFalse('/' in sanitize_filename('abc/de///'))
  97
  98         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
  99         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
 100         self.assertEqual('yes no', sanitize_filename('yes? no'))
 101         self.assertEqual('this - that', sanitize_filename('this: that'))
 102
 103         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
 104         aumlaut = 'ä'
 105         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
 106         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
 107         self.assertEqual(sanitize_filename(tests), tests)
 108
 109         self.assertEqual(
 110             sanitize_filename('New World record at 0:12:34'),
 111             'New World record at 0_12_34')
 112
 113         self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
 114         self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
 115         self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
 116         self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
 117
 118         forbidden = '"\0\\/'
 119         for fc in forbidden:
 120             for fbc in forbidden:
 121                 self.assertTrue(fbc not in sanitize_filename(fc))
 122
 123     def test_sanitize_filename_restricted(self):
 124         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
 125         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
 126
 127         self.assertEqual(sanitize_filename('123', restricted=True), '123')
 128
 129         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
 130         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
 131
 132         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
 133         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
 134         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
 135         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
 136
 137         tests = 'a\xe4b\u4e2d\u56fd\u7684c'
 138         self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
 139         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
 140
 141         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
 142         for fc in forbidden:
 143             for fbc in forbidden:
 144                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
 145
 146         # Handle a common case more neatly
 147         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
 148         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
 149         # .. but make sure the file name is never empty
 150         self.assertTrue(sanitize_filename('-', restricted=True) != '')
 151         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 152
 153     def test_sanitize_ids(self):
 154         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 155         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 156         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 157
 158     def test_sanitize_path(self):
 159         if sys.platform != 'win32':
 160             return
 161
 162         self.assertEqual(sanitize_path('abc'), 'abc')
 163         self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
 164         self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
 165         self.assertEqual(sanitize_path('abc|def'), 'abc#def')
 166         self.assertEqual(sanitize_path('<>:"|?*'), '#######')
 167         self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
 168         self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
 169
 170         self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
 171         self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
 172
 173         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 174         self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
 175         self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
 176         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 177
 178         self.assertEqual(
 179             sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
 180             'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
 181
 182         self.assertEqual(
 183             sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
 184             'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
 185         self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
 186         self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
 187         self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
 188
 189         self.assertEqual(sanitize_path('../abc'), '..\\abc')
 190         self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
 191         self.assertEqual(sanitize_path('./abc'), 'abc')
 192         self.assertEqual(sanitize_path('./../abc'), '..\\abc')
 193
 194     def test_prepend_extension(self):
 195         self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
 196         self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
 197         self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 198         self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
 199         self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
 200         self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
 201
 202     def test_replace_extension(self):
 203         self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
 204         self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
 205         self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 206         self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
 207         self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
 208         self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
 209
 210     def test_remove_quotes(self):
 211         self.assertEqual(remove_quotes(None), None)
 212         self.assertEqual(remove_quotes('"'), '"')
 213         self.assertEqual(remove_quotes("'"), "'")
 214         self.assertEqual(remove_quotes(';'), ';')
 215         self.assertEqual(remove_quotes('";'), '";')
 216         self.assertEqual(remove_quotes('""'), '')
 217         self.assertEqual(remove_quotes('";"'), ';')
 218
 219     def test_ordered_set(self):
 220         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
 221         self.assertEqual(orderedSet([]), [])
 222         self.assertEqual(orderedSet([1]), [1])
 223         # keep the list ordered
 224         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
 225
