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'), 'gasdgf')
166 self
.assertEqual(sanitize_filename('.gasdgf', is_id
=True), '.gasdgf')
170 for fbc
in forbidden
:
171 self
.assertTrue(fbc
not in sanitize_filename(fc
))
173 def test_sanitize_filename_restricted(self
):
174 self
.assertEqual(sanitize_filename('abc', restricted
=True), 'abc')
175 self
.assertEqual(sanitize_filename('abc_d-e', restricted
=True), 'abc_d-e')
177 self
.assertEqual(sanitize_filename('123', restricted
=True), '123')
179 self
.assertEqual('abc_de', sanitize_filename('abc/de', restricted
=True))
180 self
.assertFalse('/' in sanitize_filename('abc/de///', restricted
=True))
182 self
.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted
=True))
183 self
.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted
=True))
184 self
.assertEqual('yes_no', sanitize_filename('yes? no', restricted
=True))
185 self
.assertEqual('this_-_that', sanitize_filename('this: that', restricted
=True))
187 tests
= 'aäb\u4e2d\u56fd\u7684c'
188 self
.assertEqual(sanitize_filename(tests
, restricted
=True), 'aab_c')
189 self
.assertTrue(sanitize_filename('\xf6', restricted
=True) != '') # No empty filename
191 forbidden
= '"\0\\/&!: \'\t\n()[]{}$;`^,#'
193 for fbc
in forbidden
:
194 self
.assertTrue(fbc
not in sanitize_filename(fc
, restricted
=True))
196 # Handle a common case more neatly
197 self
.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted
=True), 'Song')
198 self
.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted
=True), 'Speech')
199 # .. but make sure the file name is never empty
200 self
.assertTrue(sanitize_filename('-', restricted
=True) != '')
201 self
.assertTrue(sanitize_filename(':', restricted
=True) != '')
203 self
.assertEqual(sanitize_filename(
204 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted
=True),
205 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy')
207 def test_sanitize_ids(self
):
208 self
.assertEqual(sanitize_filename('_n_cd26wFpw', is_id
=True), '_n_cd26wFpw')
209 self
.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id
=True), '_BD_eEpuzXw')
210 self
.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id
=True), 'N0Y__7-UOdI')
212 def test_sanitize_path(self
):
213 if sys
.platform
!= 'win32':
216 self
.assertEqual(sanitize_path('abc'), 'abc')
217 self
.assertEqual(sanitize_path('abc/def'), 'abc\\def')
218 self
.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
219 self
.assertEqual(sanitize_path('abc|def'), 'abc#def')
220 self
.assertEqual(sanitize_path('<>:"|?*'), '#######')
221 self
.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
222 self
.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
224 self
.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
225 self
.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
227 self
.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
228 self
.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
229 self
.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
230 self
.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
233 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
234 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
237 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
238 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
239 self
.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
240 self
.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
241 self
.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
243 self
.assertEqual(sanitize_path('../abc'), '..\\abc')
244 self
.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
245 self
.assertEqual(sanitize_path('./abc'), 'abc')
246 self
.assertEqual(sanitize_path('./../abc'), '..\\abc')
248 def test_sanitize_url(self
):
249 self
.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
250 self
.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
251 self
.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
252 self
.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')
253 self
.assertEqual(sanitize_url('foo bar'), 'foo bar')
255 def test_extract_basic_auth(self
):
256 auth_header
= lambda url
: sanitized_Request(url
).get_header('Authorization')
257 self
.assertFalse(auth_header('http://foo.bar'))
258 self
.assertFalse(auth_header('http://:foo.bar'))
259 self
.assertEqual(auth_header('http://@foo.bar'), 'Basic Og==')
260 self
.assertEqual(auth_header('http://:pass@foo.bar'), 'Basic OnBhc3M=')
261 self
.assertEqual(auth_header('http://user:@foo.bar'), 'Basic dXNlcjo=')
262 self
.assertEqual(auth_header('http://user:pass@foo.bar'), 'Basic dXNlcjpwYXNz')
264 def test_expand_path(self
):
266 return '%{0}%'.format(var
) if sys
.platform
== 'win32' else '${0}'.format(var
)
268 compat_setenv('yt_dlp_EXPATH_PATH', 'expanded')
269 self
.assertEqual(expand_path(env('yt_dlp_EXPATH_PATH')), 'expanded')
270 self
.assertEqual(expand_path(env('HOME')), compat_getenv('HOME'))
271 self
.assertEqual(expand_path('~'), compat_getenv('HOME'))
273 expand_path('~/%s' % env('yt_dlp_EXPATH_PATH')),
274 '%s/expanded' % compat_getenv('HOME'))
276 def test_prepend_extension(self
):
277 self
.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
278 self
.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
279 self
.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
280 self
.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
281 self
.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
282 self
.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
284 def test_replace_extension(self
):
285 self
.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
286 self
.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
287 self
.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
288 self
.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
289 self
.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
290 self
.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
292 def test_subtitles_filename(self
):
293 self
.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt'), 'abc.en.vtt')
294 self
.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt', 'ext'), 'abc.en.vtt')
295 self
.assertEqual(subtitles_filename('abc.unexpected_ext', 'en', 'vtt', 'ext'), 'abc.unexpected_ext.en.vtt')
297 def test_remove_start(self
):
298 self
.assertEqual(remove_start(None, 'A - '), None)
299 self
.assertEqual(remove_start('A - B', 'A - '), 'B')
300 self
.assertEqual(remove_start('B - A', 'A - '), 'B - A')
302 def test_remove_end(self
):
303 self
.assertEqual(remove_end(None, ' - B'), None)
304 self
.assertEqual(remove_end('A - B', ' - B'), 'A')
305 self
.assertEqual(remove_end('B - A', ' - B'), 'B - A')
307 def test_remove_quotes(self
):
308 self
.assertEqual(remove_quotes(None), None)
309 self
.assertEqual(remove_quotes('"'), '"')
310 self
.assertEqual(remove_quotes("'"), "'")
311 self
.assertEqual(remove_quotes(';'), ';')
312 self
.assertEqual(remove_quotes('";'), '";')
313 self
.assertEqual(remove_quotes('""'), '')
314 self
.assertEqual(remove_quotes('";"'), ';')
316 def test_ordered_set(self
):
317 self
.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
318 self
.assertEqual(orderedSet([]), [])
319 self
.assertEqual(orderedSet([1]), [1])
320 # keep the list ordered
321 self
.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
323 def test_unescape_html(self
):
324 self
.assertEqual(unescapeHTML('%20;'), '%20;')
325 self
.assertEqual(unescapeHTML('/'), '/')
326 self
.assertEqual(unescapeHTML('/'), '/')
327 self
.assertEqual(unescapeHTML('é'), 'é')
328 self
.assertEqual(unescapeHTML('�'), '�')
329 self
.assertEqual(unescapeHTML('&a"'), '&a"')
331 self
.assertEqual(unescapeHTML('.''), '.\'')
333 def test_date_from_str(self
):
334 self
.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
335 self
.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
336 self
.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
337 self
.assertEqual(date_from_str('20200229+365day'), date_from_str('20200229+1year'))
338 self
.assertEqual(date_from_str('20210131+28day'), date_from_str('20210131+1month'))
340 def test_datetime_from_str(self
):
341 self
.assertEqual(datetime_from_str('yesterday', precision
='day'), datetime_from_str('now-1day', precision
='auto'))
342 self
.assertEqual(datetime_from_str('now+7day', precision
='day'), datetime_from_str('now+1week', precision
='auto'))
343 self
.assertEqual(datetime_from_str('now+14day', precision
='day'), datetime_from_str('now+2week', precision
='auto'))
344 self
.assertEqual(datetime_from_str('20200229+365day', precision
='day'), datetime_from_str('20200229+1year', precision
='auto'))
345 self
.assertEqual(datetime_from_str('20210131+28day', precision
='day'), datetime_from_str('20210131+1month', precision
='auto'))
346 self
.assertEqual(datetime_from_str('20210131+59day', precision
='day'), datetime_from_str('20210131+2month', precision
='auto'))
