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,
99 from youtube_dl.compat import (
101 compat_etree_fromstring,
110 class TestUtil(unittest.TestCase):
111 def test_timeconvert(self):
112 self.assertTrue(timeconvert('') is None)
113 self.assertTrue(timeconvert('bougrg') is None)
115 def test_sanitize_filename(self):
116 self.assertEqual(sanitize_filename('abc'), 'abc')
117 self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
119 self.assertEqual(sanitize_filename('123'), '123')
121 self.assertEqual('abc_de', sanitize_filename('abc/de'))
122 self.assertFalse('/' in sanitize_filename('abc/de///'))
124 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
125 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
126 self.assertEqual('yes no', sanitize_filename('yes? no'))
127 self.assertEqual('this - that', sanitize_filename('this: that'))
129 self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
131 self.assertEqual(sanitize_filename(aumlaut), aumlaut)
132 tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
133 self.assertEqual(sanitize_filename(tests), tests)
136 sanitize_filename('New World record at 0:12:34'),
137 'New World record at 0_12_34')
139 self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
140 self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
141 self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
142 self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
146 for fbc in forbidden:
147 self.assertTrue(fbc not in sanitize_filename(fc))
149 def test_sanitize_filename_restricted(self):
150 self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
151 self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
153 self.assertEqual(sanitize_filename('123', restricted=True), '123')
155 self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
156 self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
158 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
159 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
160 self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
161 self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
163 tests = 'aäb\u4e2d\u56fd\u7684c'
164 self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
165 self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
167 forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
169 for fbc in forbidden:
170 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
172 # Handle a common case more neatly
173 self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
174 self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
175 # .. but make sure the file name is never empty
176 self.assertTrue(sanitize_filename('-', restricted=True) != '')
177 self.assertTrue(sanitize_filename(':', restricted=True) != '')
179 self.assertEqual(sanitize_filename(
180 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
181 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYPssaaaaaaaeceeeeiiiionooooooooeuuuuuypy')
183 def test_sanitize_ids(self):
184 self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
185 self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
186 self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
188 def test_sanitize_path(self):
189 if sys.platform != 'win32':
192 self.assertEqual(sanitize_path('abc'), 'abc')
193 self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
194 self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
195 self.assertEqual(sanitize_path('abc|def'), 'abc#def')
196 self.assertEqual(sanitize_path('<>:"|?*'), '#######')
197 self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
198 self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
200 self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
201 self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
203 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
204 self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
205 self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
206 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
209 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
210 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
213 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
214 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
215 self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
216 self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
217 self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
219 self.assertEqual(sanitize_path('../abc'), '..\\abc')
220 self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
221 self.assertEqual(sanitize_path('./abc'), 'abc')
222 self.assertEqual(sanitize_path('./../abc'), '..\\abc')
224 def test_sanitize_url(self):
225 self.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
226 self.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
227 self.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
228 self.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')
230 def test_expand_path(self):
232 return '%{0}%'.format(var) if sys.platform == 'win32' else '${0}'.format(var)
234 compat_setenv('YOUTUBE_DL_EXPATH_PATH', 'expanded')
235 self.assertEqual(expand_path(env('YOUTUBE_DL_EXPATH_PATH')), 'expanded')
236 self.assertEqual(expand_path(env('HOME')), compat_getenv('HOME'))
237 self.assertEqual(expand_path('~'), compat_getenv('HOME'))
239 expand_path('~/%s' % env('YOUTUBE_DL_EXPATH_PATH')),
240 '%s/expanded' % compat_getenv('HOME'))
242 def test_prepend_extension(self):
243 self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
244 self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
245 self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
246 self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
247 self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
248 self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
250 def test_replace_extension(self):
251 self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
