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 (
81 from youtube_dl.compat import (
83 compat_etree_fromstring,
89 class TestUtil(unittest.TestCase):
90 def test_timeconvert(self):
91 self.assertTrue(timeconvert('') is None)
92 self.assertTrue(timeconvert('bougrg') is None)
94 def test_sanitize_filename(self):
95 self.assertEqual(sanitize_filename('abc'), 'abc')
96 self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
98 self.assertEqual(sanitize_filename('123'), '123')
100 self.assertEqual('abc_de', sanitize_filename('abc/de'))
101 self.assertFalse('/' in sanitize_filename('abc/de///'))
103 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
104 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
105 self.assertEqual('yes no', sanitize_filename('yes? no'))
106 self.assertEqual('this - that', sanitize_filename('this: that'))
108 self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
110 self.assertEqual(sanitize_filename(aumlaut), aumlaut)
111 tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
112 self.assertEqual(sanitize_filename(tests), tests)
115 sanitize_filename('New World record at 0:12:34'),
116 'New World record at 0_12_34')
118 self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
119 self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
120 self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
121 self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
125 for fbc in forbidden:
126 self.assertTrue(fbc not in sanitize_filename(fc))
128 def test_sanitize_filename_restricted(self):
129 self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
130 self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
132 self.assertEqual(sanitize_filename('123', restricted=True), '123')
134 self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
135 self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
137 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
138 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
139 self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
140 self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
142 tests = 'aäb\u4e2d\u56fd\u7684c'
143 self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
144 self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
146 forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
148 for fbc in forbidden:
149 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
151 # Handle a common case more neatly
152 self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
153 self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
154 # .. but make sure the file name is never empty
155 self.assertTrue(sanitize_filename('-', restricted=True) != '')
156 self.assertTrue(sanitize_filename(':', restricted=True) != '')
158 self.assertEqual(sanitize_filename(
159 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ', restricted=True),
160 'AAAAAAAECEEEEIIIIDNOOOOOOUUUUYPssaaaaaaaeceeeeiiiionoooooouuuuypy')
162 def test_sanitize_ids(self):
163 self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
164 self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
165 self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
167 def test_sanitize_path(self):
168 if sys.platform != 'win32':
171 self.assertEqual(sanitize_path('abc'), 'abc')
172 self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
173 self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
174 self.assertEqual(sanitize_path('abc|def'), 'abc#def')
175 self.assertEqual(sanitize_path('<>:"|?*'), '#######')
176 self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
177 self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
179 self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
180 self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
182 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
183 self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
184 self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
185 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
188 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
189 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
192 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
193 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
194 self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
195 self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
196 self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
198 self.assertEqual(sanitize_path('../abc'), '..\\abc')
199 self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
200 self.assertEqual(sanitize_path('./abc'), 'abc')
201 self.assertEqual(sanitize_path('./../abc'), '..\\abc')
203 def test_prepend_extension(self):
204 self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
205 self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
206 self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
207 self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
208 self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
