4 from __future__
import unicode_literals
6 # Allow direct execution
10 sys
.path
.insert(0, os
.path
.dirname(os
.path
.dirname(os
.path
.abspath(__file__
))))
13 # Various small unit tests
16 import xml
.etree
.ElementTree
18 from youtube_dl
.utils
import (
38 get_element_by_attribute
,
39 get_elements_by_class
,
40 get_elements_by_attribute
,
100 cli_valueless_option
,
104 from youtube_dl
.compat
import (
106 compat_etree_fromstring
,
115 class TestUtil(unittest
.TestCase
):
116 def test_timeconvert(self
):
117 self
.assertTrue(timeconvert('') is None)
118 self
.assertTrue(timeconvert('bougrg') is None)
120 def test_sanitize_filename(self
):
121 self
.assertEqual(sanitize_filename('abc'), 'abc')
122 self
.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
124 self
.assertEqual(sanitize_filename('123'), '123')
126 self
.assertEqual('abc_de', sanitize_filename('abc/de'))
127 self
.assertFalse('/' in sanitize_filename('abc/de///'))
129 self
.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
130 self
.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
131 self
.assertEqual('yes no', sanitize_filename('yes? no'))
132 self
.assertEqual('this - that', sanitize_filename('this: that'))
134 self
.assertEqual(sanitize_filename('AT&T'), 'AT&T')
136 self
.assertEqual(sanitize_filename(aumlaut
), aumlaut
)
137 tests
= '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
138 self
.assertEqual(sanitize_filename(tests
), tests
)
141 sanitize_filename('New World record at 0:12:34'),
142 'New World record at 0_12_34')
144 self
.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
145 self
.assertEqual(sanitize_filename('--gasdgf', is_id
=True), '--gasdgf')
146 self
.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
147 self
.assertEqual(sanitize_filename('.gasdgf', is_id
=True), '.gasdgf')
151 for fbc
in forbidden
:
152 self
.assertTrue(fbc
not in sanitize_filename(fc
))
154 def test_sanitize_filename_restricted(self
):
155 self
.assertEqual(sanitize_filename('abc', restricted
=True), 'abc')
156 self
.assertEqual(sanitize_filename('abc_d-e', restricted
=True), 'abc_d-e')
158 self
.assertEqual(sanitize_filename('123', restricted
=True), '123')
160 self
.assertEqual('abc_de', sanitize_filename('abc/de', restricted
=True))
161 self
.assertFalse('/' in sanitize_filename('abc/de///', restricted
=True))
163 self
.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted
=True))
164 self
.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted
=True))
165 self
.assertEqual('yes_no', sanitize_filename('yes? no', restricted
=True))
166 self
.assertEqual('this_-_that', sanitize_filename('this: that', restricted
=True))
168 tests
= 'aäb\u4e2d\u56fd\u7684c'
169 self
.assertEqual(sanitize_filename(tests
, restricted
=True), 'aab_c')
170 self
.assertTrue(sanitize_filename('\xf6', restricted
=True) != '') # No empty filename
172 forbidden
= '"\0\\/&!: \'\t\n()[]{}$;`^,#'
174 for fbc
in forbidden
:
175 self
.assertTrue(fbc
not in sanitize_filename(fc
, restricted
=True))
177 # Handle a common case more neatly
178 self
.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted
=True), 'Song')
179 self
.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted
=True), 'Speech')
180 # .. but make sure the file name is never empty
181 self
.assertTrue(sanitize_filename('-', restricted
=True) != '')
182 self
.assertTrue(sanitize_filename(':', restricted
=True) != '')
184 self
.assertEqual(sanitize_filename(
185 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted
=True),
186 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy')
188 def test_sanitize_ids(self
):
189 self
.assertEqual(sanitize_filename('_n_cd26wFpw', is_id
=True), '_n_cd26wFpw')
190 self
.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id
=True), '_BD_eEpuzXw')
191 self
.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id
=True), 'N0Y__7-UOdI')
193 def test_sanitize_path(self
):
194 if sys
.platform
!= 'win32':
197 self
.assertEqual(sanitize_path('abc'), 'abc')
198 self
.assertEqual(sanitize_path('abc/def'), 'abc\\def')
199 self
.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
200 self
.assertEqual(sanitize_path('abc|def'), 'abc#def')
201 self
.assertEqual(sanitize_path('<>:"|?*'), '#######')
202 self
.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
203 self
.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
205 self
.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
206 self
.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
208 self
.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
209 self
.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
210 self
.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
211 self
.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
214 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
215 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
218 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
219 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
220 self
.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
221 self
.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
222 self
.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
224 self
.assertEqual(sanitize_path('../abc'), '..\\abc')
225 self
.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
226 self
.assertEqual(sanitize_path('./abc'), 'abc')
227 self
.assertEqual(sanitize_path('./../abc'), '..\\abc')
229 def test_sanitize_url(self
):
230 self
.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
231 self
.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
232 self
.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
233 self
.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')
235 def test_expand_path(self
):
237 return '%{0}%'.format(var
) if sys
.platform
== 'win32' else '${0}'.format(var
)
239 compat_setenv('YOUTUBE_DL_EXPATH_PATH', 'expanded')
240 self
.assertEqual(expand_path(env('YOUTUBE_DL_EXPATH_PATH')), 'expanded')
241 self
.assertEqual(expand_path(env('HOME')), compat_getenv('HOME'))
242 self
.assertEqual(expand_path('~'), compat_getenv('HOME'))
244 expand_path('~/%s' % env('YOUTUBE_DL_EXPATH_PATH')),
245 '%s/expanded' % compat_getenv('HOME'))
247 def test_prepend_extension(self
):
248 self
.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
249 self
.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
250 self
.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
251 self
.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
252 self
.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
253 self
.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
255 def test_replace_extension(self
):
256 self
.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
257 self
.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
258 self
.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
