4 from __future__
import unicode_literals
6 # Allow direct execution
10 sys
.path
.insert(0, os
.path
.dirname(os
.path
.dirname(os
.path
.abspath(__file__
))))
13 # Various small unit tests
16 import xml
.etree
.ElementTree
18 from youtube_dl
.utils
import (
37 get_element_by_attribute
,
38 get_elements_by_class
,
39 get_elements_by_attribute
,
102 from youtube_dl
.compat
import (
104 compat_etree_fromstring
,
113 class TestUtil(unittest
.TestCase
):
114 def test_timeconvert(self
):
115 self
.assertTrue(timeconvert('') is None)
116 self
.assertTrue(timeconvert('bougrg') is None)
118 def test_sanitize_filename(self
):
119 self
.assertEqual(sanitize_filename('abc'), 'abc')
120 self
.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
122 self
.assertEqual(sanitize_filename('123'), '123')
124 self
.assertEqual('abc_de', sanitize_filename('abc/de'))
125 self
.assertFalse('/' in sanitize_filename('abc/de///'))
127 self
.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
128 self
.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
129 self
.assertEqual('yes no', sanitize_filename('yes? no'))
130 self
.assertEqual('this - that', sanitize_filename('this: that'))
132 self
.assertEqual(sanitize_filename('AT&T'), 'AT&T')
134 self
.assertEqual(sanitize_filename(aumlaut
), aumlaut
)
135 tests
= '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
136 self
.assertEqual(sanitize_filename(tests
), tests
)
139 sanitize_filename('New World record at 0:12:34'),
140 'New World record at 0_12_34')
142 self
.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
143 self
.assertEqual(sanitize_filename('--gasdgf', is_id
=True), '--gasdgf')
144 self
.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
145 self
.assertEqual(sanitize_filename('.gasdgf', is_id
=True), '.gasdgf')
149 for fbc
in forbidden
:
150 self
.assertTrue(fbc
not in sanitize_filename(fc
))
152 def test_sanitize_filename_restricted(self
):
153 self
.assertEqual(sanitize_filename('abc', restricted
=True), 'abc')
154 self
.assertEqual(sanitize_filename('abc_d-e', restricted
=True), 'abc_d-e')
156 self
.assertEqual(sanitize_filename('123', restricted
=True), '123')
158 self
.assertEqual('abc_de', sanitize_filename('abc/de', restricted
=True))
159 self
.assertFalse('/' in sanitize_filename('abc/de///', restricted
=True))
161 self
.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted
=True))
162 self
.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted
=True))
163 self
.assertEqual('yes_no', sanitize_filename('yes? no', restricted
=True))
164 self
.assertEqual('this_-_that', sanitize_filename('this: that', restricted
=True))
166 tests
= 'aäb\u4e2d\u56fd\u7684c'
167 self
.assertEqual(sanitize_filename(tests
, restricted
=True), 'aab_c')
168 self
.assertTrue(sanitize_filename('\xf6', restricted
=True) != '') # No empty filename
170 forbidden
= '"\0\\/&!: \'\t\n()[]{}$;`^,#'
172 for fbc
in forbidden
:
173 self
.assertTrue(fbc
not in sanitize_filename(fc
, restricted
=True))
175 # Handle a common case more neatly
176 self
.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted
=True), 'Song')
177 self
.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted
=True), 'Speech')
178 # .. but make sure the file name is never empty
179 self
.assertTrue(sanitize_filename('-', restricted
=True) != '')
180 self
.assertTrue(sanitize_filename(':', restricted
=True) != '')
182 self
.assertEqual(sanitize_filename(
183 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted
=True),
184 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYPssaaaaaaaeceeeeiiiionooooooooeuuuuuypy')
186 def test_sanitize_ids(self
):
187 self
.assertEqual(sanitize_filename('_n_cd26wFpw', is_id
=True), '_n_cd26wFpw')
188 self
.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id
=True), '_BD_eEpuzXw')
189 self
.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id
=True), 'N0Y__7-UOdI')
191 def test_sanitize_path(self
):
192 if sys
.platform
!= 'win32':
195 self
.assertEqual(sanitize_path('abc'), 'abc')
196 self
.assertEqual(sanitize_path('abc/def'), 'abc\\def')
197 self
.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
198 self
.assertEqual(sanitize_path('abc|def'), 'abc#def')
199 self
.assertEqual(sanitize_path('<>:"|?*'), '#######')
200 self
.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
201 self
.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
203 self
.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
204 self
.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
206 self
.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
207 self
.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
208 self
.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
209 self
.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
212 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
213 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
216 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
217 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
218 self
.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
219 self
.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
220 self
.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
222 self
.assertEqual(sanitize_path('../abc'), '..\\abc')
223 self
.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
224 self
.assertEqual(sanitize_path('./abc'), 'abc')
225 self
.assertEqual(sanitize_path('./../abc'), '..\\abc')
227 def test_sanitize_url(self
):
228 self
.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
229 self
.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
230 self
.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
231 self
.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')
233 def test_expand_path(self
):
235 return '%{0}%'.format(var
) if sys
.platform
== 'win32' else '${0}'.format(var
)
237 compat_setenv('YOUTUBE_DL_EXPATH_PATH', 'expanded')
238 self
.assertEqual(expand_path(env('YOUTUBE_DL_EXPATH_PATH')), 'expanded')
239 self
.assertEqual(expand_path(env('HOME')), compat_getenv('HOME'))
240 self
.assertEqual(expand_path('~'), compat_getenv('HOME'))
242 expand_path('~/%s' % env('YOUTUBE_DL_EXPATH_PATH')),
243 '%s/expanded' % compat_getenv('HOME'))
245 def test_prepend_extension(self
):
246 self
.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
247 self
.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
248 self
.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
249 self
.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
250 self
.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
251 self
.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
253 def test_replace_extension(self
):
254 self
.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