 226     def test_unescape_html(self):
 227         self.assertEqual(unescapeHTML('%20;'), '%20;')
 228         self.assertEqual(unescapeHTML('&#x2F;'), '/')
 229         self.assertEqual(unescapeHTML('&#47;'), '/')
 230         self.assertEqual(unescapeHTML('&eacute;'), 'é')
 231         self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
 232
 233     def test_daterange(self):
 234         _20century = DateRange("19000101", "20000101")
 235         self.assertFalse("17890714" in _20century)
 236         _ac = DateRange("00010101")
 237         self.assertTrue("19690721" in _ac)
 238         _firstmilenium = DateRange(end="10000101")
 239         self.assertTrue("07110427" in _firstmilenium)
 240
 241     def test_unified_dates(self):
 242         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
 243         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
 244         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
 245         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
 246         self.assertEqual(unified_strdate('1968 12 10'), '19681210')
 247         self.assertEqual(unified_strdate('1968-12-10'), '19681210')
 248         self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
 249         self.assertEqual(
 250             unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
 251             '20141126')
 252         self.assertEqual(
 253             unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
 254             '20150202')
 255         self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
 256         self.assertEqual(unified_strdate('25-09-2014'), '20140925')
 257         self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
 258
 259     def test_determine_ext(self):
 260         self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
 261         self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
 262         self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
 263         self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
 264         self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
 265
 266     def test_find_xpath_attr(self):
 267         testxml = '''<root>
 268             <node/>
 269             <node x="a"/>
 270             <node x="a" y="c" />
 271             <node x="b" y="d" />
 272             <node x="" />
 273         </root>'''
 274         doc = compat_etree_fromstring(testxml)
 275
 276         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
 277         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
 278         self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
 279         self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
 280         self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
 281         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
 282         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
 283         self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
 284         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
 285         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
 286         self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
 287
 288     def test_xpath_with_ns(self):
 289         testxml = '''<root xmlns:media="http://example.com/">
 290             <media:song>
 291                 <media:author>The Author</media:author>
 292                 <url>http://server.com/download.mp3</url>
 293             </media:song>
 294         </root>'''
 295         doc = compat_etree_fromstring(testxml)
 296         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
 297         self.assertTrue(find('media:song') is not None)
 298         self.assertEqual(find('media:song/media:author').text, 'The Author')
 299         self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
 300
 301     def test_xpath_element(self):
 302         doc = xml.etree.ElementTree.Element('root')
 303         div = xml.etree.ElementTree.SubElement(doc, 'div')
 304         p = xml.etree.ElementTree.SubElement(div, 'p')
 305         p.text = 'Foo'
 306         self.assertEqual(xpath_element(doc, 'div/p'), p)
 307         self.assertEqual(xpath_element(doc, ['div/p']), p)
 308         self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
 309         self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
 310         self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
 311         self.assertTrue(xpath_element(doc, 'div/bar') is None)
 312         self.assertTrue(xpath_element(doc, ['div/bar']) is None)
 313         self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
 314         self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
 315         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
 316         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
 317
 318     def test_xpath_text(self):
 319         testxml = '''<root>
 320             <div>
 321                 <p>Foo</p>
 322             </div>
 323         </root>'''
 324         doc = compat_etree_fromstring(testxml)
 325         self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
 326         self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
 327         self.assertTrue(xpath_text(doc, 'div/bar') is None)
 328         self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
 329
 330     def test_xpath_attr(self):
 331         testxml = '''<root>
 332             <div>
 333                 <p x="a">Foo</p>
 334             </div>
 335         </root>'''
 336         doc = compat_etree_fromstring(testxml)
 337         self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
 338         self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
 339         self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
 340         self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
 341         self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
 342         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
 343         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
 344
 345     def test_smuggle_url(self):
 346         data = {"ö": "ö", "abc": [3]}
 347         url = 'https://foo.bar/baz?x=y#a'
 348         smug_url = smuggle_url(url, data)
 349         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
 350         self.assertEqual(url, unsmug_url)
 351         self.assertEqual(data, unsmug_data)
 352
 353         res_url, res_data = unsmuggle_url(url)
 354         self.assertEqual(res_url, url)