252 self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
253 self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
254 self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
255 self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
256 self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
258 def test_remove_start(self):
259 self.assertEqual(remove_start(None, 'A - '), None)
260 self.assertEqual(remove_start('A - B', 'A - '), 'B')
261 self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
263 def test_remove_end(self):
264 self.assertEqual(remove_end(None, ' - B'), None)
265 self.assertEqual(remove_end('A - B', ' - B'), 'A')
266 self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
268 def test_remove_quotes(self):
269 self.assertEqual(remove_quotes(None), None)
270 self.assertEqual(remove_quotes('"'), '"')
271 self.assertEqual(remove_quotes("'"), "'")
272 self.assertEqual(remove_quotes(';'), ';')
273 self.assertEqual(remove_quotes('";'), '";')
274 self.assertEqual(remove_quotes('""'), '')
275 self.assertEqual(remove_quotes('";"'), ';')
277 def test_ordered_set(self):
278 self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
279 self.assertEqual(orderedSet([]), [])
280 self.assertEqual(orderedSet([1]), [1])
281 # keep the list ordered
282 self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
284 def test_unescape_html(self):
285 self.assertEqual(unescapeHTML('%20;'), '%20;')
286 self.assertEqual(unescapeHTML('/'), '/')
287 self.assertEqual(unescapeHTML('/'), '/')
288 self.assertEqual(unescapeHTML('é'), 'é')
289 self.assertEqual(unescapeHTML('�'), '�')
290 self.assertEqual(unescapeHTML('&a"'), '&a"')
292 self.assertEqual(unescapeHTML('.''), '.\'')
294 def test_date_from_str(self):
295 self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
296 self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
297 self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
298 self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
299 self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
301 def test_daterange(self):
302 _20century = DateRange("19000101", "20000101")
303 self.assertFalse("17890714" in _20century)
304 _ac = DateRange("00010101")
305 self.assertTrue("19690721" in _ac)
306 _firstmilenium = DateRange(end="10000101")
307 self.assertTrue("07110427" in _firstmilenium)
309 def test_unified_dates(self):
310 self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
311 self.assertEqual(unified_strdate('8/7/2009'), '20090708')
312 self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
313 self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
314 self.assertEqual(unified_strdate('1968 12 10'), '19681210')
315 self.assertEqual(unified_strdate('1968-12-10'), '19681210')
316 self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
318 unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
321 unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
323 self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
324 self.assertEqual(unified_strdate('25-09-2014'), '20140925')
325 self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
326 self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
327 self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
328 self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
329 self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
330 self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
332 def test_unified_timestamps(self):
333 self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
334 self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
335 self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
336 self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
337 self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
338 self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
339 self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
341 unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
344 unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
346 self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
347 self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
348 self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
349 self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
350 self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
351 self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
352 self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
353 self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
354 self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
356 def test_determine_ext(self):
357 self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
358 self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
359 self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
360 self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
361 self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
363 def test_find_xpath_attr(self):
371 doc = compat_etree_fromstring(testxml)
373 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
374 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
375 self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
376 self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
377 self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
378 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
379 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
380 self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
381 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
382 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
383 self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
385 def test_xpath_with_ns(self):
386 testxml = '''<root xmlns:media="http://example.com/">
388 <media:author>The Author</media:author>
389 <url>http://server.com/download.mp3</url>
392 doc = compat_etree_fromstring(testxml)
393 find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