259 self
.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
260 self
.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
261 self
.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
263 def test_remove_start(self
):
264 self
.assertEqual(remove_start(None, 'A - '), None)
265 self
.assertEqual(remove_start('A - B', 'A - '), 'B')
266 self
.assertEqual(remove_start('B - A', 'A - '), 'B - A')
268 def test_remove_end(self
):
269 self
.assertEqual(remove_end(None, ' - B'), None)
270 self
.assertEqual(remove_end('A - B', ' - B'), 'A')
271 self
.assertEqual(remove_end('B - A', ' - B'), 'B - A')
273 def test_remove_quotes(self
):
274 self
.assertEqual(remove_quotes(None), None)
275 self
.assertEqual(remove_quotes('"'), '"')
276 self
.assertEqual(remove_quotes("'"), "'")
277 self
.assertEqual(remove_quotes(';'), ';')
278 self
.assertEqual(remove_quotes('";'), '";')
279 self
.assertEqual(remove_quotes('""'), '')
280 self
.assertEqual(remove_quotes('";"'), ';')
282 def test_ordered_set(self
):
283 self
.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
284 self
.assertEqual(orderedSet([]), [])
285 self
.assertEqual(orderedSet([1]), [1])
286 # keep the list ordered
287 self
.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
289 def test_unescape_html(self
):
290 self
.assertEqual(unescapeHTML('%20;'), '%20;')
291 self
.assertEqual(unescapeHTML('/'), '/')
292 self
.assertEqual(unescapeHTML('/'), '/')
293 self
.assertEqual(unescapeHTML('é'), 'é')
294 self
.assertEqual(unescapeHTML('�'), '�')
295 self
.assertEqual(unescapeHTML('&a"'), '&a"')
297 self
.assertEqual(unescapeHTML('.''), '.\'')
299 def test_date_from_str(self
):
300 self
.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
301 self
.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
302 self
.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
303 self
.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
304 self
.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
306 def test_daterange(self
):
307 _20century
= DateRange("19000101", "20000101")
308 self
.assertFalse("17890714" in _20century
)
309 _ac
= DateRange("00010101")
310 self
.assertTrue("19690721" in _ac
)
311 _firstmilenium
= DateRange(end
="10000101")
312 self
.assertTrue("07110427" in _firstmilenium
)
314 def test_unified_dates(self
):
315 self
.assertEqual(unified_strdate('December 21, 2010'), '20101221')
316 self
.assertEqual(unified_strdate('8/7/2009'), '20090708')
317 self
.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
318 self
.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
319 self
.assertEqual(unified_strdate('1968 12 10'), '19681210')
320 self
.assertEqual(unified_strdate('1968-12-10'), '19681210')
321 self
.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
323 unified_strdate('11/26/2014 11:30:00 AM PST', day_first
=False),
326 unified_strdate('2/2/2015 6:47:40 PM', day_first
=False),
328 self
.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
329 self
.assertEqual(unified_strdate('25-09-2014'), '20140925')
330 self
.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
331 self
.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
332 self
.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
333 self
.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
334 self
.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
335 self
.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
337 def test_unified_timestamps(self
):
338 self
.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
339 self
.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
340 self
.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
341 self
.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
342 self
.assertEqual(unified_timestamp('1968 12 10'), -33436800)
343 self
.assertEqual(unified_timestamp('1968-12-10'), -33436800)
344 self
.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
346 unified_timestamp('11/26/2014 11:30:00 AM PST', day_first
=False),
349 unified_timestamp('2/2/2015 6:47:40 PM', day_first
=False),
351 self
.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
352 self
.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
353 self
.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
354 self
.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
355 self
.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
356 self
.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
357 self
.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
358 self
.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
359 self
.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
360 self
.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363)
362 def test_determine_ext(self
):
363 self
.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
364 self
.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
365 self
.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
366 self
.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
367 self
.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
368 self
.assertEqual(determine_ext('foobar', None), None)
370 def test_find_xpath_attr(self
):
378 doc
= compat_etree_fromstring(testxml
)
380 self
.assertEqual(find_xpath_attr(doc
, './/fourohfour', 'n'), None)
381 self
.assertEqual(find_xpath_attr(doc
, './/fourohfour', 'n', 'v'), None)
382 self
.assertEqual(find_xpath_attr(doc
, './/node', 'n'), None)
383 self
.assertEqual(find_xpath_attr(doc
, './/node', 'n', 'v'), None)
384 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x'), doc
[1])
385 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x', 'a'), doc
[1])
386 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x', 'b'), doc
[3])
387 self
.assertEqual(find_xpath_attr(doc
, './/node', 'y'), doc
[2])
388 self
.assertEqual(find_xpath_attr(doc
, './/node', 'y', 'c'), doc
[2])
389 self
.assertEqual(find_xpath_attr(doc
, './/node', 'y', 'd'), doc
[3])
390 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x', ''), doc
[4])
392 def test_xpath_with_ns(self
):
393 testxml
= '''<root xmlns:media="http://example.com/">
395 <media:author>The Author</media:author>
396 <url>http://server.com/download.mp3</url>
399 doc
= compat_etree_fromstring(testxml
)
400 find
= lambda p
: doc
.find(xpath_with_ns(p
, {'media': 'http://example.com/'}
))
401 self
.assertTrue(find('media:song') is not None)
402 self
.assertEqual(find('media:song/media:author').text
, 'The Author')
403 self
.assertEqual(find('media:song/url').text
, 'http://server.com/download.mp3')
405 def test_xpath_element(self
):
406 doc
= xml
.etree
.ElementTree
.Element('root')
407 div
= xml
.etree
.ElementTree
.SubElement(doc
, 'div')
408 p
= xml
.etree
.ElementTree
.SubElement(div
, 'p')
410 self
.assertEqual(xpath_element(doc
, 'div/p'), p
)
411 self
.assertEqual(xpath_element(doc
, ['div/p']), p
)