347 self
.assertEqual(datetime_from_str('now+1day', precision
='hour'), datetime_from_str('now+24hours', precision
='auto'))
348 self
.assertEqual(datetime_from_str('now+23hours', precision
='hour'), datetime_from_str('now+23hours', precision
='auto'))
350 def test_daterange(self
):
351 _20century
= DateRange("19000101", "20000101")
352 self
.assertFalse("17890714" in _20century
)
353 _ac
= DateRange("00010101")
354 self
.assertTrue("19690721" in _ac
)
355 _firstmilenium
= DateRange(end
="10000101")
356 self
.assertTrue("07110427" in _firstmilenium
)
358 def test_unified_dates(self
):
359 self
.assertEqual(unified_strdate('December 21, 2010'), '20101221')
360 self
.assertEqual(unified_strdate('8/7/2009'), '20090708')
361 self
.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
362 self
.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
363 self
.assertEqual(unified_strdate('1968 12 10'), '19681210')
364 self
.assertEqual(unified_strdate('1968-12-10'), '19681210')
365 self
.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
367 unified_strdate('11/26/2014 11:30:00 AM PST', day_first
=False),
370 unified_strdate('2/2/2015 6:47:40 PM', day_first
=False),
372 self
.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
373 self
.assertEqual(unified_strdate('25-09-2014'), '20140925')
374 self
.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
375 self
.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
376 self
.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
377 self
.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
378 self
.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
379 self
.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
380 self
.assertEqual(unified_strdate('November 3rd, 2019'), '20191103')
381 self
.assertEqual(unified_strdate('October 23rd, 2005'), '20051023')
383 def test_unified_timestamps(self
):
384 self
.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
385 self
.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
386 self
.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
387 self
.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
388 self
.assertEqual(unified_timestamp('1968 12 10'), -33436800)
389 self
.assertEqual(unified_timestamp('1968-12-10'), -33436800)
390 self
.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
392 unified_timestamp('11/26/2014 11:30:00 AM PST', day_first
=False),
395 unified_timestamp('2/2/2015 6:47:40 PM', day_first
=False),
397 self
.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
398 self
.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
399 self
.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
400 self
.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
401 self
.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
402 self
.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
403 self
.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
404 self
.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
405 self
.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
406 self
.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363)
408 def test_determine_ext(self
):
409 self
.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
410 self
.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
411 self
.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
412 self
.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
413 self
.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
414 self
.assertEqual(determine_ext('foobar', None), None)
416 def test_find_xpath_attr(self
):
424 doc
= compat_etree_fromstring(testxml
)
426 self
.assertEqual(find_xpath_attr(doc
, './/fourohfour', 'n'), None)
427 self
.assertEqual(find_xpath_attr(doc
, './/fourohfour', 'n', 'v'), None)
428 self
.assertEqual(find_xpath_attr(doc
, './/node', 'n'), None)
429 self
.assertEqual(find_xpath_attr(doc
, './/node', 'n', 'v'), None)
430 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x'), doc
[1])
431 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x', 'a'), doc
[1])
432 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x', 'b'), doc
[3])
433 self
.assertEqual(find_xpath_attr(doc
, './/node', 'y'), doc
[2])
434 self
.assertEqual(find_xpath_attr(doc
, './/node', 'y', 'c'), doc
[2])
435 self
.assertEqual(find_xpath_attr(doc
, './/node', 'y', 'd'), doc
[3])
436 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x', ''), doc
[4])
438 def test_xpath_with_ns(self
):
439 testxml
= '''<root xmlns:media="http://example.com/">
441 <media:author>The Author</media:author>
442 <url>http://server.com/download.mp3</url>
445 doc
= compat_etree_fromstring(testxml
)
446 find
= lambda p
: doc
.find(xpath_with_ns(p
, {'media': 'http://example.com/'}
))
447 self
.assertTrue(find('media:song') is not None)
448 self
.assertEqual(find('media:song/media:author').text
, 'The Author')
449 self
.assertEqual(find('media:song/url').text
, 'http://server.com/download.mp3')
451 def test_xpath_element(self
):
452 doc
= xml
.etree
.ElementTree
.Element('root')
453 div
= xml
.etree
.ElementTree
.SubElement(doc
, 'div')
454 p
= xml
.etree
.ElementTree
.SubElement(div
, 'p')
456 self
.assertEqual(xpath_element(doc
, 'div/p'), p
)
457 self
.assertEqual(xpath_element(doc
, ['div/p']), p
)
458 self
.assertEqual(xpath_element(doc
, ['div/bar', 'div/p']), p
)
459 self
.assertEqual(xpath_element(doc
, 'div/bar', default
='default'), 'default')
460 self
.assertEqual(xpath_element(doc
, ['div/bar'], default
='default'), 'default')
461 self
.assertTrue(xpath_element(doc
, 'div/bar') is None)
462 self
.assertTrue(xpath_element(doc
, ['div/bar']) is None)
463 self
.assertTrue(xpath_element(doc
, ['div/bar'], 'div/baz') is None)
464 self
.assertRaises(ExtractorError
, xpath_element
, doc
, 'div/bar', fatal
=True)
465 self
.assertRaises(ExtractorError
, xpath_element
, doc
, ['div/bar'], fatal
=True)
466 self
.assertRaises(ExtractorError
, xpath_element
, doc
, ['div/bar', 'div/baz'], fatal
=True)
468 def test_xpath_text(self
):
474 doc
= compat_etree_fromstring(testxml
)
475 self
.assertEqual(xpath_text(doc
, 'div/p'), 'Foo')
476 self
.assertEqual(xpath_text(doc
, 'div/bar', default
='default'), 'default')
477 self
.assertTrue(xpath_text(doc
, 'div/bar') is None)
478 self
.assertRaises(ExtractorError
, xpath_text
, doc
, 'div/bar', fatal
=True)
480 def test_xpath_attr(self
):
486 doc
= compat_etree_fromstring(testxml
)
487 self
.assertEqual(xpath_attr(doc
, 'div/p', 'x'), 'a')
488 self
.assertEqual(xpath_attr(doc
, 'div/bar', 'x'), None)
489 self
.assertEqual(xpath_attr(doc
, 'div/p', 'y'), None)
490 self
.assertEqual(xpath_attr(doc
, 'div/bar', 'x', default
='default'), 'default')
491 self
.assertEqual(xpath_attr(doc
, 'div/p', 'y', default
='default'), 'default')
492 self
.assertRaises(ExtractorError
, xpath_attr
, doc
, 'div/bar', 'x', fatal
=True)
493 self
.assertRaises(ExtractorError
, xpath_attr
, doc
, 'div/p', 'y', fatal
=True)
495 def test_smuggle_url(self
):
496 data
= {"ö": "ö", "abc": [3]}
497 url
= 'https://foo.bar/baz?x=y#a'
498 smug_url
= smuggle_url(url
, data
)
499 unsmug_url
, unsmug_data
= unsmuggle_url(smug_url
)
500 self
.assertEqual(url
, unsmug_url
)
501 self
.assertEqual(data
, unsmug_data
)
503 res_url
, res_data
= unsmuggle_url(url
)
504 self
.assertEqual(res_url
, url
)
505 self
.assertEqual(res_data
, None)
507 smug_url
= smuggle_url(url
, {'a': 'b'}
)
508 smug_smug_url
= smuggle_url(smug_url
, {'c': 'd'}
)
509 res_url
, res_data
= unsmuggle_url(smug_smug_url
)
510 self
.assertEqual(res_url
, url
)
511 self
.assertEqual(res_data
, {'a': 'b', 'c': 'd'}
)
513 def test_shell_quote(self
):
514 args
= ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
517 """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name
!= 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
519 def test_float_or_none(self
):
520 self
.assertEqual(float_or_none('42.42'), 42.42)
521 self
.assertEqual(float_or_none('42'), 42.0)
522 self
.assertEqual(float_or_none(''), None)
523 self
.assertEqual(float_or_none(None), None)
524 self
.assertEqual(float_or_none([]), None)
525 self
.assertEqual(float_or_none(set()), None)
527 def test_int_or_none(self
):
528 self
.assertEqual(int_or_none('42'), 42)
529 self
.assertEqual(int_or_none(''), None)
530 self
.assertEqual(int_or_none(None), None)
531 self
.assertEqual(int_or_none([]), None)
532 self
.assertEqual(int_or_none(set()), None)
534 def test_str_to_int(self
):
535 self
.assertEqual(str_to_int('123,456'), 123456)
536 self
.assertEqual(str_to_int('123.456'), 123456)
537 self
.assertEqual(str_to_int(523), 523)
538 # Python 3 has no long
539 if sys
.version_info
< (3, 0):
540 eval('self.assertEqual(str_to_int(123456L), 123456)')
541 self
.assertEqual(str_to_int('noninteger'), None)
542 self
.assertEqual(str_to_int([]), None)
544 def test_url_basename(self
):
545 self
.assertEqual(url_basename('http://foo.de/'), '')
546 self
