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
17 import xml
.etree
.ElementTree
19 from yt_dlp
.utils
import (
44 get_element_by_attribute
,
45 get_elements_by_class
,
46 get_elements_by_attribute
,
47 get_element_html_by_class
,
48 get_element_html_by_attribute
,
49 get_elements_html_by_class
,
50 get_elements_html_by_attribute
,
51 get_elements_text_and_html_by_attribute
,
52 get_element_text_and_html_by_tag
,
114 parse_dfxp_time_expr
,
117 cli_valueless_option
,
123 from yt_dlp
.compat
import (
125 compat_etree_fromstring
,
127 compat_HTMLParseError
,
133 class TestUtil(unittest
.TestCase
):
134 def test_timeconvert(self
):
135 self
.assertTrue(timeconvert('') is None)
136 self
.assertTrue(timeconvert('bougrg') is None)
138 def test_sanitize_filename(self
):
139 self
.assertEqual(sanitize_filename(''), '')
140 self
.assertEqual(sanitize_filename('abc'), 'abc')
141 self
.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
143 self
.assertEqual(sanitize_filename('123'), '123')
145 self
.assertEqual('abc_de', sanitize_filename('abc/de'))
146 self
.assertFalse('/' in sanitize_filename('abc/de///'))
148 self
.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
149 self
.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
150 self
.assertEqual('yes no', sanitize_filename('yes? no'))
151 self
.assertEqual('this - that', sanitize_filename('this: that'))
153 self
.assertEqual(sanitize_filename('AT&T'), 'AT&T')
155 self
.assertEqual(sanitize_filename(aumlaut
), aumlaut
)
156 tests
= '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
157 self
.assertEqual(sanitize_filename(tests
), tests
)
160 sanitize_filename('New World record at 0:12:34'),
161 'New World record at 0_12_34')
163 self
.assertEqual(sanitize_filename('--gasdgf'), '--gasdgf')
164 self
.assertEqual(sanitize_filename('--gasdgf', is_id
=True), '--gasdgf')
165 self
.assertEqual(sanitize_filename('--gasdgf', is_id
=False), '_-gasdgf')
166 self
.assertEqual(sanitize_filename('.gasdgf'), '.gasdgf')
167 self
.assertEqual(sanitize_filename('.gasdgf', is_id
=True), '.gasdgf')
168 self
.assertEqual(sanitize_filename('.gasdgf', is_id
=False), 'gasdgf')
172 for fbc
in forbidden
:
173 self
.assertTrue(fbc
not in sanitize_filename(fc
))
175 def test_sanitize_filename_restricted(self
):
176 self
.assertEqual(sanitize_filename('abc', restricted
=True), 'abc')
177 self
.assertEqual(sanitize_filename('abc_d-e', restricted
=True), 'abc_d-e')
179 self
.assertEqual(sanitize_filename('123', restricted
=True), '123')
181 self
.assertEqual('abc_de', sanitize_filename('abc/de', restricted
=True))
182 self
.assertFalse('/' in sanitize_filename('abc/de///', restricted
=True))
184 self
.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted
=True))
185 self
.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted
=True))
186 self
.assertEqual('yes_no', sanitize_filename('yes? no', restricted
=True))
187 self
.assertEqual('this_-_that', sanitize_filename('this: that', restricted
=True))
189 tests
= 'aäb\u4e2d\u56fd\u7684c'
190 self
.assertEqual(sanitize_filename(tests
, restricted
=True), 'aab_c')
191 self
.assertTrue(sanitize_filename('\xf6', restricted
=True) != '') # No empty filename
193 forbidden
= '"\0\\/&!: \'\t\n()[]{}$;`^,#'
195 for fbc
in forbidden
:
196 self
.assertTrue(fbc
not in sanitize_filename(fc
, restricted
=True))
198 # Handle a common case more neatly
199 self
.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted
=True), 'Song')
200 self
.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted
=True), 'Speech')
201 # .. but make sure the file name is never empty
202 self
.assertTrue(sanitize_filename('-', restricted
=True) != '')
203 self
.assertTrue(sanitize_filename(':', restricted
=True) != '')
205 self
.assertEqual(sanitize_filename(
206 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted
=True),
207 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy')
209 def test_sanitize_ids(self
):
210 self
.assertEqual(sanitize_filename('_n_cd26wFpw', is_id
=True), '_n_cd26wFpw')
211 self
.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id
=True), '_BD_eEpuzXw')
212 self
.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id
=True), 'N0Y__7-UOdI')
214 def test_sanitize_path(self
):
215 if sys
.platform
!= 'win32':
218 self
.assertEqual(sanitize_path('abc'), 'abc')
219 self
.assertEqual(sanitize_path('abc/def'), 'abc\\def')
220 self
.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
221 self
.assertEqual(sanitize_path('abc|def'), 'abc#def')
222 self
.assertEqual(sanitize_path('<>:"|?*'), '#######')
223 self
.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
224 self
.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
226 self
.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
227 self
.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
229 self
.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
230 self
.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
231 self
.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
232 self
.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
235 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
236 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
239 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
240 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
241 self
.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
242 self
.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
243 self
.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
245 self
.assertEqual(sanitize_path('../abc'), '..\\abc')
246 self
.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
247 self
.assertEqual(sanitize_path('./abc'), 'abc')
248 self
.assertEqual(sanitize_path('./../abc'), '..\\abc')
250 def test_sanitize_url(self
):
251 self
.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
252 self
.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
253 self
.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
254 self
.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')
255 self
.assertEqual(sanitize_url('foo bar'), 'foo bar')
257 def test_extract_basic_auth(self
):
258 auth_header
= lambda url
: sanitized_Request(url
).get_header('Authorization')
259 self
.assertFalse(auth_header('http://foo.bar'))
260 self
.assertFalse(auth_header('http://:foo.bar'))
261 self
.assertEqual(auth_header('http://@foo.bar'), 'Basic Og==')
262 self
.assertEqual(auth_header('http://:pass@foo.bar'), 'Basic OnBhc3M=')
263 self
.assertEqual(auth_header('http://user:@foo.bar'), 'Basic dXNlcjo=')
264 self
.assertEqual(auth_header('http://user:pass@foo.bar'), 'Basic dXNlcjpwYXNz')
266 def test_expand_path(self
):
268 return '%{0}%'.format(var
) if sys
.platform
== 'win32' else '${0}'.format(var
)
270 compat_setenv('yt_dlp_EXPATH_PATH', 'expanded')
271 self
.assertEqual(expand_path(env('yt_dlp_EXPATH_PATH')), 'expanded')
272 self
.assertEqual(expand_path(env('HOME')), compat_getenv('HOME'))
273 self
.assertEqual(expand_path('~'), compat_getenv('HOME'))
275 expand_path('~/%s' % env('yt_dlp_EXPATH_PATH')),
276 '%s/expanded' % compat_getenv('HOME'))
278 def test_prepend_extension(self
):
279 self
.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
280 self
.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
281 self
.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
282 self
.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
283 self
.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
284 self
.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
286 def test_replace_extension(self
):
287 self
.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
288 self
.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
289 self
.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
290 self
.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
291 self
.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
292 self
.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
294 def test_subtitles_filename(self
):
295 self
.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt'), 'abc.en.vtt')
296 self
.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt', 'ext'), 'abc.en.vtt')
297 self
.assertEqual(subtitles_filename('abc.unexpected_ext', 'en', 'vtt', 'ext'), 'abc.unexpected_ext.en.vtt')
299 def test_remove_start(self
):
300 self
.assertEqual(remove_start(None, 'A - '), None)
301 self
.assertEqual(remove_start('A - B', 'A - '), 'B')
302 self
.assertEqual(remove_start('B - A', 'A - '), 'B - A')
304 def test_remove_end(self
):
305 self
.assertEqual(remove_end(None, ' - B'), None)
306 self
.assertEqual(remove_end('A - B', ' - B'), 'A')
307 self
.assertEqual(remove_end('B - A', ' - B'), 'B - A')
309 def test_remove_quotes(self
):
310 self
.assertEqual(remove_quotes(None), None)
311 self
.assertEqual(remove_quotes('"'), '"')
312 self
.assertEqual(remove_quotes("'"), "'")
313 self
.assertEqual(remove_quotes(';'), ';')
314 self
.assertEqual(remove_quotes('";'), '";')
315 self
.assertEqual(remove_quotes('""'), '')
316 self
.assertEqual(remove_quotes('";"'), ';')
318 def test_ordered_set(self
):
319 self
.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
320 self
.assertEqual(orderedSet([]), [])
321 self
.assertEqual(orderedSet([1]), [1])
322 # keep the list ordered
323 self
.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