412 self
.assertEqual(xpath_element(doc
, ['div/bar', 'div/p']), p
)
413 self
.assertEqual(xpath_element(doc
, 'div/bar', default
='default'), 'default')
414 self
.assertEqual(xpath_element(doc
, ['div/bar'], default
='default'), 'default')
415 self
.assertTrue(xpath_element(doc
, 'div/bar') is None)
416 self
.assertTrue(xpath_element(doc
, ['div/bar']) is None)
417 self
.assertTrue(xpath_element(doc
, ['div/bar'], 'div/baz') is None)
418 self
.assertRaises(ExtractorError
, xpath_element
, doc
, 'div/bar', fatal
=True)
419 self
.assertRaises(ExtractorError
, xpath_element
, doc
, ['div/bar'], fatal
=True)
420 self
.assertRaises(ExtractorError
, xpath_element
, doc
, ['div/bar', 'div/baz'], fatal
=True)
422 def test_xpath_text(self
):
428 doc
= compat_etree_fromstring(testxml
)
429 self
.assertEqual(xpath_text(doc
, 'div/p'), 'Foo')
430 self
.assertEqual(xpath_text(doc
, 'div/bar', default
='default'), 'default')
431 self
.assertTrue(xpath_text(doc
, 'div/bar') is None)
432 self
.assertRaises(ExtractorError
, xpath_text
, doc
, 'div/bar', fatal
=True)
434 def test_xpath_attr(self
):
440 doc
= compat_etree_fromstring(testxml
)
441 self
.assertEqual(xpath_attr(doc
, 'div/p', 'x'), 'a')
442 self
.assertEqual(xpath_attr(doc
, 'div/bar', 'x'), None)
443 self
.assertEqual(xpath_attr(doc
, 'div/p', 'y'), None)
444 self
.assertEqual(xpath_attr(doc
, 'div/bar', 'x', default
='default'), 'default')
445 self
.assertEqual(xpath_attr(doc
, 'div/p', 'y', default
='default'), 'default')
446 self
.assertRaises(ExtractorError
, xpath_attr
, doc
, 'div/bar', 'x', fatal
=True)
447 self
.assertRaises(ExtractorError
, xpath_attr
, doc
, 'div/p', 'y', fatal
=True)
449 def test_smuggle_url(self
):
450 data
= {"ö": "ö", "abc": [3]}
451 url
= 'https://foo.bar/baz?x=y#a'
452 smug_url
= smuggle_url(url
, data
)
453 unsmug_url
, unsmug_data
= unsmuggle_url(smug_url
)
454 self
.assertEqual(url
, unsmug_url
)
455 self
.assertEqual(data
, unsmug_data
)
457 res_url
, res_data
= unsmuggle_url(url
)
458 self
.assertEqual(res_url
, url
)
459 self
.assertEqual(res_data
, None)
461 smug_url
= smuggle_url(url
, {'a': 'b'}
)
462 smug_smug_url
= smuggle_url(smug_url
, {'c': 'd'}
)
463 res_url
, res_data
= unsmuggle_url(smug_smug_url
)
464 self
.assertEqual(res_url
, url
)
465 self
.assertEqual(res_data
, {'a': 'b', 'c': 'd'}
)
467 def test_shell_quote(self
):
468 args
= ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
471 """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name
!= 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
473 def test_float_or_none(self
):
474 self
.assertEqual(float_or_none('42.42'), 42.42)
475 self
.assertEqual(float_or_none('42'), 42.0)
476 self
.assertEqual(float_or_none(''), None)
477 self
.assertEqual(float_or_none(None), None)
478 self
.assertEqual(float_or_none([]), None)
479 self
.assertEqual(float_or_none(set()), None)
481 def test_int_or_none(self
):
482 self
.assertEqual(int_or_none('42'), 42)
483 self
.assertEqual(int_or_none(''), None)
484 self
.assertEqual(int_or_none(None), None)
485 self
.assertEqual(int_or_none([]), None)
486 self
.assertEqual(int_or_none(set()), None)
488 def test_str_to_int(self
):
489 self
.assertEqual(str_to_int('123,456'), 123456)
490 self
.assertEqual(str_to_int('123.456'), 123456)
492 def test_url_basename(self
):
493 self
.assertEqual(url_basename('http://foo.de/'), '')
494 self
.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
495 self
.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
496 self
.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
497 self
.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
499 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
502 def test_base_url(self
):
503 self
.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
504 self
.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
505 self
.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
506 self
.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
507 self
.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
509 def test_urljoin(self
):
510 self
.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
511 self
.assertEqual(urljoin(b
'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
512 self
.assertEqual(urljoin('http://foo.de/', b
'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
513 self
.assertEqual(urljoin(b
'http://foo.de/', b
'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
514 self
.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
515 self
.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
516 self
.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
517 self
.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
518 self
.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
519 self
.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
520 self
.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
521 self
.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
522 self
.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
523 self
.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
524 self
.assertEqual(urljoin('http://foo.de/', None), None)
525 self
.assertEqual(urljoin('http://foo.de/', ''), None)
526 self
.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
527 self
.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
528 self
.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de')
529 self
.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de')
531 def test_url_or_none(self
):
532 self
.assertEqual(url_or_none(None), None)
533 self
.assertEqual(url_or_none(''), None)
534 self
.assertEqual(url_or_none('foo'), None)
535 self
.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
536 self
.assertEqual(url_or_none('https://foo.de'), 'https://foo.de')
537 self
.assertEqual(url_or_none('http$://foo.de'), None)
538 self
.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
539 self
.assertEqual(url_or_none('//foo.de'), '//foo.de')
541 def test_parse_age_limit(self
):
542 self
.assertEqual(parse_age_limit(None), None)
543 self
.assertEqual(parse_age_limit(False), None)
544 self
.assertEqual(parse_age_limit('invalid'), None)
545 self
.assertEqual(parse_age_limit(0), 0)
546 self
.assertEqual(parse_age_limit(18), 18)
547 self
.assertEqual(parse_age_limit(21), 21)
548 self
.assertEqual(parse_age_limit(22), None)
549 self
.assertEqual(parse_age_limit('18'), 18)
550 self
.assertEqual(parse_age_limit('18+'), 18)
551 self
.assertEqual(parse_age_limit('PG-13'), 13)
552 self
.assertEqual(parse_age_limit('TV-14'), 14)
553 self
.assertEqual(parse_age_limit('TV-MA'), 17)
554 self
.assertEqual(parse_age_limit('TV14'), 14)
555 self
.assertEqual(parse_age_limit('TV_G'), 0)
557 def test_parse_duration(self
):
558 self
.assertEqual(parse_duration(None), None)
559 self
.assertEqual(parse_duration(False), None)
560 self