255 self
.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
256 self
.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
257 self
.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
258 self
.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
259 self
.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
261 def test_remove_start(self
):
262 self
.assertEqual(remove_start(None, 'A - '), None)
263 self
.assertEqual(remove_start('A - B', 'A - '), 'B')
264 self
.assertEqual(remove_start('B - A', 'A - '), 'B - A')
266 def test_remove_end(self
):
267 self
.assertEqual(remove_end(None, ' - B'), None)
268 self
.assertEqual(remove_end('A - B', ' - B'), 'A')
269 self
.assertEqual(remove_end('B - A', ' - B'), 'B - A')
271 def test_remove_quotes(self
):
272 self
.assertEqual(remove_quotes(None), None)
273 self
.assertEqual(remove_quotes('"'), '"')
274 self
.assertEqual(remove_quotes("'"), "'")
275 self
.assertEqual(remove_quotes(';'), ';')
276 self
.assertEqual(remove_quotes('";'), '";')
277 self
.assertEqual(remove_quotes('""'), '')
278 self
.assertEqual(remove_quotes('";"'), ';')
280 def test_ordered_set(self
):
281 self
.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
282 self
.assertEqual(orderedSet([]), [])
283 self
.assertEqual(orderedSet([1]), [1])
284 # keep the list ordered
285 self
.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
287 def test_unescape_html(self
):
288 self
.assertEqual(unescapeHTML('%20;'), '%20;')
289 self
.assertEqual(unescapeHTML('/'), '/')
290 self
.assertEqual(unescapeHTML('/'), '/')
291 self
.assertEqual(unescapeHTML('é'), 'é')
292 self
.assertEqual(unescapeHTML('�'), '�')
293 self
.assertEqual(unescapeHTML('&a"'), '&a"')
295 self
.assertEqual(unescapeHTML('.''), '.\'')
297 def test_date_from_str(self
):
298 self
.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
299 self
.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
300 self
.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
301 self
.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
302 self
.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
304 def test_daterange(self
):
305 _20century
= DateRange("19000101", "20000101")
306 self
.assertFalse("17890714" in _20century
)
307 _ac
= DateRange("00010101")
308 self
.assertTrue("19690721" in _ac
)
309 _firstmilenium
= DateRange(end
="10000101")
310 self
.assertTrue("07110427" in _firstmilenium
)
312 def test_unified_dates(self
):
313 self
.assertEqual(unified_strdate('December 21, 2010'), '20101221')
314 self
.assertEqual(unified_strdate('8/7/2009'), '20090708')
315 self
.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
316 self
.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
317 self
.assertEqual(unified_strdate('1968 12 10'), '19681210')
318 self
.assertEqual(unified_strdate('1968-12-10'), '19681210')
319 self
.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
321 unified_strdate('11/26/2014 11:30:00 AM PST', day_first
=False),
324 unified_strdate('2/2/2015 6:47:40 PM', day_first
=False),
326 self
.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
327 self
.assertEqual(unified_strdate('25-09-2014'), '20140925')
328 self
.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
329 self
.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
330 self
.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
331 self
.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
332 self
.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
333 self
.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
335 def test_unified_timestamps(self
):
336 self
.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
337 self
.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
338 self
.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
339 self
.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
340 self
.assertEqual(unified_timestamp('1968 12 10'), -33436800)
341 self
.assertEqual(unified_timestamp('1968-12-10'), -33436800)
342 self
.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
344 unified_timestamp('11/26/2014 11:30:00 AM PST', day_first
=False),
347 unified_timestamp('2/2/2015 6:47:40 PM', day_first
=False),
349 self
.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
350 self
.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
351 self
.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
352 self
.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
353 self
.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
354 self
.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
355 self
.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
356 self
.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
357 self
.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
358 self
.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363)
360 def test_determine_ext(self
):
361 self
.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
362 self
.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
363 self
.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
364 self
.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
365 self
.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
366 self
.assertEqual(determine_ext('foobar', None), None)
368 def test_find_xpath_attr(self
):
376 doc
= compat_etree_fromstring(testxml
)
378 self
.assertEqual(find_xpath_attr(doc
, './/fourohfour', 'n'), None)
379 self
.assertEqual(find_xpath_attr(doc
, './/fourohfour', 'n', 'v'), None)
380 self
.assertEqual(find_xpath_attr(doc
, './/node', 'n'), None)
381 self
.assertEqual(find_xpath_attr(doc
, './/node', 'n', 'v'), None)
382 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x'), doc
[1])
383 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x', 'a'), doc
[1])
384 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x', 'b'), doc
[3])
385 self
.assertEqual(find_xpath_attr(doc
, './/node', 'y'), doc
[2])
386 self
.assertEqual(find_xpath_attr(doc
, './/node', 'y', 'c'), doc
[2])
387 self
.assertEqual(find_xpath_attr(doc
, './/node', 'y', 'd'), doc
[3])
388 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x', ''), doc
[4])
390 def test_xpath_with_ns(self
):
391 testxml
= '''<root xmlns:media="http://example.com/">
393 <media:author>The Author</media:author>
394 <url>http://server.com/download.mp3</url>
397 doc
= compat_etree_fromstring(testxml
)
398 find
= lambda p
: doc
.find(xpath_with_ns(p
, {'media': 'http://example.com/'}
))
399 self
.assertTrue(find('media:song') is not None)
400 self
.assertEqual(find('media:song/media:author').text
, 'The Author')
401 self
.assertEqual(find('media:song/url').text
, 'http://server.com/download.mp3')
403 def test_xpath_element(self
):
404 doc
= xml
.etree
.ElementTree