209 self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
211 def test_replace_extension(self):
212 self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
213 self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
214 self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
215 self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
216 self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
217 self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
219 def test_remove_quotes(self):
220 self.assertEqual(remove_quotes(None), None)
221 self.assertEqual(remove_quotes('"'), '"')
222 self.assertEqual(remove_quotes("'"), "'")
223 self.assertEqual(remove_quotes(';'), ';')
224 self.assertEqual(remove_quotes('";'), '";')
225 self.assertEqual(remove_quotes('""'), '')
226 self.assertEqual(remove_quotes('";"'), ';')
228 def test_ordered_set(self):
229 self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
230 self.assertEqual(orderedSet([]), [])
231 self.assertEqual(orderedSet([1]), [1])
232 # keep the list ordered
233 self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
235 def test_unescape_html(self):
236 self.assertEqual(unescapeHTML('%20;'), '%20;')
237 self.assertEqual(unescapeHTML('/'), '/')
238 self.assertEqual(unescapeHTML('/'), '/')
239 self.assertEqual(unescapeHTML('é'), 'é')
240 self.assertEqual(unescapeHTML('�'), '�')
242 def test_date_from_str(self):
243 self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
244 self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
245 self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
246 self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
247 self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
249 def test_daterange(self):
250 _20century = DateRange("19000101", "20000101")
251 self.assertFalse("17890714" in _20century)
252 _ac = DateRange("00010101")
253 self.assertTrue("19690721" in _ac)
254 _firstmilenium = DateRange(end="10000101")
255 self.assertTrue("07110427" in _firstmilenium)
257 def test_unified_dates(self):
258 self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
259 self.assertEqual(unified_strdate('8/7/2009'), '20090708')
260 self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
261 self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
262 self.assertEqual(unified_strdate('1968 12 10'), '19681210')
263 self.assertEqual(unified_strdate('1968-12-10'), '19681210')
264 self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
266 unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
269 unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
271 self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
272 self.assertEqual(unified_strdate('25-09-2014'), '20140925')
273 self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
275 def test_determine_ext(self):
276 self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
277 self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
278 self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
279 self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
280 self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
282 def test_find_xpath_attr(self):
290 doc = compat_etree_fromstring(testxml)
292 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
293 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
294 self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
295 self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
296 self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
297 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
298 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
299 self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
300 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
301 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
302 self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
304 def test_xpath_with_ns(self):
305 testxml = '''<root xmlns:media="http://example.com/">
307 <media:author>The Author</media:author>
308 <url>http://server.com/download.mp3</url>
311 doc = compat_etree_fromstring(testxml)
312 find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
313 self.assertTrue(find('media:song') is not None)
314 self.assertEqual(find('media:song/media:author').text, 'The Author')
315 self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
317 def test_xpath_element(self):
318 doc = xml.etree.ElementTree.Element('root')
319 div = xml.etree.ElementTree.SubElement(doc, 'div')
320 p = xml.etree.ElementTree.SubElement(div, 'p')
322 self.assertEqual(xpath_element(doc, 'div/p'), p)
323 self.assertEqual(xpath_element(doc, ['div/p']), p)
324 self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
325 self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
326 self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
327 self.assertTrue(xpath_element(doc, 'div/bar') is None)
328 self.assertTrue(xpath_element(doc, ['div/bar']) is None)
329 self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
330 self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