.assertEqual(parse_duration('invalid'), None)
561 self
.assertEqual(parse_duration('1'), 1)
562 self
.assertEqual(parse_duration('1337:12'), 80232)
563 self
.assertEqual(parse_duration('9:12:43'), 33163)
564 self
.assertEqual(parse_duration('12:00'), 720)
565 self
.assertEqual(parse_duration('00:01:01'), 61)
566 self
.assertEqual(parse_duration('x:y'), None)
567 self
.assertEqual(parse_duration('3h11m53s'), 11513)
568 self
.assertEqual(parse_duration('3h 11m 53s'), 11513)
569 self
.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
570 self
.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
571 self
.assertEqual(parse_duration('62m45s'), 3765)
572 self
.assertEqual(parse_duration('6m59s'), 419)
573 self
.assertEqual(parse_duration('49s'), 49)
574 self
.assertEqual(parse_duration('0h0m0s'), 0)
575 self
.assertEqual(parse_duration('0m0s'), 0)
576 self
.assertEqual(parse_duration('0s'), 0)
577 self
.assertEqual(parse_duration('01:02:03.05'), 3723.05)
578 self
.assertEqual(parse_duration('T30M38S'), 1838)
579 self
.assertEqual(parse_duration('5 s'), 5)
580 self
.assertEqual(parse_duration('3 min'), 180)
581 self
.assertEqual(parse_duration('2.5 hours'), 9000)
582 self
.assertEqual(parse_duration('02:03:04'), 7384)
583 self
.assertEqual(parse_duration('01:02:03:04'), 93784)
584 self
.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
585 self
.assertEqual(parse_duration('87 Min.'), 5220)
586 self
.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
587 self
.assertEqual(parse_duration('PT00H03M30SZ'), 210)
588 self
.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)
590 def test_fix_xml_ampersands(self
):
592 fix_xml_ampersands('"&x=y&z=a'), '"&x=y&z=a')
594 fix_xml_ampersands('"&x=y&wrong;&z=a'),
595 '"&x=y&wrong;&z=a')
597 fix_xml_ampersands('&'><"'),
598 '&'><"')
600 fix_xml_ampersands('Ӓ᪼'), 'Ӓ᪼')
601 self
.assertEqual(fix_xml_ampersands('&#&#'), '&#&#')
603 def test_paged_list(self
):
604 def testPL(size
, pagesize
, sliceargs
, expected
):
605 def get_page(pagenum
):
606 firstid
= pagenum
* pagesize
607 upto
= min(size
, pagenum
* pagesize
+ pagesize
)
608 for i
in range(firstid
, upto
):
611 pl
= OnDemandPagedList(get_page
, pagesize
)
612 got
= pl
.getslice(*sliceargs
)
613 self
.assertEqual(got
, expected
)
615 iapl
= InAdvancePagedList(get_page
, size
// pagesize
+ 1, pagesize
)
616 got
= iapl
.getslice(*sliceargs
)
617 self
.assertEqual(got
, expected
)
619 testPL(5, 2, (), [0, 1, 2, 3, 4])
620 testPL(5, 2, (1,), [1, 2, 3, 4])
621 testPL(5, 2, (2,), [2, 3, 4])
622 testPL(5, 2, (4,), [4])
623 testPL(5, 2, (0, 3), [0, 1, 2])
624 testPL(5, 2, (1, 4), [1, 2, 3])
625 testPL(5, 2, (2, 99), [2, 3, 4])
626 testPL(5, 2, (20, 99), [])
628 def test_read_batch_urls(self
):
629 f
= io
.StringIO('''\xef\xbb\xbf foo
632 # More after this line\r
635 self
.assertEqual(read_batch_urls(f
), ['foo', 'bar', 'baz', 'bam'])
637 def test_urlencode_postdata(self
):
638 data
= urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'}
)
639 self
.assertTrue(isinstance(data
, bytes))
641 def test_update_url_query(self
):
643 return compat_parse_qs(compat_urlparse
.urlparse(url
).query
)
644 self
.assertEqual(query_dict(update_url_query(
645 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']}
)),
646 query_dict('http://example.com/path?quality=HD&format=mp4'))
647 self
.assertEqual(query_dict(update_url_query(
648 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']}
)),
649 query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
650 self
.assertEqual(query_dict(update_url_query(
651 'http://example.com/path', {'fields': 'id,formats,subtitles'}
)),
652 query_dict('http://example.com/path?fields=id,formats,subtitles'))
653 self
.assertEqual(query_dict(update_url_query(
654 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')}
)),
655 query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
656 self
.assertEqual(query_dict(update_url_query(
657 'http://example.com/path?manifest=f4m', {'manifest': []}
)),
658 query_dict('http://example.com/path'))
659 self
.assertEqual(query_dict(update_url_query(
660 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'}
)),
661 query_dict('http://example.com/path?system=LINUX'))
662 self
.assertEqual(query_dict(update_url_query(
663 'http://example.com/path', {'fields': b'id,formats,subtitles'}
)),
664 query_dict('http://example.com/path?fields=id,formats,subtitles'))
665 self
.assertEqual(query_dict(update_url_query(
666 'http://example.com/path', {'width': 1080, 'height': 720}
)),
667 query_dict('http://example.com/path?width=1080&height=720'))
668 self
.assertEqual(query_dict(update_url_query(
669 'http://example.com/path', {'bitrate': 5020.43}
)),
670 query_dict('http://example.com/path?bitrate=5020.43'))
671 self
.assertEqual(query_dict(update_url_query(
672 'http://example.com/path', {'test': '第二行тест'}
)),
673 query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
675 def test_multipart_encode(self
):
677 multipart_encode({b'field': b'value'}
, boundary
='AAAAAA')[0],
678 b
'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
680 multipart_encode({'欄位'.encode('utf-8'): '值'.encode('utf-8')}
, boundary
='AAAAAA')[0],
681 b
'--AAAAAA\r\nContent-Disposition: form-data; name="\xe6\xac\x84\xe4\xbd\x8d"\r\n\r\n\xe5\x80\xbc\r\n--AAAAAA--\r\n')
683 ValueError, multipart_encode
, {b'field': b'value'}
, boundary
='value')
685 def test_dict_get(self
):
693 d
= FALSE_VALUES
.copy()
695 self
.assertEqual(dict_get(d
, 'a'), 42)
696 self
.assertEqual(dict_get(d
, 'b'), None)
697 self
.assertEqual(dict_get(d
, 'b', 42), 42)
698 self
.assertEqual(dict_get(d
, ('a', )), 42)
699 self
.assertEqual(dict_get(d
, ('b', 'a', )), 42)
700 self
.assertEqual(dict_get(d
, ('b', 'c', 'a', 'd', )), 42)
701 self
.assertEqual(dict_get(d
, ('b', 'c', )), None)
702 self
.assertEqual(dict_get(d
, ('b', 'c', ), 42), 42)
703 for key
, false_value
in FALSE_VALUES
.items():
704 self
.assertEqual(dict_get(d
, ('b', 'c', key
, )), None)
705 self
.assertEqual(dict_get(d
, ('b', 'c', key
, ), skip_false_values
=False), false_value
)
707 def test_merge_dicts(self
):
708 self
.assertEqual(merge_dicts({'a': 1}
, {'b': 2}
), {'a': 1, 'b': 2}
)
709 self
.assertEqual(merge_dicts({'a': 1}
, {'a': 2}
), {'a': 1}
)
710 self
.assertEqual(merge_dicts({'a': 1}
, {'a': None}
), {'a': 1}
)
711 self
.assertEqual(merge_dicts({'a': 1}
, {'a': ''}
), {'a': 1}
)
712 self
.assertEqual(merge_dicts({'a': 1}
, {}), {'a': 1}
)
713 self
.assertEqual(merge_dicts({'a': None}
, {'a': 1}
), {'a': 1}
)
714 self
.assertEqual(merge_dicts({'a': ''}
, {'a': 1}
), {'a': ''}
)
715 self
.assertEqual(merge_dicts({'a': ''}
, {'a': 'abc'}
), {'a': 'abc'}
)
716 self
.assertEqual(merge_dicts({'a': None}
, {'a': ''}
, {'a': 'abc'}
), {'a': 'abc'}
)
718 def test_encode_compat_str(self
):
719 self
.assertEqual(encode_compat_str(b