394 self.assertTrue(find('media:song') is not None)
395 self.assertEqual(find('media:song/media:author').text, 'The Author')
396 self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
398 def test_xpath_element(self):
399 doc = xml.etree.ElementTree.Element('root')
400 div = xml.etree.ElementTree.SubElement(doc, 'div')
401 p = xml.etree.ElementTree.SubElement(div, 'p')
403 self.assertEqual(xpath_element(doc, 'div/p'), p)
404 self.assertEqual(xpath_element(doc, ['div/p']), p)
405 self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
406 self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
407 self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
408 self.assertTrue(xpath_element(doc, 'div/bar') is None)
409 self.assertTrue(xpath_element(doc, ['div/bar']) is None)
410 self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
411 self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
412 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
413 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
415 def test_xpath_text(self):
421 doc = compat_etree_fromstring(testxml)
422 self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
423 self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
424 self.assertTrue(xpath_text(doc, 'div/bar') is None)
425 self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
427 def test_xpath_attr(self):
433 doc = compat_etree_fromstring(testxml)
434 self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
435 self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
436 self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
437 self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
438 self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
439 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
440 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
442 def test_smuggle_url(self):
443 data = {"ö": "ö", "abc": [3]}
444 url = 'https://foo.bar/baz?x=y#a'
445 smug_url = smuggle_url(url, data)
446 unsmug_url, unsmug_data = unsmuggle_url(smug_url)
447 self.assertEqual(url, unsmug_url)
448 self.assertEqual(data, unsmug_data)
450 res_url, res_data = unsmuggle_url(url)
451 self.assertEqual(res_url, url)
452 self.assertEqual(res_data, None)
454 smug_url = smuggle_url(url, {'a': 'b'})
455 smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
456 res_url, res_data = unsmuggle_url(smug_smug_url)
457 self.assertEqual(res_url, url)
458 self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
460 def test_shell_quote(self):
461 args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
464 """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
466 def test_str_to_int(self):
467 self.assertEqual(str_to_int('123,456'), 123456)
468 self.assertEqual(str_to_int('123.456'), 123456)
470 def test_url_basename(self):
471 self.assertEqual(url_basename('http://foo.de/'), '')
472 self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
473 self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
474 self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
475 self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
477 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
480 def test_base_url(self):
481 self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
482 self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
483 self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
484 self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
485 self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
487 def test_urljoin(self):
488 self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
489 self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
490 self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
491 self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
492 self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
493 self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
494 self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
495 self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
496 self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
497 self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
498 self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
499 self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//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(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
502 self.assertEqual(urljoin('http://foo.de/', None), None)
503 self.assertEqual(urljoin('http://foo.de/', ''), None)
504 self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
505 self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
507 def test_parse_age_limit(self):
508 self.assertEqual(parse_age_limit(None), None)
509 self.assertEqual(parse_age_limit(False), None)
510 self.assertEqual(parse_age_limit('invalid'), None)
511 self.assertEqual(parse_age_limit(0), 0)
512 self.assertEqual(parse_age_limit(18), 18)
513 self.assertEqual(parse_age_limit(21), 21)
514 self.assertEqual(parse_age_limit(22), None)
515 self.assertEqual(parse_age_limit('18'), 18)
516 self.assertEqual(parse_age_limit('18+'), 18)
517 self.assertEqual(parse_age_limit('PG-13'), 13)
518 self.assertEqual(parse_age_limit('TV-14'), 14)
519 self.assertEqual(parse_age_limit('TV-MA'), 17)
521 def test_parse_duration(self):
522 self.assertEqual(parse_duration(None), None)
523 self.assertEqual(parse_duration(False), None)
524 self.assertEqual(parse_duration('invalid'), None)
525 self.assertEqual(parse_duration('1'), 1)
526 self.assertEqual(parse_duration('1337:12'), 80232)
527 self.assertEqual(parse_duration('9:12:43'), 33163)
528 self.assertEqual(parse_duration('12:00'), 720)
529 self.assertEqual(parse_duration('00:01:01'), 61)
530 self.assertEqual(parse_duration('x:y'), None)
531 self.assertEqual(parse_duration('3h11m53s'), 11513)
532 self.assertEqual(parse_duration('3h 11m 53s'), 11513)
533 self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