 355         self.assertEqual(res_data, None)
 356
 357     def test_shell_quote(self):
 358         args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
 359         self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
 360
 361     def test_str_to_int(self):
 362         self.assertEqual(str_to_int('123,456'), 123456)
 363         self.assertEqual(str_to_int('123.456'), 123456)
 364
 365     def test_url_basename(self):
 366         self.assertEqual(url_basename('http://foo.de/'), '')
 367         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
 368         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
 369         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
 370         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
 371         self.assertEqual(
 372             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
 373             'trailer.mp4')
 374
 375     def test_parse_duration(self):
 376         self.assertEqual(parse_duration(None), None)
 377         self.assertEqual(parse_duration(False), None)
 378         self.assertEqual(parse_duration('invalid'), None)
 379         self.assertEqual(parse_duration('1'), 1)
 380         self.assertEqual(parse_duration('1337:12'), 80232)
 381         self.assertEqual(parse_duration('9:12:43'), 33163)
 382         self.assertEqual(parse_duration('12:00'), 720)
 383         self.assertEqual(parse_duration('00:01:01'), 61)
 384         self.assertEqual(parse_duration('x:y'), None)
 385         self.assertEqual(parse_duration('3h11m53s'), 11513)
 386         self.assertEqual(parse_duration('3h 11m 53s'), 11513)
 387         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
 388         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
 389         self.assertEqual(parse_duration('62m45s'), 3765)
 390         self.assertEqual(parse_duration('6m59s'), 419)
 391         self.assertEqual(parse_duration('49s'), 49)
 392         self.assertEqual(parse_duration('0h0m0s'), 0)
 393         self.assertEqual(parse_duration('0m0s'), 0)
 394         self.assertEqual(parse_duration('0s'), 0)
 395         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
 396         self.assertEqual(parse_duration('T30M38S'), 1838)
 397         self.assertEqual(parse_duration('5 s'), 5)
 398         self.assertEqual(parse_duration('3 min'), 180)
 399         self.assertEqual(parse_duration('2.5 hours'), 9000)
 400         self.assertEqual(parse_duration('02:03:04'), 7384)
 401         self.assertEqual(parse_duration('01:02:03:04'), 93784)
 402         self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
 403         self.assertEqual(parse_duration('87 Min.'), 5220)
 404
 405     def test_fix_xml_ampersands(self):
 406         self.assertEqual(
 407             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 408         self.assertEqual(
 409             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 410             '"&amp;x=y&amp;wrong;&amp;z=a')
 411         self.assertEqual(
 412             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 413             '&amp;&apos;&gt;&lt;&quot;')
 414         self.assertEqual(
 415             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 416         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 417
 418     def test_paged_list(self):
 419         def testPL(size, pagesize, sliceargs, expected):
 420             def get_page(pagenum):
 421                 firstid = pagenum * pagesize
 422                 upto = min(size, pagenum * pagesize + pagesize)
 423                 for i in range(firstid, upto):
 424                     yield i
 425
 426             pl = OnDemandPagedList(get_page, pagesize)
 427             got = pl.getslice(*sliceargs)
 428             self.assertEqual(got, expected)
 429
 430             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
 431             got = iapl.getslice(*sliceargs)
 432             self.assertEqual(got, expected)
 433
 434         testPL(5, 2, (), [0, 1, 2, 3, 4])
 435         testPL(5, 2, (1,), [1, 2, 3, 4])
 436         testPL(5, 2, (2,), [2, 3, 4])
 437         testPL(5, 2, (4,), [4])
 438         testPL(5, 2, (0, 3), [0, 1, 2])
 439         testPL(5, 2, (1, 4), [1, 2, 3])
 440         testPL(5, 2, (2, 99), [2, 3, 4])
 441         testPL(5, 2, (20, 99), [])
 442
 443     def test_struct_unpack(self):
 444         self.assertEqual(struct_unpack('!B', b'\x00'), (0,))
 445
 446     def test_read_batch_urls(self):
 447         f = io.StringIO('''\xef\xbb\xbf foo
 448             bar\r
 449             baz
 450             # More after this line\r
 451             ; or after this
 452             bam''')
 453         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
 454
 455     def test_urlencode_postdata(self):
 456         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
 457         self.assertTrue(isinstance(data, bytes))
 458
 459     def test_dict_get(self):
 460         FALSE_VALUES = {
 461             'none': None,
 462             'false': False,
 463             'zero': 0,
 464             'empty_string': '',
 465             'empty_list': [],
 466         }
 467         d = FALSE_VALUES.copy()
 468         d['a'] = 42
 469         self.assertEqual(dict_get(d, 'a'), 42)
 470         self.assertEqual(dict_get(d, 'b'), None)
 471         self.assertEqual(dict_get(d, 'b', 42), 42)
 472         self.assertEqual(dict_get(d, ('a', )), 42)
 473         self.assertEqual(dict_get(d, ('b', 'a', )), 42)
 474         self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
 475         self.assertEqual(dict_get(d, ('b', 'c', )), None)
 476         self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
 477         for key, false_value in FALSE_VALUES.items():
 478             self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
 479             self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
 480
 481     def test_encode_compat_str(self):
 482         self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
 483         self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
 484
 485     def test_parse_iso8601(self):
 486         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
 487         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
 488         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
 489         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
 490         self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
 491         self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