325 def test_unescape_html(self
):
326 self
.assertEqual(unescapeHTML('%20;'), '%20;')
327 self
.assertEqual(unescapeHTML('/'), '/')
328 self
.assertEqual(unescapeHTML('/'), '/')
329 self
.assertEqual(unescapeHTML('é'), 'é')
330 self
.assertEqual(unescapeHTML('�'), '�')
331 self
.assertEqual(unescapeHTML('&a"'), '&a"')
333 self
.assertEqual(unescapeHTML('.''), '.\'')
335 def test_date_from_str(self
):
336 self
.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
337 self
.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
338 self
.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
339 self
.assertEqual(date_from_str('20200229+365day'), date_from_str('20200229+1year'))
340 self
.assertEqual(date_from_str('20210131+28day'), date_from_str('20210131+1month'))
342 def test_datetime_from_str(self
):
343 self
.assertEqual(datetime_from_str('yesterday', precision
='day'), datetime_from_str('now-1day', precision
='auto'))
344 self
.assertEqual(datetime_from_str('now+7day', precision
='day'), datetime_from_str('now+1week', precision
='auto'))
345 self
.assertEqual(datetime_from_str('now+14day', precision
='day'), datetime_from_str('now+2week', precision
='auto'))
346 self
.assertEqual(datetime_from_str('20200229+365day', precision
='day'), datetime_from_str('20200229+1year', precision
='auto'))
347 self
.assertEqual(datetime_from_str('20210131+28day', precision
='day'), datetime_from_str('20210131+1month', precision
='auto'))
348 self
.assertEqual(datetime_from_str('20210131+59day', precision
='day'), datetime_from_str('20210131+2month', precision
='auto'))
349 self
.assertEqual(datetime_from_str('now+1day', precision
='hour'), datetime_from_str('now+24hours', precision
='auto'))
350 self
.assertEqual(datetime_from_str('now+23hours', precision
='hour'), datetime_from_str('now+23hours', precision
='auto'))
352 def test_daterange(self
):
353 _20century
= DateRange("19000101", "20000101")
354 self
.assertFalse("17890714" in _20century
)
355 _ac
= DateRange("00010101")
356 self
.assertTrue("19690721" in _ac
)
357 _firstmilenium
= DateRange(end
="10000101")
358 self
.assertTrue("07110427" in _firstmilenium
)
360 def test_unified_dates(self
):
361 self
.assertEqual(unified_strdate('December 21, 2010'), '20101221')
362 self
.assertEqual(unified_strdate('8/7/2009'), '20090708')
363 self
.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
364 self
.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
365 self
.assertEqual(unified_strdate('1968 12 10'), '19681210')
366 self
.assertEqual(unified_strdate('1968-12-10'), '19681210')
367 self
.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
369 unified_strdate('11/26/2014 11:30:00 AM PST', day_first
=False),
372 unified_strdate('2/2/2015 6:47:40 PM', day_first
=False),
374 self
.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
375 self
.assertEqual(unified_strdate('25-09-2014'), '20140925')
376 self
.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
377 self
.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
378 self
.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
379 self
.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
380 self
.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
381 self
.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
382 self
.assertEqual(unified_strdate('November 3rd, 2019'), '20191103')
383 self
.assertEqual(unified_strdate('October 23rd, 2005'), '20051023')
385 def test_unified_timestamps(self
):
386 self
.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
387 self
.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
388 self
.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
389 self
.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
390 self
.assertEqual(unified_timestamp('1968 12 10'), -33436800)
391 self
.assertEqual(unified_timestamp('1968-12-10'), -33436800)
392 self
.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
394 unified_timestamp('11/26/2014 11:30:00 AM PST', day_first
=False),
397 unified_timestamp('2/2/2015 6:47:40 PM', day_first
=False),
399 self
.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
400 self
.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
401 self
.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
402 self
.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
403 self
.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
404 self
.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
405 self
.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
406 self
.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
407 self
.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
408 self
.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363)
410 def test_determine_ext(self
):
411 self
.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
412 self
.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
413 self
.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
414 self
.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
415 self
.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
416 self
.assertEqual(determine_ext('foobar', None), None)
418 def test_find_xpath_attr(self
):
426 doc
= compat_etree_fromstring(testxml
)
428 self
.assertEqual(find_xpath_attr(doc
, './/fourohfour', 'n'), None)
429 self
.assertEqual(find_xpath_attr(doc
, './/fourohfour', 'n', 'v'), None)
430 self
.assertEqual(find_xpath_attr(doc
, './/node', 'n'), None)
431 self
.assertEqual(find_xpath_attr(doc
, './/node', 'n', 'v'), None)
432 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x'), doc
[1])
433 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x', 'a'), doc
[1])
434 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x', 'b'), doc
[3])
435 self
.assertEqual(find_xpath_attr(doc
, './/node', 'y'), doc
[2])
436 self
.assertEqual(find_xpath_attr(doc
, './/node', 'y', 'c'), doc
[2])
437 self
.assertEqual(find_xpath_attr(doc
, './/node', 'y', 'd'), doc
[3])
438 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x', ''), doc
[4])
440 def test_xpath_with_ns(self
):
441 testxml
= '''<root xmlns:media="http://example.com/">
443 <media:author>The Author</media:author>
444 <url>http://server.com/download.mp3</url>
447 doc
= compat_etree_fromstring(testxml
)
448 find
= lambda p
: doc
.find(xpath_with_ns(p
, {'media': 'http://example.com/'}
))
449 self
.assertTrue(find('media:song') is not None)
450 self
.assertEqual(find('media:song/media:author').text
, 'The Author')
451 self
.assertEqual(find('media:song/url').text
, 'http://server.com/download.mp3')
453 def test_xpath_element(self
):
454 doc
= xml
.etree
.ElementTree
.Element('root')
455 div
= xml
.etree
.ElementTree
.SubElement(doc
, 'div')
456 p
= xml
.etree
.ElementTree
.SubElement(div
, 'p')
458 self
.assertEqual(xpath_element(doc
, 'div/p'), p
)
459 self
.assertEqual(xpath_element(doc
, ['div/p']), p
)
460 self
.assertEqual(xpath_element(doc
, ['div/bar', 'div/p']), p
)
461 self
.assertEqual(xpath_element(doc
, 'div/bar', default
='default'), 'default')
462 self
.assertEqual(xpath_element(doc
, ['div/bar'], default
='default'), 'default')
463 self
.assertTrue(xpath_element(doc
, 'div/bar') is None)
464 self
.assertTrue(xpath_element(doc
, ['div/bar']) is None)
465 self
.assertTrue(xpath_element(doc
, ['div/bar'], 'div/baz') is None)
466 self
.assertRaises(ExtractorError
, xpath_element
, doc
, 'div/bar', fatal
=True)
467 self
.assertRaises(ExtractorError
, xpath_element
, doc
, ['div/bar'], fatal
=True)
468 self
.assertRaises(ExtractorError
, xpath_element
, doc
, ['div/bar', 'div/baz'], fatal
=True)
470 def test_xpath_text(self
):
476 doc
= compat_etree_fromstring(testxml
)
477 self
.assertEqual(xpath_text(doc
, 'div/p'), 'Foo')
478 self
.assertEqual(xpath_text(doc
, 'div/bar', default
='default'), 'default')
479 self
.assertTrue(xpath_text(doc
, 'div/bar') is None)
480 self
.assertRaises(ExtractorError
, xpath_text
, doc
, 'div/bar', fatal
=True)
482 def test_xpath_attr(self
):
488 doc
= compat_etree_fromstring(testxml
)
489 self
.assertEqual(xpath_attr(doc
, 'div/p', 'x'), 'a')
490 self
.assertEqual(xpath_attr(doc
, 'div/bar', 'x'), None)
491 self
.assertEqual(xpath_attr(doc
, 'div/p', 'y'), None)
492 self
.assertEqual(xpath_attr(doc
, 'div/bar', 'x', default
='default'), 'default')
493 self
.assertEqual(xpath_attr(doc
, 'div/p', 'y', default
='default'), 'default')
494 self
.assertRaises(ExtractorError
, xpath_attr
, doc
, 'div/bar', 'x', fatal
=True)
495 self
.assertRaises(ExtractorError
, xpath_attr
, doc
, 'div/p', 'y', fatal
=True)
497 def test_smuggle_url(self
):
498 data
= {"ö": "ö", "abc": [3]}
499 url
= 'https://foo.bar/baz?x=y#a'
500 smug_url
= smuggle_url(url
, data
)
501 unsmug_url
, unsmug_data
= unsmuggle_url(smug_url
)
502 self
.assertEqual(url
, unsmug_url
)
503 self
.assertEqual(data
, unsmug_data
)
505 res_url
, res_data
= unsmuggle_url(url
)
506 self
.assertEqual(res_url
, url
)
507 self
.assertEqual(res_data
, None)
509 smug_url
= smuggle_url(url
, {'a': 'b'}
)
510 smug_smug_url
= smuggle_url(smug_url
, {'c': 'd'}
)
511 res_url
, res_data
= unsmuggle_url(smug_smug_url
)
512 self
.assertEqual(res_url
, url
)
513 self
.assertEqual(res_data
, {'a': 'b', 'c': 'd'}
)
515 def test_shell_quote(self
):
516 args
= ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