.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
547 self
.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
548 self
.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
549 self
.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
551 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
554 def test_base_url(self
):
555 self
.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
556 self
.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
557 self
.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
558 self
.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
559 self
.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
561 def test_urljoin(self
):
562 self
.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
563 self
.assertEqual(urljoin(b
'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
564 self
.assertEqual(urljoin('http://foo.de/', b
'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
565 self
.assertEqual(urljoin(b
'http://foo.de/', b
'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
566 self
.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
567 self
.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
568 self
.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://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/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
571 self
.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
572 self
.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
573 self
.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
574 self
.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
575 self
.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
576 self
.assertEqual(urljoin('http://foo.de/', None), None)
577 self
.assertEqual(urljoin('http://foo.de/', ''), None)
578 self
.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
579 self
.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
580 self
.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de')
581 self
.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de')
583 def test_url_or_none(self
):
584 self
.assertEqual(url_or_none(None), None)
585 self
.assertEqual(url_or_none(''), None)
586 self
.assertEqual(url_or_none('foo'), None)
587 self
.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
588 self
.assertEqual(url_or_none('https://foo.de'), 'https://foo.de')
589 self
.assertEqual(url_or_none('http$://foo.de'), None)
590 self
.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
591 self
.assertEqual(url_or_none('//foo.de'), '//foo.de')
592 self
.assertEqual(url_or_none('s3://foo.de'), None)
593 self
.assertEqual(url_or_none('rtmpte://foo.de'), 'rtmpte://foo.de')
594 self
.assertEqual(url_or_none('mms://foo.de'), 'mms://foo.de')
595 self
.assertEqual(url_or_none('rtspu://foo.de'), 'rtspu://foo.de')
596 self
.assertEqual(url_or_none('ftps://foo.de'), 'ftps://foo.de')
598 def test_parse_age_limit(self
):
599 self
.assertEqual(parse_age_limit(None), None)
600 self
.assertEqual(parse_age_limit(False), None)
601 self
.assertEqual(parse_age_limit('invalid'), None)
602 self
.assertEqual(parse_age_limit(0), 0)
603 self
.assertEqual(parse_age_limit(18), 18)
604 self
.assertEqual(parse_age_limit(21), 21)
605 self
.assertEqual(parse_age_limit(22), None)
606 self
.assertEqual(parse_age_limit('18'), 18)
607 self
.assertEqual(parse_age_limit('18+'), 18)
608 self
.assertEqual(parse_age_limit('PG-13'), 13)
609 self
.assertEqual(parse_age_limit('TV-14'), 14)
610 self
.assertEqual(parse_age_limit('TV-MA'), 17)
611 self
.assertEqual(parse_age_limit('TV14'), 14)
612 self
.assertEqual(parse_age_limit('TV_G'), 0)
614 def test_parse_duration(self
):
615 self
.assertEqual(parse_duration(None), None)
616 self
.assertEqual(parse_duration(False), None)
617 self
.assertEqual(parse_duration('invalid'), None)
618 self
.assertEqual(parse_duration('1'), 1)
619 self
.assertEqual(parse_duration('1337:12'), 80232)
620 self
.assertEqual(parse_duration('9:12:43'), 33163)
621 self
.assertEqual(parse_duration('12:00'), 720)
622 self
.assertEqual(parse_duration('00:01:01'), 61)
623 self
.assertEqual(parse_duration('x:y'), None)
624 self
.assertEqual(parse_duration('3h11m53s'), 11513)
625 self
.assertEqual(parse_duration('3h 11m 53s'), 11513)
626 self
.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
627 self
.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
628 self
.assertEqual(parse_duration('62m45s'), 3765)
629 self
.assertEqual(parse_duration('6m59s'), 419)
630 self
.assertEqual(parse_duration('49s'), 49)
631 self
.assertEqual(parse_duration('0h0m0s'), 0)
632 self
.assertEqual(parse_duration('0m0s'), 0)
633 self
.assertEqual(parse_duration('0s'), 0)
634 self
.assertEqual(parse_duration('01:02:03.05'), 3723.05)
635 self
.assertEqual(parse_duration('T30M38S'), 1838)
636 self
.assertEqual(parse_duration('5 s'), 5)
637 self
.assertEqual(parse_duration('3 min'), 180)
638 self
.assertEqual(parse_duration('2.5 hours'), 9000)
639 self
.assertEqual(parse_duration('02:03:04'), 7384)
640 self
.assertEqual(parse_duration('01:02:03:04'), 93784)
641 self
.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
642 self
.assertEqual(parse_duration('87 Min.'), 5220)
643 self
.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
644 self
.assertEqual(parse_duration('PT00H03M30SZ'), 210)
645 self
.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)
646 self
.assertEqual(parse_duration('01:02:03:050'), 3723.05)
647 self
.assertEqual(parse_duration('103:050'), 103.05)
649 def test_fix_xml_ampersands(self
):
651 fix_xml_ampersands('"&x=y&z=a'), '"&x=y&z=a')
653 fix_xml_ampersands('"&x=y&wrong;&z=a'),
654 '"&x=y&wrong;&z=a')
656 fix_xml_ampersands('&'><"'),
657 '&'><"')
659 fix_xml_ampersands('Ӓ᪼'), 'Ӓ᪼')
660 self
.assertEqual(fix_xml_ampersands('&#&#'), '&#&#')
662 def test_paged_list(self
):
663 def testPL(size
, pagesize
, sliceargs
, expected
):
664 def get_page(pagenum
):
665 firstid
= pagenum
* pagesize
666 upto
= min(size
, pagenum
* pagesize
+ pagesize
)
667 for i
in range(firstid
, upto
):
670 pl
= OnDemandPagedList(get_page
, pagesize
)
671 got
= pl
.getslice(*sliceargs
)
672 self
.assertEqual(got
, expected
)
674 iapl
= InAdvancePagedList(get_page
, size
// pagesize
+ 1, pagesize
)
675 got
= iapl
.getslice(*sliceargs
)
676 self
.assertEqual(got
, expected
)
678 testPL(5, 2, (), [0, 1, 2, 3, 4])
679 testPL(5, 2, (1,), [1, 2, 3, 4])
680 testPL(5, 2, (2,), [2, 3, 4])
681 testPL(5, 2, (4,), [4])
682 testPL(5, 2, (0, 3), [0, 1, 2])
683 testPL(5, 2, (1, 4), [1, 2, 3])
684 testPL(5, 2, (2, 99), [2, 3, 4])
685 testPL(5, 2, (20, 99), [])
687 def test_read_batch_urls(self
):
688 f
= io
.StringIO('''\xef\xbb\xbf foo
691 # More after this line\r
694 self
.assertEqual(read_batch_urls(f
), ['foo', 'bar', 'baz', 'bam'])
696 def test_urlencode_postdata(self
):
697 data
= urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'}
)
698 self
.assertTrue(isinstance(data
, bytes))
700 def test_update_url_query(self
):
701 self
.assertEqual(parse_qs(update_url_query(
702 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']}
)),
703 parse_qs('http://example.com/path?quality=HD&format=mp4'))
704 self
.assertEqual(parse_qs(update_url_query(
705 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']}
)),
706 parse_qs('http://example.com/path?system=LINUX&system=WINDOWS'))
707 self
.assertEqual(parse_qs(update_url_query(
708 'http://example.com/path', {'fields': 'id,formats,subtitles'}
)),
709 parse_qs('http://example.com/path?fields=id,formats,subtitles'))
710 self
.assertEqual(parse_qs(update_url_query(
711 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')}
)),
712 parse_qs('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
713 self
.assertEqual(parse_qs(update_url_query(
714 'http://example.com/path?manifest=f4m', {'manifest': []}
)),
715 parse_qs('http://example.com/path'))
716 self
.assertEqual(parse_qs(update_url_query(
717 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'}
)),
718 parse_qs('http://example.com/path?system=LINUX'))
719 self
.assertEqual(parse_qs(update_url_query(
720 'http://example.com/path', {'fields': b'id,formats,subtitles'}
)),
721 parse_qs('http://example.com/path?fields=id,formats,subtitles'))
722 self
.assertEqual(parse_qs(update_url_query(
723 'http://example.com/path', {'width': 1080, 'height': 720}
)),
724 parse_qs('http://example.com/path?width=1080&height=720'))
725 self
.assertEqual(parse_qs(update_url_query(
726 'http://example.com/path', {'bitrate': 5020.43}
)),
727 parse_qs('http://example.com/path?bitrate=5020.43'))
728 self
.assertEqual(parse_qs(update_url_query(
729 'http://example.com/path', {'test': '第二行тест'}
)),
730 parse_qs('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
732 def test_multipart_encode(self
):
734 multipart_encode({b'field': b'value'}
, boundary
='AAAAAA')[0],
735 b
'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
737 multipart_encode({'欄位'.encode('utf-8'): '值'.encode('utf-8')}
