_ 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     prepend_extension,
  44     replace_extension,
  45     shell_quote,
  46     smuggle_url,
  47     str_to_int,
  48     strip_jsonp,
  49     struct_unpack,
  50     timeconvert,
  51     unescapeHTML,
  52     unified_strdate,
  53     unsmuggle_url,
  54     uppercase_escape,
  55     lowercase_escape,
  56     url_basename,
  57     urlencode_postdata,
  58     version_tuple,
  59     xpath_with_ns,
  60     xpath_element,
  61     xpath_text,
  62     xpath_attr,
  63     render_table,
  64     match_str,
  65     parse_dfxp_time_expr,
  66     dfxp2srt,
  67 )
  68
  69
  70 class TestUtil(unittest.TestCase):
  71     def test_timeconvert(self):
  72         self.assertTrue(timeconvert('') is None)
  73         self.assertTrue(timeconvert('bougrg') is None)
  74
  75     def test_sanitize_filename(self):
  76         self.assertEqual(sanitize_filename('abc'), 'abc')
  77         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
  78
  79         self.assertEqual(sanitize_filename('123'), '123')
  80
  81         self.assertEqual('abc_de', sanitize_filename('abc/de'))
  82         self.assertFalse('/' in sanitize_filename('abc/de///'))
  83
  84         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
  85         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
  86         self.assertEqual('yes no', sanitize_filename('yes? no'))
  87         self.assertEqual('this - that', sanitize_filename('this: that'))
  88
  89         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
  90         aumlaut = 'ä'
  91         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
  92         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
  93         self.assertEqual(sanitize_filename(tests), tests)
  94
  95         self.assertEqual(
  96             sanitize_filename('New World record at 0:12:34'),
  97             'New World record at 0_12_34')
  98
  99         self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
 100         self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
 101         self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
 102         self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
 103
 104         forbidden = '"\0\\/'
 105         for fc in forbidden:
 106             for fbc in forbidden:
 107                 self.assertTrue(fbc not in sanitize_filename(fc))
 108
 109     def test_sanitize_filename_restricted(self):
 110         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
 111         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
 112
 113         self.assertEqual(sanitize_filename('123', restricted=True), '123')
 114
 115         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
 116         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
 117
 118         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
 119         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
 120         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
 121         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
 122
 123         tests = 'a\xe4b\u4e2d\u56fd\u7684c'
 124         self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
 125         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
 126
 127         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
 128         for fc in forbidden:
 129             for fbc in forbidden:
 130                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
 131
 132         # Handle a common case more neatly
 133         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
 134         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
 135         # .. but make sure the file name is never empty
 136         self.assertTrue(sanitize_filename('-', restricted=True) != '')
 137         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 138
 139     def test_sanitize_ids(self):
 140         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 141         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 142         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 143
 144     def test_sanitize_path(self):
 145         if sys.platform != 'win32':
 146             return
 147
 148         self.assertEqual(sanitize_path('abc'), 'abc')
 149         self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
 150         self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
 151         self.assertEqual(sanitize_path('abc|def'), 'abc#def')
 152         self.assertEqual(sanitize_path('<>:"|?*'), '#######')
 153         self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
 154         self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
 155
 156         self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
 157         self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
 158
 159         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 160         self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
 161         self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
 162         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 163
 164         self.assertEqual(
 165             sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
 166             'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
 167
 168         self.assertEqual(
 169             sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
 170             'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
 171         self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
 172         self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
 173         self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
 174
 175         self.assertEqual(sanitize_path('../abc'), '..\\abc')
 176         self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
