_ Git - youtube-dl/blob - test/test_utils.py

   1 #!/usr/bin/env python
   2 # coding: utf-8
   3
   4 from __future__ import unicode_literals
   5
   6 # Allow direct execution
   7 import os
   8 import sys
   9 import unittest
  10 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
  11
  12
  13 # Various small unit tests
  14 import io
  15 import json
  16 import xml.etree.ElementTree
  17
  18 from youtube_dl.utils import (
  19     age_restricted,
  20     args_to_str,
  21     encode_base_n,
  22     clean_html,
  23     DateRange,
  24     detect_exe_version,
  25     determine_ext,
  26     dict_get,
  27     encode_compat_str,
  28     encodeFilename,
  29     escape_rfc3986,
  30     escape_url,
  31     ExtractorError,
  32     find_xpath_attr,
  33     fix_xml_ampersands,
  34     InAdvancePagedList,
  35     intlist_to_bytes,
  36     is_html,
  37     js_to_json,
  38     limit_length,
  39     ohdave_rsa_encrypt,
  40     OnDemandPagedList,
  41     orderedSet,
  42     parse_duration,
  43     parse_filesize,
  44     parse_count,
  45     parse_iso8601,
  46     read_batch_urls,
  47     sanitize_filename,
  48     sanitize_path,
  49     prepend_extension,
  50     replace_extension,
  51     remove_quotes,
  52     shell_quote,
  53     smuggle_url,
  54     str_to_int,
  55     strip_jsonp,
  56     struct_unpack,
  57     timeconvert,
  58     unescapeHTML,
  59     unified_strdate,
  60     unsmuggle_url,
  61     uppercase_escape,
  62     lowercase_escape,
  63     url_basename,
  64     urlencode_postdata,
  65     update_url_query,
  66     version_tuple,
  67     xpath_with_ns,
  68     xpath_element,
  69     xpath_text,
  70     xpath_attr,
  71     render_table,
  72     match_str,
  73     parse_dfxp_time_expr,
  74     dfxp2srt,
  75     cli_option,
  76     cli_valueless_option,
  77     cli_bool_option,
  78 )
  79 from youtube_dl.compat import (
  80     compat_etree_fromstring,
  81     compat_urlparse,
  82     compat_parse_qs,
  83 )
  84
  85
  86 class TestUtil(unittest.TestCase):
  87     def test_timeconvert(self):
  88         self.assertTrue(timeconvert('') is None)
  89         self.assertTrue(timeconvert('bougrg') is None)
  90
  91     def test_sanitize_filename(self):
  92         self.assertEqual(sanitize_filename('abc'), 'abc')
  93         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
  94
  95         self.assertEqual(sanitize_filename('123'), '123')
  96
  97         self.assertEqual('abc_de', sanitize_filename('abc/de'))
  98         self.assertFalse('/' in sanitize_filename('abc/de///'))
  99
 100         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
 101         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
 102         self.assertEqual('yes no', sanitize_filename('yes? no'))
 103         self.assertEqual('this - that', sanitize_filename('this: that'))
 104
 105         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
 106         aumlaut = 'ä'
 107         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
 108         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
 109         self.assertEqual(sanitize_filename(tests), tests)
 110
 111         self.assertEqual(
 112             sanitize_filename('New World record at 0:12:34'),
 113             'New World record at 0_12_34')
 114
 115         self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
 116         self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
 117         self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
 118         self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
 119
 120         forbidden = '"\0\\/'
 121         for fc in forbidden:
 122             for fbc in forbidden:
 123                 self.assertTrue(fbc not in sanitize_filename(fc))
 124
 125     def test_sanitize_filename_restricted(self):
 126         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
 127         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
 128
 129         self.assertEqual(sanitize_filename('123', restricted=True), '123')
 130
 131         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
 132         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
 133
 134         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
 135         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
 136         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
 137         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
 138
 139         tests = 'a\xe4b\u4e2d\u56fd\u7684c'
 140         self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
 141         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
 142
 143         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
 144         for fc in forbidden:
 145             for fbc in forbidden:
 146                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
 147
 148         # Handle a common case more neatly
 149         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
 150         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
 151         # .. but make sure the file name is never empty
 152         self.assertTrue(sanitize_filename('-', restricted=True) != '')
 153         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 154
 155     def test_sanitize_ids(self):
 156         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 157         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 158         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 159
 160     def test_sanitize_path(self):
 161         if sys.platform != 'win32':
 162             return
 163