'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
720 self
.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
722 def test_parse_iso8601(self
):
723 self
.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
724 self
.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
725 self
.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
726 self
.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
727 self
.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
728 self
.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
730 def test_strip_jsonp(self
):
731 stripped
= strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
732 d
= json
.loads(stripped
)
733 self
.assertEqual(d
, [{"id": "532cb", "x": 3}
])
735 stripped
= strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
736 d
= json
.loads(stripped
)
737 self
.assertEqual(d
, {'STATUS': 'OK'}
)
739 stripped
= strip_jsonp('ps.embedHandler({"status": "success"});')
740 d
= json
.loads(stripped
)
741 self
.assertEqual(d
, {'status': 'success'}
)
743 stripped
= strip_jsonp('window.cb && window.cb({"status": "success"});')
744 d
= json
.loads(stripped
)
745 self
.assertEqual(d
, {'status': 'success'}
)
747 stripped
= strip_jsonp('window.cb && cb({"status": "success"});')
748 d
= json
.loads(stripped
)
749 self
.assertEqual(d
, {'status': 'success'}
)
751 stripped
= strip_jsonp('({"status": "success"});')
752 d
= json
.loads(stripped
)
753 self
.assertEqual(d
, {'status': 'success'}
)
755 def test_uppercase_escape(self
):
756 self
.assertEqual(uppercase_escape('aä'), 'aä')
757 self
.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
759 def test_lowercase_escape(self
):
760 self
.assertEqual(lowercase_escape('aä'), 'aä')
761 self
.assertEqual(lowercase_escape('\\u0026'), '&')
763 def test_limit_length(self
):
764 self
.assertEqual(limit_length(None, 12), None)
765 self
.assertEqual(limit_length('foo', 12), 'foo')
767 limit_length('foo bar baz asd', 12).startswith('foo bar'))
768 self
.assertTrue('...' in limit_length('foo bar baz asd', 12))
770 def test_mimetype2ext(self
):
771 self
.assertEqual(mimetype2ext(None), None)
772 self
.assertEqual(mimetype2ext('video/x-flv'), 'flv')
773 self
.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
774 self
.assertEqual(mimetype2ext('text/vtt'), 'vtt')
775 self
.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
776 self
.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
778 def test_month_by_name(self
):
779 self
.assertEqual(month_by_name(None), None)
780 self
.assertEqual(month_by_name('December', 'en'), 12)
781 self
.assertEqual(month_by_name('décembre', 'fr'), 12)
782 self
.assertEqual(month_by_name('December'), 12)
783 self
.assertEqual(month_by_name('décembre'), None)
784 self
.assertEqual(month_by_name('Unknown', 'unknown'), None)
786 def test_parse_codecs(self
):
787 self
.assertEqual(parse_codecs(''), {})
788 self
.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
789 'vcodec': 'avc1.77.30',
790 'acodec': 'mp4a.40.2',
792 self
.assertEqual(parse_codecs('mp4a.40.2'), {
794 'acodec': 'mp4a.40.2',
796 self
.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
797 'vcodec': 'avc1.42001e',
798 'acodec': 'mp4a.40.5',
800 self
.assertEqual(parse_codecs('avc3.640028'), {
801 'vcodec': 'avc3.640028',
804 self
.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
808 self
.assertEqual(parse_codecs('av01.0.05M.08'), {
809 'vcodec': 'av01.0.05M.08',
813 def test_escape_rfc3986(self
):
814 reserved
= "!*'();:@&=+$,/?#[]"
815 unreserved
= 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
816 self
.assertEqual(escape_rfc3986(reserved
), reserved
)
817 self
.assertEqual(escape_rfc3986(unreserved
), unreserved
)
818 self
.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
819 self
.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
820 self
.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
821 self
.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
823 def test_escape_url(self
):
825 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
826 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
829 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
830 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
833 escape_url('http://тест.рф/фрагмент'),
834 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
837 escape_url('http://тест.рф/абв?абв=абв#абв'),
838 'http://xn--e1aybc.xn--p1ai/%D0%B0%D0%B1%D0%B2?%D0%B0%D0%B1%D0%B2=%D0%B0%D0%B1%D0%B2#%D0%B0%D0%B1%D0%B2'
840 self
.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
842 def test_js_to_json_realworld(self
):
844 'clip':{'provider':'pseudo'}
846 self
.assertEqual(js_to_json(inp
), '''{
847 "clip":{"provider":"pseudo"}
849 json
.loads(js_to_json(inp
))
852 'playlist':[{'controls':{'all':null}}]
854 self
.assertEqual(js_to_json(inp
), '''{
855 "playlist":[{"controls":{"all":null}}]
858 inp
= '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
859 self
.assertEqual(js_to_json(inp
), '''"The CW's 'Crazy Ex-Girlfriend'"''')
861 inp
= '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
862 json_code
= js_to_json(inp
)
863 self
.assertEqual(json
.loads(json_code
), json
.loads(inp
))
866 0:{src:'skipped', type: 'application/dash+xml'},
867 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
869 self
.assertEqual(js_to_json(inp
), '''{
870 "0":{"src":"skipped", "type": "application/dash+xml"},
871 "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
874 inp
= '''{"foo":101}'''
875 self
.assertEqual(js_to_json(inp
), '''{"foo":101}''')
877 inp
= '''{"duration": "00:01:07"}'''
878 self
.assertEqual(js_to_json(inp
), '''{"duration": "00:01:07"}''')
880 inp
= '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}'''
881 self
.assertEqual(js_to_json(inp
), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''')
883 def test_js_to_json_edgecases(self
):
884 on
= js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
885 self
.assertEqual(json
.loads(on
), {"abc_def": "1'\\2\\'3\"4"}
)
887 on
= js_to_json('{"abc": true}')
888 self
.assertEqual(json
.loads(on
), {'abc': True}
)
890 # Ignore JavaScript code as well
897 self
.assertEqual(d
['x'], 1)
898 self
.assertEqual(d
['y'], 'a')
900 on
= js_to_json('["abc", "def",]')
901 self
.assertEqual(json
.loads(on
), ['abc', 'def'])
903 on
= js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
904 self
.assertEqual(json
.loads(on
), ['abc', 'def'])
906 on
= js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
907 self
.assertEqual(json
.loads(on
), ['abc', 'def'])
909 on
= js_to_json('{"abc": "def",}')
910 self
.assertEqual(json
.loads(on
), {'abc': 'def'}
)
912 on
= js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
913 self
.assertEqual(json
.loads(on
), {'abc': 'def'}
)
915 on
= js_to_json('{ 0: /* " \n */ ",]" , }')
916 self
.assertEqual(json
.loads(on
), {'0': ',]'}
)
918 on