 177         self.assertEqual(sanitize_path('./abc'), 'abc')
 178         self.assertEqual(sanitize_path('./../abc'), '..\\abc')
 179
 180     def test_prepend_extension(self):
 181         self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
 182         self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
 183         self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 184         self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
 185         self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
 186         self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
 187
 188     def test_replace_extension(self):
 189         self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
 190         self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
 191         self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 192         self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
 193         self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
 194         self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
 195
 196     def test_ordered_set(self):
 197         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
 198         self.assertEqual(orderedSet([]), [])
 199         self.assertEqual(orderedSet([1]), [1])
 200         # keep the list ordered
 201         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
 202
 203     def test_unescape_html(self):
 204         self.assertEqual(unescapeHTML('%20;'), '%20;')
 205         self.assertEqual(unescapeHTML('&#x2F;'), '/')
 206         self.assertEqual(unescapeHTML('&#47;'), '/')
 207         self.assertEqual(
 208             unescapeHTML('&eacute;'), 'é')
 209
 210     def test_daterange(self):
 211         _20century = DateRange("19000101", "20000101")
 212         self.assertFalse("17890714" in _20century)
 213         _ac = DateRange("00010101")
 214         self.assertTrue("19690721" in _ac)
 215         _firstmilenium = DateRange(end="10000101")
 216         self.assertTrue("07110427" in _firstmilenium)
 217
 218     def test_unified_dates(self):
 219         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
 220         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
 221         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
 222         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
 223         self.assertEqual(unified_strdate('1968 12 10'), '19681210')
 224         self.assertEqual(unified_strdate('1968-12-10'), '19681210')
 225         self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
 226         self.assertEqual(
 227             unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
 228             '20141126')
 229         self.assertEqual(
 230             unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
 231             '20150202')
 232         self.assertEqual(unified_strdate('25-09-2014'), '20140925')
 233
 234     def test_find_xpath_attr(self):
 235         testxml = '''<root>
 236             <node/>
 237             <node x="a"/>
 238             <node x="a" y="c" />
 239             <node x="b" y="d" />
 240             <node x="" />
 241         </root>'''
 242         doc = xml.etree.ElementTree.fromstring(testxml)
 243
 244         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
 245         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
 246         self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
 247         self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
 248         self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
 249         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
 250         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
 251         self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
 252         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
 253         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
 254         self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
 255
 256     def test_xpath_with_ns(self):
 257         testxml = '''<root xmlns:media="http://example.com/">
 258             <media:song>
 259                 <media:author>The Author</media:author>
 260                 <url>http://server.com/download.mp3</url>
 261             </media:song>
 262         </root>'''
 263         doc = xml.etree.ElementTree.fromstring(testxml)
 264         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
 265         self.assertTrue(find('media:song') is not None)
 266         self.assertEqual(find('media:song/media:author').text, 'The Author')
 267         self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
 268
 269     def test_xpath_element(self):
 270         doc = xml.etree.ElementTree.Element('root')
 271         div = xml.etree.ElementTree.SubElement(doc, 'div')
 272         p = xml.etree.ElementTree.SubElement(div, 'p')
 273         p.text = 'Foo'
 274         self.assertEqual(xpath_element(doc, 'div/p'), p)
 275         self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
 276         self.assertTrue(xpath_element(doc, 'div/bar') is None)
 277         self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
 278
 279     def test_xpath_text(self):
 280         testxml = '''<root>
 281             <div>
 282                 <p>Foo</p>
 283             </div>
 284         </root>'''
 285         doc = xml.etree.ElementTree.fromstring(testxml)
 286         self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
 287         self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
 288         self.assertTrue(xpath_text(doc, 'div/bar') is None)
 289         self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
 290
 291     def test_xpath_attr(self):
 292         testxml = '''<root>
 293             <div>
 294                 <p x="a">Foo</p>
 295             </div>
 296         </root>'''