331 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
332 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
334 def test_xpath_text(self):
340 doc = compat_etree_fromstring(testxml)
341 self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
342 self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
343 self.assertTrue(xpath_text(doc, 'div/bar') is None)
344 self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
346 def test_xpath_attr(self):
352 doc = compat_etree_fromstring(testxml)
353 self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
354 self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
355 self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
356 self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
357 self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
358 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
359 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
361 def test_smuggle_url(self):
362 data = {"ö": "ö", "abc": [3]}
363 url = 'https://foo.bar/baz?x=y#a'
364 smug_url = smuggle_url(url, data)
365 unsmug_url, unsmug_data = unsmuggle_url(smug_url)
366 self.assertEqual(url, unsmug_url)
367 self.assertEqual(data, unsmug_data)
369 res_url, res_data = unsmuggle_url(url)
370 self.assertEqual(res_url, url)
371 self.assertEqual(res_data, None)
373 def test_shell_quote(self):
374 args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
375 self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
377 def test_str_to_int(self):
378 self.assertEqual(str_to_int('123,456'), 123456)
379 self.assertEqual(str_to_int('123.456'), 123456)
381 def test_url_basename(self):
382 self.assertEqual(url_basename('http://foo.de/'), '')
383 self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
384 self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
385 self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
386 self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
388 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
391 def test_parse_duration(self):
392 self.assertEqual(parse_duration(None), None)
393 self.assertEqual(parse_duration(False), None)
394 self.assertEqual(parse_duration('invalid'), None)
395 self.assertEqual(parse_duration('1'), 1)
396 self.assertEqual(parse_duration('1337:12'), 80232)
397 self.assertEqual(parse_duration('9:12:43'), 33163)
398 self.assertEqual(parse_duration('12:00'), 720)
399 self.assertEqual(parse_duration('00:01:01'), 61)
400 self.assertEqual(parse_duration('x:y'), None)
401 self.assertEqual(parse_duration('3h11m53s'), 11513)
402 self.assertEqual(parse_duration('3h 11m 53s'), 11513)
403 self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
404 self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
405 self.assertEqual(parse_duration('62m45s'), 3765)
406 self.assertEqual(parse_duration('6m59s'), 419)
407 self.assertEqual(parse_duration('49s'), 49)
408 self.assertEqual(parse_duration('0h0m0s'), 0)
409 self.assertEqual(parse_duration('0m0s'), 0)
410 self.assertEqual(parse_duration('0s'), 0)
411 self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
412 self.assertEqual(parse_duration('T30M38S'), 1838)
413 self.assertEqual(parse_duration('5 s'), 5)
414 self.assertEqual(parse_duration('3 min'), 180)
415 self.assertEqual(parse_duration('2.5 hours'), 9000)
416 self.assertEqual(parse_duration('02:03:04'), 7384)
417 self.assertEqual(parse_duration('01:02:03:04'), 93784)
418 self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
419 self.assertEqual(parse_duration('87 Min.'), 5220)
420 self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
422 def test_fix_xml_ampersands(self):
424 fix_xml_ampersands('"&x=y&z=a'), '"&x=y&z=a')
426 fix_xml_ampersands('"&x=y&wrong;&z=a'),
427 '"&x=y&wrong;&z=a')
429 fix_xml_ampersands('&'><"'),
430 '&'><"')
432 fix_xml_ampersands('Ӓ᪼'), 'Ӓ᪼')
433 self.assertEqual(fix_xml_ampersands('&#&#'), '&#&#')
435 def test_paged_list(self):
436 def testPL(size, pagesize, sliceargs, expected):
437 def get_page(pagenum):
438 firstid = pagenum * pagesize
439 upto = min(size, pagenum * pagesize + pagesize)
440 for i in range(firstid, upto):
443 pl = OnDemandPagedList(get_page, pagesize)
444 got = pl.getslice(*sliceargs)
445 self.assertEqual(got, expected)
447 iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
448 got = iapl.getslice(*sliceargs)
449 self.assertEqual(got, expected)
451 testPL(5, 2, (), [0, 1, 2, 3, 4])
452 testPL(5, 2, (1,), [1, 2, 3, 4])
453 testPL(5, 2, (2,), [2, 3, 4])
454 testPL(5, 2, (4,), [4])
455 testPL(5, 2, (0, 3), [0, 1, 2])
456 testPL(5, 2, (1, 4), [1, 2, 3])
457 testPL(5, 2, (2, 99), [2, 3, 4])
458 testPL(5, 2, (20, 99), [])
460 def test_struct_unpack(self):
461 self.assertEqual(struct_unpack('!B', b'\x00'), (0,))
463 def test_read_batch_urls(self):
464 f = io.StringIO('''\xef\xbb\xbf foo
467 # More after this line\r
470 self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
472 def test_urlencode_postdata(self):
473 data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