.Element('root')
405 div
= xml
.etree
.ElementTree
.SubElement(doc
, 'div')
406 p
= xml
.etree
.ElementTree
.SubElement(div
, 'p')
408 self
.assertEqual(xpath_element(doc
, 'div/p'), p
)
409 self
.assertEqual(xpath_element(doc
, ['div/p']), p
)
410 self
.assertEqual(xpath_element(doc
, ['div/bar', 'div/p']), p
)
411 self
.assertEqual(xpath_element(doc
, 'div/bar', default
='default'), 'default')
412 self
.assertEqual(xpath_element(doc
, ['div/bar'], default
='default'), 'default')
413 self
.assertTrue(xpath_element(doc
, 'div/bar') is None)
414 self
.assertTrue(xpath_element(doc
, ['div/bar']) is None)
415 self
.assertTrue(xpath_element(doc
, ['div/bar'], 'div/baz') is None)
416 self
.assertRaises(ExtractorError
, xpath_element
, doc
, 'div/bar', fatal
=True)
417 self
.assertRaises(ExtractorError
, xpath_element
, doc
, ['div/bar'], fatal
=True)
418 self
.assertRaises(ExtractorError
, xpath_element
, doc
, ['div/bar', 'div/baz'], fatal
=True)
420 def test_xpath_text(self
):
426 doc
= compat_etree_fromstring(testxml
)
427 self
.assertEqual(xpath_text(doc
, 'div/p'), 'Foo')
428 self
.assertEqual(xpath_text(doc
, 'div/bar', default
='default'), 'default')
429 self
.assertTrue(xpath_text(doc
, 'div/bar') is None)
430 self
.assertRaises(ExtractorError
, xpath_text
, doc
, 'div/bar', fatal
=True)
432 def test_xpath_attr(self
):
438 doc
= compat_etree_fromstring(testxml
)
439 self
.assertEqual(xpath_attr(doc
, 'div/p', 'x'), 'a')
440 self
.assertEqual(xpath_attr(doc
, 'div/bar', 'x'), None)
441 self
.assertEqual(xpath_attr(doc
, 'div/p', 'y'), None)
442 self
.assertEqual(xpath_attr(doc
, 'div/bar', 'x', default
='default'), 'default')
443 self
.assertEqual(xpath_attr(doc
, 'div/p', 'y', default
='default'), 'default')
444 self
.assertRaises(ExtractorError
, xpath_attr
, doc
, 'div/bar', 'x', fatal
=True)
445 self
.assertRaises(ExtractorError
, xpath_attr
, doc
, 'div/p', 'y', fatal
=True)
447 def test_smuggle_url(self
):
448 data
= {"ö": "ö", "abc": [3]}
449 url
= 'https://foo.bar/baz?x=y#a'
450 smug_url
= smuggle_url(url
, data
)
451 unsmug_url
, unsmug_data
= unsmuggle_url(smug_url
)
452 self
.assertEqual(url
, unsmug_url
)
453 self
.assertEqual(data
, unsmug_data
)
455 res_url
, res_data
= unsmuggle_url(url
)
456 self
.assertEqual(res_url
, url
)
457 self
.assertEqual(res_data
, None)
459 smug_url
= smuggle_url(url
, {'a': 'b'}
)
460 smug_smug_url
= smuggle_url(smug_url
, {'c': 'd'}
)
461 res_url
, res_data
= unsmuggle_url(smug_smug_url
)
462 self
.assertEqual(res_url
, url
)
463 self
.assertEqual(res_data
, {'a': 'b', 'c': 'd'}
)
465 def test_shell_quote(self
):
466 args
= ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
469 """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name
!= 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
471 def test_str_to_int(self
):
472 self
.assertEqual(str_to_int('123,456'), 123456)
473 self
.assertEqual(str_to_int('123.456'), 123456)
475 def test_url_basename(self
):
476 self
.assertEqual(url_basename('http://foo.de/'), '')
477 self
.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
478 self
.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
479 self
.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
480 self
.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
482 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
485 def test_base_url(self
):
486 self
.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
487 self
.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
488 self
.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
489 self
.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
490 self
.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
492 def test_urljoin(self
):
493 self
.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
494 self
.assertEqual(urljoin(b
'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
495 self
.assertEqual(urljoin('http://foo.de/', b
'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
496 self
.assertEqual(urljoin(b
'http://foo.de/', b
'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
497 self
.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
498 self
.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
499 self
.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
500 self
.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
501 self
.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
502 self
.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
503 self
.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
504 self
.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
505 self
.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
506 self
.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
507 self
.assertEqual(urljoin('http://foo.de/', None), None)
508 self
.assertEqual(urljoin('http://foo.de/', ''), None)
509 self
.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
510 self
.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
511 self
.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de')
512 self
.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de')
514 def test_url_or_none(self
):
515 self
.assertEqual(url_or_none(None), None)
516 self
.assertEqual(url_or_none(''), None)
517 self
.assertEqual(url_or_none('foo'), None)
518 self
.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
519 self
.assertEqual(url_or_none('https://foo.de'), 'https://foo.de')
520 self
.assertEqual(url_or_none('http$://foo.de'), None)
521 self
.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
522 self
.assertEqual(url_or_none('//foo.de'), '//foo.de')
524 def test_parse_age_limit(self
):
525 self
.assertEqual(parse_age_limit(None), None)
526 self
.assertEqual(parse_age_limit(False), None)
527 self
.assertEqual(parse_age_limit('invalid'), None)
528 self
.assertEqual(parse_age_limit(0), 0)
529 self
.assertEqual(parse_age_limit(18), 18)
530 self
.assertEqual(parse_age_limit(21), 21)
531 self
.assertEqual(parse_age_limit(22), None)
532 self
.assertEqual(parse_age_limit('18'), 18)
533 self
.assertEqual(parse_age_limit('18+'), 18)
534 self
.assertEqual(parse_age_limit('PG-13'), 13)
535 self
.assertEqual(parse_age_limit('TV-14'), 14)
536 self
.assertEqual(parse_age_limit('TV-MA'), 17)
537 self
.assertEqual(parse_age_limit('TV14'), 14)
538 self
.assertEqual(parse_age_limit('TV_G'), 0)
540 def test_parse_duration(self
):
541 self
.assertEqual(parse_duration(None), None)
542 self
.assertEqual(parse_duration(False), None)
543 self
.assertEqual(parse_duration('invalid'), None)
544 self
.assertEqual(parse_duration('1'), 1)
545 self
.assertEqual(parse_duration('1337:12'), 80232)
546 self
.assertEqual(parse_duration('9:12:43'), 33163)
547 self
.assertEqual(parse_duration('12:00'), 720)
