_ 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     encodeFilename,
  25     escape_rfc3986,
  26     escape_url,
  27     ExtractorError,
  28     find_xpath_attr,
  29     fix_xml_ampersands,
  30     InAdvancePagedList,
  31     intlist_to_bytes,
  32     is_html,
  33     js_to_json,
  34     limit_length,
  35     OnDemandPagedList,
  36     orderedSet,
  37     parse_duration,
  38     parse_filesize,
  39     parse_iso8601,
  40     read_batch_urls,
  41     sanitize_filename,
  42     sanitize_path,
  43     sanitize_url_path_consecutive_slashes,
  44     prepend_extension,
  45     replace_extension,
  46     shell_quote,
  47     smuggle_url,
  48     str_to_int,
  49     strip_jsonp,
  50     struct_unpack,
  51     timeconvert,
  52     unescapeHTML,
  53     unified_strdate,
  54     unsmuggle_url,
  55     uppercase_escape,
  56     url_basename,
  57     urlencode_postdata,
  58     version_tuple,
  59     xpath_with_ns,
  60     xpath_text,
  61     render_table,
  62     match_str,
  63     parse_dfxp_time_expr,
  64     dfxp2srt,
  65 )
  66
  67
  68 class TestUtil(unittest.TestCase):
  69     def test_timeconvert(self):
  70         self.assertTrue(timeconvert('') is None)
  71         self.assertTrue(timeconvert('bougrg') is None)
  72
  73     def test_sanitize_filename(self):
  74         self.assertEqual(sanitize_filename('abc'), 'abc')
  75         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
  76
  77         self.assertEqual(sanitize_filename('123'), '123')
  78
  79         self.assertEqual('abc_de', sanitize_filename('abc/de'))
  80         self.assertFalse('/' in sanitize_filename('abc/de///'))
  81
  82         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
  83         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
  84         self.assertEqual('yes no', sanitize_filename('yes? no'))
  85         self.assertEqual('this - that', sanitize_filename('this: that'))
  86
  87         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
  88         aumlaut = 'ä'
  89         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
  90         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
  91         self.assertEqual(sanitize_filename(tests), tests)
  92
  93         self.assertEqual(
  94             sanitize_filename('New World record at 0:12:34'),
  95             'New World record at 0_12_34')
  96
  97         self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
  98         self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
  99         self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
 100         self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
 101
 102         forbidden = '"\0\\/'
 103         for fc in forbidden:
 104             for fbc in forbidden:
 105                 self.assertTrue(fbc not in sanitize_filename(fc))
 106
 107     def test_sanitize_filename_restricted(self):
 108         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
 109         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
 110
 111         self.assertEqual(sanitize_filename('123', restricted=True), '123')
 112
 113         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
 114         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
 115
 116         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
 117         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
 118         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
 119         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
 120
 121         tests = 'a\xe4b\u4e2d\u56fd\u7684c'
 122         self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
 123         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
 124
 125         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
 126         for fc in forbidden:
 127             for fbc in forbidden:
 128                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
 129
 130         # Handle a common case more neatly
 131         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
 132         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
 133         # .. but make sure the file name is never empty
 134         self.assertTrue(sanitize_filename('-', restricted=True) != '')
 135         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 136
 137     def test_sanitize_ids(self):
 138         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 139         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 140         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 141
 142     def test_sanitize_path(self):
 143         if sys.platform != 'win32':
 144             return
 145
 146         self.assertEqual(sanitize_path('abc'), 'abc')
 147         self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
 148         self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
 149         self.assertEqual(sanitize_path('abc|def'), 'abc#def')
 150         self.assertEqual(sanitize_path('<>:"|?*'), '#######')
 151         self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
 152         self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
 153
 154         self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
 155         self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