 297         doc = xml.etree.ElementTree.fromstring(testxml)
 298         self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
 299         self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
 300         self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
 301         self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
 302         self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
 303         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
 304         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
 305
 306     def test_smuggle_url(self):
 307         data = {"ö": "ö", "abc": [3]}
 308         url = 'https://foo.bar/baz?x=y#a'
 309         smug_url = smuggle_url(url, data)
 310         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
 311         self.assertEqual(url, unsmug_url)
 312         self.assertEqual(data, unsmug_data)
 313
 314         res_url, res_data = unsmuggle_url(url)
 315         self.assertEqual(res_url, url)
 316         self.assertEqual(res_data, None)
 317
 318     def test_shell_quote(self):
 319         args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
 320         self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
 321
 322     def test_str_to_int(self):
 323         self.assertEqual(str_to_int('123,456'), 123456)
 324         self.assertEqual(str_to_int('123.456'), 123456)
 325
 326     def test_url_basename(self):
 327         self.assertEqual(url_basename('http://foo.de/'), '')
 328         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
 329         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
 330         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
 331         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
 332         self.assertEqual(
 333             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
 334             'trailer.mp4')
 335
 336     def test_parse_duration(self):
 337         self.assertEqual(parse_duration(None), None)
 338         self.assertEqual(parse_duration(False), None)
 339         self.assertEqual(parse_duration('invalid'), None)
 340         self.assertEqual(parse_duration('1'), 1)
 341         self.assertEqual(parse_duration('1337:12'), 80232)
 342         self.assertEqual(parse_duration('9:12:43'), 33163)
 343         self.assertEqual(parse_duration('12:00'), 720)
 344         self.assertEqual(parse_duration('00:01:01'), 61)
 345         self.assertEqual(parse_duration('x:y'), None)
 346         self.assertEqual(parse_duration('3h11m53s'), 11513)
 347         self.assertEqual(parse_duration('3h 11m 53s'), 11513)
 348         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
 349         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
 350         self.assertEqual(parse_duration('62m45s'), 3765)
 351         self.assertEqual(parse_duration('6m59s'), 419)
 352         self.assertEqual(parse_duration('49s'), 49)
 353         self.assertEqual(parse_duration('0h0m0s'), 0)
 354         self.assertEqual(parse_duration('0m0s'), 0)
 355         self.assertEqual(parse_duration('0s'), 0)
 356         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
 357         self.assertEqual(parse_duration('T30M38S'), 1838)
 358         self.assertEqual(parse_duration('5 s'), 5)
 359         self.assertEqual(parse_duration('3 min'), 180)
 360         self.assertEqual(parse_duration('2.5 hours'), 9000)
 361         self.assertEqual(parse_duration('02:03:04'), 7384)
 362         self.assertEqual(parse_duration('01:02:03:04'), 93784)
 363         self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
 364         self.assertEqual(parse_duration('87 Min.'), 5220)
 365
 366     def test_fix_xml_ampersands(self):
 367         self.assertEqual(
 368             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 369         self.assertEqual(
 370             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 371             '"&amp;x=y&amp;wrong;&amp;z=a')
 372         self.assertEqual(
 373             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 374             '&amp;&apos;&gt;&lt;&quot;')
 375         self.assertEqual(
 376             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 377         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 378
 379     def test_paged_list(self):
 380         def testPL(size, pagesize, sliceargs, expected):
 381             def get_page(pagenum):
 382                 firstid = pagenum * pagesize
 383                 upto = min(size, pagenum * pagesize + pagesize)
 384                 for i in range(firstid, upto):
 385                     yield i
 386
 387             pl = OnDemandPagedList(get_page, pagesize)
 388             got = pl.getslice(*sliceargs)
 389             self.assertEqual(got, expected)
 390
 391             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
 392             got = iapl.getslice(*sliceargs)
 393             self.assertEqual(got, expected)
 394
 395         testPL(5, 2, (), [0, 1, 2, 3, 4])
 396         testPL(5, 2, (1,), [1, 2, 3, 4])
 397         testPL(5, 2, (2,), [2, 3, 4])
 398         testPL(5, 2, (4,), [4])
 399         testPL(5, 2, (0, 3), [0, 1, 2])
 400         testPL(5, 2, (1, 4), [1, 2, 3])
 401         testPL(5, 2, (2, 99), [2, 3, 4])
 402         testPL(5, 2, (20, 99), [])
 403
 404     def test_struct_unpack(self):
 405         self.assertEqual(struct_unpack('!B', b'\x00'), (0,))
 406
 407     def test_read_batch_urls(self):
 408         f = io.StringIO('''\xef\xbb\xbf foo
 409             bar\r
 410             baz
 411             # More after this line\r
 412             ; or after this
 413             bam''')
 414         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
 415
 416     def test_urlencode_postdata(self):
 417         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
 418         self.assertTrue(isinstance(data, bytes))