548 self
.assertEqual(parse_duration('00:01:01'), 61)
549 self
.assertEqual(parse_duration('x:y'), None)
550 self
.assertEqual(parse_duration('3h11m53s'), 11513)
551 self
.assertEqual(parse_duration('3h 11m 53s'), 11513)
552 self
.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
553 self
.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
554 self
.assertEqual(parse_duration('62m45s'), 3765)
555 self
.assertEqual(parse_duration('6m59s'), 419)
556 self
.assertEqual(parse_duration('49s'), 49)
557 self
.assertEqual(parse_duration('0h0m0s'), 0)
558 self
.assertEqual(parse_duration('0m0s'), 0)
559 self
.assertEqual(parse_duration('0s'), 0)
560 self
.assertEqual(parse_duration('01:02:03.05'), 3723.05)
561 self
.assertEqual(parse_duration('T30M38S'), 1838)
562 self
.assertEqual(parse_duration('5 s'), 5)
563 self
.assertEqual(parse_duration('3 min'), 180)
564 self
.assertEqual(parse_duration('2.5 hours'), 9000)
565 self
.assertEqual(parse_duration('02:03:04'), 7384)
566 self
.assertEqual(parse_duration('01:02:03:04'), 93784)
567 self
.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
568 self
.assertEqual(parse_duration('87 Min.'), 5220)
569 self
.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
570 self
.assertEqual(parse_duration('PT00H03M30SZ'), 210)
571 self
.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)
573 def test_fix_xml_ampersands(self
):
575 fix_xml_ampersands('"&x=y&z=a'), '"&x=y&z=a')
577 fix_xml_ampersands('"&x=y&wrong;&z=a'),
578 '"&x=y&wrong;&z=a')
580 fix_xml_ampersands('&'><"'),
581 '&'><"')
583 fix_xml_ampersands('Ӓ᪼'), 'Ӓ᪼')
584 self
.assertEqual(fix_xml_ampersands('&#&#'), '&#&#')
586 def test_paged_list(self
):
587 def testPL(size
, pagesize
, sliceargs
, expected
):
588 def get_page(pagenum
):
589 firstid
= pagenum
* pagesize
590 upto
= min(size
, pagenum
* pagesize
+ pagesize
)
591 for i
in range(firstid
, upto
):
594 pl
= OnDemandPagedList(get_page
, pagesize
)
595 got
= pl
.getslice(*sliceargs
)
596 self
.assertEqual(got
, expected
)
598 iapl
= InAdvancePagedList(get_page
, size
// pagesize
+ 1, pagesize
)
599 got
= iapl
.getslice(*sliceargs
)
600 self
.assertEqual(got
, expected
)
602 testPL(5, 2, (), [0, 1, 2, 3, 4])
603 testPL(5, 2, (1,), [1, 2, 3, 4])
604 testPL(5, 2, (2,), [2, 3, 4])
605 testPL(5, 2, (4,), [4])
606 testPL(5, 2, (0, 3), [0, 1, 2])
607 testPL(5, 2, (1, 4), [1, 2, 3])
608 testPL(5, 2, (2, 99), [2, 3, 4])
609 testPL(5, 2, (20, 99), [])
611 def test_read_batch_urls(self
):
612 f
= io
.StringIO('''\xef\xbb\xbf foo
615 # More after this line\r
618 self
.assertEqual(read_batch_urls(f
), ['foo', 'bar', 'baz', 'bam'])
620 def test_urlencode_postdata(self
):
621 data
= urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'}
)
622 self
.assertTrue(isinstance(data
, bytes))
624 def test_update_url_query(self
):
626 return compat_parse_qs(compat_urlparse
.urlparse(url
).query
)
627 self
.assertEqual(query_dict(update_url_query(
628 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']}
)),
629 query_dict('http://example.com/path?quality=HD&format=mp4'))
630 self
.assertEqual(query_dict(update_url_query(
631 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']}
)),
632 query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
633 self
.assertEqual(query_dict(update_url_query(
634 'http://example.com/path', {'fields': 'id,formats,subtitles'}
)),
635 query_dict('http://example.com/path?fields=id,formats,subtitles'))
636 self
.assertEqual(query_dict(update_url_query(
637 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')}
)),
638 query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
639 self
.assertEqual(query_dict(update_url_query(
640 'http://example.com/path?manifest=f4m', {'manifest': []}
)),
641 query_dict('http://example.com/path'))
642 self
.assertEqual(query_dict(update_url_query(
643 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'}
)),
644 query_dict('http://example.com/path?system=LINUX'))
645 self
.assertEqual(query_dict(update_url_query(
646 'http://example.com/path', {'fields': b'id,formats,subtitles'}
)),
647 query_dict('http://example.com/path?fields=id,formats,subtitles'))
648 self
.assertEqual(query_dict(update_url_query(
649 'http://example.com/path', {'width': 1080, 'height': 720}
)),
650 query_dict('http://example.com/path?width=1080&height=720'))
651 self
.assertEqual(query_dict(update_url_query(
652 'http://example.com/path', {'bitrate': 5020.43}
)),
653 query_dict('http://example.com/path?bitrate=5020.43'))
654 self
.assertEqual(query_dict(update_url_query(
655 'http://example.com/path', {'test': '第二行тест'}
)),
656 query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
658 def test_multipart_encode(self
):
660 multipart_encode({b'field': b'value'}
, boundary
='AAAAAA')[0],
661 b
'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
663 multipart_encode({'欄位'.encode('utf-8'): '值'.encode('utf-8')}
, boundary
='AAAAAA')[0],
664 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')
666 ValueError, multipart_encode
, {b'field': b'value'}
, boundary
='value')
668 def test_dict_get(self
):
676 d
= FALSE_VALUES
.copy()
678 self
.assertEqual(dict_get(d
, 'a'), 42)
679 self
.assertEqual(dict_get(d
, 'b'), None)
680 self
.assertEqual(dict_get(d
, 'b', 42), 42)
681 self
.assertEqual(dict_get(d
, ('a', )), 42)
682 self
.assertEqual(dict_get(d
, ('b', 'a', )), 42)
683 self
.assertEqual(dict_get(d
, ('b', 'c', 'a', 'd', )), 42)
684 self
.assertEqual(dict_get(d
, ('b', 'c', )), None)
685 self
.assertEqual(dict_get(d
, ('b', 'c', ), 42), 42)
686 for key
, false_value
in FALSE_VALUES
.items():
687 self
.assertEqual(dict_get(d
, ('b', 'c', key
, )), None)
688 self
.assertEqual(dict_get(d
, ('b', 'c', key
, ), skip_false_values
=False), false_value
)
690 def test_merge_dicts(self
):
691 self
.assertEqual(merge_dicts({'a': 1}
, {'b': 2}
), {'a': 1, 'b': 2}
)
692 self
.assertEqual(merge_dicts({'a': 1}
, {'a': 2}
), {'a': 1}
)
693 self
.assertEqual(merge_dicts({'a': 1}
, {'a': None}
), {'a': 1}
)
694 self
.assertEqual(merge_dicts({'a': 1}
, {'a': ''}
), {'a': 1}
)
695 self
.assertEqual(merge_dicts({'a': 1}
, {}), {'a': 1}
)
696 self
.assertEqual(merge_dicts({'a': None}
, {'a': 1}
), {'a': 1}
)
697 self
.assertEqual(merge_dicts({'a': ''}
, {'a': 1}
), {'a': ''}
)
698 self
.assertEqual(merge_dicts({'a': ''}
, {'a': 'abc'}
), {'a': 'abc'}
)
699 self
.assertEqual(merge_dicts({'a': None}
, {'a': ''}
, {'a': 'abc'}
), {'a': 'abc'}
)
701 def test_encode_compat_str(self
):
702 self
.assertEqual(encode_compat_str(b
'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
703 self
.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
705 def test_parse_iso8601(self
):
706 self
.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