534 self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
535 self.assertEqual(parse_duration('62m45s'), 3765)
536 self.assertEqual(parse_duration('6m59s'), 419)
537 self.assertEqual(parse_duration('49s'), 49)
538 self.assertEqual(parse_duration('0h0m0s'), 0)
539 self.assertEqual(parse_duration('0m0s'), 0)
540 self.assertEqual(parse_duration('0s'), 0)
541 self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
542 self.assertEqual(parse_duration('T30M38S'), 1838)
543 self.assertEqual(parse_duration('5 s'), 5)
544 self.assertEqual(parse_duration('3 min'), 180)
545 self.assertEqual(parse_duration('2.5 hours'), 9000)
546 self.assertEqual(parse_duration('02:03:04'), 7384)
547 self.assertEqual(parse_duration('01:02:03:04'), 93784)
548 self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
549 self.assertEqual(parse_duration('87 Min.'), 5220)
550 self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
551 self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
552 self.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)
554 def test_fix_xml_ampersands(self):
556 fix_xml_ampersands('"&x=y&z=a'), '"&x=y&z=a')
558 fix_xml_ampersands('"&x=y&wrong;&z=a'),
559 '"&x=y&wrong;&z=a')
561 fix_xml_ampersands('&'><"'),
562 '&'><"')
564 fix_xml_ampersands('Ӓ᪼'), 'Ӓ᪼')
565 self.assertEqual(fix_xml_ampersands('&#&#'), '&#&#')
567 def test_paged_list(self):
568 def testPL(size, pagesize, sliceargs, expected):
569 def get_page(pagenum):
570 firstid = pagenum * pagesize
571 upto = min(size, pagenum * pagesize + pagesize)
572 for i in range(firstid, upto):
575 pl = OnDemandPagedList(get_page, pagesize)
576 got = pl.getslice(*sliceargs)
577 self.assertEqual(got, expected)
579 iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
580 got = iapl.getslice(*sliceargs)
581 self.assertEqual(got, expected)
583 testPL(5, 2, (), [0, 1, 2, 3, 4])
584 testPL(5, 2, (1,), [1, 2, 3, 4])
585 testPL(5, 2, (2,), [2, 3, 4])
586 testPL(5, 2, (4,), [4])
587 testPL(5, 2, (0, 3), [0, 1, 2])
588 testPL(5, 2, (1, 4), [1, 2, 3])
589 testPL(5, 2, (2, 99), [2, 3, 4])
590 testPL(5, 2, (20, 99), [])
592 def test_read_batch_urls(self):
593 f = io.StringIO('''\xef\xbb\xbf foo
596 # More after this line\r
599 self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
601 def test_urlencode_postdata(self):
602 data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
603 self.assertTrue(isinstance(data, bytes))
605 def test_update_url_query(self):
607 return compat_parse_qs(compat_urlparse.urlparse(url).query)
608 self.assertEqual(query_dict(update_url_query(
609 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
610 query_dict('http://example.com/path?quality=HD&format=mp4'))
611 self.assertEqual(query_dict(update_url_query(
612 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
613 query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
614 self.assertEqual(query_dict(update_url_query(
615 'http://example.com/path', {'fields': 'id,formats,subtitles'})),
616 query_dict('http://example.com/path?fields=id,formats,subtitles'))
617 self.assertEqual(query_dict(update_url_query(
618 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
619 query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
620 self.assertEqual(query_dict(update_url_query(
621 'http://example.com/path?manifest=f4m', {'manifest': []})),
622 query_dict('http://example.com/path'))
623 self.assertEqual(query_dict(update_url_query(
624 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
625 query_dict('http://example.com/path?system=LINUX'))
626 self.assertEqual(query_dict(update_url_query(
627 'http://example.com/path', {'fields': b'id,formats,subtitles'})),
628 query_dict('http://example.com/path?fields=id,formats,subtitles'))
629 self.assertEqual(query_dict(update_url_query(
630 'http://example.com/path', {'width': 1080, 'height': 720})),
631 query_dict('http://example.com/path?width=1080&height=720'))
632 self.assertEqual(query_dict(update_url_query(
633 'http://example.com/path', {'bitrate': 5020.43})),
634 query_dict('http://example.com/path?bitrate=5020.43'))
635 self.assertEqual(query_dict(update_url_query(
636 'http://example.com/path', {'test': '第二行тест'})),
637 query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
639 def test_multipart_encode(self):
641 multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0],
642 b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
644 multipart_encode({'欄位'.encode('utf-8'): '值'.encode('utf-8')}, boundary='AAAAAA')[0],
645 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')
647 ValueError, multipart_encode, {b'field': b'value'}, boundary='value')
649 def test_dict_get(self):
657 d = FALSE_VALUES.copy()
659 self.assertEqual(dict_get(d, 'a'), 42)
660 self.assertEqual(dict_get(d, 'b'), None)
661 self.assertEqual(dict_get(d, 'b', 42), 42)
662 self.assertEqual(dict_get(d, ('a', )), 42)
663 self.assertEqual(dict_get(d, ('b', 'a', )), 42)
664 self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
665 self.assertEqual(dict_get(d, ('b', 'c', )), None)
666 self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
667 for key, false_value in FALSE_VALUES.items():
668 self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
669 self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
671 def test_encode_compat_str(self):
672 self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
673 self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
675 def test_parse_iso8601(self):
676 self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
677 self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
678 self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
679 self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
680 self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
681 self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
683 def test_strip_jsonp(self):
684 stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