519 """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name
!= 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
521 def test_float_or_none(self
):
522 self
.assertEqual(float_or_none('42.42'), 42.42)
523 self
.assertEqual(float_or_none('42'), 42.0)
524 self
.assertEqual(float_or_none(''), None)
525 self
.assertEqual(float_or_none(None), None)
526 self
.assertEqual(float_or_none([]), None)
527 self
.assertEqual(float_or_none(set()), None)
529 def test_int_or_none(self
):
530 self
.assertEqual(int_or_none('42'), 42)
531 self
.assertEqual(int_or_none(''), None)
532 self
.assertEqual(int_or_none(None), None)
533 self
.assertEqual(int_or_none([]), None)
534 self
.assertEqual(int_or_none(set()), None)
536 def test_str_to_int(self
):
537 self
.assertEqual(str_to_int('123,456'), 123456)
538 self
.assertEqual(str_to_int('123.456'), 123456)
539 self
.assertEqual(str_to_int(523), 523)
540 # Python 3 has no long
541 if sys
.version_info
< (3, 0):
542 eval('self.assertEqual(str_to_int(123456L), 123456)')
543 self
.assertEqual(str_to_int('noninteger'), None)
544 self
.assertEqual(str_to_int([]), None)
546 def test_url_basename(self
):
547 self
.assertEqual(url_basename('http://foo.de/'), '')
548 self
.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
549 self
.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
550 self
.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
551 self
.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
553 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
556 def test_base_url(self
):
557 self
.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
558 self
.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
559 self
.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
560 self
.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
561 self
.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
563 def test_urljoin(self
):
564 self
.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
565 self
.assertEqual(urljoin(b
'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
566 self
.assertEqual(urljoin('http://foo.de/', b
'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
567 self
.assertEqual(urljoin(b
'http://foo.de/', b
'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
568 self
.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
569 self
.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
570 self
.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
571 self
.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
572 self
.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
573 self
.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
574 self
.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
575 self
.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
576 self
.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
577 self
.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
578 self
.assertEqual(urljoin('http://foo.de/', None), None)
579 self
.assertEqual(urljoin('http://foo.de/', ''), None)
580 self
.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
581 self
.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
582 self
.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de')
583 self
.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de')
585 def test_url_or_none(self
):
586 self
.assertEqual(url_or_none(None), None)
587 self
.assertEqual(url_or_none(''), None)
588 self
.assertEqual(url_or_none('foo'), None)
589 self
.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
590 self
.assertEqual(url_or_none('https://foo.de'), 'https://foo.de')
591 self
.assertEqual(url_or_none('http$://foo.de'), None)
592 self
.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
593 self
.assertEqual(url_or_none('//foo.de'), '//foo.de')
594 self
.assertEqual(url_or_none('s3://foo.de'), None)
595 self
.assertEqual(url_or_none('rtmpte://foo.de'), 'rtmpte://foo.de')
596 self
.assertEqual(url_or_none('mms://foo.de'), 'mms://foo.de')
597 self
.assertEqual(url_or_none('rtspu://foo.de'), 'rtspu://foo.de')
598 self
.assertEqual(url_or_none('ftps://foo.de'), 'ftps://foo.de')
600 def test_parse_age_limit(self
):
601 self
.assertEqual(parse_age_limit(None), None)
602 self
.assertEqual(parse_age_limit(False), None)
603 self
.assertEqual(parse_age_limit('invalid'), None)
604 self
.assertEqual(parse_age_limit(0), 0)
605 self
.assertEqual(parse_age_limit(18), 18)
606 self
.assertEqual(parse_age_limit(21), 21)
607 self
.assertEqual(parse_age_limit(22), None)
608 self
.assertEqual(parse_age_limit('18'), 18)
609 self
.assertEqual(parse_age_limit('18+'), 18)
610 self
.assertEqual(parse_age_limit('PG-13'), 13)
611 self
.assertEqual(parse_age_limit('TV-14'), 14)
612 self
.assertEqual(parse_age_limit('TV-MA'), 17)
613 self
.assertEqual(parse_age_limit('TV14'), 14)
614 self
.assertEqual(parse_age_limit('TV_G'), 0)
616 def test_parse_duration(self
):
617 self
.assertEqual(parse_duration(None), None)
618 self
.assertEqual(parse_duration(False), None)
619 self
.assertEqual(parse_duration('invalid'), None)
620 self
.assertEqual(parse_duration('1'), 1)
621 self
.assertEqual(parse_duration('1337:12'), 80232)
622 self
.assertEqual(parse_duration('9:12:43'), 33163)
623 self
.assertEqual(parse_duration('12:00'), 720)
624 self
.assertEqual(parse_duration('00:01:01'), 61)
625 self
.assertEqual(parse_duration('x:y'), None)
626 self
.assertEqual(parse_duration('3h11m53s'), 11513)
627 self
.assertEqual(parse_duration('3h 11m 53s'), 11513)
628 self
.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
629 self
.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
630 self
.assertEqual(parse_duration('62m45s'), 3765)
631 self
.assertEqual(parse_duration('6m59s'), 419)
632 self
.assertEqual(parse_duration('49s'), 49)
633 self
.assertEqual(parse_duration('0h0m0s'), 0)
634 self
.assertEqual(parse_duration('0m0s'), 0)
635 self
.assertEqual(parse_duration('0s'), 0)
636 self
.assertEqual(parse_duration('01:02:03.05'), 3723.05)
637 self
.assertEqual(parse_duration('T30M38S'), 1838)
638 self
.assertEqual(parse_duration('5 s'), 5)
639 self
.assertEqual(parse_duration('3 min'), 180)
640 self
.assertEqual(parse_duration('2.5 hours'), 9000)
641 self
.assertEqual(parse_duration('02:03:04'), 7384)
642 self
.assertEqual(parse_duration('01:02:03:04'), 93784)
643 self
.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
644 self
.assertEqual(parse_duration('87 Min.'), 5220)
645 self
.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
646 self
.assertEqual(parse_duration('PT00H03M30SZ'), 210)
647 self
.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)
648 self
.assertEqual(parse_duration('01:02:03:050'), 3723.05)
649 self
.assertEqual(parse_duration('103:050'), 103.05)
651 def test_fix_xml_ampersands(self
):
653 fix_xml_ampersands('"&x=y&z=a'), '"&x=y&z=a')
655 fix_xml_ampersands('"&x=y&wrong;&z=a'),
656 '"&x=y&wrong;&z=a')
658 fix_xml_ampersands('&'><"'),
659 '&'><"')
661 fix_xml_ampersands('Ӓ᪼'), 'Ӓ᪼')
662 self
.assertEqual(fix_xml_ampersands('&#&#'), '&#&#')
664 def test_paged_list(self
):
665 def testPL(size
, pagesize
, sliceargs
, expected
):
666 def get_page(pagenum
):
667 firstid
= pagenum
* pagesize
668 upto
= min(size
, pagenum
* pagesize
+ pagesize
)
669 for i
in range(firstid
, upto
):
672 pl
= OnDemandPagedList(get_page
, pagesize
)
673 got
= pl
.getslice(*sliceargs
)
674 self
.assertEqual(got
, expected
)
676 iapl
= InAdvancePagedList(get_page
, size
// pagesize
+ 1, pagesize
)
677 got
= iapl
.getslice(*sliceargs
)
678 self
.assertEqual(got
, expected
)
680 testPL(5, 2, (), [0, 1, 2, 3, 4])
681 testPL(5, 2, (1,), [1, 2, 3, 4])
682 testPL(5, 2, (2,), [2, 3, 4])
683 testPL(5, 2, (4,), [4])
684 testPL(5, 2, (0, 3), [0, 1, 2])
685 testPL(5, 2, (1, 4), [1, 2, 3])
686 testPL(5, 2, (2, 99), [2, 3, 4])
687 testPL(5, 2, (20, 99), [])
689 def test_read_batch_urls(self
):
690 f
= io
.StringIO('''\xef\xbb\xbf foo
693 # More after this line\r
696 self
.assertEqual(read_batch_urls(f
), ['foo', 'bar', 'baz', 'bam'])
698 def test_urlencode_postdata(self
):
699 data
= urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'}
)
700 self
.assertTrue(isinstance(data
, bytes))
702 def test_update_url_query(self
):
703 self
.assertEqual(parse_qs(update_url_query(
704 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']}
)),
705 parse_qs('http://example.com/path?quality=HD&format=mp4'))
706 self
.assertEqual(parse_qs(update_url_query(
707 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']}
)),
708 parse_qs('http://example.com/path?system=LINUX&system=WINDOWS'))
709 self
.assertEqual(parse_qs(update_url_query(
710 'http://example.com/path', {'fields': 'id,formats,subtitles'}
)),
711 parse_qs('http://example.com/path?fields=id,formats,subtitles'))
712 self
.assertEqual(parse_qs(update_url_query(
713 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')}
)),
714 parse_qs('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
715 self
.assertEqual(parse_qs(update_url_query(
716 'http://example.com/path?manifest=f4m', {'manifest': []}
)),
717 parse_qs('http://example.com/path'))
718 self
.assertEqual(parse_qs(update_url_query(
719 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'}
)),
720 parse_qs('http://example.com/path?system=LINUX'))
721 self
.assertEqual(parse_qs(update_url_query(
722 'http://example.com/path', {'fields': b'id,formats,subtitles'}