707 self
.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
708 self
.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
709 self
.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
710 self
.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
711 self
.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
713 def test_strip_jsonp(self
):
714 stripped
= strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
715 d
= json
.loads(stripped
)
716 self
.assertEqual(d
, [{"id": "532cb", "x": 3}
])
718 stripped
= strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
719 d
= json
.loads(stripped
)
720 self
.assertEqual(d
, {'STATUS': 'OK'}
)
722 stripped
= strip_jsonp('ps.embedHandler({"status": "success"});')
723 d
= json
.loads(stripped
)
724 self
.assertEqual(d
, {'status': 'success'}
)
726 stripped
= strip_jsonp('window.cb && window.cb({"status": "success"});')
727 d
= json
.loads(stripped
)
728 self
.assertEqual(d
, {'status': 'success'}
)
730 stripped
= strip_jsonp('window.cb && cb({"status": "success"});')
731 d
= json
.loads(stripped
)
732 self
.assertEqual(d
, {'status': 'success'}
)
734 stripped
= strip_jsonp('({"status": "success"});')
735 d
= json
.loads(stripped
)
736 self
.assertEqual(d
, {'status': 'success'}
)
738 def test_uppercase_escape(self
):
739 self
.assertEqual(uppercase_escape('aä'), 'aä')
740 self
.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
742 def test_lowercase_escape(self
):
743 self
.assertEqual(lowercase_escape('aä'), 'aä')
744 self
.assertEqual(lowercase_escape('\\u0026'), '&')
746 def test_limit_length(self
):
747 self
.assertEqual(limit_length(None, 12), None)
748 self
.assertEqual(limit_length('foo', 12), 'foo')
750 limit_length('foo bar baz asd', 12).startswith('foo bar'))
751 self
.assertTrue('...' in limit_length('foo bar baz asd', 12))
753 def test_mimetype2ext(self
):
754 self
.assertEqual(mimetype2ext(None), None)
755 self
.assertEqual(mimetype2ext('video/x-flv'), 'flv')
756 self
.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
757 self
.assertEqual(mimetype2ext('text/vtt'), 'vtt')
758 self
.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
759 self
.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
761 def test_month_by_name(self
):
762 self
.assertEqual(month_by_name(None), None)
763 self
.assertEqual(month_by_name('December', 'en'), 12)
764 self
.assertEqual(month_by_name('décembre', 'fr'), 12)
765 self
.assertEqual(month_by_name('December'), 12)
766 self
.assertEqual(month_by_name('décembre'), None)
767 self
.assertEqual(month_by_name('Unknown', 'unknown'), None)
769 def test_parse_codecs(self
):
770 self
.assertEqual(parse_codecs(''), {})
771 self
.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
772 'vcodec': 'avc1.77.30',
773 'acodec': 'mp4a.40.2',
775 self
.assertEqual(parse_codecs('mp4a.40.2'), {
777 'acodec': 'mp4a.40.2',
779 self
.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
780 'vcodec': 'avc1.42001e',
781 'acodec': 'mp4a.40.5',
783 self
.assertEqual(parse_codecs('avc3.640028'), {
784 'vcodec': 'avc3.640028',
787 self
.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
791 self
.assertEqual(parse_codecs('av01.0.05M.08'), {
792 'vcodec': 'av01.0.05M.08',
796 def test_escape_rfc3986(self
):
797 reserved
= "!*'();:@&=+$,/?#[]"
798 unreserved
= 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
799 self
.assertEqual(escape_rfc3986(reserved
), reserved
)
800 self
.assertEqual(escape_rfc3986(unreserved
), unreserved
)
801 self
.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
802 self
.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
803 self
.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
804 self
.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
806 def test_escape_url(self
):
808 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
809 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
812 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
813 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
816 escape_url('http://тест.рф/фрагмент'),
817 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
820 escape_url('http://тест.рф/абв?абв=абв#абв'),
821 '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'
823 self
.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
825 def test_js_to_json_realworld(self
):
827 'clip':{'provider':'pseudo'}
829 self
.assertEqual(js_to_json(inp
), '''{
830 "clip":{"provider":"pseudo"}
832 json
.loads(js_to_json(inp
))
835 'playlist':[{'controls':{'all':null}}]
837 self
.assertEqual(js_to_json(inp
), '''{
838 "playlist":[{"controls":{"all":null}}]
841 inp
= '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
842 self
.assertEqual(js_to_json(inp
), '''"The CW's 'Crazy Ex-Girlfriend'"''')
844 inp
= '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
845 json_code
= js_to_json(inp
)
846 self
.assertEqual(json
.loads(json_code
), json
.loads(inp
))
849 0:{src:'skipped', type: 'application/dash+xml'},
850 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
852 self
.assertEqual(js_to_json(inp
), '''{
853 "0":{"src":"skipped", "type": "application/dash+xml"},
854 "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
857 inp
= '''{"foo":101}'''
858 self
.assertEqual(js_to_json(inp
), '''{"foo":101}''')
860 inp
= '''{"duration": "00:01:07"}'''
861 self
.assertEqual(js_to_json(inp
), '''{"duration": "00:01:07"}''')
863 inp
= '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}'''
864 self
.assertEqual(js_to_json(inp
), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''')
866 def test_js_to_json_edgecases(self
):
867 on
= js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
868 self
.assertEqual(json
.loads(on
), {"abc_def": "1'\\2\\'3\"4"}
)
870 on
= js_to_json('{"abc": true}')
871 self
.assertEqual(json
.loads(on
), {'abc': True}
)
873 # Ignore JavaScript code as well
880 self
.assertEqual(d
['x'], 1)
881 self
.assertEqual(d
['y'], 'a')
883 on
= js_to_json('["abc", "def",]')
884 self
.assertEqual(json
.loads(on
), ['abc', 'def'])
886 on
= js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
887 self
.assertEqual(json
.loads(on
), ['abc', 'def'])
889 on
= js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
890 self
.assertEqual(json
.loads(on
), ['abc', 'def'])
892 on
= js_to_json('{"abc": "def",}')
893 self
.assertEqual(json
.loads(on
), {'abc': 'def'}
)
895 on
= js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
896 self
.assertEqual(json
.loads(on
), {'abc': 'def'}
)
898 on
= js_to_json('{ 0: /* " \n */ ",]" , }')
899 self
.assertEqual(json
.loads(on
), {'0': ',]'}
)
901 on
= js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
902 self
.assertEqual(json
.loads(on
), {'0': ',]'}