685 d = json.loads(stripped)
686 self.assertEqual(d, [{"id": "532cb", "x": 3}])
688 stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
689 d = json.loads(stripped)
690 self.assertEqual(d, {'STATUS': 'OK'})
692 stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
693 d = json.loads(stripped)
694 self.assertEqual(d, {'status': 'success'})
696 stripped = strip_jsonp('window.cb && window.cb({"status": "success"});')
697 d = json.loads(stripped)
698 self.assertEqual(d, {'status': 'success'})
700 stripped = strip_jsonp('window.cb && cb({"status": "success"});')
701 d = json.loads(stripped)
702 self.assertEqual(d, {'status': 'success'})
704 def test_uppercase_escape(self):
705 self.assertEqual(uppercase_escape('aä'), 'aä')
706 self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
708 def test_lowercase_escape(self):
709 self.assertEqual(lowercase_escape('aä'), 'aä')
710 self.assertEqual(lowercase_escape('\\u0026'), '&')
712 def test_limit_length(self):
713 self.assertEqual(limit_length(None, 12), None)
714 self.assertEqual(limit_length('foo', 12), 'foo')
716 limit_length('foo bar baz asd', 12).startswith('foo bar'))
717 self.assertTrue('...' in limit_length('foo bar baz asd', 12))
719 def test_mimetype2ext(self):
720 self.assertEqual(mimetype2ext(None), None)
721 self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
722 self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
723 self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
724 self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
725 self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
727 def test_month_by_name(self):
728 self.assertEqual(month_by_name(None), None)
729 self.assertEqual(month_by_name('December', 'en'), 12)
730 self.assertEqual(month_by_name('décembre', 'fr'), 12)
731 self.assertEqual(month_by_name('December'), 12)
732 self.assertEqual(month_by_name('décembre'), None)
733 self.assertEqual(month_by_name('Unknown', 'unknown'), None)
735 def test_parse_codecs(self):
736 self.assertEqual(parse_codecs(''), {})
737 self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
738 'vcodec': 'avc1.77.30',
739 'acodec': 'mp4a.40.2',
741 self.assertEqual(parse_codecs('mp4a.40.2'), {
743 'acodec': 'mp4a.40.2',
745 self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
746 'vcodec': 'avc1.42001e',
747 'acodec': 'mp4a.40.5',
749 self.assertEqual(parse_codecs('avc3.640028'), {
750 'vcodec': 'avc3.640028',
753 self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
758 def test_escape_rfc3986(self):
759 reserved = "!*'();:@&=+$,/?#[]"
760 unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
761 self.assertEqual(escape_rfc3986(reserved), reserved)
762 self.assertEqual(escape_rfc3986(unreserved), unreserved)
763 self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
764 self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
765 self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
766 self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
768 def test_escape_url(self):
770 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
771 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
774 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
775 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
778 escape_url('http://тест.рф/фрагмент'),
779 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
782 escape_url('http://тест.рф/абв?абв=абв#абв'),
783 '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'
785 self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
787 def test_js_to_json_realworld(self):
789 'clip':{'provider':'pseudo'}
791 self.assertEqual(js_to_json(inp), '''{
792 "clip":{"provider":"pseudo"}
794 json.loads(js_to_json(inp))
797 'playlist':[{'controls':{'all':null}}]
799 self.assertEqual(js_to_json(inp), '''{
800 "playlist":[{"controls":{"all":null}}]
803 inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
804 self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
806 inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
807 json_code = js_to_json(inp)
808 self.assertEqual(json.loads(json_code), json.loads(inp))
811 0:{src:'skipped', type: 'application/dash+xml'},
812 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
814 self.assertEqual(js_to_json(inp), '''{
815 "0":{"src":"skipped", "type": "application/dash+xml"},
816 "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
819 inp = '''{"foo":101}'''
820 self.assertEqual(js_to_json(inp), '''{"foo":101}''')
822 inp = '''{"duration": "00:01:07"}'''
823 self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
825 inp = '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}'''
826 self.assertEqual(js_to_json(inp), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''')
828 def test_js_to_json_edgecases(self):
829 on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
830 self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
832 on = js_to_json('{"abc": true}')
833 self.assertEqual(json.loads(on), {'abc': True})
835 # Ignore JavaScript code as well
842 self.assertEqual(d['x'], 1)
843 self.assertEqual(d['y'], 'a')
845 on = js_to_json('["abc", "def",]')
846 self.assertEqual(json.loads(on), ['abc', 'def'])
848 on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
849 self.assertEqual(json.loads(on), ['abc', 'def'])
851 on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
852 self.assertEqual(json.loads(on), ['abc', 'def'])
854 on = js_to_json('{"abc": "def",}')
855 self.assertEqual(json.loads(on), {'abc': 'def'})
857 on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
858 self.assertEqual(json.loads(on), {'abc': 'def'})
860 on = js_to_json('{ 0: /* " \n */ ",]" , }')
861 self.assertEqual(json.loads(on), {'0': ',]'})
863 on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
864 self.assertEqual(json.loads(on), {'0': ',]'})
866 on = js_to_json('{ 0: // comment\n1 }')
867 self.assertEqual(json.loads(on), {'0': 1})
869 on = js_to_json(r'["<p>x<\/p>"]')