, boundary
='AAAAAA')[0],
738 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')
740 ValueError, multipart_encode
, {b'field': b'value'}
, boundary
='value')
742 def test_dict_get(self
):
750 d
= FALSE_VALUES
.copy()
752 self
.assertEqual(dict_get(d
, 'a'), 42)
753 self
.assertEqual(dict_get(d
, 'b'), None)
754 self
.assertEqual(dict_get(d
, 'b', 42), 42)
755 self
.assertEqual(dict_get(d
, ('a', )), 42)
756 self
.assertEqual(dict_get(d
, ('b', 'a', )), 42)
757 self
.assertEqual(dict_get(d
, ('b', 'c', 'a', 'd', )), 42)
758 self
.assertEqual(dict_get(d
, ('b', 'c', )), None)
759 self
.assertEqual(dict_get(d
, ('b', 'c', ), 42), 42)
760 for key
, false_value
in FALSE_VALUES
.items():
761 self
.assertEqual(dict_get(d
, ('b', 'c', key
, )), None)
762 self
.assertEqual(dict_get(d
, ('b', 'c', key
, ), skip_false_values
=False), false_value
)
764 def test_merge_dicts(self
):
765 self
.assertEqual(merge_dicts({'a': 1}
, {'b': 2}
), {'a': 1, 'b': 2}
)
766 self
.assertEqual(merge_dicts({'a': 1}
, {'a': 2}
), {'a': 1}
)
767 self
.assertEqual(merge_dicts({'a': 1}
, {'a': None}
), {'a': 1}
)
768 self
.assertEqual(merge_dicts({'a': 1}
, {'a': ''}
), {'a': 1}
)
769 self
.assertEqual(merge_dicts({'a': 1}
, {}), {'a': 1}
)
770 self
.assertEqual(merge_dicts({'a': None}
, {'a': 1}
), {'a': 1}
)
771 self
.assertEqual(merge_dicts({'a': ''}
, {'a': 1}
), {'a': ''}
)
772 self
.assertEqual(merge_dicts({'a': ''}
, {'a': 'abc'}
), {'a': 'abc'}
)
773 self
.assertEqual(merge_dicts({'a': None}
, {'a': ''}
, {'a': 'abc'}
), {'a': 'abc'}
)
775 def test_encode_compat_str(self
):
776 self
.assertEqual(encode_compat_str(b
'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
777 self
.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
779 def test_parse_iso8601(self
):
780 self
.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
781 self
.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
782 self
.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
783 self
.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
784 self
.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
785 self
.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
787 def test_strip_jsonp(self
):
788 stripped
= strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
789 d
= json
.loads(stripped
)
790 self
.assertEqual(d
, [{"id": "532cb", "x": 3}
])
792 stripped
= strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
793 d
= json
.loads(stripped
)
794 self
.assertEqual(d
, {'STATUS': 'OK'}
)
796 stripped
= strip_jsonp('ps.embedHandler({"status": "success"});')
797 d
= json
.loads(stripped
)
798 self
.assertEqual(d
, {'status': 'success'}
)
800 stripped
= strip_jsonp('window.cb && window.cb({"status": "success"});')
801 d
= json
.loads(stripped
)
802 self
.assertEqual(d
, {'status': 'success'}
)
804 stripped
= strip_jsonp('window.cb && cb({"status": "success"});')
805 d
= json
.loads(stripped
)
806 self
.assertEqual(d
, {'status': 'success'}
)
808 stripped
= strip_jsonp('({"status": "success"});')
809 d
= json
.loads(stripped
)
810 self
.assertEqual(d
, {'status': 'success'}
)
812 def test_strip_or_none(self
):
813 self
.assertEqual(strip_or_none(' abc'), 'abc')
814 self
.assertEqual(strip_or_none('abc '), 'abc')
815 self
.assertEqual(strip_or_none(' abc '), 'abc')
816 self
.assertEqual(strip_or_none('\tabc\t'), 'abc')
817 self
.assertEqual(strip_or_none('\n\tabc\n\t'), 'abc')
818 self
.assertEqual(strip_or_none('abc'), 'abc')
819 self
.assertEqual(strip_or_none(''), '')
820 self
.assertEqual(strip_or_none(None), None)
821 self
.assertEqual(strip_or_none(42), None)
822 self
.assertEqual(strip_or_none([]), None)
824 def test_uppercase_escape(self
):
825 self
.assertEqual(uppercase_escape('aä'), 'aä')
826 self
.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
828 def test_lowercase_escape(self
):
829 self
.assertEqual(lowercase_escape('aä'), 'aä')
830 self
.assertEqual(lowercase_escape('\\u0026'), '&')
832 def test_limit_length(self
):
833 self
.assertEqual(limit_length(None, 12), None)
834 self
.assertEqual(limit_length('foo', 12), 'foo')
836 limit_length('foo bar baz asd', 12).startswith('foo bar'))
837 self
.assertTrue('...' in limit_length('foo bar baz asd', 12))
839 def test_mimetype2ext(self
):
840 self
.assertEqual(mimetype2ext(None), None)
841 self
.assertEqual(mimetype2ext('video/x-flv'), 'flv')
842 self
.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
843 self
.assertEqual(mimetype2ext('text/vtt'), 'vtt')
844 self
.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
845 self
.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
846 self
.assertEqual(mimetype2ext('audio/x-wav'), 'wav')
847 self
.assertEqual(mimetype2ext('audio/x-wav;codec=pcm'), 'wav')
849 def test_month_by_name(self
):
850 self
.assertEqual(month_by_name(None), None)
851 self
.assertEqual(month_by_name('December', 'en'), 12)
852 self
.assertEqual(month_by_name('décembre', 'fr'), 12)
853 self
.assertEqual(month_by_name('December'), 12)
854 self
.assertEqual(month_by_name('décembre'), None)
855 self
.assertEqual(month_by_name('Unknown', 'unknown'), None)
857 def test_parse_codecs(self
):
858 self
.assertEqual(parse_codecs(''), {})
859 self
.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
860 'vcodec': 'avc1.77.30',
861 'acodec': 'mp4a.40.2',
862 'dynamic_range': None,
864 self
.assertEqual(parse_codecs('mp4a.40.2'), {
866 'acodec': 'mp4a.40.2',
867 'dynamic_range': None,
869 self
.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
870 'vcodec': 'avc1.42001e',
871 'acodec': 'mp4a.40.5',
872 'dynamic_range': None,
874 self
.assertEqual(parse_codecs('avc3.640028'), {
875 'vcodec': 'avc3.640028',
877 'dynamic_range': None,
879 self
.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
882 'dynamic_range': None,
884 self
.assertEqual(parse_codecs('av01.0.05M.08'), {
885 'vcodec': 'av01.0.05M.08',
887 'dynamic_range': None,
889 self
.assertEqual(parse_codecs('vp9.2'), {
892 'dynamic_range': 'HDR10',
894 self
.assertEqual(parse_codecs('av01.0.12M.10.0.110.09.16.09.0'), {
895 'vcodec': 'av01.0.12M.10',
897 'dynamic_range': 'HDR10',
899 self
.assertEqual(parse_codecs('dvhe'), {
902 'dynamic_range': 'DV',
904 self
.assertEqual(parse_codecs('theora, vorbis'), {
907 'dynamic_range': None,
909 self
.assertEqual(parse_codecs('unknownvcodec, unknownacodec'), {
910 'vcodec': 'unknownvcodec',
911 'acodec': 'unknownacodec',
913 self
.assertEqual(parse_codecs('unknown'), {})
915 def test_escape_rfc3986(self
):
916 reserved
= "!*'();:@&=+$,/?#[]"
917 unreserved
= 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
918 self
.assertEqual(escape_rfc3986(reserved
), reserved
)
919 self
.assertEqual(escape_rfc3986(unreserved
), unreserved
)
920 self
.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
921 self
.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
922 self
.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
923 self
.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
925 def test_escape_url(self
):
927 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
928 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
931 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
932 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
935 escape_url('http://тест.рф/фрагмент'),
936 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
939 escape_url('http://тест.рф/абв?абв=абв#абв'),
940 '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'
942 self
.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
944 def test_js_to_json_realworld(self
):
946 'clip':{'provider':'pseudo'}
948 self
.assertEqual(js_to_json(inp
), '''{
949 "clip":{"provider":"pseudo"}
951 json
.loads(js_to_json(inp
))
954 'playlist':[{'controls':{'all':null}}]
956 self
.assertEqual(js_to_json(inp
), '''{
957 "playlist":[{"controls":{"all":null}}]
960 inp
= '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
961 self
.assertEqual(js_to_json(inp
), '''"The CW's 'Crazy Ex-Girlfriend'"''')
963 inp
= '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
964 json_code
= js_to_json(inp
)
965 self
.assertEqual(json
.loads(json_code
), json
.loads(inp
))
968 0:{src:'skipped', type: 'application/dash+xml'},
969 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
971 self
.assertEqual(js_to_json(inp
), '''{
972 "0":{"src":"skipped", "type": "application/dash+xml"},
973 "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
976 inp
= '''{"foo":101}'''
977 self
.assertEqual(js_to_json(inp
), '''{"foo":101}''')
979 inp
= '''{"duration": "00:01:07"}'''
980 self
.assertEqual(js_to_json(inp
), '''{"duration": "00:01:07"}''')
982 inp
= '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}'''
983 self
.assertEqual(js_to_json(inp
), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''')
985 def test_js_to_json_edgecases(self
):
986 on
= js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
987 self