 164         self.assertEqual(sanitize_path('abc'), 'abc')
 165         self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
 166         self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
 167         self.assertEqual(sanitize_path('abc|def'), 'abc#def')
 168         self.assertEqual(sanitize_path('<>:"|?*'), '#######')
 169         self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
 170         self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
 171
 172         self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
 173         self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
 174
 175         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 176         self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
 177         self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
 178         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
 179
 180         self.assertEqual(
 181             sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
 182             'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
 183
 184         self.assertEqual(
 185             sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
 186             'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
 187         self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
 188         self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
 189         self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
 190
 191         self.assertEqual(sanitize_path('../abc'), '..\\abc')
 192         self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
 193         self.assertEqual(sanitize_path('./abc'), 'abc')
 194         self.assertEqual(sanitize_path('./../abc'), '..\\abc')
 195
 196     def test_prepend_extension(self):
 197         self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
 198         self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
 199         self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 200         self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
 201         self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
 202         self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
 203
 204     def test_replace_extension(self):
 205         self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
 206         self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
 207         self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
 208         self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
 209         self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
 210         self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
 211
 212     def test_remove_quotes(self):
 213         self.assertEqual(remove_quotes(None), None)
 214         self.assertEqual(remove_quotes('"'), '"')
 215         self.assertEqual(remove_quotes("'"), "'")
 216         self.assertEqual(remove_quotes(';'), ';')
 217         self.assertEqual(remove_quotes('";'), '";')
 218         self.assertEqual(remove_quotes('""'), '')
 219         self.assertEqual(remove_quotes('";"'), ';')
 220
 221     def test_ordered_set(self):
 222         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
 223         self.assertEqual(orderedSet([]), [])
 224         self.assertEqual(orderedSet([1]), [1])
 225         # keep the list ordered
 226         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
 227
 228     def test_unescape_html(self):
 229         self.assertEqual(unescapeHTML('%20;'), '%20;')
 230         self.assertEqual(unescapeHTML('&#x2F;'), '/')
 231         self.assertEqual(unescapeHTML('&#47;'), '/')
 232         self.assertEqual(unescapeHTML('&eacute;'), 'é')
 233         self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
 234
 235     def test_daterange(self):
 236         _20century = DateRange("19000101", "20000101")
 237         self.assertFalse("17890714" in _20century)
 238         _ac = DateRange("00010101")
 239         self.assertTrue("19690721" in _ac)
 240         _firstmilenium = DateRange(end="10000101")
 241         self.assertTrue("07110427" in _firstmilenium)
 242
 243     def test_unified_dates(self):
 244         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
 245         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
 246         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
 247         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
 248         self.assertEqual(unified_strdate('1968 12 10'), '19681210')
 249         self.assertEqual(unified_strdate('1968-12-10'), '19681210')
 250         self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
 251         self.assertEqual(
 252             unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
 253             '20141126')
 254         self.assertEqual(
 255             unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
 256             '20150202')
 257         self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
 258         self.assertEqual(unified_strdate('25-09-2014'), '20140925')
 259         self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
 260
 261     def test_determine_ext(self):
 262         self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
 263         self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
 264         self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
 265         self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
 266         self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
 267
 268     def test_find_xpath_attr(self):
 269         testxml = '''<root>
 270             <node/>