= js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
919 self
.assertEqual(json
.loads(on
), {'0': ',]'}
)
921 on
= js_to_json('{ 0: // comment\n1 }')
922 self
.assertEqual(json
.loads(on
), {'0': 1}
)
924 on
= js_to_json(r
'["<p>x<\/p>"]')
925 self
.assertEqual(json
.loads(on
), ['<p>x</p>'])
927 on
= js_to_json(r
'["\xaa"]')
928 self
.assertEqual(json
.loads(on
), ['\u00aa'])
930 on
= js_to_json("['a\\\nb']")
931 self
.assertEqual(json
.loads(on
), ['ab'])
933 on
= js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
934 self
.assertEqual(json
.loads(on
), ['ab'])
936 on
= js_to_json('{0xff:0xff}')
937 self
.assertEqual(json
.loads(on
), {'255': 255}
)
939 on
= js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
940 self
.assertEqual(json
.loads(on
), {'255': 255}
)
942 on
= js_to_json('{077:077}')
943 self
.assertEqual(json
.loads(on
), {'63': 63}
)
945 on
= js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
946 self
.assertEqual(json
.loads(on
), {'63': 63}
)
948 on
= js_to_json('{42:42}')
949 self
.assertEqual(json
.loads(on
), {'42': 42}
)
951 on
= js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
952 self
.assertEqual(json
.loads(on
), {'42': 42}
)
954 on
= js_to_json('{42:4.2e1}')
955 self
.assertEqual(json
.loads(on
), {'42': 42.0}
)
957 def test_js_to_json_malformed(self
):
958 self
.assertEqual(js_to_json('42a1'), '42"a1"')
959 self
.assertEqual(js_to_json('42a-1'), '42"a"-1')
961 def test_extract_attributes(self
):
962 self
.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}
)
963 self
.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'}
)
964 self
.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'}
)
965 self
.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"}
)
966 self
.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'}
)
967 self
.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}
)
968 self
.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}
)
969 self
.assertEqual(extract_attributes('<e x="&">'), {'x': '&'}
) # XML
970 self
.assertEqual(extract_attributes('<e x=""">'), {'x': '"'}
)
971 self
.assertEqual(extract_attributes('<e x="£">'), {'x': '£'}
) # HTML 3.2
972 self
.assertEqual(extract_attributes('<e x="λ">'), {'x': 'λ'}
) # HTML 4.0
973 self
.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'}
)
974 self
.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"}
)
975 self
.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'}
)
976 self
.assertEqual(extract_attributes('<e x >'), {'x': None}
)
977 self
.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None}
)
978 self
.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'}
)
979 self
.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'}
)
980 self
.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'}
)
981 self
.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'}
)
982 self
.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'}
)
983 self
.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'}
)
984 self
.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}
) # Names lowercased
985 self
.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'}
)
986 self
.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'}
)
987 self
.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'}
)
988 self
.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'}
)
989 self
.assertEqual(extract_attributes('<e x="décomposé">'), {'x': 'décompose\u0301'}
)
990 # "Narrow" Python builds don't support unicode code points outside BMP.
993 supports_outside_bmp
= True
995 supports_outside_bmp
= False
996 if supports_outside_bmp
:
997 self
.assertEqual(extract_attributes('<e x="Smile 😀!">'), {'x': 'Smile \U0001f600!'}
)
998 # Malformed HTML should not break attributes extraction on older Python
999 self
.assertEqual(extract_attributes('<mal"formed/>'), {})
1001 def test_clean_html(self
):
1002 self
.assertEqual(clean_html('a:\nb'), 'a: b')
1003 self
.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
1004 self
.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
1006 def test_intlist_to_bytes(self
):
1008 intlist_to_bytes([0, 1, 127, 128, 255]),
1009 b
'\x00\x01\x7f\x80\xff')
1011 def test_args_to_str(self
):
1013 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
1014 'foo ba/r -baz \'2 be\' \'\'' if compat_os_name
!= 'nt' else 'foo ba/r -baz "2 be" ""'
1017 def test_parse_filesize(self
):
1018 self
.assertEqual(parse_filesize(None), None)
1019 self
.assertEqual(parse_filesize(''), None)
1020 self
.assertEqual(parse_filesize('91 B'), 91)
1021 self
.assertEqual(parse_filesize('foobar'), None)
1022 self
.assertEqual(parse_filesize('2 MiB'), 2097152)
1023 self
.assertEqual(parse_filesize('5 GB'), 5000000000)
1024 self
.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
1025 self
.assertEqual(parse_filesize('1.2tb'), 1200000000000)
1026 self
.assertEqual(parse_filesize('1,24 KB'), 1240)
1027 self
.assertEqual(parse_filesize('1,24 kb'), 1240)
1028 self
.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
1030 def test_parse_count(self
):
1031 self
.assertEqual(parse_count(None), None)
1032 self
.assertEqual(parse_count(''), None)
1033 self
.assertEqual(parse_count('0'), 0)
1034 self
.assertEqual(parse_count('1000'), 1000)
1035 self
.assertEqual(parse_count('1.000'), 1000)
1036 self
.assertEqual(parse_count('1.1k'), 1100)
1037 self
.assertEqual(parse_count('1.1kk'), 1100000)
1038 self
.assertEqual(parse_count('1.1kk '), 1100000)
1039 self
.assertEqual(parse_count('1.1kk views'), 1100000)
1041 def test_parse_resolution(self
):
1042 self
.assertEqual(parse_resolution(None), {})
1043 self
.assertEqual(parse_resolution(''), {})
1044 self
.assertEqual(parse_resolution('1920x1080'), {'width': 1920, 'height': 1080}
)
1045 self
.assertEqual(parse_resolution('1920×1080'), {'width': 1920, 'height': 1080}
)
1046 self
.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080}
)
1047 self
.assertEqual(parse_resolution('720p'), {'height': 720}
)
1048 self
.assertEqual(parse_resolution('4k'), {'height': 2160}
)
1049 self
.assertEqual(parse_resolution('8K'), {'height': 4320}
)
1051 def test_parse_bitrate(self
):
1052 self
.assertEqual(parse_bitrate(None), None)
1053 self
.assertEqual(parse_bitrate(''), None)
1054 self
.assertEqual(parse_bitrate('300kbps'), 300)
1055 self
.assertEqual(parse_bitrate('1500kbps'), 1500)
1056 self
.assertEqual(parse_bitrate('300 kbps'), 300)
1058 def test_version_tuple(self
):
1059 self
.assertEqual(version_tuple('1'), (1,))
1060 self
.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
1061 self
.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
1063 def test_detect_exe_version(self
):
1064 self
.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
1065 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
1066 configuration: --prefix=/usr --extra-'''), '1.2.1')