 419
 420     def test_parse_iso8601(self):
 421         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
 422         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
 423         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
 424         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
 425
 426     def test_strip_jsonp(self):
 427         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
 428         d = json.loads(stripped)
 429         self.assertEqual(d, [{"id": "532cb", "x": 3}])
 430
 431         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
 432         d = json.loads(stripped)
 433         self.assertEqual(d, {'STATUS': 'OK'})
 434
 435     def test_uppercase_escape(self):
 436         self.assertEqual(uppercase_escape('aä'), 'aä')
 437         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
 438
 439     def test_lowercase_escape(self):
 440         self.assertEqual(lowercase_escape('aä'), 'aä')
 441         self.assertEqual(lowercase_escape('\\u0026'), '&')
 442
 443     def test_limit_length(self):
 444         self.assertEqual(limit_length(None, 12), None)
 445         self.assertEqual(limit_length('foo', 12), 'foo')
 446         self.assertTrue(
 447             limit_length('foo bar baz asd', 12).startswith('foo bar'))
 448         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
 449
 450     def test_escape_rfc3986(self):
 451         reserved = "!*'();:@&=+$,/?#[]"
 452         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
 453         self.assertEqual(escape_rfc3986(reserved), reserved)
 454         self.assertEqual(escape_rfc3986(unreserved), unreserved)
 455         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
 456         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
 457         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
 458         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
 459
 460     def test_escape_url(self):
 461         self.assertEqual(
 462             escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
 463             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
 464         )
 465         self.assertEqual(
 466             escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
 467             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
 468         )
 469         self.assertEqual(
 470             escape_url('http://тест.рф/фрагмент'),
 471             'http://тест.рф/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
 472         )
 473         self.assertEqual(
 474             escape_url('http://тест.рф/абв?абв=абв#абв'),
 475             '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'
 476         )
 477         self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
 478
 479     def test_js_to_json_realworld(self):
 480         inp = '''{
 481             'clip':{'provider':'pseudo'}
 482         }'''
 483         self.assertEqual(js_to_json(inp), '''{
 484             "clip":{"provider":"pseudo"}
 485         }''')
 486         json.loads(js_to_json(inp))
 487
 488         inp = '''{
 489             'playlist':[{'controls':{'all':null}}]
 490         }'''
 491         self.assertEqual(js_to_json(inp), '''{
 492             "playlist":[{"controls":{"all":null}}]
 493         }''')
 494
 495         inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
 496         json_code = js_to_json(inp)
 497         self.assertEqual(json.loads(json_code), json.loads(inp))
 498
 499     def test_js_to_json_edgecases(self):
 500         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
 501         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
 502
 503         on = js_to_json('{"abc": true}')
 504         self.assertEqual(json.loads(on), {'abc': True})
 505
 506         # Ignore JavaScript code as well
 507         on = js_to_json('''{
 508             "x": 1,
 509             y: "a",
 510             z: some.code
 511         }''')
 512         d = json.loads(on)
 513         self.assertEqual(d['x'], 1)
 514         self.assertEqual(d['y'], 'a')
 515
 516         on = js_to_json('["abc", "def",]')
 517         self.assertEqual(json.loads(on), ['abc', 'def'])
 518
 519         on = js_to_json('{"abc": "def",}')
 520         self.assertEqual(json.loads(on), {'abc': 'def'})
 521
 522     def test_clean_html(self):
 523         self.assertEqual(clean_html('a:\nb'), 'a: b')
 524         self.assertEqual(clean_html('a:\n   "b"'), 'a:    "b"')
 525
 526     def test_intlist_to_bytes(self):
 527         self.assertEqual(
 528             intlist_to_bytes([0, 1, 127, 128, 255]),
 529             b'\x00\x01\x7f\x80\xff')
 530
 531     def test_args_to_str(self):
 532         self.assertEqual(
 533             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
 534             'foo ba/r -baz \'2 be\' \'\''
 535         )
 536
 537     def test_parse_filesize(self):
 538         self.assertEqual(parse_filesize(None), None)
 539         self.assertEqual(parse_filesize(''), None)
 540         self.assertEqual(parse_filesize('91 B'), 91)
 541         self.assertEqual(parse_filesize('foobar'), None)
 542         self.assertEqual(parse_filesize('2 MiB'), 2097152)
 543         self.assertEqual(parse_filesize('5 GB'), 5000000000)
 544         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
 545         self.assertEqual(parse_filesize('1,24 KB'), 1240)
 546
 547     def test_version_tuple(self):
 548         self.assertEqual(version_tuple('1'), (1,))
 549         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
 550         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
 551
 552     def test_detect_exe_version(self):
 553         self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
 554 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
 555 configuration: --prefix=/usr --extra-'''), '1.2.1')