 271             <node x="a"/>
 272             <node x="a" y="c" />
 273             <node x="b" y="d" />
 274             <node x="" />
 275         </root>'''
 276         doc = compat_etree_fromstring(testxml)
 277
 278         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
 279         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
 280         self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
 281         self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
 282         self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
 283         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
 284         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
 285         self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
 286         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
 287         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
 288         self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
 289
 290     def test_xpath_with_ns(self):
 291         testxml = '''<root xmlns:media="http://example.com/">
 292             <media:song>
 293                 <media:author>The Author</media:author>
 294                 <url>http://server.com/download.mp3</url>
 295             </media:song>
 296         </root>'''
 297         doc = compat_etree_fromstring(testxml)
 298         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
 299         self.assertTrue(find('media:song') is not None)
 300         self.assertEqual(find('media:song/media:author').text, 'The Author')
 301         self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
 302
 303     def test_xpath_element(self):
 304         doc = xml.etree.ElementTree.Element('root')
 305         div = xml.etree.ElementTree.SubElement(doc, 'div')
 306         p = xml.etree.ElementTree.SubElement(div, 'p')
 307         p.text = 'Foo'
 308         self.assertEqual(xpath_element(doc, 'div/p'), p)
 309         self.assertEqual(xpath_element(doc, ['div/p']), p)
 310         self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
 311         self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
 312         self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
 313         self.assertTrue(xpath_element(doc, 'div/bar') is None)
 314         self.assertTrue(xpath_element(doc, ['div/bar']) is None)
 315         self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
 316         self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
 317         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
 318         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
 319
 320     def test_xpath_text(self):
 321         testxml = '''<root>
 322             <div>
 323                 <p>Foo</p>
 324             </div>
 325         </root>'''
 326         doc = compat_etree_fromstring(testxml)
 327         self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
 328         self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
 329         self.assertTrue(xpath_text(doc, 'div/bar') is None)
 330         self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
 331
 332     def test_xpath_attr(self):
 333         testxml = '''<root>
 334             <div>
 335                 <p x="a">Foo</p>
 336             </div>
 337         </root>'''
 338         doc = compat_etree_fromstring(testxml)
 339         self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
 340         self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
 341         self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
 342         self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
 343         self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
 344         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
 345         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
 346
 347     def test_smuggle_url(self):
 348         data = {"ö": "ö", "abc": [3]}
 349         url = 'https://foo.bar/baz?x=y#a'
 350         smug_url = smuggle_url(url, data)
 351         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
 352         self.assertEqual(url, unsmug_url)
 353         self.assertEqual(data, unsmug_data)
 354
 355         res_url, res_data = unsmuggle_url(url)
 356         self.assertEqual(res_url, url)
 357         self.assertEqual(res_data, None)
 358
 359     def test_shell_quote(self):
 360         args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
 361         self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
 362
 363     def test_str_to_int(self):
 364         self.assertEqual(str_to_int('123,456'), 123456)
 365         self.assertEqual(str_to_int('123.456'), 123456)
 366
 367     def test_url_basename(self):
 368         self.assertEqual(url_basename('http://foo.de/'), '')
 369         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
 370         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
 371         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
 372         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
 373         self.assertEqual(
 374             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
 375             'trailer.mp4')
 376
 377     def test_parse_duration(self):
 378         self.assertEqual(parse_duration(None), None)
 379         self.assertEqual(parse_duration(False), None)
 380         self.assertEqual(parse_duration('invalid'), None)
 381         self.assertEqual(parse_duration('1'), 1)
 382         self.assertEqual(parse_duration('1337:12'), 80232)
 383         self.assertEqual(parse_duration('9:12:43'), 33163)
 384         self.assertEqual(parse_duration('12:00'), 720)