1067 self
.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
1068 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
1069 self
.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
1070 Trying to open render node...
1071 Success at /dev/dri/renderD128.
1072 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
1074 def test_age_restricted(self
):
1075 self
.assertFalse(age_restricted(None, 10)) # unrestricted content
1076 self
.assertFalse(age_restricted(1, None)) # unrestricted policy
1077 self
.assertFalse(age_restricted(8, 10))
1078 self
.assertTrue(age_restricted(18, 14))
1079 self
.assertFalse(age_restricted(18, 18))
1081 def test_is_html(self
):
1082 self
.assertFalse(is_html(b
'\x49\x44\x43<html'))
1083 self
.assertTrue(is_html(b
'<!DOCTYPE foo>\xaaa'))
1084 self
.assertTrue(is_html( # UTF-8 with BOM
1085 b
'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
1086 self
.assertTrue(is_html( # UTF-16-LE
1087 b
'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
1089 self
.assertTrue(is_html( # UTF-16-BE
1090 b
'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
1092 self
.assertTrue(is_html( # UTF-32-BE
1093 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'))
1094 self
.assertTrue(is_html( # UTF-32-LE
1095 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'))
1097 def test_render_table(self
):
1101 [[123, 4], [9999, 51]]),
1106 def test_match_str(self
):
1107 self
.assertRaises(ValueError, match_str
, 'xy>foobar', {})
1108 self
.assertFalse(match_str('xy', {'x': 1200}
))
1109 self
.assertTrue(match_str('!xy', {'x': 1200}
))
1110 self
.assertTrue(match_str('x', {'x': 1200}
))
1111 self
.assertFalse(match_str('!x', {'x': 1200}
))
1112 self
.assertTrue(match_str('x', {'x': 0}
))
1113 self
.assertFalse(match_str('x>0', {'x': 0}
))
1114 self
.assertFalse(match_str('x>0', {}))
1115 self
.assertTrue(match_str('x>?0', {}))
1116 self
.assertTrue(match_str('x>1K', {'x': 1200}
))
1117 self
.assertFalse(match_str('x>2K', {'x': 1200}
))
1118 self
.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}
))
1119 self
.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}
))
1120 self
.assertFalse(match_str('y=a212', {'y': 'foobar42'}
))
1121 self
.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}
))
1122 self
.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}
))
1123 self
.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}
))
1124 self
.assertFalse(match_str(
1125 'like_count > 100 & dislike_count <? 50 & description',
1126 {'like_count': 90, 'description': 'foo'}
))
1127 self
.assertTrue(match_str(
1128 'like_count > 100 & dislike_count <? 50 & description',
1129 {'like_count': 190, 'description': 'foo'}
))
1130 self
.assertFalse(match_str(
1131 'like_count > 100 & dislike_count <? 50 & description',
1132 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}
))
1133 self
.assertFalse(match_str(
1134 'like_count > 100 & dislike_count <? 50 & description',
1135 {'like_count': 190, 'dislike_count': 10}
))
1136 self
.assertTrue(match_str('is_live', {'is_live': True}
))
1137 self
.assertFalse(match_str('is_live', {'is_live': False}
))
1138 self
.assertFalse(match_str('is_live', {'is_live': None}
))
1139 self
.assertFalse(match_str('is_live', {}))
1140 self
.assertFalse(match_str('!is_live', {'is_live': True}
))
1141 self
.assertTrue(match_str('!is_live', {'is_live': False}
))
1142 self
.assertTrue(match_str('!is_live', {'is_live': None}
))
1143 self
.assertTrue(match_str('!is_live', {}))
1144 self
.assertTrue(match_str('title', {'title': 'abc'}
))
1145 self
.assertTrue(match_str('title', {'title': ''}
))
1146 self
.assertFalse(match_str('!title', {'title': 'abc'}
))
1147 self
.assertFalse(match_str('!title', {'title': ''}
))
1149 def test_parse_dfxp_time_expr(self
):
1150 self
.assertEqual(parse_dfxp_time_expr(None), None)
1151 self
.assertEqual(parse_dfxp_time_expr(''), None)
1152 self
.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1153 self
.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1154 self
.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1155 self
.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1156 self
.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1158 def test_dfxp2srt(self
):
1159 dfxp_data
= '''<?xml version="1.0" encoding="UTF-8"?>
1160 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1163 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1164 <p begin="1" end="2">第二行<br/>♪♪</p>
1165 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1166 <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1167 <p begin="-1" end="-1">Ignore, two</p>
1168 <p begin="3" dur="-1">Ignored, three</p>
1171 </tt>'''.encode('utf-8')
1173 00:00:00,000 --> 00:00:01,000
1174 The following line contains Chinese characters and special symbols
1177 00:00:01,000 --> 00:00:02,000
1182 00:00:02,000 --> 00:00:03,000
1187 self
.assertEqual(dfxp2srt(dfxp_data
), srt_data
)
1189 dfxp_data_no_default_namespace
= '''<?xml version="1.0" encoding="UTF-8"?>
1190 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1193 <p begin="0" end="1">The first line</p>
1196 </tt>'''.encode('utf-8')
1198 00:00:00,000 --> 00:00:01,000
1202 self
.assertEqual(dfxp2srt(dfxp_data_no_default_namespace
), srt_data
)
1204 dfxp_data_with_style
= '''<?xml version="1.0" encoding="utf-8"?>
1205 <tt xmlns="http://www.w3.org/2006/10/ttaf1" xmlns:ttp="http://www.w3.org/2006/10/ttaf1#parameter" ttp:timeBase="media" xmlns:tts="http://www.w3.org/2006/10/ttaf1#style" xml:lang="en" xmlns:ttm="http://www.w3.org/2006/10/ttaf1#metadata">
1208 <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1209 <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1210 <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1211 <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1214 <body tts:textAlign="center" style="s0">