)
904 on
= js_to_json('{ 0: // comment\n1 }')
905 self
.assertEqual(json
.loads(on
), {'0': 1}
)
907 on
= js_to_json(r
'["<p>x<\/p>"]')
908 self
.assertEqual(json
.loads(on
), ['<p>x</p>'])
910 on
= js_to_json(r
'["\xaa"]')
911 self
.assertEqual(json
.loads(on
), ['\u00aa'])
913 on
= js_to_json("['a\\\nb']")
914 self
.assertEqual(json
.loads(on
), ['ab'])
916 on
= js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
917 self
.assertEqual(json
.loads(on
), ['ab'])
919 on
= js_to_json('{0xff:0xff}')
920 self
.assertEqual(json
.loads(on
), {'255': 255}
)
922 on
= js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
923 self
.assertEqual(json
.loads(on
), {'255': 255}
)
925 on
= js_to_json('{077:077}')
926 self
.assertEqual(json
.loads(on
), {'63': 63}
)
928 on
= js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
929 self
.assertEqual(json
.loads(on
), {'63': 63}
)
931 on
= js_to_json('{42:42}')
932 self
.assertEqual(json
.loads(on
), {'42': 42}
)
934 on
= js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
935 self
.assertEqual(json
.loads(on
), {'42': 42}
)
937 on
= js_to_json('{42:4.2e1}')
938 self
.assertEqual(json
.loads(on
), {'42': 42.0}
)
940 def test_js_to_json_malformed(self
):
941 self
.assertEqual(js_to_json('42a1'), '42"a1"')
942 self
.assertEqual(js_to_json('42a-1'), '42"a"-1')
944 def test_extract_attributes(self
):
945 self
.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}
)
946 self
.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'}
)
947 self
.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'}
)
948 self
.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"}
)
949 self
.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'}
)
950 self
.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}
)
951 self
.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}
)
952 self
.assertEqual(extract_attributes('<e x="&">'), {'x': '&'}
) # XML
953 self
.assertEqual(extract_attributes('<e x=""">'), {'x': '"'}
)
954 self
.assertEqual(extract_attributes('<e x="£">'), {'x': '£'}
) # HTML 3.2
955 self
.assertEqual(extract_attributes('<e x="λ">'), {'x': 'λ'}
) # HTML 4.0
956 self
.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'}
)
957 self
.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"}
)
958 self
.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'}
)
959 self
.assertEqual(extract_attributes('<e x >'), {'x': None}
)
960 self
.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None}
)
961 self
.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'}
)
962 self
.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'}
)
963 self
.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'}
)
964 self
.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'}
)
965 self
.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'}
)
966 self
.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'}
)
967 self
.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}
) # Names lowercased
968 self
.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'}
)
969 self
.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'}
)
970 self
.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'}
)
971 self
.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'}
)
972 self
.assertEqual(extract_attributes('<e x="décomposé">'), {'x': 'décompose\u0301'}
)
973 # "Narrow" Python builds don't support unicode code points outside BMP.
976 supports_outside_bmp
= True
978 supports_outside_bmp
= False
979 if supports_outside_bmp
:
980 self
.assertEqual(extract_attributes('<e x="Smile 😀!">'), {'x': 'Smile \U0001f600!'}
)
981 # Malformed HTML should not break attributes extraction on older Python
982 self
.assertEqual(extract_attributes('<mal"formed/>'), {})
984 def test_clean_html(self
):
985 self
.assertEqual(clean_html('a:\nb'), 'a: b')
986 self
.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
987 self
.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
989 def test_intlist_to_bytes(self
):
991 intlist_to_bytes([0, 1, 127, 128, 255]),
992 b
'\x00\x01\x7f\x80\xff')
994 def test_args_to_str(self
):
996 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
997 'foo ba/r -baz \'2 be\' \'\'' if compat_os_name
!= 'nt' else 'foo ba/r -baz "2 be" ""'
1000 def test_parse_filesize(self
):
1001 self
.assertEqual(parse_filesize(None), None)
1002 self
.assertEqual(parse_filesize(''), None)
1003 self
.assertEqual(parse_filesize('91 B'), 91)
1004 self
.assertEqual(parse_filesize('foobar'), None)
1005 self
.assertEqual(parse_filesize('2 MiB'), 2097152)
1006 self
.assertEqual(parse_filesize('5 GB'), 5000000000)
1007 self
.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
1008 self
.assertEqual(parse_filesize('1.2tb'), 1200000000000)
1009 self
.assertEqual(parse_filesize('1,24 KB'), 1240)
1010 self
.assertEqual(parse_filesize('1,24 kb'), 1240)
1011 self
.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
1013 def test_parse_count(self
):
1014 self
.assertEqual(parse_count(None), None)
1015 self
.assertEqual(parse_count(''), None)
1016 self
.assertEqual(parse_count('0'), 0)
1017 self
.assertEqual(parse_count('1000'), 1000)
1018 self
.assertEqual(parse_count('1.000'), 1000)
1019 self
.assertEqual(parse_count('1.1k'), 1100)
1020 self
.assertEqual(parse_count('1.1kk'), 1100000)
1021 self
.assertEqual(parse_count('1.1kk '), 1100000)
1022 self
.assertEqual(parse_count('1.1kk views'), 1100000)
1024 def test_parse_resolution(self
):
1025 self
.assertEqual(parse_resolution(None), {})
1026 self
.assertEqual(parse_resolution(''), {})
1027 self
.assertEqual(parse_resolution('1920x1080'), {'width': 1920, 'height': 1080}
)
1028 self
.assertEqual(parse_resolution('1920×1080'), {'width': 1920, 'height': 1080}
)
1029 self
.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080}
)
1030 self
.assertEqual(parse_resolution('720p'), {'height': 720}
)
1031 self
.assertEqual(parse_resolution('4k'), {'height': 2160}
)
1032 self
.assertEqual(parse_resolution('8K'), {'height': 4320}
)
1034 def test_parse_bitrate(self
):
1035 self
.assertEqual(parse_bitrate(None), None)
1036 self
.assertEqual(parse_bitrate(''), None)
1037 self
.assertEqual(parse_bitrate('300kbps'), 300)
1038 self
.assertEqual(parse_bitrate('1500kbps'), 1500)
1039 self
.assertEqual(parse_bitrate('300 kbps'), 300)
1041 def test_version_tuple(self
):
1042 self
.assertEqual(version_tuple('1'), (1,))
1043 self
.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
1044 self
.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
1046 def test_detect_exe_version(self
):
1047 self
.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
1048 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
1049 configuration: --prefix=/usr --extra-'''), '1.2.1')