 156
 157         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 158         self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
 159         self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
 160         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 161
 162         self.assertEqual(
 163             sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
 164             'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
 165
 166         self.assertEqual(
 167             sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
 168             'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
 169         self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
 170         self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
 171         self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
 172
 173         self.assertEqual(sanitize_path('../abc'), '..\\abc')
 174         self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
 175         self.assertEqual(sanitize_path('./abc'), 'abc')
 176         self.assertEqual(sanitize_path('./../abc'), '..\\abc')
 177
 178     def test_sanitize_url_path_consecutive_slashes(self):
 179         self.assertEqual(
 180             sanitize_url_path_consecutive_slashes('http://hostname/foo//bar/filename.html'),
 181             'http://hostname/foo/bar/filename.html')
 182         self.assertEqual(
 183             sanitize_url_path_consecutive_slashes('http://hostname//foo/bar/filename.html'),
 184             'http://hostname/foo/bar/filename.html')
 185         self.assertEqual(
 186             sanitize_url_path_consecutive_slashes('http://hostname//'),
 187             'http://hostname/')
 188         self.assertEqual(
 189             sanitize_url_path_consecutive_slashes('http://hostname/foo/bar/filename.html'),
 190             'http://hostname/foo/bar/filename.html')
 191         self.assertEqual(
 192             sanitize_url_path_consecutive_slashes('http://hostname/'),
 193             'http://hostname/')
 194         self.assertEqual(
 195             sanitize_url_path_consecutive_slashes('http://hostname/abc//'),
 196             'http://hostname/abc/')
 197
 198     def test_prepend_extension(self):
 199         self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
 200         self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
 201         self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 202         self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
 203         self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
 204         self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
 205
 206     def test_replace_extension(self):
 207         self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
 208         self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
 209         self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 210         self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
 211         self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
 212         self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
 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(
 226             unescapeHTML('&eacute;'), 'é')
 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
 252     def test_find_xpath_attr(self):
 253         testxml = '''<root>
 254             <node/>
 255             <node x="a"/>
 256             <node x="a" y="c" />
 257             <node x="b" y="d" />
 258         </root>'''
 259         doc = xml.etree.ElementTree.fromstring(testxml)
 260
 261         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
 262         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
 263         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
 264
 265     def test_xpath_with_ns(self):
 266         testxml = '''<root xmlns:media="http://example.com/">
 267             <media:song>
 268                 <media:author>The Author</media:author>
 269                 <url>http://server.com/download.mp3</url>
 270             </media:song>
 271         </root>'''
 272         doc = xml.etree.ElementTree.fromstring(testxml)
 273         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
 274         self.assertTrue(find('media:song') is not None)
 275         self.assertEqual(find('media:song/media:author').text, 'The Author')
 276         self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
 277
 278     def test_xpath_text(self):
 279         testxml = '''<root>
 280             <div>
 281                 <p>Foo</p>
 282             </div>
 283         </root>'''
 284         doc = xml.etree.ElementTree.fromstring(testxml)
 285         self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
 286         self.assertTrue(xpath_text(doc, 'div/bar') is None)
 287         self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
 288
 289     def test_smuggle_url(self):
 290         data = {"ö": "ö", "abc": [3]}
 291         url = 'https://foo.bar/baz?x=y#a'
 292         smug_url = smuggle_url(url, data)
 293         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
 294         self.assertEqual(url, unsmug_url)
 295         self.assertEqual(data, unsmug_data)
 296
 297         res_url, res_data = unsmuggle_url(url)