870 self.assertEqual(json.loads(on), ['<p>x</p>'])
872 on = js_to_json(r'["\xaa"]')
873 self.assertEqual(json.loads(on), ['\u00aa'])
875 on = js_to_json("['a\\\nb']")
876 self.assertEqual(json.loads(on), ['ab'])
878 on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
879 self.assertEqual(json.loads(on), ['ab'])
881 on = js_to_json('{0xff:0xff}')
882 self.assertEqual(json.loads(on), {'255': 255})
884 on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
885 self.assertEqual(json.loads(on), {'255': 255})
887 on = js_to_json('{077:077}')
888 self.assertEqual(json.loads(on), {'63': 63})
890 on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
891 self.assertEqual(json.loads(on), {'63': 63})
893 on = js_to_json('{42:42}')
894 self.assertEqual(json.loads(on), {'42': 42})
896 on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
897 self.assertEqual(json.loads(on), {'42': 42})
899 on = js_to_json('{42:4.2e1}')
900 self.assertEqual(json.loads(on), {'42': 42.0})
902 def test_js_to_json_malformed(self):
903 self.assertEqual(js_to_json('42a1'), '42"a1"')
904 self.assertEqual(js_to_json('42a-1'), '42"a"-1')
906 def test_extract_attributes(self):
907 self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
908 self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
909 self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
910 self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
911 self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
912 self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
913 self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
914 self.assertEqual(extract_attributes('<e x="&">'), {'x': '&'}) # XML
915 self.assertEqual(extract_attributes('<e x=""">'), {'x': '"'})
916 self.assertEqual(extract_attributes('<e x="£">'), {'x': '£'}) # HTML 3.2
917 self.assertEqual(extract_attributes('<e x="λ">'), {'x': 'λ'}) # HTML 4.0
918 self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
919 self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
920 self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
921 self.assertEqual(extract_attributes('<e x >'), {'x': None})
922 self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
923 self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
924 self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
925 self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
926 self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
927 self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
928 self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
929 self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}) # Names lowercased
930 self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
931 self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
932 self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
933 self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
934 self.assertEqual(extract_attributes('<e x="décomposé">'), {'x': 'décompose\u0301'})
935 # "Narrow" Python builds don't support unicode code points outside BMP.
938 supports_outside_bmp = True
940 supports_outside_bmp = False
941 if supports_outside_bmp:
942 self.assertEqual(extract_attributes('<e x="Smile 😀!">'), {'x': 'Smile \U0001f600!'})
943 # Malformed HTML should not break attributes extraction on older Python
944 self.assertEqual(extract_attributes('<mal"formed/>'), {})
946 def test_clean_html(self):
947 self.assertEqual(clean_html('a:\nb'), 'a: b')
948 self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
949 self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
951 def test_intlist_to_bytes(self):
953 intlist_to_bytes([0, 1, 127, 128, 255]),
954 b'\x00\x01\x7f\x80\xff')
956 def test_args_to_str(self):
958 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
959 'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""'
962 def test_parse_filesize(self):
963 self.assertEqual(parse_filesize(None), None)
964 self.assertEqual(parse_filesize(''), None)
965 self.assertEqual(parse_filesize('91 B'), 91)
966 self.assertEqual(parse_filesize('foobar'), None)
967 self.assertEqual(parse_filesize('2 MiB'), 2097152)
968 self.assertEqual(parse_filesize('5 GB'), 5000000000)
969 self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
970 self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
971 self.assertEqual(parse_filesize('1,24 KB'), 1240)
972 self.assertEqual(parse_filesize('1,24 kb'), 1240)
973 self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
975 def test_parse_count(self):
976 self.assertEqual(parse_count(None), None)
977 self.assertEqual(parse_count(''), None)
978 self.assertEqual(parse_count('0'), 0)
979 self.assertEqual(parse_count('1000'), 1000)
980 self.assertEqual(parse_count('1.000'), 1000)
981 self.assertEqual(parse_count('1.1k'), 1100)
982 self.assertEqual(parse_count('1.1kk'), 1100000)
983 self.assertEqual(parse_count('1.1kk '), 1100000)
984 self.assertEqual(parse_count('1.1kk views'), 1100000)
986 def test_parse_resolution(self):
987 self.assertEqual(parse_resolution(None), {})
988 self.assertEqual(parse_resolution(''), {})
989 self.assertEqual(parse_resolution('1920x1080'), {'width': 1920, 'height': 1080})
990 self.assertEqual(parse_resolution('1920×1080'), {'width': 1920, 'height': 1080})
991 self.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080})
992 self.assertEqual(parse_resolution('720p'), {'height': 720})
993 self.assertEqual(parse_resolution('4k'), {'height': 2160})
994 self.assertEqual(parse_resolution('8K'), {'height': 4320})
996 def test_version_tuple(self):
997 self.assertEqual(version_tuple('1'), (1,))
998 self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
999 self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
1001 def test_detect_exe_version(self):
1002 self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
1003 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
1004 configuration: --prefix=/usr --extra-'''), '1.2.1')