 385         self.assertEqual(parse_duration('00:01:01'), 61)
 386         self.assertEqual(parse_duration('x:y'), None)
 387         self.assertEqual(parse_duration('3h11m53s'), 11513)
 388         self.assertEqual(parse_duration('3h 11m 53s'), 11513)
 389         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
 390         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
 391         self.assertEqual(parse_duration('62m45s'), 3765)
 392         self.assertEqual(parse_duration('6m59s'), 419)
 393         self.assertEqual(parse_duration('49s'), 49)
 394         self.assertEqual(parse_duration('0h0m0s'), 0)
 395         self.assertEqual(parse_duration('0m0s'), 0)
 396         self.assertEqual(parse_duration('0s'), 0)
 397         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
 398         self.assertEqual(parse_duration('T30M38S'), 1838)
 399         self.assertEqual(parse_duration('5 s'), 5)
 400         self.assertEqual(parse_duration('3 min'), 180)
 401         self.assertEqual(parse_duration('2.5 hours'), 9000)
 402         self.assertEqual(parse_duration('02:03:04'), 7384)
 403         self.assertEqual(parse_duration('01:02:03:04'), 93784)
 404         self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
 405         self.assertEqual(parse_duration('87 Min.'), 5220)
 406
 407     def test_fix_xml_ampersands(self):
 408         self.assertEqual(
 409             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 410         self.assertEqual(
 411             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 412             '"&amp;x=y&amp;wrong;&amp;z=a')
 413         self.assertEqual(
 414             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 415             '&amp;&apos;&gt;&lt;&quot;')
 416         self.assertEqual(
 417             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 418         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 419
 420     def test_paged_list(self):
 421         def testPL(size, pagesize, sliceargs, expected):
 422             def get_page(pagenum):
 423                 firstid = pagenum * pagesize
 424                 upto = min(size, pagenum * pagesize + pagesize)
 425                 for i in range(firstid, upto):
 426                     yield i
 427
 428             pl = OnDemandPagedList(get_page, pagesize)
 429             got = pl.getslice(*sliceargs)
 430             self.assertEqual(got, expected)
 431
 432             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
 433             got = iapl.getslice(*sliceargs)
 434             self.assertEqual(got, expected)
 435
 436         testPL(5, 2, (), [0, 1, 2, 3, 4])
 437         testPL(5, 2, (1,), [1, 2, 3, 4])
 438         testPL(5, 2, (2,), [2, 3, 4])
 439         testPL(5, 2, (4,), [4])
 440         testPL(5, 2, (0, 3), [0, 1, 2])
 441         testPL(5, 2, (1, 4), [1, 2, 3])
 442         testPL(5, 2, (2, 99), [2, 3, 4])
 443         testPL(5, 2, (20, 99), [])
 444
 445     def test_struct_unpack(self):
 446         self.assertEqual(struct_unpack('!B', b'\x00'), (0,))
 447
 448     def test_read_batch_urls(self):
 449         f = io.StringIO('''\xef\xbb\xbf foo
 450             bar\r
 451             baz
 452             # More after this line\r
 453             ; or after this
 454             bam''')
 455         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
 456
 457     def test_urlencode_postdata(self):
 458         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
 459         self.assertTrue(isinstance(data, bytes))
 460
 461     def test_update_url_query(self):
 462         def query_dict(url):
 463             return compat_parse_qs(compat_urlparse.urlparse(url).query)
 464         self.assertEqual(query_dict(update_url_query(
 465             'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
 466             query_dict('http://example.com/path?quality=HD&format=mp4'))
 467         self.assertEqual(query_dict(update_url_query(
 468             'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
 469             query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
 470         self.assertEqual(query_dict(update_url_query(
 471             'http://example.com/path', {'fields': 'id,formats,subtitles'})),
 472             query_dict('http://example.com/path?fields=id,formats,subtitles'))
 473         self.assertEqual(query_dict(update_url_query(
 474             'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
 475             query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
 476         self.assertEqual(query_dict(update_url_query(
 477             'http://example.com/path?manifest=f4m', {'manifest': []})),
 478             query_dict('http://example.com/path'))
 479         self.assertEqual(query_dict(update_url_query(
 480             'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
 481             query_dict('http://example.com/path?system=LINUX'))
 482         self.assertEqual(query_dict(update_url_query(
 483             'http://example.com/path', {'fields': b'id,formats,subtitles'})),
 484             query_dict('http://example.com/path?fields=id,formats,subtitles'))
 485         self.assertEqual(query_dict(update_url_query(
 486             'http://example.com/path', {'width': 1080, 'height': 720})),
 487             query_dict('http://example.com/path?width=1080&height=720'))
 488         self.assertEqual(query_dict(update_url_query(
 489             'http://example.com/path', {'bitrate': 5020.43})),
 490             query_dict('http://example.com/path?bitrate=5020.43'))
 491         self.assertEqual(query_dict(update_url_query(