.assertEqual(json
.loads(on
), {"abc_def": "1'\\2\\'3\"4"}
)
989 on
= js_to_json('{"abc": true}')
990 self
.assertEqual(json
.loads(on
), {'abc': True}
)
992 # Ignore JavaScript code as well
999 self
.assertEqual(d
['x'], 1)
1000 self
.assertEqual(d
['y'], 'a')
1002 # Just drop ! prefix for now though this results in a wrong value
1003 on
= js_to_json('''{
1013 self
.assertEqual(json
.loads(on
), {
1024 on
= js_to_json('["abc", "def",]')
1025 self
.assertEqual(json
.loads(on
), ['abc', 'def'])
1027 on
= js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
1028 self
.assertEqual(json
.loads(on
), ['abc', 'def'])
1030 on
= js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
1031 self
.assertEqual(json
.loads(on
), ['abc', 'def'])
1033 on
= js_to_json('{"abc": "def",}')
1034 self
.assertEqual(json
.loads(on
), {'abc': 'def'}
)
1036 on
= js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
1037 self
.assertEqual(json
.loads(on
), {'abc': 'def'}
)
1039 on
= js_to_json('{ 0: /* " \n */ ",]" , }')
1040 self
.assertEqual(json
.loads(on
), {'0': ',]'}
)
1042 on
= js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
1043 self
.assertEqual(json
.loads(on
), {'0': ',]'}
)
1045 on
= js_to_json('{ 0: // comment\n1 }')
1046 self
.assertEqual(json
.loads(on
), {'0': 1}
)
1048 on
= js_to_json(r
'["<p>x<\/p>"]')
1049 self
.assertEqual(json
.loads(on
), ['<p>x</p>'])
1051 on
= js_to_json(r
'["\xaa"]')
1052 self
.assertEqual(json
.loads(on
), ['\u00aa'])
1054 on
= js_to_json("['a\\\nb']")
1055 self
.assertEqual(json
.loads(on
), ['ab'])
1057 on
= js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
1058 self
.assertEqual(json
.loads(on
), ['ab'])
1060 on
= js_to_json('{0xff:0xff}')
1061 self
.assertEqual(json
.loads(on
), {'255': 255}
)
1063 on
= js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
1064 self
.assertEqual(json
.loads(on
), {'255': 255}
)
1066 on
= js_to_json('{077:077}')
1067 self
.assertEqual(json
.loads(on
), {'63': 63}
)
1069 on
= js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
1070 self
.assertEqual(json
.loads(on
), {'63': 63}
)
1072 on
= js_to_json('{42:42}')
1073 self
.assertEqual(json
.loads(on
), {'42': 42}
)
1075 on
= js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
1076 self
.assertEqual(json
.loads(on
), {'42': 42}
)
1078 on
= js_to_json('{42:4.2e1}')
1079 self
.assertEqual(json
.loads(on
), {'42': 42.0}
)
1081 on
= js_to_json('{ "0x40": "0x40" }')
1082 self
.assertEqual(json
.loads(on
), {'0x40': '0x40'}
)
1084 on
= js_to_json('{ "040": "040" }')
1085 self
.assertEqual(json
.loads(on
), {'040': '040'}
)
1087 on
= js_to_json('[1,//{},\n2]')
1088 self
.assertEqual(json
.loads(on
), [1, 2])
1090 def test_js_to_json_malformed(self
):
1091 self
.assertEqual(js_to_json('42a1'), '42"a1"')
1092 self
.assertEqual(js_to_json('42a-1'), '42"a"-1')
1094 def test_extract_attributes(self
):
1095 self
.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}
)
1096 self
.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'}
)
1097 self
.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'}
)
1098 self
.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"}
)
1099 self
.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'}
)
1100 self
.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}
)
1101 self
.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}
)
1102 self
.assertEqual(extract_attributes('<e x="&">'), {'x': '&'}
) # XML
1103 self
.assertEqual(extract_attributes('<e x=""">'), {'x': '"'}
)
1104 self
.assertEqual(extract_attributes('<e x="£">'), {'x': '£'}
) # HTML 3.2
1105 self
.assertEqual(extract_attributes('<e x="λ">'), {'x': 'λ'}
) # HTML 4.0
1106 self
.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'}
)
1107 self
.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"}
)
1108 self
.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'}
)
1109 self
.assertEqual(extract_attributes('<e x >'), {'x': None}
)
1110 self
.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None}
)
1111 self
.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'}
)
1112 self
.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'}
)
1113 self
.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'}
)
1114 self
.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'}
)
1115 self
.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'}
)
1116 self
.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'}
)
1117 self
.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}
) # Names lowercased
1118 self
.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'}
)
1119 self
.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'}
)
1120 self
.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'}
)
1121 self
.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'}
)
1122 self
.assertEqual(extract_attributes('<e x="décomposé">'), {'x': 'décompose\u0301'}
)
1123 # "Narrow" Python builds don't support unicode code points outside BMP.
1126 supports_outside_bmp
= True
1128 supports_outside_bmp
= False
1129 if supports_outside_bmp
:
1130 self
.assertEqual(extract_attributes('<e x="Smile 😀!">'), {'x': 'Smile \U0001f600!'}
)
1131 # Malformed HTML should not break attributes extraction on older Python
1132 self
.assertEqual(extract_attributes('<mal"formed/>'), {})
1134 def test_clean_html(self
):
1135 self
.assertEqual(clean_html('a:\nb'), 'a: b')
1136 self
.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
1137 self
.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
1139 def test_intlist_to_bytes(self
):
1141 intlist_to_bytes([0, 1, 127, 128, 255]),
1142 b
'\x00\x01\x7f\x80\xff')
1144 def test_args_to_str(self
):
1146 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
1147 'foo ba/r -baz \'2 be\' \'\'' if compat_os_name
!= 'nt' else 'foo ba/r -baz "2 be" ""'
1150 def test_parse_filesize(self
):
1151 self
.assertEqual(parse_filesize(None), None)
1152 self
.assertEqual(parse_filesize(''), None)
1153 self
.assertEqual(parse_filesize('91 B'), 91)
1154 self
.assertEqual(parse_filesize('foobar'), None)
1155 self
.assertEqual(parse_filesize('2 MiB'), 2097152)
1156 self
.assertEqual(parse_filesize('5 GB'), 5000000000)
1157 self
.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
1158 self
.assertEqual(parse_filesize('1.2tb'), 1200000000000)
1159 self
.assertEqual(parse_filesize('1,24 KB'), 1240)
1160 self
.assertEqual(parse_filesize('1,24 kb'), 1240)
1161 self
.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
1163 def test_parse_count(self
):
1164 self
.assertEqual(parse_count(None), None)
1165 self
.assertEqual(parse_count(''), None)
1166 self
.assertEqual(parse_count('0'), 0)
1167 self
.assertEqual(parse_count('1000'), 1000)
1168 self
.assertEqual(parse_count('1.000'), 1000)
1169 self
.assertEqual(parse_count('1.1k'), 1100)
1170 self
.assertEqual(parse_count('1.1 k'), 1100)
1171 self
.assertEqual(parse_count('1,1 k'), 1100)
1172 self
.assertEqual(parse_count('1.1kk'), 1100000)
1173 self
.assertEqual(parse_count('1.1kk '), 1100000)
1174 self
.assertEqual(parse_count('1,1kk'), 1100000)
1175 self
.assertEqual(parse_count('100 views'), 100)
1176 self
.assertEqual(parse_count('1,100 views'), 1100)
1177 self
.assertEqual(parse_count('1.1kk views'), 1100000)
1178 self
.assertEqual(parse_count('10M views'), 10000000)
1179 self
.assertEqual(parse_count('has 10M views'), 10000000)
1181 def test_parse_resolution(self
):
1182 self
.assertEqual(parse_resolution(None), {})
1183 self
.assertEqual(parse_resolution(''), {})
1184 self
.assertEqual(parse_resolution(' 1920x1080'), {'width': 1920, 'height': 1080}
)
1185 self
.assertEqual(parse_resolution('1920×1080 '), {'width': 1920, 'height': 1080}
)
1186 self
.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080}
)
1187 self
.assertEqual(parse_resolution('720p'), {'height': 720}
)
1188 self
.assertEqual(parse_resolution('4k'), {'height': 2160}
)
1189 self
.assertEqual(parse_resolution('8K'), {'height': 4320}
)
1190 self
.assertEqual(parse_resolution('pre_1920x1080_post'), {'width': 1920, 'height': 1080}
)
1191 self
.assertEqual(parse_resolution('ep1x2'), {})
1192 self
.assertEqual(parse_resolution('1920, 1080'), {'width': 1920, 'height': 1080}
)
1194 def test_parse_bitrate(self
):
1195 self
.assertEqual(parse_bitrate(None), None)
1196 self
.assertEqual(parse_bitrate(''), None)
1197 self
.assertEqual(parse_bitrate('300kbps'), 300)
1198 self
.assertEqual(parse_bitrate('1500kbps'), 1500)
1199 self
.assertEqual(parse_bitrate('300 kbps'), 300)
1201 def test_version_tuple(self
):
1202 self
.assertEqual(version_tuple('1'), (1,))
1203 self
.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
1204 self
.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
1206 def test_detect_exe_version(self
):
1207 self
.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
1208 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
1209 configuration: --prefix=/usr --extra-'''), '1.2.1')