 492
 493     def test_strip_jsonp(self):
 494         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
 495         d = json.loads(stripped)
 496         self.assertEqual(d, [{"id": "532cb", "x": 3}])
 497
 498         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
 499         d = json.loads(stripped)
 500         self.assertEqual(d, {'STATUS': 'OK'})
 501
 502         stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
 503         d = json.loads(stripped)
 504         self.assertEqual(d, {'status': 'success'})
 505
 506     def test_uppercase_escape(self):
 507         self.assertEqual(uppercase_escape('aä'), 'aä')
 508         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
 509
 510     def test_lowercase_escape(self):
 511         self.assertEqual(lowercase_escape('aä'), 'aä')
 512         self.assertEqual(lowercase_escape('\\u0026'), '&')
 513
 514     def test_limit_length(self):
 515         self.assertEqual(limit_length(None, 12), None)
 516         self.assertEqual(limit_length('foo', 12), 'foo')
 517         self.assertTrue(
 518             limit_length('foo bar baz asd', 12).startswith('foo bar'))
 519         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
 520
 521     def test_escape_rfc3986(self):
 522         reserved = "!*'();:@&=+$,/?#[]"
 523         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
 524         self.assertEqual(escape_rfc3986(reserved), reserved)
 525         self.assertEqual(escape_rfc3986(unreserved), unreserved)
 526         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
 527         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
 528         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
 529         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
 530
 531     def test_escape_url(self):
 532         self.assertEqual(
 533             escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
 534             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
 535         )
 536         self.assertEqual(
 537             escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
 538             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
 539         )
 540         self.assertEqual(
 541             escape_url('http://тест.рф/фрагмент'),
 542             'http://тест.рф/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
 543         )
 544         self.assertEqual(
 545             escape_url('http://тест.рф/абв?абв=абв#абв'),
 546             'http://тест.рф/%D0%B0%D0%B1%D0%B2?%D0%B0%D0%B1%D0%B2=%D0%B0%D0%B1%D0%B2#%D0%B0%D0%B1%D0%B2'
 547         )
 548         self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
 549
 550     def test_js_to_json_realworld(self):
 551         inp = '''{
 552             'clip':{'provider':'pseudo'}
 553         }'''
 554         self.assertEqual(js_to_json(inp), '''{
 555             "clip":{"provider":"pseudo"}
 556         }''')
 557         json.loads(js_to_json(inp))
 558
 559         inp = '''{
 560             'playlist':[{'controls':{'all':null}}]
 561         }'''
 562         self.assertEqual(js_to_json(inp), '''{
 563             "playlist":[{"controls":{"all":null}}]
 564         }''')
 565
 566         inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
 567         self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
 568
 569         inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
 570         json_code = js_to_json(inp)
 571         self.assertEqual(json.loads(json_code), json.loads(inp))
 572
 573     def test_js_to_json_edgecases(self):
 574         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
 575         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
 576
 577         on = js_to_json('{"abc": true}')
 578         self.assertEqual(json.loads(on), {'abc': True})
 579
 580         # Ignore JavaScript code as well
 581         on = js_to_json('''{
 582             "x": 1,
 583             y: "a",
 584             z: some.code
 585         }''')
 586         d = json.loads(on)
 587         self.assertEqual(d['x'], 1)
 588         self.assertEqual(d['y'], 'a')
 589
 590         on = js_to_json('["abc", "def",]')
 591         self.assertEqual(json.loads(on), ['abc', 'def'])
 592
 593         on = js_to_json('{"abc": "def",}')
 594         self.assertEqual(json.loads(on), {'abc': 'def'})
 595
 596     def test_extract_attributes(self):
 597         self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
 598         self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
 599         self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
 600         self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
 601         self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
 602         self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
 603         self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
 604         self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'})  # XML
 605         self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
 606         self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'}) # HTML 3.2
 607         self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'}) # HTML 4.0
 608         self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
 609         self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
 610         self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
 611         self.assertEqual(extract_attributes('<e x >'), {'x': None})
 612         self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
 613         self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
 614         self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
 615         self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
 616         self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
 617         self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
 618         self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
 619         self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}) # Names lowercased
 620         self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
 621         self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
 622         self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
 623         self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
 624         self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
 625         # "Narrow" Python builds don't support unicode code points outside BMP.