1005 self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
1006 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
1007 self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
1008 Trying to open render node...
1009 Success at /dev/dri/renderD128.
1010 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
1012 def test_age_restricted(self):
1013 self.assertFalse(age_restricted(None, 10)) # unrestricted content
1014 self.assertFalse(age_restricted(1, None)) # unrestricted policy
1015 self.assertFalse(age_restricted(8, 10))
1016 self.assertTrue(age_restricted(18, 14))
1017 self.assertFalse(age_restricted(18, 18))
1019 def test_is_html(self):
1020 self.assertFalse(is_html(b'\x49\x44\x43<html'))
1021 self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
1022 self.assertTrue(is_html( # UTF-8 with BOM
1023 b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
1024 self.assertTrue(is_html( # UTF-16-LE
1025 b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
1027 self.assertTrue(is_html( # UTF-16-BE
1028 b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
1030 self.assertTrue(is_html( # UTF-32-BE
1031 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'))
1032 self.assertTrue(is_html( # UTF-32-LE
1033 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'))
1035 def test_render_table(self):
1039 [[123, 4], [9999, 51]]),
1044 def test_match_str(self):
1045 self.assertRaises(ValueError, match_str, 'xy>foobar', {})
1046 self.assertFalse(match_str('xy', {'x': 1200}))
1047 self.assertTrue(match_str('!xy', {'x': 1200}))
1048 self.assertTrue(match_str('x', {'x': 1200}))
1049 self.assertFalse(match_str('!x', {'x': 1200}))
1050 self.assertTrue(match_str('x', {'x': 0}))
1051 self.assertFalse(match_str('x>0', {'x': 0}))
1052 self.assertFalse(match_str('x>0', {}))
1053 self.assertTrue(match_str('x>?0', {}))
1054 self.assertTrue(match_str('x>1K', {'x': 1200}))
1055 self.assertFalse(match_str('x>2K', {'x': 1200}))
1056 self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
1057 self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
1058 self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
1059 self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
1060 self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
1061 self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
1062 self.assertFalse(match_str(
1063 'like_count > 100 & dislike_count <? 50 & description',
1064 {'like_count': 90, 'description': 'foo'}))
1065 self.assertTrue(match_str(
1066 'like_count > 100 & dislike_count <? 50 & description',
1067 {'like_count': 190, 'description': 'foo'}))
1068 self.assertFalse(match_str(
1069 'like_count > 100 & dislike_count <? 50 & description',
1070 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
1071 self.assertFalse(match_str(
1072 'like_count > 100 & dislike_count <? 50 & description',
1073 {'like_count': 190, 'dislike_count': 10}))
1075 def test_parse_dfxp_time_expr(self):
1076 self.assertEqual(parse_dfxp_time_expr(None), None)
1077 self.assertEqual(parse_dfxp_time_expr(''), None)
1078 self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1079 self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1080 self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1081 self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1082 self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1084 def test_dfxp2srt(self):
1085 dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
1086 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1089 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1090 <p begin="1" end="2">第二行<br/>♪♪</p>
1091 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1092 <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1093 <p begin="-1" end="-1">Ignore, two</p>
1094 <p begin="3" dur="-1">Ignored, three</p>
1097 </tt>'''.encode('utf-8')
1099 00:00:00,000 --> 00:00:01,000
1100 The following line contains Chinese characters and special symbols
1103 00:00:01,000 --> 00:00:02,000
1108 00:00:02,000 --> 00:00:03,000
1113 self.assertEqual(dfxp2srt(dfxp_data), srt_data)
1115 dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
1116 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1119 <p begin="0" end="1">The first line</p>
1122 </tt>'''.encode('utf-8')
1124 00:00:00,000 --> 00:00:01,000
1128 self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
1130 dfxp_data_with_style = '''<?xml version="1.0" encoding="utf-8"?>
1131 <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">
1134 <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1135 <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1136 <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1137 <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1140 <body tts:textAlign="center" style="s0">