 492             'http://example.com/path', {'test': '第二行тест'})),
 493             query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
 494
 495     def test_dict_get(self):
 496         FALSE_VALUES = {
 497             'none': None,
 498             'false': False,
 499             'zero': 0,
 500             'empty_string': '',
 501             'empty_list': [],
 502         }
 503         d = FALSE_VALUES.copy()
 504         d['a'] = 42
 505         self.assertEqual(dict_get(d, 'a'), 42)
 506         self.assertEqual(dict_get(d, 'b'), None)
 507         self.assertEqual(dict_get(d, 'b', 42), 42)
 508         self.assertEqual(dict_get(d, ('a', )), 42)
 509         self.assertEqual(dict_get(d, ('b', 'a', )), 42)
 510         self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
 511         self.assertEqual(dict_get(d, ('b', 'c', )), None)
 512         self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
 513         for key, false_value in FALSE_VALUES.items():
 514             self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
 515             self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
 516
 517     def test_encode_compat_str(self):
 518         self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
 519         self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
 520
 521     def test_parse_iso8601(self):
 522         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
 523         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
 524         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
 525         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
 526         self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
 527         self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
 528
 529     def test_strip_jsonp(self):
 530         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
 531         d = json.loads(stripped)
 532         self.assertEqual(d, [{"id": "532cb", "x": 3}])
 533
 534         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
 535         d = json.loads(stripped)
 536         self.assertEqual(d, {'STATUS': 'OK'})
 537
 538         stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
 539         d = json.loads(stripped)
 540         self.assertEqual(d, {'status': 'success'})
 541
 542     def test_uppercase_escape(self):
 543         self.assertEqual(uppercase_escape('aä'), 'aä')
 544         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
 545
 546     def test_lowercase_escape(self):
 547         self.assertEqual(lowercase_escape('aä'), 'aä')
 548         self.assertEqual(lowercase_escape('\\u0026'), '&')
 549
 550     def test_limit_length(self):
 551         self.assertEqual(limit_length(None, 12), None)
 552         self.assertEqual(limit_length('foo', 12), 'foo')
 553         self.assertTrue(
 554             limit_length('foo bar baz asd', 12).startswith('foo bar'))
 555         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
 556
 557     def test_escape_rfc3986(self):
 558         reserved = "!*'();:@&=+$,/?#[]"
 559         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
 560         self.assertEqual(escape_rfc3986(reserved), reserved)
 561         self.assertEqual(escape_rfc3986(unreserved), unreserved)
 562         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
 563         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
 564         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
 565         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
 566
 567     def test_escape_url(self):
 568         self.assertEqual(
 569             escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
 570             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
 571         )
 572         self.assertEqual(
 573             escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
 574             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
 575         )
 576         self.assertEqual(
 577             escape_url('http://тест.рф/фрагмент'),
 578             'http://тест.рф/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
 579         )
 580         self.assertEqual(
 581             escape_url('http://тест.рф/абв?абв=абв#абв'),
 582             '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'
 583         )
 584         self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
 585
 586     def test_js_to_json_realworld(self):
 587         inp = '''{
 588             'clip':{'provider':'pseudo'}
 589         }'''
 590         self.assertEqual(js_to_json(inp), '''{
 591             "clip":{"provider":"pseudo"}
 592         }''')
 593         json.loads(js_to_json(inp))
 594
 595         inp = '''{
 596             'playlist':[{'controls':{'all':null}}]
 597         }'''
 598         self.assertEqual(js_to_json(inp), '''{
 599             "playlist":[{"controls":{"all":null}}]
 600         }''')
 601
 602         inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
 603         self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
 604
 605         inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
 606         json_code = js_to_json(inp)
 607         self.assertEqual(json.loads(json_code), json.loads(inp))
 608
 609     def test_js_to_json_edgecases(self):
 610         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
 611         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
 612
 613         on = js_to_json('{"abc": true}')
 614         self.assertEqual(json.loads(on), {'abc': True})
 615
 616         # Ignore JavaScript code as well