1210 self
.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
1211 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
1212 self
.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
1213 Trying to open render node...
1214 Success at /dev/dri/renderD128.
1215 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
1217 def test_age_restricted(self
):
1218 self
.assertFalse(age_restricted(None, 10)) # unrestricted content
1219 self
.assertFalse(age_restricted(1, None)) # unrestricted policy
1220 self
.assertFalse(age_restricted(8, 10))
1221 self
.assertTrue(age_restricted(18, 14))
1222 self
.assertFalse(age_restricted(18, 18))
1224 def test_is_html(self
):
1225 self
.assertFalse(is_html(b
'\x49\x44\x43<html'))
1226 self
.assertTrue(is_html(b
'<!DOCTYPE foo>\xaaa'))
1227 self
.assertTrue(is_html( # UTF-8 with BOM
1228 b
'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
1229 self
.assertTrue(is_html( # UTF-16-LE
1230 b
'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
1232 self
.assertTrue(is_html( # UTF-16-BE
1233 b
'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
1235 self
.assertTrue(is_html( # UTF-32-BE
1236 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'))
1237 self
.assertTrue(is_html( # UTF-32-LE
1238 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'))
1240 def test_render_table(self
):
1243 ['a', 'empty', 'bcd'],
1244 [[123, '', 4], [9999, '', 51]]),
1251 ['a', 'empty', 'bcd'],
1252 [[123, '', 4], [9999, '', 51]],
1261 [['1\t23', 4], ['\t9999', 51]]),
1269 [[123, 4], [9999, 51]],
1279 [[123, 4], [9999, 51]],
1280 delim
='-', extra_gap
=2),
1286 def test_match_str(self
):
1288 self
.assertFalse(match_str('xy', {'x': 1200}
))
1289 self
.assertTrue(match_str('!xy', {'x': 1200}
))
1290 self
.assertTrue(match_str('x', {'x': 1200}
))
1291 self
.assertFalse(match_str('!x', {'x': 1200}
))
1292 self
.assertTrue(match_str('x', {'x': 0}
))
1293 self
.assertTrue(match_str('is_live', {'is_live': True}
))
1294 self
.assertFalse(match_str('is_live', {'is_live': False}
))
1295 self
.assertFalse(match_str('is_live', {'is_live': None}
))
1296 self
.assertFalse(match_str('is_live', {}))
1297 self
.assertFalse(match_str('!is_live', {'is_live': True}
))
1298 self
.assertTrue(match_str('!is_live', {'is_live': False}
))
1299 self
.assertTrue(match_str('!is_live', {'is_live': None}
))
1300 self
.assertTrue(match_str('!is_live', {}))
1301 self
.assertTrue(match_str('title', {'title': 'abc'}
))
1302 self
.assertTrue(match_str('title', {'title': ''}
))
1303 self
.assertFalse(match_str('!title', {'title': 'abc'}
))
1304 self
.assertFalse(match_str('!title', {'title': ''}
))
1307 self
.assertFalse(match_str('x>0', {'x': 0}
))
1308 self
.assertFalse(match_str('x>0', {}))
1309 self
.assertTrue(match_str('x>?0', {}))
1310 self
.assertTrue(match_str('x>1K', {'x': 1200}
))
1311 self
.assertFalse(match_str('x>2K', {'x': 1200}
))
1312 self
.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}
))
1313 self
.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}
))
1314 self
.assertTrue(match_str('x > 1:0:0', {'x': 3700}
))
1317 self
.assertFalse(match_str('y=a212', {'y': 'foobar42'}
))
1318 self
.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}
))
1319 self
.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}
))
1320 self
.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}
))
1321 self
.assertTrue(match_str('y^=foo', {'y': 'foobar42'}
))
1322 self
.assertFalse(match_str('y!^=foo', {'y': 'foobar42'}
))
1323 self
.assertFalse(match_str('y^=bar', {'y': 'foobar42'}
))
1324 self
.assertTrue(match_str('y!^=bar', {'y': 'foobar42'}
))
1325 self
.assertRaises(ValueError, match_str
, 'x^=42', {'x': 42}
)
1326 self
.assertTrue(match_str('y*=bar', {'y': 'foobar42'}
))
1327 self
.assertFalse(match_str('y!*=bar', {'y': 'foobar42'}
))
1328 self
.assertFalse(match_str('y*=baz', {'y': 'foobar42'}
))
1329 self
.assertTrue(match_str('y!*=baz', {'y': 'foobar42'}
))
1330 self
.assertTrue(match_str('y$=42', {'y': 'foobar42'}
))
1331 self
.assertFalse(match_str('y$=43', {'y': 'foobar42'}
))
1334 self
.assertFalse(match_str(
1335 'like_count > 100 & dislike_count <? 50 & description',
1336 {'like_count': 90, 'description': 'foo'}
))
1337 self
.assertTrue(match_str(
1338 'like_count > 100 & dislike_count <? 50 & description',
1339 {'like_count': 190, 'description': 'foo'}
))
1340 self
.assertFalse(match_str(
1341 'like_count > 100 & dislike_count <? 50 & description',
1342 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}
))
1343 self
.assertFalse(match_str(
1344 'like_count > 100 & dislike_count <? 50 & description',
1345 {'like_count': 190, 'dislike_count': 10}
))
1348 self
.assertTrue(match_str(r
'x~=\bbar', {'x': 'foo bar'}
))
1349 self
.assertFalse(match_str(r
'x~=\bbar.+', {'x': 'foo bar'}
))
1350 self
.assertFalse(match_str(r
'x~=^FOO', {'x': 'foo bar'}
))
1351 self
.assertTrue(match_str(r
'x~=(?i)^FOO', {'x': 'foo bar'}
))
1354 self
.assertTrue(match_str(r
'x^="foo"', {'x': 'foo "bar"'}
))
1355 self
.assertFalse(match_str(r
'x^="foo "', {'x': 'foo "bar"'}
))
1356 self
.assertFalse(match_str(r
'x$="bar"', {'x': 'foo "bar"'}
))
1357 self
.assertTrue(match_str(r
'x$=" \"bar\""', {'x': 'foo "bar"'}
))
1360 self
.assertFalse(match_str(r
'x=foo & bar', {'x': 'foo & bar'}
))
1361 self
.assertTrue(match_str(r
'x=foo \& bar', {'x': 'foo & bar'}
))
1362 self
.assertTrue(match_str(r
'x=foo \& bar & x^=foo', {'x': 'foo & bar'}
))
1363 self
.assertTrue(match_str(r
'x="foo \& bar" & x^=foo', {'x': 'foo & bar'}
))
1366 self
.assertTrue(match_str(
1367 r
"!is_live & like_count>?100 & description~='(?i)\bcats \& dogs\b'",
1368 {'description': 'Raining Cats & Dogs'}
))
1371 self
.assertFalse(match_str('id!=foo', {'id': 'foo'}
, True))
1372 self
.assertTrue(match_str('x', {'id': 'foo'}
, True))
1373 self
.assertTrue(match_str('!x', {'id': 'foo'}
, True))
1374 self
.assertFalse(match_str('x', {'id': 'foo'}
, False))
1376 def test_parse_dfxp_time_expr(self
):
1377 self
.assertEqual(parse_dfxp_time_expr(None), None)
1378 self
.assertEqual(parse_dfxp_time_expr(''), None)
1379 self
.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1380 self
.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1381 self
.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1382 self