 626         try:
 627             compat_chr(0x10000)
 628             supports_outside_bmp = True
 629         except ValueError:
 630             supports_outside_bmp = False
 631         if supports_outside_bmp:
 632             self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
 633
 634     def test_clean_html(self):
 635         self.assertEqual(clean_html('a:\nb'), 'a: b')
 636         self.assertEqual(clean_html('a:\n   "b"'), 'a:    "b"')
 637
 638     def test_intlist_to_bytes(self):
 639         self.assertEqual(
 640             intlist_to_bytes([0, 1, 127, 128, 255]),
 641             b'\x00\x01\x7f\x80\xff')
 642
 643     def test_args_to_str(self):
 644         self.assertEqual(
 645             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
 646             'foo ba/r -baz \'2 be\' \'\''
 647         )
 648
 649     def test_parse_filesize(self):
 650         self.assertEqual(parse_filesize(None), None)
 651         self.assertEqual(parse_filesize(''), None)
 652         self.assertEqual(parse_filesize('91 B'), 91)
 653         self.assertEqual(parse_filesize('foobar'), None)
 654         self.assertEqual(parse_filesize('2 MiB'), 2097152)
 655         self.assertEqual(parse_filesize('5 GB'), 5000000000)
 656         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
 657         self.assertEqual(parse_filesize('1,24 KB'), 1240)
 658
 659     def test_version_tuple(self):
 660         self.assertEqual(version_tuple('1'), (1,))
 661         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
 662         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
 663
 664     def test_detect_exe_version(self):
 665         self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
 666 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
 667 configuration: --prefix=/usr --extra-'''), '1.2.1')
 668         self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
 669 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
 670         self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
 671 Trying to open render node...
 672 Success at /dev/dri/renderD128.
 673 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
 674
 675     def test_age_restricted(self):
 676         self.assertFalse(age_restricted(None, 10))  # unrestricted content
 677         self.assertFalse(age_restricted(1, None))  # unrestricted policy
 678         self.assertFalse(age_restricted(8, 10))
 679         self.assertTrue(age_restricted(18, 14))
 680         self.assertFalse(age_restricted(18, 18))
 681
 682     def test_is_html(self):
 683         self.assertFalse(is_html(b'\x49\x44\x43<html'))
 684         self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
 685         self.assertTrue(is_html(  # UTF-8 with BOM
 686             b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
 687         self.assertTrue(is_html(  # UTF-16-LE
 688             b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
 689         ))
 690         self.assertTrue(is_html(  # UTF-16-BE
 691             b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
 692         ))
 693         self.assertTrue(is_html(  # UTF-32-BE
 694             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'))
 695         self.assertTrue(is_html(  # UTF-32-LE
 696             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'))
 697
 698     def test_render_table(self):
 699         self.assertEqual(
 700             render_table(
 701                 ['a', 'bcd'],
 702                 [[123, 4], [9999, 51]]),
 703             'a    bcd\n'
 704             '123  4\n'
 705             '9999 51')
 706
 707     def test_match_str(self):
 708         self.assertRaises(ValueError, match_str, 'xy>foobar', {})
 709         self.assertFalse(match_str('xy', {'x': 1200}))
 710         self.assertTrue(match_str('!xy', {'x': 1200}))
 711         self.assertTrue(match_str('x', {'x': 1200}))
 712         self.assertFalse(match_str('!x', {'x': 1200}))
 713         self.assertTrue(match_str('x', {'x': 0}))
 714         self.assertFalse(match_str('x>0', {'x': 0}))
 715         self.assertFalse(match_str('x>0', {}))
 716         self.assertTrue(match_str('x>?0', {}))
 717         self.assertTrue(match_str('x>1K', {'x': 1200}))
 718         self.assertFalse(match_str('x>2K', {'x': 1200}))
 719         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
 720         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
 721         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
 722         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
 723         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
 724         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
 725         self.assertFalse(match_str(
 726             'like_count > 100 & dislike_count <? 50 & description',
 727             {'like_count': 90, 'description': 'foo'}))
 728         self.assertTrue(match_str(
 729             'like_count > 100 & dislike_count <? 50 & description',
 730             {'like_count': 190, 'description': 'foo'}))
 731         self.assertFalse(match_str(
 732             'like_count > 100 & dislike_count <? 50 & description',
 733             {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
 734         self.assertFalse(match_str(
 735             'like_count > 100 & dislike_count <? 50 & description',