1050 self
.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
1051 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
1052 self
.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
1053 Trying to open render node...
1054 Success at /dev/dri/renderD128.
1055 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
1057 def test_age_restricted(self
):
1058 self
.assertFalse(age_restricted(None, 10)) # unrestricted content
1059 self
.assertFalse(age_restricted(1, None)) # unrestricted policy
1060 self
.assertFalse(age_restricted(8, 10))
1061 self
.assertTrue(age_restricted(18, 14))
1062 self
.assertFalse(age_restricted(18, 18))
1064 def test_is_html(self
):
1065 self
.assertFalse(is_html(b
'\x49\x44\x43<html'))
1066 self
.assertTrue(is_html(b
'<!DOCTYPE foo>\xaaa'))
1067 self
.assertTrue(is_html( # UTF-8 with BOM
1068 b
'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
1069 self
.assertTrue(is_html( # UTF-16-LE
1070 b
'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
1072 self
.assertTrue(is_html( # UTF-16-BE
1073 b
'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
1075 self
.assertTrue(is_html( # UTF-32-BE
1076 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'))
1077 self
.assertTrue(is_html( # UTF-32-LE
1078 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'))
1080 def test_render_table(self
):
1084 [[123, 4], [9999, 51]]),
1089 def test_match_str(self
):
1090 self
.assertRaises(ValueError, match_str
, 'xy>foobar', {})
1091 self
.assertFalse(match_str('xy', {'x': 1200}
))
1092 self
.assertTrue(match_str('!xy', {'x': 1200}
))
1093 self
.assertTrue(match_str('x', {'x': 1200}
))
1094 self
.assertFalse(match_str('!x', {'x': 1200}
))
1095 self
.assertTrue(match_str('x', {'x': 0}
))
1096 self
.assertFalse(match_str('x>0', {'x': 0}
))
1097 self
.assertFalse(match_str('x>0', {}))
1098 self
.assertTrue(match_str('x>?0', {}))
1099 self
.assertTrue(match_str('x>1K', {'x': 1200}
))
1100 self
.assertFalse(match_str('x>2K', {'x': 1200}
))
1101 self
.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}
))
1102 self
.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}
))
1103 self
.assertFalse(match_str('y=a212', {'y': 'foobar42'}
))
1104 self
.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}
))
1105 self
.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}
))
1106 self
.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}
))
1107 self
.assertFalse(match_str(
1108 'like_count > 100 & dislike_count <? 50 & description',
1109 {'like_count': 90, 'description': 'foo'}
))
1110 self
.assertTrue(match_str(
1111 'like_count > 100 & dislike_count <? 50 & description',
1112 {'like_count': 190, 'description': 'foo'}
))
1113 self
.assertFalse(match_str(
1114 'like_count > 100 & dislike_count <? 50 & description',
1115 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}
))
1116 self
.assertFalse(match_str(
1117 'like_count > 100 & dislike_count <? 50 & description',
1118 {'like_count': 190, 'dislike_count': 10}
))
1119 self
.assertTrue(match_str('is_live', {'is_live': True}
))
1120 self
.assertFalse(match_str('is_live', {'is_live': False}
))
1121 self
.assertFalse(match_str('is_live', {'is_live': None}
))
1122 self
.assertFalse(match_str('is_live', {}))
1123 self
.assertFalse(match_str('!is_live', {'is_live': True}
))
1124 self
.assertTrue(match_str('!is_live', {'is_live': False}
))
1125 self
.assertTrue(match_str('!is_live', {'is_live': None}
))
1126 self
.assertTrue(match_str('!is_live', {}))
1127 self
.assertTrue(match_str('title', {'title': 'abc'}
))
1128 self
.assertTrue(match_str('title', {'title': ''}
))
1129 self
.assertFalse(match_str('!title', {'title': 'abc'}
))
1130 self
.assertFalse(match_str('!title', {'title': ''}
))
1132 def test_parse_dfxp_time_expr(self
):
1133 self
.assertEqual(parse_dfxp_time_expr(None), None)
1134 self
.assertEqual(parse_dfxp_time_expr(''), None)
1135 self
.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1136 self
.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1137 self
.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1138 self
.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1139 self
.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1141 def test_dfxp2srt(self
):
1142 dfxp_data
= '''<?xml version="1.0" encoding="UTF-8"?>
1143 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1146 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1147 <p begin="1" end="2">第二行<br/>♪♪</p>
1148 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1149 <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1150 <p begin="-1" end="-1">Ignore, two</p>
1151 <p begin="3" dur="-1">Ignored, three</p>
1154 </tt>'''.encode('utf-8')
1156 00:00:00,000 --> 00:00:01,000
1157 The following line contains Chinese characters and special symbols
1160 00:00:01,000 --> 00:00:02,000
1165 00:00:02,000 --> 00:00:03,000
1170 self
.assertEqual(dfxp2srt(dfxp_data
), srt_data
)
1172 dfxp_data_no_default_namespace
= '''<?xml version="1.0" encoding="UTF-8"?>
1173 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1176 <p begin="0" end="1">The first line</p>
1179 </tt>'''.encode('utf-8')
1181 00:00:00,000 --> 00:00:01,000
1185 self
.assertEqual(dfxp2srt(dfxp_data_no_default_namespace
), srt_data
)
1187 dfxp_data_with_style
= '''<?xml version="1.0" encoding="utf-8"?>
1188 <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">
1191 <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1192 <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1193 <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1194 <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1197 <body tts:textAlign="center" style="s0">