.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1383 self
.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1385 def test_dfxp2srt(self
):
1386 dfxp_data
= '''<?xml version="1.0" encoding="UTF-8"?>
1387 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1390 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1391 <p begin="1" end="2">第二行<br/>♪♪</p>
1392 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1393 <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1394 <p begin="-1" end="-1">Ignore, two</p>
1395 <p begin="3" dur="-1">Ignored, three</p>
1398 </tt>'''.encode('utf-8')
1400 00:00:00,000 --> 00:00:01,000
1401 The following line contains Chinese characters and special symbols
1404 00:00:01,000 --> 00:00:02,000
1409 00:00:02,000 --> 00:00:03,000
1414 self
.assertEqual(dfxp2srt(dfxp_data
), srt_data
)
1416 dfxp_data_no_default_namespace
= '''<?xml version="1.0" encoding="UTF-8"?>
1417 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1420 <p begin="0" end="1">The first line</p>
1423 </tt>'''.encode('utf-8')
1425 00:00:00,000 --> 00:00:01,000
1429 self
.assertEqual(dfxp2srt(dfxp_data_no_default_namespace
), srt_data
)
1431 dfxp_data_with_style
= '''<?xml version="1.0" encoding="utf-8"?>
1432 <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">
1435 <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1436 <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1437 <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1438 <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1441 <body tts:textAlign="center" style="s0">
1443 <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1444 <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>
1445 <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1446 <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>
1449 </tt>'''.encode('utf-8')
1451 00:00:02,080 --> 00:00:05,840
1452 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1455 00:00:02,080 --> 00:00:05,840
1456 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1457 </font>part 2</font></b>
1460 00:00:05,840 --> 00:00:09,560
1461 <u><font color="lime">line 3
1465 00:00:09,560 --> 00:00:12,360
1466 <i><u><font color="yellow"><font color="lime">inner
1467 </font>style</font></u></i>
1470 self
.assertEqual(dfxp2srt(dfxp_data_with_style
), srt_data
)
1472 dfxp_data_non_utf8
= '''<?xml version="1.0" encoding="UTF-16"?>
1473 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1476 <p begin="0" end="1">Line 1</p>
1477 <p begin="1" end="2">第二行</p>
1480 </tt>'''.encode('utf-16')
1482 00:00:00,000 --> 00:00:01,000
1486 00:00:01,000 --> 00:00:02,000
1490 self
.assertEqual(dfxp2srt(dfxp_data_non_utf8
), srt_data
)
1492 def test_cli_option(self
):
1493 self
.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}
, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1494 self
.assertEqual(cli_option({'proxy': None}
, '--proxy', 'proxy'), [])
1495 self
.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1496 self
.assertEqual(cli_option({'retries': 10}
, '--retries', 'retries'), ['--retries', '10'])
1498 def test_cli_valueless_option(self
):
1499 self
.assertEqual(cli_valueless_option(
1500 {'downloader': 'external'}
, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1501 self
.assertEqual(cli_valueless_option(
1502 {'downloader': 'internal'}
, '--external-downloader', 'downloader', 'external'), [])
1503 self
.assertEqual(cli_valueless_option(
1504 {'nocheckcertificate': True}
, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1505 self
.assertEqual(cli_valueless_option(
1506 {'nocheckcertificate': False}
, '--no-check-certificate', 'nocheckcertificate'), [])
1507 self
.assertEqual(cli_valueless_option(
1508 {'checkcertificate': True}
, '--no-check-certificate', 'checkcertificate', False), [])
1509 self
.assertEqual(cli_valueless_option(
1510 {'checkcertificate': False}
, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1512 def test_cli_bool_option(self
):
1515 {'nocheckcertificate': True}
, '--no-check-certificate', 'nocheckcertificate'),
1516 ['--no-check-certificate', 'true'])
1519 {'nocheckcertificate': True}
, '--no-check-certificate', 'nocheckcertificate', separator
='='),
1520 ['--no-check-certificate=true'])
1523 {'nocheckcertificate': True}
, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1524 ['--check-certificate', 'false'])
1527 {'nocheckcertificate': True}
, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1528 ['--check-certificate=false'])
1531 {'nocheckcertificate': False}
, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1532 ['--check-certificate', 'true'])
1535 {'nocheckcertificate': False}
, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1536 ['--check-certificate=true'])
1539 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1542 def test_ohdave_rsa_encrypt(self
):
1543 N
= 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1547 ohdave_rsa_encrypt(b
'aa111222', e
, N
),
1548 '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1550 def test_pkcs1pad(self
):
1552 padded_data
= pkcs1pad(data
, 32)
1553 self
.assertEqual(padded_data
[:2], [0, 2])
1554 self
.assertEqual(padded_data
[28:], [0, 1, 2, 3])
1556 self
.assertRaises(ValueError, pkcs1pad
, data
, 8)
1558 def test_encode_base_n(self
):
1559 self
.assertEqual(encode_base_n(0, 30), '0')
1560 self
.assertEqual(encode_base_n(80, 30), '2k')
1562 custom_table
= '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1563 self
.assertEqual(encode_base_n(0, 30, custom_table
), '9')
1564 self
.assertEqual(encode_base_n(80, 30, custom_table
), '7P')
1566 self
.assertRaises(ValueError, encode_base_n
, 0, 70)
1567 self
.assertRaises(ValueError, encode_base_n
, 0, 60, custom_table
)
1569 def test_caesar(self
):
1570 self
.assertEqual(caesar('ace', 'abcdef', 2), 'cea')
1571 self
.assertEqual(caesar('cea', 'abcdef', -2), 'ace')
1572 self
.assertEqual(caesar('ace', 'abcdef', -2), 'eac')
1573 self
.assertEqual(caesar('eac', 'abcdef', 2), 'ace')
1574 self
.assertEqual(caesar('ace', 'abcdef', 0), 'ace')
1575 self
.assertEqual(caesar('xyz', 'abcdef', 2), 'xyz')
1576 self
.assertEqual(caesar('abc', 'acegik', 2), 'ebg')
1577 self
.assertEqual(caesar('ebg', 'acegik', -2), 'abc')
1579 def test_rot47(self
):
1580 self
.assertEqual(rot47('yt-dlp'), r
'JE\5=A')
1581 self
.assertEqual(rot47('YT-DLP'), r
'*%\s{!')