)),
723 parse_qs('http://example.com/path?fields=id,formats,subtitles'))
724 self
.assertEqual(parse_qs(update_url_query(
725 'http://example.com/path', {'width': 1080, 'height': 720}
)),
726 parse_qs('http://example.com/path?width=1080&height=720'))
727 self
.assertEqual(parse_qs(update_url_query(
728 'http://example.com/path', {'bitrate': 5020.43}
)),
729 parse_qs('http://example.com/path?bitrate=5020.43'))
730 self
.assertEqual(parse_qs(update_url_query(
731 'http://example.com/path', {'test': '第二行тест'}
)),
732 parse_qs('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
734 def test_multipart_encode(self
):
736 multipart_encode({b'field': b'value'}
, boundary
='AAAAAA')[0],
737 b
'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
739 multipart_encode({'欄位'.encode('utf-8'): '值'.encode('utf-8')}
, boundary
='AAAAAA')[0],
740 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')
742 ValueError, multipart_encode
, {b'field': b'value'}
, boundary
='value')
744 def test_dict_get(self
):
752 d
= FALSE_VALUES
.copy()
754 self
.assertEqual(dict_get(d
, 'a'), 42)
755 self
.assertEqual(dict_get(d
, 'b'), None)
756 self
.assertEqual(dict_get(d
, 'b', 42), 42)
757 self
.assertEqual(dict_get(d
, ('a', )), 42)
758 self
.assertEqual(dict_get(d
, ('b', 'a', )), 42)
759 self
.assertEqual(dict_get(d
, ('b', 'c', 'a', 'd', )), 42)
760 self
.assertEqual(dict_get(d
, ('b', 'c', )), None)
761 self
.assertEqual(dict_get(d
, ('b', 'c', ), 42), 42)
762 for key
, false_value
in FALSE_VALUES
.items():
763 self
.assertEqual(dict_get(d
, ('b', 'c', key
, )), None)
764 self
.assertEqual(dict_get(d
, ('b', 'c', key
, ), skip_false_values
=False), false_value
)
766 def test_merge_dicts(self
):
767 self
.assertEqual(merge_dicts({'a': 1}
, {'b': 2}
), {'a': 1, 'b': 2}
)
768 self
.assertEqual(merge_dicts({'a': 1}
, {'a': 2}
), {'a': 1}
)
769 self
.assertEqual(merge_dicts({'a': 1}
, {'a': None}
), {'a': 1}
)
770 self
.assertEqual(merge_dicts({'a': 1}
, {'a': ''}
), {'a': 1}
)
771 self
.assertEqual(merge_dicts({'a': 1}
, {}), {'a': 1}
)
772 self
.assertEqual(merge_dicts({'a': None}
, {'a': 1}
), {'a': 1}
)
773 self
.assertEqual(merge_dicts({'a': ''}
, {'a': 1}
), {'a': ''}
)
774 self
.assertEqual(merge_dicts({'a': ''}
, {'a': 'abc'}
), {'a': 'abc'}
)
775 self
.assertEqual(merge_dicts({'a': None}
, {'a': ''}
, {'a': 'abc'}
), {'a': 'abc'}
)
777 def test_encode_compat_str(self
):
778 self
.assertEqual(encode_compat_str(b
'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
779 self
.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
781 def test_parse_iso8601(self
):
782 self
.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
783 self
.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
784 self
.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
785 self
.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
786 self
.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
787 self
.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
789 def test_strip_jsonp(self
):
790 stripped
= strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
791 d
= json
.loads(stripped
)
792 self
.assertEqual(d
, [{"id": "532cb", "x": 3}
])
794 stripped
= strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
795 d
= json
.loads(stripped
)
796 self
.assertEqual(d
, {'STATUS': 'OK'}
)
798 stripped
= strip_jsonp('ps.embedHandler({"status": "success"});')
799 d
= json
.loads(stripped
)
800 self
.assertEqual(d
, {'status': 'success'}
)
802 stripped
= strip_jsonp('window.cb && window.cb({"status": "success"});')
803 d
= json
.loads(stripped
)
804 self
.assertEqual(d
, {'status': 'success'}
)
806 stripped
= strip_jsonp('window.cb && cb({"status": "success"});')
807 d
= json
.loads(stripped
)
808 self
.assertEqual(d
, {'status': 'success'}
)
810 stripped
= strip_jsonp('({"status": "success"});')
811 d
= json
.loads(stripped
)
812 self
.assertEqual(d
, {'status': 'success'}
)
814 def test_strip_or_none(self
):
815 self
.assertEqual(strip_or_none(' abc'), 'abc')
816 self
.assertEqual(strip_or_none('abc '), 'abc')
817 self
.assertEqual(strip_or_none(' abc '), 'abc')
818 self
.assertEqual(strip_or_none('\tabc\t'), 'abc')
819 self
.assertEqual(strip_or_none('\n\tabc\n\t'), 'abc')
820 self
.assertEqual(strip_or_none('abc'), 'abc')
821 self
.assertEqual(strip_or_none(''), '')
822 self
.assertEqual(strip_or_none(None), None)
823 self
.assertEqual(strip_or_none(42), None)
824 self
.assertEqual(strip_or_none([]), None)
826 def test_uppercase_escape(self
):
827 self
.assertEqual(uppercase_escape('aä'), 'aä')
828 self
.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
830 def test_lowercase_escape(self
):
831 self
.assertEqual(lowercase_escape('aä'), 'aä')
832 self
.assertEqual(lowercase_escape('\\u0026'), '&')
834 def test_limit_length(self
):
835 self
.assertEqual(limit_length(None, 12), None)
836 self
.assertEqual(limit_length('foo', 12), 'foo')
838 limit_length('foo bar baz asd', 12).startswith('foo bar'))
839 self
.assertTrue('...' in limit_length('foo bar baz asd', 12))
841 def test_mimetype2ext(self
):
842 self
.assertEqual(mimetype2ext(None), None)
843 self
.assertEqual(mimetype2ext('video/x-flv'), 'flv')
844 self
.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
845 self
.assertEqual(mimetype2ext('text/vtt'), 'vtt')
846 self
.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
847 self
.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
848 self
.assertEqual(mimetype2ext('audio/x-wav'), 'wav')
849 self
.assertEqual(mimetype2ext('audio/x-wav;codec=pcm'), 'wav')
851 def test_month_by_name(self
):
852 self
.assertEqual(month_by_name(None), None)
853 self
.assertEqual(month_by_name('December', 'en'), 12)
854 self
.assertEqual(month_by_name('décembre', 'fr'), 12)
855 self
.assertEqual(month_by_name('December'), 12)
856 self
.assertEqual(month_by_name('décembre'), None)
857 self
.assertEqual(month_by_name('Unknown', 'unknown'), None)
859 def test_parse_codecs(self
):
860 self
.assertEqual(parse_codecs(''), {})
861 self
.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
862 'vcodec': 'avc1.77.30',
863 'acodec': 'mp4a.40.2',
864 'dynamic_range': None,
866 self
.assertEqual(parse_codecs('mp4a.40.2'), {
868 'acodec': 'mp4a.40.2',
869 'dynamic_range': None,
871 self
.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
872 'vcodec': 'avc1.42001e',
873 'acodec': 'mp4a.40.5',
874 'dynamic_range': None,
876 self
.assertEqual(parse_codecs('avc3.640028'), {
877 'vcodec': 'avc3.640028',
879 'dynamic_range': None,
881 self
.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
884 'dynamic_range': None,
886 self
.assertEqual(parse_codecs('av01.0.05M.08'), {
887 'vcodec': 'av01.0.05M.08',
889 'dynamic_range': None,
891 self
.assertEqual(parse_codecs('vp9.2'), {
894 'dynamic_range': 'HDR10',
896 self
.assertEqual(parse_codecs('av01.0.12M.10.0.110.09.16.09.0'), {
897 'vcodec': 'av01.0.12M.10',
899 'dynamic_range': 'HDR10',
901 self
.assertEqual(parse_codecs('dvhe'), {
904 'dynamic_range': 'DV',
906 self
.assertEqual(parse_codecs('theora, vorbis'), {
909 'dynamic_range': None,
911 self
.assertEqual(parse_codecs('unknownvcodec, unknownacodec'), {
912 'vcodec': 'unknownvcodec',
913 'acodec': 'unknownacodec',
915 self
.assertEqual(parse_codecs('unknown'), {})
917 def test_escape_rfc3986(self
):
918 reserved
= "!*'();:@&=+$,/?#[]"
919 unreserved
= 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
920 self
.assertEqual(escape_rfc3986(reserved
), reserved
)
921 self
.assertEqual(escape_rfc3986(unreserved
), unreserved
)
922 self
.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
923 self
.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
924 self
.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
925 self
.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
927 def test_escape_url(self
):
929 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
930 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
933 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
934 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
937 escape_url('http://тест.рф/фрагмент'),
938 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
941 escape_url('http://тест.рф/абв?абв=абв#абв'),
942 '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'
944 self
.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
946 def test_js_to_json_realworld(self
):
948 'clip':{'provider':'pseudo'}
950 self
.assertEqual(js_to_json(inp
), '''{
951 "clip":{"provider":"pseudo"}
953 json
.loads(js_to_json(inp
))
956 'playlist':[{'controls':{'all':null}}]
958 self
.assertEqual(js_to_json(inp
), '''{
959 "playlist":[{"controls":{"all":null}}]
962 inp
= '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
963 self
.assertEqual(js_to_json(inp
), '''"The CW's 'Crazy Ex-Girlfriend'"''')
965 inp
= '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
966 json_code
= js_to_json(inp
)
967 self
.assertEqual(json
.loads(json_code
), json
.loads(inp
))
970 0:{src:'skipped', type: 'application/dash+xml'},
971 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
973 self
.assertEqual(js_to_json(inp
), '''{
974 "0":{"src":"skipped", "type": "application/dash+xml"},
975 "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
978 inp
= '''{"foo":101}'''
979 self
.assertEqual(js_to_json(inp