474 self.assertTrue(isinstance(data, bytes))
476 def test_update_url_query(self):
478 return compat_parse_qs(compat_urlparse.urlparse(url).query)
479 self.assertEqual(query_dict(update_url_query(
480 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
481 query_dict('http://example.com/path?quality=HD&format=mp4'))
482 self.assertEqual(query_dict(update_url_query(
483 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
484 query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
485 self.assertEqual(query_dict(update_url_query(
486 'http://example.com/path', {'fields': 'id,formats,subtitles'})),
487 query_dict('http://example.com/path?fields=id,formats,subtitles'))
488 self.assertEqual(query_dict(update_url_query(
489 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
490 query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
491 self.assertEqual(query_dict(update_url_query(
492 'http://example.com/path?manifest=f4m', {'manifest': []})),
493 query_dict('http://example.com/path'))
494 self.assertEqual(query_dict(update_url_query(
495 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
496 query_dict('http://example.com/path?system=LINUX'))
497 self.assertEqual(query_dict(update_url_query(
498 'http://example.com/path', {'fields': b'id,formats,subtitles'})),
499 query_dict('http://example.com/path?fields=id,formats,subtitles'))
500 self.assertEqual(query_dict(update_url_query(
501 'http://example.com/path', {'width': 1080, 'height': 720})),
502 query_dict('http://example.com/path?width=1080&height=720'))
503 self.assertEqual(query_dict(update_url_query(
504 'http://example.com/path', {'bitrate': 5020.43})),
505 query_dict('http://example.com/path?bitrate=5020.43'))
506 self.assertEqual(query_dict(update_url_query(
507 'http://example.com/path', {'test': '第二行тест'})),
508 query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
510 def test_dict_get(self):
518 d = FALSE_VALUES.copy()
520 self.assertEqual(dict_get(d, 'a'), 42)
521 self.assertEqual(dict_get(d, 'b'), None)
522 self.assertEqual(dict_get(d, 'b', 42), 42)
523 self.assertEqual(dict_get(d, ('a', )), 42)
524 self.assertEqual(dict_get(d, ('b', 'a', )), 42)
525 self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
526 self.assertEqual(dict_get(d, ('b', 'c', )), None)
527 self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
528 for key, false_value in FALSE_VALUES.items():
529 self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
530 self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
532 def test_encode_compat_str(self):
533 self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
534 self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
536 def test_parse_iso8601(self):
537 self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
538 self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
539 self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
540 self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
541 self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
542 self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
544 def test_strip_jsonp(self):
545 stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
546 d = json.loads(stripped)
547 self.assertEqual(d, [{"id": "532cb", "x": 3}])
549 stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
550 d = json.loads(stripped)
551 self.assertEqual(d, {'STATUS': 'OK'})
553 stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
554 d = json.loads(stripped)
555 self.assertEqual(d, {'status': 'success'})
557 def test_uppercase_escape(self):
558 self.assertEqual(uppercase_escape('aä'), 'aä')
559 self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
561 def test_lowercase_escape(self):
562 self.assertEqual(lowercase_escape('aä'), 'aä')
563 self.assertEqual(lowercase_escape('\\u0026'), '&')
565 def test_limit_length(self):
566 self.assertEqual(limit_length(None, 12), None)
567 self.assertEqual(limit_length('foo', 12), 'foo')
569 limit_length('foo bar baz asd', 12).startswith('foo bar'))
570 self.assertTrue('...' in limit_length('foo bar baz asd', 12))
572 def test_escape_rfc3986(self):
573 reserved = "!*'();:@&=+$,/?#[]"
574 unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
575 self.assertEqual(escape_rfc3986(reserved), reserved)
576 self.assertEqual(escape_rfc3986(unreserved), unreserved)
577 self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
578 self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
579 self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
580 self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
582 def test_escape_url(self):
584 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
585 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
588 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
589 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
592 escape_url('http://тест.рф/фрагмент'),
593 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
596 escape_url('http://тест.рф/абв?абв=абв#абв'),
597 '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'