 736             {'like_count': 190, 'dislike_count': 10}))
 737
 738     def test_parse_dfxp_time_expr(self):
 739         self.assertEqual(parse_dfxp_time_expr(None), None)
 740         self.assertEqual(parse_dfxp_time_expr(''), None)
 741         self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
 742         self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
 743         self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
 744         self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
 745         self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
 746
 747     def test_dfxp2srt(self):
 748         dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
 749             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
 750             <body>
 751                 <div xml:lang="en">
 752                     <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
 753                     <p begin="1" end="2">第二行<br/>♪♪</p>
 754                     <p begin="2" dur="1"><span>Third<br/>Line</span></p>
 755                     <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
 756                     <p begin="-1" end="-1">Ignore, two</p>
 757                     <p begin="3" dur="-1">Ignored, three</p>
 758                 </div>
 759             </body>
 760             </tt>'''
 761         srt_data = '''1
 762 00:00:00,000 --> 00:00:01,000
 763 The following line contains Chinese characters and special symbols
 764
 765 2
 766 00:00:01,000 --> 00:00:02,000
 767 第二行
 768 ♪♪
 769
 770 3
 771 00:00:02,000 --> 00:00:03,000
 772 Third
 773 Line
 774
 775 '''
 776         self.assertEqual(dfxp2srt(dfxp_data), srt_data)
 777
 778         dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
 779             <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
 780             <body>
 781                 <div xml:lang="en">
 782                     <p begin="0" end="1">The first line</p>
 783                 </div>
 784             </body>
 785             </tt>'''
 786         srt_data = '''1
 787 00:00:00,000 --> 00:00:01,000
 788 The first line
 789
 790 '''
 791         self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
 792
 793     def test_cli_option(self):
 794         self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
 795         self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
 796         self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
 797
 798     def test_cli_valueless_option(self):
 799         self.assertEqual(cli_valueless_option(
 800             {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
 801         self.assertEqual(cli_valueless_option(
 802             {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
 803         self.assertEqual(cli_valueless_option(
 804             {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
 805         self.assertEqual(cli_valueless_option(
 806             {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
 807         self.assertEqual(cli_valueless_option(
 808             {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
 809         self.assertEqual(cli_valueless_option(
 810             {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
 811
 812     def test_cli_bool_option(self):
 813         self.assertEqual(
 814             cli_bool_option(
 815                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
 816             ['--no-check-certificate', 'true'])
 817         self.assertEqual(
 818             cli_bool_option(
 819                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
 820             ['--no-check-certificate=true'])
 821         self.assertEqual(
 822             cli_bool_option(
 823                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
 824             ['--check-certificate', 'false'])
 825         self.assertEqual(
 826             cli_bool_option(
 827                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
 828             ['--check-certificate=false'])
 829         self.assertEqual(
 830             cli_bool_option(
 831                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
 832             ['--check-certificate', 'true'])
 833         self.assertEqual(
 834             cli_bool_option(
 835                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
 836             ['--check-certificate=true'])
 837
 838     def test_ohdave_rsa_encrypt(self):
 839         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
 840         e = 65537
 841
 842         self.assertEqual(
 843             ohdave_rsa_encrypt(b'aa111222', e, N),
 844             '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
 845
 846     def test_encode_base_n(self):
 847         self.assertEqual(encode_base_n(0, 30), '0')
 848         self.assertEqual(encode_base_n(80, 30), '2k')
 849
 850         custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
 851         self.assertEqual(encode_base_n(0, 30, custom_table), '9')
 852         self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
 853
 854         self.assertRaises(ValueError, encode_base_n, 0, 70)
 855         self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
 856
 857 if __name__ == '__main__':
 858     unittest.main()