 556         self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
 557 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
 558         self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
 559 Trying to open render node...
 560 Success at /dev/dri/renderD128.
 561 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
 562
 563     def test_age_restricted(self):
 564         self.assertFalse(age_restricted(None, 10))  # unrestricted content
 565         self.assertFalse(age_restricted(1, None))  # unrestricted policy
 566         self.assertFalse(age_restricted(8, 10))
 567         self.assertTrue(age_restricted(18, 14))
 568         self.assertFalse(age_restricted(18, 18))
 569
 570     def test_is_html(self):
 571         self.assertFalse(is_html(b'\x49\x44\x43<html'))
 572         self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
 573         self.assertTrue(is_html(  # UTF-8 with BOM
 574             b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
 575         self.assertTrue(is_html(  # UTF-16-LE
 576             b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
 577         ))
 578         self.assertTrue(is_html(  # UTF-16-BE
 579             b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
 580         ))
 581         self.assertTrue(is_html(  # UTF-32-BE
 582             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'))
 583         self.assertTrue(is_html(  # UTF-32-LE
 584             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'))
 585
 586     def test_render_table(self):
 587         self.assertEqual(
 588             render_table(
 589                 ['a', 'bcd'],
 590                 [[123, 4], [9999, 51]]),
 591             'a    bcd\n'
 592             '123  4\n'
 593             '9999 51')
 594
 595     def test_match_str(self):
 596         self.assertRaises(ValueError, match_str, 'xy>foobar', {})
 597         self.assertFalse(match_str('xy', {'x': 1200}))
 598         self.assertTrue(match_str('!xy', {'x': 1200}))
 599         self.assertTrue(match_str('x', {'x': 1200}))
 600         self.assertFalse(match_str('!x', {'x': 1200}))
 601         self.assertTrue(match_str('x', {'x': 0}))
 602         self.assertFalse(match_str('x>0', {'x': 0}))
 603         self.assertFalse(match_str('x>0', {}))
 604         self.assertTrue(match_str('x>?0', {}))
 605         self.assertTrue(match_str('x>1K', {'x': 1200}))
 606         self.assertFalse(match_str('x>2K', {'x': 1200}))
 607         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
 608         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
 609         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
 610         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
 611         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
 612         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
 613         self.assertFalse(match_str(
 614             'like_count > 100 & dislike_count <? 50 & description',
 615             {'like_count': 90, 'description': 'foo'}))
 616         self.assertTrue(match_str(
 617             'like_count > 100 & dislike_count <? 50 & description',
 618             {'like_count': 190, 'description': 'foo'}))
 619         self.assertFalse(match_str(
 620             'like_count > 100 & dislike_count <? 50 & description',
 621             {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
 622         self.assertFalse(match_str(
 623             'like_count > 100 & dislike_count <? 50 & description',
 624             {'like_count': 190, 'dislike_count': 10}))
 625
 626     def test_parse_dfxp_time_expr(self):
 627         self.assertEqual(parse_dfxp_time_expr(None), 0.0)
 628         self.assertEqual(parse_dfxp_time_expr(''), 0.0)
 629         self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
 630         self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
 631         self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
 632         self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
 633
 634     def test_dfxp2srt(self):
 635         dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
 636             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
 637             <body>
 638                 <div xml:lang="en">
 639                     <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
 640                     <p begin="1" end="2">第二行<br/>♪♪</p>
 641                     <p begin="2" dur="1"><span>Third<br/>Line</span></p>
 642                 </div>
 643             </body>
 644             </tt>'''
 645         srt_data = '''1
 646 00:00:00,000 --> 00:00:01,000
 647 The following line contains Chinese characters and special symbols
 648
 649 2
 650 00:00:01,000 --> 00:00:02,000
 651 第二行
 652 ♪♪
 653
 654 3
 655 00:00:02,000 --> 00:00:03,000
 656 Third
 657 Line
 658
 659 '''
 660         self.assertEqual(dfxp2srt(dfxp_data), srt_data)
 661
 662         dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
 663             <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
 664             <body>
 665                 <div xml:lang="en">
 666                     <p begin="0" end="1">The first line</p>
 667                 </div>
 668             </body>
 669             </tt>'''
 670         srt_data = '''1
 671 00:00:00,000 --> 00:00:01,000
 672 The first line
 673
 674 '''
 675         self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
 676
 677
 678 if __name__ == '__main__':
 679     unittest.main()