 298         self.assertEqual(res_url, url)
 299         self.assertEqual(res_data, None)
 300
 301     def test_shell_quote(self):
 302         args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
 303         self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
 304
 305     def test_str_to_int(self):
 306         self.assertEqual(str_to_int('123,456'), 123456)
 307         self.assertEqual(str_to_int('123.456'), 123456)
 308
 309     def test_url_basename(self):
 310         self.assertEqual(url_basename('http://foo.de/'), '')
 311         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
 312         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
 313         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
 314         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
 315         self.assertEqual(
 316             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
 317             'trailer.mp4')
 318
 319     def test_parse_duration(self):
 320         self.assertEqual(parse_duration(None), None)
 321         self.assertEqual(parse_duration(False), None)
 322         self.assertEqual(parse_duration('invalid'), None)
 323         self.assertEqual(parse_duration('1'), 1)
 324         self.assertEqual(parse_duration('1337:12'), 80232)
 325         self.assertEqual(parse_duration('9:12:43'), 33163)
 326         self.assertEqual(parse_duration('12:00'), 720)
 327         self.assertEqual(parse_duration('00:01:01'), 61)
 328         self.assertEqual(parse_duration('x:y'), None)
 329         self.assertEqual(parse_duration('3h11m53s'), 11513)
 330         self.assertEqual(parse_duration('3h 11m 53s'), 11513)
 331         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
 332         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
 333         self.assertEqual(parse_duration('62m45s'), 3765)
 334         self.assertEqual(parse_duration('6m59s'), 419)
 335         self.assertEqual(parse_duration('49s'), 49)
 336         self.assertEqual(parse_duration('0h0m0s'), 0)
 337         self.assertEqual(parse_duration('0m0s'), 0)
 338         self.assertEqual(parse_duration('0s'), 0)
 339         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
 340         self.assertEqual(parse_duration('T30M38S'), 1838)
 341         self.assertEqual(parse_duration('5 s'), 5)
 342         self.assertEqual(parse_duration('3 min'), 180)
 343         self.assertEqual(parse_duration('2.5 hours'), 9000)
 344         self.assertEqual(parse_duration('02:03:04'), 7384)
 345         self.assertEqual(parse_duration('01:02:03:04'), 93784)
 346         self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
 347
 348     def test_fix_xml_ampersands(self):
 349         self.assertEqual(
 350             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 351         self.assertEqual(
 352             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 353             '"&amp;x=y&amp;wrong;&amp;z=a')
 354         self.assertEqual(
 355             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 356             '&amp;&apos;&gt;&lt;&quot;')
 357         self.assertEqual(
 358             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 359         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 360
 361     def test_paged_list(self):
 362         def testPL(size, pagesize, sliceargs, expected):
 363             def get_page(pagenum):
 364                 firstid = pagenum * pagesize
 365                 upto = min(size, pagenum * pagesize + pagesize)
 366                 for i in range(firstid, upto):
 367                     yield i
 368
 369             pl = OnDemandPagedList(get_page, pagesize)
 370             got = pl.getslice(*sliceargs)
 371             self.assertEqual(got, expected)
 372
 373             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
 374             got = iapl.getslice(*sliceargs)
 375             self.assertEqual(got, expected)
 376
 377         testPL(5, 2, (), [0, 1, 2, 3, 4])
 378         testPL(5, 2, (1,), [1, 2, 3, 4])
 379         testPL(5, 2, (2,), [2, 3, 4])
 380         testPL(5, 2, (4,), [4])
 381         testPL(5, 2, (0, 3), [0, 1, 2])
 382         testPL(5, 2, (1, 4), [1, 2, 3])
 383         testPL(5, 2, (2, 99), [2, 3, 4])
 384         testPL(5, 2, (20, 99), [])
 385
 386     def test_struct_unpack(self):
 387         self.assertEqual(struct_unpack('!B', b'\x00'), (0,))
 388
 389     def test_read_batch_urls(self):
 390         f = io.StringIO('''\xef\xbb\xbf foo
 391             bar\r
 392             baz
 393             # More after this line\r
 394             ; or after this
 395             bam''')
 396         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
 397
 398     def test_urlencode_postdata(self):
 399         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
 400         self.assertTrue(isinstance(data, bytes))
 401
 402     def test_parse_iso8601(self):
 403         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
 404         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
 405         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
 406         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
 407
 408     def test_strip_jsonp(self):
 409         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