 617         on = js_to_json('''{
 618             "x": 1,
 619             y: "a",
 620             z: some.code
 621         }''')
 622         d = json.loads(on)
 623         self.assertEqual(d['x'], 1)
 624         self.assertEqual(d['y'], 'a')
 625
 626         on = js_to_json('["abc", "def",]')
 627         self.assertEqual(json.loads(on), ['abc', 'def'])
 628
 629         on = js_to_json('{"abc": "def",}')
 630         self.assertEqual(json.loads(on), {'abc': 'def'})
 631
 632     def test_clean_html(self):
 633         self.assertEqual(clean_html('a:\nb'), 'a: b')
 634         self.assertEqual(clean_html('a:\n   "b"'), 'a:    "b"')
 635
 636     def test_intlist_to_bytes(self):
 637         self.assertEqual(
 638             intlist_to_bytes([0, 1, 127, 128, 255]),
 639             b'\x00\x01\x7f\x80\xff')
 640
 641     def test_args_to_str(self):
 642         self.assertEqual(
 643             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
 644             'foo ba/r -baz \'2 be\' \'\''
 645         )
 646
 647     def test_parse_filesize(self):
 648         self.assertEqual(parse_filesize(None), None)
 649         self.assertEqual(parse_filesize(''), None)
 650         self.assertEqual(parse_filesize('91 B'), 91)
 651         self.assertEqual(parse_filesize('foobar'), None)
 652         self.assertEqual(parse_filesize('2 MiB'), 2097152)
 653         self.assertEqual(parse_filesize('5 GB'), 5000000000)
 654         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
 655         self.assertEqual(parse_filesize('1,24 KB'), 1240)
 656
 657     def test_parse_count(self):
 658         self.assertEqual(parse_count(None), None)
 659         self.assertEqual(parse_count(''), None)
 660         self.assertEqual(parse_count('0'), 0)
 661         self.assertEqual(parse_count('1000'), 1000)
 662         self.assertEqual(parse_count('1.000'), 1000)
 663         self.assertEqual(parse_count('1.1k'), 1100)
 664         self.assertEqual(parse_count('1.1kk'), 1100000)
 665
 666     def test_version_tuple(self):
 667         self.assertEqual(version_tuple('1'), (1,))
 668         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
 669         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
 670
 671     def test_detect_exe_version(self):
 672         self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
 673 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
 674 configuration: --prefix=/usr --extra-'''), '1.2.1')
 675         self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
 676 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
 677         self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
 678 Trying to open render node...
 679 Success at /dev/dri/renderD128.
 680 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
 681
 682     def test_age_restricted(self):
 683         self.assertFalse(age_restricted(None, 10))  # unrestricted content
 684         self.assertFalse(age_restricted(1, None))  # unrestricted policy
 685         self.assertFalse(age_restricted(8, 10))
 686         self.assertTrue(age_restricted(18, 14))
 687         self.assertFalse(age_restricted(18, 18))
 688
 689     def test_is_html(self):
 690         self.assertFalse(is_html(b'\x49\x44\x43<html'))
 691         self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
 692         self.assertTrue(is_html(  # UTF-8 with BOM
 693             b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
 694         self.assertTrue(is_html(  # UTF-16-LE
 695             b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
 696         ))
 697         self.assertTrue(is_html(  # UTF-16-BE
 698             b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
 699         ))
 700         self.assertTrue(is_html(  # UTF-32-BE
 701             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'))
 702         self.assertTrue(is_html(  # UTF-32-LE
 703             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'))
 704
 705     def test_render_table(self):
 706         self.assertEqual(
 707             render_table(
 708                 ['a', 'bcd'],
 709                 [[123, 4], [9999, 51]]),
 710             'a    bcd\n'
 711             '123  4\n'
 712             '9999 51')
 713
 714     def test_match_str(self):
 715         self.assertRaises(ValueError, match_str, 'xy>foobar', {})
 716         self.assertFalse(match_str('xy', {'x': 1200}))
 717         self.assertTrue(match_str('!xy', {'x': 1200}))
 718         self.assertTrue(match_str('x', {'x': 1200}))
 719         self.assertFalse(match_str('!x', {'x': 1200}))
 720         self.assertTrue(match_str('x', {'x': 0}))
 721         self.assertFalse(match_str('x>0', {'x': 0}))
 722         self.assertFalse(match_str('x>0', {}))
 723         self.assertTrue(match_str('x>?0', {}))
 724         self.assertTrue(match_str('x>1K', {'x': 1200}))
 725         self.assertFalse(match_str('x>2K', {'x': 1200}))
 726         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
 727         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
 728         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
 729         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
 730         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
 731         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
 732         self.assertFalse(match_str(
 733             'like_count > 100 & dislike_count <? 50 & description',
 734             {'like_count': 90, 'description': 'foo'}))
 735         self.assertTrue(match_str(
 736             'like_count > 100 & dislike_count <? 50 & description',