1583 def test_urshift(self
):
1584 self
.assertEqual(urshift(3, 1), 1)
1585 self
.assertEqual(urshift(-3, 1), 2147483646)
1587 GET_ELEMENT_BY_CLASS_TEST_STRING
= '''
1588 <span class="foo bar">nice</span>
1591 def test_get_element_by_class(self
):
1592 html
= self
.GET_ELEMENT_BY_CLASS_TEST_STRING
1594 self
.assertEqual(get_element_by_class('foo', html
), 'nice')
1595 self
.assertEqual(get_element_by_class('no-such-class', html
), None)
1597 def test_get_element_html_by_class(self
):
1598 html
= self
.GET_ELEMENT_BY_CLASS_TEST_STRING
1600 self
.assertEqual(get_element_html_by_class('foo', html
), html
.strip())
1601 self
.assertEqual(get_element_by_class('no-such-class', html
), None)
1603 GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
= '''
1604 <div itemprop="author" itemscope>foo</div>
1607 def test_get_element_by_attribute(self
):
1608 html
= self
.GET_ELEMENT_BY_CLASS_TEST_STRING
1610 self
.assertEqual(get_element_by_attribute('class', 'foo bar', html
), 'nice')
1611 self
.assertEqual(get_element_by_attribute('class', 'foo', html
), None)
1612 self
.assertEqual(get_element_by_attribute('class', 'no-such-foo', html
), None)
1614 html
= self
.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
1616 self
.assertEqual(get_element_by_attribute('itemprop', 'author', html
), 'foo')
1618 def test_get_element_html_by_attribute(self
):
1619 html
= self
.GET_ELEMENT_BY_CLASS_TEST_STRING
1621 self
.assertEqual(get_element_html_by_attribute('class', 'foo bar', html
), html
.strip())
1622 self
.assertEqual(get_element_html_by_attribute('class', 'foo', html
), None)
1623 self
.assertEqual(get_element_html_by_attribute('class', 'no-such-foo', html
), None)
1625 html
= self
.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
1627 self
.assertEqual(get_element_html_by_attribute('itemprop', 'author', html
), html
.strip())
1629 GET_ELEMENTS_BY_CLASS_TEST_STRING
= '''
1630 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1632 GET_ELEMENTS_BY_CLASS_RES
= ['<span class="foo bar">nice</span>', '<span class="foo bar">also nice</span>']
1634 def test_get_elements_by_class(self
):
1635 html
= self
.GET_ELEMENTS_BY_CLASS_TEST_STRING
1637 self
.assertEqual(get_elements_by_class('foo', html
), ['nice', 'also nice'])
1638 self
.assertEqual(get_elements_by_class('no-such-class', html
), [])
1640 def test_get_elements_html_by_class(self
):
1641 html
= self
.GET_ELEMENTS_BY_CLASS_TEST_STRING
1643 self
.assertEqual(get_elements_html_by_class('foo', html
), self
.GET_ELEMENTS_BY_CLASS_RES
)
1644 self
.assertEqual(get_elements_html_by_class('no-such-class', html
), [])
1646 def test_get_elements_by_attribute(self
):
1647 html
= self
.GET_ELEMENTS_BY_CLASS_TEST_STRING
1649 self
.assertEqual(get_elements_by_attribute('class', 'foo bar', html
), ['nice', 'also nice'])
1650 self
.assertEqual(get_elements_by_attribute('class', 'foo', html
), [])
1651 self
.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html
), [])
1653 def test_get_elements_html_by_attribute(self
):
1654 html
= self
.GET_ELEMENTS_BY_CLASS_TEST_STRING
1656 self
.assertEqual(get_elements_html_by_attribute('class', 'foo bar', html
), self
.GET_ELEMENTS_BY_CLASS_RES
)
1657 self
.assertEqual(get_elements_html_by_attribute('class', 'foo', html
), [])
1658 self
.assertEqual(get_elements_html_by_attribute('class', 'no-such-foo', html
), [])
1660 def test_get_elements_text_and_html_by_attribute(self
):
1661 html
= self
.GET_ELEMENTS_BY_CLASS_TEST_STRING
1664 list(get_elements_text_and_html_by_attribute('class', 'foo bar', html
)),
1665 list(zip(['nice', 'also nice'], self
.GET_ELEMENTS_BY_CLASS_RES
)))
1666 self
.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'foo', html
)), [])
1667 self
.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'no-such-foo', html
)), [])
1669 GET_ELEMENT_BY_TAG_TEST_STRING
= '''
1670 random text lorem ipsum</p>
1672 this should be returned
1673 <span>this should also be returned</span>
1675 this should also be returned
1677 closing tag above should not trick, so this should also be returned
1679 but this text should not be returned
1681 GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML
= GET_ELEMENT_BY_TAG_TEST_STRING
.strip()[32:276]
1682 GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT
= GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML
[5:-6]
1683 GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML
= GET_ELEMENT_BY_TAG_TEST_STRING
.strip()[78:119]
1684 GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT
= GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML
[6:-7]
1686 def test_get_element_text_and_html_by_tag(self
):
1687 html
= self
.GET_ELEMENT_BY_TAG_TEST_STRING
1690 get_element_text_and_html_by_tag('div', html
),
1691 (self
.GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT
, self
.GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML
))
1693 get_element_text_and_html_by_tag('span', html
),
1694 (self
.GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT
, self
.GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML
))
1695 self
.assertRaises(compat_HTMLParseError
, get_element_text_and_html_by_tag
, 'article', html
)
1697 def test_iri_to_uri(self
):
1699 iri_to_uri('https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b'),
1700 'https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b') # Same
1702 iri_to_uri('https://www.google.com/search?q=Käsesoßenrührlöffel'), # German for cheese sauce stirring spoon
1703 'https://www.google.com/search?q=K%C3%A4seso%C3%9Fenr%C3%BChrl%C3%B6ffel')
1705 iri_to_uri('https://www.google.com/search?q=lt<+gt>+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#'),
1706 'https://www.google.com/search?q=lt%3C+gt%3E+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#')
1708 iri_to_uri('http://правозащита38.рф/category/news/'),
1709 'http://xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/')
1711 iri_to_uri('http://www.правозащита38.рф/category/news/'),
1712 'http://www.xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/')
1714 iri_to_uri('https://i❤.ws/emojidomain/👍👏🤝💪'),
1715 'https://xn--i-7iq.ws/emojidomain/%F0%9F%91%8D%F0%9F%91%8F%F0%9F%A4%9D%F0%9F%92%AA')
1717 iri_to_uri('http://日本語.jp/'),
1718 'http://xn--wgv71a119e.jp/')
1720 iri_to_uri('http://导航.中国/'),
1721 'http://xn--fet810g.xn--fiqs8s/')
1723 def test_clean_podcast_url(self
):
1724 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')
1725 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')
1727 def test_LazyList(self
):
1728 it
= list(range(10))
1730 self
.assertEqual(list(LazyList(it
)), it
)
1731 self
.assertEqual(LazyList(it
).exhaust(), it
)
1732 self
.assertEqual(LazyList(it
)[5], it
[5])
1734 self
.assertEqual(LazyList(it
)[5:], it
[5:])
1735 self
.assertEqual(LazyList(it
)[:5], it
[:5])
1736 self
.assertEqual(LazyList(it
)[::2], it
[::2])
1737 self
.assertEqual(LazyList(it
)[1::2], it
[1::2])
1738 self
.assertEqual(LazyList(it
)[5::-1], it
[5::-1])
1739 self
.assertEqual(LazyList(it
)[6:2:-2], it
[6:2:-2])
1740 self
.assertEqual(LazyList(it
)[::-1], it
[::-1])
1742 self
.assertTrue(LazyList(it
))
1743 self
.assertFalse(LazyList(range(0)))
1744 self
.assertEqual(len(LazyList(it
)), len(it
))
1745 self
.assertEqual(repr(LazyList(it
)), repr(it
))
1746 self
.assertEqual(str(LazyList(it
)), str(it
))
1748 self
.assertEqual(list(LazyList(it
, reverse
=True)), it
[::-1])
1749 self
.assertEqual(list(reversed(LazyList(it
))[::-1]), it
)
1750 self
.assertEqual(list(reversed(LazyList(it
))[1:3:7]), it
[::-1][1:3:7])
1752 def test_LazyList_laziness(self
):
1754 def test(ll
, idx
, val
, cache
):
1755 self
.assertEqual(ll
[idx
], val
)
1756 self
.assertEqual(getattr(ll
, '_LazyList__cache'), list(cache
))
1758 ll
= LazyList(range(10))
1759 test(ll
, 0, 0, range(1))
1760 test(ll
, 5, 5, range(6))
1761 test(ll
, -3, 7, range(10))
1763 ll
= LazyList(range(10), reverse
=True)
1764 test(ll
, -1, 0, range(1))
1765 test(ll
, 3, 6, range(10))
1767 ll
= LazyList(itertools
.count())
1768 test(ll
, 10, 10, range(11))
1770 test(ll
, -15, 14, range(15))
1772 def test_format_bytes(self
):
1773 self
.assertEqual(format_bytes(0), '0.00B')
1774 self
.assertEqual(format_bytes(1000), '1000.00B')
1775 self
.assertEqual(format_bytes(1024), '1.00KiB')
1776 self
.assertEqual(format_bytes(1024**2), '1.00MiB')
1777 self
.assertEqual(format_bytes(1024**3), '1.00GiB')
1778 self
.assertEqual(format_bytes(1024**4), '1.00TiB')
1779 self
.assertEqual(format_bytes(1024**5), '1.00PiB')
1780 self
.assertEqual(format_bytes(1024**6), '1.00EiB')
1781 self
.assertEqual(format_bytes(1024**7), '1.00ZiB')
1782 self
.assertEqual(format_bytes(1024**8), '1.00YiB')
1784 def test_hide_login_info(self
):
1785 self
.assertEqual(Config
.hide_login_info(['-u', 'foo', '-p', 'bar']),
1786 ['-u', 'PRIVATE', '-p', 'PRIVATE'])
1787 self
.assertEqual(Config
.hide_login_info(['-u']), ['-u'])
1788 self
.assertEqual(Config
.hide_login_info(['-u', 'foo', '-u', 'bar']),
1789 ['-u', 'PRIVATE', '-u', 'PRIVATE'])
1790 self
.assertEqual(Config
.hide_login_info(['--username=foo']),
1791 ['--username=PRIVATE'])
1794 if __name__
== '__main__':