 410         d = json.loads(stripped)
 411         self.assertEqual(d, [{"id": "532cb", "x": 3}])
 412
 413         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
 414         d = json.loads(stripped)
 415         self.assertEqual(d, {'STATUS': 'OK'})
 416
 417     def test_uppercase_escape(self):
 418         self.assertEqual(uppercase_escape('aä'), 'aä')
 419         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
 420
 421     def test_limit_length(self):
 422         self.assertEqual(limit_length(None, 12), None)
 423         self.assertEqual(limit_length('foo', 12), 'foo')
 424         self.assertTrue(
 425             limit_length('foo bar baz asd', 12).startswith('foo bar'))
 426         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
 427
 428     def test_escape_rfc3986(self):
 429         reserved = "!*'();:@&=+$,/?#[]"
 430         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
 431         self.assertEqual(escape_rfc3986(reserved), reserved)
 432         self.assertEqual(escape_rfc3986(unreserved), unreserved)
 433         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
 434         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
 435         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
 436         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
 437
 438     def test_escape_url(self):
 439         self.assertEqual(
 440             escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
 441             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
 442         )
 443         self.assertEqual(
 444             escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
 445             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
 446         )
 447         self.assertEqual(
 448             escape_url('http://тест.рф/фрагмент'),
 449             'http://тест.рф/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
 450         )
 451         self.assertEqual(
 452             escape_url('http://тест.рф/абв?абв=абв#абв'),
 453             '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'
 454         )
 455         self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
 456
 457     def test_js_to_json_realworld(self):
 458         inp = '''{
 459             'clip':{'provider':'pseudo'}
 460         }'''
 461         self.assertEqual(js_to_json(inp), '''{
 462             "clip":{"provider":"pseudo"}
 463         }''')
 464         json.loads(js_to_json(inp))
 465
 466         inp = '''{
 467             'playlist':[{'controls':{'all':null}}]
 468         }'''
 469         self.assertEqual(js_to_json(inp), '''{
 470             "playlist":[{"controls":{"all":null}}]
 471         }''')
 472
 473         inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
 474         json_code = js_to_json(inp)
 475         self.assertEqual(json.loads(json_code), json.loads(inp))
 476
 477     def test_js_to_json_edgecases(self):
 478         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
 479         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
 480
 481         on = js_to_json('{"abc": true}')
 482         self.assertEqual(json.loads(on), {'abc': True})
 483
 484         # Ignore JavaScript code as well
 485         on = js_to_json('''{
 486             "x": 1,
 487             y: "a",
 488             z: some.code
 489         }''')
 490         d = json.loads(on)
 491         self.assertEqual(d['x'], 1)
 492         self.assertEqual(d['y'], 'a')
 493
 494         on = js_to_json('["abc", "def",]')
 495         self.assertEqual(json.loads(on), ['abc', 'def'])
 496
 497         on = js_to_json('{"abc": "def",}')
 498         self.assertEqual(json.loads(on), {'abc': 'def'})
 499
 500     def test_clean_html(self):
 501         self.assertEqual(clean_html('a:\nb'), 'a: b')
 502         self.assertEqual(clean_html('a:\n   "b"'), 'a:    "b"')
 503
 504     def test_intlist_to_bytes(self):
 505         self.assertEqual(
 506             intlist_to_bytes([0, 1, 127, 128, 255]),
 507             b'\x00\x01\x7f\x80\xff')
 508
 509     def test_args_to_str(self):
 510         self.assertEqual(
 511             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
 512             'foo ba/r -baz \'2 be\' \'\''
 513         )
 514
 515     def test_parse_filesize(self):
 516         self.assertEqual(parse_filesize(None), None)
 517         self.assertEqual(parse_filesize(''), None)
 518         self.assertEqual(parse_filesize('91 B'), 91)
 519         self.assertEqual(parse_filesize('foobar'), None)
 520         self.assertEqual(parse_filesize('2 MiB'), 2097152)
 521         self.assertEqual(parse_filesize('5 GB'), 5000000000)
 522         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
 523         self.assertEqual(parse_filesize('1,24 KB'), 1240)
 524
 525     def test_version_tuple(self):
 526         self.assertEqual(version_tuple('1'), (1,))
 527         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
 528         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
 529
 530     def test_detect_exe_version(self):
 531         self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
 532 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
 533 configuration: --prefix=/usr --extra-'''), '1.2.1')