 737             {'like_count': 190, 'description': 'foo'}))
 738         self.assertFalse(match_str(
 739             'like_count > 100 & dislike_count <? 50 & description',
 740             {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
 741         self.assertFalse(match_str(
 742             'like_count > 100 & dislike_count <? 50 & description',
 743             {'like_count': 190, 'dislike_count': 10}))
 744
 745     def test_parse_dfxp_time_expr(self):
 746         self.assertEqual(parse_dfxp_time_expr(None), None)
 747         self.assertEqual(parse_dfxp_time_expr(''), None)
 748         self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
 749         self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
 750         self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
 751         self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
 752         self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
 753
 754     def test_dfxp2srt(self):
 755         dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
 756             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
 757             <body>
 758                 <div xml:lang="en">
 759                     <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
 760                     <p begin="1" end="2">第二行<br/>♪♪</p>
 761                     <p begin="2" dur="1"><span>Third<br/>Line</span></p>
 762                     <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
 763                     <p begin="-1" end="-1">Ignore, two</p>
 764                     <p begin="3" dur="-1">Ignored, three</p>
 765                 </div>
 766             </body>
 767             </tt>'''
 768         srt_data = '''1
 769 00:00:00,000 --> 00:00:01,000
 770 The following line contains Chinese characters and special symbols
 771
 772 2
 773 00:00:01,000 --> 00:00:02,000
 774 第二行
 775 ♪♪
 776
 777 3
 778 00:00:02,000 --> 00:00:03,000
 779 Third
 780 Line
 781
 782 '''
 783         self.assertEqual(dfxp2srt(dfxp_data), srt_data)
 784
 785         dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
 786             <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
 787             <body>
 788                 <div xml:lang="en">
 789                     <p begin="0" end="1">The first line</p>
 790                 </div>
 791             </body>
 792             </tt>'''
 793         srt_data = '''1
 794 00:00:00,000 --> 00:00:01,000
 795 The first line
 796
 797 '''
 798         self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
 799
 800     def test_cli_option(self):
 801         self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
 802         self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
 803         self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
 804
 805     def test_cli_valueless_option(self):
 806         self.assertEqual(cli_valueless_option(
 807             {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
 808         self.assertEqual(cli_valueless_option(
 809             {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
 810         self.assertEqual(cli_valueless_option(
 811             {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
 812         self.assertEqual(cli_valueless_option(
 813             {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
 814         self.assertEqual(cli_valueless_option(
 815             {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
 816         self.assertEqual(cli_valueless_option(
 817             {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
 818
 819     def test_cli_bool_option(self):
 820         self.assertEqual(
 821             cli_bool_option(
 822                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
 823             ['--no-check-certificate', 'true'])
 824         self.assertEqual(
 825             cli_bool_option(
 826                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
 827             ['--no-check-certificate=true'])
 828         self.assertEqual(
 829             cli_bool_option(
 830                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
 831             ['--check-certificate', 'false'])
 832         self.assertEqual(
 833             cli_bool_option(
 834                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
 835             ['--check-certificate=false'])
 836         self.assertEqual(
 837             cli_bool_option(
 838                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
 839             ['--check-certificate', 'true'])
 840         self.assertEqual(
 841             cli_bool_option(
 842                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
 843             ['--check-certificate=true'])
 844
 845     def test_ohdave_rsa_encrypt(self):
 846         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
 847         e = 65537
 848
 849         self.assertEqual(
 850             ohdave_rsa_encrypt(b'aa111222', e, N),
 851             '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
 852
 853     def test_encode_base_n(self):
 854         self.assertEqual(encode_base_n(0, 30), '0')
 855         self.assertEqual(encode_base_n(80, 30), '2k')
 856
 857         custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
 858         self.assertEqual(encode_base_n(0, 30, custom_table), '9')
 859         self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
 860
 861         self.assertRaises(ValueError, encode_base_n, 0, 70)
 862         self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
 863
 864 if __name__ == '__main__':
 865     unittest.main()