), '''{"foo":101}''')
981 inp
= '''{"duration": "00:01:07"}'''
982 self
.assertEqual(js_to_json(inp
), '''{"duration": "00:01:07"}''')
984 inp
= '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}'''
985 self
.assertEqual(js_to_json(inp
), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''')
987 def test_js_to_json_edgecases(self
):
988 on
= js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
989 self
.assertEqual(json
.loads(on
), {"abc_def": "1'\\2\\'3\"4"}
)
991 on
= js_to_json('{"abc": true}')
992 self
.assertEqual(json
.loads(on
), {'abc': True}
)
994 # Ignore JavaScript code as well
1001 self
.assertEqual(d
['x'], 1)
1002 self
.assertEqual(d
['y'], 'a')
1004 # Just drop ! prefix for now though this results in a wrong value
1005 on
= js_to_json('''{
1015 self
.assertEqual(json
.loads(on
), {
1026 on
= js_to_json('["abc", "def",]')
1027 self
.assertEqual(json
.loads(on
), ['abc', 'def'])
1029 on
= js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
1030 self
.assertEqual(json
.loads(on
), ['abc', 'def'])
1032 on
= js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
1033 self
.assertEqual(json
.loads(on
), ['abc', 'def'])
1035 on
= js_to_json('{"abc": "def",}')
1036 self
.assertEqual(json
.loads(on
), {'abc': 'def'}
)
1038 on
= js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
1039 self
.assertEqual(json
.loads(on
), {'abc': 'def'}
)
1041 on
= js_to_json('{ 0: /* " \n */ ",]" , }')
1042 self
.assertEqual(json
.loads(on
), {'0': ',]'}
)
1044 on
= js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
1045 self
.assertEqual(json
.loads(on
), {'0': ',]'}
)
1047 on
= js_to_json('{ 0: // comment\n1 }')
1048 self
.assertEqual(json
.loads(on
), {'0': 1}
)
1050 on
= js_to_json(r
'["<p>x<\/p>"]')
1051 self
.assertEqual(json
.loads(on
), ['<p>x</p>'])
1053 on
= js_to_json(r
'["\xaa"]')
1054 self
.assertEqual(json
.loads(on
), ['\u00aa'])
1056 on
= js_to_json("['a\\\nb']")
1057 self
.assertEqual(json
.loads(on
), ['ab'])
1059 on
= js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
1060 self
.assertEqual(json
.loads(on
), ['ab'])
1062 on
= js_to_json('{0xff:0xff}')
1063 self
.assertEqual(json
.loads(on
), {'255': 255}
)
1065 on
= js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
1066 self
.assertEqual(json
.loads(on
), {'255': 255}
)
1068 on
= js_to_json('{077:077}')
1069 self
.assertEqual(json
.loads(on
), {'63': 63}
)
1071 on
= js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
1072 self
.assertEqual(json
.loads(on
), {'63': 63}
)
1074 on
= js_to_json('{42:42}')
1075 self
.assertEqual(json
.loads(on
), {'42': 42}
)
1077 on
= js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
1078 self
.assertEqual(json
.loads(on
), {'42': 42}
)
1080 on
= js_to_json('{42:4.2e1}')
1081 self
.assertEqual(json
.loads(on
), {'42': 42.0}
)
1083 on
= js_to_json('{ "0x40": "0x40" }')
1084 self
.assertEqual(json
.loads(on
), {'0x40': '0x40'}
)
1086 on
= js_to_json('{ "040": "040" }')
1087 self
.assertEqual(json
.loads(on
), {'040': '040'}
)
1089 on
= js_to_json('[1,//{},\n2]')
1090 self
.assertEqual(json
.loads(on
), [1, 2])
1092 def test_js_to_json_malformed(self
):
1093 self
.assertEqual(js_to_json('42a1'), '42"a1"')
1094 self
.assertEqual(js_to_json('42a-1'), '42"a"-1')
1096 def test_extract_attributes(self
):
1097 self
.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}
)
1098 self
.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'}
)
1099 self
.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'}
)
1100 self
.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"}
)
1101 self
.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'}
)
1102 self
.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}
)
1103 self
.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}
)
1104 self
.assertEqual(extract_attributes('<e x="&">'), {'x': '&'}
) # XML
1105 self
.assertEqual(extract_attributes('<e x=""">'), {'x': '"'}
)
1106 self
.assertEqual(extract_attributes('<e x="£">'), {'x': '£'}
) # HTML 3.2
1107 self
.assertEqual(extract_attributes('<e x="λ">'), {'x': 'λ'}
) # HTML 4.0
1108 self
.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'}
)
1109 self
.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"}
)
1110 self
.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'}
)
1111 self
.assertEqual(extract_attributes('<e x >'), {'x': None}
)
1112 self
.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None}
)
1113 self
.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'}
)
1114 self
.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'}
)
1115 self
.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'}
)
1116 self
.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'}
)
1117 self
.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'}
)
1118 self
.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'}
)
1119 self
.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}
) # Names lowercased
1120 self
.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'}
)
1121 self
.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'}
)
1122 self
.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'}
)
1123 self
.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'}
)
1124 self
.assertEqual(extract_attributes('<e x="décomposé">'), {'x': 'décompose\u0301'}
)
1125 # "Narrow" Python builds don't support unicode code points outside BMP.
1128 supports_outside_bmp
= True
1130 supports_outside_bmp
= False
1131 if supports_outside_bmp
:
1132 self
.assertEqual(extract_attributes('<e x="Smile 😀!">'), {'x': 'Smile \U0001f600!'}
)
1133 # Malformed HTML should not break attributes extraction on older Python
1134 self
.assertEqual(extract_attributes('<mal"formed/>'), {})
1136 def test_clean_html(self
):
1137 self
.assertEqual(clean_html('a:\nb'), 'a: b')
1138 self
.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
1139 self
.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
1141 def test_intlist_to_bytes(self
):
1143 intlist_to_bytes([0, 1, 127, 128, 255]),
1144 b
'\x00\x01\x7f\x80\xff')
1146 def test_args_to_str(self
):
1148 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
1149 'foo ba/r -baz \'2 be\' \'\'' if compat_os_name
!= 'nt' else 'foo ba/r -baz "2 be" ""'
1152 def test_parse_filesize(self
):
1153 self
.assertEqual(parse_filesize(None), None)
1154 self
.assertEqual(parse_filesize(''), None)
1155 self
.assertEqual(parse_filesize('91 B'), 91)
1156 self
.assertEqual(parse_filesize('foobar'), None)
1157 self
.assertEqual(parse_filesize('2 MiB'), 2097152)
1158 self
.assertEqual(parse_filesize('5 GB'), 5000000000)
1159 self
.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
1160 self
.assertEqual(parse_filesize('1.2tb'), 1200000000000)
1161 self
.assertEqual(parse_filesize('1,24 KB'), 1240)
1162 self
.assertEqual(parse_filesize('1,24 kb'), 1240)
1163 self
.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
1165 def test_parse_count(self
):
1166 self
.assertEqual(parse_count(None), None)
1167 self
.assertEqual(parse_count(''), None)
1168 self
.assertEqual(parse_count('0'), 0)
1169 self
.assertEqual(parse_count('1000'), 1000)
1170 self
.assertEqual(parse_count('1.000'), 1000)
1171 self
.assertEqual(parse_count('1.1k'), 1100)
1172 self
.assertEqual(parse_count('1.1 k'), 1100)
1173 self
.assertEqual(parse_count('1,1 k'), 1100)
1174 self
.assertEqual(parse_count('1.1kk'), 1100000)
1175 self
.assertEqual(parse_count('1.1kk '), 1100000)
1176 self
.assertEqual(parse_count('1,1kk'), 1100000)
1177 self
.assertEqual(parse_count('100 views'), 100)
1178 self
.assertEqual(parse_count('1,100 views'), 1100)
1179 self
.assertEqual(parse_count('1.1kk views'), 1100000)
1180 self
.assertEqual(parse_count('10M views'), 10000000)
1181 self
.assertEqual(parse_count('has 10M views'), 10000000)
1183 def test_parse_resolution(self
):
1184 self
.assertEqual(parse_resolution(None), {})
1185 self
.assertEqual(parse_resolution(''), {})
1186 self
.assertEqual(parse_resolution(' 1920x1080'), {'width': 1920, 'height': 1080}
)
1187 self
.assertEqual(parse_resolution('1920×1080 '), {'width': 1920, 'height': 1080}
)
1188 self
.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080}
)
1189 self
.assertEqual(parse_resolution('720p'), {'height': 720}
)
1190 self
.assertEqual(parse_resolution('4k'), {'height': 2160}
)
1191 self
.assertEqual(parse_resolution('8K'), {'height': 4320}
)
1192 self
.assertEqual(parse_resolution('pre_1920x1080_post'), {'width': 1920, 'height': 1080}
)
1193 self
.assertEqual(parse_resolution('ep1x2'), {})
1194 self
.assertEqual(parse_resolution('1920, 1080'), {'width': 1920, 'height': 1080}
)
1196 def test_parse_bitrate(self
):
1197 self
.assertEqual(parse_bitrate(None), None)
1198 self
.assertEqual(parse_bitrate(''), None)
1199 self
.assertEqual(parse_bitrate('300kbps'), 300)
1200 self
.assertEqual(parse_bitrate('1500kbps'), 1500)
1201 self
.assertEqual(parse_bitrate('300 kbps'), 300)
1203 def test_version_tuple(self
):
1204 self
.assertEqual(version_tuple('1'), (1,))
1205 self
.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
1206 self
.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
1208 def test_detect_exe_version(self
):
1209 self
.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
1210 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
1211 configuration: --prefix=/usr --extra-'''), '1.2.1')