1142 <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1143 <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>
1144 <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1145 <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>
1148 </tt>'''.encode('utf-8')
1150 00:00:02,080 --> 00:00:05,839
1151 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1154 00:00:02,080 --> 00:00:05,839
1155 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1156 </font>part 2</font></b>
1159 00:00:05,839 --> 00:00:09,560
1160 <u><font color="lime">line 3
1164 00:00:09,560 --> 00:00:12,359
1165 <i><u><font color="yellow"><font color="lime">inner
1166 </font>style</font></u></i>
1169 self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)
1171 dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?>
1172 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1175 <p begin="0" end="1">Line 1</p>
1176 <p begin="1" end="2">第二行</p>
1179 </tt>'''.encode('utf-16')
1181 00:00:00,000 --> 00:00:01,000
1185 00:00:01,000 --> 00:00:02,000
1189 self.assertEqual(dfxp2srt(dfxp_data_non_utf8), srt_data)
1191 def test_cli_option(self):
1192 self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1193 self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
1194 self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1195 self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
1197 def test_cli_valueless_option(self):
1198 self.assertEqual(cli_valueless_option(
1199 {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1200 self.assertEqual(cli_valueless_option(
1201 {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
1202 self.assertEqual(cli_valueless_option(
1203 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1204 self.assertEqual(cli_valueless_option(
1205 {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
1206 self.assertEqual(cli_valueless_option(
1207 {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
1208 self.assertEqual(cli_valueless_option(
1209 {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1211 def test_cli_bool_option(self):
1214 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1215 ['--no-check-certificate', 'true'])
1218 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1219 ['--no-check-certificate=true'])
1222 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1223 ['--check-certificate', 'false'])
1226 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1227 ['--check-certificate=false'])
1230 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1231 ['--check-certificate', 'true'])
1234 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1235 ['--check-certificate=true'])
1238 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1241 def test_ohdave_rsa_encrypt(self):
1242 N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1246 ohdave_rsa_encrypt(b'aa111222', e, N),
1247 '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1249 def test_pkcs1pad(self):
1251 padded_data = pkcs1pad(data, 32)
1252 self.assertEqual(padded_data[:2], [0, 2])
1253 self.assertEqual(padded_data[28:], [0, 1, 2, 3])
1255 self.assertRaises(ValueError, pkcs1pad, data, 8)
1257 def test_encode_base_n(self):
1258 self.assertEqual(encode_base_n(0, 30), '0')
1259 self.assertEqual(encode_base_n(80, 30), '2k')
1261 custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1262 self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1263 self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1265 self.assertRaises(ValueError, encode_base_n, 0, 70)
1266 self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1268 def test_urshift(self):
1269 self.assertEqual(urshift(3, 1), 1)
1270 self.assertEqual(urshift(-3, 1), 2147483646)
1272 def test_get_element_by_class(self):
1274 <span class="foo bar">nice</span>
1277 self.assertEqual(get_element_by_class('foo', html), 'nice')
1278 self.assertEqual(get_element_by_class('no-such-class', html), None)
1280 def test_get_element_by_attribute(self):
1282 <span class="foo bar">nice</span>
1285 self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
1286 self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
1287 self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
1290 <div itemprop="author" itemscope>foo</div>
1293 self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo')
1295 def test_get_elements_by_class(self):
1297 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1300 self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
1301 self.assertEqual(get_elements_by_class('no-such-class', html), [])
1303 def test_get_elements_by_attribute(self):
1305 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1308 self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
1309 self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
1310 self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
1313 if __name__ == '__main__':