1199 <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1200 <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>
1201 <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1202 <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>
1205 </tt>'''.encode('utf-8')
1207 00:00:02,080 --> 00:00:05,839
1208 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1211 00:00:02,080 --> 00:00:05,839
1212 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1213 </font>part 2</font></b>
1216 00:00:05,839 --> 00:00:09,560
1217 <u><font color="lime">line 3
1221 00:00:09,560 --> 00:00:12,359
1222 <i><u><font color="yellow"><font color="lime">inner
1223 </font>style</font></u></i>
1226 self
.assertEqual(dfxp2srt(dfxp_data_with_style
), srt_data
)
1228 dfxp_data_non_utf8
= '''<?xml version="1.0" encoding="UTF-16"?>
1229 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1232 <p begin="0" end="1">Line 1</p>
1233 <p begin="1" end="2">第二行</p>
1236 </tt>'''.encode('utf-16')
1238 00:00:00,000 --> 00:00:01,000
1242 00:00:01,000 --> 00:00:02,000
1246 self
.assertEqual(dfxp2srt(dfxp_data_non_utf8
), srt_data
)
1248 def test_cli_option(self
):
1249 self
.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}
, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1250 self
.assertEqual(cli_option({'proxy': None}
, '--proxy', 'proxy'), [])
1251 self
.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1252 self
.assertEqual(cli_option({'retries': 10}
, '--retries', 'retries'), ['--retries', '10'])
1254 def test_cli_valueless_option(self
):
1255 self
.assertEqual(cli_valueless_option(
1256 {'downloader': 'external'}
, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1257 self
.assertEqual(cli_valueless_option(
1258 {'downloader': 'internal'}
, '--external-downloader', 'downloader', 'external'), [])
1259 self
.assertEqual(cli_valueless_option(
1260 {'nocheckcertificate': True}
, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1261 self
.assertEqual(cli_valueless_option(
1262 {'nocheckcertificate': False}
, '--no-check-certificate', 'nocheckcertificate'), [])
1263 self
.assertEqual(cli_valueless_option(
1264 {'checkcertificate': True}
, '--no-check-certificate', 'checkcertificate', False), [])
1265 self
.assertEqual(cli_valueless_option(
1266 {'checkcertificate': False}
, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1268 def test_cli_bool_option(self
):
1271 {'nocheckcertificate': True}
, '--no-check-certificate', 'nocheckcertificate'),
1272 ['--no-check-certificate', 'true'])
1275 {'nocheckcertificate': True}
, '--no-check-certificate', 'nocheckcertificate', separator
='='),
1276 ['--no-check-certificate=true'])
1279 {'nocheckcertificate': True}
, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1280 ['--check-certificate', 'false'])
1283 {'nocheckcertificate': True}
, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1284 ['--check-certificate=false'])
1287 {'nocheckcertificate': False}
, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1288 ['--check-certificate', 'true'])
1291 {'nocheckcertificate': False}
, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1292 ['--check-certificate=true'])
1295 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1298 def test_ohdave_rsa_encrypt(self
):
1299 N
= 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1303 ohdave_rsa_encrypt(b
'aa111222', e
, N
),
1304 '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1306 def test_pkcs1pad(self
):
1308 padded_data
= pkcs1pad(data
, 32)
1309 self
.assertEqual(padded_data
[:2], [0, 2])
1310 self
.assertEqual(padded_data
[28:], [0, 1, 2, 3])
1312 self
.assertRaises(ValueError, pkcs1pad
, data
, 8)
1314 def test_encode_base_n(self
):
1315 self
.assertEqual(encode_base_n(0, 30), '0')
1316 self
.assertEqual(encode_base_n(80, 30), '2k')
1318 custom_table
= '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1319 self
.assertEqual(encode_base_n(0, 30, custom_table
), '9')
1320 self
.assertEqual(encode_base_n(80, 30, custom_table
), '7P')
1322 self
.assertRaises(ValueError, encode_base_n
, 0, 70)
1323 self
.assertRaises(ValueError, encode_base_n
, 0, 60, custom_table
)
1325 def test_urshift(self
):
1326 self
.assertEqual(urshift(3, 1), 1)
1327 self
.assertEqual(urshift(-3, 1), 2147483646)
1329 def test_get_element_by_class(self
):
1331 <span class="foo bar">nice</span>
1334 self
.assertEqual(get_element_by_class('foo', html
), 'nice')
1335 self
.assertEqual(get_element_by_class('no-such-class', html
), None)
1337 def test_get_element_by_attribute(self
):
1339 <span class="foo bar">nice</span>
1342 self
.assertEqual(get_element_by_attribute('class', 'foo bar', html
), 'nice')
1343 self
.assertEqual(get_element_by_attribute('class', 'foo', html
), None)
1344 self
.assertEqual(get_element_by_attribute('class', 'no-such-foo', html
), None)
1347 <div itemprop="author" itemscope>foo</div>
1350 self
.assertEqual(get_element_by_attribute('itemprop', 'author', html
), 'foo')
1352 def test_get_elements_by_class(self
):
1354 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1357 self
.assertEqual(get_elements_by_class('foo', html
), ['nice', 'also nice'])
1358 self
.assertEqual(get_elements_by_class('no-such-class', html
), [])
1360 def test_get_elements_by_attribute(self
):
1362 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1365 self
.assertEqual(get_elements_by_attribute('class', 'foo bar', html
), ['nice', 'also nice'])
1366 self
.assertEqual(get_elements_by_attribute('class', 'foo', html
), [])
1367 self
.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html
), [])
1370 if __name__
== '__main__':