1212 self
.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
1213 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
1214 self
.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
1215 Trying to open render node...
1216 Success at /dev/dri/renderD128.
1217 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
1219 def test_age_restricted(self
):
1220 self
.assertFalse(age_restricted(None, 10)) # unrestricted content
1221 self
.assertFalse(age_restricted(1, None)) # unrestricted policy
1222 self
.assertFalse(age_restricted(8, 10))
1223 self
.assertTrue(age_restricted(18, 14))
1224 self
.assertFalse(age_restricted(18, 18))
1226 def test_is_html(self
):
1227 self
.assertFalse(is_html(b
'\x49\x44\x43<html'))
1228 self
.assertTrue(is_html(b
'<!DOCTYPE foo>\xaaa'))
1229 self
.assertTrue(is_html( # UTF-8 with BOM
1230 b
'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
1231 self
.assertTrue(is_html( # UTF-16-LE
1232 b
'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
1234 self
.assertTrue(is_html( # UTF-16-BE
1235 b
'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
1237 self
.assertTrue(is_html( # UTF-32-BE
1238 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'))
1239 self
.assertTrue(is_html( # UTF-32-LE
1240 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'))
1242 def test_render_table(self
):
1245 ['a', 'empty', 'bcd'],
1246 [[123, '', 4], [9999, '', 51]]),
1253 ['a', 'empty', 'bcd'],
1254 [[123, '', 4], [9999, '', 51]],
1263 [['1\t23', 4], ['\t9999', 51]]),
1271 [[123, 4], [9999, 51]],
1281 [[123, 4], [9999, 51]],
1282 delim
='-', extra_gap
=2),
1288 def test_match_str(self
):
1290 self
.assertFalse(match_str('xy', {'x': 1200}
))
1291 self
.assertTrue(match_str('!xy', {'x': 1200}
))
1292 self
.assertTrue(match_str('x', {'x': 1200}
))
1293 self
.assertFalse(match_str('!x', {'x': 1200}
))
1294 self
.assertTrue(match_str('x', {'x': 0}
))
1295 self
.assertTrue(match_str('is_live', {'is_live': True}
))
1296 self
.assertFalse(match_str('is_live', {'is_live': False}
))
1297 self
.assertFalse(match_str('is_live', {'is_live': None}
))
1298 self
.assertFalse(match_str('is_live', {}))
1299 self
.assertFalse(match_str('!is_live', {'is_live': True}
))
1300 self
.assertTrue(match_str('!is_live', {'is_live': False}
))
1301 self
.assertTrue(match_str('!is_live', {'is_live': None}
))
1302 self
.assertTrue(match_str('!is_live', {}))
1303 self
.assertTrue(match_str('title', {'title': 'abc'}
))
1304 self
.assertTrue(match_str('title', {'title': ''}
))
1305 self
.assertFalse(match_str('!title', {'title': 'abc'}
))
1306 self
.assertFalse(match_str('!title', {'title': ''}
))
1309 self
.assertFalse(match_str('x>0', {'x': 0}
))
1310 self
.assertFalse(match_str('x>0', {}))
1311 self
.assertTrue(match_str('x>?0', {}))
1312 self
.assertTrue(match_str('x>1K', {'x': 1200}
))
1313 self
.assertFalse(match_str('x>2K', {'x': 1200}
))
1314 self
.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}
))
1315 self
.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}
))
1316 self
.assertTrue(match_str('x > 1:0:0', {'x': 3700}
))
1319 self
.assertFalse(match_str('y=a212', {'y': 'foobar42'}
))
1320 self
.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}
))
1321 self
.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}
))
1322 self
.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}
))
1323 self
.assertTrue(match_str('y^=foo', {'y': 'foobar42'}
))
1324 self
.assertFalse(match_str('y!^=foo', {'y': 'foobar42'}
))
1325 self
.assertFalse(match_str('y^=bar', {'y': 'foobar42'}
))
1326 self
.assertTrue(match_str('y!^=bar', {'y': 'foobar42'}
))
1327 self
.assertRaises(ValueError, match_str
, 'x^=42', {'x': 42}
)
1328 self
.assertTrue(match_str('y*=bar', {'y': 'foobar42'}
))
1329 self
.assertFalse(match_str('y!*=bar', {'y': 'foobar42'}
))
1330 self
.assertFalse(match_str('y*=baz', {'y': 'foobar42'}
))
1331 self
.assertTrue(match_str('y!*=baz', {'y': 'foobar42'}
))
1332 self
.assertTrue(match_str('y$=42', {'y': 'foobar42'}
))
1333 self
.assertFalse(match_str('y$=43', {'y': 'foobar42'}
))
1336 self
.assertFalse(match_str(
1337 'like_count > 100 & dislike_count <? 50 & description',
1338 {'like_count': 90, 'description': 'foo'}
))
1339 self
.assertTrue(match_str(
1340 'like_count > 100 & dislike_count <? 50 & description',
1341 {'like_count': 190, 'description': 'foo'}
))
1342 self
.assertFalse(match_str(
1343 'like_count > 100 & dislike_count <? 50 & description',
1344 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}
))
1345 self
.assertFalse(match_str(
1346 'like_count > 100 & dislike_count <? 50 & description',
1347 {'like_count': 190, 'dislike_count': 10}
))
1350 self
.assertTrue(match_str(r
'x~=\bbar', {'x': 'foo bar'}
))
1351 self
.assertFalse(match_str(r
'x~=\bbar.+', {'x': 'foo bar'}
))
1352 self
.assertFalse(match_str(r
'x~=^FOO', {'x': 'foo bar'}
))
1353 self
.assertTrue(match_str(r
'x~=(?i)^FOO', {'x': 'foo bar'}
))
1356 self
.assertTrue(match_str(r
'x^="foo"', {'x': 'foo "bar"'}
))
1357 self
.assertFalse(match_str(r
'x^="foo "', {'x': 'foo "bar"'}
))
1358 self
.assertFalse(match_str(r
'x$="bar"', {'x': 'foo "bar"'}
))
1359 self
.assertTrue(match_str(r
'x$=" \"bar\""', {'x': 'foo "bar"'}
))
1362 self
.assertFalse(match_str(r
'x=foo & bar', {'x': 'foo & bar'}
))
1363 self
.assertTrue(match_str(r
'x=foo \& bar', {'x': 'foo & bar'}
))
1364 self
.assertTrue(match_str(r
'x=foo \& bar & x^=foo', {'x': 'foo & bar'}
))
1365 self
.assertTrue(match_str(r
'x="foo \& bar" & x^=foo', {'x': 'foo & bar'}
))
1368 self
.assertTrue(match_str(
1369 r
"!is_live & like_count>?100 & description~='(?i)\bcats \& dogs\b'",
1370 {'description': 'Raining Cats & Dogs'}
))
1373 self
.assertFalse(match_str('id!=foo', {'id': 'foo'}
, True))
1374 self
.assertTrue(match_str('x', {'id': 'foo'}
, True))
1375 self
.assertTrue(match_str('!x', {'id': 'foo'}
, True))
1376 self
.assertFalse(match_str('x', {'id': 'foo'}
, False))
1378 def test_parse_dfxp_time_expr(self
):
1379 self
.assertEqual(parse_dfxp_time_expr(None), None)
1380 self
.assertEqual(parse_dfxp_time_expr(''), None)
1381 self
.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1382 self
.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1383 self
.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1384 self
