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