599 self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
601 def test_js_to_json_realworld(self):
603 'clip':{'provider':'pseudo'}
605 self.assertEqual(js_to_json(inp), '''{
606 "clip":{"provider":"pseudo"}
608 json.loads(js_to_json(inp))
611 'playlist':[{'controls':{'all':null}}]
613 self.assertEqual(js_to_json(inp), '''{
614 "playlist":[{"controls":{"all":null}}]
617 inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
618 self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
620 inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
621 json_code = js_to_json(inp)
622 self.assertEqual(json.loads(json_code), json.loads(inp))
624 def test_js_to_json_edgecases(self):
625 on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
626 self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
628 on = js_to_json('{"abc": true}')
629 self.assertEqual(json.loads(on), {'abc': True})
631 # Ignore JavaScript code as well
638 self.assertEqual(d['x'], 1)
639 self.assertEqual(d['y'], 'a')
641 on = js_to_json('["abc", "def",]')
642 self.assertEqual(json.loads(on), ['abc', 'def'])
644 on = js_to_json('{"abc": "def",}')
645 self.assertEqual(json.loads(on), {'abc': 'def'})
647 def test_extract_attributes(self):
648 self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
649 self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
650 self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
651 self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
652 self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
653 self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
654 self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
655 self.assertEqual(extract_attributes('<e x="&">'), {'x': '&'}) # XML
656 self.assertEqual(extract_attributes('<e x=""">'), {'x': '"'})
657 self.assertEqual(extract_attributes('<e x="£">'), {'x': '£'}) # HTML 3.2
658 self.assertEqual(extract_attributes('<e x="λ">'), {'x': 'λ'}) # HTML 4.0
659 self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
660 self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
661 self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
662 self.assertEqual(extract_attributes('<e x >'), {'x': None})
663 self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
664 self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
665 self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
666 self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
667 self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
668 self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
669 self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
670 self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}) # Names lowercased
671 self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
672 self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
673 self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
674 self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
675 self.assertEqual(extract_attributes('<e x="décomposé">'), {'x': 'décompose\u0301'})
676 # "Narrow" Python builds don't support unicode code points outside BMP.
679 supports_outside_bmp = True
681 supports_outside_bmp = False
682 if supports_outside_bmp:
683 self.assertEqual(extract_attributes('<e x="Smile 😀!">'), {'x': 'Smile \U0001f600!'})
685 def test_clean_html(self):
686 self.assertEqual(clean_html('a:\nb'), 'a: b')
687 self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
689 def test_intlist_to_bytes(self):
691 intlist_to_bytes([0, 1, 127, 128, 255]),
692 b'\x00\x01\x7f\x80\xff')
694 def test_args_to_str(self):
696 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
697 'foo ba/r -baz \'2 be\' \'\''
700 def test_parse_filesize(self):
701 self.assertEqual(parse_filesize(None), None)
702 self.assertEqual(parse_filesize(''), None)
703 self.assertEqual(parse_filesize('91 B'), 91)
704 self.assertEqual(parse_filesize('foobar'), None)
705 self.assertEqual(parse_filesize('2 MiB'), 2097152)
706 self.assertEqual(parse_filesize('5 GB'), 5000000000)
707 self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
708 self.assertEqual(parse_filesize('1,24 KB'), 1240)
710 def test_parse_count(self):
711 self.assertEqual(parse_count(None), None)
712 self.assertEqual(parse_count(''), None)
713 self.assertEqual(parse_count('0'), 0)
714 self.assertEqual(parse_count('1000'), 1000)
715 self.assertEqual(parse_count('1.000'), 1000)
716 self.assertEqual(parse_count('1.1k'), 1100)
717 self.assertEqual(parse_count('1.1kk'), 1100000)
718 self.assertEqual(parse_count('1.1kk '), 1100000)
719 self.assertEqual(parse_count('1.1kk views'), 1100000)
721 def test_version_tuple(self):
722 self.assertEqual(version_tuple('1'), (1,))
723 self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
724 self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
726 def test_detect_exe_version(self):
727 self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
728 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
729 configuration: --prefix=/usr --extra-'''), '1.2.1')