.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1385 self
.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1387 def test_dfxp2srt(self
):
1388 dfxp_data
= '''<?xml version="1.0" encoding="UTF-8"?>
1389 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1392 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1393 <p begin="1" end="2">第二行<br/>♪♪</p>
1394 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1395 <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1396 <p begin="-1" end="-1">Ignore, two</p>
1397 <p begin="3" dur="-1">Ignored, three</p>
1400 </tt>'''.encode('utf-8')
1402 00:00:00,000 --> 00:00:01,000
1403 The following line contains Chinese characters and special symbols
1406 00:00:01,000 --> 00:00:02,000
1411 00:00:02,000 --> 00:00:03,000
1416 self
.assertEqual(dfxp2srt(dfxp_data
), srt_data
)
1418 dfxp_data_no_default_namespace
= '''<?xml version="1.0" encoding="UTF-8"?>
1419 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1422 <p begin="0" end="1">The first line</p>
1425 </tt>'''.encode('utf-8')
1427 00:00:00,000 --> 00:00:01,000
1431 self
.assertEqual(dfxp2srt(dfxp_data_no_default_namespace
), srt_data
)
1433 dfxp_data_with_style
= '''<?xml version="1.0" encoding="utf-8"?>
1434 <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">
1437 <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1438 <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1439 <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1440 <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1443 <body tts:textAlign="center" style="s0">
1445 <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1446 <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>
1447 <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1448 <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>
1451 </tt>'''.encode('utf-8')
1453 00:00:02,080 --> 00:00:05,840
1454 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1457 00:00:02,080 --> 00:00:05,840
1458 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1459 </font>part 2</font></b>
1462 00:00:05,840 --> 00:00:09,560
1463 <u><font color="lime">line 3
1467 00:00:09,560 --> 00:00:12,360
1468 <i><u><font color="yellow"><font color="lime">inner
1469 </font>style</font></u></i>
1472 self
.assertEqual(dfxp2srt(dfxp_data_with_style
), srt_data
)
1474 dfxp_data_non_utf8
= '''<?xml version="1.0" encoding="UTF-16"?>
1475 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1478 <p begin="0" end="1">Line 1</p>
1479 <p begin="1" end="2">第二行</p>
1482 </tt>'''.encode('utf-16')
1484 00:00:00,000 --> 00:00:01,000
1488 00:00:01,000 --> 00:00:02,000
1492 self
.assertEqual(dfxp2srt(dfxp_data_non_utf8
), srt_data
)
1494 def test_cli_option(self
):
1495 self
.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}
, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1496 self
.assertEqual(cli_option({'proxy': None}
, '--proxy', 'proxy'), [])
1497 self
.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1498 self
.assertEqual(cli_option({'retries': 10}
, '--retries', 'retries'), ['--retries', '10'])
1500 def test_cli_valueless_option(self
):
1501 self
.assertEqual(cli_valueless_option(
1502 {'downloader': 'external'}
, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1503 self
.assertEqual(cli_valueless_option(
1504 {'downloader': 'internal'}
, '--external-downloader', 'downloader', 'external'), [])
1505 self
.assertEqual(cli_valueless_option(
1506 {'nocheckcertificate': True}
, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1507 self
.assertEqual(cli_valueless_option(
1508 {'nocheckcertificate': False}
, '--no-check-certificate', 'nocheckcertificate'), [])
1509 self
.assertEqual(cli_valueless_option(
1510 {'checkcertificate': True}
, '--no-check-certificate', 'checkcertificate', False), [])
1511 self
.assertEqual(cli_valueless_option(
1512 {'checkcertificate': False}
, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1514 def test_cli_bool_option(self
):
1517 {'nocheckcertificate': True}
, '--no-check-certificate', 'nocheckcertificate'),
1518 ['--no-check-certificate', 'true'])
1521 {'nocheckcertificate': True}
, '--no-check-certificate', 'nocheckcertificate', separator
='='),
1522 ['--no-check-certificate=true'])
1525 {'nocheckcertificate': True}
, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1526 ['--check-certificate', 'false'])
1529 {'nocheckcertificate': True}
, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1530 ['--check-certificate=false'])
1533 {'nocheckcertificate': False}
, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1534 ['--check-certificate', 'true'])
1537 {'nocheckcertificate': False}
, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1538 ['--check-certificate=true'])
1541 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1544 def test_ohdave_rsa_encrypt(self
):
1545 N
= 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1549 ohdave_rsa_encrypt(b
'aa111222', e
, N
),
1550 '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1552 def test_pkcs1pad(self
):
1554 padded_data
= pkcs1pad(data
, 32)
1555 self
.assertEqual(padded_data
[:2], [0, 2])
1556 self
.assertEqual(padded_data
[28:], [0, 1, 2, 3])
1558 self
.assertRaises(ValueError, pkcs1pad
, data
, 8)
1560 def test_encode_base_n(self
):
1561 self
.assertEqual(encode_base_n(0, 30), '0')
1562 self
.assertEqual(encode_base_n(80, 30), '2k')
1564 custom_table
= '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1565 self
.assertEqual(encode_base_n(0, 30, custom_table
), '9')
1566 self
.assertEqual(encode_base_n(80, 30, custom_table
), '7P')
1568 self
.assertRaises(ValueError, encode_base_n
, 0, 70)
1569 self
.assertRaises(ValueError, encode_base_n
, 0, 60, custom_table
)
1571 def test_caesar(self
):
1572 self
.assertEqual(caesar('ace', 'abcdef', 2), 'cea')
1573 self
.assertEqual(caesar('cea', 'abcdef', -2), 'ace')
1574 self
.assertEqual(caesar('ace', 'abcdef', -2), 'eac')
1575 self
.assertEqual(caesar('eac', 'abcdef', 2), 'ace')
1576 self
.assertEqual(caesar('ace', 'abcdef', 0), 'ace')
1577 self
.assertEqual(caesar('xyz', 'abcdef', 2), 'xyz')
1578 self
.assertEqual(caesar('abc', 'acegik', 2), 'ebg')
1579 self
.assertEqual(caesar('ebg', 'acegik', -2), 'abc')
1581 def test_rot47(self
):
1582 self
.assertEqual(rot47('yt-dlp'), r
'JE\5=A')
1583 self
.assertEqual(rot47('YT-DLP'), r
'*%\s{!')