1216 <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1217 <p style="s2" begin="00:00:02.08" id="p0" end="00:00:05.84"><span tts:color="lime">part 1<br /></span><span tts:color="cyan">part 2</span></p>
1218 <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1219 <p style="s1" tts:textDecoration="underline" begin="00:00:09.56" id="p2" end="00:00:12.36"><span style="s2" tts:color="lime">inner<br /> </span>style</p>
1222 </tt>'''.encode('utf-8')
1224 00:00:02,080 --> 00:00:05,839
1225 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1228 00:00:02,080 --> 00:00:05,839
1229 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1230 </font>part 2</font></b>
1233 00:00:05,839 --> 00:00:09,560
1234 <u><font color="lime">line 3
1238 00:00:09,560 --> 00:00:12,359
1239 <i><u><font color="yellow"><font color="lime">inner
1240 </font>style</font></u></i>
1243 self
.assertEqual(dfxp2srt(dfxp_data_with_style
), srt_data
)
1245 dfxp_data_non_utf8
= '''<?xml version="1.0" encoding="UTF-16"?>
1246 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1249 <p begin="0" end="1">Line 1</p>
1250 <p begin="1" end="2">第二行</p>
1253 </tt>'''.encode('utf-16')
1255 00:00:00,000 --> 00:00:01,000
1259 00:00:01,000 --> 00:00:02,000
1263 self
.assertEqual(dfxp2srt(dfxp_data_non_utf8
), srt_data
)
1265 def test_cli_option(self
):
1266 self
.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}
, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1267 self
.assertEqual(cli_option({'proxy': None}
, '--proxy', 'proxy'), [])
1268 self
.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1269 self
.assertEqual(cli_option({'retries': 10}
, '--retries', 'retries'), ['--retries', '10'])
1271 def test_cli_valueless_option(self
):
1272 self
.assertEqual(cli_valueless_option(
1273 {'downloader': 'external'}
, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1274 self
.assertEqual(cli_valueless_option(
1275 {'downloader': 'internal'}
, '--external-downloader', 'downloader', 'external'), [])
1276 self
.assertEqual(cli_valueless_option(
1277 {'nocheckcertificate': True}
, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1278 self
.assertEqual(cli_valueless_option(
1279 {'nocheckcertificate': False}
, '--no-check-certificate', 'nocheckcertificate'), [])
1280 self
.assertEqual(cli_valueless_option(
1281 {'checkcertificate': True}
, '--no-check-certificate', 'checkcertificate', False), [])
1282 self
.assertEqual(cli_valueless_option(
1283 {'checkcertificate': False}
, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1285 def test_cli_bool_option(self
):
1288 {'nocheckcertificate': True}
, '--no-check-certificate', 'nocheckcertificate'),
1289 ['--no-check-certificate', 'true'])
1292 {'nocheckcertificate': True}
, '--no-check-certificate', 'nocheckcertificate', separator
='='),
1293 ['--no-check-certificate=true'])
1296 {'nocheckcertificate': True}
, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1297 ['--check-certificate', 'false'])
1300 {'nocheckcertificate': True}
, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1301 ['--check-certificate=false'])
1304 {'nocheckcertificate': False}
, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1305 ['--check-certificate', 'true'])
1308 {'nocheckcertificate': False}
, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1309 ['--check-certificate=true'])
1312 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1315 def test_ohdave_rsa_encrypt(self
):
1316 N
= 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1320 ohdave_rsa_encrypt(b
'aa111222', e
, N
),
1321 '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1323 def test_pkcs1pad(self
):
1325 padded_data
= pkcs1pad(data
, 32)
1326 self
.assertEqual(padded_data
[:2], [0, 2])
1327 self
.assertEqual(padded_data
[28:], [0, 1, 2, 3])
1329 self
.assertRaises(ValueError, pkcs1pad
, data
, 8)
1331 def test_encode_base_n(self
):
1332 self
.assertEqual(encode_base_n(0, 30), '0')
1333 self
.assertEqual(encode_base_n(80, 30), '2k')
1335 custom_table
= '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1336 self
.assertEqual(encode_base_n(0, 30, custom_table
), '9')
1337 self
.assertEqual(encode_base_n(80, 30, custom_table
), '7P')
1339 self
.assertRaises(ValueError, encode_base_n
, 0, 70)
1340 self
.assertRaises(ValueError, encode_base_n
, 0, 60, custom_table
)
1342 def test_urshift(self
):
1343 self
.assertEqual(urshift(3, 1), 1)
1344 self
.assertEqual(urshift(-3, 1), 2147483646)
1346 def test_get_element_by_class(self
):
1348 <span class="foo bar">nice</span>
1351 self
.assertEqual(get_element_by_class('foo', html
), 'nice')
1352 self
.assertEqual(get_element_by_class('no-such-class', html
), None)
1354 def test_get_element_by_attribute(self
):
1356 <span class="foo bar">nice</span>
1359 self
.assertEqual(get_element_by_attribute('class', 'foo bar', html
), 'nice')
1360 self
.assertEqual(get_element_by_attribute('class', 'foo', html
), None)
1361 self
.assertEqual(get_element_by_attribute('class', 'no-such-foo', html
), None)
1364 <div itemprop="author" itemscope>foo</div>
1367 self
.assertEqual(get_element_by_attribute('itemprop', 'author', html
), 'foo')
1369 def test_get_elements_by_class(self
):
1371 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1374 self
.assertEqual(get_elements_by_class('foo', html
), ['nice', 'also nice'])
1375 self
.assertEqual(get_elements_by_class('no-such-class', html
), [])
1377 def test_get_elements_by_attribute(self
):
1379 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1382 self
.assertEqual(get_elements_by_attribute('class', 'foo bar', html
), ['nice', 'also nice'])
1383 self
.assertEqual(get_elements_by_attribute('class', 'foo', html
), [])
1384 self
.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html
), [])
1387 if __name__
== '__main__':