730 self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
731 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
732 self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
733 Trying to open render node...
734 Success at /dev/dri/renderD128.
735 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
737 def test_age_restricted(self):
738 self.assertFalse(age_restricted(None, 10)) # unrestricted content
739 self.assertFalse(age_restricted(1, None)) # unrestricted policy
740 self.assertFalse(age_restricted(8, 10))
741 self.assertTrue(age_restricted(18, 14))
742 self.assertFalse(age_restricted(18, 18))
744 def test_is_html(self):
745 self.assertFalse(is_html(b'\x49\x44\x43<html'))
746 self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
747 self.assertTrue(is_html( # UTF-8 with BOM
748 b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
749 self.assertTrue(is_html( # UTF-16-LE
750 b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
752 self.assertTrue(is_html( # UTF-16-BE
753 b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
755 self.assertTrue(is_html( # UTF-32-BE
756 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'))
757 self.assertTrue(is_html( # UTF-32-LE
758 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'))
760 def test_render_table(self):
764 [[123, 4], [9999, 51]]),
769 def test_match_str(self):
770 self.assertRaises(ValueError, match_str, 'xy>foobar', {})
771 self.assertFalse(match_str('xy', {'x': 1200}))
772 self.assertTrue(match_str('!xy', {'x': 1200}))
773 self.assertTrue(match_str('x', {'x': 1200}))
774 self.assertFalse(match_str('!x', {'x': 1200}))
775 self.assertTrue(match_str('x', {'x': 0}))
776 self.assertFalse(match_str('x>0', {'x': 0}))
777 self.assertFalse(match_str('x>0', {}))
778 self.assertTrue(match_str('x>?0', {}))
779 self.assertTrue(match_str('x>1K', {'x': 1200}))
780 self.assertFalse(match_str('x>2K', {'x': 1200}))
781 self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
782 self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
783 self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
784 self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
785 self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
786 self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
787 self.assertFalse(match_str(
788 'like_count > 100 & dislike_count <? 50 & description',
789 {'like_count': 90, 'description': 'foo'}))
790 self.assertTrue(match_str(
791 'like_count > 100 & dislike_count <? 50 & description',
792 {'like_count': 190, 'description': 'foo'}))
793 self.assertFalse(match_str(
794 'like_count > 100 & dislike_count <? 50 & description',
795 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
796 self.assertFalse(match_str(
797 'like_count > 100 & dislike_count <? 50 & description',
798 {'like_count': 190, 'dislike_count': 10}))
800 def test_parse_dfxp_time_expr(self):
801 self.assertEqual(parse_dfxp_time_expr(None), None)
802 self.assertEqual(parse_dfxp_time_expr(''), None)
803 self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
804 self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
805 self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
806 self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
807 self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
809 def test_dfxp2srt(self):
810 dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
811 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
814 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
815 <p begin="1" end="2">第二行<br/>♪♪</p>
816 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
817 <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
818 <p begin="-1" end="-1">Ignore, two</p>
819 <p begin="3" dur="-1">Ignored, three</p>
824 00:00:00,000 --> 00:00:01,000
825 The following line contains Chinese characters and special symbols
828 00:00:01,000 --> 00:00:02,000
833 00:00:02,000 --> 00:00:03,000
838 self.assertEqual(dfxp2srt(dfxp_data), srt_data)
840 dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
841 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
844 <p begin="0" end="1">The first line</p>
849 00:00:00,000 --> 00:00:01,000
853 self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
855 def test_cli_option(self):
856 self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
857 self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
858 self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
860 def test_cli_valueless_option(self):
861 self.assertEqual(cli_valueless_option(
862 {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
863 self.assertEqual(cli_valueless_option(
864 {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
865 self.assertEqual(cli_valueless_option(
866 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
867 self.assertEqual(cli_valueless_option(
868 {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
869 self.assertEqual(cli_valueless_option(
870 {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
871 self.assertEqual(cli_valueless_option(
872 {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
874 def test_cli_bool_option(self):
877 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
878 ['--no-check-certificate', 'true'])
881 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
882 ['--no-check-certificate=true'])
885 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
886 ['--check-certificate', 'false'])
889 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
890 ['--check-certificate=false'])
893 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
894 ['--check-certificate', 'true'])
897 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
898 ['--check-certificate=true'])
900 def test_ohdave_rsa_encrypt(self):
901 N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
905 ohdave_rsa_encrypt(b'aa111222', e, N),
906 '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
908 def test_encode_base_n(self):
909 self.assertEqual(encode_base_n(0, 30), '0')
910 self.assertEqual(encode_base_n(80, 30), '2k')
912 custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
913 self.assertEqual(encode_base_n(0, 30, custom_table), '9')
914 self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
916 self.assertRaises(ValueError, encode_base_n, 0, 70)
917 self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
919 if __name__ == '__main__':