1585 def test_urshift(self
):
1586 self
.assertEqual(urshift(3, 1), 1)
1587 self
.assertEqual(urshift(-3, 1), 2147483646)
1589 GET_ELEMENT_BY_CLASS_TEST_STRING
= '''
1590 <span class="foo bar">nice</span>
1593 def test_get_element_by_class(self
):
1594 html
= self
.GET_ELEMENT_BY_CLASS_TEST_STRING
1596 self
.assertEqual(get_element_by_class('foo', html
), 'nice')
1597 self
.assertEqual(get_element_by_class('no-such-class', html
), None)
1599 def test_get_element_html_by_class(self
):
1600 html
= self
.GET_ELEMENT_BY_CLASS_TEST_STRING
1602 self
.assertEqual(get_element_html_by_class('foo', html
), html
.strip())
1603 self
.assertEqual(get_element_by_class('no-such-class', html
), None)
1605 GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
= '''
1606 <div itemprop="author" itemscope>foo</div>
1609 def test_get_element_by_attribute(self
):
1610 html
= self
.GET_ELEMENT_BY_CLASS_TEST_STRING
1612 self
.assertEqual(get_element_by_attribute('class', 'foo bar', html
), 'nice')
1613 self
.assertEqual(get_element_by_attribute('class', 'foo', html
), None)
1614 self
.assertEqual(get_element_by_attribute('class', 'no-such-foo', html
), None)
1616 html
= self
.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
1618 self
.assertEqual(get_element_by_attribute('itemprop', 'author', html
), 'foo')
1620 def test_get_element_html_by_attribute(self
):
1621 html
= self
.GET_ELEMENT_BY_CLASS_TEST_STRING
1623 self
.assertEqual(get_element_html_by_attribute('class', 'foo bar', html
), html
.strip())
1624 self
.assertEqual(get_element_html_by_attribute('class', 'foo', html
), None)
1625 self
.assertEqual(get_element_html_by_attribute('class', 'no-such-foo', html
), None)
1627 html
= self
.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
1629 self
.assertEqual(get_element_html_by_attribute('itemprop', 'author', html
), html
.strip())
1631 GET_ELEMENTS_BY_CLASS_TEST_STRING
= '''
1632 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1634 GET_ELEMENTS_BY_CLASS_RES
= ['<span class="foo bar">nice</span>', '<span class="foo bar">also nice</span>']
1636 def test_get_elements_by_class(self
):
1637 html
= self
.GET_ELEMENTS_BY_CLASS_TEST_STRING
1639 self
.assertEqual(get_elements_by_class('foo', html
), ['nice', 'also nice'])
1640 self
.assertEqual(get_elements_by_class('no-such-class', html
), [])
1642 def test_get_elements_html_by_class(self
):
1643 html
= self
.GET_ELEMENTS_BY_CLASS_TEST_STRING
1645 self
.assertEqual(get_elements_html_by_class('foo', html
), self
.GET_ELEMENTS_BY_CLASS_RES
)
1646 self
.assertEqual(get_elements_html_by_class('no-such-class', html
), [])
1648 def test_get_elements_by_attribute(self
):
1649 html
= self
.GET_ELEMENTS_BY_CLASS_TEST_STRING
1651 self
.assertEqual(get_elements_by_attribute('class', 'foo bar', html
), ['nice', 'also nice'])
1652 self
.assertEqual(get_elements_by_attribute('class', 'foo', html
), [])
1653 self
.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html
), [])
1655 def test_get_elements_html_by_attribute(self
):
1656 html
= self
.GET_ELEMENTS_BY_CLASS_TEST_STRING
1658 self
.assertEqual(get_elements_html_by_attribute('class', 'foo bar', html
), self
.GET_ELEMENTS_BY_CLASS_RES
)
1659 self
.assertEqual(get_elements_html_by_attribute('class', 'foo', html
), [])
1660 self
.assertEqual(get_elements_html_by_attribute('class', 'no-such-foo', html
), [])
1662 def test_get_elements_text_and_html_by_attribute(self
):
1663 html
= self
.GET_ELEMENTS_BY_CLASS_TEST_STRING
1666 list(get_elements_text_and_html_by_attribute('class', 'foo bar', html
)),
1667 list(zip(['nice', 'also nice'], self
.GET_ELEMENTS_BY_CLASS_RES
)))
1668 self
.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'foo', html
)), [])
1669 self
.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'no-such-foo', html
)), [])
1671 GET_ELEMENT_BY_TAG_TEST_STRING
= '''
1672 random text lorem ipsum</p>
1674 this should be returned
1675 <span>this should also be returned</span>
1677 this should also be returned
1679 closing tag above should not trick, so this should also be returned
1681 but this text should not be returned
1683 GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML
= GET_ELEMENT_BY_TAG_TEST_STRING
.strip()[32:276]
1684 GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT
= GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML
[5:-6]
1685 GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML
= GET_ELEMENT_BY_TAG_TEST_STRING
.strip()[78:119]
1686 GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT
= GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML
[6:-7]
1688 def test_get_element_text_and_html_by_tag(self
):
1689 html
= self
.GET_ELEMENT_BY_TAG_TEST_STRING
1692 get_element_text_and_html_by_tag('div', html
),
1693 (self
.GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT
, self
.GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML
))
1695 get_element_text_and_html_by_tag('span', html
),
1696 (self
.GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT
, self
.GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML
))
1697 self
.assertRaises(compat_HTMLParseError
, get_element_text_and_html_by_tag
, 'article', html
)
1699 def test_iri_to_uri(self
):
1701 iri_to_uri('https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b'),
1702 'https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b') # Same
1704 iri_to_uri('https://www.google.com/search?q=Käsesoßenrührlöffel'), # German for cheese sauce stirring spoon
1705 'https://www.google.com/search?q=K%C3%A4seso%C3%9Fenr%C3%BChrl%C3%B6ffel')
1707 iri_to_uri('https://www.google.com/search?q=lt<+gt>+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#'),
1708 'https://www.google.com/search?q=lt%3C+gt%3E+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#')
1710 iri_to_uri('http://правозащита38.рф/category/news/'),
1711 'http://xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/')
1713 iri_to_uri('http://www.правозащита38.рф/category/news/'),
1714 'http://www.xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/')
1716 iri_to_uri('https://i❤.ws/emojidomain/👍👏🤝💪'),
1717 'https://xn--i-7iq.ws/emojidomain/%F0%9F%91%8D%F0%9F%91%8F%F0%9F%A4%9D%F0%9F%92%AA')
1719 iri_to_uri('http://日本語.jp/'),
1720 'http://xn--wgv71a119e.jp/')
1722 iri_to_uri('http://导航.中国/'),
1723 'http://xn--fet810g.xn--fiqs8s/')
1725 def test_clean_podcast_url(self
):
1726 self
.assertEqual(clean_podcast_url('https://www.podtrac.com/pts/redirect.mp3/chtbl.com/track/5899E/traffic.megaphone.fm/HSW7835899191.mp3'), 'https://traffic.megaphone.fm/HSW7835899191.mp3')
1727 self
.assertEqual(clean_podcast_url('https://play.podtrac.com/npr-344098539/edge1.pod.npr.org/anon.npr-podcasts/podcast/npr/waitwait/2020/10/20201003_waitwait_wwdtmpodcast201003-015621a5-f035-4eca-a9a1-7c118d90bc3c.mp3'), 'https://edge1.pod.npr.org/anon.npr-podcasts/podcast/npr/waitwait/2020/10/20201003_waitwait_wwdtmpodcast201003-015621a5-f035-4eca-a9a1-7c118d90bc3c.mp3')
1729 def test_LazyList(self
):
1730 it
= list(range(10))
1732 self
.assertEqual(list(LazyList(it
)), it
)
1733 self
.assertEqual(LazyList(it
).exhaust(), it
)
1734 self
.assertEqual(LazyList(it
)[5], it
[5])
1736 self
.assertEqual(LazyList(it
)[5:], it
[5:])
1737 self
.assertEqual(LazyList(it
)[:5], it
[:5])
1738 self
.assertEqual(LazyList(it
)[::2], it
[::2])
1739 self
.assertEqual(LazyList(it
)[1::2], it
[1::2])
1740 self
.assertEqual(LazyList(it
)[5::-1], it
[5::-1])
1741 self
.assertEqual(LazyList(it
)[6:2:-2], it
[6:2:-2])
1742 self
.assertEqual(LazyList(it
)[::-1], it
[::-1])
1744 self
.assertTrue(LazyList(it
))
1745 self
.assertFalse(LazyList(range(0)))
1746 self
.assertEqual(len(LazyList(it
)), len(it
))
1747 self
.assertEqual(repr(LazyList(it
)), repr(it
))
1748 self
.assertEqual(str(LazyList(it
)), str(it
))
1750 self
.assertEqual(list(LazyList(it
, reverse
=True)), it
[::-1])
1751 self
.assertEqual(list(reversed(LazyList(it
))[::-1]), it
)
1752 self
.assertEqual(list(reversed(LazyList(it
))[1:3:7]), it
[::-1][1:3:7])
1754 def test_LazyList_laziness(self
):
1756 def test(ll
, idx
, val
, cache
):
1757 self
.assertEqual(ll
[idx
], val
)
1758 self
.assertEqual(getattr(ll
, '_LazyList__cache'), list(cache
))
1760 ll
= LazyList(range(10))
1761 test(ll
, 0, 0, range(1))
1762 test(ll
, 5, 5, range(6))
1763 test(ll
, -3, 7, range(10))
1765 ll
= LazyList(range(10), reverse
=True)
1766 test(ll
, -1, 0, range(1))
1767 test(ll
, 3, 6, range(10))
1769 ll
= LazyList(itertools
.count())
1770 test(ll
, 10, 10, range(11))
1772 test(ll
, -15, 14, range(15))
1774 def test_format_bytes(self
):
1775 self
.assertEqual(format_bytes(0), '0.00B')
1776 self
.assertEqual(format_bytes(1000), '1000.00B')
1777 self
.assertEqual(format_bytes(1024), '1.00KiB')
1778 self
.assertEqual(format_bytes(1024**2), '1.00MiB')
1779 self
.assertEqual(format_bytes(1024**3), '1.00GiB')
1780 self
.assertEqual(format_bytes(1024**4), '1.00TiB')
1781 self
.assertEqual(format_bytes(1024**5), '1.00PiB')
1782 self
.assertEqual(format_bytes(1024**6), '1.00EiB')
1783 self
.assertEqual(format_bytes(1024**7), '1.00ZiB')
1784 self
.assertEqual(format_bytes(1024**8), '1.00YiB')
1785 self
.assertEqual(format_bytes(1024**9), '1024.00YiB')
1787 def test_hide_login_info(self
):
1788 self
.assertEqual(Config
.hide_login_info(['-u', 'foo', '-p', 'bar']),
1789 ['-u', 'PRIVATE', '-p', 'PRIVATE'])
1790 self
.assertEqual(Config
.hide_login_info(['-u']), ['-u'])
1791 self
.assertEqual(Config
.hide_login_info(['-u', 'foo', '-u', 'bar']),
1792 ['-u', 'PRIVATE', '-u', 'PRIVATE'])
1793 self
.assertEqual(Config
.hide_login_info(['--username=foo']),
1794 ['--username=PRIVATE'])
1797 if __name__
== '__main__':