 534         self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
 535 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
 536         self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
 537 Trying to open render node...
 538 Success at /dev/dri/renderD128.
 539 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
 540
 541     def test_age_restricted(self):
 542         self.assertFalse(age_restricted(None, 10))  # unrestricted content
 543         self.assertFalse(age_restricted(1, None))  # unrestricted policy
 544         self.assertFalse(age_restricted(8, 10))
 545         self.assertTrue(age_restricted(18, 14))
 546         self.assertFalse(age_restricted(18, 18))
 547
 548     def test_is_html(self):
 549         self.assertFalse(is_html(b'\x49\x44\x43<html'))
 550         self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
 551         self.assertTrue(is_html(  # UTF-8 with BOM
 552             b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
 553         self.assertTrue(is_html(  # UTF-16-LE
 554             b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
 555         ))
 556         self.assertTrue(is_html(  # UTF-16-BE
 557             b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
 558         ))
 559         self.assertTrue(is_html(  # UTF-32-BE
 560             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'))
 561         self.assertTrue(is_html(  # UTF-32-LE
 562             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'))
 563
 564     def test_render_table(self):
 565         self.assertEqual(
 566             render_table(
 567                 ['a', 'bcd'],
 568                 [[123, 4], [9999, 51]]),
 569             'a    bcd\n'
 570             '123  4\n'
 571             '9999 51')
 572
 573     def test_match_str(self):
 574         self.assertRaises(ValueError, match_str, 'xy>foobar', {})
 575         self.assertFalse(match_str('xy', {'x': 1200}))
 576         self.assertTrue(match_str('!xy', {'x': 1200}))
 577         self.assertTrue(match_str('x', {'x': 1200}))
 578         self.assertFalse(match_str('!x', {'x': 1200}))
 579         self.assertTrue(match_str('x', {'x': 0}))
 580         self.assertFalse(match_str('x>0', {'x': 0}))
 581         self.assertFalse(match_str('x>0', {}))
 582         self.assertTrue(match_str('x>?0', {}))
 583         self.assertTrue(match_str('x>1K', {'x': 1200}))
 584         self.assertFalse(match_str('x>2K', {'x': 1200}))
 585         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
 586         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
 587         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
 588         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
 589         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
 590         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
 591         self.assertFalse(match_str(
 592             'like_count > 100 & dislike_count <? 50 & description',
 593             {'like_count': 90, 'description': 'foo'}))
 594         self.assertTrue(match_str(
 595             'like_count > 100 & dislike_count <? 50 & description',
 596             {'like_count': 190, 'description': 'foo'}))
 597         self.assertFalse(match_str(
 598             'like_count > 100 & dislike_count <? 50 & description',
 599             {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
 600         self.assertFalse(match_str(
 601             'like_count > 100 & dislike_count <? 50 & description',
 602             {'like_count': 190, 'dislike_count': 10}))
 603
 604     def test_parse_dfxp_time_expr(self):
 605         self.assertEqual(parse_dfxp_time_expr(None), 0.0)
 606         self.assertEqual(parse_dfxp_time_expr(''), 0.0)
 607         self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
 608         self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
 609         self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
 610         self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
 611
 612     def test_dfxp2srt(self):
 613         dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
 614             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
 615             <body>
 616                 <div xml:lang="en">
 617                     <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
 618                     <p begin="1" end="2">第二行<br/>♪♪</p>
 619                     <p begin="2" end="3"><span>Third<br/>Line</span></p>
 620                 </div>
 621             </body>
 622             </tt>'''
 623         srt_data = '''1
 624 00:00:00,000 --> 00:00:01,000
 625 The following line contains Chinese characters and special symbols
 626
 627 2
 628 00:00:01,000 --> 00:00:02,000
 629 第二行
 630 ♪♪
 631
 632 3
 633 00:00:02,000 --> 00:00:03,000
 634 Third
 635 Line
 636
 637 '''
 638         self.assertEqual(dfxp2srt(dfxp_data), srt_data)
 639
 640
 641 if __name__ == '__main__':
 642     unittest.main()